Flow heck reactions using extremely low loadings of phosphine-free palladium acetate

Article abstract of DOI:10.1021/ol4018134

High-yielding Heck reactions are demonstrated using 0.05 mol % Pd(OAc) ₂ without phosphine ligands. These reactions are run in a mesoscale flow reactor which allows precise control of reaction times and temperatures. Profiling yield and selecti

Full text of DOI:10.1021/ol4018134

ORGANIC
LETTERS
XXXX
Vol. XX, No. XX
Flow Heck Reactions Using Extremely
Low Loadings of Phosphine-Free
Palladium Acetate
000–000
Patrick Cyr, Stella T. Deng, Joel M. Hawkins, and Kristin E. Price*
Worldwide Research and Development, Pfizer Inc., Eastern Point Road,
Groton, Connecticut 06340, United States
Kristin.E.Price@pfizer.com
Received June 27, 2013
ABSTRACT
2
High-yielding Heck reactions are demonstrated using 0.05 mol % Pd(OAc) without phosphine ligands. These reactions are run in a mesoscale
flow reactor which allows precise control of reaction times and temperatures. Profiling yield and selectivity versus Pd loading shows 500 ppm to
be optimal for aryl iodides; higher loadings favor side reactions caused by Pd(II) species. Aryl halides are examined via concise Design of
Experiment to expand the scope and optimize conditions.
The Heck reaction is a versatile method for the forma-
tion of CÀC bonds between alkenes and aryl or alkyl
poisoning downstream chemistry or contamination of ac-
3
,4
tive pharmaceutical ingredients. Consequently, there re-
mains a significant need to develop new techniques for
Heck reactions with reduced palladium loading and little
or no phosphine ligand.
1
halides. Typically catalyzed by palladiumÀphosphine
complexes, this reaction has seen use across a wide variety
of substrates and has been employed on large scale for
2,3
the synthesis of pharmaceutical intermediates. More
widespread adoption of Heck chemistry on industrial
scale is challenged by the need for high catalyst loadings
The ligand-free Heck reaction is not new; Heck and
Mizoroki both introduced the reaction between aryl io-
dides and activated alkenes in the early 1970s with ligand-
5
(
1À5 mol %). This adds cost, especially as the cost of
free PdCl or Pd(OAc) . The addition of tetraalkylam-
2
2
palladium increases. High catalyst loadings also increase
the difficulty of purging palladium and ligands to prevent
monium salts improved the ligand-free Heck reaction, and
subsequent work has broadened the scope further to
6
,7
include aryl bromides and chlorides. This reaction re-
mains limited by the requirement for high temperatures
(
1) (a) Beletskaya, I. P.; Cheprakov, A. V. Chem. Rev. 2000, 100,
009–3066. (b) Heck, R. F. Acc. Chem. Res. 1979, 12, 146–151. (c)
Schmidt, A. F.; Al Halaiqa, A.; Smirnov, V. V. Synlett 2006, 18, 2861–
3
8
and, in some cases, extended reaction times.
Flow reactors directly address these challenges by pro-
viding straightforward access to the high temperatures
2878.
(2) (a) Diek, H. A.; Heck, R. F. J. Am. Chem. Soc. 1974, 96, 1133–
136. (b) Herrmann, W. A.; Brossmer, C.; Ofele, K.; Reisinger, C. P.;
1
Priermeier, T.; Beller, M.; Fischer, H. Angew. Chem., Int. Ed. Engl. 1995,
3
4, 1844–1848. (c) Littke, A. F.; Fu, G. C. J. Am. Chem. Soc. 2001, 123,
(
4) Konigsberger, K.; Chen, G. P.; Wu, R. R.; Girgis, M. J.; Prasad,
K.; Repic, O.; Blacklock, T. J. Org. Process Res. Dev. 2003, 7, 733–742.
5) (a) Heck, R. F.; Nolley, J. P., Jr. J. Org. Chem. 1972, 37, 2320–
2
4
6989–7000.
(
3) (a) Ripin, D. H. B.; Bourassa, D. E.; Brandt, T.; Castaldi, M. J.;
(
Frost, H. N.; Hawkins, J.; Johnson, P. H.; Massett, S. S.; Neumann, K.;
Phillips, J.; Raggon, J. W.; Rose, P. R.; Rutherford, J. L.; Sitter, B.;
Stewart, A. M., III; Vetelino, M. G.; Wei, L. Org. Process Res. Dev.
322. (b) Mizoroki, T.; Mori, K.; Ozaki, A. Bull. Chem. Soc. Jpn. 1971,
4, 581.
(
6) Jeffery, T. Adv. Met.-Org. Chem. 1996, 5, 153–260.
2005, 9, 440–450. (b) Jiang, X.; Lee, G. T.; Prasad, K.; Repic, O. Org.
Process Res. Dev. 2008, 12, 1137–1141. (c) Caron, S.; Vazquez, E.;
Stevens, R. W.; Nakao, K.; Koike, H.; Murata, Y. J. Org. Chem. 2003,
(7) (a) Reetz, M. T.; de Vries, J. G. Chem. Commun. 2004, 1559–1563.
(b) Gurtler, C.; Buchwald, S. L. Chem.;Eur. J. 1999, 5, 3107–3112.
(8) To reach the high temperatures, high boiling point solvents are
needed and some, including DMAc and DMF, are strongly regulated in
Europe due to their toxicity, limiting their use on scale.
(9) Mason, B. P.; Price, K. E.; Steinbacher, J. L.; Bogdan, A. R.;
McQuade, D. T. Chem. Rev. 2007, 107, 2300–2318.
68, 4104–4107. (d) Chekal, B. P.; Guinness, S. M.; Lillie, B. M.;
McLaughlin, R. W.; Palmer, C. W.; Post, R. J.; Sieser, J. E.; Singer,
R. A.; Sluggett, G. W.; Vaidyanathan, R.; Withbroe, G. J. Org. Process
Res. Dev. 2013, ASAP; DOI: 10.1021/op400088k. (e) Magano, J.;
Dunetz, J. R. Chem. Rev. 2011, 111, 2177–2250.
1
0.1021/ol4018134 r XXXX American Chemical Society

needed for the ligand-free Heck reaction without the
extended heating and cooling ramps required in batch
9
reactors. Because of their relatively low volumes, flow
reactors can be run at temperatures and pressures far
beyond those obtainable using standard glass equipment,
obviating the need for high boiling solvents which are
difficult to remove during workup. High temperatures
accelerate the reaction; thus the catalyst only needs to be
stable for 5À20 min, allowing elimination of ligands and
potentially reduced loadings. Flow Heck reactions have
been demonstrated with both hetero- and homogeneous
1
0
catalysts, and Kappe recently reported an example with
1
1
Pd(OAc)2. Here, we describe the optimization of fast
ligand-free flow Heck reactions with Pd(OAc) and show
2
that very low catalyst loadings are essential to minimize
byproducts.
Scheme 1. Heck Catalyst Cycle and Palladium Aggregation
Pathway
The primary challenge in running Heck chemistry is to
facilitate the desired coupling reaction while minimizing
byproduct formation and catalyst deactivation. The main
catalytic cycle for the Heck reaction is unimolecular in
Pd, and its rate increases with increased Pd concentra-
tion (Scheme 1, steps aÀd). The undesired reduction
product (ArÀH) and biaryl product (ArÀAr) are formed
via bimolecular pathways through reactions between
Figure 1. Yield of the Heck product, the reduction product
(ArÀH), and the biaryl product (ArÀAr) calibrated against an
1
2,13
internal standard as a function of Pd(OAc)
loading.
(A)
2
Reactions of p-iodobenzonitrile with n-butylacrylate (1.5 equiv
of olefin, 1.5 equiv of iPr
NEt, acetonitrile, 200 °C, 5 min). (B)
Reactions of p-bromobenzonitrile with n-butylacrylate
1.5 equiv of olefin, 1.5 equiv of iPr NEt, 0.1 equiv of Bu NBr,
1
c
2
(
2
4
1
4
acetonitrile, 200 °C, 5 min). The very narrow error bars
illustrate the reproducibility between runs.
[PdÀAr] species and [PdÀH] species (step e) or [PdÀAr]
species (step f). Catalytically active Pd(0) species also
convert to nonactive palladium black via agglomeration
7
in a pathway which is also higher-order in Pd (step g).
We initially used a fixed residence time (5 min) and
temperature (200 °C) to profile the reaction between
p-iodobenzonitrile and n-butyl acrylate (Figure 1A). In
this study, the palladium acetate loading was varied from
a
Since the Pd loading will affect the rate of the main
catalytic cycle, which is unimolecular in Pd, and the rate
of undesired processes, which are higher order in Pd,
differently, we focused first on optimizing the catalyst
loading.
0.002 to 2 mol %. The data are presented as the in situ
yields of the Heck product, reduction product (ArÀH),
15
and biaryl product (ArÀAr) as a function of Pd loading.
This differs from the more traditional graph of yield as a
6
1
function of time; here the reaction time is fixed at 5 min.
The use of an automated flow reactor simplified the collec-
(
(
10) Noel, T.; Buchwald, S. L. Chem. Soc. Rev. 2011, 40, 5010–5029.
11) Glasnov, T. N.; Findenig, S.; Kappe, C. O. Chem.;Eur. J. 2009,
1
5, 1001–1010. In this paper, all optimization is carried out in micro-
wave batch and then executed in flow.
12) Expanded views of Figure 1 and data are provided in the
Supporting Information (SI).
13) Points are connected by lines (for visualization) not fitted to
curves.
tion of these data, and replicates at each Pd concentration
12,17
showed excellent reproducibility.
Notably, product
(
(
(15) All in situ yields determined against an internal standard,
biphenyl, included in the reaction.
(
14) Multiple cleaning segments were executed between each reaction
segment to prevent Pd buildup in the reactor and to ensure that any
catalysis occurs with the indicated loading. See SI for more details.
(16) de Vries uses this type of graph to illustrate the impact of Pd
loading on batch Heck reactions; see refs 7 and 17a.
B
Org. Lett., Vol. XX, No. XX, XXXX

yield dramatically peaks at the low loading of 0.05 mol %
catalyst concentrations. Also, because the Pd is largely in
the Pd(0) oxidation state, stabilizing salts such as Bu NBr
12
Pd(OAc)2. At lower Pd loadings (20À500 ppm), the
product yield increases with Pd concentration. At 500 ppm,
the aryl iodide is virtually consumed (product distribution:
4
have a beneficial effect on the yield, as has been observed
1
previously.
8
8
9
5% product, 4% ArÀH, and 3% ArÀAr, accounting for
5% of the reaction mass). As the loading is further
increased from 500 to 2000 ppm Pd, the amount of
reduction product, ArÀH, and biaryl product, ArÀAr,
increase, attenuating the yield of the desired product.
Palladium loading has a different impact on the reaction
of p-bromobenzonitrile with n-butylacrylate (Figure 1B).
Quaternary ammonium salts are typically included in
ligand-free reactions with aryl bromides, and here tetra-
Table 1. Isolated Yields from Aryl Iodide Reactions
1
1,18,19
a
butylammonium bromide was used.
The Pd loading
product
R
isolated yield
was varied from 0.001 mol % (10 ppm) to 2 mol % in
reactionsrun at200 °C for5 min. Atlow Pd loadings(0.001
to 0.05 mol %), the yield of product increases from a 10%
to 88% yield as the Pd loading increases. This part of the
graph is very similar to the aryl iodide case. Interestingly,
as the catalyst loading was further increased from 0.05 to
1
2
3
4
5
H
83%
81%
85%
CN
Cl
b
Me
OMe
81%
c
78%
a
b
Chromatographed yield, ∼1 g scale. 200 °C, 9 min residence time.
2
mol %, the yield increased slightly and then leveled off.
c
1
60 °C, 20 min residence time.
Also, the biaryl product (ArÀAr) and reduction product
(
ArÀH) are observed in very low amounts (<1%) in all
These conditions were then verified using larger scale
aryl bromide reactions, in contrast to the aryl iodide case.
The differing effect of catalyst loading on product
distribution in the p-iodobenzonitrile and p-bromobenzo-
nitrile reactions in Figure 1 is consistent with a change in
the catalyst resting state due to changes in the oxidizing
power of the aryl halide and thus relative rates in the
reactions. A variety of aryl iodide substrates provided 78%
À85% isolated yields after chromatography (Table 1).
Additional investigations focused on delineating the influ-
ence of residence time and temperature on the reaction of
each substrate. To optimize time and temperature simul-
taneously, a DoE (Design of Experiment) optimization
sequence was executed for each aryl iodide. In the auto-
2
0
catalytic cycle.
The increased formation of byproducts (ArÀAr and
ArÀH) observed in reactions of p-iodobenzonitrile imply
an increasing concentration of Pd(II) [PdÀAr] species and
thus a Pd(II) resting state. This is consistent with literature
findings that the oxidative insertion is facile with aryl
iodide substrates, and the palladium will be observed as
mated flow system, the DoE requires <400 mg of aryl
22,23
iodide and <1 h of experiment setup time.
Visualiza-
24
tion of a selection of these experiments illustrates the
impact of substrate electronics on the reaction with unique
response surfaces and optimal conditions for each aryl
iodide (Figure 2).
6
,21
Pd(II) rather than Pd(0).
In the aryl bromide case, the
Because of the substrate dependence observed with the
aryl iodides, a similar DoE was applied to the aryl bro-
25
mides (Figure 2, Table 2). These experiments served to
define the method scope rapidly and to illustrate the
extremely divergent impact of time and temperature on
each substrate’s reactivity.
insertion of Pd(0) into the ArÀBr bond is slow, resulting in
a Pd(0) resting state. The Pd(0) concentration will increase
as the loading of Pd(OAc) increases, and the concentra-
2
tion of Pd(II) species will remain small. Consequently,
ArÀAr and ArÀH formation will be slow even at higher
Thus Heck reactions with a range of aryl iodides
and electron-poor aryl bromides can be run with very
low loadings of a simple palladium catalyst, 0.05 mol %
(17) (a) Conjure and Propel flow systems were used for these experi-
ments (www.accendocorporation.com). (b) Bogdan, A. R.; Sach, N. W.
Adv. Synth. Catal. 2009, 351, 849–854.
Pd(OAc) without any phosphine ligands, by a 5À20 min
2
(
18) (a) (NMP/NaOAc) de Vries, A. H. M.; Mulders, J. M. C. A.;
Mommers, J. H. M.; Henderickx, H. J. W.; deVries, J. G. Org. Lett.
003, 5, 3285–3288. (b) (NMP/DMG/NaOAc) Reetz, M. T.; Wester-
mann, E.; Lohmer, R.; Lohmer, G. Tetrahedron Lett. 1998, 39, 8449–
452. (c) (Et NCl, Cy NMe) Gurtler, C.; Buchwald, S. L. Chem.;Eur.
J. 1999, 5, 3107–3112.
19) The reaction of 4-bromobenzonitrile and butyl acrylate with
NEt in acetonitrile (0.6 M) was compared in the presence and
absence of 0.1 equiv of Bu
NBr. There was a 10À20% increase in
conversion (relative to an internal standard) when the reaction was
run in the presence of Bu NBr. Addition of Bu NI provided comparable
results to Bu NBr. The addition of Bu NBr had no impact on the yield in
the reaction of p-iodobenzonitrile (see SI).
20) For more information on resting states: Littke, A. F.; Fu, G. C.
J. Am. Chem. Soc. 2001, 123, 6989–7000.
21) Evans, J.; O’Neill, L.; Kambhampati, V. L.; Rayner, G.; Turin,
S.; Genge, A.; Dent, A. J.; Neisius, T. J. Chem. Soc., Dalton Trans. 2002,
207–2212.
exposure to 160À200 °C. Stabilizing ligands are not re-
quired because the catalyst only needs to be active for these
short residence times. Flow reactors provide easy access to
2
8
4
2
(
iPr
2
(22) Optimization experiments on the Conjure varied the tempera-
ture between 160 and 200 °C and the time between 5 and 20 min (see SI).
Reactions were 0.6 M in aryl halide.
(23) Design Expert Version 7.1.5 (Stat-Ease Inc.) was used to gen-
erate the contour plots captured in Figure 2. The data can be found in the
SI.
(24) Figure 2 is a small multiple graph. For more information on
small multiples, see: Tufte, E. R. The Visual Display of Quantitative
Information, 2nd ed.; Graphics Press LLC: Cheshire, CT, 2001; p 42.
(25) The catalyst loading has not yet been further evaluated for the
electron-rich aryl bromide substrates.
4
4
4
4
4
(
(
2
Org. Lett., Vol. XX, No. XX, XXXX
C

Figure 2. Effect of residence time and temperature on the yield of reactions between aryl iodides and aryl bromides with n-butyl acrylate.
All reactions run with 1.5 equiv of iPr NEt and 0.05 mol % Pd(OAc) in acetonitrile. The aryl bromide reactions included 0.1 equiv of
Bu NBr. Contour lines represent reaction predicted yields. Red circles represent data points.
2
2
4
26
minimized byproducts seen at higher loadings. Profiling
reactions vs time and temperature provided a unique
response surface for each substrate, allowing high yielding
conditions to be identified withrapid experiment setup and
Table 2. In situ Yield from Optimized Aryl Bromide Reactions
<400 mg of substrate. The ability to study continuous
variables easily in an automated fashion complements
traditional batch reaction catalyst screening that focuses
on categorical variables and often uses higher catalyst
loadings.
R
in situ yield
temp
time
H
8%
200 °C
180 °C
160 °C
200 °C
200 °C
160 °C
5 min
5 min
5 min
5 min
5 min
5 min
a
CN
Cl
>99%
Acknowledgment. The authors thank the following
Pfizer colleagues: John Lucas for assistance with the
Accendo Propel flow system, David Pattavina for assistance
with reaction analysis, and Neal Sach and Marty Berliner for
helpful discussions. The authors also thank Terry Long and
Jan Hughes from Accendo Inc. for assistance with the
Conjure and Propel systems. The authors thank Prof. David
Collum, CornellUniversity,andProf.StevenLey,University
of Cambridge, for helpful discussions.
52%
12%
12%
Me
OMe
b
CF
3
90%
a
b
Isolated yield is 71% at 200 °C, 5 min. Conversion.
high reaction temperatures even with low boiling solvents
such as acetonitrile. Automated flow reactors offer the
further advantage of easily profiling continuous variables
such as the catalyst loading, reaction temperature, and
residence time. Profiling the Heck reactions against the
catalyst loading showed a distinct peak in yield and
Supporting Information Available. Experimental pro-
cedures, expanded figures, and characterization data.
This material is available free of charge via the Internet
at http://pubs.acs.org.
selectivity at 0.05 mol % Pd(OAc) for aryl iodides which
2
(26) This is consistent with earlier findings in ligand-free Heck
reactions; see refs 7a and 11.
The authors declare no competing financial interest.
D
Org. Lett., Vol. XX, No. XX, XXXX