Catalytic conversion of ethane to valuable products through non-oxidative dehydrogenation and dehydroaromatization.

Chemical utilization of ethane to produce valuable chemicals has become especially attractive since the expanded utilization of shale gas in the United States and associated petroleum gas in the Middle East. Catalytic conversion to ethylene and aromatic hydrocarbons through non-oxidative dehydrogenation and dehydroaromatization of ethane (EDH and EDA) are potentially beneficial technologies because of their high selectivity to products. The former represents an attractive alternative to conventional thermal cracking of ethane. The latter can produce valuable aromatic hydrocarbons from a cheap feedstock. Nevertheless, further progress in catalytic science and technology is indispensable to implement these processes beneficially. This review summarizes progress that has been achieved with non-oxidative EDH and EDA in terms of the nature of active sites and reaction mechanisms. Briefly, platinum-, chromium- and gallium-based catalysts have been introduced mainly for EDH, including effects of carbon dioxide co-feeding. Efforts to use EDA have emphasized zinc-modified MFI zeolite catalysts. Finally, some avenues for development of catalytic science and technology for ethane conversion are summarized. This journal is © The Royal Society of Chemistry.

Introduction: chemical utilization of ethane

Chemical utilization of ethane (C2H6) has had great effects on the modern petroleum industry. Inexpensive ethane derived from shale gas and associated petroleum gases is widely available, respectively, in North America and in the Middle East.[1] Ethane is converted to ethylene (C2H4) by steam cracking without a catalyst in an ethane cracker. Ethylene is the most fundamental chemicals in petrochemical industry: 149.7 million tons of ethylene were produced in 2017 worldwide[2] to synthesize polyethylene, ethylene oxide, styrene, and other ethylene derivatives.

Ethylene production from ethane is far superior in terms of cost to production from naphtha, which is a conventional feedstock for ethylene production by the steam cracking. According to the literature,[1] costs of ethane were €46.28 and €145.96 per ton in Saudi Arabia and the United States, respectively, while those of naphtha were €708.56 and €669.58 per ton. As shown in Fig. 1, the global demand for ethylene will be expanded to 182.5 million tons in 2023 worldwide[2] due to the global increase in population and economic growth. Consequently, the feedstock for ethylene production will be shifted much from naphtha to ethane.

Fig. 1

History and projection of global demand for ethylene, benzene, toluene, and xylenes. Data originated from ref. 2.

However, ethane cracking is an energy-intensive process because it necessitates high reaction temperatures (>1073 K). Therefore, ethylene production is favorably conducted at low temperatures to lower the energy consumption and carbon dioxide (CO2) emission. Furthermore, development of ethylene production from ethane instead of naphtha has led to another difficulty. In addition to ethylene, fundamental chemicals including propylene, 1,3-butadiene, and aromatic hydrocarbons are produced by the steam cracking of naphtha. In contrast, these important building blocks are not obtained when using the ethane cracker.[1] Therefore, other methods to produce fundamental chemicals aside from ethylene must be used. For instance, propylene must then be produced through dehydrogenation of propane,[3] methanol to propylene,[4,5] or olefin metathesis of ethylene and 2-butenes.[6]

Catalytic conversion of ethane to ethylene and aromatic hydrocarbons is anticipated as key technologies to address those needs. Non-oxidative catalytic ethane dehydrogenation (EDH) over oxide or supported-metal catalysts has been investigated widely to achieve environmentally friendly ethylene production. As is the case in propylene production, aromatics including benzene, toluene, and xylenes (BTX), must be produced from feedstocks other than naphtha. BTX are important chemicals used for plastics production and pharmaceuticals synthesis. The global demand for BTX will increase from 111.2 million tons (BTX total) in 2017 to 141.0 million tons in 2023 worldwide (Fig. 1).[2] Catalytic ethane dehydroaromatization (EDA) is an attractive alternative to steam cracking of naphtha because valuable aromatics are produced from inexpensive ethane. In addition, EDA can be performed at relatively low temperatures with high BTX selectivity. Therefore, EDA is a potentially promising process for efficient BTX production from natural gas resources. Catalytic EDH and EDA under non-oxidative conditions are summarized in this review because some reviews of oxidative EDH have been published earlier.[7,8] Initially, steam cracking of ethane is introduced briefly based on its reaction mechanism. Next, details of non-oxidative EDH and EDA are described. Finally, perspectives for catalytic ethane conversion are discussed.

Steam cracking of ethane

Radical chain reaction mechanism

Steam cracking of ethane is a non-catalytic process for ethane conversion to ethylene through pyrolysis. Before pyrolysis, ethane is mixed with effluent steam to supply reaction heat and to suppress carbonaceous deposits on cracking coils, where the reaction proceeds at ca. 1073 K. Ethane pyrolysis proceeds through a radical chain mechanism first proposed by Rice and Herzfeld.[9] This mechanism comprises various elementary reactions such as homolysis, hydrogen abstraction, hydrogen transfer, and recombination.[10] The reaction is initiated by homolytic cleavage of the C–C bond in an ethane molecule because enthalpy and the activation energy for the C–C bond dissociation is lower than that of the C–H bond dissociation.[11-14] The C–C bond cleavage in an ethane molecule leads to the formation of two methyl radicals as shown below.

The methyl radical induces the subsequent propagation steps.

A hydrogen atom in an ethane molecule is first abstracted by the methyl radical, resulting in the formation of a methane and an ethyl radical (reaction (2)). Subsequent hydrogen transfer in the ethyl radical engenders the formation of ethylene and methane molecules (reaction (3)). The methyl radical also attacks ethane and ethylene (reactions (4) and (5)), resulting in the formation of propane and propylene (reactions (6) and (7)). Heavy hydrocarbons are obtainable through analogous reactions. Furthermore, termination steps proceed through reactions between two radicals.

Based on the temperature programmed reaction, pyrolysis of ethane to ethylene initiates at higher than 873 K.[15] Under a practical situation, many reactions proceed in addition to the reactions (1)–(8). Sundaram and Froment considered 49 reactions to simulate the product distribution and concentration of radicals.[13] The cracking coil is a tubular reactor. Therefore, the temperature gradient along the radial direction is an important factor to be incorporated into calculation of the product distribution and concentration of the radicals. Marin and co-workers reported that the methyl radical concentration is higher near the cracking coil wall than at its center because of high temperatures near the wall, which are the result of endothermicity.[14] This phenomenon results in methane formation in great quantities if the simulation is based on a two-dimensional model.

Formation of by-products such as methane, C3, and heavier hydrocarbons is unavoidable in the case of ethylene production through the radical chain mechanism because the reaction initiates from the C–C bond cleavage; it also terminates with methane formation. Particularly, the formation of methane and coke, which are thermodynamically stable, is remarkable at high conversion (temperature) levels.[16,17] This remarkable formation decreases the efficiency for ethylene production.

Coke formation and its suppression

Coke formation on the internal wall of the cracking coils induces severe difficulties such as a pressure drop and a decrease in heat flux to the reactor.[18] The latter particularly decreases the ethylene yield because of a kinetic restriction by decreased temperature. To maintain high ethylene yields, the outlet gas temperature is generally operated as constant, which indicates that the cracking coil is heated much more than usual to supply sufficient heat flux, thereby creating a higher temperature of the internal wall than the reaction temperature. Therefore, taking the thermal resistance of the cracking coil and the tolerable pressure drop into consideration, periodic removal of the carbonaceous deposits (decoking) is conducted every 20–60 days depending on operational conditions.[17,19] In general, coke formation mechanisms of three types have been proposed: the heterogeneous catalytic mechanism, the heterogeneous radical mechanism, and the homogeneous condensation mechanism.[20] The heterogeneous catalytic mechanism is based on coke formation on the surface of the cracking coils. The iron (Fe) and nickel (Ni) included in iron–chromium–nickel (Fe–Cr–Ni) alloys, which are used as cracking coil materials, catalytically promote coke formation on their surface. This phenomenon is observed at the initial stage of the reaction until the cracking coil surface is covered with carbonaceous deposits, indicating that the coke formation rate decreases gradually and reaches an asymptotic value.[17,21] The heterogeneous radical mechanism is based on the growth of the carbonaceous deposits through reactions with the radical species. Wauters and Marin proposed that this mechanism comprises five steps: hydrogen abstraction, substitution, addition by radicals, addition to olefins, and cyclization.[22] Once the coke forms on the wall, the growth of coke proceeds throughout the reaction. The carbonaceous deposits can also be formed through the homogeneous condensation mechanism as a result of the chain reaction. Therefore, the coke formation via this route is concomitant with the pyrolysis of ethane.

From a metallurgic perspective, coating the surface with inert oxides is an effective means of mitigating the coke formation. In principle, aluminum (Al) or manganese (Mn) is included in the Fe–Cr–Ni alloys. After oxidation of the alloy at high temperatures (1323 K for Al; 1023 K for Mn), alumina (Al2O3) or manganese chromate (MnCr2O4), which has high coking suppression capability, is deposited on the surface.[20,23] Recently, coating a catalyst for the water gas reaction has been developed to achieve operations without coke formation.

For instance, BASF developed the CAMOL® process to mitigate coke formation. In this case, tungsten-containing oxides are used as catalysts.[24] General Electric also developed gasification catalysts that include perovskite oxides containing cerium (Ce) and alkaline-earth metals.[25] However, introduction of the catalysts to cracking coils remains an immature technology. Further research and development are necessary in the domain of catalytic science.

Dehydrogenation of ethane under non-oxidative conditions

Catalytic ethane dehydrogenation, denoted as EDH, enables ethylene production at lower temperatures than those of pyrolysis of ethane.

However, the reaction is endothermic. Therefore, high reaction temperatures must achieve higher equilibrium conversion, as portrayed in Fig. 2. Furthermore, low pressure of ethane results in high equilibrium conversion based on Le Chaterlier's principle.

Fig. 2

Temperature dependence of equilibrium conversion of EDH at various ethane pressures.

The EDH reaction mechanism is intrinsically unlike pyrolysis of ethane: catalysts should have ability of selective C–H bond dissociation. Side reactions such as hydrogenolysis of ethane and decomposition of ethane and ethylene to coke must be prohibited to achieve high and stable selectivity to ethylene. In this chapter, representative catalysts for EDH are described in terms of their active sites and reaction mechanisms.

Platinum-based catalysts

For dehydrogenation of alkanes (ethane, propane and butanes), platinum (Pt)-based catalysts have been used widely: one example is known as the Oleflex process.[26] In this process, the dehydrogenation reaction are performed by multiple fluidized bed reactors. The spent catalysts are continuously regenerated after the reaction. The regeneration processes are composed of removing carbonaceous deposits and re-dispersion of the Pt particles with chlorine treatment.[27] The regenerated catalysts are, then, reused for the reaction. Since dehydrogenation reactions are endothermic, the Pt catalysts are used at high temperatures, which accelerate catalyst deactivations because of coke formation and agglomeration. Therefore, supported-Pt catalysts are modified with promoters such as tin (Sn) for improving their catalytic performance. In this section, we first introduce overview of supported-Pt catalysts. Subsequently, effects of the promoters on the nature of active sites are described in detail.

Overview of supported-platinum catalysts

A typical catalyst used for EDH is Pt/Al2O3. Highly dispersed Pt particles can exist on γ-Al2O3 by virtue of the presence of penta-coordinated Al, which stabilizes the Pt particles.[28,29] Calcined layered double hydroxides (hydrotalcite) are alternative candidates for use as a support material.[30-32] Particularly, Mg(Al)O hydrotalcite includes Al in its structure. The surface basicity can be expected to contribute to the suppression of coke formation, which results in deactivation of the catalyst. For dehydrogenation of propane, Pt/Mg(Al)O showed better performance than conventional alumina-supported catalysts, but coke formation is unavoidable.[33]

Surface active sites of two types are known to exist for Pt nanoparticles: coordinatively saturated Pt located at terrace sites and coordinatively unsaturated Pt (cus-Pt) such as step and edge sites. Both Pt sites are active for the C–H bond cleavage. However, the latter sites easily induce side reactions such as hydrogenolysis of ethane[34,35] and subsequent dehydrogenation of ethylene to coke[36,37] as shown in Fig. 3. However, Wu et al. calculated the turnover frequency (TOF) of ethane consumption using Pt/Mg(Al)O with different Pt particle sizes.[38] The TOF increased concomitantly with increased Pt particle size, whereas selectivity to ethylene was nearly constant, indicating that EDH proceeds at the terrace sites. This trend is not the case for propane dehydrogenation, where TOF decreases concomitantly with increased Pt particle size.[39]

Fig. 3

Schematic image of reactivity of ethane on Pt particle. EDH proceeds on the Pt terrace sites, whereas coke formation and hydrogenolysis of ethane proceed on the “cus-Pt” sites circled in red.

The supported-Pt catalysts deactivate in two ways: coke formation and sintering (agglomeration). The carbonaceous deposits formed during the reaction cover the Pt surface and inhibit the access of reactants to active sites. To remove the carbon deposit, the spent catalysts are regenerated under oxidative atmospheres. In addition, the sintering of the Pt nanoparticles decreases the Pt surface area and the number of active sites. In contrast with dehydrogenation of propane and butanes, EDH necessitates the use of high reaction temperatures (>873 K), at which sintering of the Pt nanoparticles is accelerated. Consequently, Pt catalysts are exposed to severe environments: reductive (dehydrogenation) and oxidative (regeneration) atmospheres at high temperatures.

Co-feeding hydrogen (H2) with ethane in the reaction system can inhibit coke formation. Galvita et al. demonstrated that positive effects can be produced for co-feeding H2 on catalytic stability, but selectivity to ethylene decreases because of methane formation through hydrogenolysis.[40] Based on density functional theory (DFT) calculations,[41-43] high hydrogen coverage on the Pt surface inhibits the formation of coke precursors such as ethylidyne (CH3C) and methylidyne (CH) species. In addition to suppression of the coke precursors, H2 addition to the reaction system promotes ethylene formation under the kinetic conditions.[40] Hansen et al. applied DFT calculations to find that H2 addition stabilizes the adsorbed ethylene and ethyl groups, leading not only to suppression of the coke precursors but to the facilitation of ethylene formation.[43]

Their practical use requires realization of the highly stable and dispersed Pt nanoparticles with coke formation in small amounts. To overcome these shortcomings, Pt catalysts are modified with promoters including boron (B), Al, copper (Cu), zinc (Zn), gallium (Ga), indium (In), and Sn.[38,40,44-58] Based on theoretical studies, silicon (Si) and germanium (Ge) are also good candidates for use as promoters.[59,60] Irrespective of the elements, the promoter effects are fundamentally classifiable into two types: geometric and electronic. In the following subsection, we describe the respective effects on the catalytic performance. Details of preparation methods, characterizations, and other applications of Pt alloy catalysts are reviewed elsewhere.[61,62]

Geometric effects of promoters on platinum

Addition of the promoters to the supported Pt catalysts strongly influences the Pt active site structures. Geometric effects of various kinds are reported, such as surface modification, alloy formation, and re-dispersion. Among the promoters, we mainly describe Pt–Sn bimetallic catalysts because the traditional dehydrogenation catalysts are investigated thoroughly in detail.

The major role of the promoter is alloy formation. Addition of the promoters induces formation of Pt-based alloys or intermetallic compounds, as shown in Fig. 4(a). In addition to Sn,[38,40,56-58] other post-transition metals such as Zn, Ga, and In are used as promoters.[46-53] Fundamentally, Pt-based alloys are prepared by addition of the promoter and by subsequent reductive treatment. For Mg(Al)O support, some fraction of the Al can be replaced with Ga or In.[46,48,49,52] During the pre-reduction step, the dopants move from the partially reduced support to Pt, resulting in Pt-based alloy formation.

Fig. 4

Typical geometric effects of Sn addition to Pt: (a) alloy formation and (b) surface modification. Face centered cubic and hexagonal close packing are denoted respectively as fcc and hcp.

Catalytic performance, particularly selectivity to ethylene and stability, is enhanced by virtue of alloy formation. Alloy formation not only decreases the number of “cus-Pt” through replacement of “cus-Pt” atoms with the promoter: it also varies the surface composition (ensemble) of terrace Pt sites. Celik and co-workers used DFT calculations to elucidate coke formation from ethane on Pt(111) and Pt–Sn(111), representative terrace sites.[42] They constructed the alloy models by replacing the surface Pt atoms with Sn atoms, indicating that the model of PtSn has a face centered cubic (fcc) structure. They reported that CH3C (coke precursor) is formed at three-fold Pt sites of Pt(111). For the PtSn(111) surface, where no three-fold Pt exists, CH3C is unstable, leading to suppression of coke formation. They also predicted that a similar effect of alloy formation would be obtained using post-transition metals such as lead (Pb).[63]

In addition to the ensemble effect, alloy formation can alter the crystalline structure of Pt. Various Pt-based alloys form depending on the preparation conditions, including Pt/promoter ratios and pre-treatment conditions. For instance, Pt3Sn and PtSn alloys can be prepared under different pre-treatment or reduction conditions, as investigated thoroughly by Deng et al.[64-67] For the dehydrogenation of ethane and propane, Pt3Sn is suitable; it has the same crystalline structure (fcc) as that of metallic Pt,[38,65] whereas PtSn with a hexagonal close packing (hcp) structure is not active for propane.[65] The activity of PtSn for EDH has not been evaluated clearly yet. This is also the case in Pt–In catalysts, where Pt3In exhibits higher activity than PtIn2 (a fluorite structure).[53] Based on several studies, it seems plausible that the crystalline structure is an important factor for EDH. However, PtZn with a tetragonal gold–copper alloy (AuCu) structure is highly active for EDH.[51] Additional studies must be undertaken to elucidate the influence of the crystalline structure on EDH activity.

Surface modification is a geometric effect, as shown in Fig. 4(b). As described in Subsection 3.1.1, side reactions such as coke formation and hydrogenolysis are facile at the “cus-Pt” sites. These active sites can be covered by Sn addition if Pt–Sn alloys are formed or not, leading to suppression of the side reactions.[64] Similar effects of Al addition were also reported by Peng et al.[37] The surface Sn species, which do not contribute to Pt–Sn alloy, exist on “cus-Pt” as Sn oxides such as SnO2. The location of Sn is evaluated using infrared (IR) spectroscopy with carbon monoxide (CO) as a probe molecule.[55,64,67,68] Broad absorption bands for linearly adsorbed CO and weak absorption bands for bridged CO on monometallic Pt are observable, respectively, at around 2080 and 1850 cm−1.[55,64] In this case, CO linearly adsorbed onto “cus-Pt” shows the absorption bands at slightly lower wavenumbers than that on Pt terrace because “cus-Pt” interacts strongly with CO, resulting in much back donation from Pt to 2π* (anti-bonding) orbitals of CO. Deng et al. reported that absorption bands attributed to CO2 are observed in the spectrum of Pt modified with SnO2.[64] However, no CO2 formation was verified without SnO2. The Pt modified with SnO2 exhibits higher selectivity for dehydrogenation of propane and butane. Therefore, they concluded that SnO2 exists on the Pt surface, where it covers “cus-Pt”. On Pt3Sn and PtSn surfaces, linearly adsorbed CO shows symmetric absorption bands because of the few “cus-Pt” sites. Furthermore, the absorption bands for bridged CO were not apparent, even on Pt3Sn. The existence of multi-coordinated CO was verified on well-defined Pt3Sn(111).[69] This difference is ascribed to the surface enrichment of Sn on Pt3Sn and PtSn nanoparticles. Deng et al. applied X-ray photoelectron spectroscopy (XPS) and found that the surface Sn/Pt ratios are higher than those of bulk material,[66,67] indicating that surface modification of SnO2 occurs concomitantly with alloy formation. It is worth noticing that segregation of Sn from Pt–Sn alloys[70,71] and surface reconstruction of metallic Pt nanoparticles[72] occur through CO adsorption. These phenomena alter the wavenumbers and shape of the CO absorption bands. Therefore, the information obtained from the IR spectra would not appropriately reflect the working state of the Pt and Pt–Sn surfaces.

The Pt particle size (dispersion) can be varied by promoter addition. For example, small Pt particles are obtained by Sn addition using silica supports.[64,73] Such is not the case for γ-Al2O3 and Mg(Al)O because these Al containing supports stabilize Pt particles,[28,29] leading to small Pt particles (38,74] In addition, Sn plays a crucially important role in re-dispersion of the Pt particles through oxidative and subsequent reduction treatments for catalyst regeneration, as presented in Fig. 5. Weckhuysen and co-workers reported characteristic behavior of the re-dispersive effect of Sn on the Pt particles during regeneration based on CO adsorption, in situ X-ray absorption fine structure (XAFS) spectroscopy, and transmission electron microscopy.[74,75] They observed atomically dispersed Sn on γ-Al2O3, which plays a role in nucleation sites for re-dispersion of Pt (Fig. 5(a)). During the oxidative treatment, Sn segregates from the alloy; platinum oxide (PtO) is formed on the support (Fig. 5(b)).[76] Subsequent reductive treatment reconstructs the Pt–Sn alloy.[75] Therefore, they argued that the agglomeration of Pt is more responsible for deactivation than carbon deposition.[74] The phenomena occurring during oxidation are expected to be crucially important for the re-dispersion of Pt–Sn alloy. However, the nucleation mechanism and the role of segregated Sn have not been clearly proved.

Fig. 5

Schematic images of (a) re-dispersion of Pt–Sn alloy through oxidation and subsequent reduction (regeneration) and (b) segregation of Sn from Pt–Sn alloy during oxidation.

Electronic effects of promoters on platinum

Modification of the electronic state of Pt is the other role of the promoters. Particularly, effects of Sn addition on the electronic state of Pt were examined mainly based on Pt LIII-edge XAFS spectroscopy.[64-67,76,77] The absorption edge energy of Sn-modified Pt catalysts is shifted to higher values than that of bare Pt catalysts. This finding apparently indicates electron-deficient Pt formation by Sn addition. However, the shift of the edge energy, which represents the transition energy of electrons from 2p3/2 orbital to vacant 5d orbital, is attributed to an increase in the energy level of the vacant 5d orbital. Therefore, the shift does not indicate the electronic density of Pt accurately. Deng et al. evaluated the electronic state of Pt by calculating d-band vacancy based on the white line intensity in Pt LIII-edge X-ray absorption near edge structure (XANES) spectra.[64] It is noteworthy that Pt catalysts with almost identical Pt particle size must be used for measurement because the metal particle size influences the white line intensity.[78-80] They argued that Pt with high electron density is obtained by Sn addition if Pt–Sn alloys composed mainly of Pt3Sn are formed or not, indicating that even modification of Pt with SnO2 induces electron donation from Sn to Pt, as shown in Fig. 6(a). Therefore, they conclude that Sn addition makes Pt electron-rich. The electron donation of Sn to Pt is also verified by DFT calculations.[42]

Fig. 6

Proposed effects of the promoters on the electronic state of Pt: (a) electron donation from the promoter (M) to Pt and (b) modification of the energy level of Pt 5d bands. Densities of states are denoted as DOS.

The electron donation from Sn to Pt can be supported by CO adsorption using IR spectroscopy. The absorption bands for CO linearly adsorbed onto Pt3Sn or PtSn show at lower wavenumbers than that on monometallic Pt terrace.[64,67,68] The shift of the absorption bands is based on formation of electron-rich Pt by alloy formation, resulting in much back-donation from Pt to CO. However, the following two factors must be considered: (i) Pt–Sn alloy has no “cus-Pt” and (ii) the spectra for CO adsorption are measured under the same coverage of CO.[81] In the former case, the absorption band shape corresponding to CO adsorbed onto Pt–Sn alloy is sharp compared to that on monometallic Pt. Ideally, the bands should be symmetric to demonstrate that Pt–Sn has environmentally uniform active sites. The latter case is related to dipole–dipole interactions between CO and adjacent CO adsorbed at Pt sites. On the Pt–Sn alloy surface, the number of Pt sites for CO adsorption is diluted with Sn. This dilution induces low coverage of CO and weak dipole–dipole interactions, resulting in low wavenumbers of bands for linearly adsorbed CO.

In contrast with electron donation, Xin et al. argued that the electronic effects of Sn on Pt are not electron donation but the change in the energy level of d-band center of Pt through the hybridization of Pt 5d bands with Sn valence (5s and 5p) bands.[77,82] Recently, Miller and co-workers investigated the electronic structure of Pt in PtZn and Pt–Fe (Pt3Fe, PtFe and PtFe3) intermetallic compounds using resonant inelastic X-ray scattering to evaluate the energy level of Pt 5d orbitals.[51,83] They reported that the average energy gap separating the filled and unfilled Pt 5d bands is expanded compared with that of monometallic Pt. In other words, Zn or Fe addition stabilized the energy level of the occupied 5d band, whereas the unoccupied 5d band is shifted to higher energy levels, as shown in Fig. 6(b). Additionally, they calculated the projected d density of states localized on Pt atoms.[83] The mean energy of filled (below the Fermi level) Pt 5d bands is stabilized by virtue of alloy formation, whereas that of unfilled (above the Fermi level) Pt 5d bands is shifted upward. Therefore, they proposed that the alloy formation effect on the electronic state of Pt is the modification of the energy level of the Pt 5d bands rather than electron donation from the promoters to Pt. This change in the Pt 5d band energy is expected to affect the platinum–ethane interaction and to enhance the catalytic activity.

The change in the electronic state of Pt modified with the promoters influences the adsorption properties of intermediates and products. Hook et al. calculated the binding energy of the intermediates on Pt and Pt–Sn(111) surfaces based on DFT.[44] From their results, they reported that ethylene adsorbs weakly on the Pt–Sn surface, indicating that re-adsorption of ethylene, which induces ethylene decomposition to coke, can be inhibited on the surface. Such is also the case in the binding energy of the coke precursors. Generally, ethylene adsorbs on Pt in either π-bonded or di-σ-bonded configuration.[84] The latter adsorbs on the Pt surface more strongly, resulting in ethylene decomposition. Sn addition to Pt inhibits ethylene adsorption on Pt, as demonstrated by Liwu et al.[85] However, large amounts of carbon deposition were verified in the case of Pt–Sn/γ-Al2O3.[74,86] This phenomenon is based on the drain-off mechanism proposed by Lieske et al. and modified by Kumar et al.[73,86] As shown in Fig. 7, the coke precursors weakly adsorbed onto Pt–Sn are facile to transfer from the Pt–Sn surface to γ-Al2O3. Then, coke formation proceeds on the support. This coke is burnable at higher temperatures than that deposited on Pt.[85] In this way, Sn addition to Pt not only contributes to the suppression of coke formation; it also keeps the Pt–Sn surface clean.

Fig. 7

Schematic image of the drain-off mechanism. Coke precursors such as ethylidyne move to the supports on which coke formation proceeds.

Chromium-based catalysts

Chromium (Cr) oxide catalysts have been used for dehydrogenation of light alkanes, as represented by the Catofin process (Mcdermott-Lummus).[26] The dehydrogenation reactions are performed by multiple fixed bed reactors operated in the cyclic mode (reaction and regeneration).[87] During the operation, several reactors are used for the reactions while catalyst regeneration is conducted in the others. Among the products, hydrogen is separated by a pressure swing adsorption unit. The unreacted feed is separated from the products and recycled for the reactions. The Catofine process is employed for dehydrogenation of propane and butanes, similarly to the Oleflex process. In terms of cost, the Cr oxide catalysts is superior to Pt-based catalysts. As described in this section, thorough investigations have been conducted of the active sites, effects of support, and influences of co-feeding CO2 in order to use the Cr oxide catalysts for EDH.

Active sites of chromium species for EDH

The nature of Cr species is an important factor affecting C–H bond activation. Fundamentally, EDH proceeds on Cr–O sites.[88] Their structures and electronic states have been investigated using various characterizations such as temperature-programmed reduction (TPR), Raman spectroscopy, ultra violet (UV)-visible (Vis) spectroscopy, X-ray diffraction, XAFS spectroscopy, and XPS.[89-96]

The Cr species can be various structures. Generally, the Cr species are classifiable into four types: monochromate (isolated mononuclear Cr), polychromate (Cr oxide cluster), amorphous Cr oxide (Cr2O3) and crystalline α-Cr2O3.[90-92,94] The Cr species structure is mainly dependent on the amount of Cr loading. At the high loading amount (>5 wt%), the amorphous Cr2O3 and crystalline α-Cr2O3 are easily formed. These Cr species are regarded as inactive for dehydrogenation.[90,97] In contrast, monochromate and polychromate species are dominant at low loading amount. The formation of Cr2O3 is distinguishable from the color of the catalysts: actually, the color is varied from yellowish to greenish because of the Cr2O3 formation.[95,98,99]

Investigations of the nature of monochromate and polychromate have been conducted widely because these Cr species are active sites for EDH.[87,100,101] Images of them are presented in Fig. 8. To identify the existence of the Cr species, UV-Vis and Raman spectroscopy are useful techniques. However, the Cr species are reduced under the reaction conditions from Cr6+ or Cr5+ to Cr3+ or Cr2+. Therefore, evaluating the active sites under in situ conditions is important. For fresh Cr oxide catalysts, Cr6+ is dominant. Many studies have specifically examined Cr6+ because the existence of Cr5+ was identified only by electron spin resonance (ESR) and IR spectroscopy.[91,98] In addition, the reduced Cr species are mainly Cr3+. Although Cr2+ was identified using XPS and XAFS,[96,100] the proportion of Cr2+ in the spent catalysts was rather small.[100] Therefore, Cr6+ is considered to be reduced to Cr3+ during the dehydrogenation reaction. The nature of Cr5+ and Cr2+ remains under discussion.

Fig. 8

Structure of (a) monochromate and (b) polychromate supported on a metal oxide.

The EDH over Cr oxide catalysts proceeds on the reduced monochromate and polychromate. As presented in Fig. 9(a), Olsbey et al. reported based on results of an isotopic study that end-on and dissociative adsorption of ethane occurs on the active sites.[88] However, Shee and Sayari observed adsorbed acetaldehyde using diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy.[102] They proposed that the ethoxide species formation is the first step in the dehydrogenation reaction (Fig. 9(b)). Determination of the adsorbed species on the respective Cr species is a challenging subject. Therefore, additional investigations based on, for instance, DFT calculations are expected to elucidate the reaction mechanism.

Fig. 9

Schematic adsorption of ethane on reduced monochromate supported on a metal oxide: formation of (a) ethyl group and (b) ethoxy group.

Effects of supports on the chromium state

As described in the preceding subsection, the ideal Cr oxide catalyst has as large an amount of monochromate or polychromate species as possible. However, the large Cr loading amount results in the formation of the inactive Cr2O3 species through agglomeration.[95] Therefore, selection of an appropriate support strongly influences the catalytic activity and ethylene selectivity.

Alumina is a typical support for this purpose.[88-92,95,102] Recently, Fridman et al. thoroughly examined the Cr species on the alumina support.[95] They reported that a part of the monochromate and polychromate was unstable and reduced during dehydrogenation of isobutane. After the reduction, small Cr2O3 clusters were agglomerated; large Cr2O3 clusters were formed. Furthermore, regeneration of the catalyst under oxidative conditions induced formation of aluminum–chromium–chromate species around the periphery of the Cr2O3 clusters, as shown in Fig. 10. Agglomeration of the monochromate and polychromate leads to irreversible deactivation of the catalysts.

Fig. 10

Structure of aluminum–chromium–chromate.

Silica (SiO2) is an alternative support to alumina. Silica support of various kinds, including silicious-zeolites and mesoporous silica, were examined.[94-100,103-106] These silica supports have a large surface area, resulting in highly dispersed Cr species on the supports. The amount of silanol (SiOH) groups is expected to play an important role in stabilizing the dispersed Cr species (monochromate or polychromate). Cheng et al. reported that the initial activity of Cr/SBA-15 (mesoporous silica) is proportional to the amount of the SiOH groups contributing to the abundantly dispersed Cr species.[105] Botavina et al. proposed detailed structures of monochromate based on XAFS measurements.[96] The reduced monochromate was bound with two short Si–O–Cr bonds and a long Si–O–Cr bond. Its electronic state is expected to be Cr2+ rather than Cr3+. These studies indicate that the density of the SiOH groups influences the Cr active site characteristics.

Co-feed of carbon dioxide in EDH over chromium-based catalysts

To enhance the catalytic performance of the Cr catalysts, EDH has been investigated in the presence of CO2.[107] In many cases, EDH in the presence of CO2 is regarded as oxidative dehydrogenation. However, the overall reaction is still endothermic, as presented below.

This chemical formula is interpreted as EDH (reaction (10)) in combination with the reverse water gas shift (RWGS) reaction.

Therefore, the reaction (11) characteristics differ completely from those of oxidative EDH using oxygen as an oxidizing agent, which is an exothermic reaction. Effects of the CO2 introduction to the dehydrogenation system are generally explained based on the following three reasons:[108-110] decrease in the partial pressure of ethane (see also Fig. 2); removal of H2 from the reaction system through RWGS, resulting in the promotion of EDH; and improvement of catalytic stability by virtue of removal of carbonaceous deposits through the reverse Boudart reaction.

The promotion of ethane conversion and suppression of coke formation have been reported.[99,109] However, the promotive effect on EDH over Cr-based catalysts is attributed not only to the contribution of RWGS but to differences in the nature of active sites and reaction mechanism. Without CO2, the active sites are presumed to be Cr3+ reduced from monochromate or polychromate (Cr6+). At the initial stage of the reaction, where the reduction of Cr6+ proceeds, ethane conversion tends to decrease,[100,101] indicating that Cr6+ is more active than Cr3+ for EDH. Indeed, Mimura et al. reported that the apparent activation energy of EDH decreased from 119.3 (without CO2) to 91.3 (with CO2) kJ mol−1 using a Cr/ZSM-5 (zeolite) catalyst.[98] Therefore, it is proposed that EDH proceeds through the redox mechanism as shown in Scheme 1. In this scheme, ethane is activated oxidatively at Cr6+O sites, resulting in the formation of ethylene and water (H2O). Then, the catalytic cycle is completed through oxidation of the reduced Cr3+ species with CO2. Among the Cr species, Baek et al. proposed based on TPR that the large amount of polychromate results in a high catalytic activity for dehydrogenation of propane in the presence of CO2.[111] Therefore, polychromate is expected to be the active site for EDH via the redox mechanism.

Scheme 1

Redox mechanism for EDH in the presence of CO2 over Cr oxide catalysts.

The redox behavior of the Cr species was verified using various techniques such as TPR, XPS, and IR spectroscopy. Based on TPR and XPS, the reduced Cr3+ can be oxidized to Cr6+ under a CO2 atmosphere.[99,104,109,110] In addition, Mimura et al. evaluated CrO bonds by IR spectroscopy,[98] demonstrating that the intensity of the absorption band attributed to CrO was decreased and increased respectively by ethane and CO2 treatments. However, the reducible Cr6+ species are not re-oxidized completely with CO2, probably because of agglomeration of monochromate and polychromate and because of the low oxidizing ability of CO2.[112] Therefore, it is necessary for the catalyst to stabilize the reducible Cr6+ species and to promote oxidation of the reduced Cr3+. The silica supports are expected to be useful to stabilize the Cr active sites because of the presence of SiOH groups as described above. Recently, Al-Awadi et al. reported that modification of Cr/MCM-41 (mesoporous silica) with titania (TiO2) nanoparticles enhances the concentration of Cr6+, leading to high activity for EDH with CO2.[113] To promote the oxidation of the reduced Cr species, the use of additives with oxygen storage capacity is one option.[114,115] In addition, CO2 activation can be expected to be a key factor because of its low reactivity. Zirconia (ZrO2)-supported Cr catalysts exhibited a high activity for EDH in the presence of CO2 because of the basicity of ZrO2 which would contribute to CO2 activation.[110,116] The Zr promoter also increases the amount of reducible Cr6+.[117]

Some difficulties remain unresolved in the case of EDH in the presence of CO2. For instance, the contribution of CO2 to oxidation of the reduced Cr species and RWGS, is not discussed in detail. One must elucidate the role of respective Cr species in dehydrogenation through the redox mechanism and RWGS. In addition, operando analyses in combination with isotopic CO2 should be performed to prove the redox mechanism.

Gallium-based catalysts

Several Ga oxide catalysts are known to have catalytic activity for EDH.[118] Actually, Ga catalysts are classified into two types: zeolite-supported Ga oxide catalysts and other metal oxide-supported Ga oxide catalysts. Comparison of zeolite supports with other metal oxide supports such as Al2O3 and TiO2 reveals that characteristics of the active sites differ because of the cation-exchange ability of zeolite. In this regard, differences in the characteristics of the active sites, which are derived from properties of the supports, are described.

Gallium-based catalysts supported on zeolites

Zeolite-supported Ga oxide catalysts were used originally for conversion of propane or butane into aromatic hydrocarbons.[119] The Cyclar process developed by BP/UOP is a representative example.[120] Among the zeolites, MFI type zeolite in proton form (H-ZSM-5) is used widely as a support. The characteristics of the Ga species, which can be formed by virtue of the ion-exchange ability of zeolite, have been investigated based on practical experiments and DFT calculations.

From a practical perspective, Ga/H-ZSM-5 catalysts prepared using various methods such as physical mixing, impregnation, and chemical vapor deposition (CVD) are used to elucidate the characteristics of the Ga species.[121-123] The pristine Ga/H-ZSM-5 usually contains highly dispersed Ga oxide species in the micropore system. At this stage, the Ga species are exchanged only insufficiently with the protons. To promote the ion-exchange, the Ga/H-ZSM-5 catalyst must be reduced under a H2 atmosphere. This phenomenon was verified using in situ IR spectroscopy. Intensity of the absorption band at ca. 3610 cm−1 derived from acidic hydroxy groups (Si–OH–Al) decreases under H2 atmospheres.[121,124] Simultaneously, a new absorption band attributed to GaO–H appeared at ca. 3660 cm−1, clearly indicating the ion-exchange of the reduced Ga species with the acidic protons. Furthermore, formation of Ga–H bonds, which is unstable at high temperatures such as 573 K, was observed at room temperature. Based on findings obtained using IR spectroscopy, Ga+, GaO+, and [GaH2]+ are candidates for active sites of the reduced Ga/H-ZSM-5.[124,125] In addition, Rane et al. evaluated the electronic state of Ga using in situ Ga K-edge XANES spectroscopy.[123] At high temperatures (>673 K) under a H2 atmosphere, a new feature was observed in a lower energy region (10 371 eV), indicating the formation of Ga+. Subsequent cooling to 373 K induces the increase of a feature at 10 377 eV with disappearance of the spectrum at the low energy feature, indicating the formation of [GaH2]+. However, Getsoian et al. reported that the absorption edge energy was dependent not only on the electronic state of Ga but also on the type of ligand.[126] For instance, the absorption energy can be shifted to the lower position through the formation of low-coordinated Ga3+ alkyl or hydride species. However, Schreiber et al. performed quantitative analyses to elucidate the Ga species.[127] They conducted H2 pulse tests at 873 K and measured the amount of the consumed H2 and formed H2O. As a result, they concluded that Ga+ was formed after reduction at 873 K.

The activity of respective Ga species was evaluated experimentally. Rane et al. performed dehydrogenation of propane after reduction or oxidation treatments.[123] After reduction treatment to form [GaH2]+, the activity increased gradually with time on stream. In contrast, the activity of the oxidized Ga/H-ZSM-5 containing GaO+ decreased rapidly in 1 h. These activities were close to that of Ga+ species. They concluded that [GaH2]+ and GaO+ are unstable active sites during the dehydrogenation of propane. Furthermore, the intrinsic activities of the Ga species are in the order of GaO+ > Ga+ > [GaH2]+ based on the initial activities. However, Ausavasukhi and Sooknoi conducted EDH in a pulse reactor.[122] They demonstrated that the activity was enhanced by the introduction of H2, indicating higher activity of [GaH2]+ for EDH than that of either Ga+ or GaO+. Moreover, the activity of [GaH2]+ decreased with subsequent C2H6 pulses under He, whereas high activity was maintained under H2, indicating the unstable characteristics of the [GaH2]+. Therefore, the Ga+ species would be stable during dehydrogenation reactions, although the activities of the Ga species for dehydrogenation of light alkanes would depend on the reactants. Nascimento and co-workers proposed that the reaction mechanisms were varied because of the size and type (linear or branched) of alkanes based on DFT studies.[128,129] In addition, differences in the reactor systems (flow or pulse reactor) must be examined because the adsorbed species, which would influence the activity, exist in a steady state using the flow reactor.

The reaction mechanism of EDH on the Ga species was investigated mainly using DFT calculations. As portrayed in Scheme 2, three reaction mechanisms have been proposed by some research groups:[128-133] the alkyl mechanism, the carbenium mechanism, and the concerted mechanism. The differences are based on the mode of ethane activation.

Scheme 2

Mechanism of ethane activation via (a) the alkyl mechanism, (b) the carbenium mechanism, and (c) the concerted mechanism. Ga+ is a representative active site of Ga/H-ZSM-5.

The alkyl mechanism begins from formation of Ga–C2H5 by hydrogen abstraction. By contrast, the carbenium mechanism proceeds through formation of an ethoxy group concomitantly with that of Ga–H. In the concerted mechanism, two hydrogen atoms are abstracted simultaneously from an ethane molecule, leading to the formation of ethylene and hydrogen in one step. The activation of ethane at Ga+, [GaH]2+, and [GaH2]+ sites is investigated by DFT calculations using cluster models. The alkyl mechanism is facile at any site.[128-132] However, Mansoor et al. recently reported that the carbenium mechanism is preferable at the [GaH]2+ site.[133] In addition, Schreiber et al. proposed that the proximity between the Ga and Brønsted acid sites promotes the dehydrogenation of propane.[127] Experimentally, Kazansky et al. observed the adsorbed ethane on the Ga+ sites using DRIFT spectroscopy.[125,134] They demonstrated the formation of Ga–C2H5 concomitantly with that of Ga–H. After heating, the absorption band of Ga–H at 2057 cm−1 was shifted to 2040 cm−1, which is attributed to H–Ga–H, with the appearance of olefinic C–H. Therefore, EDH at the Ga+ sites can be expected to proceed as the following Scheme 3.

Scheme 3

Reaction mechanism of EDH at the Ga+ site.

The dehydrogenation activity of the GaO+ species was evaluated in the presence of steam because the GaO+ species are reduced under the reaction conditions.[123] Hensen et al. performed dehydrogenation of propane over Ga/H-ZSM-5 in the presence of steam.[135] As a result, stable activity was achieved. At the mononuclear GaO+ site, the high activation barrier for hydrogen recombination was calculated among the elementary steps, resulting in the formation of a H2O molecule and the Ga+ site.[131] Therefore, they proposed that multinuclear GaO clusters are the active sites for the dehydrogenation reaction.[135-138] The Ga cations included in the clusters are favorable for tetrahedral coordination.[135] The proposed EDH at Ga2O2 clusters, for instance, is shown in Scheme 4. The dehydrogenation reaction proceeds at a Ga–O Lewis acid–base pair. Hydrogen recombination is facile at the Ga oxide clusters, although the H2O formation is still favorable. Strong Lewis basicity of the bridged oxygen stabilizes the adsorbed hydrogen, resulting in the inhibition of the hydrogen recombination.[138] Therefore, Lewis basicity should be moderate to facilitate hydrogen desorption.

Scheme 4

Reaction mechanism of EDH in the presence of steam on a Ga oxide cluster.

Ga oxide (Ga2O3) catalysts

Supported Ga oxide catalysts such as Ga/TiO2 are used for EDH, particularly, in the presence of CO2.[139] On the supports, Ga oxide is fundamentally Ga2O3, which has five polymorphs (α-, β-, γ-, κ-, ε-Ga2O3).[140] Characterizations of Ga properties such as the structure of the active sites are indispensable because the surface Ga species play an important role in the reaction. Zheng et al. performed dehydrogenation of propane over Ga2O3 polymorphs, except for ε-Ga2O3.[141] As a result, β-Ga2O3, the most stable polymorph, exhibited the highest activity among them. They concluded that the tetrahedrally coordinated Ga3+ cations facilitate the dehydrogenation reaction. Recently, Seki et al. investigated the nature of active Ga species at the β-Ga2O3 surface for ethane activation.[142] Based on DFT calculations and on results of experimental measurements, ethylene strongly adsorbs on the surface tetrahedrally coordinated Ga species, probably resulting in ethylene decomposition to coke. In contrast, dissociative adsorption of ethane is favorable at the surface octahedrally coordinated Ga with coordinatively unsaturated, for instance, located at the β-Ga2O3(002) facet.

The coordination environment of Ga can be evaluated using Ga K-edge XANES. Nishi et al. conducted a quantitative analysis of tetrahedrally and octahedrally coordinated Ga using XANES spectroscopy.[143] Hereinafter, these Ga species are denoted respectively as Ga(T) and Ga(O). The analysis is based on deconvolution of the XANES spectra. It is applicable to Ga polymorphs and supported Ga oxide catalysts.[143,144] In addition, the coordination environment of Ga2O3 can be evaluated by IR spectroscopy using H2 as a molecular probe. Collins et al. reported that the absorption bands attributed to Ga–H bonds are observed at around 2000 cm−1.[145] They further measured the spectra for hydrogen adsorbed onto α-, β-, and γ-Ga2O3.[146] The spectra consisted of absorption bands at 2003 and 1980 cm−1. The contribution of the band at 1980 cm−1 was in the order of α-Ga2O3 > γ-Ga2O3 > β-Ga2O3. This order agrees with that of the fraction of Ga(O) in their structures. Therefore, they concluded that the Ga(T)–H and Ga(O)–H bonds are observed respectively at 2003 and 1980 cm−1. The proportion of the surface Ga(T) to Ga(O) is calculable from the absorption band area. However, the band intensity is dependent on the temperatures.[147,148] At high temperatures such as 573 K, the intensity increased because of the formation of oxygen defects, resulting in the formation of new Ga–H bonds. Therefore, the coordination environment of surface Ga is expected to be evaluated appropriately at temperatures higher than 573 K because the dehydrogenation reaction is usually conducted at such higher temperatures under reductive conditions.

The reaction mechanism of EDH over Ga2O3 is similar to the alkyl mechanism over Ga/H-ZSM-5 (Scheme 3). Kazansky et al. measured the spectra for ethane adsorbed onto Ga2O3 using DRIFT spectroscopy.[149] In this case, the polymorph of Ga2O3 is not described clearly. They observed the formation of Ga–C2H5 in addition to C–H corresponding to CH2 groups derived from ethylene, and Ga–H. The formation of Ga–C2H5 was also verified by 13C cross-polarization magic angle spinning (MAS) nuclear magnetic resonance (NMR) spectroscopy.[150] In some cases, the formation of Ga–H and alkoxy groups was assumed to occur via the activation of light alkanes.[151,152] Nevertheless, no clear evidence of the formation of the alkoxy groups has been reported to date.

The dehydrogenation of ethane over Ga oxide catalysts is performed mainly in the presence of CO2.[118] Nakagawa et al. reported that Ga/TiO2 exhibited high activity when compared to other supports including Al2O3, SiO2, ZrO2, and zinc oxide (ZnO).[139] In the case of Ga/Al2O3, a negative effect on the catalytic activity was verified in the presence of CO2. The promotive effects of CO2 on EDH are attributed respectively to hydrogen and coke removal through the RWGS and the reverse Boudart reaction (reactions (12) and (13)).[153,154] In contrast, selectivity to ethylene tends to decrease with the increase in ethane activation.[155] Methane formation is promoted instead of the decrease in ethylene selectivity, indicating the subsequent conversion of ethylene to methane. Therefore, appropriate conditions such as space velocity and partial pressure of CO2 are necessary to achieve high ethylene yield.

In contrast to Cr oxides, Ga2O3 is a non-reducible oxide. Therefore, no report has described the redox mechanism over Ga oxide catalysts in the presence of CO2, although Ga species in H-ZSM-5 might have redox ability. To improve catalytic performance, designing Ga active sites is indispensable. For instance, Ga/H-ZSM-5 catalysts exhibited a higher selectivity to ethylene than β-Ga2O3.[156] Also, Ga/H-ZSM-5 exhibits the activity for RWGS, indicating that H2O co-exists during EDH with CO2. Therefore, the Ga oxide clusters, which exhibit high dehydrogenation activity viaScheme 4, would be maintained during the reaction.

Other catalysts and performance of EDH catalysts

Aside from Pt-, Cr- and Ga-based catalysts, cobalt (Co)- and molybdenum (Mo)-based catalysts exhibit catalytic activities for EDH.[157-163] Although Zn catalysts also have ethane activation ability,[134] they are applied for aromatization of light alkanes. Both are used mainly in the presence of CO2. Among Co-based catalysts, the electronic state of Co is expected to be important. The existence of Co4+ promotes EDH via the redox mechanism similarly to the Cr catalysts.[157] However, Co2+ species facilitate RWGS analogously to the Ga catalysts.[158,159] For Mo-based catalysts, molybdenum carbide (Mo2C) is rather more active for ethane activation than molybdenum oxide (MoO3) in the presence of CO2.[160] Chen and co-workers evaluated the electronic state of Mo using in situ XANES spectroscopy, revealing that Mo2C is oxidized with CO2.[161] They reported further that CO2 is converted to CO and adsorbed oxygen over β-Mo2C.[162] Therefore, the oxygen-modified Mo2C (oxycarbide) would be formed and would promote EDH with CO2.

Recently, lanthanum manganite (LaMnO3) perovskites were applied for EDH. Yang et al. performed EDH without co-feed over reduced Ni-doped LaMnO3.[164] They found that the proportion of Mn4+ increases and that Ni nanoparticles are deposited after reduction treatment with H2. Therefore, they proposed that EDH proceeds on oxygen vacancies around Mn4+ and Ni nanoparticles. In contrast, Sekine and co-workers investigated EDH with H2O over Ba-doped LaMnO3.[165,166] They demonstrated clearly that EDH is promoted in the presence of steam because the reaction proceeds via the Mars–van Krevelen mechanism. They proposed that the amount of reducible Mn3+ is a key factor for ethylene formation. Ba doping enhances the redox properties of LaMnO3 and the oxygens coordinated with Ba promote C–H activation of ethane.

Catalytic performance of the catalysts for EDH in the absence and presence of CO2 is summarized respectively in Tables 1 and 2. In addition, ethylene selectivity vs. ethane conversion is shown in Fig. 11. The noble metal catalysts are used at low temperatures without CO2, probably to suppress agglomeration and dry reforming. Yan et al. reported that modification of ceria (CeO2)-supported Ni catalysts with Fe could mitigate dry reforming of ethane by virtue of iron oxide (FeO) located around periphery of Ni particles (Ni–CeO2 interface).[167] Among the noble metals, palladium (Pd) and gold (Au) also exhibit the catalytic activity for EDH.[168,169] In order to replace the expensive noble metals, Ni3Ga alloys,[170] which exhibited comparable catalytic performance of EDH to noble metal catalysts,[171] are good candidates. In recent studies, zeolite-supported metal catalysts appeared as new candidates for EDH without CO2. The characteristic micropores and cation-exchange ability contribute to the formation of active sites such as highly dispersed iron carbides and indium hydrides.[172,173] In the presence of CO2, Cr-based catalysts supported on various materials including oxidized diamond[174] and titanosilicate[175] exhibit high ethane conversion thanks to the redox mechanism (Scheme 1), which indicates that other materials with redox properties can be good catalysts. The metal oxides with redox properties are applicable to a chemical looping process.[176] Thorium (Th)- and Ni-based oxides are also reported to exhibit the activities[177,178] although comparative investigations are necessary to elucidate advantages of these materials.

Catalytic performance for EDH in the absence of CO2

No.	Catalyst	Temperature/K	C2H6 conversion/%	C2H4 selectivity/%	C2H4 yield/%	Ref.
1	PtSn/Mg(Al)O	873	2.6	100	2.6	40
2	Pt–In/SiO2	873	15	99	15	53
3	PtSn–MgGaAlO	823	28	99	28	54
4	Pd–In/SiO2	873	15	100	15	168
5	Au/SiO2-doped TiO2	923	16	95	15	169
6	Ni3Ga/Al2O3	873	10	94	9.4	171
7	Cr/SBA-15/Al2O3/FeCrAl	1023	47	87	41	99
8	Cr/MCM-41	923	19	98	18	100
9	Cr–Ce/SBA-15	973	41	83	34	114
10	Cr2O3/Oxidized diamond	923	7.0	97	6.8	174
11	Cr/TS-1	923	52	86	45	175
12	Ga2O3/Al2O3	923	28	93	26	139
13	Ga/SiO2-doped TiO2	923	46	85	39	154
14	La0.9Mn0.8Ni0.2O3	1023	42	98	41	164
15	La0.7Ba0.3MnO3−δ	973	4.8	88	4.2	165
16	Th0.75Ca0.25O2	998	48	78	37	177
17	Fe/ZSM-5	873	22	72	16	172
18	In–CHA	933	27	97	26	173

Catalytic performance for EDH with CO2

No.	Catalyst	Temperature/K	C2H6 conversion/%	C2H4 selectivity/%	C2H4 yield/%	CO2/C2H6	Ref.
19	Cr/SBA-15/Al2O3/FeCrAl	1023	67	96	64	2	99
20	Cr2O3/SO4–SiO2	923	67	82	55	5	103
21	Cr/MSU-1	973	68	82	56	3	104
22	Cr/SBA-15	923	41	92	38	5	105
23	Cr/ZSM-5	923	66	75	49	5	106
24	Cr2O3/SiO2	923	56	93	52	5	108
25	Cr/H-ZSM-5	923	68	70	47	9	109
26	Cr/Ti/MCM-41	973	52	93	48	5	113
27	Cr–Ce/SBA-15	973	55	96	53	3	114
28	Cr2O3/HZSM-5–ZrO2	973	65	87	56	5	115
29	Fe–Cr/ZrO2	923	54	93	50	3	116
30	Cr2O3/Oxidized diamond	923	27	87	24	5	174
31	Cr/TS-1	923	62	81	50	4	175
32	Ga2O3	923	29	87	25	5	139
33	Ga2O3/ZSM-5	923	25	92	23	5	151
34	Ga/SiO2-doped TiO2	923	47	78	37	5	154
35	Ga/TiO2	973	38	57	22	2.5	155
36	Co–BaCO3	923	48	92	44	3	157
37	CoOx/SiO2	973	46	85	39	2.5	159
38	Mo2C	873	2.0	60	1.2	1	161
39	Th0.75Ca0.25O2	998	46	97	45	0.78	177
40	CaO–NiO/Al2O3	973	28	56	16	0.75	178

Fig. 11

Ethylene selectivity vs. ethane conversion of each catalyst presented in Tables 1 and 2. Dashed lines represent ethylene yield. Also, EDH in the absence and presence of CO2 are shown respectively with red circles and blue squares.

Non-thermo-catalytic EDH

Recently, electrochemical and photocatalytic EDH was emerged as new approaches for ethylene production. These systems can be operated using renewable energy; ethylene can be produced by renewable electricity and sunlight. Therefore, electrochemical and photocatalytic EDH would play a crucial role in producing petrochemicals in a sustainable society.

Electrochemical EDH

Electrochemical EDH is performed by solid oxide electrolysis cells (SOEC) as shown in Fig. 12(a). Imposing electrochemical potential, oxidation/reduction reactions proceed on anode/cathode surfaces although SOEC must be heated to 673–973 K to obtain sufficient oxygen anion or proton conductivity.[179] At such temperatures, pyrolysis of ethane would be taken into account to calculate Faraday efficiency. Characteristically, pure hydrogen can be produced without separation using proton conducting materials such as barium zirconate (BaZrO3) and barium cerate (BaCeO3). Alternatively, supplying CO2 to cathode results in CO formation.

Fig. 12

Schematic images of (a) electrochemical EDH by a solid oxide electrolysis cell and (b) photocatalytic EDH over a particulate semiconductor with a cocatalyst.

Ding et al. performed electrochemical EDH at 673–773 K using BaCeO3-based proton conducting materials as the anode.[180] They reported that electrochemical EDH proceeds without coke formation at 673 K. Increasing the reaction temperature induced coke formation and low ethylene selectivity. To mitigate coke formation, Zhang et al. used Ni and Cu doped niobium-based complexed oxides (proton conducting material) as the anode.[181] Ni–Cu alloys formed through the reduction of the anode contribute to the suppression of coke formation even at 973 K. Ethane conversion and ethylene selectivity are 66.3 and 99.7% respectively at 0.8 V. In combination with CO2 reduction in the cathode, ethane conversion increased up to 75.2%. By contrast, Song et al. performed electrochemical EDH with CO2 reduction using oxygen anion conducting materials.[182] They reported that γ-Al2O3 addition to lanthanum-based complexed oxide enhanced ethane conversion and ethylene selectivity without coke formation.

Researches on electrochemical EDH are focused on the performance and electric properties of anode materials. However, elucidation of activity–conductivity relationship is necessary for further improvement of the performance.

Photocatalytic EDH

Photocatalysis is a light-driven chemical reaction proceeding on semiconductors such as metal oxides, metal sulfides, and metal nitrides.[183] As shown in Fig. 12(b), photocatalysis is initiated by exciting electrons from the valence band (VB) to the conduction band (CB) through photon absorption. The photon energy must be greater than the band gap energy to excite the valence electrons. With the photoexcitation, holes are created in the VB. The photoexcited carriers (electrons and holes) migrate to the surface on which the redox reaction proceeds. The electron–hole pairs are easy to be recombined and, therefore, separation of the carriers from the bulk to the surface leads to a high photocatalytic activity. Metallic cocatalysts such as Pt and Pd are usually loaded on the semiconductors to promote the charge separation, namely inhibition of the recombination. Additional details of photocatalysis are reviewed in a literature.[183]

Photocatalytic EDH is performed by Zhang et al. using Pd/TiO2 catalysts in combination with CO2 reduction (Fig. 12(b)).[184] Pd/TiO2 exhibited ethane conversion of 0.14% at 308 K and 0.2 MPa. The obtained conversion is corresponded to the equilibrium conversion at 498 K. It is explained that the forward and backward reactions require different energy and reaction mechanisms (irreversible).[185] They also conducted spin-trapping ESR and demonstrated clearly that the hydroxyl radicals (·OH) are formed on TiO2 surface through oxidation of hydroxy groups with the holes. In addition to the holes, the hydroxyl radicals contribute to ethylene formation.

Producing chemicals from ethane through photocatalytic reactions are still immature research field. Practically using photocatalysts, the photocatalytic reactions must be visible light-driven. In addition to researches on materials science for controlling band gap energy, understanding the reaction mechanism is essential to design photocatalysts exhibiting high ethylene selectivity.

Dehydroaromatization of ethane

Ethane dehydroaromatization, denoted as EDA, forms BTX in one step.

This reaction is an endothermic reaction similar to EDH. Therefore, high reaction temperatures at around 873 K are favorable to achieve high equilibrium conversion. From a practical perspective, the products are readily handled in terms of separation, storage and transportation because the gaseous reactant is converted to liquid products, potentially leading to a feasible process. However, commercial processes have not been developed yet using the natural gas resources (methane and ethane) in contrast with aromatization of C3 and C4 hydrocarbons.[120] This lack of development is attributable to high reaction temperatures, resulting in the rapid deactivation of zeolite-supported metal catalysts through coke formation.[186-189] Effective zeolite topology and active metals are briefly explained at the beginning of this chapter. Next, details of the nature of Zn/H-ZSM-5, the reaction pathways and deactivation by coke formation are described. Finally, aspects of the catalytic performance of the representative EDA catalysts are summarized.

Zeolite-supported metal catalysts

Effective zeolite topology for BTX formation

For EDA, the H-ZSM-5 zeolite modified with active metals is known to be an effective catalyst. In fact, MFI-type zeolite has three-dimensional micropore channel systems, which consist of straight and sinusoidal channels.[190,191] They comprise 10-membered rings with ca. 0.55 nm diameter, which closely approximates the kinetic diameters of BTX. Because of the micropore channels, the MFI-type zeolite exhibits shape selectivity to BTX formation. In addition, H-ZSM-5 has strong Brønsted acid sites (BAS), which catalyze various reactions such as isomerization, cracking, alkylation and hydrogen transfer based on the carbenium and carbonium ion chemistry. Details of the reactions over zeolite catalysts are described in the literature.[192]

The MEL-type zeolite (ZSM-11) is another candidate for the support.[193,194] Actually, ZSM-11 has similar micropore channel systems to those of ZSM-5, which consist of two straight channels.[190,191] However, the superiority of ZSM-11 compared to ZSM-5 has not been demonstrated clearly. The MWW-type zeolite presented by MCM-22 has not been used in contrast with methane dehydroaromatization.[195] Reportedly, MCM-22 has two independent micropore channel systems.[190,191] One comprises 12-membered ring cages connected to 10-membered ring windows. The other is a two-dimensional micropore channel system. Difference between methane and ethane dehydroaromatization are attributable to the kinds of active metals. Zn or Ga catalysts are used mainly for EDA in contrast with Mo catalysts for methane dehydroaromatization. The use of Mo might be promising for BTX formation in the case of MCM-22 support because of the different reaction pathways, as described in Section 4.2.

Active metals for EDA

The H-ZSM-5 zeolite is usually modified with active metals to facilitate ethane activation. Activating ethane at BAS is difficult because the formation of primary carbenium ions is unfavorable, resulting in low ethane conversion over H-ZSM-5 without the metal modification. In Section 4.2, the roles of the active metals are discussed in relation to the reaction pathway. As reported earlier, active metals including Pt, Ga, and Zn are effective for ethane activation.[196-200] However, Pt is expensive; it induces hydrogenolysis of ethane to methane under H2 atmospheres.[35]

Also, H-ZSM-5-supported Mo and rhenium (Re) catalysts exhibit high performance for EDA[201-203] in addition to dehydroaromatization of methane.[204,205] However, these two metals are unsuitable for practical use because of sublimation of MoO3 and rhenium oxide (Re2O7),[206,207] which probably exist in the pristine and regenerated catalysts, resulting in irreversible deactivation of the catalysts.

Structure and properties of zinc species on H-ZSM-5

Based on the reasons explained above, Ga or Zn modified H-ZSM-5 is mainly used for EDA. Detailed characteristics of the Ga species are described in Subsection 3.2.1. Here, the nature of Zn species on H-ZSM-5 is presented in detail. The nature of Zn species on H-ZSM-5 has been investigated by many research groups because of its wide applications to various chemical reactions such as methane conversion and alkylation of benzene.[208]

Preparation of Zn/H-ZSM-5 can be done using various methods such as impregnation, ion-exchange, physical mixing, and CVD.[209-212] The active Zn species on H-ZSM-5 are classifiable into two types: ZnO species and ion-exchanged Zn species. The first, ZnO, exists on the external and internal surfaces denoted respectively as macrocrystalline ZnO and ZnO clusters. These ZnO species are analyzed conveniently using UV-Vis spectroscopy.[209,212] Macrocrystalline ZnO exhibits an absorption band at ca. 370 nm attributed to its band gap energy. By contrast, ZnO clusters exhibit an absorption band at around 270 nm. Chen et al. reported that ZnO clusters were identified using laser-induced luminescence spectroscopy.[213] When excited by a 244 nm laser, Zn/H-ZSM-5 containing ZnO clusters showed a purple luminescence band at ca. 440 nm. However, the ZnO species are reduced under reductive conditions at elevated temperatures. Some are exchanged with BAS (solid state ion-exchange), resulting in the formation of ion-exchanged Zn species and H2O.[214]Here, Oz represents the oxygen in the zeolite framework. This phenomenon was also observed in the case in which Zn metal was used instead of ZnO. Others are reduced to Zn metal vapor, resulting in the loss of Zn.

The existence of the ion-exchanged Zn species is roughly verified by temperature-programmed desorption of NH3 (NH3-TPD). In the TPD profile, two ammonia desorption peaks are observed respectively in low-temperature and high-temperature regions, denoted respectively as l-peak and h-peak.[215] The l-peak is attributed to the desorption of ammonia bonded to the ammonia adsorbed onto acid sites. The h-peak is ascribed to desorption of ammonia adsorbed onto the acid sites, which indicates that the peak area of l-peak does not reflect the amount of acid sites. Additionally, the desorbed ammonia derived from BAS is indistinguishable from that from Lewis acid sites.[216] Although the h-peak contains ammonia desorbed from BAS and Lewis acid sites, the h-peak intensity decreases after the ion-exchange,[209] indicating the exchange of BAS with Zn cations. The decrease in BAS can be verified using 1H MAS NMR spectroscopy or NH3-TPD with IR and mass spectrometry.[212,216] Formation of the ion-exchanged Zn species is verified by evaluating the electronic state of Zn because of its difference from that of ZnO.[210] XPS studies revealed that the ion-exchanged Zn species show a peak at ca. 1024 eV in the Zn 2p3/2 XP spectrum. In contrast, a peak corresponding to ZnO is observed at ca. 1021 eV.

Structures of the ion-exchanged Zn species are controversial. Berndt et al. proposed the structure of the ion-exchanged Zn.[211] They used temperature-programmed surface reaction (TPSR) with CO and observed formation of CO2 and H2, indicating the existence of [Zn(OH)]+, as shown in Scheme 5. During TPSR with CO, the water gas shift reaction proceeds at [Zn(OH)]+ and the nearest BAS, resulting in formation of the isolated Zn2+ (Oz–Zn2+–Oz).

Scheme 5

Structural change from [Zn(OH)]+ to isolated Zn2+ after the water gas shift reaction.

In addition, Biscardi et al. analyzed Zn/H-ZSM-5 using in situ Zn K-edge XANES spectroscopy.[217] The measurements were conducted in the presence of helium, hydrogen or propylene at up to 773 K. They concluded that [Zn(OH)]+ in pristine Zn/H-ZSM-5 is dehydrated at the nearest BAS, resulting in the isolated Zn2+ interacting with two ion-exchange sites under the propane dehydrogenation atmosphere. In their study, Zn/H-ZSM-5 with a small loading amount of Zn (<1.3 wt%) was used and, therefore, they described that two Zn2+ cations bridged by an oxygen atom (Oz–Zn2+–O–Zn2+–Oz), denoted as binuclear Zn species, could be formed at high Zn loading amount. Almutairi et al. investigated the structures of Zn in H-ZSM-5 prepared by CVD, impregnation, and ion-exchange methods.[212] The isolated Zn species are formed by CVD. In contrast, the impregnation or ion-exchange method induces formation of Zn species of various kinds, including the isolated Zn and oxygenated Zn clusters. The specific structure of the oxygenated Zn clusters was not elucidated. According to Penzien et al.,[218] the structure of Zn in *BEA-type zeolite depends on the Zn loading amount. Concentration of the binuclear Zn species increases at Zn/Al ≥ 0.15. Therefore, the large amount of Zn would result in formation of the binuclear Zn species in the case of H-ZSM-5. Furthermore, Tamiyakul et al. reported that the ion-exchanged Zn species existing in pristine Zn/H-ZSM-5 was reduced under H2 atmospheres based on XPS.[210] They proposed formation of ZnH+ in the presence of H2. Indeed, Zn hydrides were observed at room temperature using DRIFT spectroscopy.[219] Formation of ZnH+ would be attributed to the promotive H2 desorption at Zn sites through hydrogen backspillover.[220-222] In contrast, Gao et al. also evaluated the electronic state of Zn by XPS after reduction with H2 at various temperatures.[223] They described that ZnO clusters are reduced to [Zn(OH)]+ at 573 K, and reported that conversion of [Zn(OH)]+ to Zn2+ occurs at 673 K.

As described above, Zn species existing in H-ZSM-5 are diverse: ZnO, [Zn(OH)]+, Oz–Zn2+–Oz, Oz–Zn2+–O–Zn2+–Oz, and ZnH+. Under EDA conditions, ZnO and [Zn(OH)]+ are converted respectively to metallic Zn and Oz–Zn2+–Oz. The former is vaporized, resulting in the loss of Zn. The latter is anticipated as a major candidate for the active sites. Aleksandrov and Vayssilov investigated EDH at the isolated Zn2+, [Zn(OH)]+ and ZnH+ sites based on DFT.[224] The isolated Zn2+ at a paired Al site is more active for ethane activation than ZnH+ at one Al site because of the vicinity of two Al centers. The existence of the binuclear Zn species has not been demonstrated clearly. Pidko and van Santen calculated energy diagrams of EDH at binuclear Zn sites using DFT calculations.[225] The binuclear Zn species is not favorable for EDH, which is the initial reaction of EDA, because a stable intermediate is formed on the binuclear sites. However, the binuclear Zn species are presumed to be the active sites for EDA in recent studies.[226,227] Therefore, further studies to elucidate the structures of the Zn active sites must be conducted to clarify the nature of the structure–activity relation.

Reaction pathway of BTX formation

Reaction pathways of BTX formation were investigated using Pt, Zn, Ga, and Re/H-ZSM-5.[196,228-230] Because EDA is a complex reaction, the elucidation of each elementary step is challenging. Irrespective of the kind of active metal involved, the proposed reaction pathway is almost identical. As presented in Scheme 6, ethane is first dehydrogenated to ethylene at metal sites on the external or internal surface. Subsequently, ethylene is oligomerized, cyclized, and dehydrogenated to BTX in micropore channel systems. Under low conversion levels (low residence time), ethylene is the main product. However, selectivity to BTX and methane increases with decreased selectivity to ethylene under high conversion levels (high residence time). Also, C3 and C4 hydrocarbons, which are composed mainly of olefins, form during the reaction.

Scheme 6

Reaction pathway of EDA.

Hagen et al. demonstrated the formation of linear C4 olefins using a recirculation reactor.[231] They proposed that the linear C4 olefins are intermediates for BTX. Reportedly, oligomerization of ethylene proceeds even at room temperature.[232,233] In addition, ethylene is converted to propylene through oligomerization and cracking catalyzed by BAS.[234,235] Therefore, ethylene would first be oligomerized to C4 olefins and would subsequently be cracked to propylene. Propylene could also be converted to BTX through a similar pathway of ethylene aromatization.[236,237] In contrast with Scheme 6, a recent study of aromatization of ethylene proposed that hydrocarbon species also contribute to the reaction, similarly to the methanol conversion to aromatic hydrocarbons.[238] It can be demonstrated based on 13C MAS NMR that aliphatic sp3 carbon species attributable to alkylated aromatic hydrocarbons formed through 13C2H4 aromatization are consumed after 12C2H4 aromatization.

Actually, the distribution of BTX depends on the reaction temperature.[196] At lower temperatures (<723 K), their proportion is in the order of xylenes > toluene > benzene. In a middle temperature range (773–823 K), the proportion of toluene increases. Furthermore, benzene formation increases at higher temperatures. This tendency is the case in aromatization of ethylene, propylene, and methanol.[236,239,240] Iglesia et al. performed dehydroaromatization of propane using 13C-propane (12CH3–13CH2–12CH3).[241] They revealed that the number of 13C atoms included in benzene and toluene are statistically distributed, indicating that the propane carbon chain is rearranged randomly during the reaction. This is true probably because interconversion among olefins, which are the intermediates for aromatic hydrocarbons, occurs quite rapidly. Haag et al. reported that an approximate equilibrium distribution was attained for propylene conversion to olefins at 723 and 823 K.[240] Depending on the reaction temperatures, light olefins are produced at higher temperatures, resulting in formation of light aromatic hydrocarbons such as benzene and toluene. Based on these results, Keipert et al. proposed that an olefin pool in which mutual olefins are equilibrated is formed during EDA.[229] Recently, Liang et al. reported the finding of BTX formed from different olefin intermediates based on transient kinetic studies of ethane and ethylene aromatization.[242] They proposed that the distribution of BTX might depend on the partial pressure of the olefin intermediates. In addition, Choudhary et al. reported that the proportion of BTX in aromatization of olefins is dependent on the space velocity.[237,243] Dependence of the proportion of BTX on space velocity (conversion) is similar to dependence on temperature. The partial pressure of the intermediates varies with conversion of reactants, which can be controlled by the space velocity and reaction temperature because aromatization reactions are sequential reactions. Inclusively, the distribution of BTX can be controlled in the case in which the interconversion of olefins occurs under kinetic conditions. Under such conditions, other possible reactions such as hydro-dealkylation and trans-alkylation of aromatic hydrocarbons, which are catalyzed by BAS, might also affect the distribution of BTX. Fundamentally, these reactions are not considered for EDA. In addition to elucidation of the reaction pathway, the side reactions described above should be regarded as maximizing selectivity to valuable chemicals such as p-xylene.

The nature of active metals is a crucially important factor in the aromatization of light alkanes. Generally, the first dehydrogenation step (C–H bond activation) is a rate-determining step over H-ZSM-5. The active metals promote dehydrogenation of the alkanes. In contrast, Biscardi and Iglesia proposed the hydrogen desorption step as rate-determining in dehydrogenation of propane over Zn/H-ZSM-5.[220] Indeed, Zn/H-ZSM-5 and ZnO exhibit hydrogen spillover and back-spillover effects,[221,222] indicating that hydrogen desorption is promoted by virtue of Zn. In addition, the active metals influence BTX formation. Over H-ZSM-5, dienes and cyclodienes are intermediates in aromatization of ethylene, as reported by Batchu et al.[244] As presented in Scheme 7, aromatization over H-ZSM-5 fundamentally proceeds through oligomerization, cyclization, and dehydrogenation via hydrogen transfer with formation of alkanes.[245] In contrast, modification of H-ZSM-5 with Zn, Ga and silver (Ag) enhances selectivity to BTX in aromatization of ethylene,[223,226,238,246-248] indicating the promotive effect of the active metals on formation of aromatic hydrocarbons. Bandiera and Taȃrit proposed that dehydrogenation of cycloalkenes to form BTX competes with ethane activation.[249] They demonstrated that EDA was inhibited by addition of cyclohexadiene, regarded as a precursor of BTX, to the ethane feed.[250] Therefore, the active metals contribute not only activation of light alkanes but the formation of BTX because aromatization of the intermediates would readily proceed in comparison to hydrogen transfer catalyzed by BAS. A recent study of methanol to aromatic hydrocarbons also revealed that introduction of Zn to H-ZSM-5 enhances dehydrogenation ability, which leads to high yield of aromatic hydrocarbons, and which inhibits hydrogen transfer because of the small amount of BAS induced by ion-exchange with Zn.[251]

Scheme 7

Aromatization of oligomers over H-ZSM-5.

Recent studies of Mo/H-ZSM-5 for dehydroaromatization of methane have elicited new insights into the formation of aromatics.[252-255] Generally, Mo/H-ZSM-5 is presumed to be a bifunctional catalyst, as reviewed in the literature.[256-258] Briefly, activation of methane proceeds on carburized Mo species; subsequently, BAS promote aromatization of intermediates. Kosinov et al. demonstrated that methane conversion to benzene proceeds over Mo/Silicalite-1 with no Brønsted acidity.[252] They proposed that BAS contribute to high dispersion of Mo species, which efficiently convert methane to benzene. Their studies further revealed the existence of radical carbon species in micropores.[253] These confined carbon species play a role as intermediates in the dehydroaromatization of methane.[254] Therefore, in analogous to methane, EDA over Mo/H-ZSM-5 is expected to proceed through the radical carbon species, as depicted in Scheme 8. Indeed, Uslamin et al. demonstrated clearly in EDA that reduction of Mo and subsequent building up the carbon species in micropores proceed through ethane decomposition.[259] Benzene formation is initiated after the two-step induction period. Such was not observed in case of Ga/H-ZSM-5.

Scheme 8

Possible reaction pathway of EDA over Mo/H-ZSM-5.

In addition, Gascon and co-workers explored the nature of the carbon included in Mo (oxy-)carbide.[260] They performed 12CO treatment to form Mo (oxy-)carbide and conducted pulse experiments using 13CH4. They demonstrated that 12C included in Mo (oxy-)carbide is consumed to form ethylene and benzene. The nature of Mo species is also investigated elsewhere.[261-263] It is noteworthy that the characteristic nature of Mo (oxy-)carbide might not be exhibited in the aromatization of ethylene. After dehydroaromatization of methane and ethylene over Mo/H-ZSM-5, stable selectivity to aromatic hydrocarbons and reactive carbonaceous deposits was achieved in the case of methane.[264] They concluded that ethylene is unlikely to be the main intermediate for dehydroaromatization of methane. Therefore, the reaction pathway of BTX formation would be considered carefully in EDA over Mo/H-ZSM-5.

Catalyst deactivation by coke formation

Deactivation of H-ZSM-5 modified with active metals presents great difficulty. Coke formation over H-ZSM-5 is catalyzed by BAS through over-oligomerization and aromatization. In addition, elements of the active metals and their amounts influence the ethylene to aromatic hydrocarbons step.[238] Various techniques including thermogravimetric analysis, and Raman and 13C MAS NMR spectroscopy have been used to analyze coke species. Aspects of coke formation on zeolite catalysts are summarized in the literature.[265] The coke species are mainly polycyclic aromatic hydrocarbons under aromatization conditions. They induce blocking of micropore channels, leading to eventual deactivation of the catalysts. Therefore, the catalysts must be periodically regenerated under oxidative conditions.[266]

Addition of a promoter to a metal-modified H-ZSM-5 is one effective means of suppressing coke formation. Robinson et al. performed Fe addition to Mo/ZSM-5 for EDA.[266] The Fe promoter improved the catalytic stability because carbon nanotubes were formed on the H-ZSM-5 external surface. Formation of the carbon nanotubes, which have ordered structures, is expected to maintain openings of the micropore channels. Another example is Pt addition to Ga/H-ZSM-5.[267] In addition to the stability, Ga–Pt/H-ZSM-5 exhibited higher ethane conversion and selectivity to aromatic hydrocarbons than Ga/H-ZSM-5 and Pt/H-ZSM-5. The high stability of Ga–Pt/H-ZSM-5 would be attributed to consecutive removal of the coke species through hydrogenolysis during the reaction. Indeed, Marecot et al. performed TPSR with H2 using coke-deposited Pt, Re and iridium (Ir) catalysts.[268] Methane formation through hydrogenolysis of the deposited coke was verified. The methane formation temperature increased in the order of Ir < Re < Pt.

Modification of BAS is an alternative means of inhibiting coke formation. The BAS are located on the external and internal surfaces. However, the BAS on the external surface are not expected to exhibit shape selectivity to BTX formation because they are not located in the micropore channels. The formation of coke precursors such as polycyclic aromatic hydrocarbons is rather favorable. Epelde et al. investigated the deactivation pathways of H-ZSM-5 during ethylene conversion to propylene.[269] They inferred that the coke species composed of condensed aromatics and long aliphatic chains are deposited on the external surface, which indicates that the removal of BAS located at the external surface is effective. Inagaki et al. reported that nitric acid (HNO3) treatment of H-ZSM-5 synthesized without organic structure-directing agent removes Al (BAS) selectively from its external surface.[270] In addition, control of the distribution of framework Al would also be an effective means. Generally, the MFI topology has 12 T-sites, which are candidates for Al sites, as shown in Fig. 13. Among them, T1, T4, and T6 do not face the intersections. Although T4 is located at the sinusoidal channels, T1 and T6 are located at the straight channels. The distribution of the Al sites can be controlled by the synthesis conditions.[271,272] Liu et al. synthesized H-ZSM-5 with different Al distribution and performed ethane and ethylene aromatization over Pt/H-ZSM-5.[273] They concluded that BAS located at the intersections tend to contribute to the formation of aromatic hydrocarbons. H-ZSM-5 with much Al located at the intersections exhibited high and stable conversion and BTX yield. The remaining Al sites located at the straight and sinusoidal channels can be expected to induce coke formation, resulting in the blocking of micropore channels and the inhibition of diffusion of the products.

Fig. 13

Twelve T-sites of MFI topology.

The proximity of a Brønsted acid site to the nearest one is regarded as important for the suppression of coke formation. Zhang et al. prepared Zn/H-ZSM-5 using nano-ZSM-5 with crystalline size of approximately 100 nm.[227] Because of the suppression of carbon deposition, Zn/nano-ZSM-5 exhibits higher stability for EDA than Zn/micro-ZSM-5 does. In micropores of nano-ZSM-5, the short diffusion path can be expected to facilitate the escape of aromatic hydrocarbons, leading to mitigation of consecutive conversion of the products to coke. It is noteworthy that nano-ZSM-5 has a larger external surface area than micro-ZSM-5. Assuming that the density of BAS on the external surface is equal to that on the internal surface, the contribution of BAS located on the external surface would be enhanced in the case of nano-ZSM-5. Therefore, selectivity to products would also depend on the crystalline size because BAS on the external surface do not exhibit shape selectivity to BTX formation. Introduction of mesopores, which probably improve mass diffusion, to ZSM-5 enhances the catalytic performance.[274] It also contributes to an increase in coke capacity; ZSM-5 with the mesopores exhibited high activity and stability in spite of the large amount of carbon deposition.[275] In another study, Ye et al. synthesized pillared ZSM-5 with various lamellar thickness.[276] They performed EDA over Mo/pillared-ZSM-5 and demonstrated that ZSM-5 with thick layers (long diffusion path) and moderate Si/Al ratios exhibited high activity, selectivity to aromatic hydrocarbons, and stability. In their study, ZSM-5 with low Si/Al ratios (the large amount of BAS) is unfavorable because of its rapid deactivation. However, ZSM-5 with low Si/Al ratios has high ion-exchange capacity, which is rather convenient to incorporate large amounts of ion-exchanged metal cations such as Zn. In this case, dealumination as a post-synthetic method is a useful means to control the amount of BAS. Saito et al. performed steam treatment of Zn/H-ZSM-5 for EDA.[277] They revealed that the framework Al with proton (BAS) is preferentially dealuminated through the steam treatment after ion-exchange with Zn. However, the framework Al with Zn cations was unaffected by the steam treatment. The steam-treated Zn/H-ZSM-5 exhibited high and stable activity for EDA, respectively, by virtue of optimal and large amounts of BAS and Zn.

Performance of EDA catalysts

Representative performance of EDA catalysts is presented in Table 3. The reaction is conducted at around 873 K using zeolites with low Si/Al ratios. Further studies of the properties of active metals and reaction mechanism are indispensable to enhance their catalytic performance. In addition, the improvement of catalytic stability is crucially important for the practical use of EDA on a large scale. As described above, studies of various kinds have been conducted, examining promoters, zeolite synthesis, and post-synthetic methods to elucidate methods of inhibiting coke formation. From a practical perspective, the reaction system would be pressurized to make the reactor volume small because large amounts of hydrogen are formed through EDA (reaction (14)). Recent study of dehydroaromatization of methane[255] has revealed that catalytic stability can be improved at 15 bar through the rapid hydrogenation of coke at high pressure. Additionally, hydrogen removal from the reaction system can enhance BTX yield.[278] Therefore, further studies conducted from chemical engineering perspectives must be undertaken to support EDA process development.

Representative performance of EDA catalysts

Catalyst	Si/Al	Temperature/K	C2H6 conversion/%	Aromatics selectivity/%	Aromatics yield/%	Ref.
Ga/H-ZSM-5	15	873	28.2	57.3	16.2	199
Pt–Ga/H-ZSM-5	21	823	34.4	62.0	22.1	278
Zn–H-ZSM-5	11.5	773	15.4	48.0	7.4	226
Zn–H-ZSM-5	15	873	41.9	48.0	20.1	242
Mo/ZSM-11	27	973	99.7	56.7	56.5	193
Re/ZSM-5	30	873	48.0	59.0	28.2	203

Conclusion and perspectives

Recent progress in catalysts for EDH and EDA has been described from the perspective of the nature of active sites and reaction mechanisms. In terms of EDH, supported Pt-, Cr- and Ga-based catalysts have been intensively investigated.

Generally speaking, Pt-based catalysts exhibit high activity and selectivity at low temperatures (<873 K). Fundamentally, promoters such as Sn are used for Pt catalysts to enhance their performance. Modification of Pt with Sn decreases the number of “cus-Pt”, at which sites hydrogenolysis and coke formation proceed, if Pt–Sn alloys are formed or not. In addition, the promoters affect the electronic state of Pt, although the intrinsic effects are under discussion (electron donation or modification of the Pt 5d bands). The change in the electronic state of Pt enhances TOF concomitantly with fast desorption of ethylene. For Pt-based catalysts, structure–activity and electronic state–activity relations can be expected to be mutually interactive. Practically, the structure of Pt active sites (fcc or hcp) must be examined with the surface composition of Pt/promoter ratios. Determination of the active sites contributes to construction of calculation models. In theoretical studies, coverage of adsorbates such as hydrogen affects the stability of the intermediates. Therefore, calculation conditions must be close to the realistic ones to predict the active site characteristics.

By contrast to Pt catalysts, Cr-based catalysts are cost-effective. Among the Cr species, trivalent monochromate and polychromate are candidates for use as active sites in the absence of CO2. On the other hand, CO2 co-feed enhances activity by virtue of redox properties. For the Cr catalysts, it is challenging to evaluate the activities of respective Cr species because both monochromate and polychromate usually exist. In addition, the reaction mechanisms that take place in the presence of CO2 must be evaluated carefully: for instance, kinetic analyses at low conversion levels and elucidation of CO2 roles using isotopes must be supported. Additional studies are expected to be indispensable for the assessment of Cr catalysts. However, these findings reveal that ethane activation is rather more favorable through the redox mechanism than through the Langmuir–Hinshelwood-like mechanism. Such is the case with other metal oxides with redox properties, such as La–Mn perovskites. These redox materials can be applied to EDH using chemical looping.[279]

Because of their cost-effectiveness and non-toxicity compared to Pt and Cr, Ga-based catalysts are also candidates for use for EDH. However, additional improvement of the catalytic performance must be done to make the Ga-based catalysts alternative candidates for them. It is noteworthy that zeolite supports in addition to the conventional supports including Al2O3, SiO2 and TiO2 are used for Ga in contrast with Pt and Cr. Zeolite supports contribute to the formation of characteristic Ga active sites, as described in Subsection 3.3.1. The Ga species in zeolite micropores have various structures such as [GaH2]+ and GaO clusters dependent on the reaction atmospheres. In recent studies, zeolite-supported Fe and In catalysts are applied to EDH. The characteristic nature of zeolite can increase the feasible options of active metals.

As for EDA catalysts, Ga, Zn or Mo supported on ZSM-5 has been investigated. Studies of EDA have specifically pursued elucidation of the structure of ion-exchanged Zn in ZSM-5 and methods to inhibit coke formation. Among the proposed structures, the isolated Zn (Oz–Zn2+–Oz) is a probable active site. The specific structures of oxygenated Zn species such as the binuclear Zn (Oz–Zn2+–O–Zn2+–Oz) are not demonstrated clearly. However, it is plausible that the properties of Zn depend on, for example, the preparation method and loading amount. To suppress coke formation, the Al location and the proximity of BAS in ZSM-5 should be controlled. For Mo, the reaction pathway is expected to be different from that of Zn. Therefore, further studies of the reaction pathways of BTX formation are expected to be indispensable for improvement of the catalytic performance.

To implement EDA practically, the activity and selectivity to aromatic hydrocarbons must be enhanced. Considering studies of EDH, CO2 co-feed is one option. Indeed, EDA with CO2 was performed over Ga and Zn/H-ZSM-5.[280,281] Furthermore, other active metals with redox properties are expected to exhibit high catalytic performance for EDA with CO2 because the ethane activation step can proceed via the redox mechanism (Scheme 1). In the presence of CO2, Ga/H-ZSM-5 is apparently preferable to Zn/H-ZSM-5, probably because H2O is formed through RWGS. The existence of H2O promotes dehydrogenation in case of Ga/H-ZSM-5.[135] The isolated Zn can be converted to [Zn(OH)]+, which would be inactive for aromatization, in the presence of H2O. In addition, a recent study of dehydroaromatization of methane has demonstrated that addition of metallic Zr to Mo/H-ZSM-5 enhances catalytic performance by virtue of efficient removal of H2 through hydrogen absorption with ZrH1.75 formation.[282] Removal of H2 from the reaction system can shift the equilibrium of EDH forward, leading to high catalytic activity.

In conclusion, fundamental research efforts must be undertaken to develop catalytic processes for converting ethane to valuable chemicals. Recent progress in catalytic science and technology for ethane conversion has gradually revealed the intrinsic characteristics of the active sites indispensable to ethane activation. These new findings are expected to develop not only methods for efficient production of the valuable chemicals from ethane but for bridging technologies to guide future petrochemical industry development to production of a sustainable society.

Notes

The molecules and crystalline structures were visualized by GaussView 5 and VESTA softwares.[283,284]

Conflicts of interest

There are no conflicts to declare.