Direct, catalytic monofluorination of sp3 c-h bonds: A radical-based mechanism with ionic selectivity

Article abstract of DOI:10.1021/ja505136j

Recently, our group unveiled a system in which an unusual interplay between copper(I) and Selectfluor effects mild, catalytic sp³ C-H fluorination. Herein, we report a detailed reaction mechanism based on exhaustive EPR, ¹⁹F NMR, UV-vis, electrochemical, kinetic, synthetic, and computational studies that, to our surprise, was revealed to be a radical chain mechanism in which copper acts as an initiator. Furthermore, we offer an explanation for the notable but curious preference for monofluorination by ascribing an ionic character to the transition state.

Full text of DOI:10.1021/ja505136j

Article
pubs.acs.org/JACS
Direct, Catalytic Monofluorination of sp3 C−H Bonds: A Radical-
Based Mechanism with Ionic Selectivity
Cody Ross Pitts,† Steven Bloom,† Ryan Woltornist,† Dillon Jay Auvenshine,† Lev R. Ryzhkov,‡
Maxime A. Siegler,† and Thomas Lectka*,†
†Department of Chemistry, Johns Hopkins University, 3400 North Charles Street, Baltimore, Maryland 21218, United States
‡Department of Chemistry, Towson University, 8000 York Road, Towson, Maryland 21252, United States
S
* Supporting Information
ABSTRACT: Recently, our group unveiled a system in which an unusual
interplay between copper(I) and Selectfluor effects mild, catalytic sp3 C−H
fluorination. Herein, we report a detailed reaction mechanism based on
exhaustive EPR, 19F NMR, UV−vis, electrochemical, kinetic, synthetic, and
computational studies that, to our surprise, was revealed to be a radical chain
mechanism in which copper acts as an initiator. Furthermore, we offer an
explanation for the notable but curious preference for monofluorination by
ascribing an ionic character to the transition state.
INTRODUCTION
■
Selective functionalization of sp3 C−H bonds represents an
area of invaluable and economical chemistry. The direct
formations of alcohols, alkenes, alkyl halides, and other
functional groups from inactivated C−H bonds are impressive,
seemingly effortless reactions accomplished by enzymes that are
often challenging to effect in a laboratory setting. However,
selective fluorination has proven an arduous undertaking for
both Nature and the synthetic chemist alike. Biologically, very
few fluorinase enzymes are known, and none of them operate
on the basis of direct C−H functionalization.1 Synthetically, a
Figure 1. Common “N−F” reagents.
conceivable radical fluorination method using hazardous and
difficult-to-use F2, similar to the well-established bromination
and chlorination reactions, is actually highly exothermic, which
causes great selectivity and safety concerns.2 For organofluorine
chemists, this issue and other existing challenges call for a more
innovative approach to C−H fluorination (Scheme 1).
Arguably one of the most significant developments in the
field of organofluorine chemistry was the advent of the N−F
reagents (containing a nitrogen−fluorine bond) intended as
mild sources of electrophilic fluorine in the late 1980s.3
Considering that these reagents were solid, stable, and effective
compounds, they quickly superseded the use of the high-energy
electrophilic fluorinating reagents such as fluorine gas, xenon
difluoride, perchloryl fluoride, and hypofluorites, making
fluorination reactions significantly more accessible to the
synthetic chemist.4 Among the top ranks of the N−F reagents
are N-fluorobenzenesulfonimide (NFSI), N-fluoropyridinium
salts (NFPy), and 1-chloromethyl-4-fluoro-1,4-
diazoniabicyclo[2.2.2]octane bis(tetrafluoroborate) (Select-
fluor), vide infra (Figure 1). These unique and versatile
compounds have proven their worth as reagents for
fluorofunctionalization as mediators and catalysts, but are also
ideal candidates for mechanistic studies.5
Scheme 1
Recent findings suggest that some of these so-called
“electrophilic” N−F reagents can also act as F atom transfer
Received: May 22, 2014
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja505136j | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
reagents. Sammis et al. have reported the ability of NFSI to
react with alkyl radicals,6 Baran et al. have suggested the ability
of Selectfluor to participate in single-electron transfer (SET)
chemistry and the homolytic cleavage of C−H bonds,7 and
within the last year both our laboratory and the Groves
laboratory have independently published methods on metal-
catalyzed sp3 C−H monofluorination. Where the Groves
system utilizes silver(I) fluoride (a nucleophilic fluorine source)
and iodosobenzene to generate a manganese(IV) fluoride
porphyrin catalyst in situ instead of an aforementioned N−F
reagent,8 our system, as will be shown in this full paper, relies
fundamentally on radical-based chemistry between Selectfluor
and a copper(I) promoter to effect both H atom abstraction
and subsequent installation of fluorine.9
Scheme 2
Groves’s and our work were among the first direct, catalytic
methodologies for the monofluorination of aliphatic substrates.
These discoveries prompted further investigations in our
laboratory, viz., (1) simplification of the conditions for our
originally fairly complex system, (2) exploration of the
chemistry of other redox-active transition metals with
Selectfluor,10 and especially (3) in-depth mechanistic studies
of the system(s) we devised. In this article, we propose a
detailed mechanism for the copper-initiated aliphatic fluorina-
tion method that is consistent with exhaustive EPR, 19F NMR,
UV−vis, electrochemical, kinetic, synthetic, and computational
studies. Furthermore, we offer a possible explanation for the
notable, useful, but curious preference for monofluorination.
The article is structured to present a logical narrative
whereby the mechanistic studies were conducted. With this
regard, the article is organized respectively as to (1) establish
the simplified protocol used for mechanistic analysis, (2)
discuss the experiments used to determine the role of copper as
an initiator, (3) examine the H atom abstraction/fluorination
steps of the mechanism (illuminating the involvement of radical
intermediates), (4) illustrate our conclusions drawn from
kinetic analyses, (5) propose a reasonable mechanism in accord
with all experimental observations, and (6) offer an explanation
for the observed selectivity of our reaction as a manifestation of
the “polar effect” by ascribing an ionic character to the H atom
abstraction transition state and, finally, subjecting the system to
computational analysis to confirm experimental results.
Simplified Protocol. Our original discovery combined
Selectfluor and transition metal catalysts (especially copper(I)
based complexes) in effecting direct aliphatic, benzylic, and, in
special cases, allylic monofluorination.9,10 However, the copper
system that focused on aliphatic fluorination, albeit intriguing, is
admittedly less practical for large-scale applications as it
involves the use of several additives. Thus, our immediate
goal was to establish a simplified protocol that is more
accessible, cost-effective, scalable, less time-sensitive, and easier
to subject to mechanistic studies. A logical approach was to
strip the system back down to the minimum number of
necessary components (i.e., Selectfluor, a copper salt,
acetonitrile) and address possible problems more directly.
Previously, we observed that our newly fluorinated substrates
were prone to ionization in situ over time, which led to a
decrease in product yields if the reactions were not quenched at
the appropriate time intervals. Perhaps this is attributable to a
gradual accumulation of hydrogen fluoride (HF) as a byproduct
of the reaction, which was observed by 19F NMR under our
published conditions. To prevent the buildup of HF, we
screened a variety of bases and noted that whereas amines tend
to impede the reaction altogether, 0.1 equiv of potassium
carbonate is often enough to effect the reaction and eliminate
any traces of HF by 19F NMR for 16−24 h. This small
modification allows us to let a variety of substrates stir at room
temperature for longer, generalized periods of time without
having vigilantly to monitor and optimize each one individually.
To our satisfaction, we also obtained comparable conversions
to monofluorinated products in the presence of potassium
carbonate. However, at this time we did not conclude anything
about the true role of the potassium carbonate in the system.
Hoping to circumvent the dependency on potassium
tetrakis(pentafluorophenyl)borate, N-hydroxyphthalimide, and
potassium iodide for higher yields, we decided to focus on
modifying the ligand. In the original system, we had the most
success with N,N′-bis(benzylidene)ethane-1,2-diamine. Making
minor modifications to the ligand scaffold, we quickly found a
substantial increase in percent conversions at room temperature
by using N,N′-bis(2,6-dichlorobenzylidene)ethane-1,2-diamine
instead.11 At this juncture in our laboratory, we have established
that a standard reaction using 2.2 equiv of Selectfluor, 0.1 equiv
of cuprous iodide, 0.1 equiv of the aforementioned ligand, and
0.1−1.0 equiv of potassium carbonate in MeCN under N2 at
room temperature overnight was a suitable, generalized
protocol for aliphatic and benzylic monofluorination (Scheme
2). Under these conditions, the reaction has also proven
amenable to gram-scale synthesis of monofluorinated products
(e.g., 1-fluorocyclododecane was obtained in 50% yield after 8
h). Using this simplified protocol, we sought to address the
most fundamental concerns surrounding the reaction mecha-
nism, i.e., the role of copper, how the fluorine atom is installed,
how the reaction kinetics behave, and the preference for
monofluorination.
Loss of Fluoride from Copper(I)−Selectfluor Inter-
action. Intuitively, copper can be either a species actively
involved in the catalytic cycle or an initiator to the reaction.
With these potential roles in mind, a large array of experiments
were designed to probe the behavior of copper over the course
of the reaction.
Considering that the minimum necessary components to
effect sp3 C−H fluorination are simply Selectfluor and
copper(I), we first studied their interaction by NMR. A 19F
B
dx.doi.org/10.1021/ja505136j | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
NMR spectrum of Selectfluor in CD3CN displays an N−F
signal at +47.1 ppm and a BF4 signal at −152.1 ppm, relative to
3-chlorobenzotrifluoride.12 A spectrum of a 1:1 mixture of
Selectfluor and cuprous iodide in CD3CN, taken after 45 min of
stirring, displays a BF4 signal at −152.4 ppm and the standard
peak. No N−F fluorine signal is observed at +47.1 ppm, nor are
any additional signals from +400 ppm to −300 ppm present.
Preliminary EPR experiments reveal the formation of a
copper(II) species, but no Cu−F coupling is observed at room
temperature, as well. So where did the fluorine atom go? The
most logical scenario is the formation of a copper fluoride
species that is undetectable by 19F NMR due to extreme signal
broadening induced by the paramagnetic copper(II) center
(unlikely), formation of a copper(II) bifluoride exhibiting
fluxional behavior in solution,13 or the fact that, after rapid
solvolysis, it exists as a solvent separated ion pair.14 Attempts
were made to “freeze out” a copper(II) bifluoride signal at −10
°C and −40 °C, but no evidence for this type of species or any
other signal was seen. Notably, a simple 19F NMR of cupric
fluoride in MeCN supports the notion of solvent separation: no
fluorine signal is observed.
To rule out the possibility of a copper fluoride formed in situ
being the key player for H atom abstraction and subsequent
installation of fluorine, several control experiments were run
using preformed copper fluorides (cupric fluoride and
(PPh3)3CuF·2MeOH)15 in the absence of Selectfluor.16
Although these experiments provide no evidence for/against
a copper fluoride as the source of fluorine during the
fluorination step of the mechanism,17 they do help confirm
that an interaction between copper and Selectfluor is necessary
to generate the species responsible for effecting H atom
abstraction.18
Figure 3. Flat-cell liquid phase spectra of copper(II) over time.
at 426 nm, which disappears in the absence of ligand and is
conceivably a charge-transfer band from a copper−ligand
interaction. The decreasing absorbance at 456 nm was
determined to result from an interaction between iodide and
Selectfluor: this absorbance was duplicated when taking a UV−
vis spectrum upon mixing Selectfluor with tetrabutylammonium
iodide (note that the interaction between iodide and Selectfluor
alone will not effect the fluorination reaction; copper is
necessary). Interestingly, when the reaction was run in a cuvette
under standard conditions (in the presence of substrate), the
spectrum obtained was virtually identical. Furthermore, a UV−
vis spectrum taken after several hours still shows a strong
copper(II) absorbance.
X-Band CW EPR Flat-Cell Experiments. The formation of a
paramagnetic copper(II) species presents an opportunity for
analysis via EPR spectroscopy. For liquid phase EPR experi-
ments, a flat cell was used in place of a cylindrical sample
configuration in order to minimize the absorption of micro-
waves by the solvent.19 The copper(II) spectra of reaction
conditions with and without a substrate present consist of four
hyperfine lines (from copper; I = 3/2) of unequal intensities
that grow in and persist over time. Subsequent observation of a
reaction in the absence of a substrate over time revealed gradual
shifts in intensities and resonances (Figure 3). This could
indicate a change in geometry or ligand environment of the
original copper(II) species formed. For better clarification, we
turned to solid-state EPR.
Figure 2. UV−vis spectra of CuI, ligand, and Selectfluor.
UV−Vis and EPR Analyses Indicate Copper(II) Species.
This copper(I)−Selectfluor interplay may best be elucidated by
direct observation of copper. Formation of a copper(II) species
was recognized early on in the investigation by UV−vis and
EPR analyses, and was subsequently studied intently.
Figure 4. Isotopically enriched ligand for solid state EPR.
UV−Vis Spectroscopy. UV−vis spectroscopy was used to
monitor changes in the copper species early in the reaction (ca.
t = 5 min to t = 15 min displayed in Figure 2). Figure 2 displays
visible bands at 426, 456, and 692 nm upon the addition of
cuprous iodide and our bis(imine) ligand to Selectfluor in
MeCN under N2. The broad band at 692 nm, a new copper(II)
absorbance, grows in concomitantly with the sharp absorbance
Solid-State X-Band CW EPR. The added complexity of solid-
state EPR spectra due to anisotropic effects can illuminate
details about the geometry of a complex, symmetry, and the
nature of any neighboring atoms.20 In an attempt to achieve
optimal resolution, spectra were collected at 8 K using
isotopically enriched 63CuI and 15N-labeled ligand (Figure
4).21 To our knowledge, this is the best approach to determine
C
dx.doi.org/10.1021/ja505136j | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 6. Solid-state spectra of copper(II) after 180 min with (C1)
and without (C2) substrate present at 8 K.
Figure 5. Solid-state spectra of copper(II) in the absence of a substrate
at 8 K.
Scheme 3
Table 1. EPR Parameters for Complexes in Figures 5 and 6
complex
g∥
g⊥
A∥ (G)
A⊥ (G)
giso
Aiso (G)
C1
2.27
2.30
2.28
2.04
2.12
2.07
170
177
167
12.5
16.5
14.0
2.12
2.18
2.14
65
70
65
C2a
C2b
definitively whether a direct Cu−F interaction is characteristic
of the copper species at any point in the reaction.
Solid-state spectra of the reaction in the absence of a
substrate display an interesting feature. An equilibrium of two
copper(II) species is well resolved in a spectrum taken after 3 h
(Figure 5). The signatures indicate that both species are
monomeric, solely surrounded by nitrogen-containing ligands,
and tetragonal in coordination geometry (g|| > g⊥ > ge; see
Table 1).22 Although it is tempting to mistake the separation of
the hyperfine resonances for each species as “splitting”, perhaps
due to a Cu−F interaction, none is observed: these are two
separate copper complexes that both lack coupling to fluorine.
Regarding the implausibility of a Cu−F interaction, Weltner et
al. reported a hyperfine coupling constant of A(19F) = 115 G
derived from EPR spectra of cupric fluoride at 4 K in argon and
neon matrices, which is significantly higher than any supposed
splitting observed in these complexes, but may not be the most
appropriate comparison.23 In another scenario, by exposing
ceruloplasmin to 15 equiv of fluoride, Gray et al. reported
A(19F) = 40 G for a cupric fluoride,24 which seems on par with
the separation between our observed hyperfine resonances. Yet,
the additional g3 resonance that appears in our spectra shatters
the appeal of perceiving this as Cu−F coupling and solidifies
the notion of two separate copper(II) complexes.25
In the presence of substrate (under standard reaction
conditions), something even more interesting is observed: the
presence of only one of the two copper(II) species (Figure 6).
This is likely an issue of dynamic ligand activity between the
putative complexes 1 and 2 (Scheme 3). A higher
concentration of an additional amine ligand 3 (Selectfluor
minus F+) is formed under reaction conditions, which shifts the
equilibrium preferentially toward only one of the copper(II)
species.
would be a suitable ligand for copper(II). If an alkane substrate
is not present, the formation of 5 is significantly slower, the
concentration of the amine significantly lower, and thus, there
is a mixture of amine-ligated copper(II) 2 and non-amine-
ligated copper(II) 1. This is consistent with the EPR
parameters for the complexes (Table 1), which indicate that
both copper species are surrounded solely by nitrogen-
containing ligands. Under any circumstance, there is no
observed Cu−F interaction, characteristic of a copper(II)
bifluoride or otherwise. It is crucial to highlight that this by no
means rules out the possibility of a solvent separated copper(II)
fluoride being formed as a product of the reaction, which can be
inferred reasonably from our NMR experiments.
Lastly, hoping for more clarification, several attempts were
made to grow single crystals suitable for X-ray structure
determination of the unoxidized copper(I)−bis(imine) com-
plex and the oxidized copper(II) species observed by EPR. In
the former scenario, an interesting polymeric structure was
obtained exhibiting 2:1 cuprous iodide:bis(imine) ligand
stoichiometry. However, this polymer is likely just a
thermodynamic sink for the copper(I):bis(imine) ligand
interaction and does not play an active role in the chemistry;
In the catalytic cycle we ultimately propose, radical dication 4
abstracts a hydrogen atom from an alkane to form ammonium
salt 5, which would easily be deprotonated in the presence of
potassium carbonate (Scheme 3). The corresponding amine 3
D
dx.doi.org/10.1021/ja505136j | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
EPR signatures of the copper(II) species observed over the
course of the reaction do not resemble those of dimeric or
polymeric copper species.26 In the latter scenario, any attempt
to grow crystals of the oxidized copper species (in the presence
of Selectfluor) only afforded the ammonium salt 5, H-TEDA-
BF4, previously reported by the Baran group.7
unsuccessful in reducing Selectfluor while producing any
detectable fluorinated products. Third of all, in the absence
of base (i.e., potassium carbonate), we should be able to detect
an initial burst of HF by 19F NMR at room temperature, but
this was not observed. Lastly, a differential pulse voltammogram
(DPV) of a 1:1 mixture of copper:bis(imine) ligand reveals an
oxidation potential of +0.87 V vs SCE for the copper(II/I)
transition; however, the reported reduction potential of
Selectfluor, −0.296 V vs AgRE,30 would suggest an unfavorable
flow of electrons by an outer-sphere electron transfer
mechanism and further aid in the nullification of this type of
process. Thus, an inner-sphere mechanism whereby radical
dication 4 is formed may be the more likely of the two.
Inner-Sphere SET. Still, a more convincing argument would
be to show an example where the reaction proceeds through
another inner-sphere electron transfer event. Thus, we
examined an initiator that cannot fathomably form radical
dication 4 through an “outer-sphere” process accompanied by
loss of fluoride: a primary alkyl radical. The formation of ethyl
radicals in situ is well established upon reaction of
triethylborane with oxygen.31 Applying this chemistry to our
system, an ethyl radical could reasonably form 4 and
fluoroethane upon interaction with Selectfluor (Scheme 5).
Initiation by Single-Electron Transfer. Evidence of a
rapid growth and persistence of copper(II) over the course of
the reaction was observed in the liquid phase EPR studies,
whereby copper(II) is formed rapidly over the first hour of the
reaction (∼85% conversion from copper(I)) and asymptotically
approaches 100% conversion thereafter.27 It is very possible
that the copper species plays a laissez-faire role beyond
initiating the reaction and generating an unstable Selectfluor
derivative that serves as the H atom abstractor and propagator
in the reaction mechanism. Taking into account previous
observations by both our laboratory and the Baran laboratory,
we explored the supposed SET chemistry between copper and
Selectfluor. There are two potential scenarios to consider under
the reaction conditions, resembling either an outer-sphere or
inner-sphere electron transfer mechanism (Scheme 4).28
Scheme 4
Scheme 5
To our satisfaction, adding a catalytic amount of triethylborane
to a solution of Selectfluor and cyclododecane in MeCN, with
no measures taken to remove O2, resulted in the formation of
1-fluorocyclododecane in 50% yield after 4 h. The involvement
of ethyl radicals in initiating the reaction is supported by
detection of fluoroethane by 19F NMR. Furthermore, a few
other synthetic methods have been published since our original
copper system that effect an analogous fluorination reaction
using catalytic amounts of iron,10 vanadium,32 and organic-
based reagents33 that conceivably participate in inner-sphere
electron transfer chemistry with Selectfluor. (Note that other
methods have also been published recently using photocatalysts
that likely operate under much different initiation mecha-
nisms.34)
Outer-Sphere SET. In the instance of an outer-sphere
mechanism, the copper species and Selectfluor 6 would remain
separate and otherwise unchanged throughout the course of an
event where copper(I) transfers an electron to Selectfluor,
generating copper(II) and Selectfluor radical cation 7. One
could draw out a mechanism where radical cation 7 performs H
atom abstraction, forming HF and an alkyl radical, and the
newly formed alkyl radical reacts with Selectfluor to generate a
fluorinated product and a radical dication species (4) that
would be responsible for subsequent H atom abstraction.
However, a few experimental findings discount this possibility.
First of all, if this outer-sphere mechanism holds true for
initiating the reaction, other known, highly competent outer-
sphere single-electron transfer reagents, such as ferrocene,
should be able to produce similar results upon reaction with
Selectfluor.29 Running the reaction with ferrocene instead of
cuprous iodide (despite the promising color change to dark
green, indicating formation of the ferrocenium ion) gave very
poor results, yielding only a trace amount of the desired
fluorinated product. Tris(bypyridine)ruthenium(II) also proved
incompetent in effecting the reaction. Second, a controlled
potential electrolysis experiment was attempted in the presence
of an electrolyte, Selectfluor, and cyclododecane, but was
Additional efforts were made to probe the role of copper as
an initiator by attempting to remove or sequester copper during
the course of the reaction and also suggest that the reaction
does not need copper to proceed beyond initiation (see
Supporting Information for details). Lastly, an experiment
probing the potential for asymmetric inductionusing a chiral
variant of our bis(imine) ligand (derived from trans-1,2-
cyclohexanediamine)35 and the Mosher ester of 3-phenyl-
propanol36 (as benzylic fluorination of this substrate establishes
spectroscopically distinct diastereomers by 19F NMR)37
E
dx.doi.org/10.1021/ja505136j | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
resulted in a distribution of fluorinated products that was
identical to the distribution when an achiral ligand was
employed. In a small way, this helps support the notion that
fluorine may not be transferred from a copper catalyst. All
things considered, the evidence overwhelmingly insinuates that
copper(I) is, in fact, an initiator in our system that operates
through an inner-sphere electron transfer mechanism with
Selectfluor, as opposed to being necessary throughout the
catalytic cycle.
dihydroanthracene) without any substantial amount of
fluorinated variants of the scavengers detected and (2) if
dihydroanthracene is added at any point after fluorinated
products start to appear by 19F NMR, the fluorination reaction
stops. These experiments strongly infer the shutting down of a
radical pathway. Note that oxygen also quenches the reaction,
typical of many radical chain reactions.
Although we have shown the ability to interrupt the
proposed radical pathway, these experiments do not necessarily
allude to the scavenging of alkyl radicals. In fact, the
aforementioned compounds and oxygen are likely to inhibit
the reaction via cessation of the radical dication. The best way
to probe the involvement of alkyl radicals is to run the reaction
with substrates that notoriously rearrange to provide more
stable radicals or release ring strain, such as those containing a
cyclopropyl moiety. The rates of rearrangement have been
studied for several “radical clocks” and, under certain
circumstances, allow the possibility of extrapolating rate
information from the reaction. We studied a small family of
cyclopropane-based radical clocks, spanning rearrangement
rates over a few orders of magnitude (Table 2).
As suggested in Scheme 4, copper(I) is used to generate what
we propose to be the true “catalyst” from Selectfluor: a radical
dication (4).38 Conceptually, if this radical dication acts as an H
atom abstractor, an alkyl radical would be generated that could
feasibly react with Selectfluor to form the fluorinated product
and regenerate the radical dication. This idea is akin to the
mechanism established by Corey and co-workers for the
Hoffman−Loffler−Freytag reaction (Scheme 6).39 Correspond-
̈
ingly, the next set of experiments discussed focus on probing
the involvement of radicals.
Scheme 6
Table 2. Radical Clocks
Involvement of Alkyl Radicals. The reaction was run in
the presence of four radical scavengers to explore the
involvement of radical intermediates: 2,2,6,6-tetramethylpiper-
idine 1-oxyl (TEMPO) 8, 2,6-di-tert-butyl-4-methylphenol
(BHT) 9, p-quinone 10, and dihydroanthracene 11 (Scheme
7).40 Subjecting cyclododecane to normal reaction conditions
Scheme 7
The first three radical clocks studiedbenzylcyclopropane,
thujone, and norcarane41showed evidence of fluorinated
product mixtures by 19F NMR, but no detectable amount of the
expected “rearranged” fluorinated products following the
putative formation of radicals 12, 13, and 14, respectively.42
However, the rate of fluorination may be significantly faster
than their rates of rearrangement, and the latter two clocks have
multiple competing sites for H atom abstraction that would not
allow for a rearranged product anyway. Accordingly, we
examined another slightly faster clock with one favorable
benzylic site for H atom abstraction under our reaction
conditions: 2-phenylbenzylcyclopropane (to form radical 15).43
with an added 1.2 equiv of each radical scavenger, the
formation of fluorocyclododecane was inhibited by 95% in
the presence of p-quinone, 97% with BHT, and completely in
the presence of either TEMPO or dihydroanthracene. One
potential criticism of these experiments may be that some of
these compounds do not solely act as radical scavengers; rather,
some will likely also be fluorinated or oxidized, consuming a
significant amount of Selectfluor, and thus inhibiting fluorina-
tion through another venue. To elucidate the primary role of
these compounds as radical inhibitors, we also found that (1)
merely 0.15 equiv of TEMPO and dihydroanthraceneleaving
a 15-fold excess of Selectfluoralso resulted in significant
reaction inhibition (85% with TEMPO and 70% with
A
19F NMR analysis revealed that the reaction yielded four
fluorinated products in a total yield of 18.2%: one of these
signals corresponds to the (E)-isomer of rearranged product 16
(δ = −172.53 ppm, ddd, J = 47.4, 24.8, 16.5 Hz), and another
signal also has the characteristics of an “opened” fluorinated
clock (δ = −178.69 ppm, ddd, J = 48.5, 39.2, 14.4 Hz).44 The
two additional signals have slightly more difficult splitting to
decipher, but have chemical shifts that reasonably match up
with two benzylic fluorinated isomers that contain an intact
cyclopropane ring (δ = −179.81 ppm, m and δ = −185.33 ppm,
m). The identification of these compounds is also supported by
F
dx.doi.org/10.1021/ja505136j | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a crude GC/MS analysis where four similar fragmentation
patterns were found with m/z = 226.3. The ratio of total
rearranged products to intact cyclopropane products is ca.
1:1.09. This rearrangement is strong evidence for a stepwise
fluorination mechanistic pathway and for the involvement of
short-lived alkyl radicals.
As an aside, the fact that the reported rates of rearrangement
for norcarane and 2-phenylbenzylcyclopropane are very similar,
yet we found no rearranged norcarane products, is a
noteworthy result. As either rearrangement or fluorination of
the radical happens after the rate-determining step (vide infra),
this observation indicates that secondary alkyl radicals
fluorinate faster than the more delocalized secondary benzylic
radicals in this reaction.
Induction Period. A mechanistic study would not be
complete without an analysis of reaction kinetics. A preliminary
kinetic study to monitor the rate of appearance of the
fluorinated product of 3-phenylpropyl acetate by 19F NMR
under standard reaction conditions revealed a significant
induction period before the desired 3-fluoro-3-phenylpropyl
acetate began to form. Over the course of our studies, we have
noted induction periods for this same compound varying
anywhere from 20 min to 2.5 h. We also found that the length
of this induction period can vary greatly among all substrates;
for instance, the induction periods for monitored reactions with
cyclodecane or cyclohexane have varied in length on the orders
of minutes to hours, just as 3-phenylpropyl acetate has. (A
sample plot of the rate of fluorination of cyclodecane is
provided in Figure 7, illustrating the induction period.)
Thus far, these experiments paint a reasonably convincing
picture where radical dication 4 generates an alkyl radical,
which may react homolytically with Selectfluor to yield a
fluorinated product and regenerate 4. One alternative to
consider is the role that carbocations may play in the
mechanism, as cationic intermediates may also result in the
opening of the cyclopropane ring. For example, can an alkyl
radical sacrifice another electron to a suitable acceptor and then
trap fluoride? There are a number of factors from theoretical
and experimental standpoints that militate against this
possibility. Most of all, we would be considering secondary
cations, whose free existence in solution is at the very least
unfavorable, and somewhat debatable.45 In any case, a
secondary cation in MeCN solvent would rapidly collapse to
the nitrilium as opposed to trapping fluoride. Nitrilium adducts
17rather, acetamides upon aqueous workupwere observed
by Baran and co-workers in a copper(II)−Selectfluor based
system.7 However, their postulated mechanism, involving a
copper(II) reagent that is subjected to harsher conditions in the
presence of Selectfluor, invokes formation of a precedented
copper(III) species that is much more likely to be reduced by
an alkyl radical than our observed copper(II) species (Scheme
8). The fact that nitrilium derived products are minimal in our
system (aside from ex post facto solvolysis) would seem to
indicate that cations play a minor role.
Figure 7. Sample rate of fluorination plot displaying induction period.
To determine whether the substrate itself plays a significant
role in the induction period of the reaction, we looked at the
consequences of “aging” the catalyst in six reactions set up in
parallel. In this experiment, 3-phenylpropyl acetate was added
at six different time intervals (t = 0, 15 min, 30 min, 1 h, 2 h,
and 4 h) into six different reaction flasks, and an aliquot was
taken from each flask at the 4.5 h mark. In every instance where
the starting material was added at/prior to 2 h, the percent
yields of the fluorinated products by 19F NMR relative to an
internal standard were virtually identical. However, in the
reaction where the starting material was added at 4 h, well past
any previously observed induction period, the fluorinated
product had already appeared after only 30 min of stirring, and
in half the percent yield of the other reactions. Thus, the
induction period does not appear to be substrate dependent.
We noticed shorter induction periods as technique improved,
presumably with respect to excluding oxygen from the system.
In fact, suspecting the involvement of radical species, we noted
that the reaction is greatly hindered in the presence of an O2
atmosphere and also found that the induction period is typically
shorter using degassed anhydrous MeCN (with N2) over
simply anhydrous MeCN (with no measures taken to remove
dissolved oxygen).46 If oxygen is quenching 4, then the origin
of the induction period is likely attributed to a slower buildup
in concentration of 4, the effective catalyst, in situ.47 Even after
rigorous efforts to exclude oxygen, a small concentration was
Scheme 8
What about direct formation of cations through hydride
transfer? Take the well-behaved substrate 1-hexyl acetate, which
fluorinates predominately in the 5-position, as a model. Hexyl
acetate should donate hydride preferentially from the 2-
position, as this would form, after anchimeric assistance, a
stable cyclic oxonium 18 that could trap fluoride (Scheme 9).
This product is not observed to any significant extent.
Scheme 9
G
dx.doi.org/10.1021/ja505136j | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
present in each reaction: the induction periods shortened
significantly, but never disappeared.
Rate Dependence. We next sought to determine the
overall order of the reaction using the method of initial rates;
however, it is very challenging if not impossible to obtain
quantitative rate dependencies for this reaction, given its
induction period and the limited solubility of several
components.
Our model thus far involves three steps: (1) an inner-sphere
SET event between Selectfluor and copper(I) generates
copper(II) and a radical dication; (2) this radical dication
performs H atom abstraction on an alkane, which generates an
ammonium salt and an alkyl radical; and (3) the resultant alkyl
radical abstracts a fluorine atom from Selectfluor, which
regenerates the radical dication to enter the catalytic cycle.
Since the radical dication is believed to be the true catalyst (or
chain carrier), then if H atom abstraction is the rate-limiting
step, the rate of product formation (studied by 19F NMR)
would likely have a first-order dependence on both the alkane
and the radical dication. Our data show that the rate of product
formation is, in fact, strictly first-order with respect to the
substrate.
Figure 8. Competitive KIE 19F NMR overlay of the formation of 3-
fluoro-3-phenylpropyl acetate (left, ddd, J = 47.4, 30.9, 14.4 Hz) and 3-
fluoro-3-phenylpropyl-3-d acetate (right, ddt, J = 30.9, 14.4, 7.2 Hz).
Scheme 10
The rate of radical cation formation is dependent on the
concentrations of copper(I) and Selectfluor, but the observed
induction period seriously complicates the picture. Qualita-
tively, the length of the induction period is inversely
proportional to the concentration of copper and proportional
to the concentration of oxygen. We also observed that
copper(I) is not entirely expended as the reaction rate
accelerates. The total concentration of radical cation, and
thus product, is dependent on a first-order term in Selectfluor
and a reciprocal first-order term (reflecting the production of
the radical dication). An accurate mathematical analysis of the
rate dependencies of Selectfluor and copper(I) is less feasible
under these circumstances, but qualitatively they should both
be <1 (depending on the relative contributions of the two
terms), which proved to be the case. The proposed rate
equation is illustrated in eq 1.
the rate-limiting step is, in fact, H atom abstraction.51
A
transition state calculation of the radical dication 4 engaging in
H atom abstraction at B3LYP/6-311++G** supports this
notion (d(C−H) = 1.17 Å, d(N−H) = 1.69 Å)). (In order to
simplify the calculation, the aliphatic substrate used was
propane. Counterions were included in an MeCN dielectric,
as otherwise without counterions present the barrier to H atom
transfer diminished to zero.)
A second competitive KIE experiment was also conducted
using a purely aliphatic substrate, viz., a 1:1 mixture of
cyclohexane (21):cyclohexane-d12 (22) to provide 23 and 24,
which provided a slightly smaller average value of 2.0 (Scheme
10). Similar to the 3-phenylpropyl acetate result, there is a
moderate primary isotope effect and small secondary effect
from the geminal deuterium atom. On the other hand,
cyclohexane-d12 has four vicinal deuterium atoms that have an
inverse secondary effect on the rate that accounts for a notable
diminution of the phenomenological KIE value.48
Proposed Mechanism. Based on experimental observa-
tions thus far, we can propose a reasonable mechanism. EPR,
UV−vis, 19F NMR, and several synthetic experiments point to
an inner-sphere SET reaction between copper and Selectfluor
whereby copper(I) is oxidized to copper(II) accompanying a
loss in fluoride from Selectfluor. As determined by the
aforementioned KIE experiments and transition state calcu-
lation, the resultant radical dication species from the SET
reaction 4 is a reasonable actor in H atom abstraction that
occurs through an early transition state and is postulated to be
rate-determining. Radical scavenger and radical clock experi-
ments confirm the involvement of alkyl radicals that would be
formed along with ammonium salt 5 (observed) upon H atom
abstraction. Furthermore, the notion that fluorine is being
transferred directly from Selectfluor is logical, as this would
regenerate the radical dication and complete a catalytic cycle/
KIE. Kinetic isotope effect (KIE) experiments are also
capable of providing a wealth of knowledge about a reaction
mechanism, from information about the rate-determining step
to intimate details about the nature of the transition state.48 An
appropriate benzylic substrate for this experiment would be 3-
phenylpropyl acetate, as it yields only one fluorinated product
(in the benzylic position) and the corresponding mono/
dideuterio species 19 is easily accessible.49 The appearance of
fluorinated products 20 was monitored by 19F NMR in a
competitive KIE experiment, as the deuterium-induced 19F
isotopic shift is significant enough to allow independent
observation of the geminal protio and deuterio products (Δδ
= 0.59 ppm; Figure 8).50 This method also obviates misleading
results from potential inconsistencies in induction periods.
Comparison of the initial rates revealed an average kinetic
isotope effect of 2.3, which is a superposition of a moderate
primary KIE and a secondary effect from the dideuterio species
(Scheme 10). This diminished putative primary KIE value
appears to be consistent with an early or bent transition state if
radical chain reaction similar to the Hoffman−Loffler−Freytag
̈
H
dx.doi.org/10.1021/ja505136j | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 9. Mechanistic hypothesis based on experimental results.
reaction (Figure 9). We have also provided an energetic profile
of the reaction intermediates in the catalytic cycle that
illustrates a largely exothermic reaction pathway (Figure 10).
distribution of 1-hexyl acetate. Fluorination of this substrate
predominates in the 5-position, yields of the other mono-
fluorinated isomers largely decrease moving down the chain,
and there are trace (if any) monofluorinated products in the 1-
position, 6-position, and α-position to the carbonyl. Compare
this to the outcome of a reaction using n-dodecane, where an
almost equal distribution of monofluorinated products on the
methylene sites is observed. It is clear that the reaction is
sensitive to substituent effects that will provide some potent
clues.
From one vantage point, as we propose a mechanism
involving a radical chain process, we conducted a computa-
tional experiment early on that interestingly suggested that the
observed distribution of n-fluoro-1-hexyl acetate isomers
correlates with the calculated relative stabilities of the
corresponding hexyl acetate radicals. If the selectivity of the
reaction is based solely on radical stability though, which is
characteristic of a purely covalent valence bond model for the
rate-determining H atom abstraction transition state,52 then
geminal difluorination should be favored. Also consider the
isodesmic analyses of cyclohexane and cyclodecane (Table 3)
a
Table 3. Isodesmic Reactions
Figure 10. Free-energy profile for the monofluorination of cyclo-
decane through our proposed catalytic cycle.
a
All geometry optimizations were performed at B3PW91/6-311+G**-
Overall, this picture appears to be a reasonable mechanism
for this system. However, perhaps the most difficult question to
answer pertaining to the selectivity of the reaction still remains:
why is monofluorination preferred? Finally, we turned our
attention to a more in-depth theoretical analysis to try to
complete the puzzle.
Role of Valence Bond “Ionicity” in Reaction Selectiv-
ity. One of the most enlightening features regarding the
selectivity of this reaction is in the highly reproducible product
(MeCN).
that indicate favorable formations of 1,1-difluorocyclohexane
and 1,1-difluorocyclodecane over monofluorination based on
thermodynamic considerations; yet, geminal difluorinated
products are not observed experimentally, except to a minor
extent when we apply forcing conditions (but even then,
ionization/trapping of acetonitrile is a more competitive
I
dx.doi.org/10.1021/ja505136j | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
process). The desire to analyze this reaction in terms of
generating the most stable radical, a bond dissociation energy
argument, is thus a misguided instinct.
Instead, if we revisit the substituent effect observed in 1-hexyl
acetate as an effect resembling that of a radical reaction with
ionic character in the transition state, then we can begin to
rationalize the selectivity. In this light, the deactivation of sp3
C−H sites proximal to an electron-withdrawing group toward
fluorination agrees nicely with our proposed mechanism. The
species that we suggest is responsible for H atom abstraction,
radical dication 4, is an electron deficient radical that would
much prefer interaction with the more electron rich C−H sites
(hence the starting material over the newly formed fluorinated
products).
Polar Effect. Ionic-like selectivity is not unheard of in radical
reactions; there are several accounts of this phenomenon in the
literature, first noted by Walling and Mayo53 in free radical
polymerization reactions and since referred to as “the polar
effect”. By analogy of our reaction to the Hoffman−Loffler−
̈
Freytag reaction, reports demonstrating that this polar effect,
putatively at play in our fluorination reaction, is similarly
observed in free radical chlorination54 and bromination55
reactions involving intermolecular H atom abstraction also by
amine radical cations make an extremely convincing argument
for our case. These reports also indicate an overwhelming
preference for the penultimate sp3 C−H site on n-alkyl esters,
which they attribute to such polar (and also minor steric)
effects.
Figure 11. Application of Donahue’s theory.
preference for monofluorination of cyclohexane, as well. For
cyclohexane, CP is calculated to be 3.4 kcal, which is a lower
barrier than that of fluorocyclohexane at 5.5 kcal.
The last piece of the puzzle lies in further examining the
effect of ionicity on the H atom abstraction transition states of
the alkane versus the monofluorinated product. Postulating the
role of the ionic potential energy surface on dictating selectivity
and given the complexity of transition state calculations, we first
turned to Donahue’s seminal ionic curve crossing theory as a
way to study the nature of the transition states: only geometry
optimization calculations are necessary by this analysis.56 This
theory indicates that the lowering in energy of the saddle point
on the ground state potential energy surface results from an
avoided curve crossing with the ionic potential energy surface.
Succinctly stated, lower ionic state energies correlate with lower
transition state energies. Boundary conditions for an avoided
curve crossing are derived from plotting the evolution of the
ground and ionic state energies as reactants approach each
other (bear in mind that, for radical cation abstraction
reactions, the ground state is ionic, as well). In our system,
ΔE1 is the calculated difference between ground and “ionic”
states of the reactants, ΔE2 is the same for the products, ΔEa is
the activation energy, ΔHREACT is the reaction enthalpy, and CP
is the potential energy surface crossing point (eq 2).
Additionally, we calculated the transition states for formation
of the isopropyl radical and the 2-fluoroisopropyl radical,
representing pruned substrates for ease of calculation. The
result is in excellent agreement with the curve crossing analysis,
vide supra, as the transition state for the formation of the
isopropyl radical is earlier and calculated to be 2.2 kcal lower
than for the formation of the 2-fluoroisopropyl radical at
B3LYP/6-311++G**. An NBO analysis also confirms that a
positive charge has developed in the transition state (relative to
an isoenergetic H atom abstraction) that is accentuated on the
hydrogen atom. A strong electron-withdrawing group such as
fluorine would destabilize this positive charge, advocating again
for H atom abstraction of an alkane over a fluoroalkane (Figure
12).
Finally, note that all attempts to calculate the transition state
whereby Selectfluor fluorinates the isopropyl radical repeatedly
collapsed to the products, potentially signifying a barrier-less
reaction.
CONCLUSIONS
■
ΔE1(ΔE1 + ΔHREACT
)
Through in-depth analysis of experimental and theoretical data,
we are able to propose a mechanistic scenario of the copper-
initiated sp3 C−H fluorination methodology. Spectroscopic
evidence and synthetic experiments confirm a radical chain
mechanism initiated by an inner-sphere SET from copper(I) to
Selectfluor (as opposed to a mechanism where copper plays a
role in the catalytic cycle), but this alone does not explain the
observed preference for monofluorination. Analyzing the
influence of the ionic potential energy surface and applying
Donahue’s ionic curve crossing theory has allowed us to offer a
reasonable explanation for the energetics and selectivity of the
reaction.
CP =
ΔE1 + ΔE2
(2)
For cyclodecane, CP is calculated to be 4.6 kcal, whereas
fluorocyclodecane, as a precursor to the more stable 1-
fluorocyclodecyl radical, leads to CP = 5.4 kcal (B3PW91/6-
311++G**/MeCN), implying a higher activation energy for its
formation, consistently accounting for the observed selectivity
from this reaction (Figure 11).
The calculations in Figure 11 include optimized geometries
of the 1-fluorocyclodecyl and cyclodecyl cations, both of which
are found to be hydrido-bridged employing the MeCN
continuum. This model is consistent in predicting the observed
J
dx.doi.org/10.1021/ja505136j | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(6) Rueda-Becerril, M.; Sazepin, C. C.; Leung, J. C. T.; Okbinoglu,
T.; Kennepohl, P.; Paquin, J.-F.; Sammis, G. M. J. Am. Chem. Soc.
2012, 134, 4026−4029.
(7) Michaudel, Q.; Thevenet, D.; Baran, P. S. J. Am. Chem. Soc. 2012,
134, 2547−2550.
(8) Liu, W.; Huang, X.; Cheng, M.-J.; Nielsen, R. J.; Goddard, W. A.,
III; Groves, J. T. Science 2012, 337, 1322−1325.
(9) Bloom, S.; Pitts, C. R.; Miller, D.; Haselton, N.; Holl, M. G.;
Urheim, E.; Lectka, T. Angew. Chem., Int. Ed. 2012, 51, 10580−10583.
(10) Bloom, S.; Pitts, C. R.; Woltornist, R.; Griswold, A.; Holl, M. G.;
Lectka, T. Org. Lett. 2013, 15, 1722−1724.
(11) Both N,N′-bis(benzylidene)ethane-1,2-diamine and N,N′-
bis(2,6-dichloro-benzylidene)ethane-1,2-diamine were synthesized
according to literature procedure. See: Liu, H.; Zhang, H.-L.; Wang,
S.-J.; Mi, A.-Q.; Jiang, Y.-Z.; Gong, L.-Z. Tetrahedron: Asymmetry 2005,
16, 2901−2907.
(12) Naumann, D.; Kischkewitz, J. J. Fluorine Chem. 1990, 47, 283−
299.
(13) Roe, C. D.; Marshall, W. J.; Davidson, F.; Soper, P. D.; Grushin,
V. V. Organometallics 2000, 19, 4575−4582.
(14) Baxter, A. C.; Cameron, J. H.; McAuley, A.; McLaren, F. M.;
Winfield, J. M. J. Fluorine Chem. 1977, 10, 289−298.
(15) For syntheses of stabilized copper(I) fluoride complexes, see:
(a) Jardine, F. H.; Rule, L.; Vohra, A. G. J. Chem. Soc. A 1970, 238−
240. (b) Gulliver, D. J.; Levason, W.; Webster, M. Inorg. Chim. Acta
1981, 52, 153−159.
Figure 12. Transition state calculations and charge distributions
alongside isoenergetic scenarios.
(16) Another control experiment was also designed to probe the
involvement of a copper(III) fluoride by applying a (2-pyridyl)
methylamine ligand to the system, which has been shown to promote
two-electron chemistry in copper(I) complexes, but no positive ligand
effects were observed. See: Osaka, T.; Karlin, K. D.; Itoh, S. Inorg.
Chem. 2005, 44, 410−415.
(17) Another control experiment conducted was a thermolysis of tert-
butyl 2-phenylpropaneperoxoate in the presence of cupric fluoride
under Sammis’s conditions (ref 6) that could conceivably illuminate
how an alkyl radical reacts with a copper(II) fluoride. No 1-
fluoroethylbenzene was observed; however, these conditions do not
directly mimic the reaction conditions.
ASSOCIATED CONTENT
* Supporting Information
General experimental procedures, kinetic data, spectral data,
crystallographic information, and computational files. This
material is available free of charge via the Internet at http://
pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Author
*lectka@jhu.edu
■
(18) Note that the reaction does not produce fluorinated products
under the reaction conditions with Selectfluor in the absence of copper
either; this control reaction was conducted.
Notes
The authors declare no competing financial interest.
(19) MeCN is a high dielectric solvent and makes for a “lossy”
sample, which can be overcome with a flat-cell. See: (a) Hyde, J. S. Rev.
Sci. Instrum. 1972, 43, 629−631. (b) Mett, R. R.; Hyde, J. S. J. Magn.
Reson. 2003, 165, 137−152. (c) Sidabras, J. W.; Mett, R. R.; Hyde, J. S.
J. Magn. Reson. 2005, 172, 333−341. (d) Eaton, S. S.; Eaton, G. R.
Anal. Chem. 1977, 49, 1277−1278.
ACKNOWLEDGMENTS
T.L. thanks the NSF (CHE 1152996) for support. C.R.P.
thanks Johns Hopkins for a Gary H. Posner Fellowship.
■
REFERENCES
■
(1) (a) Eustaq
́
uio, A. S.; O’Hagan, D.; Moore, B. S. J. Nat. Prod.
(20) Bennati, M.; Murphy, D. M. Electron Paramagnetic Resonance
Spectra in the Solid State. In Electron Paramagnetic Resonance: A
Practitioner’s Toolkit; Brustolon, M., Giamello, E., Eds.; John Wiley &
Sons, Inc.: Hoboken, NJ, 2009; pp 195−250.
2010, 73, 378−382. (b) O’Hagan, D.; Schaffrath, C.; Cobb, S. L.;
Hamilton, J. T. G.; Murphy, C. D. Nature 2002, 416, 276.
(2) The chain propagation steps in the radical fluorination of an
alkane using F2 (H atom abstraction and subsequent fluorination) have
an overall change in enthalpy of approximately −103 kcal/mol.
(3) For examples: (a) Differding, E.; Lang, R. W. Tetrahedron Lett.
1988, 29, 6087−6090. (b) Umemoto, T.; Kawada, K.; Tomita, K.
Tetrahedron Lett. 1986, 27, 4465−4468. (c) Davis, F. A.; Han, W.
Tetrahedron Lett. 1991, 32, 1631−1634. (d) Banks, R. E. J. Fluorine
Chem. 1998, 87, 1−17 and references cited therein.
(21) The natural abundance of 63Cu:65Cu is about 70:30 (I = 3/2 in
both instances) and 14N:15N is over 99:1, but 15N (I = 1/2) gives rise
to a simpler (doublet vs triplet for 14N), more pronounced
superhyperfine pattern (A(15N)/A(14N) = 1.4). For some applications,
see: (a) Yuan, H.; Collins, M. L. P.; Antholine, W. E. Biophys. J. 1999,
76, 2223−2229. (b) Lemos, S. S.; Collins, M. L. P.; Eaton, S. S.; Eaton,
G. R.; Antholine, W. E. Biophys. J. 2000, 79, 1085−1094.
(22) Majahan, M.; Saxena, K. N.; Saxena, C. P. J. Inorg. Nucl. Chem.
1981, 43, 2148−2152.
(4) Kirsch, P. Synthesis of Complex Organofluorine Compounds. In
Modern Fluoroorganic Chemistry: Synthesis, Reactivity, Applications;
Wiley-VCH Verlag GmbH & Co. KGaA: Weinheim, Germany, 2004;
pp 203−278 and references cited therein.
(5) For some examples of studies and applications of N-F reagents,
particularly Selectfluor, see: (a) Stavber, S.; Zupan, M. Acta Chim. Slov.
2005, 52, 13−26. (b) Stavber, S. Molecules 2011, 16, 6432−6464.
(c) Vincent, S. P.; Burkart, M. D.; Tsai, C.-Y.; Zhang, Z.; Wong, C.-H.
J. Org. Chem. 1999, 64, 5264−5279. (d) Oliver, E. W.; Evans, D. H. J.
Electroanal. Chem. 1999, 474, 1−8.
(23) Kasai, P. H.; Whipple, E. B.; Weltner, W. J. Chem. Phys. 1966,
44, 2581−2591.
(24) Dawson, J. H.; Dooley, D. M.; Gray, H. B. Proc. Natl. Acad. Sci.
U.S.A. 1978, 75, 4078−4081.
(25) Despite optimal conditions with pure isotopes, no additional
information on ligand binding could be obtained via EPR spectroscopy
without access to instrumentation capable of ENDOR.
K
dx.doi.org/10.1021/ja505136j | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(26) For instance: Moncol, J.; Mudra, M.; Lonnecke, P.; Hewitt, M.;
(46) For some examples of oxygen consumption causing induction
periods and retarding reaction rates in radical processes, see:
(a) Cunningham, M. F.; Geramita, K.; Ma, J. W. Polymer 2000, 41,
5385−5392. (b) Okubo, M., Ed. Polymer Particles (Advances in Polymer
Science); Springer-Verlag: Berlin Heidelberg: The Netherlands, 2010.
(47) For some examples of induction period dependences on catalyst
concentration, see: (a) Singh, U. K.; Strieter, E. R.; Blackmond, D. G.;
Buchwald, S. L. J. Am. Chem. Soc. 2002, 124, 14104−14114.
(b) Ishiyama, T.; Takagi, J.; Ishida, K.; Miyaura, N.; Anastasi, N. R.;
̈
Valko, M.; Morris, H.; Svorec, J.; Melnik, M.; Mazur, M.; Koman, M.
Inorg. Chim. Acta 2007, 360, 3213−3225.
(27) Determined by EDTA titration.
(28) For some of the original discussion of outer-sphere and inner-
sphere electron transfer mechanisms, see: (a) Marcus, R. A. J. Chem.
Phys. 1956, 24, 966−978. (b) Marcus, R. A. J. Chem. Phys. 1956, 24,
979−989. (c) Taube, H.; Myers, H.; Rich, R. L. J. Am. Chem. Soc.
1953, 75, 4118−4119.
(29) Clegg, A. D.; Rees, N. V.; Klymenko, O. V.; Coles, B. A.;
Compton, R. G. J. Electroanal. Chem. 2005, 580, 78−86.
(30) Oliver, E. W.; Evans, D. H. J. Electroanal. Chem. 1999, 474, 1−8.
(31) Ollivier, C.; Renaud, P. Chem. Rev. 2001, 101, 3415−3434.
(32) Xia, J.-B.; Ma, Y.; Chen, C. Org. Chem. Front. 2014, 1, 468−472.
(33) Amaoka, Y.; Nagatomo, M.; Inoue, M. Org. Lett. 2013, 15,
2160−2163.
(34) For examples: (a) Bloom, S.; Knippel, J. L.; Lectka, T. Chem. Sci.
2014, 5, 1175−1178. (b) Kee, C. W.; Chin, K. F.; Wong, M. W.; Tan,
C.-H. Chem. Commun. 2014, DOI: 10.1039/C4CC01848F. (c) Xia, J.-
B.; Zhu, C.; Chen, C. J. Am. Chem. Soc. 2013, 135, 17494−17500.
(35) For some applications of N,N′-bis(2,6-dichlorobenzylidene)
cyclohexane-1,2-diamine in asymmetric catalysis, see: (a) Evans, D. A.;
Lectka, T.; Miller, S. J. Tetrahedron Lett. 1993, 34, 7027−7030. (b) Li,
Z.; Conser, K. R.; Jacobsen, E. N. J. Am. Chem. Soc. 1993, 115, 5326−
5327. (c) Wu, J.; Chen, Y.; Panek, J. S. Org. Lett. 2010, 12, 2112−
2115.
Hartwig, J. F. J. Am. Chem. Soc. 2002, 124, 390−391. (c) Mar
́
ta, F.;
Boga, E.; Matok, M. Discuss. Faraday Soc. 1968, 46, 173−183.
́
(48) (a) Anslyn, E. V.; Dougherty, D. A. Modern Physical Organic
Chemistry; University Science Books: Sausalito, CA, 2006. (b) Giagou,
T.; Meyer, M. P. Chem.Eur. J. 2010, 16, 10616−10628.
(49) Kurita, T.; Hattori, K.; Seki, S.; Mizumoto, T.; Aoki, F.; Yamada,
Y.; Ikawa, K.; Maegawa, T.; Monguchi, Y.; Sajiki, H. Chem.Eur. J.
2008, 14, 664−673.
(50) For instance, see: Osten, H. J.; Jameson, C. J.; Craig, N. C. J.
Chem. Phys. 1985, 83, 5434−5441.
(51) (a) More O’Ferrall, R. A. J. Chem. Soc. B 1970, 785−790.
(b) Strong, H. L.; Brownawell, M. L.; San Filippo, J., Jr. J. Am. Chem.
Soc. 1983, 105, 6526−6528. (c) Westheimer, F. H. Chem. Rev. 1961,
61, 265−273.
(52) Hiberty, P. C.; Megret, C.; Song, L.; Wu, W.; Shaik, S. J. Am.
Chem. Soc. 2006, 128, 2836−2843.
(53) See: Walling, C. Free Radicals in Solution; Wiley: New York, NY,
1957; pp 132−140, 365−369, 375−376, 474−491.
(54) (a) Minisci, F.; Galli, R.; Galli, A.; Bernardi, R. Tetrahedron Lett.
1967, 23, 2207−2209. (b) Bernardi, R.; Galli, R.; Minisci, F. J. Chem.
Soc. B 1968, 324−325.
(55) Minisci, F.; Galli, R.; Bernardi, R. Chem. Commun. 1967, 903−
904.
(56) Donahue, N. M.; Clarke, J. S.; Anderson, J. G. J. Phys. Chem. A
1998, 102, 3923−3933.
(36) The Mosher ester, 3-phenylpropyl (S)-3,3,3-trifluoro-2-
methoxy-2-phenylpropanoate, was synthesized via standard DCC
coupling chemistry. See: Neises, B.; Steglich, W. Angew. Chem., Int. Ed.
1978, 17, 522−524.
(37) (a) Dale, J. A.; Mosher, H. S. J. Am. Chem. Soc. 1973, 95, 512−
519. (b) Gorkom, M. v.; Hall, G. E. Q. Rev., Chem. Soc. 1968, 22, 14−
29.
(38) Radical dication 3 was observed by UV−vis spectroscopy in a
spectroelectrochemical experiment (procedure described in the
Supporting Information) whereby the corresponding amine, 1-
(chloromethyl)-1,4-diazabicyclo[2.2.2]octan-1-ium tetrafluoroborate,
is oxidized in MeCN via controlled potential electrolysis. Over 1 h,
a new absorbance grew in at 273 nm. It is impossible to observe this
absorbance under the normal reaction conditions/concentrations.
(39) For example: Corey, E. J.; Hertler, W. R. J. Am. Chem. Soc. 1960,
82, 1657−1668.
(40) Save the experiment with dihydroanthracene, a similar study was
done in: Vincent, S. P.; Burkart, M. D.; Tsai, C.-Y.; Zhang, Z.; Wong,
C.-H. J. Org. Chem. 1999, 64, 5264−5279.
(41) Benzylcyclopropane and norcarane were synthesized according
to literature procedure. See: Simmons, H. E.; Smith, R. D. J. Am. Chem.
Soc. 1959, 81, 4256−4264.
(42) For respective rates of rearrangement of benzylcyclopropane,
thujone, and norcarane, see: (a) Bowry, V. W.; Ingold, K. U. J. Am.
Chem. Soc. 1991, 113, 5699−5707. (b) He, X.; Ortiz de Montellano, P.
R. J. Org. Chem. 2004, 69, 5684−5689. (c) Auclair, K.; Hu, Z.; Little,
D. M.; Ortiz de Montellano, P. R.; Groves, J. T. J. Am. Chem. Soc.
2002, 124, 6020−6027.
(43) 2-Phenylbenzylcyclopropane was synthesized according to
literature procedure. See: Aguila, M. J. B.; Badiei, Y. M.; Warren, T.
H. J. Am. Chem. Soc. 2013, 135, 9399−9406. For rate of
rearrangement, see: Hollis, R.; Hughes, L.; Bowry, V. W.; Ingold, K.
U. J. Org. Chem. 1992, 57, 4284−4287.
(44) Although it is difficult to isolate the low-yielding products under
the reaction conditions, we confirmed the presence of the (E)-isomer
of 16 in the reaction mixture by synthesizing each isomer by other
means and comparing 19F NMR shifts, splitting, and coupling
constants.
(45) Hydrido-bridged secondary cations are considered more stable,
viable intermediates in solution. For direct observation of this
phenomenon with secondary cycloalkane cations, see: Kirchen, R.
P.; Sorensen, T. S. J. Am. Chem. Soc. 1979, 101, 3240−3243.
L
dx.doi.org/10.1021/ja505136j | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX