Aqueous phase hydrodeoxygenation of polyols over Pd/WO3-ZrO2: Role of Pd-WO3 interaction and hydrodeoxygenation pathway

Article abstract of DOI:10.1016/j.cattod.2015.10.034

Aqueous phase processing of biomass derived sugar alcohols is one of the promising routes to convert biomass into fuels and chemicals. Bifunctional catalysts are critical in the aqueous phase hydrodeoxygenation of sugar alcohol. Understanding the interaction between metal and acidic metal oxides as well as the hydrodeoxygenation pathways will help develop more efficient bifunctional catalysts. Here, tungstated zirconia supported palladium catalysts were prepared and further characterized using nitrogen sorption, X-ray diffraction, FT-IR analysis of adsorbed pyridine, CO chemisorption and diffuse reflectance UV-vis. Strong interaction between palladium and WO₃ in addition to a synergetic effect of the acidic and metallic sites were found to promote the aqueous phase hydrodeoxygenation of ethylene glycol. H-D exchange experiments using ¹³C{¹H} NMR spectroscopy confirmed that the aqueous phase hydrodeoxygenation follows a dehydration-hydrogenation pathway. The hydrogenation of the dehydration products shifts the dehydration-hydration equilibrium toward the dehydration pathway and leads to highly selective C-O cleavage.

Full text of DOI:10.1016/j.cattod.2015.10.034

G Model
CATTOD-9885; No. of Pages7
ARTICLE IN PRESS
Catalysis Today xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
Aqueous phase hydrodeoxygenation of polyols over Pd/WO3-ZrO2:
Role of Pd-WO3 interaction and hydrodeoxygenation pathway
Changjun Liua, Junming Sunb, Heather M. Brownc, Oscar G. Marin-Floresb,
J. Timothy Baysc, Ayman M. Karimd,∗∗, Yong Wangb,c,∗
a Multi-phase Mass Transfer and Reaction Engineering Laboratory, College of Chemical Engineering, Sichuan University, Chengdu 610065, China
b The Gene & Linda Voiland School of Chemical Engineering and Bioengineering, Washington State University, Pullman, WA 99164, United States
c Institute for Interfacial Catalysis, Pacific Northwest National Laboratory, Richland, WA 99352, United States
d Department of Chemical Engineering, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Aqueous phase processing of biomass derived sugar alcohols is one of the promising routes to convert
biomass into fuels and chemicals. Bifunctional catalysts are critical in the aqueous phase hydrodeoxy-
genation of sugar alcohol. Understanding the interaction between metal and acidic metal oxides as
well as the hydrodeoxygenation pathways will help develop more efficient bifunctional catalysts. Here,
tungstated zirconia supported palladium catalysts were prepared and further characterized using nitro-
gen sorption, X-ray diffraction, FT-IR analysis of adsorbed pyridine, CO chemisorption and diffuse
reflectance UV–vis. Strong interaction between palladium and WO3 in addition to a synergetic effect
of the acidic and metallic sites were found to promote the aqueous phase hydrodeoxygenation of eth-
Received 1 August 2015
Received in revised form 19 October 2015
Accepted 22 October 2015
Available online xxx
Keywords:
Aqueous phase hydrodeoxygenation
Biomass
Zirconia
Bifunctional catalyst
1
ylene glycol. H-D exchange experiments using 13C{ H} NMR spectroscopy confirmed that the aqueous
phase hydrodeoxygenation follows a dehydration–hydrogenation pathway. The hydrogenation of the
dehydration products shifts the dehydration–hydration equilibrium toward the dehydration pathway
and leads to highly selective C O cleavage.
© 2015 Elsevier B.V. All rights reserved.
1. Introduction
separating them from the insoluble lignin [5–7]. Biomass deriva-
tives have high oxygen content, low volatility, and high solubility
The depletion of fossil resources and increasing global warming
issues have urged both the scientific and industrial communities
to explore renewable resources such as biomass [1]. Biomass is
considered the only renewable carbon resource for our indus-
trial society [2]. Lignocellulosic biomass, which is composed of
65–82 wt.% cellulose and hemicellulose [3], has been extensively
studied in an effort to produce liquid fuels and valuable chemicals.
Catalytic fast pyrolysis can directly convert lignocellulosic biomass
into hydrocarbons but with low carbon yields [4]. On the other
hand, well-developed hydrolysis processes are capable of decon-
structing cellulose and hemicelluloses in lignocellulosic biomass
into their corresponding sugars (e.g., glucose, fructose, xylose) and
in water. Their low thermal stability and highly reactive nature are
more suitable to catalytic aqueous phase processing at moderate
temperatures [2]. With aqueous phase processing, cellulose and
hemicellulose derived intermediates can be more selectively con-
verted into desired fuel products via a tandem reaction pathway
[8].
Selective removal of oxygen from these biomass-derived inter-
mediates without excessively breaking C C linkages is one of the
key challenges in fuel production [9]. Hydrodeoxygenation (HDO)
is the common step in all processes that currently exist for the
thermochemical conversion of biomass to liquid fuels/chemicals
(e.g. liquefaction, fast pyrolysis, catalytic fast pyrolysis, and aque-
ous phase conversion of carbohydrates). Controlled aqueous phase
HDO of the hydrolysis products (monoses/polyols) from cellulose
and hemicellulose is the primary step to achieve high selectivity in
these processes [8]. Bifunctional catalysts with appropriate metallic
and acidic sites were found to be capable of catalyzing the aque-
ous phase HDO of polyols [10,11]. The selectivity to hydrocarbon
products can be improved by catalysts with more hydrogenating
metallic phases on supports with stronger acidity [12]. A wide range
∗
Corresponding author at: The Gene & Linda Voiland School of Chemical Engineer-
ing and Bioengineering, Washington State University, Pullman, WA 99164, United
States.
∗∗
Corresponding author at: Department of Chemical Engineering, Virginia Poly-
technic and State University, Blacksburg, VA 24061, United States.
E-mail addresses: amkarim@vt.edu (A.M. Karim), wang42@wsu.edu (Y. Wang).
http://dx.doi.org/10.1016/j.cattod.2015.10.034
0920-5861/© 2015 Elsevier B.V. All rights reserved.
Please cite this article in press as: C. Liu, et al., Aqueous phase hydrodeoxygenation of polyols over Pd/WO3-ZrO2: Role of Pd-WO3
interaction and hydrodeoxygenation pathway, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.10.034

G Model
CATTOD-9885; No. of Pages7
ARTICLE IN PRESS
2
C. Liu et al. / Catalysis Today xxx (2015) xxx–xxx
of Pt promoted solid acid catalysts has been tested [13]. However,
further understanding of the HDO reaction mechanism and opti-
mizing of the catalysts are necessary for improving the selectivity
to mono-oxygenates.
WO3 is used as a solid acid in many acid catalyzed pro-
cesses [14–17] including dehydration reactions by tungstated
zirconia [18,19]. Palladium is a cheaper hydrogenation noble
metal than platinum. The combination of Pd and tungstated zir-
conia (Pd/WO3-ZrO2) has been studied by Dedsuksophon et al.
for hydrolysis/dehydration/aldol condensation/hydrogenation of
lignocellulose and carbohydrates [20,21]. In this work, Pd/ZrO2 cat-
alysts promoted by WO3 were studied in the aqueous phase HDO of
polyols with the focus on the selectivity to C O cleavage and C
C
cleavage. Ethylene glycol was used as the surrogate for polyols to
simplify the reaction routes and products. The interaction between
palladium and WO3 as well as the aqueous phase HDO mechanism
of polyols were elucidated.
Fig. 1. XRD spectra of (a) WO3-ZrO2 and (b) 2Pd/WO3-ZrO2.
Pd atom. The temperature-programmed desorption of ethanol and
isopropylamine were carried out on a home-made TPD system
with a bubbler to introduce the vapors. The samples were first
reduced at 300 ◦C for 1 h by H2, the adsorption step was performed
at room temperature for 1 h then purged with helium for another
hour. Thereafter, the desorption profile was recorded by a Pfeif-
fer OmniStar Quadrupole Mass Spectrometer under helium with a
2. Experimental
2.1. Preparation of catalysts
Zirconia (ZrO2) and tungstated zirconia (WO3-ZrO2) purchased
from Saint-Gobain NorPro were used as the starting material. The
nominal tungsten loading of tungstated zirconia is 12.5 wt.%. Both
zirconia and tungstated zirconia supported palladium catalysts
were prepared by incipient wetness impregnation. Palladium(II)
nitrate hydrate Pd(NO3)2·xH2O purchased from Sigma–Aldrich was
used as the precursor of palladium. A given amount of palla-
dium nitrate hydrate was dissolved into the 2 M NH4NO3 solution
to achieve better Pd dispersion (determined from characteriza-
tion of catalysts prepared without NH4NO3, not shown). After
impregnation, the resultant catalysts were dried at 120 ◦C for 4 h
then calcined in static air at 400 ◦C for 6 h with a ramping rate
of 2 ◦C min−1. The nominal palladium loading is 2.0 wt.% and the
catalyst was labeled as 2Pd/WO3-ZrO2. Pd/WO3-ZrO2 catalysts of
smaller Pd particle sizes were prepared to investigate the effect of
Pd particle size on the activity and selectivity by following the same
procedure with Pd loadings of 0.5 wt.% and 1.0 wt.%. Pd/WO3-ZrO2
catalysts with Pd loadings of 0.1 wt.% and 0.25 wt.% were prepared
and tested in UV–vis-DRS to minimize the effect of the absorp-
tion due to the black color of the reduced catalysts on the pre-edge
energy analysis of WO3.
ramping rate of 10 K min−1
.
2.3. Catalytic reaction
The catalytic activity tests were carried out in a fixed bed reac-
tor with both ethylene glycol solution and hydrogen fed from the
top of the reactor. A given amount of catalyst with the particle size
about 100 ␮m was loaded in the middle of the reactor. Dense zir-
conia ceramic of similar particle size was packed below and above
the catalyst section so that the feed could be evenly distributed
across the reactor section and preheated to the reaction tempera-
ture before it reached the catalyst zone. The catalyst was reduced
in situ at 300 ◦C in 40 sccm H2 for 1 h and then cooled down to reac-
tion temperature before starting the reaction. Ethylene glycol was
diluted to 10 wt.% in water and used as liquid feedstock. The reac-
tion temperature was set to 240 ◦C and the system was pressurized
to 7.34 MPa (1050 psig) by high purity H2. The product stream from
the reactor was chilled to 4 ◦C. The liquid products were collected
and analyzed using a Waters high performance liquid chromato-
graph (HPLC) with a refractive index detector while the gases were
analyzed online by an Agilent micro 3000 chromatograph with a
thermal conductivity detector (TCD). The discharged gas flow rate
was recorded with a mass flow meter. The H-D exchange exper-
iments were performed under the same reaction conditions. The
system was completely purged with D2O and H2 before exper-
iments using 10 wt.% ODCH2CH2OD in D2O as the liquid feed,
and was purged thoroughly with H2O and H2 when using 10 wt.%
CD3OH in H2O feed. The resultant liquid reaction mixtures were
2.2. Catalyst characterization
Crystalline structures of the catalysts were characterized by
X-ray diffraction (XRD) on a Philips X’pert diffractometer with a
copper anode (K˛1 = 0.15405 nm) operating at 40 kV and 50 mA.
Nitrogen adsorption–desorption isotherms were performed at
−196 ◦C with an automatic adsorption meter (Quantachrome
instruments autosorb-6). The samples were pretreated at 150 ◦C for
6 h under vacuum. The surface areas were calculated from adsorp-
tion values at five relative pressures (P/P0) ranging from 0.10 to
0.31 using the Brunauer–Emmett–Teller (BET) method. The pore
volumes were determined from the total amount of N2 adsorbed
between P/P0 = 0.05 and P/P0 = 0.98. Infrared spectra of adsorbed
pyridine (IR-Py) were collected on a Bruker Tenser 27 FTIR spec-
trometer. Reflectance spectra of the catalysts were recorded with
a Varian Cary 5000 UV-Vis-NIR spectrophotometer in the wave-
length range from 200 to 800 nm at a resolution of 0.333 nm.
The pulse CO chemisorption experiments were performed on a
Micromeritics Autochem II 2920 Chemisorption Analyzer with a
TCD detector at 40 ◦C using 10% CO in helium. The Pd dispersion
was calculated by assuming the linear adsorption of CO on each
1
analyzed by 13C{ H} NMR on a 500 MHz Varian Inova spectrome-
ter at 125 MHz in CDCl3 with 0.05 M chromium (III) acetylacetonate
for quantitation, referenced to CDCl3.
3. Results
3.1. Characterization
Table 1 shows that tungstated zirconia has slightly higher BET
surface area than zirconia. The impregnation of palladium onto
zirconia does not significantly change the BET surface area pos-
sibly due to good dispersion of Pd on the surface as indicated by
the CO chemisorption. The impregnation of palladium onto the
Please cite this article in press as: C. Liu, et al., Aqueous phase hydrodeoxygenation of polyols over Pd/WO3-ZrO2: Role of Pd-WO3
interaction and hydrodeoxygenation pathway, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.10.034

G Model
CATTOD-9885; No. of Pages7
ARTICLE IN PRESS
C. Liu et al. / Catalysis Today xxx (2015) xxx–xxx
3
Table 1
Physicochemical properties of the supports and catalysts.
Sample
Pd loading, wt.%
W loading, wt.%
Surface area, m2 g−1
Pore volume, cm3 g−1
Pulse chemisorption, CO/Pd
Pd particle size,a nm
ZrO2
2Pd/ZrO2
WO3-ZrO2
2Pd/WO3-ZrO2
1.0Pd/WO3-ZrO2
0.5Pd/WO3-ZrO2
–
2.0
–
2.0
1.0
0.50
–
–
12.5
12.5
12.5
12.5
99
101
2
2
–
–
0.30
0.31
0.23
0.22
–
–
0.86
–
0.17
0.45
0.70
–
1.30
–
6.73
2.46
1.60
116
110
–
a
3
˚
Palladium particle sizes were estimated by D = 6(vm/am)/dVA [22], where D is the metal dispersion, dVA is the mean particle size in angstrom, vm = 14.70 A is the volume
2
˚
occupied by a Pd atom in bulk metal, am = 7.93 A is the area occupied by a surface Pd atom.
5
4
3
2
1
0
Pd size
0.00 nm
Edge Energy
3.05 eV
1.48 nm 3.00 eV
2.03 nm 2.73 eV
2.59 eV
c
b
a
3.23 eV
4
1
2
3
5
6
Photon energy (eV)
Fig. 3. Diffuse reflectance UV–vis absorption spectra of WO3-ZrO2 (a) without pal-
ladium, (b) Pd particle size of 1.48 nm, (c) Pd particle size of 2.03 nm.
Fig. 2. IR-Py spectra of (a) ZrO2, (b) WO3-ZrO2, (c) 2Pd/WO3-ZrO2. Reduced at 300 ◦
C
for 1 h, spectra collected at 150 ◦C.
absorption edge energy (Fig. 3). However the detailed mechanism
of WO3 domain size growth is not clear yet.
tungstated zirconia only leads to the slight decrease of BET sur-
face area from 116 m2 g−1 to 110 m2 g−1. The XRD spectra (Fig. 1)
shows that tetragonal tungstated zirconia is a major phase and a
peak at 40.3◦ corresponding to palladium particles of an average
diameter about 6.73 nm was observed.
3.2. Catalytic activity
2Pd/ZrO2 and 2Pd/WO3-ZrO2, having the same palladium load-
ing (Table 1), were tested in aqueous phase HDO of ethylene glycol
and Fig. 4 compares the product at a similar conversion of ∼14%.
Carbon oxides, methanol, methane, ethanol, acetaldehyde, acetic
acid and ethane were the main products of aqueous phase HDO
of ethylene glycol. Carbon oxides, methanol and methane are a
result of C C bond cleavage. Although the CO-chemisorption indi-
cates that 2Pd/WO3-ZrO2 has a larger Pd particle size than 2Pd/ZrO2
Pyridine FT-IR was employed to characterize the acidic nature
of the supports and catalyst (Fig. 2). Pyridine adsorption was per-
formed at 150 ◦C and the spectra were collected at the same
temperature after purging with helium for 30 min. The band at
1543 cm−1 is known to be due to N+ H stretching of pyridinium
ions and is used to characterize the protonic acid sites along with
auxiliary peaks at 1489 cm−1, 1620 cm−1 and 1640 cm−1 [23–25].
WO3-ZrO2 and 2Pd/WO3-ZrO2 showed both Brönsted acidity and
Lewis acidity while zirconia itself showed only Lewis acidity. The
presence of WO3 in ZrO2 introduces Brönsted acidic sites to the
catalysts.
Pd/WO3-ZrO2 catalysts of Pd loadings higher than 0.5 wt.%
were black after reduction and showed strong absorption of visible
light. It is impossible to characterize the WO3 domain size using
their pre-edge adsorption energy. To understand the effect of Pd
on WO3 aggregation, Pd/WO3-ZrO2 catalysts were prepared by
lowering the Pd loading to 0.1 wt.% and 0.25 wt.% with Pd particle
sizes estimated by CO-chemisorption of 1.48 nm and 2.03 nm
respectively. The WO3 domain sizes were determined by the
absorption edge energy with the diffuse reflectance UV–vis spectra
(Fig. 3). It is known that absorption edge energy decreases with
increasing WO3 domain size [16]. Fig. 3 shows the WO3 domain
sizes of the catalysts are somewhere between the bulk monoclinic
WO3 (2.59 eV) of six neighboring octahedra and the W12 cluster
in ammonium metatungstate (3.23 eV) [16]. The impregnation of
palladium onto WO3-ZrO2 resulted in the increase of WO3 domain
size as suggested by increased absorption edge energy (Fig. 3).
It is likely that the larger palladium particle size of 0.25Pd/WO3-
ZrO2 leads to larger WO3 domain size as indicated by the lower
Fig. 4. Products distribution with/without WO3. Feed: 10 wt.% ethylene gly-
col, T = 240 ◦C, P = 1050 psig, H2/C = 2, WHSV = 1.10 g g−1 h−1 for 2Pd/ZrO2 and
0.23 g g−1 h−1 for 2Pd/WO3-ZrO2.
Please cite this article in press as: C. Liu, et al., Aqueous phase hydrodeoxygenation of polyols over Pd/WO3-ZrO2: Role of Pd-WO3
interaction and hydrodeoxygenation pathway, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.10.034

G Model
CATTOD-9885; No. of Pages7
ARTICLE IN PRESS
4
C. Liu et al. / Catalysis Today xxx (2015) xxx–xxx
70
100
80
60
40
20
5
4
3
2
1
0
2% Pd/ZrO2
2% Pd/WO3-ZrO2
1.60 nm
2.46 nm
6.73 nm
60
50
40
30
20
10
0
l
l
l
o
n
e
d
e
e
e
d
y
h
s
r
e
n
o
o
n
n
i
x
io
n
a
n
a
a
s
a
h
t
e
a
p
o
r
h
t
o
h
t
h
t
e
r
h
t
io
e
d
l
0
e
e
v
e
d
m
n
a
p
n
m
t
o
o
e
c
b
c
a
r
a
C-O/C-C
C1 Products
Ethanol
c
Fig. 5. Products distribution over Pd/WO3-ZrO2 catalysts of different Pd particle size.
Catalysts: Pd/WO3-ZrO2 with 0.5 wt.%, 1 wt.% and 2 wt.% Pd loadings corresponding
to 1.60, 2.46 and 6.73 nm Pd particle sizes, respectively. Carbon balance: 100 2%,
Fig. 6. Comparison of the selectivity to C O and C C bond cleavage. Feed:
10 wt.% ethylene glycol, T = 240 ◦C, P = 1050 psig, H2/C = 2, WHSV = 1.10 g g−1 h−1 for
2Pd/ZrO2 and 0.23 g g−1 h−1 for 2Pd/WO3-ZrO2.
feed: 10 wt.% ethylene glycol, T = 240 ◦C, P = 1050 psig, WHSV = 0.25–0.7 g g−1 h−1
.
(Table 1), 2Pd/ZrO2 shows much higher selectivity to CO and CO2
than 2Pd/WO3-ZrO2. The presence of WO3 largely improves the
selectivity to ethanol. This suggests that the reaction pathways on
the two catalysts are different. The effect of palladium particle size
on activity and selectivity of Pd/WO3-ZrO2 catalysts was investi-
gated by preparing catalysts with 0.5 wt.%, 1 wt.% and 2 wt.% Pd
loadings corresponding to estimated Pd particle sizes of 1.60, 2.46
and 6.73 nm (Table 1), respectively. Fig. 5 shows that the products
distribution is not significantly affected by the palladium particle
size.
4. Discussion
4.1. Effect of WO3 on Pd/ZrO2 in aqueous phase HDO of ethylene
glycol
Fig. 7. Ethanol TPD profile of 2Pd/ZrO2.
Removal of oxygen without unnecessarily breaking the C
C
ethanol took place at approximately 110 ◦C and the resultant eth-
ylene desorbed at 225 ◦C. The evolution of acetaldehyde and H2 at
about 340 ◦C was observed. Further increasing the temperature, the
adsorbed species decomposed into CH4 and CO at about 415 ◦C. For
WO3-ZrO2, there are two desorption features of ethanol (Fig. 8).
The low-temperature desorption feature at 97 ◦C corresponds to
the bound ethanol molecules in excess of one per acidic site [27].
The high-temperature feature at about 219 ◦C corresponds to stoi-
chiometric adsorption of ethanol molecules on Brönsted acidic sites
[27]. The dehydration of adsorbed ethanol on WO3-ZrO2 occurred
bond is preferable in aqueous phase HDO of polyols since the
resultant higher carbon-chain alcohols are better feeds for the sub-
sequent transformation process over zeolites [26]. Here a C O to
C bond cleavage ratio was defined by Eq. (1) to differentiate the
selectivity between C O and C C bond cleavage of catalysts.
C
C–O
C–C
ethanol + acetaldehyde + ethane × 2 + acetic acid
ꢀ
=
(1)
C1 products/2
The activity test results are summarized in Fig. 6. 2Pd/ZrO2 led
to mainly C1 products (including methanol, methane, CO and CO2)
resulting from C C bond cleavage. The C O to C C bond cleavage
ratio of 2Pd/ZrO2 is as low as 0.3. By doping of WO3, the selectiv-
ity to C1 products decreased from 77% to 41% while the selectivity
to ethanol was increased from 15% to 48%. The C O to C C bond
cleavage ratio increased from 0.3 to 1.52 suggesting that WO3 sig-
nificantly changed the reaction pathway of aqueous phase HDO of
ethylene glycol. The C O bond cleavage relies heavily on the pres-
ence of WO3. Doping of WO3 enhances the C O bond cleavage and
suppresses the C C bond cleavage over Pd/ZrO2 in aqueous phase
HDO of ethylene glycol.
277 oC
WO3-ZrO2
H2
H2O
ethylene
ethanol
m/z 44
97 oC
219 oC
The doping of WO3 onto the ZrO2 introduced a large amount
of Brönsted acidic sites to ZrO2 and the resultant Pd/WO3-ZrO2
as evidenced by IR-py spectra (Fig. 2). Ethanol was employed as
the probe molecule to study the adsorption of reactant molecules
on the catalysts (Figs. 7–9). Fig. 7 shows that ethanol desorbed
as molecular ethanol at temperature below 100 ◦C on 2Pd/ZrO2.
The evolution of H2O and ethylene were observed at about 110 ◦C
and 225 ◦C respectively. This suggests that the dehydration of
0
100
200
300
400
500
600
700
Temperature, oC
Fig. 8. Ethanol TPD profile of WO3-ZrO2.
Please cite this article in press as: C. Liu, et al., Aqueous phase hydrodeoxygenation of polyols over Pd/WO3-ZrO2: Role of Pd-WO3
interaction and hydrodeoxygenation pathway, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.10.034

G Model
CATTOD-9885; No. of Pages7
ARTICLE IN PRESS
C. Liu et al. / Catalysis Today xxx (2015) xxx–xxx
5
20
6
5
4
3
2
1
0
1.6 nm
2.5 nm
6.7 nm
15
10
5
0
Ethylene
glycol
C-C
C-O
C-O/C-C
Fig. 9. Ethanol TPD profile of 2Pd/WO3-ZrO2.
at 277 ◦C as indicated by the simultaneous evolution of ethylene
and H2O. No significant acetaldehyde evolution was observed on
WO3-ZrO2. The impregnation of palladium seems to attenuate the
strength of these ethanol adsorption sites on WO3-ZrO2 since the
high-temperature desorption feature of ethanol at 219 ◦C on WO3-
ZrO2 (Fig. 8) is not present on 2Pd/WO3-ZrO2 (Fig. 9). There are
two distinct peaks of mass spectra m/z 28 at 219 ◦C and 348 ◦C
along with the evolution of H2 for 2Pd/WO3-ZrO2. The peak at
219 ◦C corresponds to ethylene possibly from ethanol dehydration.
The dehydration temperature is much lower on 2Pd/WO3-ZrO2
(<200 ◦C) than that on WO3-ZrO2 (∼277 ◦C). The peak at 348 ◦C
corresponds to CO evolution due to the decomposition of some
strongly adsorbed species resulted from the dissociative adsorption
of ethanol (similar decomposition peak was observed in acetalde-
hyde TPD, Fig. S1). The results suggest that the highly selective
hydrodeoxygenation over 2Pd/WO3-ZrO2 originates from the com-
plex and synergic effect of palladium and WO3. To summarize,
WO3 introduces Brönsted acidic sites to Pd/ZrO2 and the acidity
of WO3-ZrO2 is somehow attenuated by palladium.
Fig. 10. Effect of palladium particle size on the reaction rates and selectivity. Cata-
lysts: Pd/WO3-ZrO2 with 0.5–2 wt.% Pd; carbon balance: 100 2%; conversion: ∼5%,
feed: 10 wt.% ethylene glycol, T = 240 ◦C, P = 1050 psig, WHSV = 0.25–0.7 g g−1 h−1
.
leading to improved catalyst stability as evidenced by only slight
decrease in ethylene glycol conversion (12.1% vs. 13.9%) after 8 h
time-on-stream. It indicates that the presence of both palladium
and WO3 is crucial for the selective hydrodeoxygenation of ethyl-
ene glycol.
Isopropylamine is known to form 1:1 complex with Brönsted
acidic sites and decompose to propene and ammonia at tem-
peratures between 302 ◦C and 377 ◦C, in a reaction analogous to
solution-phase Hoffmann elimination [27,28]. The decomposition
of isopropylamine led to simultaneous evolution of propene and
NH3 (Fig. S3). Fig. 12 shows the NH3 evolution profiles of the
catalysts during the isopropylamine-TPD. The decomposition tem-
perature of isopropylamine over WO3-ZrO2 is about 360 ◦C which
is consistent with literature [27,28]. However the introduction of
palladium significantly lowers the decomposition temperature of
isopropylamine as evidenced by the evolution of NH3 at a much
lower temperature. Although the 2Pd/ZrO2 is capable of decom-
posing isopropylamine to ammonia at temperatures below 300 ◦C,
the presence of both Pd and WO3 largely enhances the decom-
position as evidenced by the intensive NH3 evolution signal over
Pd/WO3-ZrO2 catalysts. This strongly suggests a synergic effect
between palladium and WO3. Different from WO3-ZrO2, however,
only a small amount of propene was detected during isopropy-
lamine decomposition over Pd/WO3-ZrO2 (Fig. S4), likely due to
the strong adsorption or polymerization on Pd [29,30]. The high
temperature desorption feature of isopropylamine above 200 ◦C
4.2. Interaction between palladium and WO3
Pd/WO3-ZrO2 catalysts with various Pd particle sizes were
prepared and studied to understand the interaction between palla-
dium and WO3 in Pd/WO3-ZrO2. The C O and C C bond cleavage
rates were derived from the generation rates of corresponding C
O
and C C cleavage products based on per mole of accessible palla-
dium on the catalyst surface. The reaction rates of ethylene glycol,
the cleavage rates of C O and C C bonds as well as the C O/C
C
ratios are summarized in Fig. 10. The rates of C O and C C bond
cleavage increase with increasing palladium particle size with-
out significantly changing the selectivity. The diffuse reflectance
UV–vis absorption spectra showed that the WO3 domain size
increases with the increasing palladium particle size (Fig. 3). This
indicates that there is strong interaction between Pd particles and
WO3-ZrO2.
Activity tests show that ZrO2 is inactive for the aqueous
phase HDO of ethylene glycol under the experimental conditions.
Although WO3-ZrO2 led to the high selectivity to acetaldehyde,
it deactivated after 8 h time-on-stream due to the severe coke
deposition (Fig. S2). Acetaldehyde was possibly produced by the
dehydration of ethylene glycol over acidic sites. Temperature
programmed desorption of isopropylamine (isopropylamine-TPD)
was employed to characterize the nature of the acidic sites
(Fig. 11). WO3-ZrO2 has stronger acidic sites which led to the high-
temperature region desorption of isopropylamine. Those stronger
acidic sites of WO3-ZrO2 are prone to coke formation (Fig. S2).
Such strong acidic sites disappear after impregnation of palladium,
Fig. 11. Isopropylamine TPD profiles of catalysts.
Please cite this article in press as: C. Liu, et al., Aqueous phase hydrodeoxygenation of polyols over Pd/WO3-ZrO2: Role of Pd-WO3
interaction and hydrodeoxygenation pathway, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.10.034

G Model
CATTOD-9885; No. of Pages7
ARTICLE IN PRESS
C. Liu et al. / Catalysis Today xxx (2015) xxx–xxx
Table 2
6
Composition of deuterated ethylene glycol in D2O after reaction with H2.
Ethylene glycol
WO3-ZrO2, mol%
2Pd/WO3-ZrO2, mol%
DOCH2CH2OD
DOCH2CHDOD
DOCHDCHDOD
DOCD2CH2OD
96.0
1.0
2.3
27.9
17.5
42.7
12.0
0.6
Table 3
Composition of deuterated ethanol products.
Ethanola
WO3-ZrO2, mol%
2Pd/WO3-ZrO2, mol%
34.1
DOCH2CH3/DOCH2CH2D/
DOCH2CHD2/DOCH2CD3
DOCHDCH3/DOCHDCH2D/
DOCHDCHD2/DOCHDCD3
DOCD2CH3/DOCD2CH2D/
DOCD2CHD2/DOCD2CD3
100
0
67.5
6.5
0
Fig. 12. NH3 evolution during isopropylamine TPD.
a
Produced from the HDO of deuterated ethylene glycol in D2O with H2.
over WO3-ZrO2 is not present on Pd/WO3-ZrO2 catalysts (Fig. 11)
possibly due to the strong acidic sites being blocked by Pd particles.
The deposition of Pd on these strong acidic sites could induce an
electronic interaction between Pd and WO3-ZrO2 [21,31] which in
turn modifies the acidity and reducibility of the Pd/WO3-ZrO2 cat-
alysts [32,33]. As a result, a synergic effect between palladium and
WO3 facilitates the elimination of NH3 from isopropylamine. Such
an elimination process could be similar to the elimination of H2O
from polyols in aqueous phase HDO process.
Table 4
Composition of deuterated methanol products.
Methanola
WO3-ZrO2,
mol%
2Pd/WO3-ZrO2,
mol%
2Pd/WO3-ZrO2,b
mol%
CH3OD
0
0
0
0
11.1
31.4
38.0
19.5
0.6
1.6
13.7
84.0
CDH2OD
CD2HOD
CD3OD
a
Produced from the HDO of deuterated ethylene glycol in D2O with H2.
H-D scrambling using 10 wt.% CD3OH in H2O and H2 as the feed.
b
4.3. Aqueous phase HDO mechanism of ethylene glycol
should lead to ethanol with both deuterated methyl and methylene
groups (CH2DCHDOD) (Scheme 2).
With both palladium and WO3, hydrodeoxygenation of ethyl-
ene glycol and resultant alcohols could proceed via either direct
hydrogenolysis of C O bonds or a dehydration–hydrogenation
pathway. Isotope-labeled experiments were designed to obtain
more direct evidence on the aqueous phase HDO pathway of
ethylene glycol. Deuterated ethylene glycol (DOCH2CH2OD) was
employed instead of its non-deuterated counterpart in the tests.
These isotope-labeled experiments were conducted with D2O as
solvent and H2 as the hydrogen source at the same reaction tem-
perature and pressure used in the previous activity tests.
Table 2 summarizes the H-D scrambling results of ethylene gly-
col after the HDO tests. Table 3 summarizes the composition of HDO
products of ethanol. Interestingly, very limited H-D scrambling of
DOCH2CH2OD (∼4%) was observed over WO3-ZrO2. The majority of
the ethylene glycol (DOCH2CH2OD) remained untouched. Dehydra-
tion followed by isomerization lead to a deuterated methyl group
on ethanol. No C C bond hydrogenolysis product (methanol) over
WO3-ZrO2 was observed. This confirms that the acidic sites on
WO3-ZrO2 favor the dehydration of ethylene glycol.
The hypothesis is that if hydrodeoxygenation proceeds via direct
hydrogenolysis of ethylene glycol, both the ethanol and methanol
produced from deuterated ethylene glycol in D2O should have zero
or one D on their methyl group due to the scrambling of H on
metal by D from D2O (Scheme 1). If there are more than one D
on the methyl groups in ethanol or methanol, the D should be a
result of direct H-D exchange on metal. On the other hand, if the
hydrodeoxygenation proceeds via a dehydration and hydrogena-
tion pathway, vinyl alcohol (CH2 = CHOD) should be the primary
dehydration product of ethylene glycol. Isomerization of vinyl
alcohol to acetaldehyde involves the hydrogen transfer from an
– OD group to the methylene group of a vinyl alcohol leading
to an acetaldehyde intermediate with a deuterated methyl group
(CH2DCHO). However, both direct hydrogenation of vinyl alcohol
and the resultant acetaldehyde from vinyl alcohol isomerization
Compared to WO3-ZrO2, more than 70% of remaining eth-
ylene glycol (Table 2) has experienced H-D scrambling over
2Pd/WO3-ZrO2 during the test. However, the H-D scrambling of
ethylene glycol over 2Pd/WO3-ZrO2 could either proceed via the
dehydration–hydration pathway or direct H-D scrambling over
metallic sites. Deuterated methanol (CD3OH) in H2O was employed
to evaluate the direct H-D scrambling over 2Pd/WO3-ZrO2 dur-
ing the aforementioned aqueous phase HDO conditions with H2.
Table 4 summarizes the composition of methanol from HDO and the
H-D scrambling results of deuterated methanol. Methanol dehydra-
tion can only proceed between hydroxyl groups of two methanol
molecules to form the dimethyl ether (CD3OCD3). Hydrolysis of
the resultant dimethyl ether in H2O will form CD3OH and no H-
D scrambling on the methyl group is possible. If there is only H-D
scrambling of the CD3 group of methanol, it must proceed over
Scheme 1. Direct hydrogenolysis reaction pathway.
Please cite this article in press as: C. Liu, et al., Aqueous phase hydrodeoxygenation of polyols over Pd/WO3-ZrO2: Role of Pd-WO3
interaction and hydrodeoxygenation pathway, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.10.034

G Model
CATTOD-9885; No. of Pages7
ARTICLE IN PRESS
C. Liu et al. / Catalysis Today xxx (2015) xxx–xxx
7
Scheme 2. Dehydration–hydrogenation reaction pathway.
the metallic sites of 2Pd/WO3-ZrO2. Table 4 shows that only about
16% of the CD3OH has experience H-D scrambling, much less than
the H-D scrambling of ethylene glycol (>70%) over 2Pd//WO3-ZrO2
(Table 2). This suggests that the dehydration–hydration followed by
hydrogenation pathway prevails over hydrogenolysis of ethylene
glycol during aqueous phase HDO over 2Pd//WO3-ZrO2.
Appendix A. Supplementary data
Figures S1–S4 related to this article can be found, in the online
version, at http://dx.doi.org/10.1016/j.cattod.2015.10.034.
References
Table
over 2Pd//WO3-ZrO2 has deuterated methylene groups
CHD CD2 ) which suggest that the majority of ethanol
3 also shows that 70% of the ethanol produced
[1] G.W. Huber, B.E. Dale, Sci. Am. 301 (2009) 52–56.
[2] J.C. Serrano-Ruiz, R. Luque, A. Sepulveda-Escribano, Chem. Soc. Rev. 40 (2011)
5266–5281.
(
/
is produced by dehydration–hydrogenation of ethylene glycol
(DOCH2CH2OD). Thus the high degree H-D scrambling of the
remained ethylene glycol and resultant ethanol, methanol prod-
ucts over 2Pd//WO3-ZrO2 suggests that the dehydration step
over 2Pd/WO3-ZrO2 is in equilibrium and not the rate-limiting
step of HDO. In summary, the aqueous phase HDO of ethylene
glycol proceeds via the dehydration–hydrogenation reaction
pathway over the bifunctional 2Pd/WO3-ZrO2 and hydrogenation
of the dehydration products shifts the dehydration–hydration
equilibrium toward the dehydration direction.
[3] D. Mohan, C.U. Pittman, P.H. Steele, Energy Fuels 20 (2006) 848–889.
[4] C. Liu, H. Wang, A.M. Karim, J. Sun, Y. Wang, Chem. Soc. Rev. 43 (2014)
7594–7623.
[5] G.W. Huber, S. Iborra, A. Corma, Chem. Rev. 106 (2006) 4044–4098.
[6] Y. Sun, J. Cheng, Bioresour. Technol. 83 (2002) 1–11.
[7] J.-P. Lange, Biofpr 1 (2007) 39–48.
[8] E.L. Kunkes, D.A. Simonetti, R.M. West, J.C. Serrano-Ruiz, C.A. Gärtner, J.A.
Dumesic, Science 322 (2008) 417–421.
[9] L. Chen, Y. Zhu, H. Zheng, C. Zhang, B. Zhang, Y. Li, J. Mol. Catal. A: Chem. 351
(2011) 217–227.
[10] G.W. Huber, R.D. Cortright, J.A. Dumesic, Angew. Chem. Int. Ed. 43 (2004)
1549–1551.
[11] N.L. Pehlivanog˘lu, H.S.I.S. Uslu, L. Kirbas¸ lar, J. Chem. Eng. Data 55 (2010)
2970–2973.
[12] L. Vilcocq, A. Cabiac, C. Especel, S. Lacombe, D. Duprez, Catal. Today 189
(2012) 117–122.
5. Conclusions
[13] N. Li, G.A. Tompsett, T. Zhang, J. Shi, C.E. Wyman, G.W. Huber, Green Chem. 13
(2011) 91–101.
[14] C.D. Baertsch, K.T. Komala, Y.-H. Chua, E. Iglesia, J. Catal. 205 (2002) 44–57.
[15] D. Barton, S. Soled, E. Iglesia, Top. Catal. 6 (1998) 87–99.
[16] D.G. Barton, M. Shtein, R.D. Wilson, S.L. Soled, E. Iglesia, J. Phys. Chem. B 103
(1999) 630–640.
[17] G. Larsen, E.L. Lotero, A.M. Petkovic, D.S. Shobe, J. Catal. 169 (1997) 67–75.
[18] T.K. Phung, L. Proietti Hernández, G. Busca, Appl. Catal. A: Gen. 489 (2015)
180–187.
[19] S.-H. Chai, B. Yan, L.-Z. Tao, Y. Liang, B.-Q. Xu, Catal. Today 234 (2014)
215–222.
[20] W. Dedsuksophon, K. Faungnawakij, V. Champreda, N. Laosiripojana,
Bioresour. Technol. 102 (2011) 2040–2046.
[21] O.G. Marin-Flores, A.M. Karim, Y. Wang, Catal. Today 237 (2014) 118–124.
[22] G. Bergeret, P. Gallezot, Particle size and dispersion measurements, in: G. Ertl,
H. Knözinger, J. Weitkamp (Eds.), Handbook of Heterogeneous Catalysis,
Wiley-VCH Verlag GmbH & Co. KGaA, 2008.
The strong interaction between metallic and acidic sites of
2Pd/WO3-ZrO2 is required to efficiently remove oxygen from
sugars to form mono-functional oxygenates in aqueous phase
HDO. Aqueous phase HDO of ethylene glycol proceeds via a
dehydration–hydrogenation pathway on 2Pd/WO3-ZrO2 with the
dehydration step being in equilibrium. Hydrogenation of dehy-
dration products shifts the equilibrium toward the dehydration
direction, leading to selective cleavage of C O. A fundamental
understanding on the aqueous phase HDO mechanism and the
interaction between the metal and acid sites will help develop more
efficient bifunctional catalysts for this process.
Acknowledgments
[23] E.P. Parry, J. Catal. 2 (1963) 371–379.
[24] G.K. Chuah, S. Jaenicke, T.H. Xu, Surf. Interface Anal. 28 (1999) 131–134.
[25] R. Kourieh, S. Bennici, A. Auroux, React. Kinet. Mech. Catal. 105 (2012)
101–111.
[26] K.K. Ramasamy, Y. Wang, J. Energy Chem. 22 (2013) 65–71.
[27] W.E. Farneth, R.J. Gorte, Chem. Rev. 95 (1995) 615–635.
[28] T.J.G. Kofke, R.J. Gorte, W.E. Farneth, J. Catal. 114 (1988) 34–45.
[29] E.I. Altman, Surf. Sci. 547 (2003) 108–126.
We would like to acknowledge Virent for their initial work on
Pd/WO3-ZrO2 for HDO of polyols, especially Dr. Randy Cortright,
Ms. Liz Woods and Mr. Brian Blank from Virent for numerous
invaluable discussions and suggestions. We also acknowledge the
financial support from the National Advanced Biofuels Consortium
(NABC) which is funded by the Department of Energy’s Office of
Biomass Program with recovery act funds, as well as the financial
support by the U.S. Department of Energy (Grant No. DE-FG02-
05ER15712). The Pacific Northwest National Laboratory is operated
by Battelle for the United States Department of Energy under Con-
tract DE-AC05-76RL01830. We thank Berlin Sudduth (WSU) for
reading through the paper.
[30] T. Perger, T. Kovács, T. Turányi, C. Trevin˜o, Combust. Flame 142 (2005)
107–116.
[31] P. Vijayanand, K. Chakarova, K. Hadjiivanov, P. Lukinskas, H. Knozinger, Phys.
Chem. Chem. Phys. 5 (2003) 4040–4044.
[32] M. Occhiuzzi, D. Cordischi, S. De Rossi, G. Ferraris, D. Gazzoli, M. Valigi, Appl.
Catal. A: Gen. 351 (2008) 29–35.
[33] X. Wu, L. Zhang, D. Weng, S. Liu, Z. Si, J. Fan, J. Hazard. Mater. 225–226 (2012)
146–154.
Please cite this article in press as: C. Liu, et al., Aqueous phase hydrodeoxygenation of polyols over Pd/WO3-ZrO2: Role of Pd-WO3
interaction and hydrodeoxygenation pathway, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.10.034