The Electrophilic Fluorination of Enol Esters Using SelectFluor: A Polar Two-Electron Process

Article abstract of DOI:10.1002/chem.201900029

The reaction of enol esters with SelectFluor is facile and leads to the corresponding α-fluoroketones under mild conditions and, as a result, this route is commonly employed for the synthesis of medicinally important compounds such as fluorinated steroids. However, despite the use of this methodology in synthesis, the mechanism of this reaction and the influence of structure on reactivity are unclear. A rigorous mechanistic study of the fluorination of these substrates is presented, informed primarily by detailed and robust kinetic experiments. The results of this study implicate a polar two-electron process via an oxygen-stabilised carbenium species, rather than a single-electron process involving radical intermediates. The structure–reactivity relationships revealed here will assist synthetic chemists in deploying this type of methodology in the syntheses of α-fluoroketones.

Full text of DOI:10.1002/chem.201900029

A Journal of
Accepted Article
Title: The Electrophilic Fluorination of Enol Esters using SelectFluor
Occurs via a Polar Two-Electron Process
Authors: Susanna Wood, Stephen Etridge, Alan Kennedy, Jonathan
Percy, and David James Nelson
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.201900029
Link to VoR: http://dx.doi.org/10.1002/chem.201900029
Supported by

10.1002/chem.201900029
Chemistry - A European Journal
FULL PAPER
The Electrophilic Fluorination of Enol Esters using SelectFluor
Occurs via a Polar Two-Electron Process
Susanna H. Wood,[a] Stephen Etridge,[b] Alan R. Kennedy,[a] Jonathan M. Percy,[a]† and David J.
Nelson*[a]
Abstract: The reaction of enol esters with SelectFluor is facile and
leads to the corresponding α-fluoroketones under mild conditions
and, as a result, this route is commonly employed for the synthesis
of medicinally important compounds such as fluorinated steroids.
However, despite the use of this methodology in synthesis, the
mechanism of this reaction and the influence of structure on
reactivity are unclear. We present a rigorous mechanistic study of
the fluorination of these substrates, informed primarily by detailed
and robust kinetic experiments. The results of this study implicate a
polar two-electron process via an oxygen-stabilised carbenium
Scheme 1. The fluorination of steroid structures via enol ester intermediates.
species, rather than a single-electron process involving radical
intermediates. The structure/reactivity relationships revealed here
will assist synthetic chemists in deploying this type of methodology in
under very mild conditions; although 1 is ultimately prepared
from elemental fluorine, the use of this bench-stable solid is
more convenient than the direct deployment of fluorine gas,[24]
the syntheses of α-fluoroketones.
and requires no special glassware or apparatus. In addition, 1
[25-26]
Introduction
offers advantages over expensive XeF2
hypofluorites;[27-28]
or hazardous
the latter often require cryogenic
SelectFluor (1)[1] is an easy-to-handle and stable solid that is
widely used as an electrophilic fluorinating agent[2-5] and oxidant
in synthetic chemistry applications.[6-8] It is relatively cost-
temperatures which have associated cost, safety, and
environmental implications. Despite the significant utility and
widespread application of 1, there is only
a
limited
understanding of the underlying reaction mechanisms in many of
these fluorination reactions. In particular, there are many
examples where reaction via a polar two-electron mechanism is
proposed, and a corresponding set of examples where a
reaction via radical intermediates is posited.
effective
compared
to
alternatives
such
as
N-
fluorobenzenesulfonimide (NFSI) and fluoropyridinium salts, is
stable for extended periods of time,[9] and does not require
storage at low temperatures or under a protective atmosphere.
Various synthetic studies have demonstrated the ability of 1 to
fluorinate 1,3-dicarbonyl compounds, enol esters, and even
electron-rich arenes.[10-18] A variety of acyclic and cyclic alkyl and
aryl ketones can also undergo direct α-fluorination reactions
using 1 and analogous species, when heated to reflux in
acetonitrile or methanol solution.[19-21] Reagent 1 is one of the
more reactive fluorinating reagents, as judged from recent
kinetic studies.[22-23]
In support of a polar two-electron process, radical clock
methods where a pendant alkene or cyclopropyl group is
present on a substrate for fluorination using 1 suggest that either
radical species are not formed, and or if they are, that they are
not sufficiently long-lived to engage in subsequent reactions.
Citronellic ester enolates react with NFSI or with a close
analogue of 1, without ring-closing reactions that might be
expected to occur if a radical was formed (Scheme 2 (a));[29] in
contrast, XeF2 leads to ca. 10% of material undergoing such a
ring-closing process. Estimated rate constants for electron
transfer processes are ca. 107-times (or less) lower than
measured rate constants for fluorination reactions, for some
reactions between enolates or organometallic reagents and an
N-fluorosultam reagent.[30] However, in both of these studies, the
nucleophile is typically an organometallic reagent or an enolate;
emerging research shows that enolates can have intriguing
electron transfer behaviour, carrying out single-electron
processes, despite the fact that they would traditionally be
considered as two-electron nucleophiles.[31] Inner-sphere
electron transfer has also been proposed as a potential
mechanism for reactions of 1, in the context of the electrophilic
fluorination of alkenes.[32] Cyclopropyl-functionalised enol ethers
undergo reaction with 1 to produce fluoroacetals
Steroid-like structures are often fluorinated using 1 via the
corresponding (di)enol ester intermediates (Scheme 1), typically
[a]
[b]
S. H. Wood, Dr A. R. Kennedy, Prof. J. M. Percy, Dr D. J. Nelson
WestCHEM Department of Pure and Applied Chemistry
University of Strathclyde
295 Cathedral Street, Glasgow, G1 1XL, UK
E-mail: david.nelson@strath.ac.uk
Dr S. Etridge
GMS Manufacturing and Supply
GlaxoSmithKline
Cobden Street, Montrose, DD10 8EA, UK
† Retired December 2016.
Supporting information for this article is given via a link at the end of
the document. Raw data underpinning this research can be
downloaded from the following URL: http://dx.doi.org/[TBA].
This article is protected by copyright. All rights reserved.

10.1002/chem.201900029
Chemistry - A European Journal
FULL PAPER
Scheme 2. Literature studies that implicate two-electron or single-electron reaction mechanisms.
(Scheme 2 (b));[33] while no ring-opened product was obtained,
the fluoroacetal was recovered only in 45% yield, so much of the
material is unaccounted for. The reaction of this probe with NFSI
leads to 40% fluoroacetal plus 5% of ring-opened product. A
recent and detailed study of the electrophilicity of common
electrophilic fluorinating reagents used enamine and carbanion
nucleophiles (Scheme 2 (c));[22] the data obtained ruled out
mechanisms that involve single-electron transfer and are entirely
consistent with a polar two-electron mechanism.
methodology and the judicious selection of reaction conditions.
While SelectFluor is not the most expensive electrophilic
fluorinating reagent, it is still likely to be the major cost for any
industrial process that uses it. To the best of our knowledge, the
mechanism of the fluorination of enol esters by 1 has not been
unambiguously established.
Results
However, other studies implicate radical intermediates
strongly. An ESI-MS study of the reaction of tri- and
tetraphenylethene with 1 provided evidence for radical cation
intermediates, while no such radical cations were observed in
the absence of 1.[34] Similar studies of the reactions of styrene,
2-phenylpropene, TEMPO, and DMPO with 1 gave rise to
signals consistent with radical intermediates in fluorination
reactions mediated by 1.[35] Recent studies of vinyl azide
fluorination implicate a radical cation intermediate on the basis
of radical clock experiments;[36-37] however, the stability of
cyclopropyl-bearing substrates depends on how close this
Model systems based on a tetralone core provided a range of
synthetically tractable compounds for this work, and have well-
resolved and diagnostic 1H NMR spectra. The reactions of these
compounds with 1 are expected to form the corresponding α-
fluoroketones, plus by-products including ammonium salt 2, a
carboxylic acid, and tetrafluoroboric acid (Scheme 3). During this
work, systematic variation of the ester group at position 1 was
explored, as well as substitution at the 2- and 6-positions (vide
infra).
Parent compound 3 (R, R’ = H; R’’ = Me) was prepared
from tetralone by reaction with isopropenyl acetate; 3 underwent
smooth and complete fluorination at 298 K in a 95/5 v/v MeCN(-
d3)/water(-d2) mixture to produce fluoroketone 4 (93% yield) plus
2, acetic acid, and tetrafluoroboric acid (Scheme 4). Reactions in
neat acetonitrile(-d3) were often irreproducible and typically did
not reach complete conversion. An authentic sample of 2 was
prepared via an independent synthetic route to confirm the
assignment.[38] High-quality, reproducible kinetic data were
obtained from 1H NMR kinetic experiments; the use of a small
moiety is to the azide (Scheme
2 (d)). The successful
deployment of 1 in chemical processes, especially on scale in
industry, depends on a detailed understanding of how the
reaction proceeds, so that reaction outcomes can be understood
and, ideally, predicted.
We set out to understand the reaction mechanism involved
in the electrophilic fluorination of enol esters, and to identify and
quantify structure/reactivity relationships. A full understanding of
these issues is important to allow the rational deployment of this
This article is protected by copyright. All rights reserved.

10.1002/chem.201900029
Chemistry - A European Journal
FULL PAPER
Scheme 3. Fluorination of tetralone derivatives using SelectFluor.
Figure 2. Eyring-Polanyi plot for the fluorination of 3 using 1. Each point is the
average of two experiments.
Scheme 4. Fluorination of compound 3.
excess of 1 (1.5 equiv.) led to a pseudo-first order in 3 for three
or more half-lives. The reaction between 1 and 3 is first order in
1 as judged from plots of kobs versus [1] (Figure 1). Kinetic
experiments at different temperatures gave a good quality
Eyring-Polayni plot and values for ΔH‡ (15.0 kcal mol-1), ΔS‡ (-24
cal K-1 mol-1), and ΔG‡ (22.0 kcal mol-1) that are consistent with a
bimolecular reaction that occurs smoothly at 298 K (Figure 2).
Figure 3. Kinetic isotope effects measured for substrate 3 in the fluorination
reaction.
were performed with para-substituted enol benzoates 5a-h
(Scheme 5), which were typically prepared by reaction of 3 with
the corresponding benzoyl chloride. Substrates 5b-h reacted
smoothly with 1.5 equiv. 1, yielding good quality kinetic data and
forming 4 and the expected by-products. For 5a (X = NMe2) an
intensely-coloured solution was formed and no fluorination
occurred, and control experiments with N,N-dimethylaniline
confirmed that this functional group is not tolerated by 1; it has
been reported in the literature that similar compounds react with
N-fluoro-compounds to yield deeply coloured radical cations.[30]
Figure 1. Plot of kobs versus [1] for the fluorination of 3 with 1.
Reactions with 0.8 equiv. 1 gave 78% conversion of 3 to 4;
the recovered 3 was used to measure 13C kinetic isotope effects
(KIEs) using Singleton’s method.[39] The methyl signal (δC = 20.5
ppm) was used as the internal standard, assuming that this has
a KIE of 1.000 due to its distance from the reacting centre. The
only signal that showed a significant KIE was the arene carbon
at the 6-position, which is para to the carbon bearing the acetoxy
group (δC = 126.0 ppm; KIE = 1.031) (Figure 3).
Scheme 5. Reactions of benzoyl enol esters 5a-h.
A Hammett treatment of the kinetic data for 5b-h using σp
as the abscissa gave a relatively poor correlation (ρ = -0.4),
indicating a build-up of positive charge in the transition state
(Figure 4 (a)). A better plot was obtained using σp+ (Figure 4 (b);
Three Hammett studies were carried out to probe the
electronic effects of substrate structures.[40] Initial experiments
This article is protected by copyright. All rights reserved.

10.1002/chem.201900029
Chemistry - A European Journal
FULL PAPER
5f
Figure 4. Hammett plots for substrates 5b-h, constructed using (a) σp and (b)
σp+. Each point is the average of two experiments.
Figure 5. Eyring-Polanyi plots for the reactions of 5b, 5c, and 5h with 1. Each
point is the average of two experiments.
the point for 5f is included in both lines of best fit); this is
indicative of the development of a formal positive charge in
conjugation with the benzoyl ester and perhaps some change in
the rate-determining step as more electron-poor substrates are
considered (vide infra).
Table 1. Thermodynamic parameters for the reactions of 3, 5b, 5c, and 5h
with 1, obtained from kinetic studies at various temperatures.
Compound
ΔG‡/kcal mol-1
ΔH‡/kcal mol-1
ΔS‡/cal K-1 mol-1
3
22.0
21.6
22.0
22.3
15.0
15.5
14.9
15.5
-24
-20
-24
-23
Eyring-Polanyi plots were constructed for the reactions of
5b, 5c, and 5h based on kinetic studies carried out at 5 to 45 °C
(Figure 5) to further investigate this apparent break in the
Hammett plot. The free energy of activation (ΔG‡) followed the
same trend as the Hammett plot (Table 1), with the fluorination
of electron-rich 5b being the most rapid. The values of ΔS‡
obtained for 5c and 5h are almost identical (-23 and -24 cal K-1
mol-1 respectively), indicating a similar entropic penalty for
forming the transition state. The Eyring-Polanyi plot for 5c is
almost identical to that produced for enol acetate 3, consistent
with a common mechanism between enol esters.
A set of compounds was prepared with aryl substituents at
the site of fluorination (Scheme 6). While this was done to alter
the electron density at this site, it must be noted that the
resulting compounds are essentially stilbene derivatives and this
leads to some divergent behaviour compared to the previous
series of substrates (vide infra). The bromination of tetralone
using N-bromosuccinimide yielded 2-bromotetralone (6) which
gave 7 upon treatment with NaHMDS and acetic anhydride.
Compound 7 underwent palladium-catalysed cross-coupling with
5b
5c
5h
arylboronic acids to form 8a-e. Compound 7 underwent smooth
fluorination to form 9, without loss of the bromine substituent
(Scheme 6(a)). A molecular structure for 9 was obtained by X-
ray diffraction analysis (Figure 6). Consistent with previous
reports,[41] the bromide has a greater preference than fluorine for
the axial position due to unfavourable interactions with the
carbonyl group, placing the fluorine substituent in the equatorial
position.
Synthetic experiments with 8c led to the expected
fluorinated product 10c, but also to a non-fluorinated side-
product (Scheme 6 (b)). The latter was determined to be 11c on
the basis of detailed 1D and 2D NMR experiments, GC-MS and
IR analyses. The epoxidation of 8c to form 12c followed by
Scheme 6. (a) Reaction of compound 7. (b) Reactions of compounds 8a-e. (c) Independent synthesis of 11c.
This article is protected by copyright. All rights reserved.

10.1002/chem.201900029
Chemistry - A European Journal
FULL PAPER
Figure 6. Molecular structure of 9, as determined by X-ray diffraction analysis
of a single crystal.
Figure 7. Eyring-Polanyi plot for the reaction of 8c with 1. Each point is the
average of two experiments.
heating to provoke rearrangement, allowed the synthesis of an
authentic sample of 11c with NMR data that matched the
material produced during the fluorination reaction (see the
Supporting Information). The formation of 11c likely derives from
the stilbene nature of this substrate; further work is underway in
our laboratory understand more fully the origin of this side
product. Attempts to prepare 11c by exposing solutions of 8c to
oxygen and/or water were unsuccessful, so 1 is necessary for
both pathways.
The reactions of each of 8a-e gave various ratios of 10a-e
and 11a-e, as determined by integration of the 1H NMR
spectrum using an internal standard, with more electron-rich
substrates yielding more 11 (Table 2). All reactions reached
≥97% conversion. Attempts to suppress side-product formation
by performing the reaction under a nitrogen atmosphere led to
no change in the product ratio; similarly, the use of air-sparged
solvent did not increase the proportion of 11 produced.
Kinetic experiments with 8a-e allowed rate constants to be
measured for their reaction with SelectFluor in acetonitrile/water
(95/5 v/v). The correlation with σp was again rather poor (Figure
+
8 (a)), but a much better correlation was obtained with σp
+
(Figure 8 (b)); unfortunately, a value for σp is not available for
acetyl so 8d could not be included in the correlation. These
results are also indicative of the build-up of a formal positive
charge. The slope is steeper than observed in Figure 4 (b)
(above), with ρ = -0.87 versus -0.30 to -0.79. Each experiment
produced some by-product 11 but the decrease in concentration
of 8c was well-behaved pseudo-first order when 1.5 equiv. of 1
were used. This well-behaved kinetic behaviour and the results
of an Eyring-Polanyi analysis for 8c are indicative of a common
rate-determining step followed by different pathways for the
formation of 10 and 11.
Table 2. Results of the fluorination of 8a-e in 95/5 v/v MeCN-d3/D2O, as
determined from kinetic experiments. Conversions are determined by
integration versus an internal standard (cyclohexane capillary).
Entry
Substrate
Ar p-subs.
MeO
Me
Conv. to 10
Conv. to 11
1
2
3
4
5
6
8a
8b
8c
8d
8e
8f
57
70
84
85
90
87
40
27
16
15
10
13
H
Ac
Figure 8. Hammett plot for the fluorination of 8a-e in acetonitrile/water (95/5
v/v) versus (a) the σp parameter and (b) the σp+ parameter.
MeO2C
F3C
A set of substrates was prepared in which the 6-position of
the tetralone core was systematically varied (Scheme 7), as KIE
experiments (vide supra) suggest that this position is important
in the fluorination reaction. A set of compounds was prepared,
using 6-methoxytetralone (13a), 6-aminotetralone (13b), and 6-
carboxytetralone (13c) as the starting materials (see the
Supporting Information).
An Eyring analysis of the reaction of 8c with 1 revealed
very similar parameters to those for 3, 5b, 5c, and 5h, even
though two products result (ΔH‡ = 15.5 kcal mol-1, ΔS‡ = -21 cal
K-1 mol-1, ΔG‡ = 21.9 kcal mol-1; Figure 7). This is consistent with
a common rate- determining step involving both 8c and 1,
followed by a subsequent step (or steps) in which 9c and 10c
are formed.
This article is protected by copyright. All rights reserved.

10.1002/chem.201900029
Chemistry - A European Journal
FULL PAPER
Scheme 7. Reactions of enol esters 14a – 14d.
The fluorination reactions of 14a-d proceeded smoothly
+
with pseudo-first order kinetics. Hammett plots using σp and σp
+
showed a better fit with σp+ (R2 = 0.75 for σp versus 0.94 for σp )
(Figure 9). The value of ρ obtained for this series (-1.36) was
greater in magnitude than those observed for studies with
compounds 5 and 8 (vide supra).
Scheme 8. Reactions of cyclopropyl-substituted substrates.
Reactions in acetonitrile/H218O solvent were used to
identify the source of the oxygen in 4; these were analysed by
mass spectrometry.[42] The reactions of 3, 5c, and 5h led to
complete incorporation of 18O into 4 (Scheme 9(a)). For 8c,
there was no incorporation of 18O into the product (Scheme 9
(b)). The molecular ion of side-product 11c was not detected by
GC-MS (EI ionisation), but 21c was and this did not contain 18O
(Figure 10); IR analysis of a mixture of 10c and 11c from this
experiment showed the characteristic shift in the wavenumber of
the ester C=O signals that would be expected if 16O was
replaced with 18O. When 18 underwent reaction with 1 in the
presence of H218O, there was 18O incorporation into 19 and 20
(Scheme 9 (c)).
Attempts to probe the involvement of radicals by adding
TEMPO to kinetic reactions were unsuccessful. While the
reaction did indeed cease completely upon the addition of
TEMPO, this was due to the immediate and complete
destruction of 1.[33] The reaction between 1 and TEMPO leads to
complete conversion of 1, as judged by 19F NMR spectroscopy,
and the formation of oxoammonium salt [22]BF4 (Scheme 10 (a))
as judged by the comparison of IR and UV spectra with those of
an authentic sample prepared by the reaction of TEMPO with
bleach and tetrafluoroboric acid (Scheme 10 (b)).[43] The use of
TEMPO as a radical trap in reactions mediated by 1 is therefore
unreliable. However, while [22]ClO4 reacts with vinyl azides to
form various functionalised products,[44] [22]BF4 reacts with 8c
and 18 to form the corresponding naphthalenes 23 and 24
(Scheme 10 (c)).
Figure 9. Hammett plots for the reactions of 14a-e with 1, plotted using (i) σp
+
and (ii) σp
.
Cyclopropyl-substituted substrates were prepared to probe
whether radical species were involved; the formation of a radical
close to the cyclopropyl group would lead to ring-opening.
Compound 16 was prepared from tetralone and subjected to the
standard reaction conditions (Scheme 8 (a)); the products of this
reaction were α-fluoroketone 4 and cyclopropyl carboxylic acid
17. Compound 18 was prepared via Suzuki-Miyaura cross-
coupling of 7 and cyclopropylboronic acid. As observed with the
α-aryl substituted series 8, reaction with 1 produced the α-
fluoroketone 19, ester side product 20 and acetic acid (Scheme
8 (b)). There was no evidence of cyclopropyl ring-opening.
The storage of 19 in CDCl3 for three weeks resulted in
discolouration of the solution. 1H and 19F NMR analyses
indicated complete decomposition of 19 to several fluorinated
products, including some with large coupling constants (J = ca.
2
45 Hz) that are indicative of JHF coupling. It is likely that these
compounds have arisen as a result of the loss of fluoride from
19, producing a carbenium adjacent to the cyclopropane
followed by ring opening/rearrangement and reintroduction of
fluoride. Analogous compound 11c is stable under these
conditions, demonstrating that the phenyl and cyclopropyl
substituents have different effects on the stability of these
molecules.
Indanone derivative 25 was prepared via a sequence of
bromination acetylation, and palladium-catalysed cross-coupling
and underwent reaction with 1 to form product 26 and side-
product 27 in a ratio of 84:16 (Scheme 11). This ratio is identical
This article is protected by copyright. All rights reserved.

10.1002/chem.201900029
Chemistry - A European Journal
FULL PAPER
Scheme 10. (a) Formation of [22]BF4 from the reaction of TEMPO with
SelectFluor. (b) Formation of [22]BF4 by oxidation of TEMPO using bleach. (c)
Reactions of 8c and 18 with [22]BF4.
Scheme 11. Reaction of indanone derivative 25.
ester fluorination reaction is occurring via
a
common
Scheme 9. Fluorination of compounds 3, 5c, 5h, and 17 in acetonitrile/
[
18O]water. Reaction yields were not determined.
mechanism. The reaction is pseudo-first order under the
reaction conditions when a small excess of 1 is used, first order
in 1, and no longer pseudo-first order in enol ester if less than
1.5 equivalents of 1 are used. This bimolecular reaction
therefore involves 1 and the enol ester.
It is not likely that ester hydrolysis is rate-determining.
Experiments with 5c in 95/5 v/v acetonitrile/water and
acetonitrile/methanol proceed with similar rates (kobs
(MeCN/water) = 4.97 x 10-4 s-1; kobs (MeCN/MeOH) = 6.21 x 10-4
s-1; at 298 K). Water and methanol have very different
Figure 10. Des-acetate compound 21c identified by GC-MS (EI) analysis.
nucleophilicities, as determined by Mayr:[45]
a
90/10
acetonitrile/water mixture has N = 4.56 (sN = 0.94), while 90/10
acetonitrile/methanol has N = 5.55 (sN = 0.97); note that N is a
logarithmic scale and so a one unit difference in N translates to
to that obtained for the reaction of tetralone derivative 8c with 1.
a
ten-fold difference in reaction rate.[46] Other ratios of
acetonitrile/solvent (80/20, 67/33, 33/67, 20/80, and 9/91) show
the same trend, so we assume that this holds for 95/5 mixtures
also. If ester hydrolysis was rate-determining, then reactions in
methanol should be much faster. In addition, the rates of
fluorination of 5b-h increase when electron-donating groups are
present; such a structural change would render the ester
carbonyl group less electrophilic. The enol ester must therefore
be the nucleophile in the rate-determining step, not the
Discussion
The data gathered for the fluorination of 3, 5b, 5c, and 5h show
that the rate-determining step is bimolecular (ΔS‡ = -20 to -24
cal K-1 mol-1), and has a free energy of activation that is
consistent with a facile reaction that takes a few hours at room
temperature to reach completion (ΔG‡ = 20.9 to 22.3 kcal mol-1).
The consistency between each series indicates that each enol
This article is protected by copyright. All rights reserved.

10.1002/chem.201900029
Chemistry - A European Journal
FULL PAPER
electrophile. The outcomes of the H218O experiments show that
the nucleophilic attack must occur from the enol ester not the
enolate; if the enolate was the nucleophile, as a result of ester
hydrolysis, then 18O would be incorporated into the carboxylic
acid side-product and not into the ketone.
study in Figure 9 are the strongest evidence for such a
mechanism, with the substituent at the tetralone 6-position
clearly supporting a developing positive charge. The correlation
+
with σp is very poor, yet the correlation with σp is excellent; the
+
σp scale takes into account the ability of +R substituents to
Evidence from other experiments is entirely consistent with
a polar two-electron mechanism that proceeds via a carbenium
intermediate. Substrates with cyclopropyl groups (16 and 18),
which are known to undergo rapid ring-opening when radicals
are nearby (ca. 108 s-1),[47] do not lead to the types of alkene
side-products that might be expected if a single-electron process
was in operation (e.g. Scheme 12 (a)). Furthermore, if a
carbenium ion was to be formed adjacent to the cyclopropyl
group, ring-opening or ring-expansion products would also be
expected (e.g. Scheme 12 (b)). It is unlikely that, if a radical
intermediate was generated in this reaction, it could undergo a
bimolecular reaction before side-reactions occurred because at
the concentrations studied such a reaction would have to occur
faster than diffusion through solution (ca. 1010 L mol-1 s-1).
A third possibility – the formation of carbenium adjacent to
the ester after nucleophilic attack of 1 by the enol ester – is
supported by all of our experimental data (Scheme 13). Such an
intermediate would gain resonance stabilisation from the
adjacent oxygen atom, and the absence of a radical species
precludes cyclopropyl ring-opening. The results of the Hammett
stabilise cations, and so the excellent correlation in Figure 9 is
indicative of carbenium ion stabilisation. Figures 4 and 8 are
also consistent with this proposal, showing again that electron-
donating groups accelerate the reaction, with – in both cases –
better correlations to σp+ than to σp. If electron transfer to 1 from
the substrate and fluorine atom transfer from a putative radical
anion of 1 to the radical thus formed occur as discrete steps,
they must occur in such rapid succession that they are best
considered as a single step; species such as [22]BF4 that act
only as oxidants lead to formation of the corresponding
naphthalene. The 13C KIE experiments with 3 rather surprisingly
showed the greatest KIE (1.03) at the 6-position of the tetralone
core. This is consistent with a concomitant change in the
electronic structure of the aryl ring which would be expected if a
carbenium ion was formed in a position in which it was in
conjugation with this aryl ring.
The value of ρ decreases for substitution at X1 > X2 > X3
(Scheme 14), which can be rationalised by considering the
behaviour of an intermediate formed after nucleophilic attack of
1. Substitution at the 6-position of the tetralone core (X1, ρ = -
Scheme 12. Expected, but unobserved, reaction pathways if (a) a radical species was generated or (b) a carbenium ion was generated adjacent to the
cyclopropyl group via single-electron transfer from 18.
Cl
O
O
N
O
O
Rate-determining
nucleophilic
aꢀack
H
O
R1
O
R1
O
N
F
F
R1
O
R1
O
water
F
R2
R2
F
R2
R2
X
X
X
X
Scheme 13. Proposed mechanism for fluorination of enol esters by 1.
This article is protected by copyright. All rights reserved.

10.1002/chem.201900029
Chemistry - A European Journal
FULL PAPER
Scheme 14. Values of ρ for the three different Hammett studies conducted during this work.
1.4) has the strongest effect, because a resonance form can be
envisaged in which a lone pair of electrons directly stabilises the
carbenium ion, while retaining essentially the same geometry
within the compound. For substitution at the other end of the
molecule (X2, ρ = -0.9), even though the arene is not directly in
conjugation with the carbenium, neighbouring group participation
can stabilise the carbenium ion. Finally, X3 is sufficiently remote
from the carbenium ion centre that again some form of
neighbouring group participation would be the manner in which it
could stabilise a carbenium ion. The comparable rates of
benzoate and acetate fluorination (Table 1; the benzoate reacts
only slightly faster), rule out neighbouring group effects involving
Wheland-type intermediates for enol benzoate substrates.
This leads to a rationale for the change of slope in the
Hammett plot for enol benzoate fluorination (see Figure 4).
When considering the resonance (R) and field (F) parameters
for the substituents,[40] it is found that MeO, Cl, and F are +R and
CO2Me, CF3, and NO2 are –R (R = -0.56, -0.19, -0.39, 0.11,
0.16, and 0.13, respectively); the former three substituents are
therefore capable of participating in the process outlined in
Scheme 14 (c), while the latter three are not, leading to a
precipitous decline in the reaction rate as the benzoate group
becomes less electron-rich. The interception of carbenium
intermediates intra- and intermolecularly (with carboxylic acids
and esters, and with acetonitrile, respectively) has been reported
in previous studies of fluorination using 1.[48]
hydrolysis mechanism for this reaction, as no 18O is incorporated
at the ketone; the hydrolysis mechanism must therefore be
different, perhaps as a result of the increased steric bulk
adjacent to the ketone, rendering this similar to neopentyl
systems which are infamously bad electrophiles (Scheme 15
(b)).
The origins of the side products 11 (from the reactions of
8) and 20 (from 18) are currently unclear. Compound 19 is
rather unstable in solution, and so it is possible that 20 is a side-
product arising from decomposition. The formation of 11c is
likely to occur subsequent to the rate determining step between
1 and 8c, but not from ketone 10c, for a number of reasons: (i)
10c is stable in solution, even in the presence of added acid; (ii)
the formation of a carbenium adjacent to the carbonyl in 10c,
such as from the loss of fluoride, in MeCN/H218O solvent might
lead to some products from the reaction of the carbenium with
water or acetic acid, incorporating an 18O label;[49] and (iii) the
18O label is incorporated exclusively into the carbonyl oxygen of
the ester, while intramolecular reactions involving [18O]1-AcOH
would be expected to result in a ca. 1:1 ratio of R18OC(O)Me
and ROC(18O)Me. The similar ratio of product to side product
obtained from the indanone-derived analogue 25 shows that this
pathway is not limited to our tetralone system. Several
mechanistic hypotheses are currently under investigation using
further experimental studies and computational modelling.
Once fluorination has occurred, and a carbenium ion is
formed, the final step is hydrolysis using water. This could occur
either via attack of water at the ester carbonyl group, or directly
at the carbenium. When H218O was used in place of H216O,
complete incorporation of 18O at the ketone was identified by
GC-MS analysis of the reactions of 3, consistent with the
Conclusions
This work represents the most detailed study to date of
mechanism and structure/reactivity relationships in the
fluorination of enol esters using 1. The collection of high quality
kinetic data, and the use of isotopic labelling, have allowed us to
gather a significant body of evidence that points towards a polar
two-electron process for the fluorination reaction (Scheme 12,
presence of
a
highly-reactive carbenium species which
undergoes reaction with water (Scheme 15 (a)). Experiments
with 8c in acetonitrile/[18O]water suggest a rather different
This article is protected by copyright. All rights reserved.

10.1002/chem.201900029
Chemistry - A European Journal
FULL PAPER
H
18O
Cl
O
O
(a)
N
H
O
O
H
O
R1
O
R1
O
R1
N
18O
F
R2
F
R2
F
R
O
F
R2
H
X
X
X
X
H
Cl
18O
O
O
(b)
N
O
18O
H
H
O
O
R1
O
R1
O
N
R1
F
F
F
O
F
Ar
Ar
Ar
Ar
Scheme 15. Proposed mechanisms for the hydrolysis of (a) 3 and (b) 8c after fluorination.
above). All of the Hammett studies indicate that a carbenium ion
is formed as a crucial intermediate, with +R substituents leading
to significant increases in the reaction rate. Attempts to provoke
any radical species present into undergoing reactions such as
cyclopropyl group ring-opening were unsuccessful; cyclopropyl-
substituted compounds undergo smooth and facile fluorination
without the formation of side-products.
Experimental Section
General. General experimental details are provided here. Full
characterisation for all compounds can be found in the Supporting
Information. Unless stated otherwise, all compounds were obtained from
commercial sources and used as supplied. Each batch of SelectFluor (1)
was analysed by iodometric titration and confirmed to be >95% pure.
It is important to consider these results in the context of
recent studies that imply that similar polar two-electron
mechanisms are in operation for the fluorination of enamines,[22]
stabilised carbanions,[22] and 1,3-dicarbonyl species.[23] The
carbanions in particular are more electron-rich than the enol
ester substrates considered here, and the enamines would also
be expected to be more electron rich. The enol forms of the 1,3-
dicarbonyl compounds considered by Sandford and Hodgson
possibly have similar nucleophilicities to the enol esters
considered here, but unfortunately the nucleophilicities of these
species have not yet been measured. However, (silyl) enol
ethers have nucleophilicities (N = ca. 4 – 7)[50] that are ≥ 104-
times lower than enamines (N = ca. 10 – 16)[51] and ≥ 1010-times
Compounds 2 and [22]BF4 were prepared according to literature
methods.[38, 43]
Analysis. NMR characterisation was carried out using Bruker AV3-400
equipped with a liquid nitrogen Prodigy cryoprobe, or a Bruker AV400
equipped with a BBFO-z-ATMA probe; kinetic data were obtained using a
Bruker AVII-600 NMR spectrometer equipped with a BBO-z-ATMA
probe. All NMR chemical shifts are quoted in units of parts per million
(ppm). 1H NMR spectra were referenced to residual solvent signals,[55]
13C{1H} spectra were referenced to the solvent signal,[55] and 19F and
19F{1H} spectra were externally referenced to CFCl3. Assignments of the
spectra were achieved by the use of [1H, 1H] COSY, [1H, 13C] HSQC, and
[1H, 13C] HMBC experiments, as required. Mass spectrometry data was
obtained using an Agilent 7890A GC system coupled to an Agilent 5975C
mass spectrometer in electron impact mode. High resolution mass
spectrometry was carried out on a ThermoFinnigan Exactive mass
lower than
nucleophilicities of vinyl azides, which under some
circumstances undergo fluorination with via radical
the carbanions studied by Mayr.[52-54] The
1
a
mechanism,[36-37] have not yet been measured. The fact that
nucleophiles with such a wide range of reactivity all undergo
reaction via a two-electron mechanism leads us to propose that
radical reactions between organic nucleophiles and reagent 1
are the exception, rather than the rule.
spectrometer.
Infrared spectra were obtained using a Shimadzu
IRAffinity-1 IR spectrometer with an ATR accessory. Thin layer
chromatography was performed on pre-coated aluminium-backed silica
gel plates (silica gel 60 F254, 0.2 mm thick). Column chromatography
was performed on silica gel (40 – 63 μm).
Further mechanistic studies of electrophilic fluorination
reactions with 1 are currently underway in our laboratories using
a variety of substrates, as we continue to work towards a
quantitative understanding of this class of reaction.
Procedure for SelectFluor titration. A nominally 0.01 mol L-1 aqueous
solution of SelectFluor (10 mL) was added to a 0.02 mol L-1 aqueous
solution of potassium iodide (20 mL), followed by 1.9 mol L-1 aqueous
sulfuric acid (5 mL). The liberated iodine was titrated against a 0.05 mol
L-1 aqueous solution of sodium thiosulfate.
This article is protected by copyright. All rights reserved.

10.1002/chem.201900029
Chemistry - A European Journal
FULL PAPER
General procedure A for fluorination reactions with 1 (synthetic
experiments). Substrate (1 equiv.) and SelectFluor (1 equiv.) were
suspended in 95/5 v/v MeCN/H2O (0.38 mol L-1) and the mixture was
stirred at room temperature for 4-19 hours.
University of Strathclyde for a Chancellor’s Fellowship. We thank
Mr Gavin Bain, Mr Alexander Clunie, Mr Craig Irving, Ms Patricia
Keating, and Dr John Parkinson for assistance with technical
and analytical facilities. We thank Marco Smith (GSK) for
assistance with high resolution mass spectrometry. We are
grateful to Professor John Murphy (University of Strathclyde), Dr
Tony Harsanyi (GSK), and Dr Andrew Dominey (GSK) for
helpful discussions.
General procedure B for fluorination reactions with 1 (kinetic
experiments). The substrate was weighed into an NMR tube, dissolved
in 0.7 mL of the pre-made solvent mixture and a sealed capillary
containing 5 μL cyclohexane and CDCl3 was added to the tube. This
sample was used to tune, match, lock and shim the spectrometer, and to
set the receiver gain. SelectFluor was then added and 1H NMR spectra
(2 scans per spectrum) were acquired at 120 s intervals until more than 4
half-lives had elapsed.
Keywords: fluorine • reaction mechanisms • structure-activity
relationships • kinetics • hydrolysis
[1]
R. E. Banks, S. N. Mohialdin-Khaffaf, G. S. Lal, I. Sharif, R. G.
Syvret, J. Chem. Soc., Chem. Commun. 1992, 595-596.
G. G. Furin, A. A. Fainzilberg, Russ. Chem. Rev. 1999, 68, 653.
G. S. Lal, G. P. Pez, R. G. Syvret, Chem. Rev. 1996, 96, 1737-
1756.
P. T. Nyffeler, S. G. Durón, M. D. Burkart, S. P. Vincent, C.-H.
Wong, Angew. Chem. Int. Ed. 2005, 44, 192-212.
R. P. Singh, J. n. M. Shreeve, Acc. Chem. Res. 2004, 37, 31-44.
E. Differding, P. M. Bersier, Tetrahedron 1992, 48, 1595-1604.
E. W. Oliver, D. H. Evans, J. Electroanal. Chem 1999, 474, 1-8.
S. Stavber, M. Zupan, Acta Chim. Slov. 2005, 52, 13-26.
The assay of a sample of 1 from the early 1990s, recently
discovered in a laboratory at Strathclyde, confirmed that this
compound was still >95% active.
General procedure C for the synthesis of enol esters. Tetralone (1
equiv.), isopropenyl acetate (2 equiv.) and acid chloride (1.25 equiv.)
were combined and heated to 100 °C for 1-2 h in a flask equipped with
still head, condenser and a collection flask. The temperature was then
increased to 170 °C for 4 – 16 h. The resulting brown residue was cooled
to room temperature, suspended in DCM and concentrated onto silica.
The silica was then washed (10% ethyl acetate/hexane) and the filtrate
was concentrated and chilled, resulting in precipitation of the desired
product.
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
General procedure D for the synthesis of enol esters. Compound 3 (1
equiv.) was suspended in toluene and p-toluenesulfonic acid
monohydrate (0.1 equiv.) and acid chloride (1.5 equiv.) were added. The
reaction was heated to 160 °C for 7 h with a Dean-Stark trap fitted. Upon
cooling to room temperature the residue solidified. This was suspended
in toluene and filtered. The filtrate was concentrated and the residue was
dissolved in DCM. This solution was washed with saturated aqueous
sodium bicarbonate solution (to pH 8). The organic phase was collected,
dried over anhydrous MgSO4 and concentrated in vacuo onto silica. The
silica was then washed (10% ethyl acetate/hexane) to yield a filtrate from
which the compound precipitated on cooling.
[10]
[11]
[12]
G. Stavber, M. Zupan, S. Stavber, Tetrahedron Lett. 2007, 48,
2671-2673.
T. Fukuzumi, N. Shibata, M. Sugiura, S. Nakamura, T. Toru, J.
Fluor. Chem. 2006, 127, 548-551.
R. E. Banks, M. K. Besheesh, S. N. Mohialdin-Khaffaf, I. Sharif, J.
Chem. Soc., Perkin Trans. 1 1996, 2069-2076.
G. S. Lal, J. Org. Chem. 1993, 58, 2791-2796.
M. Zupan, J. Iskra, S. Stavber, J. Fluor. Chem. 1995, 70, 7-8.
M. Zupan, J. Iskra, S. Stavber, Tetrahedron 1996, 52, 11341-
11348.
A. Solladié-Cavallo, L. Jierry, A. Klein, M. Schmitt, R. Welter,
Tetrahedron: Asymmetry 2004, 15, 3891-3898.
S. L. Acebedo, J. A. Ramírez, L. R. Galagovsky, Steroids 2009,
74, 435-440.
V. Reydellet-Casey, D. J. Knoechel, P. M. Herrinton, Org. Proc.
Res. Dev. 1997, 1, 217-221.
S. Stavber, M. Zupan, Tetrahedron Lett. 1996, 37, 3591-3594.
S. Stavber, M. Jereb, M. Zupan, Chem. Commun. 2000, 1323-
1324.
S. Stavber, M. Jereb, M. Zupan, Synthesis 2002, 2002, 2609-
2615.
D. S. Timofeeva, A. R. Ofial, H. Mayr, J. Am. Chem. Soc. 2018.
N. Rozatian, I. W. Ashworth, G. Sandford, D. R. W. Hodgson,
Chem. Sci. 2018.
R. D. Chambers, J. Hutchinson, J. Fluor. Chem. 1998, 89, 229-
232.
B. Zajc, M. Zupan, J. Chem. Soc., Chem. Commun. 1980, 759-
760.
[13]
[14]
[15]
[16]
[17]
[18]
General procedure E for the synthesis of alpha-substituted enol
esters. Vinyl bromide (1 equiv.), boronic acid (1.1 equiv.), 1,1-
[bis(diphenylphosphino)ferrocene]dichloropalladium(II) (5 mol%), K3PO4
(3 equiv.) and water (10 equiv.) were suspended in toluene (0.25 mol L-1
in bromide) under a nitrogen atmosphere and heated to 85 ˚C for 2.5 - 19
h. The mixture was then allowed to cool to room temperature, filtered
through silica and the silica was washed with 10% ethyl acetate in
hexane. The filtrate was concentrated in vacuo and the residue re-
suspended in hexane resulting in the precipitation of the desired
compound.
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
B. Zajc, M. Zupan, J. Org. Chem. 1982, 47, 573-575.
S. Rozen, Y. Menahem, Tetrahedron Lett. 1979, 20, 725-728.
W. E. Barnette, R. C. Wheland, W. J. Middleton, S. Rozen, J. Org.
Chem. 1985, 50, 3698-3701.
E. Differding, G. M. Rüegg, Tetrahedron Lett. 1991, 32, 3815-
3818.
E. Differding, M. Wehrli, Tetrahedron Lett. 1991, 32, 3819-3822.
S. Zhou, E. Doni, G. M. Anderson, R. G. Kane, S. W. MacDougall,
V. M. Ironmonger, T. Tuttle, J. A. Murphy, J. Am. Chem. Soc.
2014, 136, 17818-17826.
Y. A. Serguchev, M. V. Ponomarenko, L. F. Lourie, A. A. Fokin, J.
Phys. Org. Chem. 2011, 24, 407-413.
S. P. Vincent, M. D. Burkart, C.-Y. Tsai, Z. Zhang, C.-H. Wong, J.
Org. Chem. 1999, 64, 5264-5279.
X. Zhang, Y. Liao, R. Qian, H. Wang, Y. Guo, Org. Lett. 2005, 7,
3877-3880.
General procedure F for the synthesis of alpha-ester ketones. Enol
ester (1 equiv.) and m-CPBA were dissolved in DCM and heated to reflux
for 24 h. The mixture was extracted with 1 mol L-1 NaOH solution. The
organic phase was collected, concentrated and passed through silica in
10% ethyl acetate/hexane solution. The filtrate was concentrated to yield
the desired compound.
[29]
[30]
[31]
Compounds 10 and 11. Note that the fluorination of compounds 8 leads
to the formation of mixtures of 10 and 11. Fluorinated compounds 9a, 9b,
9d and 9f were characterised by their 1H NMR spectra. Side products 11
were characterised by NMR spectroscopy and HRMS. See the
Supporting Information for full details.
[32]
[33]
[34]
[35]
[36]
X. Zhang, H. Wang, Y. Guo, Rapid Commun. Mass Spectrom.
2006, 20, 1877-1882.
S.-W. Wu, J.-L. Liu, F. Liu, Chem. Commun. 2017, 53, 12321-
Acknowledgements
12324.
[37]
[38]
S.-W. Wu, F. Liu, Org. Lett. 2016, 18, 3642-3645.
J. C. Sarie, C. Thiehoff, R. J. Mudd, C. G. Daniliuc, G. Kehr, R.
Gilmour, J. Org. Chem. 2017, 82, 11792-11798.
D. A. Singleton, A. A. Thomas, J. Am. Chem. Soc. 1995, 117,
9357-9358.
SHW thanks GlaxoSmithKline Global Manufacturing and Supply
(Montrose) and the University of Strathclyde for a studentship
via the GSK/Strathclyde Centre for Doctoral Training in
Medicinal Chemistry and Organic Synthesis. DJN thanks the
[39]
[40]
C. Hansch, A. Leo, R. W. Taft, Chem. Rev. 1991, 91, 165-195.
This article is protected by copyright. All rights reserved.

10.1002/chem.201900029
Chemistry - A European Journal
FULL PAPER
[41]
E. A. Basso, C. Kaiser, R. Rittner, J. B. Lambert, J. Org. Chem.
1993, 58, 7865-7869.
[42]
These reactions were conducted on a very small scale for analysis
by GC-MS, which shows complete consumption of starting
material, but no yields were recorded.
[43]
[44]
[45]
S.-S. Weng, J.-W. Zhang, ChemCatChem 2016, 8, 3720-3724.
J.-L. Liu, S.-W. Wu, Q.-Y. Wu, F. Liu, J. Org. Chem. 2018.
S. Minegishi, S. Kobayashi, H. Mayr, J. Am. Chem. Soc. 2004,
126, 5174-5181.
[46]
[47]
[48]
H. Mayr, A. R. Ofial, J. Phys. Org. Chem. 2008, 21, 584-595.
D. C. Nonhebel, Chem. Soc. Rev. 1993, 22, 347-359.
L. F. Lourie, Y. A. Serguchev, G. V. Shevchenko, M. V.
Ponomarenko, A. N. Chernega, E. B. Rusanov, J. A. K. Howard, J.
Fluor. Chem. 2006, 127, 377-385.
[49]
The concentration of water in these reactions is approximately 2.5
mol/L. The concentration of acetic acid will be, at most, 0.04 mol/L.
The presence of tetrafluoroboric acid in solution will mean that all
acetic acid is present as acetic acid, and not acetate.
H. Mayr, B. Kempf, A. R. Ofial, Acc. Chem. Res. 2003, 36, 66-77.
D. S. Timofeeva, R. J. Mayer, P. Mayer, A. R. Ofial, H. Mayr,
Chem. Eur. J. 2018, 24, 5901-5910.
[50]
[51]
[52]
[53]
F. Seeliger, H. Mayr, Org. Biomol. Chem. 2008, 6, 3052-3058.
S. T. A. Berger, A. R. Ofial, H. Mayr, J. Am. Chem. Soc. 2007, 129,
9753-9761.
[54]
[55]
O. Kaumanns, R. Appel, T. Lemek, F. Seeliger, H. Mayr, J. Org.
Chem. 2009, 74, 75-81.
G. R. Fulmer, A. J. M. Miller, N. H. Sherden, H. E. Gottlieb, A.
Nudelman, B. M. Stoltz, J. E. Bercaw, K. I. Goldberg,
Organometallics 2010, 29, 2176-2179.
This article is protected by copyright. All rights reserved.

10.1002/chem.201900029
Chemistry - A European Journal
FULL PAPER
Entry for the Table of Contents (Please choose one layout)
Layout 1:
FULL PAPER
Text for Table of Contents
Author(s), Corresponding Author(s)*
Page No. – Page No.
Title
((Insert TOC Graphic here: max.
width: 5.5 cm; max. height: 5.0 cm))
Layout 2:
FULL PAPER
S. H. Wood, S. Etridge, A. R. Kennedy,
J. M. Percy, and D. J. Nelson*
((Insert TOC Graphic here; max. width: 11.5 cm; max. height: 2.5 cm))
Page No. – Page No.
The Electrophilic Fluorination of Enol
Esters using SelectFluor Occurs via a
Polar Two-Electron Process
Kinetic studies of the reactions of enol esters with SelectFluor show that these
occur via a polar two-electron mechanism.
This article is protected by copyright. All rights reserved.