Radical cyclizations of acylgermanes. New reagent equivalents of the carbonyl radical acceptor synthon

Article abstract of DOI:10.1021/ja970219m

An in depth study of the capability of acylgermanes to function as acceptors in radical cyclizations is reported. Radicals add to acylgermanes, and rapid fragmentation of the resulting α- germylalkoxy radicals provides ketones and germyl radicals. The germyl radicals in turn propagate the chain by addition or abstraction, so the reaction occurs by a unimolecular chain transfer (UMCT) process. In contrast, acylsilanes also function as radical acceptors, but they do not participate in UMCT processes because a 'radical-Brook' rearrangement intervenes. Cyclizations in 5-exo and 6-exo modes show good to excellent scope, and rate constants for cyclization can be varied over 2 orders of magnitude by changing the germanium substituents. Acyltriarylgermanes are among the best radical acceptors yet identified, and this quality makes them superior reagent equivalents of the carbonyl radical acceptor synthon. Parent cyclizations in the 4-exo and 7-exo modes fall. Attempted 3-exo cyclization results in a 1,2-acyl shift, which can be conducted alone or in tandem with a subsequent cyclization to the rearranged acylgermane. The results provide a foundation for future synthetic applications of radical cyclizations to acylgermanes.

Full text of DOI:10.1021/ja970219m

VOLUME 119, NUMBER 21
MAY 28, 1997
© Copyright 1997 by the
American Chemical Society
Radical Cyclizations of Acylgermanes. New Reagent
Equivalents of the Carbonyl Radical Acceptor Synthon
Dennis P. Curran,* Ulf Diederichsen, and Michael Palovich
Contribution from the Department of Chemistry, UniVersity of Pittsburgh,
Pittsburgh, PennsylVania, 15260
ReceiVed January 23, 1997X
Abstract: An in depth study of the capability of acylgermanes to function as acceptors in radical cyclizations is
reported. Radicals add to acylgermanes, and rapid fragmentation of the resulting R-germylalkoxy radicals provides
ketones and germyl radicals. The germyl radicals in turn propagate the chain by addition or abstraction, so the
reaction occurs by a unimolecular chain transfer (UMCT) process. In contrast, acylsilanes also function as radical
acceptors, but they do not participate in UMCT processes because a “radical-Brook” rearrangement intervenes.
Cyclizations in 5-exo and 6-exo modes show good to excellent scope, and rate constants for cyclization can be
varied over 2 orders of magnitude by changing the germanium substituents. Acyltriarylgermanes are among the
best radical acceptors yet identified, and this quality makes them superior reagent equivalents of the carbonyl radical
acceptor synthon. Parent cyclizations in the 4-exo and 7-exo modes fail. Attempted 3-exo cyclization results in a
1,2-acyl shift, which can be conducted alone or in tandem with a subsequent cyclization to the rearranged acylgermane.
The results provide a foundation for future synthetic applications of radical cyclizations to acylgermanes.
Introduction
(imines,6 hydrazones,7 oximes,8 nitriles,9 heteroaromatic rings10)
multiple bonds have been used as radical acceptors in diverse
Early synthetic applications of radical cyclizations relied
heavily on the use of carbon-carbon double bonds as radical
acceptors.1 But more recently2 a wide assortment of carbon-
carbon (alkynes,3 allenes,4 aromatic rings5) and carbon-nitrogen
(6) (a) Tomaszewski, M. J.; Warkentin, J.; Werstiuk, N. H. Aust. J. Chem.
1995, 48, 291. (b) Bowman, W. R.; Stephenson, P. T.; Young, A. R.
Tetrahedron Lett. 1995, 36, 5623. (c) Bowman, W. R.; Stephenson, P. T.;
Young, A. R. Tetrahedron 1996, 52, 11445.
(7) (a) Bernardhenriet, C. D.; Grimaldi, J. R.; Hatem, J. M. Tetrahedron
Lett. 1994, 35, 3699. (b) Bowman, W. R.; Stephenson, P. T.; Terrett, N.
K.; Young, A. R. Tetrahedron Lett. 1994, 35, 6369. (c) Bowman, W. R.;
Stephenson, P. T.; Terrett, N. K.; Young, A. R. Tetrahedron 1995, 51, 7959.
(d) Sturino, C. F.; Fallis, A. G. J. Org. Chem. 1994, 59, 6514. (e) Kim, S.;
Cheong, J. H.; Yoon, K. S. Tetrahedron Lett. 1995, 36, 6069. (f) Kim, S.;
Kee, I. S. Tetrahedron Lett. 1993, 34, 4213.
(8) (a) Bartlett, P. A.; McLaren, K. L.; Ting, P. C. J. Am. Chem. Soc.
1988, 110, 1633. (b) Della, E. W.; Knill, A. M. Aust. J. Chem. 1994, 47,
1833. (c) Hatem, J.; Henriet-Bernard, C.; Grimaldi, J.; Maurin, R.
Tetrahedron Lett. 1992, 33, 1057. (d) Marco-Contelles, J.; Destabel, C.;
Chiara, J. L.; Bernabe, M. Tetrahedron-Asymmetry 1995, 6, 1547. (e) Naito,
T.; Tajiri, K.; Harimoto, T.; Ninomiya, I.; Kiguchi, T. Tetrahedron Lett.
1994, 35, 2205.
(9) (a) Griller, D.; Schmid, P.; Ingold, K. U. Can. J. Chem. 1979, 57,
831. (b) Clive, D. L. J.; Beaulieu, P. L.; Set, L. J. Org. Chem. 1984, 49,
1313.
X Abstract published in AdVance ACS Abstracts, May 1, 1997.
(1) (a) Curran, D. P. Synthesis 1988, 419. (b) Curran, D. P. Synthesis
1988, 487.
(2) Curran, D. P. In ComprehensiVe Organic Synthesis; Trost, B. M.,
Fleming, I., Eds.; Pergamon: Oxford, 1991; Vol. 4, p 779.
(3) (a) Clive, D. L. J.; Manning, H. W. J. Chem. Soc., Chem. Commun.
1993, 666. (b) Clive, D. L. J.; Tao, Y.; Khodabocus, A.; Wu, Y. J.; Angoh,
A. G.; Bennett, S. M.; Boddy, C. N.; Bordeleau, L.; Kellner, D.; Kleiner,
G.; Middleton, D. S.; Nichols, C. J.; Richardson, S. R.; Vernon, P. G. J.
Am. Chem. Soc. 1994, 116, 11275.
(4) (a) Crandall, J. K.; Ayers, T. A. Tetrahedron Lett 1991, 32, 3659.
(b) Ziegler, C. B., Jr. J. Org. Chem. 1990, 55, 2983. (c) Apparu, M.;
Crandall, J. K. J. Org. Chem. 1984, 49, 2125.
(5) (a) Bowman, W. R.; Heaney, H.; Jordan, B. M. Tetrahedron 1991,
47, 10119. (b) Ziegler, F. E.; Harran, P. G. J. Org. Chem. 1993, 58, 2768.
(10) Murphy, J. A.; Sherburn, M. S. Tetrahedron 1991, 47, 4077.
S0002-7863(97)00219-9 CCC: $14.00 © 1997 American Chemical Society

4798 J. Am. Chem. Soc., Vol. 119, No. 21, 1997
Curran et al.
Scheme 1
Figure 1. Existing reagent equivalents of carbonyl radical acceptors.
settings. Indeed, the variety of radical acceptors that can be
used in radical cyclizations is a major synthetic attraction.11
The carbonyl group is one of the central functional groups
in organic chemistry, and this renders the development of radical
cyclizations to make carbonyl compounds an important goal.
Equation 1a depicts a strategy12 for formation of a cyclopen-
tanone by reaction of an alkyl radical precursor with a carbonyl
radical acceptor. The strategy complements the traditional ionic
route, eq 1b, which is an intramolecular acylation.13 The
the bond-forming capabilities of the carbonyl group (because
it is masked). A more direct route involves cyclizations to
nitriles;9 the intermediate imines are easily hydrolyzed by mild
acid. But nitriles are very modest radical acceptors which are
useful only for fairly rapid cyclizations, and attempts to form
bridged or other moderately strained rings usually result in
fragmentation (nitrile transfer).15
Direct equivalents of the carbonyl radical acceptor are lacking
because functional groups like amides and esters are inert to
radical addition.16 Recently, acyl sulfides and selenides have
been used with some success.17 As anticipated from studies in
the vitamin B12 area,17b acyl sulfides have relatively low
reactivity, but acyl selenides show more potential. For these
substrates, a reagent like hexabutylditin is required in stoichio-
metric amounts to propagate chains. Carbon monoxide adds
to reactive radicals at high CO pressures, but it obviously cannot
be used as a radical acceptor in cyclizations. However, it is
very useful in radical additions and in tandem reactions of all
sorts.18 Aldehydes are excellent radical acceptors,19 and the
resulting alcohols can be oxidized to ketones under mild
conditions. The problem here is that radical cyclizations to
carbonyl groups are reversible, and trapping of the closed
product is not always easy. In general, cyclohexanols can be
made by radical cyclizations to aldehydes, but attempts to form
other ring sizes often results in migration or ring expansion.
These reactions are of significant preparative interest in their
own right.20
(
(
)
)
functional group tolerance of radical methods could make a
radical-based strategy the method of choice for many synthetic
applications. The alkyl radical can readily be prepared from a
variety of standard precursors including halides. The problem
with the strategy in eq 1a is that reagent equivalents of the
carbon radical acceptor synthon are often lacking in one or more
respects. In contrast, carbonyl radical precursors14 (acyl halides,
sulfides, selenides, tellurides, cobalts, etc.) are readily available,
and the derived acyl and related radicals provide powerful
synthetic options in a number of settings.
In 1990, Kiyooka and co-workers21 reported the discovery
of an interesting reaction of unsaturated acylgermanes, the
simplest example of which is shown in Scheme 1. Photolysis
of 1 with a UV lamp provided 2 in 92% yield. This
isomerization took place under very mild conditions and had
good generality, although some limitations were identified. A
mechanism was posited in which the acylgermane behaved as
the radical precursor and the alkene behaved as the radical
acceptor. We were intrigued by these observations, and we soon
garnered strong experimental support for the alternative mech-
Existing reagent equivalents of the carbonyl radical acceptor
synthon are summarized in Figure 1. Cyclizations to alkynes
have often been used,3 and oxidative cleavage is required to
reveal the ketone. This is a valuable strategy when ketone
protection before and after the radical cyclization is desired,
and it has added flexibility because the terminal alkyne
substituents can be chosen to accelerate the radical cyclization
(TMS and Ph are popular choices). However, the strategy is
indirect, ranks low on the “atom economy” scale, and sacrifices
(11) Jasperse, C. P.; Curran, D. P.; Fevig, T. L. Chem. ReV. 1991, 91,
1237.
(12) For retrosynthetic planning with radical reactions, see: Curran, D.
P. Synlett 1991, 63.
(15) Curran, D. P.; Seong, C. M. Tetrahedron 1992, 48, 2175.
(16) For the only known exception, see: Lee, E.; Yoon, C. H.; Lee, T.
H. J. Am. Chem. Soc. 1992, 114, 10981.
(13) For example, see: Molander, G. A.; Harris, C. R. J. Am. Chem.
Soc. 1995, 117, 3705. Molander, G. A.; McKie, J. A. J. Org. Chem. 1993,
58, 7216.
(17) (a) Kim, S.; Jon, S. Y. Chem. Commun. 1996, 1335. (b) Dowd, P.
Sel. Hydrocarbon Act. 1990, 265. (c) For related bimolecular reactions to
oxime ethers, see: Kim, S.; Lee, I. Y.; Yoon, J. Y.; Oh, D. H. J. Am. Chem.
Soc. 1996, 118, 5138.
(14) (a) Lucas, M. A.; Schiesser, C. H. J. Org. Chem. 1996, 61, 5754.
(b) Chen, L.; Gill, G. B.; Pattenden, G.; Simonian, H. J. Chem. Soc., Perkin
Trans. 1 1996, 31. (c) Batsanov, A.; Chen, L.; Gill, G. B.; Pattenden, G. J.
Chem. Soc., Perkin Trans. 1 1996, 45. (d) Penn, J. H.; Liu, F. J. Org. Chem.
1994, 59, 2608. (e) Gill, G. B.; Pattenden, G.; Reynolds, S. J. J. Chem.
Soc., Perkin Trans. 1 1994, 369. (f) Chen, C.; Crich, D. Tetrahedron Lett.
1993, 34, 1545. (g) Chen, C.; Crich, D.; Papadatos, A. J. Am. Chem. Soc.
1992, 114, 8313. (h) Crich, D.; Yao, Q. J. Org. Chem. 1996, 61, 3566.
(18) (a) Ryu, I.; Sonoda, N.; Curran, D. P. Chem. ReV. 1996, 96, 177.
(b) Ryu, I.; Sonoda, N. Angew. Chem., Int. Ed. Engl. 1996, 35, 1051.
(19) (a) Walton, R.; Fraser-Reid, B. J. Am. Chem. Soc. 1991, 113, 5791.
(b) Beckwith, A. L. J.; Hay, B. P. J. Am. Chem. Soc. 1989, 111, 2674.
(20) Dowd, P.; Zhang, W. Chem. ReV. 1993, 93, 2091.
(21) Kiyooka, S.; Kaneko, Y.; Matsue, Y.; Hamada, M.; Fujiyama, R.
J. Org. Chem. 1990, 55, 5562.

Radical Cyclizations of Acylgermanes
J. Am. Chem. Soc., Vol. 119, No. 21, 1997 4799
anism shown in Scheme 1.22 In this mechanism, the alkene
behaves as a radical precursor, reversibly adding the tri-
phenylgermyl radical (step 1), and the acylgermane behaves as
the radical acceptor (step 2). The key fragmentation (step 3)
occurs not with cleavage of a carbon-carbon bond (as happens
with radical cyclizations to aldehydes and ketones) but instead
with cleavage of the carbon-germanium bond. This step
transfers the chain, so the overall transformation requires only
initiation. The reaction is a heteroatom analog of the well-
known cyclization of vinylstannanes.23 The chain transfer
process is unimolecular, conferring on this transformation all
the advantages associated with unimolecular chain transfer
(UMCT) reactions.24
Scheme 2
These results suggested that acylgermanes, and possibly
silanes and stannanes as well, showed potential as reagent
equivalents of the carbonyl radical acceptor synthon. Indeed,
in preliminary experiments25 we showed that cyclizations of 3a
and 3b to cyclopentanone 4a and cyclohexanone 4b occurred
rapidly and in very high yield (eq 2). Estimates of rate constants
suggested that the acylgermanes were much more reactive
towards radicals than most other acceptors.
Herein we report the full details of our studies on the scope
and limitations of radical cyclization reactions of acylgermanes.
These studies have focused on how the germanium and radical
substituents and the ring size affect the radical cyclization. These
areas were chosen because other substituent effects in radical
cyclizations are well understood and can be superposed on
acylgermane reactions once the unique features of acylgermanes
are evident. Acylgermanes are indeed excellent radical accep-
tors. As far as we know, radical cyclizations of acylstannanes
have still not been reported, but it seems highly probable that
they will be good radical acceptors that undergo reactions
analogous to those of the acylgermanes, not silanes.
(
)
Results and Discussion
Synthesis of Acylgermanes. Acylgermanes have been much
less widely synthesized and studied than their acylsilane and
stannane counterparts, but the syntheses of all three classes of
compounds are conceptually similar.27 Scheme 2 illustrates the
synthesis of the acylgermanes that were used in rate studies to
uncover the effects of germane substitution. One of two
pathways was generally followed. In the first route a germyl-
lithium species28 (generated by deprotonation or transmetala-
tion,29 depending on the germanium substituents) was reacted
with δ-valerolactone, and the resulting unstable hydroxyl-
acylgermane was mesylated and converted into the iodide. This
route was not successful for substrates bearing electron poor
aromatic rings, and these were made instead by addition of the
germyllithium to an aldehyde followed by DIAD oxidation
according to Marshall for acylstannane synthesis.30
Our preliminary studies of acylsilanes were preempted by a
report of Tsai,26 who observed that acylsilanes are indeed
excellent radical acceptors, yet they do not behave like acylger-
manes. UMCT reactions will not propagate, but haloacylsi-
lanes5 can be reductively cyclized with reagents like tin hydride
to give silylcyclopentanols 6 (eq 3). The difference in reaction
Other acylgermanes used for the ring size studies were made
by one of these two routes, and complete details of all of these
(
)
(23) Some examples: (a) Gomez, A. M.; Lopez, J. C.; Fraser-Reid, B.
J. Chem. Soc., Perkin Trans. 1 1994, 1689. (b) Curran, D. P.; van Elburg,
P. A. Tetrahedron Lett. 1989, 30, 2501. (c) Ardisson, J.; Ferezou, J. P.;
Julia, M.; Li, Y.; Liu, L. W.; Pancrazi, A. Bull. Soc. Chim. Fr. 1992, 129,
387. (d) Kinney, W. A. Tetrahedron Lett. 1993, 34, 2715. (e) Harris, F. L.;
Weiler, L. Tetrahedron Lett. 1987, 28, 2941. (f) Lowinger, T. B.; Weiler,
L. J. Org. Chem. 1992, 57, 6099. (g) Jasperse, C. P.; Curran, D. P. J. Am.
Chem. Soc. 1990, 112, 5601.
(24) (a) Curran, D. P.; Xu, J.; Lazzarini, E. J. Am. Chem. Soc. 1995,
117, 6603. (b) Curran, D. P.; Xu, J. Y.; Lazzarini, E. J. Chem. Soc., Perkin
Trans. 1 1995, 3049.
mode between the acylsilanes and acylgermanes can be traced
to the high bond strength of the silicon-oxygen bond. The silyl
radical is not easily released from 7, but instead undergoes a
“radical-Brook” rearrangement to give 8, which in turn abstracts
hydrogen from tin hydride in a standard bimolecular chain
process. Acylsilanes are thus reagent equivalents of the
hydroxyalkyl radical acceptor synthon. In this respect, they
resemble aldehydes, but with two significant advantages: (1)
the radical-Brook rearrangement prevents fragmentation, and
(2) the rearranged radical 8 can be used for additional
transformations.26
(25) Curran, D. P.; Palovich, M. Synlett 1992, 631.
(26) (a) Tsai, Y. M.; Cherng, C. D. Tetrahedron Lett 1991, 32, 3515.
(b) Tsai, Y.-M.; Tang, K.-H.; Jiaang, W.-T. Tetrahedron Lett. 1993, 34,
1303. (c) Curran, D. P.; Jiaang, W.-T.; Palovich, M.; Tsai, Y.-M. Synlett
1993, 403. (d) Tsai, Y. M.; Chang, S. Y. J. Chem. Soc., Chem. Commun.
1995, 981.
(27) (a) Cirillo, P. F.; Panek, J. S. Org. Prep. Proced. Int. 1992, 24,
555. (b) Qi, H.; Curran, D. P. In ComprehensiVe Organic Functional Group
Transformations; Katritzky, A. R., Meth-Cohn, O., Rees, C. W., Eds.;
Elsevier Science: Oxford, U.K., 1995; Vol. 5, p 409.
(28) The desired germanes have been prepared starting with tetrasub-
stitution of germanium(IV) chloride by Grignard reaction followed by
bromination and LAH reduction of the monobrominated derivative. For
the synthesis of Ph3GeH, see: Harris, D. M.; Nebergall, W. H.; Johnson,
O. H. Inorg. Synth. 1957, 5, 76.
(22) Curran, D. P.; Liu, H. J. Org. Chem. 1991, 56, 3463.

4800 J. Am. Chem. Soc., Vol. 119, No. 21, 1997
Curran et al.
Table 1. 5-exo Cyclization Rate Constant of I(CH2)4COGePH3
(10a)
Scheme 3
entry
temp, °C
[HSnPh3], M
[12]/[11a]a
kc, s-1
1
2
3
4
5
6
7
8
80
80
80
80
25
25
25
25
0.23
0.13
0.065
0.033
0.29
0.14
0.07
0.036
1.39
2.32
4.12
7.75
0.53
0.96
1.90
4.00
6.5 × 106
6.7 × 106
6.2 × 106
6.0 × 106
6.0 × 105
6.0 × 05
6.1 × 05
6.7 × 05
5-exo cyclization
kc(80 °C) ) 6.4 × 106 s-1
kc(25 °C) ) 6.2 × 105 s-1
a Average of 2-3 trials.
syntheses are contained in the Supporting Information. Yields
for the formation of the acylgermanes were typically in the 50-
65% range, although a few instances of both higher and lower
yields were encountered. The acylgermanes for kinetic studies
were purified by flash chromatography and/or recrystallization.
Some of the acylgermanes may not be completely stable to flash
chromatography (as determined by rechromatography of pure
samples), but for the kinetic experiments, sample purity took
priority over yield. These acyltriphenylgermanes proved rea-
sonably stable to atmospheric oxygen (in contrast to acylstan-
nanes, many of which are readily air oxidized). The compounds
were handled with no special precautions; however, long-term
storage requires chilling under nitrogen or argon.
Table 2. Rate Constants for 5-exo Radical Cyclizations of
Acylgermanes
entry
R
kc(25 °C), s-1
kc(80 °C), s-1
1
2
3
4
5
6
7
C6H5
6.2 × 105
5.1 × 105
4.8 × 105
1.0 × 106
6.4 × 106
5.0 × 106
4.9 × 106
1.3 × 107
2.7 × 107
2.9 × 106
3.1 × 105
p-C6H4Me
p-C6H4OMe
p-C6H4F
p-C6H4CF3
Et
2.9 × 105
TMS
2.7 × 104
abstraction from triphenyltin hydride32 sis not as high as that
with tributyltin hydride (which could not be used because it
interfered with integration of the cyclopentanone CH2 reso-
nance), and (2) the reactions are not pseudo-first order.33
However, that the different reaction concentrations consistently
produced the same rate constant within experimental error
suggests that the error introduced by using non-first-order
conditions is not very significant. Rate constants at 25 °C were
consistently about 1 order of magnitude below those at 80 °C.
These competition experiments show that the cyclization rate
is dependent on the germanium ligand. Compared to the phenyl
derivative 10a [kc(80 °C) ) 6.4 × 106 s-1], we observed a
slightly decreased cyclization rate for the p-tolylacylgermane
10b [kc(80 °C) ) 5.0 × 106 s-1] and the p-anisyl derivative
10c [kc(80 °C) ) 4.9 × 106 s-1]. On the other hand, the
fluorinated acylgermanes 10d [kc(80 °C) ) 1.3 × 107 s-1] and
10e [kc(80 °C) ) 2.7 × 107 s-1] were slightly more reactive.
This trend supports a standard picture in which the LUMO
energy of the carbonyl group is an important factor; lowering
the LUMO accelerates the cyclization.
Effects of the Substituents on Germanium. To determine
the effects of the substituents on germanium on the rate constant
for 5-exo radical cyclization, we selected five triarylgermanes
of varying electronic properties along with a triethylgermane
and a tris(trimethylsilyl)germane (Scheme 3). Rate constants
were estimated by conducting standard competition kinetic
studies31 in the presence of 1.3 equiv of triphenyltin hydride.
Reactions (1-3 trials) were conducted at both 25 and 80 °C at
three or four different concentrations in C6D6 in sealed NMR
tubes, and the ratio of the reduced product 11 to cyclopentanone
1
(12) was measured by H NMR integration of the methylene
resonances adjacent to the carbonyl. This is facile since the
CH2 group R to the carbonyl group of acylgermanes 10 is
considerably further downfield from that of cyclopentanone. The
competition involved between cyclization (a unimolecular chain
transfer process24) and reduction (a bimolecular chain transfer
process) is a standard one, as shown in Scheme 3.
The 5-exo cyclization rate constants of the aromatic acylger-
manes correlate roughly linearly with the pKa value of the acids
The data from one representative set of experiments (with
10a) are shown in Table 1. Analogous tables in the Supporting
Information provide the data for all the other substrates. No
other products were detected in any of the experiments.
Furthermore, all the acylgermanes provided cyclopentanone and
the corresponding germyl iodide in high yield (>95%) on both
thermal and photochemical initiation (in the absence of tin
hydride). Therefore, the competition experiments were not
standardized; we assume that the combined yield of the two
products is 100%.
(XC6H4)3GeCOOH34 as well as with the Hammett constant σpara
.
Plots of the cyclization rate constants against the 13C NMR
chemical shifts of the respective carbonyl carbon are also nearly
linear (graphs shown in the Supporting Information). On the
other hand, correlation with the Hammett constants representing
the inductive effect σI or the mesomeric effect σM alone are
not linear. This indicates that both effects influence the radical
acceptor reactivity.
In addition to these trends, the absolute magnitude of these
rate constants is also noteworthy. These acylgermanes are
among the most reactive radical acceptors known for 5-exo
cyclizations. Cyclizations to acylgermanes are about 1 order
The calculated rate constants of all the substrates are compiled
in Table 2. These should be considered estimates for (at least)
two reasons: (1) the precision of the base rate constantshydrogen
(29) (a) Brook, A. G.; Dillon, P. J.; Pearce, R. Can. J. Chem. 1971, 49,
133. (b) Brook, A. G.; Abdesaken, F.; So¨llradl, H. J. Organomet. Chem.
1986, 299, 9. (c) Kraus, C. A.; Flood, E. A. J. Am. Chem. Soc. 1932, 54,
1635. (d) Anderson, H. H. J. Am. Chem. Soc. 1957, 79, 326.
(30) (a) Marshall, J. A.; Welmaker, G. S.; Gung, B. W. J. Am. Chem.
Soc. 1991, 113, 647. (b) Marshall, J. A. Chem. ReV. 1996, 96, 31.
(31) Newcomb, M. Tetrahedron 1993, 49, 1151.
(32) (a) Carlsson, D. J.; Ingold, K. U. J. Am. Chem. Soc. 1968, 90, 7047.
(b) Walling, C.; Cioffari, A. J. Am. Chem. Soc. 1972, 94, 6059. We used
kH [HSnPh3, 80 °C] ) 2.5 × 107 M-1 s-1
.
(33) As usual, the “effective tin hydride concentration” was estimated
as the concentration of tin hydride at 50% conversion of the acylgermane.
(34) Steward, O. W.; Dziedzic, J. E.; Johnson, J. S.; Frohliger, J. O. J.
Org. Chem. 1971, 36, 3480.

Radical Cyclizations of Acylgermanes
J. Am. Chem. Soc., Vol. 119, No. 21, 1997 4801
Table 3. Competitive Rate Experiments with 13
of magnitude faster than similar reactions of aldehydes and
alkenes, and closer to 2 orders of magnitude faster than additions
to nitriles and unactivated alkynes. More sterically correct
comparisons, for example, with 2-substituted alkenes or ketones,
are even more impressive. For example, simple 5-exo cycliza-
tions to unactivated ketones do not occur at all, so the
acylgermanes are probably at least 104 or 105 times more
reactive than the corresponding methyl ketones. Acylgermanes
react at rates comparable to those of some of the best radical
acceptors including hydrazones and activated alkenes. Unfor-
tunately, these high rate constants for 5-exo cyclizations did
not translate into successful bimolecular reactions. Chains did
not propagate in attempts to add several halides to a few simple
acylgermanes.35
experiment
concn of 13, M
ratio 14/15
kc, s-1
1
2
3
4
0.1
0.1
0.2
0.2
3.03/1
2.70/1
1.53/1
1.71/1
5.0 × 106
4.4 × 106
4.6 × 106
5.2 × 106
The nonaromatic germanium ligands exhibited lower reactiv-
ity than the triphenylgermane group. The radical derived from
triethylacylgermane 10f [kc(80 °C) ) 2.9 × 106 s-1] cyclized
about 3 times more slowly while that derived from tris-
(trimethylsilyl) derivative 10g [kc(80 °C) ) 3.1 × 105 s-1
]
cyclized about 20 times more slowly. Nonetheless, in the yield
experiments described above (which are directly relevant to
synthetic applications), both compounds provided cyclopen-
tanone in >95% yield. The decreased reactivity of these two
substrates could be due to a higher LUMO energy or to a steric
effect (or both).
In contrast to the good results with the secondary radical
precursor 13, the cyclization of the tertiary radical precursor
16 was not nearly as efficient. Photolysis of the acylgermane
16 with a sunlamp for 30 min at room temperature resulted in
only 44% yield of 2,2-dimethylcyclopentanone (17) as measured
by H NMR (eq 4b). The proton NMR spectrum showed that
the acylgermane 16 was consumed, but no other products
besides 17 were evident.
Because acylgermanes operate by a UMCT mechanism, they
effectively combine the radical cyclization and chain transfer
steps. The rate constants for these processes are crucial in many
applications. Excellent tuning of the radical acceptor reactivity
in 5-exo cyclizations of acylgermanes is available by varying
the germanium ligands. The efficiency of the radical acceptor
can readily be varied over about 2 orders of magnitude without
changing the yield of the final product. While a reactive
acceptor is desired for many processes, tandem reactions often
require radical traps of lower reactivity to allow time for
intermediate radicals to do other reactions prior to chain transfer.
In this respect, acylgermanes should be excellent reagents for
radical cascades.
Ring Size and Radical Substituent Effects. To study the
scope of acylgermane radical cyclizations, we prepared a series
of substrates in which both the substitution of the radical
(primary, secondary, tertiary) and the chain length connecting
the radical and the acylgermane were varied. The preparation
of the precursors for these experiments is fully described in the
Supporting Information. Preparative experiments (usually by
sunlamp photolysis) were conducted to determine the yield of
cyclization. Because the products of these simple substrates
are volatile, yields were usually determined by 1H NMR against
internal standards. Competitive experiments were conducted
with triphenyltin hydride to estimate rate constants for cycliza-
tion. With these substrates, the competitive experiments were
not as extensive. Typically one or two trials at two or three
different concentrations were conducted. All competitive
experiments were carried out at 80 °C.
1
The results of the competition experiments performed with
acylgermane 16 are shown in Table 4. From these experiments,
the rate constant was calculated to be about (1-2) × 104 s-1 at
80 °C. This is over 2 orders of magnitude lower than the
primary and secondary 5-exo cyclization rate constants. These
experiments have considerably more error than the prior
competition experiments for two reasons. First, there is a
problem with integrating the proton signals of the cyclic product
17 because even at low reaction concentrations it is formed in
only small amounts. Increasing the reaction concentration from
0.003 to 0.005 M should result in a decrease in the ratio of the
products, not an increase as was found in the two experiments
(Table 4). Second, the assumption that the mass balance of
the two products in the competition experiments is 100% may
not be a good one in this case. Considering these problems,
the estimated rate constant is probably on the high end. Errors
aside, the bottom line is clear: the tertiary radical generated
from 16 is at least 2 orders of magnitude less reactive that its
primary and secondary counterparts.
The reason for the dramatic rate reduction in the cyclization
of the tertiary radical is not immediately clear. FMO effects
cannot be invoked since the tertiary radical has a higher SOMO
than either primary or secondary radicals, and should therefore
be expected to cyclize more rapidly. It is possible that steric
effects are in play as suggested by the transition state models
shown in Figure 2. It is possible that the rate constant for
cyclization decreases when the R1 substituent facing the large
triphenylgermanium group is not hydrogen.
The results of the preparative cyclization with the secondary
radical precursor 13 are shown in eq 4a. Cyclization of 13 by
photolysis with a sunlamp proceeded smoothly over the course
of 30 min at room temperature and provided 2-methylcyclo-
pentanone (14) in 80% yield (as measured by 1H NMR against
an internal standard). The competition experiments with
triphenyltin hydride (Table 3) provided an estimated rate
This analysis suggests that there is a high preference in
secondary radicals for the methyl group to occupy a position
trans to the triphenylgermyl group. If, on top of this, the
acylgermane exhibited a high preference for either a “boatlike”
or “chairlike” orientation, then high stereoselectivity would be
expected in the cyclization. The stereoselectivity of the
cyclization of a simple acylgermane cannot be probed directly
(because the stereocenter bearing the R-germylalkyl radical is
constant for cyclization of about 5 × 106 s-1
. This is
comparable to the rate constant for cyclization of the primary
radical (entry 1 in Table 2).
(35) Curran, D. P.; Diederichsen, U. J. Organomet. Chem., in press.

4802 J. Am. Chem. Soc., Vol. 119, No. 21, 1997
Curran et al.
Table 5. Competitive Rate Experiments with 21
experiment
concn of 21, M
ratio 22/23
kc, s-1
1
2
3
0.05
0.05
0.20
0.93/1
1.10/1
0.26/1
1.1 × 106
1.4 × 106
1.4 × 106
Figure 2. Transition state models for acylgermane radical cyclizations.
Table 4. Competitive Rate Experiments with 16
experiment
concn of 16, M
ratio 17/18
kc, s-1
1
2
0.003
0.005
0.24/1
0.38/1
9.4 × 103
2.6 × 104
carbonyl acceptors is in direct contrast to that of alkenes and
alkynes, where 5-exo cyclizations are typically significantly
faster than 6-exo cyclizations. The rapidity of the 6-exo
cyclizations of aldehydes has been attributed to the “tighter”
transition states in radical additions to carbonyl groups due to
the stronger SOMO/LUMO interaction. In particular, the
forming bond in cyclizations to carbonyls is calculated to be
significant shorter than its counterpart in cyclizations to
alkenes.19b In a simple picture,36b the “looser” transition states
of alkenes (with their long forming bonds) resemble cyclohep-
tane in ring strain while the tighter transition states in 6-exo
cyclizations of carbonyl compounds more closely resemble the
cyclohexane product. This analysis is also consistent with the
postulate that the triphenylgermyl group prefers to be adjacent
to a hydrogen atom in a cyclization with a tight transition state.
The cyclizations of the secondary 6-exo substrate 24 were
not as well behaved as the primary one. Preparative cyclization
of 24 provided 2-methylcyclohexanone (25) in only 44% yield
(eq 6b). Furthermore, the rate constant measurements (Table
6) provided cyclization rate “constants” that decreased with
decreasing tin hydride concentration.37 Taken together, these
observations suggest that 1,5-hydrogen transfer is competing
with cyclization. If we assume that the ratio of cyclized to
reduced products is equal to the ratio kc/k1,5, we can estimate
that kc is in the vicinity of (5-6) × 105 s-1 and k1,5 is about
(2-3) × 105 s-1. These are just approximations; however, even
simple comparison of the yields of the primary and secondary
6-exo cyclizations suggests that more 1,5-hydrogen transfer
occurs in the secondary case. This is surprising. Competing
1,5-hydrogen transfer is a common problem in 6-exo cyclizations
of all sorts, and it also competes to some extent in the radical
cyclization reactions of acylsilanes. However, in both cases,
the cyclization products seem to predominate. In view of the
poor results with the tertiary 5-exo substrate 16, a tertiary 6-exo
analog was not investigated.
rapidly lost by fragmentation), but it can be probed indirectly
simply by adding a substituent to the chain. According to the
calculations of Houk and Spellmeyer,36 C3 substituents have a
good “pseudoequatorial” preference (see R3 in Figure 2) so they
should provide a good probe.
To test this idea, we prepared the unsaturated acylgermane
19 and cyclized it according to the procedure of Kiyooka (eq
5).21 It follows from the above analysis that the trans product
should predominate if a chairlike transition state is favored and
the cis product should predominate if a boatlike transition state
is favored. Unfortunately, the results were not very informative.
Cyclization of 19 by sunlamp (30 min) or sunlight (8 h)
photolysis provided a 1/1 mixture of stereoisomers 20 in good
yield. In view of the low selectivity, no attempt was made to
separate the isomers or assign the configuration. Assuming that
the C3 methyl group does indeed have a reasonable pseu-
doequatorial preference, the results could mean that this
cyclization occurs competitively through boatlike and chairlike
transition states, or they could mean that the postulate that the
C1 substituent and the triphenylgermyl group must be trans in
the cyclization is incorrect.
To investigate the potential for 6-exo radical cyclization to
an acylgermane, the preparative and competitive cyclizations
of the acylgermane 21 were studied (eq 6a). When the
acylgermane 21 was photolyzed with a sunlamp in benzene for
30 min at room temperature, an 87% yield of cyclohexanone
(22) was found. No other products were evident in the proton
1H NMR spectrum. Table 5 shows the results of the competition
experiments performed on the acylgermane 21 following the
standard competition experimental procedure. From these
experiments, the cyclization rate constant was estimated to be
1.3 × 106 s-1 at 80 °C. Therefore, the 6-exo primary radical
cyclization is just as efficient as the primary and secondary 5-exo
radical cyclizations.
Attempts to conduct the parent 4-exo and 7-exo cyclizations
were not successful (eqs 7a,b). Irradiation of 27 for 30 min
resulted in its complete consumption, but the crude product
1
exhibited a complex H NMR spectrum in which the protons
(36) (a) Houk, K. N.; Paddon-Row, M. N.; Spellmeyer, D. C.; Rondan,
N. G.; Nagase, S. J. Org. Chem. 1986, 51, 2874. (b) Curran, D. P.; Porter,
N. A.; Giese, B. Stereochemistry of Radical Reactions: Concepts, Guide-
lines, and Synthetic Applications; VCH: Weinheim, 1996, p 283.
(37) This could also be shown by labeling experiments with Bu3SnD.
See: Palovich, M. Ph.D. Thesis, University of Pittsburgh, 1996.
These results are reminiscent of the radical cyclizations of
aldehydes, where rate constants for 5-exo and 6-exo cyclizations
of the parent substrates are very similar. This behavior of

Radical Cyclizations of Acylgermanes
J. Am. Chem. Soc., Vol. 119, No. 21, 1997 4803
Table 6. Competitive Rate Experiments with 24
Scheme 4
experiment
concn of 24, M
ratio 25/26
kc, s-1
1
2
3
0.02
0.01
0.005
1.12/1
1.73/1
2.11/1
3.2 × 105
2.6 × 105
1.6 × 105
Figure 3. Comparison of 3-exo cyclization rate constants at 25 °C.
of cycloheptanone (28) could not be identified. Attempts at
to provide a tertiary radical, 36, which is then reduced by
triphenyltin hydride. Thus, the triphenylgermyl group is not
involved in the radical chemistry, and the acylgermane simply
participates in the standard 1,2-shift observed with many other
types of multiply bonded functional groups. The rate constant
for this shift was estimated as about 3 × 107 s-1 by the
competition experiments shown at the bottom of Scheme 4.
Comparison rate constants for related shifts are provided in
Figure 3.39 This comparison is done at 25 °C, and to estimate
the rate constant for the acylgermane at this temperature, we
simply divided the 80 °C rate constant by 10. Once again, the
acylgermane proves to be an excellent radical acceptor in
comparison with alkenes, ketones, and aldehydes.
thermal initiation at 80 °C with AIBN and 5% Ph3GeH returned
starting material even after 12 h. Chain propagation did not
occur. Reduction of 27 with a stoichiometric quantity of
triphenyltin or triphenylgermanium hydride cleanly provided the
product of reductive debromination (not shown). All these
results suggest that the 7-exo cyclization rate constant is below
the limit needed to propagate a chain in solution (probably <103
s-1).
The 4-exo cyclization precursor exhibited behavior similar
to that of 27, except that it underwent slow, thermal cyclization
to germyltetrahydrofuran 31. This reaction probably occurs by
SN2 substitution, and may be a useful method to prepare such
cyclic germyl enol ethers. Analogous reactions of acylsilanes
are known.38
Because unimolecular chain transfer reactions are especially
powerful for conducting cascade radical reactions,24 we closed
the basic study of acylgermane radical cyclizations with a few
preliminary tandem reactions. Two attempts to develop radical
annulations were not successful, as summarized in eq 9a,b.
In contrast to these disappointments, the 3-exo cyclization
precursor 32 provided especially interesting results. We choose
the gem-dimethyl-substituted precursor to preclude any pos-
sibility of â-elimination of the radical precursor in the acylger-
mane. Due to the Thorpe-Ingold effect, this group will also
dramatically accelerate the 3-exo cyclization. Attempts to
isomerize 32 by using a catalytic amount of Ph3SnH were not
successful; as with the 7-exo and 4-exo cases, 32 was not
consumed. However, when 32 was reduced with 1.3 equiv of
Ph3SnH at 0.05 M, a smooth reductive rearrangement occurred,
as shown in eq 8. Acylgermane 33 was isolated in 90% yield
by flash chromatography.
Attempts to initiate chain reactions with either 32 or 29 and
methyl acrylate resulted only in very slow conversions to
mixtures of products. However, reductive addition of 32 to
methyl acrylate by the standard (stoichiometric) tin hydride gave
40 in 65% isolated yield. This product results from a 1,2-shift
followed by addition to methyl acrylate and hydrogen transfer.
This result suggests that the annulation fails because the ester-
substituted radical does not add to the acylgermane.
The mechanism for the formation of 33 is shown in Scheme
4. Radical generation followed by 3-exo cyclization provides
35, which apparently does not eliminate the triphenylgermyl
radical to provide cyclopropanone 38. Instead, 35 fragments
In contrast, an intramolecular variant of this type of sequence
was successful. Radical precursor 42 was readily available from
dimethyl malonate by a series of standard alkylation reactions
(38) Tsai, Y. M.; Nieh, H. C.; Cherng, C. D. J. Org. Chem. 1992, 57,
7010.
(39) Giese, B.; Heinrich, N; Horler, H.; Koch, W; Schwarz, H. Chem.
Ber. 1986, 119, 3528.

4804 J. Am. Chem. Soc., Vol. 119, No. 21, 1997
Curran et al.
Scheme 5
constant for cyclization over a wide range by varying the
germanium substituent is an asset. The reactions occur by a
mechanism analogous to that for vinylstannanes, but they are
tin free. Preparative experiments are very easy to conduct, and
separation of the R3GeX byproduct by chromatography did not
prove difficult. Finally, the combination of good radical-
accepting properties and the UMCT chain mechanism21 makes
acylgermanes good candidates for conducting sequences of
radical reactions.
Limitations are also suggested by this work, and these involve
mainly the character of the attacking radical. Tertiary radicals
do not appear to be good candidates for these types of
cyclizations, and the presence of electron-withdrawing groups
on the radical may also cause problems. More work is needed
to better ascertain what radical substituent pairs work well with
acylgermane acceptors and why.
The complementary nature of acylgermane and acylsilane
radical reactions is an especially valuable asset. These func-
tional groups can be prepared in similar ways, have similar
properties, and have grossly similar behavior in nonradical
reactions. However, their radical chemistry is very different.
Both are excellent radical acceptors, but the subsequent evolu-
tion of products in acylsilane chemistry is dictated by the radical-
Brook rearrangement and ultimately produces alcohols, while
acylgermanes undergo simple fragmentation to ketones.
to introduce the side chains followed by conversion of the final
aldehyde to an acylgermane. Irradiation of â-bromoacylger-
mane for 30 min followed by flash chromatography provided
bicyclic ketone 43 in 39% yield through the sequence of steps
indicated in Scheme 5. The modest yield in this cyclization is
almost certainly attributed to the expected lack of stereoselec-
tivity in the 5-exo cyclization step.36b The trans isomer of 46
is reluctant to cyclize to a trans-fused diquinane and probably
decomposes via other pathways.
Conclusions
The results of this study suggest that acylgermanes are indeed
suitable equivalents of the carbonyl radical acceptor synthon
for many synthetic applications. For primary radicals at least,
acylgermanes are among the best of the known classes of radical
acceptors. The speed of the cyclizations suggests that many
types of 5-exo and 6-exo cyclizations will be possible, including
those which are generally considered to be slow. The simple
4-exo and 7-exo cyclizations did not work, though in especially
favorable cases it is probably possible to use acylgermanes in
these types of reactions as well. The ability to tune the rate
Acknowledgment. We thank the National Science Founda-
tion for support of this work. U.D. thanks the Deutsche
Forschungsgemeinschaft for a postdoctoral fellowship.
Supporting Information Available: Full experimental
details for all the reactions and compounds reported in the paper
(30 pages). See any current masthead page for ordering and
Internet access instructions.
JA970219M