Selective, one-pot catalytic conversion of levulinic acid to pentanoic acid over Ru/H-ZSM5

Article abstract of DOI:10.1016/j.jcat.2014.09.014

The direct conversion of levulinic acid (LA) to pentanoic acid (PA) has been studied with six 1 wt% Ru/H-ZSM5 catalysts at 40 bar H₂ and 473 K in dioxane. The influence of ZSM5 cation form, Si/Al ratio and ruthenium precursor on metal dispersion and acidity has been assessed. A highly active bifunctional 1 wt% Ru/H-ZSM5 catalyst was developed to give a PA yield of 91.3% after 10 h. The PA productivity of 1.157 molP_A g_Ru¹ h¹ is the highest reported to date. The simple preparation method allows for a significant fraction of ruthenium to be located inside the zeolite pores, providing the desired proximity between the hydrogenation function and the strong acid sites, which is key to the conversion of LA into PA. Coke buildup during reaction causes some deactivation, but activity can be almost fully restored after catalyst regeneration by simple coke burn-off.

Full text of DOI:10.1016/j.jcat.2014.09.014

Journal of Catalysis 320 (2014) 33–41
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Selective, one-pot catalytic conversion of levulinic acid to pentanoic acid
over Ru/H-ZSM5
⇑
⇑
Wenhao Luo, Pieter C.A. Bruijnincx , Bert M. Weckhuysen
Inorganic Chemistry and Catalysis, Debye Institute for Nanomaterials Science, Utrecht University, Universiteitsweg 99, 3584 CG Utrecht, The Netherlands
a r t i c l e i n f o
a b s t r a c t
Article history:
The direct conversion of levulinic acid (LA) to pentanoic acid (PA) has been studied with six 1 wt% Ru/H-
ZSM5 catalysts at 40 bar H and 473 K in dioxane. The influence of ZSM5 cation form, Si/Al ratio and
2
ruthenium precursor on metal dispersion and acidity has been assessed. A highly active bifunctional 1
Received 17 June 2014
Revised 8 September 2014
Accepted 19 September 2014
wt% Ru/H-ZSM5 catalyst was developed to give a PA yield of 91.3% after 10 h. The PA productivity of
À1 À1
1.157 molPA
g
Ru
h
is the highest reported to date. The simple preparation method allows for a significant
fraction of ruthenium to be located inside the zeolite pores, providing the desired proximity between the
hydrogenation function and the strong acid sites, which is key to the conversion of LA into PA. Coke buildup
during reaction causes some deactivation, but activity can be almost fully restored after catalyst regenera-
tion by simple coke burn-off.
Keywords:
Levulinic acid
Pentanoic acid
ZSM5
Ruthenium
Acidity
Ó 2014 Elsevier Inc. All rights reserved.
Catalyst synthesis
1
. Introduction
Levulinic acid (LA) has emerged as one of the most promising
an acidic ion exchange resin. In addition to PE synthesis, a compre-
hensive study of their fuel properties was conducted, which dem-
onstrated the compatibility of PE for both gasoline and diesel
applications and revealed a superior performance [6].
The Dumesic group also reported on the production of PA, in
2 5
this case from GVL over a bifunctional Pd/Nb O catalyst [8]. PA
renewable platform molecules. Indeed, it provides many opportu-
nities for biobased fuel and bulk chemicals production [1,2] and
can be produced easily and economically from the carbohydrate
fraction of lignocellulosic biomass through a simple and high yield-
ing acid hydrolysis process [3]. Several derivatives of LA have been
could subsequently be converted into 5-nonanone, i.e., by ketoni-
zation and ultimately to hydrocarbon fuels suitable for gasoline
and diesel applications [8]. Both approaches thus consist of a
two-step conversion of LA into PA and require a catalyst system
combining an acid functionality and a hydrogenation function for
the second, most difficult step, i.e., the ring-opening/hydrogena-
tion of GVL to PA.
Building on these results, we previously reported on the direct,
one-pot conversion of LA to PA and PE, without isolating the GVL
intermediate, using bifunctional catalysts consisting of ruthenium
supported on H-ZSM5 or H-b. A yield of 45.8 mol% of PA was
obtained with 1 wt% Ru/H-ZSM5 in dioxane at 473 K and after
4 h of reaction time [9]. An extensive catalyst deactivation study
showed that acid site loss as the result of dealumination is the
cause of deactivation under the applied conditions, which involved
elevated temperatures and a highly polar and corrosive environ-
ment. Relatedly, Pan et al. recently reported the direct conversion
of LA to valeric acid and its ester, performing the reaction in etha-
nol over various bifunctional ruthenium catalysts with 5 wt% Ru/
studied as cellulosic fuels or fuel additives; these include c-valero-
lactone (GVL) [4], 2-methyltetrahydrofuran (MTHF) [5], pentanoic
acid (PA) and its esters (PE) [6], all of which can be obtained by
sequential LA hydrodeoxygenation steps (Scheme 1). While GVL
has received most attention, two recent examples have demon-
strated that pentenoic/pentanoic acid-based value chains are also
very promising for the production of cellulosic biofuels [7].
Lange et al. showed, for instance, that so-called valeric biofuels,
i.e., pentanoic acid esters, are a superior class of cellulosic biofuels
[
6]. A multistep catalytic conversion of LA into valeric esters was
reported, involving the hydrogenation of LA to GVL over Pt/TiO
2
,
the conversion of GVL to PA by acid-catalyzed ring-opening and
the hydrogenation over a bifunctional Pt/H-ZSM5 catalyst and,
finally, esterification to the desired pentanoic acid esters (PE) over
⇑
Corresponding authors. Fax: +31 (0)30 251 1027.
E-mail addresses: p.c.a.bruijnincx@uu.nl (P.C.A. Bruijnincx), b.m.weckhuysen@
3
SBA-SO H performing best with a combined PA/PE yield of 94%
uu.nl (B.M. Weckhuysen).
http://dx.doi.org/10.1016/j.jcat.2014.09.014
0
021-9517/Ó 2014 Elsevier Inc. All rights reserved.

3
4
W. Luo et al. / Journal of Catalysis 320 (2014) 33–41
at 523 K after 6 h of reaction, after a comprehensive optimization
of the reaction conditions [10]. Also for this catalyst system, acid
site loss was found to be an issue. These first, few examples
nonetheless show that the direct conversion is feasible and the
promising results warrant further exploration.
2. Experimental section
2.1. Catalyst preparation
Six distinct 1 wt% Ru/H-ZSM5 catalysts were prepared via a wet
impregnation method. The required amount of precursor solution
(22 mL) was prepared with deionized water in a 50-mL round bot-
tom flask equipped with a magnetic stirring bar. The round bottom
flask was submerged in a temperature controlled oil bath, and the
mixture was then agitated at 303 K using a hot plate stirrer
(1000 rpm). After 10 min of stirring, the ZSM5 support was slowly
added to the impregnation solution, with agitation speed 1000 rpm
at 303 K for about 1 h. After mixing of the support with the impreg-
nation solution, the temperature was raised to 358 K and kept at
that temperature for 12 h, allowing evaporation of the solvent at
a desirable rate. A fine catalyst powder was obtained of homoge-
neous color, which was then reduced directly at 723 K with a heat-
ing ramp 2 K/min under a flow of 10 vol.% H /N for 6 h. The
The efficient conversion of LA to PA requires a careful balance
to be struck between the number, strength, and location of the
acid and hydrogenation sites, key parameters that in principle
can be controlled by careful catalyst synthesis. Herein, we further
explore 1 wt% Ru/H-ZSM5 as a catalyst for the direct conversion
of LA to PA and study the influence of catalyst preparation on
activity and selectivity. H-ZSM5 was selected as the acidic sup-
port, given its higher intrinsic acid strength than H–Y [11] and
the fact that acid leaching (dealumination) was previously found
to be more limited with H-ZSM5 than with H-b under the
demanding conditions required for the LA to PA conversion [9].
In the sequential hydrodeoxygenation of LA to PA (Scheme 1),
LA can be first converted to GVL, either via ring closure and
subsequent hydrogenation involving angelicalactone (AL) as
intermediate or via hydrogenation first to 4-hydroxypentanoic
acid (HPA) followed by ring closure; GVL can subsequently be
ring-opened to PEA and finally hydrogenated to PA; an alterna-
tive route differs in the fate of the first intermediate 4-hydroxy-
pentanoic acid (HPA), which can in principle also be directly
dehydrated to PEA, followed finally by hydrogenation to PA
2
2
catalyst sample was subsequently cooled rapidly to room temper-
ature and used without any further modification.
The six 1 wt% Ru/H-ZSM5 catalysts are labeled according to
their preparation method. The first indicator denotes the ruthe-
nium precursor used with ‘Cl’/‘NH ’ indicating RuCl (99.9%, Acros
3
3
Chemicals)/Ru(NH ) Cl (99%, ABCR), respectively; the second indi-
3
6
3
cator, ‘A’/‘H’, indicates if the cation of the ZSM5 support was either
+
+
[
12,13]. The strong acidity of the zeolite would be important
NH or H prior to impregnation; the last indicator depicts the Si/Al
4
for both routes. Putting the strong acid sites and hydrogenation
metal in close proximity might furthermore favor the second
LA–HPA–PEA–PA route and thus avoid the most difficult step of
both sequences, i.e., GVL ring-opening. To achieve this, controlled
deposition/exchange of ruthenium, especially into the zeolite
pores is required.
ratio of ZSM5. Four ZSM5 materials were used in this study with Si/
2
Al ratios of 11.5 (CBV2314, S
BET
= 425 m /g), 25 (CBV5524G,
= 425 m /g), 140 (CBV28014,
2
2
SBET = 425 m /g), 40 (CBV8014, S
BET
2
SBET = 400 m /g; all from Zeolyst).
Two 1 wt% Ru/H-ZSM5 catalysts, Ru/H-ZSM5(Cl, H, 11.5) and
Ru/H-ZSM5(Cl, A, 11.5) were prepared with RuCl as the ruthenium
3
Here, we report on the influence of preparation method and
zeolite composition on the catalytic performance and stability of
precursor and CBV2314 as the support; for Ru/H-ZSM5(Cl, H, 11.5),
the ammonium form of ZSM5 was converted to H-ZSM5 before
1
wt% Ru/H-ZSM5 catalysts for the LA to PA conversion. The influ-
impregnation by heating NH -ZSM5 at 1 K/min to 393 K for 1 h
4
ence of the type of extra-framework cations in ZSM5 before wet
impregnation, ruthenium precursor, and variation of Si/Al ratio
was examined on (1) metal dispersion and distribution, (2) H-
ZSM5 acidity, and (3) catalytic performance of the Ru/H-ZSM5
catalysts in the direct, one-pot conversion of LA to PA in dioxane
at 473 K. A much improved bifunctional catalyst was developed
by a new, simple synthesis method to yield 91.3% PA within a reac-
tion time of 10 h. A comparison of catalysts of varying acidity
showed acid strength to be key to the efficient conversion of LA
to PA. Finally, insight into the activity, stability, and deactivation
of the new 1 wt% Ru/H-ZSM5 catalysts is provided.
and at 2 K/min to 823 K for 4 h; the other four 1 wt% Ru/H-ZSM5
catalysts were prepared with Ru(NH ) Cl as precursor and used
3
6
3
the ammonium form of the four ZSM5 zeolites of different Si/Al
ratio: Ru/H-ZSM5(NH , A, 11.5), Ru/H-ZSM5(NH , A, 25), Ru/H-
3
3
ZSM5(NH , A, 40), and Ru/H-ZSM5(NH , A, 140).
3
3
For comparison, a parent ZSM5 (11.5) sample (CBV2314),
without having been impregnated, was similarly treated at 723 K
with a heating ramp of 2 K/min under a flow of 10 vol.% H /N2
2
for 6 h. Catalyst regeneration of spent Ru/H-ZSM5(NH , A, 11.5)
3
was performed by heating the spent catalyst to 723 K with a heat-
ing ramp of 2 K/min under a flow of 10 vol.% H /N for 4 h.
2
2
angelicalactone (AL)
γ-valerolactone
O
O
-
H2O
(
GVL)
OH
O
H2
O
H2
-
H2O
OH
O
methyltetrahydrofuran
(MTHF)
O
1,4-pentanediol (PD)
OH
O
-H O
2
O
levulinic acid (LA) H2
2
-H O
O
O
H2
OH
OH
OH
OH
-hydroxy-pentanoic acid (HPA)
pentenoic acid isomers
PEA)
pentanoic acid
4
(
(PA)
-
H O
ROH
2
O
OR
pentanoic ester
(
PE)
Target Products
Scheme 1. Levulinic acid hydrodeoxygenation platform.

W. Luo et al. / Journal of Catalysis 320 (2014) 33–41
35
2
2
.2. Catalyst characterization
acid (2.5 g, 21.5 mmol) in dioxane (22.5 g) over 1 wt% Ru/ZSM5
1
1.5 catalysts (0.5 g).
.2.1. Transmission electron microscopy (TEM)
The reaction products were analyzed using a Varian gas chro-
The bright field and high angle annular dark field (HAADF) TEM
matograph equipped with a VF-5 ms capillary column and FID
detector. Products were identified with a Shimadzu GCÀMS with
a VF-5 ms capillary column. The gas phase reaction products were
analyzed by an online dual channel Varian CP4900 micro-GC
images were obtained using a Tecnai 20FEG microscope operating
at 200 kV. Ruthenium particle diameters of more than 200 parti-
cles for each sample were measured using the iTEM software (soft
Imaging System GmbH). For non-symmetrical particle shapes, both
the largest and shortest diameter were measured to obtain an
average value.
equipped with a CO
X
column and TCD detector, for analysis of
H
2
, CO , CO, and CH .
2
4
3
. Results and discussion
2.2.2. N
2
physisorption
N
2
physisorption isotherms were recorded to determine surface
A series of six distinct 1 wt% Ru/H-ZSM5 catalysts were pre-
pared, by varying the cation form and Si/Al ratio of the zeolite as
areas and pore volumes using a Micromeritics Tristar 3000 set-up
operating at 77 K. All samples were outgassed for 12 h at 473 K in a
nitrogen flow prior to the physisorption measurements. BET
surface areas were determined using 10 points between 0.06 and
.25. Micropore volumes (cm /g) were determined by t-plot
well as the ruthenium precursor salt, and tested in the hydrodeox-
+
ygenation of LA. The influence of the extra-framework cation (NH
4
+
vs. H ) of the ZSM5 support with a Si/Al ratio of 11.5 and impreg-
nation salt (RuCl vs. Ru(NH Cl ) on the catalysts’ physicochem-
3
0
3
3
)
6
3
analysis for t between 3.5 and 5.0 Å to ensure inclusion of the
minimum required pressure points.
ical properties and performance were assessed by comparison of
Ru/H-ZSM5(Cl, H, 11.5), Ru/H-ZSM5(Cl, A, 11.5), and Ru/H-
3
ZSM5(NH , A, 11.5). Four different Si/Al ratios varying from 11.5
2
.2.3. Temperature-programmed desorption of ammonia (TPD-NH
Catalyst acidity was investigated by TPD-NH under He flow
25 mL/min) using a Micromeritics AutoChem II equipped with a
TCD detector. 0.15–0.2 g of catalyst was loaded and dried at
73 K for 1 h, after which the sample was cooled down to 373 K.
3
)
to 40 were examined with Ru/H-ZSM5(NH , A, 11.5), Ru/H-
3
3
ZSM5(NH , A, 25), Ru/H-ZSM5(NH , A, 40), and Ru/H-ZSM5(NH ,
3
3
3
(
3 6 3
A, 140), all prepared with the Ru(NH ) Cl precursor and the zeo-
lites in the ammonium form. As ruthenium is known to form the
8
volatile oxides RuO2 and RuO , and to severely sinter when con-
4
Subsequently, pulses of ammonia were introduced up to saturation
of the sample. The temperature-programmed desorption was per-
formed up to 873 K, with a heating ramp of 5 K/min. The total
tacted with oxygen above 373 K [14], the traditional calcination
step was omitted and the wet impregnation step was followed
directly by a prolonged reduction step of 6 h, in order to fully
decompose the ruthenium precursor and prepare catalysts with a
better dispersion of ruthenium.
3
number of acid sites (mmol NH /gram zeolite) was determined
from the total amount of desorbed ammonia.
2
.2.4. Thermal gravimetric analysis (TGA)
TGA-MS measurements of the spent catalysts were performed
3.1. Physicochemical characterization of the catalysts
with a Perkin–Elmer Pyris 1 apparatus. The sample was initially
heated to 423 K for 1 h with a temperature ramp of 10 K/min in
a 20 mL/min flow of argon to exclude physisorbed water and
acetone, followed by a ramp of 5 K/min to 973 K in a 10 mL/min
flow of oxygen to burn off the coke. Analysis was performed with
a quadrupole Pfeiffer Omnistar mass spectrometer, which was con-
nected to the outlet of the TGA apparatus. Ion currents were
recorded for m/z values of 18 and 44.
The physicochemical properties of the six catalysts under study
are given in Table 1. Ruthenium particle sizes, falling in the range
of 1.7–4.9 nm, and dispersions were determined by TEM (Fig. S1).
The number and strength of acid sites of the parent zeolite ZSM5
(
11.5) and the six zeolite-supported catalysts were determined
by temperature-programmed desorption of ammonia (TPD-NH
Fig. 1, Table 2). The TPD-NH traces could be divided into two
3
,
3
clearly distinguishable parts, with desorption being observed at a
low-temperature range (LT) from 430 to 580 K and a high-
temperature range (HT) from 590 to 740 K. These LT and HT ranges
correspond to weak and strong acid sites, respectively [15,16].
The metal dispersion as well as the acidity of the catalyst was
found to depend strongly on the cation in the zeolite used for the
impregnation step. Ru/H-ZSM5(Cl, H, 11.5), for instance, showed
a ruthenium dispersion of 0.39 (average particle size of 3.3 nm),
which was slightly better than that of Ru/H-ZSM5(Cl, A, 11.5)
0.26 (particle size of 4.9 nm). In both cases, ruthenium particles
can be mainly found at the external surface of the zeolite. While
giving a better Ru dispersion, a significant decrease in strong acid
sites (HT) with concomitant large increase in weak acid sites (LT)
2.3. Catalyst activity and stability testing
LA hydrodeoxygenation reactions were conducted in a 100 mL
Parr batch autoclave reactor equipped with a thermocouple, a
pressure transducer and gauge and overhead stirring. In a typical
run, the batch reactor was loaded with a 10 wt% levulinic acid solu-
tion (6.0 g, 51.7 mmol) in dioxane (54 g) and the 1 wt% Ru catalyst
(
0.6 g). The reactor was sealed, purged three times with argon,
heated to 473 K and subsequently charged with H to 40 bar. This
2
was taken as the starting point of the reaction. Reactions were run
for 10 h with a stirring speed of 1600 rpm; this stirring speed was
previously shown to be sufficient for avoiding external mass trans-
fer limitations, guarantying the reactions to be operated in the
kinetic regime [9]. After the reaction, the autoclave was cooled to
room temperature, H was released and 2 wt% anisole was added
2
as internal standard. The catalyst was separated by centrifugation,
filtration, and finally washed with acetone.
The experiments that aimed at higher PA yields as well as the
stability tests were run at a higher catalyst loading. These reactions
were conducted in a 50 mL Parr batch autoclave reactor at 473 K
for 10 h using a hydrogen pressure of 40 bar and a stirring speed
of 1600 rpm. These runs were performed with 10 wt% levulinic
+
was seen when H -type ZSM5 was used for the impregnation
(Fig. 1). This is undesirable as strong acid sites are essential to effi-
cient PA production. Ru/H-ZSM5(Cl, H, 11.5) actually contained the
largest amount of LT (0.23 mmol/gcat) and lowest amount of HT
(0.34 mmol/gcat) of the three catalysts made with the Si/Al = 11.5
ZSM5 support (Table 2, entry 2–4). If one takes into account that
+
a sample of bare NH
4
-ZSM5 zeolite subjected to the same reduc-
tion conditions as the other ruthenium-loaded zeolites contains
0.57 HT and 0.04 mmol/gcat LT (Table 2, entry 1), then it is clear
+
from the results above that conversion of the zeolite from NH
4
to
+
H form prior to impregnation is not beneficial, given the resulting

3
6
W. Luo et al. / Journal of Catalysis 320 (2014) 33–41
Table 1
Physicochemical properties of fresh Ru/H-ZSM5 and spent catalysts after reaction at 473 K, 40 bar H
2
after 10 h in dioxane.
2
Catalyst
BET (m /g)
Micropore surface Micropore
Pore volume
(cm /g)
Average Ru particle Ru dispersion
size (nm)
Coke contents
(wt%)e
2
3
a
3
b
c
ZSM5d
area (m /g)
volume (cm /g)
1
Non-impregnated ZSM5 (11.5) 379
295
0.14
0.22
–
–
–
Ru/H-ZSM5
Fresh Spent Fresh
Spent
Fresh
Spent
Fresh Spent Fresh
Spent
Fresh Spent
2
3
4
5
6
7
(Cl, H, 11.5)
(Cl, A, 11.5)
333
371
324
377
390
363
126
53
137
208
267
337
247
266
251
242
237
221
112
51
107
137
161
168
0.12
0.13
0.12
0.12
0.12
0.11
0.06
0.03
0.05
0.07
0.08
0.08
0.20
0.23
0.19
0.26
0.27
0.21
0.09
0.05
0.06
0.18
0.21
0.20
3.3 ± 1.4 6.6 ± 2.6 0.39
4.9 ± 1.4 6.7 ± 2.2 0.26
1.7 ± 0.4 3.8 ± 0.9 0.76
4.3 ± 1.5 5.0 ± 1.9 0.30
4.8 ± 1.6 6.1 ± 2.0 0.27
6.2 ± 3.2 7.4 ± 1.9 0.21
0.20
0.19
0.34
0.26
0.21
0.17
8.0
9.2
7.6
5.8
4.0
1.6
(NH
(NH
(NH
(NH
3
3
3
3
, A, 11.5)
, A, 25)
, A, 40)
, A, 140)
a
b
c
Data obtained by the t-plot method.
Data obtained by single point adsorption.
Data obtained by TEM.
d
is bulk metal atomic density of Ru (13.65 Â 10À3 nm ), a
3
Data estimated by TEM: D = 6 * (
m
m
/a
m
)/d. Where
m
m
m
is the surface area occupied by an atom on a polycrys-
À2
2
talline surface of Ru (6.35 Â 10 nm ), d is the cluster size of Ru metal.
e
Data determined by TGA.
small ruthenium clusters could be obtained by ion exchange of
+
Non-impregnated
ZSM5(11.5)
Na -Y zeolites with aqueous solutions of Ru(NH Cl
3
)
6
3
[14,19–21].
The higher dispersion obtained with Ru(NH Cl was thought to
3
)
6
3
be the result of strong interaction between negatively charged por-
ous framework of ZSM5 and the ruthenium cations formed upon
(Cl, A, 11.5)
(
Cl, H, 11.5)
dissolution of Ru(NH
sions and larger particle sizes observed outside of ZSM5 with RuCl
are the result of weaker interactions between metal salt and sup-
port [18]. An additional, important benefit of using Ru(NH Cl
3
is that with this precursor, the largest amount of strong acid sites
is preserved (0.48 mmol/g). Indeed, the temperature of maximum
ammonia desorption in the HT region is at 686 K, which is the clos-
est to the maximum of 719 K observed for the original, non-
3 6 3
) Cl in water; conversely, the lower disper-
3
(
NH , A, 11.5)
3
)
3 6
(
NH , A, 25)
3
(
NH , A, 40)
3
(
NH , A, 140)
3
400
500
600
700
800
3 6 3
impregnated zeolite (Table 2, entry 4). The use of Ru(NH ) Cl thus
Temperature (K)
improved metal dispersion as well as efficiently preserved the
amount and strength of strong acid sites on ZSM5.
3
Fig. 1. TPD-NH profiles of the parent ZSM5 and the six 1 wt% ZSM5-supported
catalysts.
Variation of the zeolite’s Si/Al ratio was also found to affect
ruthenium dispersion. Expectedly, the total number of acid sites
and fraction of strong acid sites of the 1 wt% Ru/H-ZSM5 catalysts
decreased with increasing the Si/Al ratio from 11.5 to 140 (Table 2,
entry 4–7). The average ruthenium particle size in turn was found
to gradually increase from 1.7 to 6.2 nm with increasing Si/Al ratio
large drop in acid strength. This transition of strong acid sites to
weak ones during the pretreatment process at 823 K was attrib-
uted to the extraction of framework aluminum species (Fig. 1)
[
17]. The use of NH+
4
-ZSM5 is therefore preferred for the synthesis
(Table 1, entry 4–7). A similar increase in average particle size from
of the bifunctional catalyst.
1
.4 to 6.1 nm upon increasing the Si/Al ratio from 11.5 to 500 was
The Ru dispersion could be further improved by using
also seen by You et al. [22] for platinum supported on ZSM5. The
difference in the distribution of strong and weak acid sites might
be responsible for the observed differences in the Ru dispersion,
with those zeolites richer in HT sites stabilizing smaller Ru clusters.
For Pt/H-ZSM5 it is known, for instance, that highly acidic zeolites
lead to higher Pt dispersion, with computational studies pointing
at the stabilizing interaction between Pt and a Brønsted acidic pro-
ton [23].
+
Ru(NH
3
)
6
Cl
3
as precursor on NH
4
-ZSM5 to give Ru/H-ZSM5(NH
3
,
A, 11.5). This preparation method gave the smallest average
particle size of 1.7 nm of all the Ru/H-ZSM5 catalysts, with con-
comitant significant increase in Ru dispersion to 0.76 (Table 1,
entry 4). Previous studies also showed that large ruthenium parti-
cles were obtained by impregnation of X, L, and ZSM5 zeolites with
2
ruthenium chloride after calcination and H -reduction [18], while
Table 2
3
Amount and type of acid sites as determined by NH -TPD.
Catalyst
LT region T of maximum
desorption (K)
Amount of weak
acid sites (mmol/gcat
HT region T of maximum
desorption (K)
Amount of strong
acid sites (mmol/gcat
Total acidity
)
)
(mmol/gcat
)
1
Non-impregnated ZSM5 (11.5)
559
0.04
719
0.57
0.61
Fresh Ru/H-ZSM5
(Cl, H, 11.5)
(Cl, A, 11.5)
2
3
4
5
6
7
544
575
517
476
467
434
0.23
0.10
0.12
0.20
0.25
0.07
657
655
686
615
636
597
0.34
0.44
0.48
0.31
0.24
0.07
0.57
0.54
0.60
0.51
0.49
0.14
(NH
(NH
(NH
(NH
3
3
3
3
, A, 11.5)
, A, 25)
, A, 40)
, A, 140)

W. Luo et al. / Journal of Catalysis 320 (2014) 33–41
37
Additional insights into the distribution of ruthenium might be
obtained from any observed decrease in pore volume and the num-
ber of acid sites, as these changes give an indication of the fraction
of small ruthenium clusters deposited/exchanged into the ZSM5
pores, i.e., those that are difficult to be visualized in TEM. Zhao
et al., for instance, previously reported on a decrease in pore vol-
ume and acid density of ZSM5 when a significant fraction of Ni
was located inside the ZSM5 pores [24]. A similar decrease,
reported by You et al. after introduction of platinum [22] on
ZSM5, was again attributed to the fraction of Pt deposited inside
the ZSM5 pores. The surface areas and pore volumes of the Ru cat-
butanol as decomposition products, among others. The secondary
PE products that are formed with these solvent-derived alcohols
(up to 25 mol% max.) should be considered as PA and are included
in the total PA yields reported below. The activity of the catalysts is
reported in Table 3 in terms of PA productivity, expressed as
À1 À1
molPA
g
Ru
h
. The time online concentration profiles of substrate
and products are depicted in Fig. 3. Trace amounts of MTHF were
found, but no PD was observed in any of the reactions, indicating
that the LA–GVL–MTHF route is insignificant over the 1 wt% Ru/H-
ZSM5 catalysts under the applied conditions. As can be seen from
the time profiles, GVL is initially formed as the primary product, with
selectivity to PA increasing over time as a result of the consecutive
hydrodeoxygenation reaction.
The variations in metal location, dispersion, and acidity as a
result of the different preparation methods are clearly reflected
in the catalytic activity and PA productivity (Fig 2a–c). For exam-
2
alysts as determined by N physisorption are listed in Table 1.
Compared to the non-impregnated zeolite ZSM5(11.5), wet
impregnation of ZSM5(11.5) with Ru with the different precursors
resulted in a drop in pore volume and acid density in most cases
(Tables 1 and 2), indicating the formation of small Ru particles
inside the zeolite pores. The pore volume of ZSM5(11.5) dropped
from 0.22 to 0.20 cm /g and 0.19 cm /g for Ru/H-ZSM5(Cl, H,
ple, Ru/H-ZSM5(Cl, A, 11.5) gave a higher PA yield of 23.3 mol%
3
3
À1 À1
PA and productivity of 0.451 molPA
11.5) did (7.2 mol% and 0.250 molPA
g
g
Ru
h
h
than Ru/H-ZSM5(Cl, H,
) after 10 h of reaction
À1 À1
1
3
1.5) and Ru/H-ZSM5(NH , A, 11.5), respectively (Table 1, entry
Ru
1
, 2, 4), whereas no loss in pore volume was observed for Ru/H-
(Fig. 2a, b and Table 3, entry 1, 2). It should be noted that Ru/H-
ZSM5(Cl, A, 11.5) has more strong acid sites but a lower metal
dispersion than Ru/H-ZSM5(Cl, H, 11.5). This implies that acidity
plays a more important role than ruthenium dispersion in PA pro-
duction, which is in an agreement with our previous study in
which the acid-catalyzed GVL ring-opening was shown to be
3
ZSM5(Cl, A, 11.5). The 0.02 cm /g decrease in pore volume
observed for Ru/H-ZSM5(Cl, H, 11.5) might also have an additional
origin. A significant increase in the fraction of weak acid sites is
seen for this catalyst, which points at a partial pore blockage by
migration of Al to extra-framework sites upon transition of the
NH
in pore volume and much better ruthenium dispersion for Ru/H-
ZSM5(NH , A, 11.5) indicate that this catalyst has the largest frac-
tion of Ru particles deposited inside the zeolite channel system.
+
+
4
-ZSM5 to the H -ZSM5 prior to impregnation [17,25]. The drop
3
rate-limiting [9]. Ru/H-ZSM5(NH , A, 11.5) gave a PA yield of
57.4 mol%, showing that the change of ruthenium precursor in
the preparation method resulted in a rather substantial increase
in PA production. More importantly, the rate of PA formation does
not seem to slow down, even after 10 h of reaction time; this in
contrast to Ru/H-ZSM5(Cl, H, 11.5) which already shows some
signs of deactivation after 3 h of reaction (Fig. 2a and c). The
3
3.2. Catalytic performance
3 6 3
use of Ru(NH ) Cl as the precursor thus clearly improves catalyst
Catalytic activity of the six 1 wt% Ru/H-ZSM5 catalysts was ini-
performance in terms of PA yield. Based on the characterization
data discussed above, this can be attributed to the good preserva-
tion of strong acid sites and the improved deposition of ruthenium
particles also inside the zeolite pores leading to close proximity of
the active sites in the bifunctional catalyst.
2
tially compared at 473 K and 40 bar H with a 10 wt% solution of LA
in dioxane and a stirrer speed of 1600 rpm, conditions that are
identical to those of our previous study. As reported, dioxane
decomposition takes place to a limited extent in the presence of
strongly acidic catalysts, yielding ethanol, 2-ethoxyethanol, and
Table 3
One-pot hydrodeoxygenation of levulinic acid (LA) to pentanoic acid (PA).
PA yield (%)a
Productivity (molPA
g
À1 À1
h
)
Catalyst
Catalyst screeningb
C-balance (%)
LA conv. (%)
GVL yield (%)
1
2
3
4
5
6
(Cl, A, 11.5)
(Cl, H, 11.5)
93.3
98.8
88.7
76.4
86.8
88.0
100
100
100
97.9
97.5
72.2
70.0
91.6
31.3
50.1
68.0
58.2
23.3 (4.7)
7.2 (0.9)
57.4 (17.7)
24.2 (8.0)
16.3 (4.5)
1.0 (0.0)
0.451
0.169
0.629
0.135
0.131
0.009
(NH
(NH
(NH
(NH
3
3
3
3
, A, 11.5)d
, A, 25)
, A, 40)
, A, 140)
PA yield optimization for (NH
3
, A, 11.5) and catalyst reusec
d
7
8
9
1st run
99.0
94.3
91.1
100
100
100
7.7
21.3
9.9
91.3 (25.0)
66.4 (17.7)
81.2 (25.2)
1.042
0.547
0.685
2nd run
3rd run after coke burn-off
Comparison with literature results
e
1
1
1
0
1
2
(NH
1 wt% Ru/H-ZSM5
5 wt% Ru/SBA-SO
3
, A, 11.5)
93.4
96.0
100
100
100
100
28.9
50.2
4.0
64.5 (18.9)
45.8 (16.3)
90.0
1.157
0.822
e
f
0.060g
3
H
h
0.160
1
3g
5 wt% Ru/H-ZSM5
93.0
99.0
43.0
49.0
0.033
a
b
c
The total PA yield is given; this value includes the PE yield which is given in parentheses.
Conditions: 1 wt% Ru/H-ZSM5, 51.7 mmol of LA, 0.6 g of catalyst, 54 g dioxane, 10 h, 473 K and 40 bar H
Conditions: 1 wt% Ru/H-ZSM5, 21.5 mmol of LA, 0.5 g of catalyst, 22.5 g dioxane, 10 h, 473 K and 40 bar H
See Fig. S2 for dependence of productivity on catalyst/substrate ratio.
2
.
2
.
d
e
f
Conditions: 21.5 mmol of LA, 0.3 g of catalyst, 22.5 g dioxane, 4 h, 473 K and 40 bar H
2
. See Ref. [9] for details.
Conditions: 4 mmol of LA, 0.2 g of catalyst, 10 mL of ethanol, 6 h, 523 K and 40 bar H
PA productivity obtained after 6 h reaction time at 523 K.
2
. See Ref. [10] for details.
g
h
Highest PA productivity obtained at 513 K. See Ref. [10] for details.

3
8
W. Luo et al. / Journal of Catalysis 320 (2014) 33–41
a. Ru/H-ZSM5 (Cl, H, 11.5)
1
1
1
00
100
80
60
40
20
0
b. Ru/H-ZSM5 (Cl, A, 11.5)
80
60
40
20
0
LA
GVL
PA+PE
LA
GVL
PA+PE
0
0
0
100 200 300 400 500 600
Time (min)
0
0
0
100 200 300 400 500 600
Time (min)
00
c. Ru/H-ZSM5(NH , A, 11.5)
100
80
60
40
20
0
d. Ru/H-ZSM5(NH , A, 25)
3
3
LA
GVL
PA+PE
LA
GVL
PA+PE
80
60
40
20
0
100 200 300 400 500 600
Time (min)
100 200 300 400 500 600
Time (min)
e. Ru/H-ZSM5 (NH , A, 40)
00
3
100
80
60
40
20
0
f. Ru/H-ZSM5 (NH , A, 140)
3
LA
LA
GVL
PA
GVL
PA+PE
8
6
4
2
0
0
0
0
0
100 200 300 400 500 600
Time (min)
100 200 300 400 500 600
Time (min)
Fig. 2. Time profiles of the catalytic hydrodeoxygenation of 10 wt% LA in dioxane at 473 K and 40 bar H
acid; GVL: -valerolactone; PA + PE: pentanoic acid and its esters.
2
pressure, 1 wt% Ru/H-ZSM5, stirrer speed of 1600 rpm. LA: levulinic
c
Clear differences in selectivity and activity are also observed
upon variation of the Si/Al ratio of ZSM5 (Fig. 2d–f and Table 3
entry 3–6). Full LA conversion was achieved with Ru/H-ZSM5
is at least six fold higher than the highest PA productivity of
À1 À1
0.16 molPA
g
Ru
h
recently reported by Pan et al. for the production
H, a reaction which was
run at the higher temperature of 513 K (Table 3, entry 12) [10]. A
4 h run with 1 wt% Ru/H-ZSM5(NH , A, 11.5) allowed for a direct
comparison with our previously reported Ru/H-ZSM5 catalyst
(Table 3, entry 10 and 11), which was prepared with RuNO(NO
of PA + PE from LA over 5 wt% Ru/SBA-SO
3
(
3
NH , A, 11.5) after 5 h (Fig. 2c), while 2.1, 2.5, and 27.8 mol%
of LA were still left after 10 h reaction with the catalysts with
Si/Al ratios of 25, 40, and 140, respectively. PA productivity of
the Ru/H-ZSM5 catalysts decreased accordingly, with Ru/H-
3
3 3
)
ZSM5(NH
3
, A, 140) having produced hardly any PA after 10 h.
as precursor via a wet impregnation method followed by both
calcination and reduction [9]. The new results show that an
These results are in line with Pan et al. who also included a 5
wt% Ru/H-ZSM5 catalyst with a high Si/Al ratio of 50 in their
screening studies and observed
improved yield and productivity can be obtained with the new
À1 À1
a
low PA productivity of
catalyst (64.5 mol% PA/1.157 molPA
g
Ru
h
vs. 45.8 mol% PA/
À1 À1
À1 À1
0
.033 molPA
g
Ru
h
at the higher reaction temperature of 513 K
0.822 molPA
g
Ru
h
).
(Table 3, entry 13) [10].
The excellent activity of the 1 wt% Ru/H-ZSM5(NH , A, 11.5)
3
The PA yield can be further improved with a slight increase in
catalyst in the direct, one-pot conversion of LA to PA can be
mainly attributed to the improved dispersion of small ruthenium
particles and the number and accessibility of the strong acid
sites that are required for this reaction. The time online profiles
catalyst loading under otherwise standard conditions of 473 K
and 40 bar H (Fig. S2). A PA yield of 91.3 mol% could thus be
achieved with 1 wt% Ru/ZSM5(NH , A, 11.5) after a 10 h reaction
2
3
with an excellent mass balance of 99.0% (Table 3, entry 7), corre-
with Ru/H-ZSM5(NH
GVL is detected as intermediate in the reaction, which indicates
3
, A, 11.5) show (Figs. 2c and 3a) that only
À1 À1
sponding to a productivity of 1.042 molPA
g
Ru
h
. This productivity

W. Luo et al. / Journal of Catalysis 320 (2014) 33–41
39
st
3.3. Catalyst stability and reuse
100
a. 1 run of Ru/H-ZSM5(NH , A, 11.5)
3
The spent catalysts were characterized to detect any changes in
8
6
4
2
0
0
0
0
0
GVL
either metal phase or zeolite support. The average ruthenium
particle sizes are given in Table 1, with TEM pictures of the spent
catalysts being depicted in Fig. S1. For all catalysts, different
extents of sintering occurred as evidenced by the slight increase
in ruthenium particle size after 10 h of reaction. We already previ-
ously observed that leaching of ruthenium is very limited under
the employed reaction conditions [9].
PA+PE
The N
surface area of 62% and 86% for the spent Ru/H-ZSM5(Cl, H,
1.5) and (Cl, A, 11.5) catalysts, as a result of 8.0 wt% and 9.2
wt% coke build-up on the catalyst surface during reaction, as
determined by TGA(-MS). Ru/H-ZSM5(NH , A, 11.5), the catalyst
2
physisorption data (Table 1) show a significant drop in
0
100
200
300
400
500
600
1
Time (min)
nd
3
1
00
b. 2 run of Ru/H-ZSM5(NH , A, 11.5)
3
giving the best PA yield, suffered a little less from coke forma-
tion (7.6 wt%). Upon increase of the Si/Al ratio, the amount of
coke deposited was found to decrease, consistent with the drop
in support acidity.
GVL
PA+PE
8
6
4
2
0
0
0
0
0
The time online concentration profiles upon reuse of 1 wt% Ru/
H-ZSM5(NH
3
, A, 11.5) are shown in Fig. 3. The 1 wt% Ru/H-
ZSM5(NH , A, 11.5) catalyst was assessed in two consecutive runs,
3
then followed by a third run after a regeneration step involving
coke burn-off at 723 K under a dilute hydrogen flow. While com-
plete LA conversion was observed within 1 h reaction time for all
three runs, the yield of PA decreased from 91.3 to 66.4 mol% upon
0
100
200
300
400
500
600
two consecutive 10 h runs, with a concomitant drop in PA produc-
Time (min)
À1 À1
tivity from 1.042 to 0.547 molPA
g
Ru
h
(Table 3). From the time-
profile (Fig. 3b), it can be seen that not the conversion to GVL, but
rather the GVL-PA step was slowed down, again pointing to GVL
ring-opening step to be most difficult. Upon regeneration, the spent
catalyst recovered after two runs gave an increase in PA yield to
81.2 mol% in the third run, indicating that PA production activity
of the catalyst could be almost completely restored by simple coke
burn-off.
rd
1
00
c. 3 run after regeneration
8
6
4
2
0
0
0
0
0
GVL
PA+PE
After regeneration, a bimodal particle size distribution was
observed for Ru/H-ZSM5(NH
dark field scanning TEM imaging (HAADF), showing small particles
<2 nm) as well as larger particles (ranging from 5 to 14 nm)
3
, A, 11.5) with high angle annular
(
located on the external surface of ZSM5 (Fig. S4). This bimodal
distribution points at a significant fraction of ruthenium being
originally located inside the zeolite pores.
0
100
200
300
400
500
600
Time (min)
While some sintering does take place, the fact that PA yields
can be almost completely restored shows that the strong acidity
of the ZSM5 must be mostly recovered upon regeneration. The
TPD measurements show that the spent catalysts in three con-
secutive runs maintain the total acidity of the fresh one (Table 4),
suggesting that Al leaching is very limited under the applied
reaction conditions and that the coke that is deposited does
not block the accessibility of the acid sites for ammonia. The
overall strength of the acid sites is reduced upon the first and
second run, however, as some of the strong acid sites are con-
verted to weak acid sites and the remaining strong sites are
on average less strong (Table 4, entry 1–3 and Fig. S3). The spent
catalyst after the 3rd run shows a number of strong acid sites
that is larger than that of the spent catalyst after the 2nd run,
and almost the same as the spent catalyst after the 1st run
(Table 4, entry 2, 3, 5). Compared to the spent catalyst of the
first run, the shift of the HT peak maximum of the spent regen-
erated catalyst from 670 K to 651 K (Table 4, entry 2, 5) sug-
Fig. 3. Time profiles of the catalytic hydrodeoxygenation of 10 wt% LA in dioxane at
73 K, 40 bar H , 10 h reaction time, with wt% Ru/H-ZSM5(NH , A, 11.5), stirrer
speed of 1600 rpm. GVL: -valerolactone; PA: pentanoic acid and its esters; LA
4
2
3
c
conversion was 100% for all three runs within 1 h and is therefore not depicted.
that PA is dominantly formed via the LA–GVL–PEA–PA route. PA
is, however, already detected at low LA conversions, which
might indicate that the LA–HPA–PEA–PA route, which avoids
the most difficult GVL–PEA step, also occurs, possibly facilitated
by the close proximity of the ruthenium and strong acid sites in
the H-ZSM5 pores. Although HPA is hard to detect due to its
instability under the applied conditions, it has been reported
that reactive intermediates can be stabilized in zeolites pores
via confinement and nest effects [26–29]. The HPA intermediate
might be stabilized by a strong acid site in the ZSM5 pores and
subsequently directly dehydrated to PEA, rather than it being
ring-closed to GVL. Dumesic et al. previously proposed the HPA
to PEA route to be thermodynamically relevant at 523 K [13].
Recently, Xin et al. achieved a 95% selectivity of PA by the elec-
trochemical reduction of LA to PA at pH = 0 on a Pb electrode in
an electrocatalytic (flow) cell reactor, also suggesting the same
LA–HPA–PEA–PA pathway [12].
gests, however, that
a
small amount of aluminum is
irreversibly converted to extra-framework species during reac-
tion and regeneration. Additionally, most of the coke (89%) could
be removed in the regeneration step, fully recovering the poros-
ity of the fresh Ru/H-ZSM5(NH , A, 11.5) (Table 4, entry 1, 4).
3
Taken together, these results show that coke formation is the

4
0
W. Luo et al. / Journal of Catalysis 320 (2014) 33–41
main reason for deactivation, as it weakens the acid strength and
blocks the accessibility of ZSM5.
4
. Conclusions
A highly active and selective bifunctional 1 wt% Ru/H-ZSM5 cat-
alyst was developed for the direct, one-pot conversion of levulinic
acid to pentanoic acid. Exploration of various synthesis parame-
ters, including ruthenium precursor salt and cation form of the
zeolite, resulted in a simple preparation method for catalysts with
an improved dispersion of ruthenium, localized also in the pores of
the zeolite, as well as an improved amount and strength of strong
acid sites. The 1 wt% Ru/H-ZSM5 catalyst gives an excellent yield of
9
1.3% PA under relatively mild conditions and shows the highest
À1 À1
productivity of 1.157 molPA
g
Ru
h
reported to date. Better control
over the density of in particular strongly acidic sites was key to
achieve this high productivity, as these sites are needed for the most
difficult step in the tandem conversion of LA to PA. The new prepa-
ration method furthermore improves the proximity between the
hydrogenation and ring-opening/dehydration functions of the cata-
lyst, as a result of the deposition of small Ru clusters into the pores
of ZSM5, which is beneficial for PA production. Deactivation of 1 wt%
3
Ru/H-ZSM5(NH , A, 11.5) is primarily caused by carbon residue
deposition on the strong acid sites.
Acknowledgments
The authors gratefully thank the Smart Mix Program of the
Netherlands Ministry of Economic Affairs and the Netherlands
Ministry of Education, Culture, and Science within the framework
of the CatchBio Program for financial support. Cor van der Spek
(
2
TEM), Nazila Masoud (N physisorption), Rafael de Lima Oliveira
(
N
2
physisorption), and Marjan Versluijs-Helder (TGA), all from
Utrecht University, are acknowledged for the measurements.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2014.09.014.
References
[
[
1] J.J. Bozell, G.R. Petersen, Green Chem. 12 (2010) 539–554.
2] J.J. Bozell, L. Moens, D.C. Elliott, Y. Wang, G.G. Neuenscwander, S.W. Fitzpatrick,
R.J. Bilski, J.L. Jarnefeld, Resour. Conserv. Recy. 28 (2000) 227–239.
[
[
3] D.W. Rackemann, W.O.S. Doherty, Biofuel. Bioprod. Bior. 5 (2011) 198–214.
4] I.T. Horvath, H. Mehdi, V. Fabos, L. Boda, L.T. Mika, Green Chem. 10 (2008) 238–
242.
[
[
5] V. Pace, P. Hoyos, L. Castoldi, P. Domínguez de María, A.R. Alcántara,
ChemSusChem 5 (2012) 1369–1379.
6] J.-P. Lange, R. Price, P.M. Ayoub, J. Louis, L. Petrus, L. Clarke, H. Gosselink,
Angew. Chem. Int. Ed. 49 (2010) 4479–4483.
[
[
[
7] R. Palkovits, Angew. Chem. Int. Ed. 49 (2010) 4336–4338.
8] J.C. Serrano-Ruiz, D. Wang, J.A. Dumesic, Green Chem. 12 (2010) 574–577.
9] W. Luo, U. Deka, A.M. Beale, E.R.H. van Eck, P.C.A. Bruijnincx, B.M. Weckhuysen,
J. Catal. 301 (2013) 175–186.
[
10] T. Pan, J. Deng, Q. Xu, Y. Xu, Q.-X. Guo, Y. Fu, Green Chem. 15 (2013) 2967–
974.
2
[
[
11] F. Lónyi, A. Kovács, J. Valyon, J. Phys. Chem. B 110 (2006) 1711–1721.
12] L. Xin, Z. Zhang, J. Qi, D.J. Chadderdon, Y. Qiu, K.M. Warsko, W. Li,
ChemSusChem 6 (2013) 674–686.
[
[
[
13] J.C. Serrano-Ruiz, R.M. West, J.A. Dumesic, Annu. Rev. Chem. Biomol. Eng. 1
(
2010) 79–100.
14] J.J. Verdonck, P.A. Jacobs, M. Genet, G. Poncelet, J. Chem. Soc., Faraday Trans. 76
1980) 403–416.
15] C.V. Hidalgo, H. Itoh, T. Hattori, M. Niwa, Y. Murakami, J. Catal. 85 (1984) 362–
69.
(
3
[
[
16] F. Lónyi, J. Valyon, Micropor. Mesopor. Mater. 47 (2001) 293–301.
17] S.M. Campbell, D.M. Bibby, J.M. Coddington, R.F. Howe, R.H. Meinhold, J. Catal.
161 (1996) 338–349.
[
18] V.I. Pârvulescu, S. Coman, P. Palade, D. Macovei, C.M. Teodorescu, G. Filoti, R.
Molina, G. Poncelet, F.E. Wagner, Appl. Surf. Sci. 141 (1999) 164–176.

W. Luo et al. / Journal of Catalysis 320 (2014) 33–41
41
[
19] R.M. Dalla Betta, M. Boudart, in: J. Hightower (Ed.), Proceedings of the 5th
International, Congress on Catalysis, Amsterdam, North-Holland, 1973, 1329.
20] D.J. Elliott, J.H. Lunsford, J. Catal. 57 (1979) 11–26.
[25] G.L. Woolery, G.H. Kuehl, H.C. Timken, A.W. Chester, J.C. Vartuli, Zeolites 19
(1997) 288–296.
[26] E.G. Derouane, J. Catal. 100 (1986) 541–544.
[27] E. Kikuchi, H. Nakano, K. Shimomura, Y. Morita, J. Jpn. Pet. Inst. 28 (1985) 210–
213.
[28] M. Brändle, J. Sauer, J. Am. Chem. Soc. 120 (1998) 1556–1570.
[29] A. Vjunov, M.Y. Hu, J. Feng, D.M. Camaioni, D. Mei, J.Z. Hu, C. Zhao, J.A. Lercher,
Angew. Chem. Int. Ed. 53 (2014) 479–482.
[
[
[
21] L.A. Pedersen, J.H. Lunsford, J. Catal. 61 (1980) 39–47.
22] S. You, I. Baek, Y. Kim, K.-E. Jeong, H.-J. Chae, T.-W. Kim, C.-U. Kim, S.-Y. Jeong,
T. Kim, Y.-M. Chung, S.-H. Oh, E. Park, Korean J. Chem. Eng. 28 (2011) 744–750.
23] P. Treesukol, K. Srisuk, J. Limtrakul, T.N. Truong, J. Phys. Chem. B 109 (2005)
[
[
(
1945) 11940–11945.
24] C. Zhao, J.A. Lercher, Angew. Chem. Int. Ed. 51 (2012) 5935–5940.