Solvent-free semihydrogenation of acetylene alcohols in a capillary reactor coated with a Pd-Bi/TiO2 catalyst

Article abstract of DOI:10.1016/j.apcata.2016.01.019

A solvent-free semihydrogenation of 2-methyl-3-butyn-2-ol (MBY) to 2-methyl-3-buten-2-ol was performed in a capillary reactor (10 m long, 0.53 mm i.d.) coated with a titania supported Pd-Bi catalyst. Several coatings with different Pd/Bi ratio have been prepared. The catalysts have been characterized with SEM, TEM, EDX, XRD analysis and N₂ adsorption-desorption measurements. The maximum alkene yield of 90% was obtained at a molar Pd/Bi ratio of 11. The yield was increased to 95% in the presence of 10 mol.% pyridine in the reaction mixture. The alkene selectivity decreased with time due to leaching of Bi. The leaching was fully suppressed in the presence of 1 vol.% acetic acid in the reaction mixture. The catalyst remained stable for 100 h of continuous operation. The results demonstrate that capillary reactors provide alkene selectivity the same compared to ideal stirred tank batch reactors.

Full text of DOI:10.1016/j.apcata.2016.01.019

Applied Catalysis A: General 515 (2016) 108–115
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Solvent-free semihydrogenation of acetylene alcohols in a capillary
reactor coated with a Pd–Bi/TiO catalyst
2
Nikolay Cherkasova,1, Alex O. Ibhadon , Evgeny V. Rebrovb,c
a,∗
a
Catalysis and Reactor Engineering Research Group, Department of Chemistry and School of Biological, Biomedical and Environmental Sciences, University
of Hull, Cottingham Road, Hull HU6 7RX, UK
b
School of Engineering, University of Warwick, Coventry CV4 7AL, UK
Department of Biotechnology and Chemistry, Tver State Technical University, Tver 170026, Russia
c
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 1 November 2015
Received in revised form
A solvent-free semihydrogenation of 2-methyl-3-butyn-2-ol (MBY) to 2-methyl-3-buten-2-ol was per-
formed in a capillary reactor (10 m long, 0.53 mm i.d.) coated with a titania supported Pd–Bi catalyst.
Several coatings with different Pd/Bi ratio have been prepared. The catalysts have been characterized
with SEM, TEM, EDX, XRD analysis and N2 adsorption–desorption measurements. The maximum alkene
yield of 90% was obtained at a molar Pd/Bi ratio of 11. The yield was increased to 95% in the presence
of 10 mol.% pyridine in the reaction mixture. The alkene selectivity decreased with time due to leaching
of Bi. The leaching was fully suppressed in the presence of 1 vol.% acetic acid in the reaction mixture.
The catalyst remained stable for 100 h of continuous operation. The results demonstrate that capillary
reactors provide alkene selectivity the same compared to ideal stirred tank batch reactors.
© 2016 Elsevier B.V. All rights reserved.
2
2 December 2015
Accepted 15 January 2016
Available online 19 January 2016
Keywords:
Semihydrogenation
Acetylene
Pd
Solvent-free
1
. Introduction
The conversion of alkynes into alkenes through semihydrogena-
dehydrolinalool (C10H16O, 3,7-dimethyl-6-octaen-1-yne-3-ol) and
dehydroisophytol (C20H38O, 3,7,11,15-tetramethyl-1-hexadecyne-
3-ol) follows the same trend while a maximum of selectivity
is observed at a slightly different ratio of Pd and second metal
in the bimetallic catalyst [6]. Still such minor change could not
be considered as significant therefore all results obtained in the
hydrogenation of 2-methyl-3-butyn-2-ol can be translated to dehy-
drolinalool and dehydroisophytol.
The state-of-the art approach to semihydrogenation includes (i)
the use of batch or semi-batch reactors, and (ii) application of Pd
catalysts partially poisoned with Pb (Lindlar catalyst) or other met-
als such as Cu, Ag, Bi, Sn, Ga [7–11]. Batch reactors provide quick
scaling up from laboratory to industrial scale and they are easily
adaptable for various reaction types [12,13]. However, their low
energy efficiency, the excessive use of solvents, and the high toxic-
ity of lead compounds in the Lindlar catalyst cause environmental
concerns. Moreover, the process optimization is usually required
even after minor changes in the substrate molecule structure to
maximize the product yield, which is particularly laborious for
batch reactors [2]. Not surprisingly, small-scale industrial chem-
ical processes in the pharmaceutical industry produce more than
tion reactions is a critical step in the synthesis of many vital fine
and speciality chemicals. For example, semihydrogenation reac-
tions are used in the synthesis of a fragrance compound linalool
and vitamins K1 and E [1,2]. Other vitamins are industrially pro-
duced via semihydrogenation steps at a scale of up to 10,000 tons
a year [1–3]. Moreover semihydrogenation reactions are exten-
sively used in the synthesis of many other important fine chemicals
including fragrances, agrochemicals, food additives, pharmaceuti-
cals and others [1,4,5]. In particular, Fig. 1 shows the scheme of
hydrogenation reactions of an important industrial intermediate
2
-methyl-3-butyn-2-ol (MBY), where along with the target semi-
hydrogenation product, 2-methyl-3-buten-2-ol (MBE), by-product
formation is possible via overhydrogenation to MBA or dimerisa-
tion to C10 products. The results presented by Rebrov et al. [6],
Yarulin et al. [7], and our previous results [6] show that MBY is
a representative example for a large number of acetylene alcohols
with 10, 15 or 20 carbon atoms. The kinetics of hydrogenation of
2
5 kilograms of waste per kilogram of the product, thus having a
very high overall environmental impact [1,14].
An elegant solution to the environmental problems described
lies in process intensification which is possible in microreactors
∗ Corresponding author.
E-mail address: a.o.ibhadon@hull.ac.uk (A.O. Ibhadon).
1
Present address: School of Engineering, University of Warwick, Coventry CV4
7
AL, UK.
http://dx.doi.org/10.1016/j.apcata.2016.01.019
926-860X/© 2016 Elsevier B.V. All rights reserved.
0

N. Cherkasov et al. / Applied Catalysis A: General 515 (2016) 108–115
109
[
15–18]. Microreactors provide high surface to volume ratio of
several orders of magnitude as high as batch reactors, which
results in high heat- and mass-transfer rates [19–22]. The high heat
transfer rates allow fast heating or cooling of reaction mixture,
so highly exothermic hydrogenation reactions can be performed
under solvent-free conditions at high rate. In microreactors, the
reactions are kinetically controlled, resulting in higher product
selectivity. Small reaction volume of microreactors allows safe
utilization of high pressure using inexpensive equipment, while
laminar flow allows precise control of the residence time [15,19,23].
In addition, continuous operation of microreactors allows simple
and quick process optimization resulting in lower labour costs
[
12,24]. All these advantages result in substantially improved
sustainability of chemical processes performed in microreactors
[
15,22,25,26].
Although microreactors have demonstrated great promise in
many organic liquid-phase reactions, application of heterogeneous
catalysts for gas–liquid reactions still remains a major challenge
Fig. 2. Schematic view of the capillary reactor.
[
19,27–29]. For example, hydrogenation reactions require an effi-
nia supported Pd–Bi catalyst. This catalyst uses Bi as a non-toxic
selectivity enhancing agent, which increases alkene semihydro-
genation selectivity by preferentially poisoning step sites of Pd
particles [8,9].
cient handling of (i) hydrogen gas, (ii) liquid substrate medium,
and (iii) solid catalyst to ensure quick mass transfer and reaction
rates. Furthermore, semihydrogenation reactions require precise
control of the reactant residence time due to the possibility of over-
hydrogenation. Therefore, essential requirements for an efficient
semihydrogenation in microreactors are (i) a consistent, control-
lable and very narrow residence time distribution, and (ii) the
absence of internal diffusion limitations [30].
2. Experimental
2.1. Synthesis of Pd–Bi/TiO2 coating
There are three ways to perform heterogeneously catalysed
multiphase reactions in microreactors: (i) using small catalyst
particles in a micro fixed bed configuration [21,31,32], (ii) with
magnetically-supported catalysts held at the reactor walls with
magnetic field [33–36], or (iii) with a catalytic coating deposited
onto the inner reactor walls [6,37–41]. The first approach seems
advantageous because any catalyst suitable for batch hydrogena-
tion can be used to make a packed-bed reactor. However, gas–liquid
reactions in packed-bed reactors have a number of problems asso-
ciated with poor heat and mass transfer. In addition, liquid tends
to form preferential pathways between the catalyst particles lead-
ing to poor radial dispersion and wide residence time distribution,
which in the case of semihydrogenation results in low alkene yield
A titanium dioxide sol was prepared by dissolving 7.3 g of tita-
nium (IV) isopropoxide (Sigma–Aldrich, 97 wt.%) in 8 mL methanol
Fisher Chemical, 99.95 wt.%); then 7.0 mL of water was added
while stirring. After 2 h of stirring under reflux, 0.7 mL of con-
centrated HNO3 (Fisher Scientific, 65 wt.%) was added to disperse
possible agglomerates, the sol was refluxed for 2 h followed by the
addition of 0.3 g Pluronic F127 (Sigma–Aldrich), and the stirring
was continued at room temperature for 12 h.
The dispersion of Pd nanoparticles poisoned with Bi was
obtained by dissolving 160 mg of palladium (II) acetate
Sigma–Aldrich, 98 wt.%) and 210 mg polyvinylpyrrolidone
Sigma–Aldrich) in 10.0 mL methanol. The resulting solution was
reduced by adding 60 mg sodium borohydride (Sigma–Aldrich,
8% wt.). After stirring for 2 h, sodium borohydride was neutral-
(
(
(
[
42–44]. Magnetically-recoverable catalysts offer a convenient way
9
of catalyst introduction and good reusability, but the price of the
catalysts and complexity of reactor operation seem to be a dis-
advantage for industrial semihydrogenation at the current stage.
Therefore, wall-coated reactors are more suitable because they
allow good control of the residence time and provide high heat
transfer rates due to an intimate contact between the catalyst
and the reactor wall. However, wall-coated reactors are usually
more expensive compared to packed-bed systems and require a
reproducible deposition method for catalytic coating. In addition,
high-performance capillary reactors require high catalyst loading,
long reactor channels and thick porous catalytic coatings [6,39]. To
the best of our knowledge, capillary reactors reported in the litera-
ture for semihydrogenation reactions have a throughput below 5 g
of alkynol a day and suffer from solvent overuse [6,39,41,45–47].
In order to address these problems, we have studied a solvent free
semihydrogenation of MBY in a capillary reactor coated with a tita-
ized with 0.2 mL nitric acid, and a solution of 52 mg of bismuth
nitrate pentahydrate (Sigma–Aldrich, 98 wt.%), 0.10 mL nitric acid
in 5.0 mL methanol was added. The dispersion was stirred in
hydrogen for 1 h to poison Pd nanoparticles with Bi [8,48]. The sol
for wall-coating was obtained by mixing 2.3 mL of the TiO2 sol,
1
.0 mL of the Pd–Bi nanoparticle dispersion with methanol to a
total volume of 10.0 mL.
A fused silica capillary (10 m × 0.53 mm i.d.) was activated by
passing a saturated aqueous NaOH solution for 2 h and then wash-
ing with water. The sol containing Pd–Bi/TiO2 was fed into the
capillary using a syringe and then the filling end of the capillary was
closed with a valve. Afterwards, the capillary was slowly introduced
◦
into an oven (200 C) using a stepper motor at a displacement speed
of 0.2 mm/s. The coating method used provides quick and highly
controllable method for solvent evaporation from the precursor
sols and minimises capillary clogging [37]. The wall-coated capil-
◦
lary obtained was dried in a vacuum oven at 200 C for 12 h, washed
−
1
with hexane and methanol (100 ␮L min for 1 h each). The mass of
−
1
the coating was 1.6 mg m with a nominal Pd loading of 2.5 wt.%.
Several capillaries were coated during the study and demonstrated
catalyst mass reproducibility within ± 7%. A glass tube (3.1 mm o.d.,
1
.6 mm i.d., 300 mm long) was coated using the same method, the
coating was mechanically removed and used for XRD and nitrogen
physisorption characterization.
Fig. 1. Scheme of 2-methyl-3-butyn-2-ol hydrogenation.

1
10
N. Cherkasov et al. / Applied Catalysis A: General 515 (2016) 108–115
Fig. 3. (a) Pore size distribution and (b) PXRD pattern of Pd–Bi/TiO2 coating obtained in a 1.6 mm glass tube.
2.2. Characterization
Powder X-ray diffraction (XRD) patterns were recorded using
an Empyrean powder X-ray diffractometer equipped with a
◦
The thickness of the catalyst coating on the capillary channels
monochromatic K␣-Cu X-ray source in the 2ꢀ range 20–70 , step
◦
was studied by using a TM-1000 Hitachi tabletop scanning elec-
tron microscope (SEM) examining 25 capillary cross-sections of the
capillary evenly distributed along the capillary length. Elemental
energy-dispersive X-ray (EDX) analysis was performed in a Ziess
EVO SEM studying the coating mechanically removed from the cap-
illary reactor and placed on a conductive adhesive pad. The content
of Pd, Bi and Ti was determined averaging the results from 4–6 areas
of the powder samples.
Transmission electron microscopy (TEM) study was performed
using a Jeol 2010 instrument equipped with an EDX spectrome-
ter produced by Oxford Instruments. For this study, the coating
removed from the capillary reactor was applied on carbon-coated
copper grids, and 5–8 regions of the grid were studied to obtain
representative data.
size 0.039 and step time 335 s. The pattern was analysed using a
PANalytical Highscore Plus software.
2.3. Catalytic test
The coated capillaries obtained were tested in the apparatus
schematically presented in Fig. 2. The setup included a continuous
flow syringe pump, a pressure transducer, and mass-flow con-
trollers for hydrogen and nitrogen (CK gas, 99.999 vol.%) connected
via a one-way valve and a cross-mixer to the capillary reactor [37].
MBY (Sigma–Aldrich, 98 wt.%) was used for the reaction undiluted
at the flow range from 10 to 100 ␮L min . The hydrogen flow rate
was constant at 10 mL min (STP), and the reaction temperature
was changed in the range from 30 to 70 C using a thermostat. At
each reaction condition studied, the system was allowed to reach
steady state within 10–30 min, then 3–5 liquid samples were taken
from the reactor outlet, diluted 50:1 with hexane and analysed
using a Varian 430 gas chromatograph equipped with an autosam-
pler, a 30 m Stabilwax capillary column and an FID detector.
−
1
−
1
◦
The surface area and pore size distribution of the coatings
were measured by N2 adsorption–desorption isotherms (Energas,
9.999 vol.%) using a TriStar 3000 micrometrics analyser using
9
standard multipoint BET analysis and BJH pore distribution meth-
ods. Prior to nitrogen adsorption, the catalyst material was dried in
◦
a nitrogen flow at 140 C for 3 h.
Fig. 4. (a) SEM image of the Pd–Bi/TiO2 wall-coated capillary, (b) axial coating thickness distribution obtained from 25 cross-sectional SEM images, and (c) TEM image of the
Pd–Bi nanoparticles.

N. Cherkasov et al. / Applied Catalysis A: General 515 (2016) 108–115
111
3
. Results and discussion
3.3. Solvent-free semihydrogenation of MBY
3
.1. Characterization of the Pd–Bi/TiO2 coating
Fig. 6a shows the effect of the reaction temperature and liquid
flow rate on conversion in the solvent-free semihydrogenation of
MBY performed in a 10 m capillary reactor coated with a Pd–Bi/TiO2
catalyst. At a constant reaction temperature, the MBY conversion
decreased at higher liquid flow rates due to lower residence time,
while the MBE selectivity increased up to 94% in the entire range
of temperatures studied (Fig. 6b). An increase in alkene selectivity
at low alkyne conversions is typical for Pd-based catalysts and is
explained by the high adsorption energy of alkyne molecules. This
results in the displacement of alkene molecules from the catalyst
surface and prevents over-hydrogenation to alkanes [56,57]. How-
ever in a hexane solution, the maximum MBE selectivity reached
almost 98% [9,37]. The lower selectivity observed in the capillary
reactor was caused by lower Bi content due to incomplete Bi reduc-
tion from the methanol solution as it was discussed above.
Fig. 3 shows the pore size distribution obtained from the des-
orption branch of the isotherm and an XRD pattern of the coating
obtained. It can be seen that the coating is mesoporous with an
average pore diameter of 2.6 nm and a specific surface area of
2
−1
1
80 m g . The XRD pattern shows that the coating consists of
anatase and brookite phases in a 40:60 ratio. These results are sim-
ilar to our previous study [37], where a titania sol with a higher
concentration was obtained from a titanium butoxide precursor
rather than titanium isopropoxide. In addition, the coating mor-
phology is similar to that reported by Bleta et al. [49] who obtained
TiO2 powders by drying titania sols. These data suggest that the
coating morphology can be tuned using the procedures reported
for sol drying [49,50].
A representative SEM image of the coated capillary is shown
in Fig. 4a. Statistical analysis of the coating thickness obtained by
studying cross-sectional SEM images (Fig. 4b) shows that a uni-
form coating was obtained with an average thickness of 2.0 ␮m
and the standard deviation of 1.0 ␮m. The TEM study showed that
Pd–Bi particles were 4–10 nm in diameter (Fig. 4c). The Pd content
determined by EDX analysis was 2.8 ± 0.3 wt.%, which is in good
agreement with the nominal Pd loading of 2.5 wt.%. However, the
Bi content was 0.5 ± 0.2 wt.% corresponding to a Pd/Bi molar ratio of
about 11. Because large amount of the material was needed for ICP
analysis (10 mg), it was performed only for the sample obtained in a
The fluctuations in the MBY conversion and MBE selectivity
observed in a 10 m capillary reactor were smaller (Fig. 6), as
compared with those of ± 4% in a 2.5 m long reactor [37]. These
discrepancies can be explained by fluctuations in the reaction
temperature, reactant flow rates, and more importantly, the pres-
sure drop changes as a result of transition between hydrodynamic
regimes in the capillary reactor. It appears that a longer reactor
length significantly decreases the pressure fluctuations [58].
The MBE yield as a function of MBY flow rate at different reaction
temperatures is shown in Fig. 7a. At every reaction temperature,
a maximum MBE yield of 87–90% was observed at the MBY flow
rate corresponding to the MBY conversion of 92–98%. An increase
or decrease in the MBY flow rate provided a lower MBE yield as a
result of either decreased conversion or over-hydrogenation. How-
ever, no products other than MBE and MBA were identified, which
shows zero selectivity towards oligomerization. More importantly,
the reactor throughput increased with the reaction temperature
1
.6 mm id glass tube. The obtained results, a Pd loading of 2.55 wt.%
and a Bi loading of 0.32 wt.%, showed excellent agreement with
EDX data obtained for a sample obtained from the capillary reac-
tor. The observed Pd/Bi ratio was substantially higher than Pd/Bi
precursor ratio used of 7, which suggest that only a partial reduc-
tion of Bi ions occurred in a methanol solution as opposed their
complete reduction in an aqueous solution [9,37]. This effect can
−
1
allowing the MBE production capacity of 40 ␮L min (or about 50 g
3+
◦
−1
be explained by the change in electrode potential of the Bi /Bi
system in a non-aqueous solvent [51,52].
day ) with the MBE yield of up to 90% in a single 10 m capillary
reactor.
The hydrogen conversion notably increased with the temper-
ature due to increased reaction rate (Fig. 7b). As the temperature
◦
increased from 30 to 70 C, the hydrogen conversion increased from
4
5 to 80% and the slug-annular flow was transformed to the slug
3.2. Flow regimes in the capillary reactor
flow regime near the reactor outlet (Fig. 5).
The solvent-free alkyne hydrogenation of MBY into MBE is
3.4. MBY semihydrogenation in the presence of pyridine
accompanied by very high hydrogen consumption. For example,
the conversion of 1 ␮L of the reactant requires about 230 ␮L (STP)
of hydrogen. Hence, the hydrogen flow rate should be more than 2
orders of magnitude higher than the liquid flow rate to provide
stoichiometric reactant ratio in high throughput hydrogenation
reactors. Moreover, as a result of high hydrogen consumption, a
flow regime can change inside the microreactor [53]. Therefore it is
necessary to obtain the flow regime map corresponding to selected
experimental conditions.
An addition of competitive adsorbates such as quinoline or pyri-
dine is a common way to increase selectivity towards alkene in a
hydrogenation reaction. These molecules, being adsorbed on the
catalyst surface, isolate Pd active sites, and decrease the heat of
adsorption of semihydrogenated alkene species [59]. This increases
alkene selectivity, often up to 99%, via the reduced rate of formation
of over-hydrogenation products, although the alkyne hydrogena-
tion rate also decreases [30,39].
The flow regimes were studied observing flows of hydrogen
and MBY coloured with methylene blue in an untreated silica
capillary with an internal diameter of 0.53 mm using an optical
microscope (Fig. 5). Several flow regimes were observed in the
studied range of gas and liquid flow rates. At the reactor inlet, a
slug-annular regime was always observed under the studied flow
conditions. The flow changed to a slug flow regime when the hydro-
The effect of pyridine addition on the MBY conversion and MBE
selectivity was studied by addition of 1 and 10 mol.% pyridine solu-
◦
tions to the reactant solution at 70 C (Fig. 8). A full MBY conversion
−1
was observed up to a liquid flow rate of 33 ␮L min in the absence
of pyridine. An addition of 1 and 10 mol.% pyridine reduced conver-
sion to 95 and 90%, respectively, due to the blocking of the active
sites [39,60]. The MBE selectivity increased up to 98% in the pres-
ence of pyridine (Fig. 8b). The MBY conversion over the Pd–Bi/TiO2
catalyst decreased by 5% and the MBE selectivity increased by 1.5%
in the presence of high concentration of pyridine (10 mol.%) as com-
pared with the low concentration case (1 mol.%). This change is
relatively low as compared to our previous study [39], where we
observed a 4% increase in the MBE selectivity on addition of pyridine
−
1
gen flow rate decreased to 1–4 mL min . For higher gas flow rates
−1
above 20 mL min , an annular flow regime was realised. A bub-
bly flow regime was not observed as small gas bubbles are quickly
consumed in the hydrogenation reaction forming a single phase
(
liquid) flow [42,54,55]. A single phase liquid flow was observed in
the case of full hydrogen consumption.

1
12
N. Cherkasov et al. / Applied Catalysis A: General 515 (2016) 108–115
Fig. 5. Images of hydrogen-MBY (coloured with methylene blue) two phase flow regimes; and an experimental flow regime map for hydrogen (atmospheric pressure)—MBY
in the capillary with 0.53 mm i.d., jG and jL are superficial liquid and gas velocities, respectively.
Fig. 6. (a) MBY conversion and (b) MBE selectivity as a function of MBY flow rate in solvent-free MBY hydrogenation in a Pd–Bi/TiO2-coated capillary reactor at a hydrogen
−1
flow of 10 mL min and different reaction temperatures.
Fig. 7. (a) MBE yield and (b) hydrogen consumption as a function of MBY flow rate in solvent-free MBY hydrogenation in a Pd–Bi/TiO2-coated capillary reactor at a hydrogen
−1
flow of 10 mL min and different reaction temperatures.
to an MBY methanol solution. In the present study, the MBY con-
centration was more than 3 orders of magnitude higher than that
in our previous study [39]. This results in the preferential occu-
pation of catalyst active sites with MBY even in the presence of
pyridine. On the contrary, in a highly diluted solution, Nijhuis et al.
rate on introduction of 20 mM quinoline, while 200 mM quinoline
fully suppressed the alkene overhydrogenation reaction.
In the presence of the high and low concentrations of pyridine,
the MBE yield increased to 95 and 93%, respectively. However, the
−
1
reactor throughput decreased from 50 to 35 g day
due to the
[
30] observed a twentyfold decrease in the alkene hydrogenation
decreased reaction rate (Fig. 9a). The reaction temperature has no

N. Cherkasov et al. / Applied Catalysis A: General 515 (2016) 108–115
113
Fig. 8. (a) MBY conversion and (b) MBE selectivity as a function of liquid flow rate in MBY hydrogenation in the presence of pyridine in a Pd–Bi/TiO2-coated capillary reactor
at a 10 mL/min hydrogen flow rate.
effect on the MBE selectivity over the Pd–Bi/TiO2 catalyst in the
absence of pyridine (Fig. 9b). It can be concluded that the changes
in flow regimes and associated changes in the mass transfer rates
do not influence alkene selectivity [53].
leads to the full coverage of Pd sites with MBY molecules even at
relatively low MBY concentrations [63].
A constant MBE selectivity of 98% was observed in the first
10 h of operation (Fig. 10). However within the next 80 h, it gradu-
ally decreased to 92%, and remained relatively constant after 90 h
on stream. The MBY conversion decreased from 49 to 42% dur-
ing 100 h on stream. EDX analysis of the catalytic coating showed
that the Pd content decreased from 3.6 ± 0.9 to 3.1 ± 1.1 wt.% and
the Pd/Bi molar ratio increased from 7.0 ± 0.6 to 9.4 ± 1.0% after
3.5. Deactivation of the Pd–Bi/TiO2 catalyst
The MBY semihydrogenation can efficiently be performed with-
1
00 h on stream. It is known that the partial blocking of Pd sites at
−
1
out a solvent with a throughput of 50 g day
in a 10 m long
higher Bi content decreases catalytic activity and at the same time
it increases MBE selectivity [8,9,64]. Therefore it can be concluded
that the decrease in activity was mainly caused by Pd leaching dur-
ing hydrogenation reactions, while the increase in Pd/Bi ratio shows
that the Bi leaching occurred to a larger extent than that of Pd.
It should be mentioned, the MBE selectivity and MBY conversion
were constant during the first 5 h, and then the MBY conversion
increased during a time-on-stream from 6 to 10 h at a constant
MBE selectivity. These data suggest that during the first 5 h, Bi
species leached from the multi-coordinated adsorption sites or
from the catalyst support and no single Pd sites were liberated.
It is known that these Bi species preferentially block step- and
edge-sites of Pd nanoparticles that are the most active in MBY
over-hydrogenation [8,9,48,65,66]. A steady decrease in selectiv-
ity in the period from 10 to 90 h on stream can be explained by
the liberation of these Pd sites as a result of Bi leaching. Following
the same reasoning, constant selectivity and an increase in activity
during the period from 5 to 10 h can be explained by Bi leaching
from more “selective” terrace sites. Finally, between 90 and 100 h
on stream, selectivity was constant at about 92%. The EDX study of
capillary reactor coated with a Pd–Bi/TiO catalyst. However, long-
term stability is an essential requirement for the application of
capillary reactors in industrial production. To study the long term
2
stability of the Pd–Bi/TiO coating, a 2 m capillary reactor (0.53 mm
2
i.d.) was wall-coated with a titania supported Pd–Bi catalyst with a
Pd/Bi molar ratio of 7 and a Pd loading of 3.4 wt.% using the previ-
ously reported method [37]. The catalyst with increased Bi content
(
Pd/Bi molar ratio of 7) was chosen for this study aiming to increase
MBE selectivity without the addition of pyridine.
Fig. 10 shows the performance of this capillary reactor in the
solvent-free hydrogenation of MBY for 100 h. The initial MBE selec-
tivity was above 98%, which is by 4% higher than that obtained over
the Pd–Bi catalyst with a Pd/Bi molar ratio of 11 (Fig. 9b). More-
over, the same 98% MBE selectivity was obtained using a 0.12 M
MBY solution in hexane [9]. This shows that the catalyst behaviour
is identical in the semihydrogenation in a hexane solution and
under solvent-free conditions. Similar behaviour was observed
over Pd/ZnO catalysts in diluted solutions and solvent-free con-
ditions [61,62]. This can be explained by the fact that supported Pd
catalysts have very high heat of adsorption of alkyne species, which
Fig. 9. (a) MBE yield as a function of liquid flow rate in MBY hydrogenation in the presence of pyridine over a Pd–Bi/TiO2-coated capillary reactor at a 10 mL/min hydrogen
flow rate, and (b) MBY conversion vs. MBE selectivity plot for the studied reaction conditions.

1
14
N. Cherkasov et al. / Applied Catalysis A: General 515 (2016) 108–115
individual Pd nanoparticles showed that the Bi content was below
the detection limit (∼0.1 wt.%). A similar selectivity in MBY semi-
hydrogenation reported over a Pd/TiO catalyst [39] suggests that
2
all Bi has been fully leached from the catalyst after 90 h. Some
Bi observed in the sample after 100 h was likely attributed to Bi
nanoparticles deposited on the support. TEM study of nanoparti-
cles was attempted, but no Bi particles were identified, likely, due
to their very small dimensions, detection limit was around 1.5 nm
for the system used.
There are three main possible mechanisms for Bi leaching from
the Pd catalysts. Bi leaching from PdBi intermetallic compounds
through the formation of Bi O3 was identified in electrochemical
2
studies [67]. Segregation of bimetallic nanoparticles was demon-
strated for a number of Pt- and Pd-based bimetallic catalysts in
various reactions and may lead to leaching of Bi in the studied
system [68–72]. Furthermore, a significant Bi leaching from Pd–Bi
catalysts, observed in the selective oxidation of glyoxal and glucose,
was associated with the formation of soluble chelate complexes of
Bi [73,74].
Based on the existing data, it is difficult to identify the leaching
mechanism realised in the studied system. However, a comparison
of the MBY semihydrogenation under solvent-free conditions and
in a non-polar hexane solution where no leaching was observed
during 100 h [37] shows that leaching of Bi significantly depends
on the polarity of solution. This suggests that the leaching was not
caused by segregation because the latter process depends mainly
on the nanoparticle composition rather than on solvent nature.
Fig. 11. MBY conversion and MBE selectivity over a Pd–Bi/TiO2 coating in a 2 m
capillary (0.53 mm i.d.) in the presence of 1 vol.% acetic acid. Liquid flow rate of
−
1
−1
◦
20 ␮L min and hydrogen flow of 5 mL min (STP), 50 C, ambient pressure, 0.2 bar
pressure drop.
techniques as fractionation. Alternatively, in-flow separation can
be performed using ion-exchange resins at the cost of below $2 per
kg of MBY feed [77–80]. The MBY conversion decreased by about
0
.1% per hour and this deactivation did not depend on the acidity
of the feedstock. It can be concluded that the primary reason of the
catalyst deactivation was associated with the leaching of Pd. How-
ever, considering that during the test the amount of hydrogenated
5
substrate was more than 10 times as high as the amount of Pd, the
deactivation rate was remarkably low.
Therefore the formation of Bi O and/or dissolution through the
2
3
The operation of wall-coated capillary reactors under conditions
when the active centre could perform the reaction cycle without
being deactivated improves sustainability of the state-of-the-art
semihydrogenation technology [15,22,25,26]. A comparison with
solvent-free hydrogenation performed on Pd/ZnO and polimer-
stabilised Pd nanoparticles shows similar maximum alkene yield of
around 95% [61,81], while capillary reactor provides much better
formation of chelate complexes are responsible for leaching.
It is known that oxidation of metallic Bi species can be sup-
pressed in an acidic solution [51]. However, an acidic solution is
unlikely to prevent leaching through complexation due to very
high values of stability constants of chelate compounds [75,76].
Therefore the hydrogenation reaction was performed over the
Pd–Bi/TiO2 coating using a 1 vol.% acetic acid reactant solution to
suppress the formation of Bi O .
scalability due to the absence of heat transfer effects.
2
3
−1
A throughput of 50 g day
exceeds the previously reported
Fig. 11 shows long-term MBY conversion and MBY selectivity
data in the MBY hydrogenation over a Pd–Bi/TiO2 catalyst in the
presence of 1 vol.% acetic acid. It can be seen that a constant MBE
selectivity of 98% was observed for 100 h on stream. The EDX data of
the catalytic coating after the reaction showed the same Pd/Bi molar
ratio (7.3 ± 0.8) as that in the fresh catalyst (7.0 ± 0.6). Therefore
one can conclude that leaching of Bi in the solvent-free MBY hydro-
genation was mainly caused by the formation of bismuth oxides or
hydroxides by the reaction with the impurities in the initial solution
such as dissolved oxygen or traces of water. For industrial appli-
cations, the acid introduced can be separated using conventional
production rate by at least an order of magnitude and shows
the feasibility of the capillary reactors for small-scale syntheses
of fine chemicals. Process intensification by increasing the pres-
sure within the reactor to at least 100 bar is expected to provide
further throughput improvement by two orders of magnitude as
the reaction has the first order in hydrogen pressure. Selective
catalyst heating by radiofrequency or microwaves is expected to
−
1
increase the throughput to the ∼10 kg day , which is suitable for
medium-scale high-value chemical syntheses [15,23]. Afterwards,
the throughput can substantially be increased via numbering-up,
i.e. increasing the number of capillaries and the length of capillaries
−
1
bringing throughput to an order of at least 100 ton year [82,83].
4. Conclusions
Solvent-free semihydrogenation of 2-methyl-3-butyn-2-ol
(MBY) was performed in 10 m long capillary reactors wall-coated
with a Pd–Bi/TiO2 catalyst. The highest MBE yield of 90% was
◦
−1
achieved at 70 C using the liquid flow rate of 40 ␮L min corre-
sponding to a reactor throughput of 50 g day−1. The addition of
1
0 mol.% pyridine to the reaction mixture increased the MBE yield
−
1
to 95%, while it decreased the reactor throughput to 35 g day
due to a partial blockage of Pd active sites. The MBE selectivity
was improved up to 98% by decreasing the Pd/Bi molar ratio in
the bimetallic catalysts from 11 to 7. The leaching of Bi species
via the formation of Bi oxides was responsible for a gradual decay
in the MBE selectivity. However, the leaching was completely
suppressed by addition of a 1 vol.% acetic acid solution to the MBY
feed resulting in a stable operation for 100 h.
Fig. 10. MBY conversion and MBE selectivity over a Pd–Bi/TiO2 coating in a 2 m
capillary (0.53 mm i.d.) in solvent-free MBY hydrogenation. Pd loading in the coating
−
1
is 3.4 wt.%, Pd/Bi molar ratio is 7. MBY flow rate is 20 ␮L min , hydrogen flow rate
−
1
◦
is 5 mL min (STP), temperature is 50 C, ambient pressure, 0.2 bar pressure drop.

N. Cherkasov et al. / Applied Catalysis A: General 515 (2016) 108–115
[34] T.K. Houlding, E.V. Rebrov, Green Process. Synth. 1 (2012) 19–31.
115
Another important conclusion is excellent reaction control
[
35] L.M. Rossi, N.J.S. Costa, F.P. Silva, R. Wojcieszak, Green Chem. 16 (2014)
906–3380.
36] T. Zhang, X. Zhang, X. Yan, L. Kong, G. Zhang, H. Liu, J. Qiu, K.L. Yeung, Chem.
Eng. J. 228 (2013) 398–404.
37] N. Cherkasov, A.O. Ibhadon, E.V. Rebrov, Lab Chip 15 (2015) 1952–1960.
38] M. Faustini, B. Louis, P.A. Albouy, M. Kuemmel, D. Grosso, J. Phys. Chem. C 114
achievable in capillary reactors. Capillary reactors provide the same
semihydrogenation selectivity as the laboratory stirred tank batch
reactors, where neither mass nor heat transfer limitations take
place. As a result, selectivity towards intermediate reaction prod-
ucts in capillary reactors is determined by the nature of the system
allowing for improved process safety, decreased labour costs and
solvent consumption.
2
[
[
[
(
2010) 7637–7645.
[39] E.V. Rebrov, E.A. Klinger, A. Berenguer-Murcia, E.M. Sulman, J.C. Schouten,
Org. Process Res. Dev. 13 (2009) 991–998.
[
40] L.N. Protasova, E.V. Rebrov, H.E. Skelton, A.E.H. Wheatley, J.C. Schouten, Appl.
Catal. A Gen. 399 (2011) 12–21.
Acknowledgements
[41] C.H. Hornung, B. Hallmark, M.R. Mackley, I.R. Baxendale, S.V. Ley, Adv. Synth.
Catal. 352 (2010) 1736–1745.
[
42] B.H. Alsolami, R.J. Berger, M. Makkee, J.A. Moulijn, Ind. Eng. Chem. Res. 52
2013) 9069–9085.
Authors are grateful to Dr. C. Wiles, Dr. M.G. Francesconi, Pro-
fessor P. Fletcher and Professor B. Binks for the access to their
equipment and to the European Commission for an FP7 Grant for
the MAPSYN Project (MAPSYN.eu No. CP-IP 309376). The financial
support provided by the European Research Council (ERC) project
(
[43] D. van Herk, P. Casta n˜ o, M. Makkee, J.A. Moulijn, M.T. Kreutzer, Appl. Catal. A
Gen. 365 (2009) 199–206.
[
[
[
[
[
44] B.A. Wilhite, R. Wu, X. Huang, M.J. McCready, A. Varma, AIChE J. 47 (2001)
548–2556.
2
45] J.J.W. Bakker, M.M.P. Zieverink, R.W.E.G. Reintjens, F. Kapteijn, J.A. Moulijn,
M.T. Kreutzer, ChemCatChem 3 (2011) 1155–1157.
46] M. Ueno, T. Suzuki, T. Naito, H. Oyamada, S. Kobayashi, Chem. Commun.
2
79867 and the Russian Science Foundation project 15-13-20015
is gratefully acknowledged.
(
2008) 1647–1649.
47] J. Kobayashi, Y. Mori, K. Okamoto, R. Akiyama, M. Ueno, T. Kitamori, S.
Kobayashi, Science 304 (2004) 1305–1308.
48] J.A. Bennett, G.A. Attard, K. Deplanche, M. Casadesus, S.E. Huxter, L.E.
Macaskie, J. Wood, ACS Catal. 2 (2012) 504–511.
49] R. Bleta, P. Alphonse, L. Lorenzato, J. Phys. Chem. C 114 (2010) 2039–2048.
50] L. Liu, X. Chen, Chem. Rev. 114 (2014) 9890–9918.
References
[
[
1] W. Bonrath, T. Netscher, Appl. Catal. A Gen. 280 (2005) 55–73.
2] W. Bonrath, J. Medlock, J. Schutz, B. Wüstenberg, T. Netscher, in: I. Karame
[
[
[
[
(
Ed.), Hydrogenation, InTech, Rijeka, 2012, pp. 69–90.
51] A. Economou, TrAC Trends Anal. Chem. 24 (2005) 334–340.
52] T.V. Nghi, F. Vydra, J. Electroanal. Chem. Interfacial Electrochem. 71 (1976)
[
[
3] M. Eggersdorfer, D. Laudert, U. Létinois, T. McClymont, J. Medlock, T. Netscher,
W. Bonrath, Angew. Chem. Int. Ed. 51 (2012) 12960–12990.
4] B. Chen, U. Dingerdissen, J.G.E. Krauter, H.G.J. Lansink Rotgerink, K. Möbus, D.J.
Ostgard, P. Panster, T.H. Riermeier, S. Seebald, T. Tacke, H. Trauthwein, Appl.
Catal. A Gen. 280 (2005) 17–46.
3
25–332.
53] J. Yue, G. Chen, Q. Yuan, L. Luo, Y. Gonthier, Chem. Eng. Sci. 62 (2007)
096–2108.
[
2
[
[
54] P. Sobieszuk, J. Aubin, R. Pohorecki, Chem. Eng. Technol. 35 (2012) 1346–1358.
55] A. Günther, S.A. Khan, M. Thalmann, F. Trachel, K.F. Jensen, Lab Chip 4 (2004)
[
[
[
5] H.-U. Blaser, C. Malan, B. Pugin, F. Spindler, H. Steiner, M. Studer, Adv. Synth.
Catal. 345 (2003) 103–151.
6] E.V. Rebrov, A. Berenguer-Murcia, H.E. Skelton, B.F.G. Johnson, A.E.H.
Wheatley, J.C. Schouten, Lab Chip 9 (2009) 503–506.
7] A. Yarulin, I. Yuranov, F. Cardenas-Lizana, D.T.L. Alexander, L. Kiwi-minsker,
Appl. Catal. A Gen. 478 (2014) 186–193.
278–286.
[
[
56] D. Mei, P. Sheth, M. Neurock, C. Smith, J. Catal. 242 (2006) 1–15.
57] G. Wowsnick, D. Teschner, M. Armbrüster, I. Kasatkin, F. Girgsdies, Y. Grin, R.
Schlögl, M. Behrens, J. Catal. 309 (2014) 221–230.
58] V. van Steijn, M.T. Kreutzer, C.R. Kleijn, Chem. Eng. J. 135 (2008) S159–S165.
59] M. García-Mota, J. Gómez-Díaz, G. Novell-Leruth, C. Vargas-Fuentes, L.
Bellarosa, B. Bridier, J. Pérez-Ramírez, N. López, Theor. Chem. Acc. 128 (2010)
[
[
[
[
8] J.A. Anderson, J. Mellor, R.K.P.K. Wells, J. Catal. 261 (2009) 208–216.
9] N. Cherkasov, A.O. Ibhadon, A. McCue, J.A. Anderson, S.K. Johnston, Appl. Catal.
A Gen. 497 (2015) 22–30.
663–673.
[
[
[
[
10] E. Esmaeili, A.M. Rashidi, A.A. Khodadadi, Y. Mortazavi, M. Rashidzadeh, Fuel
Process. Technol. 120 (2014) 113–122.
11] M. Armbrüster, K. Kovnir, M. Behrens, D. Teschner, Y. Grin, R. Schlögl, J. Am.
Chem. Soc. 132 (2010) 14745–14747.
12] D.M. Roberge, L. Ducry, N. Bieler, P. Cretton, B. Zimmermann, Chem. Eng.
Technol. 28 (2005) 318–323.
13] L.N. Protasova, M. Bulut, D. Ormerod, A. Buekenhoudt, J. Berton, C.V. Stevens,
Org. Process Res. Dev. 17 (2013) 760–791.
[
60] M. García-Mota, J. Gómez-Díaz, G. Novell-Leruth, C. Vargas-Fuentes, L.
Bellarosa, B. Bridier, J. Pérez-Ramírez, N. López, Theor. Chem. Acc. 128 (2011)
663–673.
[
[
61] M. Crespo-Quesada, M. Grasemann, N. Semagina, A. Renken, L. Kiwi-Minsker,
Catal. Today 147 (2009) 247–254.
62] N. Semagina, M. Grasemann, N. Xanthopoulos, A. Renken, L. Kiwi-Minsker, J.
Catal. 251 (2007) 213–222.
63] D. Duca, L.F. Liotta, G. Deganello, J. Catal. 154 (1995) 69–79.
64] J. Sá, J. Montero, E. Duncan, J.A. Anderson, Appl. Catal. B Environ. 73 (2007)
[
[
[
[
14] R.A. Sheldon, Green Chem. 16 (2014) 950–963.
15] V. Hessel, D. Kralisch, N. Kockmann, T. Noël, Q. Wang, ChemSusChem 6 (2013)
98–105.
7
46–789.
[
65] M. Crespo-Quesada, A. Yarulin, M. Jin, Y. Xia, L. Kiwi-Minsker, J. Am. Chem.
Soc. 133 (2011) 12787–12794.
66] E. Herrero, V. Climent, J.M. Feliu, Electrochem. Commun. 2 (2000) 636–640.
67] D.R. Blasini, D. Rochefort, E. Fachini, L.R. Alden, F.J. DiSalvo, C.R. Cabrera, H.D.
Abru n˜ a, Surf. Sci. 600 (2006) 2670–2680.
[
[
[
[
16] C. Wiles, P. Watts, Green Chem. 14 (2012) 38–54.
17] I. Huerta, J. García-Serna, M.J. Cocero, J. Supercrit. Fluids 74 (2013) 80–88.
18] V. Paunovic, J.C. Schouten, T.A. Nijhuis, Catal. Today 248 (2015) 160–168.
19] K. Jähnisch, V. Hessel, H. Löwe, M. Baerns, Angew. Chem. Int. Ed. 43 (2004)
[
[
4
06–446.
[
[
68] K.J.J. Mayrhofer, K. Hartl, V. Juhart, M. Arenz, J. Am. Chem. Soc. 131 (2009)
[20] W. Ehrfeld, V. Hessel, H. Löwe, Microreactors, Wiley-VCH Verlag GmbH & Co.
16348–16349.
KGaA, Weinheim, FRG, 2000.
69] S. Gonzalez, K.M. Neyman, S. Shaikhutdinov, H.-J. Freund, F. Illas, J. Phys.
Chem. C 2 (2007) 6852–6856.
70] O.M. Løvvik, Surf. Sci. 583 (2005) 100–106.
71] J.L. Rousset, J.C. Bertolini, P. Miegge, Phys. Rev. B 53 (1996) 4947–4957.
72] N. López, C. Vargas-Fuentes, Chem. Commun. 48 (2012) 1379–1391.
73] M. Wenkin, P. Ruiz, B. Delmon, M. Devillers, J. Mol. Catal. A Chem. 180 (2002)
[
[
[
21] K.F. Jensen, Chem. Eng. Sci. 56 (2001) 293–303.
22] M. Irfan, T.N. Glasnov, C.O. Kappe, ChemSusChem 4 (2011) 300–316.
23] V. Hessel, G. Cravotto, P. Fitzpatrick, B.S. Patil, J. Lang, W. Bonrath, Chem. Eng.
Process. Process Intensif. 71 (2013) 19–30.
[
[
[
[
[
[
24] R.A. Skilton, A.J. Parrott, M.W. George, M. Poliakoff, R.A. Bourne, Appl.
Spectrosc. 67 (2013) 1127–1131.
25] B.P. Mason, K.E. Price, J.L. Steinbacher, A.R. Bogdan, D.T. McQuade, Chem. Rev.
141–159.
[
[
[
74] F. Alardin, B. Delmon, P. Ruiz, M. Devillers, Catal. Today 61 (2000) 255–262.
75] W.R. Harris, Y. Chen, J. Stenback, B. Shah, J. Coord. Chem. 23 (1991) 173–186.
76] A.A. Frutos, L.F. Sala, G.M. Escandar, M. Devillers, J.M. Salas Peregrı ı´ n, M.
González Sierra, Polyhedron 18 (1999) 989–994.
77] H.M. Anasthas, V.G. Gaikar, Sep. Sci. Technol. 36 (2001) 2623–2646.
78] G.M. Gusler, T.E. Browne, Y. Cohen, Ind. Eng. Chem. Res. 32 (1993) 2727–2735.
79] Y.K. Hong, W.H. Hong, Sep. Purif. Technol. 42 (2005) 151–157.
80] H.M. Anasthas, V.G. Gaikar, React. Funct. Polym. 47 (2001) 23–35.
81] N. Semagina, A. Renken, D. Laub, L. Kiwi-Minsker, J. Catal. 246 (2007) 308–314.
82] M. Al-Rawashdeh, F. Yue, N.G. Patil, T.A. Nijhuis, V. Hessel, J.C. Schouten, E.V.
Rebrov, AIChE J. 60 (2014) 1941–1952.
1
07 (2007) 2300–2318.
[
[
[
26] S.G. Newman, K.F. Jensen, Green Chem. 15 (2013) 1456–1472.
27] P. Watts, S.J. Haswell, Chem. Eng. Technol. 28 (2005) 290–301.
28] T. Noël, J.R. Naber, R.L. Hartman, J.P. McMullen, K.F. Jensen, S.L. Buchwald,
Chem. Sci. 2 (2011) 287–290.
[
[
[
[
[
[
[29] A.-K. Liedtke, F. Bornette, R. Philippe, C. de Bellefon, Chem. Eng. J. 227 (2013)
1
74–181.
[30] T.A. Nijhuis, G. van Koten, J.A. Moulijn, Appl. Catal. A Gen. 238 (2003) 259–271.
[31] M.W. Losey, M.A. Schmidt, K.F. Jensen, Ind. Eng. Chem. 40 (2001) 2555–2562.
[32] S.K. Ajmera, M.W. Losey, K.F. Jensen, M.A. Schmidt, AIChE J. 47 (2001)
1
639–1647.
[
83] M. Al-Rawashdeh, F. Yu, T.A. Nijhuis, E.V. Rebrov, V. Hessel, J.C. Schouten,
Chem. Eng. J. 207–208 (2012) 645–655.
[33] R.B.N. Baig, R.S. Varma, Chem. Commun. 49 (2013) 752–770.