Moving from batch to continuous operation for the liquid phase dehydrogenation of tetrahydrocarbazole

Article abstract of DOI:10.1021/op400217d

Despite the numerous advantages of continuous processing, high-value chemical production is still dominated by batch techniques. In this paper, we investigate options for the continuous dehydrogenation of 1,2,3,4- tetrahydrocarbazole using a trickle bed reactor operating under realistic liquid velocities with and without the addition of a hydrogen acceptor. Here, a commercial 5 wt % Pd/Al₂O₃ catalyst was observed to slowly deactivate, hence proving unsuitable for continuous use. This deactivation was attributed to the strong adsorption of a byproduct on the surface of the support. Application of a base washing technique resolved this issue and a stable continuous reaction has been demonstrated. As was previously shown for the batch reaction, the addition of a hydrogen acceptor gas (propene) can increase the overall catalytic activity of the system.

Full text of DOI:10.1021/op400217d

Article
pubs.acs.org/OPRD
Moving from Batch to Continuous Operation for the Liquid Phase
Dehydrogenation of Tetrahydrocarbazole
Yan Shen,†,∥ Azman Maamor,†,⊥ Jehad Abu-Dharieh,† Jillian M. Thompson,*,† Bal Kalirai,‡ E. Hugh Stitt,§
and David W. Rooney*,†
†CenTACat, School of Chemistry and Chemical Engineering, Queen’s University Belfast, David Keir Building, Stranmillis Road,
Belfast BT9 5AG, U.K.
‡Robinson Brothers, Ltd., Phoenix Street, West Bromwich B70 0AH, U.K.
§Johnson Matthey Catalysts, PO Box 1, Belasis Avenue, Billingham, TS23 1LB, U.K.
S
* Supporting Information
ABSTRACT: Despite the numerous advantages of continuous processing, high-value chemical production is still dominated by
batch techniques. In this paper, we investigate options for the continuous dehydrogenation of 1,2,3,4-tetrahydrocarbazole using a
trickle bed reactor operating under realistic liquid velocities with and without the addition of a hydrogen acceptor. Here, a
commercial 5 wt % Pd/Al2O3 catalyst was observed to slowly deactivate, hence proving unsuitable for continuous use. This
deactivation was attributed to the strong adsorption of a byproduct on the surface of the support. Application of a base washing
technique resolved this issue and a stable continuous reaction has been demonstrated. As was previously shown for the batch
reaction, the addition of a hydrogen acceptor gas (propene) can increase the overall catalytic activity of the system.
contact time. A factor of 3 decrease in the contact time afforded
by continuous operation was also observed to improve the
selectivity for the dehydration of glycerol to acetol from 55% to
70%.2
1. INTRODUCTION
Motivated by a desire for flexibility, liquid phase processing of
fine chemicals occurs largely under batch operation, allowing
use of installed equipment for a wide variety of reactions. It is
generally considered that the initial capital costs are also
relatively low when compared to continuous operation and
installation is relatively simple, meaning that, if necessary,
additional capacity can be easily accommodated by numbering-
up or scaling out. However, the time required for starting up
each batch reaction and the subsequent downtime including
cooling as well as post reaction cleanup can be longer than the
reaction time itself, resulting in significant loss of manufacturing
time and increased labour.
Conversely, continuous operation occurs under steady state
for extended times, thereby requiring smaller equipment and
facilities as well as less labour and downtime. With decreased
manual input, there is also a reduced chance for exposure to
risk, leading to improved process safety particularly in cases
where toxic, flammable, or explosive materials are used.
Decreasing start-up and cool-down cycles also reduces the
time required to produce a given campaign, resulting in savings
in both energy and materials when compared to a batch
process. Furthermore, the more stable and consistent reaction
conditions generally result in a higher quality product when
compared to batch production.
While not all liquid phase processes are suitable for
conversion from batch to continuous operation, there are
cases where a larger scale fine chemical process could benefit
from being run continuously. To determine when continuous
operation would be advantageous, Calabrese and Pissavini
compiled a table of guidelines for initial assessment of flow
reactor applicability, in which severe reaction conditions such as
pressures greater than 120 bar, temperatures less than −10 °C,
formation of toxic byproducts, highly exothermic reactions, and
sequential reactions reducing product selectivity amongst
others are given as reasons to consider use of a flow reactor.3
Anderson discussed similar advantages amongst others for
continuous operation for the production of pharmaceutical
materials, citing improved control of mixing and heat transfer,
more facile operation for both cryogenic and high-temperature
processes, removal of reactive intermediates resulting in
improved selectivity, and the removal of toxic compounds.4
Further discussion on the topic of batch to continuous
processing is given in the article by Stitt and Rooney.5
Herein we expand on our work on the dehydrogenation of
1,2,3,4-tetrahydrocarbazole (THCZ) by investigating the liquid
phase dehydrogenation reaction under continuous operation
using a trickle bed reactor (TBR).
Rode et al. demonstrated the advantages of continuous
processing for the selective hydrogenolysis of glycerol to 1,2-
propanediol, where the conversion and selectivity to the desired
product increased from 34% to 65% and 84% to over 90%,
respectively,1 on changing from batch to continuous packed
bed operation. These improvements in rate and selectivity were
attributed to a number of factors including in situ catalyst
activation and suppression of side reactions from a lower
Carbazoles can be produced from coal tar and crude oil or
through synthetic routes such as the Graebe−Ullmann
reaction6 and the liquid-phase dehydrogenation of 1,2,3,4-
Received: August 12, 2013
Published: March 11, 2014
© 2014 American Chemical Society
392
dx.doi.org/10.1021/op400217d | Org. Process Res. Dev. 2014, 18, 392−401

Organic Process Research & Development
Article
tetrahydrocarbazole obtained from the Borsche−Drechsel
synthesis.7 The dehydrogenation has been facilitated over a
range of metal catalysts such as nickel, palladium, and
unsupported molybdenum carbide.8 Carbazoles are well-
known chemical intermediates, used in the synthesis of
pharmaceuticals, agrochemicals, dyes, pigments, and other
organic compounds. Substituted carbazoles have been used
for optical applications,9 and many others are biologically
active10 with their alkaloid derivatives having antitumor,
antibacterial, antimicrobial, and anti-inflammatory properties.11
More recently hydrocarbazoles have been investigated as
potential materials for on-board organic hydrogen storage.12
Dehydrogenation of dodecahydro-N-carbazole to tetrahydro-
carbazole and on to carbazole has been highlighted as a
potential system for this application; however, low reaction
rates attributed to strong adsorption of the substrate and
products have been observed. In contrast, dehydrogenation of
dodecahydro-N-ethylcarbazole is considerably faster than
dodecahydro-N-carbazole as the ethyl group causes a steric
hindrance, preventing the strong adsorption of the organic
materials on the catalyst surface.13 However, decomposition of
dodecahydro-N-ethylcarbazole by loss of the ethyl group at
temperatures greater than 350 K14 as well as the lower
hydrogen storage capacity of 5.8 wt % compared to 6.7 wt % for
dodecahydro-N-carbazole15 maintains interest in developing a
dehydrogenation catalyst to form the unsubstituted carbazole.
Previously, we reported both increased yields and selectivities
for the dehydrogenation of tetrahydrocarbazole to carbazole
(CZ) using a commercial 5 wt % Pd/Al2O3 catalyst with the
addition of propene into the stirred tank reactor (STR)
(Scheme 1).
the gas phase calculations, in the presence of the solvent, the
reaction was controlled by the first dehydrogenation step.17
The above results suggest that the dehydrogenation of
1,2,3,4-tetrahydrocarbazole would be a relevant, useful, and
challenging test molecule to demonstrate the potential for
continuous dehydrogenation in the fine chemical industry.
2. EXPERIMENTAL SECTION
2.1. Catalysts and Materials. Tetrahydrocarbazole
(TCHZ) (99%) and mesitylene (97%) were purchased from
Aldrich and used as received. All gases were purchased from
BOC and were of greater than 99% purity. The 5 wt % Pd/
Al2O3 catalysts used throughout this work were commercially
available catalysts obtained from Johnson Matthey as either a
powder (Al2O3 type 324, particle size 20 μm) or eggshell
cylindrical pellets of nominal 3 mm size (Al2O3 type 49). Prior
to use in a batch reactor, the pellets were crushed and sieved to
particles of less than 75 μm. Base-washed catalysts were
prepared from the pellets or crushed pellets by placing the 5 wt
% Pd/Al2O3 (5 g) into a round-bottomed flask with a 0.1 M
aqueous solution of lithium hydroxide, sodium hydroxide, or
potassium hydroxide (25 mL) and stirring for 1 to 2 h before
leaving overnight. The catalysts were removed by filtration and
washed with water until a neutral pH was obtained before
drying in an oven at 110 °C overnight. These treated catalysts
were labelled 5 wt % Pd/Al2O3 (Li, Na, or K), indicating the
base which was used to treat the 5 wt % Pd/Al2O3.
The 5 wt % Pd/C catalyst was also a commercial catalyst
obtained from Johnson Matthey. The 5 wt % Pd/α-Al2O3
catalyst was prepared on an alpha alumina support, purchased
from Alfa Aesar, by wet impregnation using an aqueous
solution of palladium nitrate containing nitric acid. After drying
overnight at 120 °C, the catalyst was calcined at 500 °C under a
flow of air for 3 h and reduced in a H2 flow at 400 °C for 5 h18
with argon used to purge the catalyst before and after reduction
for 15 min.
Scheme 1. Reaction Scheme for the Dehydrogenation of
Tetrahydrocarbazole
2.2. Catalyst Characterisation. The catalysts were
characterised by a range of techniques and the characterisation
data is summarised in Table 1. BET N2 isotherms (Micro-
Table 1. Characterisation of the Catalysts Used for the
Dehydrogenation of Tetrahydrocarbazole
BET
support
particle
size
surface
area
pore
volume
(cm3 g−1
Pd
dispersion
a
catalyst
form
(m2 g−1
)
)
5 wt % Pd/ powder
Al2O3
20 μm
143
0.53
22.4
5 wt % Pd/ crushed 3
<75 μm
82
0.19
6.4
Al2O3
mm
pellets
5 wt % Pd/ crushed 3
<75 μm
103
0.24
−
Al2O3(Na)
mm
pellets
a
Pd dispersion taken from reference 16.
It was observed that the reaction was limited by the build-up
of hydrogen within the pores of the catalyst, which favored the
hydrogenation of the intermediate product to form the initial
starting material. By adding a gaseous phase hydrogen acceptor,
such as propene, the hydrogen within the catalyst pores was
rapidly reacted out of the system, allowing the reaction to
proceed to completion at a significantly faster rate.16
meritics ASAP 2010) were used to determine the surface area
and pore volumes whilst the metal loading was measured by
ICP-OES (Perkin-Elmer Optima 4300). The metal dispersion
of each catalyst was determined by CO chemisorption, which
required the catalysts to be reduced at 150 °C in 5% H2/Ar and
cooled to room temperature in He before pulses of CO (2.38 ×
10−6 mol) were passed over the catalyst until saturated. The
metal dispersion was calculated, assuming a Pd:CO ratio of 1:1.
DFT calculations and kinetic isotope effect experiments
indicated that although the product adsorbed more strongly in
393
dx.doi.org/10.1021/op400217d | Org. Process Res. Dev. 2014, 18, 392−401

Organic Process Research & Development
Article
The total number of acidic sites (sites m−2) over each catalyst
was measured using the temperature-programmed desorption
of pyridine.19 This was performed by pretreating a sample of
catalyst (50 mg) at 250 °C for 2 h in air before exposure to
pyridine. The catalyst samples adsorbed with pyridine (15−20
mg) were heated to 600 °C, at a rate of 20 °C min−1 in dry N2
at a flow rate of 40 mL min−1. The mass loss due to desorption
of pyridine from the acidic sites was determined as a function of
the mass of the catalyst and then normalised to the surface area
of the catalyst.
X-ray diffraction patterns of the powdered catalysts were
obtained using Cu Kα radiation (PANalytical X’PERT PRO
MPD X-ray diffractometer).
2.3. Adsorption Isotherms. Adsorption of THCZ onto
untreated and sodium-hydroxide-washed 5 wt % Pd/Al2O3 was
measured by contacting samples of 5 wt % Pd/Al2O3 and 5 wt
% Pd/Al2O3(Na) (0.1 g) with solutions of THCZ in mesitylene
up to a concentration of 0.03 mol m−3. The mixtures were
stirred at room temperature overnight before filtration to
remove the catalyst with analysis of the final concentration of
THCZ by GC analysis.
Figure 1. Schematic of the Trickle Bed Reactor.
2.4. Catalyst Reuse. To study the deactivation of the
catalyst in the STR batch system, a single batch of catalyst was
reused a number of times. In this case, the catalyst was removed
from the postreaction mixture by filtration and subjected to one
of the following regeneration procedures. The catalyst was
washed with acetone (30 mL, which was sufficient to keep all
the catalyst immersed), washed in acetone overnight, washed in
acetone for 10 days with the acetone removed and changed
daily, and washed using a soxhlet extraction procedure again
with acetone or recalcined at 500 °C for 2 h.
2.4.1. Typical STR Batch Reaction Procedure. Batch
reactions were carried out in a stainless steel Parr Autoclave
reactor (300 mL capacity) containing a gasifying stirrer,
removable baffles, and a temperature-controlled heating jacket.
The gas was sparged into the vessel, and the pressure
monitored using a pressure gauge on the reactor. For a typical
reaction, THCZ (2 g), mesitylene (200 mL), and 5 wt % Pd/
Al2O3 (1 g) were placed in the reactor and a sample (1 mL)
was taken for analysis by GC. The reactor was sealed and
purged with 2 bar of nitrogen gas 5 times, before the reactor
was heated to the reaction temperature, typically 135 °C, with
moderate stirring. When at reaction temperature, the system
was pressurised with propene, typically to 4 bar, at which point
another sample was taken for analysis which corresponded to
time zero. The reaction was started immediately by starting the
impellor at typically 1200 r.p.m. Samples were taken every half
hour over the first three hours followed by every hour up to a 6
h reaction time.
The lines delivering liquid and gas reactants as well as the exit
line and receiver vessel were heated to reaction temperature
during the reaction. A pressure control valve (PCV in Figure 1)
facilitated the depressurization of the reactor exit flow, allowing
the liquid to be recycled to the receiver vessel and the gas
vented. The temperature, gas and liquid mass flow rates and
pressure were monitored on a control system.
Before reaction, the 5 wt % Pd/Al2O3, pellets were packed
into the fixed bed of 25.4 mm internal diameter with a 3 mm
diameter inert, spherical silica support, as shown in Figure S1 of
the Supporting Information, after which the reactor was
reconnected to the TBR system. The system was pressurised
to 1 bar gauge with nitrogen to ensure no leaks were present in
the system, and the catalyst bed was wetted for 1 h with the
liquid reaction mixture (300 mL) consisting of THCZ (3 g)
dissolved in mesitylene (300 mL) at a liquid flow rate of 196
mL min−1 and a gas flow rate of 100 mL min−1 at ambient
temperature and pressure. After wetting, the liquid was
discharged before the fresh reaction feed, consisting of
THCZ (4 g) dissolved in mesitylene (400 mL), was added.
The system was heated to the reaction temperature, typically
150 °C and pressurised with propene or nitrogen, typically to 1
bar, after the reaction temperature had been attained. Once the
reaction conditions were reached, the reaction was initiated by
starting the liquid flow at 167 mL min−1 (equivalent to a
superficial velocity of 5.5 mm s−1) and gas flow at 200 mL
min−1 (equivalent to a superficial velocity of 6.6 mm s−1).
Samples were taken over time for analysis by GC.
2.4.2. Typical TBR Batch and Continuous Reaction
Procedure. The TBR was operated as a cocurrent down flow
trickle bed reactor with a liquid recirculation and continuous
gas feed, shown in Figure 1.
The TBR was operated either under semibatch mode where
all the liquid flow was circulated back into the reactor by the
metering pump while the gas was vented or in continuous
mode, where a portion of the liquid feed was removed from the
reactor while 99.75% of the liquid was recycled. This protocol
has been described by the groups of Hickman20 and
Pollington21 to ensure complete wetting of the catalyst bed
and minimisation of external mass transfer, thereby allowing
reliable measurement of a rate of reaction. The inlet and outlet
flow of the liquid feed were both set to 1 mL min−1.
2.5. Analysis of Reaction Mixture. Analysis of the
reaction mixture was carried out using a GC system equipped
with an Agilent 19091J-433 HP-5 5% phenylmethylsiloxane
capillary column and FID detector. The concentration of
The system was designed to operate at a maximum
temperature of 450 °C and a maximum pressure of 40 bar.
The liquid reaction mixture was placed into a receiver vessel of
a 500 mL capacity and pumped around a closed loop using a
Bran + Luebe H2−31 Metering Pump (0−80 L h1−, discharge
pressure 40 bar). In the case of continuous operation, fresh feed
was pumped into the receiver tank while an equal volume was
extracted from the system. The liquid was introduced to the
fixed bed using a spray nozzle, while the gas was introduced
using a Brooks 5850 TR mass flow controller (0−2 L min−1).
394
dx.doi.org/10.1021/op400217d | Org. Process Res. Dev. 2014, 18, 392−401

Organic Process Research & Development
Article
THCZ and CZ were obtained by comparison with calibration
curves prepared using pure samples. The conversion to
carbazole in the reactions carried out in the batch STR and
semibatch TBR was calculated according to eq 1, where Ccz is
the concentration at any time after time zero and CTHCZ0 is the
concentration of THCZ at time zero.
75 μm shows that the lower surface area of the crushed pellets,
82 m2 g−1 compared to 143 m2 g−1 of the powder, as well as the
larger particle size of the pelletized catalyst results in a
decreased initial rate of reaction from 1.79 × 10−4 mol min−1
−1
−1
gcat to 6.94 × 10−5 mol min−1 gcat (Figure 3). Previous
CCZ
CTHCZ0
conversion of THCZ (%) =
× 100%
(1)
3. RESULTS AND DISCUSSION
3.1. Kinetic Reactions Carried out in the Batch STR.
3.1.1. General Kinetic Data. The conversion of THCZ and the
corresponding formation of CZ during a typical reaction are
shown in Figure 2, with complete conversion of THCZ
Figure 3. Graph of the formation of CZ with time using fresh
■
△
powdered catalyst ( ) and pellets crushed to less than 75 μm ( ).
Reaction conditions: in each case, the reaction was carried out over 5
wt % Pd/Al2O3 (1 g) with THCZ (2 g) in mesitylene (200 mL) at 135
°C and 4 bar propene.
studies on this system showed the rate of reaction to depend
strongly on the catalyst particle size. As was reported previously
when the catalyst was crushed and sieved to give three different
particle size ranges, the resulting reaction rates decreased with
increasing particle size and followed a linear relationship with
the ln (average particle diameter),16 indicating that the reaction
was under an internal (pore) diffusion limitation. The decrease
in reaction rate using the crushed pellets shown here is
attributed to a combination of increased internal diffusion and
the decreased dispersion of the palladium 6.4% on the eggshell
pellets compared to 22.4% on the powder. Given this result,
internal mass transfer limitation is expected to be present in the
3 mm pellets used within the TBR.
3.1.3. Catalyst Recycle. For a liquid phase process to be
carried out under continuous operation, the catalyst system
should be robust for extended periods of time so the reusability
of the crushed 5 wt % Pd/Al2O3 pellet catalyst was tested using
the batch STR. Figure 4 shows that on filtration and rinsing of
the catalyst after reaction and its subsequent reuse, the rate of
reaction decreased almost 8-fold from 6.94 × 10−5 mol min−1
gcat−1 to 9.04 × 10−6 mol min−1 gcat−1. Washing the catalyst with
acetone overnight restored some of the activity with the rate
increasing again to 1.87 × 10−5 mol min−1 gcat−1 and the initial
activity being fully obtained after 10 such treatments with
acetone. As this method of regenerating the activity of the
catalyst is not only time-consuming but also uses a lot of
solvent, Soxhlet extraction was used to regenerate the initial
activity of a used catalyst and, as shown in Figure 4, the initial
activity of the catalyst was regained.
Figure 2. Typical reaction profile showing the dehydrogenation of
■
△
THCZ ( ) to CZ ( ). Reaction conditions: the reaction was carried
out over 5 wt % Pd/Al2O3 powder (1 g) with THCZ (2 g) in
mesitylene (200 mL) at 135 °C and 4 bar propene.
occurring after about 90 min with a selectivity to CZ of greater
than 99%. Although the mass balance was low at the start of the
reaction, this was recovered as the reaction proceeded with a
mass balance of 97% at 98% conversion. The loss of THCZ
concentration at time zero is therefore due to a combination of
its adsorption onto the catalyst surface and some reaction
occurring as the temperature was increased from room
temperature to the reaction temperature of 135 °C.
Fitting of reaction equations to the data showed the reaction
to be initially close to zero order with respect to THCZ and
then positive order as the conversion is increased (Figure S2 of
the Supporting Information). This was supported by just over a
2-fold increase in initial rate from 1.79 × 10−4 mol min−1 gcat
−1
to 4.78 × 10−4 mol min−1gcat−1 when the THCZ concentration
was increased by 5-fold from 0.058 to 0.29 M. This is in
agreement with kinetic data reported previously,16 which also
describes the optimisation of this system in terms of the solvent
and hydrogen acceptor. To gain a better understanding of the
system before moving from batch to continuous operation, the
effects of the catalyst form and potential deactivation was
studied using the batch STR.
These results indicate that there is a heavy, but acetone
soluble byproduct produced, which deactivates the system.
Interestingly, the initial rate of reaction was not regained on
recalcination of a used catalyst at 500 °C due to either sintering
of the palladium particles or more likely formation of more
3.1.2. Catalyst Form. Comparison of the reaction profile
over the 5 wt % Pd/Al2O3 powdered catalyst of particle size 20
μm with a 5 wt % Pd/Al2O3 pellet catalyst crushed to less than
395
dx.doi.org/10.1021/op400217d | Org. Process Res. Dev. 2014, 18, 392−401

Organic Process Research & Development
Article
Figure 4. Graph of the formation of CZ with time using fresh crushed
Figure 5. Graph of the formation of CZ with time using fresh 5 wt %
Pd/C ( ) and used 5 wt % Pd/C catalyst washed with acetone
■
■
catalyst pellet ( ) and used catalyst subjected to different regeneration
●
procedures. No wash ( ), 12% of initial activity; used catalyst washed
●
△
overnight ( ) and fresh 5 wt % Pd/α Al2O3 ( ) and used Pd/alpha
Al2O3 catalyst washed with acetone overnight (×). Reaction
conditions: the reactions were carried out over catalyst (1 g) with
THCZ (2 g) in mesitylene (200 mL) at 135 °C and 4 bar propene.
△
with acetone overnight ( ), 25% of initial activity; used catalyst
washed with acetone for 10 days (×), 100% of initial activity; used
▲
catalyst washed with acetone using Soxhlet extraction ( ); and the
used catalyst recalcined (+). Reaction conditions: the reactions were
carried out over 5 wt % Pd/Al2O3 (1 g) with THCZ (2 g) in
mesitylene (200 mL) at 135 °C and 4 bar of propene.
pelletised 5 wt % Pd/Al2O3 was most suitable for use in the
TBR, so potential methods to modify the commercial catalyst
were investigated.
The presence of acidic sites on the commercial alumina were
considered to be potential sites for byproduct formation. To
remove these active support sites, the commercial 5 wt % Pd/
Al2O3 catalyst pellets were washed with a range of bases:
lithium, sodium, and potassium hydroxide.23 These treated
catalysts, 5 wt % Pd/Al2O3(Li, Na, or K), were subsequently
used in a standard dehydrogenation reaction, and Figure 6
shows that in all cases, the rate of reaction increased compared
strongly adsorbed coking products, which in turn were more
severely deactivating.
It should be noted that the deactivation was more severe
when using pellets compared to powder where 43% of the
initial activity was retained on the second reuse of the catalyst
with no washing (Figure S3 of the Supporting Information).
Therefore, the build up of byproducts in the pores of the
catalyst caused by poor mass transfer or metal dispersion may
play an important role in the production of this poison. Overall,
these results show that the catalyst is deactivated by strong
adsorption of a poison, formed during reaction, on the catalyst
surface and provided the poison can be removed, the activity of
the catalyst can be regenerated. A similar observation was made
by the group of Libuda during the dehydrogenation of
dodecahydro-N-ethylcarbazole above 350 K.14 Strongly bound
species were observed on the surface of the catalyst, some of
which were identified as dehydrogenation intermediate
products or decomposition products whilst others remained
unidentified. Attempts were made in this work to isolate and
identify the byproduct by Soxhlet extraction of a spent catalyst;
however, all efforts to observe material other than those present
in the reaction by GC, HPLC, or NMR were unsuccessful. With
typical mass balances throughout the reaction being in excess of
95% and no unidentified peaks being observed by GC analysis
of the liquid phase, the byproduct formed was considered to be
present in small amounts, although strongly adsorbed to the
catalyst surface.
to the untreated catalyst from 6.94 × 10−5 mol min−1 gcat
−1
using 5 wt % Pd/Al2O3 to 1.61 × 10−4 mol min−1 gcat−1 using 5
wt % Pd/Al2O3(Na).
3.2. Catalyst Modification. Secondary reactions leading to
the formation of an unidentified product may be attributed to
active sites on the alumina support itself. This hypothesis was
supported by the efficient recycle of 5 wt % Pd/C and 5 wt %
Pd/α-Al2O3 catalysts (Figure 5). Alpha alumina has been
shown to have lower acidity than the gamma form,22 and this
was supported here where the density of acid sites on the alpha
alumina catalyst was found to be negligible under pyridine
adsorption analysis. However, the commercially available
Figure 6. Graph of the formation of CZ with time using freshly
■
crushed pellet catalyst ( ) and catalyst washed with different bases:
○
lithium hydroxide, 5 wt % Pd/Al2O3(Li) ( ), potassium hydroxide 5
wt % Pd/Al2O3(K) (×), and sodium hydroxide 5 wt % Pd/Al2O3(Na)
▲
(
). Reaction conditions: in each case the reaction was carried out
over 5 wt % Pd/Al2O3 or 5 wt % Pd/Al2O3(Li, Na, or K) (1 g) with
THCZ (2 g) in mesitylene (200 mL) at 135 °C and 4 bar of propene.
396
dx.doi.org/10.1021/op400217d | Org. Process Res. Dev. 2014, 18, 392−401

Organic Process Research & Development
Article
Characterisation of 5 wt % Pd/Al2O3(Na) by XRD analysis
(Figure 7) showed that although the support remained intact
Figure 8. Adsorption isotherms of THCZ adsorbed onto the surface of
■
5 wt % Pd/Al2O3 crushed pellets ( ) and onto the surface of 5 wt %
△
Pd/Al2O3(Na) ( ).
Figure 7. XRD patterns of (a) 5 wt % Pd/Al2O3, (b) used 5 wt % Pd/
Al2O3, (c) 5 wt % Pd/Al2O3(Na), and (d) used 5 wt % Pd/Al2O3(Na).
The increase in the rate observed with the base-washed
catalysts may be due to the decrease in acid sites resulting in
fewer poisons formed on the surface of the catalyst during the
reaction, or the presence of sodium reducing the adsorption
strength of the organic compounds, or a combination of both of
these effects. Although at present the increased rate cannot be
attributed to one reason, the kinetic results and characterisation
indicate that the nature of the sodium-hydroxide-washed
catalyst may be suitably modified to decrease the deactivation
and so a recycle study was carried out.
The 5 wt % Pd/Al2O3(Na) catalyst was recycled by removing
it from the reactor, washing it with acetone, and drying it
overnight at 120 °C. Figure 9 shows a progressive decrease in
reaction rate on the second and third run; however, the
reaction rate of 5 wt % Pd/Al2O3(Na) used for the third time
was similar to that of 5 wt % Pd/Al2O3 used for the second
time. There was no change observed in the structure of 5 wt %
after base treatment, some changes did occur with the
disappearance of the peak at 2 theta = 31.5 degrees, attributed
to theta alumina, after treatment. Identification of the other
XRD peaks are given in Table S1 of the Supporting
Information and show the presence of gamma, theta, and
delta alumina in the support. In addition, the surface area of the
catalyst increased after sodium hydroxide treatment (Table 1).
The density of acid sites on the catalyst before and after
treatment was quantified using pyridine-TPD. The results,
summarised in Table S2 of the Supporting Information, show a
greater than 20% decrease in acid sites on washing with sodium
hydroxide, signifying a decrease in sites that could be
responsible for formation of the poison, resulting in a
correspondingly faster rate of reaction.
The structure and acidity of alumina is often unclear with
electronic and structural properties of the support and metal
changing with the extent of hydration, thermal treatments, and
exposure to acidic or basic media. Therefore, the decrease in
acid sites may be just part of the explanation for the increased
rate of reaction observed on the base-washed catalysts.
Adsorption of THCZ on palladium is known to occur through
the aromatic ring with all intermediates and the carbazole
product retaining this interaction throughout the dehydrogen-
ation process.17b The presence of an electron donating sodium
on the surface of the catalyst may alter the adsorption strength
of the organic substrates, intermediates, and products, thereby
decreasing the deactivating effect of strongly bound poisoning
byproducts. Here, THCZ was taken as a representative model
compound present in this reaction and an adsorption isotherm
on 5 wt % Pd/Al2O3 was compared with that for 5 wt % Pd/
Al2O3(Na). Figure 8 shows that for concentrations below
saturation of the catalyst surface, the concentration of THCZ
adsorbed per unit area of catalyst surface is less for the sodium-
hydroxide-washed catalyst than the untreated catalyst.
Figure 9. Graph of the initial rate of reaction for the formation of CZ
over a number of uses using 5 wt % Pd/Al2O3 (blacked out bar) and 5
wt % Pd/Al2O3(Na) (striped bar). Reaction conditions: the reactions
were carried out over 5 wt % Pd/Al2O3 or 5 wt % Pd/Al2O3(Na) (1 g)
with THCZ (2 g) in mesitylene (200 mL) at 135 °C and 4 bar of
propene.
Weaker adsorption of the organic substrate, intermediate, or
product will decrease the occurrence of secondary reactions
forming the poison and also decrease the adsorption strength of
any poison molecules formed.
397
dx.doi.org/10.1021/op400217d | Org. Process Res. Dev. 2014, 18, 392−401

Organic Process Research & Development
Article
Pd/Al2O3 or 5 wt % Pd/Al2O3(Na) on recycle, as shown by
XRD analysis (Figure 7).
Sodium hydroxide on alumina supports is easily removed in
the presence of polar solvents such as acetone,24 so to
overcome the effect of the acetone wash on the sodium-
hydroxide-washed catalyst, the recycle experiment was repeated
but this time included a secondary washing step with sodium
hydroxide after each acetone wash. The results shown in Figure
10 demonstrate that under these conditions the initial activity
■
Figure 11. Graph showing the dehydrogenation of THCZ ( ) and
○
the formation of CZ ( ) with time in the trickle bed reactor operated
in the semibatch mode. Reaction conditions: the reaction was carried
out over 5 wt % Pd/Al2O3 pellets (42 g) with THCZ (4 g) in
mesitylene (400 mL) at 150 °C and 1 bar propene, using a liquid flow
rate of 167 mL min−1 and a gas flow rate of 200 mL min−1.
obtain the apparent activation energy of the reaction. Figure S7
of the Supporting Information shows that the rate of reaction
increases with increasing temperature and that the reaction has
an apparent activation energy of 27.2 kJ mol−1 under these
conditions, significantly lower than 71 kJ mol−1 obtained using
crushed pellets in a STR.16 One outcome of the reduced rate of
reaction is that the build-up of hydrogen in the pores of the
catalyst is less of a problem than in the batch STR studies,
meaning the presence of a hydrogen-accepting gas has less of
an effect on the rate of reaction. Figure 12 summarises the
Figure 10. Graph of the formation of CZ with time using sodium-
■
△
washed catalyst for three uses: first use ( ), second use ( ), and third
use (×). Reaction conditions: the reactions were carried out over 5 wt
% Pd/Al2O3(Na) (1 g) with THCZ (2 g) in mesitylene (200 mL) at
135 °C and 4 bar of propene. The used catalyst was filtered from the
reaction mixture, washed with acetone followed by a secondary
washing with sodium hydroxide before being dried, and used in
subsequent runs.
of the catalyst was retained even after 3 runs. This stabilised
catalyst was considered suitable for continuous testing.
3.3. Kinetic Reactions in the Trickle Bed Reactor.
3.3.1. Evaluation of the Trickle Bed Reactor in Semibatch
Mode. To ensure the data obtained in the batch STR could be
applied to the reaction carried out continuously, a series of
experiments were carried out with the trickle bed reactor
operated in the semibatch mode, where the gas was
continuously fed through the system and the liquid feed was
recycled using the metering pump. Figure 11 shows a typical
reaction profile for a semibatch reaction with Figure S4 of the
Supporting Information showing that the order changes from
zero order to first order after around 7 hours reaction time. The
rate of reaction in the trickle bed reactor was over 70 times
slower than the STR with the rate in the TBR being 8.3 × 10−7
mol min−1 gcat compared to 6.94 × 10−5 mol min−1 gcat
−1
−1
obtained when the crushed pellets were tested in the STR.
Similarly to the STR, no other products were observed in the
TBR (Figure S5 of the Supporting Information).
Figure 12. Graph showing the formation of CZ with time in the trickle
As was previously reported for the STR,16 the reaction in the
TBR was found to be zero order with respect to propene
(Figure S6 of the Supporting Information). It is known that the
rate of reaction is strongly dependent on the size of the catalyst
particles used and here the use of the 3 mm catalyst pellets
required for the fixed bed was expected to result in mass
transfer limitations. This was confirmed by running the reaction
at different temperatures and using the Arrhenius Equation to
bed reactor operated in semibatch mode over 5 wt % Pd/Al2O3 pellets
■
with an N2 purge gas ( ), 5 wt % Pd/Al2O3 pellets with a propene
purge gas (×), 5 wt % Pd/Al2O3(Na) pellets with an N2 purge gas
○
▲
( ), and 5 wt % Pd/Al2O3(Na) pellets with a propene purge gas ( ).
Reaction conditions: in each case, the reaction was carried out over 5
wt % Pd/Al2O3 or 5 wt % Pd/Al2O3(Na) pellets (43 g) with THCZ (4
g) in mesitylene (400 mL) at 150 °C and 1 bar of propene or N2,
using a liquid flow rate of 167 mL min−1 and a gas flow rate of 200 mL
min−1.
398
dx.doi.org/10.1021/op400217d | Org. Process Res. Dev. 2014, 18, 392−401

Organic Process Research & Development
Article
effect of base washing and propene on the rate of reaction in
the TBR. It was previously reported that in the STR, a 5-fold
increase in rate could be obtained upon changing from a
nitrogen purge to remove the hydrogen to a 5 bar pressure of
propene.16 When the reaction was run in the semibatch TBR
system, only a 1.1-fold increase in rate was observed with the
rate increasing from 6.45 × 10−7 mol min−1 gcat−1, using 1 bar
of nitrogen pressure, to 7.32 × 10−7 mol min−1 gcat−1, using the
same pressure of propene. Here the propene is less significant
due to the low rate of reaction; the kinetics are controlled to a
greater extent by the reaction rather than the transport of
hydrogen from the pores.
The batch STR system showed that the sodium hydroxide
wash resulted in a 2.3-fold increase in rate, and a similar 1.9-fold
increase was observed in the semibatch TBR in the presence of
nitrogen. Here the faster rate of reaction causes a build up of
hydrogen within the pores of the catalyst, and the kinetics
become increasingly controlled by the transport of this
hydrogen out of the pores. In this case, the effect of propene
was more pronounced with a further 1.8-fold increase in rate
observed from 1.20 × 10−6 mol min−1 gcat with nitrogen to
−1
−1
2.17 × 10−6 mol min−1 gcat with propene.
It is clear, therefore, that a comparison of the rate of the
untreated catalyst in nitrogen with that observed using a
sodium-washed catalyst in the presence of propene should
result in the greatest increase in rate, and a 3-fold increase was
observed. In summary, the least active system was the untreated
catalyst used in the presence of nitrogen and by applying a
simple base wash and using a hydrogen acceptor gas, the rate of
reaction was increased by over 3-fold.
3.3.2. Continuous Dehydrogenation of THCZ. With the
results obtained in the batch STR generally translating to the
TBR and the recycle of the base-washed catalyst demonstrated
in the batch STR, the reaction was run under continuous
operation in the TBR. Here the deactivation of the untreated
catalyst and the catalyst washed with sodium hydroxide could
be compared. Conditions used for continuous operation were
similar to those used for the semibatch experiments with the
gas flow at 200 mL min−1, and the liquid flow set to 167 mL
min−1. The liquid inlet and outlet were 1 mL min−1, giving a
recycle ratio of 99.75%.
Figure 13. Graph showing (a) the formation of CZ with time in the
trickle bed reactor operated in continuous mode using a propene
purge over the untreated 5 wt % Pd/Al2O3 catalyst and (b) the change
in the rate of reaction with time on stream. Reaction conditions: the
reaction was carried out over 5 wt % Pd/Al2O3 pellets (43 g) with
THCZ (4 g) in mesitylene (400 mL) at 150 °C and 3 bar of propene,
using a liquid flow rate of 167 mL min−1 and a gas flow rate of 200 mL
min−1.
Figure 13 shows that, as expected, when using the untreated
catalyst, an initially rapid rate of reaction was observed followed
by a continuous deactivation, until the reaction ceased after 10
h on stream.
When the sodium-hydroxide-washed catalyst was used in the
TBR under continuous operation, again an initially rapid rate of
reaction was observed, followed by a slight decrease in
conversion to 22 mol m−3 (37%) (see Figure 14). However,
in this case, the conversion then leveled off to show a region of
steady state which lasted for 50 h of reaction time, at which
point the reaction was stopped.
fewer byproducts being formed and decreased adsorption
strength of organic reagents. In addition, use of a continuous
flow reactor has highlighted aspects of deactivation of this
catalyst that were not evident using a batch STR. By testing
deactivation of the catalyst in the batch STR, the results were
influenced by the effects of the regeneration treatments on the
catalyst between runs rather than just deactivation occurring
during the reaction. It is proposed that a small amount of as-yet
unidentified material forms over the untreated catalyst and
adsorbs strongly onto the catalyst surface, reducing the rate of
reaction and eventually poisoning the catalyst. However, a
simple wash with a base is sufficient to modify the catalyst
properties in such a way as to increase the rate of reaction and
limit formation and adsorption of the poison, allowing the
dehydrogenation reaction under continuous operation.
Although not the primary motivation for this work, it is
noted that the improved rate of reaction and reuse observed
with the base-washed catalyst makes this an interesting catalyst
for further investigation for a hydrogen storage application.
4. CONCLUSIONS
The ability of the sodium-hydroxide-washed catalyst to operate
under continuous operation demonstrates the viability of
continuous dehydrogenation reactions using a TBR. Treatment
of the catalyst with sodium hydroxide resulted in a catalyst with
increased activity shown by the increased rate of reaction
observed in the STR and increased stability shown by the
region of steady state observed in the TBR compared to an
untreated catalyst. The increased activity has been attributed to
a combination of the lower acidity of the support, resulting in
399
dx.doi.org/10.1021/op400217d | Org. Process Res. Dev. 2014, 18, 392−401

Organic Process Research & Development
Article
ACKNOWLEDGMENTS
■
The authors would like to acknowledge financial support from
the EPSRC and Johnson Matthey under the Castech project
(EP/G012156/1). A.M. would like to acknowledge studentship
funding from The Ministry of Higher Education of Malaysia
and The University of Malaya. The authors thank Mr Adam
Elliott for assistance in preparation of the 5 wt % Pd/α-Al2O3
catalyst.
REFERENCES
■
(1) Rode, C. V.; Ghalwadkar, A. A.; Mane, R. B.; Hengne, A. M.;
Jadkar, S. T.; Biradar, N. S. Org. Process Res. Dev. 2010, 14, 1385−1392.
(2) Mane, R. B.; Rode, C. V. Org. Process Res. Dev. 2012, 16, 1043−
1052.
(3) Calabrese, G. S.; Pissavini, S. AIChE J. 2011, 57, 828−834.
(4) Anderson, N. G. Org. Process Res. Dev. 2012, 16, 852−869.
(5) Stitt, E. H.; Rooney, D. W. Switching from batch to continuous
processing for fine and intermediate-scale chemicals manufacture. In
Novel Concepts in Catalysis and Chemical Reactors; Wiley-VCH:
Weinheim, Germany, 2010; p 2010.
(6) Ashton, B. W.; Suschitzky, H. J. Chem. Soc. 1957, 4559−4562.
(7) (a) Borsche, W.; Witte, A.; Bothe, W. Justus Liebigs Ann. Chem.
1908, 359, 49−80. (b) Drechsel, E. Journal fur Praktische Chemie 1888,
̈
38, 65−74.
(8) (a) Horning, E. C.; Horning, M. G.; Walker, G. N. J. Am. Chem.
Soc. 1948, 70, 3935−3936. (b) Zhao, L.; Sotoodeh, F.; Smith, K. J.
Catal. Commun. 2010, 11, 391−395. (c) Robinson, R. Chem. Rev.
1963, 63, 373−401.
(9) (a) Liu, L.; Wong, W.-Y.; Shi, J.-X.; Cheah, K.-W.; Lee, T.-H.;
Leung, L. M. J. Organomet. Chem. 2006, 691, 4028−4041.
(b) Pisharady, S. K.; Menon, C. S.; Kumar, C. S.; Gopinathan, T. G.
Mater. Chem. Phys. 2006, 100, 147−151.
(10) Barbieri, V.; Ferlin, M. G. Tetrahedron Lett. 2006, 47, 8289−
8292. (b) Pelaprat, D.; Oberlin, R.; Le Guen, I.; Roques, B.-P.; Le
Pecq, J. B. J. Med. Chem. 1980, 23, 1330−1335.
(11) (a) Knolker, H.-J. Top. Curr. Chem. 2005, 244, 115−148.
̈
Figure 14. Graph showing (a) the formation of CZ with time in the
trickle bed reactor operated in continuous mode, using a propene
purge over the sodium-hydroxide-washed 5 wt % Pd/Al2O3 catalyst
and (b) the change in the rate of reaction with time on stream.
Reaction conditions: the reaction was carried out over 5 wt % Pd/
Al2O3 pellets (43 g) with THCZ (4 g) in mesitylene (400 mL) at 150
°C and 1 bar of propene, using a liquid flow rate of 167 mL min−1 and
a gas flow rate of 200 mL min−1.
(b) Knolker, H.-J.; Reddy, K. R. Chem. Rev. 2002, 102, 4303−4427.
̈
(c) Kitawaki, T.; Hayashi, Y.; Ueno, A.; Chida, N. Tetrahedron 2006,
62, 6792−6801.
(12) Sotoodeh, F.; Smith, K. J. Ind. Eng. Chem. Res. 2010, 49, 1018−
1026.
(13) (a) Sotoodeh, F.; Huber, B. J. M.; Smith, K. J. Int. J. Hydrogen
Energy 2012, 37, 2715−2722. (b) Sotoodeh, F.; Smith, K. J. J. Phys.
Chem. C 2013, 117, 194−204.
(14) (a) Sobota, M.; Nikiforidis, I.; Amende, M.; Sanmartín Zanon
́
,
B.; Staudt, T.; Hofert, O.; Lykhach, Y.; Papp, C.; Hieringer, W.;
̈
Laurin, M.; Assenbaum, D.; Wasserscheid, P.; Steinruck, H.-P.;
̈
Gorling, A.; Libuda, J. Chem.Eur. J. 2011, 17, 11542−11552.
̈
ASSOCIATED CONTENT
* Supporting Information
■
(b) Amende, M.; Schernich, S.; Sobota, M.; Nikiforidis, I.; Hieringer,
W.; Assenbaum, D.; Gleichweit, C.; Drescher, H.-J.; Papp, C.;
S
Identification of XRD peaks, density of acid sites, figure of fixed
catalyst bed, curve fitting of rate equations, graph of CZ
formation, graphs of comparison in CZ formation with time.
This material is available free of charge via the Internet at
http://pubs.acs.org.
Steinruck, H.-P.; Gorling, A.; Wasserscheid, P.; Laurin, M.; Libuda,
̈
̈
J. Chem.Eur. J. 2013, 19, 10854−10865.
(15) Sotoodeh, F.; Huber, B. J. M.; Smith, K. J. Appl. Catal., A 2012,
419−420, 67−72.
(16) Hindle, K. T.; Burch, R.; Crawford, P.; Hardacre, C.; Hu, P.;
Kalirai, B.; Rooney, D. W. J. Catal. 2007, 251, 338−344.
(17) (a) Crawford, P.; Burch, R.; Hardacre, C.; Hindle, K. T.; Hu, P.;
Kalirai, B.; Rooney, D. W. J. Phys. Chem. C 2007, 111, 6434−6439.
(b) Crawford, P.; Burch, R.; Hardacre, C.; Hindle, K. T.; Hu, P.;
Rooney, D. W. J. Chem. Phys. 2008, 128, 105104-1−105104−6.
(18) Ito, K.; Tomino, T.; Ohshima, M.; Kurokawa, H.; Sugiyama, K.;
Miura, H. Appl. Catal., A 2003, 249, 19−26.
AUTHOR INFORMATION
Corresponding Authors
*E-mail: jillian.thompson@qub.ac.uk.
*E-mail: d.rooney@qub.ac.uk.
■
Present Addresses
∥Cain Department of Chemical Engineering, Louisianna State
University, Baton Rouge, Louisiana 70803.
⊥University of Malaya, 50603 Kuala Lumpur, Malaysia.
(19) Mekhemer, G. A. H.; Halawy, S. A.; Mohamed, M. A.; Zaki, M.
I. J. Phys. Chem. B. 2004, 108, 13379−13386.
(20) Hickman, D. A.; Weidenbach, M.; Friedhoff, D. P. Chem. Eng.
Sci. 2004, 59, 5425−5430.
Notes
(21) Enache, D. I.; Landon, P.; Lok, C. M.; Pollington, S. D.; Stitt, E.
H. Ind. Eng. Chem. Res. 2005, 44, 9431−9439.
The authors declare no competing financial interest.
400
dx.doi.org/10.1021/op400217d | Org. Process Res. Dev. 2014, 18, 392−401

Organic Process Research & Development
Article
(22) Gao, X.; Zhao, Y.; Wang, S.; Yin, Y.; Wang, B.; Ma, X. Chem.
Eng. Sci. 2011, 66, 3513−3522.
(23) Pekridis, G.; Kaklidis, N.; Konsolakis, M.; Iliopoulou, E. F.;
Yentekakis, I. V.; Marnellos, G. E. Top. Catal. 2011, 54, 1135−1142
and references therein.
(24) (a) Kim, M.; Yan, S.; Salley, S. O.; Ng, K. Y. S. Catal. Commun.
2009, 10, 1913−1919. (b) Arzamendi, G.; Campo, I.; Arguinarena, E.;
̃
San
130.
́
chez, M.; Montes, M.; Gandía, L. M. Chem. Eng. J. 2007, 134, 123−
401
dx.doi.org/10.1021/op400217d | Org. Process Res. Dev. 2014, 18, 392−401