β-Phosphatoxyalkyl Radical Reactions: Competing Phosphate Migration and Phosphoric Acid Elimination from a Radical Cation - Phosphate Anion Pair Formed by Heterolytic Fragmentation

Article abstract of DOI:10.1021/ja991012r

β-Phosphatoxyalkyl radical reactions were studied experimentally and computationally. The 1,1-dibenzl-2-(diphenylphosphatoxy)-2-phenylethyl radical (1) reacted to give the migration product 2-benzyl-2-(diphenylphosphatoxy)-1,3-diphenylpropyl radical (2) a

Full text of DOI:10.1021/ja991012r

J. Am. Chem. Soc. 1999, 121, 10685-10694
10685
â-Phosphatoxyalkyl Radical Reactions: Competing Phosphate
Migration and Phosphoric Acid Elimination from a Radical
Cation-Phosphate Anion Pair Formed by Heterolytic Fragmentation
Martin Newcomb,*,1a John H. Horner,1a Patrick O. Whitted,1a David Crich,*,1b
Xianhai Huang,1b Qingwei Yao,1b and Hendrik Zipse*,1c
Contribution from the Department of Chemistry, Wayne State UniVersity, Detroit, Michigan, 48202,
Department of Chemistry, UniVersity of Illinois at Chicago, 845 West Taylor Street,
Chicago, Illinois 60607-7061, and Institute fu¨r Organische Chemie, Ludwig-Maximilians-UniVersita¨t,
Butenandtstrasse 5-13, D-81377, Mu¨nchen, Germany
ReceiVed March 29, 1999. ReVised Manuscript ReceiVed September 30, 1999
Abstract: â-Phosphatoxyalkyl radical reactions were studied experimentally and computationally. The 1,1-
dibenzyl-2-(diphenylphosphatoxy)-2-phenylethyl radical (1) reacted to give the migration product 2-benzyl-
2-(diphenylphosphatoxy)-1,3-diphenylpropyl radical (2) and the elimination product 2-benzyl-1,3-diphenylallyl
radical (3) in a variety of solvents. A modest kinetic solvent effect for reactions of 1 was found. Variable
temperature studies in THF and acetonitrile gave Arrhenius functions with similar log A terms; the entropies
of activation are ∼-5 eu. A deuterated analogue of radical 1 reacted in THF and acetonitrile with rate constants
indistinguishable from those of 1, but the ratio of products 2:3 increased for the deuterated radical requiring
kinetic isotope effects (KIEs) in reactions following the rate-limiting step. In aqueous acetonitrile solutions,
the â,â-dibenzylstyrene radical cation (4) was detected as a short-lived intermediate, and the rate constants for
formation of 3 and 4 indicated that both species derived from a common intermediate. The 1,1-dimethyl-2-
(dimethylphosphatoxy)ethyl radical (C1) was studied computationally. Transition states for concerted phosphate
migrations and phosphoric acid elimination were found with energies in the order [1,2]-migration < [1,3]-
elimination < [3,2]-migration; each transition state was more polarized than radical C1. A transition state for
homolytic fragmentation of C1 to give 2-methylpropene and the dimethylphosphatoxyl radical could not be
found, but the reaction from the ensemble of these two entities to give the 2-methylallyl radical and phosphoric
acid was followed computationally. KIEs and solvent dielectric effects were computed for each concerted
reaction of C1. The results indicate that radical 1 reacts in all solvents studied by a common pathway involving
initial heterolysis. The first-formed contact pair of radical cation 4 and diphenyl phosphate anion collapses to
products 2 and 3 and, as a minor process, evolves to diffusively free radical cation 4 in aqueous acetonitrile
solutions. A model for heterolytic fragmentation of â-ester radicals involving contact and solvent-separated
ion pairs is presented.
â-Phosphatoxyalkyl radicals have received increased attention
in recent years in large part because such species are implicated
in DNA degradation by anti-cancer agents such as bleomycin
and enediyne antibiotics.2 Reactions of these radicals are
complex.3 They can react to give elimination products, substitu-
tion products, or rearrangement products (Figure 1), and multiple
reaction pathways to each product are possible. For example,
initial heterolysis of the â-phosphatoxyalkyl radical to a radical
cation and phosphate anion followed by proton transfer,
nucleophilic capture of the radical cation, or recombination
would produce the various products. Alternatively, pathways
that do not involve radical cation formation are possible; these
include concerted eliminations and migrations and direct nu-
cleophilic addition at the radical center with displacement of
the â-phosphate in a reaction that is the open-shell analogue of
an SN2′ reaction. In a seeming paradox, the consensus views of
the mechanisms of â-phosphatoxyalkyl radical reactions as
presented in a recent review3 have been selected from both types
of processes. Some details are presented in the Discussion, but
we note here that the general view of phosphate migrations is
that they are concerted, whereas elimination and substitution
reactions are thought to involve initial heterolysis to a radical
cation and phosphate anion.
Our respective groups have addressed the mechanisms of
â-phosphatoxyalkyl radical reactions from experimental and
computational perspectives. Stereochemical labeling studies
implicated concerted migration reactions.4,5 Laser flash pho-
tolysis (LFP) kinetic studies of â-phosphatoxyalkyl radicals
coupled with product analyses found that phosphate migrations,
an elimination and a nucleophilic substitution reaction proceeded
without diffusively free radical cation intermediates, and the
kinetic results were considered to be most consistent with
concerted processes.6,7 Low energy pathways for concerted
(4) Crich, D.; Yao, Q. W. J. Am. Chem. Soc. 1993, 115, 1165-1166.
(5) Crich, D.; Yao, Q. W.; Filzen, G. F. J. Am. Chem. Soc. 1995, 117,
11455-11470.
(6) Choi, S. Y.; Crich, D.; Horner, J. H.; Huang, X. H.; Martinez, F. N.;
Newcomb, M.; Wink, D. J.; Yao, Q. W. J. Am. Chem. Soc. 1998, 120,
211-212.
(1) (a) Wayne State University. (b) University of Illinois at Chicago. (c)
Ludwig-Maximilians-Universita¨t.
(2) Pogozelski, W. K.; Tullius, T. D. Chem. ReV. 1998, 98, 1089-1107.
(3) Beckwith, A. L. J.; Crich, D.; Duggan, P. J.; Yao, Q. W. Chem. ReV.
1997, 97, 3273-3312.
10.1021/ja991012r CCC: $18.00 © 1999 American Chemical Society
Published on Web 11/03/1999

10686 J. Am. Chem. Soc., Vol. 121, No. 46, 1999
Newcomb et al.
Scheme 1
Scheme 2a
a (a) (1) LDA, (2) benzaldehyde; (b) ClP(O)(OPh)2; (c) CAN; (d)
2,2′-dipyridyl disulfide bis-N,N′-oxide, Bu3P.
PTOC ester precursors. These precursors, which are isoelectronic
with diacyl peroxides, have long wavelength absorbances with
Figure 1. Reactions of â-phosphatoxyalkyl radicals.
λ
max at about 360 nm and are cleaved with high efficiency. The
[1,2]- and [3,2]-phosphate shifts, for concerted syn-[1,3]
eliminations, and for nucleophilic â-displacement of phosphate
were found computationally.8
photolysis initially gives the pyridine-2-thiyl radical (5), which
has a broad long wavelength absorbance with λmax at 490 nm,9
and an acyloxyl radical, but acyloxyl radicals decarboxylate with
subnanosecond lifetimes to give the desired alkyl radicals
(Scheme 1). In subsequent schemes, the pyridine-2-thioneoxy-
carbonyl moiety is represented as PTOC, and the short-lived
intermediate acyloxyl radicals are not shown.
The LFP work was made possible by synthetic advances that
permitted the preparation of â-phosphatoxycarboxylic acids, the
requisite precursors to the PTOC esters. Such compounds
provide a synthetic challenge because phosphoric acid elimina-
tion to give acrylate products is facile, and, in fact, â-phenyl-
â-phosphatoxycarboxylic acids could only be prepared when
the R-carbon was quaternary. The synthetic sequence for the
precursor to radical 1, PTOC ester 10, is shown in Scheme 2.
The key feature is acid protection with 2,4-dimethoxybenzyl
alcohol which was removed oxidatively with reasonable ef-
ficiency.
Low-temperature condensation of the lithium enolate of the
2,4-dimethoxybenzyl ester of 2-benzyldihydrocinnamic acid (6)
with benzaldehyde gave the aldol product 7 in good yield. This
was then treated with diphenylphosphoryl chloride to give the
stable phosphate ester 8, which could be purified by silica gel
chromatography. The 2,4-dimethoxybenzyl ester was removed
by treatment with ceric ammonium nitrate. The â-(phosphatoxy-
)alkanoic acid 9 so-obtained was relatively unstable, and
attempted passage over silica gel resulted in the formation of a
â-lactone. Thus, after removal of the 2,4-dimethoxybenzalde-
hyde byproduct, by extensive washing with sodium bisulfite,
the crude acid 9 was immediately converted to the PTOC ester
10 by treatment with 2,2′-dipyridyl disulfide N,N′-dioxide. This
method of PTOC formation10 was selected from the many
available11 as it enabled their rapid elution from silica gel ahead
In a preliminary publication,6 we reported that the 1,1-
dibenzyl-2-(diphenylphosphatoxy)-2-phenylethyl radical (1) re-
acts by a combination of migration and elimination reactions
to give product radicals 2 and 3, which, according to the
consensus views,3 arise from completely different mechanisms.
This system is, thus, well-suited for mechanistic scrutiny. We
report here details of LFP studies of reactions of radical 1 and
a deuterated analogue (1-d14 with perdeutero benzyl groups) and
computational studies of a model of radical 1 that provide
barriers for various concerted processes in a common system.
Consistent Arrhenius log A parameters for reactions of 1 and
smooth increases in rates with solvent polarity increases
implicate a common mechanism for migration and elimination
reactions in both low and high polarity solvents. The absence
of a kinetic isotope effect (KIE) in the overall reactions of 1
and the detection of radical cation 4 in reactions conducted in
aqueous solutions indicate that the mechanism involves initial
heterolytic fragmentation. We discuss a mechanistic model for
â-phosphatoxyalkyl radical reactions involving reactions in ion
pairs that appears to be consistent with extant results.
Results
LFP Studies. All radicals studied in this work were produced
by 355 nm (third harmonic of a Nd:YAG laser) photolysis of
(9) Alam, M. M.; Watanabe, A.; Ito, O. J. Org. Chem. 1995, 60, 3440-
3444.
(7) Choi, S.-Y.; Crich, D.; Horner, J. H.; Huang, X.; Newcomb, M.;
Whitted, P. O. Tetrahedron 1999, 55, 3317-3326.
(8) Zipse, H. J. Am. Chem. Soc. 1997, 119, 2889-2893.
(10) Barton, D. H. R.; Samadi, M. Tetrahedron 1992, 48, 7083-7090.
(11) Crich, D.; Quintero, L. Chem. ReV. 1989, 89, 1413-1432.

â-Phosphatoxyalkyl Radical Reactions
J. Am. Chem. Soc., Vol. 121, No. 46, 1999 10687
for 2M with a λmax at 320 nm is typical for benzylic radicals,12
and the â-phosphate group obviously has little effect on the
position of the absorbance.
Allylic radical 3 was produced from the PTOC ester precursor
12. Figure 2B shows a time-resolved decay spectrum of this
radical as it reacts relatively slowly. The spectra were obtained
by subtracting the spectrum at 250 µs from those at earlier (3-
100 µs) times.13 The extinction coefficient for radical 3 is so
large that signals from radical 5 do not obscure the spectrum,
although a small amount of decay at 490 nm from radical 5
can be seen in the figure. The spectrum for allylic radical 3
with λmax at 354 nm is similar to that previously reported for
the 1,3-diphenylallyl radical.14
Figure 2. (A) Spectrum of radical 2M in acetonitrile; see text for an
explanation of the method. (B) Time-resolved decay spectrum of radical
3 in acetonitrile. (C) Time-resolved decay spectrum of radical cation
4M produced by photoejection from the corresponding styrene in
acetonitrile. (D) Spectrum of radical cation 4 produced by triplet
chloranil oxidation of the corresponding styrene; see text for an
explanation of the method.
Styrene radical cations can be obtained by LFP methods by
direct photoejection from the styrene or by chemical oxidation
by a triplet state oxidant produced by photolysis. They have
strong absorbances in the 350-390 nm region and weaker
absorbances at longer wavelengths (>580 nm).15 Direct pho-
toejection from 2-methyl-1-phenylpropene in CH3CN by 266
nm irradiation gave the time-resolved spectrum of radical cation
4M shown in Figure 2C; radical cation 4M produced in
trifluoroethanol also has an absorbance centered at 380 nm.15
When the same conditions were used with 2-benzyl-1,3-
diphenylpropene, however, the observed spectrum was due to
multiple species. The spectrum had λmax at ∼355 nm with a
shoulder at 380-390 nm that decayed rapidly relative to the
absorbance at 355 nm, and after several µs, the observed
spectrum was similar to that from allylic radical 3. Apparently,
the predominant absorbance was due to 3 formed instantly on
the nanosecond time scale by a deprotonation reaction occurring
in the contact ion pair of radical cation 4 and the acetonitrile
radical anion.
of any more polar reagents and byproducts. The procedure was
repeated with acid 6D containing perdeuterated benzyl groups
to give 10D. PTOC ester 11, the precursor to benzylic radical
2M, was prepared by an analogous route involving aldol
condensation of 2,4-dimethoxybenzyl phenylacetate with ac-
etone, followed by phosphorylation and deprotection. PTOC
ester 12, the precursor to radical 3, was prepared as a mixture
of diastereomers from a mixture of (E)- and (Z)-3-benzyl-2,4-
diphenyl-3-butenoic acids resulting from carboxylation of the
corresponding allyllithium reagent.
The putative products from reactions of radical 1 are the
benzylic radical 2, the diphenylallylic radical 3, and the styrene
radical cation 4. One could anticipate from literature reports
that the UV spectra of 2-4 would be distinct, but authentic
spectra were desired for unambiguous identifications of the
products. We obtained spectra of radicals 2M (a model for 2)
and 3 and radical cation 4.
Figure 2A shows the spectrum of benzylic radical 2M in
acetonitrile obtained in the following manner. Photolysis of
PTOC precursor 11 gave 2M and byproduct radical 5, but
because the extinction coefficient for the benzylic radical is
relatively small, the prominent signals were from formation of
5 and bleaching of the precursor. We have observed that the
rate of decay of radical 5 is not affected by the presence of
oxygen; apparently this radical decays mainly by recombination
reactions. The benzylic radical 2M will react rapidly with
oxygen, but it is relatively persistent in the absence of oxygen.
Therefore, decay spectra from He-sparged and nonsparged
solutions of precursor 11 were obtained over the period 0.2-
4.5 µs, and the decay spectrum for the nonsparged solution was
subtracted from that of the sparged solution. This subtraction
procedure removed the absorbance signals from 5 and the
bleaching signals from the precursor to leave the residual
spectrum of 2M that is shown in Figure 2A. The absorbance
Production of radical cation 4 by oxidation of the corre-
sponding styrene with triplet excited chloranil (formed by 355
nm excitation) in CH3CN solution was possible, but the spectra
were complex. The relatively long-lived chloranil triplet has
(12) Chatgilialoglu, C. In Handbook of Organic Photochemistry; Scaiano,
J. C., Ed.; CRC Press: Boca Raton, FL, 1989; Vol. 2, pp 3-11.
(13) At short reaction times (<1 µs), a fast dynamic process was observed
for radical 3. The first-observed spectrum had λmax at 360 nm, but a rapid
6 nm hypsochromic shift of λmax and a slight ( ∼10%) increase in absorbance
occurred with a rate constant of 2 × 106 s-1 at 20 °C. Given the low
concentrations of precursor and radicals in the LFP experiments, this
dynamic process cannot be due to radical-radical or radical-precursor
reactions, and we ascribe it to a unimolecular process. Apparently, radical
3 was initially produced in a nonequilibrium distribution of conformations
and relaxed rapidly by bond rotation.
(14) Miranda, M. A.; Perez-Prieto, J.; Font-Sanchis, E.; Ko´nya, K.;
Scaiano, J. C. J. Org. Chem. 1997, 62, 5713-5719.
(15) Johnston, L. J.; Schepp, N. P. J. Am. Chem. Soc. 1993, 115, 6564-
6571.

10688 J. Am. Chem. Soc., Vol. 121, No. 46, 1999
Newcomb et al.
The spectrum in Figure 3A is a composite of signals from
two species. The strong absorbance with λmax at 354 nm is due
to allylic radical 3. In addition, a signal with λmax at ∼320 nm
(from benzylic radical 2) is present. The signal from 2 was more
apparent in time-resolved spectra from reactions of 1-d14. Figure
3B shows the results from the deuterated radical obtained under
the same conditions as used for the spectrum of the nondeu-
terated radical in Figure 3A. The 490 nm absorbances of radical
5 in A and B of Figure 3 were normalized so that the intensities
of the signals between 300 and 370 nm could be compared
directly to visualize yields. A and B of Figure 3 show the same
period of reaction time (7.6 µs), and the rate constants for
reactions of the two radicals are the same (see below). The
difference in relative intensities of the 320 and 354 nm signals
from the d0 and d14 radicals demonstrates that at least two
products are formed in reactions of radical 1 and that a KIE is
present in at least one of the product-forming steps.
The kinetics of reactions of â-phosphatoxyalkyl radical 1 were
measured in various solvents (Table S1 in Supporting Informa-
tion), and Arrhenius functions were determined in THF and in
acetonitrile (Table 1). A kinetic solvent effect was apparent with
the rate constants increasing with solvent polarity, and the
Arrhenius parameters indicate that this acceleration was due to
a reduction in the activation energy. The precision in the kinetic
measurements of first-order radical reactions with rate constants
in the range of 104 s-1 is poorer than that for rate constants of
∼105 s-1 because radical-radical reactions and reactions of
radicals with residual oxygen are competitive with the slower
processes, and we consider the rate constant measured in
benzene (2 × 104 s-1 at 20 °C) to be an upper limit. The listed
errors reflect the precision of the measurements, and the
reproducibilities of values measured at the same temperature
were good. Rate constants were the same for cases in THF where
the kinetics were determined at both 320 and 358 nm showing
that the two products formed came from the same precursor
and were not subject to secondary reactions on the time scales
of their formation. The reduced precision in the rate constants
measured in THF and the limited temperature range over which
the reactions could be studied in THF led to reduced precision
in the Arrhenius functions in that solvent in comparison to those
for reactions in acetonitrile.
Figure 3. Time-resolved spectra from reaction of radical 1-d0 (A) and
radical 1-d14 (B) in acetonitrile at 19.8 °C; the spectra show signal
evolution over 7.6 µs with the signals at 0.8 µs subtracted to give a
baseline. (C) Time-resolved spectrum from reaction of radical 1 in
water/acetonitrile (1:1, v:v) at 22 °C; signal evolution is over 1.0 µs
with the spectrum at 100 ns subtracted to give a baseline; signals from
300 to 370 nm are growing, and the signal centered at 390 nm is
decaying after rapid growth.
broad absorbances centered at 370 and 510 nm,16 and the
chloranil radical anion, formed upon oxidation of the styrene,
absorbs at 320 and 440 nm.17 In addition, a persistent signal
with λmax at ∼355 nm, apparently due to allylic radical 3, was
present in the spectra. The spectrum of radical cation 4 shown
in Figure 2D was obtained by subtracting the spectrum at 45
µs from that at 5 µs. Under the conditions employed, the
chloranil triplet signals had decayed completely by 5 µs, leaving
signals from the chloranil radical anion, allylic radical 3, and
radical cation 4. In the period 5-45 µs, signals from radical
cation 4 decayed more rapidly than those from other species.
Because the residual spectrum at 45 µs still contained the signal
from 3, the difference spectrum in Figure 2D is due mainly to
4 (λmax at 390 nm) and the chloranil radical anion (small
absorbances at 320 and 440 nm).
Rate constants and Arrhenius parameters for reactions of 1-d14
were also obtained in THF and acetonitrile (Tables S1 and 1).
In regard to the overall rate constants of the reactions, there
was no apparent (KIE). However, a KIE was present in the
product distributions as is clear from the spectra in Figure 3.
Attempts to determine the ratios of product 2 to product 3
from the signal intensities at 320 and 354 nm were complicated
by the relatively strong absorbance of radical 3 at 320 nm.
Therefore, we estimated the amount of radical 3 produced in
reactions in acetonitrile by an alternative approach. The relative
intensities of the absorbances at 354 nm (from radical 3) and at
490 nm (from radical 5) from reactions of 1 in acetonitrile were
compared to the relative intensities of these two absorbances
observed when allylic radical 3 was produced from its PTOC
ester precursor 12. These comparisons gave the percentages of
3 formed in reactions of 1-d0 and 1-d14, and the remainder of
the product was assumed to be radical 2. From these analyses,
the ratio of 2:3 from reaction of 1-d0 at 20 °C was 40:60 and
Laser photolysis of PTOC ester 10 gave radical 1, which has
no prominent chromophore at wavelengths greater than 300 nm,
and the byproduct radical 5. UV spectra obtained immediately
after photolysis showed the absorbance of radical 5 and strong
bleaching centered at 360 nm due to destruction of the PTOC
ester. As radical 1 reacted, a strong signal centered at about
350 nm evolved in all solvents studied. Figure 3A shows a
representative time-resolved spectrum obtained in acetonitrile.
The initial spectrum was subtracted from subsequent spectra to
remove the features produced immediately upon photolysis. The
UV absorbances from the products evolving with time are
positive signals, and the negative signal with λmax at 490 nm is
due to decay of radical 5.
(16) Rathore, R.; Hubig, S. M.; Kochi, J. K. J. Am. Chem. Soc. 1997,
119, 11468-11480.
(17) Andre, J. J.; Weill, G. Mol. Phys. 1968, 15, 97-99.

â-Phosphatoxyalkyl Radical Reactions
J. Am. Chem. Soc., Vol. 121, No. 46, 1999 10689
Table 1. Arrhenius Parameters for Radical Reactions
Table 2. Kinetics of Reactions of Radical 1 in Water/Acetonitrile
Solutions at 22 ( 2 °Ca
b
radical
solvent
Arrhenius functiona
k20 (s-1
)
25% water
50% water
67% water
1
THF
CH3CN
THF
(11.6 ( 0.9) - (9.6 ( 1.2)/θ
(11.9 ( 0.2) - (8.5 ( 0.3)/θ
(10.8 ( 0.4) - (8.4 ( 0.6)/θ
(12.2 ( 0.2) - (8.8 ( 0.3)/θ
(12.1 ( 0.6) - (8.1 ( 0.9)/θ
(10.9 ( 0.3) - (6.2 ( 0.5)/θ
(11.0 ( 0.5) - (5.0 ( 0.7)/θ
(10.7 ( 0.5) - (6.2 ( 0.6)/θ
(11.2 ( 0.3) - (5.9 ( 0.4)/θ
3 × 104
4 × 105
3 × 104
4 × 105
390 nm growth
390 nm decay
5.2 ( 1.0
9.4 ( 0.6
7.4 ( 0.8
4.1 ( 0.2
4.6 ( 0.4
3.03 ( 0.16
7.1 ( 1.2
2.1 ( 0.2
14.7 ( 1.2
1-d14
3.3 ( 0.5
5.2 ( 0.5
CH3CN
benzenec
THFd
1M
1.1 × 106
1.9 × 106
1.8 × 107
1.2 × 106
6 × 106
360 nm first-orderb
2.11 ( 0.02
4.5 ( 2.8
2.5 ( 0.2
3.3 ( 0.4
17 ( 2
1.8 ( 0.2
c
CH3CNd
THFd
360 nm k1
360 nm k2
c
13
CH3CNd
a 10-6 × k (s-1); errors at 1σ. b First-order fit for growth at 360 nm.
c Rate constants in Scheme 3.
a Errors at 2σ, θ ) 2.3RT in kcal/mol. b Calculated rate constant at
20 °C. c See ref 27. d See ref 7.
Scheme 3
that from reaction of 1-d14 was 70:30. We note that the ratio of
products for reaction of 1 found in the approach used here differs
from that we previously estimated from signal intensities at 320
and 354 nm,6 but we believe the method used here is more
reliable.18 The change in the product ratio 2:3 does not
necessarily translate directly to a KIE value because both
reactions could have KIE, but if we assume that the rate of
formation of the benzylic radical 2 is not affected by deuterium
substitution, the KIE for formation of allylic radical 3 is kH/kD
) 3-4.
The evolving spectra from reactions of radical 1 in acetonitrile
and less polar solvents had no appreciable absorbance in the
region 380-400 nm where radical cation 4 absorbs, and it is
apparent that diffusively free 4 was not formed in detectable
amounts. Radical 1 reacted rapidly in water/acetonitrile mixtures
at ambient temperatures (submicrosecond lifetimes), and a signal
from a new transient that we ascribe to 4 was observed in
addition to the features seen in less polar solvents.
Figure 3C shows the time-resolved spectrum from reaction
of 1 in 50% water (by volume) in acetonitrile. The time period
in this figure is between 100 ns and 1.0 µs with the signals at
100 ns subtracted from later spectra to give a baseline. Growth
of an absorbance at 390 nm was complete by 200 ns, and this
signal is decaying in later spectra, whereas the absorbances
between 300 and 370 nm are growing throughout. Similar
behavior was observed in water solutions of 25 and 67%. The
decaying absorbance in Figure 3C closely matches the UV
spectrum of radical cation 4 in Figure 2D, and the observed
kinetic behavior at 390 and 360 nm supports the assignment of
the transient as 4.19
The dynamics of the 390 nm signal were solved for double
exponential behavior, first-order growth, and first-order decay,
and the results are given in Table 2. The rate constants for the
formation of this signal increased in a manner consistent with
the expected trends in reactivity for radical 1 seen in less polar
media. The pseudo-first-order rate constants for decay of the
390 nm signal are similar to those for reactions of styrene radical
cations with alcohols reported by Johnston,15 and we consider
this to be good supporting evidence that the 390 signal is due
to radical cation 4.
ratio of the ultimate signal intensity at 360 nm to the
instantaneous signal intensity at 490 nm from byproduct radical
5 was essentially constant (ratios within 10%) in acetonitrile
and the aqueous acetonitrile reactions, indicating that, irrespec-
tive of the formation of a new intermediate, the reactions
eventually gave mainly allylic radical 3. When the kinetics of
growth in the 350-360 nm region were solved for a single-
exponential rate law, rate constants in the range of 2-3 × 106
s-1 were obtained (Table 2). However, the precision of the first-
order rate constants was increasingly poorer for increasing
percentages of water, and inspection of the residuals from the
“best first-order fit” of the traces suggested that the rate of
formation was more complex than a first-order exponential
process. Attempts to solve the data by the model of consecutive
first-order reactions did not improve the fits; the regression
analyses consistently converged to solutions in which one rate
constant was very large ( ∼108 s-1) and the other was 2-3 ×
106 s-1
.
In the Discussion, we will argue that 1 reacts by initial
heterolysis to give a contact ion pair that can react to give 2
and 3 or, in aqueous solutions, diffuse apart to give 4. If this is
the case, then 3 could be formed by two processes in aqueous
acetonitrile media, reaction within the contact ion pair and
deprotonation of diffusively free radical cation 4 by water. The
kinetics could be complex with double exponential growth
behavior of a first-order reaction combined with consecutive
first-order reactions, although the data treatment is simplified
somewhat by the fact that the rate constant for the slow initial
reaction (the heterolysis) would be both the rate constant for
the first-order process and that for the first step in the
consecutive reaction sequence. Thus, the data traces for the
signals in the 350-360 nm region were solved for the model
shown in Scheme 3 where kA, kB . k1. Inspection of the
residuals for these data analyses indicated that the solutions were
better than those for either of the other data treatments, but any
double exponential solution would have given an improvement
over the fits of the other data treatments. More importantly,
the variable rate constants for the initial reaction ultimately
giving the 350-360 nm signal (k1 in Scheme 3) were essentially
equal to those for the formation of the 390 nm signal, albeit at
poorer precision, and the second rate constants in the consecutive
reaction sequence (k2) were similar to those for signal decay at
390 nm (Table 2). The data analyses also provided ratios for
the partitioning of the initially formed intermediate (i.e., kA/kB)
which indicated that the major reaction pathway to the 350-
The kinetic behavior of the signals in the 350-360 nm region
from reactions in aqueous media was more informative. The
(18) Qualitatively, the spectra from reactions of 1-d0 and 1-d14 in THF
were similar to those observed in acetonitrile. Quantitative measurements
of the ratios of products by the method used for the product ratios in
acetonitrile were not possible because the overall reactions were too slow
in THF. Depletion of radicals by competitive radical-radical reactions in
THF would lead to overestimation of the amounts of radical 2.
(19) An experiment in an aqueous acetonitrile solution reported in our
preliminary communication (ref 6) gave the same pseudo-first-order rate
constant for growth at 360 nm as found here, but the temporal resolution
in that experiment was not adequate for detection of the fast signal growth
at 390 nm. The kinetic results in this work for aqueous solutions were
obtained with a fast oscilloscope using 0.5 or 1.0 ns resolution.

10690 J. Am. Chem. Soc., Vol. 121, No. 46, 1999
Newcomb et al.
Figure 4. Potential energy surface for reactions of the 1,1-dimethyl-2-(dimethylphosphatoxy)ethyl radical (C1) as calculated at the Becke3LYP/
6-311+G(d,p)//Becke3LYP/6-31G(d) + ∆ZPE level of theory.
Table 3. Relative Energies (in kcal/mol) for Stationary Points in
360 nm signal was the direct formation route (60-70%).
Concerted Reactions of C1
Because the consecutive reactions sequence is the minor
∆Εc [B3LYP/
∆E0d [B3LYP/
pathway and because decay of the transient absorbing at 390
nm would have a minor effect on the observed signal in the
350-360 nm region, the differences in values between k2 from
Scheme 3 and the rate constant for decay of the 390 nm signal
seem to be acceptable.
∆Εa
∆E0
[B3LYP/
6-311+G(d,p)// 6-311+G(d,p)//
b
[B3LYP/
B3LYP/
6-31G(d)]
B3LYP/
6-31G(d)]
structure 6-31G(d)] 6-31G(d)]
C1
C2
C3
C4
0.0
+24.1
+19.5
+17.3
+5.0
+10.2
+23.3
+23.8
+22.7
0.0
+19.6
+18.5
+15.9
+4.0
0.0
+19.3
+17.4
+14.9
+4.6
0.0
+14.7
+16.4
+13.5
+3.6
Computational Studies. Previous computational results
demonstrated that low-energy concerted pathways for migrations
and eliminations of â-phosphatoxyalkyl radicals existed. Be-
cause radical 1 reacts by both migration and elimination, we
have analyzed the reactions of a model radical that has both
reaction channels available, the 1,1-dimethyl-2-(dimethylphos-
phatoxy)ethyl radical (C1), which represents a reasonable
compromise between our desire to compute barriers at high
levels of theory and the large number of heavy atoms in radical
1. All barriers obtained for radical C1 should be higher than
those in radical 1 due to the phenyl group substitutions in 1.
C5
C6 + C7
C8 + C9
C10
C11
+6.9
+1.4
-1.9
+20.8
+19.9
+20.6
+20.7
+19.9
+20.4
+18.3
+16.0
+18.3
a ∆Etot(Becke3LYP/6-31G(d)//Becke3LYP/6-31G(d)). b ∆Etot(Becke3LYP/
6-31G(d)//Becke3LYP/6-31G(d) + ∆ZPE(Becke3LYP/6-31G(d)).
c ∆E (Becke3LYP/6-311+G(d,p)//Becke3LYP/6-31G(d)). d ∆E (Becke3LYP/
tot
tot
6-311+G(d,p)//Becke3LYP/6-31G(d) + ∆ZPE(Becke3LYP/6-31G(d)).
The concerted elimination through TS C2 is similar to that
found in smaller systems.8 Despite concertedness, the process
is highly asynchronous as C-O bond cleavage is essentially
complete in C2 while proton transfer from carbon to oxygen
has just begun (Figure S1 in Supporting Information). The
phosphate group in C2 carries a partial negative charge of -0.51
e excluding and of -0.16 e including the migrating hydrogen
atom, highlighting the polar character. The activation energy is
only 1.3 kcal/mol higher than that for the [1,2]-migration,
indicating a close competition between elimination and rear-
rangement. The elimination reaction is slightly exothermic and
thermodynamically favored over migration by 5.5 kcal/mol.
A transition state for homolytic C-O bond cleavage of C1
could not be located, a finding in line with earlier results for
elimination reactions of other electrophilic radicals.21 From the
total energy for homolysis (18.3 kcal/mol), this process is the
least favorable addressed. A weakly bound complex, C11,
between the phosphatoxy radical C8 and isobutene (C9) exists,
but the binding energy of 0.3 kcal/mol for C11 is so small that
differences in zero point vibrational energy are sufficient to yield
a nonbonding interaction. Starting from C11, transition state
C10 for hydrogen transfer from isobutene to the phosphatoxy
Three concerted reaction pathways for C1 were explored
(Figure 4), [1,3]-elimination through TS C2 to give allyl radical
C7, and [3,2]- and [1,2]-migrations through TS C3 and C4,
respectively, to give radical C5. Heterolytic fragmentation of
C1 was not addressed computationally in this study due to the
lack of a reliable solvent model, but a stepwise homolytic
fragmentation of C1 followed by hydrogen atom transfer was
studied. Energies of the stationary points are listed in Table 3.
In line with earlier studies of phosphatoxy rearrangements
in smaller systems,8 the barrier for the [1,2]-migration is lower
than that for the [3,2]-shift, although the additional methyl
groups in C1 compared to the nor-methyl analogue lower the
absolute reaction barriers significantly. Charge separation occurs
in both pathways with cumulative charges on the dimethyl
phosphate groups of -0.47 e in C3 and -0.42 e in C4 that are
significantly larger than that for the phosphate group in precursor
C1 (-0.33 e) and product C5 (-0.36 e). The product of
migration, primary radical C5, is less stable than the tertiary
radical C1 by 3.6 kcal/mol which agrees well with previously
estimated radical stabilization energies of approximately 2 kcal/
mol for each methyl group.20
(20) Viehe, H. G.; Janouske, Z.; Merenyi, R. Substituent Effects in
Organic Chemistry; Reidel: Dordrecht, 1986. Pasto, D. J.; Krasnansky,
R.; Zercher, C. J. Org. Chem. 1987, 52, 3062-3072.
(21) Engels, B.; Peyerimhoff, S. D. J. Phys. Chem. 1989, 93, 4462-
4470. Engels, B.; Peyerimhoff, S. D.; Skell, P. J. Phys. Chem. 1990, 94,
1267-1275.

â-Phosphatoxyalkyl Radical Reactions
J. Am. Chem. Soc., Vol. 121, No. 46, 1999 10691
radical was located, but C10 is more stable than complex C11
by 2.3 kcal/mol, again due to zero point energy corrections.
The partial negative charge of the phosphate group in C10 is
-0.33 e excluding the migrating hydrogen atom, the least
polarized reaction addressed, and TS C10 describes a typical
homolytic process.
through three-center and five-center transition states as exempli-
fied by C3 and C4 studied in this work. The caVeat, however,
is that most of the experimental work was performed in low-
polarity solvents, and thus the conclusions might not be
applicable to highly polar solvents.
In contrast to the above, the common view of elimination
reactions of â-ester radicals is that they proceed by initial
heterolysis to radical cations and anions, the model proposed
by Schulte-Frohlinde in the earliest kinetic studies of such
reactions.22,23 A considerable amount of circumstantial evidence
supports this view, including obvious polar characteristics in
the transition states of the reactions3 and product studies that
require formation of common intermediates from isomeric
reactants.24 An especially compelling case for heterolysis of an
R-oxy-â-phosphate radical to an enol ether radical cation in
methanol was reported by Rist, Giese, and co-workers who
observed CIDNP signals in the enol ether product.25 Neverthe-
less, much of the evidence supporting heterolysis does not
necessarily exclude other reactions; for example, few authors
have considered the possibility that the results might be
consistent with concerted elimination and nucleophilic substitu-
tion reactions. In addition, most of the elimination studies were
performed in high-polarity media.
KIEs were calculated for processes shown in Figure 4
according to eq 1
q
k
(∆Gd6q - ∆Gh6
)
h6 ) exp
(1)
{
}
kd6
RT
for reactant C1 containing CD3-substituents at the radical center.
Inverse secondary KIEs of 0.79 and 0.83, respectively, were
found for the [3,2]- and [1,2]-rearrangement reactions. The two
phosphate elimination pathways have different KIEs as expected.
The syn-1,3-elimination through C2 has a large normal KIE of
4.51, whereas homolytic C-O bond cleavage to give C8 and
C9 shows only a small inverse equilibrium isotope effect of
0.87. Under the condition that C-O bond fragmentation alone
is rate limiting, formation of the 1,3-elimination products
through a stepwise process will show only small isotope
effects.
The differences in solvent polarities for studies of migrations
and eliminations of â-ester radicals have left open the possibility
that two distinctly different reaction mechanisms are involved
for the two reactions. That explanation is in conflict with the
results of LFP kinetic studies of radical 1 as well as radical
1M, which reacts exclusively by phosphate migration to give
2M, and radical 13, which gives 14 by trifluoroacetate migration.
Insight into solvent effects on the reactions in Figure 4 was
provided from evaluation of the dipole moments in the ground
and transition states. Dipole moments of 0.88, 2.26, 2.31, and
2.64 D were calculated for structures C1, C2, C3, and C4 in
the gas phase, and the polarized transition states C2-C4 will
be stabilized relative to minimum C1 in polar solvents. To
quantify this expectation to some extent, we calculated the
solvation energies of structures C1-C4 using a simple Onsager
solvation model which provides an approximation of the
electrostatic component of the free energy of solvation. Solvent
effects on the reaction barriers calculated for benzene, THF,
acetonitrile, and water (Table S2 in Supporting Information)
are modest in all cases. The [3,2]-shift through TS C4 profits
most from the presence of the solvent, and elimination through
TS C2 profits least. Most notably, however, the relative ordering
of reaction barriers was not changed by solvent effects.
Because small concentrations of precursors are used in LFP
studies, a wide range of solvents is acceptable. Radical 1 reacted
by both migration and elimination reactions, and the two
pathways were competitive in solvents of either low or high
polarity. Consistent entropic factors (the log A term of the
Arrhenius functions in Table 1) were found in THF and
acetonitrile for radical 1, suggesting that the mechanism is not
altered by solvent polarity. A plot of the kinetics of reactions
of 1 in various solvents at 20 °C against the ET(30) solvent
polarity scale26 (Figure 5) strongly reinforces this conclusion;
a polarized transition state for a reaction pathway unaltered by
solvent polarity changes is implicated by the well-correlated
regression fit.27 The trifluoroacetate migration in radical 13 also
has similar log A terms in both THF and acetonitrile,7 and, for
radical 1M, the entropies of activation were essentially constant
in benzene, THF, and acetonitrile.6,7,28 An important point is
Discussion
â-Phosphatoxyalkyl radical reactions are a subset of â-ester
radical reactions, a diverse collection that has received consider-
able attention due to their synthetic utility and the biological
significance of the â-phosphate DNA radicals.2 A comprehen-
sive recent review of the general subject of â-ester radicals
provides details of several mechanistic and computational studies
of these reactions,3 and some trends are apparent. One important
generalization is that the reactions of â-acyloxyalkyl (â-
carboxylate) radicals and â-phosphatoxyalkyl (â-phosphate)
radicals are likely to be mechanistically related. Other gener-
alizations concerning migrations and eliminations of â-ester
radicals have emerged.
In the case of ester migration reactions, labeling studies and
studies aimed at demonstrations of kinetic competence indicate
that the migrations cannot involve diffusively free intermediates,
either from heterolytic or homolytic cleavages, that recombine,
nor can they proceed by initial 5-endo cyclizations followed
by ring openings. Although fragmentations (either heterolytic
or homolytic) followed by cage recombinations are not excluded
by previous results, the conclusions based on labeling and
computational work are that migrations occur mainly by a
combination of open-shell concerted reactions, proceeding
(22) Behrens, G.; Koltzenbrug, G.; Ritter, A.; Schulte-Frohlinde, D. Int.
J. Radiat. Biol. 1978, 33, 163-171.
(23) Koltzenburg, G.; Behrens, G.; Schulte-Frohlinde, D. J. Am. Chem.
Soc. 1982, 104, 7311-7312.
(24) Peukert, S.; Giese, B. Tetrahedron Lett. 1996, 37, 4365-4368.
(25) Gugger, A.; Batra, R.; Rzadek, P.; Rist, G.; Giese, B. J. Am. Chem.
Soc. 1997, 119, 8740-8741.
(26) Reichardt, C. Chem. ReV. 1994, 94, 2319-2358. Skwierczynski,
R. D.; Connors, K. A. J. Chem. Soc., Perkin Trans. 2 1994, 467-472.
(27) Using single weighting for each solvent or mixture of solvents and
excluding the results in benzene, the slope of the regression line is 0.128
( 0.003, and r2 ) 0.998.
(28) Horner, J. H.; Newcomb, M., unpublished results.

10692 J. Am. Chem. Soc., Vol. 121, No. 46, 1999
Newcomb et al.
Scheme 4
faster than 1 and which gives only the benzylic radical product
2M that has no appreciable absorbance at wavelengths greater
than 330 nm. The detection of 4 in the reactions of radical 1 in
water/acetonitrile mixtures obviously supports a heterolysis
pathway, but it is the observed kinetic behavior in aqueous
acetonitrile solutions that provides a more compelling argument
for heterolysis. In these solutions, a common intermediate for
diffusively free radical cation 4 and allylic radical 3 is implicated
by the agreement in the rate constants for formation of both
species (Table 2).
Figure 5. Measured rate constants for reactions of radical 1 in the
20-22 °C range as a function of the ET(30) solvent polarity terms.
Multiple values in THF and in acetonitrile overlap one another. The
limiting value for the rate constant for reaction in benzene was excluded
from the regression line.
that the log A terms for phosphate migration in radical 1M and
trifluoroacetate migration in 13 are quite similar to those for
the predominant phosphate elimination in radical 1. From these
behaviors, we infer that the migration and elimination reactions
must be quite closely related. If this deduction is correct, the
mechanistic question reduces to which pathway is involved, a
common intermediate formed heterolytically or similar concerted
processes.
The agreement between the relative barrier heights for
migration and elimination found computationally for radical C1
and the observation that the two processes are competitive in
reactions of radical 1 might suggest that 1 reacts by concerted
processes. However, the kinetic data for reactions of 1-d0 and
1-d14 cannot be reconciled with such a conclusion. The overall
rate constants for the reactions of these two radicals are
indistinguishable, and the LFP measurements of the first-order
processes are precise enough such that the differences in the
rate constants for the two radicals cannot be greater than 10%.
The large primary KIE found computationally for elimination
in radical C1 might be attenuated somewhat in radical 1 due to
an earlier transition state, but it will not be lost completely.
The inverse secondary KIE computed for concerted migrations
of C1 translates into an acceleration of the migration reaction
of the deuterated radical, but this effect is much smaller than
the primary KIE for the elimination and also should be
attenuated in radical 1 somewhat due to an early transition
state.
The observation of an isotope effect in the product ratios 2:3
combined with the lack of KIE in the overall rate constants is
even more damaging to the position that the reactions are
concerted because it demonstrates that transfer of a hydrogen
atom or proton must occur after the rate-limiting step of the
reaction. This is apparent in the spectra shown in A and B of
Figure 3 for reactions of 1-d0 and 1-d14, respectively, in
acetonitrile. These spectra display signal evolution over the same
amount of time for reactions that had the same overall rate
constants, but it is clear that allylic radical 3 was formed from
1-d14 less rapidly than from 1-d0.
We conclude that radical 1 gives products 2 and 3 from a
common intermediate that is produced in the rate-limiting step.
The initial process must be a fragmentation, and the fragmenta-
tion apparently is heterolytic to account for the kinetic solvent
effects found here and observed in other â-ester radical
reactions.3
The reactions of radical 1 can be cast in a model that is the
same as that for heterolysis of a closed-shell species reacting
in a solvolysis reaction (Scheme 4). The first-formed species is
a contact ion pair (CIP) which can react in various modes or
proceed to a solvent-separated ion pair (SSIP). In turn, the SSIP
can collapse back to the CIP or separate further to give the
diffusively free ions, and some reactions, such as electron
transfer, might occur in the SSIP. Sprecher proposed that such
a scheme was consistent with the reaction diversity in an
overview of â-ester radical reactions.29
The occurrence of two types of ion pairs is not required to
explain the kinetics of radical 1. The reaction pathway in
aqueous acetonitrile could involve a single intermediate that
partitions, and failure of that intermediate to partition in less
polar media would explain the absence of radical cation
formation. Nevertheless, contact pairs involving radical cations
apparently behave as encounter complexes that evolve to
solvent-separated pairs with diffusion-like rate constants, on the
order of 109-1010 s-1 16,30
If the CIP from radical 1 solvates
.
this rapidly, then another ion pair (the SSIP) provides a reservoir
for the radical cation that can partition back to the CIP in
competition with dissociation to give free ions. The ratio of
rate constants for dissociation of the SSIP and collapse back to
the CIP is likely to be highly sensitive to solvent effects with
relatively small rate constants for dissociation in nonpolar media
that preclude formation of diffusively free ions in detectable
amounts.
The dynamics of some of the processes occurring in the CIP
can be deduced in a semiquantitative sense. For example, fast
proton transfer within a CIP containing 4 must have occurred
in the photoejection experiments conducted in this work.
Photoejection produces the CIP of the styrene radical cation
and a solvated electron, in this case the radical anion of
acetonitrile. Despite the favorable energetics for back electron
transfer to return the neutral styrene, which is likely to be a
major reaction, some escape from the CIP occurred with the
â,â-dimethylstyrene radical cation to give the diffusively free
radical cation 4M that was detected. When a similar experiment
was attempted with â,â-dibenzylstyrene, however, the major
product observed “instantly” on the nanosecond time scale
apparently was allylic radical 3 instead of radical cation 4.
Therefore, the deprotonation of 4 by the acetonitrile radical anion
Radical cation 4 was not produced in nonreactive solvents
where it could have been observed easily. In a similar manner,
no radical cation was detected in THF or acetonitrile reactions
of radical 1M which reacts more than an order of magnitude
(29) Sprecher, M. Chemtracts: Org. Chem. 1994, 7, 115-119.
(30) Arnold, B. R.; Noukakis, D.; Farid, S.; Goodman, J. L.; Gould, I.
R. J. Am. Chem. Soc. 1995, 117, 4399-4400.

â-Phosphatoxyalkyl Radical Reactions
J. Am. Chem. Soc., Vol. 121, No. 46, 1999 10693
apparently was faster than the overall escape process in this
experiment.
Whether or not the relatively low-energy concerted processes
found computationally for radical C1 in this work and previously
for similar â-phosphate alkyl radicals8 are important when such
stable radical cations cannot be formed remains an open
question. That question probably will not be resolved by the
methods used in this work because diffusively free radical
cations apparently are not formed in reactions of simple â-ester
radicals even in water.23
Deprotonation of 4 by the phosphate anion to give 3 and
collapse of the ions to give 2 within the CIP formed from radical
1 also must to be quite fast.31 The kinetic model for reactions
of 1 in aqueous acetonitrile indicated that the major reaction
pathway was direct formation of allylic radical 3, and the
partitioning appeared to be relatively constant in water solutions
of 25, 50, and 67% with the direct reaction amounting to 60-
70% of the total reaction. If evolution of the CIP to the SSIP is
nearly diffusion-limited, the constant partitioning ratio with
increasing water content would require either that deprotonation
and collapse were competitive with SSIP formation or that the
rate of separation of the SSIP to free ions was not sensitive to
solvent polarity. The latter option is inconsistent with the
absence of free ions when acetonitrile and less polar solvents
were employed.
Experimental Section
Syntheses: described in the Supporting Information.
Laser Flash Photolysis. LFP studies were performed with an
Applied Photophysics LK-50 kinetic spectrometer using a Nd:YAG
laser for 266 nm (photoejection studies) or 355 nm (PTOC ester and
chloranil studies). All studies were performed with dilute, He-sparged
(unless noted) flowing solutions of precursor. Temperatures were
controlled by circulating a bath solution through the jacket of the storage
funnel, and temperatures were measured with a thermocouple placed
in the flowing stream immediately above the irradiation region of the
flow cell. For reactions in acetonitrile/water mixtures, syringe drive
pumps were used with gas-tight syringes containing a solution of PTOC
ester precursor in acetonitrile in one syringe and water or water/
acetonitrile in a second syringe; the solutions were mixed in a tee located
immediately before the flow cell, and LFP experiments were performed
within 5 s of mixing.
Kinetic Data Analyses. For kinetic runs, several traces typically
were averaged to improve signal-to-noise. First-order analyses were
performed with the Applied Photophysics software. Double exponential
analyses and analyses for the complex kinetic scheme in Scheme 3
were performed with Sigmaplot software. For the kinetic analysis of
Scheme 3, eq 2 was fit
It might be instructive to note why we originally believed6
that the results with radical 1 indicated that the reactions
involved concerted processes. A radical cation intermediate was
not observed in the original work,19 thus obviating any
demonstration of kinetic competence. The negative entropies
of activation for reaction of 1 and related radicals appear to be
inconsistent with an initial heterolysis reaction; we can only
rationalize these at this time as resulting from highly organized
transition states for heterolysis, such as shown in 15, with several
orbitals in the incipient radical cation co-aligned and the ester
group in a stereoelectronically favorable orientation. Finally,
the high degree of stereochemical integrity in phosphate
migrations of radicals similar to radical 15,32 suggested concerted
mechanisms, although the results can be accommodated in the
context of ion pair formation if collapse is exceedingly fast.
Abs ) p1 + p2 exp(-k1t) + (p3/(k2 - k1))((k2(1 - exp(-k1t))) -
(k1(1 - exp(-k2t)))) (2)
where pn are variable absorbance terms, kn are variable rate constants,
and Abs is the observed absorbance at time t. The ratio p2/p3 from eq
2 gives the ratio kA/kB in Scheme 3, assuming that a common species
is formed from both pathways.
Computational details. Geometry optimizations were performed at
the UHF/3-21G(d) level of theory and with the hybrid UBecke3LYP/
6-31G(d) density functional, or B3LYP/6-31G(d), method35 as imple-
mented in Gaussian 94.36 A large number of conformers were optimized
for radicals C1 and C5 at the UHF/3-21G(d) level of theory. The three
most favorable conformers of each radical were then reoptimized at
the B3LYP/6-31G(d) level. All stationary points were characterized
through calculation of analytical second derivatives. Improved relative
energies for the stationary points characterized at this level of theory
were then calculated using the larger 6-311+G(d,p) basis set at the
UBecke3LYP level of theory. Thermochemical quantities were calcu-
lated from the unscaled harmonic vibrational frequencies at 293.15 K
and are composed of relative energies at the Becke3LYP/6-311+G-
(d,p)//Becke3LYP/6-31G(d) level of theory and thermochemical cor-
rections derived from the Becke3LYP/6-31G(d) frequencies. KIEs were
calculated from free energy differences between isotopomers. As in
earlier studies,8 partial charges were obtained from the Mulliken
population analysis of the Becke3LYP/6-31G(d) Kohn-Sham orbitals.
Solvent effects were estimated with a simple Onsager reaction field
The CIDNP study of phosphate elimination to give an enol
ether radical cation in methanol25 and the results of this work
provide direct spectroscopic confirmation of heterolysis reactions
in polar media, and the present results indicate that the
mechanism of reaction of radical 1 is the same in less polar
media. Accordingly, heterolysis appears to be the common
pathway for reactions of â-phosphatoxyalkyl radicals when the
incipient radical cation is at least as stable as those from enol
ethers and styrenes.33 Studies aimed at further characterization
of these reactions might focus on the putative ion pairs, the
reactions of which should be highly solvent-dependent.34
(31) An alternative to two-electron processes within the CIP of radical
cation 4 and phosphate anion is possible. Electron transfer in the CIP of 4
and diphenyl phosphate anion would give the styrene and a phosphatoxyl
radical which will be quite reactive. For example, our computational results
indicated that the transition state for H-atom transfer from isobutene to the
dimethylphosphatoxyl radical was lower in energy than the weakly bound
ensemble.
(32) Crich, D.; Escalante, J.; Jiao, X. Y. J. Chem. Soc., Perkin Trans. 2
1997, 627-630.
(33) Ethyl vinyl ether and styrene have the same oxidation potentials in
acetonitrile. See: Katz, M.; Riemenschneider, P.; Wendt, H. Electrochim.
Acta 1972, 17, 1595-1607.
(34) Following the submission of this work, we found a â-phosphatoxy-
alkyl radical that reacts by heterolytic cleavage resulting in phosphate
migration and/or formation of diffusively free radical cation as a function
of solvent polarity. See: Whitted, P. O.; Horner, J. H.; Newcomb, M.;
Huang, X.; Crich, D. Org. Lett. 1999, 1, 153-156.
(35) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652. Lee, C.; Yang,
W.; Parr, R. G. Phys. ReV. B 1988, 37, 785-789. Hertwig, R. H.; Koch,
W. J. Comput. Chem. 1995, 16, 576-585.
(36) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;
Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T.; Petersson, G.
A.; Montgomery, J. A.; Raghavachari, K.; Al-Laham, M. A.; Zakrzewski,
V. G.; Ortiz, J. V.; Foresman, J. B.; Cioslowski, J.; Stefanov, B. B.;
Nanayakkara, A.; Challacombe, M.; Peng, C. Y.; Ayala, P. Y.; Chen, W.;
Wong, M. W.; Andres, J. L.; Replogle, E. S.; Gomperts, R.; Martin, R. L.;
Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart, J. P.; Head-
Gordon, M.; Gonzalez, C.; Pople, J. A. Gaussian 94, revision E.2; Gaussian,
Inc.: Pittsburgh, PA, 1995.

10694 J. Am. Chem. Soc., Vol. 121, No. 46, 1999
Newcomb et al.
model using a spherical cavity.37 The dielectric constants used were as
follows: benzene (2.27), THF (7.58), acetonitrile (35.9), and water
(78.3).
Rechenzentrum Mu¨nchen through computational resources. We
thank Dr. S. -Y. Choi for experimental assistance.
Supporting Information Available: Kinetic results for
reactions of 1-d0 and 1-d14 in low polarity solvents, computed
solvent effects, structures for ground and transition states for
radical C1, and details of the synthetic reactions (PDF). This
material is available free of charge via the Internet at
http://pubs.acs.org.
Acknowledgment. We thank the National Institutes of
Health (GM56511 to M.N. and CA60500 to D.C.), the National
Science Foundation (CHE9625256 to D.C. and CHE9614968
to M.N.), and the Volkswagen-Stiftung (to H.Z.) for financial
support. The work in Germany was supported by the Leibniz-
(37) Wong, M. W.; Frisch, M. J.; Wiberg, K. B. J. Am. Chem. Soc. 1991,
113, 4776-4782.
JA991012R