-
ww.sciencedirect.com
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e
r g y 4 2 ( 2 0 1 7 ) 9 0 8 7e9 0 9 8
Available online at w
ScienceDirect
journal homepage: www.elsevier .com/locate/he
Hydrogen production from catalytic steamreforming of glycerol
over various supported nickelcatalysts
Nurul Huda Zamzuri a, Ramli Mat b, Nor Aishah Saidina Amin
b,*,Amin Talebian-Kiakalaieh b
a Process Systems Engineering Centre (PROSPECT), Research
Institute on Sustainable Environment (RISE),
Universiti Teknologi Malaysia, 81310 UTM, Johor Bahru, Malaysiab
Chemical Reaction Engineering Group (CREG), N01-Faculty of Chemical
Engineering, Universiti Teknologi
Malaysia, 81310 UTM, Johor Bahru, Malaysia
a r t i c l e i n f o
Article history:
Received 16 October 2015
Received in revised form
4 May 2016
Accepted 10 May 2016
Available online 30 May 2016
Keywords:
Glycerol to hydrogen
Catalytic conversion
Steam reforming
Optimization
* Corresponding author.E-mail address: [email protected]
http://dx.doi.org/10.1016/j.ijhydene.2016.05.00360-3199/© 2016
Hydrogen Energy Publicati
a b s t r a c t
Supported Ni catalysts have been investigated for hydrogen
production from steam
reforming of glycerol. Ni loaded on Al2O3, La2O3, ZrO2, SiO2 and
MgO were prepared by the
wet-impregnation method. The catalysts were characterized by
nitrogen adsorptionede-
sorption, X-ray diffraction and scanning electron microscopy.
The characterization results
revealed that large surface area, high dispersion of active
phase on support, and small
crystalline sizes are attributes of active catalyst in steam
reforming of glycerol to hydrogen.
Also, higher basicity of catalyst can limit the carbon
deposition and enhance the catalyst
stability. Consequently, Ni/Al2O3 exhibited the highest H2
selectivity (71.8%) due to small
Al2O3 crystallites and large surface area. Response Surface
Methodology (RSM) could
accurately predict the experimental results with R-square ¼
0.868 with only 4.5% error. Thehighest H2 selectivity of 86.0% was
achieved at optimum conditions: temperature ¼ 692 �C,feed flow rate
¼ 1 ml/min, and water glycerol molar ratio (WGMR) 9.5:1. Also, the
optimi-zation results revealed WGMR imparted the greatest effect on
H2 selectivity among the
reaction parameters.
© 2016 Hydrogen Energy Publications LLC. Published by Elsevier
Ltd. All rights reserved.
Introduction
Limited amounts of fossil fuels, especially petroleum, and
concurrent environmental problems associated with green-
house gases have prompted the world to look for clean sus-
tainable resources as alternatives to meet increasing energy
y (N.A. Saidina Amin).84ons LLC. Published by Els
demands [1]. Among the renewable energy, biodiesel is widely
used and appears to be promising and feasible to reduce the
impact on CO2 emissions [2]. One of the main obstacles for
worldwide production of biodiesel is its production costs
[3].
Thus, production of value-added chemicals such as hydrogen
from glycerol which is the main by-product of biodiesel pro-
duction process significantly reduces the biodiesel
production
evier Ltd. All rights reserved.
mailto:[email protected]://crossmark.crossref.org/dialog/?doi=10.1016/j.ijhydene.2016.05.084&domain=pdfwww.sciencedirect.com/science/journal/03603199www.elsevier.com/locate/hehttp://dx.doi.org/10.1016/j.ijhydene.2016.05.084http://dx.doi.org/10.1016/j.ijhydene.2016.05.084http://dx.doi.org/10.1016/j.ijhydene.2016.05.084
-
i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e
r g y 4 2 ( 2 0 1 7 ) 9 0 8 7e9 0 9 89088
costs [4]. Hydrogen is considered as a clean fuel of future
due
to its excellent energy storage capacity (33 kW h/Kg) [5].
Also,
it can be considered as secondary energy source because it
can
be converted to energy (heat or electricity) by combustion
or
electrochemical reactions [6].
Hydrogen can be produced from thermo-catalytic decom-
position and single step anaerobic of methane [7e10]. The
most commonly used technology for hydrogen production
from glycerol is the steam reforming process, similar to the
conventional hydrocarbon steam reforming [11,12]. The
reforming reactions of glycerol for hydrogen production are
listed in Equations (1)e(5). A two-step global reaction
equation
in which the carbohydrate (glycerol) undergoes thermal
decomposition is presented in the first reaction (Eq. (1))
to
form CO and H2. The CO then reacts with steam (oxidizer) in
the water-gas shift reaction (Eq. (2)) to form CO2 and addi-
tional H2.
C3H8O3�����!H2O 3COþ 4H2 (1)
COþH2O4CO2 þH2 (2)Reactions (1) and (2) can be added to obtain
eq. (3):
C3H8O3 þ 3H2O/3CO2 þ 7H2 (3)Some H2 is lost via methanation of
carbon monoxide (CO)
and carbon dioxide (CO2) as shown in Eqs (4) and (5).
COþ 3H24CH4 þH2O (4)
CO2 þ 4H24CH4 þ 2H2O (5)Steam reforming of glycerol has been
investigated over a
wide variety of supported metal catalysts (Ru, Rh, Ni, Ir, Co,
Pt,
Pd and Fe) [13e17]. The activity, product distribution and
catalyst stability have been found to be dependent upon the
catalyst composition, support material, catalyst preparation
and pre-treatment technique and reaction conditions. Appli-
cation of noble metal based catalysts registered promising
re-
sults (high activity and hydrogen selectivity) in glycerol
steam
reforming reaction. However, the main obstacle for industri-
alization of this process is the exorbitant cost of noble
metal-
based catalysts synthesis and preparation. Thus, the
majority
of researchers have focused on the low cost materials such
as
copper, cobalt, and nickel. As a result, nickel based
catalysts
have attracted much attention due to their higher activity
and
lower costs compared to the other transition metals [17].
Nickel is known to have high capability of breaking the
CeC bonds and promoting the water-gas shift reactions and
thus increasing the hydrogen production [17,18]. For reform-
ing reactions, high-surface area catalysts are used [19].
The
major roles of the supports are to prepare and preserve
ther-
mally stable, well-dispersed catalytic phases during the re-
action. The supports are typically porous, high surface area
metal -oxides or carbon [20].
The aim of this work is to investigate the performance of
supported nickel catalysts that could deliver stable perfor-
mance for the steam reforming of glycerol. In an attempt to
achieve high hydrogen selectivity, the effects of operating
conditions and supported catalyst properties are studied.
The
catalysts are characterized with BET, XRD and SEM.
Optimization studies have been conducted by applying cen-
tral composite design (CCD) under the response surface
methodology (RSM). RSM is one of the methods to analyse the
significance or the influence of the factors on the response
[21]. This method is useful since it can reduce the number
of
experiments and consequently the cost and time consumed.
The variation of process conditions (reaction temperature
(T),
feed flow rate (FFR), water glycerol molar ratio (WGMR))
which
could affect the responses (H2 selectivity and glycerol con-
version) were evaluated within the range. Subsequently, the
H2 selectivity at the optimum conditionswas also evaluated
in
this study.
Experimental
Catalyst preparation
Calculated quantities of nickel nitrate hexahydrate
[Ni(NO3)2.6H2O] were dissolved in deionisedwater tomake the
precursor solution for a total nickel loading of 10 wt%. The
nitrate solution was added to the supports particles, Al2O3,
La2O3, ZrO2, SiO2 and MgO respectively. The solution was
stirred continuously for 3 h and then dried in the oven at
110 �C overnight before calcination for 5 h at 500 �C.
Catalyst characterization
The structures of the catalysts were determined by using X-
Ray Diffraction (XRD). XRD patterns were measured at Mak-
mal Sains Bahan, UTM Johor Bahru using standard Bragg-
Brentano geometry with Ni-filtered Cu Ka radiation
(l1 ¼ 1.54056 �A). The spectra were collected for a 2q range
of10e90� using a step size of 0.05 and a count time of 1 s.
Themorphology and elemental analysis of nickel catalysts on
different supports were observed by Scanning Electron Mi-
croscopy and Energy Dispersive X-ray (SEM-EDX). The anal-
ysis was performed in Hitachi S-4000 microscope with a cold
field emission gun and an energy dispersive energy detector.
The surface area of the catalyst samples was determined
using BrunauereEmmeteTeller (BET) method using a Micro-
meritics ASAP 2010.
Catalyst performance testing
The performance of the catalysts was tested in a continuous
flow process using quartz tube in a fixed bed reactor as
illus-
trated in Fig. 1. For each reaction test, 0.3 g of the
calcined
catalyst was diluted with equal amount of inert silica
carbide
(SiC). The catalysts were reduced at 500 �C in 100 ml/min of10%
H2/N2 for 1 h. The bed temperature was heated to the
desired reaction temperature, in 100% N2 flow as the carrier
gas. Glycerol and water mixture of 6:1 water to glycerol
molar
ratio (WGMR) was introduced into the vaporizer unit using a
high pressure liquid chromatography (HPLC) pump. Themolar
ration of N2 to glycerol was 1:1.6. As the injector was housed
in
the furnace, the mixture was heated at 300 �C to vaporize itand
was subsequently mixed with diluting N2 stream. The
total molar flow rate of the feed mixture was kept constant
at
100 ml/min. The reaction was carried out at atmospheric
http://dx.doi.org/10.1016/j.ijhydene.2016.05.084http://dx.doi.org/10.1016/j.ijhydene.2016.05.084
-
Fig. 1 e Catalytic testing fixed-bed reactor rig set-up.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e
r g y 4 2 ( 2 0 1 7 ) 9 0 8 7e9 0 9 8 9089
pressure. The gaseous products were cooled in a condenser
(cold trap consisting of isopropanol and water in series in
an
ice bath) and the liquid product was then collected. The
gaseous products continuously flowed into the silica bed and
analysed by the gas chromatography (GC).
The product gases passed through amoisture trap before it
was analysed using gas chromatograph (GC6890 e Agilent
Technologies) equipped with TCD with Carboxen-1010 capil-
lary column (30 m � 0.5 mm � 0.32 mm). The duration of re-action
was 5 h and the products were measured online at
30 min interval. Also, the liquid products in the condensate
were identified using a gas chromatograph-mass spectrom-
eter (GC-MS-QP20105 (Shimadzu), equipped with a column of
rtx-wax (30m� 1 mm� 0.32mm)). The catalysts performanceswere
calculated based on the H2 selectivity and glycerol con-
version. Un-reacted water, glycerol and other liquids formed
during the reaction were collected from the condenser and
analysed for determining the glycerol conversion.
H2 Selectivityð%Þ ¼ H2 moles producedTotal C atoms in gaseous
products�1RR
� 100
(6)
where RR is H2/CO2 reforming ratio. In the case of glycerol
steam reforming process the ratio is 7/3.
The glycerol conversion in gas phase is calculated by
Equation (7) [16]:
Gas
� phase Glycerol conversion ð%ÞTotal C atoms in gaseous
productsC atoms in the feedstock
(7)
The selectivity of species i ¼ CO, CO2, C2H4, C2H6 and CH4are
calculated based on equation (8):
Selectivity of i ð%Þ ¼ C atoms in species iTotal C atoms in
gaseous products
� 100 (8)
Meanwhile, the yield of hydrogen is calculated by equation
(9):
Yield of H2ð%Þ ¼ Gas� phase Glycerol Conversion� H2
Selectivityð%Þ (9)
Indeed, the main focus of this study is on gas-phase
products including H2, CH4, CO, and CO2 similar to the ma-
jority of previous studies in this field [12,16,17,22].
Experimental design
Two-level full factorial design (23) was applied in this
study
with three factors that have significant effects on the
hydrogen selectivity and glycerol conversion from steam
reforming process. The effect of reaction temperature (X1),
feed flow rate, FFR (X2) and water glycerol molar ratio,
WGMR
(X3) were investigated at three different levels (low,
medium
and high) and coded as (�1, 0, and þ1), respectively as
tabu-lated in Table 1.
Statistical analysis of the responses was performed using
Statsoft. Statistica software version 8.0. The mathematical
models for H2 selectivity and glycerol conversion were
established by using the method of least squares as given in
Eq. (9):
http://dx.doi.org/10.1016/j.ijhydene.2016.05.084http://dx.doi.org/10.1016/j.ijhydene.2016.05.084
-
Table 1 e Experimental design layout in coded variables.
Factors Symbol Range and levels
�1 0 þ1Reaction temperature,
T (�C)X1 600 650 700
Feed flow rate,
FFR (ml/min)
X2 0.5 1.0 1.5
Water glycerol molar
ratio, WGMR
X3 3 6 9
i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e
r g y 4 2 ( 2 0 1 7 ) 9 0 8 7e9 0 9 89090
Yi ¼ b0 þ b1c1 þ b2c2 þ b3c3 þ b12c1c2 þ b13c1c3 þ b23c2c3 þ
b11c21þ b22c22 þ b33c33
(10)
where the responses for hydrogen selectivity and glycerol
conversion are Y1 and Y2 respectively. X1, X2 and X3 are the
coded values of the test variables for reaction temperature,
feed flow rate (FFR) and water glycerol molar ratio (WGMR),
respectively. The terms b0 is the offset term; b1, b2, and b3
the
linear terms; b12, b12, and b23 the interaction terms; and b11,
b22,and b33 the squared terms.
Results and discussion
Catalyst characterization
X-ray diffraction (XRD)The XRD profiles of Ni/La2O3, Ni/MgO,
Ni/ZrO2, Ni/SiO2 and Ni/
Al2O3 catalysts calcined at 500 �C are illustrated in Fig. 2.
Eachcatalyst exhibits, besides the respective support peaks,
three
peaks at 2q: 37.2�, 43.3� and 63.7� which correspond to
thecharacteristic of NiO crystalline phase (JCPDS 44-1159)
[17].
Generally, the Ni on the support (La2O3, ZrO2, Al2O3, and
MgO)
is highly dispersed except for SiO2 which displays sharp
peak
of bunsenite NiO. In particular, the XRD spectrum attributed
to the Ni/La2O3 catalyst did not show clear and large peak of
Ni
due to the high dispersion of 10 wt% Ni on La2O3 support.
Indeed, a high concentration of Ni is required to saturate
the
surface of the support. Similarly, Cui et al. [23] reported
that
only >15 wt% Ni loading can saturate the La2O3 surface
anddetected by XRD. However, Song and co-workers [24]
mentioned that no lanthanum nickelate was observed in the
XRD characterization because the interaction between Ni and
the support was only limited to near the surface catalyst
and
did not form a bulk compound. Intensity peaks at 37�, 43�
and63.4� correspond to MgO while those at 74.9� and 78.8�
corre-spond to MgNiO2. It was difficult to differentiate NiO
peaks
from MgO because the XRD patterns of NiO and MgO are very
close to each other and the NiO concentration was much
lower thanMgO [25]. The spectrum for support ZrO2 highlights
the main peaks characteristic of the monoclinic phase at 2q:
28.2�, 31.5�, 49.3� and 50.2�. Finally, the diffraction peak
relatedto the bulk Al2O3 is identified to be at 18.75�, 36.9�,
44.2�, and67.3�. Similar to the Ni/La2O3 sample, Ni/Al2O3 also
exhibited ahigh dispersion of Ni on the support. Li and Chen [26]
reported
that the diffusion of nickel ions during calcination into
the
alumina lattice sites is limited to the first few outer layers
of
the support, resulting in a material without three-
dimensional long-range order.
In addition, the crystalline size of Ni in supported cata-
lysts is calculated using Scherrer's equation [27], DP ¼
0:94:lb:Cosq,where Dp, l, b, and q are average crystalline size,
X-ray
wavelength, line broadening (Peak half-width) in radius, and
diffraction angle, respectively. From the results, Ni/MgO
with 29.7 nm and Ni/Al2O3 with 3.8 nm possessed the largest
and smallest crystalline size, respectively while the other
samples registered almost similar crystalline size
(14.9e15.6 nm).
BET surface areaThe textural characteristics of the supported Ni
catalysts,
derived from nitrogen physisorption isotherms and crys-
tallite size from Scherrer equation are presented in Table
2.
The highest BET surface area (169.8 m2/g) belongs to
Ni/SiO2.
The surface area of the prepared catalysts increased in the
following order: Ni/La2O3 < Ni/ZrO2 < Ni/MgO < Ni/Al2O3
< Ni/SiO2. In general, the higher the surface area of acatalyst,
the better is the reactivity since higher surface area
provides larger contact area for the reactant gas [28]. How-
ever, it is not the only factor affecting the catalyst
reactivity.
This suggests that even though the catalyst has low overall
surface area it may have the highest amount of exposed
active Ni [29].
Scanning electron microscopy (SEM)The scanning electron
microscopy (SEM) images in Fig. 3 were
taken to observe the dispersion of the metals on the
supports.
Small dispersion of Ni on the supports was observed in Fig.
3a,
c and d. Large crystallite of Ni/MgO is evident in Fig. 3b.
Generally, metals exhibit aggregated particles with small
and
uniform sizes. Also, the images exhibited well developed
microstructure with uniform colour density. The bright parts
were identified as Ni-rich area. In SEM images, Ni species
appears as roughly spherical particles embedded in the sur-
rounding complex matrix. In studying this interaction, we
observed in Fig 3e that alumina has a greater capacity for
nickel ions than the other supports and that alumina
interacts
more strongly with nickel ions. Therefore, the quasi-meshy
structure may be attributed to the nature of the nickel sup-
ported on alumina by impregnation technique [30]. This is
because the strong interactions of nickel nitrate with
alumina
on the support surface form such a new structure. In
addition,
agglomerated nickel particles might contribute stable
perfor-
mance for hydrogen production by glycerol steam reforming
[31].
Catalyst screening performance
In the present work, Ni catalyst impregnated with five
different supports (La2O3, Al2O3, ZrO2, SiO2 and MgO) were
screened at 650 �C, FFR ¼ 1 ml/min and WGMR ¼ 6. Fig. 4adepicts
the H2 selectivity of all the catalysts tested. Ni/Al2O3registered
the highest H2 selectivity (71.8%) at 650 �C in 5 hreaction time.
The catalyst overall activity increased in the
following order: SiO2 < MgO < ZrO2 < La2O3 <
Al2O3.
http://dx.doi.org/10.1016/j.ijhydene.2016.05.084http://dx.doi.org/10.1016/j.ijhydene.2016.05.084
-
Fig. 2 e XRD patterns of a) Ni/La2O3, b) Ni/MgO, c) Ni/ZrO2 d)
Ni/SiO2 and e) Ni/Al2O3.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e
r g y 4 2 ( 2 0 1 7 ) 9 0 8 7e9 0 9 8 9091
Furthermore, the small crystallite size of Ni/Al2O3 (3.8 nm)
increased the catalyst activity for steam reforming of
glycerol
compared to the other supports. This observation is related
to
the dispersion of relatively large crystallites of nickel
oxides
Table 2e BET surface area and crystallite size of catalysts.
Catalyst BET surface area (m2/g)a Crystallite size (nm)b
Ni/La2O3 20.3 15.5
Ni/ZrO2 31.0 15.6
Ni/MgO 67.8 29.7
Ni/Al2O3 123.4 3.8
Ni/SiO2 169.8 14.9
a Obtained from BET method.b Obtained from XRD analysis.
responsible for the coke precursor adsorption [32]. Ni
supported
on Al2O3 catalyst has higher weight and atomic percentage
with good selectivity of hydrogen for the reforming process
[33].
The smaller particle size of the catalyst led to higher
catalyst
dispersion and large surface area of the supported nickel
-oxide
based catalyst. These results agree with previous finding by
Cheng et al. [34], inferring the most efficient ways to
improve
the reactivity for glycerol steam reforming is to
usemetalswith
high dispersion and large surface area. However, the sample
with the largest surface area,Ni/SiO2 registered the lowest
ac-
tivity among the supported samples. This can be due to the
low
dispersion of Ni on support (SiO2) and large crystallite size
as
confirmed by the XRD results.
Fig. 4b illustrates mixed results on glycerol conversion.
This is because the reaction pathway is complex for each
http://dx.doi.org/10.1016/j.ijhydene.2016.05.084http://dx.doi.org/10.1016/j.ijhydene.2016.05.084
-
Fig. 3 e Scanning electron microscopy of a) Ni/La2O3, b) Ni/MgO,
c) Ni/ZrO2, d) Ni/SiO2 and e) Ni/Al2O3.
i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e
r g y 4 2 ( 2 0 1 7 ) 9 0 8 7e9 0 9 89092
support and a number of undesirable side reactions occurred
thus, producing several carbonaceous by-products [35].
Furthermore, Dou et al. [36] explained high glycerol conver-
sion with Ni/Al2O3 may be caused by the active support
catalyst. The glycerol conversion for the reaction at 650
�Cwasin the order:
SiO2 < ZrO2 < La2O3 < MgO < Al2O3.Table 3 lists the
mean compositions of the products from
these reactions. Ethylene and ethane gases were not observed
over Ni supported on La2O3 and Al2O3. Methane (CH4) selec-
tivitywas sensitive to the particle size of supported Ni, and
the
smaller nickel particles favoured lower amount of CH4 [37].
Although the crystallite size of the Ni/La2O3 in this study
is
relatively large, it is evident its other characteristics
attributed
to the increased activity. Dehydration of glycerol to
ethylene
which usually contributed to rapid deactivation of the cata-
lysts through coke formation was observed over Ni supported
on ZrO2, SiO2 and MgO within the range of the operating
conditions studied.
Thermodynamic analysis of the water-glycerol system in-
dicates that at equilibrium the only additional reaction
prod-
uct in the gas phase is methane, the formation of which is
due
to the hydrogenation of CO or also called methanation reac-
tion (Eq. (4)). Since the decomposition of glycerol to CH4
is
highly favourable during the reforming process, the Ni sup-
ported on La2O3 and Al2O3 catalysts must have sufficient ca-
pacity for reforming the produced CH4 into hydrogen and
carbonmonoxide (reversed of Eqs. (4) and (5)) and the
catalyst
also must facilitate the water gas shift reaction to convert
CO
into CO2. Some of the major products in the liquid phase
were
http://dx.doi.org/10.1016/j.ijhydene.2016.05.084http://dx.doi.org/10.1016/j.ijhydene.2016.05.084
-
Fig. 4 e a) Hydrogen selectivity; b) Glycerol conversion
(Reaction conditions: WGMR ¼ 6:1, FFR ¼ 1.0 ml/min, T ¼ 650
�C).
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e
r g y 4 2 ( 2 0 1 7 ) 9 0 8 7e9 0 9 8 9093
formaldehyde, acetaldehyde, ethanol, acetic acid, acetol,
and
propylene glycol (data not shown). Based on the H2
selectivity
and glycerol conversion, Ni/Al2O3 was the best performing
catalyst in this study.
Large surface area, high dispersion of active phase on
support, and small crystalline sizes are attributes of
active
catalyst in steam reforming of glycerol to hydrogen. How-
ever, some researchers reported that support basicity is the
other important factor in catalyst activity. The basicity of
bulk supports La2O3, MgO, TiO2, and SiO2 were 31, 9.8, 6.2,
and 3.2 mmol/g.cat, respectively and ZrO2 and Al2O3 only
exhibited trace of basicity [38]. Bulk ZrO2 and Al2O3
possessed
more acidic sites than basic sites. In fact, ZrO2 and
Al2O3possessed 0.2 and 1.6 mmol/g.cat acidity. Viinikainen et
al.
[39] and Fleyset al. [40] reported that higher basicity of
cata-
lyst can limit the carbon deposition and enhance the
catalyst
stability. The significant effect of basicity can clearly be
seen
in the activity of Ni/La2O3 sample with 68.3% hydrogen
selectivity. In fact, the Ni/La2O3 registered the smallest
sur-
face area (20.3 m2/g) and the lowest weight percentage
(4.7%)
of Ni on the support. However, it performed the second
highest hydrogen selectivity among the tested catalysts in
this study. The La2O3 basicity is essential for better Ni
dis-
tribution on the support surface and to facilitate carbon
gasification. Also, Mazumder and de Lasa [41] revealed that
increasing the Ni loading from 10 to 20 wt% decreased the
total basicity due to the partial blocking of basic sites by
the
impregnated nickel.
Table 3 e Mean composition of gaseous products (%) of
glycero
Catalyst Glycerol con (%)
H2 CH4
Ni/La2O3 70 68.3 0.5
Ni/Al2O3 80 75.1 4.0
Ni/ZrO2 57 60.7 1.1
Ni/SiO2 48 45.9 16.8
Ni/MgO 73 55.4 12.8
a Reaction conditions: T ¼ 650 �C; WGMR ¼ 6; TOS ¼ 5 h.
Ni/La2O3, Ni/ZrO2and Ni/SiO2 were stable throughout the
reaction period, whereas Ni/Al2O3, Ni/MgO unveiled catalyst
deactivation with time on stream (Fig. 4a). The deactivation
may be attributed to the increased crystallinity of support,
sintering of metal particles, oxidation of metal sites and
car-
bon deposition as have been reported elsewhere [42e44]. For
instance, Bartholomew et al. [42] reported carbon deposition
caused Ni/Al2O3 catalyst deactivation. Also, Sad et al. [12]
confirmed that the catalyst deactivation is mostly due to
blockage of the active sites by coke precursors formed on
surface acid sites thereby decreasing the rate of H2 produc-
tion. Indeed, as we reported in Eqs. (4) and (5), production of
H2reduced and CH4 increased via methanation of CO and CO2.
Thus, based on the results reported in this section,
Ni/Al2O3catalyst has been selected in the optimization of the
process
by Response Surface Methodology (RSM).
Model analysis
Response surface methodology was employed to analyse the
interaction or relationship between the responses and the
variables. The matrix for the experimental and predicted re-
sults of glycerol conversion and hydrogen selectivity are
given
in Table 4. The H2 selectivity, Y1 and glycerol conversion,
Y2are the responses for the tested variables in coded units:
re-
action temperature (x1), FFR (x2) and WGMR (x3). The second
order polynomial models for H2 selectivity, Y1 and glycerol
conversion, Y2 are as follows:
l steam reforming over supported Ni catalystsa.
Gaseous products (%)
CO CO2 C2H4 C2H6
3.5 27.7 e e
8.3 12.6 e e
4.3 26.1 5.2 2.6
8.9 16.2 9.3 2.9
5.6 16.8 4.6 4.8
http://dx.doi.org/10.1016/j.ijhydene.2016.05.084http://dx.doi.org/10.1016/j.ijhydene.2016.05.084
-
Table 4 e The 23 factorial central composite design matrix in
real and coded units and the experimental response values.
NO X1 X2 X3 Experimental results RSM predicted results
T FFR WGMR Glycerolconversion,
Y2 (%)
Hydrogenselectivity,
Y1 (%)
HydrogenYield (%)
Glycerolconversion,
Y2 (%)
Hydrogenselectivity,
Y1 (%)
1 600 0.5 3 58.6 74.2 43.5 62.8 74.8
2 700 0.5 3 60.3 76.1 45.9 61.2 74.9
3 600 1.5 3 52.9 70.6 37.4 55.8 69.9
4 700 1.5 3 56.9 71.8 40.9 60.9 72.3
5 600 0.5 9 74.1 85.5 63.4 73.8 83.7
6 700 0.5 9 79.6 89.4 71.2 80.4 88.8
7 600 1.5 9 57.4 72.9 41.8 60.2 72.8
8 700 1.5 9 73.7 82.1 60.5 73.2 80.2
9 565.9 1 6 68.5 75.7 51.9 64.6 76.3
10 734.1 1 6 75.6 81.2 61.4 74.3 82.5
11 650 0.16 6 74.1 80.6 59.7 72.6 81.8
12 650 1.84 6 64.4 69.8 45.0 60.7 70.5
13 650 1 0.95 58.6 69.6 40.8 53.2 69.5
14 650 1 11.1 72.8 81.6 59.4 72.9 83.6
15 650 1 6 72.2 80.4 58.1 72.4 80.4
16 650 1 6 72.2 80.4 58.1 72.4 80.4
17 650 1 6 72.2 80.4 58.1 72.4 80.4
18 650 1 6 72.2 80.4 58.1 72.4 80.4
19 650 1 6 72.2 80.4 58.1 72.4 80.4
20 650 1 6 72.2 80.4 58.1 72.4 80.4
i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e
r g y 4 2 ( 2 0 1 7 ) 9 0 8 7e9 0 9 89094
Y1 ¼ �44:0905þ 0:3546X1 � 28:6535X2 � 1:8333X3 þ 0:0655X1X2þ
0:0134X1X3 � 1:1250X2X3 � 0:0003X21 � 7:0177X22� 0:3217X23
(11)
Y2 ¼ 6:8789þ 0:2093X1 � 14:1614X2 þ 0:1187X3 þ 0:0380X1X2þ
0:0058X1X3 þ 0:7500X2X3 � 0:0002X21 � 6:1025X22� 0:1552X23
(12)
These models signified the adequacy between the
observed and predicted results where the coefficient of
Fig. 5 e Parity plot for a) hydrogen sele
determination (R2) value for both H2 selectivity and
glycerol
conversion were closer to 1. The values of R2 were 0.868
and 0.966 for H2 selectivity and glycerol conversion as
shown in Fig. 5 indicating that 86.8% and 96.6% of the
variability in the responses can be explained by the models.
The empirical model is adequate to explain most of the
variability in the assay reading which should be at least
0.75 [45].
The analysis of variance (ANOVA) as tabulated in Table 5
was used to check the F-values by comparing it with the
tabulated F-value. The tabulated F-value was used at high
confidence level (95%) in order to obtain a good prediction
model. The F-values for H2 selectivity and glycerol
conversion
are 4.39 and 18.93; respectively, higher than the tabulated
F-
ctivity and b) glycerol conversion.
http://dx.doi.org/10.1016/j.ijhydene.2016.05.084http://dx.doi.org/10.1016/j.ijhydene.2016.05.084
-
Table 5 e Analysis of variance (ANOVA) for quadratic model.
Sources Sum of squares(SS)
Degree of freedom(DF)
Mean squares(MS)
F-value F0.05
H2 selectivity model
Regression (SSR) 915.13 9 101.68 4.39 >4.1Residual 139.02 6
23.17
Total (SST) 1054.15 15
Glycerol conversion
model
Regression (SSR) 444.62 9 49.40 18.93 >4.1Residual 15.66 6
2.61
Total (SST) 460.27 15
Fig. 6 e Pareto chart of a) hydrogen selectivity and b) glycerol
conversion.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e
r g y 4 2 ( 2 0 1 7 ) 9 0 8 7e9 0 9 8 9095
value (F0.05, 9, 6 ¼ 4.1) by rejecting the null hypothesis at
0.05significant level.
The significance of each coefficient is determined by Pareto
chart (Fig. 6). The greater the magnitude of the t-value and
the
smaller the p-value, the more significant is the
corresponding
coefficient. As illustrated, the largest effect on hydrogen
selectivity and glycerol conversion is the linear term
ofWGMR
(X3), with the largest t-value (4.4702, 9.0880) and smallest
p-
value (0.0042, 0.0001) at approximately 99% significant
level,
respectively. Linear term of FFR (X2) could also be regarded
as
a significant variable in H2 selectivity at 98% significant
level
since p-value < 0.05. Meanwhile, the linear term of FFR (X2),
T(X1), quadratic term of FFR (X2
2), and WGMR (X32) are significant
in glycerol conversion at 98% and 97% significant level,
respectively.
Variables effects on the responses
The empirical models were plotted as a three-dimensional
surface representing the responses of H2 selectivity, Y1 and
glycerol conversion, Y2 as a function of two factors within
the experimental range considered. In the presence of 6:1
WGMR, the elliptical nature of H2 selectivity (Fig. 7a) and
glycerol conversion (Fig. 7b) can be observed. The contour
plots indicated the interaction of FFR and reaction
temperature was significant on H2 selectivity and glycerol
conversion. Higher H2 selectivity and glycerol conversion
could be attained at higher reaction temperature and lower
feed flow rate implying endothermic reaction [46]. Fig. 7c
and
d illustrate the interaction of WGMR and FFR at temperature
650 �C on H2 selectivity and glycerol conversion,
respectively.Increment of H2 selectivity and glycerol conversion
with
increasing WGMR and decreasing FFR can be observed from
the figures. Meanwhile, H2 selectivity and glycerol conver-
sion increased with increasing WGMR and higher reaction
temperature at FFR ¼ 1.0 ml/min as depicted by Fig. 7e and
f.With increasing WGMR and higher reaction temperature, the
production of CH4 was completely inhibited due to water gas
shift reaction [47]. Generally, high temperature, low pres-
sure, and high water/glycerol ratio favour hydrogen pro-
duction. Indeed, methane was decreased and carbon
formation was thermodynamically inhibited under these
conditions [12].
Optimization of hydrogen selectivity
Based on the full quadratic model, the H2 selectivity was
predicted at optimum conditions in order to obtain high
H2selectivity in glycerol steam reforming. The optimization
using CCDwas conducted based on the variables in the range
http://dx.doi.org/10.1016/j.ijhydene.2016.05.084http://dx.doi.org/10.1016/j.ijhydene.2016.05.084
-
Fig. 7 e The response surface plot of hydrogen selectivity and
glycerol conversion as a function of aeb) FFR and temperature,
ced) WGMR and FFR, eef) temperature and WGMR.
i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e
r g y 4 2 ( 2 0 1 7 ) 9 0 8 7e9 0 9 89096
of experimental design. The optimum conditions predicted
at temperature ¼ 692 �C, FFR ¼ 1 ml/min, and WGMR
9.5:1,respectively corresponded to 86.0% of predicted H2 selec-
tivity. Triplicate experiments were carried out at these op-
timum conditions to confirm the accuracy and validation of
the regression model. The H2 selectivity for the observed
value was 82.3% indicating a 4.5% error between the
observed and predicted values. The error is considered small
as the observed value was within the 5% of significance
level.
The results in Table 6 confirmed that 86% hydrogen selec-
tivity obtained in the current study is better than the ma-
jority of previous results. Sad et al. [12] reported 100%
hydrogen yield in a double bed reactor by application two
catalysts in series, but in a single bed reactor the
hydrogen
yield was only 78.8%.
Conclusion
Glycerol steam reforming for hydrogen production was con-
ducted over Ni supported on La2O3, Al2O3, ZrO2, SiO2, andMgO
catalysts. Ni/Al2O3 was found to be the best catalyst with
maximumhydrogen selectivity (71.5%) was obtained at 650 �C,FFR ¼
1 ml/min and WGMR ¼ 6. Large surface area (123.4 m2/g), small
crystallite size (3.8 nm) and high dispersion of Ni on
the support were the main reason which increased the ac-
tivity of Ni/Al2O3 sample for steam reforming of glycerol
compared to the other supported catalysts.
The optimization results revealed that the water glycerol
molar ratio (WGMR) has the greatest effect on the hydrogen
selectivity and glycerol conversion compared to reaction
http://dx.doi.org/10.1016/j.ijhydene.2016.05.084http://dx.doi.org/10.1016/j.ijhydene.2016.05.084
-
Table 6 e Comparison of the current study with previously
reported results.
Catalyst Tem (�C) WGMR Glycerol con (%) H2 Ref
Ni/Al2O3 650 6:1 80 Selectivity 86% This Study
Ni/(Ca/Al2O3) 550 3:1 54.1 Concentration 85% [22]
LaNi0.9Cu0.1O3 650 3:1 73 Selectivity 67% [17]
Pt/SiO2 350 e 100 Yield 78.8% [12]
0.5Pt/SiO20.5Pt/TiO2(Double bed Reactor)
350 3:1 100 Yield 100% [12]
NieFeeCe/Al2O3 450 e 94.06 Selectivity 64.04% [16]
Ni/La2O3eSiO2 (P þ I) 600 e 79 Yield 3.8 mol/mol [48]
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e
r g y 4 2 ( 2 0 1 7 ) 9 0 8 7e9 0 9 8 9097
temperature and feed flow rate (FFR). Also, the optimization
results revealed that steam reforming of glycerol produced
86.0%maximum hydrogen selectivity at the optimum reaction
temperature 692 �C, 1 ml/min feed flow rate, andWGMR 9.5:1.
Acknowledgements
The authors would like to express their gratitude to the
Min-
istry of Higher Education for supporting this project under
the
Fundamental Research Grant Scheme (FRGS) vote number
78422 and RACE grant vote number 00M32. Furthermore, one
of the authors (NHZ) is thankful to the Ministry of Science,
Technology and Innovation for the National Science Fellow-
ship (NSF) scheme.
r e f e r e n c e s
[1] Meher LC, Churamani CP, Arif M, Ahmed Z, Naik SN.Jatropha
curcas as a renewable source for bio-fuelsdareview. Renew Sustain
Energy Rev 2013;26(0):397e407.
[2] Rahmat N, Abdullah AZ, Mohamed AR. Recent progress
oninnovative and potential technologies for glyceroltransformation
into fuel additives: a critical review. RenewSustain Energy Rev
2010;14(3):987e1000.
[3] Avasthi KS, Reddy RN, Patel S. Challenges in the
productionof hydrogen from glycerol e a biodiesel byproduct via
steamreforming process. Procedia Eng 2013;51(0):423e9.
[4] Silva JM, Soria MA, Madeira LM. Challenges and strategies
foroptimization of glycerol steam reforming process. RenewSustain
Energy Rev 2015;42(0):1187e213.
[5] Teichmann D, Arlt W, Wassersche P. Liquid organichydrogen
carriers as an efficient vector for the transport andstorage of
renewable energy. Int J Hydrogen Energy2012;37:18118e32.
[6] Pudukudy M, Yaakob Z, Mohammad M, Narayanan B,Sopian K.
Renewable hydrogen economy in Asia eopportunities and challenges:
an overview. Renew SustainEnergy Rev 2014;30:743e57.
[7] Pudukudy M, Yaakob Z. Methane decomposition over Ni, Coand
Fe based monometallic catalysts supported on sol gelderived SiO2
microflakes. Chem Eng J 2015;262:1009e21.
[8] Pudukudy M, Yaakob Z, Akmal ZS. Direct decomposition
ofmethane over Pd promoted Ni/SBA-15 catalysts. Appl SurfSci
2015;353:127e36.
[9] Pudukudy M, Yaakob Z, Takriff MS. Methane decompositionover
Pd promoted Ni/MgAl2O4 catalysts for the production ofCOx free
hydrogen and multiwalled carbon nanotubes. ApplSurf Sci
2015;356:1320e6.
[10] Pudukudy M, Yaakob Z, Akmal ZS. Direct decomposition
ofmethane over SBA-15 supported Ni, Co and Fe basedbimetallic
catalysts. Appl Surf Sci 2015;330:418e30.
[11] Pompeo F, Santori GF, Nichio NN. Hydrogen production
byglycerol steam reforming with Pt/SiO2 and Ni/SiO2 catalysts.Catal
Today 2011;172(1):183e8.
[12] Sad ME, Duarte HA, Vignatti C, Padr�o CL, Apesteguı́a
CR.Steam reforming of glycerol: hydrogen productionoptimization.
Int J Hydrogen Energy 2015;40(18):6097e106.
[13] Buffoni IN, Pompeo F, Santori GF, Nichio NN. Nickel
catalystsapplied in steam reforming of glycerol for
hydrogenproduction. Catal Commun 2009;10(13):1656e60.
[14] Cheng CK, Foo SY, Adesina AA. Glycerol steam reformingover
bimetallic Co�Ni/Al2O3. Industrial Eng Chem
Res2010a;49(21):10804e17.
[15] Chiodo V, Freni S, Galvagno A, Mondello N, Frusteri
F.Catalytic features of Rh and Ni supported catalysts in thesteam
reforming of glycerol to produce hydrogen. Appl CatalA General
2010;381(1e2):1e7.
[16] Go YJ, Go GS, Lee HJ, Moon DJ, Park NC, Kim YC. The
relationbetween carbon deposition and hydrogen production
inglycerol steam reforming. Int J Hydrogen
Energy2015;40(35):11840e7.
[17] Ramesh S, Yang EH, Jung JS, Moon DJ. Copper
decoratedperovskite an efficient catalyst for low
temperaturehydrogen production by steam reforming of glycerol. Int
JHydrogen Energy 2015;40(35):11428e35.
[18] Douette AMD, Turn SQ, Wang W, Keffer VI.
Experimentalinvestigation of hydrogen production from
glycerinreforming. Energy & Fuels 2007;21(6):3499e504.
[19] Damyanova S, Pawelec B, Arishtirova K, Fierro JLG.
Ni-basedcatalysts for reforming of methane with CO2. Int J
HydrogenEnergy 2012;37(21):15966e75.
[20] Jadid Ettaz S. Oxidation synthesis and reaction analysis of
anew arranged catalyst support. Master of Science. Universityof
Alberta; 2010.
[21] Gangadharan D, Sivaramakrishnan S, Nampoothiri KM,Sukumaran
RK, Pandey A. Response surface methodologyfor the optimization of
alpha amylase production byBacillus amyloliquefaciens. Bioresour
Technol2008;99(11):4597e602.
[22] Wang C, Dou B, Jiang B, Song Y, Du B, Zhang C, et
al.Sorption-enhanced steam reforming of glycerol on
Ni-basedmultifunctional catalysts. Int J Hydrogen
Energy2015;40:7037e44.
[23] Cui Y, Galvita V, Rihko-Struckmann L, Lorenz H,Sundmacher
K. Steam reforming of glycerol: theexperimental activity of
La1-xCexNiO3 catalyst incomparison to the thermodynamic reaction
equilibrium.Appl Catal B Environ 2009;90(1e2):29e37.
[24] Song H, Yang J, Zhao J, Chou L. Methanation of
carbondioxide over a highly dispersed Ni/La2O3 catalyst. Chin J
Catal2010;31(1):21e3.
http://refhub.elsevier.com/S0360-3199(15)31138-1/sref1http://refhub.elsevier.com/S0360-3199(15)31138-1/sref1http://refhub.elsevier.com/S0360-3199(15)31138-1/sref1http://refhub.elsevier.com/S0360-3199(15)31138-1/sref1http://refhub.elsevier.com/S0360-3199(15)31138-1/sref1http://refhub.elsevier.com/S0360-3199(15)31138-1/sref2http://refhub.elsevier.com/S0360-3199(15)31138-1/sref2http://refhub.elsevier.com/S0360-3199(15)31138-1/sref2http://refhub.elsevier.com/S0360-3199(15)31138-1/sref2http://refhub.elsevier.com/S0360-3199(15)31138-1/sref2http://refhub.elsevier.com/S0360-3199(15)31138-1/sref3http://refhub.elsevier.com/S0360-3199(15)31138-1/sref3http://refhub.elsevier.com/S0360-3199(15)31138-1/sref3http://refhub.elsevier.com/S0360-3199(15)31138-1/sref3http://refhub.elsevier.com/S0360-3199(15)31138-1/sref3http://refhub.elsevier.com/S0360-3199(15)31138-1/sref4http://refhub.elsevier.com/S0360-3199(15)31138-1/sref4http://refhub.elsevier.com/S0360-3199(15)31138-1/sref4http://refhub.elsevier.com/S0360-3199(15)31138-1/sref4http://refhub.elsevier.com/S0360-3199(15)31138-1/sref5http://refhub.elsevier.com/S0360-3199(15)31138-1/sref5http://refhub.elsevier.com/S0360-3199(15)31138-1/sref5http://refhub.elsevier.com/S0360-3199(15)31138-1/sref5http://refhub.elsevier.com/S0360-3199(15)31138-1/sref5http://refhub.elsevier.com/S0360-3199(15)31138-1/sref6http://refhub.elsevier.com/S0360-3199(15)31138-1/sref6http://refhub.elsevier.com/S0360-3199(15)31138-1/sref6http://refhub.elsevier.com/S0360-3199(15)31138-1/sref6http://refhub.elsevier.com/S0360-3199(15)31138-1/sref6http://refhub.elsevier.com/S0360-3199(15)31138-1/sref7http://refhub.elsevier.com/S0360-3199(15)31138-1/sref7http://refhub.elsevier.com/S0360-3199(15)31138-1/sref7http://refhub.elsevier.com/S0360-3199(15)31138-1/sref7http://refhub.elsevier.com/S0360-3199(15)31138-1/sref7http://refhub.elsevier.com/S0360-3199(15)31138-1/sref8http://refhub.elsevier.com/S0360-3199(15)31138-1/sref8http://refhub.elsevier.com/S0360-3199(15)31138-1/sref8http://refhub.elsevier.com/S0360-3199(15)31138-1/sref8http://refhub.elsevier.com/S0360-3199(15)31138-1/sref9http://refhub.elsevier.com/S0360-3199(15)31138-1/sref9http://refhub.elsevier.com/S0360-3199(15)31138-1/sref9http://refhub.elsevier.com/S0360-3199(15)31138-1/sref9http://refhub.elsevier.com/S0360-3199(15)31138-1/sref9http://refhub.elsevier.com/S0360-3199(15)31138-1/sref10http://refhub.elsevier.com/S0360-3199(15)31138-1/sref10http://refhub.elsevier.com/S0360-3199(15)31138-1/sref10http://refhub.elsevier.com/S0360-3199(15)31138-1/sref10http://refhub.elsevier.com/S0360-3199(15)31138-1/sref11http://refhub.elsevier.com/S0360-3199(15)31138-1/sref11http://refhub.elsevier.com/S0360-3199(15)31138-1/sref11http://refhub.elsevier.com/S0360-3199(15)31138-1/sref11http://refhub.elsevier.com/S0360-3199(15)31138-1/sref12http://refhub.elsevier.com/S0360-3199(15)31138-1/sref12http://refhub.elsevier.com/S0360-3199(15)31138-1/sref12http://refhub.elsevier.com/S0360-3199(15)31138-1/sref12http://refhub.elsevier.com/S0360-3199(15)31138-1/sref12http://refhub.elsevier.com/S0360-3199(15)31138-1/sref13http://refhub.elsevier.com/S0360-3199(15)31138-1/sref13http://refhub.elsevier.com/S0360-3199(15)31138-1/sref13http://refhub.elsevier.com/S0360-3199(15)31138-1/sref13http://refhub.elsevier.com/S0360-3199(15)31138-1/sref14http://refhub.elsevier.com/S0360-3199(15)31138-1/sref14http://refhub.elsevier.com/S0360-3199(15)31138-1/sref14http://refhub.elsevier.com/S0360-3199(15)31138-1/sref14http://refhub.elsevier.com/S0360-3199(15)31138-1/sref14http://refhub.elsevier.com/S0360-3199(15)31138-1/sref14http://refhub.elsevier.com/S0360-3199(15)31138-1/sref14http://refhub.elsevier.com/S0360-3199(15)31138-1/sref15http://refhub.elsevier.com/S0360-3199(15)31138-1/sref15http://refhub.elsevier.com/S0360-3199(15)31138-1/sref15http://refhub.elsevier.com/S0360-3199(15)31138-1/sref15http://refhub.elsevier.com/S0360-3199(15)31138-1/sref15http://refhub.elsevier.com/S0360-3199(15)31138-1/sref15http://refhub.elsevier.com/S0360-3199(15)31138-1/sref16http://refhub.elsevier.com/S0360-3199(15)31138-1/sref16http://refhub.elsevier.com/S0360-3199(15)31138-1/sref16http://refhub.elsevier.com/S0360-3199(15)31138-1/sref16http://refhub.elsevier.com/S0360-3199(15)31138-1/sref16http://refhub.elsevier.com/S0360-3199(15)31138-1/sref17http://refhub.elsevier.com/S0360-3199(15)31138-1/sref17http://refhub.elsevier.com/S0360-3199(15)31138-1/sref17http://refhub.elsevier.com/S0360-3199(15)31138-1/sref17http://refhub.elsevier.com/S0360-3199(15)31138-1/sref17http://refhub.elsevier.com/S0360-3199(15)31138-1/sref18http://refhub.elsevier.com/S0360-3199(15)31138-1/sref18http://refhub.elsevier.com/S0360-3199(15)31138-1/sref18http://refhub.elsevier.com/S0360-3199(15)31138-1/sref18http://refhub.elsevier.com/S0360-3199(15)31138-1/sref18http://refhub.elsevier.com/S0360-3199(15)31138-1/sref19http://refhub.elsevier.com/S0360-3199(15)31138-1/sref19http://refhub.elsevier.com/S0360-3199(15)31138-1/sref19http://refhub.elsevier.com/S0360-3199(15)31138-1/sref19http://refhub.elsevier.com/S0360-3199(15)31138-1/sref19http://refhub.elsevier.com/S0360-3199(15)31138-1/sref20http://refhub.elsevier.com/S0360-3199(15)31138-1/sref20http://refhub.elsevier.com/S0360-3199(15)31138-1/sref20http://refhub.elsevier.com/S0360-3199(15)31138-1/sref21http://refhub.elsevier.com/S0360-3199(15)31138-1/sref21http://refhub.elsevier.com/S0360-3199(15)31138-1/sref21http://refhub.elsevier.com/S0360-3199(15)31138-1/sref21http://refhub.elsevier.com/S0360-3199(15)31138-1/sref21http://refhub.elsevier.com/S0360-3199(15)31138-1/sref21http://refhub.elsevier.com/S0360-3199(15)31138-1/sref22http://refhub.elsevier.com/S0360-3199(15)31138-1/sref22http://refhub.elsevier.com/S0360-3199(15)31138-1/sref22http://refhub.elsevier.com/S0360-3199(15)31138-1/sref22http://refhub.elsevier.com/S0360-3199(15)31138-1/sref22http://refhub.elsevier.com/S0360-3199(15)31138-1/sref23http://refhub.elsevier.com/S0360-3199(15)31138-1/sref23http://refhub.elsevier.com/S0360-3199(15)31138-1/sref23http://refhub.elsevier.com/S0360-3199(15)31138-1/sref23http://refhub.elsevier.com/S0360-3199(15)31138-1/sref23http://refhub.elsevier.com/S0360-3199(15)31138-1/sref23http://refhub.elsevier.com/S0360-3199(15)31138-1/sref23http://refhub.elsevier.com/S0360-3199(15)31138-1/sref23http://refhub.elsevier.com/S0360-3199(15)31138-1/sref24http://refhub.elsevier.com/S0360-3199(15)31138-1/sref24http://refhub.elsevier.com/S0360-3199(15)31138-1/sref24http://refhub.elsevier.com/S0360-3199(15)31138-1/sref24http://refhub.elsevier.com/S0360-3199(15)31138-1/sref24http://refhub.elsevier.com/S0360-3199(15)31138-1/sref24http://dx.doi.org/10.1016/j.ijhydene.2016.05.084http://dx.doi.org/10.1016/j.ijhydene.2016.05.084
-
i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e
r g y 4 2 ( 2 0 1 7 ) 9 0 8 7e9 0 9 89098
[25] Adhikari S, Fernando SD, Haryanto A. Hydrogen
productionfrom glycerin by steam reforming over nickel
catalysts.Renew Energy 2008;33(5):1097e100.
[26] Li C, Chen Y-W. Temperature-programmed-reductionstudies of
nickel oxide/alumina catalysts: effects of thepreparation method.
Thermochim Acta 1995;256(2):457e65.
[27] Tsukuda E, Sato S, Takahashi R, Sodesawa T. Production
ofacrolein from glycerol over silica-supported heteropolyacids.
Catal Commun 2007;8(9):1349e53.
[28] Song HS. CH4 reforming for synthesis gas production
oversupported Ni catalysts. Master of Applied Science inChemical
Engineering. Canada: University of Waterloo; 2010.
[29] Padilla R, Benito M, Rodrı́guez L, Serrano A, Mu~noz G,
Daza L.Nickel and cobalt as active phase on supported
zirconiacatalysts for bio-ethanol reforming: influence of the
reactionmechanism on catalysts performance. Int J Hydrogen
Energy2010;35(17):8921e8.
[30] Chang F-W, Kuo M-S, Tsay M-T, Hsieh M-C. Hydrogenation
ofCO2 over nickel catalysts on rice husk ash-alumina preparedby
incipient wetness impregnation. Appl Catal A
General2003;247(2):309e20.
[31] Broda M, Manovic V, Imtiaz Q, Kierzkowska AM, Anthony
EJ,Mueller CR. High-purity hydrogen via the sorption-enhancedsteam
methane reforming reaction over a synthetic CaObased sorbent and a
Ni catalyst. Environ Sci Technol2013;47:6007e14.
[32] Bangala DN, Abatzoglou N, Chornet E. Steam reforming
ofnaphthalene on NieCr/Al2O3 catalysts doped with MgO, TiO2,and
La2O3. AIChE J 1998;44(4):927e36.
[33] Alberton AL, Souza MMVM, Schmal M. Carbon formation andits
influence on ethanol steam reforming over Ni/Al2O3catalysts. Catal
Today 2007;123(1e4):257e64.
[34] Cheng CK, Foo SY, Adesina AA. H2-rich synthesis
gasproduction over Co/Al2O3 catalyst via glycerol steamreforming.
Catal Commun 2010b;12(4):292e8.
[35] Li S, Li M, Zhang C, Wang S, Ma X, Gong J. Steam reforming
ofethanol over Ni/ZrO2 catalysts: effect of support on
productdistribution. Int J Hydrogen Energy 2012;37(3):2940e9.
[36] Dou B, Dupont V, Rickett G, Blakeman N, Williams PT,Chen H,
et al. Hydrogen production by sorption-enhancedsteam reforming of
glycerol. Bioresour Technol2009;100(14):3540e7.
[37] Chen SQ, Liu Y. LaFeyNi1-yO3 supported nickel catalysts
usedfor steam reforming of ethanol. Int J Hydrogen
Energy2009;34(11):4735e46.
[38] Muroyama H, Saburi C, Matsui T, Eguchi K.
Ammoniadecomposition over Ni/La2O3 catalyst for on-site
generationof hydrogen. Appl Catal A General
2012;443e444:119e24.
[39] Viinikainen T, Ronkkonen H, Bradshaw H, Stephenson
H,Airaksinen S, Reinikainen M, et al. Acidic and basic surfacesites
of zirconia-based biomass gasification gas clean-upcatalysts. Appl
Catal A General 2009;362:169e77.
[40] Fleys M, Simon Y, Swierczynski D, Kiennemann A,Marquaire P.
Investigation of the reaction of partial oxidationof methane over
Ni/La2O3 catalyst. Energy Fuels2006;20:2321e9.
[41] Mazumder J, de Lasa H. Fluidizable Ni/La2O3-g-Al2O3
catalystfor steam gasification of a cellulosic biomass surrogate.
ApplCatal B Environ 2014;160e161:67e79.
[42] Bartholomew CH, Weatherbee GD, Jarvi GA. Effects of
carbondeposits on the specific activity of nickel and
nickelbimetallic catalysts. Chem Eng Commun 1980;5(1):125e34.
[43] De Lima SM, Silva AM, Da Cruz IO, Jacobs G, Davis BH,Mattos
LV, et al. H2 production through steam reforming ofethanol over
Pt/ZrO2, Pt/CeO2 and Pt/CeZrO2 catalysts. CatalToday
2008;138(3e4):162e8.
[44] Lakhapatri SL. Analysis of deactivation mechanism on
amulti-component sulfur-tolerant steam reforming catalyst.Doctor of
Philosophy Degree in Engineering. The Universityof Toledo;
2010.
[45] Haaland PD. Experimental design in biotechnology. NewYork:
Marcel Dekker Inc.; 1989.
[46] Rossi CCRS, Alonso CG, Antunes OAC, Guirardello
R,Cardozo-Filho L. Thermodynamic analysis of steamreforming of
ethanol and glycerine for hydrogen production.Int J Hydrogen Energy
2009;34(1):323e32.
[47] Fatsikostas AN, Verykios XE. Reaction network of
steamreforming of ethanol over Ni-based catalysts. J
Catal2004;225(2):439e52.
[48] Thyssen VV, Assaf EM. Ni/La2O3-SiO2 catalysts applied
toglycerol steam reforming reaction: effect of the
preparationmethod and reaction temperature. J Braz Chem
Soc2014;25:2455e65.
http://refhub.elsevier.com/S0360-3199(15)31138-1/sref25http://refhub.elsevier.com/S0360-3199(15)31138-1/sref25http://refhub.elsevier.com/S0360-3199(15)31138-1/sref25http://refhub.elsevier.com/S0360-3199(15)31138-1/sref25http://refhub.elsevier.com/S0360-3199(15)31138-1/sref26http://refhub.elsevier.com/S0360-3199(15)31138-1/sref26http://refhub.elsevier.com/S0360-3199(15)31138-1/sref26http://refhub.elsevier.com/S0360-3199(15)31138-1/sref26http://refhub.elsevier.com/S0360-3199(15)31138-1/sref27http://refhub.elsevier.com/S0360-3199(15)31138-1/sref27http://refhub.elsevier.com/S0360-3199(15)31138-1/sref27http://refhub.elsevier.com/S0360-3199(15)31138-1/sref27http://refhub.elsevier.com/S0360-3199(15)31138-1/sref28http://refhub.elsevier.com/S0360-3199(15)31138-1/sref28http://refhub.elsevier.com/S0360-3199(15)31138-1/sref28http://refhub.elsevier.com/S0360-3199(15)31138-1/sref28http://refhub.elsevier.com/S0360-3199(15)31138-1/sref29http://refhub.elsevier.com/S0360-3199(15)31138-1/sref29http://refhub.elsevier.com/S0360-3199(15)31138-1/sref29http://refhub.elsevier.com/S0360-3199(15)31138-1/sref29http://refhub.elsevier.com/S0360-3199(15)31138-1/sref29http://refhub.elsevier.com/S0360-3199(15)31138-1/sref29http://refhub.elsevier.com/S0360-3199(15)31138-1/sref29http://refhub.elsevier.com/S0360-3199(15)31138-1/sref30http://refhub.elsevier.com/S0360-3199(15)31138-1/sref30http://refhub.elsevier.com/S0360-3199(15)31138-1/sref30http://refhub.elsevier.com/S0360-3199(15)31138-1/sref30http://refhub.elsevier.com/S0360-3199(15)31138-1/sref30http://refhub.elsevier.com/S0360-3199(15)31138-1/sref30http://refhub.elsevier.com/S0360-3199(15)31138-1/sref31http://refhub.elsevier.com/S0360-3199(15)31138-1/sref31http://refhub.elsevier.com/S0360-3199(15)31138-1/sref31http://refhub.elsevier.com/S0360-3199(15)31138-1/sref31http://refhub.elsevier.com/S0360-3199(15)31138-1/sref31http://refhub.elsevier.com/S0360-3199(15)31138-1/sref31http://refhub.elsevier.com/S0360-3199(15)31138-1/sref32http://refhub.elsevier.com/S0360-3199(15)31138-1/sref32http://refhub.elsevier.com/S0360-3199(15)31138-1/sref32http://refhub.elsevier.com/S0360-3199(15)31138-1/sref32http://refhub.elsevier.com/S0360-3199(15)31138-1/sref32http://refhub.elsevier.com/S0360-3199(15)31138-1/sref32http://refhub.elsevier.com/S0360-3199(15)31138-1/sref32http://refhub.elsevier.com/S0360-3199(15)31138-1/sref32http://refhub.elsevier.com/S0360-3199(15)31138-1/sref32http://refhub.elsevier.com/S0360-3199(15)31138-1/sref32http://refhub.elsevier.com/S0360-3199(15)31138-1/sref33http://refhub.elsevier.com/S0360-3199(15)31138-1/sref33http://refhub.elsevier.com/S0360-3199(15)31138-1/sref33http://refhub.elsevier.com/S0360-3199(15)31138-1/sref33http://refhub.elsevier.com/S0360-3199(15)31138-1/sref33http://refhub.elsevier.com/S0360-3199(15)31138-1/sref33http://refhub.elsevier.com/S0360-3199(15)31138-1/sref34http://refhub.elsevier.com/S0360-3199(15)31138-1/sref34http://refhub.elsevier.com/S0360-3199(15)31138-1/sref34http://refhub.elsevier.com/S0360-3199(15)31138-1/sref34http://refhub.elsevier.com/S0360-3199(15)31138-1/sref34http://refhub.elsevier.com/S0360-3199(15)31138-1/sref34http://refhub.elsevier.com/S0360-3199(15)31138-1/sref34http://refhub.elsevier.com/S0360-3199(15)31138-1/sref35http://refhub.elsevier.com/S0360-3199(15)31138-1/sref35http://refhub.elsevier.com/S0360-3199(15)31138-1/sref35http://refhub.elsevier.com/S0360-3199(15)31138-1/sref35http://refhub.elsevier.com/S0360-3199(15)31138-1/sref35http://refhub.elsevier.com/S0360-3199(15)31138-1/sref36http://refhub.elsevier.com/S0360-3199(15)31138-1/sref36http://refhub.elsevier.com/S0360-3199(15)31138-1/sref36http://refhub.elsevier.com/S0360-3199(15)31138-1/sref36http://refhub.elsevier.com/S0360-3199(15)31138-1/sref36http://refhub.elsevier.com/S0360-3199(15)31138-1/sref37http://refhub.elsevier.com/S0360-3199(15)31138-1/sref37http://refhub.elsevier.com/S0360-3199(15)31138-1/sref37http://refhub.elsevier.com/S0360-3199(15)31138-1/sref37http://refhub.elsevier.com/S0360-3199(15)31138-1/sref37http://refhub.elsevier.com/S0360-3199(15)31138-1/sref37http://refhub.elsevier.com/S0360-3199(15)31138-1/sref37http://refhub.elsevier.com/S0360-3199(15)31138-1/sref38http://refhub.elsevier.com/S0360-3199(15)31138-1/sref38http://refhub.elsevier.com/S0360-3199(15)31138-1/sref38http://refhub.elsevier.com/S0360-3199(15)31138-1/sref38http://refhub.elsevier.com/S0360-3199(15)31138-1/sref38http://refhub.elsevier.com/S0360-3199(15)31138-1/sref39http://refhub.elsevier.com/S0360-3199(15)31138-1/sref39http://refhub.elsevier.com/S0360-3199(15)31138-1/sref39http://refhub.elsevier.com/S0360-3199(15)31138-1/sref39http://refhub.elsevier.com/S0360-3199(15)31138-1/sref39http://refhub.elsevier.com/S0360-3199(15)31138-1/sref40http://refhub.elsevier.com/S0360-3199(15)31138-1/sref40http://refhub.elsevier.com/S0360-3199(15)31138-1/sref40http://refhub.elsevier.com/S0360-3199(15)31138-1/sref40http://refhub.elsevier.com/S0360-3199(15)31138-1/sref40http://refhub.elsevier.com/S0360-3199(15)31138-1/sref40http://refhub.elsevier.com/S0360-3199(15)31138-1/sref40http://refhub.elsevier.com/S0360-3199(15)31138-1/sref41http://refhub.elsevier.com/S0360-3199(15)31138-1/sref41http://refhub.elsevier.com/S0360-3199(15)31138-1/sref41http://refhub.elsevier.com/S0360-3199(15)31138-1/sref41http://refhub.elsevier.com/S0360-3199(15)31138-1/sref41http://refhub.elsevier.com/S0360-3199(15)31138-1/sref41http://refhub.elsevier.com/S0360-3199(15)31138-1/sref41http://refhub.elsevier.com/S0360-3199(15)31138-1/sref41http://refhub.elsevier.com/S0360-3199(15)31138-1/sref41http://refhub.elsevier.com/S0360-3199(15)31138-1/sref42http://refhub.elsevier.com/S0360-3199(15)31138-1/sref42http://refhub.elsevier.com/S0360-3199(15)31138-1/sref42http://refhub.elsevier.com/S0360-3199(15)31138-1/sref42http://refhub.elsevier.com/S0360-3199(15)31138-1/sref43http://refhub.elsevier.com/S0360-3199(15)31138-1/sref43http://refhub.elsevier.com/S0360-3199(15)31138-1/sref43http://refhub.elsevier.com/S0360-3199(15)31138-1/sref43http://refhub.elsevier.com/S0360-3199(15)31138-1/sref43http://refhub.elsevier.com/S0360-3199(15)31138-1/sref43http://refhub.elsevier.com/S0360-3199(15)31138-1/sref43http://refhub.elsevier.com/S0360-3199(15)31138-1/sref43http://refhub.elsevier.com/S0360-3199(15)31138-1/sref43http://refhub.elsevier.com/S0360-3199(15)31138-1/sref43http://refhub.elsevier.com/S0360-3199(15)31138-1/sref44http://refhub.elsevier.com/S0360-3199(15)31138-1/sref44http://refhub.elsevier.com/S0360-3199(15)31138-1/sref44http://refhub.elsevier.com/S0360-3199(15)31138-1/sref44http://refhub.elsevier.com/S0360-3199(15)31138-1/sref45http://refhub.elsevier.com/S0360-3199(15)31138-1/sref45http://refhub.elsevier.com/S0360-3199(15)31138-1/sref46http://refhub.elsevier.com/S0360-3199(15)31138-1/sref46http://refhub.elsevier.com/S0360-3199(15)31138-1/sref46http://refhub.elsevier.com/S0360-3199(15)31138-1/sref46http://refhub.elsevier.com/S0360-3199(15)31138-1/sref46http://refhub.elsevier.com/S0360-3199(15)31138-1/sref47http://refhub.elsevier.com/S0360-3199(15)31138-1/sref47http://refhub.elsevier.com/S0360-3199(15)31138-1/sref47http://refhub.elsevier.com/S0360-3199(15)31138-1/sref47http://refhub.elsevier.com/S0360-3199(15)31138-1/sref48http://refhub.elsevier.com/S0360-3199(15)31138-1/sref48http://refhub.elsevier.com/S0360-3199(15)31138-1/sref48http://refhub.elsevier.com/S0360-3199(15)31138-1/sref48http://refhub.elsevier.com/S0360-3199(15)31138-1/sref48http://refhub.elsevier.com/S0360-3199(15)31138-1/sref48http://refhub.elsevier.com/S0360-3199(15)31138-1/sref48http://refhub.elsevier.com/S0360-3199(15)31138-1/sref48http://dx.doi.org/10.1016/j.ijhydene.2016.05.084http://dx.doi.org/10.1016/j.ijhydene.2016.05.084
Hydrogen production from catalytic steam reforming of glycerol
over various supported nickel
catalystsIntroductionExperimentalCatalyst preparationCatalyst
characterizationCatalyst performance testingExperimental design
Results and discussionCatalyst characterizationX-ray diffraction
(XRD)BET surface areaScanning electron microscopy (SEM)
Catalyst screening performanceModel analysisVariables effects on
the responsesOptimization of hydrogen selectivity
ConclusionAcknowledgementsReferences