Reaction Intermediates and Side Products in the Methanation of CO

Dec 7, 2010 - Chem. C , 2011, 115 (4), pp 1361–1367 ... of surface formates plays only a minor role in the latter reaction, they rather act as spect...
25 downloads 0 Views 613KB Size
Download PDF
J. Phys. Chem. C 2011, 115, 1361–1367

1361

Reaction Intermediates and Side Products in the Methanation of CO and CO2 over Supported Ru Catalysts in H2-Rich Reformate Gases† Stephan Eckle,‡ Hans-Georg Anfang,§ and R. Ju¨rgen Behm*,‡ Institute of Surface Chemistry and Catalysis, Ulm UniVersity, D-89069 Ulm, Germany, and Su¨d-Chemie AG, R&D Energy & EnVironment, Waldheimer Strasse 13 D-83502 Bruckmu¨hl, Germany ReceiVed: August 26, 2010; ReVised Manuscript ReceiVed: NoVember 9, 2010

Aiming at a mechanistic understanding of the CO and CO2 methanation reaction over supported Ru catalysts and the underlying physical reasons, we have investigated the methanation of CO and CO2 over a Ru/zeolite and a Ru/Al2O3 catalyst, in idealized and CO2-rich reformate gases by in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) measurements, employing quantitative steady-state isotope transient kinetik analysis (SSITKA) techniques. On the basis of the correlation between COad band intensity/COad coverage, CH4,ad/HCOad/formate band intensity, and the CH4 formation rate under steady-state conditions, HCOad is unambiguously identified as reaction intermediate species in the dominant reaction pathway for CO methanation on the Ru/Al2O3 catalyst. On the Ru/zeolite such species could not be detected. CO2 methanation proceeds via dissociation to COad, which is subsequently methanated. Formation and decomposition of surface formates plays only a minor role in the latter reaction, they rather act as spectator species. 1. Introduction The methanation of CO has attracted increasing interest recently because of its potential as simple technique for CO removal from H2-rich feed gases for fuel cells produced by steam reforming of fossil fuels or biomass based fuels (“reformates”).1 Since these gases typically contain also considerable amounts of CO2, the reaction must be highly selective for CO methanation, with CO2 methanation essentially being inhibited; otherwise the losses of H2 would become intolerable.1,2 Therefore, the mechanisms of these reactions (CO methanation and selective methanation of CO in CO2-rich gas mixtures) as well as the reasons responsible for the high selectivity are of obvious industrial importance. The reaction pathways and mechanism of the reaction of CO or CO2 to CH4 as well as the active surface species are still under debate. The main question is whether the reaction starts with C-O bond breaking or with association of hydrogen and subsequent C-O band breaking. In earlier studies, it was proposed that CO dissociates in the first step, leading to active and inactive carbon species, where the former are stepwise methanated to CH, CH2, CH3, and finally CH4.3-9 In these studies, adsorbed CHx,ad species were proposed as reaction intermediates. Other studies, however, involving transient experiments, provided convincing evidence that the adsorbed CHx species, at least those detected by IR spectroscopy, represent side products rather than reaction intermediates.10-12 In a different concept, CO disproportionation was proposed as the initial step, followed by carbon hydrogenation to CH4 and CO2 reduction to CO via the reverse water gas shift (RWGS) reaction.3,4 More recent studies indicated that a formyl type (HCO) adspecies plays an important role in the CO methanation reaction, followed by C-O bond breaking and further hydrogenation.13,14 This species was identified also as intermediate †

Part of the “Alfons Baiker Festschrift”. * Author to whom correspondence should be addressed. E-mail: [email protected]. ‡ Ulm University. § Su¨d-Chemie AG.

in the dominant reaction pathway in density functional theory (DFT) studies by Inderwildi et al.15-17 The mechanism of the CO2 methanation reaction is similarly controversial.3,18-23 Following earlier proposals of direct CO2 methanation,3 it is nowadays generally accepted that COad is the main intermediate of the CO2 methanation.13,18,19,21-24 This COad species is subsequently hydrogenated via the mechanism for CO methanation as discussed above. Several groups proposed that CO2 first reacts to COad via the reverse water gas shift (RWGS) reaction, which then continues to react to CH4. It was suggested that the RWGS reaction proceeds via a formate intermediate;18,21-23 for CO2 conversion on oxide supported Au catalysts a redox-type mechanism was proposed for the RWGS reaction.25 In addition, dissociative CO2 adsorption to COad and Oad and subsequent reaction of COad to CH4 was considered as a third alternative for the methanation of CO2.19 It is important to realize also that almost all of the studies listed above focus on the hydrogenation of CO or CO2 under conditions relevant for methane formation from synthesis gas, at nearly stoichiometric conditions (CO:H2 ) 1:3-1:4). Only a few studies deal with reaction atmospheres with a high excess of hydrogen (CO:H2 ) 1:20-1:100), as is typical for the selective methanation reaction.6,11,14 The different reactant ratios may have considerable effects on the reaction behavior and the dominance of a specific reaction pathway. In that sense, studies performed at close stoichiometric CO:H2 ratios may lead to mechanistic conclusions that are not necessarily relevant for the reaction under conditions typical for the selective methanation in H2-rich reformate gases. This was the background of an ongoing study on mechanistic details of these reactions, which aims at a physical understanding of the CO and CO2 methanation reaction on supported Ru catalysts and the underlying reasons for their selectivity in CO2rich reformate gases under reaction conditions typical for this process (high H2 excess, atmospheric pressure).26,27 In the present paper, we focus on the identification and characterization of the active reaction intermediates of the CO and CO2 methanation reaction on a standard Ru/Al2O3 catalyst and a novel Ru/zeolite

10.1021/jp108106t  2011 American Chemical Society Published on Web 12/07/2010

1362

J. Phys. Chem. C, Vol. 115, No. 4, 2011

catalyst under these reaction conditions. This was studied in transient SSITKA (steady-state isotope transient kinetik analysis) type IR experiments, where after 1000 min of reaction in idealized CO or CO2 reformate, one educt was replaced by its isotope labeled (13CO, 13CO2) analog. Correlation of the buildup/decay of various surface species such as HCOad, CH4,ad, and COad upon the exchange step under otherwise constant reaction conditions with the steady-state activity of the catalysts allows us to distinguish between reaction intermediates in the dominant reaction pathway and stable or less active reaction side products (“spectator species”). 2. Experimental Section Catalyst Properties. A detailed description of the two catalysts investigated, a Ru/zeolite catalyst prepared by Su¨dChemie AG with 2.2 wt % Ru loading and a 5.0 wt % Ru/ Al2O3 catalyst, and of the experimental setup and procedures was given elsewhere.26 In short, BET specific surface areas of 100 m2 g-1 for the Ru/Al2O3 and of 410 m2 g-1 for the Ru/ zeolite catalyst, respectively, were determined by N2 adsorption. X-ray photoelectron spectroscopy (XPS) measurements (PHI 5800 ESCA system, monochromatized Al KR radiation), performed prior to the catalytic measurements and without any catalyst pretreatment, yielded a much lower Ru(2p) peak intensity, both absolute (∼5%) and relative to the O(1s) intensity (∼21%), on the Ru/zeolite catalyst than on the Ru/Al2O3 catalyst, indicating that the Ru particles are mainly located in the pores of the zeolite support rather than on the surface of the zeolite particles. Prior to the reaction, the Ru nanoparticles were covered by an oxide layer, as indicated by the intensity ratio of the Ru0(2p) and Run+(2p) peaks of 1:4. The Ru particle size and the active Ru surface area of the two catalysts were characterized by different techniques, including X-ray diffraction (XRD), transmission electron microscopy (TEM), hydrogen chemisorption, and pulsed hydrogen chemisorption, yielding particle sizes of around 3 nm for the Ru/Al2O3 catalyst and very small particles in the range of 1 nm or below (not detectable by XRD and TEM) for the Ru/zeolite catalyst.26 In Situ IR Spectroscopy Measurements. In situ diffuse reflectance FTIR spectroscopy (DRIFTS) measurements were performed using a commercial in situ reaction cell (Harricks, HV-DR2). The spectra were recorded in a Magna 6700 spectrometer (Thermo) equipped with a MCT narrow-band detector. The measurements were carried out with a gas flow of 41.6 NmL min-1; the reaction gas mixtures were prepared via mass flow controllers (Hastings HFC-202 and Bronkhorst F201C-FA-88 V). Prior to the experiments, the catalysts were heated up in a N2 stream to 150 °C, where the background spectra were recorded, and subsequently, the catalysts were heated within 10 min to the reaction temperature in the reaction gas atmosphere, similar to the procedure applied in the kinetic experiments reported earlier.26 The reaction measurements were performed at 190 °C reaction temperature, over 1000 min to achieve steady-state conditions in idealized reformate (0.6 kPa CO, 3 kPa N2, balance H2) or in CO2-rich reformate (15.5 kPa CO2, 3 kPa N2, balance H2). (The differences in reflectivity between background measurement at 150 °C and reaction measurements at 190 °C are marginal and can be neglected.) Subsequently, one reactant was replaced in the reaction mixture by its isotope labeled isotopomer in SSITKA like measurements (e.g., 12CO/13CO) (reaction time about 75 min). Around 30 mg of diluted catalyst (1:5 with R-Al2O3) were placed on top of a 70 mg of R-Al2O3 layer as catalyst bed. During the reaction experiments, 400 scans were coadded for one spectrum; after

Eckle et al. the isotope exchange, this was reduced to 20 scans per spectrum to yield a higher time resolution. The intensities were evaluated in Kubelka-Munk units, which are linearly related to the adsorbate concentration (for exceptions see ref 28).29 Background subtraction and normalization of the spectra were performed by subtracting spectra recorded in a flow of N2 directly after catalyst conditioning. For presentation the spectra are split into three frequency ranges, the regions of the OH (3900-3400 cm-1, top left) and CHx (3250-2800 cm-1, top right) stretch vibrations in the top panels, respectively, and the spectral range between 2200 and 1000 cm-1 including the CO stretch (2200-1900 cm-1) and the OCO bending (1600-800 cm-1) vibrations in the bottom panel. The spectra in the OH region are shown as raw data, since the significant change in the background intensity in this region with time did not allow us to determine appropriate background signals for the background subtraction. 3. Results and Discussion In a first set of experiments, reaction intermediates and side products of the CO methanation on the Ru/zeolite and Ru/Al2O3 catalyst, respectively, were investigated and characterized in idealized reformate (CO/H2/N2) by sequences of in situ DRIFTS measurements (Figure 1). The relative intensities of the respective surface species (HCO, CH4, and CO) are plotted as function of time in Figure 2. The characteristic bands arising on the Ru/zeolite and the Ru/Al2O3 catalyst from the interaction with the reaction atmosphere are evident from the spectra presented in Figure 1a,b, which were recorded during 1000 min on stream. Panels c and d show the resulting spectra of the first 40 min during the SSITKA measurement, where after 1000 min of reaction, one educt (12CO) was replaced by its isotope labeled analog (13CO). This procedure allowed us to distinguish between reaction intermediates and side products, since in the first case the signal of the respective surface species should decrease fast, and in the latter case, the respective signals will prevail while a new signal belonging to the 13C containing species will start to grow in. The assignments of the main bands (see lines) will be briefly summarized below, a detailed discussion was given in ref 26. During reaction in idealized reformate (Figure 1a,b) we find on both catalysts characteristic signals in the COad region at ∼2034 and 1969 cm-1 on the Ru/zeolite and at 2031 and 1920 cm-1 on the Ru/Al2O3 catalyst, respectively, which are generally attributed to CO adsorbed in a linear or in a bridged configuration on Ru.9,24,30-37 A signal at 2075 cm-1, only appearing on the Ru/zeolite, is usually attributed to CO adsorbed on oxidized Run+ or coadsorbed with Oad on Ru.30,35,38 Interestingly, we find no evidence for a band at 2075 cm-1 on the Ru/Al2O3 catalyst. A dynamic equilibrium of the adsorbed COad surface species is reached after 100 min on both catalysts. In the OH region, two peaks at 3740 and 3600 cm-1 appear on the Ru/zeolite catalyst, which we relate to silanol and isolated OHad groups on the support39,40 and which do not change during reaction. On the Ru/Al2O3 catalyst, the peak at 3735 cm-1 is accompanied by three additional OHad signals at 3670, 3650, and 3620 cm-1, which were attributed to OH groups on different Al2O3 crystal facets.39 With ongoing reaction, the intensity of the peaks at 3735 and 3670 cm-1, attributed to terminal and isolated OH groups, remains constant, whereas that of the other two signals decreases. The CHx spectral region shows the typical CHx,ad signals at 3017, 2956, 2928, and 2856 cm-1, arising from CH4, CH3,ad,

Methanation of CO and CO2

J. Phys. Chem. C, Vol. 115, No. 4, 2011 1363

Figure 1. DRIFT spectra recorded during 1000 min on stream at 190 °C (equilibration of surface species) over the Ru/zeolite (left panels) and the Ru/Al2O3 catalysts (right panels) in idealized reformate and subsequent change to 13CO containing atmosphere: (a), (b) 12CO containing reaction atmosphere; (c) and (d) 13CO containing reaction atmosphere. From bottom to top: (a) and (b) 0, 2, 3, 5, 7, 15, 45, 105, 195, 345, 495, 645, 795, 915 min; (c) and (d) 1000, 0.25, 0.5, 0.75, 1, 1.25, 1.5 1.75, 2, 2.25, 3, 5, 7, 10, 45 min.

Figure 2. (a) Relative intensities of (2) 12COad and (9) 12HCOad and ([) inverted relative intensity of 12HCOad (1 - I(H12COad)), (b) H13COad, and (1)13CH4,ad, during CO methanation upon changing from 12CO to 13CO containing atmosphere (data from Figure 1). (b) Intensity decay of (2) 12 COad and (9) H12COad on an expanded time scale. Lines: initial slopes of the intensity decay.

and the symmetric and asymmetric CH2,ad vibrations on the two catalysts.11,24,35 On both catalysts, these species (except for CH4) start to grow in after about 15 min of reaction and reach a steady-state situation after 340 min on the Ru/Al2O3 catalyst and after 1000 min on the Ru/zeolite catalyst. For the CH4 signal, steady state is reached much faster on both catalysts, after about 10 min. A shoulder at ∼2905 cm-1 observed on both catalysts was assigned to the C-H vibration of a surface formate.10,12 In the OCO region, bands at 1590, 1440, and 1390 cm-1, which appear after about 100 min on stream on the Ru/zeolite

catalyst, were assigned to surface formates (1590 and 1390 cm-1) and carbonates (1440 cm-1).10,12,33 On the Ru/Al2O3 catalyst, the corresponding bands start to grow in after 45 min of reaction at 1590, 1390, and 1370 cm-1 (additional formate vibration). The carbonate band is missing on this catalyst.26 An additional band at 1760 cm-1 is ascribed to adsorbed HCO species.14,41 While this band is clearly visible on the Ru/Al2O3 catalyst, it is not seen on the Ru/zeolite catalyst. On both catalysts, the bands in the OCO region (except for the HCO band) grow steadily and do not saturate over 1000 min.

1364

J. Phys. Chem. C, Vol. 115, No. 4, 2011

Sequences of DRIFT spectra, recorded after changing from CO to 13CO after 1000 min on stream on the Ru/zeolite and the Ru/Al2O3 catalysts, are shown in Figure 1c,d. For more clarity, we only show spectra recorded during the first 40 min after the exchange step, since the relevant changes of the adsorbed surface species occur on a short time scale (∼10 min). On the Ru/zeolite catalyst, the 12COad signals at 2075 and 2034 cm-1, belonging to linearly adsorbed COad on Run+ and on Ru0, respectively, vanish within the first three minutes after the substitution by 13CO, and the respective 13COad signals start to grow in at 1980 cm-1 (linearly adsorbed 13COad). The same happens on the Ru/Al2O3 catalyst with the respective 12COad signals at 2031 and 1920 cm-1 (linearly adsorbed COad and bridge bonded COad), which are replaced by signals at 1980 and 1875 cm-1 (linearly adsorbed and bridge bonded 13COad). The signal related to linear 13COad on Run+ on the Ru/zeolite catalyst, which is expected at ∼2040 cm-1, is probably obscured by the prominent 13CO gas phase signals at 2125 and 2072 cm-1. The 12CO signal at 1969 cm-1 (12COad in a bridged configuration on Ru) on the Ru/zeolite catalyst does not seem to change, and we do not observe a new signal at about 40 cm-1 lower wavenumber to grow in. The expected changes in the 1956 cm-1 band are obscured by the evolution of the prominent band at 1980 cm-1, and the same is expected also for the weak shoulder expected at ∼1940 cm-1. Hardly any changes are observed in the CHx region, except for the anticipated replacement of the 12CH4 signal at 3017 cm-1 by a 13CH4 band at ∼3005 cm-1 42 within the first three minutes, comparable to the changes in the CO region. Since the other CHx surface species do not change, at least not on this time scale, they do not seem to be involved in the main reaction pathway to CH4.10,11,26 It may, however, well be that significant amounts of inactive CHx species (e.g., CHx chains, etc.) obscure the signals of active surface CHx species. In the latter case it would hardly be possible to detect the active CHx surface species and separate it from inactive species. The OCO region shows almost no changes for the Ru/zeolite and the Ru/Al2O3 catalyst over the reaction time covered in these spectra (40 min). The formate and carbonate bands do not change in frequency, and also their intensity remains about constant. At longer reaction times (after 100 min, not shown), new formate bands related to the corresponding 13C species appear in addition to those at 1590 cm-1 (Ru/zeolite, Ru/Al2O3), 1390 and 1370 cm-1 (Ru/Al2O3), indicating that these formate species are side products in the reaction rather than reaction intermediates, as already stated earlier.26 Only for the signal at 1760 cm-1 on the Ru/Al2O3 catalyst, attributed to an HCOad species,14 the situation is different. Here, a new signal develops at a lower wavenumber at ∼1730 cm-1, while the signal at 1760 cm-1 decreases in intensity. The decrease of the HCOad signal at 1760 cm-1 agrees quantitatively with the increase of the new signal at 1730 cm-1. Furthermore, the decrease/increase of these signals with time agrees qualitatively with that of the linearly adsorbed 13COad/12COad species at 1980 and 2031 cm-1 and the 13CH4,ad/12CH4,ad products at 3005 and 3017 cm-1. This is illustrated in Figure 2a, which shows the relative intensities of the different surface species (12COad, 13CH4,ad, (inverted) 12HCOad and 13HCOad) as a function of time. Since the rates of depletion of the 12CH4,ad and 12COad signals agree quantitatively with the rate of increase of the 13 CH4,ad and 13COad signals, respectively, we only show the evolution of the 12COad, 13CH4,ad, (inverted) 12HCOad, and 13 HCOad signals for clarity. To illustrate the quantitative agreement between the intensity loss of the 12C isotopomer and 12

Eckle et al. the intensity growth of the 13C isotopomer, we include the evolution of the H13COad surface species and the inverted intensity of that species (1 - I(H12CO)), together with the intensity decay of the H12CO species, with time in Figure 2a. These plots demonstrate that the intensity change upon 12CO/ 13 CO exchange is fastest for the COad related signals, while that of the HCOad related signals are somewhat slower, though on a similar time scale. In a reaction mechanism where the reaction starts via adsorption of CO, this can be explained by the participation of two competing processes for the 12COad depletion under reaction conditions, desorption of CO and replacement by adsorbing 13CO and reaction with hydrogen to HCOad, followed by further reaction to CH4 and finally desorption. For the adsorbed formyl species (HCOad), removal is only possible by reaction with Had to CH4, and therefore slower. The above results indicate that COad desorption is faster than its further reaction to HCOad and that reaction of COad with hydrogen results in HCOad species. Though this would agree with a mechanism where HCOad acts as a reaction intermediate, it does not provide final proof for that assignment, at least not for the identification as reaction intermediate in the rate determining reaction pathway. In the latter case, the rate for HCOad conversion under steady-state reaction conditions should be identical to the reaction rate, i.e., the rate for CH4 formation under these conditions. This can be tested by determining the initial rate of the H12CO intensity decay (see Figure 2b). From the initial slope of the intensity-time curve in Figure 2b, we obtain a rate for the HCOad exchange of dI/dt ) 0.012 Iss s-1, where Iss denotes the intensity under steady-state conditions. For determining the actual reaction rate, we have to know the dispersion (fraction of active Ru surface atoms of all Ru atoms), the correlation between HCOad signal intensity and HCOad coverage, and the HCOad coverage under steady-state conditions. We assume that the intensity is proportional to the HCOad coverage, which seems to be justified in the limits of low steadystate coverages; the dispersion of the Ru/Al2O3 catalysts was determined to be 15%.26 For the steady-state HCOad coverage, we estimate a value of 0.065 monolayers (ML). This value, which would be obtained for similar cross sections of the C-O stretch vibration in HCOad and COad, can certainly only serve as a rough estimate. On the other hand, since the purpose of this evaluation is to check whether the HCOad exchange rate and the reaction (CH4 formation) rate are of similar order of magnitude, and since the coverage goes linearly into the rate, deviations by small factors are tolerable. With these data, we calculate a HCOad conversion rate of 1.2 × 10-6 mol gRu s-1, which is rather close to the reaction rate of 3.6 × 10-6 mol gRu s-1 determined for CH4 formation.26 For comparison, we also evaluated the rate for the initial intensity decay of the 12COad related signal (Figure 2b). It yields a rate of dI/dt ) 0.015 Iss s-1. Assuming that under steadystate conditions (pCO ) 0.6 kPa, 190 °C) the COad coverage is close to saturation including also the HCOad coverage (this can be derived from the adsorption isotherm, using adsorption characteristics typical for CO/Ru) and that the intensity is proportional to the COad coverage; this corresponds to a COad exchange rate of 1.5 × 10-5 mol gRu s-1. The difference in rate compared to HCOad exchange results mainly from the much higher steady-state coverage of COad compared to HCOad. As discussed above, the much higher rate for COad exchange compared to HCOad exchange and CH4 formation is mainly due to the additional fast equilibrium for CO adsorption/desorption. These results unambiguously prove that adsorbed formyl species indeed act as reaction intermediates in the rate determin-

Methanation of CO and CO2

J. Phys. Chem. C, Vol. 115, No. 4, 2011 1365

Figure 3. DRIFT spectra recorded during 1000 min on stream at 190 °C (equilibration of surface species) over the Ru/zeolite (left panels) and the Ru/Al2O3 catalysts (right panels) in CO2-rich reformate and subsequent change to 13CO2 containing atmosphere: (a), (b) 12CO2 containing reaction atmosphere; (c) and (d) 13CO2 containing reaction atmosphere. From bottom to top: (a) and (b) 0, 2, 3, 5, 7, 15, 45, 105, 195, 345, 495, 645, 795, 915 min; (c) and (d) 1000, 1, 2, 3, 5, 6, 7, 10, 15, 20. 25, 30, 35, 45 min.

ing majority reaction pathway for CO methanation over a Ru/ Al2O3 catalyst under present reaction conditions. Its direct spectroscopic identification was possible because of its significant coverage under present steady-state reaction conditions. This indicates that the rate constants for HCOad formation and its further reaction are of similar order of magnitude. If the latter were much higher, the steady-state HCOad coverage would be correspondingly lower and vanish below the detection limit of these measurements. These observations are backed by other studies, where the role of the formyl species was discussed and emphasized as well. Mitchell et al. found that on Ru(001) and at 100 K a HCOad species was formed during reaction of COad with atomic hydrogen.41 Their study, however, did not allow any conclusions on whether this species indeed acts as reaction intermediate in the CO methanation reaction under typical reaction conditions. Investigating the CO methanation on different Ni surfaces experimentally and theoretically over a wide (partial) pressure range, Andersson et al. showed that depending on the Ni sites, different surface species are favorable (HCOad or COHad).14 In all cases, however, H-assisted CO dissociation was always more facile than direct CO dissociation. Comparable results were obtained by Inderwildi et al., who using DFT techniques calculated that on Ru surfaces the methanation reaction proceeds via a H-assisted CO dissociation; reaction via the formyl mechanism was found to be more favorable than a possible CHx mechanism, where CO is dissociated right in the first step.16

Interestingly, we do not observe such a formyl species on the Ru/zeolite catalyst. While on a first view this may point to a different reaction mechanism, we think that the apparent absence of this species is related to the much higher activity of this catalyst, which is, depending on the reaction conditions, about 10-30 times higher than that of the Ru/Al2O3 catalyst.26 If the increase of the rate constant for formyl off-reaction (decomposition) on the Ru/zeolite catalyst, relative to the Ru/ Al2O3 catalyst, is more pronounced than that for formyl formation, the steady-state coverage of HCOad will be correspondingly lower on the Ru/zeolite catalyst. In that case, the signal of the formyl species may well be below the detection limit of the DRIFTS measurement. The stability of the 12C-formate related bands and the additional slow appearance of 13C-formate related bands on both catalysts clearly demonstrates that these surface formate species represent stable side products in the methanation reaction under present reaction conditions rather than reaction intermediates, in agreement with earlier reports.10-12 In a second, similar set of experiments we investigated possible intermediates and side products of the CO2 methanation reaction on the Ru/zeolite and a Ru/Al2O3 catalyst in CO2-rich reformate (CO2/H2/N2). Figure 3 shows a sequence of DRIFT spectra recorded during the CO2 methanation reaction and the subsequent SSITKA experiment on the Ru/zeolite (a and c) and on the Ru/Al2O3 (b and d) catalyst. During 1000 min of reaction, similar bands appear on both catalysts as during CO methanation

1366

J. Phys. Chem. C, Vol. 115, No. 4, 2011

(see above). In the following, we will only describe the differences to the signals described in the above section. In the OH region, CO2-related overtone signals (not marked) appear at 3450, 3540, 3600, 3640, 3700, and 3740 cm-1, in addition to the OH-related signals in this region, which blur the OH signals. The CO region only exhibits two weak signals at ∼2000 and 1865 cm-1 on the Ru/zeolite catalyst, commonly attributed to linearly adsorbed COad on Ru and COad in a bridged configuration. They appear after 45 min (Figure 3a) and reach a steadystate situation after about 100 min. The steady-state intensity is about 1/16th of that obtained during CO methanation (Figure 1) (see also ref 27). On the Ru/Al2O3 catalyst, in contrast, prominent COad signals at 2020 cm-1 (linearly COad) and 1890 cm-1 (bridge bonded COad) appear during the first minutes of reaction and increase, until reaching steady-state conditions after 100 min at an intensity which is comparable to that obtained during CO methanation (Figure 1b). The much lower intensity of the COad related bands on the Ru/zeolite catalyst compared to the Ru/Al2O3 catalyst agrees perfectly with the much higher CO2 methanation activity of the latter catalyst under similar reaction conditions reported previously.26,27 The slight deviation of the COad peak positions during CO2 methanation compared to those observed in the CO methanation experiments discussed above is attributed to the different COad coverages in both cases.43 Main difference in the OCO region compared to the CO methanation experiments described above (Figure 1a,b) are the much more pronounced formate bands at 1590, 1390, and 1370 cm-1 on both catalysts in the presence of CO2, indicating its strong effect on the growth/coverage of surface formates. Similar observations were reported and discussed previously.26,27 Also for reaction of CO2, HCOad surface species are formed on the Ru/Al2O3 catalyst, as indicated by the small peak at 1760 cm-1. Comparable differences are observed in the CH region, which reveals the typical signals at 3017, 2960, 2928, ∼2900, 2860 cm-1 as discussed above on both catalysts, but a much more pronounced band at ∼2900 cm-1 attributed to a C-H vibration of the formate. Changing from 12CO2 to 13CO2 (Figure 3c,d) after 1000 min on stream causes the following main changes in the spectra: • The COad signals at ∼2000 cm-1 on the Ru/zeolite catalyst and at 2020 and 1970 cm-1 on the Ru/Al2O3 catalyst are replaced by bands at 1970, 1895, and 1860 cm-1, respectively, within the first 3 min. • At the same time, also the 12CH4 signal decreases and disappears and the 13CH4 related signal grows in and saturates. • On the Ru/Al2O3 catalyst, the band at 1760 cm-1 related to HCOad species decreases and is replaced by a signal at 1730 cm-1, observable as a small shoulder within 3 min. • The formate bands at 1590, 1390, and 1370 cm-1 decrease only very little during that time, while new bands of a 13C containing formate at 1550, 1350, and 1320 cm-1 grow in and increase rapidly in intensity. Furthermore, a new signal at 2880 cm-1 develops, which is attributed to the C-H vibration of the 13C surface formate signal. The 12C signal is still present as a shoulder of the 13C signal. A quantitative evaluation shows that also in this case the loss of 12C related intensities of COad is identical to the gain in 13Crelated CO intensity. Furthermore, the quantitative agreement between the 12COad/13COad and the 12CH4,ad/13CH4,ad exchange characteristics confirms the earlier proposals that COad acts as reaction intermediate in the main reaction pathway for CO2

Eckle et al. methanation on Ru catalysts.13,18 For the Ru/Al2O3 catalyst, comparison of the peak intensity of COad derived from CO2 with that obtained in idealized reformate shows that CO2 dissociation leads to a COad coverage close to saturation. This explains the high CO2 methanation activity of the Ru/Al2O3 catalyst reported previously.26,27 Moreover, the decomposition rate of 12HCOad on the Ru/Al2O3 catalyst (9 × 10-7 mol gRu s-1) is of magnitude similar to that calculated above for CO methanation, which is clear proof that the reaction pathway via formation of COad and subsequent formation/off-reaction of HCOad to CH4 is the dominant reaction pathway for CO2 methanation. In addition, these observations perfectly agree with the mechanism responsible for the high CO selectivity of the Ru/zeolite catalyst during the methanation of reformate gases containing large amounts of CO2 in addition to CO, which we had proposed previously.27 In that paper we postulated that the high selectivity of the Ru/zeolite catalyst is mainly driven by its inherent low activity for CO2 dissociation. In contrast, on the Ru/Al2O3 catalyst, where CO2 is readily dissociated to COad on free active sites, the CO selectivity is determined by the abundance of free active sites. In that case, high selectivities are only reached for sufficiently high CO partial pressures (low coverage of free active sites) under steady-state conditions. Otherwise, CO2 is readily dissociated to COad on free active sites, leading to a high CO2 methanation rate at low CO partial pressures. As already discussed in the Introduction, the mechanism for COad formation during CO2 methanation over group VIII metal supported catalysts is still under debate.13,18-23 Mostly it is assumed that COad is produced via the reverse water gas shift reaction (RWGS), involving either the formation and subsequent decomposition of surface formate species18,21-23 or a redox mechanism.25 As a third alternative, dissociative adsorption of CO2 was discussed.19 Under present reaction conditions and for the Ru catalyst investigated, a redox mechanism can be excluded because of the nonreducible support material. Also the formate mechanism is highly unlikely, though it should be favored by the high H2 partial pressure. It would require a similarly rapid decrease of the formate related bands, at least of those attributed to one surface formate species, upon changing from 12CO2 to 13 CO2. This is in contrast to the experimental observations, which show a very slow decrease of the H12COO related bands only, in contrast to the rapid exchange of the COad related bands. Instead, bands related to 13C-containing formate species grow in additionally, indicating that the (total) surface formate coverage has not reached steady-state conditions after 1000 min on stream. Hence, under present experimental conditions, COad formation via formate formation/decomposition cannot represent the dominant reaction pathway, although contributions from this reaction pathway as minority pathway cannot be ruled out. In consequence, we have no evidence for COad formation via the RWGS reaction in the CO2 methanation reaction, at least not in the main reaction pathway, and therefore propose that under present reaction conditions (high H2 excess, atmospheric pressure) CO2 methanation on Ru catalysts proceeds via dissociative adsorption of CO2 to form COad and Oad, which subsequently reacts to CH4 and H2O. As the results and observations described above hold true for both catalysts, CO2 adsorbs dissociatively on the Ru/Al2O3 and on the Ru/zeolite catalyst. 4. Conclusion On the basis of time-resolved DRIFTS measurements during the methanation of CO and CO2 over two supported Ru catalyst, a Ru/zeolite catalyst, and a Ru/Al2O3 catalyst, and employing

Methanation of CO and CO2 SSITKA type exchange techniques, we could unambiguously identify adsorbed formyl species with a characteristic band at 1760 cm-1 as the reaction intermediate for CO methanation in a CO/H2/N2 mixture on the Ru/Al2O3 catalyst under reaction conditions typical for reformate purification. The identification as reaction intermediate rather than a spectator type reaction side product is based on its formation/reaction rate under steadystate conditions, which is of similar order of magnitude as the overall ration rate (CH4 formation rate). For the Ru/zeolite catalyst, we tentatively propose the same reaction mechanism. The absence of a similar band on the Ru/zeolite catalyst is attributed to a much lower steady-state coverage of that species, which is rationalized by the much higher reaction rate on that catalyst. Similar experiments performed for the CO2 methanation reaction revealed that under the present reaction conditions the reaction proceeds via formation and further reaction of COad. This, however, cannot be formed via a formate based reverse water gas shift reaction, at least not as the main reaction pathway, since the decomposition of surface formate species is too slow compared to the COad exchange rate. Since a redox RWGS mechanism involving reaction with the catalyst is equally unlikely for the present catalysts, we suggest that under present reaction conditions COad is formed via dissociative adsorption of CO2 to form COad and Oad, which subsequently reacts to CH4 and H2O. References and Notes (1) Rostrup-Nielsen, J. R.; Aasberg-Petersen, K. Fuel Cell technology and applications, 1st ed.; John Wiley and Sons: Chichester, U.K., 2003; Chapter 14, p 159. (2) Echigo, M.; Tabata, T. J. Chem. Eng. Jpn. 2004, 37, 75. (3) Mills, G. A.; Steffgen, F. W. Catal. ReV. Sci. Eng. 1974, 8, 159. (4) Nijs, H. H.; Jacobs, P. A. J. Catal. 1980, 66, 401. (5) Goodman, D. W.; Kelley, R. D.; Madey, T. E.; Yates, J. T. J. Catal. 1980, 63, 226. (6) Bonzel, H. P.; Krebs, H. J. Surf. Sci. 1980, 91, 499. (7) Bell, A. T. Cat. ReV. Sci. Eng. 1981, 23, 203. (8) Kelley, R. D.; Goodman, D. W. Surf. Sci. 1982, 123, L743–L749. (9) Gupta, N. M.; Londhe, V. P.; Kamble, V. S. J. Catal. 1997, 169, 423. (10) Dalla Betta, R. A.; Shelef, M. J. Catal. 1977, 48, 111. (11) Ekerdt, J. G.; Bell, A. T. J. Catal. 1979, 58, 170. (12) Gupta, N. M.; Kamble, V. S.; Kartha, V. B.; Iyer, R. M.; Thampi, K. R.; Gratzel, M. J. Catal. 1994, 146, 173. (13) Fisher, I. A.; Bell, A. T. J. Catal. 1996, 162, 54.

J. Phys. Chem. C, Vol. 115, No. 4, 2011 1367 (14) Andersson, M. P.; Abild-Pedersen, F.; Remediakis, I. N.; Bligaard, T.; Jones, G.; Engbæk, J.; Lytken, O.; Horch, S.; Nielsen, J. H.; Sehested, J.; Rostrup-Nielsen, J. R.; Nørskov, J. K.; Chorkendorff, I. J. Catal. 2008, 255, 6. (15) Inderwildi, O. R.; Jenkins, S. J.; King, D. A. J. Phys. Chem. C 2008, 112, 1305. (16) Inderwildi, O.; Jenkins, P. Chem. Soc. ReV. 2008, 37, 2274. (17) Inderwildi, O.; Jenkins., J. S.; King, D. Angew. Chem., Int. Ed. 2008, 47, 5253. (18) Prairie, M. R.; Renken, A.; Highfield, J. G.; Thampi, K. R.; Gra¨tzel, M. J. Catal. 1991, 129, 130. (19) Liotta, L. F.; Martin, G. A.; Deganello, G. J. Catal. 1996, 164, 322. (20) Traa, Y.; Weitkamp, J. Chem. Eng. Technol. 1999, 21, 291. (21) Yaccato, K.; Carhart, R.; Hagemeyer, A.; Lesik, A.; Strasser, P.; Volpe, A. F., Jr.; Turner, H.; Weinberg, H.; Graselli, R. K.; Brooks, C. Appl. Catal., A 2005, 296, 30. (22) Park, J.-N.; McFarland, E. W. J. Catal. 2009, 266, 92. (23) Kim, H. Y.; Lee, H. M.; Park, J.-N. J. Phys. Chem. 2010, 114, 7128. (24) Scire`, S.; Crisafulli, C.; Maggiore, R.; Minico, S.; Galvagno, S. Catal. Lett. 1998, 51, 41. (25) Burch, R. Phys. Chem. Chem. Phys. 2006, 8, 5483. (26) Eckle, S.; Denkwitz, Y.; Behm, R. J. J. Catal. 2010, 269, 255. (27) Eckle, S.; Anfang, H.-G.; Behm, R. J. Appl. Catal., A DOI: 10.1016/ j.apcata.2010.07.025. (28) Hamadeh, I. M.; Griffiths, P. R. Appl. Spectrosc. 1987, 41, 682. (29) Meunier, F.; Reid, D.; Goguet, A.; Shekhtman, S.; Hardacre, C.; Burch, R.; Deng, W.; Flytzani-Stephanopoulos, M. J. Catal. 2007, 247, 269. (30) Gonzalez, R. D.; Brown, M. F. J. Phys. Chem. 1976, 80, 1731. (31) Davydov, A. A.; Bell, A. T. J. Catal. 1977, 49, 332. (32) Solymosi, F.; Rasko´, J. J. Catal. 1989, 15, 107. (33) Guglielminotti, E.; Bond, G. C. J. Chem. Soc., Faraday Trans. 1 1990, 86, 979. (34) Gupta, N. M.; Kamble, V. S.; Iyer, R. M.; Ravindranathan, T.; Gra¨tzel, M. J. Catal. 1992, 137, 473. (35) McQuire, M. W.; Rochester, C. H. J. Catal. 1993, 141, 355. (36) Yokota, K.; Fukui, M.; Tanaka, T. Appl. Surf. Sci. 1997, 121/122, 273. (37) Todorova, S. ZH.; Kadinov, G. B. Res. Chem. Intermed. 2002, 28, 291. (38) Solymosi, F.; Erdo¨helyi, A.; Bansagi, T. J. Catal. 1981, 68, 371. (39) Kno¨zinger, H.; Ratnasamy, P. Catal. ReV. 1978, 17, 31. (40) Yong, S. T.; Hidajat, K.; Kawi, S. J. Power Sources 2004, 131, 91. (41) Mitchell, W. J.; Wang, Y.; Xie, J.; Weinberg, W. H. J. Am. Chem. Soc. 1993, 115, 4381. (42) Gupta, N. M.; Kamble, V. S.; Iyer, R. M.; Ravindranathan Thampi, K.; Gra¨tzel, M. Catal. Lett. 1993, 21, 245. (43) Pfnu¨r, H.; Menzel, D.; Hoffmann, F. M.; Ortega, A.; Bradshaw, A. M. Surf. Sci. 1980, 93, 431.

JP108106T