Coupling of Heck and hydrogenation reactions in a continuous compact reactor

Article abstract of DOI:10.1016/j.jcat.2009.07.019

A continuous multi-step synthesis of 1,2-diphenylethane was performed sequentially in a structured compact reactor. This process involved a Heck C-C coupling reaction followed by the addition of hydrogen to perform reduction of the intermediate obtained i

Full text of DOI:10.1016/j.jcat.2009.07.019

Journal of Catalysis 267 (2009) 114–120
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Coupling of Heck and hydrogenation reactions in a continuous compact reactor
Xiaolei Fan a, Maria Gonzalez Manchon a, Karen Wilson b, Steve Tennison c, Alexander Kozynchenko c,
Alexei A. Lapkin a, Pawel K. Plucinski a,
*
a Centre for Sustainable Chemical Technology, Department of Chemical Engineering, University of Bath, Bath BA2 7AY, UK
b Department of Chemistry, University of York, York YO10 5DD, UK
c MAST Carbon International Ltd., Henley Park, Guilford, Surrey GU3 2AF, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 27 May 2009
Revised 31 July 2009
Accepted 31 July 2009
Available online 10 September 2009
A continuous multi-step synthesis of 1,2-diphenylethane was performed sequentially in a structured
compact reactor. This process involved a Heck C–C coupling reaction followed by the addition of hydro-
gen to perform reduction of the intermediate obtained in the first step. Both of the reactions were cata-
lysed by microspherical carbon-supported Pd catalysts. Due to the integration of the micro-heat
exchanger, the static mixer and the mesoscale packed-bed reaction channel, the compact reactor was
proven to be an intensified tool for promoting the reactions. In comparison with the batch reactor, this
flow process in the compact reactor was more efficient as: (i) the reaction time was significantly reduced
(ca. 7 min versus several hours), (ii) no additional ligands were used and (iii) the reaction was run at lower
operational pressure and temperature. Pd leached in the Heck reaction step was shown to be effectively
recovered in the following hydrogenation reaction section and the catalytic activity of the system can be
mostly retained by reverse flow operation.
Keywords:
Flow chemistry
Compact reactors
Multifunctional reactors
Tandem reactions
Ó 2009 Elsevier Inc. All rights reserved.
1. Introduction
form hazardous reactions and reactions using toxic reagents, for
instance, selective oxidation of aromatic alcohols by molecular
Microreactor technology (reaction volume ranging in the scale
range from nanolitre to millilitre) has gained enormous interest re-
cently in both academic research and practical applications, espe-
cially for the fine chemical and pharmaceutical industry [1]. For
practical chemical transformations, this engineering technology of-
fers chemistry a potential solution to circumvent limitations of
batch processes, such as (i) temperature runaway and poor selec-
tivity in dealing with exothermic reactions, (ii) hot spots due to
inefficient mixing, in flow systems, which is often called ‘flow
chemistry’.
The advantages of switching conventional batch operations to
small flow systems are readily apparent. A prominent benefit of
this technology is high surface-to-volume ratio which enables a
narrow temperature profile along the reactor and efficient cooling.
This feature offers the opportunity to successfully execute highly
exothermic/endothermic chemical transformations which are usu-
ally inhibited in larger batch reactors. Gross et al. [2] demonstrated
the use of a flow reactor in the synthesis of NBI-75043 (an anti-
insomnia drug) to promote the key halogen–metal exchange step
whose highly exothermic nature limited this total synthesis route
in a large-scale batch. Small reaction volume also endows microre-
action technology with the inherent safety, which allows us to per-
oxygen at up to 24 atm undiluted oxygen pressure [3] and
ring-expansion reaction involved diazo compound as a reactant
[4]. Another additional interest in flow reactor technology is that,
in comparison with batch operation, multi-step synthesis can be
more facile to achieve in flow process by dosing reactants in differ-
ent reaction stages to promote the whole process without either
losing valuable unstable intermediates, product isolation at inter-
mediate stages or at least interrupting the whole process. Based
on this concept, Baxendale et al. [5] developed a micro-flow pro-
cess, which involves various packed columns containing immobi-
lized reagents, catalysts, scavengers or catch and release agents,
for the multi-step synthesis of the alkaloid natural product oxoma-
ritidine. Moreover, an important advantage of flow chemistry over
batch processing is the significant reduction of footprint of trans-
ferring technology to production scale due to the principle of num-
bering up, i.e. productivity is increased by simply adding more
identical reactors instead of expanding the volume of the reactor
[1,6].
In our former study, the use of a multichannel meso-flow com-
pact reactor as a versatile tool for designing practical heteroge-
neous catalysis process with reasonable throughput was
demonstrated. This compact reactor integrated the static mixer
and the microchannel heat exchanger which enabled an excellent
mixing and heat management ability for efficiently and safely run-
ning different kinds of reactions. Several practical examples have
* Corresponding author. Fax: +44 1225 385713.
E-mail address: p.plucinski@bath.ac.uk (P.K. Plucinski).
0021-9517/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2009.07.019

X. Fan et al. / Journal of Catalysis 267 (2009) 114–120
115
Pd(0)/C
H2
3/4 (trans/cis-isomers)
5
I
Pd(0)/C
TEA
1
2
6
Scheme 1. Consecutive synthesis of 1,2-diphenylethane (5).
been successfully demonstrated with this compact reactor, such as
(i) selective oxidation [3] and hydrogenation [7] with pure oxygen
and hydrogen at elevated pressures, (ii) the staged dosing of
gas along the length of the catalytic channel to increase product
yield [3,7], (iii) the kinetic study for fast reactions [3,7]. Further-
more, by integrating nanofluids with the micro-heat exchanger,
rapid dynamic quenching of chemical reactions was also demon-
strated [8].
Following from our previous study, in this work we report the
extension in functionality of the compact reactor and have devel-
oped of an alternative strategy for continuous multi-step synthesis
of 1,2-diphenylethane (an important intermediate for synthesising
flame retardants [9]) which is conventionally obtained by a ‘one-
pot’ reaction combining Heck coupling and hydrogenation reac-
tions, Scheme 1.
5° to 75°. The average size of the metal particle was calculated
using the Scherrer equation.
2.2. Procedure for batch Heck reactions
Eight millilitres of solvent (anhydrous dimethylformamide,
DMF) were first added in a two-necked flask and de-aerated by
N2 flow for 10 min, then 8 mmol of iodobenzene, 12 mmol of sty-
rene, 10 mmol of triethylamine (TEA) and 0.1 mol% of Pd (as heter-
ogeneous catalyst) were introduced into the reactor. The reactor
(equipped with a condenser) was placed in a preheated oil bath
at 413 K and was vigorously stirred for 4.5 h. Samples were taken
at time intervals for off-line GC analysis. All chemicals were pur-
chased from Sigma–Aldrich and used as received.
2.3. Multichannel compact reactor and continuous experiment
procedure
2. Experimental
The cross-sectional schematic diagram of the compact reactor
assembly is shown in Fig. 1. A static mixer was embedded prior
to reaction channels to pre-saturate the liquid phase with gaseous
reactant before introducing the mixed stream into the reaction
channels. The packed-bed channels were imbedded in parallel be-
tween the two micro-heat exchangers. The two micro-heat
exchangers were located underneath and above all the reaction
channels (arranged in cross-flow with respect to the reaction
channels).
The reaction channel configuration for consecutive multi-step
analysis is shown schematically in Fig. 2. Liquid reagents were pre-
mixed and placed in the feed vessels. A HPLC pump (422, Kontron
Instruments) was used to flow liquid feed through packed-bed
reaction channels, and the product was collected in another vessel
2.1. Preparation and characterisation of Pd/C catalyst
The carbon support (ꢀ150
lm particle diameter, synthetic car-
bon manufactured from phenolic resins, MAST Carbons Ltd. Guild-
ford, UK) was dispersed in distilled water (20 mL H2O for 1.5 g
carbon) at a moderate stirring rate under N2 for 30 min. Aqueous
hydrogen tetrachloropalladate(II) (H2PtCl6, 50 mL, 14.1 mM, corre-
sponding to 5 wt% of Pd) solution was introduced by portion
(10 mL each time) dropwise into the slurry and followed by the
addition of aqueous sodium borohydride solution (25 mL,
145 mM, 5 mL each time) under stirring. This sequential introduc-
tion of Pd solution and reducing agent was performed incremen-
tally with 10 min break between each addition. The slurry was
left for 2 h under a nitrogen atmosphere and gentle stirring. Fol-
lowing impregnation, the suspension was filtered and washed with
distilled water several times. The remaining solution was analysed
by the atomic absorption spectroscopy (AAS, Perkin Elmer) to
determine the quantity of deposited palladium. The catalyst was
dried under vacuum at 353 K overnight and stored in a desiccator
prior to being used for the reaction studies.
via
a
low dead-volume six-way valve (Valco Instruments)
L sample loop. The temperature of the reac-
equipped with a 250
l
tor was controlled by using a re-circulating bath (Haake DC30 with
Haake B3 tank) to circulate heat transfer fluid (glycerol) through
the micro-heat exchangers which were located underneath and
above all the reaction channels. Thus, all reaction channels have
an even temperature field (the measured temperature differences
along the reactor were less than 0.5 K). Each channel was equipped
with standard SwagelokÒ fittings for connecting the liquid and the
gas feed lines. The system was pressurised by purging gas (either
N2 or H2) through the reaction channels and the operating pressure
was set by a back pressure regulator (Brooks 5866). Details of the
design concepts and the hydrodynamic characteristics of the com-
pact reactor and details of the rig are specified elsewhere [3,11].
The gas adsorption analysis was performed using a Micromet-
rics ASAP 2010 analyser (N2 as probing gas) for obtaining nitrogen
adsorption isotherms to evaluate the specific surface area (SBET),
average pore diameter (d) and pore volume (Vp) of the carbon sup-
port and catalyst. A relative pressure range of 0.05 to 0.20 was used
to calculate the BET surface area. The median pore diameter was
calculated from the adsorption branch of the isotherm [10] using
the Barrett–Joiner–Halenda (BJH) method (relative pressure range
0.8 to 0.9).
The phase analysis and the determination of the average crys-
tallite size were carried out by X-ray diffraction (XRD) using a Bru-
2.4. Sample analysis
ker AXS D8 Discoverer X-ray diffractometer, with CuK radiation
k = 0.145 nm and a graphite monochromator, in the 2h range of
The analyses of the samples’ composition were performed using
a Varian chromatograph (CP-3800) equipped with a FID detector
a

116
X. Fan et al. / Journal of Catalysis 267 (2009) 114–120
Gas and liquid feedstocks
Packed-bed reaction channel
Static mixer
Product
Thermocouple
Micro heat exchangers
Fig. 1. Schematic cross-sectional diagram of the structured compact reactor.
N2
I
Pd(0)/C
Pd(0)/C
+
substrates
HPLC Pump
6-way valve
Heck reaction channels
intermediate
H2
static mixer
Pd(0)/C
Pd(0)/C
N2
Hydrogenation reaction channels
product
sample out
Fig. 2. A schematic diagram of the channel configuration of the compact reactor for consecutive C–C coupling/hydrogenation. The shaded area corresponds to the two micro-
heat exchangers located underneath and above all the reaction channels.
and a 15 m capillary column (CP Sil 5 CB, 0.25 mm  0.25
lm,
ca. 4.8 nm. Such nanoparticles can easily block micropores of acti-
vated carbon support, which are in the range of ca. 1 nm diameter
[12] and therefore reduce the area of the catalyst substantially.
Varian).
2.5. Determination of Pd leaching
3.2. Evaluation of the catalytic reactivity for Heck C–C coupling
reaction
Aliquots of the reaction filtrate (in batch experiments the post-
reaction mixture was filtered to remove the solids) or the liquid
product (in continuous experiments the mesh closing the reaction
channel prevented macroscopic removal of the catalyst) were col-
lected in a round bottomed flask (20 mL). The samples were evap-
orated to remove organic compounds, dissolved in Aqua Regia and
diluted with distilled water to a certain volume, and were then
analysed by AAS (Perkin–Elmer).
In order to test the activity and selectivity of the prepared 5 wt%
palladium on the carbon catalyst (Pd/C) in the Heck reaction, batch
experiments were performed initially. For comparison purposes, a
commercial palladium on an activated carbon catalyst (5 wt%, ob-
tained from Sigma–Aldrich) was employed as well. The results of
the batch Heck reaction (Table 2) show that the two catalysts have
a similar performance in terms of selectivity to the main Heck cou-
pling product (3). However, the reaction catalysed by the commer-
cial catalyst achieved a 97% of substrate (1) conversion and a 81%
yield of product (3) in 4.5 h, which were higher than that for the
synthesised Pd/C catalyst (Table 2 and Fig. 2). The higher reaction
kinetics of the commercial Pd/C catalyst may be attributed to its
smaller size and therefore it’s smaller contribution to intraparticle
diffusion.
3. Results and discussion
3.1. Catalyst characterisation
The porous structure characteristics of the carbon and the pre-
pared Pd/C catalyst are reported in Table 1. The synthesised cata-
lyst uses a new synthetic carbon support, specifically designed
for deposition of metal nanoparticles with a controlled particle size
distribution. The prepared Pd/C catalyst has a specific surface area
of 727 m2 gÀ1, which is lower than that of the carbon support
(1070 m2 gÀ1), that is, the support lost 32% of its area after impreg-
nation of the palladium. It might be that the surface area loss is due
to the pore blocking by the palladium nanoparticles. The XRD spec-
tra showed the characteristic bands of the palladium phase for the
prepared Pd/C catalyst and the calculated average particle size was
According to the mechanism of C–C coupling reactions using
aryl iodides as reactants and catalysed by supported Pd catalysts,
Pd(0) metal particles on the supports are in fact only precursors
for the soluble palladium species that are the real catalysts in this
system [12–17]. The activated carbon support of the commercial
Table 2
Results of batch Heck reactiona performed using Pd/C (5 wt%) catalysts.
Catalysts
Conversion of Yield of (3)b Selectivity to Pd leaching
(1) (%)
(3) (%)
(ppm)c
Table 1
The porous structure characteristics of the carbon support and the prepared Pd/C
catalyst.
Pd/C, Sigma
Pd/C, this work 87
97
81
74
85
85
11
6
SBET (m2 gÀ1
)
d (nm)
Vp (cm3 gÀ1
)
a
Temperature (T) = 413 K, reaction time (tR = 4.5 h, in dimethylformamide
(DMF), triethylamine (TEA) as base.
Carbon support
Pd/C (5 wt%)
1070
727
13.3
10.2
0.195
0.111
b
No cis-stilbene (4) was found in the product.
c
Aqueous solution was prepared immediately after the reaction.

X. Fan et al. / Journal of Catalysis 267 (2009) 114–120
117
catalyst is in the form of a fine powder with the mean particle size
of 20 m which might make it much easier to release Pd nanopar-
ticles into the reaction mixture as compared with the Pd/C where
ca. 150 m particles were used as the support. This hypothesis
was confirmed by measuring the Pd content in the reaction solu-
tions after the reactions. The Pd concentration was measured by
atomic absorption in the cold or hot filtrates once the reaction per-
iod was completed.
Palladium (11 ppm) was measured in the filtrate of the com-
mercial catalyst, which corresponded to 10% Pd leaching in respect
to the initial loading of Pd onto the activated carbon. This was
found to be greater than the corresponding value of the Pd/C cata-
lysts synthesised in this work (6 ppm, corresponding to 6% Pd
leaching). Because of the mechanism of the Pd-catalysed Heck
reaction [13–15], which appears to require the presence of leached
Pd, it is questionable whether a continuous Heck reaction is a fea-
sible option. However, palladium should re-deposit on the surface
of a support after the completion of the reaction due to: (i) the ab-
sence of aryl halides for oxidative addition, i.e. the absence of driv-
ing force for Pd to leave the support and (ii) the instability of the
soluble Pd complexes. The solubilization/redeposition process
was also found to be controllable by adjusting the reaction condi-
tions [13–16]. Thus, Pd recovery might be achievable under flow
conditions by using a bed of ‘‘empty” support (which could be sub-
sequently used as an additional catalyst bed) following the bed of a
supported Pd catalyst or by using an additional metal scavenger
module [16].
tion channels is due to the change of flow rate (calculated resi-
dence time of the liquid substrates in the reaction channels
reduces from ca. 3 min to ca. 0.5 min by increase flow rate from
0.25 to 2 mL minÀ1). The decrease in the residence time is the only
explanation for the observed results, since the Heck coupling reac-
tions are usually quite ‘‘slow” under batch conditions.
The selectivity to coupling products ((3) and (4)) in the contin-
uous mode was found to be slightly higher than in the batch one
(89% for the flow reactor versus 85% for the batch reactor). This
can be attributed to the presence of cis-stilbene (4) in the product
stream, which was confirmed by GC analysis. In the batch experi-
ments, however, trans-stilbene (3) was the only Heck coupling
product detected. It seems that the conditions of the continuous
reactor with different mass and heat transfer characteristics com-
pared to those of the batch reactor promoted the synthesis of cis-
stilbene (4). However, the ratio of cis to trans product was very
low and equal to ca. 1:40.
The higher conversions in the continuous reactor compared to
those in the batch one (e.g. 58% of conversion for the residence
time t = 2.88 min in the continuous reactor (Fig. 4) compared to
ca. 150 min necessary for the same conversion in the batch reactor
(see Fig. 3)) resulted in a much higher catalyst loading in the con-
tinuous reactor (500 mg of catalyst per cm3 of the channel volume
compared to 2.2 mg of catalyst per cm3 of organic phase in the
batch reactor).
l
l
Significant catalyst deactivation was found in the subsequent
experimeÀn1ts, with yields dropping by ca. 60% for 0.25, 0.5 and
1 mL min
liquid flow rates, and by ca. 53% for the 1.5, and
2 mL minÀ1 flow rates. This is related to the reaction mechanism
since Pd must be leached from the support and thus is no longer
available for consecutive catalytic runs. Pd was detected at differ-
ent levels in the products collected from continuous experiments.
3.3. Compact reactor performance in Heck C–C coupling reaction
The continuous Heck coupling reaction was conducted in a
packed-bed compact reactor. In these experiments only the cata-
lyst synthesised in this work was used; the Sigma Pd/C catalyst,
due to small size of the particles, generated a high pressure drop
across the reactor. Fig. 3 shows the reactor performances (as yield
and selectivity) for the Heck coupling reaction as a function of the
liquid feed rate. A clear dependence of the coupling products yield
on the liquid flow rate was found in this study. For the very first
continuous Heck coupling experiment with a fresh catalyst, the
yield of 58% to coupling products (66% conversion of (1)) was mea-
sured at the flow rate of 0.25 mL minÀ1, whereas, only 18% yield
3.4. Compact reactor performance in alkene hydrogenation
The Pd/C catalysed hydrogenation of intermediates (3) to the fi-
nal product (5) (see the reaction scheme in Fig. 5) was performed
continuously in the compact reactor to test the catalytic activity
of the flow system for alkene hydrogenation. Fig. 5 shows the mo-
(20% conversion of 1) was obtained for the flow rate of 2 mL minÀ1
.
Residence time / min
2.88
0.72
0.48
0.36
1.44
The variation of the residence time of liquid substrates in the reac-
92
90
88
(a)
85
60 (b)
50
I
Pd(0)/C
DMF,TEA
80
100
1
2
3/4 (trans/cis -isomers)
40
run 1
run 2
run 3
run 4
/
/
/
/
80
60
40
30
20
20
10
0
Pd/C, Sigma
Pd/C
0
0.5
1.0
1.5
2.0
Liquid flow rate / mL min-1
0
30
60
90
120 150 180 210 240 270 300
Reaction time / min
Fig. 4. Continuous Heck coupling reaction in the compact reactor. (a) Selectivity to
Fig. 3. Batch Heck C–C coupling reaction catalysed by supported palladium
catalysts. Reaction conditions: 8 mmol of (1), 12 mmol of (2), 10 mmol of TEA
and 0.1 mol% Pd in 8 mL of DMF, T = 413 K, under N2.
3 + 4, (b) yield of 3 + 4. Reaction conditions: C1,0 = 0.5 mol LÀ1, C2,0 = 0.75 mol LÀ1
,
CTEA,0 = 0.6 mol LÀ1, DMF as solvent, T = 390 K, two channels in series, approx. 0.9 g
catalyst.

118
X. Fan et al. / Journal of Catalysis 267 (2009) 114–120
The catalytic system was not found to be effective for producing
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
the main desirable product (5) by using hydrogen to pressurise the
system (Runs 1–3). Benzene (dehalogenation product of substrate
(1)) and ethylbenzene (hydrogenation product of substrate (2))
were detected in the final product. This result indicates that the
H2 might back-flush into the Heck reaction channel and the Heck
coupling reaction was suppressed due to the absence of substrates
caused by their consumption in the side reactions. This hypothesis
was supported by the measured significant increase in the amount
of benzene with the increase of system pressure, i.e. H2 pressure
(Table 3, condition b). In order to eliminate the effect of side reac-
tions, nitrogen was used to pressurise the whole system instead of
hydrogen. After pressurising the system, a small stream of nitrogen
(1 mL minÀ1) was kept to flow through the reaction channels for
providing an inert atmosphere for the Heck reaction. Then, a
hydrogen stream was introduced into the hydrogenation channels.
The system became more efficient for synthesising the target prod-
uct. The complete conversion of the substrate (1) was achieved and
a significant increase in the product yield (ca. 80% selectivity to fi-
nal product) was measured (Runs 4 and 5). This result suggested
that a stepwise conversion of the substrates/intermediate occurred
in the continuous process. Compared with the system under condi-
tion (b) the target product yield became less dependent on the to-
tal pressure for condition (c). The same level of product yield can
be achieved using moderate total pressure. High pressure is not
essential for achieving high product yield.
One disadvantage of the present system is the use of DMF as a
solvent which is toxic and difficult to evaporate during the follow-
ing workup. Therefore, finding a suitable substitute to replace DMF
is essential to make the whole system more environmentally be-
nign. Ethanol is innocuous and less expensive and was considered
in this study. Furthermore, due to the flow reactor being pressur-
ised, solvents with lower boiling points can be held in reaction
channels as superheated liquids to promote the reaction. In our
experiments (Table 3 condition d), ethanol was used as a solvent
for the hydrogenation reaction, Scheme 1, at 393 K and elevated
pressure under continuous flow (Runs 6 and 7). The process with
ethanol was as effective and efficient as the one using DMF, i.e.
the same level of substrate conversion and yield of product (5)
were obtained.
Pd(0)/C, H2
DMF
3
5
H2 supply
initial conc. of 3:
0.6 M
0.4 M
0.2 M
0
2
4
6
8
10
12
14
16
18
H2 flow rate / mL min-1
Fig. 5. Continuous alkene hydrogenation in the compact reactor. Reaction condi-
tions: T = 397 K, P = 8 bar, Fliquid = 1 mL minÀ1, two channels in series, approx. 0.9 g
catalyst in the two channels.
lar flow rate of 1,2-diphenylethane (5) in the outlet of the reactor
(measured at steady state) as a function of H2 flow rate for the
three initial substrate concentrations.
The Pd/C catalyst showed remarkable activity in the packed bed
continuous-flow system. Furthermore, the yields of product (5)
were found to correlate very well with the theoretical hydrogen
supply to the system. This indicates that efficient gas/liquid mixing
(mass transfer) was achieved by the integral static mixer within
the reactor structure. Compared with the previous literature that
used a batch reactor [17], the continuous flow packed-bed flow
reactor enables a more efficient overall hydrogenation process: a
lower temperature (398 K versus 413 K), a lower pressure (8 bar
versus 20 bar) and a much shorter reaction time (ca. 6 min to
achieve steady state with ca. 100% conversion versus 20 h reaction
time).
3.5. Consecutive synthesis of 1,2-diphenylethane in a compact reactor
To evaluate the feasibility of consecutive Heck coupling and
hydrogenation reactions under flow conditions, the reaction of
1,2-diphenylethane synthesis in four consecutively connected
channels of the compact reactor was studied. The results are
shown in Table 3. Compared with alkene hydrogenation, the Heck
coupling reaction is the slower step of the two. A liquid flow rate of
0.25 mL minÀ1, therefore, was used in the consecutive synthesis
experiments in order to promote Heck coupling reaction.
3.6. Retention of Pd in a flow reactor
The aforementioned solubilization/redeposition was well stud-
ied and supported by several batch investigations [13–16,18]. Even
in a flow reactor [19], this mechanism was employed as well for
retrieving leached species from the continuous Heck reactions
(by inserting a column containing the metal scavenger resin after
Table 3
Optimization of the flow chemistrya in the compact reactor for a desirable multi-step organic synthesis (Scheme 1).
Run
Pressure (bar)
Solvent
PhI conversion (%)
Selectivity (%)
trans/cis-Stilbene (3/4) intermediate
1,2-Diphenylethane (5) product
Benzene by-producte
1
2
3
4
5
6
7
1b
4b
8b
4c
8c
4d
8d
DMF
DMF
DMF
DMF
DMF
EtOH
EtOH
83
96
38
0
0
1
0
38
18
5
80
82
83
83
23
81
95
15
13
13
12
100
99
100
100
100
0
0
a
Reaction conditions: C1,0 = 0.4 mol LÀ1, T = 393 K, Fliquid = 0.25 mL minÀ1, Fhydrogen = 8 mL minÀ1, four channels in series, approx. 0.9 g catalyst for the Heck reaction,
approx. 0.75 g catalyst for the alkene hydrogenation.
b
System was pressurised with H2.
System was pressurised with N2.
System was pressurised with N2, ethanol was used as solvent.
Analysis of the reaction did not include the production of ethylbenzene due to the excess of styrene in the system.
c
d
e

X. Fan et al. / Journal of Catalysis 267 (2009) 114–120
119
Channel configuration for forward flow experiment
heat transfer fluid
H2
N2
liquid
reagents
product
Channel 1-2
Channel 3-4
Heck C-C coupling
Hydrogenation
Channel configuration for reverse flow experiment
product
N2
Channel 1-2
Channel 3-4
H2
Hydrogenation
liquid
reagents
Heck C-C coupling
Fig. 6. Channel configurations for normal/reverse flow experiments.
the reaction column). In the present flow reactor, the subsequent
hydrogenation channels can actually serve as scavengers to recap-
ture the leached Pd species from the Heck reaction channels. The-
oretically, the catalytic activity of the system could be maintained
for both the reactions by reverse running [20,21] the two reactions
in each other’s channels under carefully chosen conditions. The
channel arrangement for this concept is shown in Fig. 6. This
hypothesis was proved by the experimental results in Fig. 7 (only
one set of results from forward/reverse runs is presented in
Fig. 7, the same reaction profiles as for reverse run were obtained
in two consecutive experiments). The reuse of the catalyst is viable
without major deactivation by obverse/reverse running two reac-
tions in each other’s reaction channels. Most of the leached Pd spe-
cies were found to re-deposit on the support during hydrogenation
under the conditions used in this study, which was proven by XPS
analysis. Compared with the fresh catalyst in Channel 4, Pd species
on the carbon surface were found to have increased by 79% (from
1.4% to 2.5%) after one experiment. Accordingly, a decrease of oxy-
gen on the carbon surface from 26.2% to 22.3% was measured
which indicates the consumption of oxygen containing surface
groups by redeposition of Pd in hydrogenation channel. Zhao
et al. [13–15] and Heidenreich et al. [16] confirmed this reversible
transfer of Pd species between solvent and support in their studies
and asserted that a unique process is tuneable by carefully choos-
ing reaction conditions and base.
reported that in heterogeneous systems the formation of triethyl-
ammonium hydrohalides might have detrimental effects on cata-
lyst activity [13,22]. This is due to adsorption of the halide salts
on the catalyst surface, which subsequently prevents redeposition
of Pd from solution to the catalyst. Zhao et al. [13] asserted that
adsorption of the triethylammonium iodide on the support is one
of the reasons for hindering Pd redeposition; a batch reactor with
Pd-supported catalysts in a N-methylpyrrolidone (NMP) solvent
was used. However, in the current study, the presence of triethyl-
ammonium iodide species was not determined. With regard to the
flow system used here (with either DMF or ethanol as solvent), the
influence of the dissolved triethylammonium hydrohalides on cat-
alyst deactivation deserves further study. Note that triethylammo-
nium hydrohalides are known to be soluble in polar solvents;
especially in aprotic polar solvents such as NMP, DMF and 1,4-
dioxane [22]. Thus, in the reaction in a polar solvent used in this
study the deposition of salt may have been prevented. Additionally
further investigation is needed to optimise the present catalytic
system in order to minimise the loss of Pd from the reactor, and
to promote rapid redeposition of the Pd species onto the support.
4. Conclusions
The suitability of a compact multichannel reactor for multi-step
organic synthesis was investigated using a conventional ‘one-pot’
reaction, i.e. a Heck C–C coupling reaction followed by hydrogena-
tion for synthesising 1,2-diphenylethane. Compared with conven-
tional batch reactors, the compact reactor promoted a more
intensified Heck reaction and alkene hydrogenation. 1,2-Diphenyl-
ethane was synthesised using the flow compact reactor within
minutes of residence time with a lower operating temperature
and pressure. In comparison, the conventional batch synthesis took
several hours for a similar yield. A stepwise conversion of the
substrates to the final product was achieved in the compact multi-
channel reactor by coupling the Heck reaction and hydrogenation.
Compared with the batch reactor, this flow chemistry process in
the compact reactor features some advantages such as: (i) being
more intensified (much shorter residence time) due to the high
catalyst/substrates ratio, (ii) being operated under benign condi-
tions, i.e. lower temperature and pressure, (iii) using superheated
ethanol at elevated pressure which makes the process greener.
Leached Pd species were believed to be active for igniting and
promoting the Heck reaction. Due to (i) the reversible transfer
mechanism of Pd species between solvent and support, and (ii)
the unique mass transfer of Pd in different reaction channels, cata-
lytic activity of the system can be mostly maintained by reverse run-
ning the two reactions in one another’s reaction channels.
This continuous catalytic system might be a promising alternative
to replacing the conventional batch reactor for multi-step synthesis.
The present investigation demonstrated that a compact reactor
is capable of performing the multi-step synthesis in one stream.
Considering the specific reaction used in this study, i.e. C–C cou-
One consequence of using triethylamine as a base in this flow
reaction system is the formation of triethylammonium hydroha-
lide salts. These are formed by reaction of the amine with the
hydrogen halide, which is released during C–C coupling. It has been
100
80
60
40
20
forward run
reverse run
0
0
2
4
6
8
10
12
14
Time on stream / min
Fig. 7. Time yield of 1,2-diphenylethane 5 for the normal flow direction and the
reverse flow direction runs. Reaction conditions: C1,0 = 0.4 mol LÀ1
T = 393 K,
P = 4 bar, Fliquid = 0.20 mL minÀ1 Fhydrogen = 8 mL minÀ1
EtOH as solvent, four
,
,
,
channels in series, approx. 0.9 g catalyst for Heck reaction, approx. 0.75 g catalyst
for the alkene hydrogenation.

120
X. Fan et al. / Journal of Catalysis 267 (2009) 114–120
[2] T.D. Gross, S. Chou, D. Bonneville, R.S. Gross, P. Wang, O. Campopiano, M.A.
Ouellette, S.E. Zook, J.P. Reddy, W.J. Moree, F. Jovic, S. Chopade, Org. Process
Res. Dev. 12 (2008) 929–939.
[3] D.V. Bavykin, A.A. Lapkin, S.T. Kolaczkowski, P.K. Plucinski, Appl. Catal. A: Gen.
288 (2005) 175–184.
[4] X. Zhang, S. Stefanick, F.J. Villani, Org. Process Res. Dev. 8 (2004) 455–460.
[5] I.R. Baxendale, J. Deeley, C.M. Griffiths-Johns, S.V. Ley, S. Saaby, G.K. Tranmer,
Chem. Commun. (2006) 2566–2568.
[6] S. Becht, R. Franke, A. Geisselmann, H. Hahn, Chem. Eng. Technol. 30 (2007)
295–299.
[7] X.L. Fan, A.A. Lapkin, P.K. Plucinski, Catal. Today (2009), accepted for
publication, doi:10.1016/j.cattod.2009.07.048.
[8] X.L. Fan, H.S. Chen, Y.L. Ding, P.K. Plucinski, A.A. Lapkin, Green Chem. 10 (2008)
670–677.
[9] L. Ye, X. Meng, X. Liu, J. Tang, Z. Li, J. Appl. Polym. Sci. 111 (2009) 2372–2380.
[10] S.J. Gregg, K.S.W. Sing, Adsorption, Surface Area and Porosity, Academic Press,
London, 1982.
pling reactions, however, there is still much room left for optimis-
ing this flow catalytic system with compact reactor. Therefore, fu-
ture studies on C–C coupling reactions in this flow reactor system
are suggested to focus on (i) developing the ligand-free catalytic
system which is capable of effectively and efficiently catalysing
C–C coupling reactions involving aryl bromides and aryl chlorides
(less active, more practical and economical substrates), (ii) mini-
mising the Pd leaching by either optimising present flow catalytic
system or engineering new supported Pd catalysts for
true surface catalysis. The optimisation of styrene concentration
is also necessary to reduce the formation of by-products
(ethylbenzene).
[11] P.K. Plucinski, D.V. Bavykin, S.T. Kolaczkowski, A.A. Lapkin, Ind. Eng. Chem. Res.
44 (2005) 9683–9690.
Acknowledgments
[12] A. Pigamo, M. Besson, B. Blanc, P. Gallezot, A. Blackburn, O. Kozynchenko, S.
Tennison, E. Crezee, F. Kapteijn, Carbon 40 (2002) 1267–1278.
[13] F. Zhao, B.M. Bhanage, M. Shirai, M. Arai, Chem. Eur. J. 6 (2000) 843–848.
[14] F. Zhao, K. Murakami, S.M.M. Arai, J. Catal. 194 (2000) 479–483.
[15] F. Zhao, M. Shirai, Y. Ikushima, M. Arai, J. Mol. Catal. A: Chem. 180 (2002) 211–
219.
[16] R.G. Heidenreich, K. Köhler, J.G.E. Krauter, J. Pietsch, Synlett (2002) 1118–1122.
[17] M. Gruber, S. Chouzier, K. Koehler, L. Djakovitch, Appl. Catal. A: Gen. 265
(2004) 161–169.
[18] F. Zhao, M. Shirai, M. Arai, J. Mol. Catal. A: Chem. 154 (2000) 39–44.
[19] N. Nikbin, M. Ladlow, S.V. Ley, Org. Process Res. Dev. 11 (2007) 458–462.
[20] J.E. Bailey, Chem. Eng. Commun. 1 (1973) 111–124.
[21] Y.S. Matros, Catalytic Processes under Unsteady State Conditions, Elsevier
Science Publishers, Amsterdam, The Netherlands, 1989.
[22] R.L. Augustine, S.T. O’Leary, J. Mol. Catal. 72 (1992) 229–242.
We gratefully acknowledge financial support from Engineering
and Physical Sciences Research Council (Engineering Functional
Materials, EP/C519736/1). XF is grateful to ORS award and Univer-
sity of Bath Research Studentship. MGM was an Erasmus exchange
student from the Universidad Complutense de Madrid, Spain. We
appreciated the assistance of Mr. Alan Carver for performing atom-
ic adsorption spectrometry analysis.
References
[1] W. Ehrfeld, V. Hessel, H. Löwe, Microreactors, Wiley-VCH, Weinheim, 2000. p.
288.