Catalytic fast pyrolysis of furfural over H-ZSM-5 and Zn/H-ZSM-5 catalysts

Article abstract of DOI:10.1016/j.apcata.2012.01.017

Furfural, a cellulose model compound, was converted into gasoline-range fuels through catalytic fast pyrolysis. H-ZSM-5-based catalysts were employed in a continuous fixed bed system. The reaction temperature, reactant contact time, and catalytic promoter are keys to manipulate the product distribution. The first step in furfural conversion is the decarbonylation of furfural to form furan, followed by furan conversion to intermediates (e.g., cyclohexene and 3,4-dimethylbenzaldehyde) in the ZSM-5 pores. These intermediates can then be transformed into aromatics, coke, light olefins, and carbon oxides. A reaction temperature of 500°C generated the highest yield of aromatics and the lowest amount of coke. A long contact time (～1.5 s) also provided the highest aromatic selectivity. The promoter, zinc oxide, plays an important role in hydrogen atom transfer. This is attributed to the change of acid site concentration and Lewis acid sites created by anchored Zn ions.

Full text of DOI:10.1016/j.apcata.2012.01.017

Applied Catalysis A: General 419–420 (2012) 102–110
Contents lists available at SciVerse ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Catalytic fast pyrolysis of furfural over H-ZSM-5 and Zn/H-ZSM-5 catalysts
Wei-Liang Fanchiang, Yu-Chuan Lin∗
Department of Chemical Engineering and Materials Science, Yuan Ze University, Chungli, Taoyuan 32003 Taiwan
a r t i c l e i n f o
a b s t r a c t
Article history:
Furfural, a cellulose model compound, was converted into gasoline-range fuels through catalytic fast
pyrolysis. H-ZSM-5-based catalysts were employed in a continuous fixed bed system. The reaction tem-
perature, reactant contact time, and catalytic promoter are keys to manipulate the product distribution.
The first step in furfural conversion is the decarbonylation of furfural to form furan, followed by furan
conversion to intermediates (e.g., cyclohexene and 3,4-dimethylbenzaldehyde) in the ZSM-5 pores. These
intermediates can then be transformed into aromatics, coke, light olefins, and carbon oxides. A reaction
temperature of 500 ◦C generated the highest yield of aromatics and the lowest amount of coke. A long
contact time (∼1.5 s) also provided the highest aromatic selectivity. The promoter, zinc oxide, plays an
important role in hydrogen atom transfer. This is attributed to the change of acid site concentration and
Lewis acid sites created by anchored Zn ions.
Received 14 November 2011
Received in revised form 13 January 2012
Accepted 15 January 2012
Available online 25 January 2012
Keywords:
Fast pyrolysis
Furfural
Zeolite
Biofuels
© 2012 Elsevier B.V. All rights reserved.
ZSM-5
1. Introduction
couple with acetone to form dimers through aldol-condensation.
A series of hydrodeoxygenation steps can then transform these
The conversion of lignocellulosic biomass has the potential to
be a renewable and carbon neutral method of producing fuels
and chemicals [1–3]. Several groups have recently developed
novel technologies, such as reactive flash volatilization [4,5] and
catalytic fast pyrolysis [6,7], to overcome the difficulty of ligno-
cellulosic biomass processing. The first step of these approaches
is to disintegrate the solid lignocellulosic frameworks into reac-
tive intermediates, which can then be transformed into desired
products.
Furfural is a common intermediate in various biorefining pro-
cesses. Woody biomass pyrolysis yields furfural in the form of
dehydrated products from cellulose and hemicellulose [8–10].
Bio-oils derived from biomass liquefaction also contain signifi-
cant amounts of furfural [2,11]. Thus, furfural is a popular model
compound of lignocellulosic derivatives. For example, Elliott and
Hart used furfural as a cellulosic model compound in the catalytic
hydroprocessing of a bio-oil-like mixture [12]. An effective method
of furfural conversion should facilitate lignocellulosic biorefining.
Hydrogenation and aqueous phase processing are currently the
two most encountered approaches of converting furfural into fuel-
grade chemicals [2]. The major products of furfural hydrogenation
are furfuryl alcohol, methylfuran, tetrahydrofurfuryl alcohol, and
methyltetrahydrofuran (MTHF) [2], which is a key component in
P-Series fuels [13]. Aqueous phase processing allows furfural to
dimers into gasoline-range alkanes (C7–C15) [14–16].
Pyrolysis in the presence of zeolites is another method for
biomass upgrading that involves oxygen removal through a series
of dehydration, decarbonylation, and decarboxylation reactions.
This process was introduced in Mobil’s pioneering research on
methanol-to-gasoline (MTG) production [17–19]. Chen et al. [20]
showed that ZSM-5-based catalysts can convert glucose solutions
into olefins and aromatics. Diebold et al. [21,22] invented an inte-
grated process of fast pyrolysis with catalytic cracking to upgrade
pyrolysis vapors. Their research stimulated follow-up studies on
reactor design [23–26].
Zeolite design for biomass pyrolysis is another common
research topic. A research group at the University of Saskatchewan
investigated the performance of various zeolites in bio-oil upgrad-
ing [27–29]. Zeolites selection for lignocellulosics pyrolysis has
recently been investigated in numerous institutes, such as Åbo
Akademi University [30,31], Aston University [32,33], and National
Renewable Energy Laboratory [34]. ZSM-5 plays an essential role
in each case, and is almost omnivorous for different feedstocks.
Unlike the catalytic pyrolysis of biomass and its derivatives,
relatively few studies investigate the catalytic pyrolysis of fur-
fural, especially in a fast pyrolysis environment. The Dao group
[35,36] studied glucose-related compounds, including furfural,
over ZSM-5-based catalysts. They achieved poor yields of hydro-
carbon products (less than 10%), with mostly furan and CO derived
from furfural decarbonylation. Grandmaison et al. [37] employed
ZSM-5 in both furfural and furan conversion, proving that the latter
was more stable than the former. Nevertheless, converting furanic
∗
Corresponding author. Tel.: +886 3 463 8800x3554; fax: +886 3 455 9373.
E-mail address: yclin@saturn.yzu.edu.tw (Y.-C. Lin).
0926-860X/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2012.01.017

W.-L. Fanchiang, Y.-C. Lin / Applied Catalysis A: General 419–420 (2012) 102–110
103
compounds resulted in low hydrocarbon yields with long contact
times (more than 1000 s). The Huber group lately performed furan
and furfural conversion in both batch and continuous systems with
short contact times [38–40], and achieved more than a 30% yield
of aromatics. They also demonstrated that the product distribution
can be manipulated using the appropriate operating conditions.
This study investigates the catalytic pyrolysis of furfural in
producing gasoline-range fuels. The effects of reactant contact
time, pyrolysis temperature, and catalytic promoter were stud-
ied. ZSM-5-based catalysts, including H-ZSM-5 and Zn/H-ZSM-5,
were employed in a fixed bed system. Zinc cations were introduced
into H-ZSM-5 because of their capability in hydrogen atom trans-
fer, which is a major factor in aromatization [41–44]. This study
shows that it is possible to adjust the product distribution using
catalyst design and reaction environments in a continuous catalytic
pyrolysis system. This study also proposes a plausible mechanism
of furfural pyrolysis.
an hour prior to the test. The pre-treated sample was subjected to
a thermal program with a 5 ◦C/min heating rate in air (60 mL/min).
2.3. Activity evaluation
Catalytic performance was conducted using a continuous fixed
bed design (Fig. 1). This system consisted of a quartz reactor
(i.d. = 2 cm), a vapor saturator, a mass flow controller, and an ice
bath cooling trap. The saturator was kept at 40 ◦C (furfural vapor
pressure, ∼743 Pa) using a water bath. The reaction temperature
was controlled by a PID controller with a K-type thermal couple as
the sensor and heavy duty heating tape (Omega) as the heat source.
Prior to each test, blank runs were performed to evaluate the ther-
mal cracking of furfural and to estimate the molar carbon input.
Furfural decomposition was negligible (less than 5%) in the reaction
environments. Three different temperatures (300, 400, and 500 ◦C)
at gas hourly space velocity (GHSV) of 2412 h−1 and three different
GHSVs (2412, 7200, and 36,000 h−1) at 500 ◦C were applied to test
all catalysts.
2. Experimental
All catalysts were sieved to 40–80 mesh (0.42–0.18 mm) par-
ticle size. Each trial was carried out under atmospheric pressure
with no pressure drop across the catalyst bed. About 0.2 g of cat-
alyst was consumed per trial. All reactions reported were done
with 13 min time on-stream. After reaction, stripping of catalyst
was performed for 5 min using a N2 stream of 25 mL/min. Gaseous
products were collected in an air bag. Heavy products were trapped
in the cooling trap. The condensed products were washed out
with acetone before subsequent analysis. Qualitative and quanti-
tative analyses of gas and liquid products were performed using a
GC/MS (HP 5890 II GC with 5972 MSD, DB-5MS capillary column,
60 m × 0.25 mm) and a GC TCD/FID equipped with a methanizer
(SRI 8610, molecular sieve 13X and silica gel columns). Standards
such as benzene, toluene, phenol, furfural, p-xylene, benzofuran,
naphthalene, and indene were injected into GC/MS; CH4, CO, CO2,
ethylene, propylene, furan, benzene, toluene, and p-xylene were
detected by GC TCD/FID. Because of a lack of standard reagents, the
responses of methylnaphthalene and methylindene were set equal
to naphthalene and indene; fluorene and anthracene, naphthalene.
The resulting coke was analyzed by measuring the weight loss of
spent catalyst using an elemental analyzer (EA, PerkinElmer series
II 2400). The sample was treated in a 10% O2/He stream at 150 ◦C
for 30 min, followed by a 10 ◦C/min heating rate to 600 ◦C.
2.1. Material
Furfural (Sigma–Aldrich, ACS reagent, 99%) was used directly
without purification. ZSM-5 (Zeolyst, CBV 8014) with a SiO2/Al2O3
ratio of 80 was calcined in air at 600 ◦C overnight before treat-
ment. H-ZSM-5 was synthesized using the ion-exchange method
with a 1.0 M solution of NH4NO3 (Sigma–Aldrich, ACS reagent 99%)
at 70 ◦C for 1 h. The residual was filtered, washed with deionized
water, dried at 100 ◦C for 10 h, and calcined at 600 ◦C overnight.
The resulting powder was labeled as H-ZSM-5. Zn/H-ZSM-5 was
prepared using the incipient wetness impregnation with a 0.1 M
Zn(NO3)2 solution. Approximately 0.5 and 1.5 wt% loadings of Zn
were used. The remaining paste was dried in air at 100 ◦C for 10 h
then calcined at 550 ◦C overnight.
2.2. Catalyst characterization
X-ray diffraction (XRD) patterns were recorded using a Shi-
madzu Labx XRD-6000 with Cu K␣ radiation (0.15418 nm). Scans
were taken at a 5–60◦ 2Â range with a scanning rate of 4◦/min. The
voltage and current used were 40 kV and 30 mA, respectively.
Ammonia pulse chemisorption and ammonia temperature-
programmed desorption (NH3-TPD) were performed in the same
system [45]. The system consisted of a thermal conductive detec-
tor (TCD), a temperature-controlled furnace, and a U-shaped quartz
cell. The pulse chemisorption of ammonia was carried out to esti-
mate the number of acid sites on the catalysts. Each trial used
approximately 30 mg of the sample. The sample was pretreated in a
He stream at 120 ◦C for 30 min. Pulse chemisorption was executed
by injecting 1% NH3/He (10 mL) pulses through the catalyst bed at
130 ◦C until achieving breakthrough. NH3-TPD was performed after
NH3 pulse chemisorption in a He stream from 130 to 650 ◦C with
a 10 ◦C/min heating rate. Re-adsorption of NH3 was evaluated by
varying the contact time of the carrier gas as Fig. S1 shows (see
supplementary data). The re-adsorption effect can be ignored at a
carrier gas contact times of less than 5 × 10−4 g/min/mL in this case,
the first and second desorption temperatures were nearly iden-
tical with decreasing contact times. Therefore, a contact time of
5 × 10−4 g/min/mL was selected for all NH3-TPD experiments. The
baseline drift of pulse chemisorption and NH3-TPD was evaluated
by blank runs.
2.4. Soluble coke analysis
A leaching method developed by Guisnet and Magnoux [46]
was conducted to identify the species preserved in the spent
ZSM-5. Approximately 20 mg of used catalyst (H-ZSM-5 and 1.5%
Zn/H-ZSM-5 in 500 ◦C, GHSV = 2412 h−1) was soaked in 0.5 mL
of 2% hydrofluoric acid at ambient temperature. After vigor-
ous stirring for an hour, the sample was allowed to stand
overnight. Soluble compounds were then extracted by adding
0.5 mL methylene chloride to the solution. The organic phase was
then analyzed by GC/MS. Due to the lack of reference reagents, the
GC/MS responses of extracted compounds (cyclohexene and 3,4-
dimethylbenzaldehyde) were assumed to be the same as that of
p-xylene.
3. Results and discussion
3.1. Physicochemical properties of catalysts
Temperature-programmed oxidation (TPO) was conducted
using a thermogravimetric analyzer (TGA, TA Instruments Q50).
Approximately 6 mg of post-reaction catalyst was placed in a plat-
inum pan for each trial. The sample was dehydrated at 120 ◦C for
Fig. 2 depicts the XRD patterns of ZSM-5-based catalysts. Similar
profiles were recorded for the investigated catalysts. This indicates
that the crystalline structure of ZSM-5 remained intact after ion-
exchange and Zn implementation. In addition, the characteristic

104
W.-L. Fanchiang, Y.-C. Lin / Applied Catalysis A: General 419–420 (2012) 102–110
Fig. 1. Catalytic fast pyrolysis set-up.
(FWHM), acid site distribution, and the overall number of acid sites.
Two desorption peaks are apparent. These peaks can be classified as
weak and strong chemisorbed ammonia. The low desorption tem-
perature (Td) is associated with non-framework Lewis acid sites,
whereas the high Td is related to Brønsted acid sites on Si
O Al
groups [49].
As the Zn loading increased, the Td of Lewis acid sites shifted
from 177 to 216 ◦C. This suggests an increase in Lewis acidity.
Moreover, increasing Zn content widened the FWHM of desorp-
tion response of Lewis sites, suggesting the presence of different Zn
species [50]. In contrast, the Td of Brønsted acid sites decreased from
365 to 338 ◦C with similar FWHMs. This implies that introduced Zn
species suppressed the acidity of Brønsted acid sites. A possible
explanation is that the introduction of Zn2+ exchanged the protons
of Brønsted acid sites [41], leading to the decline of Brønsted acid-
ity. The sum of Brønsted and Lewis acid sites increased following
the trend of H-ZSM-5 ≤ 0.5% Zn/H-ZSM-5 ≤ 1.5% Zn/H-ZSM-5.
Fig. 2. XRD pattern of H-ZSM-5-based catalysts.
3.2. Reactivity of H-ZSM-5
responses of ZnO were absent for 0.5 and 1.5% Zn/H-ZSM-5, sug-
gesting that Zn ions were well dispersed. Various Zn species could
coexist, such as Zn2+ replaces two protons with two aluminum
sites (O− Zn O−) [41,47], zinc hydroxyl (ZnOH+) [48], and finely
dispersed ZnO [48].
Table 2 lists carbon yield and aromatic selectivity as func-
tions of temperature for furfural pyrolysis over H-ZSM-5 at
GHSV = 2412 h−1 (contact time = 1.5 s). Furfural was fully converted.
As the temperature increased, aromatics increased from 0.1 to
15.7%; CO, 18.8 to 25.1%. Coke declined from 64.1 to 27.6%. The yield
of furan initially increased from 16.5 to 30.6%, and then declined to
19.5%. Much greater amounts of ethylene, propylene, and CO2 were
detected at higher temperatures compared to low temperatures.
Based on their selectivity, aromatics can be categorized as
mono-ring aromatics or polycyclic aromatic hydrocarbons (PAHs).
Mono-ring products include benzene, toluene, phenol, and p-
xylene, whereas PAHs include naphthalene, methylnaphthalene,
indene, methylindene, benzofuran, fluorene, and anthracene. At
low and moderate temperatures, the mono-ring species accounted
for ∼40%, and was mostly phenol and benzene. Naphthalene,
methylnaphthalene, and benzofuran were the prominent prod-
ucts under the same environments. At the highest temperature,
most aromatics are mono-ring species, and especially toluene.
Heavy products such as fluorene and anthracene, which are
Fig. 3 shows the NH3-TPD profiles of freshly prepared catalysts.
Table 1 summarizes the peak position, full width at half maximum
Table 1
Characterization of acid sites of H-ZSM-5-based catalysts by NH3-TPD.
Sample
Acid site
type
Peak position FWHM Acid site
Total site
(mmol/g)
(◦C)
(◦C)
content (%)
H-ZSM-5
Lewis
Brønsted
177
365
47
132
16.2
84.8
0.36
0.37
0.39
0.5%
Zn/H-ZSM-5
Lewis
Brønsted
198
343
88
140
29.4
70.6
1.5%
Zn/H-ZSM-5
Lewis
Brønsted
216
338
98
136
47.2
52.8

W.-L. Fanchiang, Y.-C. Lin / Applied Catalysis A: General 419–420 (2012) 102–110
105
Fig. 3. NH3-TPD of H-ZSM-5-based catalysts.
usually considered as coke precursors [40], decreased with increas-
ing temperature. This agrees with the declining coke content.
Table 3 presents carbon yield and aromatic selectivity as func-
tions of contact time over H-ZSM-5 at 500 ◦C. Furfural conversion
was 100% for each trial. Excluding furan, all products showed a
downward trend with decreasing contact time. This suggests that
the outset of furfural conversion is the decarbonylation of furfural
to furan and CO, followed by furan conversion to other products.
With elevating contact times, the carbon yields of CO increased
steadily from 20.4 to 25.1%. This indicates a subsequent route for
furan decarbonylation to allene or methylacetylene [40].
Approximately 90% of mono-ring aromatics were generated
at GHSVs of 2412–36,000 h−1 (contact time = 1.5–0.1 s). Ben-
zene dominated at the shortest contact time. As the contact
time increased, benzene decreased monotonically with increasing
toluene. The same trend was observed for ethylene and propylene,
indicating that benzene is the primary product of furan conver-
sion. The formation of toluene and light olefins is highly correlated.
The Diels–Alder cycloaddition of furan and light alkenes may be
the source of PAHs [37]. However, this study provides no evidence
supporting this hypothesis. This may be attributed to different
pyrolysis environments: Kaliaguine and co-workers [37] employed
a contact time thousands of orders of magnitude greater than in this
study.
Table 2
Product carbon yield and aromatic selectivity as functions of temperature over H-
ZSM-5.
Temperature (◦C)
300
400
500
Carbon yield (%)
Aromatics
Coke
Furan
CO
CO2
CH4
C2H4
C3H6
Unidentified
Aromatic selectivity (%)
Mono-rings
Benzene
0.1
64.1
16.5
18.8
0.5
0
0
0
0
2.3
43.0
30.6
20.8
1.2
0
0.6
0.3
1.2
15.7
27.6
19.5
25.1
2.8
0.1
3.0
3.1
3.1
0
0
0
44.2
40.5
0.1
0
29.1
62.2
0
Toluene
p-Xylene
Phenol
3.7
0.2
PAHs
Naphthalene
Methylnaphthalene
Indene
Methylindene
Benzofuran
Fluorene
Anthracene
14.9
31.3
0.7
3.0
3.6
15.6
14.8
2.0
2.5
19.4
0.5
5.1
2.1
0.3
0.2
0.2
0.2
0.4
Fig. 4 shows the TPO signals of used H-ZSM-5 at 300, 400,
and 500 ◦C. These profiles represent the combustion of deposited
coke on the catalysts. According to Cheng and Huber [40], the
TPO patterns represent the combustion of both oxygenated and
graphite-like coke. Therefore, we deconvoluted the TPO pattern
through Gaussian peak fitting using two or three peaks. The peak
1.2
1.1
0.9
Note that furfural was completely converted.

106
W.-L. Fanchiang, Y.-C. Lin / Applied Catalysis A: General 419–420 (2012) 102–110
Table 3
Table 4
Product carbon yield and aromatic selectivity as functions of contact time over H-
ZSM-5.
Deconvolution of TPO profiles of used H-ZSM-5 at different temperatures.
Reaction (◦C)
300
Peak (◦C)
Ratio
Coke type
GHSV (h−1
2412
)
334
435
518
0.09
0.36
0.55
Oxygenated coke
Oxygenated coke
Graphite-like coke
7200
36,000
Carbon yield (%)
Aromatics
Coke
Furan
CO
CO2
CH4
C2H4
C3H6
Unidentified
Aromatic selectivity (%)
Mono-rings
Benzene
400
500
362
435
519
0.06
0.21
0.73
Oxygenated coke
Oxygenated coke
Graphite-like coke
15.7
27.6
19.5
25.1
2.8
0.1
3.0
3.1
3.1
5.3
23.3
41.7
23.4
3.0
0.1
1.6
1.6
0
0.8
12.6
64.2
20.4
0.7
0
0.6
0.6
0.1
–
439
527
0
0.18
0.82
–
Oxygenated coke
Graphite-like coke
dark red. The surface then turned light gray after reaction at 500 ◦C.
This agrees with earlier research [40], which shows the dark red
color was the deposit of oligomerized furan, whereas the gray color
is the graphite-like species.
29.1
62.2
0
56.1
42.1
0
89.4
0
0
Toluene
p-Xylene
Phenol
0.2
0.2
0.4
PAHs
3.3. Reactivity of Zn-promoted H-ZSM-5
Naphthalene
Methylnaphthalene
Indene
Methylindene
Benzofuran
Fluorene
Anthracene
5.1
2.1
0.3
0.2
0.2
0.2
0.4
0.5
0.6
0.4
0.1
0
4.6
2.8
0.7
0.9
0.3
0.4
0.5
Table 5 summarizes the product distributions of 0.5 and 1.5%
Zn/H-ZSM-5 as functions of temperature. Furfural was completely
converted. Like the trend observed for H-ZSM-5, aromatic yields
0
0
were enhanced with increasing temperature at GHSV = 2412 h−1
.
Furthermore, 1.5% Zn/H-ZSM-5 produced slightly higher aromat-
ics (∼5%) than 0.5% Zn/H-ZSM-5. Compared to untainted H-ZSM-5,
Zn-doped H-ZSM-5 catalysts yielded greater aromatics and olefins
with less furan and coke under the same reaction platforms.
Furan was completely consumed for 1.5% Zn/H-ZSM-5 at 500 ◦C,
Note that furfural was completely converted.
below 500 ◦C was regarded as oxygenated coke; beyond 500 ◦C,
graphite-like coke. Table 4 summarizes the coke composition
obtained by deconvoluting TPO. More and more graphite-like coke
was formed as the reaction temperature increased. A visual obser-
vation (Fig. S2, supplementary data) further indicates that different
types of coke were generated: after used at 300 ◦C, the surface was
GHSV = 2412 h−1
.
The selectivity of aromatics revealed different outcomes com-
pared to pure H-ZSM-5. Although PAHs still dominated at the
lowest temperature, benzene and toluene showed different trends
Fig. 4. TPO of post-reaction H-ZSM-5.

W.-L. Fanchiang, Y.-C. Lin / Applied Catalysis A: General 419–420 (2012) 102–110
107
Table 5
Table 7
Product carbon yield and aromatic selectivity as functions of temperature over 0.5%
and 1.5% Zn/H-ZSM-5.
Deconvolution of TPO profiles of used 0.5% Zn/H-ZSM-5 at different temperatures.
Reaction (◦C)
300
Peak (◦C)
Ratio
Coke type
Temperature (◦C)
335
418
505
0.08
0.40
0.52
Oxygenated coke
Oxygenated coke
Graphite-like coke
300
400
0.5
500
0.5
Zn (wt%)
0.5
400
500
348
422
517
0.09
0.27
0.64
Oxygenated coke
Oxygenated coke
Graphite-like coke
1.5
1.5
1.5
Carbon yield (%)
Aromatics
Coke
Furan
CO
CO2
CH4
C2H4
C3H6
Unidentified
Aromatic selectivity (%)
Mono-rings
Benzene
0.1
52.5
27.8
18.5
1.0
0
0
0.1
0
0
3.3
2.6
29.3
45.0
19.2
2.8
0.2
0.4
0.5
0
20.7
26.2
2.5
28.2
5.4
2.9
2.6
1.5
10.0
26.3
18.2
0
28.8
5.6
3.3
1.6
1.2
15.0
–
427
522
0
0.19
0.81
–
52.0
30.7
16.2
1.1
0
0
0
0
31.6
33.1
18.5
1.5
0.1
0.6
Oxygenated coke
Graphite-like coke
As contact time decreased, the production of aromatics and coke
declined, and furan became the prominent product. Unlike H-ZSM-
5, ∼10% aromatics were still generated by 1.5% Zn/H-ZSM-5 at the
shortest contact time. Benzene and toluene, in approximately a 2:1
ratio, are two major species at 500 ◦C. 1.5% Zn/H-ZSM-5 produced
slightly more benzene and less toluene than 0.5% Zn/H-ZSM-5.
Again, increasing Zn content may retard the alkylation of benzene.
Figs. 5 and 6 display the TPO of post-reaction 0.5 and 1.5% Zn/H-
ZSM-5. Tables 7 and 8 respectively show their peak position, coke
composition, and coke type. The peak positions for different types
of cokes were similar for all catalysts. However, as the Zn loading
increased, less and less graphite-like coke was formed compared
to those of H-ZSM-5. Moreover, the surface of 1.5% Zn/H-ZSM-
5 turned light brown, instead of light gray, after used at 400 ◦C
and above. For 1.5% Zn/H-ZSM-5, oxygenated coke became the
prominent species on the catalyst surface. The TPO response of
graphite-like coke cannot be specified in the post-reaction 500 ◦C
sample.
Zinc-modified H-ZSM-5 has been extensively studied in aroma-
tization of alkanes [41–44,48,50–52]. Aromatization is a sequence
of hydride abstraction that generates carbenium ions as precur-
sors of various products [43,51]. The Lewis acid sites newly-formed
by zinc implementation stimulate H-atom migration (also called
reverse spillover [43]) through C H activation [42]. Activated
hydrogen atoms are transferred to adsorbed carbenium by nearby
Brønsted acid sites, which substantially enhance aromatics yields.
The same could be stated in furfural conversion to aromatics:
relatively high aromatics are produced by Zn-doped catalysts com-
pared to parent H-ZSM-5. Moreover, the greater oxygenated coke
by Zn/H-ZSM-5 implies a lower degree of dehydration (hydrogen
loss) [40]. That is, more hydrogen atoms are available inside Zn-
promoted zeolites for aromatization.
0.5
10.8
0
0
0
6.5
0
0
0
10.4
43.3
45.7
0
45.5
35.5
0
70.5
26.5
0
71.6
25.0
0
Toluene
p-Xylene
Phenol
0.6
0.5
0
0
PAHs
Naphthalene
Methylnaphthalene
Indene
Methylindene
Benzofuran
Fluorene
Anthracene
20.7
18.4
0.1
6.2
46.6
0.8
32.0
27.2
0
11.3
19.1
0
4.4
3.1
0
0.6
1.7
0.3
0.3
5.7
3.0
1.0
1.3
6.1
0.7
0.7
2.2
0.8
0
0
0
2.5
0.8
0.1
0
0
0
0
0
0.7
0
0
Note that furfural was completely converted.
than H-ZSM-5: the former increased with elevating temperatures,
whereas the latter decreased. This suggests that the alkylation of
benzene and alkenes to toluene [40] was inhibited.
Table 6 lists the product distribution of 0.5 and 1.5% Zn/H-ZSM-5
at different contact times. No furfural was detected in the effluent.
Table 6
Product carbon yield and aromatic selectivity as functions of contact time over 0.5%
and 1.5% Zn/H-ZSM-5.
GHSV (h−1
)
2412
Zn (wt%)
0.5
7200
0.5
36,000
0.5
1.5
1.5
1.5
Carbon yield (%)
Aromatics
Coke
Furan
CO
20.7
26.2
2.5
28.2
5.4
26.3
18.2
0
28.8
5.6
15.8
21.8
15.5
24.3
5.9
17.3
23.3
5.1
24.8
10.1
3.7
7.5
15.8
50.1
22.5
2.0
0
12.2
18.1
38.7
24.2
3.1
3.4. Soluble intermediates
Spent H-ZSM-5 and 1.5% Zn/H-ZSM-5 at 500 ◦C and 0.67 s−1
space time were subjected to leaching study as described in Section
2.4. Cyclohexene and 3,4-dimethylbenzaldehyde were identified as
soluble intermediates. Nevertheless, these two contributed to less
CO2
CH4
2.9
3.3
2.0
0.9
C2H4
C3H6
2.6
1.5
10.0
1.6
1.2
15.0
1.8
2.0
10.9
1.4
1.8
12.5
1.0
1.1
0
0.9
1.2
0.7
Unidentified
Aromatic selectivity (%)
Mono-rings
Benzene
Table 8
Deconvolution of TPO profiles of used 1.5% Zn/H-ZSM-5 at different temperatures.
70.5
26.5
0
71.6
25.0
0
66.8
31.4
0
70.7
27.4
0
63.5
34.9
0
69.7
28.9
0
Toluene
p-Xylene
Phenol
Reaction (◦C)
300
Peak (◦C)
Ratio
Coke type
0
0
0
0
0.1
0
303
418
513
0.05
0.75
0.20
Oxygenated coke
Oxygenated coke
Graphite-like coke
PAHs
Naphthalene
Methylnaphthalene
Indene
Methylindene
Benzofuran
Fluorene
Anthracene
2.2
0.8
0
0
0
2.5
0.8
0.1
0
0
0
1.2
0.6
0
0
0
1.5
0.4
0
0
0
0.7
0.4
0.1
0.1
0
0.6
0.3
0.1
0.1
0.1
0.1
0.1
400
500
299
425
530
0.04
0.82
0.14
Oxygenated coke
Oxygenated coke
Graphite-like coke
0
0
0
0
0
0
0.1
0.1
–
0
–
0
480
–
1.0
0
Oxygenated coke
–
Note that furfural was completely converted.

108
W.-L. Fanchiang, Y.-C. Lin / Applied Catalysis A: General 419–420 (2012) 102–110
Fig. 5. TPO of post-reaction 0.5% Zn/H-ZSM-5.
Fig. 6. TPO of post-reaction 1.5% Zn/H-ZSM-5.

W.-L. Fanchiang, Y.-C. Lin / Applied Catalysis A: General 419–420 (2012) 102–110
109
3.6. Reaction mechanism
According to the aforementioned outcomes, the mechanism
of furfural catalytic fast pyrolysis can be proposed. The initiation
of furfural conversion was decarbonylation, which liberates furan
and CO as the products. This furan can then be transformed into
intermediates inside the pores of ZSM-5-based catalysts. Leaching
experiments indicate that parts of the intermediates are cyclo-
hexene and 3,4-dimethylbenzaldehyde. Oxygenated coke can be
formed concurrently from furfural or furan, particularly at low tem-
peratures.
The intermediates in ZSM-5 cages can also be transformed into
different products. At low temperatures, PAHs and oxygenated
coke were the major products for all catalysts, with trace amounts
of graphite-like coke. At high temperatures, H-ZSM-5 generated
mono-ring aromatics, especially benzene, along with some light
gases and graphite-like coke. The selectivities of toluene and light
gases improved as the contact time increased. Therefore, the alky-
lation of benzene or other intermediates with alkenes may be a
major pathway of toluene formation [39,40].
Incorporating zinc ions into H-ZSM-5 promoted the formation
of benzene, light gases, and oxygenated coke at high temperatures.
The yields of toluene and graphite-like coke declined as the zinc
content increased, and graphite-like coke could not be detected at
the highest amount of zinc loading. This indicates that zinc ions play
a significant role in the aromatization of the intermediates within
ZSM-5 pores; however, the activity of alkylation was suppressed.
Different types of coke also suggest diverse pathways for untainted
and promoted H-ZSM-5.
Fig. 7. Furfural conversion over H-ZSM-5 and 1.5% Zn/H-ZSM-5 as a function of time
on-stream.
4. Conclusions
Furfural can be converted into aromatics by catalytic fast pyrol-
ysis over ZSM-5-based catalysts. Reaction platforms and promoters
are keys to manipulate the reaction pathway. Low tempera-
tures favor the formation of PAHs and oxygenated coke, whereas
high temperatures lead to mono-ring aromatics, light gases, and
graphite-like coke. Zn-doped H-ZSM-5 can improve furan con-
version, yielding more benzene, carbon oxides, and alkenes than
untainted H-ZSM-5. H-atom transfer activity is the key for high
aromatic yields. However, this inhibits the alkylation of benzene to
toluene. This may be correlated to the Lewis acid sites newly syn-
thesized by exchanged Zn cations. This study shows that furfural
can be converted into valuable aromatics by continuous catalytic
fast pyrolysis. The reaction temperature, reactant contact time, and
catalytic promoter are critical factors in controlling the product
distribution.
than 1% carbon yields of trapped species in both catalysts. Note
that heavy compounds, which may have ∼106 or higher molecu-
lar weights and cannot be detected by GC/MS, may be formed [40].
Advanced analysis using gel permeation chromatography (GPC) or
ultraviolet-visible spectroscopy (UV–vis) should be able to provide
more qualitative and quantitative information of retained species
in ZSM-5.
3.5. Time on-stream testing
Fig. 7 displays the product distribution, except coke, as a func-
tion of time for H-ZSM-5 and 1.5% Zn/H-ZSM-5 at 500 ◦C and
GHSV = 2412 h−1. In both cases, furfural was fully converted in the
testing period. Furan gradually increased with time on-stream,
whereas CO, CO2, benzene, and toluene decreased. At the out-
set, CO selectivity was similar for both catalysts; however, furan
was completely converted on 1.5% Zn/H-ZSM-5, whereas approx-
imately 30% furan was detected for H-ZSM-5. The former also
displayed greater benzene selectivity (30%) than the latter (7%).
Slightly higher toluene and CO2 levels were also apparent on the
Zn-doped H-ZSM-5. After 1000 min time on-stream, only furan and
CO were identified at a 4:1 ratio. This indicates that furfural decar-
bonylation prevailed, and the active sites for aromatization were
deactivated.
Acknowledgment
This work was sponsored by the National Science Council of
Taiwan under grant #NSC 99-2221-E-155-075.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at: doi:10.1016/j.apcata.2012.01.017.
Fig. S3 (see supplementary data) shows the TPO patterns of
spent H-ZSM-5 and 1.5% Zn/H-ZSM-5 at 500 ◦C, time on-stream
studies. Again, graphite-like coke dominated on H-ZSM-5 whereas
some oxygenated coke appeared on the 1.5% Zn/H-ZSM-5. Because
furfural decarbonylation only occurred in the last stage, it is highly
plausible that this reaction was catalyzed nonselectively. Different
TPO profiles suggest diverse active phases produced by Zn substi-
tution, promoting furan conversion to aromatics.
References
[1] A.J. Ragauskas, C.K. Williams, B.H. Davison, G. Britovsek, J. Cairney, C.A. Eckert,
W.J. Frederick Jr., J.P. Hallett, D.J. Leak, C.L. Liotta, J.R. Mielenz, R. Murphy, R.
Templer, T. Tschaplinski, Science 311 (2006) 484–489.
[2] G.W. Huber, S. Iborra, A. Corma, Chem. Rev. 106 (2006) 4044–4098.
[3] Y.-C. Lin, G.W. Huber, Energy Environ. Sci. 2 (2009) 68–80.
[4] J.R. Salge, B.J. Dreyer, P.J. Dauenhauer, L.D. Schmidt, Science 314 (2006)
801–804.

110
W.-L. Fanchiang, Y.-C. Lin / Applied Catalysis A: General 419–420 (2012) 102–110
[5] P.J. Dauenhauer, B.J. Dreyer, N.J. Degenstein, L.D. Schmidt, Angew. Chem. Int.
[28] J.D. Adjaye, N.N. Bakhshi, Fuel Process. Technol. 45 (1995) 161–183.
[29] J.D. Adjaye, N.N. Bakhshi, Fuel Process. Technol. 45 (1995) 185–202.
[30] A. Aho, N. Kumar, K. Eränen, T. Salmi, M. Hupa, D.Y. Murzin, Fuel 87 (2008)
2493–2501.
[31] A. Aho, N. Kumar, A.V. Lashkul, K. Eränen, M. Ziolek, P. Decyk, T. Salmi, B.
Holmbom, M. Hupa, D.Y. Murzin, Fuel 89 (2010) 1992–2000.
[32] A. Pattiya, J.O. Titiloye, A.V. Bridgwater, J. Anal. Appl. Pyrolysis 81 (2008) 72–79.
[33] A. Pattiya, J.O. Titiloye, A.V. Bridgwater, Fuel 89 (2010) 244–253.
[34] R. French, S. Czernik, Fuel Process. Technol. 91 (2010) 25–32.
[35] L. Dao, M.I. Haniff, A. Houle, D. Lamothe, Am. Chem. Soc., Div. Fuel Chem. Prepr.
32 (1987) 308.
Ed. 46 (2007) 5864–5867.
[6] T.R. Carlson, T.P. Vispute, G.W. Huber, ChemSusChem 1 (2008) 397–400.
[7] T.R. Carlson, G. Tompsett, W. Conner, G.W. Huber, Top. Catal. 52 (2009)
241–252.
[8] C. Branca, P. Giudicianni, C. Di Blasi, Ind. Eng. Chem. Res. 42 (2003) 3190–3202.
[9] D. Mohan, C.U. Pittman, P.H. Steele, Energy Fuels 20 (2006) 848–889.
[10] C. Di Blasi, C. Branca, A. Galgano, Ind. Eng. Chem. Res. 49 (2010) 2658–2671.
[11] B. Zhang, M. von Keitz, K. Valentas, Biotechnol. Bioeng. 101 (2008) 903–912.
[12] D.C. Elliott, T.R. Hart, Energy Fuels 23 (2009) 631–637.
[13] M. Joyce, US DOE, Development in U.S. Alternative Fuel Market,
http://www.eia.doe.gov/cneaf/alternate/issues trends/altfuelmarkets.html#
pseries (accessed 14.11.2011).
[36] M.I. Haniff, L.H. Dao, Appl. Catal. 39 (1988) 33–47.
[37] J.-L. Grandmaison, P.D. Chantal, S.C. Kaliaguine, Fuel 69 (1990) 1058–1061.
[38] T.R. Carlson, J. Jae, Y.-C. Lin, G.A. Tompsett, G.W. Huber, J. Catal. 270 (2010)
110–124.
[14] G.W. Huber, J.N. Chheda, C.J. Barrett, J.A. Dumesic, Science 308 (2005)
1446–1450.
[15] G.W. Huber, J.A. Dumesic, Catal. Today 111 (2006) 119–132.
[16] R. Xing, A.V. Subrahmanyam, H. Olcay, W. Qi, G.P. van Walsum, H. Pendse, G.W.
Huber, Green Chem. 12 (2010) 1933–1946.
[17] C.D. Chang, A.J., Silvestri, R.L. Smith, Mobil Oil Corporation, U.S. patent 3 894
103, (1975).
[39] T.R. Carlson, Y.-T. Cheng, J. Jae, G.W. Huber, Energy Environ. Sci. 4 (2011)
145–161.
[40] Y.-T. Cheng, G.W. Huber, ACS Catal. 1 (2011) 611–628.
[41] J.A. Biscardi, G.D. Meitzner, E. Iglesia, J. Catal. 179 (1998) 192–202.
[42] J.A. Biscardi, E. Iglesia, J. Catal. 182 (1999) 117–128.
[43] C.P. Nicolaides, N.P. Sincadu, M.S. Scurrell, Catal. Today 71 (2002) 429–435.
[44] L.M. Lubango, M.S. Scurrell, Appl. Catal., A 235 (2002) 265–272.
[45] C.-L. Li, Y.-C. Lin, Catal. Lett. 140 (2010) 69–76.
[46] M. Guisnet, P. Magnoux, Appl. Catal. 54 (1989) 1–27.
[47] V.I. Yakerson, T.V. Vasina, L.I. Lafer, V.P. Sytnyk, G.L. Dykh, A.V. Mokhov, O.V.
Bragin, K.M. Minachev, Catal. Lett. (1989) 339–345.
[18] C.D. Chang, A.J. Silvestri, J. Catal. 47 (1977) 249–259.
[19] P.B. Weisz, W.O. Haag, P.G. Rodewald, Science 206 (1979) 57–58.
[20] N.Y. Chen, T.F. Degnan Jr., L.R. Koenig, Chemtech 16 (1986) 506–511.
[21] J.P. Diebold, H.L. Chum, R.J. Evans, T.A. Milne, T.B. Reed, J.W. Scahill, in: D.L. Klass
(Ed.), Energy from Biomass and Wastes X, IGT Chicago and Elsevier Applied
Sciences Publishers, London, 1987, p. 801.
[22] J.P. Diebold, J.W. Scahill, in: E.J. Soltes, T.A. Milne (Eds.), Pyrolysis Liquid from
Biomass, ACS, Washington, DC, 1988, pp. 297–307.
[48] H. Berndt, G. Lietz, J. Völter, Appl. Catal., A 146 (1996) 365–379.
[49] G.L. Woolery, G.H. Kuehl, H.C. Timken, A.W. Chester, J.C. Vartuli, Zeolites 19
(1997) 288–296.
[50] H. Berndt, G. Lietz, B. Lücke, J. Völter, Appl. Catal., A 146 (1996) 351–363.
[51] M.S. Scurrell, Appl. Catal. 41 (1988) 89–98.
[23] P.T. Williams, P.A. Horne, Biomass Bioenergy 7 (1994) 223–236.
[24] P.T. Williams, P.A. Horne, J. Anal. Appl. Pyrolysis 31 (1995) 15–37.
[25] P.A. Horne, P.T. Williams, J. Anal. Appl. Pyrolysis 34 (1995) 65–85.
[26] P.A. Horne, P.T. Williams, Fuel 75 (1996) 1051–1059.
[27] S.P.R. Katikaneni, J.D. Adjaye, N.N. Bakhshi, Energy Fuels 9 (1995) 1065–1078.
[52] T. Mole, J.R. Anderson, G. Creer, Appl. Catal. 17 (1985) 141–154.