Kinetics of reductive N-O bond fragmentation: The role of a conical intersection

Article abstract of DOI:10.1021/ja020768e

N-alkoxyheterocycles can act as powerful one-electron acceptors in photochemical electrontransfer reactions. One-electron reduction of these species results in formation of a radical that undergoes N-O bond fragmentation to form an alkoxy radical and a neutral heterocycle. The kinetics of this N-O bond fragmentation reaction have been determined for a series of radicals with varying substituents and extents of delocalization. Rate constants varying over 7 orders of magnitude are obtained. A reaction potential energy surface is described that involves avoidance of a conical intersection. A molecular basis for the variation of the reaction rate constant with radical structure is given in terms of the relationship between the energies of the important molecular orbitals and the reaction potential energy surface. Ab initio and density functional electronic structure calculations provide support for the proposed reaction energy surface.

Full text of DOI:10.1021/ja020768e

Published on Web 11/28/2002
Kinetics of Reductive N-O Bond Fragmentation: The Role of
a Conical Intersection
Edward D. Lorance, Wolfgang H. Kramer, and Ian R. Gould*
Contribution from the Department of Chemistry and Biochemistry, Arizona State UniVersity,
Box 871604, Tempe, Arizona 85287-1604
Received June 3, 2002. Revised Manuscript Received September 13, 2002
Abstract: N-alkoxyheterocycles can act as powerful one-electron acceptors in photochemical electron-
transfer reactions. One-electron reduction of these species results in formation of a radical that undergoes
N-O bond fragmentation to form an alkoxy radical and a neutral heterocycle. The kinetics of this N-O
bond fragmentation reaction have been determined for a series of radicals with varying substituents and
extents of delocalization. Rate constants varying over 7 orders of magnitude are obtained. A reaction potential
energy surface is described that involves avoidance of a conical intersection. A molecular basis for the
variation of the reaction rate constant with radical structure is given in terms of the relationship between
the energies of the important molecular orbitals and the reaction potential energy surface. Ab initio and
density functional electronic structure calculations provide support for the proposed reaction energy surface.
Introduction
redox potentials and the energy of the excited state. For the
case of photochemical reduction of a fragmentable X-Y
molecule using an excited-state electron donor sensitizer
molecule, S*, eq 2a, ∆Get is given by eq 2b.4 Here, EexS* is the
energy of the excited-state sensitizer, Eox is the oxidation
potential of the sensitizer, and EredX-Y is the reduction potential
of the X-Y molecule.
Studies of single-electron transfer-initiated bond fragmenta-
tion reactions have been an important theme in physical organic
chemistry in recent years.1 Kinetic studies of both radical anions
formed upon reduction and radical cations formed upon oxida-
tion have resulted in the development of several new and
important mechanistic principles.1 Photoinduced single-electron-
transfer-induced bond fragmentation reactions also find many
practical applications, particularly in the area of imaging
technology.2 For example, fragmentation of a radical cation
results in the formation of a radical (Y•) and an electrophile
(X+), eq 1a. If X+ is a proton, H+, then eq 1 forms the basis of
a photoacid system.3 Correspondingly, fragmentation of a radical
anion can form a radical (Y•) and a nucleophile or base, (X-),
eq 1b. The radicals formed in such processes have been used
in photoinitiated polymerization systems.2
S
S* + X-Y f S•+ + X-Y•-
∆Get ) (EoxS + EredX-Y) - EexS* + C
(2a)
(2b)
C is a term that corrects for Coulombic interactions in the
product geminate pair and is often negligible in polar solvents.4
To be broadly useful in technological applications, photochemi-
cally induced fragmentation of an X-Y molecule should occur
with as wide a range of excitation wavelengths as possible, that
is, the reaction should proceed with as small a sensitizer
X-Y -8 X-Y•+ f X+ Y•
(1a)
(1b)
-e
excitation energy Eex as possible.5 For the example given in
S*
X-Y -8 X-Y•- f X- Y•
+e
eq 2b, this means that the X-Y molecule should be easy to
X-Y
reduce (Ered
should be as small a negative number as
The energetics of photoinduced single-electron-transfer reac-
possible). One obvious way to achieve facile reduction is to
use a positively charged X-Y molecule (or a negatively charged
molecule in a corresponding oxidation reaction). Indeed, the
two technological applications of photoinduced bond fragmenta-
tion that have been demonstrated over the widest range of
excitation wavelengths use this approach. Photochemical oxida-
tion of borate salts has been studied extensively by Schuster et
al.5b,6 One-electron oxidation of an alkyltriarylborate (-BRAr3)
results in formation of a boranyl radical, •BRAr3, which
tions, ∆Get, are usually described in terms of the appropriate
* To whom correspondence should be addressed. E-mail: igould@asu.edu.
(1) (a) Baciocchi, E.; Bietti, M.; Lanzalunga, O. Acc. Chem. Res. 2000, 33,
243. (b) Gaillard, E. R.; Whitten, D. G. Acc. Chem. Res. 1996, 29, 292. (c)
Albini, A.; Fasani, E.; Freccero, M. AdV. Electron Transfer Chem. 1996,
5, 103. (d) Saveant, J.-M. AdV. Electron Transfer Chem. 1994, 4, 53. (e)
Maslak, P. Top. Curr. Chem. 1993, 168, 1. (f) Popielarz, R.; Arnold, D. R.
J. Am. Chem. Soc. 1990, 112, 3068. (g) Dinnocenzo, J. P.; Farid, S.;
Goodman, J. L.; Gould, I. R.; Todd, W. P.; Mattes, S. L. J. Am. Chem.
Soc. 1989, 111, 8973. (h) Mattes, S. L.; Farid, S. Org. Photochem. 1983,
6, 233.
(2) See, for example: (a) Paczkowski, J.; Neckers, D. C. In Electron Transfer
in Chemistry; Balzani, V., Ed.; Wiley-VCH: New York, 2001; Vol. 5, p
516. (b) Reiser, A. PhotoreactiVe Polymers, the Science and Technology
of Resists; Wiley: New York, 1989.
(4) Weller, A. Z. Phys. Chem. (Munich) 1982, 130, 129.
(5) (a) Gould, I. R.; Shukla, D.; Giesen, D.; Farid, S. HelV. Chim. Acta, 2001,
84, 2796. (b) Chatterjee, S.; Gottschalk, P.; Davis, P. D.; Schuster, G. B.
J. Am. Chem. Soc. 1988, 110, 2326.
(3) See, for example: (a) Saeva, F. D. AdV. Electron Transfer Chem. 1994, 4,
1. (b) DeVoe, R. J.; Olofson, P. M.; Sahyun, M. R. V. AdV. Photochem.
1992, 17, 313. (c) Crivello, J. V. AdV. Polym. Sci. 1984, 62, 1.
(6) Schuster, G. B. AdV. Electron Transfer Chem. 1991, 1, 163.
9
10.1021/ja020768e CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. 2002, 124, 15225-15238
15225

A R T I C L E S
Lorance et al.
Scheme 1
Scheme 2
from an X-Y+ compound that is easier to reduce may have a
smaller value of kfr.
One obvious question therefore relates to the extent to which
the kinetics of bond fragmentation in species such as X-Y• in
Scheme 2 actually vary with the reduction potential of the
precursor X-Y+ and what other factors influence the rates of
these reactions. In the literature, the rate constants for cleavage
of two radicals formed upon reduction of N-alkoxyheterocycles
have been reported, eq 4.11 Both can be considered to be
cleavage reactions of N-alkoxypyridyl radicals with electron-
withdrawing substituents, p-(N-ethoxypyridyl) in 4a11a and
p-cyano in 4b.11b As such, both radical precursors are relatively
easy to reduce and might be considered to be equally useful
with a wide range of sensitizers. However, the rate constants
of the two radical cleavage reactions differ by an astonishing 6
orders of magnitude, eq 4.
efficiently fragments to an alkyl radical and a triarylborane, eq
3. Alkyl radicals generated this way using visible light sensitiza-
tion have been used to initiate radical polymerization reactions
in a commercial technological application.7
-BRAr3 -8 •BRAr3 f R• + BAr3
(3)
-e
The radical initiation system that has been demonstrated with
the widest range of wavelengths involves reduction of an
N-methoxypyridinium compound, Scheme 1.5a,8,9 Exothermic
electron transfer, ket, from an excited sensitizer results in
formation of the sensitizer radical cation, S•+, and a pyridyl
radical. Fragmentation of the pyridyl radical, kfr, produces a
methoxy radical that initiates radical polymerization reactions.5a,8
Sensitizers with absorption maxima even greater than 750 nm
have been shown to be useful in this process.5a The reactions
proceed with reasonable photon efficiency, which means that
fragmentation of the pyridyl radical is rapid enough to compete
favorably with the energy-wasting return electron transfer
process in the geminate pair (k-et, Scheme 1).5a In general,
identification of large values for kfr is a common goal in studies
of photoinduced fragmentation reactions.1 For technological
applications, large values of kfr are essential for high photon
efficiency.5
Although facile reduction of X-Y is beneficial so that the
energy of the excited state can be as low as possible (eq 2b),
the thermodynamic favorability of bond fragmentation decreases
with increasing ease of reduction of X-Y. For the case of
reduction of a positively charged molecule (X-Y+) to a radical
(X-Y•) that undergoes fragmentation, the thermodynamic cycle
of Scheme 2 can be given.5a,10 Scheme 2 shows that the bond
dissociation energy for the radical X-Y•, BDE (X-Y•), will
increase (i.e., fragmentation will become less favorable) as the
reduction potential of the X-Y+ molecule (EredX-Y+) becomes
less negative, that is, as X-Y+ becomes easier to reduce. Purely
the basis of on thermodynamic arguments, the radical derived
The fragmentation reaction in 4b occurs with such a sufficiently
high rate constant that it would be expected to compete usefully
with return electron transfer in a sensitization process such as
that illustrated in Scheme 1. The reaction in eq 4a is sufficiently
slow to be essentially useless in such a reaction scheme.
Obviously, the “substituent” in eq 4a is a somewhat stronger
withdrawing group than that in eq 4b, and reduction of the
pyridinium precursor in the latter case will thus be somewhat
more difficult than in the former. According to the thermo-
dynamic arguments given above, it is understandable that the
reaction in eq 4a is slower than that in eq 4b. However, it is
also clear that the radical in eq 4a is more delocalized and
resonance stabilized than that in eq 4b. The relative importance
of these and other possible factors in determining the very large
difference in kfr values in this case is not clear.
To determine the factors that control the rate constants for
fragmentation in such radicals, the relative importance of
electron-withdrawing and delocalizing substituents needs to be
further explored. Here we describe an extensive study of the
kinetics of fragmentation of a series of N-methoxyheterocyclic
radicals, with varying substituents and extents of delocalization.
One goal of this work is to identify single-electron-transfer-
induced fragmentation reactions that occur with very high rate
constants, for the reasons discussed above. A further goal is to
develop a detailed description of the factors that control the
rate constants for fragmentation in radicals such as these and
(7) Rastogi, A. K.; Wright, R. F. Proc. SPIE-Int. Soc. Opt. Eng. 1989, 1079,
183.
(8) Farid, S. Y.; Haley, N. F.; Moody, R. E.; Specht, D. P. U.S. Patents
4,743,528, 4,743,529, 4,743,530, and 4,743,531, 1988.
(9) (a) Schnabel, W. Macromol. Rapid Commun. 2000, 21, 628. (b) Zhu, Q.
Q.; Schnabel, W. Eur. Polym. J. 1997, 3, 1325. (c) Yagci, Y.; Schnabel,
W. Macromol. Symp. 1994, 85, 115. (d) Kayaman, N.; Onen, A.; Yagci,
Y.; Schnabel, W. Polym. Bull. (Berlin) 1994, 32, 589.
(10) (a) Popielarz, R.; Arnold, D. R. J. Am. Chem. Soc. 1990, 112, 3068. (b)
Maslak, P.; Vallombroso, T. M.; Chapman, W. H., Jr.; Narvaez, J. N.
Angew. Chem., Int. Ed. Engl. 1994, 33, 73.
(11) (a) Wolfle, I.; Lodaya, J.; Sauerwein, B.; Schuster, G. B. J. Am. Chem.
Soc. 1992, 114, 9304. (b) Bockman, T. M.; Lee, K. Y.; Kochi, J. K. J.
Chem. Soc., Perkin Trans. 2 1992, 9, 1581.
9
15226 J. AM. CHEM. SOC. VOL. 124, NO. 51, 2002

Kinetics of Reductive N−O Bond Fragmentation
A R T I C L E S
Scheme 3
Scheme 5
Table 1. Arrhenius Activation Parameters for Fragmentation of
Radicals Formed by Reduction of N-Methoxyheterocyclesa
Scheme 4
a See Experimental Section for estimates of errors. b In 1,2-dichloroethane
solvent. c In acetonitrile solvent.
ence of ca. 1 M biphenyl (BP) results in electron-transfer
quenching and efficient formation within ca. 5 ns of the DCA
radical anion (DCA•-) and the biphenyl radical cation (BP•+)
in solution.13a In the presence of ca. 10-2 M 12 and 10-2
M
benzyltrimethylsilane (BTS), within another ca. 10 ns, the
DCA•- reduces 12 to the corresponding radical, and the BP•+
oxidizes the BTS to its radical cation (BTS•+). Within a few
nanoseconds of its formation, the BTS•+ fragments to form a
benzyl radical (Bz•), which absorbs only in the UV region.1g
Thus, within ca. 15 ns of the laser pulse, the only visible
absorbing transient species should be the radical derived from
12. Under these conditions, a transient species is observed with
an absorption maximum at ca. 575 nm. This transient is assigned
to the radical on the basis of the fact that it is formed at the
same rate at which the DCA•- decays and that a very similar
absorption is observed with the corresponding N-ethyl com-
pound 12E. Importantly, the absorption from 12E is relatively
long-lived (the first half-life is greater than 10 µs) and decays
by a mixture of first- and second-order processes, whereas that
from 12 decays in a rapid first-order process, consistent with
N-O bond cleavage. Measurements as a function of temperature
in acetonitrile allow a room-temperature rate constant for
fragmentation of 1.0 × 108 s-1 to be determined (Table 1).
For compound 2, electron transfer from the 9-thioxanthenone
triplet was used to generate the reduced form, in a manner
similar to that previously reported for the corresponding diethoxy
compound.11a For both 2 and 12, experiments were performed
as a function of temperature (Table 1).
other one-electron-reduced species. Specifically, a molecular
basis is sought for the relative rate effects predicted by
thermodynamic cycles such as those in Scheme 2. A description
of the radical reactions that involves the avoidance of a conical
intersection is given.12 A relationship between the energies of
the important molecular orbitals and the reaction potential energy
surface is described that is supported by density functional
electronic structure calculations.
Results and Discussion
Molecular Structures. The structures summarized in Scheme
3 have been investigated. Structures 1, 4-7, and 10 exhibit
differing degrees of delocalization. Structure 3 has a simple
electron-withdrawing group (ignoring the delocalization in the
cyano group), and 2, 8, 9, 11-13 include both delocalizing and
electron-withdrawing effects.
The N-ethyl-substituted heterocycles summarized in Scheme
4 were also used in the kinetic measurements, see further below.
Kinetic Measurements. Two methods were used to measure
the rate constants of fragmentation of the radicals. For those
with rate constants smaller than ca. 1 × 108 s-1, the radicals
were generated in diffusive second-order reduction reactions,
and the decay of the radicals was observed directly as a function
of time in a conventional nanosecond transient absorption
experiment.
For the radicals that fragmented faster than ca. 1 × 108 s-1
,
diffusive reduction could not be used since the rates of formation
of the reactive radicals would in most cases be slower than the
rates of their fragmentations. In these cases, a method involving
excitation of a charge-transfer complex was used to determine
the fragmentation rate constants.11b,14 Excitation of a suitable
CT complex forms a geminate pair containing the reactive
radical. The rate constant for fragmentation can be extracted
For compound 12, the radical was generated using the DCA/
biphenyl cosensitization system in acetonitrile solvent, Scheme
5.13a Excitation of 9,10-dicyanoanthracene (DCA) in the pres-
(12) (a) Yarkony, D. A. Acc. Chem. Res. 1998, 31, 511. (b) Yarkony, D. R. J.
Phys. Chem. 2001, 105, 6277.
(13) (a) Gould, I. R.; Ege, D.; Moser, J. E.; Farid, S. J. Am. Chem. Soc. 1990,
112, 4290. (b) Lewis, F. D.; Dykstra, R. E.; Elbert, J. E.; Gould, I. R.;
Farid, S. J. Am. Chem. Soc. 1990, 112, 8055.
(14) (a) Bockman, T. M.; Hubig, S. M.; Kochi, J. K. J. Am. Chem. Soc. 1998,
120, 6542. (b) Bockman, T. M.; Hubig, S. M.; Kochi, J. K. J. Org. Chem.
1997, 62, 2210.
9
J. AM. CHEM. SOC. VOL. 124, NO. 51, 2002 15227

A R T I C L E S
Lorance et al.
Scheme 6
from the overall dynamics of the geminate pair. The method
used here is similar to one described previously by Kochi et
al.,11b,14 except that an additional process of the geminate pair
is included in the kinetic analysis, that is, diffusive separation
of the partners. This method has not been described previously
and thus is discussed in some detail here.
The N-methoxy- and N-ethyl-substituted compounds of
Schemes 3 and 4, X-Y+, are generally strong one-electron
acceptors. Addition of a suitable electron donor compound, D,
to an X-Y+ solution results in reversible formation of a charge-
transfer (CT) complex (D X-Y+, Scheme 6), characterized by
a broad CT electronic absorption tailing into the visible
region.11b,14,15 Excitation into this CT absorption band results
in formation of the first excited state of the CT complex, that
is, a geminate radical cation/radical pair (D•+ X-Y•, Scheme
6). The geminate pair can undergo three processes, that is, return
electron transfer to regenerate the ground-state complex (D
X-Y+, k-etM),15b separation to form a separated radical cation
and radical (D•+ + X-Y•, ksep),15c and fragmentation of the
N-O bond to form a methoxy radical and the neutral hetero-
cycle (X Y•, kfr). The X-Y• pair that separates will also
fragment, but this reaction is not observed in the geminate pair.
The donor D is chosen so that the time scale for return
electron transfer is as similar to that of fragmentation as possible.
Accurate extraction of k-et and kfr from the measured rate
constant for decay of the geminate pair is only possible when
these two processes are competitive (see further below). With
the exception of two cases described separately below, three
donors have been used for all of the experiments, that is,
p-trianisyamine (TAA), stilbene (S), and biphenyl (BP). Return
electron transfer occurs in the Marcus inverted region for the
systems studied here,13 and k-et increases with decreasing
oxidation potential of the donor.13 The oxidation potential
decreases in the order BP to S to TAA.13 Hence, the slower
fragmenting radicals were generally measured using BP as the
donor, and the fastest was measured using TAA.
Figure 1. Relative absorbance spectra (arbitrary absolute units) observed
for excitation of the charge-transfer complex formed between N-methoxy-
phenanthridinium tetrafluoroborate (6) and stilbene in butyronitrile at room
temperature, showing decay of the geminate pair due to return electron
transfer. The spectra were taken at ca. 0, 25, 50, 100, and 400 ps after the
excitation pulse. (Inset) Decay of the absorbance integral from 482 to 491
nm as a function of time (A(t), filled circles). The larger open circles
represent the times at which the spectra were acquired. The smooth line
through the data shows the best fit to eq 9, using the parameters shown.
Figure 2. (Top panel) Normalized absorbance decays as a function of time,
A(t), for geminate pairs of N-methoxy-2-styrylpyridyl (open circles) and
N-ethyl-2-styrylpyridyl radicals (closed circles) with stilbene as the donor
in butyronitrile at room temperature. The curves through the data represent
best fits to eq 9 with the parameters shown. The three lower panels show
residual plots for fits of the N-methoxy radical data to (1 Exp) single-
exponential kinetics using eq 9 with the parameters shown, (FL) the Fick’s
law diffusion model of eq 11 with 1.87, 18.78, and 303 for κ, λ, and τ,
respectively, and (2 Exp) double-exponential kinetics, eq 12, with 0.11,
1.67 × 1010 s-1, 0.32, 1.76 × 109 s-1 and 0.57 for AF′, kF′, AS′, kS′ and
The donor D is also chosen so that D•+ has a characteristic
and strong electronic absorption band in the visible region. The
decay kinetics of the geminate pair can then be monitored via
the absorptions of D•+. The absorption of D•+ initially decays
due to return electron transfer and reaches a persistent value
(on the picosecond time scale) that corresponds to the number
of the geminate pairs that either separate or undergo fragmenta-
tion. Typical data are illustrated in Figures 1 and 2, for excitation
of the CT complexes formed between N-methoxyphenan-
thridinium tetrafluoroborate, 6 (Figure 1), N-methoxy-2-styryl-
pyridinium tetrafluoroborate, 7 and N-ethyl-2-styrylpyridinium
Φions, respectively.
hexafluorophosphate, 7E (Figure 2), as acceptors, and stilbene
as the donor, in butyronitrile at room temperature. The spectra
in Figure 1 are of the geminate stilbene radical cation/
phenanthridyl radical geminate pair. It is well-known that the
absorption spectra of such geminate pairs can be very well
described as the sum of the absorptions of the two partners (see,
for example, refs 14 and 15b,c and refs therein), in this case
the stilbene radical cation and the phenanthridyl radical. The
(15) (a) Mulliken, R. S.; Pearson, W. B. Molecular Complexes: A Lecture and
Reprint Volume; Wiley: New York, 1966. (b) Arnold, B. R.; Noukakis,
D.; Farid, S.; Goodman, J. L.; Gould, I. R. J. Am. Chem. Soc. 1995, 117,
4399. (c) Arnold, B. R.; Farid, S.; Goodman, J. L.; Gould, I. R. J. Am.
Chem. Soc. 1996, 118, 5482.
9
15228 J. AM. CHEM. SOC. VOL. 124, NO. 51, 2002

Kinetics of Reductive N−O Bond Fragmentation
A R T I C L E S
extinction coefficient of the stilbene radical cation in the visible
region is very large,13b and indeed the spectra in Figure 1 are
essentially identical to the reported spectrum of the stilbene
radical cation.13b We conclude that the phenanthridyl radical
absorbance is sufficiently small so that it cannot be observed
in the presence of the stilbene radical cation. Indeed, for all of
the CT complexes studied with all three donors, the ab-
sorption spectra of the geminate pairs were essentially those
of the radical cations, with absorption maxima at ca. 670 nm
(for BP•+),13a ca. 475 nm (for S•+),13b and ca. 710 nm (for
TAA•+).13a
The observed rate constant for exponential decay of the
absorption signal to the persistent level, kobsM, is equal to the
sum of the rate constants for all of the competing processes, eq
5a. The ratio of the persistent signal to that at time zero can be
termed the quantum yield for formation of separated radicals
and radical ions, ΦionsM. As shown in eq 5b, ΦionsM is equal to
the ratio of the sum of the rate constants that do not decrease
the radical cation signal to the sum of all rate constants, eq 5b.
with exponential fits to the data according to eq 9, where, A0 is
the absorbance at zero time.
A(t) ) (A0 - Φions) e-k .t + Φions
(9)
obs
As discussed above, in most cases, the absorbances of the
X-Y• radicals were too small compared to that of the D•+ ion
to be directly observed in these experiments. However, for
several of the styryl-substituted pyridiniums the X-Y• ab-
sorbances were sufficiently large to be observed. For compounds
9 and 11, a somewhat different method for obtaining kfr was
used that takes advantage of this fact. For these acceptors, CT
complexes were studied with 1,2,4-trimethylbenzene (TMB) as
the donor. Unlike the other donors, the TMB•+ has compara-
tively weak absorptions in the visible region,13a which allowed
the absorptions due to reduced 9 and 11 to be observed at ca.
440 and 470 nm, respectively. The X-Y• absorbances decay
via biexponential kinetics. A faster decay is observed due to
the radicals in the geminate pairs, followed by a slower decay
due to fragmentation of the radicals that escape the geminate
pair. The decay kinetics are described by eq 10, where AF, kf
and AS, kf refer to the amplitudes and observed rate constants
associated with the faster and slower exponential components,
respectively. At wavelengths where only the radical X-Y• is
observed and not the D•+, no persistent signal is observed, since
all of the radicals decay either by return electron transfer within
the geminate pair or by fragmentation after separation.
kobsM ) k-etM + ksep + kfr
(5a)
(5b)
ksep + kfr
M
Φions
)
k-etM + ksep + kfr
There is not enough information in the two pieces of
experimental data, kobsM and ΦionsM, to determine values for all
three elementary rate constants. The rate constant for diffusive
separation, ksep, was thus determined independently from studies
of the corresponding nonfragmenting N-ethylheterocyclic com-
pounds (Scheme 4) in otherwise identical charge-transfer
complexes. The rate constant for decay of the geminate pairs
in this case, kobsE, is given by eq 6a, and the corresponding
quantum yield for formation of separated radicals, ΦionsE, by
eq 6b.
A(t) ) AF e-k ‚t + AS e-k ‚t
kfr ) kS
(10a)
(10b)
F
S
Under these conditions, the slow observed rate, kS, is simply
the rate constant for fragmentation of the separated radicals,
kfr. The fast observed rate, kF, is equal to the sum of the
competing rates as described above, kobsM, assuming that
kobsE ) k-etE + ksep
(6a)
(6b)
fragmentation occurs with the same rate constant both in the
M
geminate pair and when separated. Values for (k-et + ksep
)
ksep
E
Φions
)
are obtained by subtracting kfr from kF.
k-etE + ksep
A possible kinetic complication that has been considered in
other studies of geminate kinetics is that the return electron
transfer rate may vary with separation distance.16 To test this
possibility, we attempted to fit the kinetic data using Shin and
Kapral’s equation for the time-dependent probability of geminate
pair survival, developed using Fick’s diffusion law and the
radiation boundary condition, eq 11.16
The rate constants for return electron transfer and separation in
E
these geminate pairs, k-et and ksep respectively, are obtained
directly using eqs 7a, b.
E
ksep ) ΦionsE‚kobs
(7a)
(7b)
k-etE ) kobsE - ksep
λ
κ - 1
P(τ) ) 1 -
erfc
-
{
Since differences in mass and shape of the N-methoxy- and
N-ethyl-substituted compounds are very small, ksep can safely
be assumed to be the same for the geminate pairs for both types
[
]
κ(1 + λ)
x
2 τ
2
κ - 1
x
e
[(1 + λ)(κ - 1)]e[(1 + λ) τ] erfc (1 + λ) τ +
(11)
}
[
]
x
2 τ
M
of compound. Values for kfr and k-et are thus obtained using
eqs 8a, b.
Here, τ is the ratio between elapsed time and the diffusive time
constant τ0 (τ0 ) σ2/Dc, where σ is the sum of the hard-sphere
collision radii and Dc is the mutual diffusion coefficient). The
parameter κ is the ratio between the initial pair separation and
the sum of the hard-sphere collision radii, λ is the ratio between
kfr ) ΦionsM‚kobsM - ksep
k-etM ) kobsM - ksep - kfr
(8a)
(8a)
Typical kinetic data are shown in Figures 1 and 2. The
absorbance of the normalized stilbene radical cation in each
geminate pair, A(t), is shown as a function of time, t, together
(16) (a) Shin, K. J.; Kapral, R. J. Chem. Phys. 1978, 69, 3685. (b) Scott, T. W.;
Liu, S. N. J. Phys. Chem. 1989, 93, 1393. (c) Hyde, M. G.; Beddard, G. S.
Chem. Phys. 1991, 151, 239.
9
J. AM. CHEM. SOC. VOL. 124, NO. 51, 2002 15229

A R T I C L E S
Lorance et al.
Table 2. Rate Constants for Fragmentation, Return Electron
Scheme 7
Transfer, and Separation, for Geminate Pairs of Reduced
N-Methoxyheterocycles at Room Temperaturea
the effective rate constant for reactions that decrease D•+ to
the rate constant for diffusional separation, and erfc is the
complimentary error function.
Inspection of the residuals corresponding to the best fit to
the data for the N-methoxy-2-styrylpyridinium (7) using this
method, Figure 2, reveals that eq 11 is actually a poorer
description of the kinetic behavior than the simple exponential
description. Similar results were obtained for the other CT
complexes studied here (data not shown). For this reason and
the fact that extraction of elementary rate constants from κ and
λ is not possible without further assumptions,16b,c we chose not
to use the kinetic approach of eq 11.
A perhaps more sophisticated approach to the problem of
variation of return electron transfer rate with distance, which
takes into account accepted radical distribution functions for
particles in condensed media,17 includes two possible geminate
species, that is, contact and solvent-separated radical ion pairs,
CRIP and SSRIP, respectively, Scheme 7.18 Return electron
transfer in a contact pair, k-etcp, may be larger than in a solvent-
separated pair, k-etss, because of larger electronic coupling.18
Formation of the SSRIP from the CRIP is understood as a
solvation process, ksolv, and overall separation occurs from the
SSRIP, that is, the meaning of ksep is somewhat different from
that in Scheme 6.15b,18 A complete description also includes the
a For charge-transfer complexes of the radical precursor (bold number)
with stilbene as the donor in butyronitrile solvent, except where noted. See
Experimental Section for estimates of errors. b Value used in the determi-
nation of kfr and k-et (see text), obtained from experiments on the
corresponding N-ethyl compound (Table 3), except where noted. c WIth
tri-p-anisylamine as the donor in acetonitrile solvent. d Too small to be
considered, see text. e With biphenyl as the donor. f With 1,2,4-trimethyl-
benzene as the donor in acetonitrile solvent. Rate constant obtained by direct
observation of the radical, does not rely on other measurements, see text.
g This value represents the sum (k-et + ksep), see text. h Assumed to be equal
to the rate constant obtained for compound 10E.
fact that the SSRIP and CRIP may interconvert via k-solv.
18 For
this kinetic scheme, double exponential kinetics should be
observed for the absorption decays of the geminate species, eq
12.
‚t
A(t) ) AF′ e-k ′ + AS′ e-k ‚t + Φions
(12)
F
S
are barely different for the N-methoxy compound, 0.9566 and
0.9552 for single and double exponential analyses, respectively.
Furthermore, a two-tailed z-test using Fischer’s Z transformation
for determining the significance of a difference between two
correlation coefficients demonstrates that the double-exponential
analysis is not statistically different from the single-exponential
analysis, up to and beyond the 99% confidence limit.19 Double-
exponential analyses of other geminate pairs (data not shown)
provided even less satisfactory results. Either one of the
preexponential factors AF′ or AS′ was very small, or kF′ and kS′
were very similar, implying that the decay kinetics were equally
well described by single-exponential kinetics. Presumably, one
of the two geminate pairs is being observed mainly in these
experiments because the lifetime of the other is too short to be
readily detected. For this reason, the single-exponential analysis
method for analyzing the data described above was used. It is
difficult to say whether the observed fragmentation occurs in
the CRIP or the SSRIP, although this is not important for the
present purposes.
For such a kinetic analysis to be useful, however, neither of
the amplitudes AF′ or AS′ should be much larger than the other,
and the values of kF′ and kS′ should not be too similar.18c In
other words, the lifetimes of both geminate pairs should be both
measurable and different. In turn this means that the CRIP
should not solvate to form the SSRIP so rapidly that the former
pair is not observed, and the SSRIP should be sufficiently long-
lived that a reasonable population builds up during the time
scale of the experiment. The lifetime of the SSRIP depends on
the relative values of k-etss, kfr, ksep, and k-solv in a complex
manner. Clearly, kinetic data that can be fitted reasonably well
using a single-exponential function must be fitted as well or
even better by using a double exponential function. Inspection
of the residuals corresponding to the best fit to the data for the
N-methoxy-2-styrylpyridinium (7), Figure 2, reveals a slightly
better fit for the double compared to that of the single-
exponential analysis (based on the sum of the squares of the
residuals), as expected. However, the correlation coefficients
Fragmentation Rate Constants. Fragmentation of the
radicals from compounds 2 and 12 was slow enough to be
measured using conventional nanosecond laser spectroscopy as
(17) Chandler, D. Introduction To Modern Statistical Mechanics; Oxford: New
York, 1987.
(18) (a) Gould, I. R.; Farid, S. Acc. Chem. Res. ,1996, 29, 522. (b) Gould, I. R.;
Young, R. H.; Farid, S. J. Phys. Chem. 1991, 95, 2068. (c) Arnold, B. R.;
Atherton, S. J.; Farid, S.; Goodman, J. L.; Gould, I. R. Photochem.
Photobiol. 1997, 65, 15.
(19) Spiegel, M. R. Theory and Problems of Statistics; McGraw-Hill: New York,
1961.
9
15230 J. AM. CHEM. SOC. VOL. 124, NO. 51, 2002

Kinetics of Reductive N−O Bond Fragmentation
A R T I C L E S
Table 3. Rate Constants for Return Electron Transfer, and
Separation, for Geminate Pairs of Reduced N-Ethylheterocycles at
Room Temperaturea
Figure 3. State correlation diagrams for a pure π* pyridyl radical (bottom
left) and a pure σ* radical (top left) as a function of N-O bond length,
rN-O. The dotted lines show the correlations (and lack of avoidance) for
the flat geometry. The solid curves show the energies of lower, L, and
upper, U, electronic energy states formed by mixing the pure radical states
at a geometry in which the N-O bond is bent out of the plane of the
aromatic ring. When rN-O is equal to rC, the curves meet at a conical
intersection in the flat conformation.
haps an exception. The somewhat smaller preexponential factor
here may be a consequence of the fact that the reactive species
is a cation, and solvent reorientation effects in the transition
state may be important. An additional factor may be the fact
that there are degenerate radical states in this case, see further
below.
a For charge-transfer complexes of the radical precursor (bold number)
with stilbene as the donor in butyronitrile solvent, except where noted. See
Experimental Section for estimates of errors. b With biphenyl as the donor.
Scheme 8
Qualitatively, the fragmentation rate constants exhibit reason-
able trends. The reaction rate constants decrease with increasing
strength of the electron-withdrawing group (compare 1 versus
3, 10 versus 9, etc.). The reaction rate constants also decrease
with increasing delocalization in the aromatic system (compare
1 versus 4-6 and 10). A combination of delocalization and
withdrawing groups decreases the rate constant further (compare
3 versus 11, 1 versus 11, and 12, etc.). The extreme case is 2,
which exhibits by far the slowest rate constant for fragmentation.
The p-(N-methoxypyridyl)) is the strongest withdrawing group,
but perhaps more importantly, delocalization of this radical is
most extensive due to degenerate resonance structures with the
radical alternately on each of the two rings, Scheme 8. This is
presumably a major reason for relatively slow fragmentation
(relatively high radical stability) in this case.
Assuming the thermodynamics and kinetics of the radical
fragmentation reactions are related, the thermodynamic cycle
of Scheme 2 provides a qualitative explanation for the electron-
withdrawing effects, if not the delocalization effects. However,
we sought a molecular explanation for the observed rate trends,
which requires an examination of the important orbital and state
correlations.
Conical Intersection Model for Fragmentation. A useful
starting point is the simple-state correlation diagram shown in
Figure 3. For simplicity, the simple N-methoxypyridinium parent
compound (not studied experimentally) is chosen as an illustra-
tive example. For most nitrogen heterocycles, the lowest-energy
unoccupied orbital is clearly π*.21 The pyridyl radical formed
upon reduction of N-methoxypyridinium (lower left in Figure
3) is thus shown in the geometry of the cation, with the extra
electron associated with the π system (the π* radical). Increasing
described above. Measurements as a function of temperature
were performed, and the Arrhenius parameters summarized in
Table 1 were determined. The rate constant for 2 in 1,2-
dichloroethane is reasonably close to the value of 1.4 × 104
s-1 reported in the literature in acetonitrile for the corresponding
N-ethoxy compound.11a
The rate constants for fragmentation of the other radicals were
determined using the CT excitation methods described above.
The elementary rate constants for the CT complexes of the
N-methoxy compounds and the corresponding N-ethyl-deriva-
tives are summarized in Tables 2 and 3. For 1, the sum of the
rate constants for fragmentation and return electron transfer was
much larger than that for separation, which was thus ignored.
The value obtained for 3 is reasonably similar to that reported
previously, 8 × 109 s-1 11b
.
The data exhibit a very wide range of values for kfr, almost
7 orders of magnitude, from 4 × 104 s-1 for 2 to 2.7 × 1011
s-1 for 1. Experimental measurements of the temperature
dependencies of the reactions for 2 and 12 gave estimates of
the barrier heights for these reactions, Table 1. The Arrhenius
preexponential factors obtained in these cases (1.75 × 1012 and
1.52 × 1013 s-1 for 2 and 12, respectively) are large. The value
for 12 is very close to (kT/h), which is consistent with a
unimolecular reaction that occurs efficiently after the energy
barrier is surmounted.20 Assuming that 12 is a reasonable model
for the other reactions of Table 2, this would imply that the
differences in the rate constants are due mainly to differ-
ences in energy barrier heights, the radical from 2 being per-
(20) Moore, J. W.; Pearson, R. G. Kinetics and Mechanism, 3rd ed.; John
(21) Turner, D. W. Molecular Photoelectron Spectroscopy: A Handbook of He
584 Å Spectra; Wiley-Interscience: New York, 1970.
Wiley: New York, 1981.
9
J. AM. CHEM. SOC. VOL. 124, NO. 51, 2002 15231

A R T I C L E S
Lorance et al.
the N-O bond length, rN-O, produces the methoxy radical and
pyridine with uncoupled electrons, one localized on an atomic
orbital on nitrogen and one in the π system. Thus, bond
stretching in the π* radical leads to an excited state of pyridine
and does not correlate with the ground-state reaction products,
which are shown on the lower right of Figure 3. Decreasing
the nitrogen-oxygen bond distance in the ground-state products,
that is, reVersing the fragmentation reaction, leads to a higher-
energy pyridyl radical in which the extra electron is in a N-O
σ* orbital (the σ* radical, top left in Figure 3). The state
correlations, indicated by the dashed lines in Figure 3, cross at
an N-O bond distance, rC. Bond fragmentation thus requires a
transition from the π* radical state to the σ* radical state. This
is not possible at the crossing point since the required mixing
of the π* and σ* orbitals is symmetry-forbidden in the flat
configuration of the radical.22
Mixing of these orbitals is allowed upon bending of the N-O
bond out of the plane of the aromatic ring.22 In the absence of
orbital mixing (pure π* and pure σ* radicals), this bending
would result in an increase in energy. This is due to increases
in the energy of some of the occupied orbitals, described in
detail below, and at very large bending angles to electron
repulsion effects. For the π* radical, these energy increases are
offset somewhat by a decrease in energy of the singly occupied
orbital, see further below. The increases in energy with
increasing bending angle, R (defined in Figure 4), for the pure
π* and σ* radicals are illustrated qualitatively as the dotted
curves in Figure 4, for two different values of rN-O. The upper
panel of Figure 4 represents the situation at the energy crossing
point, rC. Here, the energies of the pure π* and σ* radicals are
identical in the flat geometry (R ) 180°), and the two dotted
curves touch. The lower panel in Figure 4 illustrates the situation
at a smaller value of rN-O, that is, where the energy of the pure
π* radical is lower than that of the pure σ* radical.
Figure 4. (Dashed lines) Qualitative representation of the energies of pure
π* and σ* pyridyl radicals as a function of the N-O out-of-plane bending
angle R. (Solid lines) Energies of the two radical states, L ) lower and U
) upper, formed by mixing the pure π* and σ* radicals. When the N-O
bond length is equal to rC (top panel), the energies of the two radical
configurations are equal in the flat geometry. At smaller N-O bond
lengths (lower panel), the energy of the π* radical is lower than that of the
σ* radical. When rN-O is equal to rC, the curves meet at a conical
intersection.
minimal. Under these conditions, the energy splitting between
the L and U surfaces around rC is small, and an unproductive
jump from the L to the U surface is possible as rN-O approaches
rC.24 The reaction rate constant increases with increasing bending
since this increases the splitting between the U and L surfaces,
and decreases the probability of unproductive jumps from the
L to the U surface.24
At larger bending angles the π*/σ* mixing is extensive. Under
these conditions, the splitting between the L and U surfaces is
large and jumps between the surfaces have negligible prob-
ability, that is, the reaction becomes adiabatic.24 Increasing the
out-of-plane bending further increases the energy splitting
between the U and L states, thus lowering the barrier to reaction
and increasing the reaction rate. The large values for the
preexponential factors obtained in the temperature dependence
studies (Table 1), suggest that the reactions are indeed adiabatic
and that the reactions become faster with increased bending for
this second reason, that is, due to a lowering of the energy
barrier.
The important consequence of bending, however, is the π*/
σ* orbital mixing that produces two new radical states from
the π* and σ* states. The new states are designated as L for
the lower-energy radical state that actually undergoes the
fragmentation reaction, and U for the corresponding higher-
energy (excited) state, respectively. These are indicated in
Figures 3 and 4 by the solid curves. The L state is lower in
energy than either of the π* or σ* states, and at lower bending
angles, an overall stabilization results. Orbital mixing thus
decreases the energy of the radical, and opposes the energy-
increasing effects of bending described above. The amount of
bending in a particular radical will be determined by the extents
to which these energy increasing and -decreasing contributions
vary with bending angle. The point at which the radical states
cross at rC in Figure 3, and touch in the bending coordinate in
Figure 4, defines a conical intersection that the reaction must
avoid.12,23
The reaction thus proceeds along the lower-energy surface,
L, by a combination of N-O stretching and bending motions.
The reaction rate will presumably increase with increasing
bending angle (up to a point) for one of two possible reasons.
When the bending angles are small, the π*/σ* mixing is
To provide further evidence for the conical intersection
description of the reaction potential energy surface, ab initio
electronic structure calculations were performed. The protonated
N-methoxypyrazinium radical 14R, eq 13, was chosen as a
model radical for the calculations since it contains both the
(22) Michl, J.; Bonacic-Koutecky, V. Electronic Aspects of Organic Photo-
chemistry; Wiley-Interscience: New York, 1990.
(23) The touching of the surfaces in the bending coordinate is a Renner-Teller
(24) (a) Butler, L. J. Annu. ReV. Phys. Chem. 1998, 49, 125. (b) Salem, L.
Electrons in Chemical Reactions; Wiley-Interscience: New York, 1982.
type interaction.12a
9
15232 J. AM. CHEM. SOC. VOL. 124, NO. 51, 2002

Kinetics of Reductive N−O Bond Fragmentation
A R T I C L E S
Figure 5. Electronic energy surface for reaction of 4-H+-1-methoxypyrazinium cation, 14R, calculated using the UHF/6-31+G* method. The energies, kcal
mol-1, are normalized to the energy of the minimized radical structure. The locations, structures, and values for rN-O and R of the radical minimum and the
transition state are indicated. The conical intersection (CI) is located at the top right corner.
N-methoxypyridyl structure and a simple and strong withdraw-
ing group.
in each case. The calculated N-O bond lengths, rN-O, and out-
of-plane bending angles, R, are summarized in Figure 6, together
with the corresponding rate constants for fragmentation. The
results reveal some interesting trends. In general, the N-O bond
length in the radical increases with increasing reaction rate, from
1.37 Å for 2 (the slowest fragmenting radical) to 1.43 Å for 1
(the fastest fragmenting radical). Larger effects are seen in the
out-of-plane bending angles, which increases significantly with
increasing reaction rate. Radical 2 is close to flat (R ) 175°),
whereas 1 is significantly bent (R ) 148°). For a completely
sp3-hybridized nitrogen, R would be ca. 135°.
The electronic energy of the radical was computed using the
spin-unrestricted Hartree-Fock method (UHF/6-31+G*). Thirty
six points were geometry-optimized at different values of the
N-O bond length rN-O (from 1.3 to 1.8 Å at 0.1 Å intervals)
and the N-O bending angle R (from 110° to 185° at 15°
intervals) as indicated by the large, coarse grid of Figure 5.
Seventy further points were optimized for rN-O from 1.575 to
1.725 Å at 0.025 Å intervals, and for R from 110° to 155° at 5°
intervals in the same way for the fine grid covering the area
near the transition state. The points at the different values of
As discussed above, the varying degrees of bending in the
radicals can be understood as arising from varying extents of
mixing of the π* and σ* orbitals. The orbital energy diagram
of Figure 7 provides a useful basis for this discussion. Bending
transforms the important molecular orbitals from those of a
cyclic π-radical system, πc, into those of a pentadienyl system,
πp, plus a nonbonding orbital on nitrogen, n. As discussed above,
in the absence of π*/σ* orbital mixing, the overall energy
rN-O and R were used to construct the potential energy surface
indicated by the contour lines in Figure 5. In addition, the
minimum and transition state were calculated separately and
without constraints, as indicated in the Figure. The Figure very
clearly indicates that the reaction path from the minimum to
the transition state involves simultaneous N-O bond stretching
and bending. The conical interaction was not specifically located,
but is clearly at R ) 180° and between 1.75 and 1.8 Å, Figure
5.
The data also clearly indicate highly bent transition states
for the fragmentation reaction. Thus, the important factors that
control the reaction rate should be those that determine the extent
of out-of-plane bending and orbital mixing in the bent confor-
mation. The faster reactions will require less energy to reach
the bent transition state, and slower reactions will require more
energy.
c
increases with bending since two electrons (in π3 ) are raised
from a bonding to a nonbonding orbital (n) as the orbital system
transforms from πc to πp. In the pure π* radical shown in Figure
7, this is partially offset by the fact that the unpaired electron
c
p
is lowered from the antibonding π4 to the nonbonding π3 in
the pentadienyl system.
These energy-increasing contributions are compensated by
the stabilizing effect of orbital mixing. In Figure 7, the most
important mixing is illustrated, that is, that between the highest
occupied π3p orbital of the pentadienyl system and the relatively
low-energy N-O σ* orbital.26 The simplest interpretation of
the results of the electronic structure calculations is that the
extent of this mixing, and thus the energy benefit of out-of-
To investigate the role of N-O out-of-plane bending in the
reactions, more accurate UB3PW91/6-31+G* density functional
electronic structure calculations were performed on several of
the fragmenting radicals that were studied experimentally.25
Minimum-energy configurations for the radicals were obtained
(25) (a) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (b) Hehre, W. J.; Ditchfield,
R.; Pople, J. A. J. Chem. Phys. 1972, 56, 2257. (c) Krishnan, R.; Binkley,
J. S.; Seeger, R.; Pople, J. A. J. Chem. Phys. 1980, 72, 650. (d) McLean,
A. D.; Chandler, G. S. J. Chem. Phys. 1980, 72, 5639.
p
(26) The π3 orbital is roughly at the nonbonding energy level and thus is not
strictly a π* orbital. To avoid changing notation, however, we will continue
to refer to π*/σ* mixing.
9
J. AM. CHEM. SOC. VOL. 124, NO. 51, 2002 15233

A R T I C L E S
Lorance et al.
Figure 6. UB3PW91/6-31+G* minimized structures for radicals obtained by one-electron reduction of N-methoxyheterocycles (bold numbers), the calculated
radical nitrogen-oxygen bond distances, rN-O, the out-of-plane bending angles, R, and the experimental rate constants for N-O bond fragmentation, kfr.
important in this case. Furthermore, because the energy of the
p
π3 orbital is raised, mixing with the σ* orbital is increased.
Thus, radicals with electron-donating groups should have a more
bent geometry, and thus be closer to the geometry of the
transition state for fragmentation, and should react faster.
The effect of delocalizing groups is more subtle. Delocalizing
groups reduce the energy gap between the bonding and
antibonding orbitals in both the cyclic and the pentadienyl
systems. However, in the pentadienyl radical, the unpaired
electron is close to the nonbonding energy level (π3p in Figure
7), and there is no significant change in the energy of this orbital
upon delocalization. Thus, delocalization lowers the energy of
the unpaired electron in the flat cyclic radical system and not
Figure 7. Molecular orbital energy diagram for (left) a flat π* pyridyl
radical and (right) a bent π* pyridyl radical. The orbitals in the flat
in the bent pentadienyl radical. Bending in the delocalized
radicals is thus associated with a higher energy cost, and
consequently the radicals are flatter. The flatter delocalized
radicals require more energy to reach the bent geometry of the
transition state and fragment more slowly; again, all as observed.
Return Electron Transfer and Separation. In general, the
return electron transfer rate constants appear to follow expected
trends for electron-transfer reactions in the inverted region.18a
The main factors that control the rate constant for return electron
transfer is the exothermicity of the reaction and, to a lesser
extent, the molecular size.18a The reaction exothermicity is given
by the oxidation potential of the donor and the reduction
potential of the acceptor.18a Thus, only reactions with the same
donor can usefully be compared. In the inverted region, the less
negative the reduction potential, the less exothermic the return
electron transfer reaction and the larger the rate constant.
Although reduction potentials for the various N-methoxy and
N-ethyl-heterocycles are not generally available, some obvious
trends can be observed. For example, return electron transfer is
significantly faster for the styrylpyridiniums with strong with-
drawing groups, 9 and 11 as compared to 10, as expected.27 In
the inverted region, smaller compounds tend to have larger rate
constants for return electron transfer. Although it is difficult to
determine the extent to which differences in the rate constants
for 4, 5, 6, and 10 are due to differences in reduction potential,
conformation are those of a cyclic π system, those in the bent conformation
are those of a pentadienyl system plus a nonbonding orbital on nitrogen.
Mixing between the π orbitals and the antibonding σ* orbital of the N-O
bond occurs only in the bent conformation.
plane bending, is determined by the effects of the substituents
on the important molecular orbitals. The higher the energy of
p
the π3 orbital (the closer to that of the σ* orbital), the greater
the mixing.
An electron-withdrawing substituent in the para position of
c
a pyridyl radical, Z in Figure 7, will lower the energies of π1 ,
c
π2 , and π4c in the cyclic system but will only lower the energies
of π1p and π3p in the pentadienyl system. Thus, a withdrawing
group will stabilize a pure π* radical more when flat than when
bent, since the energies of more electrons are lowered in the
flat conformation. Thus, compared to an unsubstituted radical,
the energy-increasing contributions to bending are more im-
portant with an electron-withdrawing group. Furthermore, by
lowering the energy of the π3p orbital, mixing with the σ* orbital
is decreased in the bent conformation, and the stabilizing effect
of mixing is decreased. Consequently, radicals with electron-
withdrawing groups are more planar, more energy is required
to reach the bent geometry of the transition state for fragmenta-
tion, and their reactions are slower, all as observed.
c
An electron-donating group will raise the energies of π1 ,
c
c
p
p
π2 , and π4 in the cyclic system, and π1 and π3 in the
pentadienyl system. Thus, a donating group will destabilize a
pure π* radical more when flat than when bent (since the
energies of more electrons are raised in the flat conformation),
and the energy-increasing contributions to bending are less
(27) k-et was not determined for 9 and 11, only the sum (k-et + ksep) (Table 2).
However, it is clear from Table 3 that the ksep values for the other
compounds vary over a small range and are all much smaller than (k-et
sep) for 9 and 11. Therefore, k-et for 9 and 11 must be much larger than
that for 10.
+
k
9
15234 J. AM. CHEM. SOC. VOL. 124, NO. 51, 2002

Kinetics of Reductive N−O Bond Fragmentation
A R T I C L E S
it is generally true that the smaller compounds undergo return
electron transfer with larger rate constants, as expected. Interest-
ingly, return electron transfer is uniformally slower for the
N-ethyl (Table 3)- compared to the N-methoxy-substituted
compounds (Table 2). Presumably, the N-methoxy compounds
are easier to reduce than the N-ethyl compounds. Another
interesting possible reason for faster return electron transfer for
the N-methoxy compared to that for the N-ethyl radicals is that
the change in the N-O bond length upon reduction of the former
is presumably much larger than the corresponding change in
the N-C bond length in the latter. Return electron transfer for
the N-methoxy radicals will thus be characterized by a larger
internal reorganization energy, which will also contribute to an
increase in the rate constant for return electron transfer in the
inverted region.18a
best of our knowledge, none at all for reactions of neutral
organic radicals in solution. We are in the process of further
exploring the role of the conical intersection in these reactions
using more extensive electronic structure calculations.29
Experimental Section
Materials. The solvents were spectrograde from Aldrich and were
used as received. 9,10-Dicyanoanthracene and p-trianisylamine were
gifts from Samir Farid (Eastman Kodak Company) and used as received.
Stilbene, biphenyl, 1,2,4-trimethylbenzene, and benzyltrimethylsilane
were obtained from Aldrich and were used as received. 9-Thioxan-
thenone was obtained from Aldrich, and was recrystallized from ethanol.
General Synthetic Procedures. The N-ethyl- and N-methoxy
heterocycles were prepared by alkylation of the corresponding hetero-
cycles and heterocycle N-oxides. The compounds were characterized
by proton and carbon NMR spectroscopy using the Varian Gemini 300,
Varian Inova 400, and Varian Inova 500 spectrometers. The samples
were dissolved in CDCl3, DMSO-d6, CD3CN, or CD3OD (Cambridge
Isotopes). Mass verification was accomplished using a Vestec MALDI-
TOF mass spectrometer with 337-nm excitation; the compounds were
sometimes excited without a matrix (LD-TOF). Melting points are
uncorrected.
The rate constants for separation do not vary over a very wide
range, from ca. 6-30 × 108 s-1, Table 2. In general, the rate
constants are larger for the smaller compounds. A full discussion
of the factors controlling the rate constants for separation will
be published elsewhere.28
General Procedures for the Heterocycles. Quinoline, isoquinoline,
phenanthridine, 4-benzoylpyridine, and 4-cyanopyridine were obtained
from Aldrich and used as received. 2-Styrylpyridine and 2-(4-
chlorostyryl)pyridine were gifts from Samir Farid (Eastman Kodak
Company) and were used as received. The other styrylpyridines were
prepared using the condensation procedure of Williams, et al.32
General Procedures for the N-Oxides. Quinoline N-oxide
(Aldrich), isoquinoline N-oxide (Aldrich), 4,4’-dipyridyl-N,N’-dioxide
(Aldrich), and 4-cyanopyridine N-oxide (Lancaster) were obtained
commercially and used as received. Phenanthridine N-oxide was
prepared according to the method of Hayashi and Hotta.33 The other
N-oxides were prepared using the following general procedure.34 The
heteroaromatic compound (6 mmol) was dissolved in 35 mL of glacial
acetic acid in a 100-mL round-bottom flask. After the addition of 8
mmol H2O2 (30% solution) the mixture was held at 80 °C for 12 h.
After cooling to room temperature, the mixture was poured into
chloroform and extracted successively with an aqueous solution of HCl,
an aqueous solution of sodium hydroxide, and then brine. The organic
phase was dried over magnesium sulfate, and the solvent was removed.
The crude N-oxides were alkylated without further purification.
General Procedures for the N-Ethyl Heterocycles. The N-ethyl
heterocycles were prepared according to the following general proce-
dure. A mixture of 1.15 equiv of triethyloxonium hexafluorophosphate
(Aldrich) and 1 equiv of the N-heterocycle was stirred in dichloro-
methane under argon for 4 h. After the addition of 15 mL of methanol
the solvent was removed, and the crude product was recrystallized.
General Procedures for the N-Methoxy Heterocycles. N-Methoxy-
4-(4-cyanostyryl)pyridinium tetrafluoroborate and N-methoxy-4-phen-
ylpyridinium tetrafluoroborate were gifts from Samir Farid (Eastman
Kodak Company) and were used as received. The others were prepared
according to the following general procedure. A mixture of 1.15 equiv
of trimethyloxonium tetrafluoroborate (Aldrich) and 1 equiv of the
heterocycle N-oxide was stirred in dichloromethane under argon for 4
h. After the addition of 15 mL of methanol the solvent was removed,
and the crude product was recrystallized.
Conclusions
N-O bond fragmentation of radicals formed by one-electron
reduction of N-methoxy-substituted aromatic compounds occurs
with rate constants that can vary over a very wide range, from
4.0 × 104 to 2.7 × 1011 s-1 for those radicals studied here. The
reaction requires mixing of π* and σ* orbitals,26 which is
achieved by bending the N-O bond out of the plane of the
aromatic ring. Substituent and delocalization effects on the
reacting radicals can be understood in terms of varying extents
of this orbital mixing. Extensive mixing leads to a radical that
is close in geometry and energy to the bent transition state for
reaction. Strong electron-withdrawing groups or delocalizing
structures or both result in decreased π*/σ* orbital mixing and
slow the reactions. The discussion above suggests that electron-
donating groups should increase the extent of π*/σ* orbital
mixing so that fragmentation may be even faster than the fastest
reactions observed here. We are currently exploring structures
of this type to confirm this prediction, with the ultimate goal
of discovering radical fragmentation reactions that occur with
no energy barrier.29
The orbital mixing results in a conical intersection that the
reaction avoids. Although the important role of conical intersec-
tions in photochemical processes is well documented,30 there
have been very few discussions of conical intersections in
ground-state thermal reactions of organic species,31 and to the
(28) Lorance, E. D.; Gould, I. R. Manuscript in preparation.
(29) Lorance, E. D.; Kramer, W. H.; Hendrickson, K. A.; Gould, I. R. Manuscript
in preparation.
(30) See, for example: (a) Klessinger, M. J. Photochem. Photobiol. A 2001,
144, 217. (b) De Vico, L.; Page, C. S.; Garavelli, M.; Bernardi, F.; Basosi,
R.; Olivucci, M. J. Am. Chem. Soc. 2002, 124, 4124. (c) Ben-Nun, M.;
Molnar, F.; Schulten, K.; Martinez, T. J. Proc. Natl. Acad. Sci. U.S.A. 2002,
99, 1769. (d) Clifford, S.; Bearpark, M. J.; Bernardi, F.; Olivucci, M.; Robb,
M. A.; Smith, B. R. J. Am. Chem. Soc. 1996, 118, 7353. (e) Wilsey, S.;
Houk, K. N. J. Am. Chem. Soc. 2000, 122, 2651. (f) Yamamoto, N.;
Olivucci, M.; Celani, P.; Bernardi, F.; Robb, M. A. J. Am. Chem. Soc.
1998, 120, 2407. (g) Wilsey, S.; Houk, K. N.; Zewail, A. H. J. Am. Chem.
Soc. 1999, 121, 5772. (h) Zhong, D.; Diau, E. W.-G.; Bernhardt, T.; De
Feyter, S.; Roberts, J. D.; Zewail, A. H. Chem. Phys. Lett. 1998, 298, 129.
(31) (a) Blancafort, L.; Jolibois, F.; Olivucci, M.; Robb, M. A. J. Am. Chem.
Soc. 2001, 123, 722. (b) Bearpark, M. J.; Robb, M. A.; Yamamoto, N.
Spectrochim. Acta, Part A 1999, 55, 639. (c) Blancafort, L.; Adam, W.;
Gonzalez, D.; Olivucci, M.; Vreven, T.; Robb, M. A. J. Am. Chem. Soc.
1999, 121, 10583. (d) Diau, E. W.-G.; De Feyter, S.; Zewail, A. H. Chem.
Phys. Lett. 1999, 304, 134.
4-Styrylpyridine. Following the literature procedure,32 4-picoline
(6.0 g, 64.4 mmol) and benzaldehyde (6.84 g, 64.4 mmol) were refluxed
12 h in 35 mL of acetic anhydride to give 5.50 g (30.3 mmol, 47%) of
4-styrylpyridine. Mp 121-123 °C. 1H NMR (300 MHz, CDCl3) δ
(32) Williams, J. L. R.; Adel, R. E.; Carlson, J. M.; Reynolds, G. A.; Borden,
D. G.; Ford, J. A. J. Org. Chem. 1963, 28, 387.
(33) E. Hayashi, Y. Hotta, J. Pharm. Soc. Jpn. 1960, 80, 834.
(34) Katritzky, A. R.; Lagowski, J. M. Chemistry of the Heterocyclic N-Oxides;
Academic: New York, 1971.
9
J. AM. CHEM. SOC. VOL. 124, NO. 51, 2002 15235

A R T I C L E S
Lorance et al.
(ppm) ) 6.98 (d, 1 H, J ) 16.5 Hz), 7.25 (d, 1 H, J ) 16.5 Hz),
7.31-7.40 (m, 5 H), 7.51 (dd, 2 H, J ) 1.2, 8.1 Hz), 8.55 (d, 2 H, J
) 6.3 Hz). 13C NMR (75 MHz, CDCl3) δ (ppm) ) 120.8, 125.9, 126.9,
128.7, 128.8, 133.2, 136.1, 144.6, 150.0.
(d, 1 H), 9.19 (d, 1 H), 9.49 (d, 1 H). 13C NMR (75 MHz, CDCl3) δ
(ppm) ) 13.7, 52.8, 117.7, 121.5, 129.2, 129.6, 130.2, 135.2, 137.1,
146.4, 148.6. MS m/z 158.3 (M+).
N-Methoxyisoquinolinium Tetrafluoroborate (5). Following the
general procedure, trimethyloxonium tetrafluoroborate (4.48 g, 30.3
mmol) and isoquinoline N-oxide (4.00 g, 27.6 mmol) were stirred in
120 mL of dichloromethane for 4 h. Recrystallization from ethyl acetate
afforded 6.40 g (25.9 mmol, 94%) of 5. 1H NMR (300 MHz, CD3OD)
δ (ppm) ) 4.56 (s, 3 H), 8.06-8.11 (m, 1 H), 8.21-8.27 (m, 1 H),
8.34 (dd, 1 H, J ) 1.2, 8.1 Hz), 8.49 (dd, 1 H, J ) 1.2, 8.1 Hz), 8.57
(d, 1 H, J ) 7.2 Hz), 8.89-8.93 (m, 1 H), 10.20 (d, 1 H, J ) 1.8 Hz).
13C NMR (75 MHz, CD3OD) δ (ppm) ) 70.3, 118.5, 128.7, 129.0,
131.2, 132.3, 133.0, 138.1, 138.3, 145.5. MS m/z 160.6 (M+).
N-Ethylisoquinolinium Hexafluorophosphate (5E). Following the
general procedure, triethyloxonium hexafluorophosphate (5.81 g, 23.4
mmol) and isoquinoline (2.50 g, 19.4 mmol) were stirred in 60 mL of
dichloromethane for 4 h. Recrystallization from methanol/dichloro-
methane afforded 4.10 g (13.5 mmol, 70%) of 5E. 1H NMR (300 MHz,
CD3OD) δ (ppm) ) 1.71 (t, 3 H, J ) 7.2 Hz), 4.77 (q, 2 H, J ) 7.2
Hz), 8.00 (dt, 1 H, J ) 0.9, 8.1 Hz), 8.14-8.26 (m, 2 H), 8.39-8.46
(m, 2 H), 8.59 (dd, 1 H, J ) 0.9, 6.6 Hz), 9.78 (s, 1 H). 13C NMR (75
MHz, CD3OD) δ (ppm) ) 16.5, 58.3, 127.6, 128.4, 129.3, 131.5, 132.5,
135.3, 138.2, 138.9, 150.6. MS m/z 158.3 (M+).
N-Methoxyphenanthridinium Tetrafluoroborate (6). Following
the general procedure, trimethyloxonium tetrafluoroborate (1.42 g, 9.6
mmol) and phenanthridine N-oxide (1.7 g, 8.7 mmol) were stirred in
40 mL of dichloromethane for 4 h. Recrystallization from methanol
afforded 2.1 g (7.1 mmol, 81%) of 6. 1H NMR (500 MHz, CD3OD at
55 °C) δ (ppm) ) 4.67 (s, 3 H), 8.08-8.20 (m, 3 H), 8.38 (dt, 1 H, J
) 1.5, 7.8 Hz), 8.58 (dt, 2 H, J ) 1.0, 8.0 Hz), 9.02 (d, 1 H, J ) 8.5
Hz), 9.07 (d, 1 H, J ) 8.0 Hz), 10.46 (s, 1 H). 13C NMR (125 MHz,
CD3OD at 55 °C) δ (ppm) ) 70.3, 118.4, 124.5, 125.1, 125.9, 128.7,
131.9, 132.7, 133.2, 133.8, 133.9, 135.8, 139.6, 149.7. MS m/z 210.5
(M+).
4-(4-Nitrostyryl)pyridine. Following the literature procedure,32
4-picoline (3.08 g, 33.1 mmol) and 4-nitrobenzaldehyde (5.00 g, 33.1
mmol) were refluxed 12 h in 17 mL of acetic anhydride to give 4.32
g (19.1 mmol, 58%) of 4-(4-nitrostyryl)pyridine. Mp 166-168 °C. 1H
NMR (300 MHz, CDCl3) δ (ppm) ) 7.14 (d, 1 H, J ) 16.5 Hz), 7.31
(d, 1 H, J ) 16.5 Hz), 7.36-7.39 (m, 2 H), 7.63-7.66 (m, 2 H), 8.17-
8.21 (m, 2 H), 8.59 (br d, 2 H). 13C NMR (75 MHz, CDCl3) δ (ppm)
) 121.0, 124.1, 127.4, 130.3, 130.6, 142.4, 143.4, 147.4, 150.2.
4-(2,3,4,5,6-Pentafluorostyryl)pyridine. Following the literature
procedure,32 4-picoline (0.70 g, 7.65 mmol) and pentafluorobenzalde-
hyde (1.5 g, 7.65 mmol) were stirred 4 d in 15 mL of acetic anhydride.
The resulting solid was purified by column chromatography (silica gel,
methanol/dichloromethane 1:30, Rf ) 0.4) to give 0.5 g (1.84 mmol,
1
24%) of 4-(2,3,4,5,6-pentafluorostyryl)pyridine. Mp 113-115 °C. H
NMR (300 MHz, CDCl3) δ (ppm) ) 7.15 (d, 1 H, J ) 13.6 Hz), 7.33
(d, 1 H, J ) 13.6 Hz), 7.36 (d, 2 H, J ) 4.8 Hz), 8.61 (d, 2 H, J ) 4.0
Hz). 13C NMR (75 MHz, CDCl3) δ (ppm) ) 111.3, 117.2, 121.0, 134.4,
137.8, 140.5, 143.7, 145.0, 150.3. 19F NMR (470.9 MHz, CDCl3 (versus
NaF in D2O)) δ (ppm) ) -42.01 (dt, 2 F), -33.92 (t, 1 F), -21.59
(dd, 2 F). MS m/z 272.4 ((M + H)+).
N,N’-Dimethoxy-4,4’-dipyridylium Bis-tetrafluoroborate (2). Fol-
lowing the general procedure, trimethyloxonium tetrafluoroborate (847
mg, 5.70 mmol) and 4,4’-dipyridyl N,N’-dioxide (511 mg, 2.49 mmol)
were stirred in 10 mL of acetonitrile for 4 h. Recrystallization from
1
acetonitrile afforded 746 mg (1.90 mmol, 76%) of 2. H NMR (300
MHz, CD3OD(+CD3CN)) δ (ppm) ) 4.55 (s, 6 H), 8.60-8.64 (m, 4
H), 9.43-9.45 (m, 4 H). 13C NMR (75 MHz, CD3OD(+CD3CN)) δ
(ppm) ) 69.0, 116.6, 127.8, 141.0. MS m/z 187.2 ((M - OCH3)+).
N-Methoxy-4-cyanopyridinium Tetrafluoroborate (3). Following
the general procedure, trimethyloxonium tetrafluoroborate (0.72 g, 4.87
mmol) and 4-cyanopyridine N-oxide (0.56 g, 4.63 mmol) were stirred
in 30 mL of dichloromethane for 4 h. Recrystallization from methanol
afforded 0.81 g (3.65 mmol, 79%) of 3. 1H NMR (300 MHz, DMSO-
d6) δ (ppm) ) 4.48 (s, 3 H), 8.83 (d, 2 H, J ) 6.6 Hz), 9.75 (d, 2 H,
J ) 6.6 Hz). 13C NMR (75 MHz, DMSO-d6) δ (ppm) ) 69.7, 114.6,
126.1, 132.6, 142.2. MS m/z 135.3 (M+).
N-Ethylphenanthridinium Hexafluorophosphate (6E). Following
the general procedure, triethyloxonium hexafluorophosphate (3.10 g,
11.3 mmol) and phenanthridine (1.98 g, 11.0 mmol) were stirred in 25
mL of dichloromethane for 4 h. Recrystallization from chloroform/
1
acetonitrile afforded 4.10 g (13.5 mmol, 76%) of 6E. H NMR (500
MHz, DMSO-d6) δ (ppm) ) 1.71 (t, 3 H, J ) 7.2 Hz), 5.13 (q, 2 H,
J ) 7.2 Hz), 8.07-8.16 (m, 3 H), 8.38 (t, 1 H, J ) 7.2 Hz), 8.56 (d,
1 H, J ) 8.7 Hz), 8.62 (d, 1 H, J ) 8.1 Hz), 9.10 (d, 1 H, J ) 8.7 Hz),
9.13-9.17 (m, 1 H), 10.29 (s, 1 H). 13C NMR (125 MHz, DMSO-d6)
δ (ppm) ) 14.7, 53.3, 119.8, 123.0, 123.7, 124.9, 125.8, 130.2, 132.0,
132.7, 132.9, 134.2, 137.8, 155.1. MS m/z 208.3 (M+).
N-Ethyl-4-cyanopyridinium Hexafluorophosphate (3E). Following
the general procedure, triethyloxonium hexafluorophosphate (0.65 g,
2.62 mmol) and 4-cyanopyridine (0.24 g, 2.27 mmol) were stirred in
30 mL of dichloromethane for 4 h. Recrystallization from methanol/
1
dichloromethane afforded 0.36 g (1.29 mmol, 57%) of 3E. H NMR
(300 MHz, DMSO-d6) δ (ppm) ) 1.55 (t, 3 H, J ) 7.2 Hz), 4.68 (q,
2 H, J ) 7.2 Hz), 8.69 (d, 2 H, J ) 6.6 Hz), 9.35 (d, 2 H, J ) 6.6 Hz).
13C NMR (75 MHz, DMSO-d6) δ (ppm) ) 16.0, 57.5, 114.8, 126.9,
130.9, 146.0. MS m/z 133.2 (M+).
N-Methoxy-2-styrylpyridinium Tetrafluoroborate (7). Following
the general procedure, trimethyloxonium tetrafluoroborate (1.07 g, 7.30
mmol) and 2-styrylpyridine N-oxide (1.20 g, 6.08 mmol) were stirred
in 30 mL of dichloromethane for 4 h. Recrystallization from methanol
afforded 1.35 g (4.51 mmol, 74%) of 7. 1H NMR (300 MHz, CD3OD
(+CD3CN)) δ (ppm) ) 4.40 (s, 3 H), 7.49-7.52 (m, 3 H), 7.59 (d, 1
H, J ) 14.4 Hz), 7.81-7.85 (m, 2 H), 7.91 (dt, 1 H, J ) 1.5, 7.2 Hz),
8.06 (d, 1 H, J ) 14.4 Hz), 8.44 (dt, 1 H, J ) 1.8, 7.2 Hz,), 8.53 (dd,
1 H, J ) 1.8, 8.4 Hz), 9.03 (dd, 1 H, J ) 1.5, 6.6 Hz). 13C NMR (75
MHz, CD3OD (+CD3CN)) δ (ppm) ) 70.1, 113.9, 126.9, 127.6, 129.8,
130.3, 132.5, 135.8, 141.7, 145.0, 146.7, 151.6. MS m/z 212.6 (M+).
N-Ethyl-2-styrylpyridinium Hexafluorophosphate (7E). Following
the general procedure, triethyloxonium hexafluorophosphate (2.46 g,
9.93 mmol) and 2-styrylpyridine (1.50 g, 8.28 mmol) were stirred in
40 mL of dichloromethane for 4 h. Recrystallization from methanol/
N-Methoxyquinolinium Tetrafluoroborate (4). Following the
general procedure, trimethyloxonium tetrafluoroborate (4.09 g, 27.7
mmol) and quinoline N-oxide (3.65 g, 25.1 mmol) were stirred in 110
mL of dichloromethane for 4 h. Recrystallization from methanol/
1
dichloromethane afforded 3.80 g (15.4 mmol, 61%) of 4. H NMR
(300 MHz, CD3OD) δ (ppm) ) 4.60 (s, 3 H), 8.07-8.18 (m, 2 H),
8.34 (t, 1 H, J ) 7.8 Hz), 8.48 (d, 1 H, J ) 8.1 Hz), 8.59 (d, 1 H, J
) 8.7 Hz), 9.23 (d, 1 H, J ) 8.1 Hz), 9.76 (d, 1 H, J ) 6.6 Hz). 13C
NMR (75 MHz, CD3OD) δ (ppm) ) 70.5, 117.3, 123.3, 131.5, 132.3,
132.7, 137.7, 138.2, 145.1, 148.0. MS m/z 160.3 (M+).
N-Ethylquinolinium Hexafluorophosphate (4E). Following the
general procedure, triethyloxonium hexafluorophosphate (9.30 g, 37.0
mmol) and quinoline (4.00 g, 31.0 mmol) were stirred in 60 mL of
dichloromethane for 4 h. Recrystallization from methanol/dichloro-
methane afforded 6.62 g (21.8 mmol, 70%) of 4E. 1H NMR (300 MHz,
CDCl3) δ (ppm) ) 1.73 (t, 3 H, J ) 7.2 Hz), 5.15 (q, 2 H, J ) 7.2
Hz), 7.99 (dd, 1 H), 8.10 (dd, 1 H), 8.26 (dt, 1 H), 8.40 (d, 1 H), 8.56
1
dichloromethane afforded 1.90 g (12.4 mmol, 65%) of 7E. H NMR
(300 MHz, CD3CN) δ (ppm) ) 1.59 (t, 3 H, J ) 7.8 Hz), 4.75 (q, 2
H, J ) 7.8 Hz), 7.45-7.52 (m, 4 H), 7.78-7.88 (m, 4 H), 8.35-8.46
(m, 2 H), 8.73 (d, 1 H, J ) 6 Hz). 13C NMR (75 MHz, CD3CN) δ
(ppm) ) 15.8, 55.1, 117.7, 126.9, 127.4, 129.6, 130.2, 132.0, 136.0,
145.5, 145.8, 145.9, 153.9. MS m/z 210.3 (M+).
9
15236 J. AM. CHEM. SOC. VOL. 124, NO. 51, 2002

Kinetics of Reductive N−O Bond Fragmentation
A R T I C L E S
N-Methoxy-2-(4-chlorostyryl)pyridinium Tetrafluoroborate (8).
Following the general procedure, trimethyloxonium tetrafluoroborate
(0.95 g, 6.45 mmol) and 2-(4-chlorostyryl)pyridine N-oxide (1.30 g,
5.61 mmol) were stirred in 30 mL of dichloromethane for 4 h.
Recrystallization from methanol afforded 1.23 g (3.69 mmol, 66%) of
H, J ) 7.2 Hz). 13C NMR (75 MHz, CD3CN) δ (ppm) ) 70.6, 125.1,
126.7, 127.3, 129.9, 139.9, 141.3, 142.0, 149.5, 153.2. MS m/z 257.1
(M+).
N-Methoxy-4-benzoylpyridinium Tetrafluoroborate (13) Follow-
ing the general procedure, trimethyloxonium tetrafluoroborate (2.19 g,
14.8 mmol) and 4-benzoylpyridine N-oxide (2.56 g, 12.6 mmol) were
stirred in 50 mL of dichloromethane for 4 h. Recrystallization from
methanol/dichloromethane afforded 3.41 g (11.3 mmol, 90%) of 13.
1H NMR (300 MHz, CD3CN) δ (ppm) ) 4.49 (s, 3 H), 7.57-7.62 (m,
2 H), 7.73-7.85 (m, 3 H), 8.26 (dd, 2 H, J ) 1.8, 5.1 Hz), 9.14 (dd,
2 H, J ) 1.5, 4.8 Hz). 13C NMR (75 MHz, CD3CN) δ (ppm) ) 70.9,
129.6, 129.9, 131.1, 135.0, 135.8, 142.3, 152.6, 192.1. MS m/z 214.2
(M+).
1
8. H NMR (300 MHz, CD3OD) δ (ppm) ) 4.39 (s, 3 H), 7.45-7.50
(m, 2 H), 7.56 (d, 1 H, J ) 15.9 Hz), 7.75-7.82 (m, 2 H), 8.00 (d, 1
H, J ) 15.9 Hz), 8.30-8.52 (m, 2 H), 8.68 (d, 1 H, J ) 6 Hz), 9.03
(d, 1 H, J ) 6.9 Hz). 13C NMR (75 MHz, CD3OD) δ (ppm) ) 68.3,
112.8, 116.6, 125.2, 126.0, 128.5, 129.4, 132.7, 139.9, 143.3, 144.0,
149.5. MS m/z, 246.4, 248.4 (M+).
N-Ethyl-2-(4-chlorostyryl)pyridinium Hexafluorophosphate (8E).
Following the general procedure, triethyloxonium hexafluorophosphate
(217 mg, 0.88 mmol) and 2-(4-chlorostyryl)pyridine (156 mg, 0.72
mmol) were stirred in 5 mL of dichloromethane for 4 h. Recrystalli-
zation from methanol/dichloromethane afforded 215 mg (0.55 mmol,
76%) of 8E. 1H NMR (300 MHz, DMSO-d6) δ (ppm) ) 1.60 (t, 3 H,
J ) 7.8 Hz), 4.76 (q, 2 H, J ) 7.8 Hz), 7.51 (d, 1 H, J ) 15.9 Hz),
7.50-7.52 (m, 2 H), 7.78 (d, 1 H, J ) 15.9 Hz), (d, 1 H, J ) 15.9
Hz), (d, 1 H, J ) 15.9 Hz), 7.78-7.81 (m, 2 H), 7.89 (dt, 1 H, J )
1.2, 6.9 Hz), 8.38 (d, 1 H, J ) 7.2 Hz), 8.43-8.49 (m, 1 H), 9.35 (d,
1 H, J ) 6.3 Hz). 13C NMR (75 MHz, DMSO-d6) δ (ppm) ) 15.5,
53.3, 117.8, 125.8, 129.0, 130.2, 133.7, 135.2, 141.9, 144.5, 145.2,
151.4, 152.6. MS m/z 244.8, 246.8 (M+).
Methods. The nanosecond and microsecond time-resolved absorption
measurements were performed using a nanosecond transient absorption
spectrometer of conventional design.35 The excitation source was an
Opotek Vibrant OPO laser. Typical specifications were pulse width of
ca. 4 ns, variable wavelength from 385 to 410 nm and energies in the
range 0.5-2 mJ per pulse. The optical densities of the solutions at the
excitation wavelengths were adjusted to be ca. 1.0, or as close to 1.0
as possible in a 1-cm path length.
Measurements on the picosecond time scale were also performed
using an apparatus of conventional design.35 The source was a
Continuum custom model PY-61C Nd:YAG laser. The specifications
at 1064 nm were typically 30 mJ and ca. 6 ps. Frequency tripled 355-
nm light was used to excite the sample, either directly or after frequency
shifting to 395 or 445 nm using stimulated Raman scattering in
cyclohexane. Residual 1064-nm light was used to generate the quasi-
continuum white light pulse. The detector was a Princeton Instruments
dual diode array detector, controlled by a Princeton Instruments model
ST-121S controller. The white light was dispersed using an Acton
Research Corp. Spectropro-300i spectrograph. At each setting of an
optical delay line, the entire spectrum was recorded from ca. 400 to
800 nm. Integral regions were chosen from these spectra and plotted
versus time to give the kinetic data. Experimental details, including
excitation and observation wavelengths are provided as Supporting
Information.
N-Methoxy-4-(2,3,4,5,6-pentafluorostyryl)pyridinium Tetrafluo-
roborate (9). Following the general procedure, trimethyloxonium
tetrafluoroborate (199 mg, 1.35 mmol) and 4-(2,3,4,5,6-pentafluoro-
styryl)pyridine N-oxide (322 mg, 1.12 mmol) were stirred in 40 mL of
dichloromethane for 4 h. 310 mg (0.80 mmol, 71%) of 9 was obtained
1
as a colorless oil. H NMR (400 MHz, CD3CN) δ (ppm) ) 4.39 (s, 3
H), 7.62 (s, 2 H), 8.22 (d, 2 H), 8.88 (dd, 2 H). 13C NMR (75 MHz,
CD3CN) δ (ppm) ) 70.8, 126.5, 127.2, 128.6, 131.2, 139.1, 141.6,
142.9, 146.7, 153.4. 19F NMR (470.9 MHz, CDCl3 (versus NaF in D2O))
δ (ppm) ) -38.5 (dt, 2 F), -28.1 (t, 1 F), -25.9 (s, 3 F), -25.2 (s,
1 F), -15.9 (dd, 2 F).
N-Methoxy-4-styrylpyridinium Tetrafluoroborate (10). Following
the general procedure, trimethyloxonium tetrafluoroborate (1.25 g, 8.52
mmol) and 4-styrylpyridine N-oxide (1.40 g, 7.10 mmol) were stirred
in 30 mL of dichloromethane for 4 h. Recrystallization from methanol
afforded 1.26 g (4.21 mmol, 59%) of 10. 1H NMR (300 MHz, CD3CN)
δ (ppm) ) 4.35 (s, 3 H), 7.36 (d, 1 H, J ) 16.5 Hz), 7.45-7.47 (m,
3 H), 7.69-7.72 (m, 2 H), 7.81 (d, 1 H, J ) 16.5 Hz), 8.08 (d, 2 H,
J ) 7.2 Hz), 8.76 (d, 2 H, J ) 7.2 Hz). 13C NMR (75 MHz, CD3CN)
δ (ppm) ) 70.7, 123.5, 126.0, 129.3, 130.2, 131.9, 136.0, 141.1, 143.2,
154.8. MS m/z 212.4 (M+).
The measurements of the complex of 1 with p-trianisylamine on
the subpicosecond time scale were performed using an apparatus that
has been previously described.36 The excitation wavelength used was
420 nm. The kinetics of the geminate pairs were determined by plotting
the integrated absorbance of the p-trianisylamine band over the range
659-797 nm as a function of time. This integral was used to account
for time-dependent spectral shifts in the absorption due to dynamic
solvent relaxation.37
Errors in the various rate constants were estimated mainly from
repeated measurements. On this basis, the errors in the activation
parameters for 2 and 12 are estimated to be ca. 15% for the log A
values, and ca. 5% for the Ea values. For the radical precursors included
in Tables 2 and 3, some experiments were repeated multiple times,
and others were repeated with more than one donor and in some cases
more than one solvent (detailed data not reported here). Different
statistical errors were computed for different radicals where enough
repeat experiments were performed. On this basis, an average error of
ca. 10% is estimated for all of the rate constants included in the Tables.
An exception is the radical from 13. In this case, the corresponding
N-ethyl compound was not available, and it was assumed that the value
of ksep from compound 10E was appropriate. On the basis of the
variation in the ksep values in Table 3, a maximum error of ca. 25% is
estimated for the rate constant in this case.
N-Ethyl-4-styrylpyridinium Hexafluorophosphate (10E). Follow-
ing the general procedure, triethyloxonium hexafluorophosphate (6.70
g, 26.7 mmol) and 4-styrylpyridine (4.00 g, 22.1 mmol) were stirred
in 100 mL of dichloromethane for 4 h. Recrystallization from methanol
afforded 5.10 g (14.3 mmol, 65%) of 10E. 1H NMR (300 MHz,
CD3OD) δ (ppm) ) 1.62 (t, 3 H, J ) 7.2 Hz), 4.55 (q, 2 H, J ) 7.2
Hz), 7.39 (d, 1 H, J ) 16.5 Hz), 7.42-7.46 (m, 3 H), 7.71-7.74 (m,
2 H), 7.88 (d, 1 H, J ) 16.5 Hz), 8.13 (d, 2 H, J ) 6.6 Hz), 8.73 (d,
2 H, J ) 6.6 Hz). 13C NMR (75 MHz, CD3OD) δ (ppm) ) 14.5, 55.3,
121.8, 123.4, 127.2, 127.4, 128.1, 129.7, 140.8, 142.8, 153.4. MS m/z
210.3 (M+).
N-Methoxy-4-(4-nitrostyryl)pyridinium Tetrafluoroborate (12).
Following the general procedure, trimethyloxonium tetrafluoroborate
(463 mg, 3.13 mmol) and 4-(4-nitrostyryl)pyridine N-oxide (690 mg,
2.85 mmol) were stirred in 20 mL of dichloromethane for 4 h.
Recrystallization from methanol afforded 610 mg (1.77 mmol, 62%)
of 12. 1H NMR (300 MHz, CD3CN) δ (ppm) ) 4.38 (s, 3 H), 7.52 (d,
1 H, J ) 16.5 Hz), 7.84 (d, 1 H, J ) 16.5 Hz), 7.88 (d, 2 H, J ) 8.7
Hz), 8.16 (d, 2 H, J ) 7.2 Hz), 8.26 (d, 2 H, J ) 8.7 Hz), 8.84 (d, 2
(35) Herkstroeter, W. G.; Gould, I. R. In Physical Methods of Chemistry Series,
2nd ed.; Rossiter, B., Baetzold, R., Eds.; Wiley: New York, 1993; Vol. 8,
p 225.
(36) Sakomura, M.; Lin, S.; Moore, T. A.; Moore, A. L.; Gust, D.; Fujihira, M.
J. Phys. Chem. A 2002, 106, 2218.
(37) Barbara, P. F.; Jarzeba, W. AdV. Photochem. 1990, 15, 1.
9
J. AM. CHEM. SOC. VOL. 124, NO. 51, 2002 15237

A R T I C L E S
Lorance et al.
The geometry-optimized radical minima of Figure 6 were calculated
with Gaussian 98 on a personal computer38 or on a University of Illinois,
Urbana-Champaign NCSA supercomputer.39 DFT was preferred over
basic Hartree-Fock calculations because reasonably high accuracy was
desired, but the molecules proved too large for MP2 calculations. Of
the various DFT methods, B3PW91 was selected since it has been
shown to be the best three-parameter hybrid method for minima,40 and
the hybrid methods exhibit superior performance as compared to the
pure DFT methods for most computational applications. The 6-31G
basis set41 augmented with one set of d diffuse functions42 and one set
of d polarization functions43 on the heavy atoms (6-31+G*) was used.
The nature of the stationary points was verified by frequency calcula-
tions.
For the three-dimensional potential energy surface of Figure 5, less
expensive UHF/6-31+G* calculations were used as implemented in
GAMESS (Version 26, October 2000, from Iowa State University).44
Spin-restricted open-shell Hartree-Fock was also attempted in the
region of the transition state (ROHF/6-31+G*), but while the results
were qualitatively similar, the wave function (as one would expect)
proved to be unstable near the transition state.
Acknowledgment. We thank Faye L. Stump for the synthesis
of compounds 4, 5, and 6. Support from Arizona State
University and The Research Corporation is gratefully acknowl-
edged. Some of the electronic structure calculations were
performed using NCSA Grants CHE010038N, CHE020011N.
(38) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M.
A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr., Stratmann,
R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin,
K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi,
R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.;
Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz,
J. V.; Baboul, A. G.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.;
Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham,
M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.;
Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.;
Gonzalez, C.; Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian
98, Rev. A.7; Gaussian, Inc.: Pittsburgh, PA, 1998.
(39) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M.
A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr., Stratmann,
R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin,
K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi,
R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Rega, N,; Salvador,
P.; Dannenberg, J. J.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.;
Foresman, J. B.; Cioslowski, J.; Ortiz, J. V.; Baboul, A. G.; Stefanov, B.
B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin,
R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara,
A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M.
W.; Andres, J. L.; Gonzalez, C.; Head-Gordon, M.; Replogle, E. S.; Pople,
J. A. Gaussian 98, Rev. A.11.2; Gaussian, Inc.: Pittsburgh, PA, 2002.
(40) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
Supporting Information Available: Experimental details of
the picosecond and femtosecond transient absorption spectra,
including CT complexes studied, excitation and observation
wavelengths, and raw experimental data (PDF). This material
is available free of charge via the Internet at http://pubs.acs.org.
JA020768E
(41) (a) Ditchfield, R.; Hehre, W. J.; Pople, J. A. J. Chem. Phys. 1971, 54, 724.
(b) Here, W. J.; Ditchfield, R.; Pople, J. A. J. Chem. Phys. 1972, 56, 2257.
(42) Clark, T.; Chandrasekhar, J.; Spitznagel, G. W.; Schleyer, P. v. R. J.
Comput. Chem. 1983, 4, 294.
(43) (a) Harihan, P. C.; Pople, J. A. Theor. Chim. Acta 1973, 28, 213. (b) Franci,
M. M.; Pietro, W. J.; Hehre, W. J.; Binkley, J. S.; Gordon, M. S.; DeFrees,
D. J.; Pople, J. A. J. Chem. Phys. 1982, 77, 3654.
(44) Schmidt, M. W.; Baldridge, K. K.; Boatz, J. A.; Elbert, S. T.; Gordon, M.
S.; Jensen, J. H.; Koseki, S.; Matsunaga, N.; Nguyen, K. A.; Su, S. J.;
Windus, T. L.; Dupuis, M.; Montgomery, J. A. J. Comput. Chem. 1993,
14, 1347.
9
15238 J. AM. CHEM. SOC. VOL. 124, NO. 51, 2002