Catalysis Science & Technology

2 downloads 0 Views 3MB Size Report
d Energy Research Institute, Tianneng Group, Huaxi Industrial Function Zone,. Changxing ...... 6 H. Yang, Z. Xu, M. Fan, R. Gupta, R. B. Slimane, A. E. Bland.

Catalysis Science & Technology View Article Online

Published on 21 April 2016. Downloaded by Nanyang Technological University on 09/05/2016 14:49:56.

MINIREVIEW

View Journal

Catalysis mechanisms of CO2 and CO methanation Cite this: DOI: 10.1039/c6cy00478d

Bin Miao,a Su Su Khine Ma,a Xin Wang,b Haibin Su*cd and Siew Hwa Chan*ae Understanding the reaction mechanisms of CO2 and CO methanation processes is critical towards the suc-

Received 3rd March 2016, Accepted 20th April 2016

cessful development of heterogeneous catalysts with better activity, selectivity, and stability. This review provides detailed mechanisms of methanation processes and undesired catalyst deactivation. We characterize the methanation processes into two categories: (1) associative scheme, in which hydrogen atoms are

DOI: 10.1039/c6cy00478d

involved in the C–O bond breaking process, and (2) dissociative scheme, where C–O bond breaking takes place before hydrogenation. For the catalyst deactivation mechanisms, we highlight three important fac-

www.rsc.org/catalysis

tors, i.e. sulfur poisoning, carbon deposition and metal sintering.

1. Introduction Renewable energy resources are playing and will continue to play a vital role in meeting current and future energy needs. They are also key components in the sustainable development of transportation and electricity generation sectors.1,2 However, there are some issues including high cost, intermittency of energy supply, grid connection and storage challenges that should be addressed so that the penetration of renewable energy in the energy market would be significant.2 Small output renewable energy sources could easily be balanced within the energy network, but the large, incrementing percentage of global renewable power highlights the substantial need for energy storage. For example, the installation of wind energy in Asia, Europe, and the USA in 2014 was 26.0, 12.9, and 4.9 GW, which led the accumulation of wind energy installation in these regions to 142.0, 134.0 and 65.9 GW, respectively.3 The above figures unveiled great potential and urgency for the development of energy storage solutions, especially longterm and grid-scale storage. Besides available energy storage techniques, pumped hydro and water electrolysis from Solid Oxide Electrolysis Cells (SOECs) can be considered to be gridscale energy storage solutions.4,5 However, the former highly relies on geography and environment, while the latter de-

a

Energy Research Institute at NTU ([email protected]), Nanyang Technological University, 50 Nanyang Avenue, 639798, Singapore. E-mail: [email protected] b School of Chemical and Biomedical Engineering, Nanyang Technological University, 50 Nanyang Avenue, 639798, Singapore c Division of Materials Science, Nanyang Technological University, 50 Nanyang Avenue, 639798, Singapore. E-mail: [email protected] d Energy Research Institute, Tianneng Group, Huaxi Industrial Function Zone, Changxing County, Zhejiang 313100, PR China e School of Mechanical & Aerospace Engineering, Nanyang Technological University, 50 Nanyang Avenue, 639798, Singapore

This journal is © The Royal Society of Chemistry 2016

mands hydrogen storage and transmission infrastructure to be established. Power-to-gas technology converts excess electricity to methane as an energy storage intermediate through the Sabatier reaction (4H2 + CO2 → CH4 + 2H2O), to fill in the gap between grid-scale surplus power and the existing natural gas infrastructure. The feedstocks for the methanation process are hydrogen produced from water electrolysis from SOECs and CO2 captured from flue gas, biomass or other carbon-containing resources.6 Power-to-gas, accompanied by carbon capture technology, could be the key solution in tackling the challenge of climate change caused by greenhouse gas emissions.7,8 For the economic feasibility aspect of powerto-gas plants, however, the cost of synthesized methane is several times higher than that of conventional natural gas.9 Most of the existing pilot plants globally have only been operated for a short time and further work is needed to improve their efficiencies.10 The economic competitiveness of power-to-gas plants counts on technical breakthroughs. Therefore, developing high activity, high selectivity, and durable catalysts is of great importance for implementing large-scale energy storage facilities. Great progress has been made on catalyst metal, promoter and support material advancements.11–13 Generally, Ni, Ru, Co, and Fe supported on Al2O3, SiO2, TiO2, ZrO2, and CeO2 are commonly used as methanation catalysts under various reaction conditions.12 Particularly, Ni and Ru were found to be very selective to CH4 production and were intensively studied.14,15 Promoter materials are added to provide auxiliary functions, such as sulfur-resistance, sintering-resistance and carbon-resistance properties.16–20 Nevertheless, in-depth discussion and systematic categorization of existing mechanisms have not been done so far. A good understanding of the mechanism of methanation reaction will provide useful guidance to achieve the desired catalyst properties, such as good activity, selectivity, and stability. This review discusses CO2 and CO

Catal. Sci. Technol.

View Article Online

Published on 21 April 2016. Downloaded by Nanyang Technological University on 09/05/2016 14:49:56.

Minireview

Catalysis Science & Technology

methanation mechanisms through both experimental and computational aspects. Focus is given to Ni- and Ru-based catalysts for their great commercialization potential. There are four main sections including the introduction and summary sections. The second section discusses the wellaccepted methanation mechanisms proposed from the 1970s up to the present according to their categories, i.e., associative methanation and dissociative methanation.12,21,22 The third section is devoted to catalyst deactivation mechanisms, which include sulfur poisoning, surface carbon deposition, and catalyst metal sintering mechanisms.

2. Methanation mechanisms The reaction mechanism is all about the elementary steps. Specifically, there are three obstacles to be overcome: (1) what are the intermediates? (2) What are the elementary steps, especially the rate-determining steps that lead to the intermediates? (3) What are the active sites? Despite the complexity of the elementary reaction steps, the overall reaction of CO2 and CO methanation processes is represented by the following reactions: CO2 + 4H2 → CH4 + 2H2O

CO + 3H2 → CH4 + H2O

ΔH298 = −165 kJ mol−1

ΔH298 = −206 kJ mol−1

Fig. 1 below (the schematic diagram does not show all the exact intermediates for clarity). 2.1. CO2 methanation mechanisms In the arena of CO2 methanation, two types of mechanisms have been proposed. One mechanism suggested that CO2 associatively adsorbed with adatom Had forming oxygenate intermediates and was subsequently hydrogenated to CH4. The other mechanism described that CO2 first dissociated to carbonyl (COad) and Oad, followed by carbonyl hydrogenation to CH4. The detailed discussion is presented in the following two sub-sections. CO2 associative methanation CO2 associative methanation involves the associative adsorption of CO2 and H2, followed by the hydrogenation of the associated species to form methane. Evidence from direct observation using the in situ infrared technique was found to support this mechanism. Aldana et al. studied CO2 methanation on the Ni–ceria–zirconia catalyst to reveal the reaction mechanism with in situ Fourier transform infrared spectroscopy (FTIR).25 The spectra of CO2 methanation at 150 °C showed carbonate (CO3ad) on the support and carbonyl (COad) on the Ni metal, as shown in Fig. 2(A), spectrum a). With increasing temperature, carbonate hydrogenated to bicarbonate (HCO3ad), and bicarbonate quickly dehydrated to formate (HCOOad). On the other hand, the carbonyl bands on Ni remain unchanged, as shown in Fig. 2(A), spectrum b)–f).

To know what exactly is happening during the reaction and further guide the catalyst design, the direct way is in situ observation with various spectroscopy techniques23 augmented by computational modelling24 of the elementary steps. The existing proposed methanation mechanism can be divided into two categories: (1) associative scheme and (2) dissociative scheme. A schematic diagram is illustrated in

Fig. 1 In the CO2/CO associative methanation scheme, the C–O bond breaking is assisted by adatom Had, while in the dissociative scheme, the C–O bond dissociated directly on the catalyst active sites.

Catal. Sci. Technol.

Fig. 2 (A): FTIR spectra for CO2 methanation on the Ni–ceria–zirconia catalyst (150 to 400 °C, H2 : CO2 = 4 : 1). The formation of bicarbonate and formate corresponds to the formation of CH4, whereas carbonyl is merely a spectator. Reaction temperature: a) 150 °C – 1 min, b) 150 °C – 30 min, c) 200 °C, d) 250 °C, e) 300 °C, f) 350 °C, g) 400 °C – 1 min, and h) 400 °C – 20 min. (B): Transient experiment (at 400 °C) on the Ni–ceria–zirconia catalyst showed immediately ceased formation of CH4 after H2 was turned off (solid line circle), and immediately ceased formation of CO after CO2 was turned off (dot circle line). Scheme (C-1): Methane produced from continuous hydrogenation of carbonate adsorbed on the support through formate and methoxy intermediates.25 Scheme (C-2): Carbon monoxide formation from CO2 reduction at Ce3+ sites.25,29 Reproduced with permission from ref. 25. Copyright 2013 Elsevier.

This journal is © The Royal Society of Chemistry 2016

View Article Online

Published on 21 April 2016. Downloaded by Nanyang Technological University on 09/05/2016 14:49:56.

Catalysis Science & Technology

The adatoms Had were believed to be provided by the Niparticles (not shown in the figure). Based upon the above observation, the authors proposed that the CH4 and CO in the venting gas were produced by different mechanisms.25 CH4 is produced from successive hydrogenation of formate species, while CO is produced from CO2 reduction at Ce3+ sites as a by-product, as shown in Fig. 2(B) (highlighted in solid and dashed circles, respectively). The overall reaction scheme is illustrated in Fig. 2(C-1): CH4 production from hydrogenation of carbonate and formate species, and Fig. 2(C-2): CO production from CO2 reduction at Ce3+ sites and adsorption on Ni. The connection between formate and CH4 formation can be found from infrared data and mass spectroscopy results. Schild et al. observed the depletion of the formate signal, leading to an increase in methane formation, from their study of CO2 methanation on the Ni–zirconia catalyst with in situ FTIR spectroscopy.26 They concluded that formate is the essential intermediate for methane production.26 Formate may have multiple functions. Westermann et al. regarded formate as the precursor of both CH4 and CO.27 They studied the CO2 methanation mechanism over Ni supported on ultra-stable Y-type (USY) zeolite with in situ IR spectroscopy. Under methanation conditions, carbonate hydrogenated to formate and adsorbed onto Ni-particles at a temperature lower than 200 °C. As the temperature increased to 300 °C and higher, formate dehydrated to carbonyl and further hydrogenated to CH4, or desorbed as CO. Pan et al. obtained similar results in their study of CO2 methanation on the Ni–ceria catalyst with in situ FTIR.28 The spectra showed five types of adsorption species, among which monodentate carbonate and bicarbonate hydrogenated to formate. Formate then evolved to CH4 and CO. An increase in CH4 and carbonyl bands at the expense of the formate bands was observed during the temperature-programmed study, indicating that formate is the major intermediate for methane production.28 On the aspect of the computational study, Pan et al. calculated the activation energy of the associative route (formate as an intermediate) and the dissociative route (carbonyl as an intermediate) of CO2 methanation on a Ni–alumina catalyst via DFT.30 They found that surface hydroxyl (OHad) altered the pathway, wherein the dissociative route became kinetically (Ea = 0.69 eV) and thermodynamically (energy release = 0.67 eV) favorable with the hydroxylated catalyst surface. When hydroxyl was not available, formate became the major intermediate for methane production (Ea = 1.25 eV).30 The associative mechanism is also applicable on Ru-based catalysts. Prairie et al. investigated CO2 methanation on the Ru–Al2O3 and Ru–TiO2 catalysts with DRIFT spectroscopy.31 The spectra showed formate (HCOOad) adsorbed on the support, and carbonyl (COad) on Ru and Ru-oxide sites. The carbonyl was produced from reverse water gas shift (RWGS) reaction of adsorbed CO2 through formate species. The ratelimiting step is thought to be the hydrogenation of carbonyl to CH4. Upham et al. came to this conclusion with transient experiments on the Ru–ceria catalyst.29 Specifically, by intro-

This journal is © The Royal Society of Chemistry 2016

Minireview

ducing CO2 before H2, methane and CO were produced. If the injection sequence was reversed, ceria was reduced by H2 first. Ce3+ then reduced CO2 to CO, but CH4 was not produced in this case. They proposed that methane is produced from associative hydrogenation of adsorbed CO2, while CO was produced from CO2 reduction over Ce3+ sites. On the aspect of computational calculation, Zhang et al. investigated CO2 hydrogenation pathways over the Ru(0001) surface.32 The results showed that direct CO2 dissociation was only thermodynamically favorable and needed to overcome a high energy barrier to occur (Ea = 1.20 eV). The feasible pathway for C–O bond breaking was the formate (HCOOad) route (Ea = 0.37 eV) which further dissociated to the formyl (CHOad) intermediate (Ea = 0.41 eV).

CO2 dissociative methanation It is possible that CO2 directly dissociated to carbonyl (COad) and Oad as intermediates during the methanation process. COad subsequently hydrogenated or further dissociated to Cad and Oad in the next step. In situ infrared techniques provide direct observation of the catalyst surface intermediates. Eckle et al. conducted a Steady-State Isotope Transient Kinetic Analysis (SSITKA) experiment to reveal the reaction intermediates during CO and CO2 methanation on Ru–Al2O3, under H2-rich atmospheric pressure conditions.33 The isotope exchange from 12CO2 to 13CO2 resulted in the reduction of the intensity of 12COad and the increase in the intensity of the 13 COad band, which indicated that carbonyl is the intermediate during CO2 methanation.33 On the other hand, the slow response of formate bands during isotope exchange excludes the formate species as the major intermediate, as shown in Fig. 3(A) and (B).33 The overall reaction schemes are illustrated in Fig. 3(C-1) and (C-2) below. The key findings in

Fig. 3 DRIFT spectra of CO2 methanation on Ru–Al2O3 (190 °C, H2 : CO2 = 5.3 : 1), switching from (A) 12CO2 reformate gas to (B) 13CO2 reformate gas. The carbonyl and formyl bands quickly switched to new bands. The formate signal retained its H12COOad band and accumulated new H13COOad bands. Scheme (C-1): CO2 dissociated to COad and Oad on Ru sites. COad then evolved to methane in an associative way. Scheme (C-2): Surface formate species accumulated rather than hydrogenated to CH4. Reproduced with permission from ref. 33. Copyright 2011 American Chemical Society.

Catal. Sci. Technol.

View Article Online

Published on 21 April 2016. Downloaded by Nanyang Technological University on 09/05/2016 14:49:56.

Minireview

SSITKA are consistent with the temperature-programmed desorption study. De Leitenburg et al. examined the effect of ceria support during CO2 methanation on a noble-metalbased catalyst.34 The high-temperature reduced ceria (Ce3+) promotes CO2 adsorption in the form of carbonyl (COad) and oxidized ceria (Ce4+). The ceria support provides the adsorbed carbonyl, while the metal particle provides the Had for the methanation process. Akamaru et al. obtained the same result using the DFT method in their study of CO2 methanation on Ru–TiO2.35 The calculation showed that adsorbed CO2 firstly dissociated to carbonyl and Oad on Ru–TiO2 (with energy barrier Ea = 0.6 eV), as has been observed by isotopic and infrared experiments.33 The subsequent methanation of carbonyl followed the hydrogen-assisted scheme with the formyl intermediate (CHOad, Ea = 0.95 eV).35

Discussion The discrepancies in reaction mechanisms between associative methanation and dissociative methanation are related to reaction conditions. The isotope exchange experiment in ref. 33, as displayed in Fig. 3(B), clearly showed the decay of carbonyl (COad) and the accumulation of formate (HCOOad) species during CO2 methanation, which supported the CO2 dissociative scheme.33 However, the reaction conditions are only limited to the low temperature (190 °C) and H2-rich environment (H2 : CO2 = 5.3 : 1). In a wider temperature range, Fujita et al. observed two methanation peaks during temperature-programmed reaction from 100 to 400 °C on the Ni–Al2O3 catalyst (with H2 : CO2 = 9 : 1).36 The first peak that appeared at 150 °C was attributed to the methanation of bridged-carbonyl (COad) adsorbed on Ni, and the second peak that appeared at 250 °C was attributed to bidentate-formate (HCOOad) methanation on the support.36 The high-temperature peak at 250 °C was consistent with the FTIR spectra from ref. 25 (Aldana et al.), as shown in Fig. 2(A).25 But Aldana et al. did not observe low-temperature methanation.25 This suggests that not only the temperature but also the H2 : CO2 ratio play a key role. CO2 dissociative methanation is promoted in a H2-rich environment. The H2-rich environment allows the prompt reduction of metal-oxide so that the CO2 redox process is able to proceed. Most authors, however, examined the stoichiometric (H2 : CO2 = 4 : 1) situation and observed the associative scheme.25–28 Further study on the H2 : CO2 ratio is necessary to reach a quantitative conclusion. For the effect of the support material, CO2 adsorption on CeO2 was much higher than that on Al2O3.37 This may contribute to the reverse water gas shift reaction at the metal– support interface and affect the mechanism observed.

2.2. CO methanation mechanisms In the area of CO methanation during the pre-1970s, two types of methanation mechanisms were proposed. One mechanism involved the combination of Had with COad to form

Catal. Sci. Technol.

Catalysis Science & Technology

COHad, CHOad or CHOHad intermediates, followed by C–O bond breaking. The other mechanism involved the direct dissociation of adsorbed COad, forming surface-carbon (Cad) as the methanation intermediate.38

CO associative methanation Many researchers claimed that direct C–O bond breaking was not kinetically favorable without the assistance of hydrogen.39 Thus, Had combined with carbonyl to form either formyl (CHOad) or carbon-hydroxyl (COHad) to promote C–O bond breaking, as illustrated in Fig. 4(C-1) and (C-2). Eckle et al. proposed the formyl (CHOad) route in their CO methanation on a Ru-based catalyst.33 SSITKA observed a change in the surface intermediate signal in the CO methanation process over Ru–zeolite and Ru–alumina catalysts. The DRIFT spectra showed a prompt COad signal change, followed by a CHOad signal change after the switch of the 12CO to the 13CO feed, indicating that the reaction sequence starts with CO adsorption then hydrogenation of COad to form CHOad, as seen from Fig. 4(A) and (B). The steady-state exchange rate of CHOad species was also found to be consistent with the CH4 formation rate.33 The build-up of formate and carbonate bands does not saturate throughout the experiment.40 On the aspect of Ni-based catalysts, Andersson et al. studied CO methanation on a Ni(111) crystal through experimental and computational methods.39 They found that direct CO dissociation on Ni(111) would not occur if the feed gas was properly cleaned to eliminate the Ni-carbonyl (NiIJCO)4) contaminant. The C–O bond was only breakable at the step and defect sites via the hydrogen-assisted route. The reaction intermediate was thought to be COHad species. Scanning Tunneling Microscopy (STM) measurement on Ni(111) observed carbon islands near the steps, whereas oxygen species were not detected. On the aspect of computational studies,

Fig. 4 DRIFT spectra of CO methanation on Ru–Al2O3 (190 °C, H2 : CO > 100), switching from (A) 12CO feed gas to (B) 13CO feed gas. The band switches indicate that carbonyl and formyl (HCOad) are the major intermediates in the CO methanation process. CO associative methanation to form Scheme (C-1): formyl33 or Scheme (C-2): carbon hydroxyl (not observed in the IR data).39 Reproduced with permission from ref. 33. Copyright 2011 American Chemical Society.

This journal is © The Royal Society of Chemistry 2016

View Article Online

Published on 21 April 2016. Downloaded by Nanyang Technological University on 09/05/2016 14:49:56.

Catalysis Science & Technology

for the Ni(111) surface, Wang et al. calculated the activation energy of the elementary steps of CO methanation and found that the Ea for COad to evolve to Cad, COHad and CHOad is 2.91, 1.93, and 1.70 eV, respectively.41 This result was consistent with the study of Zhi et al., in which CO prefers to adsorb at the catalyst hexagonal close-packed (hcp) site and is converted to CHOad as the dominant intermediate.42 Similarly, Fajín et al. studied CO methanation on a Ni(110) catalyst using the DFT method.43 Among all the routes, H-assisted CO dissociation through COHad (Ea = 1.42 eV) species and H-assisted CO dissociation through CHOad (Ea = 1.25 eV) species are the optimal schemes for C–O bondbreaking. On the supported catalyst, Wang et al. calculated the activation energy of each elementary step of CO methanation over Ni–Al2O3 using the DFT method.44 Their results showed that hydrogen-assisted C–O bond breaking is energetically favorable than direct CO dissociation. The calculated activation energy (Ea) for CO dissociation on Ni4 hollow sites was 3.19 eV, which was higher than the COad desorption barrier, Ea = 2.03 eV. Hydrogen-assisted dissociation at the same location, CHOad → HCad + Oad (Ea = 1.32 eV), was energetically achievable.

CO dissociative methanation CO dissociative methanation suggested that C–O bond breaking takes place directly at the active sites, before the successive hydrogenation steps. The other scheme for CO dissociation is through CO disproportionation. With the in situ FTIR technique, Panagiotopoulou et al. investigated the active sites of CO methanation over the Ru–TiO2 catalyst.21 The in situ FTIR spectra of temperature-programmed reaction showed linearly and bridged bonded carbonyls adsorbed on Ru, Ru oxide, and metal–support interface, as shown in Fig. 5(A) below. The Ru-oxide spectra indicated that CO firstly dissociatively adsorbed as Cad and Oad, followed by Cad hydrogenation. In their subsequent work, they confirmed that both the dissociative and associative methanation pathways occurred with different temperature and hydrogen content ranges. 22 At low temperature (200 °C), the dissociative scheme prevailed. In addition, at a higher temperature (350 °C) associative methanation dominated, as shown in Fig. 5(B)-1 and -2.22 On the aspect of computational studies, DFT calculation provided evidence of CO dissociation on the step sites by calculating the activation energy. Tison et al. conducted ultra-high vacuum (UHV) experiments and DFT calculation of CO dissociation over a Ru-crystal.45 STM under UHV conditions showed step decoration, which was assigned as evidence of CO dissociation at the step sites (4-fold hollow sites). The oxygen adatoms were observed at the terrace, where the 3-fold hollow sites are present. Their DFT calculation confirmed that CO dissociation over 4-fold hollow sites is energetically feasible. Vendelbo et al. conducted CO dissociation over Ru with a UHV apparatus and DFT calculation.46 The calculation results suggested that CO dissociation was preferable than desorption on Ru step sites.

This journal is © The Royal Society of Chemistry 2016

Minireview

Fig. 5 (A): FTIR spectra of CO temperature-programmed methanation on Ru–TiO 2 (25 to 450 °C, H2 : CO = 10 : 1). The spectra showed carbonyl adsorbed on Ru-oxide and Ru sites. The carbonyl-Run+ band was reduced as the temperature increased. Reproduced with permission from ref. 21. Copyright 2011 American Chemical Society. (B)-1 and (B)-2: FTIR spectra of CO methanation on Ru–TiO2 (200 °C, 350 °C, H2 : CO = 0 to 24 : 1). The carbonyl was observed on Ru-oxide and Ru sites. The metal oxide band was reduced with the increase of H2 partial pressure. (C): Illustration of the CO dissociative methanation scheme. Reproduced with permission from ref. 22. Copyright 2012 Elsevier.

Transient experiments were carried out to examine the CO methanation mechanisms. Araki and Ponec studied CO methanation over Ni films at temperatures of 250–350 °C.47 They observed a short induction period of CH4 formation after a CO and H2 mixture was introduced to the Ni film while CO2 was almost immediately produced. CO2 formation without an induction period can be observed even without H2 present.47 They concluded that CO2 formation should pass through a fast route, most probably via CO disproportionation, also known as the Boudouard reaction.47 An isotopetraced experiment is shown in Fig. 6(A) and (B), in which a H2 and 12CO mixture was introduced to a 13C-coated Ni-film at 250 °C. Mass spectroscopy detected immediate 13CH4 production followed by 12CH4 and 12CO2 formation, which clearly showed that H2 firstly combined with surface carbon to form 13CH4. 12CO2 formation indicated that CO dissociation was an easy process and does not necessarily require the assistance of hydrogen.47 This conclusion was supported by Madden and Ertl in their earlier study of CO decomposition on a Ni(110) surface,48 where CO started to dissociate to Cad and Oad at temperatures as low as 177 °C, and Oad was subsequently taken away by CO.48 Goodman et al. conducted a CO methanation kinetic study over Ni single crystals.14 Auger electron spectroscopy (AES) observation supported the pathway of surface carbide or CHx species as the major intermediate. The surface carbon is formed through CO disproportionation since the Oad element is not detected on the surface.14

Catal. Sci. Technol.

View Article Online

Published on 21 April 2016. Downloaded by Nanyang Technological University on 09/05/2016 14:49:56.

Minireview

Catalysis Science & Technology

blocks these sites.51 For instance, the existing possible mechanism of sulfur poisoning was described as sulfur preferentially adsorbing on the active site and hence blocking the reaction on that site.52 Adsorbed sulfur also affects the adjacent active site by weakening the electronic density, and hence altering the site activity towards certain products.52 Coking is a mechanical process, in which carbon filaments or whiskers physically (1) foul the metal, (2) block the pores, and (3) disintegrate the catalyst metal and support.53 The temperature has a great impact on the carbon morphology: (1) surface carbide is formed below 325 °C, (2) graphite carbon starts to form upon CO decomposition at 425 °C, and (3) filamentous and crystalline graphitic carbon forms above 550 °C.53 Fig. 6 (A): Mass spectroscopy of CO methanation on the Ni-film (250 °C, H2 : CO = 5 : 1): introduction of 12CO and H2 onto the 13Ccoated Ni-film produced 13CH4 gas followed by 12 CH4 and 12CO2, as illustrated in (B). 13CH4 and 12CH4 are formed from hydrogenation of adsorbed 13C ad and 12 Cad . 12CO2 is produced from 12CO disproportionation (Boudouard reaction). Reproduced with permission from ref. 47. Copyright 1976 Elsevier.

Discussion CO is more active than CO2 in the methanation process. On the Ni film, CO2 is almost instantly produced from CO disproportionation (H2 : CO = 5, T = 250 °C).47 Besides, the left over surface carbon is able to hydrogenate to CH4 quickly, as shown in the isotope-traced experiment in Fig. 6.47 In addition, the metal-oxide was observed as evidence for CO dissociative methanation on Ru–TiO2 at 200 °C with various H2 contents (H2 : CO = 0 to 24).22 On the other hand, CHOad was observed to participate on Ru–Al2O3 in the methanation process, which is strong evidence of the associative methanation scheme (H2 : CO > 100, T = 190 °C).39 The above evidence suggests that both associative and dissociative methanation contribute a certain portion to CH4 production under the respective reaction conditions. Since associative CO methanation was only observed at very high H2 content, it could be possible that the associative adsorption of CO and H2 followed the Eley–Rideal mechanism where H2 was inserted to bridged-COad directly. It's worth mentioning that only bridged-COad can be effectively hydrogenated while linearCOad retarded the methanation process.49 The dissociative route requires under-coordinated sites to break the C–O bond and a reductive environment (either H2- or CO-rich environment) to regenerate the active sites from the oxide state.

3. Deactivation mechanism Catalyst deactivation is a complex process that combines many mechanisms together, among which poisoning, coking, and sintering are great concerns in the methanation process.50 The poisoning effect is principally caused by the strong chemisorption of species on the activity sites which

Catal. Sci. Technol.

3.1. Sulfur poisoning The sulfur poisoning mechanism was widely studied in many areas.52 In the aspect of methanation, Legras et al. examined the sulfur poisoning mechanism on Ni–Al2O3 with SSITKA and in situ FTIR techniques.54 They introduced H2S to the catalyst after CO methanation. The sulfur species seem to inhibit CO adsorption without disturbing CH4 production. The authors proposed that carbonyl (COad) or Cad (generated from CO methanation) at the step sites protected the sites from sulfur poisoning,54 as illustrated in Fig. 7(C) below. In this case, S species adsorbed on terrace sites, inhibiting CO adsorption, whereas CH4 formation sites were not affected, as evidenced in Fig. 7(A) and (B). In the absence of methanation before S introduction, S species attached to the step sites easily and caused 10-fold catalyst deactivation.54 The same phenomena were observed on the Ru-based catalyst, where the carbon deposited catalyst showed better sulfur-resistance

Fig. 7 Introducing H2S to Ni–Al2O3 after CO methanation. (A): S species occupied the CO adsorption sites, which result in a quick decay of the 12CO signal in the isotope exchange experiment (SSITKA), (B): S species have a negligible effect on CH4 isotope exchange, which indicates that S species have not poisoned the CH4 formation sites, and (C): active carbon protects the step sites from being occupied by the sulfur species. Reproduced with permission from ref. 54. Copyright 2014 American Chemical Society.

This journal is © The Royal Society of Chemistry 2016

View Article Online

Published on 21 April 2016. Downloaded by Nanyang Technological University on 09/05/2016 14:49:56.

Catalysis Science & Technology

properties.55 Bartholomew et al. proposed the site competing model in their work.56 The higher H2 to CO ratio resulted in a lower carbon coverage and hence a stronger sulfur poisoning effect. Concerning the sites poisoned by sulfur species, defects and steps are favored compared to terraces.56 Rostrup-Nielsen and Pedersen examined the sulfur poisoning of the Boudouard reaction and methanation reaction on a Ni catalyst.57 They pre-sulfided the catalyst pellets at various temperatures to test sulfur coverage variations. They found that both the Boudouard reaction and methanation reaction shared the same reaction intermediate, most likely surface carbon.57 Moreover, the non-linear poisoning effect indicated that sulfur poisoned several nickel atoms at the same time.57 It was found that only 3% monolayer sulfur completely deactivated the Ni crystal catalyst in their CO dissociation test.58 The morphology of sulfur species varies. Erekson et al. examined the effect of H2S concentration.59 Their findings showed that the catalyst deactivation rate decreased with the H2S concentration, due to the formation of multilayer sulfide. In contrast, surface sulfide (2-D sulfide) that formed with ppb level H2S deactivates the catalyst more severely. The surface sulfide poisoning effect with ppb level H2S was studied by Fitzharris et al.60 over a Ni catalyst and by Agrawal et al.55 over a Ru catalyst. They found that surface sulfide had a lower energy barrier of formation and hence bonded to the catalyst surface strongly. Sulfur poisoning does not affect the activation energy of the methanation process, indicating that sulfur poisoning is a surface geometric effect rather than an electronic effect.55 Czekaj et al. calculated the stability of various sulfur species on the Ni catalyst and support via DFT.61 The result showed that many stable structures exist on the catalyst surface, among which carbonyl sulfide and hydrogen sulfide were the most stable species. Sulfide species not only existed on the supported Ni particle but were also attached to the support. The remaining sulfide species on the support after catalyst reactivation can cause re-poisoning of the catalyst.61 Struis et al. showed that the sulfur poisoning effect was not due to H2S only.62 The sulfur K-edge X-ray adsorption near edge structure indicated that an industrial methanation catalyst had been poisoned by thiophene (C4H4S) rather than H2S, as shown in Fig. 8 wherein bringing the specimen into contact with air may transform the poisoning species. To enhance the sulfur tolerance, Yuan et al. examined the sulfur tolerance of SiO2 supported Ni and Ni–Ru methanation catalysts using experimental and computational approaches.16 Their results showed that sulfur poisons the catalyst by blocking the active sites as well as enhancing the sintering and oxidation of Ni particles. Ru promoted the sulfur resistance by weakening the bond between S and Ni. Lee et al. studied the sulfur tolerance effect of Rh–Ni binary metal using the DFT method.63 The result showed that the sulfur adatom could increase the activation barrier for CO dissociation. The binary metal catalyst reduces the sulfur adsorption strength and hence resists sulfur poisoning. Yan et al. prepared a Ni cata-

This journal is © The Royal Society of Chemistry 2016

Minireview

Fig. 8 Surface scanning of a deactivated catalyst taken from a methanation reactor in Güssing (Austria) and of the poisoned one at 400 °C. (A): Without being in contact with air: the poisoning species on the deactivated catalyst are thiophene (C4H4S) and S instead of H2S. (B): After being in contact with air, the thiophene and S spectra shifted to H2S. Reproduced with permission from ref. 62. Copyright 2009 Elsevier.

lyst via a plasma decomposition technique and discussed the sulfur-resistance effect of the catalyst.64 They found that lesser defect sites on the plasma-decomposed catalyst result in better sulfur-resistance performance.

3.2. Carbon deposition and metal sintering Carbon deposition has been intensively studied in the methanation area in the temperature range from 200 to 450 °C and the steam reforming area in the temperature range from 600 to 900 °C.53 In thermodynamic equilibrium calculation, carbon formation is favorable in the high-temperature range (above 450 °C).65 However, carbon deposition was observed at a much lower temperature due to some mechanisms.66 McCarty et al. conducted CO temperature-programmed reaction at the temperature of 227 ± 50 °C over a Ni–Al2O3 catalyst and characterized 4-types of surface carbon atoms, namely (1) atomic carbon Cα, (2) amorphous carbon Cβ, (3) bulk Nicarbide Cγ, and (4) crystalline graphic carbon.66 The atomic carbon Cα and the initial Ni-carbide Cγ were active carbon atoms, whereas amorphous carbon Cβ and crystalline carbon were relatively stable. Nonetheless, the amorphous carbon (Cβ) was found to be active for methanation to some extent on a Rh-based catalyst.67 Fig. 9(A) shows the CH4 production peak during the temperature-programmed reaction. With the increase of carbon coverage, more carbon accumulated on the catalyst surface, as illustrated in Fig. 9(B).66 The step edges on the metal catalysts are the growth centres of carbon whiskers.68 It was observed on TEM that the nucleation and growth of graphene layers are accompanied by the restructuring of the metal step edges.68 In the methanation of lean hydrogen synthesis gas where H2 : COx is between 0.3 and 1.8, the metal particle morphology is changed by the carbon deposits.69 Czekaj et al. investigated the mechanism of carbon whisker growth on Ni–Al2O3 through in situ XPS and HighResolution Transmission Electron Microscopy (HRTEM) techniques. The carbon whiskers associated with the Ni particles

Catal. Sci. Technol.

View Article Online

Published on 21 April 2016. Downloaded by Nanyang Technological University on 09/05/2016 14:49:56.

Minireview

Fig. 9 (A): Mass spectra (CH4) of deposited surface carbon methanation on Ni–Al2O3. The two peaks indicate the activity of two kinds of surface carbon species, i.e., atomic carbon Cα and amorphous carbon Cβ. (B): Carbon inactivation pathways66 from left to right: very active atomic carbon Cα’ migrates72 to the step sites forming amorphous carbon Cβ (active), or atomic carbon Cα, in which the former is converted to elemental carbon and the latter to bulk carbide Cγ. Reproduced with permission from ref. 66. Copyright 1979 Elsevier.

Fig. 10 (A): HRTEM image showing the association of Ni particles with carbon whiskers on the Ni–Al2O3 catalyst after 137 hours of methanation. (B): Illustration of carbon whisker growth and Ni particle detachment from the catalyst support. Reproduced with permission from ref. 69. Copyright 2007 Elsevier.

caused the detachment of Ni particles from the support, as can be seen in Fig. 10(A) and (B) below.69 As for the carbon inactivation, Alstrup et al. proposed a whisker carbon growth model, in which unstable carbide initiated the growth.70 During the induction period, the unstable carbide evolved and caused Ni particle reconstruction. The unstable carbide then decomposed to filamentous carbon and metal. The carbon filament grew via surfacecarbon migration.70 Gupta et al. proposed another carbon inactivation model. They observed active and inactive carbon after carbon deposition on the catalyst surface.71 The inactive carbon was initially active, which then converted to inactive carbon with time and elevated temperature (>252 °C). The inactive carbon eventually converted to the graphitic form

Catal. Sci. Technol.

Catalysis Science & Technology

and blocked the active sites.71 Goodman et al. deposited surface carbon using the Boudouard reaction in various temperature ranges.14 The high-temperature deposited (>427 °C) surface carbon is present in the graphitic form. Heating in hydrogen can hardly remove the graphic carbon.14 Computational studies provide useful insight into the carbon deposition mechanism. Abild-Pedersen et al. studied the mechanism of carbon nanofiber formation using DFT calculation over a Ni single crystal. 72 They found that the Ni edge sites were the preferred sites for graphite growth. Surface carbon is able to diffuse along the step edge sites to the graphite perimeter to fill the graphite growth. Helveg et al. summarized the mechanisms of whisker carbon growth on a Ni catalyst. 73 The classic mechanisms that involved temperature gradient or concentration gradient induced whisker growth are investigated by HRTEM observation and DFT calculations. Surface steps acted as carbon growth centres. The stronger bond between carbon and graphene acted as the driving force. Carbon transported from the step sites to the free metal surface through surface diffusion or sub-surface diffusion.73 The rapid loss of methanation activity of the Ni-based catalyst during the first 5 hours can be attributed to the metal size growth.74,75 Agnelli et al. studied the low-temperature (230 °C) metal sintering process during CO hydrogenation over a supported Ni catalyst with the DRIFT technique.74 They observed the evolution of metal particle size distribution when the reaction temperature is far below the Tammann temperature (591 °C for Ni), which means no physical sintering has taken place. They attributed this sintering process to the migration of Ni-carbonyl (NiIJCO)4) species on the silica surface, as shown in Fig. 11, where catalyst particle size growth (A) corresponds to Ni-carbonyl band formation (B). Mirodatos et al. showed that higher CO pressure contributed to Ni carbonyl formation, which further enhanced the metal sintering.76 Shen et al. studied the catalyst deactivation caused by Ni-carbonyl formation and diffusion.75 They suggested using thermodynamic calculation to predict

Fig. 11 (A): DRIFT spectra of CO methanation on Ni–SiO2 (230 °C, H2 : CO = 2 : 1) showing the formation of surface Ni-carbonyl species (the sub-carbonyl peak) corresponding to the catalyst deactivation (due to Ni-particle smoothing) process. (B): Ni-carbonyl migration induced Ni metal sintering. Reproduced with permission from ref. 74. Copyright 1998 Elsevier.

This journal is © The Royal Society of Chemistry 2016

View Article Online

Published on 21 April 2016. Downloaded by Nanyang Technological University on 09/05/2016 14:49:56.

Catalysis Science & Technology

Ni(CO)4 formation. Specifically, the equilibrium partial pressure of Ni(CO)4 which was lesser than 10−6 Pa was found to be a safe zone for the Ni–Al2O3 catalyst.75 Rostrup-Nielsen et al. concluded that in the range of low-temperature methanation, sintering of the Ni catalyst was due to Ni-carbonyl (Ni(CO)4) formation and migration.77 High-temperature methanation prohibited Ni-carbonyl formation but thermal sintering was dominant.77

4. Summary and perspectives We discussed recent CO2 and CO methanation mechanisms and catalyst deactivation mechanisms through experimental and computational approaches. The following issues were addressed: (1) what are the intermediates on the catalyst surface? (2) What are the elementary steps that lead to the intermediates? (3) What are the active sites? For CO2 methanation, intermediates in the associative scheme include carbonate (CO3ad), bicarbonate (HCO3ad), formate (HCOOad), and carbonyl (COad) species. Researchers claimed that hydrogenation of these surface species reduced the barrier of C–O bond breaking. The active sites are located at the metal–support interface. On the other hand, the dissociative pathway involved carbonyl (COad) as an intermediate, which resulted from direct C–O bond breaking at the under-coordinated sites, i.e. step and kink sites. The active sites reduce C–O bond breaking barrier so that direct CO2 dissociation is possible at these locations. For CO methanation, some researchers concluded that direct C–O bond breaking is not energetically feasible. CO and H2 associatively adsorbed as formyl (CHOad) or carbon-hydroxyl (COHad) species so that the barrier is reduced. Nevertheless, some other researchers observed surface carbon (Cad) as the result of direct CO dissociation or disproportionation. In both associative and dissociative schemes, under-coordinated sites are the critical active sites, and the rate-limiting step is either C–O bond breaking or Cad hydrogenation. For the deactivation mechanisms: (1) sulfur poisoning results from the strong chemisorption of sulfur species on the active sites, (2) carbon deposition is induced by the migration and nucleation of inactive carbon atoms. The elevated temperature and the long period of absence of Had species induced the inactivation of the active carbon. The fouling of surface carbon caused metal detachment from the support material. (3) The metal sintering is caused by the migration of metal-carbonyl and other transportable species. For the future work, first of all, the debated on mechanisms should be uncovered clearly based on the respective active metals, supports, and reaction conditions so that catalyst optimization is possible. It is suggested to focus on model catalyst testing, where the effects of the support and promoter materials can be eliminated. Secondly, the catalyst micro-level morphology should be well-controlled to enhance the kinetics of the rate-determining elementary step. This requires the advancement of the catalyst synthesis method to achieve the desired combination of active sites. Thirdly, the

This journal is © The Royal Society of Chemistry 2016

Minireview

durability of the catalyst should be increased for commercialization and large-scale implementation. A strong metal–support interaction to counter the carbon detachment effect could be an effective approach. Lastly, the optimal reaction temperature, H2 : COx ratio, and pressure must be explored to maximize the methanation output. In conclusion, CO2 and CO methanation processes have been extensively studied, but the exact mechanism is still under debate. This literature review showed that both associative and dissociative methanation hold solid evidence in proving their validity in the respective reaction conditions. Despite the numerous approaches and claims made, more work is necessary to draw a clearer boundary between the two mechanisms. All in all, the development of high activity, high selectivity, and durable methanation catalysts are the key factors to the large-scale implementation of power-to-gas to tackle the intermittency of renewable energy sources. The mechanism studies provide insight into the rational design of the catalyst rather than the trial and error method. The breakthrough in this area, as well as other technologies such as solid oxide electrolysis cells and carbon capture and storage, are of great importance for the sustainable development of the future world.

Acknowledgements The authors would like to thank Agency for Science, Technology and Research (A*STAR), Singapore for the funding support to this project and B. M. is grateful to the Ph.D. scholarship provided by Nanyang Technological University.

Notes and references 1 S. Chu and A. Majumdar, Nature, 2012, 488, 294–303. 2 J. A. Turner, Science, 1999, 285, 687–689. 3 Global Wind Energy Council, http://www.gwec.net/, (accessed 03 Jan 2016). 4 F. Díaz-González, A. Sumper, O. Gomis-Bellmunt and R. Villafáfila-Robles, Renewable Sustainable Energy Rev., 2012, 16, 2154–2171. 5 A. Sternberg and A. Bardow, Energy Environ. Sci., 2015, 8, 389–400. 6 H. Yang, Z. Xu, M. Fan, R. Gupta, R. B. Slimane, A. E. Bland and I. Wright, J. Environ. Sci., 2008, 20, 14–27. 7 S. J. Davis, K. Caldeira and H. D. Matthews, Science, 2010, 329, 1330–1333. 8 G. A. Meehl, W. M. Washington, W. D. Collins, J. M. Arblaster, A. Hu, L. E. Buja, W. G. Strand and H. Teng, Science, 2005, 307, 1769–1772. 9 S. Schiebahn, T. Grube, M. Robinius, V. Tietze, B. Kumar and D. Stolten, Int. J. Hydrogen Energy, 2015, 40, 4285–4294. 10 G. Gahleitner, Int. J. Hydrogen Energy, 2013, 38, 2039–2061. 11 W. Wang, S. Wang, X. Ma and J. Gong, Chem. Soc. Rev., 2011, 40, 3703–3727. 12 J. Gao, Q. Liu, F. Gu, B. Liu, Z. Zhong and F. Su, RSC Adv., 2015, 5, 22759–22776.

Catal. Sci. Technol.

View Article Online

Published on 21 April 2016. Downloaded by Nanyang Technological University on 09/05/2016 14:49:56.

Minireview

13 M. A. A. Aziz, A. A. Jalil, S. Triwahyono and A. Ahmad, Green Chem., 2015, 17, 2647–2663. 14 D. W. Goodman, R. D. Kelley, T. E. Madey and J. T. Yates Jr, J. Catal., 1980, 63, 226–234. 15 R. D. Kelley and D. W. Goodman, Surf. Sci., 1982, 123, L743–L749. 16 C. Yuan, N. Yao, X. Wang, J. Wang, D. Lv and X. Li, Chem. Eng. J., 2015, 260, 1–10. 17 P. W. Wentrcek, J. G. McCarty, C. M. Ablow and H. Wise, J. Catal., 1980, 61, 232–241. 18 Y. Li, G. Lu and J. Ma, RSC Adv., 2014, 4, 17420–17428. 19 J. Liu, J. Yu, F. Su and G. Xu, Catal. Sci. Technol., 2014, 4, 472–481. 20 G. Zhi, X. Guo, X. Guo, Y. Wang and G. Jin, Catal. Commun., 2011, 16, 56–59. 21 P. Panagiotopoulou, D. I. Kondarides and X. E. Verykios, J. Phys. Chem. C, 2011, 115, 1220–1230. 22 P. Panagiotopoulou, D. I. Kondarides and X. E. Verykios, Catal. Today, 2012, 181, 138–147. 23 G. A. Somorjai, Chem. Rev., 1996, 96, 1223–1235. 24 J. K. Norskov, T. Bligaard, J. Rossmeisl and C. H. Christensen, Nat. Chem., 2009, 1, 37–46. 25 P. A. U. Aldana, F. Ocampo, K. Kobl, B. Louis, F. ThibaultStarzyk, M. Daturi, P. Bazin, S. Thomas and A. C. Roger, Catal. Today, 2013, 215, 201–207. 26 C. Schild, A. Wokaun, R. A. Koeppel and A. Baiker, J. Phys. Chem., 1991, 95, 6341–6346. 27 A. Westermann, B. Azambre, M. C. Bacariza, I. Graça, M. F. Ribeiro, J. M. Lopes and C. Henriques, Appl. Catal., B, 2015, 174–175, 120–125. 28 Q. Pan, J. Peng, S. Wang and S. Wang, Catal. Sci. Technol., 2014, 4, 502–509. 29 D. C. Upham, A. R. Derk, S. Sharma, H. Metiu and E. W. McFarland, Catal. Sci. Technol., 2015, 5, 1783–1791. 30 Y. X. Pan, C. J. Liu and Q. F. Ge, J. Catal., 2010, 272, 227–234. 31 M. R. Prairie, A. Renken, J. G. Highfield, K. Ravindranathan Thampi and M. Grätzel, J. Catal., 1991, 129, 130–144. 32 S. T. Zhang, H. Yan, M. Wei, D. G. Evans and X. Duan, RSC Adv., 2014, 4, 30241–30249. 33 S. Eckle, H. G. Anfang and R. J. Behm, J. Phys. Chem. C, 2011, 115, 1361–1367. 34 C. De Leitenburg, A. Trovarelli and J. Kašpar, J. Catal., 1997, 166, 98–107. 35 S. Akamaru, T. Shimazaki, M. Kubo and T. Abe, Appl. Catal., A, 2014, 470, 405–411. 36 S. I. Fujita, M. Nakamura, T. Doi and N. Takezawa, Appl. Catal., A, 1993, 104, 87–100. 37 S. Tada, T. Shimizu, H. Kameyama, T. Haneda and R. Kikuchi, Int. J. Hydrogen Energy, 2012, 37, 5527–5531. 38 G. A. Mills and F. W. Steffgen, Catal. Rev.: Sci. Eng., 1974, 8, 159–210. 39 M. P. Andersson, F. Abild-Pedersen, I. N. Remediakis, T. Bligaard, G. Jones, J. Engbæk, O. Lytken, S. Horch, J. H. Nielsen, J. Sehested, J. R. Rostrup-Nielsen, J. K. Nørskov and I. Chorkendorff, J. Catal., 2008, 255, 6–19.

Catal. Sci. Technol.

Catalysis Science & Technology

40 S. Eckle, Y. Denkwitz and R. J. Behm, J. Catal., 2010, 269, 255–268. 41 J. Wang, Y. Kawazoe, Q. Sun, S. Chan and H. B. Su, Surf. Sci., 2016, 645, 30–40. 42 C. Zhi, Q. Wang, B. Wang, D. Li and R. Zhang, RSC Adv., 2015, 5, 66742–66756. 43 J. L. C. Fajín, J. R. B. Gomes and M. N. D. S. Cordeiro, J. Phys. Chem. C, 2015, 119, 16537–16551. 44 Y. Wang, Y. Su, M. Zhu and L. Kang, Int. J. Hydrogen Energy, 2015, 40, 8864–8876. 45 Y. Tison, K. Nielsen, D. J. Mowbray, L. Bech, C. Holse, F. Calle-Vallejo, K. Andersen, J. J. Mortensen, K. W. Jacobsen and J. H. Nielsen, J. Phys. Chem. C, 2012, 116, 14350–14359. 46 S. B. Vendelbo, M. Johansson, D. J. Mowbray, M. P. Andersson, F. Abild-Pedersen, J. H. Nielsen, J. K. Norskov and I. Chorkendorff, Top. Catal., 2010, 53, 357–364. 47 M. Araki and V. Ponec, J. Catal., 1976, 44, 439–448. 48 H. H. Madden and G. Ertl, Surf. Sci., 1973, 35, 211–226. 49 S.-i. Fujita, H. Terunuma, M. Nakamura and N. Takezawa, Ind. Eng. Chem. Res., 1991, 30, 1146–1151. 50 P. Forzatti and L. Lietti, Catal. Today, 1999, 52, 165–181. 51 C. H. Bartholomew, Appl. Catal., A, 2001, 212, 17–60. 52 J. Oudar, Catal. Rev.: Sci. Eng., 1980, 22, 171–195. 53 C. H. Bartholomew, Catal. Rev.: Sci. Eng., 1982, 24, 67–112. 54 B. Legras, V. V. Ordomsky, C. Dujardin, M. Virginie and A. Y. Khodakov, ACS Catal., 2014, 4, 2785–2791. 55 P. K. Agrawal, J. R. Katzer and W. H. Manogue, J. Catal., 1982, 74, 332–342. 56 C. H. Bartholomew, G. D. Weatherbee and G. A. Jarvi, J. Catal., 1979, 60, 257–269. 57 J. R. Rostrup-Nielsen and K. Pedersen, J. Catal., 1979, 59, 395–404. 58 J. Engbæk, O. Lytken, J. H. Nielsen and I. Chorkendorff, Surf. Sci., 2008, 602, 733–743. 59 E. J. Erekson and C. H. Bartholomrw, Appl. Catal., 1983, 5, 323–336. 60 W. D. Fitzharris, J. R. Katzer and W. H. Manogue, J. Catal., 1982, 76, 369–384. 61 I. Czekaj, R. Struis, J. Wambach and S. Biollaz, Catal. Today, 2011, 176, 429–432. 62 R. P. W. J. Struis, T. J. Schildhauer, I. Czekaj, M. Janousch, S. M. A. Biollaz and C. Ludwig, Appl. Catal., A, 2009, 362, 121–128. 63 K. Lee, C. Song and M. J. Janik, Appl. Catal., A, 2010, 389, 122–130. 64 X. L. Yan, Y. Liu, B. R. Zhao, Y. Wang and C. J. Liu, Phys. Chem. Chem. Phys., 2013, 15, 12132–12138. 65 J. Gao, Y. Wang, Y. Ping, D. Hu, G. Xu, F. Gu and F. Su, RSC Adv., 2012, 2, 2358–2368. 66 J. G. McCarty and H. Wise, J. Catal., 1979, 57, 406–416. 67 K. Nishizawa, H. Kato, K. Mimori, T. Yoshida, N. Hasegawa, M. Tsuji and Y. Tamaura, J. Mater. Sci., 1994, 29, 768–772. 68 S. Helveg, C. Lopez-Cartes, J. Sehested, P. L. Hansen, B. S. Clausen, J. R. Rostrup-Nielsen, F. Abild-Pedersen and J. K. Norskov, Nature, 2004, 427, 426–429. 69 I. Czekaj, F. Loviat, F. Raimondi, J. Wambach, S. Biollaz and A. Wokaun, Appl. Catal., A, 2007, 329, 68–78.

This journal is © The Royal Society of Chemistry 2016

View Article Online

Published on 21 April 2016. Downloaded by Nanyang Technological University on 09/05/2016 14:49:56.

Catalysis Science & Technology

70 I. Alstrup, J. Catal., 1988, 109, 241–251. 71 N. M. Gupta, V. S. Kamble, K. A. Rao and R. M. Iyer, J. Catal., 1979, 60, 57–67. 72 F. Abild-Pedersen, J. K. Nørskov, J. R. Rostrup-Nielsen, J. Sehested and S. Helveg, Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 73, 115419. 73 S. Helveg, J. Sehested and J. R. Rostrup-Nielsen, Catal. Today, 2011, 178, 42–46.

This journal is © The Royal Society of Chemistry 2016

Minireview

74 M. Agnelli, H. M. Swaan, C. Marquez-Alvarez, G. A. Martin and C. Mirodatos, J. Catal., 1998, 175, 117–128. 75 W. M. Shen, J. A. Dumesic and C. G. Hill Jr, J. Catal., 1981, 68, 152–165. 76 C. Mirodatos, H. Praliaud and M. Primet, J. Catal., 1987, 107, 275–287. 77 J. R. Rostrup-Nielsen, K. Pedersen and J. Sehested, Appl. Catal., A, 2007, 330, 134–138.

Catal. Sci. Technol.