Behaviour of nickel–alumina spinel (NiAl2O4) catalysts for isooctane steam reforming

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 5 ) 1 e8

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/he

Behaviour of nickelealumina spinel (NiAl2O4) catalysts for isooctane steam reforming
lez, Zouhair Boukha, Beatriz de Rivas, Cristina Jim
enez-Gonza
n Gonza
lez-Velasco, Jose Ignacio Guti
errez-Ortiz, Juan Ramo
pez-Fonseca* Rub
en Lo Chemical Technologies for Environmental Sustainability Group, Department of Chemical Engineering, Faculty of Science and Technology, University of The Basque Country UPV/EHU, P.O. Box 644, E-48080 Bilbao, Spain

article info

abstract

Article history:

The suitability of NiAl2O4-based catalysts for steam reforming of isooctane, which is used

Received 11 July 2014

here as a surrogate for gasoline, was examined at moderate temperatures (600e700
C)

Received in revised form

during a relatively prolonged time on stream (about 30 h). A series of catalysts with a

2 December 2014

varying nickel loading in the 10e33 wt.% range were prepared by co-precipitation. This

Accepted 12 January 2015

synthesis route was shown to be adequate to produce nickel/alumina catalysts with a

Available online xxx

small Ni particle size (8e10 nm). The catalytic behaviour and the extent of the observed loss of conversion were mainly controlled by the available metallic surface area. Coke

Keywords:

formation in the form of both amorphous and graphitic filamentous carbon was identified

Co-precipitation

as the main reason for deactivation. Thus, the best catalytic results corresponded to a

Nickel aluminate

17 wt.Ni% loading since it showed the lowest specific carbon deposition per surface area of

Isooctane

metallic sites. At elevated reforming temperatures (650 and 700
C) noticeably higher

Steam reforming

conversion values were obtained and the stability was enhanced, which suggested that

Hydrogen

coking did not induce a marked effect on the catalytic behaviour.

Coke

Copyright © 2015, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.

Introduction Catalytic steam reforming of gasoline, compared to partial oxidation and oxidative steam reforming, has the advantage of producing larger hydrogen yield which may be used in stationary and mobile fuel cells applications [1]. Noble metals catalysts offer high activity and low carbon deposition [2], but their implementation may be cost-prohibitive. Ni-based catalysts are interesting alternatives due to their competitive costs although they are susceptible to deactivation essentially due to

active phase sintering, carbon deposition and sulphur poisoning [3]. For this reason, design of new formulations capable of developing suitable metal/support interactions, with the aim of improving the activity and stability, is essential. It is widely accepted that over Ni/alumina catalysts the reforming reactions are sensitive to the structure of the Ni initial precursors. In this sense, nickel aluminate (NiAl2O4) has been proposed as a catalytic precursor that, after its reduction, leads to metallic nickel with strong Ni/Al2O3 interactions [4,5]. Moreover, in our previous studies we showed that the NiAl2O4 preparation method influenced the distribution of the metallic

* Corresponding author. Tel.: þ34 94 6015985; fax: þ34 94 6015963.
pez-Fonseca). E-mail address: [email protected] (R. Lo http://dx.doi.org/10.1016/j.ijhydene.2015.01.064 0360-3199/Copyright © 2015, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.


nez-Gonza
lez C, et al., Behaviour of nickelealumina spinel (NiAl2O4) catalysts for Please cite this article in press as: Jime isooctane steam reforming, International Journal of Hydrogen Energy (2015), http://dx.doi.org/10.1016/j.ijhydene.2015.01.064

2

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 5 ) 1 e8

Ni species and the resultant activity and stability in various methane reforming processes [6,7]. Nevertheless, little is known about the performance of these nickel catalysts in the reforming of heavy hydrocarbons, especially in the case of gasoline [8,9]. Most of the studies related to catalyst screening for the reforming of gasoline select isooctane as a model component. There are two main types of reaction to yield hydrogen from isooctane, namely partial oxidation (Reaction 1) and steam reforming (Reaction 2). Also the combined process known as oxidative steam reforming can be used. Steam reforming, although it is an endothermic process and requires an extra input of energy for vaporisation of water, is usually preferred due to its larger H2 yield.

iC8H18 þ 4O2 / 8CO þ 9H2

(1)

iC8H18 þ 8H2O / 8CO þ 17H2

(2)

The main objective of this work is therefore to analyse the efficiency of NiAl2O4 loaded alumina systems prepared by coprecipitation in isooctane (typical molecule found in gasoline) steam reforming. A special attention was paid, by varying the Ni loading, to determining correlations among catalytic behaviour, physico-chemical properties and resistance to deactivation. Results for a bulk NiAl2O4 were used as a reference for comparison. Further, spent catalysts were characterised in order to define the main deactivating phenomena.

Catalyst characterisation The catalysts were characterised by N2 physisorption at 196
C, wavelength dispersive X-ray fluorescence (WDXRF), X-ray diffraction (XRD), transmission electron microscopy (TEM), temperature programmed reduction with hydrogen (H2-TPR) and temperature programmed desorption of NH3 (NH3-TPD). The spent samples were characterised by XRD, TEM, thermogravimetry coupled to mass spectrometry (TGAMS) and Raman spectroscopy. The experimental details of each analytical technique are described elsewhere [7,11].

Catalytic tests The catalytic tests for steam reforming of isooctane were performed, on 125 mg of catalyst, in a flow reactor operating at atmospheric pressure. The reaction mixture was composed of 1.95 vol% iC8H18 and H2O (H2O/C ¼ 3) diluted in N2 with a total flow rate of 800 cm3 min1 (7500 cm3 iC8H18 g1 h1, 1.95% iC8H18/46.8%H2O/N2.). Prior to the reaction, the nickel catalysts were activated by reduction with 5%H2/N2 at 850
C for 2 h whereas the rhodium catalyst was reduced at 700
C. The experiments were carried out at constant temperature (in the 600e700
C range) for about 30 h. Feed and effluent streams were analysed online by a MicroGC (Agilent 3000) equipped with a TCD detector. On the basis of the molar flow at the inlet and outlet of the reactor, conversion and yields to the main reforming products were calculated according to the following equations: XðiC8 H18 Þ; % ¼

Experimental

(3)

YðH2 Þ ¼

FðH2 out Þ 9$FðiC8 H18 in Þ

(4)

YðCOÞ ¼

FðCOout Þ 8$FðiC8 H18 in Þ

(5)

Catalysts preparation Three alumina-supported NiAl2O4 catalysts were prepared by co-precipitation. The process was conducted by the drop-bydrop addition under constant stirring of a 0.6 M solution of NH4OH into an aqueous solution of a mixture of Ni(CH3eCOO)2$4H2O and Al(NO3)3$9H2O (1:2 Ni/Al molar ratio) and powered g-Al2O3 alumina (133 m2 g1, 0.3e0.5 mm, SA 6173, Saint-Gobain). The temperature was kept at 25
C during the precipitation and the pH was fixed at 8. Afterwards the precipitates were aged for 30 min before being filtered and washed with hot deionised water. The use of nickel acetate instead of nickel nitrate can give additional advantages in terms of dispersion and crystallite size of the resultant catalyst [10]. On the other hand, it should be noted that there were no changes in the transition alumina structure after calcination at 850
C for 8 h in comparison with the as-received sample. The catalysts were labelled as CP/A(10), CP/A(17) and CP/ A(24) and the corresponding nickel loadings were 10, 17 and 24 wt.%, respectively. Also a NiAl2O4 bulk catalyst (CP(33) sample with a 33 wt.% Ni loading) was prepared following the same route without using the commercial alumina as a support. All the samples were dried at 110
C overnight and then calcined at 850
C in static air for 4 h at a heating rate of 10
C min1. As a reference reforming catalyst a commercial Rh/A sample was used (1%Rh/Al2O3, 132 m2 g1, Alfa Aesar).

FðCOout Þ þ FðCO2 out Þ þ FðCH4 out Þ $100 8$FðiC8 H18 in Þ

YðCO2 Þ ¼

FðCO2 out Þ 8$FðiC8 H18 in Þ

(6)

YðCH4 Þ ¼

FðCH4 out Þ 8$FðiC8 H18 in Þ

(7)

This way of calculating the conversion of the reforming reaction reflects the activity of the investigated catalysts in the conversion of the feed into the main reforming products (CO, CO2 and CH4). On the other hand, the thermodynamic data were calculated via the HSC Chemistry software package by the GIBBS program using the so-called Gibbs Energy Minimization Method.

Results and discussion Characterisation of the fresh samples Table 1 summarises the characterisation results of the reduced nickel aluminate catalysts. XRD and H2-TPR analyses pointed out the efficiency of the synthesis route based on co-


nez-Gonza
lez C, et al., Behaviour of nickelealumina spinel (NiAl2O4) catalysts for Please cite this article in press as: Jime isooctane steam reforming, International Journal of Hydrogen Energy (2015), http://dx.doi.org/10.1016/j.ijhydene.2015.01.064

3

e 1.3 1.3 1.3 1.3 e 61 74 67 66 e 12 11 11.5 11 e 106 139 123 128 e 23 34 40 27

NiO

Al2O3

e 9.5 20 33 40

(a)

CP/A(10)

(c)

f

e

d

c

b

Values in parentheses corresponded to the calcined catalytic precursors. Determined by NH3-TPD. Determined by H2-TPR for the calcined catalytic precursors. Determined by TEM. Values in parentheses were determined by XRD. Determined by TGA-MS. Determined by Raman spectroscopy.

(a)

a

(133) (74) (77) (94) (109)

4.7 4.3 5.5 3.6 6.1

e 5.6 4.2 2.8 1.3

e 9/47/44 7/19/74 6/19/75 10/0/90

e 10 (11) 9.5 (9) 9.5 (9) 8.5 (11)

(b)

55 57 84 100

SBET, m g

NiAl2O4

Al2O3

Al2O3 CP(33) CP/A(24) CP/A(17) CP/A(10)

Total acidity, Hydrogen NiO species Ni Ni dispersion, Ni surface area, SBET, m g mmol m2,b consumption, contribution, crystallite size, %c m2Ni g1 ,d mmol g1,c % (a/b/g)c nmd

2 0

Freshly reduced catalysts

1,a 2

Samples

Table 1 e Characterisation results of the reduced nickel catalysts.

Ni0

C, graphite

Relative Intensity, a. u.

Ni0 Coke, wt.%e ID/IGf crystallite size, nmd

precipitation to produce samples with a predominant presence of NiAl2O4 phase with a relatively low contribution of free NiO species (<10%). Hence, the set of diffraction peaks at 2q ¼ 19.3
, 31.5
, 37.2
, 45.2
, 59.9
and 65.7
, assignable to the nickel aluminate phase (JCPS 78-1601), were clearly observed in Fig. 1((a) samples) [12]. This finding was in line with the notable H2 uptake of the samples at high temperature (>600
C) observed in the H2-TPR profiles (Supplementary information, Fig. 1S), which was associated with the reduction of the spinel phase (the reduction of the so-called b-NiO and gNiO species). In our previous study the splitting of this reduction band, was associated with the different coordination of Ni2þ ions in the inverse spinel structure [7]. Moreover, it was concluded that Ni2þ ions occupying octahedral sites (bNiO) were more prone to be reduced than the tetrahedral ones (g-NiO species). In addition, as deduced from the contribution of the b and g peaks, reported in Table 1, the Ni2þ ions preferentially adopted a tetrahedral coordination at lower Ni content (CP/A(10)) while they tended to occupy octahedral sites at higher Ni content (CP(33)). Contrastingly the

1

Used catalysts

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 5 ) 1 e8

CP/A(17) (c) (b)

(a)

CP/A(24) (c) (b)

(a)

CP(33)

(c) (b)

(a)

10

20

30

40

50

60

70

80

Angle, 2θ Fig. 1 e XRD patterns of the nickel catalysts: (a) calcined, (b) reduced and (c) used samples.


nez-Gonza
lez C, et al., Behaviour of nickelealumina spinel (NiAl2O4) catalysts for Please cite this article in press as: Jime isooctane steam reforming, International Journal of Hydrogen Energy (2015), http://dx.doi.org/10.1016/j.ijhydene.2015.01.064

4

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 5 ) 1 e8

reducibility of the samples at low temperatures (<500
C), where the reduction of free a-NiO species occurred, was comparatively less noticeable. Note that the presence of these a-NiO species was not apparently observed in the XRD patterns probably due to their low concentration and/or high dispersion [12]. The surface distribution of nickel species laying on the surface of the prepared catalysts were investigated by means of XPS analysis (Fig. 2S). The position of the main XPS peak in the Ni 2p3/2 region was located between 855.7 and 858.6 eV, thereby evidencing the presence of NiAl2O4, and seemed to depend on Ni content. Indeed, the sample with the highest Ni loading (CP(33)) shifted the Ni 2p3/2 main peak towards lower binding energies (855.7 eV). Reports in the literature have correlated this XPS peak shift with changes in the ionic coordination of Ni [13]. Furthermore, it was shown that the binding energy obtained for a cation in octahedral sites was generally lower than in tetrahedral sites. Accordingly, the shift observed in the case of the CP(33) sample was consistent with the predominance on the surface of Ni2þ ions in octahedral symmetry. By contrast, in the case of the sample presenting a higher binding energy (CP/A(10) catalyst) an enrichment in tetrahedral Ni2þ ions could be expected. It should be pointed out that, these XPS results were in good agreement with those obtained from the H2-TPR analysis. After reduction at 850
C for 2 h with 5%H2/N2 it was found that the Ni2þ species were massively reduced into metallic Ni (JCPDS 89-7128, peaks at 2q ¼ 44.6
, 52
and 76.5
). The degree of the reduction of nickel aluminate depends on the calcination temperature required to obtain the spinel. Normally, a high calcination temperature (around 800
C) is necessary to induce the formation of the stoichiometric phase [4]. If this temperature is higher the stability of the resultant spinel increases and this will lead to higher temperatures (900
C or even higher) to fully reduce the material as shown by Muroyama et al. [8]. On the other hand, the reduction of the nickel aluminate spinel phase simultaneously involved the formation of the alumina phase. Although this process was expected for all the calcined precursors, it could be more easily observed in the bulk sample, CP(33), since it did not contain alumina as a support in the calcined catalyst formulation. Hence, after reduction the alumina signals at 2q ¼ 19.7
, 31.6
, 37.8
, 39.6
, 45.7
and 67.1
(JCPDS 79-1558) were visible (Fig. 1-(b) samples). This severe reduction step, which provoked the transformation of the NiAl2O4-based precursors into active Ni/ Al2O3 systems, resulted in a decrease by about 25e60% of the surface area compared to bare alumina (Table 1). Hence, the reduced CP/A(10) catalyst showed the highest surface area (100 m2 g1). The average nickel particle size was estimated by TEM from the measurement of the size of 300 particles (Fig. 3S). The particle size distribution profiles were characterised by a relatively symmetrical peak for all the samples, being 99% of the measured particle sizes lower than 20 nm. Thus, the estimated average size was 10 nm for the bulk CP(33), 9.5 nm for CP/A(24), 9.5 nm for CP/A(17) and 8.5 nm for CP/A(10) (Table 1). These values were in good agreement with those estimated by XRD. From this average particle size the corresponding values (Table 1) for nickel dispersion (D, %) and specific nickel

surface area (SNi ; m2Ni g1 ) were calculated by the following equations:  D ¼ n$

SNi ¼

d Dat

2 $

SBET $MWNi  1020 Aimage $CNi $NA

Aat $CNi $D$NA $1022 MWNi $

(8)

(9)

where d (nm) is the average particle size, Dat is the diameter of the atomic nickel section (0.296 nm), SBET is the surface area of the reduced catalysts (m2 g1), n is the number of particles in the selected area (Aimage, nm2) of the analysed TEM micrograph, CNi is the nickel content of the sample (%wt.), NA is the Avogadro number and MWNi is the nickel molecular weight. Hence, the largest specific Ni surface area was shown by the CP/A(17) catalyst (40 m2 g1). Finally, the acid properties of the reduced samples were evaluated by NH3-TPD (Table 1). As aforementioned, the XRD results revealed that reduced CP(33) also contained alumina together with metallic nickel as products of its reduction process. This could explain the comparable values of its SBET and NH3 uptake, in comparison with the other samples. While the overall acidity of the CP/A(10) sample was slightly lower (609 mmol NH3 g1) than that of the bare support (630 mmol NH3 g1), it remarkably decreased in the samples with a higher Ni loading (about 300e315 mmol NH3 g1), suggesting that metallic Ni covered the major part of the support surface acid sites. In order to stress the evolution of the surface acidity of the analysed samples, the data corresponding to the adsorbed NH3 were referred to 1 m2 of the catalyst (Table 1). The comparison was also made between data for Ni/ Al2O3 referred to the surface corresponding to the alumina support only. In fact, when the data were compared as suggested above, the slight modification of the chemical properties of the alumina by the nickel deposition was fully confirmed in the case of CP(33), CP/A(17) and CP/A(10) (7e8.3 mmol m2). However the CP/A(24) sample seemed to present a significantly higher value (13.7 mmol m2).

Catalytic activity with time on stream and characterisation of the used catalysts Fig. 2 shows the evolution of isooctane conversion and H2 yield with time on stream, at 600
C and 7500 cm3 iC8H18 g1 h1, over the reduced nickel aluminate-based catalysts. The activity data of the commercial rhodium catalyst were also included for the sake of comparison. In these experiments the catalysts were intentionally run under conditions giving incomplete conversion of isooctane, to allow meaningful reactivity and stability comparison among the various catalysts. In this sense, note that the used volume hourly space velocity was appreciably higher than those found in the literature for the steam reforming of this liquid hydrocarbon over nickel catalysts (between 1400 and 4900 cm3 iC8H18 g1 h1) [14e16]. All the samples exhibited a marked initial conversion with values around 45e53% and their corresponding H2 yield values ranged between 0.9 and 1.3. However, conversion rapidly decreased during the first 5 h. At longer time intervals (>5 h),


nez-Gonza
lez C, et al., Behaviour of nickelealumina spinel (NiAl2O4) catalysts for Please cite this article in press as: Jime isooctane steam reforming, International Journal of Hydrogen Energy (2015), http://dx.doi.org/10.1016/j.ijhydene.2015.01.064

5

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 5 ) 1 e8

60

CP/A(10)

CP(33) CP/A(24) CP/A(17)

Rh/A

X(iC8H18), %

50

40

30

20

0

5

10

15

20

25

30

Time, h 1.4

CP/A(10)

CP(33) CP/A(24) CP/A(17)

Rh/A

Y(H2)

1.2

1.0

0.8

0.6

0

5

10

15

20

25

30

Time, h Fig. 2 e Evolution of conversion and H2 yield with time on stream for the various nickel catalysts at 600
C. Reaction conditions: W ¼ 0.125 g; 7500 cm3 iC8H18 g¡1 h¡1; Gas mixture: 1.95%iC8H18/46.8%H2O/N2.

although a continuous loss of activity was still observed, this was less pronounced and tended to reach a stationary state. The same evolution of conversion with time on line was also reported by Al-Musa et al. [16] and Westrich et al. [17] in the catalytic steam reforming over Cu/CeO2 and catalytic

decomposition of isooctane over Ni/CeZr, respectively. Thus, this conversion-time on line profile could be interpreted in terms of a fast initial accumulation of carbon (probably due to cracking reaction on the acidic sites) that eventually led to a balance between carbon deposition and gasification with water at extended time span. This in turn resulted in a decreased but stable conversion. The greatest loss of activity (%), evaluated as the difference between initial conversion and conversion at 31 h divided by the initial conversion, was observed for the CP(33) and CP/ A(10) samples (50 and 39%, respectively). By contrast, the other two samples, namely CP/A(17) and CP/A(24), showed a significantly lower loss (about 28%). Likewise, H2 yield was not stable with time on line and simultaneously suffered a significant decrease. The comparison of the nickelealumina spinel catalysts performance revealed that their activity followed this trend: CP/A(17) > CP/A(24) > CP/A(10) > CP(33). The catalytic behaviour was mainly controlled by their increased nickel surface area. Indeed, the catalysts with a higher metallic surface area, 40 m2 g1 for CP/A(17) and 34 m2 g1 for CP/A(24), showed the best catalytic behaviour. Moreover these catalysts displayed a better efficiency compared to the alumina-supported rhodium catalyst, thereby evidencing the potential of the nickel aluminate precursor prepared by coprecipitation for producing active Ni/Al2O3 samples. However, the bulk catalyst (33 wt.%Ni) showed the poorest performance since its activity and H2 yield did not exceed 27% and 0.7, respectively. Table 2 summarises the obtained results after 31 h of time on stream in terms of isooctane conversion, yield to main reforming products and H2/CO and CO/CO2 ratios. The reported data evidenced that the most active CP/A(17) and CP/ A(24) catalysts had a similar efficiency (40e42%), H2 yield (0.96), CO yield (0.07e0.09), CO2 yield (0.26e0.27) and CH4 yield (0.06e0.07). Irrespective of the used catalyst, the H2/CO ratio values were higher than 12.2. This clearly indicated that the contribution of the water gas shift Reaction (10) was considerably relevant as also evidenced by the low CO/CO2 ratios. As a result of its slightly higher activity in this reaction, the CP/ A(17) sample gave the lowest H2/CO and the higher CO/CO2 ratios. On the other hand, the formation of methane was attributed to the Reaction (11).

CO þ H2O / CO2 þ H2

(10)

CO þ 3H2 / CH4 þ H2O

(11)

Table 2 e Catalytic performance of the various nickel catalysts at 600
C after 31 h time on stream. Catalyst CP(33) CP/A(24) CP/A(17) CP/A(10) Rh/A Equilibrium

X(iC8H18), %

Y(H2)

Y(CO)

Y(CO2)

Y(CH4)

H2/CO

CO/CO2

30 40 42 27 30 100

0.70 0.96 0.96 0.80 0.80 2.15

0.05 0.07 0.09 0.06 0.08 0.35

0.20 0.27 0.26 0.21 0.16 0.56

0.05 0.06 0.07 0.00 0.06 0.09

15.0 14.5 12.2 14.8 11.3 7.0

0.27 0.28 0.34 0.29 0.48 0.61

Reaction conditions: W ¼ 0.125 g; 7500 cm3 iC8H18 g1 h1; Gas mixture: 1.95%iC8H18/46.8%H2O/N2.


nez-Gonza
lez C, et al., Behaviour of nickelealumina spinel (NiAl2O4) catalysts for Please cite this article in press as: Jime isooctane steam reforming, International Journal of Hydrogen Energy (2015), http://dx.doi.org/10.1016/j.ijhydene.2015.01.064

iC8H18 / 8C þ 9H2

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 5 ) 1 e8

55

CO þ H2 / C þ H2O

(13)

(14)

The observed deactivation of the catalysts could be related to several causes such as coke formation, which might be mainly caused by Reactions (12e14), nickel sintering and partial nickel oxidation of the metallic nickel to NiO. Therefore, in order to determine the state of the used catalysts and thus analyse the extent of these potential deactivating phenomena, the spent catalysts were characterised by XRD, TEM, TGA-MS and Raman spectroscopy. The total amount of carbon deposited on the spent catalysts and peak temperatures of oxidation were determined by TGA-MS. A notable formation of coke ranging between 61 and 74% was observed for all the examined catalysts (Table 1) displaying two oxidation peaks in the 450e620
C temperature range (Fig. 4S). According to the classification given by Westrich et al. [17], it is suggested that the traces corresponded to the combustion of filamentous carbon. Due to the absence of oxidation peaks below 450
C it seemed that the samples did not contain appreciable amounts of coating carbon. Further, on-line MS analysis of gaseous products during thermogravimetric analysis showed an absence of water signals (or hydrogen species signals) in the effluent gas, indicating the absence of CHx carbonaceous species which would be expected to form in the oxidation of coating carbon. In line with the temperature-programmed oxidation profiles, TEM images (Fig. 5S) confirmed that the post-reaction samples were mainly covered with filamentous carbon which in turn had an effect on the textural properties of the samples. Indeed, the specific surface area markedly increased by a 1.3e2.4 factor (Table 1) owing to the contribution of the porosity of the deposited carbon. On the other hand, the diffraction patterns of the used samples also evidenced the presence of large amounts of coke in all the samples as revealed by the characteristic peak of graphitic carbon at 2q ¼ 26.3
(JCPDS 89e8487) (Fig. 1, (c)samples). However, since the absence of amorphous carbon could not be evidenced by XRD, the spent catalysts were characterised by Raman spectroscopy as well (Fig. 6S). Two distinct bands were detected at 1340 cm1 (the so-called D band) attributed to the defective and disordered structures and 1580 cm1 (the so-called G band) attributed to ordered graphitic coke [18]. Thus, it was found that both amorphous and graphitic filamentous carbon were present, in all the used nickel catalysts, with a similar distribution since the ID/IG intensity ratio was always about 1.3 (Table 1). Fig. 3 includes the relationship among the activity loss, the specific carbon formation, referred to 1 m2 of surface Ni, and the specific Ni surface area of the catalysts. This correlation allowed us to conclude that an increased specific Ni surface area (observed for CP/A(17) and CP/A(24) samples) had a beneficial effect on the catalytic behaviour of the nickelealumina spinel catalysts. This improvement was related to the fact that the accessibility of active nickel species was not negatively affected by coking to a considerable extent.

Activity loss, %

2CO / C þ CO2

0.030

60

(12)

CP(33)

50

0.025

45 40 CP/A(10) 0.020

35

CP/A(17)

30 CP/A(24)

Specific carbon deposition, gC m-2Ni

6

0.015 25 22 24 26 28 30 32 34 36 38 40 42

Specific Ni surface area, m2 g-1 Fig. 3 e Influence of the specific Ni surface area on the specific carbon deposition and activity loss.

However, at lower metallic surface area values, case of CP(33) and CP/A(10) catalysts, the carbon deposition, probably due to the contribution of the uncovered support acid sites by catalysing coke formation, was so massive that provoked a substantial decreased conversion due to blockage of active sites and/or pores [19]. In sum, the larger accessible Ni surface area of CP/A(17) and CP/A(24) catalyst seemed to favour the reforming reactions instead of reactions leading to coking. XRD analysis of the used samples also revealed the partial oxidation of the active phase into inactive NiO as evidenced by the signal at 2q ¼ 43.3
. Besides the average Ni particle size estimated by TEM suggested that a slight sintering occurred with an increase in the size from 8-10 nm to 11e12 nm. These two findings may also induce a decrease in conversion with time on stream although it is believed that, judging from the extent of these phenomena, these were of secondary importance in comparison with coke formation. Given the marked deactivation and the resultant low final conversion observed for the nickel aluminate-based catalysts at relatively low reforming temperatures (600
C), some preliminary studies were also carried out at higher temperatures, namely 650 and 700
C. Fig. 4 includes the evolution of conversion as a function of time on line at these two temperatures for the CP/A(17) catalyst, which exhibited a similar performance to the CP/A(24) sample but with a substantially lower Ni content. Apart from exhibiting a remarkably higher conversion, the catalytic activity was also noticeably more stable with average conversion values of 86% at 650
C and 93% at 700
C. Stefanescu et al. [14] also found a very high conversion at 700
C (virtually 100%) over 15%Ni/Al2O3 coated on microstructured platelets but operating a significantly lower volume hourly space velocity (3250 cm3 iC8H18 g1 h1). The observed good stability was related to the lower formation of coke deposits as revealed by TGA-MS (namely 40 and 25% at 650 and 700
C, respectively). Furthermore, on the basis of the same oxidation peaks, it was evidenced that coke had a filamentous morphology as well. While it was true that coke deposition could not be avoided, this did not have a


nez-Gonza
lez C, et al., Behaviour of nickelealumina spinel (NiAl2O4) catalysts for Please cite this article in press as: Jime isooctane steam reforming, International Journal of Hydrogen Energy (2015), http://dx.doi.org/10.1016/j.ijhydene.2015.01.064

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 5 ) 1 e8

7


nchez)) and CIC bioGUNE (D. Gil and S. Delgado) is also (M.B. Sa gratefully acknowledged.

100 90 80

Appendix A. Supplementary data

X(iC8H18), %

70

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.ijhydene.2015.01.064.

60 50 40

references

30 CP/A(17)

20

600 ºC 650 ºC 700 ºC

10 0

0

5

10

15 20 Time, h

25

30

Fig. 4 e Evolution of conversion with time on stream over the CP/A(17) catalyst at 600, 650 and 700
C. Reaction conditions: W ¼ 0.125 g; 7500 cm3 iC8H18 g¡1 h¡1; Gas mixture: 1.95%iC8H18/46.8%H2O/N2.

crucial impact on the catalytic behaviour of the sample. Finally, it must be indicated that a slight sintering was noticed with nickel crystallite sizes around 13e14 nm.

Conclusions The catalytic behaviour of various nickel aluminate catalysts (10e33 wt.%Ni), in bulk and supported forms, prepared by coprecipitation was investigated in the steam reforming of isooctane operated at high volume hourly space velocity between 600 and 700
C. It was found that the reforming efficiency (conversion, H2 yield and stability) could be tuned by controlling the active nickel surface area of the sample and the reaction temperature. The higher activity of CP/A(17) and CP/A(24) samples was thus related to their lower specific carbon deposition. This in turn resulted in a less marked impact of coking on activity for samples with a larger nickel surface area. In this way the effect of the unavoidable filamentous coke formation on the performance could be minimised for a catalyst with a 17 wt.%Ni loading and reaction temperatures higher than 650
C.

Acknowledgements The authors wish to thank the financial support for this work provided by the Spanish Science and Innovation Ministry (CTQ2010-16752) and Ministry of Economy and Competitiveness (ENE2013-41187-R), the Basque Government (PRE_2013_2_453, IT657-13) and the University of the Basque Country (UFI 11/39). Technical and human support from ~ aga), WDXRF (F.J. Sangu¨esa) and XPS SGIker (XRD (A. Larran

[1] Cheekatamarla PK, Finnerty CM. Reforming catalysts for hydrogen generation in fuel cell applications. J Power Sources 2006;160:490e9. [2] Xie C, Chen Y, Engelhard MH, Song C. Comparative study on the sulfur tolerance and carbon resistance of supported noble metal catalysts in steam reforming of liquid hydrocarbon fuel. ACS Catal 2012;2:1127e37. [3] Zhang J, Wang Y, Ma R, Wu D. Characterization of aluminasupported Ni and Ni-Pd catalysts for partial oxidation and steam reforming of hydrocarbons. Appl Catal A Gen 2003;243:251e9. [4] Salhi N, Boulahouache A, Petit C, Kiennemann A, Rabia C. Steam reforming of methane to syngas over NiAl2O4 spinel catalysts. Int J Hydrogen Energy 2011;36:11433e9. [5] Kathiraser Y, Thitsartarn W, Sutthiumporn K, Kawi S. Inverse NiAl2O4 on LaAlO3-Al2O3: unique catalytic structure for stable CO2 reforming of methane. J Phys Chem C 2013;117:8120e30.
nez-Gonza
lez C, Boukha Z, de Rivas B, Delgado JJ, [6] Jime
lez-Velasco JR, et al. Structural Cauqui MA, Gonza characterisation of Ni/alumina reforming catalysts activated at high temperatures. Appl Catal A Gen 2013;466:9e20.
nez-Gonza
lez C, de Rivas B, Gonza
lez[7] Boukha Z, Jime
rrez-Ortiz JI, Lo
pez-Fonseca R. Synthesis, Velasco JR, Gutie characterisation and performance evaluation of spinelderived Ni/Al2O3 catalysts for various methane reforming reactions. Appl Catal B Env 2014;158e159:190e201. [8] Muroyama H, Nakase R, Matsui T, Eguchi K. Ethanol steam reforming over Ni-based spinel oxide. Int J Hydrogen Energy 2010;35:1575e81. [9] Fauteux-Lefebvre C, Abatzoglou N, Braidy N, Achouri IE. Diesel steam reforming with a nickel-alumina spinel catalyst for solid oxide fuel cell application. J Power Sources 2011;196:7673e80. [10] Wu G, Zhang C, Li S, Han Z, Wang T, Ma X, et al. Hydrogen production via glycerol steam reforming over Ni/Al2O3: influence of nickel precursors. ACS Sustain Chem Eng 2013;1:1052e62.
pez-Fonseca R, Jime
nez-Gonza
lez C, de Rivas B, Gutie
rrez[11] Lo Ortiz JI. Partial oxidation of methane to syngas on bulk NiAl2O4 catalyst. Comparison with alumina supported nickel, platinum and rhodium catalysts. Appl Catal A Gen 2012;437e438:53e62. [12] Li G, Hu L, Hill JM. Comparison of reducibility and stability of alumina-supported Ni catalysts prepared by impregnation and co-precipitation. Appl Catal A Gen 2006;301:16e24. [13] Drouet C, Laberty C, Fierro JLG, Alphonse P, Rousset A. X-ray photoelectron spectroscopic study of non-stoichiometric nickel and nickelecopper spinel manganites. Inter J Inorg Mater 2000;2:419e26. [14] Murata K, Wang L, Saito M, Inaba M, Takahara I, Mimura N. Hydrogen production from steam reforming of hydrocarbons over alkaline-earth metal-modified Fe- or Ni-based catalysts. Energ Fuels 2004;18:122e6.


nez-Gonza
lez C, et al., Behaviour of nickelealumina spinel (NiAl2O4) catalysts for Please cite this article in press as: Jime isooctane steam reforming, International Journal of Hydrogen Energy (2015), http://dx.doi.org/10.1016/j.ijhydene.2015.01.064

8

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 5 ) 1 e8

[15] Mayne JM, Tadd AR, Dahlberg KA, Schwank JW. Influence of thiophene on the isooctane reforming activity of Ni-based catalysts. J Catal 2010;271:140e52. [16] Al-Musa A, Al-Saleh M, Ioakeimidis ZC, Ouzounidou M, Yentekakis IV, Konsolakis M, et al. Hydrogen production by iso-octane steam reforming over Cu catalysts supported on rare earth oxides (REOs). Int J Hydrogen Energy 2014;39:1350e63.

[17] Westrich TA, Chen X, Schwank JW. Isooctane decomposition and carbon deposition over ceriaezirconia supported nickel catalysts. Appl Catal A Gen 2010;386:83e93. [18] Pompeo F, Nichio NN, Ferretti OA, Resasco D. Study of Ni catalysts on different supports to obtain synthesis gas. Int J Hydrogen Energy 2005;30:1399e405. [19] Bartholomew CH. Mechanisms of catalyst deactivation. Appl Catal A Gen 2001;212:17e60.


nez-Gonza
lez C, et al., Behaviour of nickelealumina spinel (NiAl2O4) catalysts for Please cite this article in press as: Jime isooctane steam reforming, International Journal of Hydrogen Energy (2015), http://dx.doi.org/10.1016/j.ijhydene.2015.01.064