Nickel / Doped Ceria Solid Oxide Fuel Cell Anodes for Dry Reforming of Methane

Neste trabalho, foi estudado o desempenho catalítico de materiais a base de Ni, suportado em cério dopado com Zr, Pr e Nb para serem usados como anodos em células a combustível de óxido sólido (SOFC) alimentadas com combustíveis contendo metano e CO2. Os anodos foram preparados pelo método hidrotérmico, usando um teor de Ni (14% em volume) menor do que o utilizado em anodos convencionais (30% em volume). Os materiais obtidos foram caracterizados através de análises de difração de raios X, redução à temperatura programada, espectroscopia Raman e termogravimetria. Os resultados mostraram que o material contendo Zr apresentou o menor tamanho de cristalito de níquel metálico, o que levou a um maior valor de atividade inicial na reação de reforma seca do metano a 1073 K. Entretanto, o catalisador Ni/CePr apresentou a menor quantidade de carbono formada. Isto foi atribuído à maior mobilidade de oxigênio do suporte CePr, o que promove o mecanismo de remoção do carbono.


Introduction
The solid oxide fuel cells (SOFC) technology has been considered as an attractive energy conversion system.It has several advantages such as high efficiency, relatively low sensitivity to impurities, and possibility for operation with an internal reformer. 1,2In the case of the direct internal reforming solid oxide fuel cell (DIR-SOFC), the complexity and costs of the fuel cell system are reduced, since the available fuels (hydrocarbons or alcohols) can be fed straight to the anode side of SOFC and reformed to H 2 without the use of an external reformer.][5] Associated gas or biogas fueled SOFC system is a potential technology for electric power generation, since it can contribute to reduce CO 2 emissions.The associated gas produced in the world's largest deepwater field in Brazil contains a significant amount of carbon dioxide that has to be removed before its use.In the case of biogas or landfill gas (LFG), this gaseous mixture containing, mainly methane and carbon dioxide, is produced by anaerobic digestion or fermentation of organic matter Vol. 25, No. 12, 2014   such as sewage sludge, municipal solid waste (MSW), etc.Therefore, the emission of biogas to the atmosphere may contribute significantly to the greenhouse gas effect.Biogas fuelled SOFC can produce electricity with high efficiency (30-40%) even in small size power generations (< 20 kW). 6hen associated gas or biogas is fed straight to SOFC, H 2 and CO are produced by the CO 2 reforming of methane, the so called dry reforming of methane (DRM).
8][9] Ni provides electronic conductivity, while YSZ provides the ionic conductivity and thermal stability.However, in order to exceed the percolation threshold for electronic conductivity, a high Ni content (above 30 vol.%) is necessary.Since Ni is very active to reforming as well as cracking reactions, such a high volume fraction favors carbon formation when hydrocarbons such as methane are used directly as a fuel in SOFC.The carbon deposits on metal sites of anodes, resulting in the rapid degradation of the cell performance. 7,8,10It is well known that catalyst deactivation due to carbon deposition is one of the main issues of the dry reforming of methane. 11,12Therefore, the development of anodes for SOFC running on associated gas or biogas that exhibit high catalytic activity, high stability and adequate ionic and electronic conductivity at working conditions is still a challenge.Some strategies have been proposed for suppressing carbon deposition over SOFC anodes such as: (i) the addition of an oxidant to the feed; (ii) the decrease in Ni content and; (iii) the use of redox supports.
The use of an excess of carbon dioxide in the DRM reaction could avoid carbon deposition on the anodes. 13owever, the use of high CO 2 /CH 4 ratios decreases the electrical efficiency of the SOFC by the dilution of fuel, the yield of hydrogen and the system efficiency. 13ecreasing the Ni content of the anode aims at controlling the ensemble size, while keeping the appropriate electronic conductivity.The nickel particle size significantly influences the nucleation rate of carbon. 14Therefore, controlling the number of atoms in an ensemble will most likely suppress the coke formation rate.In this sense, the anode preparation method may play a critical role in determining anode stability during DRM.Some authors 15,16 reported the use of SOFC anodes with low Ni content (ca.10-18 wt.%) prepared by alternative methods.Jasisnky et al. 15 showed that the conductivity under H 2 of anodes with 14 vol.% of Ni prepared by a net shape procedure was similar to the conductivity of the cermet with 35 vol.% of Ni synthesized by a conventional technique.On the other hand, Moddaferi et al. 16 reported that Ni based anodes with low metal content (5 wt.% Ni and 5 wt.%Ru) prepared by the hydrothermal method did not exhibit carbon formation during oxidative steam reforming (OSR) of propane under SOFC operation conditions.
Another possibility is the use of redox supports such as ceria and ceria-mixed oxides instead of YSZ in order to improve catalyst stability.][19] The support participates in the dissociative adsorption of CO 2 near the metal particles, transferring oxygen to the surface of the metallic particle containing carbon deposits.Therefore, the use of supports that exhibit a high oxygen exchange capacity, such as ceria and ceria-mixed oxides, promotes the mechanism of carbon removal.Some authors 20,21 showed that adding dopants, such as Zr, to ceria structure promotes the reduction of the bulk ceria, which was attributed to an improvement of the oxygen mobility.In addition, ceria exhibits a mixed ionic and electronic conductivity due to the reduction of Ce 4+ to Ce 3+ , which improves the electrocatalytic activity of the anode.3][24][25] It was also reported that doping ceria with elements of valence higher than 4+, such as Nb 5+ , leads to an increase of electronic conductivity. 26owever, only few studies reported the use of nickel/ doped ceria anodes of SOFC running on methane and CO 2 . 27,28Bornura et al. 27 detected the carbon formation on NiCu/GdCe catalysts during dry reforming of biogas at a reaction temperature ranging from 923 to 1073 K.The catalytic behavior of Ni supported on ceria-based anodes doped with Zr, Gd or Pr for methane reforming was evaluated by Gaudillèrea et al. 28 The effect of addition of water and CO 2 to a reactant mixture containing CH 4 and O 2 was studied.However, the catalytic tests were performed at very low temperatures (673-873 K) that are not typical of SOFC operating conditions.In addition, they observed that, in the presence of CO 2 , most of Ni/ceria catalysts are not active.
Therefore, the aim of this work is to study the catalytic performance of Ni supported on ceria doped with Zr, Pr and Nb as anodes for SOFC operating directly on methane and CO 2 .The anodes were prepared by a hydrothermal method using a Ni content (14 vol%) lower than that of a conventional SOFC anodes (30 vol%).The materials obtained were characterized by X-ray diffraction (XRD), temperature-programmed reduction (TPR) and Raman spectroscopy.Thermogravimetric analyses (TG) were carried out to measure and characterize the carbonaceous deposits formed during the reaction.The catalytic performance of the samples for DRM was evaluated, using a conventional fixed-bed reactor at the reaction temperature typical of a SOFC (1073 K).

Catalyst preparation
The supports were synthesized by a hydrothermal method previously described elsewhere. 16,29An aqueous solution of precursors salts (cerium (IV) ammonium nitrate, zirconium and praseodymium nitrate, or ammonium niobium oxalate) was prepared with a Me/Ce molar ratio of 1/9 (Me = Zr, Pr or Nb).Cerium and zirconium, praseodymium or niobium hydroxides were then co-precipitated by the addition of an excess of ammonium hydroxide.The precipitate was transferred to an autoclave and heated to 453 K for 4 h.Then, the precipitate was washed with distilled water and calcined at 573 K (CeZr-573, CePr-573, CeNb-573) for 2 h in a muffle furnace.Ni (14 vol.%) was added to CeZr-573, CePr-573 and CeNi-573 supports by wet impregnation using an aqueous solution of Ni(NO 3 ).6H 2 O.After impregnation, the samples were dried at 393 K and calcined under air (50 mL min -1 ) at 1073 K according to the following temperature program: (i) from room temperature to 673 K at 0.5 K min -1 ; (ii) from 673 to 973 K at 1.7 K min -1 and (iii) from 973 to 1073 K at 10 K min -1 , remaining at the final temperature for 5 h.Then, three catalysts were obtained: Ni/CeZr, Ni/CePr and Ni/CeNb.For comparison with the catalysts, the precipitates obtained after hydrothermal treatment were also calcined at 1073 K (CeZr-1073, CePr-1073, CeNb-1073).

BET surface area
The BET surface areas of the samples were measured using a Micromeritics ASAP 2020 analyzer by nitrogen adsorption at the boiling temperature of liquid nitrogen.

X-ray diffraction (XRD)
The X-ray powder diffraction pattern of the calcined and reduced/passivated samples were obtained with CuK a radiation (l = 1.5406Å) using a RIGAKU diffractometer.Data were collected over the 2θ range of 25 to 75° using a scan rate of 0.04 degree step -1 and a scan time of 1 s step -1 .The Scherrer equation was used to estimate the crystallite mean diameters of CeO 2 , NiO, and Ni.For the measurements of the crystallite mean diameter of Ni, the calcined samples were also reduced under pure hydrogen (30 mL min -1 ) at 1023 K for 1 h, purged under N 2 at the same temperature for 30 min and cooled to 298 K.Then, the reactor was maintained at 203 K by immersing it in a mixture of isopropyl alcohol and liquid nitrogen for 1 h, and the catalyst was passivated with a 5% O 2 /He mixture.

Raman spectroscopy
The Raman spectra were recorded at room temperature using a Horiba LabRam HR-UV800/Jobin-Yvon spectrometer, equipped with He-Ne laser (l = 632 nm) of intensity 10 mW, a CCD detector (203 K), Olympus BX41 microscope, objective lens of 100× and spot 0.96 nm.
Temperature-programmed reduction (TPR) TPR measurements were carried out in a homemade apparatus.The catalyst was pretreated at 673 K for 1 h under air flow prior to the TPR experiment in order to remove traces of water.The reducing mixture (2% H 2 /N 2 ) was passed through the sample (80 mg) at a flow rate of 30 mL min -1 and the temperature was increased to 1273 K at a heating rate of 10 K min -1 .

Thermogravimetric analysis (TG)
Thermogravimetric analysis of the used catalysts was carried out in a TA Instruments equipment (SDT Q600) in order to determine the amount of carbon formed over the catalyst.Approximately 10 mg of spent catalyst was heated under air flow from room temperature to 1273 K at a heating rate of 20 K min -1 and the weight change was measured.

Reaction conditions
CO 2 reforming of methane was performed in a quartz reactor at 1073 K and atmospheric pressure.In order to avoid hot spot formation or temperature gradients, catalyst samples (25 mg) were diluted with SiC (25 mg).Prior to the reaction, the catalyst was reduced under H 2 at 1023 K for 1 h and then heated to 1073 K under N 2 .A reactant mixture with CH 4 :CO 2 ratio of 1 and a flow rate of 100 mL min -1 was used.The exit gases were analyzed using a gas chromatograph (Agilent 6890) equipped with a thermal conductivity detector and a Carboxen 1010 column (Supelco).The methane conversion was determined as follows:

Catalyst characterization
The supports calcined at 1073 K exhibited low surface areas (CeZr: 39 m 2 g -1 ; CePr: 18 m 2 g -1 e CeNb: 22 m 2 g -1 ).However, these values are higher than that obtained by Augusto et al. 29 for gadolinium-doped ceria support (ca. 8 m 2 g -1 ) prepared by the same procedure used in this work.Regarding the nature of dopant, CeZr sample exhibited the highest surface area.Some authors reported that the increase in ceria surface area by the addition of zirconia is due to the higher thermal stability of CeO 2 , avoiding the sintering process. 20,21,30he XRD patterns of supports calcined at 1073 K are shown in Figure 1.
The diffractograms of CeZr, CePr and CeNb exhibited the diffraction lines corresponding to CeO 2 .However, these lines were slightly shifted to higher 2θ positions, when the diffractograms of these samples were compared to the diffractogram of CeO 2 with cubic structure (ICSD 72155).For CeZr and CePr, these shifts indicate the formation of Ce 0.9 Zr 0.1 O 1.95 (ICSD 152478) and Ce 0.9 Pr 0.1 O 2 (ICSD 182184) solid solutions.In the case of CeNb sample, the niobium content used (5.2 wt.%) is higher than the solubility limit of Nb in CeO 2 (1.4 wt.%), which can lead to the formation of secondary phases, such as Nb 2 O 5 and CeNbO 4 . 26Nevertheless, the lines corresponding to Nb 2 O 5 and CeNbO 4 were not detected in this work.This could be attributed to the high dispersion of these phases.
For the catalysts calcined at 1073 K, in addition to the lines related to ceria and ceria solid solutions, the XRD patterns (Figure 2) also revealed the lines characteristic of NiO phase.However, Ni/CeNb sample showed less intense lines of NiO than Ni/CeZr and Ni/CePr.This indicates that a fraction of Ni would be present as another phase such as nickel niobate.This phase was not detected in the diffractogram that is likely due to its low loading or high dispersion.Further evidences of the presence of this phase will be provided by the TPR experiments.
Table 1 lists the crystallite sizes calculated using the Scherrer equation for supports and catalysts, after calcination at 1073 K.For the supports, the ceria crystallite size of CeZr is smaller than that obtained for CePr and CeNb, which agrees with the values of BET surface area.These results confirm that, in the presence of Zr, the sintering process of ceria was less pronounced.In the case of catalysts, the sample containing Zr also exhibited the smallest ceria crystallite size.Moreover, adding Ni to CeZr support did not significantly affect the ceria crystallite size.On the other hand, the addition of Ni to CePr and CeNb supports led to a slight increase of ceria crystallite sizes.Ni/CeNb showed the largest NiO crystallite size (30 nm).Augusto et al. 29 obtained the same NiO crystallite size for nickel/Gd-doped ceria calcined at 1073 K.
The XRD patterns obtained for the reduced and passivated samples (Figure 3) exhibited the lines associated with ceria with cubic structure, CeZr and CePr solid solutions and metallic Ni phase (ICSD 43397).The reduction treatment did not significantly change the ceria crystallite size for Ni/CeZr and Ni/CePr catalysts (Table 1).For Ni/CeNb, a decrease in the ceria crystallite size was observed.The Ni crystallite size of Ni/CeZr was smaller (24 nm) than that of Ni/CePr and Ni/CeNb (33 and 32 nm, respectively).Augusto et al. 29 also obtained a Ni crystallite size of 33 nm for nickel/Gd-doped ceria calcined and reduced at the same conditions used in this work.
The Raman spectra of the supports are shown in Figure 4.The Raman spectrum of bulk CeO 2 was also added for comparison.CeO 2 and ceria-doped materials exhibited a band at around 465 cm -1 , which is associated with F2g mode, due to the symmetric stretching of Ce-O vibrational unit of CeO 2 .CePr sample showed an additional broad band at 574 cm -1 that corresponds to defect induced mode (D bond), due to oxygen vacancy. 31The relative intensity ratio of I D /I F2g is related to the degree of defect sites on CeO 2 .Then, this result indicates that the CePr sample has a higher number of oxygen vacancies.
Figure 5 shows the TPR profile of the supports calcined at 1073 K.
The TPR profile of CeZr exhibited a small H 2 consumption between 600 and 900 K and a broad peak at around 1148 K. Two peaks at 806 and 856 K and a broad H 2 uptake above 900 K were observed K in the TPR profile of CePr.For CeNb, a small H 2 consumption was also detected at 973 and 1184 K.
According to the literature, 20,21,30 the TPR profile of CeO 2 generally exhibits a H 2 consumption at around 820 K and a large H 2 uptake at 1220 K.The H 2 consumption at low temperatures is attributed to the surface reduction of CeO 2 , whereas the H 2 uptake at high temperatures is related to the formation of Ce 2 O 3 .Adding dopants, such as Zr, to ceria structure shifts the H 2 consumption to lower temperatures, which suggests that the presence of dopants promotes the reduction of ceria.This was attributed to an increase in oxygen mobility induced by the insertion of dopant into the CeO 2 lattice.
In this work, for CeZr and CePr, the H 2 consumption at low temperatures (600-900 K) was attributed to the surface reduction of CeO 2 and the H 2 uptake at higher temperatures (> 900 K) was associated with the reduction of bulk ceria.For CeNb support, the H 2 consumption at low temperatures could be also attributed to the reduction of surface ceria.However, the reduction of niobium oxide or cerium niobate at high temperature cannot be ruled out.Some authors 32 observed the reduction of Nb 2 O 5 around 900-1050 K.
A comparison between the TPR profiles obtained for all supports showed that CePr exhibited the highest H 2 consumption at low temperatures.In addition, a higher   reduction degree was obtained for this sample (Table 2).These results indicated that CePr showed the highest reducibility and consequently the highest oxygen mobility, which suggests that the promoting effect of ceria reduction was more significant for the sample containing Pr.
The TPR results obtained for all catalysts were also presented in Figure 5.The TPR of a bulk NiO was also added for comparison.The TPR profile of NiO exhibited a reduction peak at 719 K and a shoulder at 770 K.The first peak was found to be much higher than the second one.In the case of Ni/CeZr and Ni/CePr catalysts, the TPR profiles were similar to the reduction profile of NiO, showing a peak at 726-728 K and a shoulder around 805-890 K.These results show the formation of large NiO particles on these catalysts, which are in agreement with the XRD results.Similar TPR profiles were obtained by Roh et al. 33 and Augusto et al. 29 for bulk NiO and Ni supported on ceria based oxides, indicating that there is no strong interaction between NiO and CeO 2 .This was attributed to the large crystallite sizes of both support and NiO phases formed at high calcination temperatures.
The reduction degree of NiO was larger than 100% for Ni/CeZr and Ni/CePr catalysts (Table 2), which shows that the H 2 uptake at low temperature region (600-900 K) observed for these catalysts cannot be attributed only to the reduction of NiO.The H 2 consumption in this low temperature region of the TPR profiles of Ni/CeZr and Ni/CePr is likely due to the reduction of bulk NiO as well as to the reduction of CeO 2 promoted by the metallic nickel particles formed.This promoting effect is due to the hydrogen spillover from the metal particles onto the support. 21In the case of Ni/CeNb catalysts, in addition to the reduction of bulk NiO (749 K), it was also observed a reduction peak at 1025 K.The reduction degree of NiO was lower than 100% (Table 2).This could be related to the presence of Ni species with lower reducibility, in addition to bulk NiO.Some authors 34 reported that Nb 2 O 5 phase can react with nickel oxide to form a mixed niobium and nickel oxide phases, such as NiNb 2 O 6 over Ni/Nb 2 O 5 catalysts.The nickel niobate phase is reduced at high temperature region (1025 K).These phases were not detected in the XRD analyses, which is likely due to its low loading or high dispersion.

Reactions
Methane and CO 2 conversions as a function of time on stream (TOS) for the DRM at 1073 K over Ni/CeZr, Ni/CePr and Ni/CeNb catalysts are shown in Figure 6.The values of initial methane and CO 2 conversions obtained for Ni/CeZr (CH 4 = 74% and CO 2 = 81%) were slightly higher than that observed for Ni/CePr and Ni/CeNb (CH 4 = 66-69% and CO 2 = 75-77%).In addition, the methane and CO 2 conversions slightly decreased during 24 h TOS for all catalysts.Regarding the selectivity to H 2 and CO, all catalysts exhibited similar H 2 /CO ratios (around 0.54-0.67).The values of H 2 /CO ratio lower than 1.0 is likely due to the reverse of the water-gas shift reaction.

Characterization of used catalysts
TG analyses were carried out to characterize postreaction catalysts.Figure 7 shows the TPO profiles obtained after DRM.The TPO profile of all catalysts exhibited a peak at around 899-945 K.In addition, Ni/CeNb showed a small CO 2 formation around 766 K.According to the literature, [35][36][37][38] CO 2 peaks in the low temperature region (< 673 K) were attributed to oxidation of amorphous carbon overlaying the metal surface.On the other hand, the CO 2 formed at high temperatures was assigned to filamentous (773-973 K) and graphitic carbon (> 973 K).In this work, the TPO profiles after dry DRM suggest the formation of filamentous carbon for all samples.
From the TPO profiles, the amount of carbon deposited over all catalysts after DRM was calculated.Ni/CePr exhibited the lowest carbon formation (0.42 mg carbon g cat -1 h -1 ).The amount of carbon formed over Ni/CeNb and Ni/CeZr was 6.47 and 9.68 mg carbon g cat -1 h -1 , respectively.Faria et al. 39 also detected large carbon formation over Ni/CeZrO 2 /Al 2 O 3 catalysts during DRM.Some authors [17][18][19] proposed a dual path mechanism for CO 2 reforming of methane.According to this mechanism, in the first step, methane dissociates on nickel surface, producing hydrogen and highly reactive carbon species (C a ). 40In the second step, this carbon reacts with oxygen from the support and the surface of the catalyst remains free of carbonaceous residues.This oxygen is formed during the dissociative adsorption of CO 2 and reoxidation of the support.When the rate of methane dissociation is faster than the rate of carbon oxidation, C a formed may undergo polymerization to less active carbon (C β ).
Then, this carbon may accumulate on the surface or may dissolve into the Ni crystallite, which is the first step for the nucleation and growth of carbon filaments.Therefore, the stability of the catalyst is determined by the proper balance between the rate of methane decomposition and the rate of carbon removal.As a result, in this mechanism, the support participates in the dissociative adsorption of CO 2 near the metal particles, transferring oxygen to the coked metal and promoting the removal of carbon from the metal. 41,42Then, supports with a high oxygen mobility such as ceria and ceria-mixed oxides, promote the mechanism of carbon removal.In this work, the lower carbon formation over Ni/CePr catalyst could be associated with the higher oxygen mobility of the CePr solid solution.The higher oxygen mobility of CePr support, as revealed by TPR and Raman, favors the mechanism of carbon removal, improving catalyst stability.

Conclusions
All anodes exhibited a good catalytic activity on dry reforming of methane.The values of initial methane and CO 2 conversions were slightly higher for Ni/CeZr likely due to the lower Ni crystallite size, as detected in DRX analyses.However, all samples presented a slight deactivation during the reaction.The TG analyses carried out after the reaction showed that this loss of activity was due to the formation of carbon that was lower for the sample containing Pr.This result was attributed to a higher reducibility and a higher amount of oxygen vacancies on Ni/CePr catalyst, as revealed by TPR and Raman analyses.A significant

a
Calculated using the (111) ceria plane; b calculated using the (200) NiO plane; c calculated using the (111) metallic Ni plane.

Figure 1 .
Figure 1.XRD patterns of supports calcined at 1073 K.The vertical lines corresponding to cerium oxide with cubic structure (ICSD 72155).

Figure 2 .
Figure 2. XRD patterns of catalysts calcined at 1073 K.The vertical lines corresponding to cerium oxide with cubic structure (ICSD 72155).

Figure 3 .
Figure 3. XRD patterns of catalysts obtained after reduction at 1023 K and passivation.

Figure 4 .
Figure 4. Raman spectra of the supports calcined at 1073 K.

Figure 5 .
Figure 5. TPR profiles of the supports and of the catalysts calcined at 1073 K.

Figure 7 .
Figure 7. TPO profiles obtained during TG analyses of all catalysts after DRM at 1073 K for 24 h.

Table 1 .
Crystallite sizes of CeO 2 , NiO and metallic Ni phases calculated by Scherrer equation

Table 2 .
Amount of H 2 consumed and reduction degree of the supports and NiO obtained during TPR analyses Reduction degree of CeO 2 , calculated using the total amount of H 2 consumed; b reduction degree of NiO, calculated using the amount of H 2 consumed at low temperature region.