Study on the conversion of glycerol to nitriles over a Fe 19.2K0.2/γ-Al2O3 catalyst

Article abstract of DOI:10.1016/j.jcat.2014.02.014

An Fe_19.2K_0.2/γ-Al₂O₃ catalyst for the catalytic amination of glycerol to propionitrile was prepared. Acetonitrile as a major product was obtained over this catalyst from the amination of glycerol. Additionally, propionitrile, ethylene and propylene were also obtained. The parameters influencing the catalyst performance were studied thoroughly, and an optimised process for the amination of glycerol to acetonitrile and propionitrile over the catalyst was obtained. Under the optimised conditions, which were a reaction temperature of 525 °C, an atmospheric pressure with an ammonia/glycerol molar ratio of 8:1 and GHSV of 1338 h^-1, the total yield of acetonitrile and propionitrile was 58.4%, and the converted amount of glycerol over one gram of catalyst reached 0.42 g h^-1. The catalyst was characterised by XRD, XPS, TEM and IR of the adsorbed pyridine. The characterisation results indicated that the dehydration reaction in the tandem reaction mainly occurred on the Lewis acid sites and revealed that both Fe₂O₃ and Fe ₃O₄ are active species for the dehydrogenation of imines to nitriles, but the former is more active than the latter. It also revealed that the catalyst deactivation was due to carbon deposits, the transformation of Fe₂O₃ to the Fe₃O₄ phase, as well as agglomeration of the Fe₂O₃ or Fe₃O ₄ phase during the catalytic run and regeneration process.

Full text of DOI:10.1016/j.jcat.2014.02.014

Journal of Catalysis 313 (2014) 92–103
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Study on the conversion of glycerol to nitriles over a Fe19.2K0.2/
catalyst
c-Al2O3
⇑
Yuecheng Zhang, Tianqi Ma, Jiquan Zhao
School of Chemical Engineering and Technology, Hebei University of Technology, Tianjin 300130, PR China
a r t i c l e i n f o
a b s t r a c t
Article history:
An Fe19.2K0.2/c-Al2O3 catalyst for the catalytic amination of glycerol to propionitrile was prepared. Aceto-
Received 6 January 2014
Revised 26 February 2014
Accepted 28 February 2014
nitrile as a major product was obtained over this catalyst from the amination of glycerol. Additionally,
propionitrile, ethylene and propylene were also obtained. The parameters influencing the catalyst perfor-
mance were studied thoroughly, and an optimised process for the amination of glycerol to acetonitrile
and propionitrile over the catalyst was obtained. Under the optimised conditions, which were a reaction
temperature of 525 °C, an atmospheric pressure with an ammonia/glycerol molar ratio of 8:1 and GHSV
of 1338 hꢀ1, the total yield of acetonitrile and propionitrile was 58.4%, and the converted amount of glyc-
erol over one gram of catalyst reached 0.42 g hꢀ1. The catalyst was characterised by XRD, XPS, TEM and IR
of the adsorbed pyridine. The characterisation results indicated that the dehydration reaction in the tan-
dem reaction mainly occurred on the Lewis acid sites and revealed that both Fe2O3 and Fe3O4 are active
species for the dehydrogenation of imines to nitriles, but the former is more active than the latter. It also
revealed that the catalyst deactivation was due to carbon deposits, the transformation of Fe2O3 to the
Fe3O4 phase, as well as agglomeration of the Fe2O3 or Fe3O4 phase during the catalytic run and regener-
ation process.
Keywords:
Glycerol
Amination
Acetonitrile
Propionitrile
Fe19.2K0.2/c-Al2O3
Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction
and a glycerol conversion of 82.6% by an ammoxidation process
over a VSbNb/Al catalyst. Three elementary reactions are involved
Due to the continuing decline in fossil fuel resources and
increasing concerns over greenhouse gas emissions and energy
security, there has been a resurgence of interest in renewable
bio-energy. Currently, because of its environmental benefits and
little impact on engine performance, biodiesel has become one of
the best alternatives to reduce the world’s dependence on fossil
fuel. Biodiesel is produced by the transesterification of vegetable
oils, where approximately 100 kg glycerol is obtained as the major
by-product for every 900 kg of biodiesel produced [1,2]. With the
rapidly growing use of biodiesel, glycerol has become available in
excess. Attempts at converting low-cost glycerol to commercially
valued products, including 1,2-propanediol, 1,3-propanediol, acro-
lein, lactic acid, aromatics, acrylonitrile, and ethylene glycol, have
been made [3]. Among all of the transformations, the conversion
of glycerol to acrolein is the most studied [4]. The double dehydra-
tion of glycerol to acrolein can be achieved using a great variety of
acid catalysts [5–22]. In 2008, Guerrero-Pérez et al. [23] found that
glycerol can be converted to acrylonitrile with a selectivity of 58.3%
in the transformation (Scheme 1). First, glycerol is doubly dehy-
drated to acrolein. Then, acrolein condenses with ammonia to gen-
erate the intermediate imine. Finally, the imine is converted to
acrylonitrile by oxidative dehydrogenation.
Additionally, the synthesis of allyl alcohol from glycerol has
been reported very recently [24–26] and revealed that the conver-
sion of glycerol to allyl alcohol is through dehydration and consec-
utive hydrogen transfer catalysed by transition metal oxides, such
as iron oxide and methyltrioxorhenium.
Nitriles, such as acetonitrile, can be synthesised from alcohols
by two approaches: amination–dehydrogenation [27,28] and
ammoxidation [29,30]. In recent years, we have successfully syn-
thesised C2 and C4 nitriles through the amination–dehydrogena-
tion of ethanol [28,31] and butanol [32] over a CoNi/
catalyst; we have also succeeded in the synthesis of phenylaceto-
nitrile [33] and propionitrile [34] over a ZnCr/ -Al2O3 catalyst
c-Al2O3
c
through the amination–dehydrogenation of styrene oxide and allyl
alcohol, respectively. The pathways for the transformation of
allyl alcohol to propionitrile are shown in Scheme 2 [34]. First, allyl
alcohol is dehydrogenated to acrolein. After the formation of
acrolein, there are two possible pathways that yield propionitrile.
One is the condensation of acrolein with ammonia to give the
⇑
Corresponding author. Fax: +86 22 60202926.
E-mail address: zhaojq@hebut.edu.cn (J. Zhao).
http://dx.doi.org/10.1016/j.jcat.2014.02.014
0021-9517/Ó 2014 Elsevier Inc. All rights reserved.

Y. Zhang et al. / Journal of Catalysis 313 (2014) 92–103
93
OH
-2H2O
NH3
-H2O
1/2O2
-H2O
O
NH
N
HO
OH
Scheme 1. The pathway for the formation of acrylonitrile from the ammoxidation of glycerol.
NH3
- H2O
O
NH
N
- H2
H2
H2
- H2
N
O
OH
NH3
- H2O
- H2
NH
Scheme 2. Pathways of allyl alcohol to propionitrile through amination–dehydrogenation.
intermediate imine, which is then dehydrogenated to generate
acrylonitrile and hydrogen: the carbon–carbon double bond in
the acrylonitrile molecule is hydrogenated with the hydrogen
generated from the dehydrogenation of the imine and alcohol to
produce propionitrile. The other pathway involves the hydrogena-
tion of acrolein with the generated hydrogen to give propionalde-
hyde. Then, the propionaldehyde condenses with ammonia and is
dehydrogenated to give propionitrile.
Scheme 2 shows that acrolein is a key intermediate in the prep-
aration of propionitrile from allyl alcohol. As referred to above, the
conversion of glycerol to acrolein has been achieved using various
catalysts [5–22]. It is expected that the transformation of glycerol
to propionitrile is possible if we integrate the double dehydration
of glycerol to acrolein (or through allyl alcohol) and the amination
of acrolein to propionitrile reaction over a multifunctional catalyst.
Herein, we report the selection, characterisation, and catalytic test-
ing of the catalysts to pursue the above goal.
spectroscope using Al K
C1s peak at 284.6 eV was used as a reference for the charge
correction. By fitting Gaussian–Lorentzian (GL) function,
deconvolution of the Fe 2p3/2 peak of the catalyst was performed.
The GL-function was fitted to the experiments. The Fe2+ and Fe3+
peaks were fixed at 709.27 and 710.88 eV, respectively. The
Gaussian–Lorentzian ratio was fixed at 90, meaning 90% Gaussian
and 10% Lorentzian. The full width at half maximum (FWHM) was
set to 2.2 eV.
Transmission electron micrographs (TEM) were obtained on a
JEOL 100CX-II instrument equipped with an energy dispersive X-
ray (EDX) detector (Oxford Instruments) at an accelerating voltage
of 200 kV. Specimens for TEM analysis were prepared by ultrasonic
dispersion in ethanol where a drop of the resultant suspension was
evaporated on a lacey-carbon/Cu grid. The surface area, total pore
volume and pore size distribution of the catalysts were measured
at 77 K by nitrogen adsorption using a Micromeritics ASAP 2020
Surface Area and Porosity Analyser.
a X-ray source for excitation. The carbon
a
a
The IR spectra of adsorbed pyridine were recorded using a Ther-
mo Nicliet Nicolet Nexus 470 spectrometer equipped with a heat-
able and evacuatable IR cell with CaF2 windows, connected to a gas
dosing–evacuating system. The powdered samples were pressed
into self-supporting wafers with a diameter of 20 mm and a weight
of 50 mg. Prior to analysis, all samples were pretreated at 400 °C
for 1 h, under high vacuum conditions (5 ꢂ 10ꢀ5 Pa), followed by
cooling to 200 °C. Then, pyridine was adsorbed at this temperature
for 15 min. The physisorbed pyridine was removed by evacuating
during 1 h at 200 °C, under high vacuum conditions
(5 ꢂ 10ꢀ5 Pa). Then the infrared spectra were recorded.
For determining the metal content, the sample was first dis-
solved in aqua regia (HNO3–HCl) and HF, and then the metal con-
tent of the solution was analysed by a T.J.A. ICP-9000(N + M) type
ICP-AES instrument.
2. Experimental section
2.1. Catalyst preparation
The metal nitrates Fe(NO3)3ꢁ9H2O (98.5%), Cu(NO3)2ꢁ3H2O
(99%), Co(NO3)2ꢁ6H2O (99%), Zn(NO3)2ꢁ6H2O (99%), Cr(NO3)3ꢁ9H2O
(99%), Ni(NO3)2ꢁ6H2O (98%), KNO3 (99%), NaNO3 (99%), Mg(NO3)2
ꢁ6H2O (99%), Ca(NO3)2ꢁ4H2O (99%), Sr(NO3)2 (99.5%) and other re-
agents used for preparation of the catalysts were obtained from
KRS Chemical Reagent Cooperation, Tianjin, China. The catalyst
support
c-Al2O3 (99.5%) was provided by the Tianjin Research
and Design Institute of Chemical Industry, Tianjin, China. TiO2
(99%) and ZrO2 (99%) were obtained from Tianjin Chemical Reagent
Factory, Tianjin, China. The catalysts with the
were prepared by kneading a mixture of -Al2O3 with an aqueous
solution of the corresponding transition-metal nitrates, followed
by extruding, drying and calcination. For example, the catalyst
Fe19.2K0.2/c-Al2O3 was prepared as follows: in a beaker placed in
a water bath at 100 °C, 28.9 g of Fe(NO3)3ꢁ9H2O and 0.1 g of
KNO3 were added, and a small amount of distilled water (less than
2 mL) was added to dissolve the nitrates. To the solution was
c-Al2O3 as support
c
Thermogravimetric-different scanning calorimetry (TG-DSC)
measurement was carried out on a Perkin Elmer-7 thermogravi-
metric analyser from 20 to 790 °C with the rate of 10 °C minꢀ1
under air atmosphere.
2.3. Catalyst test
added 16.0 g c-Al2O3. The mixture was kneaded for 3 h in a knea-
Catalytic tests were conducted in a continuous fixed-bed reac-
tor. Specifically, 15.0 mL of the catalyst sample was loaded into a
reactor (i.d. = 15 mm; length = 1000 mm), which created a catalyst
zone of approximately 85 mm in the middle part of the reactor. The
temperature in the catalyst zone was kept constant and was mea-
sured using a thermocouple located in the centre of the catalyst
bed. An aqueous solution of 20 wt% glycerol was prepared from
99.3 wt% glycerol obtained from KRS Chemical Reagent Coopera-
tion, Tianjin, China. In the catalytic run, the aqueous solution of
20 wt% glycerol was pumped into the reactor by a syringe pump,
and the flux of ammonia (99%) was regulated by a PID cascade con-
troller (STP 220 mL/min). The liquid products were separated with
der and the resulting kneaded material was processed in an extru-
der to obtain extrudates with a diameter of 2 mm and a length of
2.5 mm. The catalyst precursors were dried at 100 °C for 6 h and
calcined at 550 °C for 6 h to obtain the catalyst.
2.2. Catalyst characterisation
The X-ray diffraction (XRD) patterns of the samples were re-
corded with a Rigaku D/max 2500 X-ray diffractometer using Cu
Ka
radiation (40 kV, 150 mA) in the range 2h = 10–90°. X-ray pho-
toelectron spectroscopy (XPS) was performed with a PHI 1600

94
Y. Zhang et al. / Journal of Catalysis 313 (2014) 92–103
a gas–liquid separator after collection in a condenser, and the gas-
eous products generated at certain periods of time were collected
after passing through an absorber with water to remove the en-
trained low boiling point liquid products for analysis. The products
were identified by gas chromatography–mass spectrometry
(HP5971 GC–MS) with a 30 m SE-30 capillary column.
propionitrile. Theoretically, the reaction should proceed as shown
in Scheme 3. Therefore, the catalyst should have the functions of
the double dehydration of glycerol to acrolein, dehydrogenation
of imines and hydrogenation of carbon–carbon double bonds with
the hydrogenation generated in situ from the above
dehydrogenation.
The products were analysed by a gas chromatograph equipped
with a 30-metre DB-5 capillary column. The content of each liquid
product was calculated using calibration curves with n-butanol as
an internal standard. The GC temperature program was 70 °C for
3 min and 40 °C/min up to 100 °C. To analyse accurately, the content
of the glycerol was analysed with cyclohexanol as an internal stan-
dard. The GC temperature program was as follows: 100 °C for 2 min,
40 °C/min up to 200 °C. The contents of the gaseous products, except
carbon dioxide, were determined by the external standard method.
The content of CO2 in the reaction solution (present in the form
of ammonium carbonate) was determined by a titration method
(ISO 3422-1975 (E)): 5.00 g (weighed to the nearest 0.001 g) of
the reaction solution was treated with a carbonate-free sodium
hydroxide solution (400 g/L) and then boiled for 15 min to elimi-
nate ammonia. After cooling to room temperature, 25 mL of a bar-
ium chloride solution (100 g/L) and 0.2 mL of a thymolphthalein
solution (1 g/L) were added to obtain sedimentation of CO2. This
solution was neutralised with a standard volumetric hydrochloric
acid solution (0.5 N) until the solution was decolourised. Then,
0.2 mL of methyl red solution and an excess standard volumetric
hydrochloric acid solution (0.5 N) of V1 mL were added to the solu-
tion and boiled for 5 min to eliminate carbonic acid. The cooled
solution was back-titrated using the standard volumetric sodium
hydroxide solution (0.5 N) until the yellow end-point of the indica-
tor was reached. The volume of the standard volumetric sodium
hydroxide solution consumed was V2 mL. The content of CO2 in
the reaction solution was then calculated using the following equa-
tion from V1 and V2.
c-Al2O3 [35], TiO2 [8] and ZrO2 [7] can catalyse the double dehy-
dration of glycerol to acrolein, and some transition metals, such as
cobalt, zinc, iron and copper, can dehydrogenate imines to nitriles
in the amination of alcohols and can promote the hydrogenation of
carbon–carbon double bonds [31,34,36,37]. Therefore,
c-Al2O3,
TiO2, and ZrO2 were chosen as supports, and several transition
metals, including chromium, iron, cobalt, nickel, copper, and zinc,
were used as dehydrogenation–hydrogenation active components
to prepare the multifunctional catalysts for the transformation of
glycerol to propionitrile. Initially, several mono-metallic catalysts
containing 20% different metals on c-Al2O3 were prepared. Their
catalytic performances in the reaction of interest were evaluated,
and the results are presented in Table 1 (Table 1, entries 1–6).
The glycerol conversions over the catalysts were higher than 94%
under atmospheric ammonia pressure at 475 °C. However, the
selectivity towards propionitrile was low for all of the catalysts.
Among the catalysts, those with iron and chromium as active com-
ponents (Fe19.5/c-Al2O3 and Cr19.8/c-Al2O3) gave better results: the
selectivity towards propionitrile was higher than 11% (entries 1, 5)
in these cases. The reaction mixture was subjected to GC–MS anal-
ysis. The results indicated that acetonitrile, ethylene, and propyl-
ene were the organic products in the reaction mixture and that
acetonitrile was the main product. Additionally, carbon dioxide
was produced. Quantitative analysis of the identified products
was performed and the carbon balance was obtained for the cata-
lysts. The results indicated that the catalyst Fe19.5/c-Al2O3 was the
best. Its total selectivity towards propionitrile and acetonitrile was
higher than 51%, and the overall selectivity of the identified organ-
ics and carbon balance reached up to 66.9% and 75.0%, respectively.
The other catalysts showed low selectivity towards organic prod-
ucts and carbon balance. Due to the good performance of iron com-
pared to other transition metals, catalysts with iron supported on
ZrO2 and TiO2 were also prepared. The catalytic results revealed
c1V1 ꢀ c2V2
Moles of CO2 in reaction solution ¼
ꢂ mout
1000 ꢂ 2 ꢂ 5:00
where c1 is the actual concentration of the 0.5 N standard volumet-
ric hydrochloric acid solution; c2 is the actual concentration of the
0.5 N standard volumetric sodium hydroxide solution; mout is the
mass of reaction solution
that Fe19.5/c-Al2O3 showed the best performance of the three cata-
lysts (Table 1, entries 1, 7, 8). Therefore, the effect of iron content
on the catalysis was studied in more detail to optimise the catalytic
performance. A series of catalysts with different iron contents were
prepared and tested. The catalytic test results are given in Table 1.
Iron is necessary for the transformation of glycerol to propionitrile
and acetonitrile (Table 1, entry 9). The selectivity of both propioni-
trile and acetonitrile increased with the increase in the iron con-
tent (Table 1, entries 1, 9–12), reaching maximums of 40.7% and
11.5%, respectively, when the iron content was 20%. The selectivity
then decreased as the iron content was further increased. Addition-
ally, the selectivity of carbon dioxide increased with the iron con-
tent within the test range.
Glycerol conversion ð%Þ ¼ ½ðmoles of glycerol pumped ꢀ moles
of unconverted glycerolÞ=mole glycerol
pumpedꢃ ꢂ 100
Product selectivity ð%Þ ¼ ½moles of the product=ðmoles of
glycerol pumped ꢀ moles of unconverted
glycerolÞꢃ ꢂ 100
In the case of carbon dioxide, the selectivity is calculated by the
following equation.
Further investigation was focused on improving the perfor-
mance of the catalyst Fe19.5/c-Al2O3 by doping with alkali or alka-
Selectivity of CO2 ð%Þ ¼ ½moles of CO2=3ðmoles of glycerol pumped
ꢀ moles of unconverted glycerolÞꢃ ꢂ 100
line earth metals. The test results of the catalysts doped with
different metals are presented in Table 2. The doping of the metals
had no obvious effect on the selectivity of propionitrile but de-
creased the selectivity of acetonitrile in most cases. Meanwhile,
the selectivity of propylene was improved in many cases. In all
cases, the formation of carbon dioxide was suppressed to some de-
gree. The doping of 0.2% potassium not only increased the selectiv-
ity of acetonitrile by 1% but also increased the selectivity of
propylene by more than 3% (Table 2, entry 5). In this case, the total
selectivity towards propionitrile and acetonitrile was higher than
53%, and the overall selectivity of the identified organics reached
70.7%. Increasing or decreasing the potassium content did not re-
Carbon balance ð%Þ ¼ ½sum of moles of carbon in the identified
products=moles of carbon in converted glycerolꢃꢂ100
3. Results and discussion
3.1. Selection of catalyst
The initial goal of this work was to find an integrated catalyst
that could catalyse the reaction of glycerol with ammonia to

Y. Zhang et al. / Journal of Catalysis 313 (2014) 92–103
95
OH
-H2
-2H2O
NH3
O
NH
N
HO
OH
-H2O
+H2
+H2
-H2
N
NH
Scheme 3. The transformation of glycerol to propionitrile in principle.
Table 1
Performances of the selected catalysts on the glycerol amination reaction.
b
Entry
Catalyst
Conversion (%)
Selectivity (%)
MeCN
Yield (%)
Carbon balance (%)
EtCN
C2H4
C3H6
CO2
1
2a
3
Fe19.5
Co19.4
Ni20.1
Cu19.7
/
/
/
/
c
c
c
c
-Al2O3
-Al2O3
-Al2O3
-Al2O3
99.0
99.4
97.7
98.6
94.4
97.0
98.9
99.3
97.7
98.7
98.2
99.8
40.7
40.5
41.0
29.1
37.9
28.7
26.4
27.2
27.2
35.2
37.1
40.1
11.5
8.6
9.7
3.8
11.2
6.0
5.3
5.9
3.6
4.9
4.4
3.0
2.3
4.8
2.0
0.4
0.4
3.4
3.6
4.1
3.6
9.8
2.2
2.0
0.4
2.4
1.0
1.4
9.5
1.6
5.4
5.9
8.4
23.3
20.6
11.8
10.3
12.2
15.7
25.6
8.6
51.7
48.8
49.5
32.4
33.7
34.1
31.4
32.9
30.1
45.0
46.7
51.0
75.0
61.2
52.0
35.4
54.2
43.2
50.2
42.4
35.1
57.7
66.2
73.8
4a
5a
6a
7
Cr19.8
/
c
-Al2O3
Zn20.2/c-Al2O3
Fe19.6/ZrO2
Fe19.1/TiO2
8
9
10
11
12
c
-Al2O3
9.5
Fe4.8
Fe9.7
/
/
c
c
-Al2O3
-Al2O3
10.4
10.5
11.1
16.1
22.4
25.3
Fe29.7/c-Al2O3
Reaction conditions: catalyst dosage 15 ml, reaction temperature 475 °C, atmospheric pressure, 20 wt% glycerol aqueous solution 0.25 mL/min, NH3 70 mL/min, time on
stream 2-4 h.
a
Pyridine and 3-picoline were both found in the products of these entries.
Total yield of acetonitrile and propionitrile.
b
Table 2
Performances of catalysts doped with alkali or alkaline earth metals.
Entry
Catalyst
Conversion (%)
Selectivity (%)
MeCN
Yielda (%)
Carbon balance (%)
EtCN
C2H4
C3H6
CO2
1
2
3
4
5
6
7
8
9
K0.2
Fe19.9Li0.2
Fe19.4Na0.2
/
c
-Al2O3
-Al2O3
-Al2O3
99.6
99.3
99.6
99.1
99.6
98.8
99.5
99.6
99.9
26.4
36.0
39.4
40.3
41.7
40.9
38.2
36.6
35.5
3.6
11.4
11.7
10.2
11.7
11.6
11.6
10.8
11.6
3.2
3.3
4.8
4.7
4.4
2.8
3.7
3.1
4.6
2.2
10.6
9.0
11.4
13.0
12.5
12.2
10.3
11.0
6.0
18.9
19.1
23.2
20.4
16.5
19.9
20.7
22.0
29.9
47.1
50.9
50.1
53.1
51.8
49.5
47.3
47.0
31.6
67.1
69.2
74.8
75.8
69.6
71.6
68.4
71.3
/
c
/
c
Fe19.6K0.1
Fe19.2K0.2
Fe19.3K0.3
/
/
/
c
c
c
-Al2O3
-Al2O3
-Al2O3
Fe19.7Mg0.2
Fe19.5Ca0.2
Fe19.0Sr0.2
/
c
c
c
-Al2O3
-Al2O3
-Al2O3
/
/
Reaction conditions: catalyst dosage 15 mL, reaction temperature 475 °C, atmospheric pressure, 20 wt% glycerol aqueous solution 0.25 mL/min, NH3 70 mL/min, time on
stream 2–4 h.
a
Total yield of acetonitrile and propionitrile.
sult in further improvement of the total yield of nitriles (Table 2,
entries, 4–6).
higher than 400 °C were required to ensure the synthesis of nitriles
in high yield from the amination of alcohols or styrene oxide.
Therefore, the influence of the reaction temperature on the perfor-
Doping bare alumina with 0.2% potassium did not improve the
performance of
c-Al2O3 (Table 1, entry 9 and Table 2, entry 1),
mance of the catalyst Fe19.2K0.2/c-Al2O3 was investigated in the
which indicates that the doping of potassium had some effect on
the active species associated with Fe to improve the performance
of the catalyst.
temperature range of 425–550 °C under atmospheric ammonia
pressure, which was maintained with a molar ratio of ammonia
to glycerol of 6:1 and a GHSV of 1338 hꢀ1. The results shown in Ta-
ble 3 indicated that the conversion of glycerol was almost quanti-
tative under at all temperatures. A decreasing trend of the
selectivity towards propionitrile was observed with an increase
in temperature. However, the acetonitrile yield increased with an
increase in the temperature from 425 to 525 °C and reached a max-
imum of 47.1% at 525 °C; a total yield of 56.2% of nitriles was ob-
tained in this case (Table 3, entry 5). Above that temperature, the
yield decreased with increasing temperature. The yields of ethyl-
ene and carbon dioxide increased with an increase in temperature.
In contrast, the selectivity towards propylene decreased with an
3.2. Performance of Fe19.2K0.2/c-Al2O3 under different reaction
conditions
Initially, the reaction was run at 365 °C, which is a favourable
temperature for the formation of acrolein [17,38–40]. However,
very low selectivity towards propionitrile and acetonitrile was
achieved due to the formation of other by-products, including
methylamine, allyl alcohol and other unidentified compounds.
Our previous work [31,34] indicated that reaction temperatures

96
Y. Zhang et al. / Journal of Catalysis 313 (2014) 92–103
Table 3
Effect of the temperature on the performance of the Fe19.2K0.2
/c
-Al2O3 catalyst.
Entry
Temperature (°C)
Conversion (%)
Selectivity (%)
MeCN
Yielda (%)
Carbon balance (%)
EtCN
C2H4
C3H6
CO2
1
2
3
4
5
6
425
450
475
500
525
550
99.3
98.3
99.6
100
99.7
99.4
34.5
34.6
41.7
42.5
47.1
43.0
12.8
13.4
11.7
10.6
9.3
0.1
2.5
4.4
5.4
6.6
8.5
1.2
16.9
13.0
11.9
6.4
16.5
18.1
20.4
21.3
25.1
27.0
46.9
47.2
53.1
53.1
56.2
50.0
53.6
73.1
75.8
75.7
76.5
70.8
7.3
2.1
Reaction conditions: atmospheric pressure, ammonia/glycerol molar ratio 6:1, GHSV 1338 hꢀ1, time on stream 2–4 h.
a
Total yield of acetonitrile and propionitrile.
increase in temperature. Overall, high temperature favoured the
formation of the products derived from the pyrolysis of glycerol,
but low temperature favoured the formation of propionitrile in
the experimental temperature range.
and the selectivity of the target products. The results indicated that
the conversion of glycerol stayed close to 100% as the GHSV was
varied from 1722 to 918 hꢀ1. The selectivity of acetonitrile had a
slight increase with the decrease in GHSV from 1722 to 1146 hꢀ1
(Table 5, entries 4–7) and then decreased after further reducing
GHSV from 1146 to 918 hꢀ1. The selectivity of propionitrile
reached its maximum of 11.9% at the GHSV of 1545 hꢀ1 (Table 5,
entries 7, 8). Meanwhile, the selectivity of propylene decreased
with the decrease in GHSV. A low GHSV indicates a long residence
time of the reactants on the catalyst. Therefore, an appropriate
GHSV favoured the formation of acetonitrile from glycerol decom-
position and the tandem reactions that followed; however, if the
GHSV was too low, the intermediates derived from glycerol under-
went deep decomposition into small molecules, such as ethylene,
H2, and CO2.
The influence of the molar ratio of ammonia to glycerol on the
reaction was investigated under atmospheric ammonia pressure
and a GHSV of 1338 hꢀ1 at 525 °C. The results are presented in Ta-
ble 4. As shown in the table, the yields of both propionitrile and
acetonitrile increased with an increase in the molar ratio of ammo-
nia to glycerol from 4:1 to 8:1, and they reached their maxima of
10.5% and 47.9% at a ratio of 8:1 (Table 4, entry 3). The yields then
decreased gradually with further increases in the molar ratio of
ammonia to glycerol. The possible reason may be that appropriate
excess ammonia can promote the condensation of acrolein and
acetaldehyde with ammonia to form imines, which are the key
intermediates in the formation of nitriles. However, when the ex-
cess ammonia was too large, for example, when the molar ratio
of ammonia to glycerol was higher than 10, the ammonia could
be adsorbed on the acid centres of the catalyst, weakening the
acidity of the catalyst and impeding the condensation of aldehydes
with ammonia to form imines. Additionally, a decreasing trend in
the selectivity towards propylene was observed with the increase
in the molar ratio of ammonia to glycerol.
3.3. Lifetime and regeneration of the Fe19.2K0.2
/c-Al2O3 catalyst
The optimised conditions—catalyst Fe19.2K0.2
/c-Al2O3 15 ml
(7.1 g), molar ratio of NH3 to glycerol 8:1, atmospheric pressure,
reaction temperature of 525oC, and GHSV of 1338 hꢀ1—were deter-
mined from the investigation of the parameters. The lifetime under
the optimised conditions and the regeneration of Fe19.2K0.2/c-Al2O3
The effect of the molar ratio of ammonia to glycerol was inves-
tigated under different K content. Similar results to those of 0.2% of
K content were obtained. The detailed results are given in Supple-
mentary material.
were then investigated. The results are shown in Fig. 1. The conver-
sion of glycerol was always higher than 99%; however, the selectiv-
ity towards total nitriles decreased with the amount of time that
the catalyst was on the stream. The selectivity towards total ni-
triles dropped from 58.4% to 51.3% with the catalyst on stream
from 4 h to 16 h. The regeneration of the catalyst was performed
by calcination in air at 550 °C for 6 h, and the reaction was run
again using the regenerated catalyst. Though the amorphous sub-
stances were removed by calcination, as revealed by TEM, the per-
formance of the catalyst was not recovered. The selectivity towards
total nitriles was lower than before the regeneration of the cata-
lyst. The reasons leading to deactivation of the catalyst will be dis-
cussed in the characterisation.
GHSV is also an important parameter that is generally consid-
ered for a catalytic reaction occurring in a fixed-bed reactor. There-
fore, the influence of GHSV on the catalytic performance was
investigated under atmospheric ammonia pressure and a molar ra-
tio of ammonia to glycerol of 8:1 at 525 °C. The results are shown
in Table 5. The reaction was run first at a very large GHSV to main-
tain a low conversion of glycerol to elucidate the reaction path-
ways (Table 5, entries 1 and 2). The composition of the reaction
mixture was analysed by GC–MS, several intermediates including
methanol, acetaldehyde, ethanol, allyl alcohol, 1-propanol and N-
ethyl methylenimine were detected. The GHSV was then lowered
to an appropriate range to increase the conversion of glycerol
Under the optimised conditions, the converted amount of glyc-
erol over 1 g of catalyst was 0.42 g hꢀ1
, corresponding to
0.089 g hꢀ1 of acetonitrile and 0.027 g hꢀ1 of propionitrile received
Table 4
Effect of the molar ratio of ammonia to glycerol on the performance of the Fe19.2K0.2
/
c
-Al2O3 catalyst.
Entry
Ammonia/glycerol
Conversion (%)
Selectivity (%)
MeCN
Yielda (%)
Carbon balance (%)
EtCN
C2H4
C3H6
CO2
1
2
3
4
5
4:1
6:1
8:1
10:1
12:1
99.7
99.7
100
99.7
99.9
41.4
47.1
47.9
47.2
46.4
8.0
9.3
10.5
9.8
5.7
6.6
5.2
6.6
7.3
4.8
6.3
4.1
3.8
2.9
24.8
25.1
25.6
26.4
26.5
49.4
56.4
58.4
56.9
55.4
69.1
76.5
75.7
75.9
74.3
9.0
Reaction conditions: atmospheric pressure, reaction temperature 525 °C, GHSV 1338 hꢀ1, time on stream 2–4 h.
a
Total yield of acetonitrile and propionitrile.

Y. Zhang et al. / Journal of Catalysis 313 (2014) 92–103
97
Table 5
Effect of GHSV on the performance of the Fe19.2K0.2/c-Al2O3 catalyst.
Entry
GHSV (hꢀ1
)
Conversion (%)
Selectivity (%)
MeCN
Yielda (%)
Carbon balance (%)
EtCN
C2H4
C3H6
CO2
1
2
3
4
5
6
7
8
13,739
8017
4014
1722
1545
1338
1146
918
69.3
76.7
93.3
100
99.9
100
13.9
24.9
37.5
45.8
46.1
47.9
47.9
45.4
5.4
7.1
9.6
10.9
11.9
10.6
9.8
0.7
0.9
2.5
4.6
4.3
5.2
5.5
6.8
0.4
1.1
2.6
4.9
4.7
4.1
3.8
2.2
15.2
16.3
16.6
19.9
24.8
25.6
27.8
26.9
13.4
24.5
43.9
56.7
58.0
58.5
57.7
54.0
30.7
41.7
55.5
69.3
75.0
75.7
77.1
72. 6
99.9
99.9
8.6
Reaction conditions: atmospheric pressure, reaction temperature 525 °C, ammonia/glycerol 8:1, time on stream 2–4 h.
a
Total yield of acetonitrile and propionitrile.
Propionitrile, acetonitrile, propylene and CO2 were the major
products during catalytic conversion of 1-propanol over the
Fe19.2K0.2/c-Al2O3 catalyst (Table 6, entry 2). When allyl alcohol
100
90
80
70
60
50
was employed as reactant (Table 6, entry 4), the main products ob-
tained were also propionitrile, acetonitrile, propylene and CO2. In
the case of propionaldehyde as reactant, propionitrile, acetonitrile
and CO2 as main products were obtained, and ethylene and propyl-
ene were also detected (Table 6, entry 3). These results suggest that
1-propanol, allyl alcohol and propionaldehyde as intermediates
may be present in the catalytic conversion of glycerol.
When the mixture of formaldehyde and acetaldehyde was fed
as reactant, acetonitrile as main product was obtained, and ethyl-
ene and propylene were also received (Table 6, entry 5). High yield
of CO2 was also obtained, which is mainly derived from the decom-
position of formaldehyde over the catalyst. The propylene is de-
rived from 3-hydroxypropanal formed from aldol reaction of
formaldehyde and acetaldehyde. Generally, aldol reaction is
reversible. These results confirm that acetaldehyde and 3-hydroxy-
propanal are also intermediates in the catalytic conversion of
glycerol.
conversion of glycerol
yield of total nitriles
8
16
1
2
3
5
4
2
14
4
6
10
12
Reaction time (h)
Fig. 1. The glycerol conversion and the yield of total nitriles as functions of reaction
time.
When hydroxyacetone was employed as reactant, the main
products obtained were acetonitrile, propionaldehyde and CO2,
and ethylene and propylene were also observed (Table 6, entry 6).
Taking into account the catalytic results stated in Table 6, and
the mechanism proposed for the transformation of glycerol to
acrolein and by-products in literature [11,22,41], we were able to
propose pathways for glycerol conversion (Scheme 4).
in the first 4 h. In this case the conversion ammonia was about 12%.
Throughout the process, the converted amount of glycerol was
maintained, but the respective yields of acetonitrile and propioni-
trile dropped with time and reached approximately 0.078 g hꢀ1
and 0.021 g hꢀ1 at the period of 14–16 h of catalyst on stream.
3.4. Pathways for product generation
During the catalytic amination of glycerol, dehydration reac-
tions occurred on the acid sites, producing water and dehydrated
compounds. Dehydrogenation reactions also occurred on the cata-
lyst’s active sites of hydrogenation–dehydrogenation, producing
hydrogen and carbonyl compounds. Decarbonylation of the dike-
tones from the dehydration or dehydrogenation of hydroxyl car-
bonyl compounds generated acetaldehyde, which condensed
with ammonia to generate the intermediate ethyleneimine. Then,
ethyleneimine was dehydrogenated to give acetonitrile. Mean-
while, acetaldehyde could also be hydrogenated to ethanol, which
was further dehydrated to give ethylene. Acrolein was produced
from the double dehydration of glycerol. Once acrolein was
formed, there were two possible pathways to produce propionitrile
from acrolein. In one path, acrolein condensed with ammonia to
generate the intermediate imine. Then, the imine was dehydroge-
nated to give acrylonitrile and hydrogen. Lastly, the carbon–carbon
double bond in the acrylonitrile molecule was hydrogenated with
the hydrogen generated in situ to give propionitrile. In the other
path, acrolein was first dehydrogenated to give propylaldehyde.
Then, propylaldehyde condensed with ammonia, and dehydroge-
nation followed to give propionitrile. In the same way that ethyl-
ene was formed, propylaldehyde from the hydrogenation of
acrolein was hydrogenated to propanol, which was further dehy-
To elucidate the pathways for the catalytic conversion of glyc-
erol over the Fe19.2K0.2/c-Al2O3 catalyst, the reaction was run under
low glycerol conversion. Apart from the four major products (pro-
pionitrile, acetonitrile, propylene and ethylene), methanol, acetal-
dehyde, ethanol, allyl alcohol, 1-propanol and N-ethyl
methylenimine were also detected by GC–MS. Moreover, Hutch-
ings [22] and others [11] have disclosed that hydroxyacetone, eth-
anol, propanal, allyl alcohol are by-products in the acid-catalysed
double dehydration of glycerol to acrolein. Therefore, ethanol,
1-propanol, propionaldehyde, allyl alcohol, the mixture of formal-
dehyde and acetaldehyde, and hydroxyacetone were used as reac-
tants under the same operation conditions performed for the
catalytic conversion of glycerol over the Fe19.2K0.2/c-Al2O3 catalyst
in order to investigate the reaction pathway. The results are shown
in Table 6.
The main products obtained when ethanol was employed as
reactant (Table 6, entry 1) were acetonitrile, ethylene and CO2.
The acetonitrile is formed through the amination–dehydrogena-
tion process [28,31], and ethylene is obtained from the dehydra-
tion of ethanol. The CO2 is generated from the decomposition of
the intermediates derived from ethanol in the catalytic process.

98
Y. Zhang et al. / Journal of Catalysis 313 (2014) 92–103
Table 6
Catalytic amination results of speculated intermediates over the Fe19.2K0.2/c-Al2O3 catalyst.
Entry
Reactants
Conversion (%)
Selectivity (%)b
MeCN
EtCN
C2H4
C3H6
CO2
1
2
3
4
5
6
Ethanol
100
100
100
100
100
100
25.9
5.3
10.3
19.3
61.0
34.4
–
16.6
–
0.2
–
5.5
1.0
–
27.2
6.9
17.3
8.5
61.6
17.7
1-Propanol
13.6
37.0
23.9
–
7.5
0.2
11.8
1.2
0.3
Propionaldehydec
Allyl alcohol
Formaldehyde + Acetaldehydea
Hydroxyacetone
13.4
Reaction conditions: catalyst Fe19.2K0.2/c
-Al2O3 dosage 15 mL, reaction temperature 525 °C, atmospheric pressure, ammonia/substrate 8:1, GHSV 1338 hꢀ1, time on stream 2–
4 h, aqueous solution of 20 wt% of intermediate.
a
Mixture of aqueous solutions of formaldehyde (10 wt%) and acetaldehyde (10.8 wt%), molar ratio of formaldehyde/acetaldehyde 1:1.4.
The selectivity of acetonitrile, propionitrile, ethylene and propylene is based on acetaldehyde, and the selectivity of carbon dioxide is based on formaldehyde in entry 5.
10 wt% Aqueous solution.
b
c
OH
HO
OH
- H2O
- H2O
- H2
O
O
OH
OH
OH
O
HO
- H2O
- H2
O
- H2O
- H2
O
O
O
+ H2
+ NH3
O
- CO
O
O
NH
+ H2
+ NH3
+ H2
+ H2
+ NH3
NH
O
H
HO
oligomers
NH
N
- H2
-
- H2
H2O
- H2O
carbon deposit
N
Scheme 4. Possible formation pathways of the products during the catalytic amination of glycerol.
drated to give propylene. The unsaturated glycol or aldehyde spe-
cies from the dehydration and dehydrogenation of glycerol and the
imines from the condensation of aldehydes with ammonia under-
went oligomerisation to generate oligomers, which finally trans-
formed to carbon deposit.
Finally, steam cracking of glycerol and the intermediates may
produce CO and H2 under the reaction conditions over the catalyst
as shown in Eq. (1), and then CO was converted into CO2 and more
H2 was produced through water-gas-shift (WGS) reaction as
shown in Eq. (2) [11].
analysed by N2 adsorption–desorption. The adsorption–desorption
isotherms are shown in Fig. 2. All of the samples displayed type
IV isotherms, which is indicative of the existence of mesopores in
the catalysts. In addition, the hysteretic loops of the samples are
similar to the type H4, which is indicative of slit-like pores [42].
Table 7 presents the specific surface areas and the range of the
pore structural parameters of the Fe19.5
and regenerated Fe19.2K0.2 -Al2O3. The fresh Fe19.2K0.2
showed a slightly higher surface area compared to Fe19.5
/
c
-Al2O3 and the fresh, used
-Al2O3
-Al2O3.
/c
/c
/c
The observed slight increase in the specific surface area can be
attributed to the fact that doping with potassium facilitates the
high dispersion of the catalyst crystallites and, as a result, in-
creased the BET surface of the iron oxide-based catalyst [43,44],
which is one of the factors improving the performance of the cat-
alyst. In fact, the mean pore diameter values for the catalysts de-
creased from 7.0 to 6.2 nm upon doping with 0.2 wt% potassium.
The increase in the specific surface area as a result of doping with
potassium can also be attributed to the creation of pores produced
from the liberation of gaseous nitrogen oxides during the thermal
CxHyOz þ ðx ꢀ zÞH2O ! xCO þ ðy=2 þ x ꢀ zÞH2
CO þ H2O ! CO2 þ H2
ð1Þ
ð2Þ
3.5. Characterisation of the catalysts
The porous structure and specific surface areas of fresh Fe19.5/c-
Al2O3 and fresh, used, and regenerated Fe19.2K0.2/c-Al2O3 were

Y. Zhang et al. / Journal of Catalysis 313 (2014) 92–103
99
4.0nm
a
b
c
d
0.08
desorption
adsorption
5.2nm
4.5nm
0.06
0.04
0.02
0.00
5.2nm
d
c
b
a
0
10 20 30
0
10 20 30
0
10 20 30
0
10 20 30
Pore Diameter (nm)
0.0
0.2
0.4
0.6
0.8
1.0
Fig. 3. Pore size distribution curves as a function of pore diameter. (a) Fe19.5
Al2O3, (b) Fe19.2K0.2 -Al2O3 (fresh), (c) Fe19.2K0.2 -Al2O3 (used), (d) Fe19.2K0.2
Al2O3 (regenerated).
/
/
c
c
-
-
/c
/c
Relative Pressure ( P/P0)
Fig. 2. Adsorption–desorption isotherms of the catalyst samples. (a) Fe19.5
/
c
-Al2O3,
(b) Fe19.2K0.2 -Al2O3 (fresh), (c) Fe19.2K0.2 -Al2O3 (used), (d) Fe19.2K0.2/c-Al2O3
/
c
/
c
(regenerated).
Fig. 4. The image of the sample of fresh Fe19.5
particles with an average diameter of 4–8 nm (Fig. 4a). The particle
size displayed in the image of the fresh Fe19.2K0.2 -Al2O3 is a diam-
eter of 2–5 nm, smaller than that of fresh Fe19.5 -Al2O3. It is also
well dispersed, which is attributed to the doping of potassium into
Fe19.5 -Al2O3. The particle size displayed in the image of the
regenerated Fe19.2K0.2 -Al2O3 is a diameter of 5–10 nm, larger
than that of the fresh catalyst, indicating that agglomeration of iron
oxides occurred in the regeneration process, which is consistent
with the N2 adsorption–desorption analysis.
/c-Al2O3 shows nano-
Table 7
/c
Textural properties of catalysts.
/c
SBET (m2 gꢀ1
)
V b (cm3 gꢀ1
)
dp (nm)
a
c
Catalyst
/c
Fe19.5
/
c
-Al2O3
173
178
141
141
0.32
0.31
0.32
0.26
7.0
6.2
7.5
6.7
Fe19.2K0.2
Fe19.2K0.2
Fe19.2K0.2
/
/
/
c
c
c
-Al2O3 (fresh)
-Al2O3 (used)
-Al2O3 (regenerated)
/c
a
b
c
BET surface area.
BJH cumulative desorption pore volume.
Mean pore diameter = 4 V/SBET
Amorphous substances covering the surface of the catalyst can
be seen by comparing the image of the used Fe19.2K0.2/c-Al2O3 with
.
that of the fresh catalyst, which shows active species particles that
are not clearly displayed. No amorphous substances were observed
decomposition of KNO3 during the thermal treatment of doped
solid at 550 °C [45].
The pore size distributions of the catalyst samples are displayed
in Fig. 3. The mesopores were mostly concentrated at 5.2 nm in
in the TEM image of the regenerated Fe19.2K0.2/c-Al2O3, which indi-
cated that the carbonaceous deposits can be removed by calcina-
tion at 550 °C for 6 h. However, the catalytic performance of the
catalyst was not recovered, as shown in Fig. 1. The TEM and N2
adsorption–desorption analysis characterisation in combination
with the catalytic test results indicated that the formation of car-
bon deposits is not the only cause of the deactivation of the cata-
lyst during the catalytic run, which is different from that of the
zinc-based catalysts for the amination of allyl alcohol to propioni-
trile reported by us [34]. Agglomeration of iron oxides during the
catalytic run or in the regeneration process may be another cause
of the deactivation of the catalyst.
The EDX analysis revealed the presence of Fe, K, Al and O in the
samples of the fresh, used, and regenerated catalyst. The elemental
carbon detected in the used sample of the catalyst can be attrib-
uted to the carbon deposition in the catalytic run, and the contam-
ination of the sample in the TEM analysis.
The amount of carbonaceous deposits on the catalyst was deter-
mined by TG-DSC and the results are shown in Fig. 5. A total weight
loss of 10% mainly taking place in the temperature range of 270 °C
to 500 °C was found, which indicated that the carbonaceous depos-
its were removed by combustion. The carbonaceous deposits ac-
count for some of the poor carbon balances in the catalytic
conversion of glycerol over the catalyst.
Fe19.5
Fe19.2K0.2
imately 4.0 nm, indicating a better dispersion after the doping with
potassium. After the catalyst Fe19.2K0.2 -Al2O3 was on stream for
/c-Al2O3, whereas the pore size distributions of the fresh
/c-Al2O3 were fairly narrow, with a maximum at approx-
/c
4 h, the BET surface area decreased from 178 m2/g to 141 m2/g,
corresponding to a decrease in the total selectivity of nitriles of
the catalyst in the catalytic run, which can be ascribed to the depo-
sition of carbon and other nonvolatile materials on the pores of the
catalyst hindering the contact of reactants with active sites. How-
ever, the pore volume and the mean pore diameter of the sample of
the used catalyst increased to some degree compared to those of
the sample of the fresh catalyst, which can be attributed to the
pore size distributions changes of the catalyst before and after
use. After the catalyst was used, the pores with small diameters
were blocked due to the deposition of carbon and other non-vola-
tile materials, corresponding to the shift in the pore size distribu-
tions towards large ones. As shown in Fig. 3, the maximum of
the pore size distribution changes from 4.0 nm to 5.2 nm. After
the used catalyst was calcined at 550 °C in air for 6 h, the maxi-
mum of the pore size distributions changed from 5.2 nm to
4.5 nm. However, the surface areas of the sample of the regener-
ated catalyst did not increase, indicating the agglomeration of iron
oxides in the regeneration process, as confirmed by the TEM
analysis.
Fig. 6 shows the XRD patterns of
fresh, used, and regenerated Fe19.2K0.2/c
the catalysts, the peaks representing the support
c
-Al2O3, Fe19.5
-Al2O3. For all samples of
-Al2O3 were
found. Both the patterns of the samples of the fresh Fe19.5
Al2O3 and Fe19.2K0.2 -Al2O3 have an Fe2O3 crystalline phase. How-
ever, no Fe2O3 crystalline phase was observed in the patterns of the
samples of the used Fe19.2K0.2 -Al2O3. Instead, an Fe3O4 crystalline
/c-Al2O3 and
c
/c-
The deposition of carbon was further confirmed by TEM images.
/c
Representative TEM images of the samples of fresh Fe19.5
/c-Al2O3
and fresh, used, and regenerated Fe19.2K0.2 -Al2O3 are shown in
/c
/c

100
Y. Zhang et al. / Journal of Catalysis 313 (2014) 92–103
e
Al
O
Fe
C
C
Cu
Fe
Fe
Fe
K
3
O
Cu
Al
Al
Fe
Fe
Fe
K
K
2
1
O
C
Cu
8
Fe
Fe
0
1
2
3
4
5
6
7
9
10
Energy (keV)
Fig. 4. TEM micrographs of the catalyst samples and EDX spectra at points 1, 2, 3. (a) Fe19.5
/c-Al2O3, (b) Fe19.2K0.2/c-Al2O3 (fresh), (c) Fe19.2K0.2/c-Al2O3 (used), (d) Fe19.2K0.2/c-
Al2O3 (regenerated), (e) EDX spectra at points 1, 2 and 3. The carbon contents are 11.2%, 24.6% and 14.8% at points 1, 2 and 3, respectively.
phase was displayed. The results indicated that the Fe2O3 was
transformed to Fe3O4 during the catalytic run. The transformation
of the Fe2O3 crystalline phase to the Fe3O4 crystalline phase is due
to the reduction of Fe2O3 by the in situ-generated H2 as shown in
Eq. (3) [46].
100
98
96
94
92
90
88
86
84
1.0
Exothermic
0.5
3Fe2O3 þ H2 ! 2Fe3O4 þ H2O
ð3Þ
0.0
The patterns of the sample of the regenerated Fe19.2K0.2
/
c-Al2O3
-0.5
-1.0
-1.5
-2.0
were almost the same as those of the sample of the used catalyst,
except for a few weak peaks of Fe2O3, which indicated that only a
part of the Fe2O3 crystalline phase was recovered by calcination
at 550 °C in the air for 6 h, even though the carbon deposits were
removed by this process. These results revealed that Fe2O3 was
one of the dehydrogenation–hydrogenation active species of the
catalyst and that the deactivation of the catalyst was in part due
to the transformation of Fe2O3 to Fe3O4 by H2 reduction in the cat-
alytic run.
0
100
200
300
400
500
o
600
700
800
Temperature
(
C
)
The average diameters of the Fe2O3 crystallites in the Fe19.5
/c-
Fig. 5. TG-DSC analysis of the used Fe19.2K0.2
/
c
-Al2O3 catalyst.
Al2O3 and fresh Fe19.2K0.2 -Al2O3 were not calculated by using
/c

Y. Zhang et al. / Journal of Catalysis 313 (2014) 92–103
101
and 23.3 nm, respectively, larger than those observed from TEM
images. The difference can be attributed to the still low crystallin-
ity of iron oxide crystallites in the used and regenerated catalyst.
The surface compositions of the samples of fresh, used, and
*
*
+
#
Fe3 O4
Fe2 O3
Al2 O3
*
*
e
*
*
+
*
+
*
+
regenerated Fe19.2K0.2/c-Al2O3 were determined by XPS. The full
+
+
+
*
spectrum is presented in the Supplementary material. The result-
ing XPS Fe 2p peak was curve-fitted as shown in Fig. 7. Due to
broader, less intense peak structure in the 2p1/2 region and to er-
rors that build up in higher-binding energy regions, concentration
analysis is based on the Fe main 2p3/2 peak intensities only.
For the fresh sample, the spectrum can be successfully fitted to
two main peaks and a satellite peak at 715.2 eV. The binding en-
ergy peaks at 710.9 eV and 712.9 eV are attributed to the Fe3+ spe-
cies, respectively [47]. No peak binding energy peak attributed to
Fe2+ is appeared. These confirm further that Fe2O3 is the dehydro-
*
d
c
b
*
+
#
#
#
#
*
*
*
+
+
+
+
+
+
+
+
+
genation–hydrogenation active species in the catalyst Fe19.2K0.2
Al2O3.
/c-
#
#
#
#
For the sample of the used catalyst, the spectrum is fitted to
three main peaks and two satellite peaks. The binding energy
peaks at 710.9 eV and 712.3 eV and the satellite peak at 713.9 eV
are attributed to the Fe3+ species [47]. The lowest binding energy
peak at 709.3 eV attributed to the Fe2+, with a corresponding satel-
lite at 715.8 eV [48]. The mean relative areas of each constituent
peak assigned to Fe2+ and Fe3+ were calculated, and the atomic ra-
tio of Fe2+ to Fe3+ was 1:2.2. Because stoichiometric Fe3O4 can also
be expressed as FeOꢁFe2O3, the Fe2+:Fe3+ ratio should be 1:2. The
results indicated that Fe2O3 was almost completely reduced to
Fe3O4 by the in situ-generated H2, which is consistent with the
XRD analysis.
For the regenerated sample, the spectrum is also fitted to two
main peaks and a satellite peak at 714.4 eV. The binding energy
peaks at 710.4 eV and 712.2 eV are attributed to the Fe3+ species,
respectively [47], no binding energy peak assigned to Fe2+ is ob-
served. The results indicated that the Fe3O4 on the surface of the
a
#
#
#
20
40
60
80
°
θ
2
Fig. 6. XRD patterns of the catalyst samples. (a)
Fe19.2K0.2 -Al2O3 (fresh), (d) Fe19.2K0.2 -Al2O3 (used), (e) Fe19.2K0.2
(regenerated).
c
-Al2O3, (b) Fe19.5
/
c
-Al2O3, (c)
/c
/
c
/
c
-Al2O3
Scherrer formula due to poor crystallinity of Fe2O3 crystallites. Be-
cause of agglomeration of iron oxides during the catalytic run or in
the regeneration process as revealed by the TEM analysis, the crys-
tallinity of the iron oxide crystallites was improved, and the aver-
age diameters of iron oxide crystallites in the used and regenerated
Fe19.2K0.2
/c-Al2O3 can be calculated by using the Scherrer formula.
The average diameters of the iron oxide crystallites are 19.5 nm
Pos.
Sep. %Area
Pos.
Sep. %Area
a
710.9 0.00 60.89
712.9 1.99 29.77
715.2 4.29 9.33
709.3 0.00 23.71
712.3 3.00 24.02
710.9 1.59 37.80
713.9 4.67
715.8 6.57
b
Fe3+
6.97
7.50
Fe3+
Fe3+
Fe3+
shake-up
shake-up
shake-up
Fe2+
718
716
714
712
710
708
718
716
714
712
710
708
706
Binding Energy (eV)
Binding Energy (eV)
Pos.
Sep. %Area
c
710.4 0.00 52.71
712.2 1.81 34.52
714.4 4.02 12.77
Fe3+
Fe3+
shake -up
718
716
714
712
710
708
706
Binding Energy (eV)
Fig. 7. Fe 2p XPS spectra of the catalyst samples. (a) Fe19.2K0.2
/
c
-Al2O3 (fresh), (b) Fe19.2K0.2
/
c
-Al2O3 (used), (c) Fe19.2K0.2
/
c
-Al2O3 (regenerated).

102
Y. Zhang et al. / Journal of Catalysis 313 (2014) 92–103
Table 8
Surface and bulk elemental compositions of the samples of the Fe19.2K0.2
/c
-Al2O3 catalyst.
O (%)
Methods
XPSa
Samples
C (%)
Al (%)
Fe (%)
Fe: Alc
Fresh
45.4
59.2
48.0
39.2
29.2
30.0
12.1
8.6
17.6
3.3
2.2
4.4
0.27
0.26
0.25
Used
Regenerated
ICPb
Fresh
Regenerated
42.6
42.7
19.2
19.1
0.22
0.22
a
b
c
Atomic%.
Mass%.
Molar ratio.
The significant changing of the spectrum of the used catalyst
sample compared with that of the fresh catalyst indicated that
some substances deposited on the surface of the catalyst in the cat-
alytic run. However, the peaks at 1545 cmꢀ1 and 1450 cmꢀ1 are
still visible by careful observation of the spectrum, indicating that
acid sites are present in the catalyst. Therefore, the dehydration of
glycerol to the intermediates is still feasible. Meanwhile, the
deposited substances, as confirmed by TEM, also masked the active
species Fe2O3, blocking the dehydrogenation of imine to nitriles to
some degree, which is one of the reasons leading to a decrease in
the selectivity towards nitriles. Although the deposited substances
were removed by calcination, the performance of the catalyst was
not recovered.
From the characterisation results of XRD, XPS and IR spectra of
adsorbed pyridine, as well as the catalytic performance of the cat-
alyst shown in Fig. 1, it can be concluded that both Fe2O3 and Fe3O4
can act as dehydrogenation–hydrogenation active species in the
amination of glycerol to nitriles, but the activity of Fe3O4 is lower
than that of Fe2O3. It can also be concluded that the failure to re-
cover the performance of the catalyst is due to the agglomeration
of Fe2O3 and Fe3O4 phases in the catalytic run and the regeneration
process in view of the TEM analysis.
1450
1545
d
c
b
a
1700
1650
1600
1550
1500
1450
1400
1350
1300
Wavenumber-1
Fig. 8. IR spectra of pyridine adsorbed on the catalysts. (a) Fe19.5
Fe19.2K0.2 -Al2O3 (fresh), (c) Fe19.2K0.2 -Al2O3 (used), (d) Fe19.2K0.2
(regenerated).
/
c
-Al2O3, (b)
/
c
/
c
/c
-Al2O3
used catalyst was completely converted to Fe2O3 after calcination
at 550 °C in the air for 6 h. However, the Fe3O4 in the bulk was re-
mained, that is why only weak peaks of Fe2O3, but strong peaks of
Fe3O4 were observed in the XRD patterns of the regenerated
sample.
Table 8 presents the surface and bulk elemental compositions of
the samples of the fresh, used and regenerated Fe19.2K0.2/c-Al2O3
catalyst from the XPS and ICP analysis. It revealed that the Fe:Al
in the bulk is lower than that on the surface, indicating migration
of Fe from bulk to surface in the calcination during the preparation
process of the catalyst. It can also be seen that the molar ratio of Fe
to Al in the bulk remained constant in the catalytic run and regen-
eration process, and the value on the surface decreased slightly,
which indicated that the catalyst was stable and no further migra-
tion of Fe from bulk to surface took place in the catalytic run and
regeneration process.
4. Conclusion
An Fe19.2K0.2/c-Al2O3 catalyst was prepared and showed activity
for the amination of glycerol to propionitrile and acetonitrile. The
parameters that affected the catalyst performance were studied
thoroughly, and an optimised process for synthesising nitriles from
glycerol and ammonia over the catalyst was obtained. Under the
optimised conditions, the total yield of propionitrile and acetoni-
trile was higher than 58%. The characterisation results indicated
that the dehydrogenation reaction mainly occurred on the Lewis
acid sites and that Fe2O3 is the main active species for the dehydro-
genation of intermediate imines to nitriles and carbon–carbon
double hydrogenation. Doping with 0.2% potassium to Fe19.5/c-
Al2O3 decreased the size of the Fe2O3 crystallites, which favoured
the dehydrogenation of imines to nitriles and carbon–carbon dou-
ble hydrogenation. The characterisation results revealed that the
catalyst deactivation was mainly caused by carbon deposition on
the catalyst and the reduction of Fe2O3 to Fe3O4 during the cata-
lytic run. Meanwhile, agglomeration of the Fe2O3 and Fe3O4 phases
during the catalytic run and regeneration process was the cause of
the failed performance recovery of the catalyst by calcination.
Fig. 8 shows the IR spectra of absorbed pyridine of the Fe19.2K0.2
/
c
-Al2O3 and fresh, used and regenerated Fe19.2K0.2/c-Al2O3 catalyst
samples in the region 1700–1300 cmꢀ1. Two peaks at 1545 cmꢀ1
and 1450 cmꢀ1 are attributed to the characteristic absorptions of
pyridine adsorbed on the Brønsted and Lewis acid centres of the
catalyst respectively [49–51]. The spectra show that both Fe19.5
/
c
-Al2O3 and Fe19.2K0.2 -Al2O3 have Lewis acidity but weak
/c
Brønsted acidity. The doping of a small amount of potassium had
no obvious effect on the acidity of the catalyst. The acid centres
provide the catalyst with the ability to convert glycerol to acrolein
by dehydration. The peaks at 1545 cmꢀ1 and 1450 cmꢀ1 in the
spectrum of the used catalyst were almost destroyed and then
recovered after regeneration, as seen by comparing the spectra of
Acknowledgments
We thank the financial support of the National Natural Science
Foundation of China (Grant Nos. 20976034, 21276061), the Natural
Science Foundation for Young Scientists of Hebei Province of China
(Grant No. B2009000009) and the Foundation for Innovative Tal-
ents of Hebei Province of China (Grant No. CPRC012).
the fresh, used and regenerated samples of Fe19.2K0.2/c-Al2O3.

Y. Zhang et al. / Journal of Catalysis 313 (2014) 92–103
[25] J. Yi, S. Liu, M.M. Abu-Omar, ChemSusChem 5 (2012) 1401.
103
Appendix A. Supplementary material
[26] A. Konaka, T. Tago, T. Yoshikawa, A. Nakamura, T. Masuda, Appl. Catal. B:
Environ. 146 (2014) 267.
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2014.02.014.
[27] C.J. Roger, S.J. Lawrence, J. Org. Chem. 46 (1981) 754.
[28] Y. Zhang, Y. Zhang, C. Feng, C. Qiu, Y. Wen, J. Zhao, Catal. Commun. 10 (2009)
1454.
[29] B.M. Reddy, B. Manchar, J. Chem. Soc., Chem. Commun. 3 (1993) 234.
[30] S.J. Kulkarni, R. Ramachandra Rao, M. Subrahmanyam, A.V. Rama Rao, J. Chem.
Soc. Chem. Commun. 3 (1994) 273.
References
[31] C. Feng, Y. Zhang, Y. Zhang, Y. Wen, J. Zhao, Catal. Lett. 141 (2011) 168.
[32] C. Feng, F. Guo, Y. Zhang, J. Zhao, Finechemicals (in Chinese) 27 (2010) 567.
[33] Y. Zhang, W. Xu, J. Zhao, RSC Adv. 2 (2012) 6590.
[1] A. Yuksel, H. Koga, M. Sasaki, M. Goto, Ind. Eng. Chem. Res. 49 (2010) 1520.
[2] Y. Zheng, X. Chen, Y. Shen, Chem. Rev. 108 (2008) 5235.
[3] J.T. Dam, U. Hanefeld, ChemSusChem 4 (2011) 1017.
[34] Y. Zhang, T. Wei, Y. Pian, J. Zhao, Appl. Catal. A: Gen. 467 (2013) 154.
[35] M. Massa, A. Andersson, E. Finocchio, G. Busca, J. Catal. 307 (2013) 170.
[36] Y.B. Kryulov, A.N. Bashkiov, N.S. Zakirov, F.I. Novak, Dokl. Akad. Nauk SSSR 170
(1966) 852.
[37] Y. Hu, J. Cao, J. Deng, B. Cui, M. Tan, J. Li, H. Zhang, React. Kinet., Mechanism.
Catal. 106 (2012) 127.
[4] M. Massa, A. Andersson, E. Finocchio, G. Busca, F. Lenrick, L.R. Wallenberg, J.
Catal. 297 (2013) 93.
[5] H. Atia, U. Armbruster, A. Martin, J. Catal. 258 (2008) 71.
[6] L. Shen, Y. Feng, H. Yin, A. Wang, L. Yu, T. Jiang, Y. Shen, Z. Wu, J. Ind. Eng. Chem.
17 (2011) 484.
[7] A. Ulgen, W. Hoelderich, Catal. Lett. 131 (2009) 122.
[38] B. Katryniok, S. Paul, V. Bellière-Baca, P. Rey, F. Dumeignil, Green Chem. 12
(2010) 2079.
[39] S.-H. Chai, H.-P. Wang, Y. Liang, B.-Q. Xu, Green Chem. 10 (2008) 1087.
[40] E. Kraleva, R. Palcheva, L. Dimitrov, U. Armbruster, A. Brückner, A. Spojakina, J.
Mater. Sci. 46 (2011) 7160.
[8] A. Ulgen, W.F. Hoelderich, Appl. Catal. A: Gen. 400 (2011) 34.
[9] F. Cavani, S. Guidetti, L. Marinelli, M. Piccinini, E. Ghedini, M. Signoretto, Appl.
Catal. B: Environ. 100 (2010) 197.
[10] P. Lauriol-Garbay, J.M.M. Millet, S. Loridant, V. Bellière-Baca, P. Rey, J. Catal.
280 (2011) 68.
[41] Z.Y. Zakaria, N.A.S. Amin, J. Linnekoski, Biomass Bioenergy 55 (2013) 370.
[42] S. Lowell, J.E. Shields, M.A. Thomas, M. Thommes, Characterization of Porous
Solids and Powders: Surface Area, Pore Size and Density, Kluwer Academic
Publishers, Dordrecht/Boston/London, 2004.
[43] M. Feyzi, M. Irandoust, A.A. Mirzaei, Fuel Process. Technol. 92 (2011) 1136.
[44] S.-R. Yan, K.-W. Jun, J.-S. Hong, M.-J. Choi, K.-W. Lee, Appl. Catal. A: Gen. 194–
195 (2000) 63.
[45] G.A. Fagal, G.A. El-Shobaky, S.M. El-Khouly, Colloid. Surf. A: Physicochem. Eng.
Asp. 178 (2001) 287.
[46] S.-C. Lee, J.-H. Jang, B.-Y. Lee, M.-C. Kang, M. Kang, S.-J. Choung, Appl. Catal. A:
Gen. 253 (2003) 293.
[11] A. Corma, G.W. Huber, L. Sauvanaud, P. O’Connor, J. Catal. 257 (2008) 163.
[12] C.-J. Jia, Y. Liu, W. Schmidt, A.-H. Lu, F. Schüth, J. Catal. 269 (2010) 71.
[13] Y.T. Kim, K.-D. Jung, E.D. Park, Appl. Catal. A: Gen. 393 (2011) 275.
[14] F. Wang, J.-L. Dubois, W. Ueda, J. Catal. 268 (2009) 260.
[15] F. Wang, J.-L. Dubois, W. Ueda, Appl. Catal. A: Gen. 376 (2010) 25.
[16] W. Yan, G.J. Suppes, Ind. Eng. Chem. Res. 48 (2009) 3279.
[17] S.-H. Chai, H.-P. Wang, Y. Liang, B.-Q. Xu, J. Catal. 250 (2007) 342.
[18] N.R. Shiju, D.R. Brown, K. Wilson, G. Rothenberg, Top. Catal. 53 (2010) 1217.
[19] J. Deleplanque, J.-L. Dubois, J.F. Devaux, W. Ueda, Catal. Today 157 (2010) 351.
[20] J. Lourenço, M.I. Macedo, A. Fernandes, Catal. Commun. 19 (2012) 105.
[21] W. Suprun, M. Lutecki, H. Papp, Chem. Eng. Technol. 34 (2011) 134.
[22] M.H. Haider, N.F. Dummer, D. Zhang, P. Miedziak, T.E. Davies, S.T. Taylor, D.J.
Willock, D.W. Knight, D. Chadwick, G.J. Hutchings, J. Catal. 286 (2012) 206.
[23] M.O. Guerrero-Pérez, M.A. Bañares, ChemSusChem 1 (2008) 511.
[24] Y. Liu, H. Tuysuz, C.J. Jia, M. Schwickardi, R. Rinaldi, A.H. Lu, W. Schmidt, F.
Schuth, Chem. Commun. 46 (2010) 1238.
[47] N.S. McIntyre, D.G. Zetaruk, Anal. Chem. 49 (1977) 1521.
[48] C.R. Clayton, Y.C. Lu, J. Electrochem. Soc. 133 (1986) 2465.
[49] C.E. Volckmar, M. Bron, U. Bentrup, A. Martin, P. Claus, J. Catal. 261 (2009) 1.
[50] C.A. Emeis, J. Catal. 141 (1993) 347.
[51] E. Modrogan, M.H. Valkenberg, W.F. Hoelderich, J. Catal. 261 (2009) 177.