Continuous production of the renewable ρ-cymene from α-pinene

Article abstract of DOI:10.1016/j.jcat.2013.08.007

The aim of this work was to demonstrate the feasibility to produce ρ-cymene, an important commodity chemical, in a continuous, one-pot reaction system from abundant α-pinene, available e.g. as a by-product of pulping industry. The isomerization reactions of α-pinene over bimetallic heterogeneous catalysts, 3 and 5 wt% Pd-Zn (1:1, 1:4, 4:1, 1:0, and 0:1), supported on Al-SBA15 were studied. The principal reaction products were identified as ρ- and m-cymenes, limonene, camphene, and ρ-menthene, respectively. The highest concentration of ρ-cymene reached 77 wt% under the optimized reaction conditions: 300 C and α-pinene feed of 0.03 mL/min. Two main reaction pathways toward ρ- and m-cymenes were described, and a mechanistic kinetic model, based on a plausible reaction network in line with Langmuir-Hinshelwood approach, was developed. The catalyst characterization revealed the reduction in Pd(II) sites, catalyst coking, and decline of surface area over the course of time. The catalyst recovery and reuse was addressed.

Full text of DOI:10.1016/j.jcat.2013.08.007

Journal of Catalysis 307 (2013) 305–315
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Continuous production of the renewable
q
-cymene from
a-pinene
a
b
c
d
a,c,
M. Golets a, , S. Ajaikumar , M. Mohln , J. Wärnå , S. Rakesh , J.-P. Mikkola
⇑
⇑
a Technical Chemistry, Department of Chemistry, Chemical-Biological Centre, Umeå University, SE-90187 Umeå, Sweden
b Microelectronics and Materials Physics Laboratories, EMPART Research Group of InfoTech Oulu, University of Oulu, P.O.BOX 4500, FI-90014 Oulu, Finland
c Laboratory of Industrial Chemistry and Reaction Engineering, Process Chemistry Centre, Åbo Akademi University, Biskopsgatan 8, FI-20500 Åbo-Turku, Finland
d Department of Chemistry, Podhigai College of Engineering and Technology, Tirupattur 635601, Tamilnadu, India
a r t i c l e i n f o
a b s t r a c t
Article history:
The aim of this work was to demonstrate the feasibility to produce
q
-cymene, an important commodity
-pinene, available e.g. as a by-product
-pinene over bimetallic heterogeneous catalysts, 3 and
5 wt% Pd–Zn (1:1, 1:4, 4:1, 1:0, and 0:1), supported on Al-SBA15 were studied. The principal reaction prod-
ucts were identified as - and m-cymenes, limonene, camphene, and -menthene, respectively. The high-
est concentration of -cymene reached 77 wt% under the optimized reaction conditions: 300 °C and
pinene feed of 0.03 mL/min. Two main reaction pathways toward - and m-cymenes were described,
Received 30 April 2013
Revised 6 August 2013
Accepted 11 August 2013
chemical, in a continuous, one-pot reaction system from abundant
of pulping industry. The isomerization reactions of
a
a
q
q
q
a-
Keywords:
q
q-Cymene
and a mechanistic kinetic model, based on a plausible reaction network in line with Langmuir–Hinshel-
wood approach, was developed. The catalyst characterization revealed the reduction in Pd(II) sites, catalyst
coking, and decline of surface area over the course of time. The catalyst recovery and reuse was addressed.
Ó 2013 Elsevier Inc. All rights reserved.
a-Pinene isomerization
Heterogeneous catalyst
Flow-through reactor
1. Introduction
natural, renewable feedstocks [6,7]. On the other hand, the applica-
tion of liquid mineral acids and soluble Lewis acid catalysts is dis-
The possibility of application for
q
-cymene as a solvent in dyes
advantageous due to problems associated with corrosion, handling,
and safety as well as due to the difficulties encountered upon sep-
aration and reutilization of the products and the catalysts [7]. Evi-
and varnishes, as an additive for flavors and fragrances, and as an
industrial odorant and raw material for chemical synthesis demon-
strates the high importance of this product for chemical industry
[1]. Moreover, prospectively terephthalic acid could be produced
dently, it would be elegant to design a process leading to
that utilizes renewable raw materials and, at the same time, is a
simple one-pot process. In fact, -pinene and related terpenes
q-cymene
from
converted to
ation of toluene with propylene followed by oxygen mediated
hydroperoxide cleavage. The further treatment of -cresol leads
to 2,6-di-tert-butyl- -cresol – a well-known antioxidant [1,3]. Fur-
ther, -cymene could be converted to isopropyl phenols that are
q
-cymene via oxidation [2]. Alternatively,
q-cymene can be
a
q-cresol, which is conventionally produced via alkyl-
are easily available from the Nordic pulping industry and could
alternatively be utilized for this purpose. Chemical pulping and
biorefining results in e.g. tall oil and turpentine fractions that pre-
dominantly contain a-pinene, along with other terpenes. Today,
these fractions are undervalued and to a large extent burned in
q
q
q
major intermediates in resins synthesis [4]. Non-nitrated musks
and polymers, such as synthetic fibers or antiseptics, are also pro-
duced from the aforementioned chemical [5].
the recovery boilers to supply energy.
Previously, straight dehydrogenation and dehydroisomerization
of
a
-pinene to
q
-cymene were studied by several groups [1,6,8].
-isomer, b-pinene was observed to
Conventionally,
q-cymene is produced via alkylation of toluene
Also, in comparison with the a
with propene or 2-propanol. Friedel – Crafts catalysts, for example,
HCl, H2SO4, AlCl3, or BF3 are commonly applied. In the second step,
possess similar reactivity [1,9,10]. Limonene conversion over sup-
ported Pd catalysts [2,11–16], zeolites [16,17], and Lewis acids
the obtained mixture of
q
, o-, and m-isomers is separated. How-
[18] to
promising results for a mixture of terpenes, including
limonene, terpinolene, and cineoles aiming at -cymene over a
q
-cymene has also been investigated. Buhl et al. reported
ever, nowadays, due to both environmental and sustainability
related reasons, the endeavor is that the share of petroleum-de-
rived chemicals should continuously diminish and be replaced by
a-pinene,
q
Pd/SiO2 catalyst. An analogous study of 3-carene was reported by
Krishnasamy [19].
The well-known mesoporous SBA-15 support is characterized
by its high surface area, relatively large-sized pores and high
stability. In an earlier study taking advantage of this versatile
material, Du et al. described the dehydrogenation of dipentene
⇑
Corresponding authors. Address: Technical Chemistry, Department of Chemis-
try, Chemical-Biological Centre, Umeå University, SE-90187 Umeå, Sweden (M.
Golets).
E-mail addresses: Mikhail.Golets@chem.umu.se (M. Golets), Jyri-Pekka.Mikko-
la@chem.umu.se (J.-P. Mikkola).
0021-9517/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcat.2013.08.007

306
M. Golets et al. / Journal of Catalysis 307 (2013) 305–315
over Al/SBA-15 and Zn/SBA-15, leading to
q-cymene as the main
with an HP-PONA capillary column (50 m, 0.2 mm I.D., 0.25 lL film
product. According to Du et al., the presence of Zn, in fact, mini-
mized the formation of cracking products due to the lack of
Brønsted acidity of the catalyst [20]. On the other hand, the pres-
ence of moderate Lewis acidity in Zn/SBA15 contributed to
thickness) and a Flame Ionization Detector (FID). Nitrogen was
used as the carrier gas (1.0 mL/min). The temperature program-
ming was as follows: the injection port and detector temperature
were set to 250 °C, column temperature was adjusted to rise from
70 up to 230 °C, with a holding time of 0.5 min at the initial tem-
perature, and a heating ramp of 10 °C/min was applied. The reac-
tant and product retention times (min) were evaluated as
87 wt% yield of
another study, Chinayon et al., observed upon hydrogenation of
acetylene over Pd/ -Al2O3 that a significant improvement of acet-
q-cymene according to their results [20]. In a yet
a
ylene conversion and ethylene selectivity could be obtained when
the Pd sites were modified with Zn. In particular, the authors
follows:
-menthene (9.72), m-cymene (10.45),
nene (10.75), and -terpinene (11.45) (Fig. 1). To evaluate the pos-
sible formation of dimers [22], a prolonged GC analysis method
was tried (30 min) at an elevated temperature of 320 °C. However,
no additional products were detected in the GC chromatograms
upon extended run times. For a detailed product analysis and iden-
tification, GCMS analysis (Gas Chromatography coupled to Mass
Spectrometry; Thermo Trace DSQ) was applied (column VF-5MS,
a
-pinene (8.63), camphene (8.90), trans-pinane (9.50),
q
q
-cymene (10.68), limo-
claimed that less coking could be observed for Pd–Zn/
for Pd/ -Al2O3 [21].
In this study, we investigated the possibility to design a one-
stage continuous process for efficient conversion of -pinene to
-cymene over Pd–Zn/Al-SBA15 catalysts with various ratios and
a-Al2O3 than
c
a
a
q
loadings of Pd and Zn metals. The structure of the catalysts, char-
acterization results, as well as the optimized reaction conditions
was discussed. Still, a kinetic model was proposed that accounts
for the described reaction mechanism.
30 m., 0.25 mm I.D., phase 0.25 lm). Constant nitrogen flow rate
of 1.5 mL/min was utilized. The temperature of the injection and
initial oven temperatures were 230 °C and 50 °C, respectively.
The maximum temperature applied was 300 °C (ramping 20 °C/
min), after one (1) minute of hold time. A representative sample
GC chromatogram with specified retention times is presented in
Fig. 1.
2. Experimental
2.1. Materials
The concentrations were expressed in terms of wt% in reference
The hydrogen gas used was purchased from a commercial
source (AGA AB) and had a purity of 99.999%. All the catalysts were
in-house prepared and characterized at Technical Chemistry,
Department of Chemistry, Umeå University. High purity
to the whole amount of the reacted
a-pinene and other reactants
in the product mixture. Prior to this, each and every standard spe-
cies was calibrated in order to identify and quantify the reaction
products. In addition,
q-cymene was distinguished from m-cym-
(P95 wt%) standards of camphene, limonene,
q-cymene, q-
ene with the help of earlier observations [1]. The following formu-
las were used for the calculation of the conversion, product
selectivities, and yields, respectively:
menthene, -pinene, and -terpinene were obtained from
a
c
Sigma–Aldrich and used as received for calibration of the gas chro-
matograph (GC) to obtain quantitative and qualitative analysis of
the reaction mixtures.
X ð%Þ
initial reactantðsÞ GC peak areaꢀfinal reactantðsÞ GC peak area
¼
ꢁ100
ð1Þ
2.2. Apparatus and general reaction procedure
initial reactantðsÞ GC peak area
The reaction was carried out in a tailor-made flow-through
reactor made of quartz glass and equipped with a heating jacket
and a reflux condenser operated at atmospheric pressure and the
equipped with control unit (CAL 9500P, CAL Controls) for accurate
temperature control of the catalyst bed. The catalyst was loaded in
the reactor tube (0.25 g in a typical experiment) along with the
thermocouple. The catalytic bed was immobilized in the middle
of the tube with the help of glass wool and glass balls. The tube
was fitted inside of the heater and connected with a collection flask
and a reflux condenser coupled to a collection flask. The reactant
and gas inlet feed pipes were connected at the top of the glass tube
for top-down flow, and the temperature was adjusted to the preset
level. Prior to the each and every experimental run, the system was
exposed for 2 h to hydrogen flow to facilitate catalyst activation
and to obtain thermal steady state of the system. Thereafter, the
GC peak area of the product Y
ꢁ 100
SY ð%Þ ¼
ð2Þ
R
GC peak area of all products
SY
100
Yield ð%Þ ¼ X ð%Þ ꢁ
ð3Þ
The weight hourly space velocity (WHSV) was expressed in (g/h)/g
and calculated by the following formula:
mass of reactant=h
mass of catalyst
WHSV ¼
ð4Þ
2.4. Catalyst preparation
reactant feed was initiated by continuously pumping
a-pinene to
The synthesis of SBA-15 material was carried out as described
elsewhere [20]. At first, for the preparation of Al-SBA15 (Si:Al ratio
of 25), 20 g of PEO–PPO–PEO triblock copolymer (P123, 99.0%, Sig-
ma–Aldrich) was dissolved in 150 mL of deionized water and
mixed with 600 mL of 2.0 M HCl. Consequently, the mixture was
left under stirring for 3 h, at 40 °C. As the next step, tetraethyl
orthosilicate (TEOS, 99.0%, Sigma–Aldrich) and AlCl3 (98.0%, Sig-
ma–Aldrich) were added in amounts of 42.5 and 1.93 g, respec-
tively. The mixture was stirred for 24 h at the above-mentioned
temperature. At the next stage, the solution was aged in an oven
(90 °C, 24 h). The obtained support structure was filtered and
washed thoroughly with deionized water and 250 mL of ethanol.
At the final stage, the formed Al-SBA15 was dried at 90 °C over-
night and calcined at 550 °C for 6 h.
the top of the upper section of the reactor tube acting as a vapor-
izer by means of an HPLC pump (high-precision liquid chromatog-
raphy pump), and thus, the reaction was commenced. For the
reliable investigation of the reaction kinetics, continuous sampling
from the collection flask was performed. The optimal temperature
and reactant flow were evaluated during the preliminary experi-
ments (Section 3.1). In all cases, a constant hydrogen flow of
10 mL/min was applied.
2.3. Product analysis
The reaction products were analyzed by means of a gas chroma-
tography (GC, Agilent Technologies, Model No. 7820A), equipped

M. Golets et al. / Journal of Catalysis 307 (2013) 305–315
307
Fig. 1. GC chromatogram of the isomerized reaction products with their specific retention times.
For the metal impregnation, PdCl2 and ZnCl2 (99.0% and 98.0%,
respectively, Merck Millipore) were selected as precursors. The re-
quired amounts of PdCl2 and ZnCl2 were diluted in HCl (0.1 M in
deionized water). The amount of Pd or Zn loaded on the support
was 3 and 5 wt%, respectively. The ratios of 1:1, 1:4, 4:1, 1:0, and
0:1 were prepared on a mass basis.
The X-ray diffraction (XRD) analysis revealed the structural fea-
tures of the studied catalyst, and a PANalytical diffractometer
using CuK
a (k = 0.154 nm) radiation was utilized. The X-ray dif-
fraction patterns of Pd–Zn/Al-SBA15 were recorded in the 2h range
of 0.5–10°. During the analysis, steps of 0.015° were set with count
time of 15 s at each point.
The support was dispersed in deionized water after which the
solution of precursors and Al-SBA15 was stirred for 5 h to reach a
uniform dispersion (wet impregnation) and dried in an oven at
100 °C for overnight. Consequently, the catalytic materials were
calcined in a tubular oven (400 °C, 3 h, 10 mL/min, synthetic air).
To obtain the final active form of the catalysts, a reduction in a
tubular oven was carried out (300 °C, 3 h, 50 mL/min, H2). The
in situ catalyst activation was performed each and every time prior
to the actual reaction (300 °C, 2 h, 10 mL/min H2).
Temperature-programmed desorption (TPD) of ammonia was
carried out using Micromeritics Instrument, AutoChem 2910. The
catalyst was dried at 110 °C overnight. Certain amount of catalyst
(50 mg) was placed in quartz made U-shaped tube. The catalyst
was first pretreated in a flow of helium (AGA AB, 99.999%) at
550 °C for 1 h, followed by saturation with anhydrous ammonia
in a flow of 10% NH3-90% He mixture at 120 °C (20 cm3/min).
The flushing of physisorbed NH3 was performed at 105 °C during
2 h, and temperature-programmed desorption was carried out
from ambient temperature to 500 °C (10 °C/min).
The thermo-gravimetric analysis (TGA) was carried out to quan-
tify the coke and residual organic compounds located in the cata-
lytic bed. The SDT-Q-600 (TA Instruments) analyzer was utilized
at 650 °C with the ramping of 10 °C/min and air flow rate of
100 mL/min.
2.5. Catalyst characterization methods
Samples of the catalyst taken before and after the reaction, car-
ried out under the optimized conditions, were characterized by
means of nitrogen physisorption measurements. The catalyst sur-
face area, pore volume, and diameter were measured in accordance
with the Brunauer–Emmett–Teller (BET) method in a Micromeri-
tics Tristar 3000 analyzer. About 0.1 g of the catalyst was degassed
at 200 °C for 1 h prior to the actual measurement. The obtained
adsorption isotherms served for the calculation of the specific sur-
face area, pore volume, and pore size distribution, respectively.
Additionally, X-ray Photoelectron Spectroscopy (XPS) technique
was applied to examine the chemical state of different species on
the catalyst. For the XPS measurements, a Kratos Axis Ultra Elec-
tron Spectrometer equipped with a delay line detector was used
3. Results and discussion
3.1. Optimization of the reaction conditions
The reaction conditions were optimized during the preliminary
studies. The most promising catalyst, Pd–Zn/Al-SBA15 1:1 (5 wt%),
was investigated at various temperatures (250, 275, 300, 325, 340,
and 400 °C) and flow rates (0.02, 0.03, 0.06, and 0.09 mL/min). The
optimization results are presented in Tables 1 and 2, respectively.
In accordance with earlier literature [1], dehydroisomerization
with the following parameters: monochromated AlK
a source
(150 W), hybrid lens system (magnetic lens), charge neutralizer,
and analysis area of 0.3 ꢂ 0.7 mm2. The C1s line of aliphatic carbon
(set at 285.0 eV) served as the reference for binding energy (BE)
value. The Kratos software was used for the processing of the
spectra.
To evaluate the distribution of the active metal sites on the cat-
alyst surface and to analyze the structures formed, Transmission
Electron Microscopy (TEM) was performed. First, the samples were
suspended in ethanol, and consequently, drop casted on the TEM
grid. The TEM images were taken using an Energy Filtered Trans-
mission Electron Microscopy (EFTEM, Leo 912 Omega) operating
at an acceleration voltage of 120 kV.
of a-pinene could not be performed at low reaction temperatures,
and mostly bicyclic products, such as camphene, were produced.
Earlier studies by Buhl et al. and Akpolat et al. suggested that a
temperature range around 300 °C should yield optimal results,
although catalyst poisons in the form of cleavage products would
form [5,23]. Wadaani et al. indicated that lower flow rates that in-
crease the reactant retention time would have a positive influence
on the
q-cymene selectivity when excess of camphene and limo-
nene was fed [6].
Also in our case, the temperature of 300 °C was close to the opti-
mal one. For the sake of comparison, an experimental runs were
carried out at 340 and 400 °C. Despite the complete conversion of

308
M. Golets et al. / Journal of Catalysis 307 (2013) 305–315
Table 1
evaluated flow conditions (Table 2). In conclusion, the reaction
temperature of 300 °C and feed rate of 0.03 mL/min were selected
as the optimal parameters for the subsequent testing.
q
-Cymene yield (wt%) upon temperature optimization (flow rate of
a-pinene 0.3 mL/
min).
Time (h)
Temperature (°C)
3.2. Long-time reaction behavior
250
275
300
325
340
400
0.5
1
2
3
4
54.1
52.8
51.9
50.6
49.7
49.2
74.4
73.0
68.0
67.8
64.9
66.6
77.1
75.6
74.1
69.2
67.1
66.9
73.6
71.2
71.8
70.7
68.4
66.5
58.4
56.9
55.9
55.1
54.8
54.2
50.0
42.4
40.7
39.8
39.2
38.1
In case of the bimetallic catalyst formulation (Pd–Zn/Al-SBA15),
-cymene was obtained as the main reaction product. Additionally,
q
limonene, camphene, m-cymene, and -menthene were formed in
q
small amounts (Fig. 1). Other non-identified products were formed
in concentrations lower than 1 wt% and were not considered here.
5
The highest yield of
conversion of -pinene was complete (100 wt%) in all cases stud-
ied. The catalytic activity was slowly deteriorating in terms of
the selectivity in a time frame of 10 h although -cymene yield still
q-cymene obtained was 77 wt%, whereas the
a
Table 2
q
q
-Cymene yield (wt%) upon the flow rate optimization (at 300 °C).
remained at around 60–65 wt% at this time. The performance of all
bimetallic catalysts observed for the first 5 h on stream is depicted
in Figs. 2–4, respectively. Moreover, for the sake of comparison,
monometallic catalysts impregnated with 5 wt% of Pd or Zn were
evaluated as well (Fig. 5). All experiments were performed at
Time (h)
Flow rate (mL/min)
0.02a
0.03b
0.06c
0.09d
0.5
1
2
3
4
66.7
71.9
71.4
70.8
70.6
69.6
77.1
75.6
74.1
69.2
67.1
67.0
73.4
73.3
73.4
71.3
69.7
67.1
76.4
73.6
71.1
69.1
66.9
66.0
300 °C and 0.03 mL/min of
reaction under similar conditions resulted in zero (0%) conversion.
As can be seen, the yield of -cymene was lower for 3 wt% of
a-pinene flow rate. A non-catalytic
q
5
Pd–Zn catalyst in comparison with the one having 5 wt% loading
(Fig. 2). However, a strong correlation was observed when looking
at the 3 wt% Pd–Zn catalyst (Fig. 2), and all catalysts prepared with
1:4 and 4:1 ratios of Pd and Zn, respectively (Figs. 3 and 4). In all
a
b
c
WHSV = 4.128 (g/h)/g.
WHSV = 6.192 (g/h)/g.
WHSV = 12.384 (g/h)/g.
WHSV = 18.576 (g/h)/g.
d
cases, the
q-cymene concentration obtained was within 5 wt%
interval. Hence, the metal loading had no significant influence on
the catalytic activity. On the other hand, as was finally established,
apparently, a Pd:Zn ratio of 1:1 and a metal loading of 5 wt% were
substrate, yields of
q-cymene were suppressed (Table 1). After
30 min on stream, at 400 °C, roughly 10 wt% of cracking by-prod-
ucts (retention times yielding from 12 to 14 min) were observed.
In effect, 42 wt% of undesired by-products were formed, and
optimal in terms of maximal
q-cymene yield. The formation of
by-products was also minimized upon selection of the aforemen-
tioned ratio and loading. Evidently, the metal loading had no
substantial influence on the catalytic stability as depicted in
Figs. 2–4, respectively.
50 wt% of
q-cymene and 8 wt% of other main reaction products
(camphene,
q-menthene, and limonene) were found. Also, in gen-
eral, the lower the reaction temperature, the better the production
economics. Low temperature operations gave rise to elevated
The influence of inert nitrogen atmosphere was tested as well.
Interestingly, upon use of the best (1:1) catalyst, after 30 min on
stream under nitrogen atmosphere, 15 wt% of camphene and only
camphene and
q-menthene yields up to 20 wt% but, at the same
time, significantly hampered the selectivity toward
q
-cymene.
When the flow rate was only 0.02 mL/min, numerous by-prod-
ucts were observed. On the contrary, the studied catalyst demon-
strated a solid performance at increased flow rates (from 0.06 to
55 wt% of
q-cymene were observed, indicating that the hydrogen
atmosphere was beneficial and it evidently improved the catalytic
performance by maintaining the metal in its active form for ex-
tended periods of time.
0.09 mL/min). However, q-cymene concentration profiles observed
at the outlet of the reactor were fluctuating more. Thus, an
intermediate feed rate of 0.03 mL/min chosen in the present study
in order to obtain reliable kinetic data. Furthermore, for this flow
rate, the highest concentration of q-cymene was obtained (Table 2).
The weight hourly space velocity (WHSV) was calculated for the
As illustrated by the XPS analysis (Section 3.3), palladium sites
were present on the catalytic surface in both oxidized and reduced
forms. Thus, we decided to evaluate which of the oxidation states
was actually active and responsible for the dehydrogenation reac-
tion. The unreduced (non-activated) catalyst (Pd–Zn/Al-SBA15 1:1,
100
90
80
70
60
50
40
30
20
10
0
100
90
80
70
60
50
40
30
20
10
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
Time, h
Time, h
Fig. 2. The evolution of reaction products, Pd–Zn/Al-SBA15 1:1 (5 wt% – left, 3 wt% – right) (300 °C, 0.03 mL/min). Symbols: (d) total cymene; (s)
(h) m-cymene; ( -menthene.
q-cymene; (j) limonene;
D) q

M. Golets et al. / Journal of Catalysis 307 (2013) 305–315
309
100
90
80
70
60
50
40
30
20
10
0
100
90
80
70
60
50
40
30
20
10
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
Time, h
Time, h
Fig. 3. The evolution of reaction products, Pd–Zn/Al-SBA15 1:4 (5 wt% – left, 3 wt% – right) (conditions and symbols – Fig. 2.).
100
90
80
70
60
50
40
30
20
10
0
100
90
80
70
60
50
40
30
20
10
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
Time, h
Time, h
Fig. 4. The evolution of reaction products, Pd–Zn/Al-SBA15 4:1 (5 wt% – left, 3 wt% – right) (conditions and symbols – Fig. 2.).
100
90
80
70
60
50
40
30
20
10
0
100
90
80
70
60
50
40
30
20
10
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
Time, h
Time, h
Fig. 5. The evolution of reaction products, Zn/Al-SBA15 5 wt% – left, Pd/Al-SBA15 5 wt% – right (conditions and symbols – Fig. 2, (+) – trans-pinane).
5 wt%) was immediately tested after the calcination under the
standard conditions (Fig. 2). A catalyst reduction in the tubular
oven (Section 2.4) at 400 °C and subsequent in situ activation
was run for the sake of comparison. In principle, the excess of
hydrogen is released during the dehydrogenation of the substrate.
Thus, it was also interesting to evaluate the necessity of the contin-
uous H2 flow in the reactor. A test run over non-activated, non-re-
duced sample of Pd–Zn/Al-SBA15 1:1 (5 wt%) was performed
under the standard conditions and in the absence of hydrogen
(Table 3).
ready after 3 h on stream. In comparison with the pre-reduced
sample, the initial selectivity toward -cymene was slightly higher
q
in case of reactant-activated sample. Nevertheless, low catalyst
stability was demonstrated in all three cases in contrast to opti-
mized conditions. Thus, a standard procedure including reduction
at 300 °C followed by the in situ activation is recommended. In
addition, the increased formation of camphene was favored upon
presence of the pre-reduced Pd sites.
Upon use of monometallic Pd or Zn catalysts (Fig. 5), the q-cym-
ene concentration was suppressed in comparison with bimetallic
In fact, self-activation by the reactant feed was observed since
catalysts (Figs. 2–4). Approximately two times more (38 wt%)
run 3 demonstrated high initial selectivity toward
q
-cymene
q-cymene was formed when pure Zn was used instead of pure
(75 wt%). However, the -cymene selectivity dropped to 64 wt% al-
q
Pd. Thus, apparently, Zn sites seem more prone to facilitate

310
M. Golets et al. / Journal of Catalysis 307 (2013) 305–315
Table 3
depending on the catalyst used, could be seen in very minor
amounts (<0.025 wt%). Indeed, side reactions such as cracking
reactions of a-pinene occurred to some extent. Interestingly, upon
Isomerization products obtained with various samples of Pd–Zn/Al-SBA15 1:1, 5 wt%
(complete conversion, optimized conditions).
Time (h)
q-Cymene
Camphene
q-Menthene
high palladium loading (ex. 4:1 Pd–Zn), less by-products were ob-
Sample 1a
Sample 2b
Sample 3c
0.5
1
2
3
4
72.1
67.6
63.0
61.9
61.7
60.1
7.6
7.0
9.5
11.2
11.6
12.3
4.4
3.5
5.7
6.8
7.0
7.3
served. On the other hand, lower relative amounts of Pd (e.g., in the
case of Pd–Zn 1:4 catalyst) enhanced
q-cymene yields but, at the
same time, also lead to increased amounts of undesired reaction
products.
5
3.3. Catalyst characterization
0.5
1
2
3
4
70.0
67.6
66.1
61.9
60.2
59.8
2.4
3.8
5.2
6.8
7.1
7.6
6.2
6.9
6.8
6.0
5.7
5.5
The XPS spectra were recorded for the freshly prepared Pd–Zn/
Al-SBA15 1:1 (5 wt%). Spent catalyst analyzed after third cycle
(Section 3.5) was analyzed for comparison. At the surface of fresh
catalyst, Pd was present as Pd(II). Nevertheless, the reduction step
was performed after the preparation. The high reactivity of the cat-
alytic material under study presumably gave rise to spontaneous
formation of palladium oxides due to the access of humidity and
air. As claimed by Albers et al., such a transformation took place
for Pd/SiO2 upon the hydrogenation of acetylene to ethylene under
moderate temperatures (<200 °C) [29]. Further, other studies [30]
have indicated the surface adsorption of oxygen monolayer and
bulk oxidation of supported palladium within the temperature
range of ꢀ73 to 200 °C, respectively. Delidovich et al. observed pal-
ladium oxide formation upon storage of Pd/C catalyst at room tem-
perature and also confirmed this by means of XPS analysis [31]. In
fact, the binding energy (BE) of Pd 3d5/2 decreased from 336.0 to
335.6 eV upon catalyst exposure to the reaction conditions. A de-
crease in BE of Pd 3d5/2 line indicates that Pd was primarily present
at reduced forms in the spent catalyst [32]. Moreover, no Pd–Zn al-
loy peak was detected upon XPS analysis [32]. In addition, up to
17 at% (or 3 wt%) of the catalytic surface was coked in the spent
catalyst. The major bond types of coke were C–C and C–H bonds
(approx. 14 at%). Minor amounts (3 at%) of C–O bonds were also
detected. The BE of Zn 2p3/2 remained at the constant value of
1022.7 eV. In fact, the monometallic palladium catalyst displayed
no changes in BE that also corresponded to the suggested oxidation
state of Pd(II) sites. The XPS analysis of monometallic Pd/Al-SBA15
and Zn/Al-SBA15 catalysts (Fig. 5) demonstrated the BE values of
336.0 and 1022.7 eV for Pd and Zn lines, respectively. Thus, there
was no evidence of direct chemical interaction between Pd and
Zn ions. The individual XPS spectra for fresh and spent catalysts
are shown in Figs. 7 and 8, respectively.
5
0.5
1
2
3
4
75.2
74.1
66.1
64.0
60.8
59.7
1.1
1.1
4.0
6.4
6.6
7.1
2.5
3.5
6.2
3.5
3.4
3.4
5
a
b
c
Sample 1: pre-reduced under 400 °C, standard H2 flow (10 mL/min).
Sample 2: non-activated and non-reduced, standard H2 flow (10 mL/min).
Sample 3: non-activated and non-reduced, no make-up gas.
dehydrogenation, as observed earlier [20]. The hydrogenation reac-
tion was promoted by use of the pure Pd, and hydrogenation prod-
ucts, such as trans-pinane (35 wt%), were mostly formed.
Native Al-SBA15 was also tested in order to evaluate the direct
influence of the support acidity on the reaction performance. Con-
sequently, Si-SBA15 support was prepared in line with the analo-
gous procedure as before (Section 2.4). The results are presented
in Fig. 6. On the basis of the data, the suitability of Al-SBA15 was
further supported. The substantial acidity caused by alumina
incorporation in Al-SBA15 gave rise to favorable properties of final
catalyst in terms of acid-catalyzed reactions, including those of
isomerization and dehydrogenation. Previous studies have referred
to the lack [24–25] and absence [26] of acidity for pure siliceous
mesoporous materials. Evidently, in case of Al-SBA15, both Lewis
and Brønsted acid sites were present, and such medium acidic cen-
ters are typically related to the amount of alumina incorporated
[24–28]. Consequently, the Brønsted acid sites could provide pro-
tons for the
a-pinene internal ring cracking. Presumably, the
Brønsted acidity also contributed to the oligomerization of the ole-
finic compounds; this being the main reason for coking [20].
All in all, 10–38 various by-products in concentrations less than
1 wt% could be seen in the GC chromatogram. Nevertheless, upon
more detailed analysis, even more by-products (up to 170),
The TEM measurement demonstrated perfectly formed catalytic
structure (Pd–Zn/Al-SBA15 1:1, 5 wt%) with an even dispersion of
Pd and Zn metals. Moreover, the catalytic properties remained
unaltered throughout the experiment. Unfortunately, the metal
100
90
80
70
60
50
40
30
20
10
0
1
2
3
4
5
334
336
338
340
342
344
Time, h
Binding energy (eV)
Fig. 6. Comparison of the reaction performance for Al-SBA15 (bricks) and Si-SBA15
(triangles) (conditions – Fig. 2.). Symbols: j, N –
cymene yield.
a-pinene conversion; h,
D
–
q-
Fig. 7. XPS spectra for the fresh Pd–Zn/Al-SBA15 1:1 (5 wt%). The peak with binding
energy 336.0 eV is referred to the Pd(II) sites.

M. Golets et al. / Journal of Catalysis 307 (2013) 305–315
311
Table 4
Acid and textural properties data for Pd–Zn/Al-SBA15 1:1 (5 wt%): NH3-TPD total
acidity; BET surface area, pore size and pore volume; XRD d100 spacing and unit cell
parameter.
Catalyst
NH3-TPD
BET-N2 adsorption
XRD
Total
acidity
SBET
(m2/
g)
Pore
size
(nm)
Pore
d100
(nm) cell-a0
Unit
volume
(
lmol/g)
(cm3/g)
(nm)
Fresh
Spent
Recovered 1013
831a
1254
868.7 7.5
417.1 9.7
782.9 5.5
1.5
1.0
1.1
–
9.0
9.9
–
10.4
11.4
a
Same as for the unloaded Al-SBA15.
334
336
338
340
342
344
Binding energy (eV)
onstrated by Si-SBA15 as described previously (Section 3.2). Inter-
estingly, the number of Pd–Zn/Al-SBA15 1:1 (5 wt%) acid sites was
Fig. 8. XPS spectra for the spent Pd–Zn/Al-SBA15 1:1 (5 wt%). Peaks with binding
energies 336.0 and 335.6 eV are referred to the Pd(II) and Pd(0) sites, respectively.
increased to 1250
lmol/g in the spent catalyst. The recovered cat-
alyst demonstrated a total acidity of 1010
lmol/g. Presumably,
immobilization of acidic by-products took place in the catalytic
bed and facilitated the growth in total acidity. According to Zhao
et al., the olefin-containing products could polymerize on the
acidic catalysts causing their deactivation [34]. As supposed, the
initial acidity of the studied catalytic material was optimal in terms
particles were not visible in the fresh catalyst due to the fact that
the Pd was present in oxidized form. The visible Pd sites, in case
of the spent catalyst, could be explained by the reduction in the
Pd(II). This was also confirmed by the XPS analysis. The high-reso-
lution TEM images, taken at 200 nm scale, are presented in Fig. 9.
As can be seen in TEM images, relatively large-sized, spherically
shaped nanoparticles simultaneously accumulated on the external
surface of the support and resided in the pore system.
The nitrogen physisorption measurements confirmed the de-
cline in surface area of the catalyst after the reaction, while
changes in the pore volumes were negligible. The original values
could, however, to a large extent, be restored in reactivated cata-
lysts (Table 4). Thus, the nitrogen physisorption supported the
XPS observations in terms of coking as the primary cause of cata-
lyst deactivation.
of
q-cymene formation.
Water, coke, and residual organics were altogether detected in
the spent catalyst during the TGA analysis in approximate amounts
of 1.5, 3, and 5 wt%, respectively (Fig. 11). In general, the XPS
observation concerning coking and that of TPD concerning residual
organics localized in the catalyst framework were confirmed by
TGA.
3.4. Reaction network
The small-angle XRD patterns of spent and recovered Pd–Zn/Al-
SBA15 1:1 (5 wt%) catalysts taken from the recovery trials
(Section 3.5) are presented in Fig. 10. Three specific peaks were ob-
served in the low angle (2h) range at 2h around 1.0°, 1.6°, and 1.8°,
respectively, indicating the formation of well-ordered hexagonal
An overall reaction mechanism was proposed capable of
describing the experimental observations and also supported by
the literature (Fig. 12) [1,5,6,9,11,17]. At the first stage,
is involved in the fast acid-catalyzed isomerization to form limo-
nene and -menthadienic compounds. The conjugated endocyclic
a-pinene
q
structure of SBA15 [33]. The d100 spacing (nk ¼ 2d sin h) and unit
menthadienes are characterized by strong adsorptivity on the cat-
alytic surface and high reactivity in terms of both dehydrogenation
and oligomerization. Presumably, the mentioned species were sub-
ject to dehydrogenation before the actual desorption; thus, they
could not be observed in the product mixture [1,11]. According
pffiffiffi
cell parameter (a0 ¼ 2d100
= 3) were calculated for the catalytic
material in analogy with our previous study (Table 4) [33]. The
hexagonal structure of the support was not affected during the
isomerization reaction. However, the peak intensities, unit cell
parameter, and d100 spacing values are increased during the recov-
ery of spent catalyst that might be the result of the coke removal
and restoration of the original properties of fresh catalyst.
to Roberge et al. [1], acid cracking of the cyclobutane ring in
a-
pinene C–C bond between the carbon atoms 1 and 6 is most prob-
able to occur. The second stage was considered the rate-limiting
step, i.e. the hydrogenation/dehydrogenation reactions occurring
As evaluated by the NH3-TPD measurement, the total acidity of
Al-SBA15 support remained constant after the metal impregnation
on the metal sites to yield
[6,35]. Additionally, minor amounts of
tained via the dehydrogenation of previously formed intermediate
q-menthene and
q-cymene, respectively
(Table 4). The mentioned total acidity value of 830
lmol/g was in
q-cymene could be ob-
line with the literature [24]. On the contrary, zero acidity was dem-
Fig. 9. TEM images taken for the fresh (left) and spent (right) catalyst (200 nm).

312
M. Golets et al. / Journal of Catalysis 307 (2013) 305–315
with the monometallic catalyst (Section 3.2 and Fig. 5). As depicted
in Fig. 5, zinc sites were involved facilitating dehydrogenation of
m- and -menthadienes to m- and -cymenes, respectively [20].
The Pd sites interacted with -menthadienes providing hydrogena-
tion function responsible for the -menthene formation. The com-
q
q
q
q
plementary dehydrogenation function is also referred to the
palladium sites, as also claimed in the literature [12]. As described
by Buhl et al., the decline in activity or accessibility of Pd sites
stimulated the formation of menthadienes and decreased the
cymene selectivity [5]. In a reaction studied by Roberge et al., Pd
was supported on high surface area carriers: zeolites, silica, -alu-
q-
c
mina, and carbon. As described by Roberge et al., in case of
monometallic catalyst, the absence of Pd eliminated the dehydro-
genation function, and isomerization reaction only toward
1
2
3
4
5
2 theta (degree)
camphenic, fenchenic, and q-menthadienic compounds was possi-
ble. In opposite, only hydrogenated and aromatic compounds were
formed in case when the Pd containing catalysts were utilized [1].
Thus, the individual activities of Pd and Zn improved the
performance of bimetallic catalysts; nevertheless, no Pd–Zn alloy
structure was observed (Section 3.3).
Fig. 10. Low-angle X-ray diffraction pattern of the Pd–Zn/Al-SBA15 catalyst
(recovered – top, spent – bottom).
c
-terpinene [1,9]. The formation of
concentrations less than 0.1 wt%. The presence of m-cymene and
-menthene, respectively, confirms the plausibility of the pro-
posed reaction network. Furthermore, the concentrations of
menthene (Figs. 2–4, respectively) substantially exceeded that of
-terpinene, so the route via the formation of limonene was rather
preferable.
In the presence of camphene, also m-cymene was observed.
c-terpinene was detected in
q
q-
3.5. Catalyst reuse
c
The catalyst reuse was studied for Pd–Zn/Al-SBA15 1:1 (5 wt%)
after 10 h of reaction. The reactor tube, containing the reduced cat-
alyst, was placed in the tubular oven and treated at 500 °C in air
flow of 50 mL/min for 5 h to burn off the coke formed. The recov-
ered catalyst was activated and tested under the optimized reac-
tion conditions (Section 3.1) upon which three consecutive cycles
were performed. Data describing the performance of the recycled
catalyst are presented in Fig. 13.
Initially, camphene isomerized to various intermediate m-mentha-
dienic structures in analogy with the main pathway [1,11]. The
camphene, which we assume as the intermediate product, was
identified by the GC analysis, along with reaction products at the
concentration range of 2.5–5 wt%, during the first 3 h on stream.
The previously mentioned assumption was supported by the sub-
stantial growth in the camphene concentration (up to 10 wt% after
10 h of reaction) and also evidently contributed to the catalyst
deactivation.
The catalyst investigated demonstrated high stability. The ini-
tial conversion remained unchanged (100 wt%) even after the third
regeneration cycle, although the selectivity toward
q-cymene de-
To explain the synergic properties of Pd–Zn, the reaction net-
work (Fig. 12) was studied in light of the experiment carried out
creased upon prolonged reaction times (10 h) for recycled cata-
lysts. Upon the third recycling, the conversion decreased slightly
Fig. 11. TGA profile for spent Pd–Zn/Al-SBA15 1:1 (5 wt%), catalyst weight loss. Estimated temperature intervals: 0–150 °C – water; 150–450 °C – organics; 450–620 °C –
coke.

M. Golets et al. / Journal of Catalysis 307 (2013) 305–315
313
Fig. 12. The overall reaction network.
100
90
80
70
60
50
40
30
20
10
0
3
1
C
B
F
E
1
1
2
1
2
2
3
A
1
1
3
2
3
4
2
3
3
2
D
5
Fig. 14. Description of the reaction network used in the kinetic modeling (symbols:
A – -pinene, B – -cymene, C – m-cymene, D – -menthene, E – limonene, F –
camphene).
a
q
q
0.5
1
2
3
10
Time, h
(system of ordinary differential equations) with the backward dif-
ference method and optimizes the parameter values using a hybrid
method involving Simflex and Levenberg–Marquardt methods. The
goal function in the optimization is the sum of square function:
Fig. 13. Long-term catalyst stability test cycles (from left to the right). Reaction
conditions: Fig. 2.
after 10 h on stream. At the same time, up to 15 wt% of camphene
and 15 wt% of limonene were detected. The decline in the catalytic
activity could be explained by the coking of the catalytic surface
and the organic residues accumulation in the catalyst framework
(Section 3.3) [29]. A thin, visible layer of coke was also formed
on the reactor walls.
X
2
SSQ ¼
ðyexp ꢀ ymodelÞ w
ð6Þ
The weight factor w was set to 1 for all experimental points. For
the optimization of the experimental data, the components
a-
pinene, -cymene, -menthene, and camphene were used. The
q
q
experimental values for m-cymene and limonene were not used,
but the concentrations of them were calculated by the model
according to the reaction scheme.
3.6. Kinetic modeling
The temperature dependency was expressed with the normal-
ized Arrhenius equation as follows:
The kinetic model was proposed in accordance with the pro-
duced experimental data. Reaction network was organized in line
with Fig. 14, which represented main pathways of the actual reac-
tion network (Fig. 12).
Ea
j
1:0d0 1:0d0
temp ꢀ
Aj ¼ kj ꢁ eꢀ
ð7Þ
R ð
TmeanÞ
To model the catalytic packed-bed reactor, the ideal plug flow
model was used:
where Eaj are the activation energies (J/mol); R is the gas constant
(J/mol K).
The reaction rates were expressed by means the following set of
equations:
_
dni
¼ rimcat
ð5Þ
dz
cA
r1 ¼ k1 ꢁ
r2 ¼ k2 ꢁ
ð8Þ
ð9Þ
where ri denote the component generation rates and mcat is the cat-
alyst mass (g) in the reactor.
D ꢁ deact
For the modeling of the reaction kinetics, the software ModestÒ
6.1 was used [36]. The software solves the reactor model equations
cA
D ꢁ deact

314
M. Golets et al. / Journal of Catalysis 307 (2013) 305–315
cF
r3 ¼ k3 ꢁ
ð10Þ
D ꢁ deact
cE
r4 ¼ k4 ꢁ
r5 ¼ k5 ꢁ
ð11Þ
ð12Þ
D ꢁ deact
cE
D ꢁ deact
where ri denote the reaction rates (mol/mcat min) toward the
respective product; Ci are the concentrations (mol/dm3); and ki is
the reaction rate constant (minꢀ1) and the parameter deact ac-
counts for catalyst deactivation.
The flow-through reactor model is expressed as follows
(Fig. 14):
ðꢀr1 ꢀ r2Þ
dsð1Þ ¼
dsð2Þ ¼
dsð3Þ ¼
dsð4Þ ¼
dsð5Þ ¼
dsð6Þ ¼
A
ð13Þ
ð14Þ
ð15Þ
ð16Þ
ð17Þ
ð18Þ
f ꢁ mcat
r4
f ꢁ mcat
B
Fig. 15. Simulation of catalytic packed-bed reactor with the concentration of
reactant as function of residence time (for temperatures 275 and 300 °C). Symbols:
(o) a-pinene, (+) q-cymene, (h) q-menthene.
r3
f ꢁ mcat
C
r5
f ꢁ mcat
D
scale high-pressure reactor (PARR) at 150–250 °C with 0.25 g of
Pd–Zn/Al-SBA15 1:1 (5 wt%) on the pure -pinene substrate.
a
r2 ꢀ r4 ꢀ r
f ꢁ mcat
5 E
Although a hydrogen pressure of 10 bar was applied, the conver-
sion did not exceeded 5 wt%, and no dehydrogenation products
were detected.
r1 ꢀ r3
F
On the contrary to a batch reactor, a continuous reactor config-
uration allows for the simplicity of modifications, solventless oper-
ations, and good process control in both laboratory and industrial
scales [37,38]. On the other hand, the use of continuous process
and active catalysts is crucial in terms of providing high reaction
rate per unit volume of reactor. Moreover, low liquid-to-solid vol-
ume ratio is needed to suppress the homogeneous side reactions
[38,39]. Therefore, continuous flow-through reactor is the only fea-
sible choice for the investigated reaction in terms of both improved
f ꢁ mcat
where ds(i) denote the generation rates (mol/mcat min) for each and
every component and f denotes the flow rate (mol/min).
The degree of explanation was very good – 99.6%, and the calcu-
lated errors of the estimated parameters are all on an acceptable
level (Table 5).
Fig. 15 represents a parity plot including all data points showing
how well the model fits the data. Fig. 16 shows the calculated con-
centration profiles as a function of the longitudinal coordinate of
the catalyst bed, the experimentally obtained concentrations are
also shown at the outlet of the reactor.
selectivity toward q-cymene and suitability for the kinetic studies
[38]. In addition, the camphene route could be suppressed by
adjusting the retention time by means of tuning of the catalytic
bed length at the suitable WHSV. The evaluation of the aspects
such as the influence of the reactor walls, heat recovery, mixing
3.7. Continuous reactors for feasible operations
Generally, to obtain good q-cymene yields in liquid phase, high
catalyst loading is required [5], and the isomerization reactions to-
ward camphene and related products are preferred [1]. A previous
study by Grau et al. demonstrated only minor formation of q-cym-
ene in liquid phase, although supported Pd catalysts were applied
[11]. Respectively, upon use of our catalysts in liquid phase gave
worthless results. Several trials were conducted in the laboratory
Table 5
The numerical values of parameters and their errors.
Estimated Estimated
Est. relative error of
parameters
Std. error
(%)
Std error
(%)
kde
k1
k2
k3
k4
k5
Ea1
Ea2
0.252Eꢀ01 0.388Eꢀ03
1.5
7.1
2.2
30.1
4.9
24.4
5.8
64.8
14.1
45.5
3.3
20.2
4.1
0.686E+00
0.265E+01
0.142E+01
0.974E+01
0.389E+00
0.329E+05
0.402E+05
0.486Eꢀ01
0.583Eꢀ01
0.429E+00
0.481E+00
0.950Eꢀ01
0.192E+04
0.693E+03
17.1
58.0
1.7
Fig. 16. Concentration profiles inside catalyst bed.

M. Golets et al. / Journal of Catalysis 307 (2013) 305–315
315
[2] P.R. Makgwane, PhD Thesis, Faculty of Science, Nelson Mandela Metropolitan
University, November 2009.
[3] H. Fiege, Ullmann’s Encyclopedia of Industrial Chemistry, vol. A8, VCH,
Weinheim, 1985, p. 25.
as well as separation aspects should be considered for the potential
scale-up operations of the present process [38].
[4] K.S.N. Reddy, B.S. Rao, V.P. Shiralkar, Appl. Catal. A: Gen. 121 (1995) 191–201.
[5] D. Buhl, P.A. Weyrich, W.M.H. Sachtler, W.F. Holderich, Appl. Catal. A: Gen. 171
(1998) 1–11.
[6] F.A. Wadaani, E.F. Kozhevnikova, I.V. Kozhevnikov, Appl. Catal. A: Gen. 363
(2009) 153–156.
[7] N.N. Binitha, S. Sugunan, Catal. Commun. 8 (2007) 1793–1797.
[8] A. Stanislaus, L.M. Yeddanapalli, Can. J. Chem. 50 (1972) 61–74.
[9] B.G. Harvey, M.E. Wright, R.L. Quintana, Energy Fuels 24 (2010) 267–273.
[10] C. Fernandes, C. Catrinescu, P. Castilho, P.A. Russo, M.R. Carrott, C. Breen, Appl.
Catal. A: Gen. 318 (2007) 108–120.
[11] R.J. Grau, P.D. Zgolicz, C. Gutierrez, H.A. Taher, J. Mol. Catal. A: Chem. 148
(1999) 203–214.
[12] P.A. Weyrich, W.F. Holderich, Appl. Catal. A: Gen. 158 (1997) 145–162.
[13] P.A. Weyrich, PhD Thesis, RWTH Aachen, Germany, 1995.
[14] D. Buhl, D.M. Roberge, W.F. Hölderich, Appl. Catal. A: Gen. 188 (1999) 287–
299.
[15] P. Lesage, J.P. Candy, C. Hirigoyen, F. Humblot, J.M. Basset, J. Mol. Catal. A:
Chem. 112 (1996) 303–309.
[16] M.A. Martın-Luengo, M. Yates, M.J. Martınez-Domingo, B. Casal, M. Iglesias, M.
Esteban, E. Ruiz-Hitzky, Appl. Catal. B: Environ. 81 (2008) 218–224.
[17] M.A. Martin-Luengo, M. Yates, E.S. Rojo, D. Huerta Arribas, D. Aguilar, E. Ruiz
Hitzky, Appl. Catal. A: Gen. 387 (2010) 141–146.
4. Conclusions
The results of the current study are summarized as follows:
ꢃ The proposed process design is clearly advantageous due to the
complete (100%) conversion of
able
ene, m-cymene, and
a-pinene and high yield of valu-
q-cymene – 77% (by weight). Moreover, limonene, camph-
q
-menthene were detected as by-products.
The modeling results further supported the hypothesis of the
predominant reaction pathways.
ꢃ The best catalyst, Pd–Zn/Al-SBA15 1:1 (5 wt%), demonstrated
good stability at high flow rates and a reasonable recoverability
with prolonged times on stream. According to the TEM and XRD
analyses, perfect hexagonal structure of the SBA15 was
observed and retained after the reaction. Despite that XPS anal-
ysis detected no chemical interaction or alloy structure between
Pd and Zn sites, their synergy was uncovered due to individual
dehydrogenation activities provided by both metals. In oppo-
site, upon use of monometallic catalysts and non-catalytic trials
poor results were obtained.
[18] C. Catrinescu, C. Fernandes, P. Castilho, C. Breen, Appl. Catal. A: Gen. 311 (1)
(2006) 172–184.
[19] V. Krishnasamy, Aust. J. Chem. 33 (1980) 1313–1321.
[20] J. Du, H. Xu, J. Shen, J. Huang, W. Shen, D. Zhao, Appl. Catal. A: Gen. 296 (2005)
186–193.
ꢃ Coke and residual organics accumulated on the catalyst frame-
work after longer experimental trials, which was confirmed by
the XPS, NH3-TPD, and TGA analyses; catalyst coking could evi-
dently contribute to the observed decline in catalytic perfor-
mance. Still, the initial acidity and textural properties of the
catalyst could be restored with the recovery procedure.
ꢃ The described catalytic reaction represents a plausible example
of a one-step green conversion of renewable feedstocks into
value-added chemicals. The proposed approach could render a
cheaper and more environmentally friendly process than the
catalytic dehydrogenation strategy presently used by the indus-
try. In summary, further studies are targeted toward the poten-
tial application of Pd–Zn/Al-SBA15 catalysts for the value-added
upgrading of the industrial turpentine.
[21] S. Chinayon, O. Mekasuwandumrong, P. Praserthdam, J. Panpranot, Catal.
Commun. 9 (2008) 2297–2302.
[22] H.A. Meylemans, R.L. Quintana, B.G. Harvey, Fuel 97 (2012) 560–568.
[23] O. Akpolat, G. Gunduzb, F. Ozkanc, N. Besun, Appl. Catal. A: Gen. 265 (2004)
11–22.
[24] M. Gomez-Cazalilla, J.M. Merida-Robles, A. Gurbani, E. Rodriguez-Castellon, A.
Jimenez-Lopez, J. Solid State Chem. 180 (2007) 1130–1140.
[25] A.K. Medina-Mendoza, M.A. Cortés-Jácome, J.A. Toledo-Antonio, C. Angeles-
Chávez, E. López-Salinas, Ignacio Cuauhtémoc-López, María C. Barrera, J.
Escobar, J. Navarrete, I. Hernández, Appl. Catal. B: Environ. 106 (2011) 14–25.
[26] Y. Li, W. Zhang, L. Zhang, Q. Yang, Z. Wei, Z. Feng, C. Li, J. Phys. Chem. B 108
(2004) 9739–9744.
[27] A. Gedeon, A. Lassoued, J.L. Bonardet, J. Fraissard, Micropor. Mesopor. Mater.
44–45 (2001) 801–806.
[28] G.V. Shanbhag, S.M. Kumbar, S.B. Halligudi, J. Mol. Catal. A: Chem. 284 (2008)
16–23.
[29] P. Albers, J. Pietsch, S.F. Parker, J. Mol. Catal. A: Chem. 173 (2001) 275–286.
[30] Z. Li, G.B. Hoflund, J. Nat. Gas. Chem. 12 (2003) 153–160.
[31] I.V. Delidovich, O.P. Taran, L.G. Matvienko, A.N. Simonov, I.L. Simakova, A.N.
Bobrovskaya, V.N. Parmon, Catal. Lett. 140 (1–2) (2010) 14–21.
[32] N. Takezawa, N. Iwasa, Catal. Today 36 (1997) 45–56.
[33] S. Ajaikumar, M. Golets, W. Larsson, A. Shchukarev, K. Kordas, A.-R. Leino, J.-P.
Mikkola, Micropor. Mesopor. Mater. 173 (2013) 99–111.
[34] G. Zhao, J. Teng, Z. Xie, W. Jin, W. Yang, Q. Chen, Y. Tang, J. Catal. 248 (2007)
29–37.
Acknowledgments
Processum Biorefinery Initiative AB, Domsjö Aditya Birla AB,
Metsä Board, Holmen Skog, Kempe Foundations, and the Bio4E-
nergy program are gratefully acknowledged for the financial sup-
port. COST CM0903 (UbioChem) is also acknowledged. Dr.
Andrey Shchukarev (Umeå University) and PhD. Stud. Eero Salmi-
nen (Åbo Akademi) are acknowledged for their assistance during
catalyst characterization experiments. This work is also associated
with the Academy of Finland Research Program Nr. 268937.
[35] G.A. Somorjai, Introduction to Surface Chemistry and Catalysis, Wiley, New
York, 1994.
[36] H. Haario, Modest 6.0 A., User’s Guide, ProfMath, Helsinki, 2001.
[37] E. Jones, K. McClean, S. Housden, G. Gasparini, I. Archer, Chem. Eng. Res. Des.
90 (6) (2012) 726–731.
[38] F.S. Mederos, J. Ancheyta, J. Chen, Appl. Catal. A: Gen. 355 (2009) 1–19.
[39] Z. Kuzeljevic, PhD Thesis, School of Engineering and Applied Science,
Department of Energy, Environmental and Chemical Engineering,
Washington University, May 2010.
References
[1] D.M. Roberge, D. Buhl, J.P.M. Niederer, W.F. Hölderich, Appl. Catal. A: Gen. 215
(2001) 111–124.