Journal of Catalysis 306 (2013) 116–128

Contents lists available at SciVerse ScienceDirect

Journal of Catalysis

journal homepage: www.elsevier .com/locate / jcat

Mathematical modeling of Ni/GDC and Au–Ni/GDC SOFC anodesperformance under internal methane steam reforming conditions

0021-9517/$ - see front matter � 2013 Elsevier Inc. All rights reserved.http://dx.doi.org/10.1016/j.jcat.2013.06.015

⇑ Corresponding authors. Address: Department of Chemical Engineering, Univer-sity of Patras, Caratheodory 1 St., GR-26504 Patras, Greece (C.G. Vayenas).

E-mail addresses: souentie@chemeng.upatras.gr (S. Souentie), cgvayenas@upatras.gr (C.G. Vayenas).

S. Souentie a,⇑, M. Athanasiou a,b, D.K. Niakolas b, A. Katsaounis a, S.G. Neophytides b, C.G. Vayenas a,c,⇑a Department of Chemical Engineering, University of Patras, Caratheodory 1 St., GR-26504 Patras, Greeceb Foundation for Research and Technology Hellas, Institute of Chemical Engineering Sciences (FORTH/ICE-HT), GR-26504, Rion Achaias, Patras, Greecec Academy of Athens, Division of Natural Sciences, Panepistimiou 28 Ave., GR-10679 Athens, Greece

a r t i c l e i n f o

Article history:Received 11 April 2013Revised 16 June 2013Accepted 18 June 2013Available online 23 July 2013

Keywords:SOFCInternal methane steam reformingSolid oxide fuel cellsNi/GDC anodesSOFC modeling

a b s t r a c t

A simple kinetic model has been developed to describe the catalytic and electrocatalytic performance ofNi/GDC and Au–Ni/GDC anodes of SOFCs operating under internal methane-steam reforming reactionconditions, at low and high steam-to-carbon ratio values. The model accounts for the surface dissociationof CH4 to form methyl species which then react with H2O to form CO and H2. Under fuel cell operationconditions, two cases have been distinguished according to the observed electrochemical behavior; thehigh and the low steam-to-carbon ratio feed conditions. The former is characterized by electrochemicalconsumption of H2 and CO, produced by internal CH4 steam reforming, while the latter by electrochem-ical partial oxidation of CH4, to form H2 and CO, and oxidation of H2. Interestingly, the coverage ofmethyl-type species of the catalyst surface, as extracted from the model and the catalytic kinetic data,was found to coincide with the methyl species coverage at the three-phase boundaries, as extracted fromthe electrocatalytic experiments. The model is in good agreement with experiment under both open-cir-cuit and fuel cell operation conditions.

� 2013 Elsevier Inc. All rights reserved.

1. Introduction

Fuel cells can be potentially one of the most important energyconversion tools of the 21st century, as they convert the chemicalenergy of fuels into electricity and heat. This technology showsadvantages concerning fuel efficiency, emissions, maintenance,and noise pollution compared to conventional heat engines, steam,and gas turbines [1–3]. Among the different types of fuel cells, theSolid Oxide Fuel Cells (SOFCs) have attracted strong attentionmainly due to the higher outlet temperature (ease of waste heatutilization), the wide range of operating temperature (500–1000 �C) that enhances the reaction kinetics, the fuel flexibilityand tolerance to poisonous compounds, etc. Thus, SOFCs can havepractical applications in both stationary and mobile applications[1,4,5].

One of the key aspects for the efficient operation of SOFCs is thesuitable selection of fuel. Hydrogen is the most frequently usedfuel, which is mainly produced by steam reforming or direct oxida-tion of several hydrocarbons [6] on Ni-based catalysts at the tem-perature range of SOFCs operation. Nevertheless, SOFCs can alsooperate directly with natural gas or other hydrocarbon fuels feed

[3,7–17], without the need of an external reformer, which im-proves and simplifies the system integration. This type of opera-tion, known as Internal Reforming process, relies on the catalyticactivity of the fuel exposed anode.

In recent reviews [4,18–20], a significant number of anodematerials have been discussed. Still, Ni-based anodes are consid-ered as the most active ones for electrochemical oxidation andfor reforming of hydrocarbons. Nickel containing ceramic–metalcomposites (cermets) with YSZ, ScSZ [9,14] and/or rare earth oryttrium doped ceria as the predominant anode materials. Ni/YSZcermets fulfill most of the requirements for an efficient SOFC an-ode. Apart from their superior catalytic activity, they exhibit goodelectrical conductivity, high gas diffusivity, as well as excellentstructural integrity under operating conditions [21]. However,their disadvantages are the poor redox stability, the tendency ofNi for agglomeration after prolonged operation, and its lowtolerance to sulfur, which is a common impurity in natural gas.In addition, Ni catalyzes the formation of carbon deposits fromhydrocarbons under reducing conditions [22]. With respect tothe selection of alternative anode materials, introduction ofCeO2�d based additives and layers has been considered as one ofthe most promising approaches in anode development [19,23,24],and Gadolinium Doped Ceria [CeO2(Gd2O3) or GDC] is frequentlyused [25] as a substitute of YSZ. The advantages of ceria comparedto YSZ are associated with its high catalytic activity for combustionreactions, e.g., CO and hydrocarbon oxidation [18]. In addition, GDC

S. Souentie et al. / Journal of Catalysis 306 (2013) 116–128 117

is known to suppress coke formation, while its catalytic behavior insteam reforming of methane has been reported in several studies[26,27]. However, the electrocatalytic activity of GDC, withoutany additives, has been proved insufficient and the addition ofnickel resulted in Ni/GDC as effective anode for CH4 fueled SOFCs[16,28–30]. This has been mainly attributed to ceria’s ability totransform carbon deposits to CO or CO2, during or after CH4

decomposition, through the cermet’s mobile bulk lattice oxygencontent [29,31,32]. Moreover, Ni/YSZ anodes experience severedegradation in fuels containing only a few ppm of H2S, due tothe high vulnerability of Ni to sulfur poisoning. Studies with Ni/GDC [33–36] have shown that this composition might also be avery good candidate for operation in the presence of H2S, sincethe degradation in performance for the H2 oxidation, in H2S/H2

fuels, is substantially smaller compared to that on Ni/YSZ [36].The performance of Ni/GDC cermets as anodes in hydrocarbon

fueled SOFCs can be further improved by uniformly dispersingtrace amounts of noble-metal dopants such as Ru, Pd, Rh, or Au[12,37–39]. Along this direction, it was recently reported that mod-ification of NiO/GDC anode powder via deposition–precipitation offinely dispersed Au nanoparticles results in high carbon toleranceand improved catalytic and electrocatalytic activity under reducingconditions. Specifically, the Au-modification process allowed forthe examined cell, a carbon tolerant operation at T = 850 �C underfuel-rich internal steam reforming of methane, with a stable powerdensity of 0.41 W/cm2 at 810 mV for over 200 h. This performancehas been attributed to an interaction between the reduced Ni sur-face and the Au nanoparticles, which results in the formation of asolid solution and modifies the carbon tolerance of the Ni-basedanode [12,40,41]. Ni-alloys (or solid solutions) can help mitigatecarbon- and/or sulfur-induced deactivation, while maintainingthe inherent material structure and performance of the SOFC[21,41,42]. Furthermore, there are several experimental studiesof binary Ni-alloys and specifically on Ni/GDC anodes, which havedemonstrated the beneficial effects of the alloying element (e.g.,Mo, Fe, and Bi) on the fuel cell performance [21,41,43–45].

Methane is a very promising fuel for SOFC systems due to itsabundance in natural gas (80–95% CH4). Steam reforming of CH4

is the main industrially used process due to its high thermal effi-ciency, compared to other fuel processing technologies [17,46],while it practically relies on Ni-based catalysts due to cost andavailability concerns of noble metals [47]. However, it is a complexprocess, since it involves several reaction steps that take place inparallel or in series, where hydrogen, carbon monoxide, and carbondioxide are formed [46,48].

In the case where CH4 or natural gas is directly introduced atthe anode of a SOFC, the internal steam reforming (ISR) (Eq. (1))takes place over the catalyst to form syngas [3,49].

CH4 þH2O! COþ 3H2 ð1Þ

In addition to syngas formation at low temperatures, the water–gas shift reaction can take place in parallel (Eq. (2)) forming CO2

and H2:

COþH2O! CO2 þH2 ð2Þ

Hydrogen and CO can then be electrochemically oxidized at thesolid electrolyte/metal/gas three-phase boundaries (tpb) to formH2O and CO2.

H2 þ O2� ! H2Oþ 2e� ð3Þ

COþ O2� ! CO2 þ 2e� ð4Þ

Furthermore, CH4 can be partially oxidized electrochemically tosyngas [50,51] (Eq. (5)):

CH4 þ O2� ! COðgÞ þ 2H2ðgÞ þ 2e� ð5Þ

Until now, numerous studies have been published on the ki-netic behavior of the CH4 steam reforming reaction mainly overNi-supported catalysts [4,47,49,52–56], while many studies havefocused on the reaction kinetics over Ni/YSZ [46,49,57–59] anodes.However, there are only few studies on the CH4 steam reformingkinetic investigation over Ni/GDC SOFC anodes [60–62]. The major-ity of these studies can be classified, depending on the kineticexpression used, in three categories [4]: Those using general Lang-muir–Hinshelwood kinetics, those assuming first order reactionrate with respect to methane, and those utilizing power lawexpressions derived from data fitting.

A general remark is that the reported kinetic expressions differsignificantly. Despite considerable efforts, there is still consider-able debate on the main catalytic steps that determine the reactionrate, while the effective reaction order with respect to both CH4

and H2O is still a contradictory topic. This can be primarily attrib-uted to the various reaction conditions and to the different cata-lytic materials studied. Moreover, variations in catalytic activitycan also be ascribed to the limited reproducibility of catalyst prep-aration and activation methods, while it is plausible that the Nisurface area, dispersion, and crystallite size have a key role in theactivation of the reaction steps, which take place on the catalystsurface [49].

In this study, the performance of Ni/GDC and 3 wt.% Au–Ni/GDCcatalyst/electrodes has been studied under open-circuit conditionsand 1 wt.% Au–Ni/GDC catalyst/electrodes under fuel cell operationconditions at temperatures 800–900 �C. The latter catalyst/elec-trode has been examined under both fuel-lean and fuel-rich gasfeed conditions, where different electrocatalytic reactions were ob-served. A kinetic and electrocatalytic mathematical model hasbeen developed to describe the reaction performance in both cases.

2. Experimental

2.1. Preparation of Au–NiO/GDC anode cermet

The NiO/GDC cermet powder is commercially available fromPraxair and contained 65 wt.% NiO. A series of binary Au–NiO/GDC anode powders with nominal Au loading 1 and 3 wt.% wereprepared via the deposition–precipitation (D.P.) method usingHAuCl4 solution. In particular, the NiO/GDC commercial powderwas immersed into 3D-H2O, and the suspension was mildly stirred,while the temperature was adjusted to 70 �C. Appropriate volumeof hydrogen tetrachloroaurate (HAuCl4) (Sigma–Aldrich) wasadded to the suspension. During the deposition of gold, the pHvalue of the suspension was continuously adjusted in the region7.0–7.1 with the addition of proper amounts of aqueous NH3 1 Msolution, which is the precipitant agent. At this pH value, goldhas been reported [41,63] to deposit as [Au(OH)4]� complexes.After filtering, the precipitate was repeatedly washed to eliminatethe Cl� anions and dried at 110 �C for 24 h. All dried powders werecalcined at 1100 �C for 90 min. The specific surface areas of thepowders were determined from the corresponding N2 absorptionisotherms at 77 K in a QuantaChrome Autosorb-1 BET and chemi-sorption system. Specifically, SBET = 4 m2/g for NiO/GDC and 1 wt.%Au–NiO/GDC, while SBET = 3.3 m2/g for 3 wt.% Au–NiO/GDC.

2.2. Preparation of Solid Oxide Electrode Assemblies

The Solid Oxide Electrode Assemblies (SOEAs) comprisedcircular shaped planar electrolyte-supported SOFC membranesmanufactured by Kerafol with a diameter of 25 mm. The mechan-ical support consisted of approximately 300 lm thick 8YSZ electro-lyte. The anode was deposited by means of stepwise addition ofdroplets of slurry with a precision micropipette. The anode slurry

118 S. Souentie et al. / Journal of Catalysis 306 (2013) 116–128

contained an amount of NiO/GDC (or Au–NiO/GDC) anode powder,terpineol (Sigma–Aldrich) as the dispersant, PVB (polyvinylbutyral,Sigma–Aldrich) as binder and iso-propanol as solvent. The slurrywas homogenized and then was stepwise dropped on the electro-lyte. The volume of the droplet was adjusted via the precisionmicropipette depending on the desired final loading and the anodelayer thickness. After the deposition of the slurry, the anode assem-bly was dried at room temperature and at 90 �C and was finallysintered at 1250 �C with a heating and cooling ramp rate of 2 �C/min. In all the examined cells, the final anode fuel electrode con-sisted of an 80 lm anode functional layer (AFL). In the case ofthe catalytic–kinetic studies, the electrode had a thickness of�40–50 lm, while there was not any cathode side. In the case ofthe electrocatalytic experiments, the cathode side consisted of por-ous LSM, which was commercially available from Fuel Cell Materi-als (FCM) and deposited by means of screen printing. The aboveAFL thicknesses are similar to the commonly used anodes in sev-eral SOFC studies. Indicatively, there is a range between 17 lm[37] and 100 lm [1] that has been used by several research groups.Moreover, preliminary activity tests with various AFL thicknessesshowed that anodes thinner than 50 lm exhibited low catalyticperformance, while limiting current effects were observed in thecase of anodes thicker than 80 lm.

Fig. 1 shows Scanning Electron Micrographs and back-scatteredelectron (BSE) compositional images of a 3 wt.% Au–NiO/GDC anodelayer. The images were collected in a HR-SEM (Zeiss SUPRA 35VP)equipped with a Centaurus BS-detector (K.E. Developments). Themicrograph of the top side (Fig. 1a) shows a porous electrode withthe size of particles in the order of 1 lm, while (1b) BSE image de-picts the top side of a reduced cell. The cross-section micrograph(Fig. 1c) shows that the thickness of the electrode is in the order

YSZ electrolyte

YSZ electrolyte

(a)

(a)

(c)

Fig. 1. Scanning Electron Micrographs of (a) the top side of a 3 wt.% Au–NiO/GDC anodecross-section perpendicular to the anode/YSZ electrolyte, and (c) SEM–BSE composition

of 80 lm. The thin layer of approximately 15 lm that is observedto be in first contact with the electrolyte is an adhesion layer ofthe same material with the electrode, which was deposited firstfor achieving a good adhesion and connectivity between the elec-trolyte and the electrode. Finally, in the SEM–BSE compositionalimage (Fig. 1d), the brighter spots correspond to the particles ofgold with higher mean atomic number, and it is observed that Auis homogeneously dispersed on the porous NiO/GDC electrode. Onthe other hand, Au is no longer detectable in (1b) BSE image, andthis is attributed to the dissolution of gold into Ni particles, duringreduction, and formation of surface Ni–Au solid solution [41].

2.3. Catalytic and electrocatalytic measurements

The open-circuit kinetic results shown in Figs. 2 and 3 were ob-tained with a Ni/GDC and a 3 wt.% Au–Ni/GDC anode, respectively.The anodes had a geometric surface area of �1 cm2 and masses of�6 and �4 mg, respectively. The electrocatalytic experiments werecarried out on a 1 wt.% Au–Ni/GDC catalyst electrode of �30.8 mgmass and �1.7 cm2 surface area. A Pt-mesh was used as currentcollector on the cathode side, while a Ni-mesh on the anode side.The effect of Ni-mesh on the catalytic rate was investigated at850 �C, and it was found to be insignificant.

The planar fuel cell system was attached on a YSZ tube andsealed airtight using a gold ring. The anode was exposed tohydrogen or methane/water mixtures, while the cathode was fedwith pure oxygen. Within the gas flow rate range used in thekinetic experiments, i.e., between 200 and 260 cm3 STP/min(equivalently SV between 20 � 105 and 39 � 105 cm3 h�1 g�1 or1.4 � 10�4 mol s�1 and 1.8 � 10�4 mol s�1), the reactor was operat-ing under differential conditions with reactant conversion being

Au particles

(b)

(d)

layer, (b) SEM–BSE compositional image of the top side in its reduced form, (c) SEMal image of the top side of the anode.

Fig. 2a. Steady-state effect of CH4 molar fraction ðyCH4Þ on the H2 formation rate

under open-circuit conditions and fixed H2O molar fraction ðyH2 O ¼ 0:033Þ. Ni/GDC,differential conditions.

Fig. 2b. Steady-state effect of H2O molar fraction ðyH2 OÞ on the H2 formation rateunder open-circuit conditions and fixed CH4 molar fraction ðyCH4

¼ 0:03Þ. Ni/GDC,differential conditions.

Fig. 3a. Steady-state effect of CH4 molar fraction ðyCH4Þ on the H2 formation rate

under open-circuit conditions and fixed H2O molar fraction ðyH2 O ¼ 0:038Þ. 3 wt.%Au–Ni/GDC, differential conditions.

S. Souentie et al. / Journal of Catalysis 306 (2013) 116–128 119

between 4% and 15% and was found to exhibit CSTR (ContinuousStirred Tank Reactor) behavior. The absence of external mass-transfer limitations and reaction thermodynamic equilibrium con-straints was also checked. In particular, for the cell tested underfuel cell conditions, the reaction rate of H2, CO, and CO2 productionwas found to remain practically constant for total gas flow ratesvarying from 100 to 260 cm3 STP/min (equivalently SV from1.9 � 105 to 5.1 � 105 cm3 h�1 g�1 or from 0.7 � 10�4 to1.8 � 10�4 mol s�1). The presented experiments were conductedin this range of gas flow rates.

The catalytic–kinetic measurements were carried out at differ-ent temperatures and different methane and water partial pres-sures. The electrocatalytic experiments were performed at 850 �C

under two different steam-to-carbon (S/C) ratios; the high S/C = 1and the low S/C = 0.25. In particular, He carrier gas was saturatedwith water using a thermostated saturator and was then mixedwith the CH4/He mixture. The reactor inlet steam concentrationwas controlled by varying both the saturator temperature andthe He flow. Pure CH4 (99.995 vol.%, L0 Air Liquide) and ultrapureHe (99.999 vol.%, L0 Air Liquide) were used. All lines and valveswere heated at 150 �C to prevent water condensation. Reactantsand products were analyzed by online gas chromatography usinga Varian CP-3800 gas chromatograph with a thermal conductivitydetector. A Porapak Q column (80–100 mesh, 1.8 m � 1/8 in. � 2 mm) was used for the analysis of CH4, CO, CO2, and H2Oat 120 �C, while a Carbosieve S-11 column (80–100 mesh,2 m � 1/8 in. � 2 mm) was used for the analysis of H2 and CO (inparallel with the Porapak Q). The electrochemical measurementswere carried out by using an AUTOLAB potentiostat/galvanonstatmodel 84693.

The apparent activation energies for the catalytic CH4 dissocia-tion reaction were determined experimentally via isothermal TGAexperiments using a TA Q50 instrument. The anode cermets werereduced in situ for 2 h at 800 �C under 80 vol.% H2/Ar. The isother-mal measurements were carried out under dry 11 vol.% CH4/Ar gasfeed and flow rate of 100 cm3 min�1. Carbon deposition kineticconstant values were calculated by determining the initial slopesof the deposited carbon weight change as a function of time, fromthe isothermal TGA measurements.

3. Results

3.1. Catalytic performance under open-circuit conditions

3.1.1. Ni/GDC and 3 wt.% Au–Ni/GDC catalystsThe effect of CH4 and H2O molar fractions on the catalytic rate

of H2 formation is shown in Figs. 2a and 2b respectively, using aNi/GDC catalyst electrode at 800, 850, and 900 �C. One observesthat the formation rate of H2 exhibits a positive order dependenceon yCH4

, and an initially positive and then slightly negative to zeroorder dependence on yH2O, at each temperature. Similar is the ki-netic behavior with respect to CH4 and H2O for the 3 wt.% Au–Ni/

Fig. 3b. Steady-state effect of H2O molar fraction of ðyH2 OÞ on the H2 formation rateunder open-circuit conditions and fixed CH4 molar fraction ðyCH4

¼ 0:04Þ. 3 wt.%Au–Ni/GDC, differential conditions.

Fig. 4. Effect of current density, i, on the potential, V, and on the reaction rates of H2,CH4, CO and CO2. yH2 O=yCH4

¼ 1, T = 850 �C, 1 wt.% Au–Ni/GDC anode, T = 850 �C.

120 S. Souentie et al. / Journal of Catalysis 306 (2013) 116–128

GDC anode, as shown in Figs. 3a and 3b, where the very shallowrate maximum with respect to yH2O is a little more pronounced,meaning that H2O adsorption reaches equilibrium after a certainpartial pressure. This kinetic behavior of Ni/GDC cermets suggeststhat dissociative adsorption of H2O takes place on different sitesthan that of CH4 and ceria possibly offers these sites. Moreover,it can be attributed to the formation of a monolayer of adsorbedH2O that hinders further dissociative adsorption of this molecule.In this respect, in order to simplify the model, we did not considerthe equilibrium adsorption process as a separate reaction step.

3.1.2. Electrocatalytic performance under fuel cell operation conditionsAs already noted, in these experiments, the 1 wt.% Au–Ni/GDC

catalyst electrode served as the SOFC anode, deposited over aYSZ solid electrolyte disk, while commercial LSM was used forthe cathode, as described in detail in the Experimental section.In these experiments, two cases have been distinguished accord-ing to the observed electrochemical behavior; the high and thelow steam-to-carbon (S/C) ratio cases. As described below indetail, the former is characterized by electrochemical consump-tion of H2 and CO, produced by the CH4-steam reformingcatalytic reaction, while the latter by the partial electrochemicaloxidation of CH4 leading to formation of H2 and CO. All electro-chemical experiments presented in this study were performed at850 �C, while in both cases, the ionic current density is in thesame order of magnitude with the oxygen supplied by H2O inthe gas phase.

3.1.2.1. High steam-to-carbon ratio, S/C = 1. The effect of currentdensity on cell potential (V) and on the reaction rates of CH4

consumption and of H2, CO, and CO2 formation is shown inFig. 4. Current densities as high as �230 mA/cm2 are obtainedwithout any evidence for diffusional limitations, since the cell po-tential decreases almost linearly with current (Fig. 4).

Under open-circuit conditions (i.e. for i = 0), the H2 formationcatalytic rate was �4 lmol/(cm2 s), while the formation rates ofCO and CO2 were 1.15 and 0.2 lmol/(cm2 s), respectively, and theconsumption rate of CH4 was 1.35 lmol/(cm2 s). These valuessatisfy the mass balance of carbon as this is expressed by:

rH2 ¼ 3rCO þ 4rCO2 ð6Þ

Blank experiments with Ni-mesh did not show any detectablecatalytic reaction rates.

Under fuel cell operation conditions, an increase in current den-sity results in an increase in the CO and H2 consumption rates andin an increase in the CO2 formation rate, while CH4 consumptionrate decreases slightly. This behavior suggests that the electro-chemical oxidation of H2 (Eq. (3)) and CO (Eq. (4)), formed catalyt-ically via Eq. (1), is favored under high S/C ratios, rather than theelectrochemical complete or partial oxidation of CH4.

The current-induced changes in the rates of formation or con-sumption of the species, j, involved in the present system (CH4,CO, CO2, H2, H2O), can be expressed, in mol O/(cm2 s) or mol H2/(cm2 s), as:

Drj ¼ rfcj � roc

j ð7Þ

where rfcj is the rate under fuel cell operating conditions and roc

j isthe measured rate under open-circuit conditions. The correspond-ing apparent faradaic efficiencies, Kj, can then be defined from:

Kj ¼Drj

i=ð2FÞ ð8Þ

where i is the current density and F is the Faraday constant. In gen-eral, the value of Kj is limited to or below unity in the case of apurely electrochemical process.

The effect of ionic current density, i/(2F), on the electrochemicalrate of H2 and CO oxidation and CO2 formation is shown in Fig. 5.One observes that the rates of H2 and CO consumption and the rateof CO2 formation increase with i/(2F), while at the same time, therate of CH4 consumption is weakly suppressed. It is worth notingthat the faradaic efficiency value of H2 consumption, KH2 , estimatedby the corresponding slope in Fig. 5, is larger than unity, while theone of CO, KCO is larger than that for the formation of CO2, KCO2 .

Fig. 5. Effect of ionic current density, i/(2F), on the electrochemical reaction rates ofCO, CH4 and H2 consumption and CO2 formation. yH2 O=yCH4

¼ 1, T = 850 �C, 1 wt.%Au–Ni/GDC.

Fig. 6. Effect of ionic current density, i/(2F), on the total electrochemical rate of COand H2 consumption. yH2 O=yCH4

¼ 1, T = 850 �C, 1 wt.% Au–Ni/GDC.

Fig. 7. Effect of current density, i, on the potential, V, and on the reaction rates of H2,CH4, CO and CO2. S/C = 0.25, T = 850 �C, 1 wt.% Au–Ni/GDC.

S. Souentie et al. / Journal of Catalysis 306 (2013) 116–128 121

The apparent non-faradaicity of the process is also shown inFig. 6, where the total electrochemical, or current-induced, rate of

the two reactions, DrH2 þ DrCO, is plotted as a function of i/(2F).One observes that DrH2 þ DrCO is almost a factor of two larger thani/(2F), and thus, the faradaic efficiency is higher than 100%. This sug-gests that the current has also an effect on the rate of some catalyticprocess involved in H2 and/or CO production or consumption.

3.1.2.2. Low steam-to-carbon ratio, S/C = 0.25. Fig. 7 shows the effectof current density on potential and on the reaction rates of H2, COand CO2 production and on the rate of CH4 consumption again at850 �C. One observes that under open-circuit conditions, i.e., fori = 0, H2 formation catalytic rate is a factor of three larger than theformation rate of CO, in agreement with the reaction stoichiometry(Eq. (1)). The effect of Ni-mesh on the catalytic rate was found to beminor. The conversion of methane was approximately 3%, and theformation rates of H2, CO, and CO2 were 0.47, 0.14, and 0 lmol/srespectively; thus, 20 times lower compared to those during theelectrocatalytic measurement. Moreover, no CO2 formation is ob-served. In this case also, the mass balance of carbon was satisfied.

Under fuel cell operation conditions, the electrochemicalformation rate of CO and H2 increases linearly with current,accompanied by a linear increase in the CH4 consumption rate,in contrast to the high S/C ratio case where increasing currentsuppresses slightly the consumption of CH4 (Figs. 4 and 5). It isworth noting that no CO2 formation rate is observed even underfuel cell operation conditions. Thus, in this case, it appears thatthe main anodic reaction is the partial oxidation of CH4 (Eq. (9))to form CO and H2, i.e.:

CH4 þ O2� ! COðgÞ þ 2H2ðgÞ þ 2e� ð9Þ

Fig. 8 shows the effect of ionic current density, i/(2F) on thecurrent-induced change in the rate of CH4 consumption and COand H2 formation. One observes that the electrochemical rate of

Fig. 8. Effect of ionic current density, i/(2F), on the electrochemical rates of CH4

consumption and CO and H2 formation. S/C = 0.25, T = 850 �C, 1 wt.% Au–Ni/GDC.

Fig. 9. DrH2 =DrCO and DrCO=ð2DrCH4 � DrH2 Þ ratios as a function of yCH4=yH2 O.

T = 850 �C, 1 wt.% Au–Ni/GDC.

122 S. Souentie et al. / Journal of Catalysis 306 (2013) 116–128

CO formation is in good agreement with Faraday’s law (KCO = 0.9).However, the electrochemical rate of H2 formation is smaller thanthat expected according to the reaction stoichiometry (Eq. (9)). Thelatter possibly indicates that H2 electrochemical oxidation takesplace in parallel with reaction (9).

Fig. 9 shows that, within experimental error, there is a linearcorrelation between DrCO=DrH2 and also DrCO=ð2DrCH4 � DrH2 Þ andthe methane to steam ratio, yCH4

=yH2O, in the examined gas compo-sition range. As shown in the figure, the resulting slope is �0.25.

Table 1aModel-predicted values of the kinetic constants for the dissociation of CH4 and forthe surface reaction steps, as calculated by Eqs. (15)–(17) at 800, 850, and 900 �C.Ni/GDC.

4. Discussion – model development

4.1. Catalytic performance under open-circuit conditions

The kinetic behavior shown in Figs. 2a–3b suggests that theconversion of CH4 to CO and H2 proceeds predominantly via atwo-step mechanism:

CH4ðgÞ þ S!ka CH�2 þH2ðgÞ ð10Þ

T (�C) 800 850 900

ka/10�6 For yCH4! 0 25 38 64

For yH2 O !1 25 43 67Average value 25 40 65

kr/10�6 For yH2! 0 100 180 280

k = ka/kr 0.25 0.22 0.23

CH�2 þH2OðgÞ!kr COðgÞ þH2ðgÞ þ S ð11Þ

where S denotes a vacant adsorption site and ka and kr are, respec-tively, the kinetic constants for the adsorptive dissociation of CH4

(Eq. (10)), leading to the formation of carbonaceous deposits, suchas adsorbed methyl-species (CH�2) and for the removal of these

species from the catalyst surface via reaction with H2O, probablythrough the intermediate formation of low coverage adsorbed hy-droxyl species.

One can use mass action kinetics to describe the rates of reac-tions (10) and (11) which, at steady state, have to be equal:

�rCH4 ¼ kayCH4ð1� hCÞ ð12Þ

�rCH4 ¼ kryH2OhC ð13Þ

where yCH4and yH2

are the gas-phase mol fractions of CH4 and H2O,respectively, and hC denotes the coverage of the carbonaceous spe-cies on the catalyst surface.

From Eqs. (12) and (13), one obtains:

hC ¼kðyCH4

=yH2OÞ1þ kðyCH4

=yH2OÞ; 1� hC ¼

11þ kðyCH4

=yH2OÞð14Þ

where k = ka/kr. This in view of Eqs. (12) and (13) leads to:

�rCH4 ¼kayCH4

1þ kðyCH4=yH2OÞ

ð15Þ

For high steam to CH4 ratios, Eq. (15) reduces to:

�rCH4 ¼ kayCH4ð16Þ

while for low S/C ratios, it reduces to:

�rCH4 ¼ kryH2O ð17Þ

Indeed, as shown in Figs. 2a–3b, the rate of H2 productionðrH2 ¼ �3rCH4 Þ is first order in CH4 and zero order in H2O at highS/C ratios and shifts to first order in H2O and zero order in CH4

for low S/C ratios.

Table 1bModel-predicted values of the kinetic constants for the dissociation of CH4 and for thesurface reaction steps, as calculated by Eqs. (15)–(17) at 800, 850, and 900 �C. 3 wt.%Au–Ni/GDC.

T (�C) 800 850 900

ka/10�6 For yCH4! 0 15 27 45

For yH2O !1 15 28 45Average value 15 28 45

kr/10�6 For yH2! 0 70 120 200

k = ka/kr 0.22 0.23 0.23

Fig. 10b. Experimental and model-predicted steady-state effect of H2O molarfraction ðyH2 OÞ on the H2 formation rate under open-circuit conditions and fixed CH4

molar fraction ðyCH4¼ 0:03Þ. Ni/GDC, differential conditions.

S. Souentie et al. / Journal of Catalysis 306 (2013) 116–128 123

Utilizing the experimental data in Figs. 2a–3b and applying Eqs.(16) and (17), one can calculate ka and kr for each temperature. Theresulting values of ka and kr are summarized in Tables 1a and 1b forthe Ni/GDC and for the Au–Ni/GDC anodes, respectively, as derivedby each analysis. An average value of ka is also given which hasbeen used in the rate data fitting shown in Figs. 10a–11b. As shownin these figures, the model is in good agreement with experimentand the fit is better at high S/C ratio values. Increasing temperaturecauses both rate constants to increase, while the resulting values ofk, as defined in Eq. (14), are also given in Tables 1a and 1b. One ob-serves that k is close to 0.23 and is almost independent of temper-ature and Au content.

Table 2 presents the activation energies of the CH4 dissociationand the surface reaction steps over the Ni/GDC and the Au–Ni/GDCanodes, and these values are close to those reported in the litera-ture [4,49,61]. The activation energy for the CH4 dissociation isslightly higher (by �5 kcal/mol) over the Au-modified catalyst.The activation energy for the surface reaction was found to bealmost the same for both catalysts, i.e., �26 kcal/mol. In supportof these results, thermogravimetric analysis (TGA) experiments,under 11% CH4/Ar feed, were performed for the estimation of thecarbon deposition activation energy on the Ni/GDC and Au–Ni/GDC catalysts and is shown in Fig. 12. In the inset of Fig. 12, onecan see the good agreement between the activation energies foundby TGA analysis and those calculated by the model (shown inTable 2).

Fig. 13 shows the effect of S/C ratio on the model-predictedmethyl-species coverage of the surface, hC, by Eq. (14) at 800,

Fig. 10a. Experimental and model-predicted steady-state effect of CH4 molarfraction ðyCH4

Þ on the H2 formation rate under open-circuit conditions and fixed H2Omolar fraction ðyH2 O ¼ 0:033Þ. Ni/GDC, differential conditions.

Fig. 11a. Experimental and model-predicted steady-state effect of CH4 molarfraction ðyCH4

Þ on the H2 formation rate under open-circuit conditions and fixed H2Omolar fraction ðyH2 O ¼ 0:038Þ. 3 wt.% Au–Ni/GDC, differential conditions.

850, and 900 �C. One observes that hC decreases with S/C ratio,while it increases with temperature, indicating an activatedadsorption process. It is worth noting that in the case ofT = 850 �C, hC decreases from 0.45 to 0.2 as the S/C ratio increasesfrom 0.25 to 1.

4.2. Electrocatalytic performance under fuel cell conditions

While the catalytic performance can be described by a singlemodel, both for low and for high S/C ratios, as already shown inFigs. 4–8, the electrocatalytic behavior changes significantly be-tween high and low S/C ratios and thus requires a separatediscussion.

Table 2Activation energy (Eact) of the CH4 dissociation and the surface reaction steps for theNi/GDC and the 3 wt.% Au–Ni/GDC anodes, as calculated by the model.

Sample CH4 dissociation(kcal/mol)

Surface reaction(kcal/mol)

Ni/GDC �23 �263 wt.% Au–Ni/GDC �28 �26

Fig. 13. Effect of steam-to-carbon ratio on the model-predicted methyl-typespecies coverage of the surface, hC, by Eq. (17) at 800, 850, and 900 �C for the3 wt.% Au–Ni/GDC anode.

Fig. 11b. Experimental and model-predicted steady-state effect of H2O molarfraction of ðyH2 OÞ on the H2 formation rate under open-circuit conditions and fixedCH4 molar fraction ðyCH4

¼ 0:04Þ. 3 wt.% Au–Ni/GDC, differential conditions.

124 S. Souentie et al. / Journal of Catalysis 306 (2013) 116–128

4.2.1. High steam-to-carbon ratio, S/C = 1As already shown in Fig. 4, under high S/C ratios, increasing

current causes a decrease in CO and H2 production rates and an in-crease in CO2 formation rate. Moreover, CH4 consumption slightly

Fig. 12. Arrhenius plots of the CH4 dissociation reaction as calculated by TGAanalysis using 0, 5.3 and 10.6 wt.% Au–Ni/GDC anodes. Inset: Effect of Au content onthe activation energy of CH4 dissociation. Comparison between TGA analysis resultswith model-predicted values.

decreases. This behavior suggests that under high S/C ratios, theelectrochemical oxidation of H2 (Eq. (18)) and CO (Eq. (19)), formedcatalytically via Eq. (1), is favored, i.e.,

H2 þ O2� ! H2Oþ 2e� ð18Þ

COþ O2� ! CO2 þ 2e� ð19Þ

As shown in Figs. 4, 5 and 9, the ratio DrCO=DrH2 remains prac-tically constant and equal to 0.3 (=0.4/1.35) as the anodic current isvaried. This implies that the ratio DrCO=ðDrH2 þ DrCOÞ is also con-stant and equal to 0.23. Surprisingly, the latter practically coincideswith the value of hC (�0.25) at S/C = 1 (Fig. 14). This is shown inFigs. 14a and 14b and strongly suggests the physical model pre-sented in Fig. 15: Hydrogen is oxidized anodically at carbon-freeadsorption sites, while CO is oxidized on sites covered by a carbo-naceous adsorbate. This covered by carbonaceous species active

Fig. 14a. Effect of ionic current density, i/(2F), on the model-predicted methylspecies coverage of the catalyst tpb, hC,tpb, calculated by Eq. (20). S/C = 1, T = 850 �C,1 wt.% Au–Ni/GDC.

Fig. 14b. Comparison between the model-predicted methyl-type species coverageof the catalyst surface, hC, (Eq. (14)) and methyl-type species coverage of thecatalyst tpb, hC,tpb, (Eq. (20)). S/C = 1, T = 850 �C, 1 wt.% Au–Ni/GDC.

Fig. 15. Schematic of the physical model developed for high steam-to-carbon ratiocases. It suggests H2 electrochemical oxidation at carbon-free adsorption sites andCO electrochemical oxidation on sites covered by a carbonaceous adsorbate.

Fig. 16. Effect of ionic current density, i/(2F), on the apparent faradaic efficiency ofCO and H2 consumption and CO2 formation. S/C = 1, T = 850 �C, 1 wt.% Au–Ni/GDC.

S. Souentie et al. / Journal of Catalysis 306 (2013) 116–128 125

site can be viewed as a macroscopic area near the tpb, which in-cludes the metal, the support, and the adsorbed carbonaceousspecies.

Denoting by hC,tpb the coverage of the carbonaceous adsorbateat the three-phase boundaries (tpb), it follows in view of Figs. 9and 14b that:

hC;tpb ¼ hC ¼DrCO

DrCO þ DrH2

ð20Þ

Upon combining Eq. (20) with Eq. (14), one obtains:

DrCO

DrH2

¼ kðyCH4=yH2OÞ ð21Þ

which is in good agreement with experiment as shown in Fig. 9. Notonly the straight line defined by the two points, corresponding tohigh and low S/C ratio, passes from the (0,0) point, but also theslope is 0.23 which practically coincides with the value of k at thetemperature of the experiments (Table 1).

In Fig. 6, it has been shown that the observed total electrochem-ically induced rate increase (DrH2 þ DrCO) is typically a factor of 2larger than i/(2F), and thus, the apparent faradaic efficiency ofthe process is typically 200% as also shown in Fig. 16, where Kvalues as high as 2.5 are obtained for H2 consumption. This is mostlikely due to a parallel catalytic process for H2 and CO productionor consumption, which is affected by the O2� flux. This catalyticprocess appears to be the steam reforming reaction (Eq. (1)), viawhich CO and H2 are formed, which is suppressed by O2� flux,and consequently, less CO and H2 is formed. The latter isfurther manifested by the observed decrease in the rate of CH4

consumption shown in Figs. 4 and 5. The fact that oxygen is not di-rectly involved in the steam reforming reaction, i.e.,

CH4 þH2O! COþ 3H2 ð22Þ

suggests that the observed reversible non-faradaic inhibiting effectof O2� supply on the rate of CH4 consumption (Eq. (22)) is a mani-festation of the effect of electrochemical promotion of catalysis(EPOC) [64–68], as discussed in the next section. However, thegradual formation of NiO at the proximity of the tpb reported byNakagawa et al. [61], who ascribed the deterioration of the catalyticactivity to the re-oxidation of Ni to NiO from the produced H2O bythe electrochemical oxidation of H2 may also have a role.

The electrocatalytic behavior in the case where only CO and H2

are fed in the cell is shown in Fig. 17. CO and H2 partial pressuresnear to those at open-circuit state under CH4 steam reforming reac-tion conditions were used at 850 �C. As shown in the figure, H2 ismainly (92%) electrochemically oxidized rather than CO. This is inagreement with the proposed model, where sites covered by a car-bonaceous adsorbate are required for selective CO electrochemicaloxidation. However, one should also account for the contribution ofthe water gas shift (WGS), utilizing the H2O formed by the electro-chemical oxidation of H2. To compare the observed rate with theone of the WGS, an experiment feeding only CO and H2O, at partialpressures (0.025 kPa each) that correspond to those obtained atcurrent density 0.7 lmolO/cm2 s, was performed under open-circuit conditions. The observed WGS catalytic rate (shown inFig. 17) was found to be near the observed CO2 formation rate dur-ing CO and H2 feed under fuel cell operating conditions.

4.2.2. Low steam-to-carbon ratio, S/C = 0.25Under low S/C ratio values, already shown in Figs. 7 and 8, an

increase in the CO, H2, and CH4 consumption rates is observed withincreasing ionic current. In this case, no CO2 formation is observed;neither under open-circuit nor under O2� flux conditions. Thisbehavior possibly indicates that the electrochemical partial oxida-tion of CH4 (Eq. (23)) is favored under low S/C ratio values, i.e.,

CH4ðgÞ þ O2� ! COðgÞ þ 2H2ðgÞ þ 2e� ð23Þ

Moreover, the observed electrochemical formation rate of H2 issmaller than expected by reaction stoichiometry (Eq. (23)), possi-bly due to its further electrochemical oxidation (Eq. (24)), i.e.,

H2ðgÞ þ O2� ! H2OðgÞ þ 2e� ð24Þ

Fig. 18b. Comparison between the model-predicted methyl-type species coverageof the catalyst surface, hC, (Eq. (14)) and methyl-type species coverage of thecatalyst tpb, hC,tpb, (Eq. (26)). S/C = 0.25, T = 850 �C, 1 wt.% Au–Ni/GDC.

Fig. 19. Schematic of the physical model developed for low steam-to-carbon ratiocases. It suggests H2 electrochemical oxidation at carbon-free adsorption sites andCH4 electrochemical oxidation on sites covered by a carbonaceous adsorbate.

Fig. 17. Effect of ionic current density, i/(2F), on the potential, V, and on theelectrochemical rates of CO and H2 consumption and CO2 formation under CO andH2 feed. The water gas shift reaction catalytic rate is also shown in case of 0.025 kPaCO and 0.025 kPa H2O (dashed line). T = 850 �C, 1 wt.% Au–Ni/GDC.

126 S. Souentie et al. / Journal of Catalysis 306 (2013) 116–128

As shown in Figs. 7–9, the ratio DrCO=ð2jDrCH4 j � DrH2 Þ remainspractically constant and equal to 0.9 (=0.9/(2�1.2–1.4)) as i/(2F) isvaried. This implies that the ratio DrCO=ð2jDrCH4 j � DrH2 þ DrCOÞ isalso constant and equal to 0.47, which practically coincides withthe value of hC (�0.45) at S/C = 0.25, as shown in Fig. 13. This isshown in Figs. 18a and 18b and suggests the physical model pre-sented in Fig. 19, which is similar to that in Fig. 15 for high S/Cratios: H2 is anodically oxidized at carbon-free adsorption sites,while CH4 is partially oxidized on sites covered by a carbonaceous

Fig. 18a. Effect of ionic current density, i/(2F), on the model-predicted methylspecies coverage of the catalyst tpb, hC,tpb, calculated by Eq. (26). S/C = 0.25,T = 850 �C, 1 wt.% Au–Ni/GDC.

adsorbate, instead of CO under high S/C ratio (Fig. 15). In this case,for the carbonaceous adsorbate coverage at the tpb, hC,tpb, in viewof Figs. 9 and 18b, one can write:

hC;tpb

1� hC;tpb¼ DrCH4 ðeq: 23Þ

DrH2 ðeq:24Þ ¼DrCO

2jDrCH4 j � DrH2

ð25Þ

where DrCO, DrCH4 , and DrH2 are the observed changes in total CO,CH4, and H2 electrochemical rates.

In view of Eq. (25), one can write for hC,tpb and hC:

hC;tpb ¼ hC ¼DrCO

2jDrCH4 j � DrH2 þ DrCOð26Þ

Upon combining Eq. (25) with Eq. (13), one obtains:

DrCO

2jDrCH4 j � DrH2

¼ kðyCH4=yH2OÞ ð27Þ

which, similarly to the high S/C ratio case (Eq. (20)), is in goodagreement with experiment as shown in Fig. 9, where the slopepractically coincides with the experimentally estimated value of kat the temperature of the experiments (�0.23, Table 1).

The direct partial electrochemical oxidation of CH4 at low S/Cratios is the essential difference of Au–Ni/GDC anode from the cor-responding performance of Au–Ni/YSZ under similar conditions. Asit was previously shown by Gavrielatos et al. [3] in the case ofAu–Ni/YSZ cermet, the electrochemical oxidation in all S/C ratiosranging from S/C = 1:1 to 1:3 involves only the electrochemicaloxidation of the catalytically produced H2 and CO for the produc-tion of H2O and CO2 (Eqs. (18) and (19)). On the contrary in the caseof Au–Ni/GDC, CO electrochemical oxidation appears with almost

S. Souentie et al. / Journal of Catalysis 306 (2013) 116–128 127

zero selectivity as this is revealed under low S/C = 0.25 depicted inFig. 7. This possibly shows that CO is not only directly electrochem-ically oxidized to CO2 as this is assumed by reaction (19), but alsoW.G.S. partly contributes. In addition and according to the pro-posed model validation, regarding the coverage of thecarbonaceous species hC (Figs. 9, 14b and 18b), both CO and CO2

formation must originate from the same carbonaceous speciesthrough two parallel electrochemical processes involving theoxidation of the adsorbed CH�2 species. As has been previously pro-posed by Triantafyllopoulos and Neophytides [69], CO formation isdriven through the oxygenation and decomposition of adsorbedmethyl-species, while CO2 is formed by the complete oxidationof the highly reactive carbidic carbon originating from the com-plete dehydrogenation of CH2

⁄ species. It can be considered thatthe complete dehydrogenation of the adsorbed CH2

⁄ species canbe facilitated either by stronger interactions of the methyl dagglingH with the Ni surface or their enhanced removal by the high oxidecoverage originating either by the high partial pressure of steam orthe high O2� flux supplied electrochemically. These processes canbe described by the reaction schemes of:

CH�2 þ O2� ! CH2Oþ 2e� ð28aÞ

CH2O!NiCOþH2 ð28bÞ

CH�2 þ Oad!Ni

Cc þH2O ð29aÞ

Cc þ 2Oad!Ni

CO2 ð29bÞ

The effect of Au doping was shown to be vital in the case ofNi/YSZ anodes, which have no tolerance on carbon formation.However, in the case of Ni/GDC within the framework of thepresent experiment and under these reaction conditions, nodegradation phenomena were observed. In light of the aforemen-tioned TGA experiments, this behavior reveals that the rate-determining step of the CH4 reforming reaction is not thecomplete dehydrogenation of CH4, but instead as already statedand shown by the model, Eq. (11) is the H2O assisted oxidativedissociation of methyl-species CH2

⁄ into CO and H2. The presenceof Au and its interaction with Ni further retards CH4 dissociationinto carbon species thus improving the carbon tolerance of thecatalyst.

Recent experimental evidence by Papaefthimiou et al. [70],based on ambient pressure X-ray photoelectron and near edgeX-ray absorption fine structure spectroscopies (APPES and NEXAFS,respectively) combined with online electrochemical and gas-phasemeasurements, concluded that Ni/GDC cermet anodes play a vitalcatalytic and electrocatalytic role in the activation of CH4 towardits electrochemical oxidation on SOFC electrochemical interfaces.The authors concluded that CH4 reacts and reduces surface Ce4+

into Ce3+ thus enhancing its electrocatalytic performance. In com-bination with DFT calculations, they proposed that CH4 is partiallyoxidized electrochemically into CO and H2 on the reduced Ce3+

sites.Thus, it is plausible that either CH�2 species or oxygenate CH2O�

species diffuse onto the Ni surface, where they are correspondinglyoxidized into CO2 and H2O or decompose into CO and H2 accordingto reactions (28) and (29), respectively. On the contrary under CH4

lean conditions, CH4 activation can take place rather on Ni, whileceria can be partially reduced, thus promoting the selectivity ofreaction (29) (oxidation). This can be rationalized by the observa-tion [70] that Ni is encapsulated in GDC especially under methanereducing atmosphere that causes total reduction in surface Ce4+

into Ce3+, while its degree of exposure depends on the degree ofreduction in ceria.

The observed non-faradaic rate changes, particularly underfuel-lean conditions, can be rationalized according to the effectof Electrochemical Promotion of Catalysis (EPOC) or non-faradaicElectrochemical Modification of Catalytic Activity (NEMCA effect)[64–68]. Normally, one would not anticipate the appearance ofelectrochemical promotion at such elevated temperatures, sinceEPOC with Pt and Ag catalyst-electrodes is limited to tempera-tures below 600 �C due to desorption of the oxygen anionic spill-over species which cause EPOC with O2� conductors. However,the metal-oxygen interaction is stronger with Ni rather than withPt and Ag, and thus, recently in their APPES experiments, Papaef-thimiou et al. [70] detected the accumulation of ionic oxygen spe-cies during current application (i.e., O2� pumping onto the Ni/GDC). The Ni/GDC electrode was exposed to CH4, and the oxida-tion state of Ni and Ce3+ remained unaffected. In this respect, itcan be claimed that the accumulated excess of oxygen ionic spe-cies can act as the Od� promoting species (as proposed by the the-ory of the EPOC phenomenon [64,66,68]) and affect the catalyticproperties of the Ni/GDC anode. This can be substantiatedthrough the enhancement of the oxidation and decompositioncatalytic reaction steps (Eq. (28)), thus promoting the overallsteam reforming catalytic reaction rate. The non-faradaic opera-tion of Au–Ni/GDC anode under CH4 rich conditions can be pro-ven as a potential example pointing to the wide application ofSOFCs as chemical co-generation reactors for the production ofelectricity and synthesis gas.

5. Conclusions

A mathematical model was developed to describe the catalyticand electrocatalytic performance of Ni-based anodes operatingunder internal CH4 steam reforming reaction conditions, underboth low and high steam-to-carbon ratio values. The model ac-counts for surface dissociation of CH4 and formation of methylspecies which then react with H2O from the gas phase to formCO and H2. Under high steam-to-carbon ratio values, it was foundthat the electrochemical oxidation of H2 and CO is favored. Themodel accounts for H2 selective oxidation at carbon-free adsorp-tion sites and CO oxidation on sites covered by a carbonaceousadsorbate. Under low steam-to-carbon ratio values, electrochem-ical partial oxidation of CH4 to CO and H2 in parallel with electro-chemical oxidation of H2 was found to take place. In this case,the model suggests that H2 is oxidized at carbon-free adsorptionsites, while CH4 on sites covered by a carbonaceous adsorbate.Interestingly, in both cases, the coverage of methyl-species ofthe catalyst surface coincides with that at the three-phaseboundaries. In each case, the model was found to be in goodagreement with experiment.

Acknowledgments

We thank Dr. G. Sourkouni and Professor C. Argirusis of theTechnical University of Clausthal, TUC and the National TechnicalUniversity of Athens, NTUA for the preparation of the LSM cathodepart of the SOFCs with the screen printing technique. The authorsalso thank Dr. V. Dracopoulos at FORTH/ICEHT for the SEM images.The research leading to these results was funded by the EuropeanUnion’s Seventh Framework Programme (FP7/2007-2013) for theFuel Cells and Hydrogen Joint Technology Initiative under theROBANODE project with Grand agreement number: 245355.

Sincere thanks are also expressed by C.G. Vayenas and A. Kat-saounis to the Greek General Secretary of Research & Technologyand the program ‘‘ARISTEIA’’ for financial support. We are alsothankful to our reviewers for their constructive and very usefulcomments.

128 S. Souentie et al. / Journal of Catalysis 306 (2013) 116–128

References

[1] A. Singh, J.M. Hill, J. Power Sources 214 (2012) 185.[2] J.T.S. Irvine, J. Power Sources 136 (2004) 203.[3] I. Gavrielatos, V. Drakopoulos, S.G. Neophytides, J. Catal. 259 (2008) 75.[4] D. Mogensen, J.D. Grunwaldt, P.V. Hendriksen, K. Dam-Johansen, J.U. Nielsen, J.

Power Sources 196 (2011) 25.[5] Fuel Cell Handbook, EG&G Technical Services Inc., 2004.[6] S. McIntosh, R.J. Gorte, Chem. Rev. 104 (2004) 4845.[7] H. Kim, C. Lu, W.L. Worrell, J.M. Vohs, R.J. Gorte, J. Electrochem. Soc. 149 (2002)

A247.[8] R.J. Gorte, J.M. Vohs, J. Catal. 216 (2003) 477.[9] H. Kishimoto, K. Yamaji, T. Horita, Y.P. Xiong, N. Sakai, M.E. Brito, H. Yokokawa,

J. Electrochem. Soc. 153 (2006) A982.[10] X.-F. Ye, B. Huang, S.R. Wang, Z.R. Wang, L. Xiong, T.L. Wen, J. Power Sources

164 (2007) 203.[11] V. Modafferi, G. Panzera, V. Baglio, F. Frusteri, P.L. Antonucci, Appl. Catal. A:

Gen. 334 (2008) 1.[12] D.K. Niakolas, J.P. Ouweltjes, G. Rietveld, V. Dracopoulos, S.G. Neophytides, Int.

J. Hydrogen Energy 35 (2010) 7898.[13] Y. Shiratori, T. Ijichi, T. Oshima, K. Sasaki, Int. J. Hydrogen Energy 35 (2010)

7905.[14] H. Sumi, Y.H. Lee, H. Muroyama, T. Matsui, K. Eguchi, J. Electrochem. Soc. 157

(2010) B1118.[15] Y. Chen, F. Chen, W. Wang, D. Ding, J. Gao, J. Power Sources 196 (2011)

4987.[16] J.-H. Myung, H.-J. Ko, J.-J. Lee, J.-H. Lee, S.-H. Hyun, Int. J. Hydrogen Energy 37

(2012) 11351.[17] R. Peters, R. Dahl, U. Klüttgen, C. Palm, D. Stolten, J. Power Sources 106 (2002)

238.[18] C. Sun, U. Stimming, J. Power Sources 171 (2007) 247.[19] E.V. Tsipis, V.V. Kharton, J. Solid State Electrochem. 12 (2008) 1367.[20] J.B. Goodenough, Y.-H. Huang, J. Power Sources 173 (2007) 1.[21] W. An, D. Gatewood, B. Dunlap, C.H. Turner, J. Power Sources 196 (2011)

4724.[22] D.H. Prasad, H.I. Ji, H.R. Kim, J.W. Son, B.K. Kim, H.W. Lee, J.H. Lee, Appl. Catal. B

101 (2011) 531.[23] B.C.H. Steele, Nature 400 (1999) 619.[24] H. Tu, H. Apfel, U. Stimming, Fuel Cells 6 (2006) 303.[25] E. Ramı́rez-Cabrera, A. Atkinson, D. Chadwick, Appl. Catal. B 47 (2004) 127.[26] O.A. Marina, C. Bagger, S. Primdahl, M. Mogensen, Solid State Ionics 123 (1999)

199.[27] B. Mosqueda, J. Toyir, A. Kaddouri, P. Gélin, Appl. Catal. B 88 (2009) 361.[28] W. Wang, S.P. Jiang, A.I.Y. Tok, L. Luo, J. Power Sources 159 (2006) 68.[29] T.-J. Huang, C.-H. Wang, J. Power Sources 163 (2006) 309.[30] T.-J. Huang, M.-C. Huang, Chem. Eng. J. 135 (2008) 216.[31] P. Hofmann, K.D. Panopoulos, L.E. Fryda, A. Schweiger, J.P. Ouweltjes, J. Karl,

Int. J. Hydrogen Energy 33 (2008) 2834.[32] T. Horita, K. Yamaji, T. Kato, N. Sakai, H. Yokokawa, J. Power Sources 131 (2004)

299.[33] J.P. Trembly, A.I. Marquez, T.R. Ohrn, D.J. Bayless, J. Power Sources 158 (2006)

263.[34] D.G. Ivey, E. Brightman, N. Brandon, J. Power Sources 195 (2010) 6301.[35] P. Lohsoontorn, D.J.L. Brett, N.P. Brandon, J. Power Sources 183 (2008) 232.

[36] L. Zhang, S.P. Jiang, H.Q. He, X. Chen, J. Ma, X.C. Song, Int. J. Hydrogen Energy 35(2010) 12359.

[37] T. Hibino, A. Hashimoto, M. Yano, M. Suzuki, S. Yoshida, M. Sano, J.Electrochem. Soc. 149 (2002) A133.

[38] I.V. Yentekakis, J. Power Sources 160 (2006) 422.[39] M. Boaro, V. Modafferi, A. Pappacena, J. Llorca, V. Baglio, F. Frusteri, P. Frontera,

A. Trovarelli, P.L. Antonucci, J. Power Sources 195 (2010) 649.[40] F. Besenbacher, I. Chorkendorff, B.S. Clausen, B. Hammer, A.M. Molenbroek, J.K.

Norskov, I. Stensgaard, Science 279 (1998) 1913.[41] D.K. Niakolas, M. Athanasiou, V. Dracopoulos, I. Tsiaoussis, S. Bebelis, S.G.

Neophytides, Appl. Catal. A 456 (2013) 223.[42] N.M. Galea, D. Knapp, T. Ziegler, J. Catal. 247 (2007) 20.[43] T.-J. Huang, M.-C. Huang, M.-S. Huang, Appl. Catal. A 354 (2009) 127.[44] C.J. Fu, S.H. Chan, X.M. Ge, Q.L. Liu, G. Pasciak, Int. J. Hydrogen Energy 36 (2011)

13727.[45] H. Kan, H. Lee, Catal. Commun. 12 (2010) 36.[46] Y. Wang, F. Yoshiba, M. Kawase, T. Watanabe, Int. J. Hydrogen Energy 34

(2009) 3885.[47] J. Wei, E. Iglesia, J. Catal. 224 (2004) 370.[48] L. Maier, B. Schädel, K. Herrera Delgado, S. Tischer, O. Deutschmann, Top. Catal.

54 (2011) 845.[49] S. Bebelis, A. Zeritis, C. Tiropani, S.G. Neophytides, Ind. Eng. Chem. Res. 39

(2000) 4920.[50] S. Bebelis, S. Neophytides, N. Kotsionopoulos, N. Triantafyllopoulos, M.T.

Colomer, J. Jurado, Solid State Ionics 177 (2006) 2087.[51] M.A. Buccheri, A. Singh, J.M. Hill, J. Power Sources 196 (2011) 968.[52] J.R. Rostrup-Nielsen, Catal. Sci. Technol. 5 (1984) 1.[53] M.I. Temkin, Adv. Catal. 28(c) (1979) 173.[54] J. Xu, G.F. Froment, AIChE 35 (1989) 88.[55] L.M. Aparicio, J. Catal. 165 (1997) 262.[56] D. Blaylock, Y.-A. Zhu, W. Green, Top. Catal. 54 (2011) 828.[57] E. Achenbach, E. Riensche, J. Power Sources 52 (1994) 283.[58] M. Boder, R. Dittmeyer, J. Power Sources 155 (2006) 13.[59] D.L. King, J.J. Strohm, X. Wang, H.-S. Roh, C. Wang, Y.-H. Chin, Y. Wang, Y. Lin, R.

Rozmiarek, P. Singh, J. Catal. 258 (2008) 356.[60] S.H. Clarke, A.L. Dicks, K. Pointon, T.A. Smith, A. Swann, Catal. Today 38 (1997)

411.[61] N. Nakagawa, H. Sagara, K. Kato, J. Power Sources 92 (2001) 88.[62] H. Timmermann, D. Fouquet, A. Weber, E. Ivers-Tiffée, U. Hennings, R. Reimert,

Fuel Cells 6 (2006) 307.[63] S. Ivanova, C. Petit, V. Pitchon, Appl. Catal. A 267 (2004) 191.[64] C.G. Vayenas, S. Bebelis, C. Pliangos, S. Brosda, D. Tsiplakides, Electrochemical

Activation of Catalysis: Promotion, Electrochemical Promotion and Metal–Support Interactions, Kluwer Academic/Plenum Publishers, New York, 2001.

[65] S. Brosda, C.G. Vayenas, J. Wei, Appl. Catal. B 68 (2006) 109.[66] C.G. Vayenas, C.G. Koutsodontis, J. Chem. Phys. 128 (2008) 182506.[67] A. Katsaounis, J. Appl. Electrochem. 40 (2010) 885.[68] P. Vernoux, L. Lizarraga, M.N. Tsampas, F.M. Sapountzi, A. De Lucas-Consuegra,

J.L. Valverde, S. Souentie, C.G. Vayenas, D. Tsiplakides, S. Balomenou, E.A.Baranova, Chem. Rev. (2013). dx.doi.org/10.1021/cr4000336.

[69] N.C. Triantafyllopoulos, S.G. Neophytides, J. Catal. 239 (2006) 187.[70] V. Papaefthimiou, M. Shishkin, D.K. Niakolas, M. Athanasiou, Y.T. Law, R.

Arrigo, D. Teschner, M. Hävecker, A. Knop-Gericke, R. Schlögl, T. Ziegler, S.G.Neophytides, S. Zafeiratos, Adv. Energy Mater. 3 (2013) 762.