Stereochemistry of hexenyl radical cyclizations with tert-butyl and related large groups: Substituent and temperature effects

Article abstract of DOI:10.1021/ja042595u

The long held notion that hexenyl radicals bearing large substituents on the radical carbon cyclize to give 1,2-trans-substituted cyclopentanes is experimentally disproved by study of the radical cyclization of an assortment of simple and complex substrat

Full text of DOI:10.1021/ja042595u

Published on Web 03/25/2005
Stereochemistry of Hexenyl Radical Cyclizations with
tert-Butyl and Related Large Groups: Substituent and
Temperature Effects
Jonathan C. Tripp,† Carl H. Schiesser,‡ and Dennis P. Curran*,†
Contribution from the Department of Chemistry, UniVersity of Pittsburgh,
Pittsburgh, PennsylVania 15260, and School of Chemistry, Bio21 Molecular Science and
Biotechnology Institute, UniVersity of Melbourne, Victoria, Australia 3010
Received December 9, 2004; E-mail: curran@pitt.edu
Abstract: The long held notion that hexenyl radicals bearing large substituents on the radical carbon cyclize
to give 1,2-trans-substituted cyclopentanes is experimentally disproved by study of the radical cyclization
of an assortment of simple and complex substrates coupled with careful product analysis and rigorous
assignment of configurations. X-ray studies and syntheses of authentic samples establish that the published
assignments for cis- and trans-1-tert-butyl-2-methylcyclopentane must be reversed. The original assignment
based on catalytic hydrogenation of 1-tert-butyl-2-methylenecyclopentane was compromised by migration
of the double bond prior to hydrogenation. The cyclization of 1-tert-butylhexenyl radical is moderately cis
selective, and the selectivity is increased by geminal substitution on carbon 3. This selectivity trend is
general and extends to relatively complex substrates. It has allowed Ihara to reduce the complexity of an
important class of round trip radical cyclizations to make linear triquinanes to the point where two tricyclic
productsscis-syn-cis and cis-anti-cissaccount for about 80% of the products. However, the further increase
in selectivity that was proposed by lowering the temperature is shown to be an artifact of the analysis
methods and is not correct. This work solidifies “1,2-cis selectivity” in cyclizations of 1-subsituted hexenyl
radicals as one of the most general stereochemical trends in radical cyclizations.
Introduction
of polyquinanes from acyclic precursors by a “round trip radical
cyclization”7 strategy. Generation of the radical from 1 can be
Cascade radical cyclizations are powerful reactions for
making condensed five-membered carbo- and heterocyclic
rings.1 For reasons of stereocontrol, tricyclic and higher rings
are usually constructed by fashioning new rings about at least
one preexisting ring.2 However, strategies for making polycycles
from acyclic precursors3 are especially direct and would be
appealing if stereochemical issues could be resolved.4
Our model experiments toward the crinipellin class of
tetraquinanes5 summarized in Figure 16 are illustrative of both
the power and the problems associated with a direct synthesis
followed by zero, one, two, or three radical cyclizations, with
the third cyclization completing the “round trip”. In the
experiment, six major products were formed, including four
tricyclic products (2a,b; 3a,b) and two bicyclic ones (4a,b).
Due to the complexity of the mixture, these products could not
be isolated individually and were instead characterized by GC-
MS and 13C NMR spectroscopy on isotopically labeled samples.6
All six of the products of this reaction arise as a consequence
of the lack of stereoselectivity in the second cyclization (6 f
7). Four stereoisomers of 7 are possible because of face
selectivity with respect to the existing stereocenter and simple
diastereoselection (cis/trans selectivity). The cis-isomers of 7
close rapidly to form less strained (cis-syn-cis and cis-anti-cis)
tricycloundecanes 2a,b, while trans isomers of 7 close more
slowly to give more strained trans-fused tricycloundecanes 3a,b.
Direct reduction of 7-trans by tin hydride competes with these
† University of Pittsburgh.
‡ University of Melbourne. Direct inquires regarding calculations to this
author.
(1) (a) Dhimane, A. L.; Fensterbank, L.; Malacria, M. In Radicals in Organic
Synthesis; Renard, P., Sibi, M., Eds.; Wiley-VCH: Weinheim, 2001; Vol.
2, pp 350-382. (b) A¨ıssa, C.; Delouvrie, B.; Dhimane´, A. L.; Fensterbank,
L.; Malacria, M. Pure Appl. Chem. 2000, 72, 1605-1613. (c) Jasperse, C.
P.; Curran, D. P.; Fevig, T. L. Chem. ReV. 1991, 91, 1237-1286. (d) Curran,
D. P. In ComprehensiVe Organic Synthesis; Trost, B. M., Fleming, I., Eds.;
Pergamon: Oxford, U.K., 1991; Vol. 4, pp 715-831.
(2) Curran, D. P. In AdVances in Free Radical Chemisty; Tanner, D. D., Ed.;
JAI: Greewich, CT, 1990; Vol. 1, pp 121-157.
(5) (a) Piers, E.; Renaud, J.; Rettig, S. J. Synthesis 1998, 590-602. (b) Mehta,
G.; Rao, K. S.; Reddy, M. S. J. Chem. Soc., Perkin Trans. 1 1991, 693-
700. (c) Schwartz, C. E.; Curran, D. P. J. Am. Chem. Soc. 1990, 112, 9272-
9284.
(6) Curran, D. P.; Sun, S. Aust. J. Chem. 1995, 48, 261-267.
(7) Branchaud, B. P.; Glenn, A. G.; Stiasny, H. C. J. Org. Chem. 1991, 56,
6656-6659.
(3) Beckwith, A. L. J.; Roberts, D. H. J. Am. Chem. Soc. 1986, 108, 5893-
5901.
(4) Review on stereochemistry of radical cyclizations: Curran, D. P.; Porter,
N. A.; Giese, B. Stereochemistry of Radical Reactions: Concepts,
Guidelines, and Synthetic Applications; VCH: Weinheim, Germany, 1996;
Chapter 2, pp 23-115.
9
5518
J. AM. CHEM. SOC. 2005, 127, 5518-5527
10.1021/ja042595u CCC: $30.25 © 2005 American Chemical Society

Stereochemistry of Hexenyl Radical Cyclizations
A R T I C L E S
Figure 2. Comparison of radicals in the second cyclization with each other
(top) and with literature benchmarks (bottom).
trans partitioning, 100/0). Similar temperature effects were
reported with tris(trimethylsilyl)silane.
Figure 1. Round trip radical cyclization generating up to three rings but
with multiple isomers formed.
cyclizations, so each trans radical gives one tricyclic (3) and
one bicyclic (4) product in a ratio that depends on the tin hydride
concentration. About 67% of the radicals partition through cis
cyclization pathways (to give 2a,b) and 33% partition through
trans cyclization pathways (to give 3a,b and 4a,b). Clearly,
methods are needed to control both face selectivity and cis/
trans selectivity in these types of reactions.
In 2001, Takasu, Ihara, and co-workers reported what
appeared to be a method to control the cis/trans selectivity in
related cyclizations through substituent and temperature effects
(eq 1).8 Cyclization of substrate 8a under standard conditions
with tributyltin hydride at 80 °C was reported to form cis-syn-
cis isomer 9a, cis-anti-cis isomer 10a, and “other diastereomers”
in a ratio of 36/28/36. While the structure of 9a was fully
assigned, the configuration of the ester-bearing center (C7) in
10a was not assigned, and the structures of the “other diaster-
eomers” were not discussed. By analogy to the cyclization of
1, we considered the possibility that the other diastereomers
were trans-isomers 11a (if so, then the cis/trans partitioning at
80 °C is 64/36). Alternatively, the other diastereomers could
also be stereoisomers of 9a and 10a at C7. Intriguingly,
conducting the cyclization at room temperature was reported
to result in the disappearance of the other diastereomers and
provide only 9a and 10a in 83% yield in a 57/43 ratio (cis/
Direct comparison of Ihara’s results8 with ours6 is complicated
because the substrates are materially different in three ways,
all of which could affect the key second cyclization. Radical
12 (Figure 2) has a malonate in the connecting chain (E ) CO2-
Me), a very large (tert-butyl-like) substituent on the radical (R
) 1-(1-methylcyclopent-2-enyl)), and an ester on the radical
acceptor (E′ ) CO2Me). In comparison, E and E′ are H in the
parent radical 6 and R is a secondary group and not a tertiary
one. While the combined effect of these non-hydrogen substit-
uents is difficult to access, a number of benchmarks in simpler
systems are available. There are many examples of radical
cyclizations with malonates in the connecting chain,9 and these
are accelerated by the Thorpe-Ingold effect.10 Typical cycliza-
tions of radicals bearing small or medium groups on the radical
carbon have low to moderate cis-selectivity. This selectivity is
increased by the presence of a malonate, as shown by examples
13a,b in Figure 2.11 The presence of an ester (E′) on the radical
(8) Takasu, K.; Maiti, S.; Katsumata, A.; Ihara, M. Tetrahedron Lett. 2001,
42, 2157-2160.
(9) (a) De Riggi, I.; Nouguier, R.; Surzur, J. M.; Lesueur, C.; Bertrand, M.;
Jaime, C.; Virgili, A. Bull. Soc. Chim. Fr. 1993, 130, 229-235. (b)
Feldman, K. S.; Berven, H. M.; Weinreb, P. H. J. Am. Chem. Soc. 1993,
115, 11364-11369.
(10) Jung, M. E. Synlett 1999, 843-846.
(11) (a) Curran, D. P.; Shen, W. J. Am. Chem. Soc. 1993, 115, 6051-6059. (b)
Robertson, J.; Pillai, J.; Lush, R. K. Chem. Soc. ReV. 2001, 30, 94-103.
(c) Beaufils, F.; De´ne`s, F.; Renaud, P. Org. Lett. 2004, 6, 2563-2566.
9
J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005 5519

A R T I C L E S
Tripp et al.
alkylations, and details of their syntheses and the associated
characterization data are contained in the Supporting Informa-
tion.
Substrates 15 and 16a are benchmarks to assess the effect of
the malonate on cyclization of a tert-butyl-substituted hexenyl
radical, while substrate 16b allows a structural relay between
products of the prior two cyclizations for rigorous assignment
of configuration. For synthetic convenience, we rely on radical
translocation (1,5-hydrogen atom transfer) to generate the radical
of interest in this series. Substrate 17 is very closely related to
Ihara’ssit lacks only a crucial double bond of the vinyl iodide
that prevents the round trip from being completed. This greatly
simplifies the analysis because no tricyclic products are possible,
and there are only two bicyclic products (cis and trans). Substrate
8a is Ihara’s precursor, while 18 is the analogue lacking the
ester on the radical acceptor (E′ ) H) and benzyl ester 8b is
the structural relay substrate in this series. Finally, Z-unsaturated
esters have been shown to give higher stereoselectivity in radical
cyclizations than E-isomers.4,14 So to assess the effect of alkene
geometry, we also prepared 8c, which is the Z R,â-unsaturated
ester isomer of 8a.
Cyclizations of Benchmark Substrates. The cyclizations of
the benchmark substrates were all conducted under similar
conditions, and the results of these cyclizations are summarized
in Table 1. For reactions at 80 °C, a mixture of the cyclization
precursor (0.1 mmol), tributyltin hydride (0.12 mmol), and
AIBN (0.05 mmol) in benzene (50 mL, 2 mM with respect to
the precursor) was heated at reflux for 4 h. After cooling and
solvent evaporation, the product mixture was subjected to rapid
chromatography to remove most of the tin and provide a crude
product mixture. This mixture was subjected to GC and NMR
analysis to determine product ratios. The two methods gave very
similar ratios, and only the GC results are recorded in the table.
Combined isolated yields of cis/trans mixtures of cyclic products
were obtained by careful chromatography of the crude mixture.
Room-temperature (25 °C) reactions were conducted by a
similar procedure with initiation by Et3B15 (0.05 mmol) in place
of AIBN.
Each of the four benchmark substrates in Table 1 exhibited
a moderate level of selectivity in favor of the cis stereoisomer
at 80 °C (76/24-80/20, see entries 2, 4, 6, and 8). The directly
reduced product (not shown) was not observed in the reductions
of 15, 16a, or 16b, but it was observed in small amounts (∼5%)
by GC in the reduction of 17. Reactions conducted at room
temperature resulted in detectable but small increases in
selectivity (1-5%) for the cis isomer (entries 3, 5, 7). We
conclude that the temperature dependence of these cyclizations
is not very significant. The level of selectivity (about (3-4)/1)
is strikingly similar to that of the methyl-substituted malonyl
radical 13b (5/1; see Figure 2) despite the presence of the large
groups on the radical-bearing carbon.
Figure 3. Radical cyclization substrates.
acceptor accelerates the cyclization,12 but it is not expected to
dramatically alter the stereoselectivity.
Beckwith and co-workers reported that the cyclization of the
tert-butyl-substituted hexenyl radical 14 (Figure 2) is an
exception to the broad trend of 1,2-cis selectivity and provides
the trans isomer as the major product.13 Intriguingly, they also
reported that there was a significant temperature dependence
on the cyclization with cis/trans ratios changing from 47/53 at
90 °C to 7/93 at -33 °C. If one momentarily accepts the
assumption that Ihara’s “other diastereomers” are trans isomers,
then both Ihara’s and Beckwith’s radicals with related large
substituents exhibit powerful temperature dependences but for
opposite stereoisomers.
To unravel these puzzling temperature and substituent effects,
we have conducted a series of tricyclizations related to Ihara’s
and extended the study of benchmark systems related to
Beckwith’s. During this study, we have identified problems with
the prior work with respect to configuration assignment and
product analysis. We now report that Ihara’s radical cyclization
is cis selective but does not show a significant temperature effect
while the cyclization of the 1-tert-butylhexenyl radical 14 is
also cis-selective and not trans-selective as reported. These
revised results and assignments along with our new results
provide a more comprehensive understanding of how very large
substituents affect stereoselectivity in radical cyclizations, both
alone and in combination with other substituents.
Assignment of Configurations of Benchmark Products.
We were not able to preparatively separate the cis and trans
isomers of any of the benchmark products in Table 1, and
overlapping of resonances in the 1H NMR spectra coupled with
concerns about ambiguities in interpretations of NOE experi-
ments led us to pursue a rigorous approach to configuration
Results
Substrate Selection and Synthesis. Figure 3 shows the
substrates used for radical cyclizations in this study. These
compounds were all prepared by straightforward malonate
(12) Giese, B.; Kopping, B.; Go¨bel, T.; Dickhaut, J.; Thoma, G.; Kulicke, K.
J.; Trach, F. In Organic Reactions; Wiley: New York, 1996; Vol. 48, pp
301-856.
(13) Beckwith, A. L. J.; Cliff, M. D.; Schiesser, C. H. Tetrahedron 1992, 48,
4641-4648.
(14) Wilcox, C. S.; Thomasco, L. M. J. Org. Chem. 1985, 50, 546-547.
(15) Yorimitsu, H.; Oshima, K. In Radicals in Organic Synthesis, 1st ed.;
Renaud, P., Sibi, M. P., Eds.; Wiley-VCH: Weinheim, Germany, 2001;
Vol. 1, pp 11-27.
9
5520 J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005

Stereochemistry of Hexenyl Radical Cyclizations
A R T I C L E S
Table 1. Radical Cyclization Products from Benchmark Substrates
entry
precursor
R
E
′
T (°
C)a
prod
cis/transb
% yieldc
1
2
3
4
5
6
7
8
15
15
CH3
CH3
CH3
CH3
CH3
CH3
H
H
25
80
25
80
25
80
25
80
19
19
76/24
83/17
78/22
79/21
77/23
81/19
80/20
54
45
52
35
67
62
69
16a
16a
16b
16b
17
CO2Me
CO2Me
CO2(CH2)2TMS
CO2(CH2)2TMS
CO2Me
20a
20a
20b
20b
21
-(CH2)4-
-(CH2)4-
17
CO2Me
21
a Initiation at 80 °C with AIBN and at 25 °C with Et3B/O2. b Raw GC ratio. c Isolated yield of a cis/trans mixture after flash chromatography.
assignment by correlation. Equation 2 shows the synthesis of
an authentic sample of 19-trans. Reductive addition of tert-
butyl bromide to diethyl mesaconate 22 mediated by vitamin
B12a and Zn provided known syn-diester 23 (96/4 ratio, syn/
anti) in 46% isolated yield.16 Reduction of the diester to the
diol, bis-mesylation, Finkelstein reaction to make the diiodide
24, and finally double malonate alkylation provided an authentic
sample of 19-trans. This was identical by NMR spectroscopy
and GC co-injection to the minor product from the cyclization
of 15 (Table 1, entry 2). To rigorously confirm the configuration
assignment from the radical addition, we reduced diester 19-
trans to the diol 25. This was crystallized by slow vapor-phase
diffusion of hexane into an ethyl acetate solution of 25, and
the crystal structure was solved (see Supporting Information).
With the configuration of 19 secure, we assigned configura-
(trimethylsilyl)diazomethane ((TMS)CHN2) gave a 77/23 mix-
ture of 20a-cis/trans. This proves that the major products in
the cyclizations of all three of these substrates are cis. Accord-
ingly, we assigned cis configuration to the major product 21-
cis of the interrupted substrate 17 by analogy.
Comparing the results of Table 1 (cis selective) with those
reported for the parent tert-butyl-substituted radical 1413 (Figure
2, trans selective) suggests that the malonyl group enforces a
reversal of stereoselectivity. This conclusion is not evidently
inconsistent with literature observations in related malonyl
radicals. However, we were discomfited by a comparison of
our spectroscopic data with the published data of 28sthe
resonances assigned to 28-cis (eq 4) fit the pattern of our trans
isomers while those assigned to 28-trans fit the pattern of our
cis isomers.17 This lack of pattern matching suggested that the
assignments of 28-cis/trans might be incorrect.
tions to 20a-cis/trans with the relay products 20b. Desilylation
of a 77/23 mixture of 20b-cis/trans to give 26 followed by
Barton reductive decarboxylation provided a 77/23 mixture of
19-cis/trans (eq 3). Correspondingly, treatment of 26 with
(16) Erdmann, P.; Scha¨fer, J.; Springer, R.; Zeitz, H.-G.; Giese, B. HelV. Chim.
Acta 1992, 75, 638-644.
9
J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005 5521

A R T I C L E S
Tripp et al.
yield (%)
Table 2. Product Ratios in Cyclizations of Triesters 8a,b
% composition
entry
substrate
reducing agent concn (mM)
reducing agent
T (°
C)
cis-syn-cis 9a
cis-anti-cis 10a
cis-anti-trans 29a
cis-syn-trans 30a
selectivity (cis/trans)
1
2
3
4
5
6
7
8
9
8a
8a
8a
8a
8a
8a
8b
8b
8cc
syringe pumpa
Bu3SnH
Bu3SnH
Bu3SnH
Bu3SnH
Ph3SnH
Ph3SnH
Bu3SnH
Bu3SnH
Bu3SnH
25
80
25
80
25
80
25
80
80
48
42
47
44
44
47
47
42
40
38
35
39
34
35
37
40
35
51
8
9
8
9
7
9
6
15
8
6
14
6
13
14
7
7
8
1
86/14
77/23
86/14
78/22
79/21
77/23
87/13
77/23
91/09
nd
nd
syringe pumpa
2
2
2
2
2
2
2
52, 69b
50, 70b
48
46
50
35
62
a A 24 mM solution of reducing agent is added to a 2 mM solution of the iodide over 3 h. b Yield determined using octadecane as an internal GC standard.
c Z-unsaturated ester isomer of 8a. Nd ) Not determined.
To revisit the assignment of the products of the radical
cyclization of the tert-butyl-substituted hexenyl radical 14, we
first prepared thionocarbonate 27 and cyclized it according to
the published procedure13 at 66 °C to provide a 55/45 ratio (GC
analysis) of stereoisomers 28 (eq 4). This result matches the
55/45 ratio reported at that temperature, and the matching of
spectral data shows that the major products are the same. Next,
hydrolysis and microwave-assisted decarboxylation of a 76/24
ratio of 19-cis/trans provided a mixture of mono acids in 28%
overall yield over two steps. This was not analyzed but was
directly subjected to Barton decarboxylation to provide a 76/
24 ratio of 28-cis/trans. The major products from both of these
exercises were identical, as were the minor ones. Accordingly,
we conclude that the literature assignments of 28-cis and 28-
trans must be reversed. This means that, like most other types
of radicals, the cyclization of the parent tert-butyl-substituted
hexenyl radical 14 is also moderately cis-selective (not trans
selective).
Cyclizations of Round Trip Radical Substrates. With a
good understanding of both new and existing benchmark
cyclizations, we next moved to careful study of the round trip
radical cyclization substrates 8a-c and 18. Our results for
cyclizations of Ihara’s substrate 8a under varying conditions
are shown in Figure 4 and Table 2. In a repeat of one of Ihara’s
key experiments, we reduced 8a with Bu3SnH at 80 °C (Figures
Figure 4. Structure assignments for cyclization products from Ihara’s
precursor 8a (E, E′ ) CO2Me).
4 and 5 and Table 2, entry 4). A relatively complex mixture
resulted. Flash chromatography was not helpful in separating
the mixture, but GC analysis provided a powerful tool.
After exposure of the crude reaction mixture to a standard
DBU workup18 to remove most of the tin, GC analysis of the
crude product provided the chromatogram shown in Figure 5
(top). In addition to the two major peaks at 8.1 and 8.4 min
(45% and 35%, respectively), there were at least six minor peaks
in the chromatogram, one of which (7.6 min, 3%) was readily
(17) After structural correction, the relative chemical shifts of the methyl and
tert-butyl resonances exhibit the same trends for 19 and 28: 28-cis, 0.93
(s, 9 H), 0.85 (d, 3 H, J ) 7.1 Hz); 28-trans 0.97 (d, 3 H, J ) 6.8 Hz),
0.84 (s, 9 H); 19-cis, 0.96 (s, 9 H), 0.91 (d, 3 H); 19-trans, 1.00 (d, 3 H,
J ) 7 Hz), 0.88 (s, 9 H). In the cis isomer, the methyl doublet resonates
upfield from the tert-butyl singlet, while in the trans isomer the positions
of these resonances are reversed.
9
5522 J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005

Stereochemistry of Hexenyl Radical Cyclizations
A R T I C L E S
cis-syn-cis isomer 9a and confirm Ihara’s structural proposal
by a combination of 2D NMR experiments to assign resonances
followed by NOE experiments. At the same time, we were able
to confirm that the peak at 8.4 min corresponds to the cis-anti-
cis isomer 10a by recording NMR spectra of the other prep
GC fraction enriched in that isomer.
The assignments of the two minor isomers as cis-anti-trans
29a and cis-syn-trans 30a are tentative because neither was
isolated in pure enough form for detailed characterization. We
considered the possibility that one or both of these compounds
could have no trans ring fusion and instead be stereoisomers of
9a and 10a at the ester-bearing C7; however, the relay
experiments with substrate 8b lacking that ester suggest that
this is not the case (see below).
With a good analytical method in hand, we then conducted
a series of cyclizations with Bu3SnH and Ph3SnH at 80 and 25
°C to determine the temperature dependence on the stereose-
lectivity in cyclizations of methyl ester 8a (Table 2, entries 1-6)
and benzyl ester 8b (entries 7, 8). Each crude product was
treated with mCPBA, chromatographed to remove alkene-
derived (epoxide) products, and then subjected to GC analysis.
Eight reactions were conducted under diverse conditions varying
concentration (syringe pump addition or fixed 0.0024 M in tin
hydride), reagent (Bu3SnH, Ph3SnH), and temperature (80 °C,
AIBN initiation; 25 °C, Et3B initiation). The ratios of the four
products formed in these experiments hardly varied at all as
shown in Table 2. Also shown are the cis/trans ratios of the
second cyclization calculated by dividing the sum of the two
major cis/cis products (structures firmly assigned) by the sum
of the two minor cis/trans products (structures tentatively
assigned). Consistent with the results in Table 1, but inconsistent
with the results of Ihara, there is a very small temperature
dependence on these ratios with the average cis/trans ratio being
78/22 at 80 °C and 86/14 at 25 °C. As a caveat, these ratios
may be slightly overstated since the minor products removed
in the mCPBA step are likely to result predominately or even
exclusively from trans cyclization. Because the total of these
products is relatively small (<10%) and because the cis/trans
selectivities are quite similar to the benchmark cases and are
not very concentration dependent, we conclude that this error
is small.
Figure 5. GC chromatograms of cyclization products from 8a (E, E′ )
CO2Me) before (a) and after (b) mCPBA treatment.
identified as the directly reduced product 31 (Figure 4). We
tentatively assigned the two major products as cis-syn-cis 9a
and cis-anti-cis 10a on the basis of Ihara’s work. We later
confirmed that assignment (see below) and showed that the peak
with the shorter retention time corresponded to the cis-syn-cis
isomer 9a. Proton NMR spectra of this crude product showed
small resonances in the alkene region in addition to those
belonging to the directly reduced product 31. We tentatively
attribute these resonances to monocyclized product 32 and
doubly cyclized products 33-trans in which the second and third
stages of the round trip journey were interrupted by tin hydride
reduction.
To remove the doubly cyclized and other alkene-containing
products from the tricyclic products (non-alkene-containing),
we exposed the crude product mixture to meta-chloroperoxy-
benzoic acid (mCPBA) with the expectation that alkene-
containing products would be epoxidized and thereby rendered
more polar. Flash chromatography of the resulting mixture then
provided a nonpolar fraction (52% isolated yield) whose NMR
spectrum exhibited no epoxide or alkene resonances and whose
GC chromatogram is shown in Figure 5 (bottom). Several of
the minor peaks are gone, and the remaining four peaks (two
major and two minor) all exhibited virtually identical mass
spectra in a GCMS experiment, suggesting that they are all
tricyclic.
Results for cyclization of the benzyl ester relay substrate at
25 and 80 °C are shown in Table 2, entries 7 and 8, and are
very similar to each other and to those of the methyl ester; the
products were assigned by relay (see below). Figures S1 and
S2 in the Supporting Information show GC chromatograms for
these products before and after mCPBA treatment. Finally,
cyclization of the Z-isomer 8c at 80 °C gave the same four
tricyclic products as E-isomer 8a but in an improved cis/trans
ratio of 91/9.
In short, we have confirmed the formation of Ihara’s two
major products and located two candidates for the “other
diastereomers”. But at the same time we have shown that there
is no large temperature dependence on stereoselectivity for this
reaction. These results call into question Ihara’s identification
of other diastereomers.
Ihara reported that the major product 9a from this cyclization
was partially separable from the others, and he kindly provided
spectra of this product. We were not able to preparatively
separate 9a by column chromatography; however, in a comple-
mentary experiment conducted in association with Prof. M.
Newcomb and R. E. P. Chandrasena, we were able to isolate
by preparative GC a fraction enriched to the level of 93% in
To study the effect of ester on the radical acceptor, we
conducted radical cyclizations of 18 in a manner similar to that
for 8a,b. As before, several minor peaks (<10% total) were
eliminated by treatment of the crude product with mCPBA (see
(18) Curran, D. P.; Chang, C.-T. J. Org. Chem. 1989, 54, 3140-3157.
9
J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005 5523

A R T I C L E S
Tripp et al.
Table 3. Product Ratios in Cyclizations of Diesters 18
% composition
entry
reducing agent concn (mM)
reducing agent
T (°
C)
cis-syn-cis 34
cis-anti-cis 35
cis-anti-trans 36
cis-syn-trans 37
selectivity (cis/trans)
yield (%)
1
2
3
4
5
6
7
8
9
syringe pumpa-c
Ph3SnH
Ph3SnH
25
80
25
80
25
80
25
80
25
80
43
40
44
42
40
40
43
41
47
45
39
36
39
38
42
37
38
41
36
38
15
19
11
14
13
19
12
11
13
14
3
5
6
6
5
4
7
7
6
3
82/18
76/24
83/17
80/20
82/18
77/23
81/19
82/18
83/17
83/17
nd
nd
51
52
nd
nd
nd
nd
nd
nd
syringe pumpa,b
syringe pumpa
Bu3SnH
Bu3SnH
Bu3SnH
Bu3SnH
(TMS)3SiH
(TMS)3SiH
Ph3SnH
syringe pumpa
2c
2
2
2
2
10
2
Ph3SnH
a A 24 mM solution of reducing agent is added to a 2 mM solution of the iodide over 3 h. b Ihara’s protocol. c Required the addition of 1 equiv of PPh3
to avoid exclusive formation of a directly reduced product.
GC chromatograms in Supporting Information, Figures S3 and
S4), leaving two major peaks and two minor peaks, whose ratios
are shown in Table 3. Again, extensive variation of concentra-
tion, reducing reagent (now including TTMSH), and temperature
had little effect on the product ratio.
The products from these two series of experiments were
correlated through the benzyl ester relay substrate 8b (eq 5).
Hydrogenation of a mixture of 9b/10b/29b/30b provided an acid
mixture, which was converted to the methyl ester by TMSCHN2
(not shown). GC injection of this sample with the product from
the radical cyclization of 8a showed the same four products in
about the same ratios. Barton decarboxylation of this acid
mixture (eq 5) provides a mixture of four products 34-37; the
three major products were identical by GC with those produced
from the cyclization of 18.19
Revisiting Prior Work To Identify Problems. We have
shown that a key structure assignment by Beckwith and co-
workers is incorrect13 and that a large temperature dependence
on a radical cyclization proposed by Takasu and Ihara8 does
not exist. As stated in the Introduction, it was easily possible at
the outset to accommodate the incorrect assignment and
Figure 6. Revisiting the stereochemical assignment of 28.
nonexistent temperature effect together within the framework
of existing results to make a plausible picture; indeed, this
accommodation was a premise of our work. Ironically, separate
accommodation of one error without the other seemed less
plausible (though certainly not impossible).
Beckwith used the cis/trans assignments of 28 to support two
sets of force field calculations (Houk/Spellmeyer20 and Beck-
with/Schiesser), both of which predicted a trans-selective
cyclization of radical 14. This calculated trans selectivity was
supported by hydrogenation and NOE experiments. In the
hydrogenation approach to configuration assignment, alkene 38
was reduced to give a 95/5 mixture of products (Figure 6). The
major product was assigned as cis by adopting the hypothesis
that hydrogenation occurs trans to the tert-butyl group. Molander
and Winterfeld have also posited cis-selectivities in related
(19) One minor product peak from this experiment did not match the GC peak
systems.21
for the very minor product 37 formed in the reduction of 18. This suggests
either that the structure of 37 might be wrong or that 10a and 11a might
To shed light on the hydrogenation stereoselectivity issue,
we prepared 38 and repeated its hydrogenation with palladium
be isomers at the ester bearing carbon rather than the ring fusion. Since
the structures of all these minor product are tentative and since 37 was
only formed in trace amounts, we decided not to pursue this issue further.
(20) (a) Spellmeyer, D. C.; Houk, K. N. J. Org. Chem. 1987, 52, 959-974. (b)
Beckwith, A. L. J.; Schiesser, C. H. Tetrahedron 1985, 41, 3925-3941.
(21) Molander, G.; Winterfeld, J. J. Organomet. Chem. 1996, 524, 275-279.
9
5524 J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005

Stereochemistry of Hexenyl Radical Cyclizations
A R T I C L E S
on carbon. Indeed, we ultimately observed a 95/5 ratio of 28-
trans/28-cis. So the literature ratio for this hydrogenation is
correct, but the assumption of cis selectivity is not. Further
insight was gained by following the time course of the
hydrogenation by GC. Alkene 38 disappeared quickly, and
alongside the isomers of 28 were formed two new major
products, which we tentatively assigned as isomerized structures
39 and 40. Isomerizations of hindered double bonds during
palladium-catalyzed hydrogenations are well-known.22 After
several hours, the peaks assigned to 39 and 40 disappeared and
the 95/5 ratio of 28-trans/cis resulted.
Tetrasubstituted alkene 39 is a known compound that was
independently synthesized23,24 and shown to have the same
retention time by co-injection with one of the intermediates in
the hydrogenation of 38. Independent hydrogenation of 39
provided a 98/2 ratio of 28-trans/28-cis. Direct hydrogenation
of 39 cannot possibly give 28-trans because such hydrogenations
are well-known to be stereospecific for the cis isomer.25 So 39
must also isomerize prior to hydrogenation. Taken together,
these results suggest that 38 (exclusively), 39 (to a large extent),
and possibly also 40 isomerize prior to hydrogenation and that
this isomerization process determines the trans/cis selectivity.
We speculate that disubstituted cyclopentenes (not shown)
resulting from further isomerizations of 38-40 are the species
undergoing hydrogenation and that the trans selectivity is the
result of a thermodynamically controlled isomerization. What-
ever the case, 38 is not directly hydrogenated, so assumptions
about its cis hydrogenation selectivity are probably correct but
are certainly not relevant.
Figure 7. Comparison of methyl group resonances in 1H NMR spectra
(CDCl3) of reaction products of cyclization of 8a: (a) Ihara’s spectrum,
Bu3SnH, 80 °C; (b) Ihara’s spectrum, Bu3SnH, 25 °C; (c) this work, Bu3-
SnH, 80 °C, prior to mCPBA treatment; (d) this work, Bu3SnH, 80 °C,
after mCPBA treatment. Spectra similar to (c) and (d) were obtained when
conducting reactions at 25 °C. Boxes highlight the key resonances at 3.68
ppm that are assigned to methyl esters of alkene-containing products.
and therefore, NOE experiments might not be a reliable method
to differentiate these isomers.
It is difficult to see how Beckwith and co-workers could have
assigned structures correctly without substantial additional
experimentationseven in hindsight, a view of their calculations
and the experiments seems to suggest that reasonable conclu-
sions were drawn on the basis of the available information.
The problem with Ihara’s work, in contrast, turns on
analysis: how were the amounts of “other diastereomers”
quantitated? Prof. Ihara kindly responded to this question with
helpful data that were not in the original paper. The top of Figure
7 shows expansions of the methyl ester regions of two 1H NMR
spectra that Ihara provided to us. The left spectrum (a)
corresponds to a tin hydride cyclization conducted at 80 °C while
the right one (b) corresponds to a similar cyclization at 25 °C.
Ihara assigned the signals at 3.67 and 3.66 ppm to one of the
three methyl esters of the cis-syn-cis 9a and cis-anti-cis 10a
products, respectively, and we confirmed this assignment with
the enriched samples isolated from the preparative GC separa-
tion. Ihara further assigned the signals at about 3.68 ppm to
single methyl esters of the “other diastereomers”. Product ratios
then follow from integration.
New experiments show that this assignment is not correct.
The lower part of Figure 7 shows expanded regions of two of
our spectra that are qualitatively similar to Ihara’s. In our hands,
the spectra like that on the left were reliably produced by simply
working up the reaction and removing the tin, as Ihara did. These
spectra (c) have small amounts of the key resonances that Ihara
assigned to “other diastereomers”. In contrast, spectra (d) like
that on the right are reliably produced by treating the crude
reaction products with mCPBA followed by flash chromatog-
raphy to remove polar products. Accordingly, we conclude the
resonances around 3.68 ppm do not belong to “other diastereo-
mers” but instead to constitutional isomers that result from
incomplete round trip cyclizationsthese are the resonances that
correspond to the small peaks in the GC that are removed on
mCPBA oxidation.
NOE experiments showed a 10% enhancement of the tert-
butyl singlet of the major product of the hydrogenation of 38
when the methyl doublet was irradiated.13 This was taken as
evidence for formation of the cis isomer, yet we now know
that 28-trans was used in the experiment. NOE studies on the
actual cis isomer (28-cis) were not reported, presumably because
it was not available in pure form. How does the trans isomer
exhibit such a large NOE? Figure 6 shows that both cis and
trans isomers of this compound have approximate syn-pentane
interactions. Such interactions should result in healthy NOEs,
(22) Chan, K.-K.; Cohen, N.; De Noble, J. P.; Specian, A. C., Jr.; Saucy, G. J.
Org. Chem. 1976, 41, 3497-3505.
(23) Hupe, E.; Denisenko, D.; Knochel, P. Tetrahedron 2003, 59, 9187-9198.
(24) The alcohol shown below is formed by hydroboration of 40, and the
stereoselectivity of the hydroboration (addition trans to the tert-butyl group)
suggests that the Beckwith assumption of (kinetic) addition of hydrogen
trans to the tert-butyl group is reasonable. To confirm the configuration of
this alcohol, we converted it to the thionocarbonate followed by Barton-
McCombie deoxygentation. This gave 28-cis as the only product. Further,
the crystal structure of the derived 3,5-dinitrobezoate ester was also solved.
See the Supporting Information for details.
(25) Siegel, S. In ComprehensiVe Organic Synthesis; Trost, B. M., Fleming, I.,
Eds.; Pergamon Press: Oxford, U.K., 1991; Vol. 8, pp 417-442.
9
J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005 5525

A R T I C L E S
Tripp et al.
While these experiments add a few more pieces, the puzzle
here is not quite complete. We could reliably observe the
absence of presence of the small extra methyl ester resonances
and GC peaks on the basis of whether the crude product was
treated with mCPBA, and these effects did not exhibit a
significant temperature dependence. In contrast, Ihara did not
treat any crude reactions products with mCPBA and reports that
the peaks appear or not as a function of temperature. Thus, we
cannot explain this experimental discrepancy between our work
and Ihara’s.
Nonetheless, the correlation of the minor NMR peaks with
the minor GC peaks and the disappearance of all these peaks
on mCPBA treatment support our conclusion that these arise
from constitutional isomers and not other diastereomers. Further,
we have identified and tentatively assigned other diastereomers
29a and 30a by GC, and NMR experiments show that they do
not have any methyl ester resonances in the region integrated
by Ihara. We conclude that NMR integration is simply not up
to the task of determining product ratios because there are two
major products and at least five minor products, each of which
has three nonequivalent methyl esters. Even though the GC
analysis is incomplete because we cannot rigorously assign
structures to all the minor peaks, it is still much more
informative than NMR integration because the GC peaks can
be individually classified as tricyclic or not and can be
individually integrated.
Figure 8. BHLYP/6-311G** optimized transition structures 41-44 for
cyclization of radical 14. MP2/6-311G** data are in parentheses, and
BHLYP/cc-pVDZ data are in brackets.
Table 4. Calculated Differences in Activation Energy (kJ mol-1
for the Different Modes of Ring-Closure of Radical 14
)
entry
method
cis-chair 41 trans-chair 42 cis-boat 43 trans-boat 44
Calculations. Clearly then, the computational methods
employed in the earlier study of Beckwith and Schiesser predict
the incorrect stereochemical outcome for the cyclization of 14
and would presumably also fail for other hexenyl radicals
bearing bulky substituents at position 1. What is special about
these 1-substituted systems such that simple steric arguments
fail to compute the correct reaction outcome? To explore this
question further, it is necessary to employ computational
methods that are designed to provide a more holistic approach
to this chemistry. Accordingly, we sought recourse to ab initio
and density functional quantum techniques.
1
2
3
4
5
6
MP2/6-311G**
BHLYP/6-311G**
BHLYP/6-311G**+ZPE
BHLYP/cc-pVDZ
BHLYP/cc-pVDZ+ZPE
BHLYP/aug-cc-pVDZ
0
0
0
0
0
0
1.1
2.2
2.5
2.7
3.0
1.5
9.0
12.5
12.8
13.4
13.5
12.6
4.6
4.5
3.8
5.0
4.3
4.2
Inspection of Figure 8 reveals the length of the forming C-C
bond in each transition state to lie in a narrow range around
2.2 Å, with BHLYP calculations predicting a slightly later
transition state in each case, as has been observed previously
for other systems. Inspection of Table 4 clearly indicates that
higher level quantum calculations do indeed predict the correct
stereochemical outcome for the ring-closure of 14; the cis-chair
structure 41 is calculated to be the lowest energy transition state
at all levels of theory. The magnitude of the preference for the
cis mode of cyclization is somewhat underestimated, the largest
energy difference of 3.0 kJ mol-1 calculated at the BHLYP/
cc-pVDZ + ZPE level of theory being some 10 kJ mol-1 less
than that determined for the ring-closure of 14 in hexane. The
cyclization of 14 displays a pronounced solvent dependence,
while the calculated data reflect gas-phase chemistry more
closely. Alternatively, very accurate data may only be achievable
Previous work has shown that, without going to very high
levels of theory (e.g. G3-RAD),26 the BH and HLYP (BHLYP,
density functional) method provides a very cost-effective
computational technique for studying free-radical reactions,27
while B3LYP often provides poor data for radicals and MP2
calculations are often highly spin contaminated;25,28 indeed, the
MP2/6-311G** calculated transition states for 14 provided s
values approaching unity, while the analogous BHLYP data
were typically below 0.82.
Searching for the C10H19 potential energy surface at all levels
of theory29 employed in this study, we located four transition
states for the cyclization of 14: cis-chair (41), trans-chair (42),
cis-boat (43), and trans-boat (44), in keeping with the earlier
publication by Beckwith and Schiesser.13 The BHLYP/
6-311G** optimized structures (41-44) are displayed in Figure
8, while the calculated differences in activation energy are listed
in Table 4. Full details (Gaussian Archive Entries) are available
as Supporting Information.
(29) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M.
A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.;
Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.;
Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.;
Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa,
J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.;
Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Adamo, C.; Jaramillo,
J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.;
Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.;
Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels,
A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.;
Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.;
Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz,
P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.;
Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson,
B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03,
revision B.05; Gaussian, Inc.: Pittsburgh, PA, 2003.
(26) Go´mez-Balderas, R.; Coote, M. L.; Henry, D. J.; Radom, L. J. Phys. Chem.
A 2004, 108, 2874-2883.
(27) Mohr, M.; Zipse, H.; Marx, D.; Parrinello, M. Phys. Chem. A 1997, 101,
8942-8948.
(28) For examples, see: Morihovitis, T.; Schiesser, C. H.; Skidmore, M. A. J.
Chem. Soc., Perkin Trans. 2 1999, 2041-2047.
9
5526 J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005

Stereochemistry of Hexenyl Radical Cyclizations
A R T I C L E S
using higher level methods such as G3-RAD. In any case, it
would appear that the stereochemical preferences exhibited by
hexenyl radicals bearing bulky substituents at position 1 are
dictated by electronic as well as steric phenomena because
quantum computational techniques are able to correctly predict
the outcome of cyclization, while the molecular mechanics
techniques used previously are unable to predict the outcome.
Z-R,â-unsaturated ester. However, the increase in selectivity that
was proposed by lowering the temperature has been shown to
be an artifact of the analysis methods and is not correct.
There have been very few exceptions to the generalization
that hexenyl radicals bearing 1-substituents cyclize with cis
selectivity. Now there is one fewer.
Acknowledgment. We thank the National Science Foundation
for funding this work, and we thank Dr. Steven J. Geib for
solving the crystal structures. We thank Prof. M. Ihara for
providing copies of key NMR spectra, and we thank Professor
M. Newcomb and R. E. P. Chandrasena for the preparative GC
purification. Support of the Victorian Institute for Chemical
Sciences High Performance Computing Facility is gratefully
acknowledged. We dedicate this paper to Professor Keith U.
Ingold in honor of his 75th birthday and in celebration of his
50 years at the National Research Council of Canada.
Conclusions
This work helps to solidify our understanding of both
stereoselectivity and temperature effects in the cyclizations of
hexenyl radicals bearing very large substituents on the radical
carbon. The long held notion that such cyclizations are trans
selective is shown to be incorrect; the cyclization of the 1-tert-
butylhexenyl radical is moderately cis selective, and the
selectivity is increased by geminal substitution on carbon 3.
Modern ab initio calculations reproduce this selectivity satis-
factorily. This selectivity trend is very general and extends to
relatively complex substrates such as 8a-c, 17, and 18. It has
allowed Ihara to reduce the complexity of round trip radical
cyclizations to the point where two tricyclic productsscis-syn-
cis and cis-anti-cissaccount for about 80% of the products. We
further increased this to about 90% by using the corresponding
Supporting Information Available: Complete experimental
procedures, compound characterization data, and Gaussian
Archive entries for all optimized structures (PDF). This material
is available free of charge via the Internet at http://pubs.acs.org.
JA042595U
9
J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005 5527