Mechanistic Study and Active Sites Investigation of Hydrogen Production from Methane and H₂O Steady-State and Transient Reactivity with Ir/GDC Catalyst

Abstract

Catalytic activity, mechanisms, and active sites were determined for methane steam reforming (MSR) over gadolinium-doped ceria (GDC) supported iridium (0.1 wt%) prepared by impregnation of GDC with iridium acetylacetonate. Isothermal steady-state rate measurements followed by micro-gas chromatography analysis were performed at 660 and 760 °C over Ir/GDC samples pretreated in N₂ or H₂ at 900 °C. Transient responses to CH₄ or H₂O step changes in isothermal conditions were carried out at 750 °C over Ir/GDC pretreated in He or H₂ using online quadrupole mass spectrometry. In the proposed mechanism, Ir/GDC proceeds through a dual-type active site associating, as follows: (i) Ir metallic particles surface as active sites for the cracking of CH₄ into reactive C species, and (ii) reducible (Ce⁴⁺) sites at GDC surface responsible for a redox mechanism involving Ce⁴⁺/Ce³⁺ sites, being reduced by reaction with reactive C into CO (or CO₂) depending on the oxidation state of GDC and re-oxidized by H₂O. Full reduction of reducible oxygen species is possible with CH₄ after He treatment, whereas only 80% is reached in CH₄ after H₂ treatment.

1. Introduction

Solid Oxide Fuel Cells (SOFCs) are promising candidates for high efficiency power generation [1,2,3,4,5,6]. Moreover, their high operating temperatures allow for a real flexibility of fuels such as fossil sources (natural gas) or bioresources (biogas, bioethanol, etc.), since these fuels can be converted into hydrogen at the anode compartment (direct internal reforming) [7,8,9,10,11,12]. Methane, which has mostly been used and tested so far as a fuel, is reformed by using steam as the co-feed. Over an Ni-based catalyst (Ni/YSZ cermet) [13,14], the reactions of methane steam reforming (MSR, Equation (1)) and water–gas shift mainly proceed as follows [15]:

CH_4(g) + H₂O_(g) ⇆ CO_(g) + 3H_2(g)   ΔH°₂₉₈ = +206 KJ/mol

(1)

CO_(g) + H₂O_(g) ⇆ CO_2(g) + H_2(g)   (water–gas shift) ΔH°₂₉₈ = −41 KJ/mol

(2)

The cermet also ensures the electrochemical oxidation of the produced hydrogen, thus generating electrical current. However, Ni is a very good catalyst for methane cracking (MC) reactions [16]:

CH_4(g) ⇆ C_(s) + 2H_2(g)   ΔH°₂₉₈ = +75 KJ/mol

(3)

Operating with S/C ratios higher than unity is used as a solution against the risk of extensive C deposition, finally leading to irreversible damage of the cell and the rapid decrease of its performances. However, increasing the amount of steam in the feed has the following drawbacks: (i) lowering the performance of the cell, and (ii) inducing strong mechanical constraints altering the lifetime. Indeed, when using high S/C ratios, the highly endothermic SMR mainly proceeds near the inlet of the cell. The temperature of the anode at the inlet thus strongly decreases, while the exothermic electrochemical H₂ oxidation causes an increase in temperature along the anode [17,18]. The resulting inappropriate distribution of endothermicity and exothermicity along the anode is detrimental to the integrity of the anode. Alternative solutions such as a Gradual Internal Reforming (GIR) [19,20,21,22] have been proposed. In GIR mode, low amounts of steam are admitted with methane to the cell inlet to reduce the conversion of methane; the progressive conversion of methane by SMR is achieved along the anode by the coupling of the catalytic and electrochemical reactions, and, since the latter reaction produces steam, the co-reactant in the SMR reaction. This, however, requires the use of new anode materials exhibiting specific catalytic properties: (i) moderate catalytic activity in CH₄ steam reforming, and (ii) high resistance against the thermodynamically favoured formation of carbon deposits because of high CH₄/H₂O molar ratios (near to 10) at the anode compartment inlet. The feasibility of the GIR concept was first demonstrated by using a lanthanum chromite anode in a tubular set-up, but very low current densities were obtained due to the poor adhesion of the chromite onto the YSZ electrolyte [23]. Another solution to perform reforming of hydrocarbons without coking was proposed by adding a catalytic layer onto a conventional anode. This proved to be successful with a catalyst layer active for CO₂-reforming of iso-octane [24]. An inert barrier layer was also found to extend the coke-free operation range using Ni/YSZ anodes fed with pure methane [25]. An Ru/SDC catalyst layer was successfully used for methane electro-catalytic oxidation [26].

More recently, it was shown that a conventional Ni/YSZ cermet anode could be modified with a catalytic layer active in methane steam reforming, resistant to carbon formation, and operating with dry methane [27,28]. The cell, built up from a conventional (LSM/YSZ/Ni-YSZ) assembly to which a Ir-CeO₂ catalyst layer was added at the anode side, could be operated with pure dry methane (without the need for a steam supply) at 900 °C at 0.6 V yielding a current density of about 0.1 A/cm² at max power for 120 h [29]. This unique behaviour was attributed to the remarkable ability of the Ir/Ce_0.9Gd_0.1O_2-x (Ir/GDC) catalyst (0.1 wt % Ir) to convert CH₄ in GIR mode without formation of carbon deposits [30,31]. The home-made cell used, although exhibiting very promising performances, was clearly not optimized, and future work is currently in progress to improve the performance.

In the configuration combining a classical anode and a catalytic layer on the anode side, an optimized commercial Ni cermet anode is used. Specific treatments at high temperature in H₂ prior operation are applied. It has to be mentioned that, during operation of the cell, the atmosphere above the Ni cermet and the active catalytic layer deposited on it also contains significant amounts of unconverted H₂. As a result, the catalytic layer, if used, will be submitted to H₂-containing atmospheres. It has been previously shown that treatment of GDC in H₂ before catalytic testing in CH₄/H₂O mixtures could have a negative influence on catalytic properties [32]. GDC shows the same activity and the same slow deactivation for N₂ or H₂O pretreated samples, while the H₂-pretreated sample was less active and did not deactivate [32]. The deactivation process was mainly explained by irreversible changes in the redox properties of GDC taking place during the reaction process or upon H₂ pretreatment.

The present work addresses the study of the CH₄/H₂O reaction over Ir/GDC catalysts, and especially the better understanding of the reaction mechanism and catalytic sites. The influence of hydrogen pretreatment on the catalytic behaviour of Ir/GDC in CH₄/H₂O reaction is also examined. Various techniques are used: steady-state rate measurements, transient responses to successive CH₄, and H₂O step changes in isothermal conditions. Possible changes in the structure and redox properties are addressed. The respective roles of iridium and GDC in MSR reactions are thoroughly described and discussed.

2. Materials and Methods

2.1. Sample Preparation

The Ir-containing Ce_0.9Gd_0.1O_2-x catalyst (Ir/GDC) was prepared by impregnation technique. The appropriate amount of an iridium acetylacetonate (Alfa Aesar) solution in toluene was added to a suspension of GDC (Praxair, 40.9 m²/g) in toluene maintained under stirring at 50 °C for 4 h. After complete evacuation of the solvent under reduced pressure, the catalyst was dried overnight at 120 °C and calcined in flowing O₂ at 350 °C for 6 h. It was checked by TPO experiments over the dried impregnated catalyst that the decomposition of the Ir precursor was complete in O₂ at 350 °C. The catalyst sample was finally treated at 900 °C for 2 h in flowing N₂.

2.2. Physico-Chemical Characterizations

Elemental analysis of the catalyst was performed by dissolving the sample in H₂SO₄/HNO₃, and the resulting solution was analysed by ICP-AES. Specific surface area was determined by BET nitrogen adsorption using a Micromeritics ASAP2000 (Norcross, GA, USA) after outgassing the sample during 2h at 200 °C. The crystal structure was checked by X-ray diffraction (XRD, Model: INEL 120) using Cu K_α1 radiation in the 2θ range of 20–90°.

Temperature-programmed oxidation experiments in O₂ (O₂-TPO) were performed using a Pfeiffer Omnistar quadrupole after catalytic tests according to the following procedure. After each test, the reactor was purged in N₂ at 900 °C, cooled down to room temperature, and isolated with 2 valves before being removed from the experimental set up. This allowed preventing sample exposure to air after testing. The reactor was then installed on the O₂-TPO experimental set up and purged in He at room temperature. The sample was then exposed to a 1% O₂/He flow (1.8 L/h) at room temperature, and O₂ adsorption was monitored. After completion of O₂ adsorption, the temperature was linearly increased from room temperature to 1000 °C at the rate of 20 °C/min. Calibrations were performed according to the procedure used for transient experiments.

2.3. Catalytic Testing

Catalytic tests were carried out in a continuous-flow system at atmospheric pressure using a tubular U-shaped quartz reactor. Twenty mg of catalyst sample were introduced onto a quartz-wool plug inside the reactor (4 mm inner diameter). A thermocouple in contact with the external wall of the reactor at the position where the catalyst bed is located allowed for monitoring the catalyst temperature. In order to simulate the conditions for a Gradual Internal Reforming, the reactant mixture consisted of 50 mol % CH₄ and 5 mol % H₂O in N₂ as balance (total flow rate of 6.5 L/h). Suitable water-vapour concentration in the reaction mixture was obtained by flowing the adequate mixture of CH₄ and N₂ dry gases throughout a saturator containing distilled water and maintained at 33 °C (thermostated bath). Water content in the reactor outflow was determined using an Edgetech Dew Prime I dew point monitor. An M&C ECP gas cooler maintained at 3 °C was used to reduce steam concentration in the feed in order to allow reliable analysis by gas chromatography. CH₄, H₂, CO, and CO₂ were analysed with a Varian micro-GC equipped with appropriate columns (molecular sieve 5A and Porapak) and thermal conductivity detector (TCD).

Prior to the catalytic tests, the samples were treated in N₂ (6.5 L/h) at 900 °C or pure H₂ at 800 °C for 2 h. The temperature was then decreased to 760 or 660 °C for reaction study.

Isothermal transient experiments of reduction by CH₄ and subsequent re-oxidation by H₂O were performed using a Pfeiffer Omnistar quadrupole mass spectrometer (QMS). Prior to the experiment, the samples (20 mg) were treated for 2 h in He at 900 °C or in 1% H₂/He up to 800 °C (linear heating rate of 10 °C/min) before reaching the reaction temperature equal to 750 °C. Experiments with dry CH₄ were carried out using a CH₄ (4900 ppm)/Ar (528 ppm)/He mixture (flow rate of 1.8 L/h). Subsequent re-oxidation by H₂O was achieved in an H₂O (2300 ppm)/Ar (7700 ppm)/He mixture. The QMS signals for H₂ (m/e = 2), CH₄ (m/e = 15), H₂O (m/e = 18), CO (m/e = 28), O₂ (m/e = 32), CO₂ (m/e = 44), and Ar (m/e = 40) were continuously monitored as a function of time on stream. Calibrations for different gases were performed using 1% H₂/He, 1% CH₄/He, 1% O₂/He, 1% CO/He, 1% CO₂/He, and 1% Ar/He mixtures (Air Liquide). For the experiment with H₂O and calibration of the H₂O signal, suitable water-vapour concentration in the reaction mixture was obtained by mixing adjusted flows of dry Ar/He and wet He (flowing through a saturator containing distilled water and maintained at 7 °C). Ar, being not adsorbed by the samples, was used as a tracer. This allowed for obtaining the response of the catalytic bed to a step change without any contribution of chemical reactions and correcting the signals for slight calibration variations during the experiments.

3. Results

3.1. Characterization

The iridium content of the sample measured by ICP-AES is very close to nominal Ir content (0.1 wt%). The impregnation of GDC support by an iridium precursor did not affect the surface area of the catalyst (43 m²/g). After calcination, the BET surface area of GDC decreased as a result of sintering (

Table 1. Loading catalyst and specific surface areas of the samples measured before reaction.

Catalyst	% Ir Given by Elemental Analysis	Specific Surface Area (m2/g)
GDC	-	43
GDC activated under N2, 900 °C, 2 h	-	6
Ir/GDC	0.07	42

).

Calcination of iridium catalysts was performed with a flow of 1% O₂ in He (1.7 L/h) and a heating rate of 0.5 °C/min from room temperature to 350 °C, based on the decomposition profile of Ir acetylacetonate precursor showing a maximum consumption of oxygen at 270 °C (Figure 1). This result indicates that, after calcination at 350 °C, the catalyst is free of carbon.

The results of XRD analysis are presented in Figure 2. XRD diffractograms show that the obtained catalysts are crystalline, and all the peaks of the powders correspond to the fluorite structure of CeO₂ (PDF card number: 34-0394). No crystalline phase corresponding to Gd₂O₃ could be found, suggesting an intimate mixing of Ce³⁺ and Gd³⁺ cations in the support.

Small signals in the 1% Ir/GDC spectrum at 2θ = 40.6° and 83.4° correspond to (111) and (311) of metallic iridium, respectively. Too small sizes of Ir crystallites cause the absence of iridium signals in the spectra of the catalysts with 0.1% Ir/GDC.

3.2. CH₄/H₂O Reaction

The catalytic behaviour of the Ir/GDC catalyst in methane steam reforming under water-deficient conditions was studied by measuring the evolution of reactant and products concentration as a function of time on stream. Two reaction temperatures (660 and 760 °C) as well as the presence or not of H₂ in the pretreatment atmosphere were considered.

Figure 3 and Figure 4 show the evolution of H₂, CO, and CO₂ concentrations obtained at 760 °C over Ir/GDC samples respectively treated in N₂ at 900 °C or H₂ at 800 °C prior to the reaction. Without pre-exposure to H₂, the reaction steady state was reached within 4 min (i.e., since the first experimental point after exposure to reactants). H₂ (8.7 mol%) and CO (2.6 mol%) formed as main products, while CO₂ formation (0.1 mol%) was low. No change in catalytic behaviour could be observed within 90 min and after 16 h. On the contrary, treating the sample in an H₂-containing atmosphere prior to testing at 760 °C induced a slow progressive increase of H₂ and CO concentrations corresponding to a strong activation of the catalyst for steam-reforming reaction. Within 6 h reaction in CH₄/H₂O mixture, the H₂ (resp. CO) concentration increased from 2.7 mol% (resp. 0.4 mol% CO) up to 6.7 mol% (resp. 2 mol% CO). On the other hand, the CO₂ formation followed an opposite trend, decreasing from 0.4 to 0.2 mol%. It was therefore obvious that treatment in an H₂-containing atmosphere had a severe negative effect on Ir/GDC catalytic behaviour in steam reforming, similar to what was previously observed with GDC alone [32]. Interestingly, the H₂-treated Ir/GDC sample reached, at steady state, only slightly lower levels of H₂ and CO production than the one treated in N₂. It might be thus inferred that H₂ at 800 °C strongly modified surface-active sites responsible for methane steam reforming. These changes were at least partially reversible, since prolonged exposure to reactants allowed for mostly the regeneration of catalytic steam-reforming activity.

Performing catalytic tests at 660 °C on Ir/GDC samples treated or not in H₂ revealed in both cases an activation period for H₂ production (Figure 5 and Figure 6). For the N₂-treated sample, this period lasted ca. 450 min, and the H₂ production increased from 0.5 to 2.2 mol%. While CO and CO₂ production increased during this period, the CO/CO₂ ratio strongly varied, being close to zero in the early stage of the reaction and increasing up to 0.9 after reaching a steady state. Contrary to what was observed at 760 °C, profound changes in the surface state of the catalyst were suggested to occur under reactants at 660 °C. Upon H₂ pretreatment, basically the same trend was observed, but the activation period turned out to be longer (600 min), essentially because of the slower activation process in the first 5 h compared to the sample treated in N₂. The steady-state reaction was also less selective into CO, the H₂-treated sample producing less CO (0.1 mol%) than the N₂-treated sample (0.3mol%) but with equivalent CO₂ amounts (0.3 mol%). This suggested slight irreversible surface changes due to H₂ pretreatment.

After reaction for ca. 5 h, the sample was purged overnight in N₂ at 660 °C before re-introducing the reaction mixture. This intermediate purge did not change the catalytic properties.

O₂-TPO experiments performed after catalytic testing revealed the absence of any detectable carbon deposits, irrespective of the pretreatment and reaction temperature. The lower limit of CO₂ amount that could be detected by O₂-TPO in our experimental conditions was estimated to be less than 1 µmol/g.

It could be concluded from these experiments that pretreatment in H₂-containing atmosphere always had a negative impact on the catalytic CH₄/H₂O reaction over the Ir/GDC catalyst for reaction temperatures in the interval 660–760 °C with respect to treatment in inert gas. However, this has to be qualified by the fact that the catalytic activity always increased during reaction, reaching at steady state only slightly lower rates of H₂ (resp. CO) production compared to the sample pretreated in N₂. The period during which the catalyst re-activates following contact with H₂ turns out to strongly depend on the reaction temperature, being shorter upon increasing reaction temperature.

Finally, the catalytic activity of Ir/GDC was measured after a treatment procedure given by Julich (FZJ procedure) for the activation of the commercial NiYSZ anode (

Table 2. Julich “FZJ” procedure employed for Ir/GDC pretreatment (heating up to 900 °C in Ar at 1 °C/min).

Step	Holding Time (min)	H2 (mL/min)	H2O (mL/min)	Ar (mL/min)	Air (mL/min)
-	-	-	0	500	500
1st	75	80	0	500	580
2nd	15	160	10	500	660
3rd	5	320	20	500	820
4th	5	640	30	360	1000
5th	5	1000	30	0	1000

). The steady-state catalytic activity and products distribution measured at 650 °C are shown in

Table 3. Catalytic activity of Ir/GDC as a function of pretreatment.

Pretreatment	Rate of CH4 Consumption (µmol/s.gcat)	CO Formed (% mol)	CO2 Formed (% mol)	H2 Formed (% mol)
N2, 900 °C, 2 h	24.0	0.3	0.35	2.2
H2, 900 °C, 2 h	14.7	0.1	0.30	1.7
Julich FZJ	13.7	0.1	0.39	1.3

. Almost the same activity was obtained for the samples tested respectively at 650 °C after the FZJ activation and at 660 °C after the “homemade” pretreatment in H₂.

3.3. Transient Reaction of Dry CH₄ over Ir/GDC at 750 °C After Treatment in He

Transient experiments consisting of reacting dry CH₄ with Ir/GDC samples pretreated in inert gas (He) at 900 °C were carried out. Transient gas concentration response curves obtained for CH₄, H₂, CO, CO_2, and H₂O after the switch He → 0.5%CH₄/He at 750 °C are shown in Figure 7.

It could be observed that CH₄ was consumed immediately after contact with the catalyst, which led to CO₂ release (no CO nor H₂ formed). H₂O is likely to form simultaneously with CO₂. Indeed, H₂O is detected with some delay with respect to CO₂ appearance, which could be explained by H₂O adsorption on line walls at the outlet of the reactor. We then suggest the complete oxidation of methane by strongly oxidising surface oxygen species proceeding at this stage according to Equation (4):

(CeO₂)s + n/4 CH₄ → (CeO₂ − n)s + n/4 CO₂ + n/2 H₂O

(4)

Thus creating intrinsic defects (oxygen vacancies) likely at the surface or near-surface layers. In Kröger and Vink notation, this can be rationalized by the following equation:

CeO₂ + n/4 CH₄ → 2n Ce’Ce + (1 − 2n) CeCe + n VÖ + (2 − n) OO + n/4 CO₂(g) + n/2 H₂O(g)

(4’)

In which a doubly positive-charged oxygen vacancy (VÖ) is created for each consumed CH₄ molecule, thus leading to the release of one molecule of CO gas and two molecules of H₂ gas. The charge neutrality requires the reduction of two Ce⁴⁺ ions of the lattice into Ce³⁺ (Ce’Ce). The complete reduction of oxidised Ir sites into Ir metal might also occur according to the following:

IrO_x + x/4 CH₄ → Ir⁰ + x/4 CO₂ + x/2 H₂O

(5)

The existence of such oxidised Ir sites can neither be proved nor excluded after treatment in inert gases at 900 °C due to the very low Ir loading. It is worth pointing out that in another study on the same material using X-ray photoelectron spectrometry (XPS) and aberration-corrected high-resolution transmission electron microscopy (HRTEM), it was reported that the catalyst activation is due to the formation of Ir nanoparticles involving both mobile O species of GDC and oxidized Ir species in the catalytic reaction [15].

The total amounts of H₂O and CO₂ formed during the experiment are respectively equal to 0.92 and 1.2 µmol. According to Equations (4) and (5), the production of H₂O should be twice as much as that of CO₂, which is not observed. The excess of CO₂ with respect to the expected amount formed by complete oxidation reactions can then be reliably attributed to another type of reaction. Considering the complete oxidation reaction of CH₄ into CO₂ and H₂O, the formation of 0.9 µmol H₂O strongly exceeds the only reduction of 0.1 µmol IrO_x species contained in the sample. Indeed, assuming x = 2 (maximum value possible), and taking into account the 0.1 µmol of Ir contained in the sample, only 0.1 µmol H₂O from reaction (5) should form. This suggests that reaction (4), i.e., the reaction involving O species of the ceria, mostly contributes to H₂O formation. This implies the existence of sites at ceria surface specifically responsible for the complete oxidation of methane and being present in the fully oxidised GDC structure. The rate of CO₂ formation went through a maximum (1.9 µmol/s.gcat) after 12 s reaction and rapidly decreased with time. The decrease of the CO₂ formation rate could be related to the disappearance of fully oxidised sites at the surface due to their consumption by CH₄ and to O diffusion limitations from the bulk to the surface.

After a small delay (a few seconds), CO and H₂ were formed and released together with CO₂. As previously indicated, it has to be mentioned that CO₂ to some extent has to be produced together with H₂ in addition to H₂O (reaction 6 and 6’):

CeO₂ + n/2 CH₄ → CeO₂ − n + n/2 CO₂ + n H₂

(6)

CeO₂ + n/2 CH₄ → 2n Ce’Ce + (1 − 2n) CeCe + n VÖ + (2 − n) OO + n/2 CO_2(g) + n H_2(g)

(6’)

It must be mentioned that after 12 s, reaction 4 and 6 can proceed simultaneously, both being limited by O diffusion limitations to the surface, which explains the rapid decrease of CO₂ formation rate via these two reactions.

After about 200 s, CO₂ was no longer detected, CO and H₂ being the only products. After 200 s and up to about 800 s, the H₂ to CO ratio is constant and equal to 2. The reaction can be written according to Equations (7) and (7’):

CeO₂ + n CH₄ → CeO₂-n + n CO + 2n H₂

(7)

CeO₂ + n CH₄ → 2n Ce’Ce + (1 − 2n) CeCe + n VÖ + (2 − n) OO + n CO_(g) + 2n H_2(g)

(7’)

Equations (6), (6’), (7), and (7’) must involve bulk O (respectively Ce⁴⁺) species, and the process should be dependent on the rate of diffusion of lattice O (resp. O vacancies) to the surface.

According to Figure 8, reaction (7, 7’) proceeds alone up to ca. 70–75% of reduction of potentially reducible Ce⁴⁺ → Ce³⁺ species in the CeO₂ fraction of GDC.

During this period, the reaction rate of CO formation first (12–50 s) rapidly reached a value close to its maximum (3.7 µmol/s.gcat). Sites being responsible for this reaction are partially oxidised, requiring less O species from the bulk to be fully regenerated. O diffusion from the bulk to the surface is not limiting. Beyond 200 s, the CO formation rate started to progressively decrease, which can be attributed to the occurrence of diffusion limitations due to the decreased availability of O species from the bulk.

Beyond 70–75% reduction of O species potentially reducible, H₂ was also produced by CH₄ cracking (reaction 3).

Accordingly, the H₂ to CO ratio sharply increased, as shown in Figure 9. This reaction continued after complete reduction of Ce sites and stopped. Carbon deposits in limited amounts were thus produced as previously shown [31], i.e., 12 µmol C on the sample surface.

Finally, the total amounts of CO and CO₂ formed through reactions (4), (6), and (7) can be summarized in

Table 4. Characteristics of CO and CO₂ production (amounts, sequence of appearance, reaction) during the transient step reaction of CH₄ at 750 °C with 0.020 g Ir/GDC pretreated in He at 900 °C.

Gas Produced	Time Interval of Formation (s)	Type of Reaction	Amount (µmol)	Reduced Ceria O (µmol)
CO2	0–200	(4)	0.45	1.8
CO2	12–200	(6)	0.75	1.5
CO	12–4000	(7)	49.9	49.9
Total amount of reduced ceria O				53.2

.

It can be observed that the total amount of reducible O species present in the CeO₂ fraction of Ir/GDC (52.6 µmol) is fairly close to the total amount of ceria O species reduced by CH₄, which leads to the conclusion that CH4 can fully reduce CeO₂ into Ce₂O₃ in the Ir/GDC sample after He treatment at 900 °C.

3.4. Transient Reaction of Dry CH₄ over Ir/GDC at 750 °C After Treatment in H₂/He

Upon pretreatment in 1% H₂/He (Figure 9), neither CO₂ nor H₂O formation was observed in the early stages of the reaction with dry methane. This indicates that sites responsible for complete CH₄ oxidation were all reduced during treatment in H₂ and transformed into sites responsible for partial CH₄ oxidation. As a matter of fact, the H₂ reduction of the CeO₂ fraction in GDC is usually represented by Equations (8) and (8’) [33]:

CeO₂ + n H₂ → CeO_2-n + n H₂O

(8)

CeO₂ + n H₂ → 2n Ce’Ce + (1 − 2n) CeCe + n VÖ + (2 − n) OO + n H₂O_(g)

(8’)

In which, for n moles of H₂ consumed, n moles of doubly positive-charged oxygen vacancies (VÖ) are created with 2n moles of Ce³⁺ ions (Ce’Ce) and n moles of H₂O gas released. The H₂ consumption during the pretreatment being equal to 7 µmol, it could be derived that 14 µmol of Ce⁴⁺ are reduced, which corresponds to only 13% reduction extent based on the complete Ce⁴⁺ → Ce³⁺ transition, in agreement with previous data [34].

Upon reaction of Ir/GDC with CH₄, CO instantly formed and rapidly (0–6 s) reached a maximum production rate close to 2.9 µmol/s.gcat. Simultaneously, the H₂ formation sharply increased. It is worth noting that the maximum H₂ concentration measured after treatment in H₂ (0.53 mol %) was slightly lower than the one measured upon treatment in He (0.74 mol %). In fact, the maximum rate of CO formation measured after H₂ treatment and corresponding to 0.13 reduction extent (2.8 µmol/s.gcat) is lower than the one which was obtained at 0.13 reduction extent in CH₄ after He treatment (3.6 µmol/s.gcat). This suggests that the reduction by H₂ would have a negative effect on the reactivity of mobile O species with CH₄. Figure 10 shows the variations of the total extent of Ce⁴⁺ → Ce³⁺ reduction (including reduction by H₂) with time. The total amount of CO produced (36 µmol) is equivalent to a 68% reduction of potentially reducible O species in Ir/GDC sample. The total fraction of Ce⁴⁺ → Ce³⁺ reduction in H₂ treatment and CH₂ step is therefore equal to 81%, which is significantly less than that obtained for the sample pretreated in He (100%). This confirms that H₂ pretreatment diminishes the capability of CH₄ to reduce CeO₂ into Ce₂O₃ at 750 °C in the Ir/GDC sample.

As observed during reaction with CH₄ after treatment in He, the H₂ to CO ratio strongly increases beyond 60% Ce⁴⁺ → Ce³⁺ reduction due to C formation. As shown in Figure 9, the H₂ formation rate progressively decreases and finally stops, which indicates that CH₄ cracking (responsible for C formation) stops, too.

3.5. Transient Reaction of H₂O with Ir/GDC at 750 °C After Reaction with Dry CH₄

Figure 11 shows the transient response of Ir/GDC pretreated in He at 900 °C to a He → 0.23% H₂O/0.5%Ar/He step change at 750 °C subsequent to the reaction in dry CH₄ at 750 °C. In the first stage, H₂O was totally consumed by the sample, while H₂ was mainly produced (0.23 mol%) together with small amounts of CO (0.007 mol%). The production of H₂ could be attributed to the re-oxidation of reduced Ce³⁺ sites of GDC according to reactions (9) and (9’):

CeO_1.5 + n H₂O → CeO_1.5+n + n H₂

(9)

Using the defect notation, the re-oxidation process of the ceria fraction in GDC solid solution is written as follows:

2n Ce’Ce + n VÖ + n H₂O → 2n CeCe + n OO + n H_2(g)

(9’)

The formation of CO would be attributed to the so-called carbon gasification, which is the direct reaction of carbon deposits with H₂O, according to Equation (10):

C + H₂O → CO + H₂     ΔG < 0 for T > 700 °C

(10)

It can be inferred that the re-oxidation of the reduced GDC (reaction 9) is much more favoured than the C gasification (reaction 10). In a subsequent step (1000–1200 s), H₂O consumption sharply decreased, thus leading to a sharp decrease of H₂ formation, which got stabilized to a lower level (ca. 0.07 mol% H₂). Simultaneously, CO₂ started to form and its concentration reached half that of produced H_2, while CO was still formed in the same amount as during the preceding step. It was checked that the onset of CO₂ formation coincided with the completion of ceria re-oxidation. This would suggest that active sites for CO₂ formation are fully oxidized sites at the GDC surface. The reaction could be written as follows:

C + 2 H₂O → CO₂ + 2H₂

(11)

The sites catalysing reaction (11) are likely the same as those responsible for complete CH₄ oxidation during the early stages of the transient reaction of dry CH₄ with the Ir/GDC sample. The fact that CO formation did not change throughout the whole experiment suggested reaction (8) as being totally independent on reaction (11), and the water–gas shift contribution as being negligible. The total number of carbon deposits deduced from CO and CO₂ formation in the experiment was equal to 550 µmol/g. This amount fairly well agrees with that deduced from TPO experiments following the reaction of Ir/GDC with dry methane at 750 °C [31].

It has to be mentioned that, despite the lower concentration of H₂O with respect to dry methane in this sequence, the rate of ceria re-oxidation is very high as a result of always being limited by the H₂O supply. It is derived that this step is not rate limiting, which agrees with the first order rate of CH₄/H₂O reaction over Ir/GDC with respect to CH₄, and almost zero order with respect to H₂O [17].

4. Discussion

4.1. Mechanism and Active Sites for CH₄/H₂O Reaction over Ir/GDC

The reaction of a CH₄/H₂O/N₂ mixture above 650 °C over Ir/GDC catalysts involves many equilibrated reactions, as listed below:

CH_4(g) + 2 H₂O_(g) ⇆ CO_2(g) + 4 H_2(g)   ΔH°298 = +165 KJ/mol   (reverse methanation)

(12)

CH_4(g) + CO_2(g) ⇆ 2 CO_(g) + 2 H_2(g)   ΔH°298 = +247 KJ/mol   (dry methane reforming)

(13)

CO_(g) + H_2(g) ⇆ H₂O_(g) + C_(s)   ΔH°298 = −131 KJ/mol  (reverse gasification of carbon by steam)

(14)

CO_2(g) + 2 H_2(g) ⇆ 2 H₂O_(g) + C_(s)   ΔH°298 = −90 KJ/mol   (reverse gasification of carbon by steam)

(15)

2CO_(g) ⇆ C_(s) + CO_2(g)   ΔH°298 = −172 KJ/mol   (Boudouard reaction)

(16)

Only reactions 1, 2, and 12 are proposed to occur over Ir/GDC since no carbon species could be detected after reaction [31]. Reactions 1 and 12 require the activation of CH₄. From the literature, CH₄ activation can proceed through different pathways depending on the type of catalyst.

CH₄ activation was proposed to proceed through dissociative adsorption at the surface of metal particles, such as supported Rh or Ir particles [35,36,37]. CH₄ activation at the surface of oxides is still unclear.

Concerning the metal C–H activation approach, Wei and Iglesia found that reaction rates were proportional to CH₄ partial pressure but independent of CO₂ and H₂O pressures, which led to the conclusion of sole kinetic relevance of C–H bond activation steps [35,36]. Their data indicate that co-reactant (CO₂ or H₂O) activation and its kinetic coupling with CH₄ activation via the formation of chemisorbed carbon intermediates are fast steps and lead to Rh or Ir surfaces essentially uncovered by reactive intermediates. Particularly, any involvement of support in the activation of co-reactants was found not to be kinetically relevant.

In the case of ceria-based catalysts, the methane activation was also proposed as the first step of the reaction mechanism [34]. This step would involve the direct reaction between methane and ceria surface O species to form CO, H₂, and O vacancies in ceria lattice. The formation of O vacancies would be accompanied by the reduction of Ce⁴⁺ to Ce³⁺. The second step of the process would correspond to the regeneration of ceria-oxidized species by H₂O by insertion of O into O vacancies and reoxidation of Ce³⁺ to Ce⁴⁺ [33,34,38]. According to Otsuka et al. [34], on the basis of CH₄/CD₄ isotopic exchange reaction, the activation of CH₄ was proposed not to be the rate determining step (rds), but on the contrary, a metal-based mechanism. In addition, O diffusion in the ceria lattice would be not limiting. Instead, the recombination of H-adsorbed species originating from CH₄ dissociation into C and H species into H₂ and/or desorption of H₂ would be the slow step, this step being facilitated by the presence of Pt metal. O vacancies were proposed as the active sites for CH₄ dissociation. It was reported by other authors that the activity of ceria-based materials stems from oxygen-vacancy formation and their migration with reversible transition Ce⁴⁺/Ce³⁺ [39].

Considering the Ir/GDC catalyst and assuming the Otsuka mechanism, CH₄ activation would not be rds; the recombination and desorption of H adsorbed species into H₂ being rds. On the other hand, if we suppose that CH₄ activation operates on Ir metal without the contribution of the support as from Iglesia et al., the catalytic activity would depend only on Ir dispersion, being first order with respect to CH₄ partial pressure. In previous works [17], the first order rate dependence with respect to methane was confirmed by steady-state kinetic measurements. This result is again confirmed in the present study with transient experiments using dry methane. The rate of CO formation in the initial stage of the reaction between Ir/GDC and dry 0.5 mol% CH₄ is maximum and does not vary significantly (3.7 µmol/s.gcat at 750 °C). From steady-state CH₄/H₂O reaction at 750 °C, assuming first order rate dependence with respect to CH₄, the value of 1 µmol of consumed CH₄ per s and per gcat can be derived. These values (from isothermal steady-state and transient experiments) are close, taking into account the very different conditions in terms of CH₄ partial pressures (two orders of magnitude difference). The lower value obtained in CH₄/H₂O reaction could also be attributed to some inhibiting effect of H₂O on the reaction [17,40]. These data confirm that, in both experiments (steady-state CH₄/H₂O and transient dry CH₄ reactions), the CH₄ activation on Ir/GDC is rds, which excludes the mechanism proposed by Otsuka observations for CeO₂ and Pt/CeO₂. The “metal” mechanism could then be considered. According to Iglesia, this would exclude any involvement of the support in the reaction mechanism. The catalytic activity of Ir/GDC was found to be higher by two orders of magnitude than Ir/Al₂O₃ containing the same Ir content (0.1 wt%), which strongly suggests the participation of the support in the case of Ir/GDC [13]. Only variations of Ir dispersion are unlikely to explain such differences.

The transient experiments with dry CH₄ over Ir/GDC confirm the involvement of GDC support through reducible O species in the CH₄/H₂O reaction mechanism. The following dual-site (Ir and O ceria species) mechanism can be proposed:

CH₄ → C_ad + 2 H₂ (at Ir surface)

(17)

C_ad + O_s → CO + VÖ (at the interface between Ir and GDC).

(18)

CH₄ activates on Ir sites to form adsorbed C species and H₂ gas. C species react with surface mobile O species of the support close to Ir, liberating CO gas and forming an O vacancy (VÖ) in the ceria lattice. This would require a high mobility of O and a high dispersion of Ir. The high Ir dispersion in Ir/GDC is supported by the fact that Ir particles could not be detected by HRTEM. However, we recall here the works reported by Cheah et al. that show the formation of Ir metal nanoparticles with narrow-size distribution (2.5–6 nm, mean size of 4 nm in diameter) after a pretreatment of Ir/GDC in He at 900 °C [15]. It is proposed that the active site closely associates Ir surface atoms and neighbouring O mobile species at the ceria surface. Surface study using XPS analysis demonstrated the coexistence of Ir³⁺ and Ir⁴⁺ at the interface with GDC [15]. The role of ceria mobile O species would be to “clean up” Ir sites from decorating carbon formed by CH₄ dissociation. The observation that carbon accumulates in the transient experiments after 75% Ce⁴⁺ → Ce³⁺ reduction (Figure 8) agrees with this view. When oxygen availability becomes low at the GDC surface, carbon forms without further transformation into CO, thus accumulating on the Ir surface and possibly migrating to the support surface. The existence of specific active sites at the GDC surface with properties being modified by Ir is not totally excluded. It is well known that the addition of metal particles strongly improves the O mobility in ceria-based materials [41,42,43,44,45,46,47,48].

Addressing the role of H₂O in the mechanism, it is shown from the transient experiment that the reaction of H₂O with O vacancies is a very fast step, which allows for fast regeneration of mobile O species in GDC. This agrees with the almost zero order rate dependence with respect to H₂O obtained by steady-state kinetic measurements [17].

An interesting point of the transient reaction of dry CH₄ with Ir/GDC is the sequential production of CO₂/H₂O, CO₂/H_2, and CO/H₂ depending on the oxidation state of the GDC support. Figure 7 clearly shows that CO₂ in the very early stage of the reaction forms without H₂ release. H₂O is also formed as a product. This strongly suggests the occurrence of a step during which CH₄ is completely oxidized by surface ceria O species (reaction 4), which would require 4 O species per converted CH₄ molecule. This is possible on the fully oxidized GDC, that is after He treatment and during the early stage of the reaction process. As reaction 4 proceeds, it is expected that the concentration of surface mobile O species progressively decreases. As a result, only partially oxidized sites would form so that the reactions leading first to CO₂/H₂ (reaction 6) and then to CO/H₂ (reaction 7) would prevail. Reactions 6 and 7 would require, respectively, active sites with 2 O and 1 O species. The sequence involving successively sites with 4, 2, and 1 surface mobile O species is shown in Figure 12. Considering the CH₄/H₂O reaction over Ir/GDC and based on these considerations, the selectivity in CO and CO₂ will depend on the distribution of these corresponding active sites at the surface of the GDC support. In fact, the observation that CO mainly forms in a mixture of 50 mol% CH₄ and 5 mol% H₂O would suggest that sites with 1 O species prevail. A partial reduction of ceria surface is expected due to the very large excess of reductant.

All the surface reactions involving methane and water vapour, together with the successive steps describing the CH₄/H₂O reaction over Ir/GDC, are reported in Scheme 1. According to the mechanism proposed in Scheme 1, the first step of the CH₄/H₂O reaction consists in methane activation over Ir particles leading to H₂ and carbonaceous species formation, which in turn react with mobile surface O of ceria at the Ir/GDC interface and produce surface O vacancies together with CO or CO₂ (step 2). Step 3 consists of replenishing surface O species by reaction of vacancies with H₂O. It is important to notice that, for proper anode operation, H₂O can be supplied in low amounts as co-feed (gradual internal reforming mode). However, the reservoir of mobile O species in GDC might be sufficient to initiate the production of H₂ to a certain extent, allowing the electrochemistry to proceed and to provide the water vapour necessary for the catalytic cycle. It is then expected that using Ir/GDC as a layer on the top of the conventional cermet Ni/YSZ anode would thus allow proper operation of the cell on dry methane without carbon formation.

4.2. Influence of H₂ on the Reactivity of Ir/GDC with CH₄/H₂O

The influence of two different treatments in H₂ prior to catalytic testing in CH₄/H₂O was examined: treatment in pure H₂ for 2 h at 800 °C and Julich FZJ procedure (Table 2). For both treatments, the following observations can be drawn:

-

The appearance of an activation period before reaching steady-state activity,

-

A decrease in the reaction rate at stationary state with respect to that obtained after treatment in inert gas.

These effects depend on the reaction temperature: the higher the reaction temperature, the lower the influence of the H₂ treatment. H₂ treatment is known to create OH on reduced ceria surfaces. Surface OH species Ce(III)(OH)₂ are reported to form on ceria upon H₂-treatment with oxidised species [49]. OH groups on Ce3+ sites are highly stable, which inhibits the reoxidation process [50].

Transient experiments confirm:

-

Some effect on the Ce⁴⁺ → Ce³⁺ reduction. Full Ce⁴⁺→Ce³⁺ reduction is possible with CH₄ after He, whereas only 80% is reached in CH₄ after H₂.

-

A decrease of the CH₄ consumption rate at equivalent reduction extent after H₂ compared to the corresponding rate after He.

Considering the influence on the CH₄/H₂O reaction, only 30% loss of activity is observed after H₂ treatments. Minor impact is expected on the application for a potential use in SOFC.

5. Conclusions

A mechanistic study of catalytic steam reforming of CH₄ was undertaken over 0.1 wt% Ir/GDC using various techniques, such as steady-state rate measurements, transient responses to CH₄, or H₂O step changes in isothermal conditions. The methane steam-reforming reaction over Ir/GDC proceeds through a bifunctional mechanism. The active site can be regarded as an intimate association between the following:

• Sites at metallic Ir particles’ surface as active sites for the cracking of CH₄ into reactive C species.
• Reducible (Ce⁴⁺) sites at GDC surface in the close vicinity of well-dispersed Ir particles and responsible for a redox mechanism involving Ce⁴⁺/Ce³⁺ sites being reduced by reaction with reactive C into CO (or CO₂) and re-oxidized by H₂O, replenishing surface O mobile species.

The existence of sites I, II, and III, respectively, corresponding to four, two, or one reactive surface O species in the vicinity of Ir sites is proposed to explain the transient reactivity of Ir/GDC with dry methane to form CO₂, CO, H₂, and H₂O according to the complete and partial oxidation of methane. The formation of carbon occurs on site IV consisting of only O vacancies around the iridium, obtained after extensive Ce⁴⁺ → Ce³⁺ reduction.

The relative amounts of CO and CO₂ formed in the CH₄/H₂O reaction might be related to the distribution and density of O mobile species at the GDC surface, leading to the complete methane oxidation (fully oxidized surface) or to the partial oxidation of methane (partially reduced surface). In the presence of water, site IV cannot be formed, which explains the high resistance of the catalyst against carbon deposition.

Using Ir/GDC in a layer deposited on a conventional Ni/YSZ membrane in the anode would furthermore not require co-feeding of water vapour, since the ceria lattice can provide the oxygen necessary to prevent carbon formation during the startup sequence before regeneration by H₂O produced at the cermet.

H₂ pretreatment used in anode activation induces a moderate loss of catalytic activity in methane steam reforming at steady state due to the decrease of the oxygen mobility but does not influence the reaction mechanism. This small effect would not significantly impact the use of Ir/GDC as a reforming catalyst in SOFCs anodes directly fed with dry methane.