Organic Process Research & Development
. EXPERIMENTAL SECTION
.1. Materials. Phenylboronic acid and all aryl halides were
Communication
2
Scheme 2. Heterogeneous Suzuki−Miyaura coupling of 2-
chlorobenzonitrile over SiliaCat DPP-Pd
2
purchased from Aldrich. Ultrapure water was used throughout
the experiments. All chemicals used were of analytical grade and
were used as received without any further purification.
Characteristics of the SiliaCat DPP-Pd heterogeneous
catalyst: particle size: d50 = 224, d90/d10 = 5.9, analyzed by
2
a Malvern Mastersizer 2000; 573 m /g BET specific surface
area; 1.13 cm /g total pore volume; 7.5 nm mesopore
diameters; 0.354 g/mL density; 0.25 mmol/g palladium
loading.
3
conditions (1 mol % Pd, 1 equiv 2-chlorobenzonitrile, 1.1
equiv 4-methyl phenylboronic acid, 1.1 equiv K CO in EtOH
2
3
1
M (molar concentration with respect to the reagents)) were
tested in flow, the inorganic base was not completely soluble in
ethanol employed as solvent. Hence, to fully dissolve K CO ,
2.2. Typical Catalytic Procedure. Reaction was per-
formed in a modular flow microreactor (Syrris Asia 220, a fully
automated device combining smooth flow and a wide range of
2
3
water was added resulting in a 70% (v/v) aqueous ethanol
solution. Under these new conditions, complete conversion in
batch was now obtained in 60 min (Entry 2 in Table 1).
8
temperatures and pressures and flow rates) that under the
laminar flow typical of microreactors allows a fast and
reproducible mixing that occurs only via diffusion phenomena
3
.2. Process Intensification. Hence, we tested similar
9
conditions in flow starting from a single 144 mL solution
containing 72 mmol substrate. The column reactor (0.785 cm
ID × 1.4 cm) was charged with 0.345 g SiliaCat DPP-Pd which
leads to a 0.06 mol % catalytic amount. The homogeneous
solution was processed directly through the solid phase reactor
heated at 77 °C using a single pump channel. At flow rate = 1
mL/min the conversion after 2.5 h was 79% (Entry 3).
establishing a uniform reaction environment. In the following,
“residence time” denotes the time during which the catalyst is
in contact with the reaction mixture, and “reaction time”
denotes the time of operation (namely the time period over
which reaction mixture is pumped through the microreactor).
Two solutions were prepared. The solution 1 of the aryl
halide (55 mmol, 1 equiv) in 70 mL tetrahydrofuran (THF)
was combined with the solution 2 of phenylboronic acid (68.75
mmol, 1.25 equiv) and K CO (82.5 mmol, 1.5 equiv) in 70
3
.3. Catalyst Stability. To assess the catalyst stability, we
repeated the reaction in the same reactor charged with the
catalyst used in the first run under flow. Entry 4 shows that the
catalytic activity markedly decreased with time of usage. We
therefore repeated the experiment using a higher amount of
catalyst and water (50% aqueous EtOH) to ensure the
homogeneous nature of the solution. The column reactor
2
3
mL ethanol HPLC grade (EtOH) + 80 mL distilled water
(
H O). The mixture was passed at 1 atm pressure through a
2
Syrris column reactor with an adjustable end (0.785 cm ID ×
6.5 cm) charged with 1.76 g SiliaCat DPP-Pd at 70 °C. Flow
rate used was 0.5 mL/min (0.16 mL/min Solution 1 and 0.34
mL/min Solution 2). The product was collected in water and
extracted with ethyl acetate or diethyl ether. The organic layer
(
0.785 cm ID × 4 cm) was charged with 1.00 g SiliaCat DPP-
Pd which leads to a 0.7 mol % catalytic amount. Entry 5 in
Table 1 shows that at 1.00 mL/min flow rate almost full
substrate conversion was obtained even after 72 min (49
turnovers) of reagent flow. Entry 6, however, shows that
continuing the Entry 5 experiment with another 144 mL
reaction mixture caused a similar drastic decrease in activity.
We did not explicitly determine the loss of palladium and
silicon from the catalyst system due to leaching, but we did
determine the palladium and silicon residues in some of the
crude isolated products. Although this may not directly
correlate to the leaching from the catalyst, it is an important
measure with respect to product purity, especially in the
context of molecules going forwards, for instance for
pharmaceuticals preparation, where elemental residues are
subject to regulatory limits. The crude products isolated from
the experiments in entries 5 and 6 in Table 1 were thus
analyzed by ICP-OES in DMF solvent (100 mg/mL
concentration). Results are given in mg/kg biaryl product.
Very low (2 mg/kg) Pd residues were found in the biaryl
product pointing to minimal leaching from the solid phase,
lower than for the reactions in batch (entries 1 and 2). The Si
leaching values were more pronounced when water was used as
cosolvent. Indeed, in the batch reaction in the presence of
water, 29.25 mg/kg of leached silicon (Entry 2) were found
versus 4.51 mg/kg leached in absence of water (Entry 1),
namely almost 1 order of magnitude higher leaching of silicon.
Under flow the amount of Si leached from the catalyst (Entry 5
in Table 1) is comparable with that leached under batch (Entry
2); with further increase by reducing the catalyst amount
(Entry 6). These findings point to hydrolysis of the organosilica
matrix in aqueous base, with increasing time.
was separated, dried over MgSO , and filtered. The organic
solvent was evaporated in vacuum to obtain the crude product,
which usually did not require further purification.
4
2
.3. Analysis. Conversion and selectivity were assessed by
using a PerkinElmer GC−MS system equipped with a Clarus
00 gas chromatograph (RTx-5MS, 30 m, 0.25 mm ID, 0.25
6
μm df) and a Clarus C mass spectrometer.
The leaching in Pd and Si was assessed by ICP-OES analysis
(
PerkinElmer Optima 2100 DV) in the isolated crude product,
in DMF solvent (100 mg/mL concentration) and reported to
mg/kg product. LODPd = 0.05 ppm in solution or 0.50 mg/kg
in the crude product; LOD = 0.02 ppm in solution or 0.20
Si
mg/kg in the crude product.
The nitrogen adsorption and desorption isotherms (BET) at
7
7 K of the heterogeneous catalyst were measured using an
ASAP 2020 system from Micromeritics, analyzing the resulting
data with a Tristar 3000 software (version 4.01). The
desorption branch was used to calculate the pore size
distribution.
3. RESULTS AND DISCUSSION
3.1. Continuous vs Batch Reaction. The first reaction
model studied is the Suzuki coupling between 2-chlorobenzo-
nitrile and 4-methyl-phenylboronic acid (Scheme 2), an
important step reaction in the synthesis of valsartan, including
10
the high-throughput process reported in 2007.
Entry 1 in Table 1 shows that this reaction over SiliaCat
DPP-Pd, carried out in batch starting from 6 mmol substrate,
11
requires 30 min for complete conversion. When these
B
dx.doi.org/10.1021/op4003449 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX