Absolute Kinetics of Amidyl Radical Reactions

Article abstract of DOI:10.1021/ja981244a

An absolute kinetic scale for amidyl radical was established by a combination of direct laser flash photolysis (LFP) and indirect competition kinetic studies. Six amidyl radicals were studied by LFP. Arrhenius parameters were determined for 6-exo cyclizat

Full text of DOI:10.1021/ja981244a

7738
J. Am. Chem. Soc. 1998, 120, 7738-7748
Absolute Kinetics of Amidyl Radical Reactions
John H. Horner,* Osama M. Musa, Anne Bouvier, and Martin Newcomb*
Contribution from the Department of Chemistry, Wayne State UniVersity, Detroit, Michigan 48202
ReceiVed April 13, 1998
Abstract: An absolute kinetic scale for amidyl radical reactions was established by a combination of direct
laser flash photolysis (LFP) and indirect competition kinetic studies. Six amidyl radicals were studied by
LFP. Arrhenius parameters were determined for 6-exo cyclizations of the N-butyl-6,6-diphenyl-5-hexenamidyl
radical (2) and the N-butyl-6-(trans-2-phenylcyclopropyl)-5-hexenamidyl radical (5) and for 1,5-hydrogen transfer
in the N-(6,6-diphenyl-5-hexenyl)acetamidyl radical (3); rate constants for these reactions at 20 °C are 3.0 ×
107 s-1 (2), 5.5 × 106 s-1 (3), and 1 × 107 s-1 (5). Kinetic limits were established by LFP for the fast
cyclizations of the N-methyl-5,5-diphenyl-4-pentenamidyl radical (1) and the N-methyl-5-(trans-2-phenylcy-
clopropyl)-4-pentenamidyl radical (4) (k > 2 × 108 s-1 at ambient temperature) and for the slow fragmentation
of the N-(2,2-diphenylethyl)acetamidyl radical (6) (k ) 2 × 104 s-1 at 48 °C). Rate constants for amidyl
radical reactions with Bu3SnH and PhSH were determined by competition kinetics; respective rate constants
are 1.3 × 109 M-1 s-1 at 20 °C and 9 × 107 M-1 s-1 at 23 °C. Cyclizations of simple amidyl radicals were
determined from competition kinetic studies by employing Bu3SnH and N-(phenylthio)amide radical precursors
using data from the literature and from studies in this work. Rate constants at 65 °C for 5-exo cyclizations of
the N-butyl-4-pentenamidyl radical and the N-(4-pentenyl)butanamidyl radical and for 6-exo cyclization of the
N-ethyl-5-hexenamidyl radical are 3 × 109 s-1, 7 × 108 s-1, and 1.0 × 107 s-1, respectively. The kinetic
values determined in this work can be employed in synthetic planning involving amidyl radicals, and the
simple amidyl radical clocks can be used for measuring rate constants of bimolecular reactions. A compilation
of the kinetics of nitrogen-centered radical cyclizations and bimolecular reactions of nitrogen radicals with
Bu3SnH and PhSH is presented.
Radical-based methods have become important components
clizations.1 Given the wide occurrence of amides and lactams
in natural products and the ease of removal of the acyl group
of amides to provide amines, one might expect that these mildly
electrophilic species3 would have enjoyed a considerable amount
of attention. This was the case years ago, and early reviews
suggested the potential importance of amidyl radicals in
synthesis. However, amidyl radical chemistry has not kept pace
with carbon-centered radical chemistry in the intervening period,
in part a result of two shortcomings. A major limitation was
that the available precursors for amidyl radicals were highly
reactive N-halo- or N-nitrosoamides that were produced under
conditions that precluded many functional groups in the substrate
and often were employed in nonchain reactions that gave high
radical concentrations. In addition, the absolute kinetics of
amidyl radical reactions were not well established.
In recent years, several new amidyl radical precursors have
been reported that are available from mild reaction conditions
and react cleanly in radical chain propagation steps. These
include members of the PTOC4 family of radical precursors5,6
(see Figure 1 for examples), N-(phenylthio)amides,7 and thio-
carbazone derivatives.8 Each of these amidyl radical precursors
will react in chain propagation steps with stannyl (and one
of the organic synthetic repertoire in recent years. Most of the
methodology involves carbon-centered radicals where the utility
of the methods is due to the facility of radical production in
chain reactions, the fast rates of radical reactions in general and
the ease of five-membered ring production in particular, and
the stability of unprotected polar functional groups to the radical
reaction conditions. Heteroatom-centered radicals have received
considerably less attention than carbon-centered radicals in part
because polar reactions of heteroatoms are so prevalent. In the
case of nitrogen-centered radicals,1 however, there exists a
feature in addition to the typical advantages of radical-based
methods that is attractive from the synthetic perspective. The
more reactive and inherently more useful families of nitrogen-
centered radicals are typically electrophilic in character and
provide an Umpolung reactivity that complements the nucleo-
philic character of nitrogen in polar reactions.
Amidyl radicals are known to be highly reactive intermediates
that display virtually exclusive reactivity at nitrogen1 despite
the fact that they apparently are π-type radicals.2 Common
reactions of amidyl radicals include remote functionalization
of unactivated positions δ to nitrogen (similar to a Hofmann-
Lo¨ffler-Freytag reaction), intermolecular additions, and cy-
(3) Caron, G.; Lessard, J. Tetrahedron 1993, 49, 8039-8058.
(4) The acronym PTOC is from pyridine-2-thioneoxycarbonyl. For a
general overview of this family of radical precursors, see: Barton, D. H.
R.; Crich, D.; Motherwell, W. B. Tetrahedron 1985, 41, 3901-3924.
(5) (a) Newcomb, M.; Esker, J. L. Tetrahedron Lett. 1991, 32, 1035-
1038. (b) Esker, J. L.; Newcomb, M. J. Org. Chem. 1993, 58, 4933-4940.
(6) Esker, J. L.; Newcomb, M. Tetrahedron Lett. 1992, 33, 5913-5916.
Esker, J. L.; Newcomb, M. J. Org. Chem. 1994, 59, 2779-2786.
(7) Esker, J. L.; Newcomb, M. Tetrahedron Lett. 1993, 34, 6877-6880.
(8) Callier-Dublanchet, A.-C.; Quiclet-Sire, B.; Zard, S. Z. Tetrahedron
Lett. 1995, 36, 8791-8794.
(1) Reviews of nitrogen-centered radicals: Neale, R. S. Synthesis 1971,
1, 1-15. Mackiewicz, P.; Furstoss, R. Tetrahedron 1978, 34, 3241-3260.
Goosen, A. S. Afr. J. Chem. 1979, 32, 37-44. Stella, L. Angew. Chem.,
Int. Ed. Eng. 1983, 22, 337-350. Esker, J.; Newcomb, M. In AdVances in
Heterocyclic Chemistry; Katritzky, A. R., Ed.; Academic Press: New York,
1993; Vol. 58, pp 1-45. Fallis, A. G.; Brinza, I. M. Tetrahedron 1997, 53,
17543-17594.
(2) Sutcliffe, R.; Griller, D.; Lessard, J.; Ingold, K. U. J. Am. Chem.
Soc. 1981, 103, 624-628.
S0002-7863(98)01244-X CCC: $15.00 © 1998 American Chemical Society
Published on Web 07/24/1998

Absolute Kinetics of Amidyl Radical Reactions
J. Am. Chem. Soc., Vol. 120, No. 31, 1998 7739
Scheme 1
Kinetic Methods. Direct kinetic studies were accomplished
with a nanosecond-resolution kinetic spectrometer employing
a Nd:YAG laser. Members of the PTOC family of radical
precursors have a long wavelength λmax centered at about 360
nm and are efficiently cleaved by the third harmonic of the Nd:
YAG laser (355 nm);10 all studies with N-acyl PTOC carbamates
and PTOC imidate esters employed 355 nm light for radical
production. The N-(phenylthio)amides do not absorb ap-
preciably at 355 nm, and these precursors were cleaved with
the fourth harmonic of the Nd:YAG laser (266 nm).
Figure 1. Radicals studied by LFP.
assumes also silyl) radicals, and the PTOC derivatives also will
propagate chain reactions involving thiyl and selenyl radicals.
In this work, we report the basis reactions for an amidyl radical
kinetic scale that can be used for synthetic planning. Laser flash
photolysis (LFP) methods were used to measure directly the
kinetics of cyclizations of amidyl radicals which served as
“radical clocks”.9 Rate constants for second-order trapping
reactions of these radical clocks by the hydrogen atom donors
Bu3SnH and PhSH were then determined by indirect kinetic
methods. Finally, the Bu3SnH trapping reaction was combined
with relative rate constants from the literature and from studies
reported herein to determine rate constants for cyclizations of
simple amidyl radicals that more closely resemble those that
one would employ in synthesis.
LFP reactions were performed with helium-sparged, dilute
solutions of the precursors (1-5 × 10-5 M) that flowed through
a cell in the kinetic spectrometer. Most of the LFP kinetic
studies involved first-order processes that were measured with
high precision because hundreds of data points were collected
in a run. Each LFP kinetic value listed in this work results
from a computer solution of averaged data collected in 5-15
experiments. The precision of the kinetic determinations was
often better than the temperature control and is indicated for
those cases where two “sets” of data are reported at a given
temperature; in these cases, each set is completely independent.
LFP studies were conducted with radicals 1-6 (Figure 1).
5-Exo cyclization of amidyl radical 1 and 6-exo cyclization of
radical 2 were expected to give diphenylalkyl radicals that are
readily detected by UV spectroscopy. Radical 3 also might have
cyclized, but the 1,5-hydrogen atom transfer reaction shown in
Figure 1 occurred instead. Radicals 4 and 5 contain a “reporter
group” 11 for use in direct LFP studies. Cyclizations of 4 and
5 initially produce the (2-phenylcyclopropyl)carbinyl radicals
shown in Figure 1. These first-formed product radicals rapidly
open to benzylic radicals that can be observed by UV spec-
troscopy; the ring opening is so fast that one only measures the
kinetics of the initial cyclization reactions.12 Radical 6 was
expected to suffer a â-fragmentation reaction that gives the UV-
detectable diphenylmethyl radical and an enamide.
Results
Syntheses. Three types of amidyl radical precursors were
used in this work, N-acyl PTOC carbamates6 (A), PTOC imidate
esters5 (I), and N-(phenylthio)amides7 (P). Each was prepared
from the corresponding amide by literature methods as shown
in Scheme 1. The preparations of the amides used in this work
are provided in the Supporting Information, and the radical
precursor preparations are given in the Experimental Section.
As a general rule, members of the PTOC family of radical
precursors are unstable with respect to reactions with nucleo-
philes and are decomposed by visible light. The N-acyl PTOC
carbamates are somewhat more stable than the PTOC imidate
esters and survive silica gel chromatography with partial
decomposition. Synthetic intermediates were characterized
spectroscopically and by HRMS. The PTOC carbamates and
PTOC imidate ester radical precursors were characterized by
NMR spectroscopy, and the more robust N-(phenylthio)amides
were characterized NMR spectroscopy and by HRMS.
Indirect kinetic studies were accomplished by standard
competition kinetics methods.13 Reactions were run with an
(10) For early LFP kinetic studies with PTOC dervatives, see: Bohne,
C.; Boch, R.; Scaiano, J. C. J. Org. Chem. 1990, 55, 5414-5418. Ha, C.;
Horner, J. H.; Newcomb, M.; Varick, T. R.; Arnold, B. R.; Lusztyk, J. J.
Org. Chem. 1993, 58, 1194-1198.
(11) Newcomb, M.; Tanaka, N.; Bouvier, A.; Tronche, C.; Horner, J.
H.; Musa, O. M.; Martinez, F. N. J. Am. Chem. Soc. 1996, 118, 8505-
8506.
In most cases, preparative-scale reactions were conducted with
the radicals studied by LFP. Products were either isolated and
characterized or shown to be identical by GC to authentic
samples prepared by alternative methods.
(12) Newcomb, M.; Manek, M. B. J. Am. Chem. Soc. 1990, 112, 9662-
9663. Newcomb, M.; Johnson, C. C.; Manek, M. B.; Varick, T. R. J. Am.
Chem. Soc. 1992, 114, 10915-10921.
(9) Griller, D.; Ingold, K. U. Acc. Chem. Res. 1980, 13, 317-323.
(13) Newcomb, M. Tetrahedron 1993, 49, 1151-1176.

7740 J. Am. Chem. Soc., Vol. 120, No. 31, 1998
Horner et al.
Scheme 2
Figure 2. Time-resolved spectra from reactions of radical 2 from
precursor 2A (A) and from precursor 2P (B). Spectra in A are at 34,
54, and 134 ns after the laser flash, with data at 24 ns subtracted to
give a baseline. Spectra in B are at 51, 71, and 101 ns after the laser
flash. Symbols for the late time spectrum in both A and B show the
wavelengths that were monitored.
excess of hydrogen atom transfer trapping agent, but true
pseudo-first-order conditions could not be maintained in tin
hydride reactions due to the high reactivity of this agent. Ratios
of rate constants were determined from product mixtures
according to eq 1 where U is unrearranged product, R is
rearranged product, kR is the first-order rate constant for
rearrangement, kT is the second-order rate constant for trapping,
and [YH]m is the average concentration of trapping agent. For
the amounts of Bu3SnH employed in this work, the use of [YH]m
in the pseudo-first-order form of eq 1 introduces an error of
<5%.13
precursor acyloxyl radical 8, the good yield of pyrrolidinone
from this reaction requires that the rate constant for cyclization
of 1 at room temperature exceeded 1 × 108 s-1
.
6-Exo radical cyclizations are typically 2 orders of magnitude
less rapid than 5-exo cyclizations of analogous radicals, and
this generalization holds for several pairs of radicals that contain
the terminal diphenylethenyl moiety present in radicals 1-3.16
Amidyl radical 2 reacted slowly enough such that kinetics could
be measured with a nanosecond-resolution unit.
Preliminary studies of radical 2 indicated that it reacted with
a rate constant in the low 107 s-1 range at ambient temperature,
which is optimal for radical kinetic studies by LFP because
radical self-reactions and radical reactions with residual oxygen
are too slow to compete. The potential problems with LFP
studies using various amidyl radical precursors are discussed
later. Because of these experimental uncertainties, we employed
three precursors to radical 2, N-acyl PTOC carbamate 2A, PTOC
imidate ester 2I, and N-(phenylthio)amide 2P, in order to provide
[U]/[R] ) (kT/kR)[YH]m
(1)
LFP Kinetic Studies. The 5-exo cyclization of radical 1,
produced from the corresponding N-acyl PTOC carbamate 1A,
was initially studied. Scheme 2 illustrates the sequence of
reactions which occur upon laser photolysis of precursor 1A.
Photochemical cleavage of the weak N-O bond gives the (2-
pyridine)thiyl radical (7) and amidylacyloxyl radical 8. De-
carboxylation of 8 gives amidyl radical 1 which can cyclize to
give the diphenylalkyl product radical that was expected to have
λmax at about 335 nm.14
The diphenylethenyl moiety accelerates the cyclization of
radical 1, and this reaction was expected to be the fastest amidyl
radical reaction studied. In fact, it was so fast that it exceeded
the dynamic resolution of our nanosecond kinetic unit. Because
the PTOC precursors have a long wavelength λmax at 360 nm
and because radical 7 does not have a strong absorbance in the
region of 300-350 nm,15 the initial photochemical cleavage
reaction should result in bleaching in the region of 330-335
nm where the diphenylalkyl radical product absorbs strongly.
However, we observed an “instant” growth in signal in the 330-
335 nm region following the laser flash and no further increase
in signal intensity with time, indicating that the cyclization
reaction of 1 was essentially complete within the 7-ns duration
of the laser pulse. The rate constant for cyclization of 1
assurance that the observed kinetics resulted from the cyclization
reaction of 2. Figure 2 shows time-resolved spectra of the
product formed upon the irradiation of precursors 2A and 2P.
The amount of radical 2 produced by 266-nm irradiation of 2P
was considerably less than that formed from 2A, but the major
feature at 335 nm is the same in both spectra. The spectrum
observed to grow in following irradiation of 2I was similar to
that observed from 2A. The absorbance with λmax at about 335
nm is characteristic of diphenylalkyl radicals,14 indicating that
radical 2 cyclizes in a 6-exo fashion.
exceeded 2 × 108 s-1
.
Confirmation that cyclization of radical 1 occurred rapidly
was obtained in a preparative scale reaction of PTOC carbamate
1A. Tin hydride trapping of amidyl radicals is quite fast (k )
1.3 × 109 M-1 s-1 at room temperature, see below), but reaction
of 1A in the presence of 0.05 M Bu3SnH gave mainly the cyclic
product N-methyl-5-(diphenylmethyl)-2-pyrrolidinone. Given
that the acyclic amide product is ultimately produced by tin
hydride trapping of both amidyl radical 1 and its immediate
An analogue of radical 2 lacking the phenyl groups has been
reported to react by an internal hydrogen atom abstraction to
give an allylic radical in competition with a 6-exo cyclization,17
and radical 3 also reacts by an intramolecular hydrogen
(16) (a) Johnson, C. C.; Horner, J. H.; Tronche, C.; Newcomb, M. J.
Am. Chem. Soc. 1995, 117, 1684-1687. (b) Newcomb, M.; Horner, J. H.;
Filipkowski, M. A.; Ha, C.; Park, S. U. J. Am. Chem. Soc. 1995, 117, 3674-
3684. (c) Horner, J. H.; Martinez, F. N.; Musa, O. M.; Newcomb, M.;
Shahin, H. E. J. Am. Chem. Soc. 1995, 117, 11124-11133. (d) Ha, C.;
Musa, O. M.; Martinez, F. N.; Newcomb, M. J. Org. Chem. 1997, 62, 2704-
2710.
(14) Chatgilialoglu, C. In Handbook of Organic Photochemistry; Scaiano,
J. C., Ed.; CRC Press: Boca Raton, FL, 1989; Vol. 2, pp 3-11.
(15) Alam, M. M.; Watanabe, A.; Ito, O. J. Org. Chem. 1995, 60, 3440-
3444.
(17) Sutcliffe, R.; Ingold, K. U. J. Am. Chem. Soc. 1982, 104, 6071-
6075.

Absolute Kinetics of Amidyl Radical Reactions
J. Am. Chem. Soc., Vol. 120, No. 31, 1998 7741
Table 1. Arrhenius Parameters for Radical Reactionsa
Scheme 3
radical
source
log A
Ea (kcal/mol)
k20b (s-1
)
2
2A
2I
9.30 ( 0.13
9.03 ( 0.35
9.35 ( 0.06
9.33 ( 0.05
9.89 ( 0.20
9.67 ( 0.59
9.86 ( 0.18
2.46 ( 0.16
2.09 ( 0.57
2.49 ( 0.07
2.48 ( 0.07
4.22 ( 0.27
3.89 ( 0.80
4.18 ( 0.26
3.45 ( 0.41
2.64 ( 0.12
2.9 × 107
2.9 × 107
3.1 × 107
3.0 × 107
5.5 × 106
5.8 × 106
5.5 × 106
10.6 × 106
8.3 × 106
2P
WAc
3
5
3A (312)d
3A (330)d
WAc
cis,trans-5Ae 9.60 ( 0.31
trans-5A
8.89 ( 0.08
a Stated errors are at 2σ. b Rate constant at 20 °C calculated from
Arrhenius function. c Weighted average. d Monitoring wavelength in
nanometers. e A ca. 1:1 mixture of diastereomers was employed.
abstraction (see below). Because we incorporated the kinetics
of reaction of amidyl radical 2 into tin hydride trapping kinetics
discussed later, it was necessary that we determine the extent
of hydrogen abstraction in this species. In preparative scale
reactions and in competition kinetic studies with Bu3SnH
employing precursor 2P, the cyclic product N-butyl-6-(diphe-
nylmethyl)piperidin-2-one was obtained. More importantly, a
preparative reaction with the N-(phenylthio)amide precursor 2P
was conducted in the presence of Bu3SnD at a concentration of
trapping agent small enough such that the ratio of acyclic to
cyclic product was ca. 2:1. If intramolecular hydrogen abstrac-
tion occurred in competition with cyclization of radical 2 and
the allylic radical thus formed reacted with the tin deuteride,
the ultimate acyclic amide product would contain deuterium.
The acyclic amide isolated from this reaction was analyzed by
1H, 2H, and 13C NMR spectroscopy, and the NMR results
indicated no detectable amount of deuterium at the allylic
position (<1%).
It is possible in principle that the putative diphenylallylic
radical formed via translocation in radical 2 was unreactive with
Bu3SnD, as apparently was the case for a related species
discussed below, and ultimately coupled with other radicals or
reacted with oxygen. In the Bu3SnH-mediated reactions of 2P
discussed below, the total yields of products were high,
comparable to those obtained in reactions of related N-
(thiophenyl)amides lacking the aryl groups. Nevertheless, we
make the conservative estimate that errors introduced in the tin
hydride kinetics due to intramolecular hydrogen abstraction in
2 are less than 10%.
Table S1 in the Supporting Information contains the variable
temperature kinetic results for radical 2 produced from the three
precursors, and Table 1 contains the Arrhenius parameters for
the reaction. Because the PTOC imidate derivatives are
especially unstable, the sample of 2I could not be purified by
chromatography, and kinetic studies with this precursor were
conducted over a limited temperature range with a crude sample.
Studies with the N-acyl PTOC carbamate were conducted over
such a broad temperature range that the Arrhenius function
should be highly reliable. The agreement in the kinetics from
the different precursors is excellent, and the Arrhenius param-
eters for 2 from each of the precursors are within 2σ of the
weighted average values.
Radical 3 mainly reacted by an intramolecular hydrogen
abstraction, a 1,5-translocation (Scheme 3). LFP studies with
precursor 3A showed fast growth of a species with a UV signal
following laser irradiation, but a broad spectrum was observed
with λmax at about 313 nm rather than at 335 nm that would
correspond to the absorbance for a diphenylalkyl radical (Figure
3A). We conclude that the spectrum is that of the diphenyl-
allylic radical 9 (Scheme 3) because a similar spectrum was
observed when the 1,1-diphenyl-3-methylallyl radical (12) was
Figure 3. (A) Time-resolved growth spectrum of product from radical
3. The traces are for 68, 108, 188, and 348 ns after the laser flash,
with the data at 28 ns subtracted to give a baseline; the 348 ns data
contains symbols showing the wavelengths monitored. (B) Spectrum
of radical 12; the radical was produced “instantly” from the PTOC
ester precursor 11 in nonsparged THF, and the residual spectrum after
45 µs was subtracted from the spectrum observed at 1.5 µs.
produced from PTOC ester precursor 11 (Figure 3B).18 In
addition, the reported spectrum of the 2-methyl-1-phenylallyl
radical has λmax at 305 nm.19 The identity of the intermediate
was also indicated by the fact that, in preparative reactions of
N-acyl PTOC carbamate 3A in the presence of Bu3SnH, we
obtained alcohol 10 (Scheme 3) but no N-acylpiperidine product.
We assume that alcohol 10 arose from reduction of a hydro-
peroxide formed by initial oxygen trapping of the persistent
radical 9 which, apparently, reacts only slowly with tin hydride.
Kinetic studies for radical 3 were obtained using N-acyl
PTOC carbamate 3A. Table S2 in the Supporting Information
contains kinetic results at two monitoring wavelengths, 312 and
330 nm, and the Arrhenius parameters are in Table 1. Signal
growth at 312 nm was a single exponential process, but the
growth at 330 nm was better fit by a double exponential solution.
The faster rate constant of the two values in the double
exponential solution of the 330 nm data is listed in Table S2,
and although the precision in the Arrhenius parameters for these
data is only fair (Table 1), the values are within the 95%
confidence interval of the data collected at 312 nm. Second-
order processes (radical-radical and radical-oxygen reactions)
cannot interfere with the kinetics of unimolecular reactions faster
than 1 × 105 s-1 because the concentrations of radicals and
(18) Emanuel, C. J.; Newcomb, M. Unpublished results.
(19) Choi, S.-Y.; Crich, D.; Horner, J. H.; Huang, X.; Martinez, F. N.;
Newcomb, M.; Wink, D. J.; Yao, Q. J. Am. Chem. Soc. 1998, 120, 211-
212.

7742 J. Am. Chem. Soc., Vol. 120, No. 31, 1998
Horner et al.
oxygen are so small in our experimental design that even
diffusion-controlled bimolecular reactions have pseudo-first-
Scheme 4
order rate constants < 1 × 104 s-1 20
.
We believe it is most
likely that the “multiple processes” at 330 nm involved (1) initial
production of allylic radical 9 in non-Boltzmann populations
of conformers, (2) conformational relaxation (allyl radical bond
rotation) with a rate constant on the order of 106 s-1, and (3)
slightly different UV spectra for the conformers of radical 9.
This type of behavior has been observed for the related 2-benzyl-
1,3-diphenylallyl radical where the spectrum produced “in-
stantly” displayed a 5-nm hypsochromic shift in the UV
spectrum and a ca. 10% increase in absorbance at λmax, with a
rate constant of 2 × 106 s-1 at ambient temperature, in what
was attributed to a conformational relaxation process.19
(see below). Both 13 and 14 were obtained as the trans isomers
as deduced by the observation that the 1H NMR spectra obtained
at 500 MHz contained vinyl proton signals only for trans-
alkenes. In addition, GC analysis of products from reactions
of 5A with PhSH did not indicate the presence of a second cyclic
isomer in detectable amounts.
The use of the “reporter groups” in amidyl radicals 4 and 5
provides a means for measuring radical kinetics by LFP with
UV detection without introducing a large kinetic acceleration
that arises with aryl substitution directly on the alkene moiety.
The reporter group in these radicals has a minor accelerating
effect that, for the case of the 5-hexenyl radical cyclization,
amounts to about a factor of 2 at ambient temperature.11
Nevertheless, the 5-exo cyclization of radical 4 was too fast to
measure. Upon irradiation of N-acyl PTOC carbamate precursor
4A at 12 or 20 °C, we observed an “instant” growth in the region
around 320 nm (λmax for benzylic radicals)14 as opposed to an
initial bleaching, and there was no further signal growth with
Amidyl radical 6 also was evaluated as a potential radical
clock. In this case, the fragmentation reaction shown in Figure
1 produces the diphenylmethyl radical. Irradiation of either
N-acyl PTOC carbamate 6A or PTOC imidate ester 6I at
ambient temperature resulted in no appreciable growth in signal
intensity at 335 nm, indicating that the fragmentation reaction
had a rate constant < 1 × 104 s-1 and was slow with respect to
bimolecular radical-radical and radical-oxygen reactions. At
48 °C, irradiation of 6A was followed by growth of a weak
time. The cyclization reaction of 4 is faster than 2 × 108 s-1
.
Radical 5 reacted in a 6-exo cyclization slowly enough to
permit kinetic studies.21 Irradiation of N-acyl PTOC carbamate
precursor 5A was followed by growth of a signal with λmax at
about 322 nm, consistent with a benzylic radical.14 Kinetic
studies of radical 5 were performed with two samples of 5A,
one that was a ca. 1:1 mixture of cis and trans isomers and one
that was diastereomerically pure trans-5A. One can assume
that LFP of the mixture of isomers of 5A produced about a 1:1
mixture of isomeric radicals 5, but the signal-to-noise ratio was
not adequate to permit a meaningful double exponential solution
of the data. Therefore, the data was solved for a single
exponential process. The kinetic values are collected in Table
S3 in the Supporting Information, and the Arrhenius parameters
are in Table 1. From the differences in the kinetic values and
Arrhenius parameters for trans-5 and the cis,trans-5 mixture,
it seems apparent that cis-5 reacted somewhat faster than trans-
5, but rate constants for cis-5 cannot be obtained because the
exact composition of the mixture of radicals cannot be deter-
mined precisely.
signal with λmax ) 330 nm and with a rate constant of kobs
)
2.3 × 104 s-1
. These results are self-consistent because
fragmentation reactions typically have large, positive entropies
of activation (log A terms of about 15).22 Accordingly, the
unimolecular fragmentation process will be increasingly com-
petitive with bimolecular reactions (log A of ca. 10) as the
temperature is increased. If the fragmentation of 6 does have
log A )15, then the 48 °C kinetic results give a calculated rate
constant for fragmentation of 6 at 20 °C of k ) 2 × 103 s-1, a
value too slow to measure with our LFP experimental design.20
Indirect Kinetic Studies. Second-order rate constants for
intermolecular reactions of amidyl radicals are necessary for
establishing a kinetic scale. One of the more common reagents
used synthetically in radical-based methods is Bu3SnH, and the
N-(phenylthio)amides react with tin hydride in radical chain
reactions.7 We have used amidyl radical 2 as a radical clock
for timing the tin hydride trapping reaction.
Preparative scale reactions confirmed that cyclizations of 4
and 5 occurred. N-acyl PTOC carbamates 4A and 5A were
Precursor 2P reacted in the presence of tin hydride in toluene
by the sequence of reactions shown in Scheme 4 to give,
ultimately, amide 15 and piperidinone 16. Relative rate
constants for trapping of the amidyl radical (kT) and cyclization
(kC) were determined by eq 1, and the results are listed in Table
2. Because tin hydride reacts rapidly with the amidyl radical,
on the order of 109 M-1 s-1, only small concentrations of the
trapping agent could be employed, and even then the amide
product 15 predominated. The data in Table 2 provides the
relative Arrhenius function in eq 2 where θ ) 2.3RT in kcal/
mol and the errors are at 2σ. Combination of the relative
Arrhenius function in eq 2 with that for cyclization of radical
2 given in Table 1 gives the absolute Arrhenius function for
reaction of tin hydride with the amidyl radical in eq 3 that can
allowed to react in the presence of dilute Bu3SnH. Pyrrolidinone
13 was obtained from 4A in 43% yield, and piperidinone 14
was obtained from 5A in 14% yield. The low yield of 14 is in
part a reflection of the very fast trapping reaction of tin hydride
(20) Musa, O. M.; Horner, J. H.; Shahin, H.; Newcomb, M. J. Am. Chem.
Soc. 1996, 118, 3862-3868.
(21) Preliminary results for radical 5 were reported in ref 11.
(22) Benson, S. W. Thermochemical Kinetics, 2nd ed.; Wiley: New York,
1976.

Absolute Kinetics of Amidyl Radical Reactions
J. Am. Chem. Soc., Vol. 120, No. 31, 1998 7743
Table 3. Results of PhSH Trapping of Radical 5a
Table 2. Results of Bu3SnH Trapping of Radical 2a
tempb (°C)
[Bu3SnH]c (M)
[U]/[R]d
(kT/kC) (M-1
)
[PhSH]b (M)
[U]/[R]c
(kT/kC) (M-1
)
50
0.097
0.081
0.098
0.082
0.210
0.239
0.261
0.193
0.167
0.248
3.27
3.12
3.40
2.80
6.89
7.88
8.20
6.13
5.08
7.67
34
38
35
34
33
33
31
32
30
31
0.09
0.16
0.24
0.32
0.53
0.70
1.23
1.87
2.24
4.06
7.8
7.7
7.8
7.0
7.7
65
76
84
92
a Reactions run in THF at 23 ( 1 °C. b Mean concentration of PhSH.
c Observed ratio of acyclic to cyclic amide products. The total yields
of products determined by GC ranged from 81% to 94%.
hydride reacts with an “electrophilic” amidyl radical more than
an order of magnitude faster than does PhSH.
a Reactions run in toluene with AIBN initiation. b (1 °C. c Mean
concentration of tin hydride. d Observed ratio of acyclic to cyclic amide
products. The total yields of products determined by GC ranged from
63% to 81%.
With the calibration of the rate constants for the reaction of
tin hydride with an amidyl radical, it is possible to use ratios of
rate constants for simple amidyl radical cyclizations in competi-
tion with tin hydride trapping to calculate rate constants for the
cyclization reactions; the Discussion contains examples where
rate constants for cyclizations were calculated from previously
reported results. There was an obvious conflict between our
data for 6-exo cyclization of radical 5 and results from ESR
studies,17 however, that required that we study the reactions of
amidyl radical 17. Whereas we observed that radical 5 reacted
virtually exclusively via a 6-exo cyclization, Ingold’s group
reported that the closely related radical 17 reacted exclusively
by an intramolecular hydrogen abstraction reaction that gives
an allylic radical product.17
be used to predict rate constants for tin hydride trapping, such
as k ) 1.3 × 109 M-1 s-1 at 20 °C.
log((kT/kC) × M) ) (0.93 ( 0.21) + (0.93 ( 0.33)/θ (2)
log((kT) × Ms) ) (10.26 ( 0.22) - (1.55 ( 0.34)/θ (3)
The rate constants calculated from eq 3 will be reasonably
accurate for temperatures close to those of the competition study,
but the Arrhenius parameters in eq 3 are probably slightly in
error due to partial diffusion control conditions. Using estimated
radii23 for tin hydride and amidyl radical 2 and the dynamic
viscosity of toluene and assuming stick boundary conditions,
the diffusional rate constant for tin hydride reaction with 2 at
20 °C is about 1.1 × 1010 M-1 s-1 or about 9 times faster than
that for tin hydride trapping. Benzene and cyclohexane are more
viscous than toluene, and the diffusional rate constants in these
solvents estimated by the same method would be about 8
(benzene) and 5 (cyclohexane) times faster than that for tin
hydride trapping at 20 °C. The diffusional rate constants are
lower limits because stick boundary conditions were assumed,
but they show that the tin hydride trapping reaction can become
partially diffusion-controlled. The apparent activation energies
for diffusional rate constants from an Arrhenius-type treatment
are in the range of 2-3 kcal/mol, and partial diffusion control
of tin hydride trapping of amidyl radicals will be increasingly
important at lower temperatures. The net result of partial
diffusion control in regard to eq 3 is that both the log A and Ea
terms in the equation are likely to be slightly greater than the
actual values for the chemical process as opposed to composite
diffusional and chemical processes.
When radical 17 was generated by reaction of N-(thiophenyl)-
amide 17P in the presence of Bu3SnH, we observed formation
of N-ethyl-6-methyl-2-piperidinone, the ultimate product from
6-exo cyclization of radical 17. The rate constant for cyclization
of radical 17 at 65 °C, determined by competition between tin
hydride trapping and cyclization, was k ) 1.0 × 107 s-1. A
more important result from the mechanistic standpoint was
obtained when precursor 17P was allowed to react in the
presence of Bu3SnD. The reaction was conducted with a
concentration of tin deuteride small enough such that the
cyclization reaction competed efficiently with trapping; the
acyclic to cyclic product ratio was 2:1. Analysis of the mixture
2
of products by H NMR spectroscopy showed that the acyclic
amide contained <2% deuterium at C4. Accordingly, the
translocation reaction must be less than 4% as fast as the
cyclization reaction, a result completely opposite that reported
by Sutcliffe and Ingold.17 A possible explanation of the
difference in results for radical 17 is given below.
Thiophenol trapping of an amidyl radical also was studied,
in this case with amidyl radical 5. A number of experimental
difficulties arise in attempting to calibrate the reaction of an
amidyl radical with a thiol trapping agent as opposed to tin
hydride. Some of these can give rise to systematic errors in
the kinetic determinations, but the kinetic value appears to be
reliable (see Discussion). N-Acyl PTOC carbamate 5A was
allowed to react in the presence of varying concentrations of
PhSH at 23 °C in THF to give the results listed in Table 3.
Analysis of the data via the linear form of eq 1 gave a slope,
equal to kT/kC, of (7.5 ( 0.7) M-1 and an intercept of (0.0 (
Discussion
The LFP-based approach to developing an amidyl radical
scale employed in this work is similar to that we have used
previously. The major difference between this approach and
that commonly employed is that we used LFP for calibrations
of unimolecular reactions, the kinetics of which can be
determined with high precision, rather than for direct studies
of bimolecular processes. The bimolecular rate constants
necessary for developing a kinetic scale were obtained by
indirect, competition kinetics using the calibrated radical clocks.
Because the indirect studies typically are conducted over wide
temperature ranges, the precisions in the rate constants and
0.2) M-1
. Combining the slope with the rate constant for
cyclization of radical 5 gives a rate constant for the PhSH
trapping reaction of k ) (9 ( 1) × 107 M-1 s-1 at 23 °C (stated
errors at 2σ). With alkyl radicals, PhSH reacts 50 times faster
than Bu3SnH at ambient temperature,13 but the “hydridic” tin
(23) Edward, J. T. J. Chem. Educ. 1970, 47, 261-270.

7744 J. Am. Chem. Soc., Vol. 120, No. 31, 1998
Horner et al.
Arrhenius parameters for bimolecular reactions determined by
the composite method we employ is as good or better than those
usually reported for direct LFP studies.
ambient temperature)27 due to the relative instability of amidyl
radicals in comparison to alkyl radicals. If the rates of
decarboxylation were too slow, however, LFP studies with these
radicals would have been complicated, and it is clear that the
rate constants for decarboxylation are greater than 2 × 108 s-1
at ambient temperature and did not affect the kinetics. The
excellent agreement in the kinetics of cyclization of radical 2
(Table S1) produced from the three different precursors indicates
that the same reaction was measured in each case. Importantly,
even though we could not measure the rate constants for
cyclizations of amidyl radicals 1 and 4, the fact that the rate
constants for cyclizations were greater than 2 × 108 s-1 requires
that the decarboxylation reactions that produced these amidyl
radicals were essentially complete within a few nanoseconds.
We estimate, therefore, that decarboxylations of radicals 18 have
rate constants in the 109-1010 s-1 range at ambient temperature.
Amidyl Radical Cyclizations. The LFP results provided
lower kinetic limits for the 5-exo cyclizations of radicals 1 and
4 of 2 × 108 s-1 and rate constants and Arrhenius functions for
the 6-exo cyclizations of radicals 2 and 5 (Table 1). By
combining the Bu3SnH trapping rate constants with extant
competition kinetic results, one can also calculate rate constants
for the 5-exo cyclizations of radicals 20 and 21. These rate
Few absolute rate constants for amidyl radical reactions were
previously available. An early LFP study of N-nitrosoamides
provided rate constants for decay of amidyl radicals and some
second-order kinetic values.24 Kinetic ESR studies by Ingold’s
group provided amidyl radical self-termination rate constants
that were at or near diffusion control and a rate constant for
hydrogen abstraction from cyclohexane by an amidyl radical.25
Sutcliffe and Ingold also reported a kinetic ESR study aimed
at the calibration of amidyl radical clocks that established some
approximate rate constants and some kinetic limits.17 The
kinetic ESR method usually requires that reactions are conducted
at low temperatures, and extrapolations of the results to ambient
temperature can result in considerable errors.26 The kinetics
determined in this work for amidyl radical cyclizations are in
fair to poor agreement with the results of the kinetic ESR studies
as discussed later.
Reactions of Amidylacyloxyl Radicals. Among the three
radical precursors employed for LFP studies in this work, the
N-acyl PTOC carbamates are quite useful due to their ease in
preparation, their relative stability, and their reactivity in chain
reactions with thiyl radicals.6 In addition, they are cleaved by
355-nm light which does not interfere with the observation of
UV absorbances from benzylic and diphenylalkyl radicals that
have λmax values in the 320- and 335-nm range, respectively.
However, the first-formed radicals from these precursors are
amidylacyloxyl radicals 18 that can react by an intramolecular
hydrogen atom transfer to give R-amide radicals 19 in addition
to the desired decarboxylations that give amidyl radicals. In
constants are valuable for an amidyl radical kinetic scale not
only because radicals 20 and 21 resemble those one would
expect to see employed in synthetic applications but also because
the absence of aryl and reporter groups in these species precludes
any special substituent-derived kinetic effects that one might
imagine for the electrophilic amidyl radicals.
The relative rate constants at 65 °C for cyclization (kC) and
tin hydride trapping (kT) of radicals 20 and 21, generated from
the corresponding N-(phenylthio)amides, were (1.7 ( 0.1) M
and (0.41 ( 0.01) M, respectively.7 These values can be
combined with the rate constant for tin hydride trapping at 65
°C determined in this work or directly converted to absolute
rate constants because the ratio of rate constants (kT/kC) for
radical 2 was determined to be 34.5 M-1 at 65 °C (Table 2).
The rate constant at 65 °C for cyclization of 20 is kC ) (3.1 (
0.2) × 109 s-1, and that for cyclization of 21 is kC ) (7.4 (
0.2) × 108 s-1. One notes that at 65 °C the 5-exo cyclization
of 20 is 300 times as fast as the 6-exo cyclization of 17,
consistent with the established pattern for related pairs of
radicals. For example, the 5-hexenyl radical cyclizes about 2
orders of magnitude faster than the 6-heptenyl radical.28
direct LFP studies, the competing radical translocation reaction
might reduce sensitivity by diverting some of the reactant, but
it has no affect on the kinetics. In indirect competition kinetic
studies, however, the translocation reaction would result in the
ultimate formation of unrearranged products and would intro-
duce an error in the apparent rate constants for amidyl radical
trapping reactions.
The translocation in radicals 18 competes with decarboxy-
lation and is a function of the substitution at the center adjacent
to the carbonyl group, with more highly alkyl-substituted
positions demonstrating increased translocation.6 In addition,
its extent is influenced by the initial conformational population
of radicals 18, as demonstrated by a dramatic change in the
amount of translocation as a function of the presence or absence
of Mg2+; this is consistent with the idea that the rotation about
the C-N bond that interconverts conformations must be slower
than decarboxylation.
The Arrhenius log A terms for the 6-exo cyclizations of
radicals 2 and 5 were typical for such reactions, and it is
reasonable to assume that the log A values for the 5-exo
cyclizations of radicals 20 and 21 and the 6-exo cyclization of
radical 17 also will be typical with log A in the range of 10.
With this assumption, one would estimate rate constants for
cyclization at 20 °C of 2 × 109 s-1 (20), 5 × 108 s-1 (21), and
The rate constants for decarboxylation of radicals 18 to give
amidyl radicals are not known, but one can establish some limits.
It is likely that the rate constants are smaller than those for
decarboxylation of alkylacyloxyl radicals (ca. 1010-1011 s-1 at
(27) Falvey, D. E.; Schuster, G. B. J. Am. Chem. Soc. 1986, 108, 7419-
7420. Hilborn, J. W.; Pincock, J. A. J. Am. Chem. Soc. 1991, 113, 2683-
2686. Bockman, T. M.; Hubig, S. M.; Kochi, J. K. J. Am. Chem. Soc. 1996,
118, 4502-4503.
(28) Beckwith, A. L. J.; Moad, G. J. Chem. Soc., Chem. Commun. 1974,
472-473.
(24) Tam, J. N. S.; Yip, R. W.; Chow, Y. L. J. Am. Chem. Soc. 1974,
96, 4543-4549.
(25) Sutcliffe, R.; Anpo, M.; Stolow, A.; Ingold, K. U. J. Am. Chem.
Soc. 1982, 104, 6064-6070.
(26) Griller, D.; Ingold, K. U. Acc. Chem. Res. 1980, 13, 193-200.

Absolute Kinetics of Amidyl Radical Reactions
J. Am. Chem. Soc., Vol. 120, No. 31, 1998 7745
4 × 106 s-1 (17).29 For comparison, radical trans-5 cyclizes
with a rate constant of 8 × 106 s-1 at 20 °C.
coupling values are clearly consistent with those expected for
a monosubstituted allyl radical.17 However, the radical precur-
sors used by Sutcliffe and Ingold were N-chloroamides provided
to them by Lessard’s group, and no characterization data was
in the report.17 Our best guess as to the origin of the conflicting
results is that the reactant claimed to be chloramide 25 actually
was chloramide 26, the precursor to radical 27 that was the
species that reacted to give an allylic radical product. This
proposal is consistent with our observation of translocation as
the major reaction for radical 3.
The rate constants for cyclizations of the simple amidyl
radicals 5, 17, 20, and 21 are in fair to poor agreement with
values for related reactions obtained by kinetic ESR studies.17
In that work, rate constants at ambient temperature were
estimated from low-temperature ESR results and assumed log
A values. The reported rate constants for cyclizations at 300 K
were k > 1 × 107 s-1 for radicals 22, k ) 5 × 104 s-1 for 23,
and k ) 1 × 106 s-1 for 24.17 The 4 orders of magnitude
Intramolecular Hydrogen Abstractions. Amidyl radical 3
reacted mainly by a 1,5-hydrogen atom transfer instead of a
6-exo cyclization. The rate constants for the translocation
reaction (e.g., 5.5 × 106 s-1 at 20 °C) are of the same order of
magnitude as that one might have expected for a 6-exo
cyclization reaction, given that the 5-exo cyclization of radical
20 was only 4 times as fast as the 5-exo cyclization of radical
21 and the rate constant for 6-exo cyclization of 2 is 3 × 107
s-1 at 20 °C. Thus, the observation of the hydrogen atom
transfer in radical 3 and our failure to isolate a piperidine product
from preparative reactions of 3 indicate that the 6-exo cyclization
reaction is disfavored. That is, some special steric effects appear
to be at play in the transition structure of the 6-exo cyclization
reaction that appreciably slow this process. It is of interest to
note that a similar phenomenon was observed in a related
carbon-radical system. Specifically, the 6-exo cyclization of
the R-amide radical 28 was found to be only 0.03 times as fast
as the cyclization of R-ester radical 29 even though 5-exo
cyclizations of secondary R-amide and R-ester radicals have
about the same rate constants.30 Although the resemblance
between 3 and 28 might be superficial, the results in total suggest
that the transition states for 6-exo cyclizations of radicals are
considerably more sensitive to steric effects than those of 5-exo
cyclizations.
difference in rate constants for cyclization of 21 determined in
this work and for 23 is especially glaring and cannot be
attributed to an error due to extrapolation. The ESR-based rate
constant for 23 was actually an estimate resulting from the fact
that no spectrum of a cyclic product radical could be observed
at low temperature.17 We have previously noted that the absence
of the expected ESR spectrum for the product from 23 probably
was due to a translocation reaction following the cyclization
reaction, a sequence that gives an R-amide radical that might
not be observed in the ESR spectrum due to line broadening.
Such a reaction was previously implicated by product studies
of radical 21.5 Those who might wish to employ amidyl radical
cyclizations for the synthesis of N-acyl pyrrolidines and
piperidines should be aware of this possible complicating
process. The factor of 4 difference in rate constants at ambient
temperature for cyclization of 17 determined in this work and
for 24 probably is a result of the extrapolation from low-
temperature results in the ESR study. This conclusion is
supported by the observation that at 20 °C radical trans-5
cyclizes twice as fast as radical 17 and the previous observation
that the same reporter group used in 5 gives a 2-fold acceleration
in the cyclization of the 5-hexenyl radical.11
The situation with radical 17 deserves special comment.
Sutcliffe and Ingold reported that whereas 24 cyclized, radical
17 reacted at low temperature by a 1,5-hydrogen atom transfer
reaction.17 We found that at 65 °C radical 17 cyclized in a
6-exo fashion and that the translocation reaction must represent
<4% of the total reaction. Because the translocation and
cyclization reactions should have similar log A values, these
results are completely at odds with one another. From the
description in the account of the ESR studies, there seems to
be little doubt that an allylic radical was produced; the ESR
signal was described as strong, and the reported hyperfine
Bimolecular Hydrogen Transfer Reactions. The rate
constants for bimolecular reactions of amidyl radicals with Bu3-
SnH and PhSH provide the essential data for expanding an
amidyl radical kinetic scale. With these values, others can
readily calibrate amidyl radical reactions by competition kinetic
methods that require no special apparatus.13 To determine the
rate constants for reactions of amidyl radicals with hydrogen
atom transfer agents, we needed sources of amidyl radicals that
reacted cleanly in radical chain reactions with the byproduct
radical produced from the hydrogen atom donor. The N-
(phenylthio)amides clearly meet this requirement when the donor
is Bu3SnH.7 The reactions appeared to proceed within minutes,
indicating that the rate-controlling step of the chain reaction
(29) We note that a competition study performed at 20 °C with Bu3SnH
trapping of radical 20 from the PTOC imidate ester precursor (ref 5b) gives
an apparently small rate constant of 1.1 × 109 s-1 for cyclization of 20. If
this rate constant is accurate, then the log A value for cyclization of radical
20 would be 12.4, which appears to be too large. The PTOC imidate esters
are readily hydrolyzed, and we believe it is possible that small amounts of
hydrolysis of the precursor, giving the acyclic amide product, resulted in
an apparently small rate constant for cyclization.
(30) Musa, O. M.; Choi, S.-Y.; Horner, J. H.; Newcomb, M. J. Org.
Chem. 1998, 63, 786-793.

7746 J. Am. Chem. Soc., Vol. 120, No. 31, 1998
Horner et al.
sequence, the relatively slow step that one assumes is reaction
of the stannyl radical with the amide derivative, was quite rapid.
The internal consistency of the kinetics of cyclization of radicals
20 and 21, determined via tin hydride trapping using N-
(phenylthio)amide precursors, with the cyclization kinetics
measured by LFP in this work further indicates that the reactions
were uncomplicated.
N-acyl PTOC carbamate 5A, but we believe the rate constant
we obtained is reasonably reliable.
The potential translocation reaction in the amidylacyloxyl
radical precursor to radical 5 is demonstrably insignificant. If
it occurred, then this unimolecular process, which competes with
unimolecular decarboxylation, would have introduced a constant
amount of unrearranged amide product irrespective of PhSH
concentration. The manifestations of this phenomenon would
be (1) the individual values of (kT/kC) calculated for each
concentration of PhSH would differ and (2) a plot of the ratio
of unrearranged to rearranged products versus PhSH concentra-
tion would have a nonzero intercept reflecting the amount of
translocation. In fact, the individual values of (kT/kC) at all
concentrations of PhSH agree (Table 3), and the least-squares
fit of that data has an intercept of (0.0 ( 0.2) M at 2σ. Any
translocation that occurred would introduce a percentage error
in (kT/kC) equal to the percentage of translocation, that is, 20%
translocation would give a 20% error in the ratio and in the
calculated value of kT. However, for the data in Table 3 with
a slope of (kT/kC) ) 7.5 M, 20% translocation would give an
intercept of 0.25 M which is outside the 95% confidence interval
for the intercept we measured.
The potential for reaction of the amidylacyloxyl radical from
5A with PhSH before decarboxylation is more difficult to
evaluate. The rate constant for decarboxylation of the amidy-
lacyloxyl radicals must be greater than 2 × 108 s-1 at ambient
temperature because the cyclizations of radicals 1 and 4 were
found to occur faster than this, and we estimated that the rate
constant is in the 109-1010 s-1 range at room temperature. It is
known that PhSH at up to 2 M concentrations does not react
with alkylacyloxyl radicals faster than decarboxylation,36 which
occurs with rate constants in the 1010-1011 s-1 range. If one
assumes that PhSH reacts with these two types of acyloxyl
radicals with about the same rate constants, then PhSH trapping
of the amidylacyloxyl radicals would not be a major reaction
in the studies conducted here where the PhSH concentrations
were 0.1-0.5 M. Nevertheless, we caution that the rate constant
for reaction of PhSH with an amidyl radical should be
considered as an upper limit.
The reaction of Bu3SnH with amidyl radical 2 was very fast,
indicative of the high exothermicity of the reaction. The bond
dissociation energy (BDE) of the N-H bond in a secondary
amide is comparable to the N-H BDE of ammonia31 or about
107 kcal/mol, whereas an R3Sn-H BDE is only about 74 kcal/
mol.32 In addition, an electrophilic character of the radical, or
at least a lack of nucleophilic character, is suggested by the
fact that the reaction with tin hydride is about 3 orders of
magnitude faster than reaction of a primary alkyl radical with
Bu3SnH.33 Specifically, the ∆G° values for tin hydride reacting
with an amidyl radical (-34 kcal/mol) and with an alkyl radical
(-26 kcal/mol)32 are large enough such that the transition states
should be quite early, yet ∆Gq for the amidyl radical reaction
is about 4 kcal/mol smaller than that for the alkyl radical
reaction, suggesting polar stabilization in the transition state for
the amidyl radical reaction. That the amidyl radical reaction
with thiophenol is slower than reaction with Bu3SnH, whereas
a nucleophilic carbon radical reacts with PhSH more rapidly34
than it reacts with Bu3SnH,33 further indicates the electrophilic
character of an amidyl radical.
The studies of thiophenol trapping were considerably more
complicated than the tin hydride studies. In a manner similar
to alkyl halides and pseudohalides, the N-(phenylthio)amides
will not react with thiyl radicals in chain propagation steps. In
studies of carbon-centered radicals reacting with thiols or
benzeneselenol, it is possible to use a silane or tin hydride as a
sacrificial reductant which reacts with the chalcogen-centered
radical to give a Group 14 radical that will propagate a chain
reaction with an alkyl halide,35 but this approach is impractical
for amidyl radicals which react more rapidly with tin hydride
than with a thiol. The N-acyl PTOC carbamates and PTOC
imidate esters do react in chain propagation steps with thiyl
radicals, but problems exist in the applications of either of these
types of precursors for indirect studies. The more robust N-acyl
PTOC carbamates present the potential problems that the first-
formed amidylacyloxyl radicals 18 could react with the donor
intermolecularly or by intramolecular translocation to give an
R-amide radical; both processes will introduce an error by
leading ultimately to an unrearranged amide product. The
PTOC imidate esters do not have this potential problem, but
these compounds cannot be purified readily nor handled without
some hydrolysis which, again, gives the unrearranged amide
product and will introduce an error in the kinetics. Further, in
an attempt to use a PTOC imidate ester in a competition kinetic
study with a thiol, it appeared as if the activated acyl precursor
reacted with the thiol by a polar addition reaction. The kinetic
study of the reaction of thiophenol with radical 5 employed the
Conclusion
Rate constants for amidyl reactions measured directly by LFP
methods provide the absolute kinetic values necessary for
establishing an amidyl radical kinetic scale. These have been
incorporated into rate constants for bimolecular reactions of
amidyl radicals and cyclizations of simple amidyl radicals to
give kinetic values that can be used in synthetic planning or
for timing other amidyl radical reactions. Cyclizations of amidyl
radicals are much faster than those of carbon-centered radicals,
indicating that 5-exo cyclizations leading to either N-acylpyr-
rolidine or γ-butyrolactam products and 6-exo cyclizations
leading to δ-valerolactam products can be incorporated into
synthetic schemes, but 6-exo cyclizations that ultimately give
N-acylpiperidine products are slow relative to competing 1,5-
hydrogen atom transfer reactions. The development of nitrogen-
centered radical kinetic scales has been an objective of our group
for some time, and Table 4 contains an overview of the kinetic
results from this work and previously reported studies.16,20,37
(31) Bordwell, F. G.; Harrelson, J. A.; Lynch, T. Y. J. Org. Chem. 1990,
55, 3337-3341.
(32) Griller, D.; Kanabus-Kaminska, J. M.; Maccoll, A. THEOCHEM
1988, 40, 125-131.
(33) Chatgilialoglu, C.; Ingold, K. U.; Scaiano, J. C. J. Am. Chem. Soc.
1981, 103, 7739-7742.
Experimental Section
(34) Franz, J. A.; Bushaw, B. A.; Alnajjar, M. S. J. Am. Chem. Soc.
1989, 111, 268-275.
General Methods. Commercially available reagents were purchased
from either Sigma or Aldrich Chemical Co. and were used as received.
Tetrahydrofuran (THF) and diethyl ether were distilled under a nitrogen
(35) Allen, R. P.; Roberts, B. P.; Willis, C. R. J. Chem. Soc., Chem.
Commun. 1989, 1387-1388. Crich, D.; Yao, Q. W. J. Org. Chem. 1995,
60, 84-88. Crich, D.; Jiao, X. Y.; Yao, Q. W.; Harwood: J. S. J. Org.
Chem. 1996, 61, 2368-2373.
(36) Newcomb, M.; Johnson, C. C.; Manek, M. B.; Varick, T. R. J. Am.
Chem. Soc. 1992, 114, 10915-10921.

Absolute Kinetics of Amidyl Radical Reactions
J. Am. Chem. Soc., Vol. 120, No. 31, 1998 7747
Table 4. Radical Rate Constants at 20 °C
The methods employed in this work are described here. The specific
preparations of the radical precursors used in this work and the amide
precursors to these compounds are presented in the Supporting
Information.
Preparation of N-Acyl PTOC Carbamates (Method A). To 1
equiv of the amide in Et2O (15 mL) were added 1.1 equiv of TEA and
1.1 equiv of TMSOTf. The mixture was stirred at ambient temperature
for 12 h, and ammonium triflate was removed via syringe. Phosgene
in toluene (1.5 equiv, 1.9 M solution) was added via syringe at ca. 5
°C, and the mixture was stirred for 3-4 h while warming to ambient
temperature. After addition of benzene (10 mL), the solution was
concentrated to ca. 10 mL at reduced pressure on a rotary evaporator
containing a trap for the excess phosgene and TMSCl. 2-Mercapto-
pyridine-N-oxide sodium salt (1.2 equiv) was added to the flask, which
was shielded from light with aluminum foil, and the mixture was stirred
for 12 h. Filtration of the salts and dilution of the filtrate in benzene
gave an organic phase which was washed with saturated aqueous
NaHCO3 solution (3 × 10 mL) and saturated aqueous NaCl solution
(3 × 10 mL). The organic layer was dried over MgSO4, filtered, and
concentrated to yield the crude N-acyl PTOC carbamate which was
purified by column chromatography on silica gel (eluent, hexanes:
EtOAc).
Preparation of PTOC Imidate Esters (Method B). To a 0.1 M
solution of a secondary amide in benzene containing a catalytic amount
of DMF at ca. 5 °C was transferred phosgene (1.5 equiv, 1.9 M solution
in toluene). The solution was allowed to warm slowly to ambient
temperature with stirring. The slow evolution of CO2 was monitored
via a mineral oil bubbler. After gas evolution was complete, excess
phosgene and solvent were removed under reduced pressure to give
the corresponding imidoyl chloride as a clear liquid. Dry Et2O was
added to the imidoyl chloride to form a 0.1 M solution. The flask was
wrapped in aluminum foil, and anhydrous 2-mercaptopyridine-N-oxide
sodium salt (1.2 equiv) was added via a solids addition tube under
nitrogen. The mixture was stirred for 4-8 h. Residual salts were
removed by filtration under N2 and were rinsed with dry Et2O.
Concentration of the combined organic phase under vacuum gave the
imidate esters as yellow oils. These products were not further purified.
Preparation of N-(Phenylthio)alkenylamides (Method C). To
NaH (2.75 mmol, 60% in oil) in dry THF (15 mL) was added a
secondary amide (2.50 mmol), and the mixture was heated at reflux
for 4 h. The mixture was cooled to -78 °C, and 0.30 mL of
phenylsulfenyl chloride, prepared by the method of Brower and
Douglass,38 was added dropwise until the yellow color of the sulfenyl
chloride was persistent. The solvent was removed under reduced
pressure, and the residue was partitioned between Et2O (30 mL) and
H2O (10 mL). The ether layer was dried over MgSO4 and concentrated
under reduced pressure. The residue was purified by chromatography
on silica gel (eluent, hexanes:EtOAc) to give the N-(phenylthio)amides
as colorless oils.
Photolysis of N-Hydroxypyridine-2-thione Derivatives (Method
D). A 0.025-0.05 M solution of the N-acyl PTOC carbamate in dry
benzene or dry THF was degassed under N2 for 15-30 min in the
dark. Trapping agents were added via syringe from a 0.05-0.5 M
stock solution in the same solvent. The sample was irradiated with
visible light from a 150 W tungsten filament flood lamp at a distance
of 2-3 ft until disappearance of the PTOC carbamate (followed by
TLC). Removal of the solvent at reduced pressure was followed by
separation of the products using radial chromatography (silica gel,
hexanes:EtOAc elution).
LFP kinetic studies were performed on a modified Applied
Photophysics LK-50 kinetic spectrometer using the Applied Photo-
physics software for instrument control and data analysis. A Spectron
Nd:YAG laser operating at a nominal power rating of 40 mJ (355 nm)
or 45 mJ (266 nm) was used for photolysis. Hammamatsu 1P28
photomultiplier tubes and a Hewlett-Packard 54522 oscilloscope were
employed. Solutions of the desired precursor in THF with an
absorbance of 0.4-0.5 at the laser wavelength were sparged with helium
and thermally equilibrated in a jacketed addition funnel by circulation
a Rate constant for 5-exo or 6-exo cyclization. b Rate constant for
reaction with Bu3SnH. c Rate constant for reaction with PhSH. d Trans-
location (1,5-hydrogen transfer) to give the allylic radical occurs; see
text. e Estimated from results with the terminal diphenyl-substituted
analogue.
atmosphere from sodium benzophenone ketyl. Methylene chloride was
distilled under nitrogen from phosphorus pentoxide. Dimethyl forma-
mide (DMF) was distilled under reduced pressure from calcium hydride.
The sodium salt of N-hydroxypyridine-2-thione was purified as
described previously.5b 1H NMR spectra were obtained at 300 or 500
MHz on Varian Gemini 300 or Unity 500 spectrometers, respectively.
13C NMR spectra (75 or 125 MHz) were obtained on the same
instruments. Melting points were obtained on a Thomas-Hoover
capillary melting point apparatus and are uncorrected. High-resolution
mass spectral analyses were performed by the Central Instrumentation
Facility at Wayne State University. GC analyses were performed on
Varian model 3400 FID-equipped chromatographs unless otherwise
noted.
(37) (a) Chatgilialoglu, C.; Ingold, K. U.; Scaiano, J. C. J. Am. Chem.
Soc. 1981, 103, 7739-7742. (b) Franz, J. A.; Bushaw, B. A.; Alnajjar, M.
S. J. Am. Chem. Soc. 1989, 111, 268-275. (c) Newcomb, M.; Musa, O.
M.; Martinez, F. N.; Horner, J. H. J. Am. Chem. Soc. 1997, 119, 4569-
4577. (d) Le Tadic-Biadatti, M. H.; Callier-Dublanchet, A. C.; Horner, J.
H.; Quiclet-Sire, B.; Zard, S. Z.; Newcomb, M. J. Org. Chem. 1997, 62,
559-563.
(38) Brower, K. R.; Douglass, I. B. J. Am. Chem. Soc. 1951, 73, 5787-
5789.

7748 J. Am. Chem. Soc., Vol. 120, No. 31, 1998
Horner et al.
of a water/ethylene glycol solution from a thermally regulated bath.
For reactions conducted below 0 °C, the flow cell was positioned in a
nitrogen-filled box fitted with quartz windows. The solutions were
allowed to flow through a quartz flow cell with an 8 mm × 8 mm i.d.
Temperatures were measured with a thermocouple placed in the flowing
stream approximately 1 cm above the irradiation zone. Multiple runs
(5-15) at a given temperature were averaged. The observed temper-
ature variation over the course of multiple runs was (0.2 °C. Observed
rate constants had standard errors of 1-5%.
Indirect kinetic studies of radical 2 were achieved by reactions of
N-(phenylthio)amide 2P with different concentrations of Bu3SnH in
toluene. Stock solutions of amide 2P (0.024 M), Bu3SnH (0.50 M),
and AIBN (0.01 M) were prepared under N2. Appropriate volumes of
these solutions were combined in tubes that were sealed under vacuum.
After heating for 5 h at the temperatures indicated in Table 2, the tubes
were opened, hexadecane was added as a standard, and the yields of
15 and 16 were determined by GC analysis (DB-17 column, TCD
detector GC). Indirect kinetic studies of radical 17 were conducted in
a similar manner using precursor 17P at 65 °C.
Acknowledgment. This work was supported by a grant to
M.N. from the National Science Foundation (CHE-9614968).
We are grateful to Dr. Felix N. Martinez for some preliminary
LFP studies and to Mr. Calvin J. Emanuel for providing the
UV spectrum of radical 12.
Supporting Information Available: Tables of LFP kinetic
results for radicals 2, 3, and 5, synthetic details for the
preparation of radical precursors, synthetic details for preparation
of the requisite amides, and NMR spectra of new compounds
(58 pages, print/PDF). This material is contained in libraries
on microfiche, immediately follows this article in the microfilm
version of the journal, and can be ordered from the ACS. See
any current masthead page for ordering information and Web
access instructions.
JA981244A