Evidence for in situ catalyst modification during the Pd-catalyzed conversion of aryl triflates to aryl fluorides

Article abstract of DOI:10.1021/ja208461k

A mechanistic investigation of the Pd-catalyzed conversion of aryl triflates to fluorides is presented. Studies reveal that C-F reductive elimination from a LPd^II(aryl)F complex (L = t-BuBrettPhos or RockPhos) does not occur when the aryl group is electron rich. Evidence is presented that a modified phosphine, generated in situ, serves as the actual supporting ligand during catalysis with such substrates. A preliminary study of the reactivity of a LPd^II(aryl)F complex based on this modified ligand is reported.

Full text of DOI:10.1021/ja208461k

COMMUNICATION
pubs.acs.org/JACS
Evidence for in Situ Catalyst Modification during the Pd-Catalyzed
Conversion of Aryl Triflates to Aryl Fluorides
Thomas J. Maimone, Phillip J. Milner, Tom Kinzel, Yong Zhang, Michael K. Takase, and
Stephen L. Buchwald*
Department of Chemistry, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States
S Supporting Information
b
ABSTRACT: A mechanistic investigation of the Pd-cata-
lyzed conversion of aryl triflates to fluorides is presented.
Studies reveal that CÀF reductive elimination from a LPdII-
(aryl)F complex (L = t-BuBrettPhos or RockPhos) does not
occur when the aryl group is electron rich. Evidence is
presented that a modified phosphine, generated in situ, serves
as the actual supporting ligand during catalysis with such
substrates. A preliminary study of the reactivity of a LPdII-
(aryl)F complex based on this modified ligand is reported.
Figure 1. Presumed mechanism of Pd-catalyzed nucleophilic fluorina-
tion and ligands used in this study.
wing to their desirable metabolic properties and unique
O
electronic characteristics, aryl fluoride-containing compounds
are highly valued in a number of fields,1 but the construction of
aryl CÀF bonds remains quite challenging.2 Metal-catalyzed
coupling of heavier aryl halides and pseudohalides (X = Cl, Br, I,
OTf) with a simple metal fluoride salt would be an ideal route
to generate aryl fluorides with respect to waste, simplicity, and
generality (Figure 1).
Figure 2. Previously reported CÀF reductive elimination from a Pd(II)
fluoride.
Elegant studies by Grushin3 and Yandulov4 have shed consider-
able light on the challenges associated with developing a Pd-catalyzed
nucleophilic fluorination. A difficult CÀF reductive elimination step,
the formation of stable Pd(II) fluoride-bridged dimers, and myriad
fluoride-induced ligand decomposition pathways have cast doubt as
to whether accessing the catalytic cycle shown in Figure 1 is possible.
To avoid these problems, processes in which CÀF reductive
elimination occurs from a higher oxidation state metal fluoride,
notably Pd(IV), have attracted significant interest.5 We recently
described the catalytic conversion of aryl triflates to aryl fluorides,
which we believe operates by a Pd(0)/Pd(II) cycle;6 key to its
success was the use of bulky, monodentate biaryl phosphines—
catalysts based on these ligands appear to circumvent many of the
aforementioned problems. While CÀF reductive elimination was
arene became more electron rich (Figure 3).6 Because the product
ratios obtained differ from those reported for a process involving
external fluoride attack on a benzyne-type intermediate,3g we felt
it was possible that a discrete LPdII(Ar)F complex is involved in
the fluorination of electron-rich aryl triflates. To address this
question, a better understanding of the CÀF reductive eli-
mination process from electron-rich LPdII(Ar)F complexes sup-
ported by ligands relevant to the catalytic reaction was required.
Herein we report results that cast doubt as to whether 2 (or 3)
serves as the actual supporting ligand in many of these reactions
but suggest that reductive elimination from LPd(Ar)F complexes
(with electron-rich aryl groups) is possible and likely occurs as
part of the catalytic cycle.
We began by preparing the 3 Pd(Ar)F complex 5, presumed
demonstrated from BrettPhos-ligated 1 Pd(Ar)F complexes,6 it
3
3
to be an intermediate in the catalytic fluorination of 4-nBuPhOTf
(Scheme 1). In general, Pd complexes derived from 3 have
proven superior to those originating from 2 in terms of isolation,
characterization, and synthetic manipulation. Simply by stirring
(COD)Pd(CH2TMS)2, 3, and 4-nBuPhBr in a minimum quantity
of cyclohexane, the desired oxidative addition complex 4 pre-
cipitated as a bright yellow solid in good yield (76%). X-ray
crystallographic analysis showed that 4 adopts a C-bound con-
formation in the solid state with a PdÀBr bond length of
was only observed when the aryl group was electron deficient and
possessed an ortho alkyl group, two structural features known to
favor reductive elimination (Figure 2).7 In addition, Pd catalysts
based on BrettPhos (1) are not active in most fluorination
reactions; only catalysts derived from the di-tert-butyl-based
ligand t-BuBrettPhos (2) were able to transform a wide range
of aryl triflates to their corresponding fluorides. We have now
discovered that catalysts based on the structurally similar Rock-
Phos (3) also perform well in these fluorination reactions with
product yields similar to those obtained with 2 (vide infra).8
In our previous report, we described the formation of regio-
isomeric aryl fluoride products whose quantities increased as the
Received: September 13, 2011
Published: October 14, 2011
r
2011 American Chemical Society
18106
dx.doi.org/10.1021/ja208461k J. Am. Chem. Soc. 2011, 133, 18106–18109
|

Journal of the American Chemical Society
COMMUNICATION
Figure 3. Regioisomer formation.
Scheme 1. Preparation of 3 Pd(Ar)F Complex 5a,b
3
Figure 4. Synthesis and isomerization of t-BuBrettPhos oxidative addi-
tion complex 6.
a Isolated yields are given. b Thermal ellipsoid plot at 50% probability;
hydrogen atoms are omitted.
Table 1. Thermolysis of 3 Pd(Ar)F Complex 5
3
Figure 5. X-ray structure of 7 and relevant bond lengths (thermal
ellipsoid plot at 50% probability; hydrogen atoms are omitted).
of 2.511 and 2.298 Å, respectively (Figure 5).10 In 7, the Pd atom
is σ-bound to C2 and has an additional interaction with C3,
analogous to the one-carbon PdÀarene interaction seen in 4 and
previously observed.11 Despite its twisted, dearomatized struc-
ture, 7 is air-stable and thermally robust. Dissolving pure crystal-
line 7 in CD2Cl2 re-established a nearly 6:1 mixture of 7:6;
remarkably, these compounds appear to be in equilibrium.
Although the origin of the difference in reactivity between 4
and 6 is not yet known,12 we have observed that complexes based
on electron-deficient aryl groups show far less propensity to
rearrange than those bearing electron-rich arenes. For example,
2.4663(3) Å. It is worth noting that, unlike with 1, oxidative
addition complexes derived from 3 show no signs of an O-bound
conformation in solution.9 Halide exchange of 4 with AgF
afforded the desired LPd(Ar)F complex 5.
When 5 was heated in toluene, no product of CÀF reductive
elimination was detected, with or without a variety of additives,
despite 5 being fully consumed (Table 1). 19F and 31P NMR
analysis of the crude reaction mixtures indicated the ab-
sence of aryl fluorides of any kind and no evidence for PÀF
bond formation.3 The major fluorine-containing species detected
by 19F NMR (δ À136 and À148 ppm) vanished upon addition
of Et3N, leading us to speculate that a formal loss of HF may have
occurred (vide infra).
While preparing the analogous t-BuBrettPhos-ligated LPd-
(Ar)Br complex 6, we observed that the initially formed bright
yellow complex (31P NMR: δ 69 ppm) that had precipitated
from cyclohexane began to rapidly convert to a new dark red
compound (31P NMR: δ 83 ppm) when dissolved in CD2Cl2,
eventually reaching a ∼6:1 mixture as determined by 1H NMR
(Figure 4). From this mixture, the major component could be
crystallized, and X-ray analysis identified it as dearomatized
Pd(II) bromide complex 7 with PdÀBr and PdÀP bond lengths
after 60 h in CD2Cl2, 3 Pd(Ar)Br (Ar = 4-cyanophenyl) shows
3
no detectable rearrangement, and the corresponding complex
with 2 shows only ∼10% isomerization.
Treating 7 with DBU (1.2 equiv) and 4-nBuPhBr (3 equiv) in
THF led to the clean formation of a new bright yellow compound
(31P NMR: δ 71 ppm) that was identified as the oxidative addi-
tion complex 8 by X-ray diffraction analysis (Figure 6). Impor-
tantly, this demonstrates that if complexes similar to 7 are formed
in cross-coupling reactions using 2, there exists a pathway by
which they can return to a LPd(0) state if base is present. In
contrast to 6, 8 could be heated to 100 °C without any detectable
decomposition or rearrangement as judged by 31P NMR.
With the reactivity of complexes of 2 and 3 brought to light, we
began to probe their relevance to the catalytic fluorination reac-
18107
dx.doi.org/10.1021/ja208461k |J. Am. Chem. Soc. 2011, 133, 18106–18109

Journal of the American Chemical Society
COMMUNICATION
Table 2. Synthesis and Reactivity of LPd(Ar)F Complex 11a-d
Figure 6. Rearomatization of 7 with concurrent oxidative addition
(thermal ellipsoid plot at 50% probability; hydrogen atoms are omitted).
a Isolated yields are given. b Yields based on 11 anddeterminedby19FNMR.
Figure 7. Arylated RockPhos ligand 9 isolated from the catalytic
fluorination reaction.
c Yield in cyclohexane = 20%. d Yield in cyclohexane = 40%, selectivity = 2:1.
Figure 9. Catalytic competence of 11.
Figure 8. Fate of the ligand during experiments that failed to give
products of CÀF reductive elimination.
initial mass loss of aryl triflate is less, and the zeroth-order decay
begins rapidly compared to when 3 is employed (see SI).
Upon re-examination of the studies on stoichiometric reductive
elimination of CÀF bonds (Table 1), we observed that substantial
quantities of ligand 9 were present in the crude reaction mixtures
(Figure 8). For example, at the end of the reaction shown in entry 1,
there was a ∼1:1 mixture of 3 and 9 (31P NMR). When excess aryl
bromide was included in the reaction mixture (entry 3), 3 had been
mostly consumed, and two new, virtually identical ligands were
detected by 31P NMR—9 and, presumably, 10.17 Although we were
unable to directly observe a rearranged Pd(II) fluoride analogous to 7,
it is now clear that formal net loss of HF had taken place (vide supra).
The arylated LPd(Ar)F complex 11 could be readily prepared
from 8 and its reactivity compared to 5 (Table 2). Upon heating
11 in toluene or cyclohexane, 4-nBuPhF was formed in yields of
15 and 20%, respectively. Heating 11 in the presence of excess
PhBr produced primarily PhF (40%, entry 2). Similarly, in the
presence of 1-naphthyltriflate, 1-fluoronaphthalene was formed
in 75% yield (entry 3). The precise mechanism leading to formal
aryl exchange is currently under investigation. Perhaps most rele-
vant to the catalytic reaction was the fate of 11 when heated in the
presence 4-nBuPhOTf (entry 4). A 1.6:1 mixture of 4-nBuPhF
and 3-nBuPhF was produced in 52% combined yield—nearly the
tion. Performing a CÀF bond-forming reaction (utilizing 3 as
ligand and 4-nBuPhOTf as the substrate) and re-isolating the
phosphine present at the end of the reaction gave arylated
phosphine 9, whose structure and connectivity were confirmed
by X-ray analysis (Figure 7).13,14 The very similar chemical shifts
(31P NMR) of 9 relative to 3 had allowed it to previously evade
our detection.15 We then compared the performance of 9 relative
to 3 as supporting ligand. Using isolated ligand 9 in place of 3 led
to an improvement in yield of 13% in the fluorination of
4-nBuPhOTf (73% vs 60%). No further arylation of 9 was
detected as judged by 31P NMR of the crude reaction mixture.16
Thus the yield increase observed is likely due to factors beyond
just the inability of 9 to undergo arylation.
To date, the kinetic profile of every fluorination reaction we
have studied shows a substantial initial mass loss of ArOTf with-
out the formation of an equivalent amount of ArF. During this
stage, ArCl (chloride from the [(cinnamyl)PdCl]2 precursor)
and ArOAr are the only other products formed detectable by
GC-MS analysis. After this induction period, zeroth-order overall
decay of the remaining ArOTf is observed. Using ligand 9, the
18108
dx.doi.org/10.1021/ja208461k |J. Am. Chem. Soc. 2011, 133, 18106–18109

Journal of the American Chemical Society
COMMUNICATION
same ratio of products seen in catalytic reactions employing 2. This
demonstrates that the formation of regioisomeric aryl fluorides in
the catalytic reaction does not require the presence of highly basic
CsF nor any other additional fluoride source.18 Notably, when this
stoichiometric reaction was conducted in cyclohexane, a slightly
improved ratio (2:1) of 4-nBuPhF to 3-nBuPhF was observed. This
is similar to improvements in selectivity in the catalytic reaction we
have seen when utilizing cyclohexane in lieu of toluene.6 If
4-nBuPhBr was used as the additive during thermolysis (entry 5),
very little regioisomer was formed. Finally, inclusion of 4-MeO-
PhOTf led to a 1.7:1 mixture of 4-nBuPhF and 3-nBuPhF along
with small amounts of 3-MeOPhF (7%); interestingly, no
4-MeOPhF could be detected (entry 6). 11 was found to be
catalytically competent in the fluorination of 4-n-BuPhOTfwiththe
highest yield we have seen to date for this substrate (Figure 9).
It has not escaped our attention that the ability of ligands 2 and
3 to undergo arylation may play a role in their success (or failure)
in other previously reported transformations.8,19 The findings
reported herein also imply that there is a slightly different catalyst
for each substrate and that processes using 3 and especially 2 may
be more complicated than previously assumed.
2008, 37, 320. Ametamey, S. M.; Honer, M.; Schubiger, P. A. Chem. Rev.
2008, 108, 1501. Kirk, K. L. Org. Process Res. Dev. 2008, 12, 305.
(2) Recent review: Furuya, T.; Kamlet, A. S.; Ritter, T. Nature 2011,
473, 470.
(3) (a) Fraser, S. L.; Antipin, M. Y.; Khroustalyov, V. N.; Grushin, V. V.
J. Am. Chem. Soc. 1997, 119, 4769. (b) Grushin, V. V. Organometallics 2000,
19, 1888. (c) Grushin, V. V. Chem.—Eur. J. 2002, 8, 1006. (d) Grushin,
V. V.; Marshall, W. J. Organometallics 2007, 26, 4997. (e) Grushin, V. V. U.
S. Patent 7,202,388, 2007. (f) Grushin, V. V. Acc. Chem. Res. 2010, 43, 160.
(g) Grushin, V. V.; Marshall, W. J. Organometallics 2008, 27, 4825.
(4) Yandulov, D. V.; Tran, N. T. J. Am. Chem. Soc. 2007, 129, 1342.
(5) Synthesis and study of high-valent Pd(IV) fluorides relevant to
ArÀF bond formation: Furuya, T.; Ritter, T. J. Am. Chem. Soc. 2008,
130, 10060. Ball, N. D.; Sanford, M. S. J. Am. Chem. Soc. 2009, 131, 3796.
Furuya, T.; Benitez, D.; Tkatchouk, E.; Strom, A. E.; Tang, P.; Goddard,
W. A.; Ritter, T. J. Am. Chem. Soc. 2010, 132, 3793. Additional examples of
ArÀF bond formation involving electrophilic fluorinating reagents: Hull,
K. L.;Anani, W. Q.;Sanford, M. S.J. Am. Chem. Soc. 2006, 128, 7134. Furuya,
T.; Kaiser, H. M.; Ritter, T Angew. Chem., Int. Ed. 2008, 47, 5993. Wang, X.;
Mei, T.-S.; Yu, J.-Q. J. Am. Chem. Soc. 2009, 131, 7520. Chan, K. S. L.; Wasa,
M.; Wang, X.; Yu, J.-Q. Angew. Chem., Int. Ed. 2011, 50, 9081.
(6) Watson, D. A.; Su, M.; Teverovskiy, G.; Zhang, Y.; García-
Fortanet, J.; Kinzel, T.; Buchwald, S. L. Science 2009, 325, 1661. No€el, T.;
Maimone, T. J.; Buchwald, S. L. Angew. Chem., Int. Ed. 2011, 50, 8900.
(7) Roy, A. H.; Hartwig, J. F. J. Am. Chem. Soc. 2001, 123, 1232.
(8) Wu, X.; Fors, B. P.; Buchwald, S. L. Angew. Chem., Int. Ed. 2011,
50, 9943.
(9) “O-bound” refers to binding of the Pd(II) center to the oxygen
of the OMe group in the 3 position as opposed to the lower arene
(i.e. “C-bound”): Fors, B. P.; Watson, D. A.; Biscoe, M. R.; Buchwald,
S. L. J. Am. Chem. Soc. 2008, 130, 13552.
(10) Doyle recently observed a seemingly similar dearomatization of
1 involving Ni(0) and an epoxide substrate. The connectivity is slightly
different with respect to the point of lower ring dearomatization:
Nielsen, D. K.; Doyle, A. G. Angew. Chem., Int. Ed. 2011, 50, 6056.
(11) Barder, T. E.; Biscoe, M. R.; Buchwald, S. L. Organometallics
2007, 26, 2183.
In conclusion, we have shown that a 3 Pd(Ar)F complex does
3
notundergoCÀF reductive elimination when the arene is electron-
rich. The observed facile and reversible rearrangement of oxidative
addition complex 6 led us to discover the in situ formation of
terarylphosphine ligands in the fluorination reaction when starting
with 2 or 3.20 Addition of a third aryl ring to the phosphine ligand
confers marked stability to the Pd complexes subsequently formed
—this may be required for CÀF bond formation to occur.
Although we believe these results are interesting and informative,
they do not directly relate to the mechanism of formation of
different regioisomers in the catalytic CÀF bond-forming process.
This is a topic of ongoing investigations in our laboratory.
(12) After 2 h at room temperature, a CD2Cl2 solution of RockPhos
complex 4 showed only ∼5% of a rearranged compound.
’ ASSOCIATED CONTENT
(13) t-BuBrettPhos (2) is quantitatively arylated in the catalytic
fluorination reaction; however, we have been unable to isolate arylated 2
in pure form from the crude mixture to resubject it to a catalytic reaction.
(14) Since chloride is also present in the catalytic reaction, we cannot
rule out that the arylated ligand is formed via rearrangement of an
initially formed LPd(Ar)Cl complex as opposed to a LPd(Ar)F complex.
(15) 31P NMR shifts of arylated ligands based on 2 or 3 have
characteristic downfield shifts of ∼1.2 ppm relative to the parent ligand.
(16) Spiking the crude reaction mixture with an authentic sample of
9 showed the presence of only 9 by 31P NMR. We assume a doubly
arylated ligand would possess a slightly different 31P chemical shift.
(17) The 31P NMR shifts of 9 and what we presume to be 10 differ
by only 0.04 ppm. LC-MS analysis of the crude mixture indicates the
presence of a compound with the mass of 10 (MW = 545).
(18) Fluoride retains considerable basicity even when bound to Pd:
Grushin, V. V.; Marshall, W. J. Angew. Chem., Int. Ed. 2002, 41, 4476.
(19) Fors, B. P.; Dooleweerdt, K.; Zeng, Q.; Buchwald, S. L. Tetra-
hedron 2009, 65, 6576. Fors, B. P.; Buchwald, S. L. J. Am. Chem. Soc.
2010, 131, 12898. Maimone, T. J.; Buchwald, S. L. J. Am. Chem. Soc.
2010, 132, 9990. Shen, X.; Hyde, A. M.; Buchwald, S. L. J. Am. Chem. Soc.
2010, 132, 14076. Dooleweerdt, K.; Fors, B. P.; Buchwald, S. L. Org. Lett.
2010, 12, 2350. Breitler, S.; Oldenhuis, N. J.; Fors, B. P.; Buchwald, S. L.
Org. Lett. 2011, 13, 3263.
S
Supporting Information. Procedural, spectral, and X-ray
b
crystallographic (CIF) data. This material is available free of
charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
Corresponding Author
sbuchwal@mit.edu
’ ACKNOWLEDGMENT
We thank the National Institutes of Health for financial
support of this project (GM46059) and a postdoctoral fellowship
to T.J.M. (1F32GM088931). T.K. thanks the Alexander von
Humboldt Foundation for a Fedor Lynen postdoctoral fellow-
ship. Mingjuan Su is acknowledged for helpful discussions as well
as assistance with LC-MS measurements. The Varian 300 MHz
NMR spectrometer used for portions of this work was purchased
with funds from the National Science Foundation (Grants CHE
9808061 and DBI 9729592). The departmental X-ray diffraction
instrumentation was purchased with the help of funding from the
National Science Foundation (CHE-0946721).
(20) For a notable example of ligand modification during Pd-
catalyzed cross-coupling, see: Shelby, Q.; Kataoka, N.; Mann, G.;
Hartwig, J. J. Am. Chem. Soc. 2000, 122, 10718.
’ REFERENCES
(1) M€uller, K.; Faeh, C.; Diederich, F. Science 2007, 317, 1881.
Purser, S.; Moore, P. R.; Swallow, S.; Gouverneur, V. Chem. Soc. Rev.
18109
dx.doi.org/10.1021/ja208461k |J. Am. Chem. Soc. 2011, 133, 18106–18109