Photoredox Activation of SF6for Fluorination

Article abstract of DOI:10.1002/anie.201608792

We report the first practical use of SF₆as a fluorinating reagent in organic synthesis. Photoredox catalysis enables the in situ conversion of SF₆, an inert gas, into an active fluorinating species by using visible light. Under these conditions, deoxyfluorination of allylic alcohols is effected with high chemoselectivity and is tolerant of a wide range of functional groups. Application of the methodology in a continuous-flow setup achieves comparable yields to those obtained with a batch setup, while providing drastically increased material throughput of valuable allylic fluoride products.

Full text of DOI:10.1002/anie.201608792

Angewandte
Communications
Chemie
International Edition: DOI: 10.1002/anie.201608792
German Edition: DOI: 10.1002/ange.201608792
Fluorination
Hot Paper
Photoredox Activation of SF6 for Fluorination
T. Andrew McTeague and Timothy F. Jamison*
Abstract: We report the first practical use of SF6 as a fluorinat-
ing reagent in organic synthesis. Photoredox catalysis enables
the in situ conversion of SF6, an inert gas, into an active
fluorinating species by using visible light. Under these con-
ditions, deoxyfluorination of allylic alcohols is effected with
high chemoselectivity and is tolerant of a wide range of
Figure 1. A) Commonly employed sulfur-based deoxyfluorination
functional groups. Application of the methodology in a con-
tinuous-flow setup achieves comparable yields to those
obtained with a batch setup, while providing drastically
increased material throughput of valuable allylic fluoride
products.
À
reagents. B) Examples of C F bond formation with SF6.
T
he incorporation of fluorine into a molecule can dramat-
ically alter its properties and impart unique or desirable
characteristics.[1] Despite their utility, currently available
fluorinating reagents for preparing organofluorines have
significant limitations and fluorination remains an important
challenge. In particular, the widely utilized sulfur-based
nucleophilic reagents, such as DAST (1; Figure 1A), can be
explosive, lack functional-group tolerance, generate signifi-
cant side products, and be prohibitively expensive.[2,3] The
recently disclosed 2-pyridinesulfonyl fluoride (PyFluor, 2) is
currently the exception within this class in that it is accessible
from inexpensive reagents, and exhibits both high stability
and chemoselectivity.[3] Owing to the scarcity of suitable
nucleophilic fluorinating reagents we sought to develop a new
chemoselective, safe, and inexpensive alternative that could
offer complementary reactivity to available methods. Herein
we report the first practical use of sulfur hexafluoride (SF6) as
a fluorinating reagent in organic synthesis.
Figure 2. Deoxyfluorination of allylic alcohols by photoredox catalysis
using SF6.
Figure 3. Photoredox activation of SF6.
(Figure 3), a photocatalyst in its excited state would first
transfer an electron to SF6, generating an SF6 radical anion
(3). Subsequent fragmentation of 3 would lead to the in situ
formation of several anionic and radical fluorine species[9]
which could perform fluorination.
Sulfur hexafluoride is an inexpensive, inert gas, and thus
would be an attractive source of fluorine atoms. Although it is
widely available owing to its industrial applications, examples
of its use in organic synthesis remain scarce.[4,5] Furthermore,
Given the reactivity of the established sulfur-based
nucleophilic fluorinating reagents,[2] we envisioned that 3 or
its fragmentation products may effect deoxyfluorination. We
therefore began our investigations by exposing various
alcohols to commonly employed photoredox conditions
using Ru(bpy)3(PF6)2 as a catalyst, diisopropylethyl amine
(DIPEA) as a stoichiometric reductant, and blue LED
(470 nm) irradiation. Allylic alcohol 4 successfully underwent
deoxyfluorination to yield the corresponding allylic fluorides
5l and 5b in a combined 27% yield and a 1.7:1 ratio of
linear (l) to branched (b) isomers (Table 1, entry 1). As
a result of the particularly high value of allylic fluorides as
versatile synthons and bioactive compounds,[10] we targeted
deoxyfluorination of allylic alcohols by SF6 for further
development.
À
the only successful examples of C F bond formation with SF6
involve gas-phase reaction conditions or extremely high
temperatures and pressures,[6] highlighting the difficulty of
using this otherwise inert gas as a fluorination reagent
(Figure 1B)
A mechanistically distinct approach to accessing highly
reactive species is photoredox catalysis (Figure 2).[7] Based on
previous reports describing the use of SF6 as an electron
scavenger[8] we hypothesized that SF6 may engage in single-
electron-transfer processes. In the proposed pathway
[*] T. A. McTeague, Prof. Dr. T. F. Jamison
Department of Chemistry
Systematic optimization studies[11] revealed that DIPEA
was critical for promoting the desired reactivity. Replacement
of DIPEA with less-hindered reductants led to decreased
yields and increased formation of ammonium adducts[12]
(entry 2). In MeCN, additional acetamide byproducts, pre-
sumably formed through Ritter-type reaction with the
Massachusetts Institute of Technology
77 Massachusetts Ave., Cambridge, MA 02139 (USA)
E-mail: tfj@mit.edu
Supporting information for this article can be found under:
http://dx.doi.org/10.1002/anie.201608792.
Angew. Chem. Int. Ed. 2016, 55, 1 – 5
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
These are not the final page numbers!

Angewandte
Communications
Chemie
Table 1: Optimization of deoxyfluorination using SF6.
(Table 2). Deoxyfluorination of both primary (6) and secon-
dary (7) allylic alcohols proceeded efficiently with SF6.
Additionally, the reaction could be performed on a variety
of substitution patterns (8–10). More complex and sterically
congested alcohols (11, 14–15) also performed well. In
general, most substrates exhibited a preference for direct
substitution (a) over allylic substitution (g) (e.g. 6 and 7 arise
from the linear and branched allylic alcohols, respectively).
This selectivity is enhanced with increased substitution at the
g-position (9, 12) giving rise to linear:branched ratios as high
as 11.3:1. Tertiary allylic alcohols, however, are unreactive
(13).
This method is also tolerant of a wide range of functional
groups including arenes (5), unprotected amines (16), hetero-
cycles (17, 19), amides (18), carbamates (19) and aldehydes
(21). Notably, substrates possessing carbonyl groups (18–20)
do not undergo competing gem-difluorination or form acyl
fluorides as with alternative reagents.[15]
Entry
Catalyst
Solvent
Conv.[a]
Yield (l:b)[a]
1[b,c]
2[b,d]
3[b,c]
4[b,c]
5[b,c]
6[e]
Ru(bpy)3(PF6)2
Ru(bpy)3(PF6)2
Ru(bpy)3(PF6)2
Ru(bpy)3(PF6)2
Ru(bpy)3(PF6)2
Ru(bpy)3(PF6)2
Ir(ppy)2(dtbbpy)PF6
Ir(ppy)2(dtbbpy)PF6
MeCN
MeCN
DMSO
tAmOH
DCE
DCE
DCE
DCE
74
94
8
34
90
93
96
75
27 (1.7:1)
8 (3:1)
4 (1:1)
14 (2.5:1)
32 (1.3:1)
32 (1.3:1)
55 (1.2:1)
38 (1.7:1)
7[e]
8[e,f]
[a] Determined by 1H NMR with dimethyldiphenylsilane as an internal
standard. [b] Reaction run for 24 h. [c] 10 equiv DIPEA. [d] 10 equiv N(n-
Bu)3. [e] 5 mol% catalyst. [f] NEt3·3HF (1 equiv) added.
solvent, were also isolated. Based on this observation and the
significant solvent effects exhibited in heterogeneous[13] and
nucleophilic fluorination reactions,[1] several solvents were
evaluated. While decreased conversions and yields were
observed for highly polar (DMSO, entry 3) and protic
solvents (tAmOH, entry 4), the use of the less polar dichloro-
ethane (DCE) gave an improved 32% yield (1.3:1 l:b) of 5
(entry 5).
The reaction was also greatly accelerated in DCE
providing identical yields in nearly half the time (14 vs. 24 h,
entries 56). These results suggest that the reaction outcome
relies on both the solvation of the active fluorinating reagent
and the solubility of SF6, which is higher in nonpolar
solvents.[14] Further improvement in reactivity was provided
by changing the photocatalyst to Ir(ppy)2(dtbbpy)PF6, which
has a higher solubility in DCE, allowing for the amount of
DIPEA to be reduced from 10 equivalents to 3 equivalents
(entry 7). Under these conditions, a 55% yield of 5 (1.2:1 l:b)
was achieved. To increase reaction yields, exogenous fluoride
sources, such as triethylamine-HF and TBAT,[11] were also
investigated, but no improvement in yield was observed
(entry 8).
Remarkably, both propargylic (22; Scheme 1A) and non-
allylic alcohols (24; Scheme 1B) could be used without
significantly hindering the formation of 6 and 25, respectively,
or undergoing deoxyfluorination, which allows the reaction to
be performed in the presence of unprotected non-allylic
alcohols (Scheme 1). To our knowledge, this is the first
example of the ability to differentiate between alcohols in this
way, demonstrating the complementarity of deoxyfluorina-
tion with SF6.
One area of chemistry that can provide considerable
improvements to both photochemical and multiphasic reac-
Scheme 1. Selective deoxyfluorination. [a] SF6, Ir(ppy)2(dtbbpy)PF6
(5 mol%), DIPEA (3 equiv), DCE (0.075m), irradiation with blue LEDs,
room temperature, 14 h. Linear:branched (l:b) ratio determined by
NMR spectroscopic analysis.
With the optimized reaction conditions in hand, we
investigated the substrate scope of this transformation
Table 2: Scope of deoxyfluorination.
[a] Run at 608C. [b] From (À)-trans-carveol. [c] From (À)-cis-carveol. Linear:branched (l:b) ratio determined by NMR spectroscopic analysis.
www.angewandte.org ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2016, 55, 1 – 5
2
These are not the final page numbers!

Angewandte
Communications
Chemie
tions is continuous flow.[16] Owing to decreased path lengths
be achieved safely and controllably. Additionally, the assem-
and increased interfacial area, mass- and energy-transfer rates
are significantly enhanced in such systems, often providing
superior results relative to batch reactions.[17] The ability to
precisely control reaction conditions can also improve scal-
ability and safety.
bled setup (2.7 mL in volume) was capable of almost five
times the material throughout (0.19 mmol product/h vs.
0.04 mmol/h in batch).
We also carried out preliminary investigations of the
mechanism of this transformation. In particular, we looked to
À
To examine these advantages with regard to the devel-
oped methodology, a simple flow setup was assembled
(Scheme 2A). Within it, substrate, catalyst, and reductant
are pumped as a single stream and then combined with
gaseous SF6 via a Y-mixer (M1). The flow rate of SF6 is
adjusted to create a uniform 1:1 gas:liquid segmented (“slug”)
flow in the reaction line. Prior to irradiation, the two phases
mix in R1 (residence time: tR = 1 min) before entering R2 (tR =
15 min), which is illuminated by blue LEDs. Following the
reaction, a back-pressure regulator (BPR) is used to place the
entire system under 100 psi of pressure and increase the
solubility of SF6 in solution.
gain better understanding of the nature of C F bond
À
formation. Prior to C O bond cleavage, we anticipated that
the alcohol would likely attack the sulfur atom of some SFx
species, since a similar step has been proposed for other
sulfur-based deoxyfluorination reagents. As would be
expected based on this proposal, when oxygen was protected
(27), no reactivity was observed (Scheme 3). Furthermore, the
À
formation of an O S bond is supported by a dialkyl sulfite
side product[13] observed in small quantities under the
reaction conditions.
Under the optimized conditions, 6 was furnished in 79%
yield relative to conversion of starting material, with a total
residence time of 16 min, as compared to a 55% yield
obtained in 14 h in batch (entries 1 and 3 in Scheme 2).
Similar improvements were also observed for 11 and 12
(Scheme 2B). Interestingly, in continuous flow, starting
material is consumed within 2 min, generating the desired
products as well as a previously unobserved intermediate,
most consistent with (RO)2SF4. Over the course of the
reaction, this intermediate is gradually converted back into
starting material, preventing undesired side reactions. Given
that elevated pressure (100 psi) did not produce similar
conversions and yields in batch, it is likely that the increased
pressure, mass-transfer, and interfacial area of the continu-
ous-flow system all contribute to the formation of the
intermediate and the improved yield-to-conversion ratio.
Also noteworthy are the safety and throughput enabled under
continuous flow. In the system, SF6 can be easily handled with
use of a BPR, allowing for the elevated pressures utilized to
Scheme 3. Standard conditions (std. cond.): SF6, Ir(ppy)2(dtbbpy)PF6
(5 mol%), DIPEA (3 equiv), DCE (0.075m), irradiation with blue LEDs,
room temperature, 14 h. Linear:branched (l:b) ratio determined by
NMR spectroscopic analysis.
À
We further hypothesized that C O bond cleavage could
occur via a one- or two-electron process. The addition of one
equivalent of TEMPO, however, had little effect on the
reaction, giving a 47% yield (1.4:1 l:b) of 6 compared to 55%
yield obtained in its absence. Additionally, no evidence for
the formation of TEMPO adducts was observed in these
reactions suggesting the absence of fluorine radicals and
sufficiently long-lived radical intermediates of the starting
À
material. The two-electron nature of C F bond formation was
further supported by radical clock experiments (Scheme 3). If
during the course of the reaction an allylic radical is generated
À
via C O bond homolysis, a 5-exo-trig cyclization or radical
cyclopropane ring opening would be expected to occur for
substrates 28 and 31, respectively. However, neither 30 nor 33
were observed (Scheme 3).
Although further studies are necessary to reveal mecha-
nistic details, preliminary experiments indicate that the
mechanism of this deoxyfluorination may share features
with that of the chlorination of alcohols with thionyl chloride.
À
À
In this case, C O bond cleavage precedes C X bond
formation from an undissociated ion pair. As noted, the
formation of both ammonium salts and Ritter-type products
derived from the allylic alcohol were observed, suggesting the
presence of an intermediate with significant cationic charac-
ter. Furthermore, deoxyfluorination of trans and cis-(À)-
carveol proceeded with overall retention of stereochemistry
Scheme 2. Continuous-flow deoxyfluorination. [a] Determined by
1H NMR with dimethyldiphenylsilane as an internal standard. [b] Yield
relative to conversion of starting material.
Angew. Chem. Int. Ed. 2016, 55, 1 – 5
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3
These are not the final page numbers!

Angewandte
Communications
Chemie
(Table 2, 14 and 15 respectively), which is also observed in the
reaction of allylic alcohols with thionyl chloride.[18]
Taken together these results support a mechanism in
À
which the alcohol is first activated via O S bond formation
resulting in a R-O-SFx intermediate. An intimate ion pair is
[1] a) T. Liang, C. N. Neumann, T. Ritter, Angew. Chem. Int. Ed.
2013, 52, 8214; Angew. Chem. 2013, 125, 8372; b) S. Purser, P. R.
Moore, S. Swallow, V. Gouverneur, Chem. Soc. Rev. 2008, 37,
320; c) C. N. Neumann, T. Ritter, Angew. Chem. Int. Ed. 2015,
54, 3216; Angew. Chem. 2015, 127, 3261.
[2] C. Ni, M. Hu, J. Hu, Chem. Rev. 2015, 115, 765.
[3] M. K. Nielsen, C. R. Ugaz, W. Li, A. G. Doyle, J. Am. Chem. Soc.
2015, 137, 9571.
[4] a) K. Seppelt, Chem. Rev. 2014, 115, 1296; b) B. G. Harvey,
A. M. Arif, A. Glçckner, R. D. Ernst, Organometallics 2007, 26,
2872; c) “Fluorine-Containing Compounds of Sulfur”: G. Cady
in Advances in Inorganic Chemistry and Radiochemistry (Eds.:
H. J. Emelꢀus, A. G. Sharpe), Academic Press, New York, 1960,
p. 105.
[5] a) L. Zꢁmostnꢁ, T. Braun, Angew. Chem. Int. Ed. 2015, 54,
10652; Angew. Chem. 2015, 127, 10798; b) P. Holze, B. Horn, C.
Limberg, C. Matlachowski, S. Mebs, Angew. Chem. Int. Ed. 2014,
53, 2750; Angew. Chem. 2014, 126, 2788.
[6] a) F. R. Cruickshank, L. Batt, J. Phys. Chem. 1966, 70, 723;
b) A. P. Hagen, B. W. Callaway, Inorg. Chem. 1975, 14, 2825.
[7] C. K. Prier, D. A. Rankic, D. W. C. MacMillan, Chem. Rev. 2013,
113, 5322.
[8] P. Giacomello, F. Pepi, J. Phys. Chem. 1993, 97, 4421.
[9] a) S. Solovev, A. Palmentieri, N. D. Potekhina, T. E. Madey, J.
Phys. Chem. C 2007, 111, 18271; b) M. Fenzlaff, R. Gerhard, E.
Illenberger, J. Chem. Phys. 1988, 88, 149.
[10] a) M. C. Pacheco, S. Purser, V. Gouverneur, Chem. Rev. 2008,
108, 1943 – 1981; Recent methods for the preparation of allylic
fluorides: b) M.-G. Braun, A. G. Doyle, J. Am. Chem. Soc. 2013,
135, 12990; c) Q. Zhang, D. P. Stockdale, J. C. Mixdorf, J. J.
Topczewski, H. M. Nguyen, J. Am. Chem. Soc. 2015, 137, 11912;
d) E. Benedetto, M. Tredwell, C. Hollingworth, T. Khotavivat-
tana, J. M. Bornw, V. Gouverneur, Chem. Sci. 2013, 4, 89.
[11] See Supporting Information for additional optimization and
control data.
À
then formed by C O bond cleavage allowing for fluoride to
be delivered from the generated [OSFx]À anion, which serves
as the active fluorinating reagent, producing the observed
products with retention of stereochemistry. The additional
stabilization provided by the adjacent alkene facilitates
formation of the proposed ion pair, enabling allylic alcohols
to be uniquely reactive under these conditions. Of particular
note, this methodology offers a stereochemical outcome
complementary to most currently available nucleophilic
fluorinations. When deoxyfluorination of trans and cis-(À)-
carveol is performed with DAST, cis/trans ratios of 1:1.5 and
1:5.7, respectively, were observed, as compared to ratios of 1:5
(14) and 5:1 (15) obtained using SF6.
In summary, we have developed the first example of the
use of SF6 as a selective fluorinating reagent in organic
synthesis via a novel activation method. As a reagent, SF6 is
unparalleled in its safety, and is only a fraction of the cost of
comparable reagents (under $4molÀ1, Airgas). In application
to the deoxyfluorination of allylic alcohols, SF6 was shown to
be highly chemoselective and tolerant of a wide array of
functional groups, supplying good to moderate yields of the
corresponding allylic fluorides, linear to branched ratios as
high as 11.3:1, and reactivity patterns complementary to
current methods. The continuous-flow format was also shown
to be beneficial, providing improved yields, operational
simplicity, and substantially higher product throughput.
Lastly, the photoredox activation of SF6 demonstrated in
this study provides a basis to explore the broader applications
of SF6 as a reagent in organic synthesis.
[12] See Supporting Information for a list of identified reaction
products.
[13] P. J. Dyson, P. G. Jessop, Catal. Sci. Technol. 2016, 6, 3302.
[14] Y. Marcus, J. Phys. Chem. B 1997, 101, 8617.
[15] a) R. P. Singh, J. M. Shreeve, Synthesis 2002, 2561; b) K. L. Kirk,
Org. Process Res. Dev. 2008, 12, 305.
[16] C. J. Mallia, I. R. Baxendale, Org. Process Res. Dev. 2015, 20,
327.
[17] R. L. Hartman, J. P. McMullen, K. F. Jensen, Angew. Chem. Int.
Ed. 2011, 50, 7502; Angew. Chem. 2011, 123, 7642.
[18] W. G. Young, F. F. Caserio, Jr., D. D. Brandon, Jr., J. Am. Chem.
Soc. 1960, 82, 6163.
Acknowledgements
We thank the Novartis-MIT Center for Continuous Manu-
facturing for financial support and several Novartis colleagues
for very helpful suggestions, in particular, Berthold Schenkel,
Benjamin Martin, Jçrg Sedelmeier, Gerhard Penn, Francesco
Venturoni, and Julien Haber. We are also grateful to Li Li for
HR-MS data, Elizabeth Kelley, and Drs. Matthew Katcher,
Evan Styduhar, Eric Standley, Christina Dai, Rachel
Beingessner for helpful discussions.
Keywords: alcohols · continuous flow · fluorination ·
Received: September 8, 2016
photoredox catalysis · sulfur hexafluoride
Published online: && &&, &&&&
4
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1 – 5
These are not the final page numbers!

Angewandte
Communications
Chemie
Communications
Fluorination
T. A. McTeague,
T. F. Jamison*
&&&& — &&&&
Photoredox Activation of SF6 for
Fluorination
Unlocking SF6’s potential: Photoredox
catalysis enables the conversion of SF6,
an inert gas, into an active fluorinating
species using visible light. Under both
batch and continuous-flow conditions,
deoxyfluorination of allylic alcohols is
effected with high chemoselectivity and is
tolerant of a wide range of functional
groups.
Angew. Chem. Int. Ed. 2016, 55, 1 – 5
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5
These are not the final page numbers!