Trimethyl phosphite as a trap for alkoxy radicals formed from the ring opening of oxiranylcarbinyl radicals. Conversion to alkenes. Mechanistic applications to the study of C-C versus C-O ring cleavage

Article abstract of DOI:10.1021/ja020761x

Trimethyl phosphite, (MeO)₃P, is introduced as an efficient and selective trap in oxiranylcarbinyl radical (2) systems, formed from haloepoxides 8-13 under thermal AIBN/n-Bu₃SnH conditions at about 80 °C. Initially, the transformations of 8-13, in the absence of phosphite, to allyl alcohol 7 and/or vinyl ether 5 were measured quantitatively (Table 1). Structural variations in the intermediate oxiranylcarbinyl (2), allyloxy (3), and vinyloxycarbinyl (4) radicals involve influences of the thermodynamics and kinetics of the C-O (2 → 3, k₁) and C-C (2 → 4, k₂) radical scission processes and readily account for the changes in the amounts of product vinyl ether (5) and allyl alcohol (7) formed. Added (MeO)₃P is inert to vinyloxycarbinyl radical 4 and selectively and rapidly traps allyloxy radical 3, diverting it to trimethyl phosphate and allyl radical 6. Allyl radicals (6) dimerize or are trapped by n-Bu₃SnH to give alkenes, formed from haloepoxides 8, 9, and 13 in 69-95% yields. Intermediate vinyloxycarbinyl radicals (4), in the presence or absence of (MeO)₃P, are trapped by n-Bu₃SnH to give vinyl ethers (5). The concentrations of (MeO)₃P and n-Bu₃SnH were varied independently, and the amounts of phosphate, vinyl ether (5), and/or alkene from haloepoxides 10, 11, and 13 were carefully monitored. The results reflect readily understood influences of changes in the structures of radicals 2-4, particularly as they influence the C-O (k₁) and C-C (k₂) cleavages of intermediate oxiranylcarbinyl radical 2 and their reverse (k_-1, k_-2). Diversion by (MeO)₃P of allyloxy radicals (3) from haloepoxides 11 and 12 fulfills a prior prediction that under conditions closer to kinetic control, products of C-O scission, not just those of C-C scission, may result. Thus, for oxiranylcarbinyl radicals from haloepoxides 11, 12, and 13, C-O scission (k₁, 2 → 3) competes readily with C-C cleavage (k₂, 2 → 4), even though C-C scission is favored thermodynamically.

Full text of DOI:10.1021/ja020761x

Published on Web 02/22/2003
Trimethyl Phosphite as a Trap for Alkoxy Radicals Formed
from the Ring Opening of Oxiranylcarbinyl Radicals.
Conversion to Alkenes. Mechanistic Applications to the Study
of C-C versus C-O Ring Cleavage
Bangwei Ding and Wesley G. Bentrude*
Contribution from the Department of Chemistry, UniVersity of Utah, Salt Lake City, Utah 84112
Received May 30, 2002; Revised Manuscript Received October 17, 2002 ; E-mail: bentrude@chemistry.utah.edu
Abstract: Trimethyl phosphite, (MeO)3P, is introduced as an efficient and selective trap in oxiranylcarbinyl
radical (2) systems, formed from haloepoxides 8-13 under thermal AIBN/n-Bu3SnH conditions at about
80 °C. Initially, the transformations of 8-13, in the absence of phosphite, to allyl alcohol 7 and/or vinyl
ether 5 were measured quantitatively (Table 1). Structural variations in the intermediate oxiranylcarbinyl
(2), allyloxy (3), and vinyloxycarbinyl (4) radicals involve influences of the thermodynamics and kinetics of
the C-O (2 f 3, k1) and C-C (2 f 4, k2) radical scission processes and readily account for the changes
in the amounts of product vinyl ether (5) and allyl alcohol (7) formed. Added (MeO)3P is inert to
vinyloxycarbinyl radical 4 and selectively and rapidly traps allyloxy radical 3, diverting it to trimethyl phosphate
and allyl radical 6. Allyl radicals (6) dimerize or are trapped by n-Bu3SnH to give alkenes, formed from
haloepoxides 8, 9, and 13 in 69-95% yields. Intermediate vinyloxycarbinyl radicals (4), in the presence or
absence of (MeO)3P, are trapped by n-Bu3SnH to give vinyl ethers (5). The concentrations of (MeO)3P and
n-Bu3SnH were varied independently, and the amounts of phosphate, vinyl ether (5), and/or alkene from
haloepoxides 10, 11, and 13 were carefully monitored. The results reflect readily understood influences of
changes in the structures of radicals 2-4, particularly as they influence the C-O (k1) and C-C (k2) cleavages
of intermediate oxiranylcarbinyl radical 2 and their reverse (k-1, k-2). Diversion by (MeO)3P of allyloxy
radicals (3) from haloepoxides 11 and 12 fulfills a prior prediction that under conditions closer to kinetic
control, products of C-O scission, not just those of C-C scission, may result. Thus, for oxiranylcarbinyl
radicals from haloepoxides 11, 12, and 13, C-O scission (k1, 2 f 3) competes readily with C-C cleavage
(k2, 2 f 4), even though C-C scission is favored thermodynamically.
contrast, primary and secondary alkyl radicals (R•′′) react
reversibly with trialkyl phosphites [(R′O)3P] to form phos-
Introduction
The differences in reactivities of free radicals toward tri-
coordinate phosphorus (Z3P) have been used effectively in the
preparation of vinylphosphonates from 6-bromo-1-alkynes1 and
substituted cyclopentanes and cyclohexanes from 5- and 6-bromo-
aldehydes.2 In both cases, an alkyl radical, essentially unreactive
toward Z3P, was allowed to cyclize to a very reactive species,
a vinyl1 or alkoxy2 radical, which was then rapidly trapped by
Z3P to give a preparatively useful product.
More specifically, alkoxy radicals (RO•) react at near diffusion
controlled rates with trialkyl phosphites to give phosphoranyl
radicals [ROP(OR′)3]• that typically are rapidly deoxygenated
to generate alkyl radicals (R•) and phosphate [OP(OR′)3].3 In
phoranyl radicals [R′′-P(OR′)3]•. These intermediates only form
product when the intermediate has available a very rapid,
product-forming â-scission step to give a relatively stable
radical, R′•, such as benzyl, and a product alkylphosphonate,
R-P(O)(OR′)2. We report here the use of these reactivity
differences, under typical thermal AIBN/Bu3SnH conditions in
the presence of (MeO)3P, to effect the conversion of bromo-
epoxides 8, 9, and 13 to the corresponding alkenes: indene (17),
3-phenylpropene (23), and isomeric indenes 45 and 46.
Because these transformations defunctionalize 8, 9, and 13,
they likely are of more mechanistic (Scheme 1) than method-
ological interest. In this regard, (MeO)3P is found to be a very
reactive and selective trap for the allyloxy radical (3) of
mechanistic Scheme 1. Previously, hydrogen-donors PhSH4 and
n-Bu3SnH5 have been used in mechanistic studies to trap free
radical intermediates, 2-4. Trimethyl phosphite has the advan-
tage that it selectively traps intermediate 3, but not 2 or 4, and,
(1) Kim, S.; Oh, D. H. Synlett 1998, 525.
(2) Jiao, X.-Y.; Bentrude, W. G. J. Am. Chem. Soc. 1999, 121, 6088.
(3) For reviews of the chemistry of phosphoranyl radicals, see: (a) Bentrude,
W. G. In The Chemistry of Organophosphorus Compounds; Hartley, F.
R., Ed.; Wiley: Sussex, 1990; Vol. 1, p 531. (b) Bentrude, W. G. In
ReactiVe Intermediates; Abramovitch, R. A., Ed.; Plenum: London, 1983;
Vol. 3, p 199. (c) Bentrude, W. G. Acc. Chem. Res. 1982, 15, 117. (d)
Roberts, B. P. In AdVances in Free Radical Chemistry; Williams, G. H.,
Ed.; Heyden and Sons: London, 1980; Vol. 6, p 225. Schipper, P.; Janzen,
E. H. J. M.; Buck, H. M. Top. Phosphorus Chem. 1977, 191, 407. Bentrude,
W. G. In Free Radicals; Kochi, J. K., Ed.; Wiley-Interscience: New York,
1973; p 54.
(4) Krishnamurthy, V.; Rawal, V. H. J. Org. Chem. 1997, 62, 1572. For a
lower limit of the rate constant for C-O scission (Scheme 1) of 1 × 1010
s-1 at 70 °C, see: Krosley, K. W.; Gleicher, G. J. 1993, 6, 228.
(5) Ziegler, F. E.; Petersen, A. K. J. Org. Chem. 1995, 60, 2666.
9
3248
J. AM. CHEM. SOC. 2003, 125, 3248-3259
10.1021/ja020761x CCC: $25.00 © 2003 American Chemical Society

Trimethyl Phosphite as a Trap for Alkoxy Radicals
A R T I C L E S
control conditions, allyl alcohol 7 is formed (R1 ) H or Ph; R2
) H or alkyl). Radicals 3 and 4 are readily reclosed to
oxiranylcarbinyl radical 2, and, ultimately, an equilibrium ratio
of 4/3, determined primarily by the influence of R1 and R2 on
the stabilities of radicals 3 and 4, potentially can be established.
Thus, for bromoepoxide 10 (1, R1 ) Ph, R2 ) H), at relatively
low concentrations of n-Bu3SnH, close to thermodynamic
control conditions prevail,5 and allyl ether 24 (5, R1 ) Ph, R2
) H) is the dominant product.5 Conversely, at high n-Bu3SnH
concentrations, nearly kinetic control is operative, which gives
predominantly allyl alcohol (7, R1 ) Ph, R2 ) H).
Scheme 1
The thermal AIBN/n-Bu3SnH-induced rearrangements of all
but bromoepoxide 9 have been investigated previously; yet,
except for bromoepoxide 10, careful quantitative studies of
product distributions have not been reported. We have examined
the product distribution from haloepoxide radical precursors
8-13 with and without added trimethyl phosphite and find that
trimethyl phosphite at high concentrations traps allyloxy radical
3 in competition with its reclosure to intermediate 2 with
formation of alkene (or allyl radical dimers), instead of alcohol
7, along with trimethyl phosphate. Moreover, the equilibrium
of Scheme 1 can be shifted by phosphite strongly in the direction
of 3 to give a reduced yield of vinyl ether 5 along with
phosphate.
as we show, it can be used in conjunction with n-Bu3SnH in
experiments in which the concentrations of the two traps are
varied independently. In fact, both (MeO)3P and n-Bu3SnH react
with allyloxy radical 3. However, the phosphite, a volatile liquid,
can be used as solvent (8 M). Even at equal concentrations, the
phosphite is considerably more reactive than n-Bu3SnH (kP/kH
ca. 8, Scheme 1, see Discussion). Moreover, n-Bu3SnH is
normally used at concentrations less than 1 M, which makes
trimethyl phosphite the essentially exclusive trap for allyloxy
radical 3, while vinyloxycarbinyl radical 4 is trapped only by
n-Bu3SnH.6
The kinetics and thermodynamics of the competition between
C-C and C-O bond cleavage for oxiranylcarbinyl radicals
(Scheme 1), generated under thermal AIBN/n-Bu3SnH condi-
tions, have received extensive recent study. For haloepoxide 1
with R1 ) H or Ph and R2 ) H or alkyl, the formation from 2
of allyloxy radical 3 instead of C-C scission product 4 has
been found both experimentally5,9 and theoretically10,11 to be
kinetically favored. Carbon-carbon scission gives the thermo-
dynamically more stable radical 4.5,9-11 Under kinetic product
Furthermore, changes in n-Bu3SnH and phosphite concentra-
tions strongly affect product distributions and provide useful
insights into the effects of variations in the structure of the
oxiranylcarbinyl radical (2) on the kinetic array represented by
Scheme 1. In this regard, it was suggested in the pivotal work
of Ziegler and Petersen5 that under conditions of kinetic control,
products from C-O cleavage of radical 2 (Scheme 1), rather
than the reported products of C-C scission, might be observed
for haloepoxides 1112 (1, R1 ) R2 ) Ph) and 1212 (1, R1 )
1-naphthyl, R2 ) H), just as they had seen with 10.5,12 Indeed,
we confirm that at low n-Bu3SnH concentrations in the absence
of phosphite, haloepoxides 11, 12, and also 1313 give the
thermodynamic product, allyl ether 5, exclusively or very
predominantly (Table 1). However, with 5 M (MeO)3P present
(conditions which are closer to kinetic control), the reactions
of haloepoxides 11-13 result in deoxygenation of allyloxy
radical 3, to yield (MeO)3PO in 52, 49, and 70% yields,
respectively (Table 1). Consequent reductions in vinyl ether 5
yields are noted along with the formation of alkenes (Scheme
1). The deoxygenation of 3 from 12 (R1 ) R2 ) Ph) occurs,
although when R2 is phenyl (radicals from chloroepoxide 12),
ring reclosure of 3 to the resonance stabilized oxiranylcarbinyl
radical (2) should be strongly favored both kinetically and
thermodynamically (see Discussion). For 1 with R1 ) Ph, R2
) H (compound 10) at high trimethyl phosphite concentrations,
(6) Neat n-Bu3SnH corresponds to a 3.7 M concentration, while neat (MeO)3P
is 8.5 M in phosphite. At the same time, the first-order rate constant7 for
abstraction of hydrogen from Bu3SnH (kH) by tert-butoxy radical at 22 °C
is 2.0 × 108 s-1 M-1. For reaction of the same radical with (EtO)3P, kP at
28 °C is measured8 to be 1.7 × 109 s-1 M-1. (MeO)3P is generally somewhat
more reactive than (EtO)3P toward free radicals.3 This makes neat (MeO)3P
at least an order of magnitude more reactive than neat n-Bu3SnH toward
tert-butoxy radical. Moreover, the deoxygenation experiments of Tables 1
and 2, where deoxygenation of 3 is maximized, were run at (MeO)3P/n-
Bu3SnH ratios of 500 and 400, respectively.
(7) Scaiano, J. C. J. Am. Chem. Soc. 1980, 102, 5399.
(8) Griller, D.; Ingold, K. U.; Patterson, L. K.; Scaiano, J. C.; Small, R. D., Jr.
J. Am. Chem. Soc. 1979, 101, 3780. Roberts, B. P.; Scaiano, J. C. J. Chem.
Soc., Perkin Trans. 2 1981, 905.
(9) Without resonance stabilization of 4, alcohol from allyloxy radicals (3) is
formed exclusively. See, for example: Edwards, A. J.; Hird, N. W.;
Marples, B. A.; Rudderman, J. A.; Slawin, A. M. Z. Tetrahedron Lett.
1997, 38, 3599. Rawal, W. H.; Zhong, H. M. Tetrahedron Lett. 1993, 55,
5181. Breen, A. P.; Murphy, J. A. J. Chem. Soc., Chem. Commun. 1993,
191. Barton, D. H. R.; Motherwell, R. S. H.; Motherwell, W. B. J. Chem.
Soc., Perkin Trans. 1 1981, 2363. Johns, A.; Murphy, J. Tetrahedron Lett.
1988, 837. Davies, A. G.; Muggleton, B. J. J. Am. Chem. Soc. 1979, 101,
9. Davies, A. G.; Tse, M. W. J. Organomet. Chem. 1978, 155, 25. Dobbs,
A. J.; Gilbert, B. C.; Laue, H. A.; Norman, R. O. C. J. Chem. Soc., Perkin
Trans. 2 1976, 1044.
(10) Pasto, D. J. J. Org. Chem. 1996, 61, 252.
(11) Smith, D. M., Nicolaides, A.; Golding, B. T.; Radom, L. J. Am. Chem.
Soc. 1998, 120, 10223.
(12) Dickinson, J. M.; Murphy, J. A.; Patterson, C. W. Wooster, N. F. J. Chem.
Soc., Perkin Trans. 1 1990, 1179.
(13) Murphy, J. A.; Patterson, C. W. J. Chem. Soc., Perkin Trans. 1 1993, 405.
9
J. AM. CHEM. SOC. VOL. 125, NO. 11, 2003 3249

A R T I C L E S
Ding and Bentrude
Table 1. Summary of Effects of Trimethyl Phosphite on Reactions
of 8-13, One-Pot Conditionsa
(16), was isolated in 70% yield. The formation of 15 by
rearrangement of alcohol 16 under free radical (thermal AIBN/
n-Bu3SnH), thermal, or weakly basic conditions is well docu-
mented.14 The high yield of 1-indanone (15) was verified in a
separate syringe-pump reaction (87% GC yield) and in a one-
pot reaction (Method B without phosphite, 91% GC yield, Table
1).
yield of alcohol
yield of vinyl ether
without
with
without
(MeO)3P
with
(MeO)3P
yield of
alkene
yield of
(MeO)3PO
reactant (MeO)3P (MeO)3P
8
9
10
11
12
13
15, 91b
21, 93
25, 4
4b
<1d
0
0
0
14, 0
22, 0
24, 89
33, 92e
30, 93j
0
0
13
43
52k
10
17, 73
89
97
82
52
49
70
23a,b,c, 95h,i
27, 33f
In contrast, addition via syringe pump of n-Bu3SnH (0.04
M) and AIBN to bromoepoxide 8 (0.02 M) in refluxed benzene
that contained only modest amounts of (MeO)3P (ca 0.4 M)
resulted in a greatly reduced yield of 1-indanone, 15 (5%), which
was replaced by indene, 17 (67%), and 1-indenyl radical dimer,
18 (9%), all quantified by GC analysis. Evidently, the kinetically
favored (eq 1) allyloxy radical 19 (radical 3 of Scheme 1) is
deoxygenated (kP ) 1.7 × 109 s-1 M-1 for reaction of tert-
butoxy radical with (EtO)3P at 28 °C8) to generate 1-indenyl
radicals which are primarily reduced by n-Bu3SnH15 to form
indene (17). Use of an increased concentration of n-Bu3SnH in
a one-pot reaction with higher phosphite concentration (5 M,
Experimental Section, one pot - Method B, but with 0.6 M
n-Bu3SnH) greatly reduced the yield of dimer 18 by trapping a
higher percentage of 1-indenyl radicals. The amount of 15 (1%)
formed was decreased. However, the net yield of indene (17,
74%) was not greatly increased, because side product 20, from
the free radical addition of n-Bu3SnH to indene (17), was formed
in 11% yield (GC).
34, 0e
31, 0
35-37,m 16
32,n 34
40, 12d,e
4c,d 41, 42, 84e,l
45, 46, 69g
a Unless otherwise indicated, reactions were run by Method B at 0.01
M substrate, 0.02 M n-Bu3SnH, 5 M (MeO)3P, (n-PrO)3PO internal standard.
Reactions without phosphite were run at the same concentrations of tin
hydride (except for 11e and 13e), substrate, and AIBN in benzene solvent.
b As 1-indanone. c As 2-methyl-1-indanone. d Detected by GC, estimated
by comparison to 1% stock solution of 21. e At 0.005 M n-Bu3SnH.
f Cinnamyl radical dimers also formed (28 and 29), 42% total yield. g Total
2-methylindene (45) plus 2-methyleneindane (46); ratio 45/46 ) 1.5.
h Combined yield of three isomers; 21% trans-4-phenyl-2-propene (23b);
54% cis-23a; 20% 4-phenyl-1-propene (23c). i At 0.02 M (TMS)3SiH, 92%
23a. j 52% cis-30, 41% trans-30. k 30, cis plus trans. l 46% 41, 38% 42.
m GC yields: 35 (3%), 36 (6%), 37 (7%). n Cis plus trans.
Table 2. Summary of Effects of Trimethyl Phosphite on Reactions
of 8-13, Syringe Pump Methoda
substrates
8
10
11
12
13
yield of vinyl ether
yield of trimethyl
phosphate
0 (0)b
8 (13) 34 (43) 42 (52) 23c(10)
96 (89) 85 (82) 58 (52) 42 (49) 69 (70)
a Method A. Addition of 0.04 M n-Bu3SnH and AIBN (6 mg) to neat
(MeO)3P (ca. 8 M) containing 0.01 M substrate at 80 °C; (n-PrO)3PO
internal standard. b Numbers in parentheses are from Table 1. c Combined
yield of 41 plus 42.
and with sufficient amounts of added radical trap n-Bu3SnH,
the C-C and C-O scission steps appear to become close to
being irreversible. From the ratio of phosphate to vinyl ether,
an estimation of k1/k2 (5.8) is obtained that is similar to the
value reported for oxiranylcarbinyl radical 2 from 10 previously
(k1/k2 ) 4.2),5 with n-Bu3SnH as the sole radical trap for both
3 and 4. For 11 and 13, k1/k2 evidently is closer to one.
Results
Under syringe-pump-addition conditions in the absence of
(MeO)3P (low [n-Bu3SnH]), bromoepoxide 9 (1, R1 ) PhCH2,
R2 ) H), which had not previously been studied, was converted
exclusively (GC) to allyl alcohol (21) in 76% isolated yield,
presumably via oxiranyl radical 2 (R1 ) PhCH2, R2 ) H) and
allyloxy radical 3 (R1 ) PhCH2, R2 ) H). No peak assignable
to allyl ether 22 (5, R1 ) PhCH2, R2 ) H) could be detected
by GC/MS. In a carefully monitored one-pot reaction (Method
B, but with no phosphite added, Table 1), the yield of allyl
alcohol 21 was 93%.
Syringe-pump addition of a benzene solution of AIBN and
n-Bu3SnH to a solution of bromoepoxide 9 in (MeO)3P as
solvent (ca. 8 M) at 80 °C resulted in nearly quantitative
deoxygenation of 3. The resulting allyl radical 6 on reaction
with n-Bu3SnH generated a mixture of isomeric alkenes (23)
The reactions of haloepoxides are reported in the order 8-13
which, except for 13, reflects decreasing amounts of trimethyl
phosphate formation. All reactions with trimethyl phosphite
present were run with exclusion of oxygen. Phosphate yields
recorded (Tables 1 and 2), therefore, contain at most 1-2% of
material from air oxidation of the phosphite (phosphate yield
based on starting haloepoxide; see control reactions in the
Experimental Section).
Debromodeoxygenations of Bromoepoxides 8 and 9. These
are reported together because in both cases phosphite brings
about total deoxygenation of allyloxy radical intermediate 3 and
formation of reasonably high yields of alkene. In the absence
of (MeO)3P, the conversion of bromoepoxide 8 to 1-indanone
is readily accomplished, as reported previously.13 Thus, syringe-
pump addition under argon of a benzene solution of n-Bu3SnH
(0.05 M) and AIBN (0.003 M) to a refluxed 0.03 M benzene
solution of 8 over a 5 h period, followed by a further 2 h reflux,
gave none of the thermodynamic C-C cleavage product, vinyl
ether 14 (GC). The primary product, 1-indanone (15), presum-
ably formed from the initial C-O scission product, 1-indenol
(14) Rearrangement of 1-indenol (16) to 1-indanone (15) has been pre-
viously reported under (1) weakly basic conditions: Friedrich, E. C.;
Taggart, D. B. J. Org. Chem. 1975, 40, 720. Eisch, J. J.; Galle, J. E. J.
Org. Chem. 1990, 55, 4835-4840; and (2) thermal and/or free radical
conditions.13
(15) Chatgilialoglu, C.; Ingold, K. U.; Scaiano, J. C. J. Am. Chem. Soc. 1981,
103, 7739.
9
3250 J. AM. CHEM. SOC. VOL. 125, NO. 11, 2003

Trimethyl Phosphite as a Trap for Alkoxy Radicals
A R T I C L E S
in high yield. In a one-pot version of the reaction (Method B,
Table 1), a 95% total GC yield of three alkenes resulted (23a:
23b:23c ) 57:22:21). Alcohol 21 was present, but in less than
1% yield. Use of the sterically demanding H-transfer agent
(TMS)3SiH afforded trans-23 (23a) as the only alkene (92%
GC yield) by regioselective reaction at the methylene terminus
of the intermediate 1-methyl-4-phenylallyl radical (6, R1 )
PhCH2, R2 ) H).
pot reactions at a high phosphite concentration (ca. 8 M) were
unsuccessful. Dimer 28 was reduced in yield to 16%, but the
yield of 27 did not exceed 13% (see Experimental Section).
Dehalodeoxygenations of Haloepoxides 11, 12, and 13. As
noted above, it had been proposed earlier5 that the reported12
exclusive formation from certain other haloepoxides (1, R1 )
Ph, R2 ) Me; 11, R1 ) 1-naphthyl, R2 ) H; and 12, R1 ) R2
) Ph) of the C-C scission product, that is, vinyl ether 5, may
be the result of working under close to thermodynamic
conditions. Ziegler and Petersen further suggested5 that under
kinetic conditions, products from C-O cleavage might be
encountered. The aboVe results indicate that (MeO)3P should
be an excellent trap of intermediate 3 (Scheme 1) with which
to test this postulate.
Under syringe-pump conditions in the absence of phosphite,
slow addition of a 5 mL benzene solution of n-Bu3SnH (0.04
M) containing AIBN (6 mg) to 15 mL of a 7 mM solution of
chloroepoxide 12 in benzene at reflux afforded vinyl ether 30
in 80% isolated yield as a mixture of two isomers in ratio 1.2/
1.0. Alcohol 31 was not detected by GC analysis, a finding in
agreement with an earlier report.12 A one-pot reaction (Table
1, Method B, no phosphite) gave a 93% yield (GC) of vinyl
ether 30 (cis/trans ) 1.3). In contrast, when 5 mL of a 0.2 M
benzene solution of n-Bu3SnH containing 6 mg of AIBN was
added by syringe pump to 15 mL of a refluxed solution of 12
(7 mM) and excess (MeO)3P (2.6 M) in benzene, the isolated
yield of vinyl ether 30 was reduced to 48%. In a GC-quantitated
one-pot reaction with 5 M phosphite present (Table 1, Method
B), GC yields for vinyl ether 30 were 23% cis and 29% trans.
The deoxygenation product, alkene 32, was formed in 34% GC
yield (two isomers), along with a 49% yield of trimethyl
phosphate.
In analogous fashion, in a one-pot reaction on a one millimole
scale at the concentrations used for Method B (no phosphite),
the previous report12 that bromoepoxide 11 yields exclusively
vinyl ether 33 (which we isolated in 73% yield) was confirmed.
At low n-Bu3SnH concentration (0.005 M) in a smaller-scale
one-pot reaction (Method B except for tin hydride concentration,
no phosphite, Table 1), the GC yield of 33 was 92%. No GC/
MS peak assignable to allyl alcohol 34 was seen.
For the reaction of bromoepoxide 8, under conditions similar
to those described for 9, replacement of n-Bu3SnH by (TMS)3SiH
(0.6 M) did not give an increased yield of indene (17, 46%).
The yield of 1-indenyl radical dimer (8), however, was
somewhat increased (47%).
Debromodeoxygenation of Bromoepoxide 10. In contrast
to the exclusive formation of C-O scission products, 15 and
21, from 8 and 9 in the absence of MeO3P, bromoepoxide 10
(1, R1 ) Ph, R2 ) H) was transformed under analogous syringe-
pump-addition conditions almost exclusively to allyl ether 24
(5, R1 ) Ph, R2 ) H) in 85% isolated yield. In a parallel reaction
(syringe pump) on a smaller scale, the yield of 24, the
thermodynamic C-C scission product, was 89% (GC). A minor
amount of allyl alcohol 25 (4%) also was found. Nearly identical
numbers were obtained in a one-pot variation of the reaction
(Method B without phosphite, Table 1). The predominance of
the C-C scission pathway in the reaction of 10 under similar,
near-thermodynamic, thermal/AIBN conditions was reported
previously in a very thorough study of this system.5
Added (MeO)3P (5 M) very rapidly traps intermediate 3 (R1
) Ph, R2 ) H), formed by C-O scission of the intermediate
oxiranylcarbinyl radical 2 from 10. (This scission has been
clearly shown5 to be the kinetically favored process for 2 with
R1 ) Ph, R2 ) H; k1/k2 ) 4.2.5) Thus, a one-pot reaction of an
argon-saturated benzene solution of 10 (0.015 M), n-Bu3SnH
(0.03 M), a large excess of (MeO)3P (5 M), and a small amount
of AIBN, heated at 80 °C until all bromoepoxide 10 was
consumed (150 min), gave a much-reduced amount of vinyl
ether 24 (18% vs 89% without phosphite, GC). Products that
result from the n-Bu3SnH reduction or dimerization of allyl
radical 26 (6, R1 ) Ph, R2 ) H) include (GC) alkene 27 (15%),
and radical dimers 28 (41%) and 29 (14%). The trimethyl
phosphate yield from a separate one-pot reaction (Method B,
Table 1) was 82%. (See also Table 1 for yields of 24, and 27-
29.)
In the presence of 5 M trimethyl phosphite in a one-pot
reaction (Method B, Table 1), the yield of vinyl ether (33) from
bromoepoxide 11 was reduced to 43% (GC, Table 1). In
addition, the three alkenes 35, 36, and 37 were seen but in only
3, 6, and 7% GC yields, respectively. Presumably, the low yields
of alkenes are the result of the formation from the 1-naphthyl-
propenyl radical (6, R1 ) 1-naphthyl, R2 ) H) of large amounts
of dimer, as is seen for the allyl radicals (6) from the parallel
reaction of 10 (28 and 29 formed). Such dimers from bromo-
epoxide 11 are no doubt too high boiling to be detected by gas
chromatography.
Carbon-carbon scission (eq 2) of the oxiranylcarbinyl radical
(38) from bromoepoxide 13 (a bicyclic analogue of 10) gives a
methylene-bridged cyclic benzylic species (39) that is otherwise
analogous to that from 10 (4, R1 ) Ph, R2 ) H). Column
chromatography of a syringe-pump-addition reaction afforded
Attempts to maximize the yield of alkene 27 by increasing
the concentrations of n-Bu3SnH (0.02, 0.04, 0.08 M) in one-
9
J. AM. CHEM. SOC. VOL. 125, NO. 11, 2003 3251

A R T I C L E S
Ding and Bentrude
haloepoxides 8-13. Also recorded are the yields of alkene 6
generated in the presence of 5 M phosphite. Table 2 lists yields
of vinyl ether and phosphate from parallel reactions of halo-
epoxides 8, 10, and 11-13 under syringe-pump conditions with
high concentrations of phosphite present (Experimental Section,
Method A). In each table, all data are obtained at the same
concentrations of reactants, except as noted.
2-methyl-1-indanone (40) and vinyl ether 42 in 8 and 48%
isolated yields, respectively. We determined in a quantitative
one-pot reaction (no phosphite, Method B, but at low 0.005 M
Bu3SnH, Table 1) that indeed major amounts of two carbon-
carbon scission products, vinyl ethers 41 (GC/MS evidence;
46%, GC) and 42 (38%, GC), were formed (Table 1). A lesser
amount of C-O scission product, 2-methyl-1-indanone, 40
(12%, GC), that was not reported in the previous study,13 also
was seen. Vinyl ether 41 also was not found in the earlier
research.13 It seems likely that allyl ether 41 is formed initially
and then is isomerized to 42 during isolation or otherwise
destroyed, as 41 was not found on column chromatography.
Deoxygenation of allyloxymethyl radicals (3, Scheme 1) by
trimethyl phosphite generates trimethyl phosphate and allyl
radicals (6) that dimerize or are trapped by n-Bu3SnH to give
alkene. The yield of trimethyl phosphate formed should cor-
respond to the fraction of 3 that is deoxygenated and be equal
to the yield of alkene and other products of the allyl radical 6.
The yield of vinyl ether (5), in the absence of phosphite,
measures the degree to which C-C scission occurs and the
success of trapping of vinyloxy radical 4 by n-Bu3SnH in
competition with reformation of oxiranylcarbinyl radical 2. In
theory, the yield of phosphate should equal the combined
reduction in yields of vinyl ether 5 and allyl alcohol 7. The
total yield of phosphate plus 5 should be 100% if all 3 is trapped
by high concentrations of phosphite such that no 7 is formed.
For Table 2, the sum of phosphate plus vinyl ether 5 ranges
from 92 to 96%, except for the reaction of bromoepoxide 13.
Using the data in Table 1, we find that this quantity runs from
89 to 101%, except again for 13 which gives a sum of phosphate
plus 5 (41, 42) of 80%. As noted above, the yield of vinyl ether
41 from 13 was seen to be reduced at higher tin hydride
concentrations, likely because of the susceptibility of the
exocyclic double bond to radical addition reactions. The syringe-
pump method keeps the tin hydride concentration very low and
preserves a higher yield of vinyl ethers 41 and 42 (23% in Table
2), an effect noted in the earlier work5 on the influence of
n-Bu3SnH concentration on the yields of products 24 and 25
from bromoepoxide 10.
It had been suggested13 that there is a 1,5-hydrogen shift in
the benzylic allyloxycarbinyl radical precursor to vinyl ether
41 to form the isomeric allyl radical that yields vinyl ether 42.
Postulation of this unusual process appears to be unnecessary
and unlikely in view of our finding that 41 is in fact formed.
(See also Discussion.)
Excess (MeO)3P largely deoxygenates the allyloxy radicals,
43, from 13 (eq 3). Alkenes 45 and 46 are the predominant
The phosphate and alkene yields in Table 1 are close to being
equal for the reactions of bromoepoxides 9 and 13. Moreover,
if the yield of dimers is included for 10, the yield of allyl radical
6 products rises to 75%, as compared to an 82% phosphate yield.
For bromoepoxides 10 and 11, as mentioned earler, extensive
allyl radical dimerization is expected, as is seen for 10.
Presumably, dimers abound for the allyl radicals from chloro-
epoxide 12 as well. As also noted above, in addition to indene
(17) and 1-indenyl radical dimers 28 and 29, 8 yields an indene/
n-Bu3SnH adduct (20) at higher concentrations of n-Bu3SnH.
Clearly, trimethyl phosphate yields are a much more reliable
measure of the amount of deoxygenation of oxiranylcarbinyl
radical 3 than are the yields of alkene.
products from subsequent n-Bu3SnH reduction of 44. Thus, in
a one-pot reaction (Method B, 0.02 M n-Bu3SnH, Table 1), 45
and 46 were formed in 69% total GC yield (45/46 ) 1.5), along
with about 1% of what appears to be 47 (GC/MS) generated
by dimerization of the 2-methyl-1-indenyl radical (44, eq 3).
Only 4% of 2-methyl-1-indanone (40) was formed. The
combined yield of vinyl ethers 41 and 42 was reduced to 10%
(83% overall product accountability). Presumably, terminal
alkenes 41 and 46 would be found in somewhat higher amounts
except for the relatively high concentrations of n-Bu3SnH (0.02
M vs 0.005 M without phosphite) with which they most
probably form adducts. Several unidentified side product peaks
were seen by GC/MS.
Effects of Variations in n-Bu3SnH and Phosphite Con-
centrations on Product Distributions. The equilibria of
Scheme 1, and the quantities of products formed thereby, may
be strongly influenced by changes in concentrations of n-Bu3SnH
or (MeO)3P leading to variations in the yields of (MeO)3PO,
vinyl ether 5, and alcohol 7. Differences in the effect of added
phosphite from one haloepoxide to another (10, 11, and 13)
may give insightful inferences concerning changes in the various
rate constants of Scheme 1 as a function of structure.
Comparisons of Product Yields from Haloepoxides 8-13.
In Table 1, the yields of alcohol 3 and vinyl ether 5 from one-
pot reactions, with and without trimethyl phosphite (Experi-
mental Section, Method B), are compared for the series of
(a) Bromoepoxide 10. In Table 3 are listed results that show
the decrease in yield of vinyl ether 24 and parallel increase in
phosphate yield with increased trimethyl phosphite concentra-
tion. The 0.04 M solution of n-Bu3SnH was added under
9
3252 J. AM. CHEM. SOC. VOL. 125, NO. 11, 2003

Trimethyl Phosphite as a Trap for Alkoxy Radicals
A R T I C L E S
Table 3. Effect of Trimethyl Phosphite Concentration on
Formation of Vinyl Ether 24 and Trimethyl Phosphate on
Debromodeoxygenation of 10 at 0.04 M n-Bu3SnH Concentration
(Syringe-Pump Addition)a
by reaction with relatively high concentrations of tin hydride
(see above, Table 8, Supporting Information). Thus, the
combined yields of 41 and 42 are maximized at 84% in Table
1 in a reaction without phosphite that was run at low tin hydride
(0.005 M). The yields of phosphate and alkene (Tables 8, 9
(Supporting Information)) are close to equal, as in theory they
should be. The phosphate yields and combined yields of 45 and
46 from bromoepoxide 13, at a relatively high tin hydride
concentration (0.6 M), are nearly constant (47-50%) over the
range 4.5-6.7 M phosphite. The same effect on phosphate and
vinyl ether 24 yields was seen for 10 (Table 4, Supporting
Information).
molar concentration
of (MeO)3P
yield of
(MeO)3PO
yield of 24
0
89
52
34
24
22
14
12
12
10
8
0
17
50
57
66
72
76
81
82
82
0.1
0.5
1.1
2.0
4.2
5.1
5.6
6.9
8.5
Discussion
a Method C. 0.04 M n-Bu3SnH in benzene added to 0.02 M 10 in refluxed
benzene containing trimethyl phosphite; (n-PrO)3PO internal standard.
Except for 9, the n-Bu3SnH/AIBN-initiated reactions of the
haloepoxides (8-13) of this study have been investigated
previously. However, only for 10 have quantitative investiga-
tions of products been carried out.5 Also, as noted above, product
indanone 40 from C-O scission of oxiranylcarbinyl radical 38
was not found in the earlier study of the AIBN/n-Bu3SnH-
initiated reaction of 1313 nor was allyl ether 41. Furthermore, a
systematic correlation of structure and product distribution
(Tables 1 and 2) for the series 8 and 10-13 has not been
attempted. The use of trimethyl phosphite as a trap for allyloxy
radicals 3 is novel. Phosphite is shown to react rapidly enough
with 3 from 11 and 12 to divert the equilibrium of Scheme 1 in
favor of products of C-O scission as suggested previously5
under conditions more nearly appoaching kinetic control of
product formation. The Results outlined above for haloepoxides
8-13, including those for the new haloepoxide 9, will be
discussed in terms of the relation of product distribution to the
energetics and rate constants of the reactions of Scheme 1.
Zeigler and Petersen5 thoroughly examined the thermal
n-Bu3SnH/AIBN-induced reaction of 10 (R1 ) Ph, R2 ) H) at
various concentrations of tin hydride with n-Bu3SnH as sole
trap of both radicals 3 and 4. Their careful kinetic analysis, based
on Scheme 1 (without phosphite), showed that the ratio of allyl
alcohol (7) to vinyl ether (5) under conditions of thermodynamic
control, that is, at very low tin hydride concentration, should
correspond to the equation:
syringe-pump-addition conditions to keep its concentration very
low. The sum of the phosphate and vinyl ether 24 yields at 2.0
M phosphite concentrations and above is 86-92%. The
continued increase in yield of 24 as phosphite concentration
increases shows that the equilibrium of Scheme 1 continues to
be drawn toward 3.
In contrast, in one-pot reactions at a higher constant concen-
tration of n-Bu3SnH (0.6 M), both phosphate (81%) and 24
(14%) yields become constant at 4.4 M phosphite concentrations
and above (see Table 4, Supporting Information). Furthermore,
at constant 5 M phosphite, no effect is seen of variation in
n-Bu3SnH concentration in the range from 0.02 to 1.2 M (Table
5, Supporting Information; phosphate 81-82%, 24 14-15%;
sum 95-97%).
(b) Bromoepoxide 11. The phosphate and vinyl ether (33)
yields from reaction of 11 at constant, 0.02 M n-Bu3SnH
concentration in one-pot reactions with increasing phosphite
concentrations also change dramatically (Table 6, Supporting
Information). Phosphate yield gradually increases up to about
6 M phosphite (34% at 1.2 M, 55% at 6.2 and 7.8 M) along
with a concomitant decrease from 84% in yield of 33 (56% at
1.2 M; 37% at 6.2 and 7.8 M). The yield of phosphate plus
vinyl ether 33 is 90-93%.
Furthermore, with 11, phosphate yields at high phosphite
concentrations (7 M), over a 10-fold range (0.06-0.64 M) of
concentration of n-Bu3SnH, remain constant (45-47%; Table
7, Supporting Information). The yield of vinyl ether 33, however,
decreases from 44 to 28%, most likely because of 1,2-addition
of the hydride to 33, and the combined yield of phosphate plus
33 decreases progressively from ca. 90% at 0.06 M n-Bu3SnH
to 73% at 0.64 M tin hydride. Phosphite does not draw the
equilibrium of Scheme 1 from 11 as far toward phosphate (45-
47%) as it does the radical system from 10 (81-82% phosphate,
Table 5 of Supporting Information).
7/5 ) (k1/k-1)(k-2/k2)(k3/k4)
(4)
Expression (4) arises from simplification of the more general
expression5 (eq 5) under the assumptions [n-Bu3SnH]k3 , k-1
and [n-Bu3SnH]k4 , k-2. Clearly, the ratio of products is
7/5 ) (k1k3(k-2 + k4[n-Bu3SnH]))/
(k2k4(k-1 + k3[n-Bu3SnH])) (5)
affected by all six rate constants of Scheme 1. If, on the other
hand, the concentration of n-Bu3SnH is such that [n-Bu3SnH]k3
. k-1 and [n-Bu3SnH]k4 . k-2, the product ratio will be
kinetically controlled, and eq 5 is simplified to eq 6.5 However,
as indicated below, purely kinetic or thermodynamic control
(c) Bromoepoxide 13. The yields from bromoepoxide 13 of
both phosphate (from 34 to 50%) and alkenes 45 and 46 (from
32 to 50%) increase on an increase in phosphite concentration
(1.3-6.7 M) at a relatively high n-Bu3SnH concentration (0.06
M, Table 8 of Supporting Information; same conditions as for
10, Table 4). A change in tin hydride concentration (0.02-0.60
M) at relatively high, 6.7 M phosphite decreases phosphate yield
(70 to 50%), as well as the sum of 45 and 46 (69 to 50%); see
Supporting Information, Table 9. Alkenes 45 and 46 were
assayed, because vinyl ethers 41 and 42 are evidently consumed
7/5 ) k1/k2
(6)
may not be operative for haloepoxides 8-13.
With n-Bu3SnH as H-transfer agent,5 the ratio k3/k4 (Scheme
1) will be large in all cases, thus favoring formation of allyl
alcohol 7. This conclusion arises from the consideration of rate
9
J. AM. CHEM. SOC. VOL. 125, NO. 11, 2003 3253

A R T I C L E S
Ding and Bentrude
constants for closely related reactions: k3 (tert-butoxy radical,
2.2 × 108 M-1 s-1 at 25 °C; 4 × 108 M-1 s-1 at 80 °C7); and
k4 (at 20 °C, ethyl radical,15 2.0 × 106 M-1 s-1; benzyl radical,
3.0 × 105 M-1 s-1). In these systems, one moves toward nearly
kinetic conditions by working at higher concentrations of
n-Bu3SnH (eq 4) and/or (MeO)3P. However, strictly kinetic
control (eq 6) was not observed in the kinetic study5 of
bromoepoxide 10, even when the H-transfer agent n-Bu3SnH
was used as solvent.
with R1 ) R2 ) H, ab initio calculations11 give at 25 °C: k1 )
5.2 × 109 s-1 (Ea ) 4.4 kcal/mol); k-1 ) 3.9 × 108 s-1 (Ea )
5.5 kcal/mol); ∆°2f3 ) -1.1 kcal/mol; k2 ) 1.5 × 104 s-1 (Ea
) 12.1 kcal/mol); and, for k-2 (Ea ) 15.9 kcal/mol); ∆°2f4
)
-3.8 kcal/mol. At 70 °C, an experimental estimate5 of k-1(3
f 2) of 2 × 109 s-1 has been made for the case R1 ) Ph, R2
) H (radicals from 10), which also is consistent with the
reversibility of C-O scission (2 a 3). A phenyl R1 substituent
should not significantly perturb the value of k1, which, therefore,
should be applicable generally to cases, including 9, where R1
) H, alkyl, aryl and R2 ) H. For such cases, the C-O scission
product, alcohol 21, should be formed exclusively because
k1 . k2.
It is arguable that for bromoepoxide 9, C-C cleavage of
oxiranylcarbinyl radical 2 occurs essentially irreversibly, because
vinyloxy carbinyl radical 4, once formed, is trapped by 0.02 M
n-Bu3SnH. In contrast, trapping by 0.02 M n-Bu3SnH likely
will not compete readily with reclosure of allyloxy radical 3 to
2. Thus, theoretical calculations11 for oxiranylcarbinyl radical
2 with R1 ) R2 ) H give k2 ) 1.5 × 104 s-1 (Ea ) 12.1 kcal/
mol). A value for k-2 (Ea ) 15.9 kcal/mol11) at 25 °C of 30 s-1
can be estimated from the difference (3.9 kcal/mol) in calcu-
lated11 Ea values for k2 and k-2, with the assumption that the
preexponential A values are nearly the same for both processes.
This means that at a concentration of n-Bu3SnH of 0.02 M
(Table 1) k4[n-Bu3SnH] . k-2, that is, (2 × 106 M-1 s-1 × (2
× 10-2 M) ) 4 × 104 s-1) . 30 s-1, where kH ) 2 × 106 M-1
s-1 is the rate constant for reaction of ethyl radical with
n-Bu3SnH.15 Furthermore, comparison of calculated and meas-
ured values of k-1 (calculated11 for R1 ) R2 ) H, k-1 ) 4 ×
108 M-1 s-1; experimentally estimated5 for R1 ) Ph, R2 ) H,
k-1 ) 2 × 109 M-1 s-1 at 70 °C) and k3 (calculated,11 2 × 108
M-1 s-1) shows that k-1 . k3[n-Bu3SnH]. Equation 5, therefore,
is simplified at 0.02 M n-Bu3SnH to eq 7:
The success of trimethyl phosphite as a selective trap for
allyloxy radicals (3, Scheme 1) is based on the fact that alkyl
radicals, such as 4, at best add reversibly to trimethyl phosphite
to give phosphoranyl radical intermediates ([4-P(OMe)3]•) but
fail to give product, because the subsequent â-scission of the
phosphoranyl radical to yield product 4-P(O)(OMe)2 and methyl
radical is too slow.3 In contrast, alkoxy radicals, such as 3, react
by rapid, irreversible3 addition to trimethyl phosphite (for tert-
butoxy radical with (EtO)3P, kP ) 1.7 × 109 M-1 s-1 at room
temperature8). Deoxygenation of 3 occurs cleanly by subsequent
irreversible â-scission of the phosphoranyl radical [3-P(OMe)3]•
to give the relatively stable allyl radical 6 and highly stable
trimethyl phosphate, readily assayed by GC. This reaction is
severalfold faster than abstraction of hydrogen from n-Bu3SnH
(for tert-ButO•, kH ) 2.0 × 108 M-1 s-1 at 22 °C7). Furthermore,
trimethyl phosphite is a relatively volatile liquid that even can
be used neat as solvent (ca. 8 M). Thus, neat phosphate has a
30- to 1000-fold kinetic advantage in the interception of allyloxy
radical 3 over tin hydride when the latter is present in the range
of concentrations, 0.02-0.6 M, used in this research. Phosphate
yields in these studies, therefore, account for essentially all of
the trapping of 3 and need not be adjusted for alcohol formation.
Control reactions showed at most 1-2% of phosphate to be
formed in the absence of haloepoxide (see Results).
The results for 8-13 will be discussed in the order that allows
them to be best correlated with the mechanism set forth in
Scheme 1. Special emphasis is given to the first-time use of
trimethyl phosphate as a trap of the allyloxy radical 3 (Scheme
1) and to the information gained thereby concerning the
competition between the C-C and C-O scission for the series
of oxiranylcarbinyl radicals (2) studied.
7/5 ) k1k3[n-Bu3SnH]/k2k-1
(7)
Use of the calculated11 rate constants listed above for R1 ) R2
) H in eq 7 gives a predicted 7/5 ratio of >103, consistent
with the exclusive formation of allyl alcohol 21 from bromo-
epoxide 9.
It is intuitively obvious from the above considerations
(Scheme 1; k1 . k2, k1 > k-1, k2 . k-2) that trimethyl phosphite,
especially at 1 M and above (kP[phosphite] > k-1), should be
an excellent trap for the allyoxy radicals 3 from bromoepoxide
9. It is not surprising, therefore, that trimethyl phosphate is
formed in high yields (Table 1, 5 M phosphite) to the near-
exclusion of alcohol 21 and elimination of allyl ether 22. At
Bromoepoxide 9. Bromoepoxide 9 has not been studied
previously. At modest n-Bu3SnH concentrations in the absence
of (MeO)3P (Tables 1 and 2), bromoepoxide 9 (along with 8)
gives only the product of C-O scission of oxiranylcarbinyl
radical 2, that is, allyl alcohol 21, in >90% yield. This outcome
is predicted, on the basis of previous experimental studies5,9
and calculations10,11 for oxiranylcarbinyl radicals (2, R1 ) H
or alkyl; R2 ) H or alkyl) that undergo potential scission to
intermediates 3 and 4 when radical 4 is not resonance stabilized
by a vinyl, naphthyl, or phenyl substituent. Calculated and
experimentally estimated values for k1 near room temperature
are in the range 109-1010 s-1. Thus, an experimental value for
k1 of 2 × 1010 s-1 at 25-30 °C has been measured for a case
in which the carbinyl center of oxiranylcarbinyl radical 2 is
tertiary.4 It has been pointed out11,16 that this value in fact may
be low because k1 was determined under the assumption that
step 2 f 3 (C-O scission) is irreversible, which has been shown
generally not to be true.5,16-23 For oxiranylcarbinyl radical 2
5-8 M phosphite concentrations, kP[(MeO)3P] is 109-1010 s-1
,
and deoxygenation should compete readily with reclosure of
vinyloxycarbinyl radical 3 to allyloxycarbinyl radical 2 under
conditions that begin to approach kinetic control. (As noted
earlier, at room temperature, the rate of reaction5,6,8 of tert-
BuO• with (EtO)3P is kP[(EtO)3P] ) 5 M × 1.7 × 109 M-1 s-1
) 8 × 109 s-1, whereas the calculated11 k-1 ) 4 × 108 s-1 for
R1 ) R2 ) H.)
(18) Nussbaum, A. L.; Wayne, R.; Yuan, E.; Zagneetko, O.; Oliveto, E. P. J.
Am. Chem. Soc. 1962, 84, 1070.
(19) Rawal, V. H.; Iwasa, S. Tetrahedron Lett. 1992, 33, 4687.
(20) Galatsis, P.; Millian, S. D. Tetrahedron Lett. 1991, 32, 7493.
(21) Galatsis, P.; Millan, S. D.; Faber, T. J. Org. Chem. 1993, 58, 8, 1215.
(22) Suginome, H.; Wang, J. B. J. Chem. Soc., Chem. Commun. 1990, 1629.
(23) Weinberg, J. S.; Miller, A. J. Org. Chem. 1979, 44, 4722.
(16) Ziegler, F. E.; Petersen, A. K. J. Org. Chem. 1995, 60, 2666.
(17) Amaudrut, J.; Wiest, O. Org. Lett. 2000, 2, 1251.
9
3254 J. AM. CHEM. SOC. VOL. 125, NO. 11, 2003

Trimethyl Phosphite as a Trap for Alkoxy Radicals
A R T I C L E S
n-Bu3SnH at 0.02 M and greater should trap essentially all
oxiranylcarbinyl radicals 2 that undergo C-C scission (see
above). It is evident from the lack of formation of vinyl ether
22, both with and without phosphite present, that C-C scission
does not compete kinetically with C-O cleavage. This is as
predicted by the calculated11 ratio for k1/k2 (>105) for 2 with
R1 ) R2 ) H.
M-1 s-1 at 22 °C; 4 × 108 M-1 s-1 at 80 °C; calculated11 for
2 with R1 ) R2 ) H k-1, 4 × 108 s-1 M-1 s-1; experimental5
for 2 with R1 ) Ph, R2 ) H k-1, 2 × 109 s-1 at 70 °C). The
value for 2 from 10 of k1/k2 4.2 favors C-O cleavage. However,
at the relatively low concentration of n-Bu3SnH (0.02 M) of
Table 1, conditions closer to thermodynamic control should
prevail. The dominant amounts of allyl ether encountered in
Table 1 and in previously reported research5 are, therefore,
predictable. Indeed, at closer to kinetic conditions (3.72 M
Bu3SnH), a predominance of alcohol 25 (7) was reported (25/
24 ) 1.94).5
Haloepoxides 10 and 11. These compounds differ from 9
in that the ring opening of 2 (R1 ) Ph or 1-naphthyl; R2 ) H)
by C-C scission yields a radical (4) which is stabilized by Ph
or 1-naphthyl with potential kinetic and thermodynamic con-
sequences. Compared to 9, these compounds give reduced
phosphate yields (10, 82%; 11, 52%; 5 M phosphite, Table 1;
10, 85%; 11, 58%; 8 M phosphite, Table 2). This is in keeping
with the formation from 10 and 11 in the absence of phosphite
of very predominant amounts of vinyl ether (89% 24 from 10
and 92% 33 from 11, Table 1), a process with which deoxy-
genation of 3 must compete when phosphite is added (Scheme
1). In the absence of phosphite, a small amount (4%) of alcohol
25 is formed from 10 which is consistent with earlier reports;5,17
yet formation of alcohol 34 from 11 was not detected by GC/
MS in our study or in the previous work.17 Vinyl ether 33
formation from 11 in the absence of phosphite (Table 1) may
have been enhanced by the very low n-Bu3SnH concentrations
(0.005 M) employed, which allows radical 3 (R1 ) 1-naphthyl,
R2 ) H) of Scheme 1 to reclose to 2 to a greater extent than
does its analogue from bromoepoxide 10.
In comparing the results from the reaction of 10 with those
from 11, it is significant that the 1-naphthylmethyl radical (1-
•
naphthyl-CH2 ) has been calculated24 to be 14 kcal/mol more
stable (resonance energy difference) than the benzyl radical. This
energy difference can be applied to the oxiranylcarbinyl radicals,
2, formed from bromoepoxide 11 (4, R1 ) 1-naphthyl, R2 )
H) and its benzylic counterpart from 10 (4, R1 ) Ph, R2 ) H).
Consequently, for 2 from 11 (R1 ) 1-naphthyl, R2 ) H), C-C
scission should be favored by about 18 kcal/mol (3.811 + 14
kcal/mol). The 1-naphthylmethyl-type radical (4) will be formed
more rapidly as well (increased k2). Consequently, k1/k2 will be
decreased in the radical system from 11, as compared to the
4.2 value5 for 10, as will k-2/k2. The greatly reduced value of
k-2 will again make k-2 , k4[n-Bu3SnH] at 0.005 M n-Bu3SnH
(Table 1), even though k4 will doubtless be reduced from what
it is with 2 from 10. The condition k-1 . k3[n-Bu3SnH] will
also apply, as will eq 7. The ring reclosures of the allyloxy
radicals 3 from 10 and 11 (as well as from 9) to regenerate 2
likely occur with very similar rate constants (k-1), as 2 is a
primary radical in both instances. Together, these factors, and
primarily an expected increase in k2 (k1/k2 increased), account
in terms of eq 7 for the now exclusive formation (Table 1) of
vinyl ether 33 from 11.
The changes in product distributions from 10 and 11, as
compared to those from 9, can be readily understood in terms
of the thermodynamics of the reactions of 2 and the rate
constants of eq 7. For the oxiranylcarbinyl species (2) with R1
) H, R2 ) Ph, a stabilization of 5-7 kcal/mol has been
calculated to be imparted by the phenyl group (benzylic type
radical).16 A similar stabilization is to be expected on phenyl
substitution of the vinyloxymethyl radical 4 from 10 (R1 ) Ph,
R2 ) H). The stabilization in 4 (R1 ) Ph, R2 ) H) will make
the C-C scission for 2 f 4 about 10 kcal/mol exothermic (3.811
+ 6 ) 10 kcal/mol) and will necessarily result in a large
decrease in the equilibrium expressed by k-2/k2. Moreover, the
benzylic stabilization of 4 will be felt in the transition state for
The ideas concerning the rate constants of Scheme 1 set forth
in the preceding paragraphs are confirmed by the use of
phosphite as a trap for radical 3 from 10 (R1 ) Ph, R2 ) H)
and 11 (R1 ) 1-naphthyl, R2 ) H) and further illustrate its
usefulness. Thus, at high concentrations of phosphite, large
amounts of phosphate are encountered from 10 (82, 85%; Tables
1 and 2), although less than from 9, and vinyl ether 24 is still
formed (13, 8%; Tables 1 and 2). At a 5 M concentration of
C-C scission greatly increasing k2, but drastically reducing k-2
.
The latter will compensate for a reduction in k4, perhaps by a
factor of 10 (kH for reaction of benzyl radical with n-Bu3SnH15
is reduced to 3.0 × 105 M-1 s-1 at 20 °C), so that k4[n-Bu3SnH]
. k-2, as is required for eq 7 to be applicable. There should be
little difference in k1 and k-1 for the radicals from 10 and 11 as
compared to those from 9. Thus, k-1 . k3[n-Bu3SnH], which
is the second requirement for eq 7 to apply.
phosphite (Table 1), as stated above for 9, kP[phosphite] > k-1
,
which also applies to 10 and 11, because k-1 should be nearly
the same for all three. Moreover, the reduced ratio of k1/k2 for
radical 2 from 10 (R1 ) Ph, R2 ) H) decreases the fraction of
initial 2 that becomes available as 3 for deoxygenation by
phosphite.
The equilibrium of Scheme 1 for the radicals from 11, because
of the stabilizing effect of the 1-naphthyl substituent on k2 and
k-2, lies even more in the direction of 4 than that from 10 and,
therefore, is less readily diverted by phosphite toward phosphate
formation. Consequently, at high phosphite concentrations, only
moderate amounts of phosphate result from 11 (52, 58%; Tables
1, 2), which still affords a considerable yield of vinyl ether 33
(43, 34%; Tables 1 and 2).
The above makes understandable the difference in the
calculated value11 for k1/k2 (3 × 105) for R1 ) R2 ) H (and
presumably applicable to 9) and k1/k2 (4.2) determined experi-
mentally5 for 10 (R1 ) Ph, R2 ) H). Notably, radical 3 from
10 (R1 ) Ph, R2 ) H) still undergoes C-O scission several
times more rapidly that it does C-C cleavage. Use of the
experimental value5 for k1/k2 (4.2), along with the above
estimated or calculated k3 and k-1 values and n-Bu3SnH
concentration of 0.02 M, allows calculation by eq 7 of 7/5 (25/
24) percentage ratios of 7/93 and lower, depending on which
of the temperature-dependent values of k3 and k-1 is used (based
on tert-butoxy radical as a model: experimental15 k3, 2 × 108
Haloepoxide 12. The vinyloxy carbinyl radicals (4) from 10
(R1 ) Ph, R2 ) H) and 12 (R1 ) R2 ) Ph), formed on C-C
(24) Calculated greater resonance energy of the 1-naphthylmethyl radical as
compared to the benzyl radical: Herndon, W. C. J. Org. Chem. 1981, 46,
2119.
9
J. AM. CHEM. SOC. VOL. 125, NO. 11, 2003 3255

A R T I C L E S
Ding and Bentrude
scission, possess essentially identical benzylic radical sites
(Scheme 1). A key difference in these radical systems, however,
is that for 12 the C-O scission of 2 to yield 3 sacrifices the
benzylic resonance stabilization, which is then regained on
presumably very rapid reclosure to 2 (increased k-1). As noted
earlier, a recent calculation11 predicts that phenyl subsitution at
the radical center of 2 with R1 ) H, R2 ) Ph will stabilize 2 by
5-7 kcal/mol. This makes the conversion of 2 to 3 for the
radical system from 12 (R1 ) R2 ) Ph) endothermic by about
5 kcal/mol (∆H° for 2 f 3 R1 ) R2 ) H, calculated11 to be
-1 kcal/mol) and renders (k1/k-1) < 1. However, conversion
of 2 from 12 (R1 ) R2 ) Ph) to 4 should still be favored by
about 4 kcal/mol,25 the same as is calculated11 for the same
interconversion with R1 ) R2 ) H, because a benzylic radical
site is present in both 2 and 4 from 12. Therefore, (k-2/k2) < 1.
Clearly, for 12, the increase in k-1 in the now exothermic
step 3 f 2 renders k-1 . k3[n-Bu3SnH] at the concentration of
n-Bu3SnH of Table 1 (0.02 M). As for the allyoxycarbinyl
radical 4 from bromoepoxide 10, [n-Bu3SnH]k4 . k-2. There-
fore, eq 7 again applies. Under the conditions of Table 1, vinyl
ether 30 (92%), not surprisingly, is the sole product. This is
the result of (1) the very small value of k1/k-1 (eq 7) associated
with the favorable 5 kcal/mol exothermic step 3 f 2; (2) little
change in k3 from the other cases; and (3) a C-C scission step
(k2) with energetics similar to those of 2 f 4 for the radicals
from 9, but with increased k2 (resonance-stabilized transition
state) such that the ratio k1/k2 is reduced.
constraints on the C-C scission of radicals 2 from 10 or 13
(eq 2). Equation 7 would appear to apply to a discussion of
product formation with both 10 and 13. The products from 13,
formed at 0.005 M n-Bu3SnH concentrations (Table 1), show a
marginally larger proportion of C-O scission (12% vinyl
alcohol-derived 2-methylindanone (40), 84% vinyl ethers 41 and
42, 96% product accountability) as compared to those from 10
(4% alcohol 25 and 89% vinyl ether 24, 94% product account-
ability). Furthermore, the formation of 41 and 42 (C-C scission)
in the one-pot reaction of Table 1 from 13 is enhanced (eq 7)
by the very low concentration of n-Bu3SnH (0.005 M) em-
ployed. If run under the tin hydride concentration (0.02 M)
conditions used for 10, 13 likely would have given a greater
amount of C-O scission product (40). The greater advantage
of C-O scission with radical 2 from 10 is in fact seen in Table
2, where the syringe-pump addition technique keeps hydride
concentration low in both cases with the vinyl ether yield (23%)
from 13 being markedly greater than that from 10 (8%). The
yields of phosphate follow those of the vinyl ethers.
Consistent with the above, deoxygenation of radical 3 (R1 )
Ph, R2 ) H) from 10 also is measurably easier than deoxygen-
ation of 43 from 13 (eq 3), as judged by comparisons of the
yields of phosphate (Tables 1 and, especially, 2). A possible
explanation is that formation of 39 (eq 2) may be assisted
somewhat kinetically by the fixed geometries about the develop-
ing benzylic radical center in 39 in which the benzene ring is
optimally prealigned to conjugate with the developing carbon
radical center. In contrast, phenyl stabilization of the C-C
scission of 3 from 10 (R1 ) Ph, R2 ) H) requires a kinetically
unfavorable loss of entropy (rotational freedom) in the transition
state.
Because of the reduction in k1, k1/k-1, and k1/k2 for 2 from
12, as a result of the endothermicity of the formation of allyloxy
radical 3, not only is vinyl ether 30 the only product in the
presence of n-Bu3SnH, but, even at high concentrations,
trimethyl phosphite will compete ([phosphite]kP) less well for
radical 3 and its enhanced reclosure to 2 (k-1 increased). Most
likely, the radical equilibrium of Scheme 1 (3 a 2 a 4) lies in
the direction of 4 and is not readily diverted toward deoxygen-
ation of 3. These factors result in the least efficient deoxygen-
ation of 3 from 12 of the whole series of haloepoxides 8-13
(49% phosphate; 52% remaining vinyl ether 30, Table 1; 42%
phosphate, 42% 30, Table 2). This contrasts, for example, to
the very high diversion of 3 from 10 (R1 ) H, R2 ) Ph) to
phosphate (82%) accompanied by much-reduced vinyl ether 24
(13%) formation (Table 1). As noted earlier, the rate constant
(k-1) for reclosure of 3 to 2 has been measured5 at 70 °C for
the case R1 ) Ph, R2 ) H (k-1 ) 2 × 109 s-1) and calculated11
for the case R1 ) R2 ) H (k-1 ) 4 × 108 s-1). It seems likely
that k-1 for 3 from 12, as a result of the strong exothermicity
Bromoepoxide 8. The oxiranylcarbinyl radical 2 formed from
8 that opens to an allyloxy radical 19 (eq 1) that is benzylic
and is likely stabilized by 5-7 kcal/mol, based on calculations
noted earlier for the stabilization of the oxiranylcarbinyl radical
with R1 ) H, R2 ) Ph,17 just as is radical 2 (R1 ) R2 ) Ph)
from 12 (R1 ) R2 ) Ph) discussed above. As with 2 from 12,
formation of 19 should be about 5 kcal/mol unfavorable because
the benzylic stabilization is lost (eq 1). However, like the radical
(4) from C-C scission of 2 from 12 (R1 ) R ) Ph), 48 retains
benzylic stabilization. Therefore, its formation from 2 should
still be favored by about 4 kcal/mol, the energy change
calculated11 for unsubstituted 2 and 4 with R1 ) R2 ) H and,
as stated above, assumed for 2 f 4 for the radicals from 12 for
which both 2 and 4 are phenyl-substituted (R1 ) R2 ) Ph).
This radical system at 0.02 M tin hydride concentrations,
therefore, would be expected to behave like that from 12 (which
gives only enol ether 30) and yield exclusively enol ether 14.
of the process, may be on the order 1010-1011 s-1
.
Haloepoxide 13. At first examination, the radical intermedi-
ates (Scheme 1) from 13 appear to be analogous to those from
10, with 13 being merely a bicyclic version of 10. Primary
oxiranylcarbinyl radicals (2) are cleaved by C-O scission to
allyloxy radicals (3) that are structurally analogous except for
the cyclic nature of 43 from 13 (eq 3). Both oxiranylcarbinyl
radicals (2) yield similar oxygen-substituted benzylic radicals,
that is, 4 (R1 ) Ph, R2 ) H) from 10 and 39 from intermediate
38 (eq 2). Moreover, unlike C-C scission for 2 generated from
8 (see discussion of 8 to follow), there are no stereoelectronic
However, carbon-oxygen scission to give 19 (eq 1) and
ultimately 15 (91%, Table 1, by rearrrangement14 of first-formed
allyl alcohol 16), although endothermic, occurs to the exclusion
of vinyl ether 14, as reported previously.13 It has been
suggested13 that the thermodynamically favored ring-opening
C-C cleavage of the oxiranylcarbinyl species (2) of eq 1 to
benzylic radical 48 is disfavored kinetically by stereoelectronic
factors (k1 . k2, Scheme 1). Thus, ring constraints prevent the
proper alignment of the SOMO on carbon in 2 with the C-C
bond to be cleaved, a geometric requirement also noted for the
â-scission of other radicals.26
(25) The difference in energy between 2 with R2 ) H and 4 with R1 ) H is
calculated to be about 4 kcal/mol.11 This difference should remain when a
phenyl group is substituted in place of hydrogen at the radical sites of 2
(R2 ) Ph) and 4 (R1 ) Ph), as it does in the radical system from 12.
9
3256 J. AM. CHEM. SOC. VOL. 125, NO. 11, 2003

Trimethyl Phosphite as a Trap for Alkoxy Radicals
A R T I C L E S
As a result, high activation energies should be associated with
both k2 and k-2 for 2 a 48 as they are11 for 2 a 4 from
bromoepoxide 9. The latter is a C-C scission process for which
the transition state is not resonance stabilized. In the absence
of phosphite, eq 7 likely applies to the radicals from 8, as was
argued earlier for those from 9. As with 9, kinetically favored
C-O scission (k1 > k2) to give 19 (eq 1) is thermodynamically
less favored than C-C scission to give 48.
a sufficient concentration of both phosphite (4.4 M or greater)
and tin hydride (0.02 M or greater) in one-pot reactions, the
amounts of vinyl ether 24 and trimethyl phosphate reach
constant values and high accountabilities, 14 and 81%, respec-
tively (Tables 4 and 5, Supporting Information).
The ratio of (phosphate)/(vinyl ether 24), 81/14 ) 5.8, is
reasonably close to the ratio of k1/k2 (4.2) reported earlier,5 which
was based on a careful kinetic analysis of the variation in allyl
alcohol/vinyl ether (25/24) with concentrations of n-Bu3SnH.
In the range of concentrations of tin hydride employed in the
published report (0.06-3.7 M),5 the 25/24 ratio did not become
constant but was maximized at 1.94. Clearly, the conversion
from thermodynamic control of the competition (eq 4) to kinetic
control (eq 6) in which reversal of both C-C and C-O scission
However, unlike 2 a 3 for 9 (R1 ) PhCH2, R2 ) H),
reformation of 2 from 19 (eq 1), following C-O scission, is
both kinetically and thermodynamically favored (k-1 > k1).
Trapping of 19 by 0.02 M tin hydride (Table 1) to yield 15
(from 16), to the exclusion of vinyl ether 14, stems largely from
the small value of k2 (k1 . k2, eq 7) and despite the fact that
k-1 > k1 (eq 7).
no longer occurs (k3[n-Bu3SnH] . k-1; k4[n-Bu3SnH] . k-2
)
Allyloxy radical 19, formed kinetically from radical 2 from
8 (eq 1), is also efficiently trapped by phosphite (Tables 1, 89%,
and 2, 96%) such that 8, along with 9, affords trimethyl
phosphate in the highest yields observed (Tables 1 and 2). As
noted above, it is very likely that phosphite (kP[phosphite]) at
5-8 M concentrations (Tables 1 and 2) in many instances
competes successfully with k-1 for reclosure of 3 to give 2
(Scheme 1). Moreover, the exothermic reclosure of 19 (eq 1)
should be even faster than that for 3 f 2 for R1 ) Ph, R2 ) H
from bromoepoxide 10, which was found experimentally5 to
be 2 × 109 s-1 at 70 °C. The high yield of phosphate from 8 in
both Tables 1 and 2 (like the high yield of 15 in its absence)
doubtless is in part dependent on the trapping of 19 formed by
reopening of oxiranylcarbinyl radicals (eq 1).
Allyl radical reduction product indene (17) and indenyl radical
dimer, 18, account for the majority of the radical products of
deoxygenation of 19. The failure of phosphite to totally prevent
formation of allyl alcohol 15 (4% 15, Table 1) is likely the
result of the rapid reclosure of 19 to benzylic radical 2 (eq 1;
k-1/k1 > 1) and its trapping by tin hydride.
Effects of n-Bu3SnH and (MeO)3P Concentration Changes
on Product Distribution. Changes in product distributions from
haloepoxides 8, 10, and 13 on increasing phosphite and/or
n-Bu3SnH concentrations (Results and tables discussed therein)
reflect readily understandable influences of structure on the
radical equilibria of Scheme 1. In general, the ease of shift of
products in the direction of deoxygenation by increased phos-
phite concentration is determined by the same rate constant
scenario that leads to a increase in ratio 7/5, (alcohol)/(vinyl
ether), especially increased k1 and decreased k2.
(a) Bromoepoxide 10. The extremely low effective concen-
tration of hydride under syringe-pump addition conditions for
10 (Table 3) likely does not prevent reclosure of 4 and 3 to 2
(k-2 . k4[n-Bu3SnH]; k-1 . k3[n-Bu3SnH]). In the absence of
phosphite, nearly thermodynamic control conditions (eq 4) for
trapping of 3 and 4 by n-Bu3SnH apply with vinyl ether 24 as
the dominant product (89%) as found previously5 at low tin
hydride concentrations. The yield of vinyl ether 24 decreases
progressively from 89 to 8% in response to the presence of
phosphite in increasing concentrations, as the equilibrium of
Scheme 1 is drawn toward 3 and its deoxygenation. Perhaps
unexpectedly, the phosphate yield does eventually level off. At
had not been accomplished5 at 3.7 M n-Bu3SnH concentration
((25)/(24) ) 1.9 vs k1/k2 ) 4.2 for kinetic control). The
asymptotic approach to constant (phosphate)/24 seen at high
concentrations of phosphite and tin hydride (Tables 4 and 5,
Supporting Information) suggests that both 3 and 4 become
efficiently and near-irreversibly trapped, and more than co-
incidentally the ratio (phosphate)/24 of 5.8 approximates the
previous value for k1/k2 (4.2). Clearly, the data of this paper
show C-O scission (k1) to be kinetically favored over the C-C
(k2) proccess, as previously determined.5
It should be reiterated that k-1 for the conversion 2 f 3 for
bromoepoxide 10 (R1 ) Ph, R2 ) H) has been estimated
experimentally5 to be 2 × 109 s-1 at 70 °C. Furthermore, kP
experimental8 for reaction of (EtO)3P with tert-BuO• at room
temperature is 1.7 × 109 M-1 s-1, making the rate of
deoxygenation with 6-8 M phosphite at 80 °C >1010 s-1
.
Arguments for the irreversible trapping of the benzylic type
radical 4 from 10 (R1 ) Ph, R2 ) H) by n-Bu3SnH at
concentrations 0.02 M and above were made earlier in this
paper.
(b) Bromoepoxide 11. At a constant tin hydride concentration
of 0.02 M (Table 6, Supporting Information), the yield of
phosphate levels out on increase in phosphite concentration at
about 55% at ca. 6 M phosphite concentration, a comparatively
smaller number than what was seen with 10 (82%) at 0.02 M
n-Bu3SnH and above (Table 5). Furthermore, to attain a constant
phosphate yield (45%) from 11 at 7 M phosphite, on increasing
tin hydride concentration, the latter must be 0.06 M or greater
(Table 7), somewhat higher than that required for the radical
system from 10 (0.02 M, Table 5).
As noted above, the expected increased k2 and reduced k-2
values for the intermediates from 11, as compared to those from
10, account for the exclusive formation from 11 of vinyl ether
33 (92%, Table 1) at low tin hydride concentration (eqs 4 and
7) in the absence of phosphite. They also predict that, as
observed, the radical system (Scheme 1) from 11 should be less
readily perturbed away from formation of vinyl ether 33 toward
3 and its deoxygenation by increasing concentrations of phos-
phite. The data for 11 are less-well behaved than those for 10
(Tables 3-5). Nonetheless, it is clear that radical 2 (R1 )
1-naphthyl, R2 ) H) from 11 gives a considerably larger
proportion of initial C-C cleavage than does 2 (R1 ) Ph, R2 )
H) from 10, that is, k2(11) > k2(10). If trapping of 3 and 4
from 11 approaches irreversibility (kinetic control), the data
(26) Beckwith, A. L. J.; Easton, C. J.; Serlis, A. K. J. Chem. Soc., Chem.
Commun. 1980, 482. Friedrich, E. C.; Holmstead, R. L. J. Org. Chem.
1972, 37, 2550. Dauben, W. G.; Wolf, R. E.; Deviny, E. J. J. Org. Chem.
1969, 34, 2512.
9
J. AM. CHEM. SOC. VOL. 125, NO. 11, 2003 3257

A R T I C L E S
Ding and Bentrude
suggest that k1/k2 is closer to 1, rather than the 4.2 ratio found5
for the radicals from 10.
Conclusions
Quantitative product studies for the thermal AIBN/n-Bu3SnH-
induced radical reactions of 8, 9, 11, and 13 at low concentra-
tions of tin hydride are reported (Table 1) for the first time.
Consideration of the potential effects of variation in the structure
(R1 and R2) of intermediates 2-4 on the thermodynamics and
kinetics (eqs 4-7) of the reactions of Scheme 1 leads to
reasonable structure/product correlations. The formation of 40
and 41 from 13 had not previously been reported. It is proposed
that 41 is isomerized to 42 under the reaction conditions which
avoid the unusual rearrangement of radical 39 previously
proposed.
The suggestion of Ziegler and Petersen5 that C-O cleavage
products from 11 and 12 might be seen under conditions that
approach kinetic control is in fact realized for the first time in
our studies of 11 and 12 at high phosphite concentrations.
However, unlike what is seen with 10, the predominance of
the reaction of 11 or 12 (Tables 1 and 2), via 3 and its
deoxygenation, is not observed, even at highest phosphite
concentrations.
(c) Bromoepoxide 13. The effects of increasingly high
phosphite or n-Bu3SnH concentrations on the reaction of 13
parallel what is seen with 10 (Tables 8 and 9, Supporting
Information). However, the asymptotic phosphate yield from
10 was about 80% as compared to the 50% value for 13.
Following the arguments made earlier for 10 and 11, it appears
that k1/k2 for the radicals from 13 may be reasonably close to
1, similar to that for 11. This is consistent with the arguments
made above that rationalize the somewhat greater propensity
of oxiranylcarbinyl radical 38 to isomerize to 39 (eq 2) than
that of the analogous radical 2 (R1 ) Ph, R2 ) H) from 10 to
undergo C-C scission to its vinyloxycarbinyl radical 4. It
appears that being closer to kinetic control is realized in the
system from 13 (i.e., irreversible trapping of radicals from C-C
and C-O scission) at high concentrations of both phosphite
and 0.6 M tin hydride than when the n-Bu3BuH concentration
is only 0.02 M, even at high concentrations of phosphite.
Alkenes from Haloepoxides 8-13. Precursors 8, 9, and 13
give the highest yields of alkenes: 73, 95, and 69%, respectively
(Table 1), which is in line with the accompanying high yields
of trimethyl phosphate found, especially for 8 and 9. Although
the defunctionalization of a haloepoxide to form an alkene would
likely be of synthetic use only in highly specialized instances,
the yield from 9 of 23 as a single, regio-, and stereoisomer
(trans-23a), when (TMS)3SiH is employed,27 is high (92%).
Evidently, the sterically highly hindered silane transfers hydro-
gen to ally radical 6 (R1 ) PhCH2, R2 ) H) at the primary
carbon terminus in a fashion to generate only the trans 2-alkene.
As indicated by the reactions of the other haloepoxides of
this study, regioselective, high-yield formation of alkene will
occur only when allyloxy radical 3 is the kinetically favored
scission product from oxiranylcarbinyl radical 2, and the allyl
radical 6 is sufficiently reactive to be rapidly trapped by the
sterically hindered silane ((TMS)3SiH). In this regard, the
1-indenyl radical, generated in the radical system from 8, is
evidently too stable to be readily reduced to indene (17) by
(TMS)3SiH (indene yield 47%, see Results), although 0.02 M
n-Bu3SnH (Table 1) gives a reasonably good yield of indene
(73%). However, indene yields at high concentrations of tin
hydride, as noted earlier, are reduced by the formation of tin
hydride-indene adduct 20.
Trimethyl phosphite is found to be a very efficient and
selective trap for the allyloxy radicals (3) of Scheme 1 to divert
the equilibria of Scheme 1 toward the deoxygenation (kP) of 3
to give allyl radical 6 and trimethyl phosphate. Radical 6
dimerizes or is trapped by n-Bu3SnH to form alkenes. The
dehalodeoxygenations of 8, 9, and 13 afford alkenes in 73, 95,
and 69% yields, respectively (Table 1), along with 89, 97, and
70% yields of trimethyl phosphate. Yields of vinyl ethers (5)
formed from haloepoxides 10-13 also are markedly diminished
by added phosphite. The amounts of trimethyl phosphate, 5,
and 7 (Tables 1 and 2), formed at high concentrations of
phosphite, vary systematically with changes in radical structure
in a fashion consistent with predictable trends in the rate
constants and thermodynamics for the individual reactions of
Scheme 1. This is particularly true for the C-O (k1) and C-C
(k2) cleavages of intermediate 2, to form radicals 3 and 4, and
their reversal to regenerate 2. Deoxygenation of 3 is especially
efficient for the radical systems from 8 and 9 which give alcohol
7 to the exclusion of allyl ether 5 in the absence of phosphite.
Notably, the diversion and deoxygenation of the allyloxy radicals
3 from 11 and 12 by trimethyl phosphite fulfill the prediction
of Ziegler and Petersen5 that, under conditions closer to kinetic
control, 11 and 12 may afford products of C-O scission. Use
of varying amounts of trimethyl phosphite and n-Bu3SnH with
10 gave product trends (phosphate/vinyl ether) consistent with
a k1/k2 ratio of approximately 5.8, which is reasonably consistent
with the reported value of 4.25 determined with n-Bu3SnH alone
as a radical trap. For 2 from 11 and 13, k1 and k2 appear to be
nearly equal. Nonetheless, even when intermediate 4 is stabilized
by a 1-naphthyl substituent (4 from 12), C-O scission is
kinetically at least competitive with C-C scission, as revealed
by trapping of 3 by trimethyl phosphite.
Experimental Section
General. Reactions were performed under Ar or N2 atmosphere.
For air-sensitive compounds, standard Schlenk techniques were used.
Chemicals were obtained from Fluka, Aldrich, and ARCOS and used
as purchased. All solvents were dried according to literature methods.
Silica gel (Gel 60, 230-400 mesh ASTM) for column chromatography
was purchased from EM Science. GC analyses were performed on a
temperature-programed, FID-equipped Hewlett-Packard HP 5890 in-
strument with a DB-1 Ultra capillary column. NMR spectra were
recorded on Varian XL300 and XVR300 spectrometers [1H, 300 MHz;
13C, 75 MHz]. 1H and 13C chemical shifts are given in δ (ppm) relative
to internal Si(CH3)4 δ (0.00 ppm). Electron ionization (70 eV) GC/MS
data were acquired on a Hewlett-Packard 5971A Mass Selective
Detector model 5971A instrument, equipped with an HP 5890 Series
II gas chromatograph. Melting points were measured on a Thomas-
Hoover apparatus and are uncorrected.
Bromoepoxide 13 also yields a relatively large amount of
alkene (45, 46, 69%, Table 1) in the presence of phosphite at
0.02 M n-Bu3SnH concentration. (TMS)3SiH was not employed
in this system but would be expected to react regioselectively
with radical 44 to give 2-methylindene (45). However, allyl
radical 44 (eq 3) may not react sufficiently rapidly with the
silane to give a high yield of 45.
(27) kH for (Me3Si)3SiH at 24 °C is 1.1 × 108. Chatgilialoglu, C.; Rossini, S.
Bull. Soc. Chim. Fr. 1988, 298.
9
3258 J. AM. CHEM. SOC. VOL. 125, NO. 11, 2003

Trimethyl Phosphite as a Trap for Alkoxy Radicals
A R T I C L E S
Systematic Study of Dehalodeoxygenation of Oxiranylcarbinyl
Radicals by (MeO)3P, Method A. A solution of tributyltin hydride
(58 mg, 0.20 mmol), internal standard (n-PrO)3P(O) (0.1 mmol), and
AIBN (6 mg) in 5 mL of benzene (deoxygenated by purging with Ar
for 20 min) was added by a syringe pump into a solution of a
haloepoxide (0.1 mmol) in neat trimethyl phosphite (10 mL, 80 mmol)
over a 5 h period. (The phosphite had been carefully distilled just before
use to remove all phosphate, GC evidence.) Reflux was continued for
another 2 h, after which the yields of trimethyl phosphate and vinyl
ether were measured by GC. An otherwise identical control reaction
without added tin hydride produced only a 1-2% yield of trimethyl
phosphate, based on haloepoxide.
Systematic Study of Dehalodeoxygenation of Oxiranylcarbinyl
Radicals by (MeO)3P, Method B. Under argon, n-Bu3SnH (58 mg,
0.20 mmol, 0.020 M), AIBN (7 mg), trimethyl phosphite (5 M final
concentration), the haloepoxide (0.1 mmol), and internal standard (n-
C3H7O)3PO, 1 mL, 0.1 M solution) were added to a volumetric flask
and then diluted with benzene to 10 mL. (The phosphite had been
carefully distilled before use to remove all phosphate, GC evidence.)
The solution was transferred to an ampule, subjected to four freeze-
thaw vacuum degassings (high vacuum/liquid nitrogen), flame-sealed,
and then held at 80 °C for at least 5 h and sometimes overnight. The
yields of trimethyl phosphate, alkene, vinyl ether, and/or allyl alcohol
were determined by GC. A control reaction run under the same
conditions but without substrate gave a yield of trimethyl phosphate
of 1-2% based on haloepoxide.
Debromodeoxygenation of 10 at Different Phosphite Concentra-
tions, Syringe-Pump Addition, Method C. A 5 mL solution of
n-Bu3SnH (58 mg, 0.20 mmol), AIBN (6 mg), tri-n-propyl phosphate
(1 mL of 0.1 M solution, 0.1 mmol), and a known amount of trimethyl
phosphite was added by a syringe pump to a refluxed, 5 mL solution
of 10 (21 mg, 0.10 mmol) in benzene containing the same concentra-
tion of trimethyl phosphite. The yields of 24 and trimethyl phos-
phate were determined by GC analysis. A control reaction, run under
identical conditions, but without substrate 10, gave a 1-2% phosphate
yield.
Acknowledgment. The authors acknowledge generous sup-
port of this work by the National Science Foundation and the
Public Health Service (N.I.H.)
Supporting Information Available: Preparations of 9, 18, 20.
Rearrangements of haloepoxides 8-13. Deoxyhalogenations of
8-13. Product identification and quantitation for Tables 1-3,
one-pot and syringe-pump conditions. Quantitative effects of
(MeO)3P and n-Bu3SnH concentrations (Tables 4-9). Methods
A-F (PDF). This material is available free of charge via the
Internet at http://pubs.acs.org.
JA020761X
9
J. AM. CHEM. SOC. VOL. 125, NO. 11, 2003 3259