Liquid-phase glycerol hydrogenolysis to 1,2-propanediol under nitrogen pressure using 2-propanol as hydrogen source

Article abstract of DOI:10.1016/j.jcat.2011.06.020

2-Propanol was studied as a hydrogen donor molecule in the transfer hydrogenation process to selectively convert glycerol into 1,2-propanediol under N₂ pressure and using Ni or/and Cu supported on Al₂O ₃ catalysts. The results were compared to those obtained under the same operating conditions but under H₂ pressure. The results of the activity tests and catalyst characterization techniques (N₂- physisorption, H₂-chemisorption, TPD of NH₃, TPR, TPO and XPS) suggest that glycerol hydrogenolysis to yield 1,2-propanediol occurred through a different mechanism regarding the origin of the hydrogen species. When atomic hydrogen came from dissolved molecular hydrogen dissociation, glycerol was first dehydrated to acetol and then hydrogenated to 1,2-propanediol. On the other hand, when the hydrogen atoms were produced from 2-propanol dehydrogenation, glycerol was directly converted to 1,2-propanediol through intermediate alkoxide formation.

Full text of DOI:10.1016/j.jcat.2011.06.020

Journal of Catalysis 282 (2011) 237–247
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Liquid-phase glycerol hydrogenolysis to 1,2-propanediol under nitrogen
pressure using 2-propanol as hydrogen source
⇑
I. Gandarias , P.L. Arias, J. Requies, M. El Doukkali, M.B. Güemez
School of Engineering (UPV/EHU), Calle Alameda Urquijo s/n, 48013 Bilbao, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
2-Propanol was studied as a hydrogen donor molecule in the transfer hydrogenation process to selec-
Received 16 February 2011
Revised 26 May 2011
Accepted 17 June 2011
Available online 26 July 2011
tively convert glycerol into 1,2-propanediol under N
Al catalysts. The results were compared to those obtained under the same operating conditions but
under H pressure. The results of the activity tests and catalyst characterization techniques (N -physi-
sorption, H -chemisorption, TPD of NH , TPR, TPO and XPS) suggest that glycerol hydrogenolysis to yield
,2-propanediol occurred through a different mechanism regarding the origin of the hydrogen species.
When atomic hydrogen came from dissolved molecular hydrogen dissociation, glycerol was first dehy-
drated to acetol and then hydrogenated to 1,2-propanediol. On the other hand, when the hydrogen atoms
were produced from 2-propanol dehydrogenation, glycerol was directly converted to 1,2-propanediol
through intermediate alkoxide formation.
2
pressure and using Ni or/and Cu supported on
2 3
O
2
2
2
3
1
Keywords:
Glycerol
Hydrogenolysis
Propanediol
Catalytic transfer hydrogenation
Ni–Cu/Al
Hydride
2
O
3
Ó 2011 Elsevier Inc. All rights reserved.
1
. Introduction
Nowadays, there is a tendency to reduce the use of fossil hydro-
hydrogenolysis at near ambient hydrogen pressure [4,5]. Liquid-
phase glycerol hydrogenolysis has also been studied [6–10], as it
has both energy- and plant-scale reduction benefits. Nevertheless,
reported results are not as promising as the ones from vapour
phase and two-step processes. In liquid-phase reactions, high
hydrogen pressures are needed due to the low solubility of hydro-
carbon sources in the production of commodity chemicals. The
main reasons are increasing and fluctuating oil prices, the drive
to renewable resources and a global concern to reduce CO emis-
2
sions. Glycerol, obtained as a by-product in biodiesel manufacture,
is a versatile feedstock for the production of a whole range of
chemicals, polymers and fuels [1]. The glycerol hydrogenolysis
process for obtaining propanediols (PDOs) has aroused consider-
able interest due to the attractive applications of 1,2-PDO, an
important commodity chemical traditionally derived from propyl-
ene oxide, and 1,3-PDO, a monomer that can be used to produce
polyester fibres. In the reaction pathway, glycerol is first dehy-
drated to 1-hydroxypropan-2-one (acetol) or 3-hydroxypropanal
2
gen in glycerol/water solutions, and limited H availability may
cause undesired side reactions such as cracking and coking.
Hydrogen accessibility problems could be avoided if the hydro-
gen required for glycerol hydrogenolysis was to be generated di-
rectly in the active sites of the catalyst, allowing a process with
inert atmosphere and lower working pressure. Among others,
two different processes can be considered to generate hydrogen:
aqueous-phase reforming (APR) and catalytic transfer hydrogena-
tion (CTH). Hydrogen production from glycerol APR is a well-re-
ported process [11,12].
(
3-HPA), which are subsequently hydrogenated to 1,2 and 1,3-
PDO, respectively [2]. Several reaction systems have been studied
to maximize glycerol conversion and PDO selectivity. Dasari et al.
developed a two-step reaction process using a copper chromite
catalyst. In the first step, acetol was produced through glycerol
reactive distillation at 473 K and 0.65 bar, while in the second step
the acetol was further hydrogenated to 1,2 PDO at 473 K and
C
3
H
5
ðOHÞ þ 3H
2
O ! 3CO
2
þ 7H
2
3
The production of PDOs by APR of glycerol has already been
studied [13]. The process initially involves hydrogen and CO for-
mation. The hydrogen produced in the reforming step is then con-
sumed in the hydrogenolysis of glycerol, leading to an overall
2
1
3.8 bar hydrogen pressure [3]. Interesting results have been re-
2
conversion of glycerol to 1,2-PDO, CO and water [14].
ported using Cu/Al in single-step vapour-phase glycerol
2 3
O
CTH, in which hydrogen is transferred from a hydrogen donor
moleculeto an acceptor, is an interestingprocess for reducingorgan-
ic compounds, as it has real advantages compared to processes with
molecular hydrogen. Molecular hydrogen has high diffusibility,
being easily ignited and presents considerable hazards on a large
⇑
Corresponding author. Address: Escuela Técnica Superior de Ingeniería, Ala-
meda Urquijo s/n, PC 48013 Bilbao, Spain. Fax: +34 946 014 179.
E-mail address: inaki_gandarias@ehu.es (I. Gandarias).
0
021-9517/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2011.06.020

2
38
I. Gandarias et al. / Journal of Catalysis 282 (2011) 237–247
scale; the use of hydrogen donors obviates these difficulties [15].
Alcohols have been widely used in CTH processes [16], and 2-
propanol is a suitable hydrogen donor for glycerol hydrogenolysis
Conversion of glycerol: %
sum of C-based mol of all liquid prod: t ¼ t
¼
C-based mol of glycerol: t ¼ 0
[
17].
This paper studies aqueous-phase reforming and catalytic
The selectivity of the products was calculated on a carbon basis.
transfer hydrogenation processes as sources of hydrogen for glyc-
erol hydrogenolysis to PDOs.
Selectivity of liquid products %
C-based mol of the product
¼
2
. Experimental
Sum of C-based mol of all liquid products
0
Initial Turnover Number (TON ) was calculated as the ratio between
the converted amount of glycerol in the first 2 h per gram of catalyst
and per hour.
2
.1. Catalysts
Amorphous silica-alumina loaded with 1 wt.% Pt (Pt/ASA) was
ðAÞ
kindly supplied by Shell, while Ni—Cu=Al
by the Boreskov Institute of Catalysis. CuCr
Süd-Chemie. Ni=Al
pared by the sol–gel method. Aluminium isopropoxide (Aldrich)
was dissolved in deionised water (9 mL of H O per gram of alumin-
2
O
was kindly supplied
3
2.3. Catalyst characterization
2
O
3
was purchased from
ðSÞ
ðSÞ
ðSÞ
2
O
3
; Cu=Al
2
O
3
and Ni—Cu=Al
2
O
3
were pre-
The catalysts were chemically analysed by Inductively Coupled
Plasma Atomic Emission (ICP-AES) using a Perkin–Elmer Optima
000 instrument. The solid samples were first digested with a mix-
ture of HF, HCl and HNO at 453 K in a microwave oven. Surface
area, pore volume and pore size distributions were determined
with N physisorption at 77 K on a Quantachrome AUTOSORB-1C
2
2
ium isopropoxide) by vigorous stirring of the solution at 313 K. The
pH was measured and kept between 3.8 and 4.2 adding the re-
3
quired amounts of HNO
hexahydrate (Aldrich) and/or Copper(II) nitrate hemi pentahydrate
Alfa Aesar) were dissolved in ethanol. The precursor solution was
3
(0.5 M). Simultaneously, nickel(II) nitrate
2
instrument. Prior to the analysis, all samples were dried at 393 K
overnight under high vacuum. The surface area was calculated
using the Brunauer, Emmett and Teller (BET) method, while pore
size distributions were calculated using the Barrett–Joyner–Halen-
da (BJH) method applied to the desorption leg of the isotherms.
The reducibility of catalysts was studied by hydrogen tempera-
ture-programmed reduction (TPR) on a Quantachrome AUTO-
SORB-1C apparatus with TPR capability. The catalyst sample was
reduced in flowing gas containing 5 vol.% H in Ar at a total flow
2
rate of 50 mL/min, using a heating rate of 10 K/min up to a final
temperature of 1223 K. A TCD detector downstream of the sample
2
monitored changes in the concentration of H . Hydrogen chemi-
sorption was performed at 313 K with the Quantachrome AUTO-
SORB-1C volumetric system. All the catalysts were reduced at
(
slowly added to an aluminium isopropoxide solution. The mixture
was stirred for 30 min at 313 K and then introduced into the ultra-
sonic apparatus for another 30 min. The mixture was then rested
for 24 h at 313 K and subsequently for another 12 h at 375 K. The
product obtained was crushed and calcined from room tempera-
ture to 723 K at a heating rate of 2 K/min. The temperature was
maintained for 4 h. Catalyst samples for activity tests were used
in powdered form with a granule size between 320 and 500
.2. Activity test
The hydrogenolysis of glycerol was carried out in a 50 mL
lm.
2
stainless steel autoclave with a magnetic stirrer. The catalyst pow-
der (166 mg catalyst/g of glycerol) was introduced into the auto-
clave and the reactor was then purged with H
8
73 K in pure H
2
flow for 4 h prior to the measurements. The quan-
tity of H adsorbed at monolayer coverage was estimated by
2
2
or N
, Ni—Cu=Al
2
. After
ðAÞ
ðSÞ
extrapolating the linear portion of the isotherms to zero pressure.
Temperature-programmed oxidation analyses of fresh and used
catalysts were carried out using a thermo-gravimetric analyser
purging, some of the catalysts (Ni—Cu=Al
2
O
2
O
,
3
3
ðSÞ
ðSÞ
Ni=Al
2
O
and Cu=Al
/N
Pt/ASA and CuCr
perature was set to 493 K and the N
2
O
Þ were pre-treated, reducing them under
3
3
a 50 vol.% H
2
2
flow for 4 h at 593 or 723 K, while the others
) were used as received. Next, the reactor tem-
or H pressure was increased
(
Mettler Toledo TGA/SDTA 851e). The standard protocol involved
(
2 3
O
the pre-treatment of the sample (45–50 mg) in 125 mL/min of N
2
2
2
flow from 297 K to 673 K at a heating rate of 10 K/min. The sample
was then cooled to 323 K and the weight change of the sample was
to 45 bar. The aqueous solution (41 mL) with the reactants was
placed on a feed cylinder and heated to the reaction temperature.
The reaction starting time was established when the line connect-
ing the feed cylinder and the reactor was opened. During the reac-
tion, changes in the system pressure were observed: a decrease in
the experiments under H
an increase in the experiments under N
donor molecule was added (due to the formation of H
continuously monitored during its heating in 125 mL/min of N
purge gas and 75 mL/min of 10 vol.% O in He as reactive gas from
23 to 1173 K at a heating rate of 5 K/min.
The acidity of the freshly reduced samples was determined by
2
as
2
3
2
pressure (due to H
pressure when a hydrogen
and other
2
consumption), and
ammonia temperature-programmed desorption (TPD) measure-
ments. The sample was pre-treated in a He stream at 673 K for
2
2
0
.5 h and then cooled to 373 K and ammonia-saturated using a
gaseous species). Nevertheless, as diluted glycerol and donor feeds
were used, the pressure variations were negligible.
stream of 5 vol.% NH /He flow (50 mL/min) for 0.5 h. Following cat-
3
alyst equilibration in a helium flow, the ammonia was desorbed
using a linear heating rate of 10 K/min to 723 K. The area under
the curve was integrated to determine the total acidity of the sam-
Five liquid samples were taken throughout the reaction in order
to obtain the time evolution of reactant and product concentrations.
These compounds were analysed using a gas chromatograph (Agi-
lentTechnologies, 7890A)equippedwithaflameionizationdetector
ple from its NH
X-ray photoelectron spectroscopy (XPS) studies were per-
formed with Physical Electronics PHI 5700 spectrometer
equipped with a hemispherical electron analyser (model 80-
65B) and a Mg K (1253.6 eV) X-ray source. High-resolution spec-
3
desorption profile.
(
FID) and a thermal conductivity detector (TCD). A Meta-Wax capil-
a
lary column (diameter 0.53 mm, length 30 m) was used for product
separation. After reaction, the gas phase was collected in a gas bag
and analysed with another GC-TCD-FID (Agilent Technologies,
3
a
tra were recorded at a 45° take-off-angle by a concentric hemi-
spherical analyser operating in the constant pass energy mode at
7
890 A) equipped with a molecular sieve column (HP-MOLESIEVE,
diameter 0.535 mm, length 30 m) and a capillary column (HP-
PLOT/Q, diameter 0.320 mm, length 30 m). The conversion of the
reactants was calculated according to following equation:
2
9.35 eV, using a 720 mm diameter analysis area. Charge referenc-
ing was done against adventitious carbon (C 1s 284.8 eV). The pres-
ꢁ
6
sure in the analysis chamber was kept below 5 ꢀ 10 Pa. The PHI

I. Gandarias et al. / Journal of Catalysis 282 (2011) 237–247
239
ðAÞ
ACCESS ESCA-V6.0 F software package was used for data acquisi-
tion and analysis. A Shirley-type background was subtracted from
the signals. Recorded spectra were always fitted using Gauss–
Lorentz curves in order to more accurately determine the binding
energy of the different element core levels.
conversion for Pt/ASA and Ni—Cu=Al
2
O
catalysts was achieved
3
in the tests with the hydrogen donor, indicating that generating
hydrogen directly in the active sites benefits glycerol conversion.
Concerning the effect of the catalyst, it can be observed that Cu-
2 3
Cr O catalyst had extremely low activity under all the operating
conditions tested. In the case of Pt/ASA, lower selectivities to the
first hydrogenolysis products (acetol + 1,2-PDO) were observed as
3
. Results and discussion
ðAÞ
compared to Ni—Cu=Al
2
O
. It is well known that Pt combined
3
with strong acid sites catalyses the C–C bond cleavage [11]; there-
fore, cracking product formation and further 1,2-PDO hydrogenol-
ysis to 1-propanol were responsible for the lower selectivity values
to the initial hydrogenolysis products obtained with Pt/ASA. Cu-
3
.1. Hydrogen donor selection
In order for cross-oxidation/reduction to occur, it has been re-
ported that adjacent sites may be necessary for donor and acceptor
18]. Therefore, the first criterion to be fulfilled by the selected
2 3
Cr O catalyst did not effectively dehydrogenate 2-PO (low 2-PO
[
conversion and low selectivity to acetone). Pt and Ni–Cu metal
sites were more effective in 2-PO dehydrogenation and therefore
in the CTH process.
In the light of these results, the use of a hydrogen donor seems
to be a more effective source of hydrogen for glycerol hyd-
hydrogen donor molecules was to be soluble in glycerol. Moreover,
in order to improve the process yield, reactions other than dehy-
drogenation of the donor should be minimized under the operating
conditions.
Alcohols have shown to be suitable for the CTH of ketones and
are soluble in glycerol. Within a series of alcohol isomers, it is
reported that 2-alkanols have higher catalytic activity than 3-
alkanols [19]. Use was made of 2-propanol (2-PO) as it is soluble
in both glycerol and water, and it has already been used as a hydro-
gen donor molecule in glycerol hydrogenolysis [17]. Under these
operating conditions, 2-PO can be dehydrogenated to acetone or
dehydrated to propene.
rogenolysis than glycerol APR. It was therefore decided to pre-
ðSÞ
pare three catalysts by the sol–gel method: Cu=Al
2
O
,
3
ðSÞ
ðSÞ
Ni—Cu=Al
2
O
and Ni=Al
2
O
in order to perform a more thorough
3
3
study of glycerol hydrogenolysis by CTH.
2 3
3.3. Glycerol hydrogenolysis by CTH on Ni and/or Cu Al O
3.3.1. N
2
physisorption and metal dispersion
3.2. Initial catalyst screening
Table 1 summarises the metal content together with the main
textural properties of the calcined catalysts and the chemisorp-
tion analysis results. All samples presented Type IV isotherms
with the hysteresis loops typical of mesoporous materials. The
The choice of a suitable catalytic system for the hydrogenation
processes selected was by no means random, as it had to fulfil sev-
eral requirements. The glycerol hydrogenolysis process requires a
bi-functional catalyst with active sites for glycerol dehydration
and active sites for acetol hydrogenation. Interesting results for
lower surface area of the impregnated samples with respect to
ðSÞ
bare Al
2
O
(also prepared by sol–gel) is consistent with the rel-
3
ðSÞ
ðSÞ
ative high metal loading (7–35%). Ni—Cu=Al
2
O
2
and Cu=Al O
3
3
glycerol hydrogenolysis have been obtained with CuCr
2
O
3
catalyst
had similar metal loadings; however, the measured
ðSÞ
[
3,5,20]. Ni–Cu-based catalysts have been used for heterogeneous
Ni—Cu=Al
2
O3 surface area and cumulative pore volume were
ðSÞ
transfer reactions [16]. On the other hand, Pt silica–alumina cata-
lysts have proven to be effective for alkane production by the
APR of sorbitol [21]. Taking all these considerations into account,
significantly higher as compared to Cu=Al
2
O
. As observed in
3
Table 1, the metal dispersion values obtained from hydrogen
chemisorption analysis follow the order: Ni > Ni–Cu > Cu. This
trend suggests that Cu particles are significantly bigger than Ni
and Ni–Cu particles. Therefore, big Cu particles could block the
ðAÞ
three commercial catalysts were tested: Pt/ASA, Ni—Cu=Al
and CuCr
Three activity tests were arranged for each of the selected cata-
lysts: under H pressure, under N pressure and under N pressure
2
O
3
2 3
O .
2 3
pores of the Al O structure, reducing the surface accessible for
2
2
2
N
2
adsorption.
but adding 2-PO as a hydrogen donor molecule (1:1 glycerol/2-PO
ðAÞ
molar ratio). In the experiments with CuCr
2
O
3
and Ni—Cu=Al
2
O
,
3.3.2. Temperature-programmed reduction
3
C3 products (acetol, 1,2-PDO, 1,3-PDO and 1-propanol) and prod-
ucts resulting from C–C bond cleavage (ethylene glycol, ethanol,
The TPR spectra of the investigated samples are shown in Fig. 2.
ðSÞ
Calcined Ni=Al
2
O
recorded one reduction peak at 650 K and an-
3
methane, and CO
2
) were detected. In the experiments using Pt/
other broad peak reduction profile in the 820–1200 K temperature
range. This fact suggests the presence of a mixture of NiO species
due to the different interaction of NiO particles with the support
[22]. On Ni-based catalysts, the low-temperature H uptakes are
attributed to the reduction of NiO particles weakly interacting with
the support, while the high temperature ones are assigned to the
reduction of NiO species in close contact with the support or form-
ASA, and besides those products, methanol, ethane, propanal and
propanoic acid were also detected. In the experiments where 2-
PO was added as a hydrogen donor molecule, acetone was detected
coming from 2-PO dehydrogenation, and propane and propene
were observed in the gas phase coming from 2-PO and acetone
dehydration.
x
2
First, the effect of the hydrogen source will be discussed. Under
ing species such as NiAl
2
O
4
[23]. The XPS analysis (discussed later)
ðSÞ
ðSÞ
H
2
pressure, most of the acetol formed from glycerol dehydration
was hydrogenated to 1,2-PDO (see Fig. 1A), showing that under a
5 bar H atmosphere, the hydrogenation of acetol was effective
with all three catalysts and that the controlling step in this hydrog-
enolysis process under H pressure seems to be the glycerol dehy-
dration step. In the experiments under N pressure, part of the
of Ni=Al
NiAl
2
O
and Ni—Cu=Al
2
O
samples proved the presence of
3
3
ðSÞ
2
O
4
in the surface of the catalyst. Calcined Cu=Al
2
O
3
recorded
0
4
2
two reduction peaks at 580 and 690 K. The reduction of CuO to Cu
generally proceeds in one step, so the peaks at 580 and 690 K could
be attributed to the reduction of two different oxidized copper spe-
cies. Previous works noted the presence of two reduction peaks for
Cu/zeolite supported catalysts [24,25]. It is suggested that the low-
er peak corresponds to the reduction of CuO crystallites and the
2
2
acetol formed from glycerol dehydration was not hydrogenated;
nevertheless, higher selectivities to 1,2-PDO were measured with
the addition of 2-PO (Fig. 1A). Therefore, it seems that CTH using
-PO was more effective than glycerol APR for propanediol forma-
tion. Moreover, as can be observed in Fig. 1B, the greater glycerol
2
+
higher one to the reduction of Cu fixed to the support. A compar-
ðSÞ
2
ison of the TPR profile of Ni—Cu=Al
2
O
with the profile of
3
ðSÞ
Ni=Al
2
O
reveals that the former records a displacement to lower
3

2
40
I. Gandarias et al. / Journal of Catalysis 282 (2011) 237–247
1
00
100
80
60
40
20
0
35
A
B
3
2
2
1
1
0
5
0
5
0
5
0
8
6
4
2
0
0
0
0
0
Pt/ASA
Ni-Cu/Al O (A)
CuCr O3
Pt/ASA
Ni-Cu/Al O (A)
CuCr O
2 3
2
3
2
2
3
Fig. 1. Selectivity values (A) to first hydrogenolysis products: 1,2-PDO (stripes) and acetol (plain), and glycerol conversion values (B) as a function of the catalyst used and the
reaction atmosphere: atmosphere. atmosphere. atmosphere + 2-PO (1:1 glycerol:2-PO molar ratio) 24-h reaction time, 45 bar pressure, 493 K, 41 mL (4 wt.%)
H
2
N
2
N
2
glycerol aqueous solution, 166 mg catalyst/g glycerol.
Table 1
Characteristics of the catalysts.
Nia (wt.%)
Cua (wt.%)
SBETb (m /g Al
2
O
Total pore volume (mL/g)
Dispersion (%)
MSAc (m /g)
2
Crystallite sized (nm)
Sample
2
3
)
ðSÞ
–
31.7
28.0
–
161
208
262
298
0.12
0.16
0.23
0.21
0.9
1.6
7.4
–
1.9
3.7
3.4
–
112
65.6
13.7
–
2
Cu=Al O
3
ðSÞ
7.7
6.9
–
2
Ni—Cu=Al O
3
ðSÞ
2
Ni=Al O
3
ðSÞ
–
2
Al O
3
a
b
c
Chemical composition determined with ICP.
BET surface area.
Metal Surface Area.
d
2
Assuming spherical geometry and H :M = 2:1 stoichiometry.
7
20–1050 K that can be associated with the reduction of Ni-based
species promoted by the presence of Cu [26].
3.3.3. Temperature-programmed oxidation
The left-hand column in Fig. 3 shows the results for the temper-
ature-programmed oxidation of the three freshly reduced samples.
ðSÞ
ðSÞ
As can be observed, both Cu=Al
2
O
and Ni—Cu=Al
2
O
samples
3
3
underwent oxidation within the 400–730 K temperature range.
ðSÞ
However, no weight increments were detected for the Ni=Al
2
O
3
sample during TPO, suggesting that the reduction treatment ap-
plied at 723 K did not lead to any significant reduction of Ni spe-
cies. This result is consistent with the TPR profile shown
previously. The three samples lost weight in the high-temperature
range (T > 700 K), which might be related to the decomposition of
Cu/Al O (S)
2
3
the nitrates remaining in the catalyst after calcination at 723 K
ðSÞ
[
27]. Concerning Cu=Al
2
O
freshly reduced samples, the derivative
3
profile presents two main oxidation peaks at 480 and 643 K, and a
smaller one at 730 K. Li and Inui [28] also observed two oxidation
Ni-Cu/Al O (S)
2
3
2 3
peaks in the TPO profiles of reduced Cu/Zn/Al O , which were
0
attributed to the stepwise oxidation of Cu . The low-temperature
peak, located in the 423–473 K region, was attributed to the oxida-
tion of Cu0 to Cu and the high-temperature peak (523–573 K) to
+
Ni/Al O (S)
2
3
+
2+
the oxidation of Cu to Cu . In agreement with this study [28],
ðSÞ
the two main peaks of Cu=Al
2
O
catalyst can be ascribed to the
3
0
stepwise oxidation of Cu . In this work, the peaks are shifted to-
wards higher temperatures, indicating a larger copper particle size
[29]. Finally, the highest temperature oxidation peak can be attrib-
uted to the oxidation of the copper species interacting strongly
with the support; the presence of two types of copper species
was already observed in the TPR profile previously discussed. For
4
00 500 600 700 800 900 100011001200
Temperature (K)
ðSÞ
ðSÞ
ðSÞ
Fig. 2. TPR profiles of calcined Cu=Al
2
O
, Ni-Cu=Al
2
O
and Ni=Al
2
O
.
3
3
3
ðSÞ
temperatures of the peak related to NiO in close contact with the
support. It seems that the addition of Cu reduces the interaction
between NiO and the support. Hence, the bimetallic system had
two reduction peaks at 580 and 660 K related to the reduction of
the reduced fresh Ni—Cu=Al O catalyst, the same three Cu oxida-
2
3
tion peaks are observed, but shifted to lower temperatures. As ob-
served in H2 chemisorption analysis, and also in the XPS analysis
discussed below, it seems that the presence of Ni improves copper
2
+
0
CuO and Cu fixed to the support, and a broad peak in the range
dispersion, and Cu is therefore oxidized at lower temperatures.

I. Gandarias et al. / Journal of Catalysis 282 (2011) 237–247
241
Cu/Al O (S); Reduced fresh
Cu/Al O (S)
2 3
2
3
1
1
1
08
06
04
1
1
1
1
1
08
06
04
02
00
4
80 K
0
0
0
.04
.02
.00
643 K
7
30 K
102
00
1
-
-
0.02
0.04
9
8
9
8
6
4
Reduced; used under H pressure
Reduced; used under N2 pressure+ 2-PO
Reduced; fresh
2
96
9
9
9
4
-0.06
400
600
800
1000
1200
4
00
600
800
1000
1200
Temperature [K]
Temperature [K]
Ni/Al O (S); Reduced fresh
Ni/Al O (S)
2
3
2
3
1
1
1
04
02
00
1
04
02
Reduced; used under H pressure
Reduced; used under N pressure + 2-PO
Reduced; fresh
2
0
0
0
.04
.02
.00
1
2
100
9
8
6
4
2
0
8
9
9
9
9
9
8
8
6
4
2
0
8
9
9
9
9
8
-0.02
-0.04
-0.06
4
00
600
800
1000
1200
400
600
800
1000
1200
Temperature [K]
Temperature [K]
Ni-Cu/Al O (S)
Ni-Cu/Al O (S); Reduced fresh
2
3
2
3
1
08
1
1
1
1
1
08
06
04
02
00
4
70 K
550 K
0
.04
106
6
30 K
0.02
1
1
04
02
0.00
-
-
0.02
0.04
100
9
8
6
4
Reduced; used under H pressure
2
9
9
8
6
Reduced; used under N pressure + 2-PO
2
9
9
Reduced; fresh
-0.06
4
00
600
800
1000
1200
4
00
600
800
1000
1200
Temperature [K]
Temperature [K]
2 2
Fig. 3. TGA corresponding to the TPO of reduced fresh samples (left-hand column). Weight change profile of fresh samples and samples used under H or N pressure with 2-
PO (right-hand column).
As the thermogravimetric equipment used was not coupled to
mass spectrometry, it was not possible to quantify the exact
amount of coke formed in the spent samples. Nevertheless, it is
possible to carry out a qualitative discussion comparing the varia-
tion for Cu-based catalysts was independent of the reaction atmo-
ðSÞ
sphere. In the case of Ni=Al
2
O
, coke formation was slightly
3
enhanced in the activity test under N
pared to the test under H pressure.
2
pressure and 2-PO as com-
2
tion in sample weight between fresh and spent samples after TPO
ðSÞ
(
see Fig. 3, right-hand column). For both Cu=Al
2
O
and
3.3.4. Temperature-programmed desorption of ammonia
3
ðSÞ
Ni—Cu=Al
2
O
spent samples, similar weight variations were mea-
The total acidity of the fresh samples, which was calculated as
3
sured regardless of whether they were used under H
2
pressure or
the integrated area of the NH
3
desorption profile, is shown in Ta-
N
2
pressure and 2-PO. It can therefore be assumed that coke forma-
ble 2. There is a decrease in the total acidity of the catalyst with

2
42
I. Gandarias et al. / Journal of Catalysis 282 (2011) 237–247
higher metal loading. As is to be expected, metallic components oc-
cupy the acid sites [30].
Table 3
Bulk theoretical ratios and surface atomic ratios obtained by XPS in reduced fresh and
reduced spent samples.
Sample
O/Ala C/Al Cu/Alb Ni/Alb
3
.3.5. X-ray photoelectron spectra
ðSÞ
ðSÞ
ðSÞ
ðSÞ
Bulk theoretical ratios
Reduced fresh
1.50
–
0.44
–
–
–
–
For each Cu=Al
2
O
; Ni=Al
2
O
and Ni—Cu=Al
2
O
catalyst,
2
Cu=Al O
3
3
3
3
1.09 0.19 0.08
three samples were analysed by XPS: (i) reduced at 723 K, (ii) re-
duced at 723 K and used under H atmosphere and (iii) reduced
at 723 K and used under N atmosphere with 2-PO as the hydrogen
donor molecule. Atomic surface ratios were computed from XPS
peak intensities and the results are presented in Table 3. These re-
sults are compared with the bulk theoretical ratios, obtained from
ICP values for Cu and Ni content and stoichiometrically for O and
Al. As can be observed, Ni/Al surface ratios in reduced fresh sam-
ples are close to theoretical bulk Ni/Al ratios, while Cu/Al surface
Used under N
Used under H
2
pressure + 2-PO 2.15 0.23 0.01
pressure 2.20 0.39 0.01
2
2
2
ðSÞ
Bulk theoretical ratios
Reduced fresh
1.50
1.76 0.19
Used under N2 pressure + 2-PO 1.98 0.34
Used under H pressure 2.24 0.35
–
–
–
–
–
0.07
0.07
0.01
0.003
2
Ni=Al O
3
2
ðSÞ Bulk theoretical ratios
1.50
–
0.42
0.10
Ni—Cu=Al
2
O
3
Reduced fresh
1.13 0.22 0.12
pressure + 2-PO 1.13 0.25 0.02
pressure 2.28 0.18 0.01
0.06
0.006
0.001
Used under N
Used under H
2
2
ratios in reduced samples are smaller than theoretical bulk ratios,
with the ratio being higher in the Ni—Cu=Al
Cu=Al
ðSÞ
a
Obtained stoichiometrically.
DataobtainedfromICPvaluesforCuandNicontentandstoichiometricallyforAl.
2
O
3
sample than in the
b
ðSÞ
2
O
3
sample. These results are in agreement with the metal
dispersion data presented above (Ni > Ni–Cu > Cu). Most of the
well-dispersed Ni atoms were accessible for X-ray radiation. On
ðSÞ
3.3.6. Activity test results and mechanistic discussion
Samples of the three catalysts were tested at 493 K reaction
2 2
temperature under H pressure and under N pressure adding 2-
the other hand, Cu=Al
2
O
catalyst had low Cu dispersion and
3
therefore comparatively big Cu particles. Those Cu atoms far from
the surface of the big particles were not detected in the XPS mea-
PO as a hydrogen donor molecule, with the 2-PO/glycerol molar ra-
tio being 1 or 1.5. All the samples were pre-treated prior to the test,
reducing them under a 50 vol.% H /N flow for 4 h at 593 or 723 K.
2 2
Fig. 4 shows the glycerol conversion values recorded in each activ-
ity test. Major increments in glycerol conversion were observed for
all three catalysts and the two reaction atmospheres when the cat-
alyst samples were reduced at 723 K. This fact indicates that, under
the reaction conditions investigated, the active sites for the glyc-
erol hydrogenolysis process are reduced sites. For the samples acti-
surement, leading to a low Cu/Al ratio. The higher Cu/Al ratio mea-
ðSÞ
sured for Ni—Cu=Al
2
O
catalyst indicates that the addition of Ni
3
enhanced Cu dispersion.
Similar C/Al ratios were measured for the samples run under H
or N pressure. It therefore seems that coke formation was unaf-
2
2
fected by the gaseous-phase composition under these experimen-
tal conditions, as was also observed in the TPO analysis of spent
samples. Concerning metal/Al ratios, the Cu/Al and Ni/Al surface
ratios for all three catalysts recorded a significant decrease in the
vated at 723 K, glycerol conversion values recorded under H
atmosphere follow the order NiCu > Ni > Cu – the same trend as
the measured MSA of the catalysts (see Table 1) – while under
2
2 2
samples used in reaction under H or N pressure. Due to the mild
operating conditions (493 K), it does not seem plausible that metal
particles underwent significant sinterization, and hence the de-
crease in the metal/Al surface ratio is probably due to coke
deposits.
The chemical species found on the catalyst surfaces and their
proportions were also evaluated. For spent catalysts, very low peak
intensities were measured for Cu and Ni, with a high noise to peak
ratio, as the metal particles were covered by coke. It was not there-
fore possible to make suitable deconvolutions. For this reason, only
2
N atmosphere and 2-PO, the order is NiCu > Cu > Ni. The different
trend in glycerol conversion regarding the hydrogen source seems
to indicate that different mechanisms and/or different active sites
might play a role regarding whether glycerol hydrogenolysis is
2 2
conducted under H pressure or under N pressure adding 2-PO
as hydrogen donor. Focusing on the effect of the amount of added
hydrogen donor, when the amount of 2-PO was increased by 50%,
the conversion of glycerol increased only slightly.
the binding energies of core electrons of freshly reduced samples
ðSÞ
It is interesting to also compare the results obtained with
are reported in Table 4. The Cu 2p3/2 level of Cu=Al
2
O
and
3
ðSÞ
ðSÞ
Ni—Cu=Al
2
O
reduced at 593 K (see Table 5) and the results ob-
3
Ni—Cu=Al
2
O
reduced samples had two components, a main peak
at 931.7–932.2 and another one at 933.9–934.2 eV, related to Cu
and Cu , respectively [31,32]. Therefore, almost all the surface
CuO was reduced after catalyst activation. Concerning Ni 2p3/2
3
ðAÞ
0
tained with Ni—Cu=Al
2
O
3
also reduced at 593 K (see Fig. 1). It
2
+
can be observed that a higher glycerol conversion was obtained
ðSÞ
with Ni—Cu=Al
2
O
2 2
(31.0% under H pressure and 41.2% under N
3
ðSÞ
ðSÞ
pressure plus 2-PO) as compared to the ones obtained with
in Ni=Al
2
O
and Ni—Cu=Al
2
O
reduced samples, three peaks were
3
3
ðAÞ
Ni—Cu=Al
2
O
2 2
(24.4% under H pressure and 27.7% under N pres-
3
obtained after the deconvolution at 852.4–852.7, 854.5 and 856.5–
0
2+
sure). Due to a confidentiality agreement, no characterization
8
56.7 eV. The first peak corresponds to Ni , the second one to Ni
ðAÞ
information regarding Ni—Cu=Al
2
O
catalyst can be provided.
3
and the peak at the highest binding energy to NiAl [33,34]. It
2 4
O
should be noted that after the activation of the samples, only 11–
ðSÞ
1
3% of the surface Ni oxide was reduced in Ni=Al
2
O
and
3
ðSÞ
Table 4
Ni—Cu=Al
2
O
samples. Taking into account the TPR spectra shown
3
Binding energies (eV) and relative abundance (%) of core electrons of Cu 2p and Ni 2p
in reduced fresh samples.
before, it is clear that under the activation conditions used (723 K),
it was not possible to reduce the NiO in close contact with the sup-
port and the NiAl O .
2 4
Sample
Cu 2p3/2 (%)
931.7 (97)
Ni 2p3/2 (%)
–
ðSÞ
2
Cu=Al O
3
Table 2
9
–
33.9 (3)
Acidity of the fresh samples from temperature-pro-
grammed desorption of ammonia.
ðSÞ
852.4 (13)
854.5 (27)
2
Ni=Al O
3
4
ꢁ1
Catalyst
Des. NH
3
ꢀ 10 (mol g
)
856.5 (60)
ðSÞ
3.64
3.39
3.95
2
Cu=Al O
3
ðSÞ
932.2 (90)
934.2 (10)
852.7 (11)
2
Ni—Cu=Al O
ðSÞ
3
2
Ni—Cu=Al O
3
854.5 (30)
856.7 (58)
ðSÞ
2
Ni=Al O
3

I. Gandarias et al. / Journal of Catalysis 282 (2011) 237–247
243
Fig. 4. Glycerol conversion values after 24-h reaction time, 45 bar pressure, 493 K, 41 mL (4 wt.%) glycerol aqueous solution and 166 mg catalyst/g glycerol. (A) 593 K
activation temperature, (B) 723 K activation temperature.
glycerol:2-PO molar ratio).
H
2
atmosphere.
N
2
atmosphere + 2-PO (1:1 glycerol:2-PO molar ratio).
2
N atmosphere + 2-PO (1:1.5
Table 5
Initial Turnover Number (TON
0
), and glycerol conversion and selectivity values obtained after 24-h reaction time as a function of the reduction temperature, the catalyst utilized
and the reacting atmosphere. 45 bar pressure, 493 K, 41 mL (4 wt.%) glycerol aqueous solution, 166 mg catalyst/g glycerol.
Sample
Activ. temp. (K)
Atm.
Hydrogen donor
Glyc:donor)
ꢁ3
ꢁ1
ꢁ1
Glyc. conv. (%)
Selectivity (%)
TONo (10 molglyc
g
h
)
catal
(
Crackeda
Restb
Acetol
1,2-PDO
ðSÞ
593
593
593
723
723
723
H2
–
0.7
1.5
32.9
14.6
0.1
90.1
40.5
9.8
6.1
0.0
5.9
2
Cu=Al O
3
N
2
2-PO (1:1)
47.5
ðSÞ
H2
–
0.5
0.1
5.2
2.0
7.2
73.7
7.5
10.5
8.6
8.4
2
Ni=Al O
3
N
2
2-PO (1:1)
60.3
22.3
ðSÞ
H2
–
1.0
4.3
31.0
41.2
0.9
84.7
48.3
6.7
5.6
7.7
4.4
Ni—Cu=Al
2
O
O
3
N
2
2-PO (1:1)
41.7
ðSÞ
H2
–
2.0
5.4
57.0
39.1
0.8
72.8
59.4
20.0
9.1
6.4
5.3
2
Cu=Al O
3
N
2
2-PO (1:1)
26.3
ðSÞ
H2
–
4.0
5.6
69.0
31.8
0.9
61.3
50.4
31.6
12.3
6.2
4.7
2
Ni=Al O
3
N
2
2-PO (1:1)
32.6
ðSÞ
H2
–
6.1
70.5
0.6
66.9
30.1
2.4
Ni—Cu=Al
2
3
N
N
2
2-PO (1:1)
2-PO (1:1.5)
7.3
3.6
57.3
60.4
27.0
17.6
62.1
64.6
8.3
12.6
2.6
5.1
2
a
2
Ethane, ethylene glycol, methane and CO .
b
1
,3-PDO and 1-Propanol.
Nevertheless, it can be addressed that both catalysts have similar
The decrease in the glycerol reaction rate over reaction time in
the activity tests under N pressure and donor could be attributed
ðSÞ
Ni content but Ni—Cu=Al
2
O
was prepared to reach a significant
2
3
higher copper content, and hence higher MSA, as compared to
to the fact that acetone formed through 2-PO dehydrogenation
competes for active sites with glycerol or that for a high acetone
ðAÞ
Ni—Cu=Al
2
O
. Bienholz et al. have recently observed a linear rela-
3
tionship between copper MSA and glycerol conversion in liquid-
2
concentration the 2-PO M acetone + H reaction could reach equi-
phase glycerol hydrogenolysis [35]. Therefore, the higher glycerol
librium and reduce the capacity of 2-PO to donor hydrogen. In or-
der to determine the effect of the presence of acetone in glycerol
ðSÞ
conversion achieved with Ni—Cu=Al
2
O
can be related to its high-
3
er MSA.
conversion, another activity test was performed under N
2
atmo-
ðSÞ
The time evolution of glycerol conversion for the activity tests
with 723 K pre-treatment is shown in Fig. 5. Higher initial reaction
rates were observed for all three catalysts in the activity tests under
sphere using Ni—Cu=Al
2
O
catalyst activated at 723 K, and with
3
2-PO, glycerol and acetone in the feed. As observed in Fig. 5D,
the initial reaction rate was not affected by the presence of acetone
ꢁ
3
ꢁ3
N
2
pressure and 2-PO compared to the activity tests under H
sure (see Fig. 5A–C and Initial Turnover Number, TON
Table 5). Nevertheless, the glycerol reaction rate under N
2
pres-
, from
atmo-
in the feed, with the TON
0
being 7.3 ꢀ 10
and 7.2 ꢀ 10
ꢁ
1
ꢁ1
0
molglyc
g
catalyst
h
for the activity test without and with acetone,
2
respectively. This means acetone has no inhibiting effect on the
glycerol reaction rate. Concerning acetone concentration
(Fig. 5D), it increased initially due to the acetone formed from 2-
PO dehydrogenation, but lower acetone concentrations were mea-
sured in the next samples, indicating that under these operating
conditions acetone reacted (being dehydrated to propene). There-
fore, above a certain acetone concentration, acetone dehydration
to propene was faster than 2-PO dehydrogenation to acetone.
sphere decreased faster over reaction time and the final glycerol
conversion values reached were higher for the tests under H atmo-
sphere. As observed in Fig. 5C and in Table 5, when the 2-PO/glyc-
2
erol ratio was increased to 1.5 glycerol, Initial TON
0
values
ꢁ3
ꢁ3
significantly decreased (from 7.3 ꢀ 10
to 3.6 ꢀ 10
ꢁ
1
ꢁ1
molglyc
g
h ). This finding seems to indicate that 2-PO and
catalyst
glycerol compete for the same active sites, and hence there must
be an optimum proportion between the donor and the acceptor to
enhance glycerol conversion.
2
Process deactivation in the activity tests under N pressure plus
2-PO could also be attributed to active site deactivation. Previously

2
44
I. Gandarias et al. / Journal of Catalysis 282 (2011) 237–247
7
6
5
0
Ni/Al O3(S)
75
60
45
30
15
0
Cu/Al O (S)
2
2
3
A
B
45
30
N +2-PO (1:1)ª
H2
N +2-PO (1:1)ª
H2
2
2
1
5
0
0
200
400
600
800
1000
1200
1400
1600
0
200
400
600
800
1000
1200
1400
1600
Time (min)
Time (min)
75
60
45
30
15
0
0.140
7
5
Ni-Cu/Al O3(S)
D
C
2
Ni-Cu/Al O3(S)
2
0
.120
.100
6
4
3
1
0
5
0
5
0
0
0.080
.060
0
N +2-PO (1:1)ª
2
H2
Glycerol Conversion
Acetone Concentration
N +2-PO(1:1)ª + acetone
0.040
0.020
0.000
b
N +2-PO (1:1.5)
2
2
0
200
400
600
800
1000 1200 1400 1600
0
200
400
600
800
1000
1200
1400
1600
Time (min)
Time (min)
Fig. 5. Time evolution of glycerol conversion for the different catalysts and reaction atmosphere. Forty-five bar pressure, 493 K, 4 wt.% glycerol aqueous solution, 723 K
activation temperature and 166 mg catalyst/g glycerol. (1:1 glycerol:2-PO molar ratio). b(1:1.5 glycerol:2-PO molar ratio).
a
reported TGA and XPS results for spent catalyst samples showed
that similar coke content and a drop in metal/Al ratios were ob-
tained regardless of whether the process was conducted under
2-PO conversion was slightly lower to the one obtained under
the same operating conditions but with glycerol as reactant (78%
and 89%, respectively). Therefore, although hydrogen species were
available they could not be transferred to acetol. Moreover, in the
H
2
or N
2
pressure plus hydrogen donor. However, in the process
pressure plus hydrogen donor the reaction rate decreased
under N
2
2
experiments under N pressure, acetol reacted to yield many dif-
faster than in the process under H
mechanisms/active sites are involved when the process is per-
formed under H or N pressure, and that the pathway for glycerol
hydrogenolysis using hydrogen from the donor is more sensitive to
coke depositions than the route for glycerol hydrogenolysis using
2
pressure. It seems that different
ferent products. Identification by GC–MS revealed the formation
of a wide range of C5–C6 compounds, such as 3-hexanol 5 methyl,
3-hexanone, 3,5-hexadien-2-ol. It is interesting to point out that
these products were not observed in the activity tests where glyc-
erol was the reactant, even though acetol was formed from glycerol
dehydration. It seems that the competitive adsorption of acetol
with glycerol and hydrogenolysis products prevents acetol from
undergoing additional reactions to form C6 compounds.
At this point, two questions need to be answered: (i) the reason
acetol is not hydrogenated when hydrogen comes from 2-PO and
(ii) the nature of the mechanism through which glycerol is con-
verted to 1,2-PDO when hydrogen is transferred from 2-PO.
It is well known that not all chemical species can be hydroge-
nated through transfer hydrogenation processes using hydrogen
donors. Huang et al. proposed that transfer hydrogenation requires
the consumption speed of hydrogen species in the metal sites to be
quicker in the form of acceptor hydrogenation than in the form of
molecular hydrogen formation and desorption [36]. When the pro-
2
2
2
hydrogen from dissolved molecular H .
Table 5 summarises the selectivity values recorded in each
activity test. Focusing first on the effect of the activation tempera-
ture, it is noted that the selectivity to cracked products increased
when the catalysts were reduced at the higher temperature. This
indicates that C–C bond cleavage is catalysed by reduced Ni and
Cu sites. An analysis of the effect of the hydrogen source in the
activity tests under H pressure revealed very low selectivities to
2
acetol. This means that the acetol formed from glycerol dehydra-
tion was quickly hydrogenated to 1,2-PDO and, therefore, that
the reaction controlling step in glycerol hydrogenolysis under H
pressure was glycerol dehydration to acetol. However, under N
2
2
atmosphere and 2-PO, much higher selectivities to acetol were
measured.
cess is conducted under N
from 2-PO can be transferred to an acceptor, or combine to produce
. When the hydrogen species are close to acetol, the formation of
molecular hydrogen occurs faster than the hydrogenation of acetol,
and the H formed readily escapes to the gas phase. In the tests un-
der H pressure, molecular hydrogen dissolved in the aqueous
2
pressure, hydrogen species formed
In order to study the acetol hydrogenation step, two activity
ðSÞ
tests were performed using Ni—Cu=Al
2
O
catalyst activated at
H
2
3
7
23 K and acetol aqueous solution as the feed; the first one under
pressure, and the second one under N pressure and 2-PO as
hydrogen source. As observed in Fig. 6, acetol reacted quickly in
both activity tests. Nevertheless, in the test under H pressure,
most of the acetol was hydrogenated to 1,2-PDO, while under N
pressure and 2-PO, very low selectivity to 1,2-PDO was recorded.
H
2
2
2
2
2
phase is in equilibrium with molecular hydrogen in the gas phase.
Due to the different formation pathways and adsorbtion/desorp-
tion rates, the behaviour of the hydrogen atoms formed from the
2

I. Gandarias et al. / Journal of Catalysis 282 (2011) 237–247
245
1
00
100
80
60
40
20
0
8
6
4
2
0
0
0
0
0
Acetol conversion (H
Acetol conversion (N
2
)
2
+ 2-PO)
Selectivity to 1,2-PDO (H
Selectivity to 1,2-PDO (N
2
)
2
+2-PO)
0
100
200
300
400
500
600
Time (min)
ðSÞ
Fig. 6. Time evolution of acetol conversion and selectivity to 1,2-PDO using Ni—Cu=Al
2
O
catalyst activated at 723 K.
3
OH
OH
OH
OH
HO
-H2
O
+
2
H2
OH
-H O
O
OH
OH
Glyceraldehyde
2
Glycerol
2-Hydroxyacrolein
1,2-propanediol
Fig. 7. Glycerol hydrogenolysis to 1,2-PDO through the glyceraldehyde pathway [37].
Table 6
dissociation of dissolved molecular hydrogen differs from the
behaviour of hydrogen atoms formed from 2-PO, and they are ac-
tive in acetol hydrogenation.
Glycerol or glyceraldehyde conversion, and yield to acetol and 1,2-PDO after 24-h
reaction time. Reduction temperature 723 K, 45 bar pressure, 493 K reaction
temperature, 41 mL aqueous solution and 0.27 g of catalyst.
Concerning the mechanism, Montassier et al. [37] suggested an-
other route for the conversion of glycerol to propylene glycol under
non-acid conditions. According to this mechanism, glycerol could
first dehydrogenate to glyceraldehyde, followed by dehydration
to 2-hydroxyacrolein and then hydrogenation to 1,2-PDO (see
Fig. 7). Moreover, this route is enhanced by the addition of a base,
as glyceraldehyde dehydration is supposed to be catalysed by ad-
sorbed OH species. In order to prove whether glycerol hydrogenol-
ysis takes place through the glyceraldehyde route when the
Catalyst
Feed (mmol)
Atmos. Initial Conv
Yield%
pH
%
Acetol 1,2-
PDO
ðSÞ Glyceraldehyde
N2
–
1.9
0.0
0.0
Ni—Cu=Al
2
O
3
(
13.3) 2-PO (17.8)
Glycerol (17.8) 2- N2
PO (17.8)
ammonia (7.3)
Glycerol (17.8)
ammonia (7.3)
11.2
26.1
1.8
19.2
(57.3) (15.4)a (35.0)a
a
H
2
11.2
17.1
0.0
13.8
(70.5) (0.4)a
a
(47.2)a
process is conducted under N
2
pressure and 2-PO, three more
a
ðSÞ
Results obtained under the same operation conditions but without ammonia
addition.
activity tests were performed using Ni—Cu=Al
2
O
catalyst. Results
3
and operating conditions are provided in Table 6, showing that
glyceraldehyde was not reactive and neither 1,2-PDO nor 2-
hydroxyacrolein was detected as a reaction product. Furthermore,
in the experiments where ammonia was added to increase the pH,
the glycerol conversion rate decreased as compared to the tests un-
der the same conditions but adding the base (Table 6, entries 2 and
dihydroxyisopropoxide can interact with an acid site and release
an OH radical from the primary carbon to produce acetol [4].
In the experiments with 2-PO as hydrogen donor, there are high
amounts of hydrides coming from 2-PO, and they attack 1,3-
dihydroxyisopropoxide converting it directly into 1,2-PDO. There-
fore, only a small proportion of 1,3-dihydroxyisopropoxide is con-
verted to acetol in the acid catalysed reaction. In this case, acetol
remains in the media, as hydrogen species formed from 2-PO can-
not hydrogenate acetol to 1,2-PDO. On the other hand, in the tests
3
), indicating that acid sites have a role to play regardless of
whether the process is conducted with molecular hydrogen or
hydrogen donor. This is clear evidence to suggest that under the
operating conditions tested glycerol hydrogenolysis did not take
place through the glyceraldehyde path.
Shinmi et al. recently proposed that glycerol hydrogenolysis to
1
,2-PDO under H
2
pressure and using Rh–ReO
x
/SiO
2
catalyst pro-
under H
2
pressure, the amount of hydrogen species adsorbed in the
in water solu-
ceeds in a single step through alkoxide adsorption [38]. We suggest
that a similar mechanism occurs for glycerol hydrogenolysis using
2
formation from 2-PO, isopropoxide, and from glycerol, 1,3-dihydr-
oxyisopropoxide. In the second step, isopropoxide either dehydro-
genates to yield acetone (Fig. 8-2A) or dehydrates to propene
interacting with an acid site (Fig. 8-2B). Propene can then be
hydrogenated to propane on metal sites [39] or escape to the gas
phase. In the final step, the hydrogen species formed from 2-PO at-
tacks the C–O bond of 1,3-dihydroxyisopropoxide to give 1,2-PDO
metal sites is lower due to the low solubility of H
2
tions and they differ from hydrogen species formed from 2-PO,
being less active for the direct hydrogenolysis of 1,3-dihydroxyiso-
propoxide. Therefore, most of the 1,3-dihydroxyisopropoxide
formed is dehydrated to acetol through the acid catalysed reaction.
Next, the hydrogen species from the dissolved molecular hydrogen
rapidly hydrogenates the acetol formed to 1,2-PDO.
It seems that direct glycerol hydrogenolysis is more sensitive to
the deactivation of the metal active sites by coke deposition as
compared to the acetol pathway, due to the fact that the hydrogen
donor and the acceptor must be adsorbed in adjacent sites in order
-PO as hydrogen donor (see Fig. 8). The first step involves alkoxide
(Fig. 8-3A). When there are no close hydrogen species, 1,3-

2
46
I. Gandarias et al. / Journal of Catalysis 282 (2011) 237–247
Fig. 8. Proposed reaction pathway for glycerol hydrogenolysis using 2-PO as a hydrogen donor molecule. M ? Metal sites;
? hydrogen species from glycerol.
A
? Acid sites;
H
? hydrogen species from 2-PO;
H
to enable transfer hydrogenation. At the beginning of the process, a
high amount of adjacent active sites are available in which the do-
nor and the acceptor can adsorb and interact, but as time on stream
elapses, coke formation reduces the amount of active metal sites,
and the interaction between the donor and the acceptor decreases,
as does the glycerol conversion rate (see Fig. 5A–C). In the process

I. Gandarias et al. / Journal of Catalysis 282 (2011) 237–247
247
under H
2
pressure, glycerol dehydration in the acid sites is the
Acknowledgments
reaction rate controlling step, as the acetol formed readily interacts
with the hydrogen species formed from molecular hydrogen, being
rapidly hydrogenated in the metal sites. Hence, although the num-
ber of active metal sites is reduced with time on stream due to coke
formation, there are still enough remaining ones for the fast hydro-
genation of the acetol. So the decrease in metal active sites by coke
deposition does not significantly affect 1,2-PDO formation through
the acetol pathway.
This work was supported by funds from the Spanish Ministry of
Science and Innovation ENE2009-12743-C04-04, and from the
Basque Government (Researcher Training Programme of the
Department of Education, Universities and Research). The authors
also gratefully acknowledge the University of the Basque Country
and the Inorganic Chemistry Department at the University of
Malaga for their technical support.
When ammonia was added to the aqueous solution, both under
N
2
and 2-PO or H
fall was more marked in the test under H
understandable, as 1,2-PDO formation under H
2
pressure, lower conversions were recorded. The
pressure, which is
atmosphere re-
References
2
[
1] M. Pagliaro, M. Rossi, The Future of Glycerol: New Usages for a Versatile Raw
material, RSC Publishing, Cambridge, 2008.
2
quires initial glycerol dehydration to acetol, where acid sites are
necessary and a basic medium can deactivate them. Nevertheless,
the decrease in glycerol conversion was also observed for the test
[
2] T. Miyazawa, Y. Kusunoki, K. Kunimori, K. Tomishige, J. Catal. 240 (2006) 213.
[3] M.A. Dasari, P. Kiatsimkul, W.R. Sutterlin, G.J. Suppes, Appl. Catal A: Gen. 281
2005) 225.
(
[
4] S. Sato, M. Akiyama, R. Takahashi, T. Hara, K. Inui, M. Yokota, Appl. Catal. A:
Gen. 347 (2008) 186.
2
under N and 2-PO, with a fall in both acetol yield and 1,2-PDO
yield. Hence, it seems that acid sites also have a role to play in
the direct glycerol hydrogenolysis process to 1,2-PDO. Studies on
CTH processes using 2-PO as hydrogen donor molecule have re-
vealed that alkoxide formation also occurs on the alumina support
[5] M. Akiyama, S. Sato, R. Takahashi, K. Inui, M. Yokota, Appl. Catal. A: Gen. 371
(2009) 60.
[
6] T. Miyazawa, S. Koso, K. Kunimori, K. Tomishige, Appl. Catal. A: Gen. 329
2007) 30.
(
[
7] T. Miyazawa, S. Koso, K. Kunimori, K. Tomishige, Appl. Catal. A: Gen. 318
(2007) 244.
[
39,40]. Therefore, the addition of ammonia meant that the iso-
[
[
8] E.P. Maris, W.C. Ketchie, M. Murayama, R.J. Davis, J. Catal. 251 (2007) 281.
9] L. Ma, D. He, Z. Li, Catal. Commun. 9 (2008) 2489.
propoxide and 1,3-dihydroxy isopropoxide formation on acid sites
decreased, as did 1,2-PDO formation.
[
10] D.G. Lahr, B.H. Shanks, J. Catal. 232 (2005) 386.
0
The proposed mechanism also explains the reason TON de-
[11] G.W. Huber, R.D. Cortright, J.A. Dumesic, Angew. Chem. Int. Ed. 43 (2004) 1549.
[
[
12] R.D. Cortright, R.R. Davda, J.A. Dumesic, Nature 418 (2002) 964.
13] E. D’Hondt, S. Van de Vyver, B.F. Sels, P.A. Jacobs, Chem. Commun. 45 (2008)
creased when a higher amount of 2-PO was added to the feed:
there is a competitive interaction for the metal sites between the
OH group both in glycerol and in the donor. Further experiments
should be conducted to obtain the optimum ratio between the do-
nor and the acceptor in which glycerol conversion to yield 1,2-PDO
is maximized.
6011.
[14] I. Gandarias, P.L. Arias, J. Requies, M.B. Güemez, J.L.G. Fierro, Appl. Catal. B:
Environ. 97 (2010) 248.
[
[
15] A. Wolfson, C. Dlugy, Y. Shotland, D. Tavor, Tetrahedron Lett. 50 (2009) 5951.
16] R.A.W. Johnstone, A.H. Wilby, I.D. Entwistle, Chem. Rev. 85 (1985) 129.
[17] M.G. Musolino, L.A. Scarpino, F. Mauriello, R. Pietropaolo, Green Chem. 11
2009) 1511.
(
[
[
18] R.A.W. Johnstone, A.H. Wilby, Tetrahedron 37 (1981) 3667.
19] M.A. Aramendía, V. Borau, C. Jiménez, J.M. Marinas, J.R. Ruiz, F.J. Urbano, J. Mol.
Catal. A: Chem. 171 (2001) 153.
4
. Conclusions
[
20] A.D. Dhale, L.K. Myrant, S.P. Chopade, J.E. Jackson, D.J. Miller, Chem. Eng. Sci. 59
2
-Propanol proved to be more effective as a hydrogen source for
(
2004) 2881.
the glycerol hydrogenolysis process under N pressure to yield 1,2-
2
[
21] G.M. Kramer, G.B. McVicker, J.J. Ziemiak, J. Catal. 92 (1985) 355.
PDO as compared to glycerol aqueous-phase reforming. The trans-
fer hydrogenation process with 2-Propanol as hydrogen donor
molecule and glycerol as acceptor was studied on Ni or/and Cu
[22] L. De Rogatis, T. Montini, A. Cognigni, L. Olivi, P. Fornasiero, Catal. Today 145
2009) 176.
23] H. Roh, K. Jun, W. Dong, J. Chang, S. Park, Y. Joe, J. Mol. Catal. A: Chem. 181
2002) 137.
[24] M. Richter, M.J.G. Fait, R. Eckelt, E. Schreier, M. Schneider, M.-. Pohl, R. Fricke,
Appl. Catal. B: Environ. 73 (2007) 269.
(
[
(
Al
tween the results of glycerol hydrogenolysis and the same operat-
ing conditions but performed under H pressure suggested that
2
O
3
catalysts prepared by the sol–gel method. A comparison be-
[
25] J. Álvarez-Rodríguez, M. Cerro-Alarcón, A. Guerrero-Ruiz, I. Rodríguez-Ramos,
A. Arcoya, Appl. Catal. B: Gen. 348 (2008) 241.
2
different mechanisms are involved regarding the origin of the ac-
tive hydrogen species. When hydrogen species come from molecu-
lar hydrogen, 1,3-dihydroxyisopropoxide is first dehydrated to
acetol, which is subsequently hydrogenated to yield 1,2-propane-
diol. When hydrogen species come from 2-propanol dehydrogena-
tion, 1,3-dihydroxyisopropoxide is directly converted into
[26] J. Lee, E. Lee, O. Joo, K. Jung, Appl. Catal. A: Gen. 269 (2004) 1.
[27] M. Korac, Z. Andjic, M. Tasic, Z. Kamberovic, J. Serb. Chem. Soc. 72 (2007) 1115.
[
[
28] J.-L. Li, T. Inui, Appl. Catal. A: Gen. 137 (1997) 105.
29] J. Agrell, H. Birgersson, M. Boutonnet, I. Melián-Cabrera, R.M. Navarro, J.L.G.
Fierro, J. Catal. 219 (2003) 389.
[30] X. Wang, X. Pan, R. Lin, S. Kou, W. Zou, J.-X. Ma, Int. J. Hydr. Energy 35 (2010)
060.
4
[
[
31] I. Melián-Cabrera, M.L. Granados, J.L.G. Fierro, J. Catal. 210 (2002) 285.
32] L. Kundakovic, M. Flytzani-Stephanopoulos, Appl. Catal. A: Gen. 171 (1998) 13.
1
,2-propanediol and the acetol formed from 1,3-dihydroxyisoprop-
oxide dehydration on acid sites does not react any further and
remains in the media.
[33] A. Velon, I. Olefjord, Oxid. Metals 56 (2001) 415.
[
[
[
34] A.A. Lemonidou, M.A. Goula, I.A. Vasalos, Catal. Today 46 (1998) 175.
35] A. Bienholz, H. Hofmann, P. Claus, Appl. Catal. A: Gen. 391 (2011) 153.
36] L. Huang, Y. Zhu, C. Huo, H. Zheng, G. Feng, C. Zhang, Y. Li, J. Mol. Catal. A:
Chem. 288 (2008) 109.
The glycerol reaction rate using 2-propanol as hydrogen source
decreased faster with time on stream as compared to the process
using molecular hydrogen. As adjacent sites are required for the
donor and the acceptor for transfer hydrogenation, glycerol
hydrogenolysis with 2-propanol is more sensitive to active site
deactivation by coke deposition than glycerol hydrogenolysis with
molecular hydrogen.
[
[
37] C. Montassier, D. Giraud, J. Barbier, Heter. Catal. Fine Chem. (1988) 165.
38] Y. Shinmi, S. Koso, T. Kubota, Y. Nakagawa, K. Tomishige, Appl. Catal. B:
Environ. 94 (2010) 318.
[39] M.I. Zaki, M.A. Hasan, L. Pasupulety, Langmuir 17 (2001) 4034.
[
40] J.A. Wang, X. Bokhimi, O. Novaro, T. López, F. Tzompantzi, R. Gómez, J.
Navarrete, M.E. Llanos, E. López-Salinas, J. Mol. Catal. A: Chem. 137 (1999) 239.