A study of furfural decarbonylation on K-doped Pd/Al2O 3 catalysts

Article abstract of DOI:10.1016/j.molcata.2010.11.016

Vapor-phase decarbonylation of furfural to furan was performed on K-doped Pd/Al₂O₃ catalysts in a fixed-bed reactor. A series of K-doped catalysts were prepared by impregnation method with various potassium salt precursors and different potassium content. The catalyst evaluation results revealed that the doping of K not only promoted the furfural decarbonylation, but also suppressed the hydrogenation side reaction. Among the investigated catalysts, Pd-K₂CO₃/Al₂O₃ (K% = 8 wt.%) presented the best performance and achieved the furan yields up to 99.5% at 260 °C. The selectivity of 2-methylfuran would approach zero at high K content (K% > 4 wt.%). Furthermore, the catalysts were characterized by BET, SEM, ICP, H₂-TPR, H₂-TPD, H₂-chemisorption, CO-FTIR, furfural-TPSR, and furfural-FTIR experiments. It was shown that the dopant phase had a remarkable effect on the precursor Pd²⁺ and the reduced Pd metal particles. Furfural-TPSR and furfural-FTIR exhibited that K-doped catalysts enhanced the furfural adsorption and promoted the decarbonylation reaction, which was due to the different adsorption mode and intensity of furfural on K-doped and undoped samples. Besides, the decrease of C-O adsorption of furfural (η₂(C,O)-furfural) by K doping is the main reason for suppressing the furfural hydrogenation.

Full text of DOI:10.1016/j.molcata.2010.11.016

Journal of Molecular Catalysis A: Chemical 335 (2011) 71–81
Contents lists available at ScienceDirect
Journal of Molecular Catalysis A: Chemical
journal homepage: www.elsevier.com/locate/molcata
A study of furfural decarbonylation on K-doped Pd/Al2O3 catalysts
Wei Zhanga,b, Yulei Zhua,c,∗, Shasha Niua,b, Yongwang Lia,c
a State Key Laboratory of Coal Conversion, Institute of Coal Chemistry, Chinese Academy of Sciences, P.O. Box 165, Taiyuan 030001, PR China
b Graduate University of the Chinese Academy of Sciences, Beijing 100039, PR China
c Synfuels China Co. Ltd, Taiyuan 030001, PR China
a r t i c l e i n f o
a b s t r a c t
Article history:
Vapor-phase decarbonylation of furfural to furan was performed on K-doped Pd/Al2O3 catalysts in a fixed-
bed reactor. A series of K-doped catalysts were prepared by impregnation method with various potassium
salt precursors and different potassium content. The catalyst evaluation results revealed that the doping of
K not only promoted the furfural decarbonylation, but also suppressed the hydrogenation side reaction.
Among the investigated catalysts, Pd–K2CO3/Al2O3 (K% = 8 wt.%) presented the best performance and
achieved the furan yields up to 99.5% at 260 ◦C. The selectivity of 2-methylfuran would approach zero
at high K content (K% > 4 wt.%). Furthermore, the catalysts were characterized by BET, SEM, ICP, H2-TPR,
H2-TPD, H2-chemisorption, CO-FTIR, furfural-TPSR, and furfural-FTIR experiments. It was shown that the
dopant phase had a remarkable effect on the precursor Pd2+ and the reduced Pd metal particles. Furfural-
TPSR and furfural-FTIR exhibited that K-doped catalysts enhanced the furfural adsorption and promoted
the decarbonylation reaction, which was due to the different adsorption mode and intensity of furfural
on K-doped and undoped samples. Besides, the decrease of C–O adsorption of furfural (ꢀ2(C,O)-furfural)
by K doping is the main reason for suppressing the furfural hydrogenation.
Received 15 September 2010
Received in revised form
10 November 2010
Accepted 14 November 2010
Available online 21 November 2010
Keywords:
Decarbonylation
Furfural
Furan
Pd/Al2O3 catalysts
2-Methylfuran
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
mation of heavy products resulting in deactivation of the catalysts
[6,7]. In contrast to the non-noble catalyst systems, the supported
Transformation of non-food biomass to chemicals is an impor-
tant sustainability issue and has attracted much attention in recent
years [1]. Furfural, as an important chemical readily accessible from
biomass, appears to be the only unsaturated large-volume organic
chemical prepared from carbohydrate resources and is a key deriva-
tive for the production of non-petroleum-derived chemicals [2].
One important chemical coming from furfural is furan [1–5], which
is a useful intermediate and employed primarily in the synthesis of
tetrahydrofuran. Both vapor [3] and liquid phase [5] decarbonyla-
tion of furfural can produce furan, but vapor phase decarbonylation
is usually preferred for its simple operation, easy separation, han-
dling and recycling of catalysts.
Some metal oxides such as iron, zinc, manganese, molybdenum,
chromium oxides, or their mixtures [6–8] and noble metals such as
Pd, Pt, Rh used as supported form [9,10] are employed as the cat-
alysts for decarbonylation of furfural to furan. The main drawback
of the non-noble oxide catalysts is the drastic operating conditions,
which may bring about asignificant breakdownoffuranandthe for-
noble metal catalysts exhibit high conversion and yield at much
lower temperatures [9]. For the supported Pd catalysts, the most
effective and widely used supports for the decarbonylation reaction
are active carbon and alumina [11]. Furthermore, some additives,
such as nickel, rhodium, ruthenium, alkali and alkaline-earth met-
als, are added to modify the properties of supports and interact
with the metal, among which alkali metals are the best promoters
to the reaction [12].
It should be noted that the vapor-phase decarbonylation must
be carried out in the presence of hydrogen as carrier gas, which
has the function of prolonging the catalysts life [9]. However, the
life expectancy of the catalysts is still not long enough for indus-
trial application and the reason may be the lower ratio of H2 to
furfural. Most of the references reported that the molar ratio of
H2 to furfural should not be higher than 2, and the best range
is from 0.5 to 1 [9]. Srivastava and Guha [13] reported that the
main reason for catalysts deactivation is the formation of coke, and
explained that the rate of deactivation is presumed to be controlled
by the reaction between two adsorbed furfural to form coke pre-
cursor followed by coking. Therefore, the ratio of H2 to furfural
is chosen as 20 in this study, which obviously prolongs the cata-
lysts life and shows no sign of deactivation through hundreds of
hours. However, the high content of hydrogen existed in the reac-
tant gases can produce significant quantities of C O hydrogenation
∗
Corresponding author at: State Key Laboratory of Coal Conversion, Institute of
Coal Chemistry, Chinese Academy of Sciences, P.O. Box 165, Taiyuan 030001, PR
China. Tel.: +86 351 7117097; fax: +86 351 7560668.
E-mail address: zhuyulei@sxicc.ac.cn (Y. Zhu).
1381-1169/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.molcata.2010.11.016

72
W. Zhang et al. / Journal of Molecular Catalysis A: Chemical 335 (2011) 71–81
Different content of potassium was doped to the Al2O3 using
(1)
K2CO3 as K precursor and the loading range was from 0.75 to
10 wt.%. The preparation procedures of the catalysts were the same
as PA–K2CO3. These catalysts are labeled as PA–K(x), while x is
denoting the K content. The PA–K(1%) means the content of K in
the Pd–K/Al2O3 catalyst is 1 wt.%.
CHO
(3)
-CO
O
O
+H2
(4)
+H2
-H2O
CH3
O
O
(5)
2.2. Catalysts characterization
CH2OH
+2H2
CH2OH
O
O
The surface area and textural propertied of the fresh catalysts
were measured by nitrogen physical adsorption at 77.4 K with ASAP
2420 (Micromeritics, USA). The catalyst samples were degassed at
350 ◦C for 9 h prior to the measurement.
Fig. 1. Main reactions existed in furfural decarbonylation.
The morphologies of as-prepared samples were investigated by
SEM (Quanta 400F). From this technique the surface morphology
of the catalysts can be obtained.
ICP optical emission spectroscopy (Optima 2100DV,
PerkinElmer) was used to analyze the chemical composition
of the catalysts. From this technique the exact composition of
metal components can be obtained.
The temperature-programmed reduction of H2 (H2-TPR) was
performed in a conventional apparatus (Micromeritics Auto Chem.
2920, USA) with a thermal conductivity detector. The catalysts
were heated at a linear rate of 10 ◦C/min from room temperature
or −80 ◦C to 500 ◦C in a 5% H2/Ar mixture (mol/mol) flowing at
50 ml/min. During the TPR, a liquid nitrogen/isopropanol trap was
used to condense the product water in the effluent.
by-products. The main reactions occur in furfural decarbonylation
as shown in Fig. 1, including the decarbonylation of furfural to furan
(1), hydrogenation of furan to tetrahydrofuran (2), hydrogenation
of furfural to 2-methylfuran (3), furfuryl alcohol (4) and tetrahy-
drofurfuryl alcohol (5). Accordingly, this complication may arise
during separation and product purification. As the final product,
tetrahydrofuran can be obtained by furan hydrogenating without
separating it from the gaseous products of furfural decarbonyla-
tion. Among these products, furfuryl alcohol (Tboil = 171 ◦C) and
tetrahydrofurfuryl alcohol (Tboil = 178 ◦C) can be easily separated
with furan (Tboil = 32 ◦C) and tetrahydrofuran (Tboil = 66 ◦C). How-
ever, 2-methylfuran (Tboil = 63.5 ◦C) has a similar boiling point with
tetrahydrofuran, whichmakes it verydifficult for separation. There-
fore, it should be considered to reduce the by-products of furfural
hydrogenation, especially 2-methylfuran. Potassium promoter can
be employed to inhibit the hydrogenation function of catalyst,
which has been proved to be effective in the iron catalysts for F-
T synthesis [14]. So, it is expected that the addition of potassium
promoter can suppress the hydrogenation of Pd/Al2O3.
To the best of our knowledge, few reports have studied the vapor
decarbonylation of furfural on high H2/furfural ratio (≥20) and the
low reaction temperature range (180–260 ◦C). Besides, the effect of
K promoter on Pd catalysts is not systematically investigated in this
reaction. Therefore, the objectives of this work are to achieve high
yield of furan at low temperature and study the effect of K promoter.
More specifically, this work is devoted: (a) to choose the best K pre-
cursor; (b) to study the effect of K content; (c) to investigate the role
of K on the catalytic behavior of Pd catalyst. The structure and redox
properties of Pd–K/Al2O3 catalysts are characterized by BET, ICP,
H2-TPR, CO-FTIR, H2 chemisorption, H2-TPD, SEM, furfural-TPSR,
and furfural-FTIR aimed at defining the interaction between metals
and adsorption mode of furfural.
H2 chemisorption was performed in a surface area and porosity
analyzer (Micromeritics. ASAP 2020, USA). The method was used
to determine the dispersion of Pd and the amount of hydrogen
chemisorption. Prior to measurements, the catalysts were degassed
for several hours, and then reduced in hydrogen at 350 ◦C for 2 h fol-
lowed by degassing at the same temperature for 2 h. After cooling
under vacuum to the adsorption temperature, the total H2 uptake
(adsorption at the Pd surface and adsorption in the bulk) was mea-
sured. Next, inert gas (He) was used to remove the physisorbed
hydrogen, and then the amount of chemisorbed hydrogen was
obtained.
The temperature-programmed desorption of H2 (H2-TPD) was
performed in the same apparatus as H2-TPR. The catalysts were
first reduced by H2 at 350 ◦C for 2 h, and then purged with Ar at the
same temperature for 30 min followed by cooling down to 50 ◦C.
After adsorption of H2 at 50 ◦C for 30 min, the sample was purged
with Ar again in order to remove the weakly adsorbed species until
the baseline leveled off. Following this, the H2 desorption was being
carried out from 50 ◦C to 800 ◦C at a linear rate of 10 ◦C/min.
CO-FTIR spectra were collected using an infrared spectrome-
ter (VERTEX70, Bruker, Germany), equipped with KBr optics which
works at the liquid nitrogen temperature. The infrared cell with
ZnSe windows was connected to a gas-feed system with a set
of stainless steel gas lines, which allowed the in situ measure-
ment for the adsorption of CO. Before measurements, the catalysts
were reduced in situ at 350 ◦C for 2 h. After the reduction proce-
dure, the system was cooled to 20 ◦C and the CO-FTIR spectra were
recorded.
The temperature-programmed surface reaction of furfural
(furfural-TPSR) was performed in a self-made reactor connected
to an on-line mass spectrometer (OmniStar TM). Firstly, the cata-
lysts were reduced by H2 at 350 ◦C for 2 h, and then purged with He
at the same temperature for 30 min and cooled to 40 ◦C overnight
in order to diminish the effect of H2. Next, the absorption of furfural
was in situ performed at 40 ◦C by a stream of furfural (diluted with
carried gas He) for 1.5 h. After the absorption, the catalysts were
purged with He for removing the physically adsorbed furfural at
40 ◦C until the values of m/e became constant. The TPSR experi-
ments were performed with a heating rate of 10 ◦C/min in a flow
2. Experimental
2.1. Catalyst preparation
␥-Al2O3, obtained from the Aluminum Co., Ltd, of China (surface
area = 233 m2/g, 20–40 mesh), was used as support. 1 wt.% Pd/␥-
Al2O catalyst was prepared by incipient wetness impregnation.
Firstly, Pd(NO3)2 solution was impregnated to the Al2O3 at ambi-
ent temperature for 5 h. After that, the sample was dried overnight
at 120 ◦C and then calcined at 450 ◦C for 6 h. The 1 wt.% Pd/Al2O3 is
referred to as PA.
Five different 1–1 wt.% for Pd–K/Al2O3 catalysts were prepared
using different potassium salts (K2CO3, KOH, KCl, KNO3, CH3COOK)
as K precursors. Firstly, potassium solutions were impregnated to
the Al2O3 for 0.5 h before Pd(NO3)2 solutions were added in. The
subsequent drying and calcination procedures were the same as
those of the Pd/Al2O3. The catalysts are designed as PA–X, with
X denoting the different potassium salts. For example, PA–K2CO3
designates the K in Pd–K/Al2O3 catalyst coming from K2CO3.

W. Zhang et al. / Journal of Molecular Catalysis A: Chemical 335 (2011) 71–81
73
Table 1
as a negligible factor for the different catalytic activities of furfural
decarbonylation, which will be shown in catalytic performance.
Table 2 displays the morphology and sample composition of Pd
catalysts with different K content. With the increase of K loading,
the surface area and pore volume are decreased monotonous due
to pore blocking. The ICP results indicate that the content of Pd does
not change too much upon different K content. However, there is a
more rapid decline in K content at high K loading.
Fig. 2 exhibits that the outer surface of the catalysts with higher
K content (such as PA–K(8%)) is covered by elongated crystals other
than the support that are not observable before doping K or lower K
content, which is constituted mainly by alkaline metal carbonates
as found by EDS analysis [15].
Surface area and pore structure of the catalysts with doping different K salts.
Catalyst
Surface area
(m2/g)
Pore volume
(cm3/g)
Average pore
size (nm)
␥-Al2O3
PA
196.6
0.619
0.611
0.605
0.602
0.601
0.605
0.610
9.07
9.36
9.73
9.59
9.73
9.92
9.26
191.12
184.34
182.70
178.48
176.42
193.11
PA–K2CO3
PA–KOH
PA–KCl
PA–KNO3
PA–KAc
of He stream, and the desorbing products were monitored with the
online MS.
Furfural-FTIR spectra were collected with a VERTEX70 spec-
trometer cooled by liquid nitrogen. A vacuum pump was used to
pull the furfural vapor through the chamber for in situ FTIR mea-
surements. Before the measurements, the catalysts were reduced
in situ at 350 ◦C for 2 h. After the reduction procedure, the system
was cooled to 30 ◦C and the Furfural-FTIR spectra were recorded.
Then, the temperature was increased from 30 ◦C to 200 ◦C so as to
further evaluate the strength of adsorption and the capability of
furfural decarbonylation.
3.1.2. H2-TPR
TPR experiments can be used to investigate the reducibility of
the Pd2+ species loaded on a given support. By analyzing the TPR
results of the undoped and doped samples, it is possible to deter-
mine whether dopants have interacted with the supported Pd2+
precursor before or during reduction [15].
The TPR profiles of PA and the whole set of K-doped samples
(PA–K(x)) are shown in Fig. 3, in which (a) and (b) represent that the
experiment starts from room temperature and far below ambient
temperature (−80 ◦C), respectively. As displayed in Fig. 3a, the cat-
alysts with low K content or without K can be easily reduced from
the first beginning, and a negative peak at about 72.5 ◦C is appeared
on PA catalyst, which may be due to ␤-palladium hydride desorp-
tion [16]. In hydrogen atmosphere, PdO is reduced easily at ambient
temperature to Pd metal, further interacts with hydrogen resulting
in the formation of PdHx species. This progress appears to have
occurred during the passage of reducing gas prior to the start of the
temperature program, and then the PdHx will be decomposed with
the increase of temperature [17]. However, the desorption peak is
weak or disappeared with the increase of K content and the reason
may be the presence of electropositive alkali atoms in the vicinity
Pd sites, which exhibits higher electron-donating properties and
causes the Pd–H bond strength weaker.
2.3. Catalytic test
The activity tests for catalytic decarbonylation of furfural over
the Pd catalysts were performed in a fixed bed reactor under atmo-
spheric pressure. Prior to the reactions, the catalysts (20–40 mesh,
2.5 g) were in situ reduced in a stream of pure H2 (150 ml/min)
at 200 ◦C for 12 h. After reduction, furfural and H2 were introduced
into the fixed-bed reactor through a saturator, and the downstream
flow lines of the saturator were maintained at 140 ◦C to prevent
condensation of vaporized furfural. The decarbonylation activity
tests were performed in the temperature range of 180–260 ◦C, and
the liquid products were obtained by a cold trap filled with water.
The reactants and the products were analyzed using a GC-950 gas-
chromatograph (GC) equipped with flame ionization detector and
a capillary column (OV-101).
In order to avoid the partial reduction of Pd2+ species during sta-
bilization of the baseline, a low-temperature H2-TPR analysis seems
to be a feasible approach (Fig. 3b). As displayed in Fig. 3b, the reduc-
tion curve of the PA catalyst shows no negative peak and the high
temperature peak has a little shift to low temperature compared
with Fig. 3a. The PA sample shows a single reduction peak start-
ing at 25 ◦C and terminating at 75 ◦C (the peak at 50 ◦C). With the
incorporation of small amounts of K, an additional peak appears
at about 155 ◦C. Further increasing the K content, the intensity of
the high temperature peak is progressive increased, and the peak
systematically shifts to lower temperature. For example, the peak
undergoes a shift from the value 122 ◦C in PA–K(4%) to 107 ◦C in
PA–K(8%). Contrarily, with the increase of K loading in the cata-
lyst, the intensity of low temperature peak is decreased to a great
extent, and the peak simultaneous shifts to higher temperature. The
3. Results
3.1. Catalysts characterization
3.1.1. Morphology and sample composition (BET, ICP, SEM)
The BET surface area, pore volume and average pore size are
summarized in Tables 1 and 2.
Table 1 shows a decrease in surface area and pore volume, as
well as an increase in average pore size compared with the support
in all the catalysts, which may be due to pore blocking. However,
the deposit of Pd and different K salt precursors on ␥-Al2O3 support
has no obvious effects on the surface areas changes. That is, the
variations in the surface areas of these catalysts can be considered
Table 2
Characterization of the catalysts with doping different K content: surface area and pore structure, chemical composition, hydrogen chemisorption and metallic dispersion.
Catalyst
Surface area
(m2/g)
Pore volume
(cm3/g)
Average pore
size (nm)
Pd (wt.%)
K (wt.%)
ꢁ (K/wt.%)a
H2 chemisorption
(mmol/g)
Pd dispersionb (%)
PA
191.12
191.82
186.46
179.77
166.28
146.62
140.10
0.611
0.609
0.590
0.563
0.520
0.497
0.456
9.36
9.42
9.23
9.03
9.09
9.02
9.40
0.86
0.80
0.89
0.89
0.92
0.87
0.88
0
0
0.00988
0.00948
0.01398
0.01644
0.01811
0.01567
0.02379
21.0233
20.1753
29.7578
34.9739
38.5475
33.3399
50.6189
PA–K (0.75%)
PA–K (2%)
PA–K (4%)
PA–K (6%)
PA–K (8%)
PA–K (10%)
0.66
1.76
3.67
4.93
6.64
8.43
0.09
0.24
0.33
1.07
1.36
1.57
a
The difference value of calculated value and true value determined by ICP.
Pd dispersion determined from H2-chemisorption.
b

74
W. Zhang et al. / Journal of Molecular Catalysis A: Chemical 335 (2011) 71–81
Fig. 2. SEM images of the surface of catalyst particles (10 ␮m): (a) PA, (b) PA–K(2%), (c) PA–K(4%), and (d) PA–K(8%).
results indicate that more and more K+ will be interacted with Pd2+
with the increase of K content, which lead to change the reduction
temperature. The behavior up to K/Pd = 4 reflects a strong chemi-
cal interaction between K+ and Pd2+ species, which may eventually
lead to the formation of a mixed compound (KPdO) even at low K
content [15].
of Pd metal particles. From the results, we may draw the conclu-
sion that K has interacted with the reduced Pd. Then, CO-FTIR was
carried out as a complement experiment to further investigate this
interaction.
3.1.4. CO-FTIR
FTIR spectroscopy of adsorbed CO is a method for the characteri-
zation of the electronic states of supported metals. According to the
ꢂ–п binding mode reported by Davydov [19], the carbon monox-
ide molecules adsorbing on metal atoms or ions interact with their
valence d-elections and the frequency of the stretching vibration
of C–O bond of the surface M–CO complex depends directly on the
effective charge of the adsorption.
3.1.3. H2-TPD and H2 chemisorption
The H2-TPR results exhibit that the dopant phase has already
interacted with the precursor Pd2+ phase. One of the questions is
whether the interaction between K and Pd is still holding on the
reduced Pd metal particles, so the experiments of H2-TPD and H2
chemisorption were carried out on the reduced Pd particles. The
results are shown in Fig. 4 and Table 2.
The FTIR spectra of CO adsorbed at 20 ◦C on Pd/Al2O3 and the
K-doped Pd/Al2O3 samples are displayed in Fig. 5a–g. The spec-
tra of the catalysts show three main peaks from high-frequency
region (2000–2100 cm−1) to low-frequency (1800–2000 cm−1).
The interpretation of CO absorption bands on supported palladium
catalysts is well defined in the literature [19,20]. The signals in the
range 1850–1980 and 2050–2100 cm−1 are attributed to bridge
Pd–CO–Pd and linear Pd–CO species. Moreover, the two bridged
adsorption bands, from high-frequency to low-frequency, can be
attributed to the compressed and isolated bridged carbonyls [21]
or the twofold bridged carbonyls on Pd(1 0 0) and Pd(1 1 1) faces
[22], respectively. As shown in Fig. 5, the progressive increase of K
content causes a gradual modification of the CO spectra, suggesting
that the dopant has interacted with the reduced Pd metal particles.
As implied by the CO-FTIR characterization, the changes in CO
spectra with the increase of K loading can be summarized in three
main phenomena. (1) With the increase of K content, the inten-
sity of adsorption band is increased, especially for the catalyst
PA–K(8%), which shows the highest CO adsorption band. (2) The
bands I and II are progressive decrease of the intensity with respect
to the PA catalyst, and a simultaneous increase of band III. At the
8% K loadings, the spectra are dominated by the last band, which is
It is observed from Fig. 4 that hydrogen desorbs from the sam-
ples exhibiting three peaks centered. The low-temperature peak
is due to hydrogen chemisorption on the surface of Pd crystal-
lites, whereas the high-temperature feature can be attributed to
spillover hydrogen associated with the support. The medium-
temperature peak has been assigned to hydrogen adsorbed on the
sites located at the periphery of Pd crystallites, which are con-
tacted with the support [18]. K addition does not appreciably affect
the position of the low-temperature peak (at ca. 85 ◦C). However,
the high-temperature peaks (at ca. 480 ◦C) decrease in intensity
and have a little shift to higher temperature in the presence of
K. The reason is that K can induce the dehydroxylation of sup-
port, which is known to suppress the hydrogen spillover. Regarding
the medium-temperature peaks (at ca. 300 ◦C), they decrease in
intensity or overlap with the high-temperature peaks except the
catalyst PA—(6%). This implies that the adsorption strength of the
corresponding sites toward hydrogen decreases in the presence
of K. However, it is interesting to see that these effects do not
decrease the quantity of adsorbed hydrogen. On the contrary, the
adsorbed hydrogen is increased as indicated by the results of H2-
chemisorption (Table 2), which may be due to the higher dispersion

W. Zhang et al. / Journal of Molecular Catalysis A: Chemical 335 (2011) 71–81
75
a
PA
PA-K(0.75)
PA-K(10%)
PA-K(2)
PA-K(8%)
PA-K(4)
PA-K(6%)
PA-K(6)
PA-K(4%)
PA-K(8)
PA-K(2%)
PA-K(10)
PA-K(0.75%)
0
100
200
300
400
500
Temperature (ºC)
PA
b
PA
100
200
300
400
500
600
700
800
Temperature(ºC)
PA-K(0.75)
PA-K(2 )
PA-K(4)
PA-K(6)
PA-K(8)
PA-K(1 0)
Fig. 4. H2-TPD profiles on reduced catalysts.
downward shift of −116 cm−1 (from 1940 cm−1 in PA to 1824 cm−1
in PA–K(8%), value taken at peak). The phenomenon shows that the
doping of K results in a red shift of the ꢃ(CO) stretching frequency
of species adsorbed on these sites, which implies strengthening of
the Pd–CO bond with the increase of K content.
3.1.5. Furfural-TPSR
Programmed thermodesorption of furfural and the decarbony-
lation products was done to verify the stability of surface species
occurring during desorption and decomposition on PA and PA–K
catalysts. These experiments were performed with the adsorption
of furfural/He flow on reduced catalysts at 40 ◦C for 1.5 h, then
purging the physisorption of furfural, after that, monitoring the
desorbing products with the online MS. Masses characteristic of
different desorption products are shown in Fig. 6, including fur-
fural (C5H4O2, 96 amu), furan (C4H4O, 68 amu), carbon monoxide
(CO, 28 amu) and hydrogen (H2, 2 amu).
-100
0
100
200
300
400
500
Temperature (ºC)
Fig. 3. H2-TPR profiles of the investigated catalysts: (a) start at ambient temperature
and (b) start at −80 ◦C.
quite sharp and intense. (3) The change in intensity is accompanied
by a general downward shift of all the bands: for example, band I
undergoes a shift from the value 2083 cm−1 in PA to 2063 cm−1
in PA–K(8%) which is taken at peak, whereas band III undergoes a
2200
2100
2000
1900
1800 2200
2100
2000
1900
1800 2200
2100
2000
1900
1800 2200
2100
2000
1900
1800
PA-K(0.75%)
(b)
PA-K(10)
(a) PA
(c)
PA-K(2%)
1979
1878
1984
PA-K(8)
PA-K(6)
PA-K(4)
PA-K(2)
PA-K(0.75)
PA
2083
1959
2065
1940
2080
1934
1824
(g)
(d)
PA-K(10%)
(e) PA-K(6%)
PA-K(4%)
(f)
PA-K(8%)
1830
1855
1828
2065
1943
1953
2071
2065
2063
1948
1942
2200
2100
2000
1900
1800 2200
2100
2000
1900
1800 2200
Wavenumber (cm–1
2100
2000
1900
1800 2200
2100
2000
1900
1800
)
Fig. 5. FTIR spectra of CO dosed at 20 ◦C on the catalysts.

76
W. Zhang et al. / Journal of Molecular Catalysis A: Chemical 335 (2011) 71–81
a
b
(96 amu)
Furfural
O
C
Furan
(68 amu)
PA
H
O
O
PA
PA-K(4%)
PA-K(4%)
PA-K(8%)
PA-K(8%)
Al2O3
500
Al2O3
500
0
100
200
300
400
600
0
100
200
300
400
600
Temperature (ºC)
Temperature (ºC)
c
d
H2 (2 amu)
CO (28 amu)
PA
PA
PA-K(4%)
PA-K(4%)
PA-K(8%)
Al2O3
PA-K(8%)
Al2O3
0
100
200
300
400
500
600
0
100
200
300
400
500
600
Temperature (ºC)
Temperature (ºC)
Fig. 6. Furfural-TPSR on the catalysts: (a) furfural desorption, (b) furan desorption, (c) CO desorption, and (d) H2 desorption.
Fig. 6a shows that furfural desorption takes place at ca. 67 ◦C,
350 ◦C before the absorption of furfural, so the desorbed hydro-
and the intensity of furfural desorption peak is decreased with the
increase of K content, which may be due to the strong adsorption
between furfural and Pd in the presence of K. Also, the support can
adsorb a certain amount of furfural, with its intensity higher than
PA–K and lower than PA catalyst, indicating that the adsorbability
of furfural on Al2O3 is weak and the adsorption quantity is small.
Fig. 6b and c expresses the desorption peaks of furan and CO
(the products of furfural decarbonylation), respectively. In Fig. 6b,
two desorption peaks of furan are presented in PA catalyst: a small
peak at ca. 67 ◦C and a large peak at ca. 195 ◦C. In comparison, the
PA–K(4%, 8%) catalysts also have two desorption peaks: the low
temperature peaks are decreased of the intensity, but keeping the
same temperature as PA; the high temperature peaks are at ca.
152 ◦C, which is 43 ◦C lower than that of PA catalyst, implying that
furan can be easier desorbed on PA–K catalysts than on PA catalyst.
It is interesting to see that the support also presents two desorption
peaks of furan at 66 ◦C and 208 ◦C. However, the result of blank
test shows that the decarbonylation reaction cannot occur on the
support (Al2O3). Fig. 6c also exhibits that the desorption quantity
of CO on Al2O3 is obviously less than other Pd-containing catalysts.
Furthermore, the first two desorption peaks of CO on these catalysts
exhibit the similar desorption temperature as furan desorption, in
other words, furan desorption is accompanied by desorption of CO.
The CO desorption peaks on PA appear at 68, 214 and 290 ◦C, while
the peaks on PA–K are at 68, 158 and 290 ◦C. The third desorption
peak (290 ◦C) may be due to the strong interaction between Pd and
CO, which is decreased with the increase of K content. However,
an additional desorption peak of CO on PA–K (8%) is present at ca.
530 ◦C, which may be due to the decomposition of polymer.
gen was mainly coming from furfural. As shown in the figure,
H2 is desorbed more easily from PA–K catalysts than from PA
catalyst. In addition, it is obvious that there are no sign of H2 des-
orption on Al2O3, which suggests that the desorbed hydrogen is
formed in the presence of Pd. During the process of reduction,
H2 can be adsorbed dissociatively on the Pd surface and further
spill to the support. As revealed by H2-TPD, the spillover hydro-
gen is hardly desorbed completely lower than 350 ◦C (Fig. 4).
Therefore, the first desorption peak may be due to the hydro-
gen atoms released to the surface by the decomposition of the
furfural adlayer recombined and desorbed, while the second des-
orption peak may be due to the combination of two hydrogen atoms
derived from dehydrogenation of furfural and spillover hydrogen,
respectively.
3.1.6. Furfural-FTIR
To better understand the promoter effects on the decarbony-
lation reaction, we investigated the interaction of reduced PA and
PA–K(8%) with furfural. The in situ FTIR spectra of furfural adsorbed
at different temperature on the samples are displayed in Fig. 7.
Fig. 7a shows the adsorbed acyl species of furfural on PA cata-
lyst. At 30 ◦C, two bands observed at 1888 cm−1 and 1695 cm−1 can
be attributed to physisorbed and chemisorbed furfural, which are
similar to the adsorption of benzaldehyde on Pd/Al2O3 [23]. With
increasing the temperature, the physisorbed furfural is decreased
the intensity and the bands shift to lower frequency, while the
chemisorbed furfural only decreases the intensity. In addition, a
new band appeared at 80 ◦C (1957 cm−1), which is assigned to the
CO adsorption, is progressive increased the intensity with further
increase the temperature. It indicates that the increase of temper-
ature promotes the decarbonylation of furfural.
Fig. 6d displays the H2 desorption from the investigated cata-
lysts. In this characterization, the samples were purged with He at

W. Zhang et al. / Journal of Molecular Catalysis A: Chemical 335 (2011) 71–81
77
c
a
b
PA-K(8%)
PA
1695
1676
1807
1624
1888-1850
PA-K(8%)
200oC
30oC
50oC
80oC
30oC
50oC
80oC
PA
200oC
100oC
150oC
200oC
2021 1913
100oC
PA-K(8%)
30oC
150oC
200oC
PA
30oC
1957-1942
2000 1800
2000
1800
1600
1400
1600
1400
2000 1800 1600 1400
Wavenumber (cm-1)
Fig. 7. In situ FTIR of furfural adsorbed on the catalysts at different temperature: (a) adsorbed on PA catalyst, (b) adsorbed on PA–K(8%) catalyst, and (c) comparison of
K-doped and undoped catalysts.
Fig. 7b exhibits the adsorbed acyl species of furfural on PA–K(8%)
catalyst. The band at 1807 cm−1 is attributed to the physisorbed
furfural, which is also decreased the intensity with increasing the
temperature. It is interesting to see that the chemisorbed fur-
fural on the K-doped catalyst shows two bands at 1676 cm−1
and 1624 cm−1, suggesting the presence of two different adsorbed
modes. With increasing the temperature, the band at 1676 cm−1
is decreased the intensity, which is the same as the change of
adsorbed furfural at 1695 cm−1 on PA catalyst (Fig. 7A). However,
the band at 1624 cm−1 has no significant change. Furthermore, the
CO bands can be seen at 1913 cm−1 and 2021 cm−1 at the temper-
ature higher than 100 ◦C.
As indicated in Fig. 7c, the differences of furfural in situ IR spec-
tra on the K-doped and undoped samples can be summarized in
three main phenomena. (1) The intensity of furfural adsorption
bands (both physisorbed and chemisorbed furfural) on PA–K(8%)
are higher than that on PA, which indicates that the adsorption of
furfural is stronger on K-doped samples. (2) The PA–K(8%) catalyst
shows a conspicuous shift of all furfural adsorption bands to lower
frequency compared with PA. For example, the chemisorbed band
undergoes a shift from the value 1695 cm−1 in PA to 1676 cm−1
in PA–K(8%). (3) In contrast to PA, the PA–K(8%) displays two
chemisorbed bands: the first band is at 1676 cm−1, having the same
adsorption mode with band 1695 cm−1 on PA; and the “new band”
is at 1624 cm−1, which is gradually dominant with increasing the
temperature. Accordingly, the doping of K can change the adsorp-
tion mode of furfural on Pd catalysts, which has promoting effect
on the catalytic performance.
comparison with PA catalyst, the conversions of furfural on the
K-doped catalysts are all decreased to a certain extent, whereas
the furan selectivities are all increased (Fig. 8a and b). In other
words, the doping of K can suppress the hydrogenation of furfural
and promote the decarbonylation reaction on Pd catalyst. Fig. 8c
exhibits that the total yield of furan and tetrahydrofuran does not
all increase with K doping, for example, PA–KCl catalyst shows the
lowest yield, which may be due to the residual presence of chlorine
poisoned the active sites. The total yields of furan and tetrahydro-
furan on the K-doping catalysts with different K salts follow the
order of PA–K2CO3 > PA–KOH > PA > PA–KNO3 > PA–KAc > PA–KCl.
The last graph (Fig. 8d) shows that PA–KCl, PA–KAc and PA–K2CO3
catalysts have excellent behavior on suppressing hydrogenation of
furfural to 2-methylfuran. Therefore, considering the yield of furan
and 2-methylfuran selectivity, K2CO3 as K precursor can obtain the
best catalyst performance in the decarbonylation reaction.
Next, the experiments of the decarbonylation reaction were car-
ried out on different content of K (from K2CO3), and the results can
be seen from Fig. 9. The first graph (Fig. 9a) shows that the cat-
alyst PA–K(6%) has the highest furfural conversion, which is not
the expected result that the furfural conversion is decreased with
the increase of K content. H2-TPD (Fig. 4) shows that the PA–K(6%)
has the largest amount of hydrogen desorption below 300 ◦C fol-
lowed by PA–K(4%), PA–K(2%) and PA–K(8%), which is coincident
with the furfural conversion. It is revealed that the decrease of des-
orbed hydrogen would decrease the hydrogenation activity, which
is eventually decreased the furfural conversion. However, the oppo-
site trend is observed for furan selectivity (Fig. 9b). PA–K(8%)
exhibits the highest furan selectivity, while PA–K(0.75%) shows the
lowest. As shown in Fig. 9c, PA–K(8%) displays the highest yield
of furan (99.5%), while PA–K(0.75%) and PA–K(10%) exhibit lower
yields than other catalysts above 220 ◦C. It is also noticed that the
selectivity of 2-methylfuran is zero when the loading of K is above
4% (Fig. 9d).
3.2. Catalytic performance
Sample catalysts were tested for furfural decarbonylation to
examine the effect of added different K sources and different K con-
tent on the performance of Pd/Al2O3 (PA) catalyst. The results are
plotted in Figs. 8 and 9, which have four small graphs represented
as furfural conversion, furan selectivity, furan and tetrahydrofuran
yield, 2-methylfuran selectivity, respectively. Again, we empha-
size that the focus of our research is to improve the yield of furan
and tetrahydrofuran at low temperature (below 300 ◦C), as well as
minimize 2-methylfuran selectivity.
Fig. 8 shows the results of conversions, selectivities and yields
about the catalytic tests, and it can be clearly seen that PA cat-
alyst reveal a maximum for furfural conversion and in contrast,
lowest furan selectivity and highest 2-methylfuran selectivity. In
4. Discussion
Furfural is an unsaturated aldehyde of five carbons, which is
readily hydrogenated under hydrogen atmosphere. Generally, ther-
modynamic studies favor hydrogenation of C C bond over the
C
O bond for hydrogenation reaction of unsaturated aldehydes
[24,25]. Reaction kinetics also prefer hydrogenation of the C
bond over C O bond for small molecules, whereas steric constraints
C

78
W. Zhang et al. / Journal of Molecular Catalysis A: Chemical 335 (2011) 71–81
b 100
a
100
90
80
70
60
50
40
30
20
10
0
90
80
70
PA
PA-K 2CO3
PA-KOH
PA-KCl
PA-KNO 3
60
PA
PA-K 2CO3
50
PA-KOH
PA-KCl
40
PA-KNO 3
PA-KAc
PA-KAc
30
20
180
200
220
240
260
180
200
220
240
260
Temperature (ºC)
Temperature (ºC)
c
100
90
80
70
60
50
40
30
20
10
0
d
16
14
12
10
8
PA
PA-K 2CO3
PA-KOH
PA-KCl
PA-KNO 3
PA-KAc
PA
PA-K 2CO3
6
PA-KOH
PA-KCl
PA-KNO3
4
2
PA-KAc
0
180
200
220
240
260
180
200
220
240
260
Temperature (ºC)
Temperature (ºC)
Fig. 8. Catalytic activity of Pd–K catalysts with doping different K salts in decarbonylation of furfural: (a) conversion of furfural, (b) selectivity of furan, (c) total yield of furan
and tetrahydrofuran, and (d) selectivity of 2-methylfuran. Reaction conditions: atmospheric pressure, WHSV = 0.77 h−1, H2/furfural = 20 (molar ratio).
for larger molecules decrease the rates for hydrogenation of C
C
and simultaneously decrease the selectivity of hydrogenation by-
products, especially 2-methylfuran.
bond. Accordingly, the hydrogenation of C C bond in furfural is
more difficult than that of C O bond, which may be due to the
steric effects or the aromatic character of furfural. In our work, how-
ever, the hydrogenation of furfural should be decreased as much as
possible, especially for the C O bond hydrogenation. It has been
According to the BET results of PA–X catalysts, the PA–KAc
catalyst exhibits the highest surface area but shows less active
in furfural decarbonylation compared with other catalysts except
PA–KCl (Fig. 8a). A reversed trend is observed for PA–KNO3 sam-
ples with lower surface area and higher catalytic activity. Besides,
with the increase of K content (using K2CO3 as the K precursor), the
surface area and pore volume are decreased monotonous, whereas
the conversion of furfural does not follow the same order. There-
fore, the catalytic activity should depend on other factors, rather
than the specific surface area. Fig. 10 shows that large amounts of
crystalline alkali carbonate are deposited on the surface, but the
promoter phase is always an alkali compound other than carbon-
ate [29]. In other words, the promoter phase is part of dispersed K in
some chemical state, which may be formed through the interaction
between K and the activity species.
H2-TPR shows that the addition of K changes the reduction
profile of PA catalyst, favoring the reduction at higher tempera-
tures. It also indicates that the high temperature reduction peak is
attributed to the interaction between K and Pd2+. When the K con-
tent is over 4%, the high temperature peak is dominated, and shifts
to lower temperature with further increase of K content. Besides,
the decrease of low temperature peak proves that “free Pd atoms”
(do not interact with K) are decreased with increasing K content.
When the Kcontent is higherthan 4%, the “free Pdatoms” arealmost
disappeared, which presents the same trend as the selectivity of
2-methylfuran. In other words, the selectivity of 2-methylfuran is
decreased with the decrease of “free Pd atoms” and is approach to
zero at the K content of 4 wt.%. For the PA–K(10%), the reduction
reported that Pd exhibits a low rate for hydrogenation the C
O
bond compared with other metals commonly used for hydrogena-
tion [26]. In addition, the catalytic performance shows that a clear
and positive effect for the suppressing hydrogenation of furfural
is present by doping of K. The role of K in the suppressing hydro-
genation on the Pd catalysts can be better interpreted taking into
account the interaction between Pd particles and the dopant. Usu-
ally, alkali metals are employed as catalytic promoters or modifiers
to increase the activity and lifetime of heterogeneous catalysts, but
the role played by alkali metals is not unequivocal, depending on
the type of solid and catalytic reaction. Firstly, alkali oxides can neu-
tralize the surface acidic sites, thus enhancing their resistance to
poisoning by acid. Furthermore, many studies [27,28] suggest that
the alkali metal atoms modify the local electron density of the tran-
sition metal either directly or through the support, which can be
approved by XPS. However, the alkali metals are always introduced
as oxides, hydroxides, carbonates and nitrates. Thus, it is explained
that K strongly segregates at the surface, forming a submonolayer
of adsorbed K and O having a stoichiometric ratio of unity.
4.1. Characters and activity effect of K on Pd catalysts
The catalyst evaluation results exhibit that the doping of K on
Pd catalysts can enhance the decarbonylation of furfural to furan

W. Zhang et al. / Journal of Molecular Catalysis A: Chemical 335 (2011) 71–81
79
a
100
b
100
90
80
70
60
90
80
70
PA-K(0.75%)
PA-K(2)
PA-K(4%)
PA-K(6%)
PA-K(8%)
PA-K(10%)
60
50
40
30
20
PA-K(0.75%)
PA-K(2)
PA-K(4%)
PA-K(6%)
PA-K(8%)
PA-K(10%)
180
200
220
240
260
180
200
220
240
260
Temperature (ºC)
Temperature (ºC)
c
100
90
80
70
60
50
40
30
20
d
5
4
3
2
1
0
PA-K(0.75%)
PA-K(2)
PA-K(4%)
PA-K(6%)
PA-K(8%)
PA-K(10%)
PA-K(0.75%)
PA-K(2)
PA-K(4%)
PA-K(6%)
PA-K(8%)
PA-K(10%)
180
200
220
240
260
180
200
220
240
260
Temperature (ºC)
Temperature (ºC)
Fig. 9. Catalytic activity of Pd–K catalysts with doping different K content in decarbonylation of furfural: (a) conversion of furfural, (b) selectivity of furan, (c) total yield of
furan and tetrahydrofuran, and (d) selectivity of 2-methylfuran. Reaction conditions: atmospheric pressure, WHSV = 0.77 h−1, H2/furfural = 20 (molar ratio).
peak shifts to high temperature compared with PA–K(8%), and the
activity is obviously decreased. One reason might be that the active
sites are blocked by carbonate, and another reason might be that
the promoting effect is suppressed by some poisoning function of
the higher alkali content [30].
The modification induced by alkali addition on metal adsorption
properties has been widely reported in the literature. Kiskinova
studying hydrogen adsorption on alkali modified nickel indicated
that these additives give rise to an important decrease in hydro-
gen dissociation ability and saturation coverage [31]. Weatherbee
indicates that K produced an increase in the activation energy for
hydrogen adsorption on K promoted Fe/SiO2 catalyst [32]. Prali-
aud reported that K caused a decrease of the hydrogen adsorption,
explained this by a decrease in metallic phase accessibility or by a
rise of hydrogen bond stoichiometry, and claimed this occurrence
induced by an alkali “sieving effect” that limits the accessibility
of the metal surface to a larger probe molecules [33]. The H2-TPD
results (Fig. 4) show that the doping of K can suppress the spillover
hydrogen and H2 may interact with the alkali-modified support, in
other words, interact with hydroxyl groups on support. Therefore,
K-doping acts to enhance the adsorption strength of hydrogen, and
this effect should tend to reduce the rate of hydrogenation reaction,
either by increase the hydrogenation-reaction activation energy,
decreasing the concentration of “free” Ha [34]. Besides, increasing
of K content induces different behavior towards the adsorption of
CO on Pd particles and the CO-FTIR spectra show that the inten-
sity is increased and accompanied by a general red-shift of all the
bands. The sample PA–K(8%) shows the strongest interaction with
CO and the main CO absorption band is isolated bridged carbonyls.
Also, the PA–K(8%) exhibits the best performance on furfural decar-
bonylation, which may be owing to the adsorption mode and the
strong interaction between furfural and the catalyst.
4.2. Electronic effect of K on Pd catalysts
As indicated by the results of H2-TPR, a mixed compound (KPdO)
may be formed through the strong chemical interaction between
K+ and Pd2+ species, which is more difficult to reduce than PdO.
With the increase of K content (from 2% to 8%), the mixed com-
Fig. 10. SEM images of the surface of Pd–K(8%)/Al2O3 particles (5 ␮m).

80
W. Zhang et al. / Journal of Molecular Catalysis A: Chemical 335 (2011) 71–81
pound is more likely to be reduced and the reason might be that
increasing of K content can share more electrons with Pd2+ during
the process of reduction. Due to the higher electronegative of Pd
atom, an electronic density transfer is readily from alkali metal to
Pd, which is proved by XPS with a shift towards negative values of
the Pd 3d binding energies [35]. Also, alkali adsorbates also cause a
dramatic enhancement of the electronic polarizability of the metal
surface extending it several angstroms into the vacuum [36]. This
effect also produces a general shift towards lower frequencies of
the CO bands as indicated by the CO-FTIR spectra on these cata-
lysts, owing to an enhanced transfer of electron density from the
metal to the ␲* molecular orbitals of CO.
Electronic effects of K promoter on Pd can take place either
through direct K–Pd interaction or through a modification of the
support properties which induces a change in metal–support inter-
actions. The addition of K2CO3 can make the support properties
change from acidic to basic [37] and the electron transfer between
metals and supports has been suggested by several researchers
[38,39]. Figoli and L’argeniere studied the selective hydrogenation
of styrene over Pd/Al2O3 catalysts and discovered the changes in
the electronic state of Pd via XPS when sodium was added to the
support. Free electrons on Pd were transferred to lower electron
density sites on acidic Al2O3 and electron transfer was attenuated
with added sodium [38]. Figueras et al. studied benzene hydro-
genation over Pd supported on several different acidic oxides and
claimed that the support influences the electronic properties of Pd
[39]. In our work, more basic nature of K-doped Al2O3 favors an
electron transfer to Pd, resulting in a higher Pd electron density,
and ultimately a reduced strength of hydrocarbon adsorption on
Pd. This interaction leads to an increase in furan selectivity from
furfural because the furan desorbs more readily before it can be fur-
ther hydrogenated to tetrahydrofuran, and simultaneously results
in a decrease of the furfural hydrogenation by-products.
H
C
O
C
η1(O)
(1)
(2)
O
H
O
O
Pd
H
O
C
η2(C,O)
C
Pd
+
O
H
O
O
Pd
H
C
a
CO
O
O
O
O
O
Pd
Pd
H
O
η1(C)
C+
C
O
O
Pd
Pd
Pd
O
C+
b
CO
+
O
Pd
O
O
η1'(C)
c
(3)
C+
Pd-K
CO
H
C
O
O
O
C+
+
O
O
Pd-K
Fig. 11. Adsorption mode of furfural on Pd/Al2O3 and Pd–K/Al2O3 catalysts.
configuration [40]. Then, a small part of furfural decarbonylates
through this adsorption mode [42] (Fig. 11a) which is labeled as
ꢀ2(C,O)-furfural. Accordingly, the decarbonylation product furan
can be obtained at ca. 67 ◦C, which is observed as the first furan
desorption peak occurred both on PA and PA–K catalysts (Fig. 6b).
Secondly, the lone-pair electrons of carbonyl oxygen displays neg-
ative charge and the surface of Pd also exhibits a negative electric
field induced by K, so the adsorption of carbonyl oxygen is more
difficult on PA–K than on PA. As a result, the amount of ꢀ2(C,O)-
furfural will be decreased by K doping, which suggests that the
furan produced through ꢀ2(C,O)-furfural will be decreased with the
increase of K content. As shown in Fig. 6b, it is obvious that the
intensity of the first furan desorption peaks is decreased accom-
panied by increasing K content. Furthermore, the hydrogenation
of furfural always occurs through ꢀ2(C,O) configuration, that is the
4.3. Adsorption mode of furfural on PA and PA-K catalysts
The results of CO-FTIR indicate that the doping of K can trans-
fer electronic density to Pd and change the adsorption mode of
CO. During the decarbonylation reaction, the surface of Pd displays
negative charge in the presence of K, while the adsorbed furfural
simultaneously dehydrogenates and then forms carbonyl C+ ions.
As a result, the adsorption of furfural is stronger on PA–K than
that of PA catalyst, which can be proved by the results of furfural-
FTIR (Fig. 7) and is in agreement with the results that the intensity
of furfural desorption peaks are decreased with the increase of K
content (Fig. 6a). Also, the strong binding between furfural and
PA–K can promote the cleavage of C–C, that is to say the strong
adsorption facilitates the furfural decarbonylation, as indicated by
the furan desorption (Fig. 6b). Barteau have studied the adsorp-
tion mode of aldehydes on Pd(1 1 1) and Pd(1 1 0) surfaces by a
combination of TPD and HREELS, mainly for adsorption in the low
temperature regime[40,41]. At lower temperatures aldehydes tend
to bond in the ꢀ1(O) configuration (that is, adsorb on the metal sur-
face through the oxygen lone pair electrons), a weakly held form
which desorbs at low temperatures, typically ca. 200 K. A more
strongly bonded ꢀ2(C,O) configuration (ꢄ-bonded through carbon
and oxygen and flat-lying) is identified above 200 K, and it appears
that decomposition or decarbonylation occurs through this state
[42]. However, Barteau proposed that the decarbonylation process
of aldehydes is carried out through the state of ꢀ1(C) on Pd(110),
which is transformed from ꢀ2(C,O) [43].
C
O bond hydrogenation [23], so the decrease of ꢀ2(C,O)-furfural
caused by K doping will inhibit the hydrogenation reaction. For PA
catalyst, the main decarbonylation occurs through ꢀ1(C) configu-
ration (Fig. 11b), which is transformed from ꢀ2(C,O) configuration
[43]. However, for PA–K catalysts, the results of furfural-FTIR (Fig. 7)
show that the adsorption mode of acyl species is different on K-
doped and undoped catalysts, and furfural-TPSR indicates that the
ꢀ2(C,O) configuration is difficult to form and unstable on PA–K cat-
alysts. Accordingly, the main decarbonylation of furfural on the
K-doped samples may occur through ꢀꢀ1(C) configuration (Fig. 11c),
which is more stable than ꢀ1(C) configuration.
5. Conclusions
In this work, the catalytic behaviors of K-doped Pd/Al2O3 cat-
alysts for furfural decarbonylation have been investigated in the
temperature range of 180–260 ◦C and ambient pressure. According
to the results, the Pd–K/Al2O3 catalyst doped with K2CO3 and with
a K-loading of 8 wt.% shows the best performance in the decarbony-
lation reaction (the conversion of furfural is 100% and the yield of
furan is 99.5% at 260 ◦C). Besides, the selectivity of 2-methylfuran
will approach zero when the loading of K is above 4 wt.%. The effect
On the basis of furfural-TPSR and furfural-FTIR studies, the
following bond activation sequence is proposed to occur during
furfural decarbonylation on Pd catalysts (Fig. 11). Firstly, furfural
adsorption in the ꢀ1(O) configuration is not stable under our oper-
ational temperature and can be transformed easily to ꢀ2(C,O)

W. Zhang et al. / Journal of Molecular Catalysis A: Chemical 335 (2011) 71–81
81
of K is systematically investigated through several characterization
methods, which can be summarized as follows: (1) K doping has a
great influence on Pd2+ precursor phase and the reduced Pd atom.
TPR data indicates that there exists a strong interaction between K
and Pd2+, which inhibits the reduction of Pd2+ species. The results
of CO-FTIR and H2 chemisorption also suggest that the strong inter-
action still exists between K and the reduced Pd atoms. (2) K doping
has caused the red-shift of CO adsorption, which suggests that more
basic nature of K-doped Al2O3 results in an electron transfer to Pd,
leading to a higher Pd electron density, and ultimately promoting
the decarbonylation of furfural. (3) The furfural-TPSR and furfural-
FTIR results indicate that the doping of K can change the adsorption
mode of furfural on Pd catalysts. Finally, it is important to mention
that the decrease of ꢀ2(C,O) configuration might be the main reason
for suppressing hydrogenation of furfural on K-doped Pd catalysts.
[10] P. Lejemble, A. Gaset, Philippe Kalck, Biomass 4 (1984) 263–274.
[11] A.P. Dunlop, G.W. Huffman, US 3,257,417 (1966).
[12] K.J. Jung, A. Gaset, Biomass 16 (1988) 89–96.
[13] R.D. Srivastava, A.K. Guha, J. Catal. 91 (1985) 254–262.
[14] Y. Yang, H.W. Xiang, Y.Y. Xu, L. Bai, Y.W. Li, Appl. Catal. A 266 (2004) 181–194.
[15] R. Pellegrini, G. Leofanti, G. Agostini, L. Bertinetti, S. Bertarione, E. Groppo, A.
Zecchina, C. Lamberti, J. Catal. 267 (2009) 40–49.
[16] G. Neri, G. Rizzo, L. De Luca, A. Donato, M.G. Musolino, R. Pietropaolo, Appl.
Catal. A 356 (2009) 113–120.
[17] A.H. Padmasri, A. Venugopal, J. Krishnamurthy, K.S. Rama Rao, P. Kanta Rao, J.
Mol. Catal. A 181 (2002) 73–80.
[18] P. Paraskevi, I.K. Dimitris, J. Catal. 267 (2009) 57–66.
[19] A.A. Davydov, Infrared Spectroscopy of Adsorbed Species on the Surface of
Transition Metal Oxides, Wiley, New York, 1990.
[20] X.Z. Jiang, Y.H. Sua, B.J. Lee, S.H. Chien, Appl. Catal. A 211 (2001) 47–51.
[21] W.J. Kim, J.H. Kang, I.Y. Ahn, S.H. Moon, Appl. Catal. A 268 (2004) 77–82.
[22] S. Bertarione, D. Scarano, A. Zecchina, V. Johanek, J. Hoffmann, S. Schauermann,
M.M. Frank, J. Libuda, G. Rupprechter, H.J. Freund, J. Phys. Chem. B 108 (2004)
3603.
[23] C. Keresszegi, D. Ferri, T. Mallat, A. Baiker, J. Catal. 234 (2005) 64–75.
[24] B.B. Baeza, I.R. Ramos, A.G. Ruiz, Appl. Catal. A 205 (2001) 227–237.
[25] R. Zanella, C. Louis, S. Giorgio, R. Touroude, J. Catal. 223 (2004) 328–339.
[26] T.B.L.W. Marinelli, S. nabuurs, V. ponec, J. Catal. 151 (1995) 431–438.
[27] A. Daz, D.R. Acosta, J.A. Odriozola, M. Montes, J. Phys. Chem. B 101 (1997)
1782–1790.
Acknowledgement
The authors gratefully thank the financial supports of Natural
Science Foundation of China (No. 20976185).
[28] W.D. Mross, Catal. Rev. 25 (4) (1983) 591–637.
[29] A.J. Urquhart, J.M. Keel, F.J. Williams, R.M. Lambert, J. Phys. Chem. B 107 (2003)
10591–10597.
References
[30] C.P. Huang, J.T. Richardson, J. Catal. 51 (1978) 1–8.
[31] M.P. Kiskinova, Studies in Surface Science and Catalysis, vol. 70, 1992, 149.
[32] G.D. Weatherbee, J.L. Rankin, C.H. Bartholimew, Appl. Catal. 11 (1984) 73.
[33] H. Praliaud, M. Primet, G.A. Matin, Appl. Surf. Sci. 17 (1983) 107.
[34] J.W. Federico, P. Alejandra, T. Samuel, S.T. Mintcho, M.L. Richard, J. Phys. Chem.
B 106 (2002) 5668–5672.
[35] F.L. Leonarda, A.M. Guy, G.D. Giulio, J. Catal. 164 (1996) 322–333.
[36] S. Stolbov, T.S. Rahman, Phys. Rev. Lett. 96 (2006) 186801–186804.
[37] H.P. Yeung, L.P. Geoffrey, Ind. Eng. Chem. Res. 31 (1992) 2.
[38] N.S. Figoli, P.C. L’argeniere, Catal. Today 5 (1989) 403.
[39] F. Figueras, R. Gomez, M. Primet, Adu. Chem. Ser. 121 (1973) 480.
[40] M. Mavrikakis, M.A. Barteau, J. Mol. Catal. A 131 (1998) 135.
[41] J.L. Davis, M.A. Barteau, J. Am. Chem. Soc. 111 (1989) 1782–1792.
[42] M. Bowker, R. Holroyd, N. Perkins, J. Bhantoo, J. Counsell, A. Carley, C. Morgan,
Surf. Sci. 601 (2007) 3651–3660.
[1] A. Corma, S. lborra, A. Velty, Chem. Rev. 107 (2007) 2411–2502.
[2] K.J. Zeitsch, The Chemistry and Technology of Furfural and Its Many By-
products, vol. 13, 1st ed., Elsevier, Amsterdam, 2000.
[3] H. Singh, M. Prasad, R.D. Srivastaval, J. Chem. Technol. Biotechnol. 30 (1980)
293–296.
[4] J. Coca, E.S. Morrondo, J.B. Parra, H. Sastre, React. Kinet. Catal. Lett. 20 (1982)
3–4.
[5] K.J. Jung, A. Gaset, Biomass 16 (1988) 63–76.
[6] P.A. Moshkin, E.A. Preobrazhenskaya, B.B. Berezina, V.E. Markovich, E.G. Kud-
rina, V.P. Papsueva, E.A. Sokolina, Khimiya Oeterotsiklicheskikh Soedinenii. 2
(1966) 3–7.
[7] J. Coca, E.S. Morrondo, H. Sastre, J. Chem. Technol. Biotechnol. 32 (1982)
904–908.
[8] G.M. Whitman, US 2,374,149 (1945).
[9] D.G. Manly, J.P. O’Halloran, US 3,223,714 (1965).
[43] R. Shekhar, M.A. Barteau, R.V. Plank, J.M. Vohs, J. Phys. Chem. B 101 (1997)
7939–7951.