A review on catalytic methane combustion at low temperatures: Catalysts, mechanisms, reaction conditions and reactor designs

Natural gas (with methane as its main component) provides an attractive energy source because of its large abundance and its high heat of combustion per mole of carbon dioxide generated. However, the emissions released from the conventional flame combustion (essentially NOx) have harmful impacts on the environment and the human health. Within the scope of rational and clean use of fossil energies, the catalytic combustion of natural gas appears as one of the most promising alternatives to flammable combustion. The presence of catalysts enables complete oxidation of methane at much lower temperatures (typically 500 C), so that the formation of pollutants can be largely avoided. This work presents a literature review on the catalytic methane combustion. Various aspects are discussed including the catalyst types, the reaction mechanisms and kinetic characteristics, effects of various influencing operational factors and different reactor types proposed and tested. This paper may serve as an essential reference that contributes to the development of well-designed reactors, equipped with appropriate catalysts, and under well-handled operating conditions to realize the favorable (kinetic) performance, for their future applications and propagation in different industrial sectors.


Introduction
One of the major concerns over economic growth and social development nowadays is the constantly increasing energy demand [1]. The study of U.S. Energy Information Administration has forecasted an increase of 28% in the world's energy consumption from 2015 to 2040 [2]. While there is a constant progress year by year for the development of renewable energies, the use of fossil sources (petroleum, coal and natural gas) is still dominant, and remains indispensable in the near future [3].
Among the fossil energy resources, the natural gas presents a particular interest because of its higher energy content (55.7 kJ g 1 if fully based on methane as its main component) than coal (39.3 kJ g 1 ) and petroleum (43.6 kJ g 1 ) as well as its reduced CO 2 emission (50% less than coal and 30% less than petroleum). Moreover, the proven natural gas reserves worldwide are abundant, reaching about 193.5 trillion cubic meters at the end of 2017 [4]. As a result, natural gas has accounted for the largest increment (24%) in the main energy consumption in the past decade until 2017, and has been suggested as a substitute for oil and coal as a future leading energy source for the next 20 years [5]. In response to this, there is a rapidly growing number of research & development efforts yearly on the deployment of natural gas for their use in various sectors including industrial, residential, power, transport and many others [6].
Besides the natural gas fields, the synthetic natural gas (SNG) can also be derived from coal gasification, CO 2 methanation and biomass gasification/digestion [7,8]. Fig. 1 shows the main reaction network of SNG in the industry. Biomass is particularly promising as a substitute for fossil resources owing to its benefits of energy security and environmental friendliness. On one hand, the SNG can be obtained from Abbreviations: 3DOM, Three-dimensionally ordered microporous; AB x Al (12-x) O 19 (x ¼ 1, 3,6,9,12), Hexaaluminate formula; ABO 3 (or A I B V O 3 , A II B IV O 3 , or A III B III O 3 ), Perovskites formula; BET, Brunauer, Emmett and Teller method, specific surface area of catalyst, unit: m 2 ⋅g 1 ; BHA, Barium hexaaluminate; CMC, Catalytic methane combustion; DP, Deposition precipitation; DRIFT, Diffuse reflectance infrared spectroscopy; HDP, Homogeneous deposition precipitation; HTF, Heat transfer fluid; ITM, Ion transport membrane reactor; NGVs, Natural gas vehicles; SEM, Scanning electron microscope; SOFCs, Solid oxide fuel cells; SNG, Synthetic natural gas.
upgrading biogas that is generated from biomass digestion (e.g. manure) and/or from carbohydrate fermentation by bacteria in an anaerobic environment [9][10][11]. On the other hand, the SNG can be produced via gasification of biomass (e.g. wood, straw and crops) followed by the process of methanation [12,13]. The syngas and methanol can be synthesized by partial oxidation and steam reforming reaction, producing consequently synthetic fuels and hydrogen. Meanwhile, the combustion of methane can provide the heat and electricity due to the strongly exothermic nature of the reaction.
The conventional flame combustion of (synthetic) natural gas occurs typically at above 1400 � C and releases harmful pollutants (such as NO x , CO and hydrocarbon). The impact of NO x on human health (respiratory diseases) has been widely recognized [14]. Its emission also has harmful environmental impacts including the formation of photochemical smog and acid rain [15]. More and more stringent regulations are thus applicable over European countries. For example, in September 2018, the maximum NO x emission level has been reduced from 70 mg⋅kWh 1 (class 5) to 56 mg⋅kWh 1 for all domestic boilers sold in Europe [16]. As a result, the complete oxidation of natural gas in the presence of catalysts (i.e. the catalytic combustion) appears as one of the most promising alternative solutions for the rational and clean use of fossil energies. The activation energy is reduced from 100-200 kJ mol 1 (conventional combustion) to 40-80 kJ mol 1 (catalytic combustion), leading to a lower working temperature (<600 � C). In this regard, less pollutant emissions could be reached (~5 ppm compared with 150-200 ppm for conventional combustion). Hence, the catalytic combustion of methane or natural gas as a clean technology has received increasing research attention [17], indicated by the significantly increasing number of yearly publications over the past two decades (Fig. 2).
Various application areas of catalytic methane combustion (CMC) have been proposed and attempted, as illustrated in Fig. 3 and briefly described below.
(i) Natural gas vehicles (NGVs) [18][19][20][21] (ca. 300-700 � C): NGVs have the advantages in the abatement of greenhouse gas emissions and smog emissions compared to gasoline or diesel-driven vehicles. Three-way catalysts are applied on NGVs mainly for exhaust purification in practice. (ii) Gas turbine [22][23][24][25][26][27] (ca. 700-1400 � C): Methane combustion is widely used as the fuel on the gas turbine. The combusted gas is used to drive a turbine for power generation. For example, 25 kW electricity output can be obtained with 0.8 vol% methane in the air [24]. (iii) Solid oxide fuel cells (SOFCs) [28] (ca. 500-1000 � C): The preheated compressed air passes into the cathode of the battery while the compressed methane mixed with the overheated steam enters into the anode of the fuel cell. The methane electrochemical conversion in SOFCs, if properly controlled, could obtain a high conversion efficiency and an environmental benefit due to a significant decrease in pollutant emissions. (iv) Domestic heating systems [29][30][31][32] (ca. 300-700 � C): the heat released from the exothermic CMC reaction is utilized to drive the domestic heating systems, such as the central boilers or gas stoves. A high energy conversion efficiency and eco-friendly water boiler prototype with a hot water yield of 11.5 kg min 1 has been reported [33]. (v) Coupling with endothermic reaction (ca. 300-700 � C): the reaction heat from CMC is commonly used to drive an endothermic reaction so as to maintain the continuous autothermal operation [34]. Novel reactor designs have been proposed for coupling the CMC with an endothermic reaction (methane steam reforming [35,36], dehydrogenation of propane to propylene [37], dehydrogenation of ethane to ethylene [38,39], etc.), owing to the optimized energy integration and the process intensification.
It may be discovered that compared to the conventional flame combustion, the presence of catalysts enables a decrease of the working temperature (<1400 � C). Depending on the target application, the operational temperature for CMC can be further divided into a relatively lower range (about 300-700 � C) and a relatively higher one (about 700-1400 � C). The low-temperature CMC becomes more attractive due to the remarkable abatement of pollutant emissions and the prolonged catalyst lifetime. For instance, the reusability and the reproducibility of catalysts, especially for noble metal catalysts, are shortened at high temperatures. In this field, developing catalysts with high catalytic activity, low light-off temperature and good thermal stability even for such low temperature operations is still a challenging issue.
A great number of researches have been devoted to catalyst development [18,40,41] and reactor design [42][43][44] for CMC. Noble metal catalysts (e.g. Pt, Pd and Rh) have been widely investigated owing to their high catalytic activity. Hexaaluminate and perovskite catalysts,  due to their relatively lower catalytic activity and high thermal stability, are commonly used for high temperature applications (600-1400 � C). Optimization of reaction conditions over various catalysts has been broadly investigated [19], such as the effect of light-off temperature, reactant concentration, oxygen to methane molar ratio, residence time, etc. Moreover, the mechanistic studies mainly focusing on kinetic models for various catalysts have been well elaborated in earlier literatures [45][46][47][48]. The reactor designs (e.g. micro/mini-structured reactor)   [68]; (b) perovskites (La 0.5 Sr 0.5 CoO 3-δ ) [69]; (c) noble metal-based (Pd-Ru) catalyst [70].  [49][50][51]. Reviews papers related to CMC have also been published, as summarized in Table 1. Nevertheless, most of them primarily focus on the improvement of catalytic activity (e.g. noble metal-based catalysts [17,18,40,41], hexaaluminates/perovskite catalysts [52,53]). Other review papers may involve the CMC in one or several sub-sections, but they are mainly devoted to a specific topic, e.g. heating system [54,55], SOFCs [28], coupling exothermic/endothermic reactions [34], SNGs [9,56], etc. The present review on CMC aims at filling the literature gap by providing a comprehensive and combined understanding of catalysts, mechanisms, reaction conditions and reactor designs. In particular, the present paper has the following objectives: � A brief introduction of the catalyst types, their advantages/disadvantages, associated reaction mechanisms and kinetic characteristics. � A complete survey on the effects of various operational factors on the performance of CMC, including temperature, space velocity, O 2 /CH 4 ratio, natural gas composition and pressure. � A review on different reactor types used for CMC, with a special focus on microchannel reactor-heat exchangers.
This paper may serve as an essential reference that contributes to the development of well-designed reactors, equipped with appropriate catalysts and under well-handled operating conditions, towards realizing their favorable (kinetic) performance and for their future application and propagation in different industrial sectors.

Catalyst category
Catalysts play an important role in terms of catalytic activity and reaction rate on the CMC, and are mainly categorized into metal oxide catalysts (e.g. hexaaluminate, perovskites, and single-metal oxides) and noble metal-based catalysts. The research interests on perovskites and noble metal catalysts are remarkably increasing over the years, with the latter being the most popular. The main advantages and disadvantages of catalysts are summarized in Table 2 [65][66][67]. Hexaaluminate has been applied for CMC since 1987, owing to its exceptionally high thermal stability and strong resistance to thermal shock [71]. Thus, hexaaluminate is considered as the most suitable catalyst for high temperature applications (e.g. for gas turbines). Other main applications include the methane partial oxidation, the dry reforming of methane and the decomposition of N 2 O. Although a great improvement of specific surface has been achieved, efforts are still required so as to synthesize hexaaluminate with simple procedures, as well as an excellent catalytic activity.
(2) Perovskites [72][73][74] are represented by a standard formula as ABO 3 (or more complicated as A I B V O 3 , A II B IV O 3 or A III B III O 3 ). A as a larger cation is commonly composed of alkaline/rare earth elements (e. g. of La, Sr, Bi, etc.), residing on the edge of the structure for its stabilization with less effect on the catalytic activity. B as a smaller cation consists of transition metal that is surrounded by octahedral of oxygen anions, functioning as the main catalytic center. Their structure is schematically shown in Fig. 4b. The microstructure of mixed oxide catalyst is beneficial for their oxygen mobility and catalytic activity [75]. The presence of defect structure in oxygen vacancies, the existence of unusual valence and the availability of reversibly released oxygen have been considered relevant to the enhanced catalytic activity, even comparable to that of noble metal catalysts [76,77]. It can be explained by the fact that the oxygen vacancies are directly relevant to the adsorbed oxygen species over the catalyst surface. The more oxygen vacancies, the more adsorbed oxygen formed over the surface, leading to the higher catalytic activities in methane oxidation. A recent work reported by Miao et al. [78] reveals that more active oxygen species could be obtained using La(Mn, Fe)O 3þλ perovskite catalyst, and the catalytic Note. a Average specific surface area measured by BET (Brunauer, Emmett and Teller) method is shown here, but may vary depending on the preparation method.
activity of CMC thereby was significantly improved. The cation-substitution of perovskites effectively increases the oxygen vacancies by varying the distribution of B oxidation state [79,80]. Different aspects of perovskites have been addressed in several review papers, including the structure, synthesis and applications [53,59,81], the acid-base catalytic properties of perovskites [82], and the lanthanum-based perovskites [83]. The mechanism and kinetics may be found in the book of Granger et al. [84]. A lower calcination temperature is required for the perovskite phase than the hexaaluminate phase [85]. Perovskite catalysts are featured by their high thermal stability as well as the improved specific surface area, displaying a better catalytic activity in CMC. The higher catalytic performance is mainly ascribed to the foreign-cation substitution, the produced oxygen lattice and the deficiency over catalyst surface. A promising direction of improvement is designing perovskite catalysts with featured morphologies (e.g., nano-sized, porous, hollow), favoring their potential industrial applications.

Noble metal catalysts
Noble metal catalysts have been most intensively investigated for CMC, owing to their high catalytic activity at low temperatures [57,[86][87][88]. Their basic structure is shown in Fig. 4c. Pd, Pt, Rh, Au and Co as the active component have been widely studied in the literature. Among them, Pd and Pt-based catalysts were reported as the most active one by far. Various support materials, such as ZrO 2 , CeO 2 , Al 2 O 3 , SnO 2 , TiO 2 , were considered. The base/acid properties of the support affect the catalytic activity by interacting with the oxidized/metallized state of noble metals. It was reported that the decreased acidity strength of Al 2 O 3 support (with Pd as the active component) could enhance the performance of CMC [89]. Moreover, the introduction of additives (e.g. of La, Mn, Ce, Mg, V) could stabilize the catalyst support and active sites, and prolong the catalyst life. It has been reported by Farrauto et al. [90,91] that the CeO 2 addition is favorable to prevent the catalyst deactivation. The PdO species on the catalyst surface thereby are stabilized due to the increased temperature of PdO decomposition. Moreover, the improved storage and exchange of oxygen species in the presence of CeO 2 effectively promote the Pd reoxidation, resulting in a higher catalytic performance [92]. The recent study by Toso et al. [93] illustrated that the stability of Pd/Ce 0.75 Zr 0.25 O 2 catalyst exposure to the water was improved by well-dispersed small Pd nanoparticles. More detailed reviews can be found in the literature [17,18,41].
composition, properties, and the preparation method. With respect to Pd and Pt-based catalysts, Pd is supposed to be superior to Pt, not only for the CMC but also for the oxidation of higher alkanes and olefins [94]. It is commonly considered that Pd in the oxidized state (PdO) is the most active and stable (up to 800 � C) [95]. Farrauto et al. [90] proposed that at least two different PdO species were present on the Al 2 O 3 support. Dispersed PdO decomposed in a temperature range between 750 and 800 � C, whereas crystalline PdO decomposed from 800 to 850 � C. Hicks et al. [96] identified at least two different phases by infrared spectra. The crystalline palladium with a smaller size presented 10 to 100 higher catalytic activity than dispersed PdO phase. Similarly, the Pt crystalline phase has a higher catalytic activity than that in the dispersed PtO 2 phase due to the formation of chemisorbed oxygen in the crystalline phase [96]. Moreover, bi-or trimetallic catalysts have been reported to have higher catalytic activity and stability compared to monometallic ones [97][98][99][100][101]. For example, Pd-Pt/Al 2 O 3 catalyst is more active and stable than Pd/Al 2 O 3 [102,103]. It has been reported that Pt-Pd catalysts showed a higher activity even than Pd-Ag, Pd-Co, Pd-Ni and Pd-Rh over the Al 2 O 3 support [97]. A better synergetic effect and the formation of bi-metal structure have proved to improve the catalyst activity and life-time. Other factors such as the support structure, the particle size and the surface morphology also have significant influence on the catalytic performance. More details on the influence of these factors can be found in the references [19,[104][105][106].
The electrochemical field-assisted CMC is a relatively novel direction in recent years owing to the synergetic effect. Electrocatalysis process commonly involves the oxidation and reduction reactions via direct electrons transformation (i.e. the produced electrical current). Electrolytes as promoting species can modify the electronic properties of the catalyst surface via the formation of favorable bonds between reactants and the electrodes. The decrease of the activation energy through the synergetic effect between electric field and catalysis results in the enhancement of reaction rate for CMC [107][108][109]. Li et al. [109] reported that the reaction rate of CMC over the Mn x Co y catalyst was remarkably accelerated by the improved reducibility of Co 3þ in the electric field, promoting the methane activation at low temperature. The light-off temperature (T 50 ¼ 255 � C) over PdCe 0.75 Zr 0.25 O x catalyst can be significantly reduced because of the enhanced reducibility of PdO x species in electric field (e.g. 3 mA current) [110]. More details on the electrochemical-assisted CMC may be found in a recent reference [111].
Although noble metal catalysts present advantages such as high specific surface area, high dispersion of active component and mild reaction conditions, the catalyst deactivation (due to sintering, particle size growth, poisoning, etc.) and the high cost are the main limitations for their large-scale application in the industry.

Shaping of catalyst
The shaping of catalysts could significantly affect the pressure drop and the reactant-catalyst mass transfer in the reactor. Fig. 5 shows a variety of catalyst shapes used for CMC. Fine powders are more suitable for being incorporated into minireactors or microreactors with higher catalyst surface area. However, powder catalysts ( Fig. 5a) could lead to a high pressure drop if packed in a long (e.g., several meters) fixed-bed reactor, or possibly be blown out when used in a fluidized-bed reactor. To decrease the pressure drop, the catalyst is commonly shaped into larger bodies, e.g. pellet, round ball, cylindrical shape ( Fig. 5b-d). Moreover, a sufficient mechanical strength of the catalyst support is essential for the catalyst's long-term structural durability.
Washcoated catalysts have received an increasing attention owing to its high surface area, low pressure drop and better usage of catalyst. This type of catalyst is usually used in monolithic reactors (Fig. 5e) [112,116], foam reactors ( Fig. 5f) [113,117], multichannel microreactors ( Fig. 5g) [118][119][120] and tube reactors (Fig. 5h) [32,121]. The recent progress of washcoated and packed-bed microreactors is reviewed in the reference [50]. The washcoated catalyst is commonly deposited as a thin layer on structured surfaces using typically a dip-coating method [122]. Other methods, such as suspension method [120,[123][124][125], sol-gel technique [126,127], chemical vapor deposition [128,129], physical vapor deposition [130,131]are also widely used. In our previous work [120], the preparation of a well-adhered Pt/γ-Al 2 O 3 catalytic coating in microreactors has been elaborated by applying various binders, particle size, pH conditions, etc. The preparation of the suspension and the pretreatment of the substrate hosting the catalytic layer have to be adapted to obtain a high thermal stability and a well dispersion of the coating [118,120,132].

Mechanism and kinetic study of CMC
Compared to other higher alkanes, methane is the most stable alkane molecule with high ionization potential (12.5 eV), low electron affinity (4.4 eV) and high C-H bond energy (434 kJ mol 1 ), rendering it extremely difficult to be activated under mild conditions. A high reaction temperature (>1400 � C) is often required for carrying out conventional methane flame combustion. Hence, mechanistic and kinetic studies are important for guiding the catalyst design and the process optimization in order to achieve an efficient combustion at relative low temperature levels (<600 � C) [133][134][135]. The reaction has been reported to be zero order in oxygen and first order in methane [136]. The kinetic model and the elementary steps were elaborated in the literature [137][138][139][140][141][142][143], and the main kinetics parameters are summarized in Table 3.
Regarding the noble metal catalyst, a great number of studies have been devoted to revealing the mechanism of catalytic methane oxidation [48,[144][145][146]. The classic reaction routes over noble metal catalysts are shown in Fig. 6 [147]. CH 4 molecules are first adsorbed on the catalyst and dissociated to the adsorbed methyl (CH 3 ⋅) or methylene (CH 2 ⋅) species, which further interact with the adsorbed oxygen, either to directly produce CO 2 and H 2 O, or to form the adsorbed CO and H 2 via formaldehyde (HCHO) as the intermediate [148,149]. The adsorbed CO and H 2 further interact with the adsorbed oxygen to form the final product (CO 2 and H 2 O) based on the reactant ratios (theoretically, partial oxidation occurs at O 2 /CH 4 molar ratio < 2). The adsorbed CO is predominant with the increasing methane coverage, whereas CO 2 formation is more favorable at high oxygen coverages. However, due to the swift dissociation of CO, the variation of the surface concentrations of methane and oxygen is negligible. Experimental measurements over Pt/Al 2 O 3 catalysts have indicated that the reaction rate determining step was shifted from the oxygen desorption to the methane adsorption with the increasing catalyst surface temperature [141]. Given the higher methane adsorption energy than oxygen [150][151][152], at the beginning the oxygen adsorption reaction (O 2 þ 2 Pt(*) → 2O(*) þ 2 Pt; * is the molecule adsorbed on the surface), this rate determining step may be additionally due to the competitive adsorption of oxygen that inhibits the methane oxidation by excluding the weakly adsorbed methane on the active sites [147]. At high oxygen atom coverages, methane is converted through the proposed reaction (CH 4 þ O(*) þ Pt(*) → CH 3 (*) þ OH(*) þ Pt). As a result, the surface temperature increases due to the release of the reaction heat. The number of the adsorbed oxygen atoms is generally decreased with the increasing temperature and the reaction of the methane adsorption (CH 4 þ 2 Pt(*) → CH 3 (*) þ H(*) þ 2 Pt) becomes more prominent. The light-off phenomenon thus happens once the favorable coverage of methane and oxygen on the catalyst surface is reached [133,153].
Three types of mechanism and the corresponding kinetic models have been proposed for CMC in the literature, including the Langmuir-Hinselwood mechanism [154][155][156], the Eley-Rideal mechanism [157] and the Mars-van Krevelen mechanism [158][159][160][161]. The rate-determining step for both the Langmuir-Hinselwood and Eley-Rideal mechanisms is commonly considered as the superficial reaction. The reaction rate is associated to the electronic properties of transition ions over the catalyst surface. On the contrary, the CMC is considered as the interfacial reaction by the Mars-van Krevelen mechanism; the reaction rate is mainly correlated to the lattice oxygen vacancies.
Regarding the Langmuir-Hinselwood mechanism, the molecules of both gas phase reactants are adsorbed on the catalyst surface and react via surface diffusion. The formed products are then desorbed from the catalyst surface to complete the reaction. The kinetic models of CMC over Pt/Al 2 O 3 catalyst proposed by Trimm and Lam [162] well fit the Langmuir-Hinselwood mechanism, indicating that both the adsorbed methane and oxygen were involved in the reaction. Their study confirmed that the temperature increase was mainly to change the reaction path from the oxygen adsorption to methane adsorption [162]. However, Jodłowski et al. [163] observed that methane over Co-Pd/γ-Al 2 O 3 catalyst was only adsorbed with pre-adsorbed oxygen over the surface (under oxygen-rich conditions) by using the DRIFT (diffuse reflectance infrared spectroscopy), suggesting that the Langmuir-Hinshelwood mechanism should not be recommended.
The Eley-Rideal mechanism suggests that only one gas phase reactant has to be adsorbed onto the catalyst surface. The adsorbed reactant then interacts with the other reactant which is still in the gas phase. Subsequently, the formed products are desorbed from the catalyst surface. Seimanides and Stoukides [157] reported that this mechanism could well predict the CMC over Pd/ZrO 2 catalyst in the range of 450-600 � C. It is likely to be the only adsorbed atomic oxygen that reacts with the gaseous methane. Veldsink et al. [164] illustrated that the Eley-Rideal mechanism was adequate to describe the experiment data, and the reaction rate equation over CuO/γ-Al 2 O 3 catalyst was proposed without the limitation of heat and mass transfer.
The Mars-van Krevelen mechanism is widely supported by a large amount of experimental results on CMC [165,166]. Different from the above two mechanisms, the Mars-van Krevelen mechanism suggests that the adsorbing surface is an active participant. Firstly, one of the reactants in the gas phase forms a chemical bond with the catalyst surface in the form of a thin layer (e.g. of metal oxide). Then, the remaining gas phase reactant can interact with the chemically bonded reactant, leaving behind a vacancy upon desorption of the products. However, it is not easy to distinguish between Mars-van Krevelen and Eley-Rideal mechanisms because of the existence of both the lattice and adsorbed oxygen species on the catalyst surface. Pfefferle et al. [160] further reported that one 16 O atom (lattice phase) in PdO was bounded to two Pd atoms, using the in-situ technology of isotopically labeled reaction. It was found that the 16 O atom in PdO was responsible to oxidize methane rather than the adsorbed 18 O atom in the gas phase. This conclusion is in line with the findings of Au-Yeung et al. [167]. In addition, the variation in the oxidation valence of Pd plays an important role in the reaction, indicating that the Mars-van Krevelen mechanism is more adequate to be used for the CMC [57,168,169]. Similarly for NiCo 2 O 4 perovskite catalyst, Tao et al. [77] reported that the chemisorbed lattice oxygen played an important role. The oxidized products (CO 2 and H 2 O) were generated           by the competitive adsorption, and surface vacancies subsequently left behind by the fast re-oxidation on the catalyst surface. Note that the DRIFT associated with Raman and X-ray fluorescence spectroscopies has been applied as important in-situ technologies by Jodłowski et al. [163] for the CMC over Co-Pd/Al 2 O 3 catalyst. The proposed mechanism is slightly different from the Mars-van Krevelen mechanism, in that only the adsorbed active oxygen species on the catalyst surface was responsible to oxidize methane instead of the bulk oxygen atoms. Furthermore, the presence of -OCH 3 species was detected, rather than HCHO and H 2 in the gas phase. These results are also supported by other studies [170,171]. Therefore, there is no unanimous mechanism so far to fully elaborate the CMC, given the whole processes being rather complex and strongly dependent on the reaction conditions and used catalysts [172,173]. The Mars-van Krevelen mechanism seems to be more widely accepted than the Langmuir-Hinselwood and Eley-Rideal mechanisms. In this respect, more in-depth understanding is still needed to better elucidate the reaction pathway [174]. More details on the reaction mechanisms may be found in the literature [138][139][140][141][142][143].

Effect of operational conditions on CMC
In this section, main factors that should be carefully assessed are reviewed so as to determine the appropriate working conditions for Notes.
a Mentioned studies in Table 4 are listed in the order of published year. b T 10 /T 50 /T 90 : the temperature at which the methane conversion is 10%, 50% and 90%. Fig. 7. Reaction rate as a function of temperature [186]. CMC, including the temperature, the ratio of methane to oxygen, the flow rate, the reactant composition and the pressure. The reaction behavior addressed in this section, unless otherwise specified, is assumed to be intrinsic. A summary of the studies over various catalysts and the key influential factors can be found in Table 4.

Effect of temperature
In general, the intrinsic reaction rate for CMC is correlated to temperature and activation energy according to the Arrhenius equation [182,183]. The reaction rate presents a great increase with the increasing reaction temperature.
Light-off temperature is one of the most crucial parameters used to indicate the catalyst activity [118,184,185]. Fig. 7 shows the reaction regime as a function of the temperature. At the beginning, the temperature over the catalyst is low and thus the reaction rate is limited by the intrinsic kinetics (regime A-B). The light-off is commonly defined by the regime where the temperature has no significant change when the conversion is increased from 10%, 20% up to 50% (regime B-C) [186]. As light-off happens at comparatively high temperature levels, the intrinsic reaction rate rapidly increases but gradually the mass transfer rate cannot keep up with (regime B-C). Hence, the overall reaction rate tends to be limited more by the mass transfer rate. Upon further increase of the reaction temperature, the intrinsic reaction rate is increased so fast that the reaction falls in the mass transfer controlled regime (regime C-D). In this regime, the rate of increase in the overall reaction rate becomes slower. Eventually at substantially high temperature levels, the homogeneous combustion of methane dominates (regime D). The latter two regimes impose high requirements on the reactor design in terms of maintaining the catalyst stability and enhancing the mass and heat transfer rates. The effect of high working temperatures on the shortened life-time due to the catalyst sintering should be considered, especially for noble metal catalysts.
As for wall-coated plate-type/microchannel reactor, the temperature distribution along the catalytic reactor commonly displays a rapid increase at the entrance region, and a smooth decrease along the reactor length thereafter, as shown in Fig. 8. This is because that methane is mainly converted at the front section of the reactor [187]. However, the peak temperature region moves slowly towards the downstream with the increasing oxygen/methane ratio. The variation of temperature along the reactor mainly depends on the reactant flow pattern and the reactor structure, which will be elaborated in the following section 5.
Moreover, the pollutant emission could be significantly affected by the operational temperature. With the presence of N 2 in the reactant feed, it was reported that NO x emission showed a growing trend with the increasing temperature [188]. The formation of CO is often due to the incomplete methane combustion under low temperatures and/or with a methane rich mixture.
In addition, it is worth noting that the catalyst deactivation is significantly influenced by the operational temperature. For instance, the active PdO phase is decomposed to the less active metallic palladium (PdO → Pd) at above 800 � C, followed by the agglomeration and deactivation of catalysts [95]. It is thus essential to maintain the operational temperature below the thermal decomposition temperature of PdO. Another solution is to increase the decomposition temperature by optimizing the catalyst, e.g. by introducing metal oxides [189]. Farrauto et al. [90] pointed out that the addition of ZrO 2 into PdO/Al 2 O 3 catalyst exhibited a superior synergistic effect between the temperature hysteresis and the active site reformation. The decomposition temperature of PdO was thus increased to ca. 900 � C. In order to regenerate the catalyst, Farrauto et al. [90] suggested that the reoxidation temperature of metallic Pd occurred at ca. 650 � C, and methane conversion increased after reoxidation process [90]. Another reason causing the catalyst deactivation at low temperatures (e.g. below 450 � C) may be the overwhelming accumulation of hydroxyl group over the catalyst surface [190]. It hinders the migration and exchange of the active oxygen species between the catalyst support and active sites (PdO/Pd), resulting in the catalyst sintering. This is further confirmed by the water vapor effect on the methane conversion to be discussed in the section 4.4.

Effect of space velocity and residence time
Theoretically, the methane conversion presents an inverse proportion to the space velocity, as shown in Fig. 9. It was observed that the methane conversion at 400 � C increased from 72% to 100% over Au-Pd catalyst when decreasing the space velocity of reactants from 40 000 to 10 000 mL⋅g cat 1 ⋅h 1 [179] (cf. the detailed experimental data in Table 4).
A similar tendency over various catalysts has also been reported by other researchers [88,113]. This can be theoretically explained by the fact that the more accessibility of reactants (at longer residence times) over the active sites of catalyst favors the conversion of the adsorbed methane molecules. However, the side reactions and coke deposition are more likely to occur under long residence times [191]. The methane conversion is directly relevant to the intrinsic reaction kinetics and mass transfer. The internal mass transfer mostly depends on the catalyst properties (e.g. diffusion in the pore structure, size and volume). The external mass transfer from gas phase to the catalyst surface is limited by the residence time (i.e., the flow rate of reactants). As for the exothermic reaction, the effect of the temperature gradient on the Fig. 9. CH 4 conversion as a function of the space velocity or residence time. Fig. 10. CH 4 conversion as a function of temperature under different oxygento-methane molar ratios over Pt/Al 2 O 3 catalyst [19].
reaction rate should also be considered. It has been reported that at a certain temperature, a steady-state conversion rate could be reached with the increasing flow rate. After that, a further increase in the flow rate could not have any influence on the methane conversion. The mass transfer limitation is thus negligible at higher flow rates [192]. Moreover, in order to obtain sufficient reaction heat for heating purposes, higher reactant space velocity is required, but usually accompanied with a lower methane conversion. Hence, a compromise between the space velocity and the methane conversion should be reached for a certain application.

Effect of oxygen to methane molar ratio
The selectivity variation of products strongly depends on the molar ratio of oxygen to methane. Mouaddib et al. [193] investigated the effect of oxygen to methane molar ratio in the range of 4 to 0.66 over Pd/Al 2 O 3 catalyst. It was observed that CO was formed under oxygen deficient conditions (O 2 /CH 4 < 2) due to the side reaction of methane steam reforming. And CO became a main product when O 2 /CH 4 molar ratio reached 0.66 [193]. Similarly, Lee et al. [186] reported that CO was formed over Pt/Al 2 O 3 , Pd/Al 2 O 3 and Rh/Al 2 O 3 catalysts in the presence of oxygen deficient mixtures as feedstock. The selectivity of CO also depended on the reaction temperature. With the increasing reaction temperature, the CO selectivity constantly increased and became a main product under low O 2 /CH 4 molar ratio. But the methane conversion was independent to the presence of CO in the feedstock [186].
Burch et al. [19] investigated the effect of different molar ratios of oxygen to methane on Pt/Al 2 O 3 and Pd/Al 2 O 3 catalysts. Pt/Al 2 O 3 catalyst became more active from the oxygen-rich (O 2 /CH 4 ¼ 5:1) condition to methane rich (O 2 /CH 4 ¼ 1:1) condition, because the less oxidized platinum was more active than the oxidized platinum. As for Pt-based catalyst, Drozdov et al. [194] further explained that a weaker bond existed between oxygen and metallic platinum than that with platinum oxides. Thus, Pt-based catalyst is less active in oxygen-rich mixtures. On the contrary, as for Pd-based catalyst, it has been found that the oxygen is weakly bound to palladium oxides compared with metallic palladium [194]. Thus, the Pd-based catalyst is more active in oxygen-rich mixture [19,195].
The methane conversion at different oxygen-to-methane molar ratios does not display the same augmentation as the reaction temperature increases. More interestingly, the light-off phenomenon was observed to be significantly influenced by the oxygen to methane molar ratio and the resulted degree of reactant surface coverages [19]. It was found that the methane conversion over Pt/Al 2 O 3 increased with the decreasing oxygen-to-methane molar ratio (e.g., from 5:1 to 1:1 as shown in Fig. 10; cf. detailed experimental results listed in Table 4) at lower temperatures (ca. 300-425 � C, in terms of different Pt loading) [19]. At higher temperatures (ca. 450-550 � C), the methane conversion at such a molar ratio of 1:1 became lower than that at 2:1, due to the insufficient oxygen supply, and the light-off was observed in the latter case. This phenomenon could be explained by the fact that the favorable concentration of CH 4 and O 2 over the catalyst surface and/or the more significant local heat release greatly accelerate the reaction rate. However, no light-off was found under an oxygen-to-methane molar ratio of 5:1 because the excessively adsorbed oxygen resulted in a non-optimized surface coverage of methane, eventually hindering the light-off. The competitive adsorption between the adsorbed methane and oxygen species could affect the methane conversion. The lower methane conversion at the case of 5:1 ratio (Fig. 10) could be explained by the fact that more adsorbed oxygen species over the catalytic surface suppress the methane absorption due to the high methane adsorption energy [147,150].
Regarding the pollutant emissions, a lean reactant mixture is preferred in order to lower the emissions of CO, NO x and the unburned hydrocarbon [196]. It has been reported that CO was detected at a methane concentration higher than 9.5 vol% and carbon deposition occurred on catalyst surfaces at a methane concentration above 26.5 vol % [197].

Effect of (synthetic) natural gas composition
Besides methane, natural gas is commonly composed of varying amounts of higher alkanes (e.g., ethane, propane) and other species depending on the resources. In this section, the effect of H 2 S, H 2 O, NO and CO 2 on the catalytic methane conversion is discussed, more details  can be found in the references [179,198]. Noble metal catalysts are easily poisoned when exposed to natural gas containing sulfur compounds (Fig. 11a). For instance, the irreversible deactivation by SO 2 might be attributed to its occupation of the active sites, and/or the transformation of the highly active compound PdO on the catalyst surface to less active PdSO 3 or PdSO 4 . Similar results were also reported in the literature [199,200]. The highly challenging issue of the resistance to poisoning by sulfur containing compounds present in trace amounts in the natural gas (odorizer) should be fully addressed. Developing catalysts with improved resistance to sulfur poisoning and possible desulfurization pretreatment are possible measures.
The presence of water vapor acts as an inhibitor for CMC [165,202]. It was observed that the methane conversion over the Au-Pd/Co 3 O 4 catalyst dropped remarkably when water vapor was introduced, but this process is reversible when the water vapor is removed (Fig. 11b) [201]. Specifically, the hydroxyl group was formed by reaction of the chemisorbed oxygen species with the water vapor over the catalyst surface, preventing thereby the exchange of the oxygen species between the active sites and the support [190,203]. Moreover, the hydroxyl group also hindered the desorption of CO 2 and H 2 O over the catalyst surface due to the competitive adsorption. The kinetic study by Geng et al. [203] illustrated that the reaction order with respect to methane decreased from 1.07 to 0.86 after the water vapor addition, and the reaction order in water also decreased from 0.72 to 0.95 with the increasing water concentration. Thus, improving the catalyst stability in the presence of water vapor becomes one important challenge to deal with wet CMC. Recently, Toso et al. [93] reported that the solution combustion synthesis could improve the stability and water resistance of Pd-ceria and Pd-ceria-zirconia catalysts compared to the traditional impregnation method. Ciuparu et al. [204] also suggested that the influence of water vapor became insignificant at above 723 K. Sadokhina et al. [205] observed the enhanced activity with NO addition under wet conditions. This can be explained by the reaction between NO and hydroxyl species to form HNO 3 on the catalyst surface, compensating the inhibition effect of water addition.
The reversible inhibition effect of CO 2 addition is depicted in Fig. 11c. The negative influence of CO 2 addition on the methane conversion is ascribed to the accumulation of carbonate species on the catalyst surface, thus preventing a further adsorption of CH 4 and O 2 over the surface [201] (cf. detailed experimental data in Table 4).

Effect of operating pressure
Most published studies have performed the CMC at the atmospheric pressure. High working pressure conditions (up to 30 bar or higher) are primarily applied for gas turbine purpose.
The methane conversion decreased (from ~7.2% to ~3.5%) with the increasing pressure (from 5 bar to 15 bar) over Pd-Pt/Al 2 O 3 catalyst (cf. more details listed in Table 4) [103]. Moreover, it was reported that the effect of working pressure varied with the temperature [206]. At lower temperatures (500-600 � C), an increase in the pressure led to the decreased methane conversion and the lower combustion efficiency. In this case, the higher specific surface area was commonly required at high pressures so as to obtain a higher conversion. At higher temperatures (>700 � C), homogenous combustion takes place. A higher combustion efficiency could thereby be obtained with an increasing pressure [206,207], probably due to the increasing mass throughputs [208].

Types of catalytic reactors
In this section, some representative types of reactor used for CMC are discussed. Special focuses are laid on the geometry, the way of reaction heat recovery and the improvement of thermal efficiency by the enhancement of heat and mass transfer. Table 5 recapitulates some key facts of the relevant studies reported in the literature and Table 6 provides a comparison on the advantages and disadvantages of these reactors.

Fixed-bed reactor
A fixed-bed catalytic reactor is commonly made up of a cylindrical  tube with a certain amount of catalysts being fixed in certain locations inside the tube. The shapes of catalyst can be powder, spherical, cylindrical or randomly shaped pellets. This type of reactor is the most commonly used, suitable not only for catalyst activity test and kinetic study, but also for practical applications in the chemical and process industries. The advantages of fixed-bed reactors include easy operation, low cost, high catalyst spatial density, etc. Poor temperature distribution, low surface area, and high-pressure drop are their main disadvantages. Besides being used individually as a single-stage reactor, several fixed-bed reactor modules may be combined to form a multi-stage reactor system. Different catalysts may be used for each stage. A twostage combustor combining a conventional flame reactor with a catalytic reactor has been proposed by Sadamori et al. [206,233], with the purpose of improving the combustion efficiency and further reducing the gas exhaust emission. In their study, three monolithic catalysts with Pd and/or Pt based catalysts were loaded in the catalytic stage. Combustion efficiency higher than 99.5% and the NO x emission less than 2 ppm (80% reduction) could be achieved by controlling the inlet temperature between 500 and 700 � C. In terms of thermal behavior, the conventional stationary fixed-bed reactor with a separated heat exchanger (Fig. 12a) has the problem such that the reaction heat is difficult to be fully utilized. Multifunctional or extended autothermal reactors having higher interfacial surface areas and more efficient heat transfer thus become more attractive, especially for strongly or moderately exothermic reactions [234]. This concept combines catalytic reaction with an effective thermal management (i.e. heat removal or addition) in an integrated and compact device (Fig. 12b), so as to largely reduce the heat losses. Compared to the ex-situ configuration, lower CO and NO x emissions and higher heat power could be achieved for the multifunctional reactor with an integrated heat exchanger [31]. Detailed information about the autothermal fixed-bed reactor can be found in the literature [235][236][237].
In order to further enhance the heat transfer capability, the fixed-bed reactor can be operated in the reversal flow mode [238,239], where the feeds are alternately introduced from either end of the reactor (Fig. 12c). The continuous heat extraction is a viable way to provide necessary amount of heat in the reactor, and to maintain the catalyst activity without overheating. A high amount of heat withdrawal could result in a high instability of the reaction due to a remarkably decreased temperature of the reactor. Moreover, a periodic steady state in temperature can be achieved by switching the direction of reactant flow. The reversal time is one of the significant factors not only to stabilize the reactor and the heat exchanger but also to maintain a stable temperature in the catalytic section. A longer reversal time is commonly accompanied with a higher heat loss. On the contrary, a shorter reversal time can result in a lower heat generation and a reduced methane conversion [240,241]. Many simulations studied the relative position of the catalyst bed and the heat exchanger as well as the flow pattern in order to find out the most efficient way to recover heat from reversal-flow reactors [242][243][244]. Gosiewski et al. [242] found that a higher heat recovery efficiency can be approached by the hot gas withdrawal (Fig. 12d) than by the central heat exchange (Fig. 12c). Detail experimental data are listed in Table 5. Their results showed that more than 70% of heat can be recovered when withdrawing only around 30% of the hot gas. Furthermore, Hunter et al. [245] studied different patterns of heat recovery with respect to the heat withdrawal location (central or end) and the possibilities of returning the cooled gas to exchange heat. It was  [251]; (c) folded sheet reactor with corrugated spacers as catalyst carriers [44]; (d) pilot-scale multifunctional folded sheet reactor [252].
found that a higher heat recovery efficiency could be obtained by positioning the heat exchanger at the two side ends of the catalytic bed without any returned cold gas (Fig. 12e). Hence, a high thermal recovery efficiency and the best methane conversion may be achieved by considering both the flow reversal mode (e.g. the hot gas withdrawal at the end of the catalyst bed) and the reversal time. More description of the flow-reversal reactors can be found in the references [236,242,[246][247][248].

Wall-coated reactor
Conventional fixed-bed reactors have the drawback of limited heat transfer rate due to the low surface-to-volume ratio. On the contrary, the wall-coated reactors have features like lower pressure drop and higher surface area by depositing the catalyst layer on the reactor wall surface, so as to augment the catalyst contact surface area and to enhance the mass/heat transfer. Hence, this type of reactor is more suitable to be employed for coupling exothermic reactions with endothermic reactions. A lower temperature difference and a shorter distance between the heat source (exothermic reaction) and heat sink (endothermic reaction) are beneficial to increase the heat exchange efficiency. The socalled wall-coated reactor could be further divided into (folded) platetype, tubular-type, micro/mini channel plate-type and monolithic reactors. Fig. 13a shows a basic plate-type reactor coupling CMC with methane steam reforming reaction. The two reactions happen on the opposite sides of the same plate with overlapped temperature profiles. The two flow modes (i.e., co-current and counter-current) were intensively investigated [249]. The hot spots could be avoided by optimizing the overlapping zone of reactions with the proper co-current arrangement. However, the complete overlapping is not able to be achieved by utilizing the counter-current flow mode. Regarding the distance between plates (thickness: 0.5 mm), it has been reported that no significant effect on the catalytic performance was observed when the plate distance varied between 1 and 4 mm [250].

(Folded) plate-type reactor
A folded sheet structure reactor for the coupled CMC and steam reforming reaction was investigated by Polman et al. [251]. A relatively simple design with two separated chambers was formed by the folded metal sheet, as shown in Fig. 13b. In order to avoid the high heat loss and to enhance the heat transfer, a novel folded sheet reactor with rectangular adjacent channels was proposed by Kolios et al. [44]. The spacers with rectangular ducts in between are responsible to support the whole structure, providing additional effective heat exchange areas as well (increased by a factor of 3). The gas distributor with channels is located on the right side of the entire reactor, as shown in Fig. 13c [44]. Fig. 13d shows a pilot-scale multifunctional folded sheet reactor combining methane combustion with methanol reforming [252].

Coated tubular reactor
A novel tube-type reactor (i.d. 25 mm, o.d. 38.5 mm) with external fins (fin height 14.3 mm) was proposed by Seo et al. (Fig. 14a) [32]. The external surfaces of tube with fins were coated with Pd/ZrO 2 catalysts for CMC while the cold air as the heat transfer fluid (HTF) was circulated inside the tube to remove the released reaction heat. From practical point of view, this compact design with a high surface area is capable of achieving 100% methane conversion and a high heat transfer rate under high methane flow rate conditions. Moreover, nearly 100% methane conversion could be obtained by 16 finned-tube reactors even under high gas velocity conditions (0.12-0.31 m s 1 ). However, the same conversion for 10 finned-tube reactors is required under a much lower velocity (<0.11 m s 1 ). Moreover, it has been reported that the conversion at 400 � C decreased from 88% to 70% with the increasing Fig. 14. Coated-tubular reactor. (a) tube-fin geometry and laboratory prototype [32]; (b) external surface coated with Ni-Cr catalyst, and internal surface coated with Ni catalyst [121].
catalyst thickness from 1 mm to 3 mm (cf. more details shown in Table 5). A washcoated finned tubular reactors with thinner catalyst thickness (e.g., <1 mm) was more favorable to reach a high conversion.
Ismagilov et al. [121] investigated a tubular reactor (i.d. 18 mm, o.d. 20 mm) with coated metal foams both on the external (Ni-Cr) and internal (Ni) tube surfaces (Fig. 14b). A stable catalytic performance and uniform temperature distribution in the reactor could be obtained by optimizing the gas mixture composition and the catalyst thickness. Moreover, the thickness of catalytic layer in the range of 4-5 mm was favorable to reach a more stable combustion than that of 2.5 mm. Thus, a suitable surface area and catalyst thickness are required not only for increasing the diffusion rate of reactants, but also for higher catalytic combustion efficiency.
Unlike the traditional straight combustor, Yan et al. [253] numerically investigated three different designs of the micro tube combustors for CMC. The improved combustion efficiency can be obtained from the design of multi-step separated baffles (two groups of three separated zones). One of the key advantages of this design is that the separated baffles provided chances for the premixed methane and air to enter the reaction zone from different locations, which enhanced the combustion efficiency and heat recirculation.

Micro/mini channel plate-type reactor
Micro/mini channel reactor has attracted more attention in the past several decades [254][255][256]. The high surface area of microchannel reactors as well as excellent mass/heat transfer presents great benefits for the catalytic performance [257][258][259]. Thus, highly exothermic reactions are better handled in the microchannel reactor, due to the suppression of the hot spot formation. However, it is worth noting that the high surface area may also result in the thermal quenching problem due to the high  [118]; (c) Pt/γ-Al 2 O 3 catalyst on multichannel reactor for CMC [120]; (d) coupling steam reforming over a microchannel reactor [262]; (e) integration of multi-plates coupling with methane steam reforming over a microchannel reactor [119].
heat loss if the reactor is not properly insulated. An extra heat source may have to be provided to the microchannel reactor when the released heat from CMC is insufficient to compensate the heat loss to maintain a continuous combustion in practice [258,260,261]. Fig. 15a shows a basic geometry of microreactor with washcoated catalyst on multiple straight channels, where the CMC reaction takes place. O'Connell et al. [118] investigated methane combustion on a microstructured reactor (51 cm length � 14 cm width) with the microchannels (500 μm � 250 μm, 14 channels in total) over washcoated Pt-W/Mo-Al 2 O 3 catalyst, as shown in Fig. 15b. A methane conversion of 50% has been obtained at 493 � C under the total flow rate of 107 mL min 1 . The CMC was experimentally investigated by He et al. [120] in a parallel microchannels reactor (317.5 mm length � 50 mm width � 3 mm thickness) over washcoat Pt/γ-Al 2 O 3 catalyst, as shown in Fig. 15c. A methane conversion of 95.75% could be obtained at 450 � C and 110 mL min 1 (at a residence time of 14.41 s). A compact microchannel reactor (15 cm � 3.9 cm � 1.5 cm) was used for the coupled CMC with methane steam reforming, as shown in Fig. 15d [262]. Each plate consists of 5 parallel straight channels (10 cm � 0.5 cm � 0.5 cm). The methane catalytic reactor with Pt-Sn/Al 2 O 3 catalyst is located on the two sides of the steam reforming reactor with Ni/CaAl 2 O 4 catalyst. The heat released from the CMC was provided for steam reforming reaction to produce hydrogen. The improved heat efficiency of 67% and methane conversion of 96% were obtained under optimized feed ratio (1.5) of combustion to reforming and at 700 � C. Enough hydrogen was expected to be generated to operate a 30 W fuel cell. Mundhwa et al. [119,263] proposed a microstructured reactor design (Fig. 15e) composed of two methane combustion reactors with segmented channels (1 mm � 5 cm � 20 μm, 20 channels) and two reforming reactor without channels. The plates were stacked alternatively one above another to form the autothermal microstructured reactor. Washcoated Pt/Al 2 O 3 catalysts were applied in methane combustion microchannels while Ni/Al 2 O 3 catalysts were coated on the steam reforming side. Based on this design, about 7-8% less reactants and 70% less catalysts were required for methane combustion to power a 1 kW fuel cell. A better methane conversion and heat transfer efficiency (in terms of better temperature distribution) could be obtained under the co-current flow mode (Fig. 15e). Contrarily, the counter-current flow mode generated undesirable high temperatures, resulting in the degraded catalyst life-time.
In order to improve the stability of methane combustion, Nui et al. [264] proposed different trapezoidal bluff bodies in the microchannel reactor. The numerical results presented that the combustion recirculation zone was broadened due to the formation of vortex by increasing the blockage ratio of bluff bodies.
The thickness of the catalyst layer has a significant impact on the catalytic performance. The increased thickness of the catalyst layer usually results in a decrease in the methane conversion due to the increased internal mass transfer limitation [121]. However, Rodrigues et al. [261] reported that a higher catalytic activity could be obtained by the thicker catalyst film. The porous catalyst prepared by the electrodeposited method offered the reactants easier access to the inner surface.

Monolithic reactor
Monolithic reactors with open structures present a low pressure drop, high surface-to-volume ratio, high mass/heat transfer rates and high thermal stability. Thus, this type of reactor is widely applied to gas turbines and automobiles for power generation and/or for pollutant emission purification [265,266]. Numerous types of interconnected channels (e.g. square/round/triangle, as shown in Fig. 16a) and various types of substrate (e.g., honeycombs, foams or fibers) are oriented to different applications [267,268]. A higher monolithic surface area can be obtained by triangular interconnected channels. It was reported that monolith with a high specific surface area (400 cells⋅cm 2 ) presented enhanced mass and heat transfer characteristics compared to the lower one (60 cells⋅cm 2 ) [269]. Ceramic and metallic substrates are the most common materials used in studies due to the high mechanical stability and the low thermal expansion.
Usually, the catalyst was loaded on the upper section of the reactor and the heat exchanger was located on the bottom. The thermal efficiency was found to be between 20% and 30% for such system [270]. In order to better utilize the heat released from reaction, a two-stage monolith prototype with two heat exchangers was designed. This compact design of premixed burner comprises two honeycombs in between the two heat exchangers on the opposite side, similar model as shown in Fig. 16b [29]. The thermal efficiency for this design could reach 101.1%, implying an effective way to satisfactorily recover the released reaction heat [33]. Fig. 16c shows a monolith and folded wall reactor [35] for the CMC coupled with the steam reforming reaction, the inner pipe for CMC reaction and the annular part for steam reforming  [29]; (c) folded-wall monolithic reactor coupled with steam reforming reaction [35]. reaction (co-flow model). The temperature difference can be managed by adjusting the feed ratio of methane steam reforming to combustion. Furthermore, the heat transfer can be obviously improved by proper channel arrangement or catalyst distribution.

Membrane reactor
A membrane reactor commonly comprises a membrane coated with catalysts or as a barrier that only allows certain component(s) to pass through. Lanthanum cobaltite perovskite ceramic is one of the most widely used materials for membranes. Other materials with improved properties, such as thin dual-phase membranes, ceramic metal dualphase membranes and ion transport membranes are also very promising for enhanced oxygen permeation [271]. The increasing motivation for their industrial applications is the reduction of CO 2 emissions from methane combustion.
The basic principle of membrane reactor for CMC is shown in Fig. 17a. CH 4 and air pass through the membrane, and the permeated O 2 reacts with CH 4 . CO 2 in the products can be successfully separated and captured. It has been reported that the nitrogen with a purity of 98-99% could be produced and the system remained stable over 120 h [272].The membrane reactor used for CMC could achieve a high methane conversion mainly by varying the partial pressure of oxygen permeation. However, the high costs of membrane reactors potentially limit their industrial applications.
The membrane reaction efficiency is affected by the feed flow rate, the temperature, and the permeability of oxygen. Fig. 17b depicts the oxygen transport over the membrane reactor [273]. The oxygen is firstly adsorbed on the surface of the membrane. The charged oxygen vacancy (O 2 ) is diffused to the other side, due to the formation of the chemical potential gradient across the membrane. The electrons on the other side are transferred in a reversed-direction so as to compensate for the oxygen vacancies. The results of Falkenstein et al. [273] present that the oxygen permeation flux increased with the increasing methane flow rate and reaction temperature. However, the oxygen permeation flux remained fairly constant at high methane flow rates (e.g. >20 mL min 1 ), probably due to the limitation of the effective membrane surface area. CO 2 selectivity was thus not significantly varied under higher flow rates in this case. This is in agreement with the results reported by Tan et al. [276], that is, the high methane and air flow rates resulted in a lower oxygen permeation and reaction rate over La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3-α hollow fiber membrane reactor. Moreover, the membrane, coated with  [273]; (c) ion transport membrane reactor [274]; (d) multi-channel membrane reactor coupled with steam reforming reaction [275]. platinum catalyst, showed that the membrane reactor effectively facilitated the oxygen permeation and improved the methane conversion, owing to the reduced oxygen permeation resistance [276].
Habib et al. [274] proposed a two-pass ion transport membrane reactor (ITM) for CMC, as shown in Fig. 17c. The first pass is responsible for oxygen permeation, where methane combustion and partial heat exchange between the mixture gas and the water also happened. The second pass is for further permeation. Moreover, the counter-current flow configuration in this ITM provided high methane conversion than that with the co-current flow configuration, owing to the higher oxygen permeation in the first case. In fact, the increased oxygen partial pressure and the accumulated oxygen flux may lead to the reduced oxygen permeation under the co-current flow mode. Moreover, the membrane reactors have also been applied for coupling with steam reforming [277] (Fig. 17d) or ammonia decomposition [278] reactions. An effective way to improve the performance for this system is to increase the membrane effectiveness and to reduce the membrane thickness.

Fluidized bed reactor
Fluidized bed reactor is a kind of typical catalytic reactors in which solid catalysts (frequently with a diameter of 10-300 μm) are fluidized during the reaction. It is capable of handling a larger amount of reactants or catalysts owing to the large reactor size, and the feed flow rate is required to suspend the catalysts. A porous plate as gas distributor is responsible for supporting the material in the fluidized bed, as shown in Fig. 18. The high gas flow results in an efficient contact between the reactants and the catalysts, leading to the enhanced heat and mass transfer rates on the catalyst surfaces. As a result, the non-uniform temperature distribution that commonly exists in fixed-bed reactor could be avoided. One of its main disadvantages is the great mass loss of the catalyst due to the in-bed attrition after long-term operation.
The experiment measurement by Yang et al. [279] illustrated that the methane conversion increased with the increasing temperature in the fluidized bed reactor, and decreased with the increasing methane inlet concentration. The methane conversion was also reported to decrease with the increasing gas velocity [228]. The fluctuation of temperature and the variation of mixture concentration may occur, due to the intensive motion of solid particles rising up and falling back [280]. Meanwhile, the enhanced mass and heat transfer can be realized between the reactants and the catalyst particles, due to the strong oscillations in the fluidized bed. Furthermore, the kinetic experiments conducted by Yang et al. [279] in the fluidized bed confirmed that the reaction was only controlled by the kinetics at a bed temperature below 450 � C, and by the mass transfer and kinetics together at temperatures above. Dubinina et al. [281] reported CuO/Al 2 O 3 MgO-Cr 2 O 3 catalyst to be the one of the most promising catalysts in the fluidized bed reactor for CMC.

Summary and prospect
This work provides an extensive review on the CMC. Different catalysts, mechanisms, effect of operational parameters and reactor types are discussed. The main conclusions may be summarized as follows.
� Noble metal catalysts with high activity are favorable for CMC at low temperatures (<700 � C). The bi-metallic catalysts have a better catalytic activity due to the more active sites and electronic synergy effects. The hexaaluminate and perovskite mixed metal-oxide catalysts with different microstructured features exhibit a high thermal stability, thus more suitable for high temperature applications (700-1300 � C). � The Mars-van Krevelen mechanism is more widely accepted than the Langmuir-Hinselwood and Eley-Rideal mechanisms. It has been observed by in-situ (spectroscopic) technologies that the adsorbed oxygen species in PdO catalyst are responsible for the methane oxidation, rather than oxygen in gas phase. � The light-off temperature is mainly influenced by the operating temperature and the oxygen to methane molar ratio. It varies depending on different catalyst properties. The optimized O 2 /CH 4 molar ratio is beneficial for a full methane conversion due to the optimized coverage of the adsorbed mixtures over the surface. � The natural gas containing the sulfur compound, carbon dioxide and water vapor can suppress the catalytic activity due to the competitive adsorption and the blockage of active sites. The deactivation due to water and carbon dioxide is reversible whereas the sulfur poisoning is irreversible. � The reversal flow mode for fixed-bed reactors with hot gas withdrawal at the end of the bed and without the return of cold gas is highly recommended in order to maximize the heat recovery and methane conversion. � The recent development of mini/microstructured reactor with compact design has been broadly investigated for coupling the CMC with endothermic reaction (methane steam reforming) at different flow modes. The co-current flow mode presents a better thermal efficiency than the counter-current flow one.
Some scientific and technological barriers remain to be overcome for the widespread industrial application of CMC, which are also the key issues and challenges of the current research and development: � To improve the thermal stability and to prolong the lifetime of noble metal catalysts by avoiding the possible sintering at high temperature levels. � To improve the current understanding into CMC reaction mechanisms and the corresponding kinetic models, including the deactivation mechanism (e.g. CMC with the presence of water and sulfides etc.). � To lower the light-off temperature and to maintain the long-term high catalytic activity of hexaaluminate and perovskite catalysts. � To develop effective desulfurization pre-treatment and measures in order to extend the catalyst lifetime. � To improve mechanical and chemical stabilities of the catalyst coating, and to cope with the deactivation issue by proposing replacement and/or regeneration methods. � To further enhance the heat and mass transfer in compact and integrated catalytic reactor-heat exchangers.

Declaration of competing interest
None.

Acknowledgement
This work was supported by the Region Pays de la Loire (Chaire Connect Talent ODE) from French side; and the Ubbo-Emmius Fund 2015 from University of Groningen and the start-up package in the area of green chemistry and technology (for Jun Yue) from the Netherlands' side. For their permission to reproduce the figures, we thank the ACS (Figs. 4a and 6), John Wiley and Sons (Figs. 4b and 15c) and Elsevier (Figs. 4c, 5f and 5g, h, 7 and 11 and 12a and b, 13b and c, Figs. 13d, 14a  and 14b, 15b, 15c and 15d, 16b and 16c, 17b and 17c, d).