Rake mechanism for the deoxygenation of ethanol over a supported Ni 2P/SiO2 catalyst

Article abstract of DOI:10.1016/j.jcat.2012.02.001

The catalytic conversion of ethanol was studied on a metallic Ni ₂P/SiO₂ catalyst, and comparison was made with an acidic HZSM-5 catalyst. Chemisorption probes indicated that the Ni₂P had substantial CO adsorption sites (134 μmol g^-1), while the HZSM-5 catalyst had large quantities of NH₃ adsorption sites (565 μmol g^-1). The catalytic activity in ethanol deoxygenation of the Ni ₂P/SiO₂ was higher than that of the HZSM-5 catalyst on the basis of these chemisorptions sites. In steady-state catalysis, contact time experiments indicated that for Ni₂P, acetaldehyde was a primary product and ethylene was a secondary product. However, this was not the result of a sequential oxidation reaction followed by a reduction process, but rather, it was due to the formation of a surface intermediate that could desorb as acetaldehyde or react further to produce ethylene. This rake-type mechanism was supported by a simulation of the reaction sequence that produced good agreement with the experimentally determined acetaldehyde and ethylene yields. The mechanism was also supported by in situ Fourier transform infrared measurements, which revealed the presence of signals compatible with adsorbed acetaldehyde, the likely surface intermediate species involved in the reaction. The present studies indicate that the reactions of alcohols on metallic catalysts like Ni₂P involve dehydrogenation/hydrogenation steps, rather than simple acid/base-catalyzed dehydration steps as occur in HZSM-5.

Full text of DOI:10.1016/j.jcat.2012.02.001

Journal of Catalysis 290 (2012) 1–12
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Rake mechanism for the deoxygenation of ethanol over a supported
Ni P/SiO catalyst
2
2
D. Li , P. Bui , H.Y. Zhao b,1, S.T. Oyama b,c, , T. Dou , Z.H. Shen
a,b
b
⇑
a
a
a
The Key Laboratory of Heavy Oil Processing, China University of Petroleum, Beijing 102249, PR China
Department of Chemical Engineering, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, United States
Department of Chemical Systems Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan
b
c
a r t i c l e i n f o
a b s t r a c t
Article history:
The catalytic conversion of ethanol was studied on a metallic Ni P/SiO2 catalyst, and comparison was
2
Received 17 September 2011
Revised 31 December 2011
Accepted 1 February 2012
Available online 17 April 2012
made with an acidic HZSM-5 catalyst. Chemisorption probes indicated that the Ni
2
P had substantial
adsorption
was higher than
ꢀ1
CO adsorption sites (134
l
mol g ), while the HZSM-5 catalyst had large quantities of NH
3
ꢀ1
sites (565
l
mol g ). The catalytic activity in ethanol deoxygenation of the Ni
2
P/SiO
2
that of the HZSM-5 catalyst on the basis of these chemisorptions sites.
In steady-state catalysis, contact time experiments indicated that for Ni
2
P, acetaldehyde was a primary
Keywords:
product and ethylene was a secondary product. However, this was not the result of a sequential oxidation
reaction followed by a reduction process, but rather, it was due to the formation of a surface intermediate
that could desorb as acetaldehyde or react further to produce ethylene. This rake-type mechanism was
supported by a simulation of the reaction sequence that produced good agreement with the experimen-
tally determined acetaldehyde and ethylene yields. The mechanism was also supported by in situ Fourier
transform infrared measurements, which revealed the presence of signals compatible with adsorbed
acetaldehyde, the likely surface intermediate species involved in the reaction. The present studies indi-
Ni P/SiO
2 2
HZSM-5
Ethanol deoxygenation
Rake mechanism
In situ FTIR
Kinetic simulation
cate that the reactions of alcohols on metallic catalysts like Ni
steps, rather than simple acid/base-catalyzed dehydration steps as occur in HZSM-5.
Ó 2012 Elsevier Inc. All rights reserved.
2
P involve dehydrogenation/hydrogenation
1
. Introduction
It is also of interest to study ethanol for fundamental reasons. First,
it is one of the simplest oxygenated compounds, and its deoxygen-
ation can provide insight into oxygen removal from more compli-
cated molecules. Second, it can react to produce different products
and for this reason can be used as a probe to relate the reaction
pathway to the properties of the catalysts, in particular their
acid/base properties versus their metallic nature. This study is thus
motivated in part to explore the catalytic chemistry of deoxygen-
ation and in part to determine the role of intermediates in the reac-
tion pathway. Use is made of a traditional acid catalyst, HZSM-5,
which has been shown to be an effective agent for dehydration,
The utilization of biomass for the production of fuels and chem-
icals is currently an area of great activity because of the recognition
that fossil fuels are a finite resource and because of the imminent
threat of global warming from the release of carbon dioxide [1–6].
Among the major products of biomass conversion is ethanol, a
commodity chemical derived from the fermentation of sugarcane
or energy-rich crops such as corn [7,8]. Although ethanol is pro-
duced in large quantities as a substitute or supplement for fuels
derived from crude oil, it is currently unlikely to replace much less
expensive chemical feedstocks derived from petroleum or natural
gas such as ethylene. Nevertheless, in a future with limited hydro-
carbon supplies, there may be a role for ethanol as a primary feed-
2
and a metallic catalyst, Ni P, which has been demonstrated to have
hydrogen transfer capabilities. The use of metallic catalysts is
important because studies have shown that biomass conversion
cannot be carried out by acid catalysts alone [2–4].
The decomposition of ethanol on solid catalysts typically occurs
through competing reactions [8,11–13]: (1) intermolecular dehy-
dration, which gives diethyl ether and water; (2) intramolecular
dehydration, which yields ethylene and water; (3) dehydrogena-
tion, which produces acetaldehyde and hydrogen; and (4) total
stock [9] as recently demonstrated by the start-up of
00,000 metric ton ethylene from ethanol plant in Brazil [10],
and for this reason, research in ethanol conversion is warranted.
a
2
⇑
Corresponding author at: Department of Chemical Systems Engineering, The
University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan.
2 4
decomposition into CO, H , CH , C, and O [7]. The acidity and basi-
city of the solid catalysts are two important factors that influence
their activity and selectivity. Dehydration generally occurs on
E-mail address: oyama@vt.edu (S.T. Oyama).
Present address: X-ray Science Division, Argonne National Laboratory, 9700 S.
1
Cass Ave., Argonne, IL 60439, United States.
0
021-9517/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2012.02.001

2
D. Li et al. / Journal of Catalysis 290 (2012) 1–12
acidic catalysis [14–16], and dehydrogenation usually proceeds on
basic sites [17,18].
6 h, then ground with a mortar and pestle, pelletized with a press
(Carver, Model C), and sieved to particles of 650–1180 m diame-
l
An important pathway for ethanol conversion is dehydration to
ethylene. Previous studies reported that HZSM-5 zeolite with
strong Brønsted acid sites was an effective catalyst for this transfor-
mation [19]. However, deactivation by formation of coke on its sur-
face led to decreasing activity and selectivity toward ethylene [20]
and therefore made the process unsuitable for industrial applica-
tions [21]. Phillips and Datta [22] investigated the production of
ethylene from hydrous ethanol over HZSM-5 under mild conditions
and demonstrated that strong Brønsted acid sites led to rapid cata-
lyst deactivation in the initial stages through the oligomerization of
ethylene and the formation of carbonaceous species. However,
moderating the acidity could reduce coke formation and enhance
the steady-state catalytic activity of HZSM-5.
Another important pathway for ethanol decomposition is dehy-
drogenation to acetaldehyde and hydrogen, which is favored on
basic [23] or metallic catalysts [24]. Chang et al. reported that cop-
per catalysts supported on rice husk ash displayed high catalytic
activity and selectivity toward dehydrogenation [24]. Much atten-
tion has been focused on preparing bifunctional catalysts with
both acidic and basic sites in order to promote both dehydrogena-
tion and dehydration [25–28]. Aramendía et al. [29,30] claimed
that basic sites were responsible for dehydrogenation and both
weak acid and basic sites were associated with dehydration.
Although there are a wide variety of catalysts for ethanol con-
version, there has been no report on the decomposition of ethanol
over transition metal phosphide catalysts. Transition metal phos-
phide catalysts possess excellent hydrogen transfer properties
and have been investigated extensively in hydrotreating applica-
tions. They have also been used in hydrodeoxygenation studies
and have shown promising results [31]. Therefore, it is of interest
to determine the activity of transition metal phosphides in ethanol
ter (16/20 mesh). Second, temperature-programmed reduction
(TPR) was carried out on the pelletized precursor phosphate (typ-
ically 200 mg) placed in a U-shaped quartz reactor. The sample was
ꢀ1
heated from room temperature to 1073 K at 2 K min in flowing
3
ꢀ1 ꢀ1
hydrogen at 100 cm (NTP) min
g
to reduce the metal phos-
phate to the desired phosphide and to determine the peak maxi-
mum for reduction. A portion of the exit gas flow was sampled
through a leak valve into a mass spectrometer (MS, Dycor/Ametek,
2 2 3
model MA100), and the masses 2(H ), 18(H O), 31(P), and 34(PH )
were monitored during the experiment. For large the sample sizes,
the sample was kept at the determined peak reduction tempera-
ture (883 K) for 2 h, followed by cooling to room temperature un-
3
ꢀ1
der He flow [100 cm (NTP) min ], and then passivated at room
temperature in a 0.5% O /He for 4 h. The Ni molar loading was
1.16 mmol g (mmol per g of support), corresponding to a weight
loading of Ni P of 7.9 wt.% with an initial Ni/P ratio of 1/2.
2
ꢀ1
2
2.3. Characterization
Irreversible CO uptake measurements were used to provide an
estimate of the active sites on the catalysts. Usually, 200 mg of a
passivated Ni
2
P/SiO
2
catalyst was loaded into a U-shaped quartz
reactor, and then the sample was reduced in flowing H
2
[200 cm3
ꢀ1
(NTP) min ] at 723 K for 2 h. For HZSM-5, the pretreatment tem-
perature was 773 K using He as the purging gas. After cooling in He
3
ꢀ1
ꢀ1
ꢀ1
[200 cm (NTP) min ], pulses of CO in a He carrier at 43 lmol s
3
[65 cm (NTP) min ] were injected at room temperature through
a sampling valve. The mass 28 (CO) signal was monitored with a
mass spectrometer (MS, Dycor/Ametek, model MA100). CO uptake
was calculated by measuring the decrease in the peak areas caused
by adsorption in comparison with the area of a calibrated volume
2 2
conversion. In the present study, a Ni P/SiO catalyst was chosen
for catalyzing the ethanol decomposition process because it is gen-
erally the most active of the transition metal phosphides for a vari-
ety of reactions (HDS, HDN, HDO) [32–36,31]. Comparison is made
to HZSM-5, the most widely studied acid catalyst.
(19.5 lmol).
Surface areas of the samples were obtained using the BET meth-
od based on adsorption isotherms at liquid nitrogen temperature,
using a value of 0.162 nm for the cross-sectional area of a N
2
2
mol-
ecule. The measurements were taken in a volumetric adsorption
unit (Micromeritics ASAP 2000). X-ray diffraction (XRD) patterns
of the samples were obtained with a PANalyticalX’pert Pro powder
2
. Experimental
diffractometer operated at 45 kV, using CuK
radiation (k = 0.154178 nm).
a monochromatized
2.1. Materials
Transmission infrared spectra of pyridine adsorbed on the Ni
2
P/
The HZSM-5(Si/Al = 15) commercial catalyst was obtained from
Zeolyst International. The transition metal phosphide Ni P/SiO
2 2
was prepared on a fumed silica EH-5 support provided by the Ca-
bot Corp. The chemicals used in the synthesis of the catalyst were
SiO catalyst and HZSM-5(Si/Al = 15) were collected to characterize
2
the surface acidic sites. Fourier transform infrared (FTIR) spectra
measurements were carried out with a Digilab Excalibur Series
FTS 3000 spectrometer equipped with a liquid N cooled mer-
2
cury–cadmium–telluride detector. The IR cell was equipped with
water-cooled KBr windows, connections for inlet and outlet flows,
and thermocouples connected to a temperature controller to mon-
itor and control the sample temperature. For the experiments,
Ni(NO
The chemical utilized in the reactivity study was ethanol (Decon
Laboratories, Inc., 200 Proof). The gases employed were H (Airco,
Grade 5, 99.99%), He (Airco, Grade 5, 99.99%), CO (Linde Research
Grade, 99.97%), 0.5% O /He (Airco, UHP Grade, 99.99%), O (Airco,
UHP Grade, 99.99%), 10% CO (Airco, UHP Grade, 99.99%), and
NH (Alexander Chemical Corporation, AH200).
3
)
2
ꢁ6H
2 4 2 4
O (Alfa Aesar, 99%) and (NH ) HPO (Aldrich, 99%).
2
2
2
2 2
about 25 mg of finely ground Ni P/SiO catalyst, or HZSM-5 samples
2
were pressed into self-supporting wafers with a diameter of 13 mm
3
ꢀ2
(
18.8 mg cm ). Wafers were mounted vertically in a quartz sample
holder to keep the incident IR beam normal to the sample. Before
dosing pyridine, the sample of Ni P/SiO was reduced in H , and
HZSM-5 was purged in He flow at 723 K and 773 K, respectively,
2
.2. Nickel phosphide (Ni
2
P/SiO
2
) synthesis
2
2
2
ꢀ
1
3
ꢀ1
The Ni P/SiO catalyst was prepared by temperature-pro-
2
2
for 2 h at a flow rate of 100
lmol s [150 cm (NTP) min ]. After
grammed reduction (TPR), following procedures reported previ-
ously [38–40]. Briefly, the synthesis of the catalysts involved two
stages. First, a solution of the corresponding metal phosphate
precursor was prepared by dissolving appropriate amounts of
pretreatment, the samples were slowly cooled in flowing He
3
ꢀ1
[200 cm (NTP) min ], and background spectra were collected at
423 K under He flow. The samples were dosed at atmospheric pres-
sure and room temperature with 1.0 mol% pyridine in He carrier at
ꢀ1
3
ꢀ1
3
Ni(NO )
2
ꢁ6H
2
O with ammonium phosphate in distilled water, and
a total flow rate of 140 lmol s [200 cm (NTC) min ] until satu-
the solution was used to impregnate silica by the incipient wetness
method. The obtained samples were dried and calcined at 773 K for
ration was achieved. The samples were then purged with carrier gas
for 0.5 h to remove gaseous and weakly adsorbed pyridine. Spectra

D. Li et al. / Journal of Catalysis 290 (2012) 1–12
3
were acquired at 423 K and are shown with subtraction of the back-
ground contribution to highlight the pyridine adsorbate peaks.
The reaction products were analyzed using an online mass spec-
trometer (Dycor/Ametek, model MA100). The conversion and prod-
uct yield were evaluated by determining the amount of ethanol
reacted and products formed using the formulas below.
2 2
Ethanol FTIR measurements were carried out with Ni P/SiO
catalyst in the same manner as with the pyridine experiments.
After pretreatment, background spectra were collected under He
flow, and then ethanol was introduced with He flow until satura-
tion was achieved.
N ðethanolÞin ꢀ N ðethanolÞout
Conversion ð%Þ ¼
ꢂ 100
N ðethanolÞin
Ammonia, pyridine, and carbon dioxide temperature-pro-
3 2
grammed desorption (NH -TPD, pyridine-TPD, and CO -TPD) were
Yield ð%Þ ¼ Fractional conversion ꢂ Selectivity ꢂ 100
used to determine the quantities of acid and base sites on the cat-
alysts. TPD experiments were carried out by loading catalyst sam-
ples (200 mg) in U-shaped quartz reactors connected to a MS
pretreating as in the CO uptake experiments, and then dosing
3
. Results and discussion
3 2
NH or CO or a mixture of 2.6 mol% pyridine in dry He flow from
3
.1. Synthesis and basic characterization
a bubbler containing pyridine at room temperature onto the sam-
ples at room temperature until saturation. Then, samples were
3
ꢀ1
The supported nickel phosphide was prepared in two stages as
purged with pure He for 1 h at a flow rate of 200 cm (NTP) min
to remove physisorbed NH or pyridine or CO and then heated
from room temperature to 973 K at 10 K min . The masses 17
NH ) or 79 (pyridine) or 44 (CO ) were monitored by the MS dur-
described in the experimental part. First, solutions of the nickel
and phosphorous components were impregnated on the silica sup-
port, and the material was dried to form supported phosphate pre-
cursors. Second, the phosphate was transformed into a phosphide
by temperature-programmed reduction (TPR). The TPR experiment
was carried out to understand the phenomena involved in the
reduction process and to determine the optimum reduction condi-
tion used for large-scale catalyst preparation (Fig. 1a). A sample of
HZSM-5 was also included in the studies because it is among the
most effective of reported catalysts and serves as a comparison
3
2
ꢀ
1
(
3
2
ing the experiments. Peak areas were quantified by comparison
with the areas of calibrated pulses of the pure compounds.
Ethanol temperature-programmed desorption (EtOH-TPD) was
used to compare the catalytic activity and product selectivity of
ethanol over the two different catalysts. To start a reaction, cata-
lysts (200 mg) were placed in a U-shaped reactor and pretreated
at the same conditions as used for chemisorption. After pretreat-
ment, a flow of He saturated with ethanol was introduced at
2
with the Ni P material. Modification of ZSM-5, though interesting,
would detract from the treated subject.
3
ꢀ1
2
00 cm (NTP) min for 1 h at room temperature. After purging
ꢀ1
in He flow for 1 h, the temperature was raised at 10 K min to
ꢀ1
1
073 K in He flow [200 cm3 (NTP) min ]. The reaction products
a
were analyzed using an online mass spectrometer (Dycor/Ametek,
model MA100). The conversion of ethanol and the product selectiv-
ity for each sample were calculated and compared. The products
were quantified using reaction response factors (RRF). The RRF
were determined experimentally calibrating the mass spectrome-
ter signals with pure standards of known concentration.
883K
2.4. Reactivity studies
Activity tests were conducted in a U-shaped reactor at atmo-
spheric pressure using catalyst amounts corresponding to
21 mol of active sites. For nickel phosphide, the catalyst amount
903 mg) was based on the CO uptake (134
HZSM-5, the catalyst amount (214 mg) was based on the NH
1
(
l
ꢀ1
l
mol g ), and for
up-
lmol g ). To start, the catalyst was pretreated at the
3
400
500
600
700
800
900
1000 1100
ꢀ1
take (565
Temperature / K
same conditions as used for chemisorption. After pretreatment, a
mixture of 1.5 mol% ethanol in dry He flow from a bubbler contain-
ing ethanol at 273 K was passed onto the sample, and then the cat-
alyst was stabilized for 0.5 h after the feed was introduced.
Reactivity testing was performed as a function of temperature,
starting at the highest temperature of 523 K, and was varied down-
ward and upward with the initial temperature repeated at the end
to establish catalyst stability. Mass balances closed within 5%.
Reactivity tests were also investigated as a function of contact
time. The contact time is defined by the following equation, with
the quantity of sites obtained from CO uptake experiment.
b
Ni P/SiO
2
2
mol g 1Þ ꢂ Catalyst weight ðgÞ
ꢀ
Quantity of sites ð
l
Contact time ðsÞ ¼
ꢀ
1
PDF reference
Reactant flowrate ð
l
mol s
Þ
The experiment started at the highest contact time of 270 s and was
varied downward and upward with the initial contact time re-
peated at the end. Generally, it took about 10 h of on stream time
to collect the rate data at several reaction temperatures or contact
times, and the smoothness of the data indicated no deactivation.
2
0
30
40
50
60
70
2 2
Fig. 1. Synthesis and XRD analysis of Ni P/SiO : (a) mass 18 signal from TPR and (b)
X-ray diffraction pattern.

4
D. Li et al. / Journal of Catalysis 290 (2012) 1–12
Table 1
leave. Gayubo et al. [46] studied the selectivity of olefin products
from bioethanol and demonstrated that higher mesoporosity and
moderate acid strength were suitable for the decomposition of
bioethanol.
2 2
The true width at half-maximum of the peak and crystallite size of Ni P/SiO catalyst.
Peak (deg)
Direction
True width (rad)
Crystallite size (nm)
4
4
4
5
0.5
4.8
7.5
4.2
111
201
210
300
0.009
0.007
0.006
0.010
17
22
25
15
3.2. Temperature-programmed desorption (TPD) of NH
3
, pyridine, and
CO
2
Average
–
0.008
20
Measurements of NH
3
[47,48], pyridine, and CO
2
[49,50] tem-
perature-programmed desorption (TPD) were carried out in order
to clarify the relationship between the catalyst activity and the
amount of acidic and basic sites on the catalysts. The results are
shown in Figs. 2–4 and summarized in Table 3. Although the signal
is reported with arbitrary units, the peaks were quantitated by
The water evolution (mass 18) shows a reduction peak at 883 K.
Only the results for mass 18 (H O) are shown, because the other
monitored masses provided little additional information. This tem-
perature was chosen for preparation of larger-scale samples used
for characterization and testing.
Analysis of the product of TPR was carried out by XRD (Fig. 1b).
The diffraction pattern for the silica-supported nickel phosphide
shows three major peaks at 40.5°, 44.8°, and 47.5°, which line up
2
injection of known quantities of NH
Relevant to the NH -TPD results, previous studies showed that
[51] peak maxima, T , from 293 K to 473 K correspond to weak acid-
ity sites, T from 473 K to 673 K to intermediate acidity sites, and T
higher than 673 K to strong acidity sites. The HZSM-5 sample shows
two peaks at T of 500 K and 724 K, respectively (Fig. 2), clearly sug-
3 2
, pyridine, and CO .
3
m
m
m
well with the standard pattern for Ni
2
P. The peaks are significantly
broadened, indicating that small Ni
2
P crystals were formed. The
m
XRD pattern after reaction was essentially unchanged, indicating
that the catalyst was stable.
gesting the presence of moderate and strong acidic sites. The ratio of
the moderate acidic sites to the strong ones is 0.75. Bi et al. [52] and
Post and Hoof [53] reported that ethylene was easily polymerized
on the strong Brønsted acid sites of HZSM-5 and suggested that
elimination of the strong acidic sites could improve the stability of
the catalyst. The Ni P/SiO catalyst exhibits one broad peak from
2 2
402 K to 805 K with a maximum around 475 K, indicating that the
catalyst has plenty of moderate acidity sites. The acid sites are prob-
The width at half-maximum was corrected for instrumental
broadening (0.2°), and the crystallite size of the supported Ni
was calculated using the Scherrer equation, D = Kk/bcos(h), where
K is a constant taken as 0.9, k is the wavelength of the X-ray radiation
0.154178 nm), b is the peak width in radians at half-maximum, and
h is the Bragg angle. Table 1 reports the true width at half-maxi-
2
P
C
(
2
mum of the peak and the crystallite size.
There is no substantial anisotropy with crystallographic direc-
tion, indicating that the crystallites are spherical, in agreement
with high-resolution transmission electron microscopy results
from the group of Bussell [41].
475K
The CO chemisorption and BET characterization results are re-
ported in Table 2. Earlier studies have shown that uptakes of CO
on SiO
uptake of the Ni
2
and Al
2
O
3
were negligible [42–44]. The CO chemisorption
ꢀ1
2
P/SiO
2
catalyst was 134
l
mol g , which is in line
500K
with previous measurements. The CO uptake of HZSM-5 was small
and was probably due to CO interaction with strong Lewis acid
sites, as HZSM-5 does not contain metal. Both the phosphide and
the HZSM-5 samples have high BET surface areas, 310 and
Ni P/SiO
2 2
724K
700
2
ꢀ1
HZSM-5
4
10 m g , respectively. The BET surface area of Ni
2
P/SiO
2
is lower
2
ꢀ1
than that of the SiO
2
support of 350 m g , and this was likely
400
500
600
800
900
1000
caused by sintering during the preparation process. Additionally,
Table 2 also lists the pore volumes of the two catalysts obtained
Temperature / K
from N
types of pores in Ni
and micropores, respectively.
2
physisorption measurements and shows that the major
3
Fig. 2. Mass 17 signal from NH -TPD.
2
P/SiO and HZSM-5 catalysts are mesopores
2
It has been reported [45] that micropores are associated
with coke formation that results in the zeolite’s loss of activity dur-
ing the ethanol dehydration process. The micropore size limits the
size of the channels, cavities, and channel intersections, trapping
the species responsible for coking. Mesopores are more active in
the ethanol reaction, for the larger size allows easier passage of
the reactant for reaching the active sites and for the products to
5
30K
5
14K
Ni P/SiO
2
2
Table 2
Characterization results for HZSM-5 and Ni
2
P/SiO
2
catalysts.
Pore volume (cm3
991K
ꢀ1
Catalyst CO uptake BET area
g
)
ꢀ
1
2
ꢀ1
(l
mol g
)
(m g )
HZSM-5
a
b
V
micro
V
meso
V
tot
Ni
HZSM-5
2
P/SiO
2
134
11
310
410
0.009
0.124
0.083
0.036
0.092
0.160
4
00 500 600 700 800 900 1000 1100 1200 1300
Temperature / K
a
Micropore volume from t-plot.
b
V
total ꢀ Vmicro
.
Fig. 3. Mass 79 signal from pyridine-TPD.

D. Li et al. / Journal of Catalysis 290 (2012) 1–12
5
in the FTIR spectrum are used to determine whether pyridine is
protonated through the nitrogen atom by surface Brønsted acid
sites or bonded to coordinative unsaturated metal sites (Lewis
acids). Upon interaction with a Brønsted acid site, pyridine is pro-
tonated to a pyridinium ion and adsorbs with a characteristic band
3
89K
ꢀ1
around 1545–1540 cm . Interaction of pyridine with Lewis acid
sites leads to a coordinatively bonded pyridinium complex with a
Ni P/SiO
2
2
373K
ꢀ1
well-resolved band centered around 1452–1447 cm . A band
ꢀ1
located around 1490 cm is common to both adsorbed species.
Many studies have employed this technique to qualitatively and
quantitatively study the acidic properties of catalytic materials
and the resulting effect on catalyst properties and activity. Fig. 5
displays FTIR spectra of adsorbed pyridine in He flow on HZSM-5
HZSM-5
900
4
00
500
600
700
800
1000
and Ni
sites. HZSM-5 contains mostly Brønsted acid sites, whereas Ni
SiO has more Lewis acid sites.
2
P/SiO
2
catalysts at 423 K and shows that both possess acidic
Temperature / K
2
P/
2
2
Fig. 4. Mass 44 signal from CO -TPD.
The quantity of Brönsted acid sites and Lewis acid sites was cal-
culated using the following equations [64], and the results are pre-
sented in Table 3:
ably associated with phosphorus on the support. In summary, the
NH -TPD characterization results show that Ni P has weak and
moderate strength acid sites, while HZSM-5 has stronger acid sites
the strongest of which might lead to side reactions.
3
2
2
p
R
CðBÞ ¼ IMECðBÞꢀ ꢂ IAðBÞ ꢂ
1
W
Previous references noted that [54,55] pyridine is a strong base
in the gas phase and requires higher temperature than ammonia
for complete desorption from catalysts. Fig. 3 displays the pyri-
dine-TPD profiles and shows that all peak maxima occur at higher
2
p
R
CðLÞ ¼ IMECðLÞꢀ1 ꢂ IAðLÞ ꢂ
W
where C is the concentration (
l
mol/g catalyst), IMEC(B, L) the inte-
mol), IA(B, L) the inte-
temperature than the corresponding NH
Ni P/SiO , 514 K and 991 K for HZSM-5, respectively). The conclu-
sions are similar to those obtained with NH -TPD, with Ni P/SiO
2
3
-TPD maxima (530 K for
grated molar extinction coefficients (cm/
l
2
2
ꢀ1
grated absorbances (cm ), R the radius of catalyst disk (cm) and
3
2
W is the weight of disk (mg).
exhibiting moderate strength acidic sites and HZSM-5 possessing
both moderate and strong acidic sites in a ratio of 0.60.
Fig. 4 shows the TPD profiles of CO from the Ni P/SiO and
2 2 2
HZSM-5 samples, with peak maxima located at 389 K and 373 K,
respectively. Table 3 summarizes the quantity of acidic and basic
Comparing the quantity of acid sites obtained by the TPD and
FTIR methods, it is found that on Ni
2
P, the NH
3
TPD amount
ꢀ
1
(
(
381
326
l
l
mol g ) is slightly larger than the pyridine FTIR amount
mol g ). This also is found on HZSM-5 where the TPD
mol g ) is larger than the pyridine FTIR amount
mol g ). There are two possible reasons for this. First,
ꢀ1
ꢀ1
amount (565
l
2 2
sites. The results suggest that Ni P/SiO has fewer acidic sites
ꢀ
1
(
486
l
and more basic sites than HZSM-5. Previous FTIR characterization
of acid sites indicates that there are SiAOH and PAOH groups on
the support that contribute to the total acidity [56–58]. The quan-
tity of probe molecules on HZSM-5 is generally larger than on the
ammonia is smaller than pyridine, so it will pack more closely on
the surface as well as enter more readily into smaller micropores.
2 2
Ni P/SiO , reflecting the large contribution of the total surface of
the material to the adsorption.
The influence of acidic and basic centers on dehydration has
been well studied. Basic and acidic sites are necessary for the dis-
sociative adsorption of ethanol [59,60]. Golay et al. [61] investi-
gated the influence of the catalyst’s acid/base properties on the
catalytic ethanol dehydration. They suggested that for the forma-
tion of ethylene, a simultaneous adsorption on both acidic sites
and basic sites was necessary and that the dehydration was af-
fected by both the surface acidity and the sorption kinetics on
the basic sites.
2
Ni P/SiO2
HZSM-5
3.3. Infrared spectroscopy of pyridine
1560
1520
1480
1440
1400
Pyridine adsorption is often combined with in situ Fourier
transform infrared spectroscopy (FTIR) to probe the surface acidic
properties of supports and catalysts [62,63]. Characteristic bands
Wavenumbers / cm-1
Fig. 5. Infrared spectra of adsorbed pyridine over samples at 423 K.
Table 3
Acid and base properties of HZSM-5 and Ni
2
P/SiO
2
catalysts.
NH
mol gꢀ1)
-TPD (
l
mol g
ꢀ1
)
Pyridine FTIR
Samples
CO
2
-TPD (
l
3
B (
36
428
l
mol g 1
ꢀ
)
L (
l
mol g
ꢀ1
)
B + L (
l
mol g
ꢀ1
)
B/L
Ni
HZSM-5
2
P/SiO a
2
408
218
381
565
290
58
326
486
0.12
7.38
a
ꢀ1
ꢀ1
The Ni
2
P/SiO
2
CO
2
-TPD amount is corrected by 149
l
mol g from the SiO
2
support, and the NH
3
-TPD amount is corrected by 229
l
mol g from the SiO
2
support.

6
D. Li et al. / Journal of Catalysis 290 (2012) 1–12
Second, the purge temperature of ammonia was room temperature,
whereas that of pyridine in the FTIR measurements was 423 K, so
that some weakly bound pyridine was desorbed.
Phillips and Datta [22] reported that Brønsted acid sites were
involved in the HZSM-5-catalyzed ethanol dehydration to ethyl-
ene; however, oligomerization of ethylene also occurred at the
same sites forming carbonaceous deposits that covered these sites
and dramatically reduced the catalytic activity. Increasing Lewis
acid sites and decreasing Brønsted acid sites were favorable for
the catalytic decomposition of ethanol [65].
3.4. Ethanol temperature-programmed desorption
Figs. 6a and 6b show the signal of ethylene and acetaldehyde,
respectively, obtained from ethanol temperature-programmed
desorption (EtOH-TPD). Fig. 6a shows that ethylene formation over
HZSM-5 occurred at around 419 K and 533 K and over Ni
occurred at around 407 K and 584 K. Thus, ethylene started to form
2 2
P/SiO
and desorb from the Ni
from HZSM-5.
2 2
P/SiO catalyst at a lower temperature than
Fig. 6b reveals the desorption of acetaldehyde. The formation of
acetaldehyde over HZSM-5 occurred at around 422 K and 541 K.
Over Ni P/SiO , acetaldehyde desorbed at 411 K and 497 K.
2 2
Table 4 summarizes the distribution of ethylene and acetalde-
hyde over different temperatures in the EtOH-TPD experiments
2 2
The number of sites titrated by CO is surprisingly high. For Ni P,
ꢀ
1
ꢀ1
the quantity (408
l
mol g ) exceeds the acidic sites (381
l
mol g
)
and the surface metal sites from CO chemisorption (134
l
mol gꢀ1).
ꢀ
1
For HZSM-5, the quantity of CO
/3 the total acid amount (565
usually attributed to basic sites, but in this case for Ni
2
adsorption (218
l
l
mol g ) is about
mol g ). The adsorption of CO is
P, there is
ꢀ
1
1
2
over HZSM-5 and Ni
Over HZSM-5, strong ethylene and acetaldehyde desorption
was recorded at 530 K and 541 K, respectively. Over Ni P/SiO , an
almost similar amount of ethylene was observed at 407 K and
84 K, and acetaldehyde desorption occurred primarily at 497 K,
with a smaller peak at 411 K. The Ni P/SiO catalyst shows reactiv-
2 2
P/SiO catalysts.
2
likely to be some adsorption on the metallic sites. For both Ni
and HZSM-5, there are likely contributions from physisorption.
2
P
2
2
5
2
2
ity at lower temperature than HZSM-5 particularly for acetalde-
hyde, which indicates that it may be an initial product. This will
be confirmed later.
5
33 K
Fig. 7 shows the conversion of ethanol and the selectivity
toward products from EtOH-TPD over the two catalysts.
2 2
Overall, Ni P/SiO has a little higher conversion of ethanol (57%)
than HZSM-5 (56%), but also exhibits a higher selectivity toward
acetaldehyde (21%) than HZSM-5 (4%), and a lower selectivity t-
oward ethylene (57%) than HZSM-5 (96%). At the same time, car-
bon monoxide, a product of conversion from acetaldehyde, was
4
19 K
HZSM-5
4
07 K
2 2
detected at up to 20% in the product stream on the Ni P/SiO cat-
5
84 K
600
Ni P/SiO2
alyst along with a small quantity of methane (3%). In contrast,
there was no carbon monoxide and methane detected from
HZSM-5. The decomposition of ethanol has been studied exten-
2
300
400
500
700
sively on different metallic catalysts such as Pt/Al
2
O
3
[66], CuO
Temperature / K
with 5% CoO and 1% Cr
2
O
3
[67], Pt/ZrO
2
[68], and supported Au
[
69], and it has been reported that the dehydrogenation of ethanol
Fig. 6a. Signal of ethylene from EtOH-TPD.
yielded acetaldehyde and hydrogen: C
2
5 3 2
H OH ? CH CHO + H (1).
It is likely that the methane detected with the Ni
2
P/SiO
2
came from
541 K
1
00
100
80
60
40
20
0
Conversion of ethanol
Acetaldehyde selectivity
Methane selectivity
Ethylene selectivity
8
6
4
2
0
0
0
0
0
Carbon monoxide selectivity
4
22 K
4
97 K
411 K
300
400
500
600
700
HZSM-5
Ni P/SiO
2 2
Temperature / K
Fig. 6b. Signal of acetaldehyde from EtOH-TPD.
Fig. 7. Conversion and Selectivity from EtOH-TPD.
Table 4
Distribution of ethylene and acetaldehyde in different temperature ranges on EtOH-TPD.
Catalyst
HZSM-5
2 2
Ni P/SiO
Maximum Temperature (K)
Percentage of ethylene (%)
Percentage of acetaldehyde (%)
419
20
–
422
–
16
533
80
–
541
–
84
407
50
–
411
–
28
497
–
72
584
50
–

D. Li et al. / Journal of Catalysis 290 (2012) 1–12
7
an intermediate acetaldehyde which decomposed into methane
and carbon monoxide: CH CHO ? CH + CO (2). It should be noted
that in the presence of water (a dehydration by-product), the CH
steam-reforming reaction may also occur to produce CO and H
CH + H O ? CO + 3H (3). Farkas and Solymosi [70] studied the
adsorption, desorption, and dissociation of ethanol on MoC/Mo
100). They obtained a product mixture of 18% of ethylene, 12%
of acetaldehyde, and 44% of carbon monoxide. The formation of
methane, however, was not reported. They suggested the reason
x
for no methane formation was that no CH species formed from
Table 6
Reactivity in ethanol deoxygenation (contact time = 162 s).
3
4
4
Catalyst
Conversion (%)
48 K
2
:
4
473 K
498 K
523 K
4
2
2
2
Ni P/SiO
2
2
41
9
54
25
ꢀ
86
68
100
91
HZSM-5
(
Turnover frequency (s 1
)
2
Ni P/SiO
0.0025
0.00057
0.0033
0.0016
0.0053
0.0042
0.0062
0.0056
HZSM-5
rupture of a CAC bond on the catalyst surface during the dehydro-
genation process. These results are in agreement with the finding
from this study that the amount of methane was much smaller
than that of carbon monoxide.
Apparent activation energies were calculated from fits to the
ꢀ1
conversion curves at low conversion and were 39 kJ mol
for
ꢀ1
Under our conditions of EtOH-TPD, the Ni
2
P/SiO
2
displayed bet-
2 2
HZSM-5 and 46 kJ mol for Ni P/SiO .
ter ability for the conversion of ethanol and selectivity toward
dehydrogenation than HZSM-5. Dehydrogenation is generally ac-
Fig. 8 shows the ethanol conversion and product yields as a func-
tion of contact time over the HZSM-5 catalyst at 473 and 498 K. It
can be seen that the conversion increased with time and tempera-
ture, as expected. The product formed in highest yield was ethylene.
At 473 K (Fig. 8a), the ethylene decreased slightly at high contact
time, and this was accompanied by a corresponding growth in
butylenes, indicating that part of the ethylene might have been con-
sumed to form the C4 products. Acetaldehyde was produced at low
contact time together with ethylene, suggesting that these C2 spe-
cies were produced in parallel. However, acetaldehyde did not grow
appreciably with time, suggesting that it was an intermediate that
also reacted to form the butylenes. At 498 K (Fig. 8b), ethylene grew
monotonically with contact time. Acetaldehyde was formed in
small amounts and went through a shallow maximum at short con-
tact time, confirming that it is an intermediate. The butylenes were
cepted to occur on basic sites and metallic sites. On Ni
2
P, the up-
take of CO is substantial (Table 3), but as discussed earlier, some
2
of this may be due to chemisorption on metallic sites, and a consid-
erable portion in physisorbed mode. It is more likely from the
known hydrogen transfer capabilities of the phosphide that dehy-
drogenation is occurring on metallic sites.
In order to ensure that the TPD results were not affected by the
lag time for gas to diffuse out of pores, an analysis was carried out
using the criterion of Gorte [71] and Ibok and Ollis [72]. These
researchers suggested that the effect of diffusion limitations could
2
bl
e
be ignored for a value of less than 0.01 for the group
where b
f
0
ðT ꢀT ÞD
is the heating rate (K/s), l is the width of catalyst slab (cm),
e
is the
3
3
porosity (cm /cm ), T
f 0
and T are the final and initial temperature
2
(
K), respectively, and D is the effective diffusivity (cm /s). For our
ꢀ5
conditions, a highest value of 6 ꢂ 10 is obtained (Table 5), and
1
00
100
80
60
40
20
0
this indicates no diffusion limitations during the TPD experiments.
(a) HZSM-5
Conversion
Ethylene
8
0
3.5. Reactivity
The conversion of ethanol as a function of temperature was
60
compared for HZSM-5 and Ni
2
P/SiO
mol g ). For HZSM-5, the main product at all
temperatures was ethylene, with some acetaldehyde, and a butyl-
ene isomer (not identified) also formed. On Ni P/SiO , the main
2
with equal sites loaded in
ꢀ1
the reactor (121
l
4
2
0
0
0
2
2
product was initially acetaldehyde, but this decreased with
increasing temperature, and ethylene became the main product
at high temperature. These results are consistent with previous
Acetaldehyde
473K
Butylene
}
studies [25–30]. In general, Ni
2
P/SiO
2
exhibited higher activity
448K
498K
523K
than HZSM-5 at all temperatures. Turnover frequencies were ob-
tained using the following formula:
Temperature / K
1
00
100
80
60
40
20
0
ꢀ
ꢁ
ꢀ
1
Reactant flowrate ð
l
mols ÞꢂConversion
(b) Ni P/SiO2
Turnover frequency ðsꢀ1Þ ¼
2
ꢀ1
Quantity of sites ð mol g ÞꢂCatalyst weight ðgÞ
l
8
0
0
A summary is provided in Table 6. Calculations of the Weisz–
Conversion
Prater criterion (CWP) were made to ascertain that no mass transfer
limitations were present (see Supplementary Information). The
highest CWP gave 0.030 ꢃ 1, indicating that internal mass transfer
effects can be neglected at the conditions employed for the reactiv-
ity study in this paper.
Ethylene
6
40
2
0
Acetaldehyde
Table 5
Parameters in the Gorte criterion.
0
b
l
e
Heating rate (K sꢀ1)
0.17
0.09
0.45
110
0.1
448K
473K
498K
523K
Width of catalyst slab (cm)
Temperature / K
3
ꢀ3
Porosity (cm cm )
T
D
f
ꢀ T
0
Temperature difference (K)
Effective diffusivity
Fig. 8. Variation of ethanol conversion and product yield as a function of reaction
temperature at a contact time of 162 s: (a) HZSM-5 and (b) Ni P/SiO
2
2
.

8
D. Li et al. / Journal of Catalysis 290 (2012) 1–12
formed in small but growing amounts, indicating that they were
final products. Overall, the results indicated that ethylene and
acetaldehyde were formed in parallel and reacted by a condensa-
tion reaction to form butylenes. At higher temperatures, the ethyl-
ene was the preferred product, and the acetaldehyde was largely
consumed to form the butylenes, resulting in a smaller acetalde-
hyde yield than at lower temperatures.
ance of 100 ± 5% indicate that the maximum acetaldehyde yield is
not due to its consumption to form products such as butylenes (not
detected) or heavy materials like polymers.
It should be noted that no ethane was detected. At the condi-
tions of this study, the concentration of the ethanol reactant is
low (1.5 mol%), and since the concentration of products is signifi-
cantly lower, it is difficult for the hydrogen (from dehydrogena-
tion) and ethylene (from dehydration) to come into contact to
form ethane. Furthermore, the ethane formation reaction is
favored by high temperature and pressure, which are not the con-
ditions employed in our study. Thus, the lack of formation of eth-
ane is understandable. Other studies report acetaldehyde and
ethylene formation but not ethane [20,26].
Fig. 9 shows the ethanol conversion and yields of acetalde-
hyde and ethylene as a function of contact time at 473 and
4
2 2
98 K on Ni P/SiO . The curves are fits to the data that will be
discussed later. The figure indicates that conversion and the
yields of products increase with contact time. The data at
4
73 K (Fig. 9a) show both ethylene and acetaldehyde growing
with time. From these data alone, it is not possible to deduce
the sequence of reaction steps. Indeed, it could be that acetalde-
hyde and ethylene are produced in parallel. As will be shown,
this is not the case. The data at 498 K (Fig. 9b) this time show
that acetaldehyde goes through a maximum, while ethylene rises
monotonically, behaviors characteristic of a sequential network.
This result is in accordance with a previous report [73] in which
the conversion of ethanol increased with increasing contact
time along with a drop in acetaldehyde selectivity. As will be
discussed in the modeling section, both sets of data can be
described by a reaction sequence involving adsorbed intermediates
undergoing consecutive reactions.
Reaction temperature and contact time play important roles in
the decomposition of ethanol, and their effect on the steady-state
reactivity of HZSM-5 and Ni P/SiO was examined. To summarize,
2 2
on HZSM-5, the main product was ethylene with small amounts of
acetaldehyde and butylenes. The formation of the ethylene can be
3
attributed to acid sites, which NH and pyridine titration showed
to be plentiful, while the formation of acetaldehyde can be associ-
ated with basic sites, which also are present in smaller quantities.
Butylenes formed at long contact time are likely formed by con-
densation of ethylene and acetaldehyde on strong Brønsted acid
sites. The reaction network can be described as a parallel network
in which ethylene and acetaldehyde are produced together by dif-
ferent mechanisms.
It should be emphasized that the measurements were carried
out by varying the contact time up and down over a period of
2 2
On Ni P/SiO , the main products of the ethanol reaction were
1
0 h, so that the smoothness of the data provides evidence for
ethylene and acetaldehyde with small amounts of carbon monox-
ide and trace quantities of methane, without ethane. The reaction
likely starts with the formation of an ethoxide species on the silica
which subsequently undergoes dehydrogenation/hydrogenation
the lack of deactivation. This fact and the obtention of a mass bal-
2
steps on the Ni P surface, due to the high reactivity of the metal
1
00
100
80
60
40
20
0
centers in hydrogen transfer reactions. Acetaldehyde was formed
as an intermediate compound, so the reaction network can be de-
scribed as a sequential network.
(
a) HZSM-5
T = 473 K
8
6
4
2
0
0
0
0
0
The results in Fig. 10 suggest that there is a consecutive reac-
tion: ethanol ? acetaldehyde ? ethylene. This seems odd, because
this would entail the oxidation of ethanol before the formation of
ethylene, which is unlikely at reductive conditions as used here.
It was reasoned that a more likely possibility was that there
existed a common species at the surface that could react to form
both acetaldehyde and ethylene in a sequential manner. This led
to the scheme above (Scheme 1).
Ethanol first adsorbs onto a vacant site to form an ethoxide
species. This undergoes dehydrogenation through an alpha-H
abstraction from the carbon proximal to the oxygen to form ad-
sorbed acetaldehyde. The adsorbed acetaldehyde can desorb or
can react further through an enolization reaction involving a
beta-H abstraction from the carbon distal to the oxygen. This
can then undergo hydrogenolysis with adsorbed hydrogen on ac-
tive sites to form ethylene and an adsorbed OH group (not
shown).
Conversion
Ethylene
Butylene
Acetaldehyde
50 100
0
150
200
250
300
Contact time / s
100
100
80
(
b) HZM-5
T = 498 K
8
6
4
2
0
0
0
0
0
Conversion
60
The catalytic reaction network was simulated based on the dis-
appearance rate of ethanol and the formation rate of acetylene and
ethylene. Details on the kinetic analysis are presented in the
Supplementary Information. Ethanol, acetylene, and ethylene were
noted as A, B, and C, and the corresponding adsorbed ethoxy, acet-
Ethylene
40
20
0
ꢄ
ꢄ
ꢄ
aldehyde, and vinyl alcohol are denoted as A , B , and C ,
respectively.
Butylene
250
Acetaldehyde
50 100
}
The kinetic rate law equations are the following:
0
150
200
300
V d½Aꢅ
½Aꢅ½ꢄꢅ þ kꢀA½Aꢄꢅ
ð1Þ
ð2Þ
Contact time / s
¼ ꢀk
s
A
S d
Fig. 9. Variation of ethanol conversion and product yield as a function of contact
time on HZSM-5: (a) reaction temperature of 473 K and (b) reaction temperature of
V d½Bꢅ
¼
S ds
ꢀk
B
½Bꢅ½ꢄꢅ þ kꢀB½Bꢄꢅ
4
98 K.

D. Li et al. / Journal of Catalysis 290 (2012) 1–12
9
(7.09 g cmꢀ3) is the true density, and D
(20 nm) is the average
Experimental Data
Simulation Data
C
crystallite size. Finally,
s
is the contact time calculated from the
1
00
100
80
60
40
20
0
V
v
(
a) Ni P/SiO
volume of the catalyst bed divided by the volumetric flow rate
.
2
2
T = 473 K
Simplification gives VS ¼ 6 f. The concentrations of the intermedi-
qD
C
q
A
8
6
4
2
0
0
0
0
0
ates as well as the vacant active sites were obtained by solving
the following set of equations, assuming that at steady state, the
net rate of the formation of the intermediates is zero.
Conversion
ꢄ
d½A ꢅ
ꢄ
ꢄ
Acetaldehyde
¼
k
A
½Aꢅ½ꢄꢅ ꢀ ðkꢀA þ k
1
Þ½A ꢅ þ kꢀ1½B ꢅ ¼ 0
ð4Þ
ð5Þ
ð6Þ
d
s
ꢄ
d½B ꢅ
Ethylene
ꢄ
ꢄ
ꢄ
¼ k
B
½Bꢅ½ꢄꢅ þ k
1
½A ꢅ ꢀ ðkꢀB þ kꢀ1 þ k
2
Þ½B ꢅ þ kꢀ2½C ꢅ ¼ 0
d
s
ꢄ
d½C ꢅ
ꢄ
ꢄ
¼
k
2
½B ꢅ ꢀ ðkꢀC þ kꢀ2Þ½C ꢅ þ k
C
½Cꢅ½ꢄꢅ ¼ 0
0
50
100
150
200
250
300
d
s
Contact time / s
Use was made of an active site balance, where [L] is the total
concentration of active sites:
Experimental Data
Simulation Data
ꢄ
ꢄ
ꢄ
100
100
½
Lꢅ ¼ ½ꢄꢅ þ ½A ꢅ þ ½B ꢅ þ ½C ꢅ
ð7Þ
(
b) Ni P/SiO2
2
T = 498 K
This kind of scheme is known in the literature as a ‘‘rake’’ mech-
8
6
4
2
0
0
0
0
0
Conversion
80
60
40
20
0
anism because the adsorption and desorption steps when written
out as in Scheme 1 resemble the prongs of the device used to gath-
er leaves. The rake mechanism was first described by Cormerais
et al. [74] and has been suggested to apply to many types of reac-
tions [75,76], but in the early days of its introduction 30 years ago,
the solution of the equations was not readily possible. Even to this
day, many kinetic studies still use simple first-order analysis of
networks that ignore the existence of the adsorption steps. In this
case, the equations were solved numerically (Table 7) to obtain
plots of acetaldehyde and ethylene yields and conversion versus
time.
As can be seen by the curves shown in Fig. 10a and b, the sim-
ulation results are in good agreement with the experimental
points, even though the behavior of the species is different at each
temperature. At 473 K, the acetaldehyde yield grows together with
that of ethylene, while at 498 K, there is a clear maximum in the
acetaldehyde yield. The rate constants are reasonable, with smaller
values at lower temperature as expected. The closeness of the fits
indicates that the proposed mechanism describes the physical sit-
uation accurately.
Ethylene
Acetaldehyde
0
50
100
150
200
250
300
Residence time / s
Fig. 10. Variation of ethanol conversion and product yield as a function of contact
time on Ni P/SiO . The curves are fits from a simulation model discussed in the text:
a) reaction temperature of 473 K and (b) reaction temperature of 498 K.
2
2
(
V d½Cꢅ
¼
s
ꢀk
C
½Cꢅ½ꢄꢅ þ kꢀC½Cꢄꢅ
ð3Þ
S d
In order to further check the reasonableness of the mechanism,
measurements of adsorbed surface intermediates were taken. Data
were taken at different temperatures with ethanol adsorbed using
He as carrier (Fig. 11).
V
d½Aꢅ
ꢀ2 ꢀ1
The rate of consumption of A, ꢀr
A
¼ S
, has units of mol m
s
s ,
d
3
2
in which V (m ) is the bed volume and S (m ) is the catalyst surface
area. This rate is generally expressed as ꢀr = [L]kf(Y) where [L]
mol m ) is the total concentration of sites, k is a rate constant,
A
ꢀ2
(
Table 8 summarizes the assignments in the spectra. At high
ꢀ1
and f(Y) is a dimensionless function of concentrations.
wavenumber, two sharp negative bands at 3746 and 3665 cm
The bed volume is given by V = w/
A
q , where
q
A
(0.3 g cmꢀ3) is
P (m ) is given
P effective surface area,
are due to SiOAH and POAH stretching vibrations and appear neg-
2
the apparent density. The total surface area of Ni
by S = awf, where a (42 m g ) is the Ni
obtained from the equation a ¼ q ; w ð0:900 gÞ is the weight of
catalyst, f (0.079) is the fractional weight loading of the Ni
2
ative because the hydroxyls are being consumed [56]. At 2990,
2
ꢀ1
ꢀ1
2
2
940, and 2900 cm , a cluster of peaks appear that are due to
6
CAH stretching vibrations for methyl and methylene groups. Next
in the center of the spectrum appear bands at 2365 and 2330 due
D
C
2
P,
q
to gas-phase CO
2
interference. For the highest temperature data,
ꢀ1
weak broad bands are present at 1930 and 1830 cm , which are
attributed to combination bands. A small feature at 1750 cm is
due to adsorbed aldehyde CHO. At 1580 cm , a small peak is ob-
served that grows with temperature. This position is characteristic
of a C@C double bond. An oddly shaped feature at 1400 cm fol-
ꢀ1
ꢀ1
ꢀ1
ꢀ1
lowed by a negative peak at 1366 cm is due to background sub-
traction. A feature at 1260 cm corresponds to a CH
ꢀ1
2
wag. Peaks at
ꢀ1
1
070 and 890 cm are assigned to the SiO support. Peaks at lower
2
wavenumber are obscured by interference from the silica.
Overall, in He atmosphere, the diminution of OH bands and the
growth of CAH stretching vibrations are consistent with the forma-
tion of ethoxide species bound to SiAOH and PAOH hydroxyl func-
Scheme 1. ‘‘Rake’’ mechanism for ethanol reaction.

1
0
D. Li et al. / Journal of Catalysis 290 (2012) 1–12
Table 7
Rate constants used in the simulation.
(cm3 molꢀ1
s
ꢀ1
)
kꢀA (s
ꢀ1
)
k
B
(cm3 mol
ꢀ1 ꢀ1
s
)
k
ꢀB (s
ꢀ1
)
k
C
(cm3 mol
ꢀ1 ꢀ1
s
)
kꢀC (s
ꢀ1
)
k
1
(s
ꢀ1
)
k
ꢀ1 (s
ꢀ1
)
k
2
(s
ꢀ1
)
kꢀ2 (s
ꢀ1
)
Temperature (K)
k
A
7
7
7
4
4
73
98
2.80 ꢂ 10
3.50 ꢂ 10
8.50
9.00
3.50 ꢂ 10
4.00 ꢂ 10
6.75
12.0
2.75 ꢂ 10
3.25 ꢂ 10
2.25
30.0
9.00
17.0
4.50
5.00
3.75
5.50
3.00
3.50
7
7
7
2
990
ylene precursors on the Ni
species on Ni
ethoxide pool on the silica. Examination of the signal due to an
2
P. When the flow is switched to He, the
2
P desorb and are replaced with species from the
2
940
900
2
1
400
890
1
070
4
48K
ꢀ
1
aldehyde at 1750 cm shows that the peak increased significantly
in size at a contact time of 60 s but then disappeared at a contact
time of 300 s. The behavior is mirrored in the CAH stretch region
3665
2350
3746
1260
366
1
4
73K
ꢀ1
at 2900–2800 cm . These findings essentially agree with the reac-
tivity results for acetaldehyde in Fig. 10b, which show a maximum
4
98K
23K
close to 60 s and then a decrease. Consideration of the signal at
1
830 1750
1930
1580
ꢀ1
1
580 cm due to an olefin (C@C) shows that it is present at time
5
zero. When the atmosphere is switched to He, its intensity
decreases slightly at 60 s and then increases at 300 s. The increase
is understandable as the acetaldehyde is converted to an ethylene
precursor. These results are also consistent with the results in
Fig. 10b and the proposed rake mechanism in which the adsorbed
acetaldehyde is an intermediate that is converted to an ethylene
precursor on the surface of the catalyst. There was no growth in
the band at 2075 cm due to adsorbed carbon monoxide (CO).
This further indicates that the adsorbed aldehyde did not decom-
4
pose into CH and CO but was converted to ethylene.
4
500 4000 3500 3000 2500 2000 1500 1000
-
1
Wavenumber / cm
2 2
Fig. 11. EtOH FTIR over Ni P/SiO with 0.15% EtOH in He flow.
ꢀ1
tionalities. There is also evidence for an adsorbed acetaldehyde spe-
cies with a band at 1750 cm and the formation of an ethylene or
vinyl alcohol species contributing to the C@C signal at 1580 cm
In order to obtain more information on the behavior of the ad-
sorbed species, a transient experiment was carried out. The objec-
tive was to approximate the contact time measurements shown in
ꢀ1
ꢀ
1
.
The picture that emerges from the FTIR measurements is con-
sistent with the reaction scheme presented earlier. Ethanol adsorbs
on hydroxyl species on the silica to form ethoxide species. Substan-
tial work [77,78] indicates that alkoxide species on silica are highly
Fig. 10b. A Ni
2
P/SiO
2
sample was exposed to the ethanol reactant
2
mobile so these can readily move to the Ni P surface, which is
stream (1.5%), and then the flow was switched to He. The spectra
of ethanol FTIR at 497 K as a function of time are shown in Fig. 12.
The assignments in Fig. 12 are the same as in Fig. 11. In Fig. 12,
excellent in hydrogen transfer reactions [34–44]. A first reaction
that occurs is dehydrogenation to form an adsorbed acetaldehyde
ꢀ
1
(growth in band of 1750 cm ). Subsequently, in He, the acetalde-
hyde undergoes an enolization to vinyl alcohol (growth in band at
ꢀ1
vertical lines have been placed at 1750 cm (position for an alde-
ꢀ1
ꢀ1
hyde functionality, HC@O) and at 1580 cm (position for an olefin
double bond). The experiment starts at time zero with the steady-
steady ethanol reaction, so the expected main species is adsorbed
ethoxide on the silica surface, together with acetaldehyde and eth-
1580 cm ). This set of steps accounts for the main species formed
and the main catalytic pathway in the reaction. There is also a side
reaction, which is not major in the steady state, but which pro-
ceeds in temperature-programmed reaction, namely the formation
Table 8
Summary of FTIR assignments.
Band (cmꢀ1)
Assignment
Species
Reference
3
3
2
746 (neg)
665 (neg)
990, 2940, 2900
SiOAH stretch
POAH stretch
CAH stretch
SiAOH
PAOH
[56]
[56]
[79,80]
CH
CH
CO
3
CH
3
, CH
2
@CHAOH
3
CHO, CH
3
2 2 2
CH OH, CH @CH
2
1
1
1
1
1
075
930, 1830
750
580
400
C@O
Combination
HC@O
C@C
OH bend
[41,56]
HCHO
[77,78]
[80]
[77]
CH
2
@CHAOH
366 (neg)
Background subtraction
ꢆ1310, 1260
CH
2
wag
CH
CH
CH
SiO
SiO
CH
CH
CH
CH
CH
CH
CH
CH
3
3
2
CH
CHO, CH
@CHAOH
2
OH
[79]
[80]
ꢆ1160
CAC stretch
Combination
SiAO
SiAOASi
C@C wag
CAC stretch
3
CH
2
OH
ꢆ1070
ꢆ890
ꢆ990
2
2
2 2
@CH
3
CH
CH
@CH
CHO
@CHAOH
CHO
3
, CH
, CH
3
CH
2
OH
[79,80]
[77,79]
790–830
CH
CH
3
rock
rock
3
2
3
2
3
2
3
3
CHO
2
2
ꢆ760
ꢆ600
ꢆ500
CAH bend
CH @C OPLA
[79]
[78]
[79]
[78]
2
CACAO deform
Torsion
300–400
@CHAOH

D. Li et al. / Journal of Catalysis 290 (2012) 1–12
11
adsorbed acetaldehyde on the surface of the catalyst at reaction
conditions. Measurements at increasing contact times show
that it increases and then disappears, in agreement with the
corresponding reactivity results.
1
750 1580
0
s
60 s
Acknowledgments
300 s
This work was supported by the US Department of Energy, Office
of Basic Energy Sciences, through Grant DE-FG02-963414669, the
National Renewable Energy Laboratory through Grant DE-FG3608
GO18214, and the New Energy Development Organization.
4
500 4000 3500 3000 2500 2000 1500 1000
-
1
Wavenumber / cm
Appendix A. Supplementary material
2 2
Fig. 12. EtOH FTIR over Ni P/SiO in He at 497 K.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jcat.2012.02.001.
of adsorbed CO. This is likely the result of the decomposition of ad-
sorbed acetaldehyde in the presence of H (not He) to a methyl
References
2
species and CO. The hydrogen is likely adsorbed atomic hydrogen,
and the driving force is the formation of methane and strongly
bonded CO. Although not a primary path in the steady-state reac-
[
[
1] T.V. Choudhary, C.B. Phillips, Appl. Catal., A 397 (2011) 1.
2] J. Zakzeski, P.C.A. Bruijnincx, A.L. Jongerius, B.M. Weckhuysen, Chem. Rev. 110
(
2010) 3552.
[3] A. Wawrzetz, B. Peng, A. Hrabar, A. Jentys, A.A. Lemonidou, J.A. Lercher, J. Catal.
69 (2010) 411–420.
[
[
tion, this pathway accounts for the observation of CH
4
and CO in
2
the ethanol TPD spectra (Fig. 7).
4] O. Casanova, S. Iborra, A. Corma, J. Catal. 275 (2010) 236–242.
5] J.J. Bozell, G.R. Petersen, Green Chem. 12 (2010) 539.
The reaction of ethanol on HZSM-5 and Ni P/SiO provides
2 2
information on the reaction pathways on a typical acid catalyst
and a metal catalyst. The acid catalyst, HZSM-5, carries out simple
dehydration to form ethylene. The metal catalyst also forms mainly
ethylene, but through a non-direct pathway in which an adsorbed
ethoxide species is first dehydrogenated to a surface acetaldehyde
species, which undergoes enolization to a vinyl alkoxide and sub-
sequent hydrodeoxygenation. This route is non-direct and occurs
because of the strongly metallic hydrogenation/dehydrogenation
[6] J.C. Serrano-Ruiz, D. Wang, J.A. Dumesic, Green Chem. 12 (2010) 574.
[
[
7] A. Birot, F. Epron, C. Descorme, D. Duprez, Appl. Catal., B 79 (2008) 17–25.
8] J. Rass-Hansen, R. Johansson, M. Moller, C.H. Christensen, Int. J. Hydrogen
Energy 33 (2008) 4547–4554.
[9] E.B. Pereira, P. Ramírez de la Piscina, N. Homs, Bioresource Technol. 102 (2011)
419–3423.
3
[
[
10] A.H. Tullo, Chem. Eng. News 89 (May 16) (2011) 24–25.
11] R.L. Guenard, L.C.F. Torres, B. Kim, S.S. Perry, P. Frantz, S.V. Didziulis, Surf. Sci.
515 (2002) 103–116.
[12] P.A. Clayborne, T.C. Nelson, T.C. De Vore, Appl. Catal., A 257 (2004) 225–233.
13] B.M. Nagaraja, A.H. Padmasri, P. Seetharamulu, K.H.P. Reddy, B.D. Raju, K.S.R.
Rao, J. Mol. Catal. A 278 (2007) 29–37.
14] E.A. El-Katatny, S.A. Halawy, M.A. Mohamed, M.I. Zaki, Appl. Catal., A 1999
(2000) 83–92.
15] N.R.C.F. Machado, V. Calsavara, N.G.C. Astrath, C.K. Matsuda, A.P. Junior, M.L.
Baesso, Fuel 84 (2005) 2064–2070.
[
[
[
2
properties of the Ni P. It is important because the pathway may
also be involved in the reactions of more complex molecules, and
for this reason, its participation should be considered in the study
of their conversion.
[
[
16] V. Calsavara, M.L. Baesso, N.R.C.F. Machado, Fuel 87 (2008) 1628–1636.
17] N. Takezawa, C. Hanamaki, H. Kobayashi, J. Catal. 38 (1975) 101–109.
4
. Conclusions
[18] Y. Matsumura, K. Hashimoto, S. Yoshida, J. Catal. 117 (1989) 135–143.
[
[
19] I. Takahara, M. Saito, M. Inaba, K. Murata, Catal. Lett. 105 (2005) 249–252.
20] X. Zhang, R. Wang, X. Yang, F. Zhang, Micropor. Mesopor. Mater. 116 (2008)
The experiments described in this work lead us to the following
210–215.
conclusions:
[21] J. Schulz, F. Bandermann, Chem. Eng. Technol. 17 (1994) 179–186.
[
[
[
22] C.B. Phillips, R. Datta, Ind. Eng. Chem. Res. 36 (1997) 4466–4475.
23] J.M. Vohs, M.A. Barteau, Surf. Sci. 211 (1989) 590–608.
24] F.W. Chang, W.Y. Kuo, K.C. Lee, Appl. Catal., A 246 (2003) 253–264.
1
. Comparison between the properties of commercial HZSM-5(Si/
Al = 15) and synthesized Ni P/SiO catalysts showed that
HZSM-5 possesses higher surface area and stronger acidic sites
than the Ni P/SiO catalyst. The Ni P/SiO catalyst essentially
[25] M.A. Aramendía, V. Borau, C. Jiménez, J.M. Marinas, A. Porras, F.J. Urbano, J.
Catal. 161 (1996) 829–838.
2
2
[
[
26] Y. Shinohara, T. Nakajima, S. Suzuki, J. Mol. Struct. 460 (1999) 231–244.
27] M.M. Doheim, H.G. El-Shobaky, Colloids Surf. A 204 (2002) 169–174.
2
2
2
2
has more mesopores and moderate acid and basic sites. Infrared
spectroscopy of pyridine reveals that acidity of HZSM-5 and
[28] F.S. Ramos, A.M. Duarte de farias, L.E.P. Borges, J.L. Monteiro, M.A. Fraga, E.F.
Sousa-Aguiar, L.G. Appel, Catal. Today 101 (2005) 39–44.
[
29] M.A. Aramendía, V. Boráu, C. Jiménez, J.M. Marinas, A. Porras, F.J. Urbano,
React. Kinet. Catal. Lett. 65 (1998) 25–31.
Ni
sites, respectively. In the ethanol temperature-programmed
desorption process (EtOH-TPD), the Ni P/SiO catalyst exhibited
2 2
P/SiO catalysts consists mainly of Brønsted and Lewis acid
[30] M.A. Aramendíra, V. Boráu, I.M. García, C. Jiménez, A. Marinas, J.M. Marinas, A.
Porras, F.J. Urbano, Appl. Catal., A 184 (1999) 115–125.
2
2
[
[
31] H.Y. Zhao, D. Li, P. Bui, S.T. Oyama, Appl. Catal., A 391 (2011) 305–310.
32] Y. Shu, Y.-K. Lee, S.T. Oyama, J. Catal. 236 (2005) 112–121.
a higher selectivity toward dehydrogenation products (acetal-
dehyde) than HZSM-5.
. Steady-state reaction results confirm that the Ni P/SiO pro-
2 2
[33] T. Kawai, K.K. Bamdo, Y.-K. Lee, S.T. Oyama, W.-J. Chun, K. Asakura, J. Catal. 241
(2006) 20–24.
2
[
[
34] X. Wang, P. Clark, S.T. Oyama, J. Catal. 208 (2002) 321–331.
35] S.T. Oyama, X. Wang, Y.-K. Lee, K. Bando, F.G. Requejo, J. Catal. 210 (2002) 207–
duces both acetaldehyde and ethylene, and contact time studies
indicate that acetaldehyde is a primary product and ethylene is
a secondary product. The reaction, however, does not consist of
an oxidation reaction followed by a reduction reaction, but
involves sequential reactions on the surface, namely a rake
mechanism. Analysis of the reaction sequence and simulation
of the results give support for this interpretation. Measure-
ments by in situ Fourier transform infrared spectroscopy also
give evidence for the presence of the suggested intermediate,
217.
[36] S.T. Oyama, T. Gott, H. Zhao, Y.-K. Lee, Catal. Today 143 (2009) 94–107.
[
[
37] S.T. Oyama, J. Catal. 216 (2003) 343–352.
38] S.T. Oyama, X. Wang, F. Requejo, T. Sato, Y. Yoshimura, J. Catal. 209 (2002)
1–5.
[39] S.T. Oyama, X. Wang, Y.-K. Lee, W.-J. Chun, J. Catal. 221 (2004) 263–273.
[
[
40] S.T. Oyama, X. Wang, Y. Lee, K. Bando, F.G. Requejo, J. Catal. 210 (2004) 207–
17.
41] S.J. Sawhill, K.A. Layman, D.R. Van Wyk, M.H. Engelhard, C. Wang, M.E. Bussell,
J. Catal. 231 (2005) 300–313.
2

1
2
D. Li et al. / Journal of Catalysis 290 (2012) 1–12
[64] C.A. Emeis, J. Catal. 141 (1993) 347–354.
[
[
[
[
[
42] P.A. Clark, X. Wang, P. Deck, J. Catal. 210 (2002) 116–126.
43] P.A. Clark, S.T. Oyama, J. Catal. 218 (2003) 78–87.
44] S.T. Oyama, Y.-K. Lee, J. Catal. 258 (2008) 393–400.
45] M. Guisnet, P. Magenoux, Stud. Surf. Sci. Catal. 88 (1994) 53–68.
46] A.G. Gayubo, A. Alonso, B. Valle, A.T. Aguayo, J. Bilbao, Appl. Catal., B 97 (2010)
[65] G.A.M. Hussein, N. Sheppard, J. Chem. Soc. Faraday Trans. 87 (1991) 2661–
2668.
[66] M. Dömöka, M. Tótha, J. Raskób, A. Erd o} helyi, Appl. Catal., B 69 (2007) 262–
272.
2
99–306.
[67] J. Franckaerts, G.F. Froment, Chem. Eng. Sci. 19 (1964) 807–818.
[68] T. Yamazaki, N. Kikuchi, M. Katoh, T. Hirose, H. Saito, T. Yoshikawa, M. Wada,
Appl. Catal., B 99 (2010) 81–88.
[69] A. Gazsi, A. Koos, T. Bansagi, F. Solymosi, Catal. Today 160 (2011) 70–78.
[70] A.P. Farkas, F. Solymosi, Surf. Sci. 601 (2007) 193–200.
[71] R.J. Gorte, J. Catal. 75 (1982) 164–174.
[
[
[
[
[
[
[
[
[
[
[
[
47] P. Berteau, B. Delmon, Appl. Catal. 70 (1991) 307–323.
48] W. Wang, S.P. Wang, X.B. Ma, J.L. Gong, Catal. Today 148 (2009) 323–328.
49] S.R. Jagtap, Y.P. Patil, Appl. Catal., A 341 (2008) 133–138.
50] A.S. Ndou, N. Plint, N.J. Coville, Appl. Catal., A 251 (2003) 337–345.
51] P. Berteau, B. Delmon, Catal. Today 5 (1989) 121–137.
52] J.D. Bi, X. W Guo, M. Liu, X.S. Wang, Catal. Today 149 (2010) 143–147.
53] J.G. Post, H.G.V. Hoof, Zeolites 4 (1984) 9–14.
54] D.J. Parrillo, C. Lee, R.J. Gorte, Appl. Catal., A 110 (1994) 67–74.
55] H.G. Karge, V. Dondur, J. Weitkamp, J. Phys. Chem. 95 (1991) 283–288.
56] Y.-K. Lee, S.T. Oyama, J. Catal. 239 (2006) 376–389.
57] T. Gott, S.T. Oyama, J. Catal. 263 (2009) 359–371.
58] S.T. Oyama, T. Gott, K. Asakura, S. Takakusagi, K. Miyazaki, Y. Koike, K.K. Bando,
J. Catal. 268 (2009) 209–222.
[72] E.E. Ibok, D.F. Ollis, J. Catal. 66 (1980) 391–400.
[73] B.M. Abu-Zied, A.M. El-Awad, J. Mol. Catal. A 176 (2001) 227–246.
[74] F.X. Cormerais, G. Perot, F. Chevalier, M. Guisnet, J. Chem. Res. S (1980) 362–
363.
[75] S. Lars, T. Andersson, J. Catal. 98 (1986) 138–149.
[76] M. Stöcker, Mesopor. Micropor. Mater. 29 (1999) 3–48.
[77] M. Seman, J.N. Kondo, K. Domen, R. Radhakrishnan, S.T. Oyama, J. Phys. Chem.
B 106 (2002) 12965–12977.
[
[
[
59] K.M.A. El-Salaam, E.A. Hassan, Surf. Technol. 16 (1982) 121–128.
60] F.F. Roca, L.D. Mourgues, Y. Trambouze, J. Catal. 14 (1969) 107–113.
61] S. Golay, L. Kiwi-Minsker, R. Doepper, A. Renken, Chem. Eng. Sci. 54 (1999)
[78] M. Seman, J.N. Kondo, K. Domen, R. Radhakrishnan, S.T. Oyama, Chem. Lett. 11
(2002) 1082–1083.
[79] T. Shimanouchi, Molecular vibrational frequencies, in: P.J. Linstrom, W.G.
Mallard (Eds.), NIST Chemistry WebBook, NIST Standard Reference Database
Number 69, 37 National Institute of Standards and Technology, Gaithersburg,
MD 20899. <http://webbook.nist.gov> (retrieved 23.06.11).
[80] M. Hawkins, L. Andrews, J. Am. Chem. Soc. 105 (1983) 2523–2530.
3
593–3598.
62] R.L. Mao, T.S. Le, M. Fairbairn, A. Muntasar, S. Xiao, G. Denes, Appl. Catal., A 185
1999) 41–52.
63] M. Lenarda, M.D. Ros, M. Casagrande, L. Storaro, R. Ganzerla, Inorg. Chim. Acta
49 (2003) 195–202.
[
[
(
3