How do analogous α-chloroenamides and α-iodoenamides give different product distributions in 5-Endo radical cyclizations?

Article abstract of DOI:10.1021/ja8012962

5-Endo cyclizations of N-alkenyl carbamoylmethyl radicals provide γ-lactam radicals, which in turn evolve to reduced or non-reduced (alkene) products depending on reagents and reaction conditions. Several groups have made surprising observations that chlorides are better radical precursors than iodides in such cyclizations. Here is described a detailed study of tin and silicon hydride-mediated radical cyclizations of N-benzyl-2-halo-N-cyclohex-1- enylacetamides. The ratios of directly reduced, cyclized/reduced, and cyclized/non-reduced products depend not only on the reaction conditions and reducing reagent but also on the precursor. Prior explanations for the precursor-dependent product ratios based on amide rotamer effects are ruled out. The precursor-dependent behavior is further dissected into two different effects: (1) the ratio of cyclized/reduced products to cyclized/non-reduced products depends on the ability of the radical precursor to react with the product γ-lactam radical in competition with tin hydride (iodides can compete, chlorides cannot), and (2) the occurrence of large amounts of directly reduced (noncyclized) products in the case of iodides is attributed to a competing ionic chain reaction by which the precursor is reductively deiodinated with HI. This side reaction is not available to chlorides, thereby explaining why the chlorides are better precursors in such reactions. The ability of the iodides to provide cyclized products can be largely restored by adding base. The chlorides and iodides then become complementary precursors, with chlorides giving largely cyclized/reduced products and iodides giving largely cyclized/non-reduced products.

Full text of DOI:10.1021/ja8012962

Published on Web 06/10/2008
How Do Analogous r-Chloroenamides and r-Iodoenamides
Give Different Product Distributions in 5-Endo Radical
Cyclizations?
Dennis P. Curran,* David B. Guthrie, and Steven J. Geib
Department of Chemistry, UniVersity of Pittsburgh, Pittsburgh, PennsylVania 15260
Received February 21, 2008; E-mail: curran@pitt.edu
Abstract: 5-Endo cyclizations of N-alkenyl carbamoylmethyl radicals provide γ-lactam radicals, which in
turn evolve to reduced or non-reduced (alkene) products depending on reagents and reaction conditions.
Several groups have made surprising observations that chlorides are better radical precursors than iodides
in such cyclizations. Here is described a detailed study of tin and silicon hydride-mediated radical cyclizations
of N-benzyl-2-halo-N-cyclohex-1-enylacetamides. The ratios of directly reduced, cyclized/reduced, and
cyclized/non-reduced products depend not only on the reaction conditions and reducing reagent but also
on the precursor. Prior explanations for the precursor-dependent product ratios based on amide rotamer
effects are ruled out. The precursor-dependent behavior is further dissected into two different effects: (1)
the ratio of cyclized/reduced products to cyclized/non-reduced products depends on the ability of the radical
precursor to react with the product γ-lactam radical in competition with tin hydride (iodides can compete,
chlorides cannot), and (2) the occurrence of large amounts of directly reduced (noncyclized) products in
the case of iodides is attributed to a competing ionic chain reaction by which the precursor is reductively
deiodinated with HI. This side reaction is not available to chlorides, thereby explaining why the chlorides
are better precursors in such reactions. The ability of the iodides to provide cyclized products can be largely
restored by adding base. The chlorides and iodides then become complementary precursors, with chlorides
giving largely cyclized/reduced products and iodides giving largely cyclized/non-reduced products.
row elements or multiple sp2-hybridized atoms between the
radical precursor and acceptor.
Introduction
Once regarded as curiosities, 5-endo radical cyclizations are
increasingly important in synthetic radical chemistry.123,4 While
they are not as general as 5-exo and 6-exo cyclizations, 5-endo
radical cyclizations are surprisingly useful in a diverse assort-
ment of settings, and suitable substrates often contain second-
The 5-endo cyclization of N-vinyl carbamoylmethyl radicals
to provide γ-lactam radicals, first described by Ikeda and
Ishibashi, is probably the most common of all classes of 5-endo
cyclizations (eq 1). Such radicals can be generated from N-vinyl-
R-haloamides (R-haloenamides) either under reductive condi-
tions to provide saturated lactams, or under oxidative conditions
to produce heteroatom-substituted or unsaturated lactams. This
reaction has been developed extensively into a general approach
to γ-lactams by Ishibashi, Parsons, Clark, and others.5
(1) Ishibashi, H.; Sato, T.; Ikeda, M. Synthesis 2002, 695–713.
(2) (a) Chatgilialoglu, C.; Ferreri, C.; Guerra, M.; Timokhin, V.; Froudakis,
G.; Gimisis, T. J. Am. Chem. Soc. 2002, 124, 10765–10772. (b)
Alabugin, I. V.; Manoharan, M. J. Am. Chem. Soc. 2005, 127, 9534–
9545.
(3) Ishibashi, H.; Nakamura, N.; Sato, T.; Takeuchi, M.; Ikeda, M.
Tetrahedron Lett. 1991, 32, 1725–1728.
(4) Selected references: (a) Sato, T.; Nakamura, N.; Ikeda, K.; Okada,
M.; Ishibashi, H.; Ikeda, M. J. Chem. Soc., Perkin Trans. 1 1992,
2399–2407. (b) Sato, T.; Chono, N.; Ishibashi, H.; Ikeda, M. J. Chem.
Soc., Perkin Trans. 1 1995, 1115–1120. (c) Ikeda, M.; Ohtani, S.;
Okada, M.; Minakuchi, E.; Sato, T.; Ishibashi, H. Heterocycles 1998,
47, 181–186. (d) Ikeda, M.; Ohtani, S.; Yamamoto, T.; Sato, T.;
Ishibashi, H. J. Chem. Soc., Perkin Trans. 1 1998, 1763–1768. (e)
Ishibashi, H.; Matsukida, H.; Toyao, A.; Tamura, O.; Takeda, Y.
Synlett 2000, 1497–1499. (f) Ishibashi, H.; Kodama, K.; Higuchi, M.;
Muraoka, O.; Tanabe, G.; Takeda, Y. Tetrahedron 2001, 57, 7629–
7637. (g) Baker, S. R.; Burton, K. I.; Parsons, A. F.; Pons, J.-F.;
Wilson, M. J. Chem. Soc., Perkin Trans. 1 1999, 427–436. (h) Davies,
D. T.; Kapur, N.; Parsons, A. F. Tetrahedron 2000, 56, 3941–3949.
(i) Bryans, J. S.; Chessum, N. E. A.; Parsons, A. F.; Ghelfi, F.
Tetrahedron Lett. 2001, 42, 2901–2905. (j) Guerrero, M. A.; Cruz-
Almanza, R.; Miranda, L. D. Tetrahedron 2003, 59, 4953–4958. (k)
Clark, A. J.; Dell, C. P.; Ellard, J. M.; Hunt, N. A.; McDonagh, J. P.
Tetrahedron Lett. 1999, 40, 8619–8623. (l) Tang, Y.; Li, C. Org. Lett.
2004, 6, 3229–3231.
During the course of this development, Ishibashi and co-
workers made unusual observations in comparing cyclizations
of R-chloro-, R-bromo-, and R-iodoenamide precursors 1a-c
of the same radical. These precursors gave very different product
ratios, and surprisingly, the iodide 1c was by far the poorest
(5) Related cyclizations are also often mediated by copper: Clark, A. J.
Chem. Soc. ReV 2002, 31, 1–11.
9
10.1021/ja8012962 CCC: $40.75
2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 8437–8445 8437

A R T I C L E S
Curran et al.
Figure 2. Chatgilialoglu’s cyclizations of R-bromoenamide 1b under
pseudo-first-order conditions with (TMS)3SiH.
To further add to the quandary, Chatgilialoglu and workers
reported a thorough study of cyclization of bromoenamide 6b,
which is closely related to 1b (N-methyl instead of N-benzyl).2a
Under standard pseudo-first-order conditions for competition
radical kinetics (large excess of (TMS)3SiH), they observed only
the cyclized/reduced product 8 and the directly reduced product
7 (Figure 2). The ratios of these products varied in a standard
fashion as a function of silane concentration, thereby allowing
the calculation of the rate constant for 5-endo cyclization of 9
to 10 as 2 × 104 s-1 at 25 °C.
Figure 1. Ishibashi’s cyclizations of R-haloamides 1a-c with tin and silicon
hydrides.
performer of the three.6 Several of the key results published in
the provocatively titled article “Is an Iodine Atom Almighty as
a Leaving Group for Bu3SnH-Mediated Radical Cyclization?”
are summarized in Figure 1.
Slow syringe pump addition of Bu3SnH to R-haloenamides
1a-c at reflux in toluene provided a mixture of the expected
cyclized/reduced product 3 along with cyclized/non-reduced
products 4 and 5 and directly reduced (non-cyclized) product
2. The product distributions varied as a function of the radical
precursor. Cyclization of chloride 1a was highly efficient and
provided only cyclized/reduced product 3 in 92% yield. Bromide
1b again provided only cyclized products, but this time reduced
product 3 (55%) was accompanied by 11% of each of the
cyclized/non-reduced products 4 and 5. In stark contrast, the
iodide 1c provided none of the standard cyclized/reduced
product 3, and the cyclized/non-reduced products 4 and 5 were
produced in 13% and 11% yields, respectively. By far the major
product was directly reduced 2 in 68% yield. Changing the
reducing agent for 1c to tris(trimethylsilyl)silicon hydride
((TMS)3SiH)7 did not make matters better, but worse; no
cyclized products were formed at all, and 2 was isolated in 76%
yield. Both Ishibashi and Parsons have described similar trends
in related substrates, so the observations are not isolated.4
These unusual observations of high amounts of directly
reduced product 2 led Ishibashi and co-workers to the apparently
inescapable conclusion that “an iodine atom is not the best
radical leaving group for radical cyclizations of R-halo-
enamides”.6 But how can this be? It is well known that iodides
are exceptionally more reactive (104-106 times) than related
chlorides toward Bu3Sn•, (TMS)3Si•, and related radicals.8 If
the iodine atom is the most reactive leaving group, then why is
it not the best leaving group for forming the cyclized products?
The contrast between Chatgilialoglu’s well-behaved product
ratios and Ishibashi’s unexpected ones focuses attention on the
unusual behavior of the different radical precursors and derived
radicals under syringe pump addition conditions. To provide
more insight into this behavior, we undertook a detailed study
of both the radical precursors and the radical reactions under
strictly controlled conditions. Our observations suggest that the
ratios of reduced/non-reduced products are controlled by
concentration effects on bimolecular reactions (hydrogen transfer
and oxidation) that compete for cyclized radicals. We further
identify conditions where formation of the directly reduced
product 2 from iodide 1c can be suppressed, so there is no longer
any basis to conclude that iodides are somehow less reactive
than chlorides. Instead, we suggest a non-radical path that is
responsible for the formation of the directly reduced products
from iodides.
Results and Discussion
Structures and Rotation Dynamics of the Radical Precursors.
Rates of rapid radical cyclization reactions can exceed those of
amide bond rotations,9 so the starting rotamer ratio of an amide
radical precursor can have a significant effect on downstream
radical reactions. Along these lines, Ishibashi and co-workers
postulated that chloride 1a might have a ground-state preference
for E-amide rotamer 1aE (Figure 3), and thus its derived radical
11E would be prone to cyclization.3 In contrast, iodide 1c might
(6) Tamura, O.; Matsukida, H.; Toyao, A.; Takeda, Y.; Ishibashi, H. J.
Org. Chem. 2002, 67, 5537–5545.
(7) (a) Chatgilialoglu, C. Acc. Chem. Res. 1992, 25, 188–194. (b)
Chatgilialoglu, C.; Ferreri, C.; Gimisis, T. In Chemistry of Organic
Silicon; Rappoport, Z., Apeloig, Y. Eds.; Wiley: Chichester, 1998;
Vol. 2, pp 1539-1579.
(9) (a) Stork, G.; Mah, R. Heterocycles 1989, 28, 723–729. (b) Curran,
D. P.; Tamine, J. J. Org. Chem. 1991, 56, 2746–2750. (c) Curran,
D. P.; DeMello, N. C. J. Chem. Soc., Chem. Commun. 1993, 1314–
1317. (d) Curran, D. P.; Qi, H.; Geib, S. J.; DeMello, N. C. J. Am.
Chem. Soc. 1994, 116, 3131–3132. (e) Curran, D. P.; Liu, H. T.
J. Chem. Soc., Perkin Trans. 1 1994, 1377–1393. (f) Bremner, J. B.;
Sengpracha, W. Tetrahedron 2005, 61, 941–953.
(8) (a) Ingold, K. U.; Lusztyk, J.; Scaiano, J. C. J. Am. Chem. Soc. 1984,
106, 343–348. (b) Ballestri, M.; Chatgilialoglu, C.; Clark, K. B.;
Griller, D.; Giese, B.; Kopping, B. J. Org. Chem. 1991, 56, 678–683.
9
8438 J. AM. CHEM. SOC. VOL. 130, NO. 26, 2008

Haloenamide Radical Cyclizations
A R T I C L E S
Figure 4. X-ray crystal structure of chloroenamide 1c.
amide radical 13E/13Z (Figure 3) as 5 × 105 s-1 at 20 °C,11
and this is about 25 times faster than the rate constant for
cyclization of radical 9 measured by Chatgilialoglu.2a Radicals
9/11 and 13 are not directly comparable;12 however, the slow
rate of cyclization of 11E relative to rotation suggests that
rotamer populations might not be important in such 5-endo
cyclizations.
To provide more information on the starting rotamer ratios
of 1a-c and their rotational dynamics, we undertook spectro-
1
scopic and crystallographic studies. The H NMR spectra of
1a-c at 300 MHz in CDCl3 were all very similar and exhibited
a single resonance for each chemically different proton or group
of protons. Thus, either the compounds exist largely as a single
amide rotamer, or they exist as two rotamers in rapid equilibrium
on the NMR time scale. The latter explanation seems unlikely
since R-halo amides have rotation barriers typical of other
amides,13 and thus the C-N rotation barriers of 1a-c are
expected to be in the range of 15-17 kcal/mol.14 Processes with
barriers of this height are slow on the NMR time scale and give
rise to broadened or, more typically, well-separated resonances
in room-temperature spectra.
While there is no evidence for amide E/Z-rotamers in the
NMR spectra, pairs of protons on several methylene groups (for
example, the benzyl and halocarbonyl CH2 groups) exhibited
sharp, well-resolved signals. This resolution of diastereotopic
protons shows that the molecules are axially chiral in solution,
with a twisted N-alkenyl bond. This twisting is expected from
the analogy to anilides.10
Figure 3. Amide rotamers of 1a-c and derived radicals.
So 1a-c exist in solution as predominantly a single amide
rotamer, and the analogy to anilides suggests that this is the
E-rotamer. We supported this conclusion by crystallizing
chloride 1a and solving its X-ray structure.15 As shown in Figure
4, the amide bond in the crystal exists in the E-rotamer. As
expected from the NMR results,10 1c exhibits axial chirality.
The planes of the amide and the N-alkenyl group are nearly
have an increased ground-state preference for Z-rotamer 1cZ,
the derived radical of which 11Z would be prone to reduction.
In addition to the assumption that the chloride 1a and the
iodide 1c have significantly different ground-state rotamer ratios,
this proposal assumes that radicals 11E and 11Z are not in
equilibrium on the time scale of their onward reactions. There
is reason to doubt that either of these assumptions is valid. First,
N-alkyl anilides are reasonable models of these enamides
because they have an N-C(sp2) bond (to an aromatic ring rather
than a cyclohexenyl group). Such molecules are well known to
exhibit a healthy preference for the E-rotamer (see 12E in Figure
3);10 thus, the postulate that iodide 1c has significant amounts
of Z-rotamer at equilibrium lacks support. Second, Newcomb
and co-workers have measured the rate constant for rotation of
(11) Musa, O. M.; Horner, J. H.; Newcomb, M. J. Org. Chem. 1999, 64,
1022–1025.
(12) Resonance theory predicts that enamide radical 11E/Z should have a
lower barrier to rotation than 13E/Z due to cross-conjugation of the
lone pair of the amide nitrogen with the attached vinyl group. This
effect should decrease the importance of amide resonance and reduce
the rotation barrier.
(13) Chupp, J. P.; Olin, J. F. J. Org. Chem. 1967, 32, 2297–2303.
(14) Wunderlich, M. D.; Leung, L. K.; Sandberg, J. A.; Meyer, K. D.;
Yoder, C. H. J. Am. Chem. Soc. 1978, 100, 1500–1503.
(15) For crystal structures of other enamides, see: (a) Cornforth, J.; Patrick,
V. A.; White, A. H. Aust. J. Chem. 1984, 37, 1453–1460. (b) Erzhanov,
K. B.; Kayukova, L. A.; Umarova, Z. N.; Espenbetov, A. A.;
Struchkov, Y. T. Russ. J. Org. Chem. 1985, 21, 2304–2309. (c) Song,
F.; Snook, J. H.; Foxman, B. M.; Snider, B. B. Tetrahedron 1998, 54,
13035–13044.
(10) (a) Saito, S.; Toriumi, Y.; Tomioka, N.; Itai, A. J. Org. Chem. 1995,
60, 4715–4720. (b) Curran, D. P.; Hale, G. R.; Geib, S. J.; Balog, A.;
Cass, Q. B.; Degani, A. L. G.; Hernandes, M. Z.; Freitas, L. C. G.
Tetrahedron: Asymmetry 1997, 8, 3955–3975.
9
J. AM. CHEM. SOC. VOL. 130, NO. 26, 2008 8439

A R T I C L E S
Curran et al.
Figure 5. Rotational profile of the amide C-N and N-alkenyl bonds of
1a-c (X ) Cl, Br, I).
orthogonal, with a dihedral angle of about 74°. The crystal was
composed of a single enantiomer of 1c, and the presence of the
chlorine atom aided in solution of the absolute configuration of
the molecule as P.
Collectively, these results provide a general picture of the
rotational properties of enamides like 1a-c, and this picture
(Figure 5) closely resembles that of the related axially chiral
anilides. Enamides 1 exist predominately as E-rotamers, with
the population of Z-rotamer unknown but certainly less than
5% in non-polar solvents. In turn, each amide rotamer exists as
a pair of enantiomers that interconvert slowly on the NMR time
scale by N-alkenyl bond rotation.
Assuming that the E and Z amide rotamers of the halides
should have comparable reactivity toward Bu3Sn•, there should
be a strong preference for formation of radical 11E (Figure 3)
in reactions of all three halides 1a-c. Further, Chatgilialoglu
has calculated that radicals related to 11E are considerably more
stable than their Z-rotamers (by >4 kcal/mol).2a If 11Z does
form in small amounts, then its best option is probably rotation
to 11E. Accordingly, the contribution of products derived from
radical 11Z in Figure 3 can be neglected; 11E is likely the
immediate precursor of all cyclized and reduced products.
Competition Kinetics. Clearly the suggestion that differing
results in the radical cyclizations of chloride 1a and iodide 1c
originate from conformational differences in these precursors
is untenable. Accordingly, upon reaction with Bu3Sn•, chloride
1a and iodide 1c (and presumably bromide 1b) should generate
the same radical. To test this notion, we conducted a series of
competition kinetics experiments16 by reducing the radical
precursors 1a-c with fixed concentrations of Bu3SnH and
measuring the product ratios by gas chromatography (GC).
The mechanistic scheme for the competition kinetics experi-
ments is shown in Figure 6. Halogen abstraction from precursor
1 generates radical 11, which then partitions in the key
competition step between direct bimolecular reduction by tin
hydride to give 2 and unimolecular, 5-endo cyclization to give
14. Radical 14 can be reduced by tin hydride to give the reduced/
cyclized product 3. Or, it can react with a molecule of radical
precursor 1 either by direct electron transfer (ET) or by halogen
atom transfer (AT),17 followed by ionization of 16 to give
acyliminium ion 15. In turn, 15 loses a proton to give enamides
Figure 6. Mechanisms for product formation in radical cyclizations of
1a-c.
4 (∆7,7a and/or ∆3a,7a isomers), and subsequent double-bond
migration provides isomerized product 5.
Syntheses of authentic samples and characterization of all
the products in Figure 6 are described in the Supporting
Information. The isomers of 4 were very sensitive to equilibra-
tion in solution or on silica gel, and they were always isolated
(17) (a) Fontana, F.; Kolt, R. J.; Huang, Y.; Wayner, D. D. M. J. Org.
Chem. 1994, 59, 4671–4676. (b) Andrieux, C. P.; Save´ant, J.-M.;
Tallec, A.; Tardivel, R.; Tardy, C. J. Am. Chem. Soc. 1997, 119, 2420–
2429. (c) Cassayre, J.; Quiclet-Sire, B.; Saunier, J.-B.; Zard, S. Z.
Tetrahedron 1998, 54, 1029–1040. (d) Curran, D. P.; Ko, S.-B.
Tetrahedron Lett. 1998, 39, 6629–6632. (e) Roepel, M. G. Tetrahedron
Lett. 2002, 43, 1973–1976. (f) Aldabbagh, F.; Bowman, W. R.
Tetrahedron 1999, 55, 4109–4122. (g) Marco-Contelles, J.; Rodrıíguez-
Ferna´ndez, M. Tetrahedron Lett. 2000, 41, 381–384. (h) Miranda,
L. D.; Cruz-Almanza, R.; Pavo´n, M.; Romero, Y.; Muchowski, J. M.
Tetrahedron Lett. 2000, 41, 10181–10184.
(16) (a) Newcomb, M. Tetrahedron 1993, 49, 1151–1176. (b) Newcomb,
M. In Radicals in Organic Synthesis, 1st ed.; Renard, P., Sibi, M.
Eds.; Wiley-VCH: Weinheim, 2001, Vol. 1, pp 317-336.
9
8440 J. AM. CHEM. SOC. VOL. 130, NO. 26, 2008

Haloenamide Radical Cyclizations
A R T I C L E S
Table 1. Product Distributions in Competitive Cyclizations of 1a-c
As the Bu3SnH concentration was increased, the ratio of reduced
product 2 to cyclized products 4-6 increased for all substrates.
Likewise, the total mass balance tended to increase as well. The
starting iodide 1c was consumed in all experiments, but small
amounts of bromide 1b and chloride 1a remained in the
experiments at lower tin hydride concentrations (even though
all the tin hydride was consumed).
The ratio of the cyclized/reduced product 3 to the cyclized/
non-reduced products 4 and 5 depended on the halogen atom
of the precursor, in qualitative agreement with Ishibashi’s results.
Regardless of the tin hydride concentration, chloride 1a gave
about a 15/1 ratio of reduced to non-reduced products, while
the ratio with bromide 1b was 9/1, and that with iodide 1c was
1/1. These results suggest that acyliminium ion 15 is not reduced
ionically by Bu3SnH to give 3, but instead loses a proton to
give alkenes 4.
with Bu3SnHa
% yieldb
halide
[Bu3SnH]i,c mM
2
3
4 + 5
total
cyc:redc
1a
10.0
15.0
17.8
30.0
42.9
7.5
10.0
15.0
17.8
30.0
42.9
7.5
15
22
29
39
56
18
22
38
41
52
55
31
33
46
55
68
33
32
36
29
28
34
29
39
35
27
20
16
13
12
11
9
2
3
3
2
2
5
5
4
4
60
57
68
70
86
57
56
81
80
82
77
65
60
72
80
87
70:30
61:39
57:43
44:66
35:65
68:32
61:39
53:47
49:51
36:64
29:71
52:48
45:55
36:64
31:69
22:78
1b
1c
3
2
18
14
14
14
10
10.0
15.0
17.8
30.0
The plots of the ratio of the reduced product 2 to the combined
cyclized products 3-5 versus the tin hydride provide further
information. The experiments are not pseudo-first-order in tin
hydride, so the mean tin hydride concentrations were used for
the plots, as recommended by Newcomb.16 In principle, the three
precursors should give the same radical, so the lines in Figure
7 should coincide. In practice, they do not. There is a 3-fold
difference in the slopes of the lines for the chloride 1a and the
iodide 1c.
a Conditions: the appropriate substrate (1.0 equiv), Bu3SnH (1.2
equiv), and AIBN (0.2 equiv) in toluene were combined in a sealed tube
and heated at 110 °C for 3 h. b GC yield, using octadecane as an
internal standard. Yields are an average of three trials. c Ratio of (3 + 4
+ 5) to 2.
Like Chatgilialoglu,2 we approximated the rate constant for
hydrogen abstraction by the carbamoylmethyl radical 11 from
Bu3SnH as 9.3 × 106 M-1 s-1 at 383 K. This provides the
following rate constants for cyclization of radical 11 as a
function of precursor: from chloride 1a, 1.3 × 105 s-1; from
bromide 1b, 5.1 × 104 s-1; and from iodide 1c, 4.0 × 104 s-1
.
These compare well to the more accurately determined rate
constant for related radical 9 (Figure 2) at 383 K, 5.1 × 104
s-1. This rate constant was measured starting from the bromide
precursor only.2a
The mechanistic scenario in Figure 6 does not allow for
differing rate constants for 5-endo cyclization as a function of
radical precursor. However, we stress that the measured dif-
ferences, while outside experimental error, are rather small. For
example, if a stoichiometric amount of tin hydride is not
consumed in the formation of alkenes 4, then the mean
concentration of this reagent will be higher in the experiments
with the bromide 1b and especially the iodide 1c. More reduced
product 2 will form, and the (apparent) cyclization rate constant
will be lower.
Further, the lines in Figure 7 all converge at low tin hydride
concentration, suggesting that all the precursors should behave
similarly in syringe pump experiments. This is not at all in line
with Ishibashi’s results from preparative syringe pump addition.
The lines also share a common y-intercept of approximately
zero, which in turn supports the postulate that the 5-endo
cyclization is not reversible. Finally, the similar slope of the
lines coupled with the halogen-dependent ratios of reduced/non-
reduced cyclized products supports the notion that the pathways
for formation of these products diverge after the cyclization, as
proposed in Figure 6.
Figure 7. Kinetic plot of data in Table 1.
in a 2.3/1 ratio of ∆7,7a to ∆3a,7a. Also, all the alkenes 4 (∆7,7a),
4 (∆3a,7a), and 5 exhibited a single, coincident peak on GC
analysis, so it is possible that equilibration occurs here as well.
Thus, the GC analysis gives only a total yield of the three alkene
products without giving ratios. This is convenient because all
three products derive from the same pathway. Since 5 was the
only alkene isolated in pure form, its response factor was
determined and used to quantify the collective alkene peak in
the gas chromatogram.
Competition experiments were conducted by reacting an
R-haloenamide 1a-c (0.025 mmol), Bu3SnH (1.2 equiv), and
azobisisobutyronitrile (AIBN, 0.2 equiv) in toluene at 110 °C
in a sealed tube for 3 h. The mixture was cooled, the tube was
cracked, and the sample was analyzed by GC against octadecane
as an internal standard. Full details on the experiments and
analyses are provided in the Supporting Information. All
reactions were reproducible and clean, product peaks were well
defined, and no significant other products were formed, even
though mass balances were not quantitative (60-87%).
The results of this series of experiments are summarized in
Table 1 and plotted in Figure 7 in the usual format (ratio of
directly reduced product to the sum of all cyclized products
versus tin hydride concentration). Several trends are apparent.
Ishibashi and co-workers also reported that cyclization of
iodide 1c by syringe pump addition of tributyltin hydride was
improved by the addition of excess tributyltin chloride
(Bu3SnCl) or other tin compounds, but addition of tributyltin
iodide (Bu3SnI) did not have much effect. It was suggested that
Bu3SnCl could act as a Lewis acid, altering the rotational
behavior of the radical precursor 1c or the reactivity of the
9
J. AM. CHEM. SOC. VOL. 130, NO. 26, 2008 8441

A R T I C L E S
Curran et al.
Table 2. Product Distributions in Competitive Cyclizations of 1c
Table 3. Product Ratios in Cyclization of Variable Concentrations
of 1a at Fixed Tin Hydride Concentrationsa
with Bu3SnH (10 mM) in the Presence of Tin Halidesa
% yieldb
% yieldb
additive
equiv
2
3
4 + 5
total
cyc:redc
equiv of 1c
2
3
4 + 5 total recovered 1c (equiv) cyc:redc ox:cycd
none
13
10
3
14
9
8
9
12
9
33
33
25
33
34
32
31
60
52
36
51
59
55
52
45:55
36:64
31:69
35:65
42:58
42:58
41:59
1.00
1.50
2.00
21 12
13
15
16
46
47
48
0.00
0.15
0.46
54:46
45:55
29:71
1.1:1
2.5:1
4.0:1
Bu3SnCl
Bu3SnCl
Bu3SnCl
Bu3SnI
Bu3SnI
Bu3SnI
1.0
3.0
5.0
1.0
2.0
3.0
26
28
6
4
9
a Conditions: iodide 1c, Bu3SnH (1.0 equiv), and AIBN (0.2 equiv) in
toluene were combined in a sealed tube and heated at 110 °C for 3 h.
b GC yield, using octadecane as an internal standard. Yields are an
average of three trials. Bu3SnH was the limiting reagent. c Ratio of (3 +
4 + 5) to 2. d Ratio of (4 + 5) to 3.
13
14
13
8
a Conditions: the appropriate substrate (1.0 equiv), Bu3SnH (1.2
equiv), Bu3SnX additive, and AIBN (0.2 equiv) in toluene were
combined in a sealed tube and heated at 110 °C for 3 h. b GC yield,
using octadecane as an internal standard. Yields are an average of three
trials. c Ratio of (3 + 4 + 5) to 2.
and 2.0 equiv) of iodide 1c and measured the product yields by
GC as above. The results of these experiments are summarized
in Table 3. Total mass balances in all three experiments were
modest (41-47%), but again no new products were observed
by GC.
radical 11. Sibi has suggested that tin halides can alter the
rotamer population of N-enoyloxazolidinones,18 and other Lewis
acids are known to affect radical reactions of R-halocarbonyl
compounds.19
To learn the effects of Bu3SnCl on the radical precursor, we
recorded the H NMR spectrum of a 1/1 mixture of 1c and
As projected, the ratio of non-reduced/reduced cyclized
products ([4 + 5]/2) increased from 1.1/1 to 2.5/1 to 4.0/1 as
the substrate concentration increased. Interestingly, however,
the ratio of cyclized to reduced products ([3 + 4 + 5]/2)
also changed (from 54/46 to 45/55 to 29/71), even though
the tin hydride concentration remained constant. Finally, in
both of the experiments where tin hydride was used in
deficiency relative to the iodide 1c, the percentage of iodide
consumed exceeded the amount of tin hydride added. A control
experiment showed that the iodide 1c was stable in refluxing
toluene, so simple thermal decomposition is ruled out. These
results suggest that the ET/AT process in Figure 6 can, at least
to some extent, operate as an independent chain and consume
starting iodide without consuming tin hydride.
With the two groups of experiments in Tables 1 and 3, we
have independently probed two separate features of syringe
pump experiments: (1) that the tin hydride concentration is very
low (Table 1 results) and (2) that the tin hydride is present in
deficient amounts relative to the starting material throughout
the experiment (Table 3 results). The collective results suggest
that it is the deficiency of tin hydride, and not its low
concentration, that produces the unusual results with the iodide
precursors.
Cation and Radical Trapping Experiments. We next con-
ducted cyclization experiments with one substrate designed to
trap a cyclized cation21 and one designed to trap a cyclized
radical. Neither trapping experiment succeeded as planned, but
together they supported Ishibashi’s observations and provided
more clues about the unusual behavior of the iodides. The
syntheses of the substrates 17a,c,d and products and the analyses
of the reactions are described in the Supporting Information.
Cyclizations of the cation-trap substrates, chloride 17a and
iodide 17c, were conducted by syringe pump addition of a
toluene solution of Bu3SnH (1.2 equiv) and AIBN (0.1 equiv)
over 2 h to a refluxing solution of the precursor in toluene
(Figure 8). This procedure is similar to that followed by Ishibashi
and different from the competition experiments above. After
cooling and solvent evaporation, the crude product mixture was
treated with aqueous KF22a and then purified by flash chroma-
tography to provide isolated yields of individual products. In
1
Bu3SnCl in CDCl3; no shifts in any resonances were observed,
suggesting that little or no complexation occurred. To probe
for halogen exchange,20 we heated a toluene solution of 1c and
5 equiv of Bu3SnCl at reflux for 5 h, but only a small amount
of conversion to chloride 1a (<5%) was observed by GC. In
contrast, addition of 5% tetrabutylammonium iodide (Bu4NI)
promoted rapid halogen exchange,20 and the ratio of 1a/1c
decreased to a level of about 1.4/1 over 1 h. No further change
occurred, suggesting that equilibrium was reached. These results
suggest that, in the presence of a halide ion source, tin halides
and R-haloamides readily exchange their halogen atoms.
To learn the effect of the tin halides on the cyclization under
fixed tin hydride concentration experiments, we conducted a
second series of cyclizations of iodide 1c with increasing
amounts of Bu3SnCl (1, 3, or 5 equiv) and Bu3SnI (1, 2, or 3
equiv). In every experiment, the initial concentration of Bu3SnH
was 10 mM. The results of these experiments, summarized in
Table 2, show that the additives have little or no effect on the
product distributions. Indeed, the cyclization of 1c under
conditions of fixed tin hydride concentrations is very reproducible.
These experiments do not support the notion that Bu3SnCl
can function as a Lewis acid toward either the starting iodide
1c or the intermediate radical 11. However, it does seem
plausible that some of the effect observed by Ishibashi was due
to conversion of iodide 1c to chloride 1a under the syringe pump
addition experiments. Since the tin halides are apparently not
functioning as Lewis acids, we decided not to pursue this line
of investigation further.
The mechanism in Figure 6 suggests that increasing the
concentration of the iodide 1c while holding the tin hydride
concentration constant will provide more non-reduced cyclized
products 4/5 relative to reduced cyclized product 3. To test this
notion, we conducted three experiments with limiting quantities
of Bu3SnH (1 equiv, 0.01 M) and increasing amounts (1.0, 1.5,
(18) Sibi, M. P.; Ji, J. J. Am. Chem. Soc. 1996, 118, 3063–3064.
(19) (a) Murakata, M.; Jono, T.; Mizuno, Y.; Hoshino, O. J. Am. Chem.
Soc. 1997, 119, 11713–11714. (b) Feng, H.; Kavrakova, I. K.; Pratt,
D. A.; Tellinghuisen, J.; Porter, N. A. J. Org. Chem. 2002, 67, 6050–
6054.
(21) Maryanoff, B. E.; Zhang, H.-C.; Cohen, J. H.; Turchi, I. J.; Maryanoff,
C. A. Chem. ReV 2004, 104, 1431–1628, especially pp 1435-1438
and 1450-1454.
(20) Friedrich, E. C.; Abma, C. B.; Vartanian, P. F. J. Organomet. Chem.
1980, 187, 203–211.
(22) (a) Leibner, J. E.; Jacobus, J. J. Org. Chem. 1979, 44, 449–450. (b)
Harrowven, D. C.; Guy, I. L. Chem. Commun. 2004, 1968–1969.
9
8442 J. AM. CHEM. SOC. VOL. 130, NO. 26, 2008

Haloenamide Radical Cyclizations
A R T I C L E S
Figure 9. Product distributions in cyclization of radical-trap substrates 23.
of p-TsOH in refluxing dichloroethane for 12 h.24 This produced
tetracycle 18 in 82% yield, along with 12% recovered 22.
Clearly cationic cyclization is possible, but the acyl iminium
ion must be generated reversibly many times before it can occur.
Under the conditions of the radical cyclization, cationic cy-
clization cannot compete with deprotonation.
Figure 8. Product distributions in cyclizations of cation-trap substrates
17.
no case did we detect any of the cation cyclization product 18;
however, the observed results resembled those of Ishibashi’s
preparative syringe pump experiments rather than our fixed
concentration kinetic experiments.
As substrates for radical trapping, we selected chloride 23a
and iodide 23c because the chloride is already a known
compound, the cyclization of which has been studied by
Ishibashi.4a Cyclizations of 23a,c were again conducted by
syringe pump addition as described above. Products were
isolated by flash chromatography over KF/silcia gel,22b and the
yields are summarized in Figure 9. The Bu3SnH-mediated
cyclization of 23a went generally as expected, providing 65%
of the tandem cyclization products 26 (5-exo) and 27 (6-endo),
along with a small amount of directly reduced 24 (6%) and
competing cyclization product 25 (5%). (This results from 6-exo
radical cyclization to the N-butenyl group in competition with
5-endo cyclization to the enamide.) Ishibashi reported a similar
combined yield of tandem cyclized products (59%), though there
was more of the 5-exo and less of the 6-endo product.
Cyclization of chloride 17a provided a 75% yield of reduced/
cyclized product 20 along with 10% of the directly reduced
product 19; alkenes 21 and 22 were not detected. Reductive
cyclization of the xanthate 17d under the same conditions
provided similar results; cyclized product 20 was isolated in
62% yield and reduced product 19 in 13% yield. Alkenes 21
and 22 were not observed. In contrast, non-reductive cyclization
of 17d with lauroyl peroxide4j,23 provided only conjugated
alkene 22 in 67% isolated yield.
The cyclization of iodide 17c in toluene occurred in lower
yield, providing 9% of 21, 43% of 22, and about 12% of 19.
The disappearance of 20 at the expense of 21 and 22 is expected
because the ET mechanism is favored over reduction at low tin
hydride concentration. A similar experiment in refluxing dichlo-
roethane provided 22 in 45% yield, along with 15% reduced
19.
Most shocking were the experiments with iodide 23c: syringe
pump addition of Bu3SnH over either 3 or 6 h provided almost
exclusively the reductively deiodinated product 24 (83% and
75%, respectively), with only traces of the other products. We
also tried to reduce 23c with (TMS)3SiH, but as soon as the
addition was begun, the color of the reaction mixture turned
deep red. Insertion of pH paper into this reaction mixture showed
that it was highly acidic. These observations suggest that
diiodine (I2) and HI are present. This reaction mixture was very
difficult to purify, producing 24 still contaminated with silicon-
containing byproducts. Again however, no cyclized products
It is likely that cation intermediates are involved in the
formation of 21 and 22, so trapping by Friedel-Crafts reaction
to give 18 is too slow to compete with deprotonation of the
intermediate acyl iminium ion. This assertion was supported
by an experiment in which 22 was exposed to a catalytic amount
(23) (a) Zard, S. Z. In Radicals in Organic Synthesis, 1st ed.; Renaud, P.,
Sibi, M. P. Eds.; Wiley-VCH: Weinheim, 2001; Vol. 1, pp 91-108.
(b) Quiclet-Sire, B.; Zard, S. Z. In Radicals in Synthesis II: Complex
Molecules; Gansauer, A. Ed.; Springer-Verlag: Berlin, Germany, 2006;
Vol. 264, pp 201-236.
(24) Padwa, A.; Lee, H. I.; Rashatasakhon, P.; Rose, M. J. Org. Chem.
2004, 69, 8209–8218.
9
J. AM. CHEM. SOC. VOL. 130, NO. 26, 2008 8443

A R T I C L E S
Curran et al.
Table 4. Product Distribution in Reductive Cyclizations of 1c with
were apparent. These results are qualitatively consistent with
those reported by Ishibashi for the reaction of 1c with
(TMS)3SiH.
Base Additivesa
% yieldb
A speculative mechanism for the formation of diiodine in
reactions with (TMS)3SiH is shown in eq 2. The standard
reaction of (TMS)3SiH with 23c should produce (TMS)3SiI,
which in turn could react with precursor 23c by a precedented
ionic mechanism25 to provide 28 and I2. Hydrolysis of 28 during
workup would provide reduced product 24. In turn, HI could
be formed either by reaction of (TMS)3SiH with diiodine or
through the ET/AT pathway as in Figure 6. The reaction of
23c with (TMS)3SiH is complex, so these are not selective and
efficient pathways, but they are possible nonetheless. Chloride
23a presumably cannot react with the silyl chloride by this
pathway, and instead a standard radical chain ensues.
additive
2
3
4
5
total
cyc:redc
-
35
5
5
-
-
-
4
13
4
4
10
-
13
60
45
44
41
32
21
3
22
5
20
-
73
81
76
53
71
32
52:48
94:6
93:7
>99:1
>99:1
>99:1
pyridine
2,6-lutidine
Cs2CO3
K2CO3
Proton Sponge
a Conditions: a solution of Bu3SnH (1.2 equiv) and AIBN (0.1 equiv)
in toluene (12 mM in Bu3SnH) was added via syringe pump to a
refluxing mixture of the amide 1c in toluene (10 mM) and additive over
3 h. b Isolated yield after 10% KF/silica column chromatography. c Ratio
of (3 + 4 + 5) to 2.
To test whether the proposed ionic pathway for reductive
deiodination of 1c could be suppressed by base, we conducted
a series of cyclization reactions under typical syringe pump
conditions in the presence of base additives. The crude products
were purified by chromatography on 10% (w/w) KF/silica gel,
and the isolated product yields from these reactions are shown
in Table 4. In the control experiment with no additive, we
obtained significant amounts of directly reduced product 2
(35%), alongside cyclized/non-reduced products 4 and 5. In a
similar experiment, Ishibashi observed even more reduced
product than we did (68%, see Figure 1), but syringe pump
addition experiments are often difficult to reproduce quantita-
tively. Qualitatively, the formation of relatively large amounts
of 2 at very low tin hydride concentrations, is again consistent
with Ishibashi’s results.
All of the bases suppressed or eliminated the formation of
the directly reduced product 2 while providing increased yields
of the alkene ET/AT products, especially 4. Pyridine provided
the highest yield of 4 (60%), although there was still 5% of the
reductively deiodinated products 2. Apparently, pyridine is not
a strong enough base to cause double bond migration of the
initial products 4 to give 5. Inexpensive potassium carbonate
also performed well, providing a 61% combined yield of alkene
products 4 and 5 with no directly reduced products. Cesium
carbonate and especially Proton Sponge gave lower yields along
with decomposition products.
Effects of Acid and Base. We had assumed that any HI
generated during the reactions of these iodoesters would be
consumed by an acid/base reaction with the stannane or silane.
However, these acid/base reactions may not be sufficiently fast
to suppress other reactions, especially under syringe pump
addition conditions (where the concentration of the hydride is
always very low). The above results led us to consider that HI
might instead react ionically with the starting iodoamide by the
pathway shown in eq 3.26 The resulting molecule of diiodine
that formed would then react with tin hydride to give Bu3SnI
and another molecule of HI. The result is a non-radical chain
that reduces the iodoamide while oxidizing the tin hydride.
To test this idea, we attempted to reductively deiodinate 1c
under several different ionic conditions, and the results of these
experiments are summarized in eq 3. Treatment of 1c with
Bu3SnI alone gave no reaction, but it was slowly deiodinated
over 24 h on exposure to camphorsulfonic acid (CSA) to provide
2 in 53% yield. Treatment of 1c with both Bu4NI and CSA
provided a 68% yield of 2 in only 2 h. Accordingly, a
combination of an acid and iodide ion can apparently reduce
1c in a non-radical process. Interestingly, heating of 1c with
CSA and Bu3SnI resulted in no reaction, suggesting that the tin
iodide is not a good source of iodide ion.
Conclusions
The results reported herein provide a basis for understanding
the unusual behavior of R-iodoenamides (and R-bromoenamides)
in 5-endo cyclizations. The rotational features of the chloride,
bromide, and iodide precursors are very similar, and they do
not play a role in the differing product ratios. Furthermore, the
unusual behavior is caused not because the iodide is a poorer
radical precursor compared to chloride, but instead because it
is a much better one. In the case of chlorides, a standard tin
hydride chain occurs. In the case of iodides, the ET/AT process
that ensues from the reaction of the cyclized radical with its
own radical precursor competes with standard reaction of the
radical with Bu3SnH. The ET/AT process is a radical chain that
generates HI. In turn, this HI can set off a non-radical chain
(25) (a) Kim, M.; Kawada, K.; Gross, R. S.; Watt, D. S. J. Org. Chem.
1990, 55, 504–511. (b) Nikulin, V. I.; Rakov, I. M.; De Los Angeles,
J. E.; Mehta, R. C.; Boyd, L. Y.; Feller, D. R.; Miller, D. D. Bioorg.
Med. Chem. 2006, 14, 1684–1697. (c) Zheng, X.; Wang, X.; Chang,
J.; Zhao, K. Synlett 2006, 19, 3277–3283.
(26) (a) Zimmerman, H. E. J. Am. Chem. Soc. 1957, 79, 6554–6558. (b)
Gervay, J.; Gregar, T. Q. Tetrahedron Lett. 1997, 38, 5921–5924.
9
8444 J. AM. CHEM. SOC. VOL. 130, NO. 26, 2008

Haloenamide Radical Cyclizations
A R T I C L E S
deiodination process for the formation of the directly reduced
product 2. It is this HI-mediated background process that makes
it appear as if the iodide is a poorer radical precursor than the
chloride.
chain, while the iodide provides predominately the alkene
products via the ET/AT chain.
The light shed by this mechanistic study will allow more
controlled and predicable applications of the important class of
5-endo cyclizations of haloenamides. Furthermore, the observed
behavior is not tied to the type of radical cyclization (5-endo)
that occurs, but instead to the type of reducing radical that is
formed (an R-amido radical). So, the results can be extrapolated
to other types of radical reactions as well.
The ET/AT process and the non-radical chain that it sets off
are most efficient under syringe pump addition conditions
because they can both propagate while there is a deficiency of
tin hydride. This explains why the syringe pump addition
experiments with the iodides give results very different from
those obtained with the low-concentration experiments with
stoichiometric quantities of tin hydride. The addition of a
suitable base neutralizes the HI that is formed in the ET/AT
process with the iodides, thereby suppressing the ionic reduction
path and restoring the radical cyclization efficiency of the
iodoenamides. Nonetheless, the chloroenamide and iodoenamide
precursors are still complementary because of the divergence
after the 5-endo cyclization. The chloride provides primarily
the reductively cyclized products through a standard tin hydride
Acknowledgment. We thank the National Science Foundation
for funding this work. This paper is dedicated to the memory of
Prof. Albert I. Meyers.
Supporting Information Available: Experimental procedures,
compound characterization, and crystallographic data. This
material is available free of charge via the Internet at http://
pubs.acs.org.
JA8012962
9
J. AM. CHEM. SOC. VOL. 130, NO. 26, 2008 8445