Palladium-catalyzed regio- and enantioselective fluorination of acyclic allylic halides

Article abstract of DOI:10.1021/ja206960k

This report describes the Pd(0)-catalyzed fluorination of linear allylic chlorides and bromides, yielding branched allylic fluorides in high selectivity. Many of the significant synthetic limitations previously associated with the preparation of these products are overcome by this catalytic method. We also demonstrate that a chiral bisphosphine-ligated palladium catalyst enables highly enantioselective access to a class of branched allylic fluorides that can be readily diversified to valuable fluorinated products.

Full text of DOI:10.1021/ja206960k

COMMUNICATION
pubs.acs.org/JACS
Palladium-Catalyzed Regio- and Enantioselective Fluorination of
Acyclic Allylic Halides
Matthew H. Katcher, Allen Sha, and Abigail G. Doyle*
Department of Chemistry, Princeton University, Princeton, New Jersey 08544, United States
S Supporting Information
b
Table 1. Optimization of Branched-Selective Fluorination
ABSTRACT: This report describes the Pd(0)-catalyzed
fluorination of linear allylic chlorides and bromides, yielding
branched allylic fluorides in high selectivity. Many of the
significant synthetic limitations previously associated with
the preparation of these products are overcome by this
catalytic method. We also demonstrate that a chiral bispho-
sphine-ligated palladium catalyst enables highly enantiose-
lective access to a class of branched allylic fluorides that can
be readily diversified to valuable fluorinated products.
llylic fluorides occur in a range of pharmaceutical agents,
A
PET tracers, and agrochemicals.1 Moreover, in much the same
way that allylic alcohols are versatile intermediates in chemical
synthesis,2 chiral allylic fluorides can serve as building blocks for the
preparation of valuable fluorine-containing synthons. This privi-
leged functionality is usually assembled by allylic substitution;
however, achieving high regio- and stereoselectivity in such pro-
cesses has been a longstanding challenge. The synthesis of linear
allylic fluorides by nucleophilic displacement of halides or activated
alcohols with metal or ammonium fluorides typically proceeds with
high regioselectivity, including in a recent Pd-catalyzed example
(eq 1).3 Alternatively, the method of choice for the production of
branched allylic fluorides is deoxyfluorination of allylic alcohols with
diethylaminosulfur trifluoride (DAST) or its derivatives (eq 2).4
Unfortunately, many of these reactions proceed with only marginal
bias for the branched regioisomer. In addition, fluorinations of
chiral allylic alcohols lead to erosion in regio- and enantiopurity
through SN1 and SN20 pathways, and the reactions are intolerant of
common organic functional groups (alcohols, aldehydes,
ketones).5 To address some of these limitations, Gouverneur and
co-workers have identified an alternative approach to the synthesis
of these products by fluorodesilylation of allylsilanes.6,7 Although
the reactions are highly branched-selective, asymmetric catalytic
variants are almost completely undeveloped.8
entry
liganda
solvent
yieldb
b:lb
c
1
2
PPh3
toluene
toluene
toluene
toluene
toluene
toluene
toluene
THF
70
21
20
24
23
55
76
22
26
6
6:1
2:1
dppe
dppf
DPEphos
L1
3
2:1
4
4:1
5
3:1
6
Xantphos
L2
7:1
7
>20:1
1:1
8
L2
9
L2
CH2Cl2
CH3CN
2:1
10
L2
1:4
a Bite angles for bidentate ligands:9 dppe (86°), dppf (99°), DPEphos
(104°), L110 (107°), Xantphos (108°). b Determined by GC using
dodecane as a quantitative internal standard. c Using 30 mol % ligand.
Recent work from our laboratory has demonstrated that a
chiral bisphosphine-ligated palladium catalyst promotes enantio-
selective fluorination of cyclic allylic chlorides with AgF.11
Mechanistically, these reactions proceed in a manner analogous
to asymmetric allylic alkylations with stabilized nucleophiles:12
namely, by outer-sphere attack of fluoride on a Pd π-allyl inter-
mediate. We sought to extend this catalytic system to branched-
selective allylic fluorination of acyclic substrates and report
herein that high regio- and enantioselectivity can be obtained
for the preparation of branched allylic fluorides using Pd(0)
catalysis (eq 3). The methodology addresses many of the current
limitations for the synthesis of this motif, furnishing a diverse
collection of synthetically valuable fluorinated products under
Received: July 25, 2011
Published: September 06, 2011
r
2011 American Chemical Society
15902
dx.doi.org/10.1021/ja206960k J. Am. Chem. Soc. 2011, 133, 15902–15905
|

Journal of the American Chemical Society
COMMUNICATION
Table 2. Scope of Regioselective Allylic Fluorination
Table 3. Enantioselective Synthesis of Branched Allylic
Fluorides
a Isolated yields (combined branched and linear isomers) for reactions
carried out on a 0.5 mmol scale. b Determined by GC analysis of crude
reaction mixtures. c Determined by GC or HPLC using commercial
chiral columns. d Reaction conducted with Pd(dmdba)2 (10 mol %) for
ease of purification. e Reaction conducted in benzene. f Not determined.
mild conditions, using commercial and bench-stable reagents and
catalysts.
a Absolute configuration of 7 assigned by derivatization to a compound
with known stereochemistry; other products assigned by analogy.
b Isolated yields (combined branched and linear isomers) for reactions
carried out on 0.5ꢀ1.0 mmol scale. c Determined by GC or NMR
analysis of crude reaction mixtures. d Determined by GC or HPLC using
We began our investigation using acyclic chloride 1 as a
substrate for allylic fluorination with AgF in the presence of
Pd2(dba)3 and various ligands.13 Notably, promising levels of
regioselectivity (6:1 branched:linear) were obtained simply by
employing PPh3 as a ligand (Table 1, entry 1); in contrast,
preparation of 2 utilizing DAST affords an inseparable mixture of
isomers in a 2:1 b:l ratio.14
e
commercial chiral columns. Determined by GC using dodecane as a
f
quantitative internal standard. Reaction conducted with Pd(dmdba)2
(10 mol %) for ease of purification. g Reaction conducted with allylic
bromide as starting material.
A number of bidentate phosphines were next examined. We
found that those with larger bite angles induced higher regios-
electivity (entries 2ꢀ6), with the commercial Trost naphthyl
ligand L2 delivering product in a >20:1 b:l ratio (entry 7).15
Similar bite angle trends have been described for Pd-catalyzed
allylic substitution with outer-sphere nucleophiles,16 but reac-
tions with acyclic, unsymmetrical substrates usually afford low-to-
moderate regioselectivity in favor of the branched isomer.17ꢀ19
We hypothesize that the enhanced selectivity with fluoride arises
in part from its small size, which would facilitate nucleophilic
addition at the more hindered terminus of the Pd π-allyl
intermediate.20 Furthermore, L2 may direct regioselective allylic
substitution by hydrogen-bond donation to the nucleophile, as
proposed by Lloyd-Jones and Norrby in their stereochemical
model of Pd-catalyzed asymmetric allylic alkylations.21 As such,
the selectivity with fluoride could also be ascribed to its strong
hydrogen-bond acceptor properties.22 As would be expected
based on this proposal, nonpolar solvents such as toluene
afforded optimal regioselectivities (entries 7ꢀ10). Use of to-
luene also suppresses an unselective background reaction on
account of the low solubility of AgF in this solvent.
conditions are also tolerant of aldehydes and alkyl bromides
(entries 4 and 5). These products bear synthetically useful
functional groups, most of which are incompatible with alter-
native protocols for the preparation of allylic fluorides.4,7a
Furthermore, the Pd-catalyzed method affords minimal diene
byproducts (<5%), as determined by GC analysis of the crude
reaction mixtures.
A current limitation of the reactions in Table 2 is that, in
contrast to our previous results with cyclic allylic chlorides,
moderate to low enantioselectivity is attained with these acyclic
substrates. We reasoned that substrates possessing allylic sub-
stituents of greater steric or electronic bias might afford higher
asymmetric induction under the fluorination conditions. In the
event, we were pleased to find that various allylic chlorides and
bromides bearing α-branching or heteroatom substituents un-
dergo fluorination with 90ꢀ97% ee (Table 3, entries 1ꢀ5).24
Ethers and protected amines are well-tolerated, and motifs such
as a β-fluoroether (11) can be accessed in 90% ee and 16:1 b:l
selectivity. In contrast to these results, the synthesis of 9 by the
reaction of an enantiopure branched allylic alcohol with DAST
proceeds with significant erosion in enantiopurity (>99% ee to
49% ee) and 2:1 b:l selectivity.14 The Pd-catalyzed method
affords mainly linear product using cinnamyl chloride as substrate,
with the minor branched isomer formed in 0% ee (entry 6).
The broad substrate scope and ease of operation of the
optimized process are highlights of the catalytic methodology.
As outlined in Table 2, substrates containing benzyl and silyl
ethers undergo reaction with high regioselectivity to afford
2ꢀ4 in 78ꢀ84% yield (entries 1ꢀ3).23,14 The mild reaction
15903
dx.doi.org/10.1021/ja206960k |J. Am. Chem. Soc. 2011, 133, 15902–15905

Journal of the American Chemical Society
COMMUNICATION
Scheme 1. Regioselective Fluorination with PPh3 as Liganda
Scheme 2. Derivatization of Allylic Fluoride 10a,b
a Isolated yields (combined branched and linear isomers) for reactions
carried out on 0.5 mmol scale. b:l ratios determined by GC analysis of
crude reaction mixtures. b Determined by GC using dodecane as a
quantitative internal standard. c Reaction conducted in benzene.
Preference for this regioisomer has also been reported by
Gouverneur and co-workers in their Pd-catalyzed fluorination
of cinnamyl 4-nitrobenzoates3 and is commonly found for Pd-
catalyzed allylic alkylations with cinnamyl substrates.12,17b
Despite the broad functional group tolerance of the fluorina-
tions in Tables 1 and 2, we have discovered that certain substrates
perform poorly with L2 as ligand. For example, allylic fluoride 13
and the structurally distinct 14 and 15 are formed in low b:l
selectivity (1:1, 1:1, and 7:1, respectively).14 This problem can be
circumvented through the use of PPh3 as ligand, which provides
13ꢀ15 in 6:1 to >20:1 b:l selectivity (Scheme 1).15 In compar-
ison with alternative protocols for allylic fluorination, this
catalytic method is unique in its ability to tolerate an unprotected
alcohol. Nevertheless, 13 is produced in only moderate yield due
to competitive intramolecular allylic etherification.14 Exocyclic
methylene-containing allylic fluorides such as 14, generated in a
9:1 b:l ratio, represent a common motif found in non-natural
vitamin D analogs.25 For tertiary allylic fluoride 15, formation of a
fully substituted carbon center at the electrophile by Pd-catalyzed
allylic substitution is noteworthy.26 However, competitive diene
formation (20ꢀ30%) erodes the efficiency of this reaction, as
well as that for the synthesis of 14. A final limitation is that the
fluorinations with either L2 or PPh3 are not compatible with
secondary alkyl amines as these substrates undergo competitive
N-alkylation.
a Reagents and conditions: (a) Pd(quinox)Cl2, AgSbF6, TBHP, CH2Cl2,
0 °C to rt; (b) O3, then NaBH4, CH2Cl2/MeOH, ꢀ78 °C to rt; (c)
NBSH, Et3N, CH2Cl2, 0 °C to rt; (d) AD-mix β, NaHCO3, t-BuOH/
H2O, 0 °C to rt; (e) 9-BBN, 0 to 40 °C, then NaOH, H2O2, THF rt; (f)
HoveydaꢀGrubbs II, 5-hexenyl acetate, CH2Cl2, 100 °C. b Isolated
yields for reactions carried out on a 0.2 mmol scale. ee’s determined by
HPLC using commercial chiral columns.
or impossible to access in enantioenriched form, such as acyclic α-
fluoroketone 16,32 aliphatic fluoride 18, diol 19,28 and γ-fluoroal-
cohol 20, can be prepared in one step from allylic fluoride 10.
In conclusion, we have developed a new approach to the
highly regioselective synthesis of branched allylic fluorides that is
operationally trivial and displays unprecedented functional group
tolerance. The utility of this transformation has been highlighted
via the expedient synthesis of a wide range of valuable fluorine-
containing building blocks. Since Ag18F is readily available,33 we
intend to optimize the process for future applications in the
production of radiotracers for PET imaging.34 Our future
investigations will also focus on elucidating the origin of the
noteworthy ligand-dependent regioselectivity observed in this
Pd-catalyzed process and extending the approach to non-allylic
CꢀF bond formation.
Overall, access to fluorine-containing products by asymmetric
allylic substitution provides an opportunity to rapidly generate a
wide array of fluorinated synthons from a common, bench-stable
precursor. To illustrate this point, we elaborated allylic fluoride
10 via Wacker oxidation,27 ozonolysis/reduction, hydrogenation,
diastereoselective dihydroxylation,28 hydroboration/oxidation,
and cross-metathesis.29 Despite the versatility of the allyl func-
tional group, some of the transformations in Scheme 2 have
never been carried out on allylic fluorides, and most have not
been performed on enantioenriched material due to the paucity
of methods for their synthesis. Excellent stereofidelity is observed
in each of the transformations, with the exception of the cross-
metathesis reaction, which results in erosion from 93 to 64% ee
due to the forcing conditions required for productive bond
construction. Motifs such as those found in enantioenriched β-
fluoroalcohol 1730 and internal allylic fluoride 2131 may now be
prepared by both asymmetric catalytic electrophilic and nucleo-
philic fluorination. Notably, products that are otherwise difficult
’ ASSOCIATED CONTENT
S
Supporting Information. Experimental procedures, ad-
b
ditional reaction optimization, X-ray crystallographic structure of
10, and spectroscopic data for all new compounds. This material
is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
Corresponding Author
agdoyle@princeton.edu
’ ACKNOWLEDGMENT
We thank Scott Semproni for X-ray crystallographic structure
determination, Henry Moss for substrate synthesis, and Lotus
Separations for chiral separations. Financial support provided by
Merck and Princeton University is gratefully acknowledged.
15904
dx.doi.org/10.1021/ja206960k |J. Am. Chem. Soc. 2011, 133, 15902–15905

Journal of the American Chemical Society
COMMUNICATION
’ REFERENCES
(1) Pacheco, M. C.; Purser, S.; Gouverneur, V. Chem. Rev. 2008,
108, 1943–1981.
(2) Hoveyda, A. H.; Evans, D. A.; Fu, G. C. Chem. Rev. 1993,
93, 1307–1370.
(3) Hollingworth, C.; Hazari, A.; Hopkinson, M. N.; Tredwell, M.;
Benedetto, E.; Huiban, M.; Gee, A. D.; Brown, J. M.; Gouverneur, V.
Angew. Chem., Int. Ed. 2011, 50, 2613–2617.
(4) (a) Singh, R. P.; Shreeve, J. M. Synthesis 2002, 2561–2578.
(b) Boukerb, A.; Grꢀee, D.; Laabassi, M.; Grꢀee, R. J. Fluorine Chem.
1998, 88, 23–27.
(19) For examples where regioselectivity has been tuned by ligand
design, see: (a) Prꢀet^ot, R.; Pfaltz, A. Angew. Chem., Int. Ed. 1998,
37, 323–325. (b) Hayashi, T. Acc. Chem. Res. 2000, 33, 354–362. (c)
Dai, L.-X.; Tu, T.; You, S.-L.; Deng, W.-P.; Hou, X.-L. Acc. Chem. Res.
2003, 36, 659–667.
(20) Trost, B. M.; Machacek, M. R.; Aponick, A. Acc. Chem. Res.
2006, 39, 747–760.
(21) A hydrogen bond has been proposed as a key interaction for
achieving high enantioselectivity with L1: Butts, C. P.; Filali, E.; Lloyd-
Jones, G. C.; Norrby, P.-O.; Sale, D. A.; Schramm, Y. J. Am. Chem. Soc.
2009, 131, 9945–9957.
(22) Clark, J. H. Chem. Rev. 1980, 80, 429–452.
(5) For an example, see: Bresciani, S.; Slawin, A. M. Z.; O’Hagan, D.
J. Fluorine Chem. 2009, 130, 537–543.
(23) Synthesis of 2ꢀ4 using DAST provides 2:1 b:l selectivity.
Fluorinations with PPh3 as ligand provide 2ꢀ6 in <7:1 b:l selectivity.
(24) Allylic bromides were used as substrates when the rate of
reaction and selectivity were inferior with allylic chlorides.
(25) Kabat, M. M.; Radinov, R. Curr. Opin. Drug. Discovery Dev.
2001, 4, 808–833.
(6) Thibaudeau, S.; Gouverneur, V. Org. Lett. 2003, 5, 4891–4893.
(7) For fluorodesilylation of cyclic and non-terminal acyclic
allylsilanes, see: (a) Purser, S.; Wilson, C.; Moore, P. R.; Gouverneur,
V. Synlett 2007, 1166–1168. (b) Giuffredi, G. T.; Purser, S.; Sawicki, M.;
Thompson, A. L.; Gouverneur, V. Tetrahedron: Asymmetry 2009, 20,
910ꢀ920 and references therein.
(26) For other recent examples, see: (a) Zhang, P.; Le, H.; Kyne,
R. E.; Morken, J. P. J. Am. Chem. Soc. 2011, 133, 9716–9719. (b) Trost,
B. M.; Malhotra, S.; Chan, W. H. J. Am. Chem. Soc. 2011, 133,
7328ꢀ7331 and references therein.
(8) For the enantioselective synthesis of tertiary allylic fluorides
using NFSI and a bis-cinchona alkaloid catalyst, see: (a) Ishimaru, T.;
Shibata, N.; Horikawa, T.; Yasuda, N.; Nakamura, S.; Toru, T.; Shiro, M.
Angew. Chem., Int. Ed. 2008, 47, 4157–4161. For a stoichiometric
example, see: (b) Greedy, B.; Paris, J.-M.; Vidal, T.; Gouverneur, V.
Angew. Chem., Int. Ed. 2003, 42, 3291–3294.
(27) To the best of our knowledge, Wacker oxidations of allylic
fluorides are unknown. Michel, B. W.; Camelio, A. M.; Cornell, C. N.;
Sigman, M. S. J. Am. Chem. Soc. 2009, 131, 6076–6077.
(28) For the dihydroxylation of enantioenriched allylic fluorides, see
refs 5, 7b, and: (a) Ahmed, Md. M.; O’Doherty, G. A. Carbohydr. Res.
2006, 341, 1505–1521. (b) Kitade, Y.; Ando, T.; Yamaguchi, T.; Hori,
A.; Nakanishi, M.; Ueno, Y. Bioorg. Med. Chem. 2006, 14, 5578–5583.
(29) For cross-metathesis of racemic allylic fluorides, see: Thibau-
deau, S.; Fuller, R.; Gouverneur, V. Org. Biomol. Chem. 2004,
2, 1110–1112.
(9) Birkholz, M.-N.; Freixa, Z.; van Leeuwen, P. W. N. M. Chem. Soc.
Rev. 2009, 38, 1099–1118.
(10) To the best of our knowledge, the bite angle of L2 has not been
reported. The bite angle of an L1ꢀPt complex has been computationally
estimated at 107°: (a) Ohshima, T.; Miyamoto, Y.; Ipposhi, J.; Nakahara,
Y.; Utsunomiya, M.; Mashima, K. J. Am. Chem. Soc. 2009,
131, 14317–14328. For a crystal structure of a ligand related to L1
and L2, see:(b) Trost, B. M.; Breit, B.; Peukert, S.; Zambrano, J.; Ziller,
J. W. Angew. Chem., Int. Ed. Engl. 1995, 34, 2386–2388.
(30) Beeson, T. D.; MacMillan, D. W. C. J. Am. Chem. Soc. 2005,
127, 8826–8828.
(31) Jiang, H.; Falcicchio, A.; Jensen, K. L.; Paix~ao, M. W.; Bertelsen,
S.; Jørgensen, K. A. J. Am. Chem. Soc. 2009, 131, 7153–7157.
(32) Enantioenriched cyclic α-fluoroketones are accessible by orga-
nocatalysis: Kwiatkowski, P.; Beeson, T. D.; Conrad, J. C.; MacMillan,
D. W. C. J. Am. Chem. Soc. 2011, 133, 1738–1741.
(33) Caires, C. C.; Guccione, S. U.S. Patent Appl. 20100152502,
2010.
(34) Ametamey, S. M.; Honer, M.; Schubiger, P. A. Chem. Rev. 2008,
108, 1501–1516.
(11) Katcher, M. H.; Doyle, A. G. J. Am. Chem. Soc. 2010, 132,
17402–17404.
(12) For reviews, see: (a) Trost, B. M.; Van Vranken, D. L. Chem. Rev.
1996, 96, 395–422. (b) Lu, Z.; Ma, S. Angew. Chem., Int. Ed. 2008, 47, 258–
297. (c) Trost, B. M.; Zhang, T.; Sieber, J. D. Chem. Sci. 2010, 1, 427–440.
(13) Highly branched-selective allylic substitution can be achieved
with other transition metal catalysts, including Ir, Mo, and Cu (see ref
12); however, we found that these metals provided low regioselectivity
for fluorination with allylic chlorides.
(14) See Supporting Information.
(15) We do not yet fully understand why PPh3 provides superior
selectivity to many of the bidentate ligands in Table 1 and to L2 for
products 13ꢀ15 in Scheme 1. However, π-acceptor ligands have been
shown to induce high b:l selectivity in other Pd-catalyzed allylic
substitution reactions: (a) Åkermark, B.; Zetterberg, K.; Hansson, S.;
Krakenberger, B.; Vitagliano, A. J. Organomet. Chem. 1987, 335, 133–
142. Since PPh3 is a better π-acceptor than many of the bidentate
ligands in Table 1, this electronic effect may account for the higher
regioselectivity. Additionally, the PꢀPdꢀP angle in a [(PPh3)2Pd-
(cinnamyl)]BF4 complex is 104°, similar in magnitude to the larger bite
angles of the bidentate ligands of Table 1: (b) Baize, M. W.; Blosser,
P. W.; Plantevin, V.; Schimpff, D. G.; Gallucci, J. C.; Wojcicki, A.
Organometallics 1996, 15, 164–173.
(16) van Haaren, R. J.; Goubitz, K.; Fraanje, J.; van Strijdonck,
G. P. F.; Oevering, H.; Coussens, B.; Reek, J. N. H.; Kamer, P. C. J.;
van Leeuwen, P. W. N. M. Inorg. Chem. 2001, 40, 3363–3372.
(17) See ref 15a and: (a) Trost, B. M.; Toste, F. D. J. Am. Chem. Soc.
1999, 121, 4545–4554. (b) Trost, B. M.; Ariza, X. J. Am. Chem. Soc. 1999,
121, 10727–10737.
(18) For examples where this selectivity has been overcome with
biased substrates, see: (a) Trost, B. M.; Bunt, R. C.; Lemoine, R. C.;
Calkins, T. L. J. Am. Chem. Soc. 2000, 122, 5968–5976. (b) Trost, B. M.;
Osipov, M.; Dong, G. J. Am. Chem. Soc. 2010, 132, 15800–15807.
15905
dx.doi.org/10.1021/ja206960k |J. Am. Chem. Soc. 2011, 133, 15902–15905