13
ORIGINAL PAPER Hydrogen Production from Glycerol Over Nickel Catalysts Supported on Al 2 O 3 Modified by Mg, Zr, Ce or La A. Iriondo V. L. Barrio J. F. Cambra P. L. Arias M. B. Gu ¨emez R. M. Navarro M. C. Sa ´nchez-Sa ´nchez J. L. G. Fierro Published online: 23 April 2008 Ó Springer Science+Business Media, LLC 2008 Abstract Hydrogen production from glycerol reforming in liquid (aqueous phase reforming, APR) and vapor (steam reforming SR) phase over alumina-supported nickel cata- lysts modified with Ce, Mg, Zr and La was studied. Characterization of catalysts by temperature programmed reduction and XPS analyses revealed important structural effects: (i) the intercalation of Mg between nickel and alumina that inhibited the alumina incorporation to nickel phases, (ii) the close contact between Ni and Zr phases and, (iii) the close surface interaction of La and Ce ions with NiO phases. The catalytic activity of the samples studied in this work clearly indicated the different catalyst function- alities necessary to carry out aqueous-phase and vapor- phase steam reforming of glycerol. For aqueous phase reforming of glycerol, the addition of Ce, La and Zr to Ni/ Al 2 O 3 improves the initial glycerol conversions obtained over the Ni/Al 2 O 3 supported catalyst. It is suggested that the differences in catalytic activities are related with geo- metric effects caused by the decoration of Ni phases by Ce and La or by the close interaction between Ni and Zr. In spite that nickel catalysts showed high APR activities at initial times on stream, all samples showed, independently of support, important deactivation rates that deactivate the catalysts after few hours under operation. Catalysts char- acterization after APR showed the oxidation of the active metallic Ni during reaction as the main cause of the observed deactivation. In the case of the glycerol steam reforming in vapor phase, the use of Ce, La, Mg and Zr as promoters of Ni based catalysts increases the hydrogen selectivity. Differences in activity were explained in terms of enhancement in: surface nickel concentration (Mg), capacity to activate steam (Zr) and stability of nickel phases under reaction conditions (Ce and La). Keywords Glycerol Nickel catalyst Steam reforming Aqueous phase reforming Ceria Magnesia Lanthana Zirconia 1 Introduction The diminution of fossil fuel reserves and the pollution caused by the continuous energy demands make hydrogen an attractive alternative energy vector, especially if it is produced from renewable resources. Nowadays biodiesel has become one of the more promising carbon neutral bio- fuels. However the cost of biodiesel is the main barrier for this product commercialization. The recovery of biodiesel co-products, like the glycerol, is one of the main options to be considered to lower the overall cost of the biodiesel production [1]. Thus, in recent years, many studies have focused in the recovery of high quality glycerol by-product. In this context, the glycerol may represent a potential source for hydrogen production. Glycerol can be efficiently converted in hydrogen by means of its catalytic reaction with steam according to the following reaction: C 3 H 8 O 3 þ 3H 2 O ! 3CO 2 þ 7H 2 ð1Þ The steam reforming of hydrocarbons (SR) is a catalytic process that typically takes place in vapour phase at A. Iriondo V. L. Barrio J. F. Cambra P. L. Arias M. B. Gu ¨emez Department of Chemical and Environmental Engineering, University of the Basque Country (School of Engineering), C/Alameda Urquijo s/n, 48013 Bilbao, Spain R. M. Navarro (&) M. C. Sa ´nchez-Sa ´nchez J. L. G. Fierro Instituto de Cata ´lisis y Petroleoquı ´mica, CSIC, C/Marie Curie 2 Cantoblanco, 28049 Madrid, Spain e-mail: [email protected] 123 Top Catal (2008) 49:46–58 DOI 10.1007/s11244-008-9060-9

Hydrogen Production from Glycerol Over Nickel Catalysts Supported on Al2O3 Modified by Mg, Zr, Ce or La

Embed Size (px)

Citation preview

ORIGINAL PAPER

Hydrogen Production from Glycerol Over Nickel CatalystsSupported on Al2O3 Modified by Mg, Zr, Ce or La

A. Iriondo Æ V. L. Barrio Æ J. F. Cambra Æ P. L. Arias ÆM. B. Guemez Æ R. M. Navarro Æ M. C. Sanchez-Sanchez ÆJ. L. G. Fierro

Published online: 23 April 2008

� Springer Science+Business Media, LLC 2008

Abstract Hydrogen production from glycerol reforming

in liquid (aqueous phase reforming, APR) and vapor (steam

reforming SR) phase over alumina-supported nickel cata-

lysts modified with Ce, Mg, Zr and La was studied.

Characterization of catalysts by temperature programmed

reduction and XPS analyses revealed important structural

effects: (i) the intercalation of Mg between nickel and

alumina that inhibited the alumina incorporation to nickel

phases, (ii) the close contact between Ni and Zr phases and,

(iii) the close surface interaction of La and Ce ions with

NiO phases. The catalytic activity of the samples studied in

this work clearly indicated the different catalyst function-

alities necessary to carry out aqueous-phase and vapor-

phase steam reforming of glycerol. For aqueous phase

reforming of glycerol, the addition of Ce, La and Zr to Ni/

Al2O3 improves the initial glycerol conversions obtained

over the Ni/Al2O3 supported catalyst. It is suggested that

the differences in catalytic activities are related with geo-

metric effects caused by the decoration of Ni phases by Ce

and La or by the close interaction between Ni and Zr. In

spite that nickel catalysts showed high APR activities at

initial times on stream, all samples showed, independently

of support, important deactivation rates that deactivate the

catalysts after few hours under operation. Catalysts char-

acterization after APR showed the oxidation of the active

metallic Ni during reaction as the main cause of the

observed deactivation. In the case of the glycerol steam

reforming in vapor phase, the use of Ce, La, Mg and Zr as

promoters of Ni based catalysts increases the hydrogen

selectivity. Differences in activity were explained in terms

of enhancement in: surface nickel concentration (Mg),

capacity to activate steam (Zr) and stability of nickel

phases under reaction conditions (Ce and La).

Keywords Glycerol � Nickel catalyst � Steam reforming �Aqueous phase reforming � Ceria � Magnesia � Lanthana �Zirconia

1 Introduction

The diminution of fossil fuel reserves and the pollution

caused by the continuous energy demands make hydrogen

an attractive alternative energy vector, especially if it is

produced from renewable resources. Nowadays biodiesel

has become one of the more promising carbon neutral bio-

fuels. However the cost of biodiesel is the main barrier for

this product commercialization. The recovery of biodiesel

co-products, like the glycerol, is one of the main options to

be considered to lower the overall cost of the biodiesel

production [1]. Thus, in recent years, many studies have

focused in the recovery of high quality glycerol by-product.

In this context, the glycerol may represent a potential

source for hydrogen production. Glycerol can be efficiently

converted in hydrogen by means of its catalytic reaction

with steam according to the following reaction:

C3H8O3 þ 3H2O! 3CO2 þ 7H2 ð1Þ

The steam reforming of hydrocarbons (SR) is a catalytic

process that typically takes place in vapour phase at

A. Iriondo � V. L. Barrio � J. F. Cambra �P. L. Arias � M. B. Guemez

Department of Chemical and Environmental Engineering,

University of the Basque Country (School of Engineering),

C/Alameda Urquijo s/n, 48013 Bilbao, Spain

R. M. Navarro (&) � M. C. Sanchez-Sanchez � J. L. G. Fierro

Instituto de Catalisis y Petroleoquımica, CSIC, C/Marie Curie 2

Cantoblanco, 28049 Madrid, Spain

e-mail: [email protected]

123

Top Catal (2008) 49:46–58

DOI 10.1007/s11244-008-9060-9

atmospheric pressure and temperatures around 1073 K.

Nevertheless it has been recently reported [2] the possibility

to obtain hydrogen from oxygenated hydrocarbons having a

C:O stoichiometry of 1:1 at low temperature (500 K) and

high pressure (2–5 MPa) by aqueous-phase reforming

(APR).

Catalysts for steam reforming of hydrocarbons are

mainly based on nickel as active component supported on

oxides with high thermal stability [3]. Although noble

metals (Ru, Rh, Pt) are more effective for the steam

reforming of hydrocarbons than Ni and less susceptible to

carbon formation, such catalysts are not common in

industrial applications because of their cost [3]. Effective

catalyst for production of hydrogen by aqueous phase

reforming of oxygenated hydrocarbons must break C–C,

O–H and C–H bonds in the oxygenated hydrocarbon

reactant and facilitate the water–gas shift reaction to

remove adsorbed CO from the surface. Studies on aqueous-

phase reforming of oxygenated hydrocarbons over various

supported metals [4, 5] indicate that Pt and Pd catalysts are,

especially in the case of Pt, active and selective for the

production of hydrogen. Nevertheless, the high cost and

limited availability of noble metals make it of particular

interest to develop less expensive catalyst for aqueous-

phase reforming. From the above reasons, in the present

work nickel was selected as active phase to be included in

catalyst formulations applied to glycerol reforming.

The structural characteristics and performance of sup-

ported nickel catalysts are strongly influenced by the nature

of support where the metallic crystallites are deposited.

The use of supports with high thermal stability that stabi-

lizes nickel particles against sintering and that promotes the

carbon gasification are necessary to develop catalysts with

high activity and stability in the steam reforming of glyc-

erol. To provide these operational advantages different

promoters are usually included in the reforming catalyst

formulations. Magnesium is widely used as promoter in Ni/

Al2O3 catalysts applied to steam reforming of hydrocar-

bons since it enhances both the steam adsorption capability

and the stability of nickel against sintering [6, 7]. The

addition of ZrO2 to reforming catalysts is also claimed in

literature [8] as an element able to improve the stability of

nickel catalysts in the steam reforming of hydrocarbons.

Ceria [9] and lanthana [10] are also found in several for-

mulations of Ni steam reforming catalysts since they are

known promoters of carbon removal from metallic sur-

faces. For aqueous-phase reforming catalysts, there are no

studies reporting the effect of the addition of Mg, Zr, Ce

and La on the reforming chemistry over nickel catalysts

being therefore necessary to be explored.

With this background, the aim of the present work was

to study the role and effect of La, Mg, Zr and Ce on

the behaviour of alumina-supported Ni catalysts in the

production of hydrogen by both aqueous-phase reforming

and vapour phase steam reforming of glycerol. Careful

investigations of the structure of the catalysts were per-

formed in an attempt to understand the relationship

between activity and their structural and surface

characteristics.

2 Experimental

2.1 Supports and Catalysts Preparation

MxOy–Al2O3 supports (M=Zr, Ce, La or Mg) were pre-

pared by impregnation of a commercial c-Al2O3 (Alfa

Aesar, SBET = 212 m2/g) with aqueous solutions of metal

nitrates. MxOy loading in the different supports were

selected in order to achieve 0.3 theoretical monolayers on

alumina carrier. The impregnated solids were dried under

air at 393 K for 3 h and subsequently calcined in air at

923 K for 6 h (except for ZrO2–Al2O3 support which was

calcined at 823 K). The supports were designated as A–M

where M refers to the oxide (C=CeO2, M=MgO, Z=ZrO2

and L=La2O3) added to Al2O3 carrier.

Supported Ni catalysts were prepared by impregnation

under stirring at 333 K of each of the above supports using

aqueous solutions of Ni(NO3)2. Ni loading in the different

catalysts was selected in order to achieve 0.75 theoretical

monolayers over supports. After Ni loading, the samples

were dried at 393 K for 2 h and subsequently calcined

in air at 773 K for 4 h. The catalysts were designated as

Ni/A–M.

2.2 Catalyst Characterization

The chemical composition of the catalysts was determined

by inductively coupled plasma atomic emission spectros-

copy (ICP-AES), using a Perkin-Elmer Optima 3300DV

apparatus. The samples were first dissolved in acid solu-

tions (a mixture of HF, HCl and HNO3), microwaved for

15 min, and diluted to concentrations within the detection

range of the instrument.

N2 adsorption–desorption isotherms were obtained at

77 K over the whole range of relative pressures using a

Micromeritics ASAP 2100 automatic device on samples

previously outgassed at 423 K for 12 h. BET specific areas

were calculated from these isotherms using the BET

method and taking a value of 0.162 nm2 for the cross-

section of the physically adsorbed N2 molecule.

Temperature-programmed reduction experiments were

carried out with a semiautomatic Micromeritics TPD/TPR

2900 apparatus equipped with a TC detector. Prior to

reduction experiments, the samples, about 30 mg, were

treated thermally under an air stream at 573 K to remove

Top Catal (2008) 49:46–58 47

123

water and other contaminants. TPR profiles were obtained

by heating the samples under a 10% H2/Ar flow (50 mL/

min) from 298 to 1173 K at a linearly programmed rate of

10 K/min. The effluent gas was passed through a cold trap

to remove water before measuring the amount of hydrogen

consumed during reduction by the TC detector.

X-ray powder diffractograms were recorded following

the step-scanning procedure (step size 0.02�, 2h scanning

from 20 to 80�) using a computerised Seifert XRD 3000P

diffractometer (Cu Ka radiation, k = 0.15418 nm) equip-

ped with a PW Bragg-Brentano h/2h goniometer and a bent

graphite monochromator and automatic slit. Volume-

averaged crystallite sizes were determined by applying the

Debye-Scherrer equation.

X-ray photoelectron spectroscopy (XPS) was used to

study the chemical composition and oxidation state of the

catalyst surfaces. Photoelectron spectra were recorded with

a VG Escalab 200R electron spectrometer equipped with a

Mg Ka X-ray source (hm = 1253.6 eV) and a hemispher-

ical electron analyser operating at constant transmission

energy (50 eV). The X-ray source was operated at low

X-ray fluxes in order to minimize X-ray-induced reduction

of Ce species. The C 1s, Al 2p, Zr 3d, Mg 2p, La 3d, Ce 3d

and Ni 2p core-level spectra were recorded and the corre-

sponding binding energies were referenced to the C 1s line

at 284.6 eV (accuracy within ± 0.1 eV). Due to the partial

overlapping of La 3d5/2 and Ni 2p3/2 signals, the chemical

state of nickel element was elucidated after carefully sub-

traction of La 3d5/2 in the Ni 2p3/2 peak. The reduction

treatment was carried out ex situ at 923 K in H2/N2 (1/9

vol.) flow for 90 min followed by reduction in situ at

773 K for 30 min. After outgassing at 10-5 mbar, the

calcined or reduced samples were transferred to the ion-

pumped analysis chamber, whose residual pressure was

kept below 7 9 10-9 mbar during data acquisition. The C

1s peak at 284.6 eV was used as an internal standard for

peak position measurement. The areas of the peaks were

estimated by calculating the integral of each peak after

subtracting a Shirley background and fitting the experi-

mental peak to a combination of Lorentzian/Gaussian lines

of variable proportions.

2.3 Activity Tests

The catalytic tests for the steam reforming (SR) or aque-

ous-phase reforming (APR) of glycerol were carried out in

a bench-scale unit equipped with a stainless steel fixed-bed

catalytic reactor. In order to avoid preferential gas flow

paths and hot spots, 200 mg of catalyst were diluted with

SiC (1:9 w/w), both in the 0.42–0.5-mm particle size

range. Each catalyst was activated in situ by reduction at

atmospheric pressure under H2 flow (75 mL/min) at 973 K

for 2 h. After catalyst activation, the activity measurements

were performed. The feed was 1 wt% of glycerol in water.

Aqueous phase reforming measurements were performed at

3 MPa of total pressure, at 498 K and with a WHSV equal

to 1.25 h-1. Steam reforming measurements were con-

ducted at atmospheric pressure, at 873 K and with a

WHSV equal to 2.5 h-1. The reaction was kept running at

this space velocity until steady-state was reached. Liquid

products was collected and analyzed by a GC equipped

with a MS detector and CHN-analyzer. The gas product

was analyzed on-line by a GC equipped with a FID and a

TCD detectors.

3 Results

3.1 Chemical Composition and Textural Properties

Table 1 shows the chemical compositions obtained from

ICP analyses, expressed as weight percentages, of the

calcined catalysts. The textural properties of all catalysts

and supports were evaluated from nitrogen adsorption–

desorption isotherms. The specific surface areas of the

catalysts and supports together with the pore volume and

average pore diameter are summarized in Table 2. For the

sake of a valid comparison, the N2 adsorption–desorption

data were normalized to unit weight of Al2O3 carrier.

Textural data in Table 2 show that the surface area of

Al2O3 was almost constant after the incorporation of Ce

and Mg and increases for lanthanum and zirconium con-

taining supports. The increase in surface area observed in

the case of A–L and A–Z supports may be related to the

known capacity of La [11] and Zr [12] to stabilize the

surface area of c-Al2O3 against the thermal sintering

induced by the treatments done in the calcination step.

The specific area and pore volume data for the Ni-loa-

ded samples indicate that the surface area of the bare

MxOy–Al2O3 supports does not change significantly upon

incorporating Ni to the supports. The almost constant sur-

face area after Ni loading is probably indicating a slight

increase in the external area associated to NiO phases

which is not counterbalanced by the simultaneous filling of

the pores by nickel species as indicated by the slight

decrease in pore volume.

Table 1 Chemical composition (wt%) of calcined catalysts

Sample NiO ZrO2 CeO2 La2O3 MgO Al2O3

Ni/A 16.0 – – – – Balance

Ni/A–Z 17.0 7.0 – – – Balance

Ni/A–C 16.0 – 8.4 – – Balance

Ni/A–M 16.2 – – – 2.5 Balance

Ni/A–L 17.0 – – 5.0 – Balance

48 Top Catal (2008) 49:46–58

123

3.2 X-Ray Diffraction (XRD)

Figure 1 shows the X-ray diffraction patterns of the fresh

calcined supports. The characteristic peaks of poorly

crystalline c-Al2O3 (at 46.8� and 66.7 JCPDS 86-1410)

were observed in the bare alumina support (Fig. 1). The

XRD pattern of the La, Zr and Mg supports (A–L, A–Z and

A–M in Fig. 1) exhibited lines only attributable to the bare

c-Al2O3 support. The Ce-containing support was the

only modified alumina that showed diffraction peaks

characteristics of bulk crystalline CeO2 besides the Al2O3

phases of bare alumina.

XRD patterns of calcined and reduced catalysts depos-

ited on MxOy–Al2O3 supports are shown in Fig. 2. As

Fig. 2a shows, calcined catalysts displayed the reflections

already detected in Ni-free supports and new reflections

associated to nickel phases. All calcined catalysts exhibited

diffraction lines at 37.2� and 43.3� associated to crystalline

phases of NiO (JCPDS 44-1159). Calculated crystallite size

of the NiO phase applying the Scherrer equation are

summarized in Table 3. The size of the NiO crystallites

varies with the support and decreases following the trend:

Ni/A–Z & Ni/A & Ni/A–C [ Ni/A–M [ Ni/A–L.

The X-ray diffractograms of reduced catalysts supported

on MxOy–Al2O3 supports (after treatment under H2/N2 (1/

9 vol, 100 mL/min at 923 K for 2 h)) are displayed in

Fig. 2b. XRD patterns of reduced catalysts in Fig. 2b

showed the diffraction line corresponding to the (200)

reflection of Ni0 phase (at 2H angle of 51.8� JCPD

04-850). No diffraction lines ascribed to NiO (at 2H angles

Table 2 Textural properties of calcined supports and catalysts as

measured by N2 adsorption desorption isotherms at 77 K

SBET (m2/gAl2O3) Vpore (cm3/gAl2O3) Dpore (nm)

A 212 0.81 15.4

A–Z 229 0.81 14.2

A–C 217 0.80 14.7

A–M 214 0.80 14.9

A–L 225 0.81 14.3

Ni/A 207 0.77 14.8

Ni/A–Z 229 0.81 14.0

NiA–C 215 0.81 14.3

Ni/A–M 218 0.82 14.1

Ni/A–L 225 0.81 14.3

20 40 60 80

Inte

nsi

ty (

au)

2 θ°

A-M

A-L

A-Z

A

A-C

Fig. 1 XRD patterns of calcined supports (j Al2O3, d CeO2)

20 40 60 80

a

Ni/A

Ni/A-L

Ni/A-M

Ni/A-Z

Ni/A-C

o

o

Inte

nsi

ty (

a.u

.)

o

o

o

oo

o

o

o

o

o

Ni/A

Inte

nsi

ty (

a.u

.)

Ni/A-L

2 θ°

20 40 60 802 θ°

Ni/A-M

Ni/A-Z

Ni/A-C

b

Fig. 2 XRD patterns of Ni catalysts after: (a) calcination in air at

773 K and (b) reduction under H2/N2 flow at 923 K (¤ NiO, o Ni0, d

CeO2 , j Al2O3)

Top Catal (2008) 49:46–58 49

123

of 37.2 and 43.3) were found in these reduced samples.

Quantitative estimation of crystallite sizes of Ni0 phase by

applying the Scherrer equation (Table 3) indicates the

larger crystal size for Ni/A–Z than that of Ni/A, Ni/A–C

and Ni/A–M counterparts. It is interesting to note the

absence of any diffraction peak associated to reduced

phases of nickel for the catalysts supported on lanthana-

loaded alumina (Ni/A–L).

3.3 Temperature Programmed Reduction (TPR)

The TPR profiles corresponding to Ni catalysts deposited

on MxOy–Al2O3 supports are depicted in Fig. 3. Quantifi-

cation of TPR data are summarized in Table 4. TPR data

showed differences in the relative proportion of nickel

species depending on the support used to disperse the

nickel entities. As it is observed in Fig. 3, the catalyst

supported on pure Al2O3 exhibits a broad reduction peak in

the 650 and 1,150 K range which could be deconvoluted

into three components at reduction temperatures of 818,

942, and 1041 K. According to literature studies, the

reduction peak at 818 K is attributed to reduction of NiO

species with weak interaction with alumina support [13]

while the reduction peaks appearing at higher temperatures

were related to the reduction of highly dispersed non-

stoichiometric amorphous nickel aluminate spinels and to a

diluted NiAl2O4-like phase respectively [14]. The incor-

poration of Mg to Al2O3 (Ni/A–M in Fig. 3) provokes a

slight increase in the intensity of the reduction peak cor-

responding to the NiO species with weak interaction with

support (peak at 822 K). From data in Table 4, the pres-

ence of Mg also increases the proportion of highly

dispersed non-stoichiometric amorphous nickel aluminate

(peak at 943 K) at the expense of the diluted NiAl2O4

phase (peak at 1068 K). Similar reduction behaviour was

observed for nickel supported on La-containing Al2O3 (Ni/

A–L in Fig. 3). The Zr-containing catalyst (Ni/A–Z in

Fig. 3) showed hydrogen uptakes at 829, 935 and 1028 K.

The peak at 829 K, ascribed to NiO particles with low

interaction with support, showed a similar shape and

intensity than those corresponding to Mg- and La-loaded

catalysts. However, with the addition of Zr to Al2O3, the

reduction peaks in the high-temperature range increases

considerably in intensity (Table 4). Similar reduction

behaviour was observed for the sample including Ce in the

composition. For this sample, the low intensity of the peak

at 822 K (Table 4) indicates the low concentration of

nickel oxide species weakly interacting with aluminium

ions in this sample. The reduction peak at 958 K, that

includes the reduction of non stoichiometric nickel spinel

species and the bulk reduction of CeO2, showed the highest

intensity among the samples studied.

3.4 X-Ray Photoelectron Spectroscopy (XPS)

The binding energies of core-electrons and the surface

atomic ratios of the calcined supports are summarized in

Table 5. Zr containing support (A–Z in Table 5) shows a

single component in the Zr 3d5/2 level at 181.8 eV char-

acteristic of Zr4+ ions in an oxygen environment [15].

Surface Zr/Al ratio (Table 5) was higher than nominal Zr/

Al ratio indicating that most of the Zr has been supported

on the surface of Al2O3. Mg added support presents a

single component photoelectron Mg 2p peak at 50.3 eV.

This binding energy is close to that reported in literature for

MgO (50.2 eV) or MgAl2O4 (50.2 eV) [16]. Surface XPS

Mg/Al ratio was higher than the corresponding nominal

Table 3 Average size of NiO

and Ni0 crystalline particles

from XRD data of calcined and

reduced Ni catalysts

Calcined

NiO (nm)

Reduced

Ni0 (nm)

Ni/A 10 7

Ni/A–Z 10 9

Ni/A–C 10 6

Ni/A–M 7 7

Ni/A–L 5 n.d.

200 400 600 800 1000 1200 1400

H2

con

sum

pti

on

(a.

u.)

Ni/A

Ni/A-L

Temperature (K)

Ni/A-C

Ni/A-Z

Ni/A-M

Fig. 3 Temperature-programmed reduction profiles of Ni catalysts

(10% vol H2/Ar, heating rate 10 K/min)

50 Top Catal (2008) 49:46–58

123

one indicating, as occurred previously for Zr, the disper-

sion, without diffusion, of Mg across the Al2O3 surface.

For La-containing alumina, the La 3d5/2 peak envelopes are

centred at a binding energy of 835.7 eV. This BE is higher

than the characteristic value for La2O3 (834.3 eV) and

LaAlO3 (833.8) [17, 18] and can be assigned to well-dis-

persed lanthanum species on alumina [19]. The La/Al

atomic ratio obtained from XPS (Table 5) was similar to

the values corresponding to the bulk composition. This fact

may be indicative of diffusion at surface level of lanthanum

ions into the alumina framework [19]. The calculated XPS

Ce/Al atomic ratio for the A–C support (Table 5) was close

to the expected value derived from its chemical composi-

tion (0.033). This fact may indicate some diffusion of Ce

on Al2O3 but as previously commented, the low tempera-

ture of calcination used in the preparation and the presence

of CeO2 entities detected by XRD discards this possibility.

The presence of this agglomeration of CeO2 particles may

be the reason of the low Ce/Al ratio since Ce atoms in the

bulk particles will not contribute to the Ce signal.

The characteristic Ni 2p3/2 levels for the calcined cata-

lysts are displayed in Fig. 4a. The binding energies of Ni

2p3/2 peaks and surface atomic concentrations are sum-

marized in Table 6. As it is seen in Fig. 4a, all calcined

catalysts show the main line of Ni 2p3/2 level at 856.4 eV.

This binding energy was slightly higher than the value

reported for pure NiO species (854.4 eV) and close to

nickel aluminate (855.6 eV). Calculated XPS Ni/Al and M/

Al (M=Zr, Ce, Mg and La) surface atomic ratios in Table 6

indicate differences in the relative metal exposition

depending on the support used. The relation between Ni/Al

surface ratio decreases according the sequence: Ni/A–L [Ni/A–M [ Ni/A [ Ni/A–Z [ Ni/A–C. Regarding the

support additives Zr and Mg, their XPS M/Al surface ratio

decreased after Ni deposition respect to the values

observed on the bare support. The loss of surface exposure

of additives on supports after nickel deposition was much

lower in the case of cerium and negligible in the case of

lanthanum.

Chemical changes in the catalysts after reduction in H2

at 923 K were also investigated by XPS and the results are

summarized in Table 7. Figure 4b shows the Ni 2p3/2 core

level spectra of catalysts after reduction. Reduced catalysts

showed binding energies characteristic of Ni0 (852.6 eV)

and Ni2+ ions in nickel aluminate entities (856.2 eV). As it

can be observed in Table 7, it was clear that none of the

samples were completely reduced after the reduction pro-

tocol applied. This fact confirms the difficulties observed in

TPR for the reduction of nickel ions with interactions with

ions from the supports. After reduction, all the catalysts

showed a strong decrease in the intensity of the Ni 2p3/2

core level that provokes a decrease in the surface exposi-

tion of metallic Ni as it was derived from the Ni/Al surface

ratios presented in Table 7. The loss in Ni XPS signal after

reduction observed for all the analyzed catalysts may be

indicative of sintering phenomena of nickel particles dur-

ing the reduction process and/or ‘decoration’ of metallic

particles by dispersed species from support. Surface

exposition of metallic nickel decreases in the order: Ni/A–C

[ Ni/A–M [ NiA–Z & Ni/A. In the case of Ni/A–L

catalysts the Ni 2p XPS level signal was very weak after

reduction and hence not quantification was done.

3.5 Aqueous Phase Reforming (APR) and Steam

Reforming (SR) Activity Tests

The conversion of glycerol in the aqueous phase reforming

over nickel based catalysts is presented in Fig. 5. Table 8

shows gaseous and liquid product compositions corre-

sponding to the 2.5 hour on stream as representative values

Table 4 Quantitative TPR data of calcined catalysts

H2/Ni Free NiO NiO–Al Surface NiAl2O4 NiAl2O4

Ni/A 0.79 – 809 K––62% 940 K––24% 1,043 K––13%

Ni/A–Z 1.14a – 829 K––50% 935 K––24%b 1,028 K––25%b

Ni/A–C 0.86b 539 K––2.3% 822 K––33% 958 K––48%a 1,062 K––15%

Ni/A–M 0.88 – 822 K––64% 943 K––29% 1,069 K––7%

Ni/A–L 0.92 – 808 K––61% 929 K––30% 1,054 K––9%

a Includes partial surface zirconia reductionb Includes ceria reduction

Table 5 Binding energies (eV) of core electrons and surface atomic

ratios of calcined supports

Al

2p

La

3d5/2

Mg

2p

Zr

3d5/2

Ce 3d5/2

u0 0 0 (% in

Ce 3d)

M/Al at

A–Z 74.5 181.9 0.090 (0.035)a

A–C 74.5 883.0 (12.1%) 0.033 (0.033)a

A–M 74.5 50.3 0.065 (0.039)a

A–L 74.5 835.7 0.021 (0.020)a

a In parenthesis the value corresponding to bulk M/Al ratio

Top Catal (2008) 49:46–58 51

123

of the first hours in the aqueous phase reforming of glyc-

erol over the different tested catalysts. In spite of the low

stability under reaction conditions showed by the catalysts,

differences in glycerol conversion at the initial activity

measurements were observed for the different nickel cat-

alysts. As Fig. 5 shows, initial glycerol conversions were

found to decrease following the sequence: Ni/A–L [ Ni/

A–C & Ni/A–Z [ Ni/A & Ni/A–M. However, all the

catalysts exhibit severe deactivation phenomena. For all

catalysts the glycerol conversion falls below 3% after 30 h

on-stream. When comparing the initial gas product com-

position of the various catalysts shown in Table 8, the

845 850 855 860 865 870 875 845 850 855 860 865

BE (eV)

Ni/A

Ni/A-M

Ni/A-C

Ni/A-Z

BE (eV)

Ni/A-L

a b

cou

nts

per

sec

on

d (

a.u

.)

Ni 2p3/2 Ni 2p

3/2

Ni/A

Ni/A-M

Ni/A-C

Ni/A-Z

cou

nts

per

sec

on

d (

a.u

.)

Fig. 4 Ni 2p3/2 photoelectron

spectra of Ni catalysts: (a) fresh

calcined in air at 773 K and (b)

after reduction in H2/N2 flow at

923 K

Table 6 Binding energy (eV) of Ni 2p3/2 core electrons and surface

atomic ratios of calcined catalysts

Ni 2p3/2 (eV) Ni/Al M/Al (M=Zr, Ce, Mg, La)

XPS Nominal XPS Nominal

Ni/A 856.4 0.110 0.130

Ni/A–Z 856.4 0.123 0.153 0.026 (0.090)a 0.035

Ni/A–C 856.4 0.115 0.148 0.029 (0.033)a 0.033

Ni/A–M 856.4 0.149 0.136 0.053 (0.065)a 0.039

Ni/A–L 856.2 0.172 0.147 0.020 (0.021)a 0.020

a In parentheses values corresponding to bare supports

Table 7 Binding energy (eV) of Ni 2p3/2 core electrons and surface

atomic ratios of reduced catalysts

Ni 2p3/2 (eV) Ni/Al M/Al (M=Zr, Ce,

Mg, La)

XPS Nominal XPS Nominal

Ni/A 852.7 (30.6%) 0.026 0.130

856.4 (69.4%)

Ni/A–Z 852.0 (50.1%) 0.016 0.153 0.019 0.035

856.5 (49.9%)

Ni/A–C 852.4 (47.2%) 0.02 0.148 0.026 0.033

856.4 (52.8%)

Ni/A–M 852.3 (31.5%) 0.028 0.136 0.038 0.039

856.3 (66.5%)

Ni/A–L n.d. – – 0.022 0.020

0

20

40

60

0 5 10 15 20 25 30

Time on stream (h)

Gly

cero

l co

nve

rsio

n (

mo

l %)

Fig. 5 Glycerol conversion as a function of time-on-stream (TOS

(h)) for aqueous phase reforming of glycerol over nickel catalysts: (¤)

Ni/A–Z, (j) Ni/A–C, (D) Ni/A–M, (d) Ni/A–L, (9) Ni/A.

(T = 498 K, P = 3 MPa, WHSV = 1.25 h-1)

52 Top Catal (2008) 49:46–58

123

catalysts supported on Ce- and La–Al2O3 exhibit relatively

higher rates of H2 production. The catalyst supported on

Mg–Al2O3 presents a higher hydrogen molar composition

although its glycerol conversion is rather low. For all the

catalysts the gas product is saturated by water vapour and

contains no significant CO amounts. Analyses of the

effluent liquid from the aqueous-phase reforming of glyc-

erol showed, for all the catalysts tested, the presence of

propylenglycol and ethylenglycol as the main products.

In a second set of experiments, the activity of these

nickel catalysts was also evaluated in the steam reforming

of glycerol performed, as indicated in the experimental

section, at atmospheric pressure and 873 K. Figure 6 pre-

sents the product composition of the glycerol steam

reforming over the different nickel based catalysts. For all

catalysts, glycerol conversion was 100 % and hydrogen

was the major component in the gas phase. From the

comparison of the molar fractions towards the various

reaction products (Fig. 6) it can be seen that the Ni/A–Z

sample was the most selective catalyst with a gas

composition after steam reforming similar to the values

predicted by equilibrium. On the contrary, the catalyst

supported on bare alumina showed the lower hydrogen

selectivity among the tested samples. Compared to the Ni/

A–Z catalyst, lower hydrogen selectivities were found for

the catalysts supported on alumina modified with Mg (Ni/

A–M), Ce (Ni/A–C) and La (Ni/A–L). Among the results

of these last samples, slight differences in hydrogen

selectivity were observed, showing the Ni/A–M and Ni/

A–C samples the higher and lower H2 selectivities,

respectively.

Long-term activity test was performed for the Zr-sup-

ported catalysts in order to study its stability under

reforming conditions. The evolution of gaseous products as

a function of reaction time is presented in Fig. 7. It can be

see that in spite that total glycerol conversion maintains

with reaction time on Ni/A–Z catalyst, the product distri-

bution changes with time-on-stream. The reaction starts

with high conversion and selectivity to H2 and CO2 but the

product distribution changes in the first hours on stream

with a decrease in the hydrogen formation and concomitant

CO production. Nevertheless, after several hours under

reaction stream the reverse was observed with increasing

CO2 formation and decreasing of CO levels. After this

change, the gas composition maintains stable for the last

25 h of the test: high H2 selectivity and low CO contents.

3.6 Characterization of Used Catalysts

XPS analyses of Ni/A–Z catalyst used in the glycerol

reforming in both aqueous and vapour phase were also per-

formed in order to know the evolution of active phases after

reaction. Binding energy for Ni 2p3/2 core of Ni/A–Z catalyst

used in the aqueous phase reforming of glycerol appears at

Table 8 Glycerol conversion and product compositions in the

aqueous phase reforming of glycerol over nickel catalysts after 2.5 h

on stream (T = 498 K, P = 3 MPa, WHSV = 1.25 h-1)

Ni/A Ni/A–Z Ni/A–C Ni/A–M Ni/A–L

Glycerol conversion

(%)

25 30 36 15 37

Gas product (%)

H2 32 32 48 32

CO2 31 42 40 40

CO – – – –

CH4 37 26 12 28

C2+ – – – –

Liquid products (%)

Propylenglycol 66 75 57 57

Ethylenglycol 24 15 33 34

Acetol 3 1 7 6

Ethanol 7 8 3 3

0%

20%

40%

60%

80%

100%

Ni/A-M Ni/A-L Ni/A-C Ni/A-Z Ni/A Equilibrium

Pro

du

ct d

istr

ibu

tio

n (

mo

lar

%)

H2 CO2

CO CH4

Fig. 6 Product distribution for steam reforming of glycerol over Ni

catalysts (T = 873 K, P = 0.1 MPa, WHSV = 2.5 h-1)

0%

20%

40%

60%

80%

100%

0 10 20 30 40 50 60

Time on stream (h)

Co

nve

rsio

n, p

rod

uct

dis

trib

uti

on

(m

ol %

)

Fig. 7 Glycerol conversion and product distribution as a function of

time-on stream (TOS (h)) for steam reforming glycerol over the Ni/

A–Z catalyst ((X) glycerol conversion, (j) H2, (¤) CO2, (D) CH4,

(d) CO, T = 873 K, P = 0.1 MPa, WHSV = 2.5 h-1)

Top Catal (2008) 49:46–58 53

123

856.4 eV (Fig. 8). This binding energy, as it was previously

indicated, corresponds to Ni2+ ions in species close to nickel

aluminate structures. Calculated XPS Ni/Al ratio for the

sample used in the liquid reforming revealed an increase in

Ni surface exposition after reaction (Table 9). The observed

changes in binding energy and surface Ni exposition points

to the formation, under reaction conditions, of the nickel

aluminate entities over Ni particles that leads to a surface

situation very close to that observed for the calcined non

activated sample. In the case of the sample used in the gas-

phase reforming of glycerol, the Ni 2p3/2 profile did not

change in comparison with the binding energies and per-

centages of Ni species detected on the fresh reduced sample.

The main difference observed for the used sample is related

to the quantity of carbon present on the surface of the sam-

ples, being much higher in the used sample.

4 Discussion

4.1 Structure of Supports

Physichochemical characterization of supports revealed the

different interaction degree of Zr-, Ce-, La- and Mg-pro-

moters with alumina carrier that modifies the properties of

the supports at different level. For lanthana–alumina sup-

port, the absence of XRD peaks characteristics of discrete

La phases (A–L in Fig. 1) together with the XPS results

showing a La 3d5/2 binding energy and a La/Al intensity

ratio (Table 5) similar to the value corresponding to bulk

composition suggest that lanthanum is highly dispersed

over alumina support. The characterization of the ceria–

alumina support by XRD and XPS showed the coexistence

of large CeO2 particles (9 nm) and highly dispersed ceria

entities in close contact with the Al2O3 surface. For

magnesia–alumina support, both the absence of XRD dif-

fraction lines corresponding to Mg phases together with the

high Mg/Al ratio point out to a high dispersion of Mg ions

over the surface of alumina. From XPS analysis was not

possible to identify any specific magnesium phase because

the BEs values of Mg 2p in MgAl2O4 and MgO are indis-

tinguishable. Nevertheless magnesium surface spinels

might well be the dominant magnesium phase on the sur-

face taking into account that, at low Mg/Al ratios, the

reaction between Mg and Al2O3 is very favourable to form

surface spinels [20]. For zirconia–alumina support, the

measured XPS Zr/Al ratio indicates that ZrO2 appears well

dispersed on Al2O3 surface. The good dispersion of zirconia

on the alumina surface was also corroborated by the

absence, in spite of the relatively high ZrO2 loading, of

XRD peaks and by the absence of loss in specific surface

area indicative of pore plugging of alumina by ZrO2

crystallites.

4.2 Structure of Nickel Catalysts

The modifications of alumina support due to the addition of

Zr, Ce, La and Mg directly affect the structure and mor-

phology of Ni particles in both oxidized and reduced forms.

Although the Scherrer’ equation should not be used for

absolute determination of particle size distribution, the

sizes it provides allows to compare average crystalline

sizes. XRD profiles of calcined catalysts (Fig. 2a) showed

NiO crystallites with size increasing following the trend:

Ni/A & Ni/A–Z & Ni/A–C [ Ni/A–M [ Ni/A–L. This

result agrees with several studies in literature that indicates

the ability of lanthanum [21, 22] and Mg [23] to disperse

nickel crystallites. TPR data of calcined Ni catalysts pro-

vided additional information about Ni oxidized entities not

detected by XRD. According to literature [24, 25] a num-

ber of factors can influence on the reduction of NiO

850 860 870

Co

un

ts p

er s

eco

nd

(a.

u.)

(c)

(b)

Binding energy (eV)

(a)

x 0.25

Ni 2p3/2

Fig. 8 Ni 2p3/2 core-level spectra of Ni/A–Z catalyst after: (a)

reduction, (b) glycerol reforming reaction in gas phase and (c)

aqueous phase glycerol reforming

Table 9 Binding energy (eV) of Ni 2p3/2 core electrons and surface

atomic ratios of used Ni/A–Z catalyst

Ni2p3/2 (eV) Ni/Al C/Al

After aqueous phase reforming 856.5 (100%) 0.093 0.429

After steam reforming 851.9 (29%) 0.017 1.513

856.5 (71%)

54 Top Catal (2008) 49:46–58

123

including particle size and/or the level of interactions of

these oxides with the support. Stronger interactions of

nickel oxide entities with supports provoke higher diffi-

culty in their reducibility. The origin of the interaction

between NiO particles and support remains a matter of

controversy, and not unambiguous theoretical explanation

of this phenomenon has yet been offered. Two different

explanations about the origin of this interaction appear in

the studies published in literature: (i), nickel ocupying

different sites in the alumina, octahedral or tetrahedral, this

latter more difficult to reduce [26, 27]; or (ii), the incor-

poration of Al3+ [13], Ce3+ [28] or La3+ [29] ions on the

surface of NiO particles during impregnation. The latter

possibility is in accordance with XPS characterization of

calcined catalysts (Table 6) since the absence of Ni 2p3/2

signal corresponding to NiO species together with the low

Ni/Al surface XPS ratios seem to indicate that in all cata-

lysts the surface of NiO particles is covered by dispersed

support species. XPS characterization of Ni/A–L and Ni/

A–C catalysts showing low Ni/Al ratios and the similarity

in La/Ce surface exposure before and after nickel deposi-

tion supports the notion that a part of the NiO particles

could be covered by surface nickel bound to aluminum and

lanthanum/cerium ions. The catalyst containing Zr shows

intense reduction peaks in the high temperature range

(Table 4). This fact may be indicative, as indicated previ-

ously for Ce- and La-added catalysts, of some surface

interaction between Ni Al and Zr ions that retarded the

reduction of nickel entities. Nevertheless XPS analysis of

Ni/A–Z does not support the above assumption. On the

contrary, the decreasing of Zr/Al XPS ratio after Ni

deposition respect to the values observed on the bare

support (Table 6) was indicative of the partial coverage

and the close contact of Zr with the deposited nickel oxide

entities. In the case of Ni/A–M catalyst, the presence of Mg

in the support affects to the proportion of Ni strongly

interacting with alumina as it is indicated by the lower

intensity of reduction peak at 1069 K respect to that cor-

responding to bare alumina (Table 4). One possible reason

for this reduction behavior could be related with the pro-

pensity of Mg to stabilize on Al2O3 lattice, as it was

previously suggested from the characterization data of A–M

support, that reduces the strong Ni–Al interaction

inhibiting the incorporation of Ni into the alumina lattice

[30].

The above changes in reducibility and interaction of

NiO particles with supports imply variations in exposition

of metallic Ni after reduction. XRD data in Table 3 showed

differences in particle size of Ni crystallites after reduction.

Crystalline sizes of Ni0 phase decreased following the

sequence: Ni/A–Z [ Ni/A = Ni/A–C = Ni/A–M [ Ni/

A–L. The differences on crystalline sizes were in good

agreement with Ni/Al ratios from XPS measurements

(Table 5). However, taking into account the differences in

nominal Ni/Al compositions, the sequence of surface Ni/Al

concentration was the reverse to that previously determined

from XRD: Ni/A = Ni/A–M [ Ni/A–Z. For Ni/A–C and

Ni/A–L catalysts the XPS surface Ni concentrations did not

also fit with XRD sequence, showing a lower Ni/Al ratio

than that expected from their Ni crystalline size. This lower

Ni surface concentration may be related, as it was previ-

ously shown, to the fact that calcined Ni/A–C and Ni/A–L

catalysts presented NiO phases partially covered by entities

with aluminum, cerium and lanthanum ions that after

reduction make difficult or even impossible the access to

metallic function.

4.3 Catalytic Structure and Activity Correlations

4.3.1 Aqueous Phase Reforming (APR)

The activity of the catalysts studied in this work clearly

indicated the different catalyst functionality necessary to

carry out aqueous-phase and vapor-phase steam reforming

reactions of glycerol. According to the mechanism

described for aqueous reforming of oxygenated hydrocar-

bons [31], the organic molecule adsorbs on metallic phases

that activate the desired C–C cleavage to form adsorbed

CO and H2 or the undesired C–O cleavage to produce

propyleneglycol, ethylenglycol, ethanol,…. It is apparent

from Fig. 6, despite the strong instability of the catalytic

systems under reaction, that with the exception of Mg, the

addition of La-, Ce- and Zr-oxides to Al2O3 improves the

initial glycerol conversions with respect the ones obtained

on bare Al2O3 supported catalyst. Results in literature [32]

report the important effect of the support on activity of

platinum based catalysts for hydrogen production by APR

of oxygenated hydrocarbons. In these studies evidence was

found that supports promoting dehydration reactions lead

to lower hydrogen productions. In our case, the composi-

tion of the liquid products collected in Table 8 does not

show significant differences on the selectivity toward the

formation of dehydration products with the type of the

support, indicating therefore that other parameters than

dehydration capacity of the support affect to catalytic

behavior of nickel based catalysts. Effects of the support on

nickel dispersion may be discarded because metallic Ni

surface concentrations from XPS and XRD analyses of the

different catalysts do not correlate with the initial glycerol

conversion values derived from Fig. 5. Therefore, the

observed differences in catalytic activities may be related

with geometric effects caused by the presence of promoters

on Ni surfaces as proposed by Shabaker et al. [33], who

found a promotional effect of Sn in the production of

hydrogen by APR of oxygenated compounds over Ni/

Al2O3 catalyst that they assign to the selective poison by

Top Catal (2008) 49:46–58 55

123

Sn of Ni defect sites responsible for methanation of CO or

CO2. Our characterization studies of Ni/A–C and Ni/A–L

catalysts suggest that cerium and lanthanum atoms partic-

ipate in nickel metal particles coverage after reduction.

Therefore in a similar way than that proposed by Shabaker

et al [33], the improvement in activity observed for the Ni/

A–C and Ni/A–L catalyst might be related to geometric

effects caused by the decoration of metallic Ni phases by

Ce and La atoms. In the case of Ni/A–Z and Ni/A–M

catalysts, however, no surface modification of Ni particles

by Zr or Mg atoms was observed. As characterization data

indicate, the main difference of Ni/A–M and Ni/A–Z cat-

alysts respect to Ni/A one are related to the propensity of

Mg to stabilize on Al2O3 that reduces the proportion of

nickel strongly interacting with aluminum and the close

interaction between Ni and ZrO2, respectively. Therefore,

the differences in APR activity between Ni/A and Ni/A–M

and Ni/A–Z samples strongly suggest that, apart of metallic

nickel, the interaction between Ni and ZrO2 and the pro-

portion of nickel strongly interacting with aluminum take

part in the APR reaction. However, the question as to how

these interactions enhance the activity was not evidenced

from the data presented here requiring therefore a further

investigation.

In spite that nickel catalysts showed differences on their

initial APR activities depending on the support used, the

stability under reaction conditions were very low for all the

catalysts prepared. Severe deactivation phenomena of Ni

catalysts supported on Al2O3, ZrO2 and SiO2 were previ-

ously reported in high-pressure aqueous environments [33].

According to these studies, the sintering of Ni particles has

been suggested as the cause of the deactivation observed

for Ni catalysts after exposure to reaction conditions.

Contrary to that, the increase of Ni/Al XPS signal after

reaction observed in our case over the representative Ni/

A–Z catalyst (Table 9) was not indicative of some sintering

phenomena. The changes in the Ni 2p XPS spectra col-

lected in Fig. 8 indicates that the main reason for the

observed deactivation phenomena on supported nickel

catalyst after APR reaction was ascribed to the gradually

change of nickel under reaction conditions from metallic to

oxidized state with formation of surface nickel aluminate

spinels. This result is in accordance with results in litera-

ture [34] that reported similar formation of nickel

aluminate spinels as the main cause of deactivation of

supported nickel catalyst subjected to high temperatures in

presence of steam.

4.3.2 Steam Reforming

For glycerol steam reforming in vapor phase, the hydrogen

selectivity over the different catalysts was higher when

nickel was supported on modified alumina compared to the

reference catalyst supported on bare alumina (Fig. 6). This

improvement in performance of Ni catalysts supported on

modified alumina was shown as higher H2 and CO2 yields

and lower CO yield. For steam reforming of hydrocarbons

over nickel catalysts a bifunctional mechanism [35, 36] has

been proposed in which Ni and support participates in the

activation of hydrocarbon and water molecules. Nickel

metal activates the organic molecule, by means of O–H,

–CH2–, C–C and CH2 bond breaking [37] and promotes the

reaction between the organic fragments with the OH

groups from the water reagent. Bearing this fact in mind,

the increase in activity observed for catalysts supported on

modified alumina may be caused by an increase in the

exposed metal surface area and/or in the capacity to acti-

vate steam. As stated above, XPS and XRD analyses of the

different Ni catalysts show changes in the metallic nickel

surface concentrations with the type of support used to

disperse Ni (Tables 3 and 7). Thus the increase in the

hydrogen production observed for the sample Ni/A–M may

be related to the better surface nickel concentration

achieved on this sample catalyst respect to that corre-

sponding to the Ni/A reference. However, the absence of

correlation between activity and surface nickel concentra-

tion in the steam reforming of glycerol for the Ni/A–C, Ni/

A–L and Ni/A–Z samples suggests that additional proper-

ties to the metallic phase exposition play a role in driving

the catalytic activity over these samples. In spite of the low

Ni dispersion and taking into account that ZrO2 by itself is

not active for steam reforming [38], the high activity

observed for the Ni/A–Z catalyst suggests that ZrO2,

modifying the Al2O3 and interacting with Ni active phase,

is able to enhance the H2O activation. A similar result was

reported by Takanabe et al. for steam reforming of acetic

acid over ZrO2-based supported Pt catalysts [38]. These

authors suggest that the reaction occurs at the periphery of

the Pt particles in the proximity of ZrO2. As our results

suggest that a part of the Ni atoms in Ni/A–Z catalyst have

an enhanced activity, we propose in line with the above

explanation, that steam reforming of glycerol occurs

probably at those Ni sites which are in close proximity of

the ZrO2 entities. In the case of Ce- and La-containing

catalysts, there were not clear evidences of the ability of

ceria and lanthana to promote the activation of water. In

literature the positive effect of lanthanum and cerium in the

activity of nickel catalysts applied to steam reforming of

hydrocarbons is generally ascribed to the good contact

between lanthana/ceria and Ni phases that inhibits the

effects, coking and/or sintering, that reduce the loss of

metal surface area under reaction [39, 40]. In our case,

characterization of Ni/A–C and Ni/A–L catalysts showed

the presence of ceria and lantana entities on top of metallic

nickel crystallites that may explain, as suggested previ-

ously, their better hydrogen production respect to the Ni/A

56 Top Catal (2008) 49:46–58

123

reference by an enhancement in the stability of the cata-

lysts that means an increase of the amount of active sites

available during the reforming reaction.

5 Conclusions

Aqueous phase reforming and steam reforming of glycerol

were investigated over nickel catalysts supported on

MxOy–Al2O3 (M=Ce, La, Mg and Zr). Characterization of

Ni catalysts showed that the modifications of alumina

support induced by the addition of Zr, Ce, La and Mg

directly affect the structure and morphology of Ni particles

in both oxidized and reduced forms. The presence of Mg

modifies the interaction degree of Ni with alumina by

intercalation of the promoter between nickel and alumina

that inhibited the incorporation of nickel to Al2O3 phase.

The addition of Zr to the Al2O3 support implies a decrease

in the dispersion of Ni phases respect to that achieved on

bare Al2O3 and the development of strong Ni–ZrO2 inter-

actions. The use of Ce- and La-added Al2O3 supports leads

to nickel phases partially covered by ceria and lanthana

entities. The activity of the catalysts studied in this work

clearly indicated the different catalyst functionality nec-

essary to carry out aqueous-phase and vapor-phase steam

reforming reaction of glycerol. For aqueous phase

reforming of glycerol (APR), the addition of Ce, La and Zr

to Al2O3 improves the initial glycerol conversions obtained

over the bare Al2O3 supported catalyst. For Ce- and La-

promoted catalysts, the improvement in catalytic perfor-

mance is suggested to be related with geometric effects

caused by the presence of cerium and lanthanum on Ni

surfaces. In this respect, the decoration of Ni phases by Ce

and La may be in the origin of the improvement in activity

observed for these samples. The differences in APR

activity between Ni/A and Ni/A–Z and Ni/A–M catalysts

strongly suggest that, apart of metallic nickel, the interac-

tion between Ni and ZrO2 and the proportion of nickel

strongly interacting with alumina take part in the APR

reaction. In spite that nickel catalysts showed high APR

activities at initial times of reaction they deactivate quickly

under reaction conditions. Characterization of catalysts

after APR indicates that the main reason for the observed

deactivation phenomena is ascribed to the gradually change

of nickel under reaction conditions from metallic to oxi-

dized state with formation of surface NiAl2O4. For glycerol

steam reforming in vapour phase, the use of Mg, Zr, Ce and

La as promoter of Ni based catalysts increases the hydro-

gen selectivity compared to the reference catalyst

supported on bare alumina. Differences in activity were

explained in terms of surface nickel concentration (Mg),

capacity to activate steam (Zr) and stability of nickel

phases under reaction conditions (Ce and La).

Acknowledgments The authors thank the financial support to

Ministerio de Educacion y Ciencia of Spain (Projects MAT2003-

08348-C04-01 and ENE2007-6753-C02-01) and the University of

the Basque Country. R.M.N also acknowledges the Ministerio de

Educacion y Ciencia for a Ramon y Cajal research program.

References

1. Ma F, Hanna MA (1999) Biosource Technol 70:1

2. Cortright RD, Davda RR, Dumesic JA (2002) Nature 418:964

3. Trimm DL (1997) Catal Today 37:233

4. Davda RR, Shabaker JW, Huber GW, Cortright RD, Dumesic JA

(2003) Appl Catal B Environm 43:13

5. Shabaker JW, Davda RR, Huber GW, Cortright RD, Dumesic JA

(2003) J Catal 215:344

6. Parmaliana A, Arena F, Frusteri F, Coluccia S, Marchese L,

Martra G, Chuvilin A (1993) J Catal 141:34

7. Choudhary VR, Uphade BS, Mamman AS (1995) Catal Lett

32:387

8. Souza MMV, Schmal M (2004) Stud Surf Sci Catal 147:133

9. Wang X, Gorte RJ (2001) Catal Lett 73:15

10. Bangala DN, Abatzoglou N, Chornet E (1998) AICHE J 44:927

11. Shaper H, Doesburg EBM, Van Reijen LL (1983) Appl Catal

7:211

12. Horiuchi T, Teshima Y, Osaki T, Sugiyama T, Suzuki K, Mori T

(1999) Catal Lett 62:107

13. Richardson JT, Twigg MV (1998) Appl Catal A Gen 167:57

14. Sheffer B, Molhoek P, Moulijn JA (1989) Appl Catal 46:11

15. Morant C, Sanz JM, Galan L, Soriano L, Rueda F (1989) Surf Sci

218:331

16. Briggs D, Seah MP (Eds) (1990) Practical surface analysis by

auger and X-ray photoelectron spectroscopy, 2nd edn. Wiley,

Chinchester

17. Chen X, Liu Y, Niu G, Yang Z, Bian M, He A (2001) Appl Catal

A Gen 205:159

18. Haack LP, de Vries JE, Otto K, Chatta MS (1992) Appl Catal A

Gen 82:199

19. Ledford JS, Houalla M, Proctor A, Hercules DM, Petrakis L

(1989) J Phys Chem 93:6770

20. Morterra C, Ghiotti G, Bocuzzi F, Coluccia S (1978) J Catal

51:299

21. Wang S, Lu GQM (1998) Energy Fuels 12:248

22. Chou TY, Leu CH, Yeh CT (1995) Catal Today 26:53

23. Shishido T, Sukenobu M, Morioka H, Kondo M, Wang Y, Takaki

K, Takehira K (2002) Appl Catal A Gen 223:35

24. Richardson JT, Lei M, Turk B, Forster K, Twigg MV (1994)

Appl Catal A General 110:217

25. Richardson JT, Turk B, Twigg MV (1996) Appl Catal A General

148:97

26. Dufresne P, Payen E, Grimblot J , Bonelle JP (1981) J Phys Chem

85:2344

27. Wu M, Hercules DM (1979) J Phys Chem 83:2003

28. Shan W, Luo M, Lin P, Shen W, Li C (2003) Appl Catal A Gen

246:1

29. Blom R, Dahl IM, Slagtern A, Sortland B, Spjelkavik A,

Tangstad E (1994) Catal Today 21:535

30. Guo J, Zhao H, Chai D, Zheng X (2004) Appl Catal A Gen

273:75

31. Davda RR, Shabaker JW, Huber GW, Cortright RD, Dumesic JA

(2005) Appl Catal B Environm 56:171

32. Shabaker JW, Huber GW, Davda RR, Cortright RD, Dumesic JA

(2003) Catal Lett 88(1–2):1

33. Shabaker JW, Huber GW, Dumesic JA (2004) J Catal 222:180

Top Catal (2008) 49:46–58 57

123

34. Oh YS, Roh HS, Jun KW, Baek YS (2003) Int J of Hyd Energy

28:1387

35. Ross JRH, Steel MCF, Zeini-Isfahani A (1978) J Catal 52:280

36. Rostrup-Nielsen JR (1984) Catalysis science and technology, vol

5. Springer/Verlag, Berlin

37. Gates SM, Russel JN, Yates JTJ (1986) Surf Sci 171:111

38. Takanabe K, Aika K, Seshan K, Lefferts L (2004) J Catal

227:101

39. Natesakhawat S, Watson RB, Wang X, Ozkan US (2005) J Catal

234:496

40. Zhang ZL, Verykios XE, MacDonald JM, Affrosman S (1996) J

Phys Chem 100:744

58 Top Catal (2008) 49:46–58

123