Steam reforming of ethanol for hydrogen production over MgO - Supported Ni-based catalysts

G Model

APCATA-15647; No. of Pages14

ARTICLE IN PRESS

Applied Catalysis A: General xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Applied Catalysis A: General

journal homepage: w ww.elsevier.com/locate/apcata

Steam reforming of ethanol for hydrogen production over

MgO—supported Ni-based catalysts

Gleicielle T. Wurzler^a,b, Raimundo C. Rabelo-Neto^a, Lisiane V. Mattos^c, Marco A. Fraga^a,b

,

Fábio B. Noronha^a,b,∗

^aNational Institute of Technology, Catalysis Division, Av. Venezuela 82, Rio de Janeiro 20081-312, Brazil

^bMilitary Institute of Engineering, Chemical Engineering Department, Prac¸ a Gal. Tiburcio 80, Rio de Janeiro 22290-270, Brazil

^cFluminense Federal University, Chemical Engineering Department, Rua Passo da Pátria, 156, Niterói 24210-240, Brazil

a r t i c l e i n f o

a b s t r a c t

Article history:

Received 13 July 2015

Received in revised form 5 November 2015

Accepted 11 November 2015

Available online xxx

This work studied the effect of preparation method of MgO on the performance of Ni/MgO catalysts for

steam reforming of ethanol. Three different MgO were prepared by precipitation and aging, precipitation

and decomposition of precursor salt. Depending on the synthesis conditions, the basicity and the type of

Ni species signiﬁcantly varied. TPD of adsorbed CO₂showed that the MgO prepared by precipitation and

aging possess the highest amount of basic sites. TPR and XANES revealed the presence of two different

Ni species: NiO particles and Ni²⁺inserted in the lattice of a NiO–MgO solid solution, which is hardly

reducible. Ni supported on MgO obtained by precipitation and aging exhibited the highest reduction

degree. The preparation method of MgO also affected the amount of carbon formed during SR of ethanol at

773 K under H₂O/ethanol molar ratio of 3.0. Increasing basicity decreased the amount of carbon deposits,

which was attributed to the increase of carbon gasiﬁcation rate.

Keywords:

Hydrogen production

Steam reforming of ethanol

Ni/MgO catalyst

Deactivation mechanism

1. Introduction

Hydrogen can be produced through the steam reforming of

biomass-derived liquids such as bioethanol, a water and ethanol

Today, the majority of hydrogen is produced in reﬁneries

to upgrade crude oil (hydrocracking and hydrotreating process),

in the petrochemical industry to synthesize different chemical

compounds (such as ammonia and methanol), for oil and fat hydro-

genation and, in metallurgical processes (as a reduction gas) [1].

However, the rising concern with the reduction of greenhouse gas

emissions and atmospheric pollution increased the interest in using

hydrogen as an energy carrier for power generation with fuel cells

as well as for the conversion of biomass into liquid fuels.

Hydrogen can be electrochemically converted in PEM fuel cells

to produce electricity for use in transportation applications and

portable power devices and also for residential combined heat and

power systems [2]. Hydrogen is also required in the hydrodeoxy-

genation (HDO) process of bio-oil produced from the fast pyrolysis

of biomass for gasoline and diesel production [3] as well as in HDO

of fermentation products obtained in a sugarcane bioreﬁnery for

jet fuel and lubricants production [4].

mixture that may be obtained by biomass fermentation [5–8]. The

steam reforming of ethanol is an attractive route to hydrogen pro-

duction because: (i) it is a renewable and CO₂-neutral source that

can readily be obtained from biomass fermentation; (ii) ethanol is

signiﬁcantly less toxic than methanol and gasoline; (iii) the infras-

tructure required for ethanol production and distribution is already

established in countries like Brazil and USA since ethanol is cur-

rently distributed and blended with gasoline.

Different technologies may be applied to generate hydrogen

from ethanol, including steam reforming (SR) (Eq. (1)), partial oxi-

dation (POX) (Eq. (2)), and oxidative steam reforming (OSR) (Eq.

(3)) [8]:

C₂H₅OH + 3H₂O → 2CO₂+ 6H₂

C₂H₅OH + 1.5O₂→ 2CO₂+ 3H₂

(1)

(2)

C₂H₅OH + (3 − 2x)H₂O + xO₂→ 2CO₂+ (6 − 2x)H₂0 < x < 0.5 (3)

However, various reaction pathways may occur depending on

the reaction conditions and the choice of the catalyst. Some of these

reactions lead to the formation of coke and, consequently, induce

catalyst deactivation.

The type of support directly inﬂuences the product distribution

and catalyst stability during ethanol conversion reactions since it

∗

Corresponding author at: National Institute of Technology, Catalysis Division,

Av. Venezuela 82, 20081-312 Rio de Janeiro, Brazil. Fax: +55 21 2123 1166.

E-mail address: fabio.bellot@int.gov.br (F.B. Noronha).

http://dx.doi.org/10.1016/j.apcata.2015.11.020

Please cite this article in press as: G.T. Wurzler, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.11.020

G Model

APCATA-15647; No. of Pages14

ARTICLE IN PRESS

G.T. Wurzler et al. / Applied Catalysis A: General xxx (2015) xxx–xxx

2

also exhibits activity for this reaction. Al₂O₃is generally used as

a support but its acid sites promote the dehydration of ethanol to

ethylene, which is considered a precursor of coke [9]. MgO con-

tains strong basic sites, which are proposed to be highly active for

ethanol dehydrogenation to acetaldehyde that is considered the

primary intermediate of SR of ethanol [9]. It is well known that

basic metal oxides minimize carbon formation increasing adsorp-

tion of water and promoting the rate of carbon gasiﬁcation reaction

[10]. Therefore, there are many studies in the literature about the

SR of ethanol over MgO supported catalysts [9,11–20]. However,

carbon formation is still observed for SR of ethanol over Ni/MgO

catalysts [11–15,18–20]. In fact, there is no work in the literature

that studies the inﬂuence of the basic properties of magnesia on the

rate of carbon formation during SR of ethanol. Then, the design of

a catalyst resistant to carbon formation for SR of ethanol requires a

better understanding of the effect of the surface basic properties of

magnesia on catalyst deactivation. This can be done by tailoring the

surface basic properties of magnesia and, consequently, the amount

and strength of basic sites, by controlling the synthesis method.

Menezes et al. [21] studied the effect of the preparation method

on the surface basicity of MgO. Different MgO were synthesized

by precipitation and hydrothermal treatments and decomposition

of magnesium nitrate. The samples presented different basic site

distributions, revealing the important role of the preparation con-

ditions on tuning the surface basicity.

analyses were performed with the samples (300 mg) in powder

form using a semi-quantitative method (QUANT-EXPRES/Bruker).

2.2.2. X-ray diffraction (XRD)

The X-ray powder diffraction pattern of the calcined samples

were obtained with Cu K␣ = 1,5406^◦using a RIGAKU diffractometer.

Data were collected over the 2ꢀ range of 30^◦–130^◦using a scan rate

of 0.02^◦/step and a scan time of 1 s/step. The lattice parameters

of MgO and NiO were calculated from the (200) reﬂections (Eq.

(4)). The composition of the NiO–MgO solid solution was calculated

from Vegard’s rule (Eq. (5)) [22]. The Scherrer equation was used

to estimate the crystallite mean diameter of MgO and metallic Ni

particles (Eq. (6)).

ꢀ

h²+ k²+ l²× ꢁ

a =

(4)

(5)

2senꢀ

a_{Ni Mg}

= x × a_NiO+ (1 − x) × a_MgO

O

x

1−x

k × ꢁ

ˇ × cos ꢀ

d =

(6)

where a is the lattice parameter; h, k and l are the Miller indices;

ꢁ is the wave length; ꢀ is the diffraction angle; x is the composi-

tion of the NiO–MgO solid solution; d is the crystallite size; k is a

constant (0.9); and ˇ is the width at half-maximum intensity of the

diffraction line.

Therefore, the aim of this work is to study the effect of prepara-

tion method of MgO on the performance of Ni/MgO catalysts during

ethanol conversion reactions. MgO was prepared by precipitation,

precipitation with aging and decomposition of the precursor salt

in order to vary the surface basicity. A correlation was established

between the density of basic sites and the catalyst resistance to

carbon deposition.

In situ XRD was carried out at the XPD-10B beamline of the

Brazilian Synchrotron Light Laboratory (LNLS). The samples were

placed in a furnace installed into a Huber goniometer operat-

ing in Bragg–Brentano geometry (ꢀ − 2␪). The XRD patterns were

obtained by a Mythen—1 K detector (Dectris) located 1 m from the

furnace, in a 2ꢀ interval from 23^◦to 56^◦, using a wavelength of

1.55002 Å. The measurements were made while the sample under-

went the following conditions: (i) Reduction under a 5% H₂/He

mixture from 298 to 1023 K at a heating of 10 K/min, remain-

ing at this temperature for 1 h. After reduction the sample was

purged with helium at the same temperature for 30 min (ii) SR

of ethanol—reaction mixture containing 98% He, 1.5% H₂O, 0.5%

ethanol for at 773 K for 1 h. The average crystallite size of metallic Ni

for the reduced catalyst and for the used catalyst after SR of ethanol

reaction was calculated using the Scherrer equation (Eq. (6)). An

Omnistar/Pfeiffer Vacuum mass spectrometer (MS) was used for

on-line monitoring of efﬂuent gas composition.

2. Experimental

2.1. Catalyst preparation

The MgOpa sample was prepared by precipitation followed by

aging. Mg(NO₃)₂·6H₂O and NaOH solutions were slowly added to

a Na₂CO₃solution under vigorous stirring. The precipitate formed

was aged at pH 10 for 12 h. The gel was centrifuged and extensively

washed with distilled water until constant pH. Then, it was dried at

373 K for 12 h and calcined at a heating rate of 5 K/min up to 773 K

for 5 h.

Another sample (MgOp) was also synthesized by precipitation

from the same Mg(NO₃)₂·6H₂O and Na₂CO₃precursors. In this case,

an aqueous solution of Mg(NO₃)₂·6H₂O was quickly added to a

container with the base solution. The magnesium hydroxide pre-

cipitate was formed instantly. It was then ﬁltered, washed with

distilled water until no pH change could be detected, and then

calcined at 773 K for 5 h (5 K/min).

2.2.3. Temperature-programmed desorption of CO₂(TPD-CO₂)

The basic surface sites were probed by temperature pro-

grammed desorption of adsorbed CO₂. TPD-CO₂was carried out

in a ﬁxed-bed reactor coupled to a quadrupole mass spectrometer

(Balzers). Prior to TPD analyses, the samples were treated under

ﬂowing H₂(30 mL/min) up to 1023 K (10 K/min), remaining at that

temperature for 1 h. The system was then purged with ﬂowing He at

the treatment temperature for 30 min and cooled to room temper-

ature. The adsorption of CO₂was performed at room temperature

by ﬂowing CO₂through the sample. Then, the sample was purged

under He for 30 min. After adsorption, the catalyst was heated at a

heating rate of 20 K/min up to 773 K in ﬂowing He (50 mL/min).

A third sample (MgOd) was obtained by thermal decomposition

of Mg(NO₃)₂in a mufﬂe at 773 K for 5 h.

The catalysts were prepared by incipient wetness impregna-

tion of the supports with an aqueous solution of Ni(NO₃)₂·6H₂O

(Sigma–Aldrich) to obtain 5 wt% Ni. The samples were dried at 373 K

and calcined under air (50 mL/min) at 673 K for 3 h. Three catalysts

were obtained: Ni/MgOpa, Ni/MgOp, Ni/MgOd.

2.2.4. Temperature programmed reduction (TPR)

TPR experiments were performed in

a TPR/TPD 2900

Micromeritics system equipped with thermal conductivity

a

detector (TCD). The catalyst was pretreated at 473 K for 1 h under

a ﬂow of air prior to the TPR experiment in order to remove

adsorbed species from the catalyst surface. The reducing mixture

(10% H₂/N₂) was passed through the sample (100 mg) at a ﬂow

rate of 30 mL/min and the temperature was increased to 1273 K at

a heating rate of 10 K/min.

2.2. Catalyst characterization

2.2.1. X-ray ﬂuorescence spectrometry (XRF)

The chemical composition of each sample was determined on

Wavelength Dispersive X-Ray Fluorescence Spectrometer (WD-

XRF) S8 Tiger, Bruker with a rhodium tube operated at 4 kW. The

Please cite this article in press as: G.T. Wurzler, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.11.020

G Model

APCATA-15647; No. of Pages14

ARTICLE IN PRESS

G.T. Wurzler et al. / Applied Catalysis A: General xxx (2015) xxx–xxx

3

Fig. 1. XRD patterns of calcined MgO and Ni/MgO samples.

2.2.5. In-situ XAFS

2.3. SR and OSR of ethanol

In-situ XAS experiments were performed at the D08B-XAFS—2

beamline of the Brazilian Synchrotron Light Source (LNLS, Camp-

inas). The samples were previously reduced under pure H₂at

1023 K for 1 h and then passivated with a 5% O₂/He mixture for

1 h at 209 K as previously described for XRD experiments. Then,

the samples were packed into a 1.0 (I.D.) x1.2 (O.D.) mm capillary

reactor, placed inside a furnace, whose temperature was controlled

via a thermocouple placed inside the reactor right close to the

catalyst bed. Ni K-edge spectra were recorded in the transmis-

sion mode, while the sample underwent the following treatment:

(i) heating ramp under a 5% H₂/He mixture at 10K min⁻¹up to

1023 K, (ii) temperature hold for 30 min at 1023 K, (iii) sample cool-

ing down to room temperature, under a 5% H₂/He mixture ﬂow.

Data reducing was made using the ATHENA software package [23].

The background subtracted, normalized XANES spectra were then

trimmed at a suitable energy range (c.a.100 eV above the absorp-

tion edge). The composition of metallic Ni phase during sample

activation (reduction under H₂/He ﬂow) was monitored by linear

combination of normalized Ni K-edge XANES spectra (LC-XANES)

within the 8.3–8.43 keV energy range, using Ni foil, NiO and a solid

solution NiMgO as references for Ni⁰and Ni²⁺, respectively.

SR and OSR of ethanol reactions were performed in a quartz tube

reactor at 773 K and atmospheric pressure. Prior to reaction, sam-

ples were reduced under pure hydrogen (30 mL/min) at 1023 K for

1 h and then purged with N₂at the same temperature for 30 min.

For SR, H₂O/ethanol molar ratio of 3.0 and 10.0 were used. The reac-

tant mixture was obtained by ﬂowing two N₂streams (30 mL/min)

through each saturator containing ethanol and water separately.

OSR was performed employing a H₂O/ethanol molar ratio of 3.0 and

an O₂/ethanol molar ratio of 0.5. A ﬂow of 5.6% O₂/N₂(30 mL/min)

and a ﬂow of N₂(30 mL/min) were passed through the saturators

containing ethanol and water, respectively. The partial pressure of

ethanol was maintained constant for all experiments. In order to

observe the deactivation of the catalyst within a short timeframe,

a small amount of catalyst was used (20 mg). The samples were

diluted with inert SiC (SiC mass/catalyst mass = 3.0). The reaction

products were analyzed by gas chromatography (Micro GC Agilent

3000 A) containing three channels for three thermal conductivity

detectors (TCD) and three columns: a molecular sieve, a Poraplot Q

and OV-1 column. The ethanol conversion and products distribu-

tion were determined as follows:

2.2.6. Thermogravimetric analysis (TG)

(n_ethanol

)

_fed− (n_ethanol

)

exit

Thermogravimetric analysis of the used catalysts was car-

ried out in a TA Instruments equipment (SDT Q600) in order to

determine the amount of carbon formed over the catalyst. Approx-

imately 10 mg of spent catalyst was heated under air ﬂow from

room temperature to 1273 K at a heating rate of 20 K/min and the

weight change was measured.

X_ethanol

=

× 100

(7)

(8)

(n_ethanol

)

fed

n

x

(

)

produced

S_x

=

× 100

produced

n

(

)

total

2.2.7. Scanning electron microscopy (SEM)

SEM analyses of the spent catalysts were carried out using a ﬁeld

emission scanning electron microscope (FE-SEM) Quanta FEG 450

FEI operating with an accelerating voltage of 20 kV. The microscope

was also equipped with an EDAX analytic system energy dispersive

spectrometer (Oxford Instruments—model X-MAX).

where (n_x)_produced= moles of x produced (x = hydrogen, CO, CO_2,

methane, acetaldehyde or ethene) and (n_{total produced}= moles of

H₂+ moles of CO + moles of CO₂+ moles of methane + moles of

acetaldehyde + moles of ethene (the moles of water produced are

not included).

)

Please cite this article in press as: G.T. Wurzler, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.11.020

G Model

APCATA-15647; No. of Pages14

ARTICLE IN PRESS

G.T. Wurzler et al. / Applied Catalysis A: General xxx (2015) xxx–xxx

4

Table 1

The TPR proﬁle of MgO supported catalysts and NiO reference

were shown in Fig. 2. Ni/MgOpa catalyst exhibited three peaks

at 555, 725 and 956 K. The reduction proﬁles of Ni/MgOd and

Ni/MgOp exhibited peaks at 565 and 587 K, respectively, and at

943 and 1055 K, respectively, and a shoulder at around 710 K. The

assignment of reduction peaks of Ni-based catalysts remains con-

troversial in the literature and this is likely due to the differences

in the experimental conditions applied for the analysis and in the

synthesis method used to obtain the samples. For example, the TPR

proﬁle of bulk NiO exhibit peaks at 589 K [31], 593 K [32], 600 K

[33], 693 K [34], or a peak and a shoulder at 662/723 K [35] or at

715/770 K [36]. In the present work, the TPR proﬁle of NiO showed

a peak at 718 K and a shoulder at around 748 K (Fig. 2).

Chemical composition of Ni/MgO catalysts and BET surface area of the supports.

Catalyst

Chemicalcomposition(%)

Support surface area (m²/g)

NiO (Ni⁰)

MgO

Ni/MgOpa

Ni/MgOp

Ni/MgOd

6.75 (5.30)

6.58 (5.17)

7.18 (5.64)

93.25

93.42

92.82

149

71

3

Composition of metallic Ni in parentheses.

3. Results and discussion

There are some studies in the literature that report the TPR

proﬁle of Ni/MgO catalyst [37,38]. The reduction proﬁle of 21%

Ni/MgO showed two peaks at 553 and 973 K that were attributed

to the reduction of NiO particles and Ni²⁺ions located in the MgO

lattice, revealing the presence of a NiO–MgO solid solution [37].

Parmaliana et al. [38] carried out a systematic study about the

effect of the calcination temperature (673–1273 K) and nickel load-

ing (2.8, 10.8 and 18.0 wt% Ni) on the reduction of Ni/MgO catalysts

using TPR. The reduction proﬁle of Ni/MgO calcined at 673 K exhib-

ited 3 peaks and a shoulder. The small peak at 533 K was attributed

to the presence of Ni³⁺species. The peak at 613 K was associated

with bulk NiO on the surface of MgO. The peak at 863 K and the

shoulder at 1011 K were assigned to Ni²⁺located in the outermost

and inner layers of the MgO lattice. When the calcination tem-

perature was increased, the low temperature peak disappeared.

In addition, the peak corresponding to bulk NiO was shifted to

higher temperatures, indicating that the interaction of NiO with

MgO increased, probably forming a NiO–MgO solid solution. This

peak was no longer detected after calcination at 1073 K. The peak at

high temperature increased and it was shifted to higher tempera-

ture. This result reveals that the increase of calcination temperature

favors the diffusion of NiO into MgO lattice, promoting the forma-

tion of a NiO–MgO solid solution and, consequently, the reduction

of Ni²⁺becomes more difﬁcult. The reducibility of Ni/MgO catalysts

decreases when the metal loading decreases. For the sample con-

taining 2.8% Ni, only a broad peak at high temperature is observed

in the TPR proﬁle.

Taking into account the peak assignments of TPR proﬁles of Ni

catalysts from the literature, the types of Ni species on Ni/MgO cat-

alysts of our study were identiﬁed. Therefore, the peak at 725 K in

the TPR proﬁle of Ni/MgOpa could be attributed to the reduction of

bulk NiO particles (Fig. 2). In our work, the diffractograms did not

detect the presence of NiO phase. However, further evidence for the

formation of NiO particles was provided by linear combination of

normalized Ni K-edge XANES spectra that will be presented in the

next session. The broad consumption of hydrogen at 956 K might be

assigned to the reduction of Ni²⁺cations inserted into the lattice of

MgO due to the formation of a solid solution during calcination, as

it was demonstrated by XRD data. The low reduction peak deserves

additional comments. Ewbank et al. [39] performed a comprehen-

3.1. Catalyst characterization

Table 1 presents the NiO (Ni⁰in parentheses) and MgO con-

tent of all samples. The Ni loading obtained by XRF analysis was

close to the nominal value for all samples studied (5.0 wt.% Ni).

Furthermore, traces of sodium were not detected.

Fig. 1 shows the XRD patterns of the supports prepared by dif-

ferent methods. The diffractograms of all samples exhibited the

diffraction lines characteristic of the periclase MgO phase with

cubic structure (PDF#45-0946). However, sharper lines were reg-

istered for MgOd, indicating that this sample presented the highest

crystallinity.

The XRD patterns of the Ni/MgO obtained after calcination

(Fig. 1) were quite similar to those of the parent supports, show-

ing the lines corresponding to the periclase MgO phase with cubic

structure as well. It was not possible to distinguish the NiO phase

(PDF#47-1049). Nevertheless, when the diffractograms of Ni/MgO

were compared to the diffractograms of the respective supports, it

was observed that the lines of MgO phase were shifted to higher

2ꢀ positions. The lattice constant a of MgO and Ni/MgO was calcu-

lated from the (200) reﬂections and listed in Table 2. The results

obtained were in agreement with data reported in the literature

[24–26]. Regardless the MgO preparation method, the addition of Ni

decreased the lattice constant. Accordingly, [24–30], the shift of the

MgO peak position towards higher 2ꢀ positions and the decrease in

the MgO lattice constant suggests the formation of a NiO–MgO solid

solution during calcinations for all samples. The composition of the

Ni_xMg_{1 − x}O solid solution formed was estimated based on Vegard’s

rule [22] and the results were reported in Table 2. For Ni/MgOp and

Ni/MgOd, these values were close to the theoretical one (x = 0.035),

whereas it was higher for Ni/MgOpa. This result was also obtained

by Arena et al. [26] and it was attributed to a gradient of Ni²⁺across

the MgO particle. The crystallite size of MgO was calculated using

the line corresponding to (200) reﬂection and the results obtained

are listed in Table 2. MgOpa exhibited the smallest MgO crystal-

lite size (8.4 nm), whereas the largest value was found for MgOd

(46.5 nm). In the case of Ni/MgO samples prepared by hydrothermal

and precipitation methods, the addition of Ni slightly increased the

MgO crystallite size. For the sample synthesized by decomposition,

the MgO crystallite size slightly decreased.

Table 2

Diffraction angle, lattice parameter, composition of the Ni_xMg_1−xO solid solution and crystallites size of MgO obtained for all calcined samples.

Samples

Difraction angle 2ꢀ (^◦)

a^a(Å)

Solid^bsolution composition

d^c(nm)

MgOpa

MgOp

MgOd

Ni/MgOpa

Ni/MgOp

Ni/MgOd

42.779

42.827

42.846

42.812

42.844

42.860

4.2263

4.2218

4.2200

4.2229

4.2202

4.2187

–

8.4

16.3

46.5

13.0

19.2

44.3

Ni_0.066Mg_0.934

Ni_0.038Mg_0.962

Ni_0.032Mg_0.968

O

a

Lattice constant.

b

Solid solution composition calculated from Vegard’s rule: a_NixMg1−xO= x.a_NiO+ (1−x). a_MgO

.

c

Crystallite size of MgO calculated using the (2 0 0) plane.

Please cite this article in press as: G.T. Wurzler, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.11.020

G Model

APCATA-15647; No. of Pages14

ARTICLE IN PRESS

G.T. Wurzler et al. / Applied Catalysis A: General xxx (2015) xxx–xxx

5

943

565

710

587

Ni/MgOd

Ni/MgOp

Ni/MgOpa

1055

555

725

956

718

748

NiO

400

600

800

1000

1200

Temperature (K)

Fig. 2. TPR proﬁles of Ni/MgO catalysts.

Table 3

as well and then, the H₂consumption in this temperature region

was not taken into account in the total hydrogen uptake values

reported in Table 3. Comparing the catalysts prepared from MgO

obtained by different methods, it is observed that the fraction of

hydrogen uptake at high temperature is higher on Ni/MgOp and

Ni/MgOd catalysts, indicating a higher formation of solid solution.

Ni/MgOpa catalyst revealed a higher fraction of bulk NiO particles.

This might explain the higher reduction degree observed for this

sample.

Hydrogen consumption from TPR proﬁles.

Sample

H₂consumption(area%)

Total H₂uptake

(␮mol/g)

Degree of

reduction (%)

600–800 K

>800 K

Ni/MgOpa

Ni/MgOp

Ni/MgOd

47

11

5

53

89

95

87

55

40

74

31

42

In order to conﬁrm the peak assignments of TPR proﬁles of our

Ni/MgO catalysts, in situ H₂-TPR XAFS studies were performed at

Ni K-edge.

sive study about the nature of the Ni species in Ni/Al₂O₃catalysts

prepared by controlled adsorption and dry impregnation. TPR pro-

ﬁle exhibited several peaks depending on the catalyst preparation

method. The ﬁrst peak appeared at 523 and 613 K for the samples

prepared by dry impregnation and controlled adsorption, respec-

tively. This peak was attributed to the decomposition of residual

nitrate. This low temperature peak at 490 K in the TPR proﬁle of 10%

Ni/CeO₂was also assigned to the decomposition of residual nitrates

[40]. Sheffer et al. [33] also reported the presence of a peak at 570 K

in the reduction proﬁle of Ni/Al₂O₃catalysts that was attributed to

the decomposition and reduction of nitrates.

3.2. In situ characterization

Fig. 3a shows the TPR-XANES spectra near the Ni K-edge for

Ni/MgOpa catalyst. Increasing the temperature under H₂ﬂow up

to 573 K did not change the spectra. This result is in agreement with

TPR proﬁle that showed only the decomposition of carbonates in

this temperature range. Further increase in the temperature caused

the decrease in the intensity of the white line in XANES spectra,

indicating the progressive reduction of Ni²⁺species.

Fig. 3b displays the evolution of the Ni species during reduc-

tion determined by the linear combination of normalized Ni K-edge

XANES spectra, using Ni foil, NiO and a solid solution NiMgO as ref-

erences. Initially, nickel is mainly in the Ni²⁺oxidation state in the

structure of the solid solution. Only a small fraction is present as

NiO phase, which agrees well with the TPR results (Table 3). The

reduction of NiO to metallic Ni started at around 573 K, whereas

the Ni²⁺in the solid solution is reduced at higher temperatures

(above 773 K), in agreement with TPR experiments. The fraction

of metallic Ni steadily increased up to the ﬁnal reduction temper-

ature (1023 K), and continuously increased during 1 h at 1023 K

(Fig. 3b).

In order to shed more light on the reduction peak at low temper-

ature, the reduction treatment was followed by mass spectrometry

coupled to the same set up used to perform the in situ XRD experi-

ments that will be described next.

Signals corresponding to NO₂, NO or NH₃were not detect during

treatment under H₂/He mixture. However, an intense signal due

to CO₂was clearly observed at the same temperature of the ﬁrst

peak in the TPR proﬁle. This results suggests that residual carbon-

ates decompose in this temperature range and could be responsible

for the peaks observed in the temperature range of 555–587 K for

Ni/MgO catalysts. In fact, the CO₂-TPD experiments revealed the

desorption of carbonate species above 673 K and thus, they could

remain adsorbed on the support even after the pretreatment car-

ried out before the TPR experiment.

The H₂consumption of TPR proﬁles are listed in Table 3. Since

the ﬁrst peak corresponds to the decomposition of residual carbon-

ates species, not only might the evolution of hydrogen occur but

also CO₂[39]. This compound could contribute to the TCD signal

The XANES spectra of Ni-K-edge after reduction at 1023 K for

1 h of all Ni/MgO catalysts and references are shown in Fig. 4.

The presence of an intense white line for all catalysts is quite

clear, indicating that the reduction of all samples was not com-

Please cite this article in press as: G.T. Wurzler, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.11.020

G Model

APCATA-15647; No. of Pages14

ARTICLE IN PRESS

G.T. Wurzler et al. / Applied Catalysis A: General xxx (2015) xxx–xxx

6

Fig. 3. (A) In situ XANES spectra and (B) LC-XANES of Ni/MgOpa catalyst during reduction under 5% H₂/He ﬂow.

Table 4

Fraction of metallic Ni and Ni in the solid solution after reduction at 1023 K calculated

by the linear combination of Ni K-edge XANES spectra of references.

Sample

% Ni⁰

% Solid solution

Ni/MgOpa

Ni/MgOp

Ni/MgOd

61

37

50

39

63

50

solid solution for all calcined samples, LC-XANES indicate that

NiO was completely reduced to metallic Ni whereas only par-

tial reduction of the NiMgO solid solution took place. In addition,

Ni/MgOpa catalyst exhibited the highest percentage of metallic Ni,

which agrees very well with the TPR discussion presented herein-

before.

The diffractograms obtained during the reduction of Ni/MgOpa

sample under the 5% H₂/He mixture are shown in Fig. 5A. The

diffractogram of the calcined sample exhibited lines located at

2ꢀ = 37.2 and 43.1^◦suggesting the formation of a NiO–MgO solid

solution. There are also lines at 2ꢀ = 38.3 and 51.1^◦corresponding

to Mg(OH)₂phase, which disappeared when the temperature was

increased. By raising the temperature, typical lines correspond-

ing to MgO were shifted to lower 2ꢀ positions. For example, the

line positioned at 2ꢀ = 43.1^◦at 298 K moved to 42.8^◦after reduc-

tion at 1023 K. This result indicates that Ni was removed from the

solid solution, generating metallic Ni particles, which were initially

observed at around 873 K. It is noticed the presence of a shoul-

der at 2ꢀ = 44.1^◦and a small peak at 2ꢀ = 51.4^◦that corresponds to

the diffraction lines characteristic of metallic Ni (1 1 1) and (2 0 0)

planes (Fig. 5 B, curve a). The diffractogram obtained after 1 h under

SR of ethanol reaction remained unchanged (Fig. 5B, curve b).

For Ni/MgOp (Fig. 6) and Ni/MgOd (Fig. 7) samples, the diffrac-

tograms obtained were quite similar to that one of Ni/MgOpa

(Fig. 5). The lines corresponding to MgO were also shifted to lower

2ꢀ positions during the reduction, which suggested that the solid

solution was partially destroyed. The lines characteristic of metal-

lic Ni appeared at 1023 K (Ni/MgO_p) and 973 K (Ni/MgO_d). These

temperatures are higher than that observed for Ni/MgO_p, which

agrees very well with the lower reduction degree of these samples

as calculated by TPR (Table 3).

Fig. 4. In situ XANES spectra of MgO supported Ni after reduction at 1023 K and Ni⁰

and NiMgO references.

plete. The distribution of Ni species after reduction of all samples

at 1023 K was determined by linear combination of normalized

Ni K-edge XANES spectra (LC-XANES) using Ni foil, NiO and a

solid solution NiMgO as references for Ni⁰and Ni²⁺, respectively

(Table 4). These result showed the presence of metallic Ni and

Ni²⁺in the solid solution on the reduced samples. Considering that

the TPR experiments revealed the existence of NiO and NiMgO

Please cite this article in press as: G.T. Wurzler, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.11.020

G Model

APCATA-15647; No. of Pages14

ARTICLE IN PRESS

G.T. Wurzler et al. / Applied Catalysis A: General xxx (2015) xxx–xxx

7

Fig. 5. (A) Diffractograms of Ni/MgOpa obtained during reduction at different temperatures; (B) XRD diffraction patterns of (a) reduced sample; (b) after SR of ethanol

reaction at 773 K.

3.3. Characterization of surface basic sites

or less resolved multiple peaks between 300 and 800 K, suggesting

a wide distribution of basic sites.

A qualitative analysis of the distribution of basic sites (i.e.,

strength and population) can be obtained by TPD of adsorbed CO₂

(Fig. 8). All samples presented a rather complex proﬁle with more

The desorption proﬁle of CO₂have been related to the distri-

bution of basic sites, including both density and strength [41–43].

Cosimo et al. [42] associated the different species of adsorbed CO₂

Please cite this article in press as: G.T. Wurzler, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.11.020

G Model

APCATA-15647; No. of Pages14

ARTICLE IN PRESS

G.T. Wurzler et al. / Applied Catalysis A: General xxx (2015) xxx–xxx

8

Fig. 6. (A) Diffractograms of Ni/MgOp obtained during reduction at different temperatures; (B) XRD diffraction patterns of (a) reduced sample; (b) after SR of ethanol reaction

at 773 K.

on MgO, Al₂O₃and Mg–Al mixed oxides to different types of surface

basic sites. The low temperature desorption peak was correlated

with hydrogen carbonate species adsorbed on weakly basic OH

groups (low basic strength site), while bidentate carbonates were

suggested to be adsorbed on Mg–O site pairs of moderate basic

strength₋, and ﬁnally monodentate carbonate was assigned to iso-

lated O₂anions of high basic strength site, remaining adsorbed

even at high temperature.

Please cite this article in press as: G.T. Wurzler, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.11.020

G Model

APCATA-15647; No. of Pages14

ARTICLE IN PRESS

G.T. Wurzler et al. / Applied Catalysis A: General xxx (2015) xxx–xxx

9

Fig. 7. (A) Diffractograms of Ni/MgOd obtained during reduction at different temperatures; (B) XRD diffraction patterns of (a) reduced sample; (b) after SR of ethanol reaction

at 773 K.

Please cite this article in press as: G.T. Wurzler, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.11.020

G Model

APCATA-15647; No. of Pages14

ARTICLE IN PRESS

10

G.T. Wurzler et al. / Applied Catalysis A: General xxx (2015) xxx–xxx

Ni/MgOd

Ni/MgOp

363

467

553

650

Ni/MgOpa

300

400

500

600

700

800

900

1000

Temperature (K)

Fig. 8. TPD of CO₂proﬁles of Ni/MgO catalysts.

Table 5

Table 6

Distribution of weak, medium and strong surface basic sites and total amount of

CO₂desorbed.

Amount of carbon deposited on Ni/MgOpa, Ni/MgOp and Ni/MgOd catalysts after

SR and OSR of ethanol at 773 K.

Sample

Basic sites distribution (area%)

Total evolved CO₂

Sample

Reaction

condition

mgC gcat⁻¹h⁻¹

mgC gcat⁻¹h⁻¹mole

reacted ethanol⁻¹

Weak

Medium

Strong

(␮mol g⁻¹

)

SR 3:1

SR 10:1

OSR

SR 3:1

6.2

4.5

0.3

6.2

6.9

0.103

0.063

0.003

0.163

0.230

Ni/MgOpa

Ni/MgOp

Ni/MgOd

22

11

15

31

39

40

47

50

45

19966

14801

4575

Ni/MgOpa

Ni/MgOp

Ni/MgOd

signiﬁcantly affected the amount of basic sites, following the order:

Ni/MgOpa > Ni/MgOp > Ni/MgOd.

Therefore, in this work, the peaks corresponding to the evolution

of CO₂during the TPD experiment can be related to the different

carbonate species observed. The peaks at low temperature region

(300 and 500 K) are assigned to decomposition of hydrogen car-

bonate and bidentate carbonate species, respectively, while the CO₂

released above 500 K is ascribed to the decomposition of monoden-

tate and polydentate carbonate species. The relative contribution

of each individual desorption peak was obtained by decomposition

of the desorption curves, considering the adsorption of four differ-

ent carbonate species. A multiple-Gaussian function was selected

for ﬁtting the experimental data. The experimental curve was ﬁt-

ted quite well as it was shown in Fig. 8. The quantitative data

obtained by integrating the decomposed peaks are reported in

Table 5. Regardless the MgO preparation method, all samples exhib-

ited a higher fraction of high basic strength sites (monodentate

and polydentate carbonate species). There is only a slight differ-

ence between the distribution of weak and medium basic strength

sites. This result did not agree with the work of Menezes et al.

[21], who reported signiﬁcant differences in basic site distribu-

tion as a function of the MgO preparation method. The apparent

contradictory results could be attributed to the pretreatment con-

ditions. In the present work, the samples were previously reduced

before the adsorption of CO₂, which could expose or create coor-

dinatively unsaturated sites that are responsible for the basicity of

surface oxides. Ni could also promote the reduction of the support

by hydrogen spill over from the metallic particle to the support.

The total amount of basic sites were also calculated and listed in

Table 6. The results showed that the preparation method of MgO

3.4. SR of ethanol over Ni/MgO catalysts

Ethanol conversion and product distributions as a function of

time on stream (TOS) for SR of ethanol at 773 K over Ni/MgO cata-

lysts are shown in Fig. 9. Ni/MgOpa exhibited the highest initial

ethanol conversion, which is likely due to the higher reduction

degree as revealed by TPR and XANES experiments. The main prod-

ucts formed were H₂, CO₂, CO, and CH₄, indicating that SR of ethanol

and ethanol decomposition are the main reactions taking place.

Small amounts of acetaldehyde were also observed for Ni/MgOp

(3%) and Ni/MgOd (5%) catalysts. The activity of the MgO sup-

ports for the SR of ethanol is low. In addition, H₂, acetaldehyde and

ethylene were the main products formed, suggesting that ethanol

dehydrogenation and ethanol dehydration reactions are favored on

the supports. Frusteri et al. [11] studied the performance of 3 wt%

Rh/MgO, 3 wt% Pd/MgO, 21 wt%Ni/MgO, and 21 wt% Co/MgO cata-

lysts for the SR of ethanol. These catalysts exhibited similar product

distributions characterized mainly by H₂, CO₂, and CO. CH₄and

acetaldehyde were also observed. All catalysts deactivated whereas

the supports remained quite stable during reaction. These results

suggest that catalyst deactivation occurs on the surface of metallic

particle or on the metal-support interface.

Catalyst deactivation during SR of ethanol is generally attributed

to the deposition of carbonaceous species as well as either the

sintering or oxidation of metallic particles [8]. Carbon formation

Please cite this article in press as: G.T. Wurzler, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.11.020

G Model

APCATA-15647; No. of Pages14

ARTICLE IN PRESS

G.T. Wurzler et al. / Applied Catalysis A: General xxx (2015) xxx–xxx

11

100

90

80

70

60

50

40

30

20

10

0

100

A

B

90

80

H₂

70

60

X ethanol

50

40

X ethanol

30

CO₂

CO

20

CO

10

CH₄

acetaldehyde

CH₄

0

2

4

6

8

10 12 14 16 18 20 22 24 26 28

Time (h)

0

2

4

6

8

10 12 14 16 18 20 22 24 26 28

Time (h)

100

90

80

70

60

50

40

30

20

C

H₂

X ethanol

CO₂

CO

10

acetaldehyde

CH₄

0

2

4

6

8

10 12 14 16 18 20 22 24 26 28

Time (h)

Fig. 9. Ethanol conversion and product distribution as a function of TOS for SR of ethanol at 773 K over Ni/MgO catalysts, using H2O/ethanol molar ratio of 3.0. (A) Ni/MgOpa;

(B) Ni/MgOp; (C) Ni/MgOd.

has been reported over Ni/MgO catalyst during SR of ethanol

[11–15,18–20]. In order to investigate the main causes of Ni/MgO

catalysts deactivation during SR of ethanol, SEM and TG of used

catalysts and in situ XRD were carried out and the results obtained

are discussed in the following section.

This result suggested that the higher basicity of the support mini-

mized carbon deposition.

According to the mechanism proposed, ethanol adsorbs disso-

ciatively as ethoxy species on a Lewis acid sites-strong basic sites

pairs. Then, ethoxy species is dehydrogenated, producing acetalde-

hyde that may be further oxidized to acetate species with either

surface oxygen or hydroxyl group of the support [8]. These acetate

species may be decomposed to CH_xand carbonate species and

ﬁnally CO₂. The unbalance between the rate of the decomposition

of acetate species to CO₂and CH_xspecies and the rate of desorption

of CH_xspecies as CH₄leads to the accumulation of carbon deposits.

This in turn obstructs the metal-support interface, which results in

catalyst deactivation, because the acetate demethanation reaction

is promoted by the metal. As it is proposed for the SR of methane,

increasing the basicity of the support increases the rate of carbon

gasiﬁcation reaction, decreasing carbon accumulation [10].

Metal sintering is also another possible cause of catalyst deacti-

vation. Then, the changes in the Ni⁰crystallite size during reduction

and SR of ethanol reaction were monitored by in situ XRD. Table 7

listed the average crystallite size of Ni⁰after reduction at 1023 K and

after SR of ethanol reaction at 773 K for Ni supported catalysts. After

reduction, the average crystallite size of Ni⁰was approximately

the same for all catalysts (7.9–10.7 nm). The Ni⁰crystallite size

remained unchanged during the reaction for all catalysts, indicating

that Ni sintering does not occur during SR of ethanol at 773 K.

3.5. Characterization of post-reaction Ni supported catalysts

The SEM images of post-reaction catalysts showed the presence

of carbon ﬁlaments for all Ni/MgO catalysts (Fig. 10). However, the

extent of the formation of carbon ﬁlaments was very limited on

Ni/MgOpa. From TG analysis, the amount of carbon deposited on

Ni/MgO catalysts prepared by different methods was calculated and

the results are listed in Table 6.

After SR of ethanol at 773 K, the amount of carbon formed

was approximately the same for all catalysts (6.2–6.9 mgC/gcat/h).

However, any comparison concerning the amount of carbon formed

between all samples is somewhat hampered by the fact that the

samples present rather different deactivation behavior. Then, the

amount of carbon deposited was calculated in mgC/gcat/h/mol of

converted ethanol and the results are listed in Table 6. The amount

of carbon deposited was signiﬁcantly affected by the preparation

method of MgO. The lowest carbon deposition was observed on

Ni/MgOpa. In fact, the amount of carbon deposited followed the

reverse order observed for the total amount of basic sites (Table 5).

Please cite this article in press as: G.T. Wurzler, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.11.020

G Model

APCATA-15647; No. of Pages14

ARTICLE IN PRESS

G.T. Wurzler et al. / Applied Catalysis A: General xxx (2015) xxx–xxx

Table 7

12

Average crystallite size of Ni⁰after reduction at 1023 K and after SR of ethanol

reaction.

Sample

Condition

Crystallite size Ni⁰(nm)

Reduced to 750 ^◦

Post SRE

C

8.4

8.9

10.7

7.9

Ni/MgOpa

Reduced to 750 ^◦

Post SRE

Ni/MgOp

Ni/MgOd

Reduced to 750 ^◦

Post SRE

7.3

Therefore, these results demonstrate that Ni/MgO catalysts

deactivation is mainly due to carbon deposition.

Different strategies have been proposed to minimize or inhibit

carbon formation such as the increase of water/ethanol ratio and

the addition of oxygen to the feed [8]. Since Ni/MgOpa exhibited the

lowest carbon formation for SR of ethanol with H₂O/ethanol = 3.0,

this catalyst was tested on the SR of ethanol with a H₂O/ethanol

ratio of 10.0 and on the OSR of ethanol with a O₂/ethanol ratio of

0.5. The results obtained were shown in Fig. 11.

Increasing the H₂O/ethanol molar ratio increased the selec-

tivity to H₂and CO₂and decreased the formation of CO and

CH₄. These results are likely due to the promotion of the water

gas-shift and steam reforming of methane reactions in the

presence of higher water content in the feed. In addition, it was

also noticed that ethanol conversion only slightly decreased during

reac-

tion.

The addition of oxygen to the feed increased the ethanol con-

version. Compared with SR, the selectivity to H₂and CO decreased,

whereas the formation of CO₂increased during OSR. These results

could be attributed to the combustion reaction. The simultaneous

addition of O₂and H₂O to the feed signiﬁcantly improved catalyst

stability.

TG analysis of the used catalysts after SR of ethanol with

a H₂O/ethanol = 10 and OSR at 773 K was carried out and the

amount of carbon formed is also listed in Table 6. Increasing the

H₂O/ethanol molar ratio signiﬁcantly decreased the formation of

carbon, whereas the addition of oxygen to the feed almost com-

pletely eliminated the deposition of carbon.

Increasing the H₂O/ethanol molar ratio or the addition of oxygen

to the feed promotes the gasiﬁcation of carbon by water and the car-

bon oxidation, and this, in turn, decreases the catalyst deactivation

rate through continuous carbon removal [8,16,44–47].

4. Conclusions

Ni supported on MgO catalysts were prepared by three differ-

ent synthesis routes: precipitation followed by aging; precipitation

and decomposition of the precursor salt. The preparation method

of MgO signiﬁcantly affected the total amount of basic sites and

the reducibility of the sample. TPD of adsorbed CO₂showed that

the basicity followed the order: Ni/MgOpa > Ni/MgOp > Ni/MgOd.

TPR proﬁles revealed the presence of NiO particles and a NiO–MgO

solid solution, which was identiﬁed by XRD. The reducibility

depended on the amount of NiO–MgO solid solution. TPR and

XANES showed that the reducibility increased when the fraction of

NiO particles increased. Ni/MgOpa exhibited the highest reduction

degree.

Regardless the MgO preparation method, all Ni/MgO catalysts

deactivated during SR of ethanol at 773 K. SEM and TG of post-

reaction catalysts revealed that catalyst deactivation was mainly

due to carbon deposition. However, the amount of carbon deposited

varied signiﬁcantly depending on the preparation method of MgO.

Carbon amount formed during SR of ethanol decreased as the basic-

Fig. 10. SEM images of Ni/MgO samples after SR of ethanol at 773 K, using

H₂O/ethanol molar ratio of 3.0; (A) Ni/MgOpa; (B) Ni/MgOp; (C) Ni/MgOd.

Please cite this article in press as: G.T. Wurzler, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.11.020

G Model

APCATA-15647; No. of Pages14

ARTICLE IN PRESS

G.T. Wurzler et al. / Applied Catalysis A: General xxx (2015) xxx–xxx

13

100

90

80

70

60

50

40

30

20

10

0

A

X ethanol

H₂

CO₂

CO

CH₄

0

2

4

6

8

10 12 14 16 18 20 22 24 26 28

Time (h)

100

90

80

70

60

50

40

30

20

10

0

X ethanol

B

H₂

CO₂

CO

CH₄

10 12 14 16 18 20 22 24 26 28

Time (h)

0

2

4

6

8

Fig. 11. Ethanol conversion and product distribution as a function of TOS over Ni/MgOpa catalyst for: (A) SR of ethanol at 773 K, using H₂O/ethanol molar ratio of 10.0; (B)

OSR of ethanol at 773 K, using H₂O/ethanol/O₂molar ratio = 3.0/1.0/0.5.

ity increased, which is likely due to the promotion of the rate of

carbon gasiﬁcation.

Acknowledgements

In order to minimize or inhibit carbon formation, the

water/ethanol ratio of the feed was increased or oxygen was added

to the feed. The increase of H₂O/ethanol molar ratio favored the

water gas-shift and steam reforming of methane reactions. The

addition of oxygen to the feed decreased the selectivity to H₂

due to the combustion reaction. Increasing the H₂O/ethanol molar

ratio or the addition of oxygen to the feed promoted the gasi-

ﬁcation of carbon formed by water and the carbon oxidation,