Iron-Catalyzed, Fluoroamide-Directed C-H Fluorination

Article abstract of DOI:10.1021/jacs.6b08171

This communication describes a mild, amide-directed fluorination of benzylic, allylic, and unactivated C-H bonds mediated by iron. Upon exposure to a catalytic amount of iron(II) triflate (Fe(OTf)₂), N-fluoro-2-methylbenzamides undergo chemoselective fluorine transfer to provide the corresponding fluorides in high yield. The reaction demonstrates broad substrate scope and functional group tolerance without the use of any noble metal additives. Mechanistic and computational experiments suggest that the reaction proceeds through short-lived radical intermediates with F-transfer mediated directly by iron.

Full text of DOI:10.1021/jacs.6b08171

Communication
pubs.acs.org/JACS
Iron-Catalyzed, Fluoroamide-Directed C−H Fluorination
Brian J. Groendyke, Deyaa I. AbuSalim, and Silas P. Cook*
Department of Chemistry, Indiana University, 800 East Kirkwood Avenue, Bloomington, Indiana 47405-7102, United States
S
* Supporting Information
ABSTRACT: This communication describes a mild,
amide-directed fluorination of benzylic, allylic, and
unactivated C−H bonds mediated by iron. Upon exposure
to a catalytic amount of iron(II) triflate (Fe(OTf)2), N-
fluoro-2-methylbenzamides undergo chemoselective fluo-
rine transfer to provide the corresponding fluorides in high
yield. The reaction demonstrates broad substrate scope
and functional group tolerance without the use of any
noble metal additives. Mechanistic and computational
experiments suggest that the reaction proceeds through
short-lived radical intermediates with F-transfer mediated
directly by iron.
ue to its unique properties, fluorine offers unparalleled
D
opportunities in drug discovery,1 crop sciences,2 and
materials3 research. However, fluorine incorporation remains
challenging; the most common strategy employs fluorinated
building blocks in the established synthesis of a given target. This
approach is limited by the availability of fluorinated building
blocks and the complexity of the synthesis. In this context, late-
stage fluorination provides a powerful strategy to access
fluorinated analogs of lead compounds.4 While standard
fluorination chemistry uses a preactivated position in the form
of an alcohol5 or tin derivative,6 the directed fluorination of
resident C−H bonds promises a particularly attractive approach
to late-stage fluorination.
Figure 1. Directed radical chemistry as a route to Csp3−H fluorination.
tertiary C−H bonds (Figure 1c). Controlling the fate of radicals
is a well-documented phenomenon18 that is underutilized in
modern C−H functionalization approaches. To implement this
strategy, we envisioned creating a high-energy, heteroatom-based
radical that could abstract a hydrogen atom. The newly formed
carbon-based radical could then be fluorinated under suitable
conditions.
The current Csp3−H fluorination methodology can be divided
into directed and undirected categories (Figure 1a−b). The
reported undirected fluorination reactions generally proceed
through carbon-based radical intermediates and have been
demonstrated with iron,7 copper,8 manganese,9 uranium,10
tungsten,11 silver,12 or organocatalysts13 (Figure 1a). A current
limitation of the existing radical C−H fluorination technology is
the inherent preference for the weakest C−H bond,14 which may
not provide the desired product. Directed fluorination can
overcome this shortcoming, yet current reports remain limited in
scope and still use palladium.15
While directed C−H functionalization has developed quickly
over the past 15 years,16 strict steric and electronic requirements
imposed by the transition-metal mediated C−H-activation step
has limited extension of this strategy to Csp3−H bonds. Here, we
propose a directed, radical C−H fluorination reaction that can
overcome such limitations and enable a late-stage fluorination.
To test the hypothesis, fluoroamide 1a was prepared from the
treatment of the parent amide with n-butyllithium and NFSI.19
Currently, our protocol for N-fluorination is limited to N-tert-
butyl amides since less-hindered amides undergo N-sulfonation
when treated with NFSI.20 While N-fluoroamides have been
studied extensively as a source of F+,21 their stability in the
context of other synthetic manipulations is poorly understood. In
general, compounds 1 are clear oils that are thermally stable to at
least 80 °C, as well as air- and silica-stable. They can be stored at
room temperature for several months and are stable to a variety
of reaction conditions (see Supporting Information, SI, for full
details). Exposure of compound 1a to a variety of initiators
revealed the surprising stability of the fluoroamide moiety, with
most reactions returning unadulterated starting material (Figure
1c and SI). Interestingly, Fe(OTf)2 in acetonitrile at 80 °C
converted compound 1a to desired product 2a in 17% yield
(entry 1, Table 1). Among all iron(II) and (III) salts tested,
Fe(OTf)2 was uniquely effective in promoting the fluorine
transfer (see SI for detailed optimization). Changing the solvent
The Hofmann−Loffler−Freytag (HLF) reaction is the earliest
̈
example of a selective C−H functionalization.17 We hypothe-
sized that, by harnessing the facile 1,5-H atom abstraction
observed in the HLF reaction, we could provide a general
solution to the functionalization of primary, secondary, and
Received: August 11, 2016
© XXXX American Chemical Society
A
DOI: 10.1021/jacs.6b08171
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
a
Table 1. Optimization of Pertinent Reaction Parameters
Table 2. Substrate Scope for Fluorine Transfer
a
entry
solvent
temp (°C) concn (M) time (h) yield (conv)
b
1
MeCN
DME
DME
DME
80
80
rt
0.1
0.1
0.5
0.5
16
16
1
17% (73%)
32% (71%)
69% (78%)
92% (96%)
b
2
3
4
40
1
a
1
Determined by H NMR with 1,3,5-trimethoxybenzene as internal
b
standard. Run with 20 mol % Fe.
to 1,2-dimethoxyethane (DME) nearly doubled the reaction
yield (entry 2, Table 1). By reducing the catalyst loading to 10
mol % and shortening the reaction to 1 h, we increased the
reaction yield and achieved much cleaner conversion to product
(entry 3, Table 1). Finally, by running the reaction at 40 °C, the
desired product could be obtained in 92% NMR yield and 86%
isolated yield (entry 4, Table 1). The reaction is iron mediated
and not promoted by visible light. Fe(OTf)2 or Fe(acac)2 does
not catalyze benzylic fluorination of the free tert-butyl amide with
NFSI or Selectfluor (see SI).
With this unique directed Csp3−H fluorination, we next
evaluated the performance of the reaction across a number of
substrates. The reaction is seemingly unaffected by the steric
environment at the benzylic carbon; primary, secondary, and
tertiary substrates perform well (2a−c, Table 2). Tertiary
fluoride 2c is prone to a small amount of HF elimination (5−
10%) under the reaction conditions, and the product was purified
on neutralized silica to minimize further elimination. Fluorine
transfers with complete chemoselectivity when multiple benzylic
C−H bonds are present, as demonstrated by 2d and 2e. The
electronic properties of the aromatic ring had a small but
noticeable effect on the reaction.
For example, while both electron-deficient and electron-rich
aryl rings provided products in good yield (2f−h, Table 2),
electron-deficient substrates react slower and require a longer
reaction time for full conversion. The reaction proceeds well in
the presence of internal and terminal alkynes and alkenes (2i−
m), which often can be problematic functional groups in
fluorination chemistry. In the case of styrenyl substrate 1l,
significant polymerization unrelated to the F transfer is observed
but can be mitigated by the addition of 0.5 mol % 3,5-di-tert-
butylcatechol. Arylboronic ester 1n provides the benzyl-fluoride
product 2n in near-quantitative yield before chromatography.
Notably, the reaction proceeds well even when the C−H bond is
distanced from the fluoroamide functionality and in the presence
of a Lewis basic sulfur as evidenced by thiophene product 2o.
Since difluoromethyl moieties often display enhanced
physiological properties relative to their fluoromethyl counter-
part,1b,22 exploring whether a fluoromethyl can undergo a second
fluorination warrants investigation. To that end, both mono-
(1p) and difluoro-2-methylisophthalamide were prepared and
subjected to the reaction conditions.23 Satisfyingly, 1q undergoes
difluorination to provide 2q as the major product (5:1 2q:2p).
Additionally, the reaction conditions allow directed allylic
fluorination as demonstrated by the selective allylic fluorination
of 1r with four competing allylic sites. Interestingly, a small
amount of α-fluorination is observed, likely due to isomerization
of the presumed intermediate allylic radical.
a
All reactions were run on 0.5 mmol scale unless otherwise noted.
b
c
Isolated yield. Isolated an 8:1 mixture of 3° fluoride/styrene. 3 h
reaction time. 6 h reaction time. With 0.5 mol % 3,5-di-tert-
butylcatechol to minimize polymerization. Product unstable to silica,
NMR yield > 95%. 2 h reaction time. 0.37 mmol scale. Isolated a
5:1 mixture of 2r: α-fluorination; observe 18% of other fluorinated
products.
d
e
f
g
h
i
Unfortunately, 2,6-disubstituted substrates provide only trace
product and extensive decomposition of the N−F bond (see SI).
While not obvious from conformational analysis, the presence of
a resident ortho substituent apparently prevents or slows H atom
abstraction, thereby allowing off-cycle pathways. Counter-
intuitively, known intermolecular C−H fluorination methods
provide lower yields for benzylic substrates relative to
unactivated C−H bonds.7−10 The work here is notable for faster
reaction times, efficacy with electron-rich arenes, and its selective
fluorination in the presence of competing activated sites.
Next, we attempted to extend the reaction to unactivated C−
H bonds. Delightfully, aliphatic 1s transfers fluorine selectively to
Scheme 1. Fluorination of an Unactivated C−H Bond
B
DOI: 10.1021/jacs.6b08171
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
form primary fluoride 2s albeit in 32% isolated yield, with 7% of
the unfluorinated secondary amide as a side product (Scheme 1).
The presumption of a radical-transfer mechanism guided the
conceptual development of this fluorination reaction. To test this
mechanistic hypothesis, experiments were conducted to uncover
evidence for the presence of radical intermediates. Interestingly,
the reaction proceeds well in the presence of 1 equiv of BHT with only
slightly lower conversion and yield. The reaction does not proceed
with 0.5 equiv of TEMPO, and the addition of TEMPO to the
reaction in progress immediately stops any further conversion.
When TEMPO is used to quench a reaction, approximately 3% of
the benzylic TEMPO-substituted product can be detected by 1H
NMR. Although the TEMPO results ostensibly suggest a radical
pathway, they cannot offer conclusive evidence since the N-oxyl
radical can be reduced by iron(II) to generate a nitroxide anion.24
Next, ortho-cyclopropylmethyl substrate 3 was synthesized
and subjected to the reaction conditions. Surprisingly, we
observed secondary fluoride 5a as the major product, instead of
the expected ring-opened product 5b (Scheme 2), suggesting a
2b, the other possible crossover product, was not detected by 1H
or 19F NMR (Scheme 3b). This surprising result is inconsistent with
a free-radical, atom-transfer mechanism initiated by iron.
Counterintuitively, the weaker N−Cl bond reacts more slowly
than the N−F bond (38% vs 89% conversion at 1 h). The high
fidelity of the atom transfer is consistent with a mechanism
wherein the iron is involved in C−X bond formation. The
discrete Fe−X intermediate formed after halogen abstraction
may react directly with the newly formed benzylic radical or be
responsible for C−H bond cleavage. The small amount of
chloride crossover may be the result of Fe−F/Fe−Cl mixing in
an off-cycle dimeric species, or from the interception of an
Fe(OTf)2Cl intermediate.
To compare the energetics of a free-radical mechanism to an
organometallic pathway, we computed the reaction profile using
DFT (Figure 2; see SI for computational details). Upon cleavage
Scheme 2. Radical-Clock Experiments
relatively short-lived radical intermediate. Although the primary
cyclopropylmethyl radical rearrangement is exceptionally fast
(1.3 × 108/s),25 benzylic stabilization likely increases the lifetime
of this radical. To mitigate the benzylic rate attenuation, we
employed trans-phenyl-substituted 4, which provided a 1.3:1
mixture of unopened and opened fluorides. With benzylic
stabilization at both sides of the cyclopropyl, these experiments
provide strong evidence for an intermediate carbon-based radical
with a lifetime close to 1 × 108/s.
To gain additional support for an intramolecular, directed
mechanism, several crossover experiments were conducted
(Scheme 3). As expected, no crossover was observed when
fluoroamide 1a reacted in the presence of 0.5 equiv of a free
amide 7 (Scheme 3a). When a 1:1 mixture of fluoroamide 1a and
chloroamide 8 was subjected to the reaction, a small amount of
crossover product 10 was detected (7%), but secondary fluoride
Figure 2. Computed (with uM06/cc-pVTZ(-f)−LACV3P**//uM06/
LACVP** level of theory) relative Gibb’s free energies for key
intermediates along the reaction pathway (for complete reaction
coordinate, see SI).
of the N−F bond by iron, the resulting N-based radical (not
shown) undergoes 1,5-hydrogen atom abstraction to form 1aC.
From benzylic radical 1aC, F-atom abstraction could occur from
either starting fluoroamide 1a or FeIIIF. The free-radical atom
transfer pathway provides transition state 1aCTS with a 25.2
kcal/mol barrier. Interestingly, removing the F atom directly
from FeIIIF converges to product 2a without an energetic barrier.
This result is consistent with the crossover experiments in
Scheme 2 and suggests an organometallic pathway, and not a
free-radical atom transfer, is operative for this reaction. Such a
mechanism would also explain the lack of reactivity with light or
other radical initiators (Figure 1). An alternative mechanism,
involving oxidation to the benzylic cation, was ruled out due to
lack of lactam formation and high energies observed computa-
tionally.
Scheme 3. Crossover Experiments
To probe the C−H cleavage step, monodeuteromethyl
fluoroamide d-1a and trideuteromethyl fluoroamide d3-1a were
prepared for kinetic-isotope-effect (KIE) analysis (Scheme 4).
With d-1a, a striking intramolecular KIE of 6.7 is almost at the
theoretical maximum,26 indicating that C−H cleavage is very
nearly complete in the transition state. The intermolecular
reaction with 1a and d3-1a was run to 3% conversion. In this case,
a primary KIE of 1.6 again is consistent with C−H cleavage, or
some prior event, as the turnover-limiting step.27
While benzylic fluorides are attractive for a number of
applications, conversion of the tert-butyl amide directing group
would offer greater versatility of the products. Gratifyingly, the
C
DOI: 10.1021/jacs.6b08171
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
Hill, M. D.; Donnelly, D. J.; Meanwell, N. A. J. Med. Chem. 2015, 58,
8315.
Scheme 4. Kinetic Isotope Effect
(2) Fujiwara, T.; O’Hagan, D. J. Fluorine Chem. 2014, 167, 16.
(3) Berger, R.; Resnati, G.; Metrangolo, P.; Weber, E.; Hulliger, J.
Chem. Soc. Rev. 2011, 40, 3496.
(4) (a) Campbell, M. G.; Ritter, T. Org. Process Res. Dev. 2014, 18, 474.
(b) Neumann, C. N.; Ritter, T. Angew. Chem., Int. Ed. 2015, 54, 3216.
(5) (a) Markovskij, L. N.; Pashinnik, V. E.; Kirsanov, A. V. Synthesis
1973, 1973, 787. (b) Al-Maharik, N.; O’Hagan, D. Aldrichimica Acta
2011, 44, 65. (c) Nielsen, M. K.; Ugaz, C. R.; Li, W.; Doyle, A. G. J. Am.
Chem. Soc. 2015, 137, 9571.
(6) (a) Adam, M. J.; Pate, B. D.; Ruth, T. J.; Berry, J. M.; Hall, L. D. J.
Chem. Soc., Chem. Commun. 1981, 733. (b) Teare, H.; Robins, E. G.;
Kirjavainen, A.; Forsback, S.; Sandford, G.; Solin, O.; Luthra, S. K.;
Gouverneur, V. Angew. Chem., Int. Ed. 2010, 49, 6821.
(7) Bloom, S.; Pitts, C. R.; Woltornist, R.; Griswold, A.; Holl, M. G.;
Lectka, T. Org. Lett. 2013, 15, 1722.
(8) Bloom, S.; Pitts, C. R.; Miller, D. C.; Haselton, N.; Holl, M. G.;
Urheim, E.; Lectka, T. Angew. Chem., Int. Ed. 2012, 51, 10580.
(9) (a) Liu, W. H. X.; Cheng, M.-J.; Nielsen, R. J.; Goddard, W. A., III;
Groves, J. T. Science 2012, 337, 1322. (b) Liu, W.; Groves, J. T. Angew.
Chem., Int. Ed. 2013, 52, 6024.
benzyl fluoride products can be reduced to the 2-(fluoromethyl)-
benzyl alcohol or benzaldehyde selectively upon treatment with
Schwartz’s reagent (see SI).28 To date, very few compounds of
this type exist in the literature.29
In conclusion, we developed a mild, directed C−H
fluorination reaction catalyzed by low-cost Fe(II) triflate. The
reaction proceeds in high yield, with broad functional-group
tolerance, under simple reaction conditions. Notably, the
methodology enables the direct fluorination of a cyclo-
propylmethyl group and provides access to 2-fluoromethylbenzyl
alcohols. Additionally, we have shown that the N-fluoroamide is a
robust, kinetically stable oxidant/functional group, despite its
high potential energy. Although the exact mechanism is currently
uncertain, the reaction most likely proceeds through short-lived
radical intermediates. Control and crossover reactions suggest
that the reaction is directed by the amide group and proceeds via
an intermediate Fe−F complex. Crossover and DFT experiments
suggest that an organometallic pathway is more likely than a free-
radical mechanism. Efforts to extend this methodology to the
fluorination of other substrate scaffolds are currently ongoing.
(10) West, J. G.; Bedell, T. A.; Sorensen, E. J. Angew. Chem., Int. Ed.
2016, 55, 8923.
(11) Nodwell, M. B.; Bagai, A.; Halperin, S. D.; Martin, R. E.; Knust,
H.; Britton, R. Chem. Commun. 2015, 51, 11783.
(12) Xu, P.; Guo, S.; Wang, L.; Tang, P. Angew. Chem., Int. Ed. 2014, 53,
5955.
(13) (a) Amaoka, Y.; Nagatomo, M.; Inoue, M. Org. Lett. 2013, 15,
2160. (b) Xia, J. B.; Zhu, C.; Chen, C. J. Am. Chem. Soc. 2013, 135,
17494.
(14) Hartwig, J. F.; Larsen, M. A. ACS Cent. Sci. 2016, 2, 281.
(15) (a) Hull, K. L.; Anani, W. Q.; Sanford, M. S. J. Am. Chem. Soc.
2006, 128, 7134. (b) McMurtrey, K. B.; Racowski, J. M.; Sanford, M. S.
Org. Lett. 2012, 14, 4094. (c) Miao, J.; Yang, K.; Kurek, M.; Ge, H. Org.
Lett. 2015, 17, 3738. (d) Zhang, Q.; Yin, X. S.; Chen, K.; Zhang, S. Q.;
Shi, B. F. J. Am. Chem. Soc. 2015, 137, 8219. (e) Zhu, Q.; Ji, D.; Liang, T.;
Wang, X.; Xu, Y. Org. Lett. 2015, 17, 3798. (f) Zhu, R. Y.; Tanaka, K.; Li,
G. C.; He, J.; Fu, H. Y.; Li, S. H.; Yu, J. Q. J. Am. Chem. Soc. 2015, 137,
7067.
(16) Lyons, T. W.; Sanford, M. S. Chem. Rev. 2010, 110, 1147.
(17) Wolff, M. E. Chem. Rev. 1963, 63, 55.
(18) (a) Jasperse, C. P.; Curran, D. P.; Fevig, T. L. Chem. Rev. 1991, 91,
1237. (b) Studer, A.; Curran, D. P. Angew. Chem., Int. Ed. 2016, 55, 58.
(c) Cekovic, Z. J. Serb. Chem. Soc. 2005, 70, 287. (d) Quiclet-Sire, B.;
Zard, S. Z. Pure Appl. Chem. 2010, 83, 519.
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/jacs.6b08171.
■
S
Experimental details and spectroscopic data (PDF)
(19) Differding, E.; Bersier, P. M. Tetrahedron 1992, 48, 1595.
(20) (a) Collman, J. P.; Zhong, M.; Boulatov, R. J. Chem. Res. 2000,
2000, 230. (b) Roy, A.; Schneller, S. W. Org. Lett. 2005, 7, 3889.
(21) (a) Purrington, S. T.; Jones, W. A. J. Org. Chem. 1983, 48, 761.
(b) Satyamurthy, N.; Bida, G. T.; Phelps, M. E.; Barrio, J. R. J. Org. Chem.
1990, 55, 3373.
AUTHOR INFORMATION
Corresponding Author
*sicook@indiana.edu
Notes
■
The authors declare no competing financial interest.
(22) (a) Erickson, J. A.; McLoughlin, J. I. J. Org. Chem. 1995, 60, 1626.
(b) Meanwell, N. A. J. Med. Chem. 2011, 54, 2529.
(23) When exposed to the N-fluorination conditions, amides 2
undergo intramolecular displacement to form the corresponding
isoindolinone.
(24) (a) Elouarzaki, K.; Mandoc, L.-R. P.; Gorgy, K.; Holzinger, M.;
Amarandei, C.-A.; Ungureanu, E.-M.; Cosnier, S. Electrochem. Commun.
2015, 60, 131. (b) Maiti, S.; Aydin, Z.; Zhang, Y.; Guo, M. Dalton Trans.
2015, 44, 8942. (c) Xu, F.; Song, X.-N.; Sheng, G.-P.; Luo, H.-W.; Li, W.-
W.; Yao, R.-S.; Yu, H.-Q. ACS Sustainable Chem. Eng. 2015, 3, 1756.
(25) Griller, D.; Ingold, K. U. Acc. Chem. Res. 1980, 13, 317.
(26) Wiberg, K. B. Chem. Rev. 1955, 55, 713.
ACKNOWLEDGMENTS
■
We acknowledge funds from Indiana University in partial
support of this work. Jenna E. Bingham is acknowledged for
assistance in preparing radical clock 4. Prof. Mu-Hyun Baik is
acknowledged for granting access to computational software and
training. We also gratefully acknowledge the American Chemical
Society Petroleum Research Fund (PRF52233-DNI1) and the
NSF CAREER Award (CHE-1254783). Eli Lilly & Co. and
Amgen supported this work through the Lilly Grantee Award and
the Amgen Young Investigator Award.
(27) Simmons, E. M.; Hartwig, J. F. Angew. Chem., Int. Ed. 2012, 51,
3066.
(28) Yi, J.; Yang, L.; Xia, C.; Li, F. J. Org. Chem. 2015, 80, 6213.
(29) (a) Ni, C.; Zhang, L.; Hu, J. J. Org. Chem. 2008, 73, 5699. (b) Hu,
J.; Gao, B.; Li, L.; Ni, C.; Hu, J. Org. Lett. 2015, 17, 3086.
REFERENCES
■
(1) (a) Champagne, P. A.; Desroches, J.; Hamel, J. D.; Vandamme, M.;
Paquin, J. F. Chem. Rev. 2015, 115, 9073. (b) Gillis, E. P.; Eastman, K. J.;
D
DOI: 10.1021/jacs.6b08171
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX