Current status and perspectives in oxidative, non-oxidative and CO₂-mediated dehydrogenation of propane and isobutane over metal oxide catalysts

Abstract

Conversion of propane or butanes from natural/shale gas into propene or butenes, which are indispensable for the synthesis of commodity chemicals, is an important environmentally friendly alternative to oil-based cracking processes. Herein, we critically analyse recent developments in the non-oxidative, oxidative, and CO₂-mediated dehydrogenation of propane and isobutane to the corresponding olefins over metal oxide catalysts. Particular attention is paid to (i) comparing the developed catalysts in terms of their application potential, (ii) structure–activity–selectivity relationships for tailored catalyst design, and (iii) reaction-engineering aspects for improving product selectivity and overall process efficiency. On this basis, possible directions for further research aimed at the development of inexpensive and environmentally friendly catalysts with industrially relevant performance were identified. In addition, we provide general information regarding catalyst preparation and characterization as well as some recommendations for carrying out non-oxidative and CO₂-mediated dehydrogenation reactions to ensure unambiguous comparison of catalysts developed in different studies.

---

Tatiana Otroshchenko

Tatiana Otroshchenko graduated from the Moscow State University in 2012. In 2014 she started her PhD on the development of the catalysts for non-oxidative dehydrogenation of light alkanes in the Leibniz-Institute for Catalysis in Rostock under the supervision of associate Professor Evgenii V. Kondratenko. Her PhD was partially supported by DAAD scholarship and by the scholarship of the president of the Russian Federation for study abroad. In 2018 she defended her PhD and won a personal grant from DFG. Now she leads her own research project on metathesis of ethylene with butenes in the Leibniz-Institute for Catalysis.

Guiyuan Jiang

Guiyuan Jiang received his BS and MS degrees from China University of Petroleum, Beijing under the supervision of Prof. Chunming Xu, and PhD degree from the Institute of Chemistry, Chinese Academy of Sciences with Prof. Yanlin Song. He is currently a Professor in University of Petroleum, Beijing. He was a visiting postdoctor at the University of California, Riverside in 2010 with Professor Pingyun Feng, and visiting scholar in Prof. Yadong Li's group in Tsinghua University in 2013–2014. His research interests mainly focus on energy catalysis, including catalytic conversion of light hydrocarbons and artificial photosynthesis.

Vita A. Kondratenko

Vita A. Kondratenko graduated from the Novosibirsk State University in 1992. In 2000, she has started to work at the Institute for Applied Chemistry Berlin-Adlershof. She received her PhD from the Humboldt-Universität zu Berlin under supervision of Professor M. Baerns in 2005. In 2006 she won a personal grant from DFG. She is currently a senior scientist at the Leibniz-Institute for Catalysis in Rostock. Her research interests are mainly focused on the evaluation of mechanistic aspects of catalytic transformations of light hydrocarbons and determining kinetics parameters of single steps of these reactions.

Uwe Rodemerck

Uwe Rodemerck studied chemistry at the Technische Hochschule Merseburg and received his diploma in homogeneous catalysis under the supervision of Prof. R. Taube in 1985. After working on reaction engineering at the Academy of Sciences of the GDR he worked at the Institute for Applied Chemistry Berlin-Adlershof and received 1995 his PhD in heterogeneous catalysis from the University Bochum under the supervision of Prof. M. Baerns. Currently, he is leader of the research group “High-Throughput Technologies” at the Leibniz-Institute for Catalysis in Rostock. His research interests focus on high-throughput catalysis, reactions of small hydrocarbons (C1–C4) and CO2-based Fischer–Tropsch synthesis.

Evgenii V. Kondratenko

Evgenii V. Kondratenko graduated from the Novosibirsk State University in 1991. He earned his PhD in 1995 from the Institute of Chemistry of Natural Materials in Krasnoyarsk. In 1997, he was awarded Alexander von Humboldt Foundation fellowship at the Institute for Applied Chemistry Berlin-Adlershof (group of Prof. Baerns). In 2007, he obtained his habilitation degree from the Technical University Berlin. He is currently associate Professor at the University Rostock and leader of the group “Reaction mechanisms” at the Leibniz-Institute for Catalysis in Rostock. He is engaged in the development and understanding of catalysts for valorization of C1–C4 alkanes, CO2 and NH3.

---

1 Introduction

The modern chemical industry is heavily relying on catalysis for the conversion of raw materials into various building blocks.^1,2 Among them, propene and butenes are highly important due to their versatile applications (Fig. 1). Their demand is continuously increasing. At the present time, there are several commercial technologies for their production. They utilize natural gas, crude oil, or coal as raw materials (Fig. 1). Steam or fluid-catalytic-cracking of different oil fractions are widely used and provide the desired olefins but suffer from low and not easily adjustable selectivity to these products. Moreover, they cannot fulfil the steady increasing demand for propene. Another shortcoming of these processes is finite natural resources of crude oil. Natural gas and coal are expected to be longer available and can also be used for production of low olefins. Technologies on the basis of these raw materials can be divided into direct and indirect routes (Fig. 1). The latter option starts with the generation of syngas (a mixture of CO and H₂) followed by Fischer–Tropsch (FT) or methanol synthesis. Methanol is further converted into lower olefins or propene through methanol to olefins^3–5 or methanol to propene⁶ technologies. The FT process is not selective to the desired lower hydrocarbons, as long-chain hydrocarbons and unfortunately methane are also co-produced. The economy and environmental impact of the methanol and FT processes suffer from syngas production requiring high amounts of energy and resulting in strong CO₂ emissions because of high endothermicity of steam reforming of methane or gasification of coal. The latter raw material is, in addition, not environmentally friendly. Some examples of damaging environmental influences of coal are emissions of particulate matter and methane upon mining, impurities such as heavy metals, sulphur- and nitrogen-containing compounds and radioactive as well as high CO₂ emissions upon its gasification. Under this consideration, conversion of C₂–C₄ alkanes present in natural gas appears to be attractive. Steam cracking of ethane, second main component of natural/shale gas, is a mature technology for production of ethylene. This chemical can be dimerized to n-butenes followed by the metathesis of 2-butenes with ethylene to propene (Fig. 1). Although the metathesis reaction is highly selective and efficient,^7,8 the above approach can be profitable only when an excess of ethylene is available.

Fig. 1 Raw materials, their main conversion ways into propene and isobutene as well as products formed from these olefins.

In comparison with the above-mentioned commercial technologies, direct dehydrogenation of C₃–C₄ alkanes present in natural/shale gas or formed in oil-based cracking processes is attractive for on-purpose direct production of the corresponding olefins. There are three options for carrying out alkane dehydrogenation: (i) without any oxidant (non-oxidative dehydrogenation), (ii) with oxygen or air (oxidative dehydrogenation) and (iii) using CO₂ as mild oxidant or for shifting the equilibrium in option (i) via reaction with H₂. The oxidative approach is free of thermodynamic limitations, while thermodynamic constraints limit alkane conversion in the non-oxidative or CO₂-mediated dehydrogenation. The latter reaction has the lowest equilibrium constant in the temperature range between 400 and 650 °C (Fig. 2).

Fig. 2 Equilibrium constants (K_eq.) of non-oxidative, oxidative, or CO₂-mediated dehydrogenation of propane or isobutane.

The CO₂-mediated dehydrogenation (CO₂-DH) of C₃–C₄ alkanes has attracted attention of researchers due to possible application for control of global CO₂ emissions. Such potential advantage of this approach can, however, be convincing if green (CO₂-neutral) energy will be used for this strongly endothermic reaction. In addition, an efficient and CO₂-neutral technology for separation of CO formed from CO₂ in this reaction is required. These aspects, state of the art catalysts and mechanistic aspects of the CO₂-DH of propane are discussed in recent reviews.^9–11

Despite the energetic advantages, the oxidative alkane dehydrogenation is not commercialized. The main reason is too low selectivity to the desired olefins at industrially relevant degrees of alkane conversion and the necessity to use pure oxygen if the non-reacted alkane should be recycled. Large-scale air separation to obtain O₂ is an expensive and energy-demanding technology. A most recent extensive review about the oxidative dehydrogenation of ethane and propane was published 13 years ago in 2007.¹⁰ Activity- and selectivity-governing properties as well as reaction engineering aspects have been reviewed. Seven years later, Carrero et al.¹² have reviewed kinetic aspects of the oxidative dehydrogenation of propane over vanadium-based catalysts. In addition, approaches for catalyst preparation and characterization have also been discussed. Experimental requirements for carrying out and evaluating kinetic experiments have been defined. A brief analysis of the oxidative propane dehydrogenation to propene over V-based catalysts is given in a review published in 2018 and dealing with aerobic oxidations of C₁–C₄ alkanes.¹³ In general, due to a very large number of available data and sometimes contradictive statements in different studies, no statistically proven relationships between catalyst physicochemical property, reaction conditions and performance could be established up to now.

The non-oxidative dehydrogenation of propane or isobutane is applied on large scale and its share is expected to grow in the coming years.^14–17 From a fundamental viewpoint, non-oxidative alkane dehydrogenation is also of great significance because it is a good model reaction for studying fundamentals of the activation of C–H bond. Commercially applied catalysts are composed of Al₂O₃ with supported Pt or CrO_x species responsible for alkane dehydrogenation.^15–18 There are, however, drawbacks related to the commercial catalysts. According to the U.S. Occupational Safety and Health Administration, workplace exposure to Cr(VI) may cause various health effects.¹⁹ High cost of platinum and the necessity to redisperse sintered Pt species in spent catalysts by ecologically harmful Cl₂ or Cl-containing compounds are the main drawbacks of Pt-based catalysts.¹⁸ Nevertheless, a major part of the ongoing research is still dealing with Pt-based catalysts and is aimed at reducing Pt amount without losing high selectivity at improved on-stream stability and durability. To account for the weaknesses of the commercial catalysts, various groups around the globe try to develop alternative catalysts with supported VO_x, GaO_x, InO_x, CoO_x, ZnO_x or SnO_x species of different structure. Recently, bulk catalysts based on ZrO₂, TiO₂ or Al₂O₃ were also demonstrated to show promising performance. Challenges and developments in this area before 2019 are discussed in several representative reviews.^{15–18,20,21}

Compared with published accounts on specific ways of production of propene and isobutene, this review discusses developments in the oxidative, non-oxidative, and CO₂-mediated dehydrogenation of propane and isobutane over supported and bulk metal oxides. In addition, reaction engineering aspects are also analysed. The general aims are (i) to critically evaluate progress in catalyst development, (ii) to identify unifying guidelines for improving these processes, and (iii) to provide our personal view on future research directions concerning catalyst development and reactor operation. To avoid repetitions and to further distinguish our review from the relevant previous assays, we will analyse the developments in the non-oxidative and CO₂-mediated dehydrogenation made in the last 5 years. As there are no recent reviews on the oxidative dehydrogenation, the relevant studies after 2007 will be reviewed. In particular, the present review is focused on (i) the application potential of catalysts developed, (ii) the structure–activity–selectivity relationships for tailored catalyst design, (iii) molecular level aspects of alkane dehydrogenation, (iv) reaction-engineering aspects, (v) factors relevant for preparation of certainly structured supported MO_x (M stands for a metal) species or defective bulk MO_x, and (vi) methods for catalyst characterisation to identify active sites and to analyse coke formation. As contradictory information is obtained from tests under different experimental conditions, we also provide some recommendations for carrying out alkane dehydrogenation tests and for their correct evaluation.

2 Non-oxidative propane dehydrogenation

2.1 General information about PDH process and applied catalysts

The non-oxidative propane dehydrogenation to propene (PDH) is a thermodynamically limited, highly endothermic reaction that requires temperatures above 550 °C to achieve industrially relevant degrees of propane conversion. According to Le Chatelier's principle, the equilibrium conversion increases with a decrease in partial pressure of propane, while co-feeding of H₂ that is often used to suppress coke formation, has a negative effect on the conversion.¹⁸Fig. 3 shows the equilibrium conversion of propane as a function of partial pressure of this alkane in the absence of co-fed hydrogen at 550 °C or 600 °C. Commercial PDH processes typically operate at propane partial pressures above 0.3 bar.²² The Oleflex process utilizes a feed consisting of propane and hydrogen-rich recycle gas.²² The reaction feed used in the Snamprogetti/Yarsintez process can also be considered to contain hydrogen due to back mixing in a fluidized-bed reactor.^23,24

Fig. 3 Equilibrium propane conversion as a function of propane content at total pressure of 1 bar and temperatures of 550 °C or 600 °C. No-co-fed H₂ or C₃H₆, which negatively affect the conversion. The numbers stand for propane conversion degrees achieved using feeds with 10 or 40 vol% C₃H₈.

Classically, catalysts composed of a supporting material and supported catalytically active metal or metal oxide species are used for the PDH reaction and dehydrogenation of other lower alkanes (Fig. 4). Their activity and selectivity are largely determined by the fine structure of supported active species and their interaction with the support (Fig. 4). One of the challenges upon catalyst development is to synthesize specific supported structures where the atoms are arranged in a predetermined way. Even when such structures are prepared, they may alter under severe reaction conditions thus resulting in a change of the catalyst performance. In addition, it is difficult to controllably place dopants/promoters in the vicinity of active PDH species to tune performance of the latter. Against this background, an alternative concept for design of alkane dehydrogenation catalysts has been introduced a few years ago.²⁵ The dehydrogenation function is reserved to surface defects of bulk oxides of typically non-reducible metals and not to supported metal oxides or metals (Fig. 4). Representatives of such materials are oxides of zirconium, titanium, aluminium, europium, and gadolinium.^26–29 The below discussion is aimed to demonstrate and to analyse recent developments including mechanistic aspects of propane dehydrogenation over conventional supported (Section 2.2) and alternative-type bulk (Section 2.3) catalysts. A major focus is put on the latter materials as they have not been thoroughly reviewed up to now.

Fig. 4 Conventional and alternative-type catalysts as well as principles of their functioning in non-oxidative alkane dehydrogenation.

2.2 Conventional supported catalysts and their active species

Based on the available literature, we prepared Table 1 showing the nature of catalytically active species in the conventional catalysts with supported CrO_x, VO_x, ZnO_x, GaO_x, SnO_x or CoO_x species. As seen in this table, there are some contradictions in this regard. Thus, further studies are required to unambiguously identify the active sites and the ways for their purposeful creation as required for tailored catalyst design. Promoters applied in the last 5 years for the modification of such catalysts with respect to their physicochemical and catalytic properties are summarized in Table 2. After this general introduction, individual features of the above-mentioned catalysts are discussed in the below sections.

Table 1 Proposed active sites and species of supported catalysts for PDH

Catalyst	Suggested active species	Ref.
CrOx-Based	Four-coordinated Cr^3+	30 and 31
	Redox and non-redox Cr^3+	32 and 33
VOx-Based	V^3+ is more active than V^5+ and V^4+	34 and 35
	V^3+ and V^4+	36 and 37
	Isolated VOx species are more active than polymerized VOx species	36 and 37
	Polymerized VOx species are more active than isolated VOx species	38 and 39
	Monomeric and low-polymerized VOx species	40
	Al–O–V (for monomeric VOx), Al–O–V and coordinatively unsaturated V (for dimeric VOx), three-coordinated surface oxygen (for crystalline VOx)	41
ZnOx-Based	Zn^2+Ox	42 and 43
	Zn(OH)^+	44
GaOx-Based	Isolated four-coordinated Ga^3+Ox	45
	[GaH]^2+	46–48
SnOx-Based	Metallic Sn	49
	Isolated Sn^2+Ox	50
	Polymeric Si–O–Sn^2+	51
	SnOx species with oxygen vacancies adjacent to the Lewis acidic sites	52
	Highly dispersed SnOx	53
CoOx-Based	Metallic Co	54 and 55
	Tetrahedral Co^2+Ox species	56–59

---

Table 2 Promoters applied in the last 5 years for preparation of supported metal oxide catalysts and their effect on physicochemical and catalytic properties

Active species	Promoter	Promoter effect on physico-chemical properties	Promoter effect on catalytic properties	Ref.
CrOx	Zn	Enhancing dispersion of CrOx due to favouring the formation of spinel ZnCr2O4 phase	Enhancing activity	60
	Ni	Favouring formation of oligomeric chromium species	Enhancing activity	30
	Ce	Decrease in the amount of inactive isolated Cr^6+Ox, increase in the amount of oxygen vacancies, strong interaction between Ce and Cr	Enhancing activity, selectivity, stability	61
VOx	P	Moderating surface acidity	Improving stability	62
	Mg	Dispersing V2O5 nanoparticles into 2D VOx species	Improving stability	63
SnOx	K	(1) Potassium interacts with Sn–O–Sn bonds	(1) Decreasing activity	53
		(2) Decrease in surface acidity	(2) Enhancing selectivity
	Ni	(1) Promoting complete reduction of SnOx to metallic Sn	(1) Decreasing activity	51
		(2) Formation of Ni3Sn2 alloy which promotes migration and recombination of hydrogen atoms	(2) Improving stability
	Co, Cu	Promoting complete reduction of SnOx to metallic Sn	Decreasing activity	51
	Pd	Suppressing coke formation due to hydrogenation ability, Improving resistance of well-dispersed SnOx species against sintering	Improving on-stream stability	49
ZnOx	Pt	Increasing dispersion of ZnOx, stabilizing Zn^2+	Enhancing activity and selectivity	42
CoOx	Fe	Formation of solid solution, generation of tetrahedral Co^2+ species	Enhancing activity and selectivity	59
	Mn, Cu	Generation of octahedral Co^3+ species	Decreasing activity	59
	Zr	Co–O bond weakening	Enhancing activity and selectivity	58

---

2.2.1 Cr-Containing catalysts. Catalysts with supported CrO_x species are commonly used in the large-scale Catofin process.^24,64 To improve their activity, selectivity, and on-stream stability various studies have been carried out. The kind of support and promoter as well as chromium loading were varied to investigate their effects on catalytic performance and the structure of supported species. They can be isolated Cr⁶⁺O_x, Cr⁵⁺O_x, Cr³⁺O_x, Cr²⁺O_x, oligomeric CrO_x and crystalline Cr₂O₃.²¹ It is, however, generally accepted that Cr³⁺O_x species are responsible for catalyst activity.²¹ The PDH reaction is initiated through the breaking of the C–H bond on Cr(III)–O site followed by β-H elimination and H₂ release on the same site.³¹

It has been demonstrated that surface acidity strongly influences the performance of CrO_x-based catalysts. On the one hand, the catalysts with low acidity exhibit high on-stream stability and anti-coking ability.⁶⁵ On the other hand, Brønsted acidic sites of support play an important role in propane activation thus enhancing catalyst activity.⁶⁶ As for many supported catalysts, the nature of supported CrO_x species and accordingly their PDH performance can be tuned through preparation method,³¹ support,^32,65 loading of CrO_x^32,67 or promoter.^30,60,61 Some examples about the influence of various promoters on physico-chemical and catalytic properties of CrO_x-based catalysts are shown in Table 2. It should be mentioned that when CrO_x is combined with ZrO₂ to prepare supported CrO_x/M(La or Y)ZrO_x⁶⁸ and Cr–Zr–O_x/MO_x(SiO₂)^69,70 or bulk Cr–Zr–O_x^68,71 materials, extremely high activity can be achieved as a result of synergetic effect between ZrO₂ and CrO_x. As surface defects of ZrO₂ in these materials are crucial for the high activity of such catalysts, they were classified by us as alternative-type materials and will be discussed in Section 2.3.

2.2.2 V-Containing catalysts. VO_x/support catalysts are highly active, selective to propene and demonstrate good durability in a series of PDH/regeneration cycles.⁷² Fresh catalysts possess VO_x species containing V⁵⁺, V⁴⁺ and V³⁺ cations. During a few minutes on propane stream V⁵⁺ cations are reduced to V⁴⁺ and V³⁺ which are supposed to be the active sites.^35–37 Among differently structured VO_x species, V₂O₅ crystallites are practically inactive, while V³⁺O_x and/or V⁴⁺O_x species in polymeric and/or isolated form show high activity.¹⁸ Which latter species show higher intrinsic activity and their exact structure under reaction conditions are, however, still under debate. It is often suggested that V³⁺ is the most active site.^34,35 Some authors provide evidence that isolated VO_x species possess higher intrinsic activity,^36,37 other suggest that polymerized VO_x species are more active.^38,39

It has been recently demonstrated that hydroxyl groups on supported VO_x species strongly influence catalytic activity and coke formation.⁷³ The catalysts containing hydroxyl groups on VO_x species created after H₂ treatment demonstrate lower PDH activity and lower coke deposition. It has been also shown that the performance of VO_x-based catalysts is strongly influenced by the preparation method,^74,75 the kind of vanadium precursor³⁹ and support,^76,77 vanadium loading,^36,37,40 and the kind of promoter (Table 2). All these parameters affect the structure of supported VO_x species and must be considered upon catalyst preparation (Section 8.1).

2.2.3 Zn-Containing catalysts. Several studies dealt with supported Zn-containing catalysts in the PDH reaction.^{42,44,78–82} However, such catalysts are mostly characterized by low activity and fast deactivation due to carbon deposition and/or ZnO reduction to metallic Zn followed by evaporation of the latter. Zn²⁺O_x species are believed to serve as active sites (Table 1). Therefore, preventing/suppressing transformation of Zn²⁺O_x to Zn⁰ is a decisive task upon preparation of ZnO-based catalysts. This can be achieved by the usage of appropriate preparation method (atomic layer deposition,^44,82 encapsulation of ZnO nanoparticles⁸⁰), modification of ZnO with Pt (Table 2)^42,81 or by selection of a proper support material.^43,78,79 Pt was reported to inhibit the reduction of ZnO due to electronic interactions between Pt and ZnO.⁴² Electron density transfer from ZnO to Pt increases Lewis acidity of ZnO. As a result, C–H bond activation and H₂ desorption are enhanced.

Zeolites were reported to stabilize small ZnO nanoclusters.^78,79 The amount of highly dispersed ZnO clusters interacting with framework oxygen atoms increases with an increase in the SiO₂/Al₂O₃ ratio in the zeolite.⁷⁹ Accordingly, high activity of ZnO supported on high-silica HZSM-5 is related to the high dispersion of ZnO_x species.

Very recently, Han et al.⁴³ developed highly active ZrO₂-supported catalysts with catalytically active ZnO_x species. Binary MZrO_x (M = Ce, La, Ti or Y) materials were used as supports. The most active Zn(4 wt%)/TiZrO_x catalyst revealed higher activity than the state-of-the-art catalysts with supported CrO_x, GaO_x, ZnO_x or VO_x species as well as bulk ZrO₂-based catalysts without ZnO. In contrast to ZrO₂-based catalysts in Section 2.3, coordinatively unsaturated Zr cations were not concluded to be the main active sites in these novel catalysts. Isolated tricoordinated Zn²⁺O_x species are responsible for the high activity and on-stream stability. Their formation and intrinsic activity depend on the kind of metal oxide promoter for ZrO₂ and the structure of the latter oxide. The stabilization of isolated Zn²⁺O_x species is favoured when ZrO₂ has an amorphous structure. It should be also mentioned that the presence of TiO₂ seems to increase the intrinsic activity of isolated Zn²⁺O_x species. Very recently, the same research group developed TiO₂-supported catalysts with ZnO_x and ZrO₂.⁸³ Small ZnO_x clusters with 1 to 3 Zn atoms were suggested to be responsible for propane dehydrogenation, while ZrO₂ enhances the intrinsic activity of ZnO_x.

2.2.4 Ga-Containing catalysts. Gallium oxide is known to be effective in the activation of various hydrocarbons and their dehydrogenation.^45,84 Thus, propane aromatization in the typical industrial Cyclar process is performed over bifunctional Ga-doped H-ZSM-5 catalysts.^85,86 The role of GaO_x species is believed to increase the rate of dehydrogenation steps, while cyclization of the formed olefin is proposed to occur on Brønsted acidic sites of the zeolite.⁸⁶

Although Ga-containing zeolites are known to be active in PDH, such catalysts show low propene selectivity at high degrees of propane conversion.⁸⁷ Thereby, many recent works were focused on increasing the selectivity.^87–89 The performance of Ga-containing catalysts could be mostly tuned by preparation method^87–89 and/or support.⁸⁴ It was suggested that strong metal–support interactions lead to higher dispersion of GaO_x species and therefore play a crucial role for catalyst activity.⁸⁴ Concerning propene selectivity, the presence of strong acidic sites has a negative effect since such sites provoke deep dehydrogenation, aromatization and coking reactions.⁸⁴

The nature of the active sites and the mechanism of propane dehydrogenation over Ga-based catalysts are still under debate. Only four-coordinated Ga³⁺–O_x Lewis acidic sites were observed under reaction conditions on the surface of single-site GaO_x/SiO₂ catalyst. When the coordination of GaO_x decreased, a significant decrease in catalytic activity was observed.⁴⁵ Such phenomenon supports the assumption that four-coordinated Ga³⁺O_x sites are catalytically active. Recently, Schreiber et al.⁴⁶ provided some evidence that Lewis–Brønsted acid pairs (Ga⁺ – Brønsted acidic site of support) are active sites for propane dehydrogenation over Ga/H-ZSM-5. Based on the results of DFT calculations, those authors proposed a bifunctional Lewis–Brønsted acid mechanism where the Brønsted acidic site protonates Ga⁺ forming [GaH]²⁺. A basic framework oxygen of Brønsted acidic site together with so-formed [GaH]²⁺ heterolytically breaks two C–H bonds in propane followed by elimination of H₂ and propene in two sequential steps. Similarly, experimental data and DFT calculations carried out for Ga/H-MFI suggest that [GaH]²⁺ cations are the most active sites for dehydrogenation of light alkanes over Ga/H-MFI.^47,48 It is proposed that the dehydrogenation reaction occurs via a heterolytic dissociation of C₃H₈ at [GaH]²⁺ site with a formation of [C₃H₇–GaH]⁺–H⁺ cation pairs.

2.2.5 Sn-Containing catalysts. Typically, Sn is used as a promoter in the commercial Pt-containing catalysts. It enhances both activity and on-stream stability due to geometric (isolation of Pt clusters) and electronic (electron transfer from Sn to Pt in PtSn particles) effects. It has also been shown that SnO_x/SiO₂ catalysts without Pt exhibit good catalytic performance in PDH.^49–51,53 For example, Sn-HMS containing 5 wt% Sn demonstrates 40% propane conversion and STY(C₃H₆) of 0.13 kg(C₃H₆) kg(cat)⁻¹ h⁻¹ with a selectivity to propene of 90% at 600 °C.⁵² There are some discrepancies concerning the nature of active sites (Table 1). Metallic Sn was suggested to be the active species.⁴⁹ A more detailed investigation led to the conclusion that polymeric Si–O–Sn²⁺ is the active species rather than isolated Sn⁴⁺O_x or metallic Sn species.⁵¹ In their further work, Chunyi Li and co-workers⁵⁰ have concluded that Sn²⁺ species are highly active and stable for PDH, while transformation of SnO_x species into metallic Sn can lead to irreversible catalyst deactivation due to Sn sintering and loss of metallic Sn, which melts at 232 °C. Therefore, stabilization of SnO_x species is an important task upon designing and preparation of Sn-containing catalysts. It has been shown that embedding SnO_x species into SiO₂ matrix results in a strong interaction between SnO_x and SiO_x species and leads to highly active and stable catalysts.^52,53 Compared with SnO_x/SiO₂ catalysts prepared by a conventional impregnation method, Sn-containing hexagonal mesoporous silica exhibits higher activity and stability due to the formation of strong Sn–O–Si bonds and thus inhibition of SnO_x reduction to metallic Sn.⁵²

The performance of Sn-containing catalysts can be affected by the presence of various additives. Some examples of their influence on physicochemical and catalytic properties of such catalysts are given in Table 2.

2.2.6 Co-Containing catalysts. Nowadays, catalysts with supported CoO_x species attract more and more attention because of their excellent ability to activate C–H bonds in alkane molecules.^56,90 There are two contradictory theories about the nature of active species in propane dehydrogenation (Table 1). According to the first theory based on the results of catalyst characterization by XANES, UV-vis and XPS, tetrahedral Co²⁺O_x species are the active sites for PDH, while metallic Co promotes side reactions such as cracking and coke formation.^{56–59,90–92} Contrarily, the second theory implies that highly dispersed metallic Co species formed in situ during the PDH reaction are the active species.^54,55 The adherents of this theory confirm their hypothesis by the existence of an induction period for the most of Co-based catalysts. During such period, CoO_x species are supposed to be reduced into metallic Co. The idea was also supported by a comparison of catalytic performance of freshly prepared and pre-reduced Co/Al₂O₃ catalysts.⁵⁴ Unlike fresh catalyst which showed an inductive period upon propane dehydrogenation, the reduced sample containing metallic Co did not have such period and revealed high activity.

2.3 Alternative-type bulk catalysts

Alternative-type bulk catalysts composed of oxides of zirconium, titanium, aluminium, europium or gadolinium are a relatively new class of alkane dehydrogenation catalysts (Fig. 4). Unlike the conventional supported catalysts, such catalysts were only scarcely described in a recent review of Hu et al.²¹ Below, we discuss the kind of active sites in such materials, factors affecting their creation and intrinsic activity, molecular-level insights and catalytic performance. In a separate Section 2.4, such catalysts will be compared with conventional supported catalysts in terms of selectivity–conversion values and productivity under industrially relevant conditions. For benchmarking purposes, the most active catalysts from a previous review of Weckhuysen and co-workers¹⁸ will also be discussed.

2.3.1 Active sites and mechanistic insights into PDH. ZrO₂-Based catalysts were firstly introduced as promising alternative catalysts for PDH in the end of 2015.²⁵ Since that time, 15 articles have been published and more than 100 different catalysts have been developed. They are environmentally friendly and demonstrate high activity and selectivity comparable to those of an industrial analogue of CrO_x–K₂O/Al₂O₃. The thorough investigation of bare ZrO₂ and ZrO₂ doped with different metal oxides brought to the conclusion that the presence of oxygen vacancies is essential for the high activity of the catalysts. It was originally suggested that coordinatively unsaturated zirconium cations (Zr_cus) and neighbouring lattice oxygen are catalytically active sites responsible for propene formation.⁹³ However, further deeper mechanistic studies combining both experimental techniques and density functional theory (DFT) calculations have concluded that two neighbouring Zr_cus⁴⁺ sites located at an oxygen vacancy form the active site.²⁶ Thus, by comparing the computational results obtained for the most stable monoclinic non-defective (ZrO₂(1̄11)) and oxygen-defective (d-ZrO₂(1̄11)) surfaces, it was established that Zr_cus sites present in d-ZrO₂(1̄11) open an alternative pathway for PDH with a lower activation barrier in comparison with ZrO₂(1̄11).²⁶ Two Zr_cus⁴⁺ sites are required for the homolytic C–H bond activation on d-ZrO₂(1̄11), while Zr⁴⁺ and neighbouring lattice oxygen participate in this process on ZrO₂(1̄11) (Fig. 5). It is noticeable that the rate-determining step for PDH over d-ZrO₂(1̄11) is the formation of H₂ (endothermic by 1.40 eV), while over ZrO₂(1̄11) it is methylene C–H bond activation (endothermic by 0.94 eV with an activation barrier of 1.25 eV).

Fig. 5 Schematic representation of C–H activation, dissociation, and propene formation steps of PDH catalysed by oxygen-defective m-ZrO₂(1̄11), oxygen-defective TiO₂(101), and γ-Al₂O₃(110).

The discovery of ZrO₂-based catalysts prompted further search for other bulk catalysts. Bare TiO₂, Al₂O₃, Eu₂O₃ and Gd₂O₃ were established to show reasonable activity in the non-oxidative dehydrogenation of propane and isobutane.^27–29,94 Similarly to ZrO₂, the activity of pristine TiO₂ is strongly influenced by the presence of oxygen vacancies experimentally detected by the EPR technique.²⁷ Based on the results of DFT calculations, the authors concluded that the presence of oxygen vacancies on the surface of TiO₂ facilitates the adsorption of propane and its further dehydrogenation. Two fold-coordinated oxygen atoms were suggested to be responsible for the adsorption of the dissociated derivatives of propane on a perfect TiO₂(101), while fourfold-coordinated titanium atoms surrounding the oxygen vacancy were the active sites for PDH over defective TiO₂(101) (Fig. 5).

Similar to ZrO₂- and TiO₂-based catalysts, the active species for bare Al₂O₃ were also concluded to be coordinatively unsaturated Al³⁺ sites.²⁸ Since such sites are Lewis acids, their presence was confirmed by Py-FTIR. Accordingly, the activity of bare Al₂O₃ was in a good correlation with the amount of Lewis acidic sites. DFT calculations performed by Dixit et al.⁹⁵ on non-hydroxylated (100) and hydroxylated (110) facets of γ-Al₂O₃ for the PDH reaction considered two different mechanisms, namely concerted and stepwise. The former mechanism implies propene formation in a single step, while the latter considers a sequential abstraction of hydrogen atoms. Thus, the authors revealed that the PDH mechanism is site-dependent. The highest computed TOF of the reaction at 600 °C was obtained for Al^III–O^III (III means tricoordinated) site pair (Fig. 5) on the hydroxylated (110) facet implying a concerted mechanism.

Very recently, Perechodjuk et al.²⁹ have demonstrated the application potential of Eu₂O₃ and Gd₂O₃ as catalysts for the PDH reaction. Similar to other previously discussed bulk catalysts, coordinatively unsaturated metal cations were suggested to participate in propane dehydrogenation.

Although coordinatively unsaturated cations in ZrO₂, TiO₂ and Al₂O₃ are essential for propane dehydrogenation, the nature of the active site on the surface of these oxides seems to be different (Fig. 5). Thus, two adjacent Zr_cus⁴⁺ sites (five- and sixfold-coordinated) participate in the formation of propene over oxygen-defective m-ZrO₂(1̄11).²⁶ Contrarily, one fourfold-coordinated Ti_cus⁴⁺ and twofold-coordinated O²⁻ on the surface of oxygen-defective TiO₂(101) are required for propane dehydrogenation to propene.²⁷ For the most active Al^III–O^III site pair on γ-Al₂O₃(110), the formation of propene occurs in a single step with participation of threefold-coordinated Al³⁺ and threefold-coordinated O²⁻ atoms.⁹⁵ Noticeably, for the latter case, only perfect γ-Al₂O₃(110) surface facet was considered. The reaction mechanism and the kind of active site might change on an oxygen-defective surface.

2.3.2 Factors influencing activity and selectivity.
Influence of phase composition and size of crystallites. Zhang et al.⁹⁶ have recently shown that the activity of bare ZrO₂ in PDH and the selectivity to propene can be tuned through the phase composition and/or the size of crystallites. Monoclinic ZrO₂ was found to be more active and selective than ZrO₂ stabilized in the tetragonal phase. To check if the phase composition influences the nature of active sites and individual reaction steps, elementary steps of propane activation and product formation over oxygen-defective monoclinic (m-ZrO₂(1̄11)) and oxygen-defective tetragonal (t-ZrO₂(101)) ZrO₂ were elucidated by means of DFT calculations.⁹⁶ Regardless of the phase composition, energy profiles during PDH reaction have the same shape and trend (Fig. 6) and two Zr_cus⁴⁺ cations are required for the PDH reaction. However, the energy values are different for the individual pathways. The formation of propene over m-ZrO₂(1̄11) occurs easier than over t-ZrO₂(101).

Fig. 6 The calculated energy profiles and optimized structures of intermediates along the pathway of propane dehydrogenation to propene over defective t-ZrO₂(101) (red) and m-ZrO₂(1̄11) (black) surfaces (Zr, light blue; C, grey; O, red; H, white). Reproduced from ref. 96 with permission from Elsevier, copyright 2020.

The results of catalytic tests demonstrated that for both phases the activity and the selectivity to propene increased with a decrease in the size of crystallites (Fig. 7(a and b)). The reason for the different catalytic activity of ZrO₂ samples was explained by their different ability to release lattice oxygen during reductive treatment. This catalyst property was determined by CO-TPR tests. The amount of consumed CO represents ZrO₂ reducibility. The activity of the catalysts was in a good correlation with this parameter; the higher the reducibility, the more active the catalyst is.

Fig. 7 (a) The rate of propene formation (r(C₃H₆)) over monoclinic, tetragonal, and amorphous ZrO₂ (m-ZrO₂, t-ZrO₂ and a-ZrO₂, respectively) versus the size of crystallites; (b) the selectivity–conversion relationships for propene over m-ZrO₂ with 9.1 nm crystallites and t-ZrO₂ samples with 3.7, 4.6 or 5.7 nm crystallites. Reaction conditions: T = 550 °C, feed composition: 40 vol% C₃H₈ in N₂. Adapted from ref. 96.

The effect of ZrO₂ reducibility on catalyst activity can be explained as follows. As above mentioned, the presence of oxygen vacancies is crucial for the activity of bulk catalysts because the active sites are coordinatively unsaturated metal cations located near such surface defects. Therefore, catalytic activity strongly depends on the concentration of oxygen vacancies. They can be produced during reductive catalyst treatment or in situ during the first minutes of PDH. It is obvious that the concentration of such defects would be higher for the sample with higher reducibility.

The increase in the selectivity to propene with a decrease in the size of crystallites was related to the lower ability of the samples with small crystallites to form coke. Propane conversion was about 15% for all samples with the exception for one t-ZrO₂ which could achieve the conversion value of only 8.5%. Thus, it was shown that the ratio of the rate of propene formation to that of coke formation increases with decreasing crystallite size implying that the active sites for propene formation and coke formation are different. The desired reaction is catalysed by Zr_cus sites, while regular surface Zr sites are responsible for coke formation.

Influence of treatment conditions. Since the active sites (coordinatively unsaturated metal cations) of bulk catalysts are located at oxygen vacancies, their creation should happen when lattice oxygen is removed. Such removal is possible during reductive treatment before catalytic reaction. The influence of different treatment conditions on the activity of bare and doped ZrO₂ as well as of bare TiO₂ was demonstrated in ref. 26, 27, 93, 97 and 98. In comparison with the air-treated samples, their H₂-treated counterparts revealed significantly higher activity. For example, the rate of propene formation measured at 550 °C over bare ZrO₂ pre-reduced in H₂ at 600 °C was more than 4 times higher than that determined for the same catalyst but treated in air at 600 °C (0.109 versus 0.043 mmol(C₃H₆) g_cat⁻¹ min⁻¹). For the bare TiO₂, the rate of propene formation at 600 °C over the sample treated in H₂ at 600 °C was about 3 times higher than over the sample treated in air at the same temperature (0.06 versus 0.02 mmol(C₃H₆) g_cat⁻¹ min⁻¹). The positive effect of the reductive treatment is due to the removal of lattice oxygen and therefore formation of active coordinatively unsaturated Zr_cus⁴⁺ or Ti_cus⁴⁺, which are present in much lower amounts in oxidized catalysts.

The amount of lattice oxygen removed during reductive treatment can be increased by (i) increasing reductive treatment temperature; (ii) increasing reductive treatment duration and/or (iii) using reducing agent(s) stronger than H₂. The influence of reductive treatment temperature on the activity of bare ZrO₂ or TiO₂ is shown in Fig. 8. The rate of propene formation at 550 °C over bare ZrO₂ gradually increases with increasing the temperature (Fig. 8(a)). It is however worth mentioning that for TiO₂ the dependence is opposite (Fig. 8(b)). The negative effect of high-temperature treatment is due to the over-reduction of TiO₂. Despite the removal of higher amount of lattice oxygen at higher reduction temperature, the creation of Ti_cus⁴⁺ does not happen since Ti⁴⁺ is reduced into inactive Ti³⁺. The negative effect of over-reduction was also observed for ZrO₂-based catalysts with supported Rh or Ru species.^93,99

Fig. 8 The rate of propene formation (a) at 550 °C over ZrO₂ as a function of the temperature of H₂ pre-treatment⁹⁷ and (b) at 600 °C over TiO₂ as a function of the temperature of H₂ pre-treatment.²⁷

The influence of reductive treatment time and the nature of the reducing agent on the rate of propene formation at 550 °C over bare ZrO₂ was investigated in ref. 26. A gradual increase in the activity was observed when the duration of hydrogen treatment was extended to 7 hours. Such long treatment is not attractive from an applied viewpoint. When H₂ was replaced by CO for the reduction purpose, the maximal activity was achieved after only 20 min treatment. Moreover, the CO-treated catalysts revealed about 3.5 times higher activity than their counterparts treated in H₂ for 6 hours. In comparison with H₂, CO can efficiently remove lattice oxygen and, more importantly, surface OH groups to generate Zr_cus sites.

Influence of promoter for ZrO₂. The kind of metal oxide dopant for ZrO₂ is another important factor affecting the intrinsic activity of Zr_cus sites. Thus, the presence of La³⁺, Y³⁺ or Sm³⁺ cations in the lattice of ZrO₂ positively influences the activity of ZrO₂, while the presence of Ca²⁺, Mg²⁺ or Li⁺ results in a decrease in the activity if compared with bare ZrO₂ (Fig. 9(a)). The different effects of the dopants were explained by their dissimilar ability to generate/stabilize Zr_cus active sites and their different effects on the intrinsic activity of Zr_cus sites. Thus, it was shown that the activation energy (E_a) of propene formation over Y-doped ZrO₂ is about 10 kJ mol⁻¹ lower than that over bare ZrO₂, La- or Sm-doped ZrO₂, while the E_a values determined for Ca- or Mg-doped ZrO₂ are 46 and 80 kJ mol⁻¹ higher (Fig. 9(b)). Such results clearly demonstrate that the presence of Y³⁺ positively influences the intrinsic activity of Zr_cus, while the presence of Ca²⁺ or Mg²⁺ has a negative effect. It is also worth mentioning that the dopant influences product selectivity. The selectivity to propene over doped ZrO₂ is higher than over bare ZrO₂ at a similar propane conversion. Such effect was related to a decrease in the concentration of strong acidic sites on the surface of ZrO₂. Consequently, secondary transformations of propene into coke are partially hindered.

Fig. 9 (a) The initial rate of propene formation at 550 °C over reduced ZrO₂-based catalysts; (b) apparent activation energy of propene formation determined over reduced catalysts. Data are from ref. 98.

Promoting ZrO₂ with chromium oxide results in highly active catalysts, which significantly outperform a commercial analogue of CrO_x–K/Al₂O₃.⁶⁸ Moreover, in comparison with the latter catalyst, such catalysts have up to 40 times lower Cr content but show superior activity. Bulk binary CrZrO_x or supported CrO_x/LaZrO_x catalysts which contain both Zr_cus⁴⁺ and Cr³⁺ showed much higher rate of propene formation at similar Cr surface densities than CrO_x/Al₂O₃ catalysts possessing only Cr³⁺ active sites (Fig. 10). Supported CrO_x/LaZrO_x catalysts demonstrated slightly higher propene selectivity (84–85%) than bulk binary CrZrO_x (80–84%) but worse durability at a degree of propane conversion of about 30%. The on-stream stability of the latter materials can be improved through promoting with Cs, Ca and/or P.⁷¹

Fig. 10 The rate of propene formation obtained at 550 °C over bulk CrZrO_x, supported CrO_x/LaZrO_x and CrO_x/Al₂O₃ and an analogue of industrial CrO_x–K/Al₂O₃. Reproduced from ref. 68 with permission from Elsevier, copyright 2020.

The origins of the high activity of CrO_x-promoted ZrO₂-based catalysts were thoroughly elucidated by Han et al.⁶⁹ Those authors prepared catalysts with supported CrZrO_x species. Mechanistic and kinetic tests were combined with characterization tests using the state-of-the art techniques, while molecular-level insights were derived from DFT calculations. Zr_cus sites were concluded to be the main active sites. The role of CrO_x promoter was explained as follows. Firstly, the promoter improves the ability of ZrO₂ to release its lattice oxygen and accordingly to generate Zr_cus sites. This positive effect also depends on the strength of interaction between CrO_x, ZrO₂ and the support. SiO₂ weakly interacting with CrO_x favours the stabilization of supported binary CrZrO_x species that is also essential for the ability of ZrO₂ to release its lattice oxygen. Secondly, CrO_x enhances the intrinsic activity of Zr_cus sites as concluded from DFT calculations. Fig. 11 shows the energy profile along the pathways in the course of propane dehydrogenation over CrZrO_x and defective t-ZrO₂(101). ZrO₂ in CrZrO_x is stabilized in the tetragonal phase. Regardless of the presence of promoter, two Zr_cus cations form the active site. As seen in Fig. 11 the highest apparent barrier for PDH over 2O_v–CrZrO_x(101) is lower than that over 1O_v–t-ZrO₂(101) (0.95 versus 1.13 eV) implying that the presence of chromium provides more energetically favourable reaction pathway. It is noticeable that unlike PDH over t-ZrO₂(101), where H₂ formation is the rate-limiting step, PDH over 2O_v–CrZrO_x(101) is limited by the breaking of the C–H bond in propane.

Fig. 11 Calculated energy profiles and optimized structures of intermediates along the pathway of PDH to propene over 2O_v–CrZrO_x(101) and 1O_v–t-ZrO₂(101) surfaces (Cr, purple; Zr, light blue; C, grey; O, red; and H, white). Reproduced from ref. 69 with permission from American Chemical Society, copyright 2020.

Influence of supported hydrogenation-active metal. It is well-known that the presence of hydrogenation-active metal on the surface of metal oxides enhances their reduction due to an easier generation of surface active hydrogen species from gas-phase H₂.^100–102 In case of bulk catalysts, where oxygen vacancies are crucial for the high activity, such property of the supported metals could be useful for promoting the generation of such defects. Indeed, recent studies on ZrO₂-based catalysts proved that the presence of tiny (0.05 wt% and lower) amounts of Ru or Rh significantly increases the activity of ZrO₂-based materials.^93,99 In addition to this promoting effect, the supported metal can also indirectly participate in the PDH reaction. DFT calculations predict that the energy for H₂ desorption (the rate-determining step) over Rh/d-ZrO₂(1̄11) is lower than that over bare d-ZrO₂(1̄11).⁹⁹ This result suggests that H₂ desorption is accelerated in the presence of Rh. It should, however, be mentioned that the experimentally determined rate of propene formation over Rh/ZrO₂ passes a maximum with Rh loading (Fig. 12(a)). According to DFT calculations, the positive effect of Rh on H₂ formation disappears above a certain reduction degree of ZrO₂ due to strong propene adsorption on Rh resulting in the blockage of sites required for hydrogen recombination. Very recently, this explanation was experimentally validated for bare ZrO₂ and MZrO_x (M = La or Y) with supported Ru or Rh nanoparticles.¹⁰³

Fig. 12 (a) The rate of propene formation (r(C₃H₆)) versus Rh loading of Rh/ZrO₂; (b) dependence of selectivity to propene on propane conversion over bare ZrO₂ and Rh/ZrO₂ with different Rh loadings. Reproduced from ref. 99 with permission from American Chemical Society, copyright 2020.

The presence of supported metal also influences product selectivity.^99,103,104 As seen in Fig. 12(b), the selectivity to propene decreases with an increase in Rh loading in Rh/ZrO₂ at a similar degree of propane conversion. The reason for such phenomenon is related to stronger propene adsorption on Rh than on bare ZrO₂. Adsorbed propene can participate in undesired side reactions leading to coke. When using supported Pt or Ir species weakly interacting with propene, no negative effect of metal loading on propene selectivity was determined over bare Pt(Ir)/ZrO₂ or Pt(Ir)/MZrO_x (M = La or Y).¹⁰³

2.4 Benchmarking and demonstrating catalyst development progress

As reviewed above, a large number of supported and bulk catalysts have been prepared and tested for their activity and selectivity during the last 5 years. Here, to demonstrate if there is any progress in catalyst development, we compare the most efficient catalysts with each other and with those described in a review of Weckhuysen and co-workers in 2014.¹⁸ The target characteristics were the selectivity to propene and the space time yield of propene formation STY(C₃H₆) at a propane conversion above 20%. Moreover, the catalysts tested with diluted reaction feeds, which are not attractive from an industrial viewpoint, were not used for such comparison. Only studies, which used reaction feeds with propane content of at least 20 vol%, were considered.

STY(C₃H₆) values fulfilling the above criteria are summarized in Fig. 13(a). These criteria were also applied to the catalysts from ref. 18. The corresponding catalyst compositions and further relevant catalytic data are provided in Table 3. The highest STY(C₃H₆) values were achieved over recently developed binary Cr–Zr–O_x-based catalysts; 3.2CrO_x/LaZrO_x⁶⁸ produced 2.6 kg(C₃H₆) kg(cat)⁻¹ h⁻¹ at 550 °C, while at 600 °C the value of about 3.5 kg(C₃H₆) kg(cat)⁻¹ h⁻¹ was obtained over Cr₃₀Zr₉₀/SiO₂.⁶⁹ ZrO_x-, VO_x-, CoO_x- and ZnO_x-based catalysts demonstrated lower productivity but still comparable with that of Pt-based catalysts described in ref. 18.

Fig. 13 (a) Space time yield (STY(C₃H₆)) of propene obtained over the most perspective catalysts described for the last 5 years (circles) and over selected best-performing catalysts from ref. 18 (stars). All data were obtained at propane conversion of at least 20% using reaction feeds with at least 20 vol% C₃H₈; (b) dependence of selectivity to propene on propane conversion at 550 °C (closed symbols) and 600 °C (open symbols) for selected catalysts from (a). Catalyst composition and reaction conditions are given in Table 3.

Table 3 Catalytic data of the most promising catalysts described in 2016–2020 and of the Pt-based catalysts from ref. 18. All data were obtained at propane conversion of at least 20% using reaction feeds with at least 20 vol% C₃H₈

Catalyst	T/°C	Feed composition	WHSV of C3H8/h^−1	X(C3H8)/%	S(C3H6)/%	Y(C3H6)/%	STY(C3H6)/kg(C3H6) kgcat^−1 h^−1	Ref.
K–CrOx/Al2O3	550	40 vol% C3H8–60 vol% N2	4.71	29.0	89.5	26.0	1.17	68
Cr-Al-800	600	100 vol% C3H8	9.43	33.2	90.4	30.0	2.70	65
Cr-Al-Ref	600	100 vol% C3H8	9.43	40.4	87.7	35.4	3.19	65
20CrOx/Al2O3	600	50 vol% C3H8–50 vol% N2	0.02	37.0	90.0	33.3	0.01	105
K–CrOx/Al2O3	550	40 vol% C3H8–60 vol% N2	1.6	42.0	87.0	36.5	0.56	98
Cr10ZrOx	550	40 vol% C3H8–60 vol% N2	9.24	30.0	81.0	24.3	2.14	68
3.2CrOx/LaZrOx	550	40 vol% C3H8–60 vol% N2	11.22	29.0	84.5	24.5	2.63	68
Cr10Zr90/SiO2	550	40 vol% C3H8–60 vol% N2	4.32	35.9	81.0	29.1	1.20	69
Cr10Zr90/SiO2	600	40 vol% C3H8–60 vol% N2	11.52	25.8	84.5	21.8	2.40	69
Cr20Zr90/SiO2	550	40 vol% C3H8–60 vol% N2	5.18	38.2	79.5	30.4	1.50	69
Cr20Zr90/SiO2	600	40 vol% C3H8–60 vol% N2	13.82	27.1	84.0	22.8	3.00	69
Cr30Zr90/SiO2	550	40 vol% C3H8–60 vol% N2	5.44	39.0	79.0	30.8	1.60	69
Cr30Zr90/SiO2	600	40 vol% C3H8–60 vol% N2	14.52	29.7	85.0	25.2	3.50	69
Cr–Zr–Ox (Cr/Zr molar ratio = 1/9)	550	40 vol% C3H8–60 vol% N2	5.89	34.0	86.0	29.2	1.64	71
1.25 wt% P/CZ	550	40 vol% C3H8–60 vol% N2	5.89	31.0	90.0	27.9	1.57	71
1.5 wt% Cs/CZ	550	40 vol% C3H8–60 vol% N2	5.89	29.0	94.0	27.3	1.53	71
1.5 wt% Cs–1.25 wt% P/CZ	550	40 vol% C3H8–60 vol% N2	5.89	30.0	93.5	28.1	1.58	71
0.6 wt% Ca/CZ	550	40 vol% C3H8–60 vol% N2	5.89	32.5	90.0	29.3	1.64	71
0.6 wt% Ca–1.25 wt% P/CZ	550	40 vol% C3H8–60 vol% N2	5.89	33.5	89.0	29.8	1.68	71
CrZrOx/SiO2_450	550	40 vol% C3H8–60 vol% N2	3.77	22.7	90.8	20.6	0.74	70
CrZrOx/SiO2_500	550	40 vol% C3H8–60 vol% N2	5.66	22.0	95.3	21.0	1.13	70
CrZrOx/SiO2_550	550	40 vol% C3H8–60 vol% N2	7.54	20.9	97.3	20.3	1.47	70
CrZrOx/SiO2_600	550	40 vol% C3H8–60 vol% N2	8.08	23.0	94.7	21.8	1.68	70
CrZrOx/SiO2_650	550	40 vol% C3H8–60 vol% N2	8.08	24.8	92.0	22.8	1.76	70
10 wt% V–K/meso-Al2O3-373 K	610	80 vol% C3H8–20 vol% N2	2.83	69.0	87.0	60.0	1.62	38
6V/Al2O3	600	28 vol% C3H8–28 vol% H2–44 vol% N2	3.3	23.0	94.0	21.6	0.68	34
12V/Al2O3	600	28 vol% C3H8–28 vol% H2–44 vol% N2	3.3	32.0	94.0	30.1	0.95	34
20V/Al2O3	600	28 vol% C3H8–28 vol% H2–44 vol% N2	3.3	29.0	87.0	25.2	0.79	34
6 wt% V/Al2O3 treated 1 h in H2	600	28 vol% C3H8–72 vol% N2	8.25	25.0	70.0	17.5	1.38	73
6 wt% V/Al2O3 treated 3 min in H2	600	28 vol% C3H8–72 vol% N2	8.25	36.0	55.0	19.8	1.56	73
12 wt% V/Al2O3	600	28 vol% C3H8–28 vol% H2–44 vol% N2	3.3	33.0	75.0	24.8	0.78	63
12V1Mg/Al2O3	600	28 vol% C3H8–28 vol% H2–44 vol% N2	3.3	33.0	83.0	27.4	0.86	63
10 wt% V/Al2O3	610	20 vol% C3H8–80 vol% N2	0.71	73.0	80.5	58.8	0.40	62
10 wt% Zn/Al2O3	600	28 vol% C3H8–28 vol% H2–44 vol% N2	3	22.0	95.0	20.9	0.60	42
15 wt% Zn/Al2O3	600	28 vol% C3H8–28 vol% H2–44 vol% N2	3	28.0	94.0	26.3	0.75	42
10 wt% Zn0.1 wt% Pt/Al2O3	600	28 vol% C3H8–28 vol% H2–44 vol% N2	3	30.0	95.0	28.5	0.82	42
15 wt% Zn0.1 wt% Pt/Al2O3	600	28 vol% C3H8–28 vol% H2–44 vol% N2	3	35.0	94.0	32.9	0.94	42
4Zn/TiZrOx	550	40 vol% C3H8–5 vol% H2–55 vol% N2	4.71	30.0	95.0	28.5	1.28	43
0.05Ru/YZrOx	550	40 vol% C3H8–60 vol% N2	1.57	36.2	82.0	29.7	0.44	93
0.05Ru/YZrOx	600	40 vol% C3H8–60 vol% N2	6.29	43.1	90.6	39.0	2.34	93
Y9Zr91Ox	550	40 vol% C3H8–60 vol% N2	1.6	36.0	90.0	32.4	0.49	98
t-ZrO2-1	550	40 vol% C3H8–60 vol% N2	0.32	26.5	78.1	20.7	0.06	96
m-ZrO2-1	550	40 vol% C3H8–60 vol% N2	1.89	27.1	86.1	23.3	0.42	96
m-ZrO2-1	550	40 vol% C3H8–60 vol% N2	1.13	33.0	83.5	27.6	0.30	96
10.0SnO2/Si-2	580	99.9 vol% C3H8	0.65	20.0	96.0	19.2	0.12	51
1.5Ni10.0SnO2/Si-2	580	99.9 vol% C3H8	0.65	20.0	96.0	19.2	0.12	51
1.36-Sn/Si(2)_reduced	600	99.9 vol% C3H8	0.65	23.0	91.0	20.9	0.13	50
1.78-Sn/Si(4)-(ws)_reduced	600	99.9 vol% C3H8	0.65	28.0	91.0	25.5	0.16	50
2.57-Sn/Si(4)_oxidized	600	99.9 vol% C3H8	0.65	20.0	50.0	10.0	0.06	50
3.82-Sn/Si(5)_reduced	600	99.9 vol% C3H8	0.65	29.0	91.5	26.5	0.16	50
Sn-HMS	600	99.87 vol% C3H8–0.13 vol% C2H6	0.39	40.0	90	36.0	0.13	52
5Co–Al2O3–HAT	590	20% vol C3H8–16 vol% H2–64 vol% N2	2.9	24.8	97.1	24.1	0.67	90
Co/Al2O3-IMP	590	20 vol% C3H8–16 vol% H2–64 vol% N2	2.9	21.2	95.5	20.2	0.56	90
7Co–Al2O3–HAT	590	20 vol% C3H8–16 vol% H2–64 vol% N2	2.9	30.0	93.0	27.9	0.77	90
Co/Al2O3	600	33.3 vol% C3H8–66.7 vol% N2	0.91	44.7	93	41.6	0.36	54
0.35 wt% Pt–1.26 wt% Sn/Al2O3	519	30 vol% C3H8–70 vol% N2	3.5	31.0	95.0	29.5	0.98	18
0.6 wt% Pt–5 wt% Ga/MgAl2O4	605	73 vol% C3H8–27 vol% H2	3.9	31.0	97.0	30.1	1.12	18
0.5 wt% Pt/Zn-Beta	555	100 vol% C3H8	2.6	40.0	55	22.0	0.55	18
0.5 wt% Pt–Na/Sn-ZSM-5	590	75 vol% C3H8–25 vol% H2	3.0	41.7	95.3	39.7	1.14	18
0.5 wt% Pt–Zn/Na-Y	555	100 vol% C3H8	2.6	24.8	91.6	22.7	0.56	18
0.5 wt% Pt–Sn–Na/Al-SBA-15	590	75% vol C3H8–25 vol% H2	3.0	27.5	94.0	25.9	0.74	18
0.7 wt% Pt/Mg(In)(Al)O	600	20 vol% C3H8–25 vol% H2–55 vol% He	2.6	20.4	98.0	20.0	0.50	18

---

For industrial application, not only productivity but also propene selectivity is an important parameter. To compare the catalysts from Fig. 13(a) in this performance, the reported selectivity–conversion values are shown in Fig. 13(b). It is obvious that all Zn-containing catalysts show propene selectivity above 80% at degrees of propane conversion up to about 40%. No data at higher conversion degrees are available. Some Zr-, Co-, Sn- or Cr-containing catalysts also show propene selectivity above 80% at degrees of propane conversion up to 50%.

In summary, an obvious progress in the development of metal oxide catalysts in terms of their productivity and selectivity at industrially relevant degrees of propane conversion was achieved during the last 5 years. However, to further check their application potential, tests with concentrated reaction feeds are required. It is also important to check catalysts durability, i.e. catalysts ability to restore their initial performance after oxidative regeneration in several dehydrogenation/regeneration cycles. Under this consideration, ZrO₂-based catalysts should be especially mentioned since they demonstrate good durability under industrially relevant conditions.^25,68,69,93

3 Non-oxidative isobutane dehydrogenation

As in the commercial PDH processes, catalysts with supported Pt or CrO_x species are used for the non-oxidative dehydrogenation of isobutane to isobutene (BDH) on large scale. High cost of Pt and toxicity of Cr(VI) compounds motivated the researchers to develop low-cost and environmentally friendly alternatives. Among them, the catalysts based on oxides of vanadium,^36,106–110 gallium,^111–114 molybdenum,¹¹⁵ zirconium,^98,116,117 iron^118–120 and aluminium⁹⁴ have been developed and investigated in the BDH reaction. None of them has so far found its commercial application. The most important tasks in the development of such catalysts include (i) maximizing their dehydrogenation performances (activity, selectivity and stability), (ii) identifying catalytically active sites, and (iii) establishing structure–activity–selectivity relationships. Herein, we describe and discuss the most recent achievements (2016–2020) related to catalysts based on metal oxides. Similar to the PDH reaction, the developed catalysts can be divided into two groups: conventional supported and alternative-type bulk catalysts (Fig. 4).

3.1 Conventional supported catalysts

3.1.1 Cr-Containing catalysts. Although Cr-containing catalysts are commercially applied, the nature of catalytically active species and their contribution to total activity are still not completely understood. Fridman et al.^121,122 made an attempt to determine the relative activity of different surface CrO_x species. At least five different Cr-containing species have been identified in the fresh CrO_x(13.2 wt%)/Al₂O₃ catalyst.¹²¹ They are isolated mono-/poly-aluminium chromate(s), redox chromium–chromate(s), redox aluminium–chromium–chromate(s), non-redox Cr³⁺ on the surface of amorphous Cr₂O₃, and non-redox Cr³⁺ on the surface of crystalline Cr₂O₃. In contrast to other species, isolated mono-/poly-aluminium chromate(s) are not stable under reaction conditions and transformed into isolated non redox Cr³⁺ and/or small clusters of Cr₂O₃. Fig. 14 shows the relative activity of chromium species, which are stable under BDH conditions.¹²² Non-redox Cr³⁺ on the surface of crystalline Cr₂O₃ seems to be the most active species. This conclusion should, however, be valid for highly loaded catalysts, where crystalline Cr₂O₃ is present. For their counterparts with predominantly dispersed CrO_x, the nature of active sites should be different. It can be expected that non-redox Cr³⁺ species should be responsible for isobutane dehydrogenation. From a selectivity viewpoint, redox aluminium–chromium–chromates are undesired as they catalyse hydrocracking and coke formation.¹²²

Fig. 14 Relative activity of different Cr-containing species on the surface of the CrO_x/Al₂O₃ catalysts. Adapted from ref. 122.

Besides studies aimed at elucidating the nature of catalytically active sites, the on-going research also deals with analysing the influence of promoter(s) and support on catalytic performance. The promoters are selected to modify overall catalyst acidity for hindering coke formation and to stabilize specifically structured CrO_x species and/or phases. Wang et al.¹²³ reported that promoting CrO_x/γ-Al₂O₃ with Ca improves both isobutane conversion and isobutene selectivity. These positive effects were related to regulation of acidic and redox properties. Lowering catalyst acidity inhibited side reactions such as cracking and isomerization. The presence of Ca is also decisive for suppressing over-reduction of CrO_x species thus ensuring high activity.

K is another basic promoter widely used for preparation of CrO_x/Al₂O₃ catalysts. The effect of K loading was investigated in ref. 124. When the loading is lower than 2%, both isobutane conversion and isobutene selectivity increase due to an improved dispersion of CrO_x species. The promoters prefer to interact with CrO_x species at higher loadings. This interaction decreases the intrinsic activity of such species without changing the selectivity.

The usage of non-basic promoters mainly affects redox properties of CrO_x-species. Salaeva et al.¹²⁵ established a synergistic effect between Cu and Zn on catalytic performance of CrO_x/Al₂O₃. A decrease in the dehydrogenation activity of CrO_x/γ-Al₂O₃ was found after promoting with Zn. When co-adding Cu, the activity was improved in comparison with CrO_x/γ-Al₂O₃. Cu alone also has a positive effect. The introduction of Zn and/or Cu was concluded to contribute to the formation of defective spinels (CuAl₂O₄, ZnAl₂O₄). Such transformation promotes the formation of higher amounts of Cr(VI) species on the catalyst surface. It should also be noted that in contrast to Zn, which decreases the reducibility of supported chromium species, the presence of Cu or Cu–Zn facilitates the reduction of such species that is important for catalyst activity.

The origins of the positive effect of Cu promoter on catalytic properties of CrO_x/Al₂O₃ were further investigated in a separate study.¹²⁶ For Cu loading below 2.6 wt%, Cu is mainly incorporated into the structure of the alumina support. As a consequence, monomeric and dimeric Cr(VI) are stabilized on the catalyst surface, which are transformed into catalytically active mononuclear and dimeric Cr³⁺ sites under reaction conditions. A CuO phase is formed at higher Cu contents. This phase was suggested to decrease the amount of redox Cr(VI)O_x species and thus results in a decrease in the activity.

The kind of support is another decisive parameter affecting catalyst activity. The support typically affects the dispersion of CrO_x species. For example, the usage of CeO₂ as a support was established to worsen the activity due to favouring the formation of low-active α-Cr₂O₃ phase.¹²⁷ In contrast, high dispersion can be achieved when using Al₂O₃, ZrO₂ or Ce_xZr_1−xO₂ as supports. Catalysts prepared on the basis of the two latter materials demonstrated the highest activity possibly due to the stabilization of Cr(VI)O_x strongly bound to the support surface. It cannot, however, be excluded that the high activity of ZrO₂-based catalysts can be related to the synergy effect between Cr and Zr_cus sites as recently was demonstrated for the PDH reaction over binary bulk or supported Cr–Zr–O_x.^68,69

Promoting Al₂O₃ with SiO₂ results in an increase in isobutane conversion and isobutene selectivity over K–CrO_x/SiAlO_x.¹²⁸ In comparison with K–CrO_x/Al₂O₃, these catalyst parameters were improved from 60.1 to 62.7% and from 87.3 to 90%, respectively. Moreover, the yield of coke decreased from 1.1 to 0.7%. The positive effect of SiO₂ was related to the stabilization of χ-Al₂O₃ phase.

3.1.2 V-Containing catalysts. Vanadium-based catalysts attract a lot of attention due to their relatively low price, good on-stream stability and high selectivity. Recent studies on VO_x-based catalysts were mainly focused on the investigation of the nature of active sites as well as on regulation of the structure of VO_x species via varying metal loading, the kind of support or by applying different preparation methods.

To elucidate the influence of the structure of VO_x species on their intrinsic activity, isobutene selectivity and on-stream stability, a series of VO_x/MCM-41 catalysts with different V loadings were prepared and investigated.³⁶ MCM-41 was chosen because it does not have strong acidic or basic sites. V³⁺O_x and V⁴⁺O_x were concluded to be the main catalytically active sites. A correlation between the number of V–O–V bonds around a central V cation determined from UV-vis spectroscopic analysis and the apparent turnover frequency of isobutene formation (TOF(isobutene)) was established (Fig. 15). The higher the polymerisation degree (number of V–O–V bonds), the lower the TOF(isobutene) values are. This was explained by stronger Lewis acidity of isolated sites. This catalyst property has, however, a negative effect on isobutene selectivity (Fig. 15). Nevertheless, isolated VO_x species deactivate slower than their polymerized VO_x counterparts (Fig. 15). The increase in deactivation rate with an increase in the degree of VO_x polymerization was related to an increase in the number of neighbouring V³⁺ or V⁴⁺ cations. For catalyst development, it was suggested that isolated VO_x sites with balanced Lewis acidity are required for achieving high activity, selectivity and on-stream stability.

Fig. 15 Influence of mean number of V–O–V bonds around a central V cation in VO_x/MCM-41 on (left figure) turnover frequency of isobutene formation (TOF(i-C₄H₈)) at 525 °C (red circles), 550 °C (black circles), 575 °C (blue circles), and 600 °C (green circles), (middle figure) selectivity to isobutene and n-butenes at about 30% isobutane conversion at 550 °C and (right figure) on-stream stability at 550 °C. Adapted from ref. 36.

Rodemerck et al.¹²⁹ investigated the influence of the kind of support (MCM-41, Al₂O₃, mixed SiO₂–Al₂O₃ oxides) on the nature of VO_x species as well as physico-chemical and catalytic performance. The degree of polymerization of VO_x species at similar V loading was concluded to depend on the support material. Thus, the lowest VO_x polymerization degree was determined for VO_x/SiO₂ while it increased with aluminium content in the support. The support also influences overall catalyst acidity. All the prepared catalysts possess Lewis acidic sites. However, due to a similar electronegativity of V and Al, VO_x/Al₂O₃ does not have Brønsted acidic sites, while such sites are present on the surface of VO_x/AlSiO_x because the electronegativity of Si is higher than that of V. Noticeably, the highest selectivity to isobutene of 85% at a degree of isobutane conversion of about 50% was achieved over VO_x/Al₂O₃. A detailed study of the dependence of product selectivity on isobutane conversion using feeds with 1-butene, isobutene or isobutane showed that the low selectivity to isobutene over VO_x/AlSiO_x catalysts in BDH is related to their higher activity for isobutene isomerization to n-butenes that depends on the degree of polymerization of VO_x and surface acidity. Catalyst isomerization ability affects the pathways of coke formation. This undesired reaction product is formed from isobutene over VO_x/Al₂O₃ catalyst. Contrarily, butadiene is the main coke precursor in the course of BDH over VO_x/AlSiO_x.

Tian et al.¹⁰⁶ reported that the distribution of VO_x species such as isolated monomeric species, oligomeric species and polymeric vanadium oxide chains in K₂O–VO_x/γ-Al₂O₃ (vanadium loading of 5, 10, 15 or 20 wt%) depends on vanadium loading. Reduced VO_x species containing V³⁺ and V⁴⁺ were proven to be the main catalytically active sites. Based on the results of FTIR, NMR and H₂-TPR studies, it could be concluded that catalysts with vanadium loading up to 5 wt% possess only isolated monomeric and dimeric VO_x species. Higher loaded materials contain polymerized VO_x species. In contrast to ref. 36, oligomeric VO_x species demonstrated an optimal dehydrogenation activity. This conclusion also disagrees with the results of Wang et al.,¹¹⁰ who investigated the effect of B modification of VO_x/MCM-41 on catalytic performance. B was found to mainly exist in the form of [BO₃]³⁻ on the surface of the catalysts, with a small fraction residing in the framework of MCM-41. The introduction of B increased the dispersion of VO_x sites as well as isobutane conversion and isobutene selectivity.

With respect to isobutene selectivity, this catalyst property decreases with an increase in the ratio of Lewis acidic sites to Brønsted acidic sites. Promoting K₂O–VO_x/γ-Al₂O₃ with sulphur was reported to also increase isobutene selectivity.¹⁰⁹ XPS, Raman and ⁵¹VMAS NMR confirmed the existence of V–S bonds after sulfidation. As the bond energy of V–S is lower than that of V–O, vanadium species could be reduced more easily after sulfidation. Consequently, more catalytically active V³⁺/V⁴⁺ sites are produced. Noticeably, sulfidation led to an increase in the strength of acidic sites but to a decrease in their concentration. Such changes in the acidic properties resulted in increasing isobutane conversion and inhibiting cracking reactions.

3.1.3 Zn-Containing catalysts. ZnO is a typical semi-conductor and has Lewis acidic sites, which may be potential sites for alkane dehydrogenation. Liu et al.¹³⁰ investigated the BDH reaction over silicalite-1-based catalysts with different amounts of Zn. The latter was introduced through a simple wetness-impregnation method. The kind of ZnO_x species varies with zinc loadings. At low content of Zn (<3.0 wt%), ZnO_x species are highly dispersed inside silicalite-1 pores in the form of sub-nanometric ZnO clusters. Increasing Zn loading to 6–12 wt% results in an additional formation of crystalline ZnO on the external surface of silicalite-1. Acidic sites determined by NH₃-TPD were concluded to be decisive for catalyst activity. It should, however, be mentioned that acidic sites of the support (zeolites) are detrimental for product selectivity. Zn-HZSM-5 (SiO₂/Al₂O₃ molar ratio of 26) showed only 0.2% isobutene selectivity at near complete isobutane conversion at 550 °C. The selectivity over Zn–silicalite-1 was 97.2–84.6% at isobutane conversion of 27.5–66.7%.

Zn/ordered mesoporous alumina was studied in isobutane dehydrogenation by Cheng et al.¹³¹ It was found that the introduction of Zn with a content below 10% did not change the mesoporous structure and textural properties of the support. ZnO_x species existed in highly dispersed forms and were incorporated into the support framework. With increasing the Zn content, the total number of acidic sites and that of weak sites increased, while those of strong acidic sites decreased. A correlation between the rate of isobutane conversion and the number of weak and medium acidic sites was established. The highest initial isobutane conversion and isobutene selectivity of 46.6% and 81.8%, respectively, were achieved over ZnO(10 wt%)/Al₂O₃ at 580 °C. This catalyst also showed good durability in 5 dehydrogenation–regeneration cycles.

3.1.4 Ga-Containing catalysts. Gallium oxide, which is a typical aromatization component, has also been considered as a promising alternative catalyst for alkane dehydrogenation. Taking into account its disadvantage caused by scarcity and accordingly high costs, the usage of Ga₂O₃ as a promoter and/or the development of Ga₂O₃-containing catalysts with a low content of gallium can be considered as a promising strategy for the development of GaO_x-based catalysts for alkane dehydrogenation. Wang et al.¹¹¹ discovered that isobutane dehydrogenation can occur on a physical mixture of Ga₂O₃ and ZnO, while the individual metal oxides revealed lower activity. The highest isobutene yield of about 55% was achieved over the ZnO/Ga₂O₃ = 10 mixture. From a mechanistic viewpoint, it was suggested that Lewis acidic sites of Ga₂O₃ promote the heterolytic cleavage of C–H bonds in isobutane on ZnO.

Alumina-supported Ga₂O₃– or Ga₂O₃–Cr₂O₃-containing catalysts were investigated in a fluidized-bed reactor.¹¹² It was found that the catalytic performances of Ga₂O₃/Al₂O₃ vary with gallium content. The conversion of isobutane at 580 °C increased from 42% to 55% with an increase in the content of gallium in Ga₂O₃/Al₂O₃ from 3 to 9 wt%. The conversion could be further improved after promoting Ga₂O₃(6 wt%)/Al₂O₃ with Cr (6 wt%) and ZrO₂ (1 wt%). This catalyst showed same conversion but higher isobutene selectivity (90 vs. 87%) in comparison with an industrial CrO_x-based catalyst. An outstanding stability during 60 cycles without any loss in catalyst performance was achieved.

Two recent studies dealt with kinetic aspects of the BDH reaction over Ga₂O₃/Al₂O₃^113,114 in the temperature range from 520 to 580 °C. Based on the experimental results and proposed reaction network, six major components including isobutane, n-butane, isobutene, butene, propene and methane were included in the kinetic modelling. The rate equations were obtained, and the corresponding kinetic parameters were estimated. The much smaller value of apparent activation energy for isobutane dehydrogenation (195 kJ mol⁻¹) than that of cracking (300 kJ mol⁻¹) indicated the efficiency of Ga₂O₃/Al₂O₃ for this reaction.

3.1.5 Fe-Containing catalysts. There are only two studies reporting about the BDH reaction over Fe-containing catalysts. Cheng et al.¹²⁰ prepared and tested SBA-15 based catalysts with FeO_x species incorporated into the support structure (Fe–SBA-15). For comparative purposes, two additional catalysts were prepared through a simple impregnation of SBA-15 (FeO_x/SBA-15) or amorphous SiO₂ (FeO_x/SiO₂) with an aqueous solution of FeCl₃. The two latter catalysts showed only slightly lower activity than the similarly loaded Fe–SBA-15 catalyst. FeO_x/SiO₂ was significantly less stable. Moderate and strong acidic sites as well as Fe³⁺ species were suggested to be the main active sites.

Ordered mesoporous Al₂O₃-based catalysts containing both ZnO_x and FeO_x supported species were also tested in the BDH reaction.¹¹⁹ They contain FeO_x species incorporated into the framework of Al₂O₃ and highly dispersed surface FeO_x species. Zn species were present in the form of hexagonal ZnO. Fe³⁺ and Zn²⁺ are considered to dehydrogenate isobutane. The isobutane conversion of 50.7% and the isobutene yield of 37.8% were achieved at 580 °C over the catalyst with 1.9 and 7.2 at% of Fe and Zn respectively.

3.1.6 Mo-Containing catalysts. There are only a few papers investigating Mo-containing catalysts in the BDH reaction.^115,132,133 Molybdenum carbides can catalyse the desired reaction but show rather low activity.¹²⁷ The highest reported rate of isobutene formation was about 2.8 mmol g⁻¹ h⁻¹ at 600 °C. The corresponding selectivity was about 70% at about 14% isobutane conversion. η-MoC seems to show lower ability for cracking reactions in comparison with MoO₂ and β-Mo₂C. Very recently reported bulk and supported molybdenum sulphide catalysts also show low activity.¹³³ No data about isobutene selectivity at industrially relevant degrees of isobutane conversion were provided. In comparison with the above-mentioned catalysts, Al₂O₃-supported catalysts with Mo in the structure of the support or in the form of surface MoO_x species show higher activity;¹¹⁵e.g. about 20 mmol g⁻¹ h⁻¹ at 600 °C. Nevertheless, the selectivity to isobutene is low and does not exceed 70% at isobutane conversion above 20%.

3.2 Alternative-type bulk catalysts

3.2.1 ZrO₂-Based catalysts. ZrO₂-Based catalysts were recently introduced as perspective low-cost and ecologically friendly alternative-type catalysts for isobutane dehydrogenation.¹¹⁶ Thorough characterization of the catalysts and analysing the influence of different factors on catalyst activity helped to reveal that coordinatively unsaturated zirconium cations (Zr_cus⁴⁺) are the active sites for dehydrogenation reaction. Since Zr_cus⁴⁺ sites are located at oxygen vacancies, the creation of oxygen vacancies through removal of lattice oxygen from ZrO₂ is an attractive way to create Zr_cus⁴⁺ active sites. Accordingly, reductive catalyst treatment positively influences the rate of isobutene formation.

In order to understand the effect of dopant for ZrO₂ on catalyst activity and product selectivity, a series of doped ZrO₂ samples containing La³⁺, Y³⁺, Sm³⁺, Ca²⁺, Mg²⁺ or Li⁺ were prepared, characterized and investigated in the BDH reaction.⁹⁸ The presence of dopant enhances the selectivity to isobutene in comparison with undoped ZrO₂ due to a decrease in the concentration of strong acidic sites, which provoke coke formation. With respect to the activity, the introduction of La³⁺ or Y³⁺ results in a decrease in the activation energy of isobutene formation indicating that the presence of such dopants in the lattice of ZrO₂ positively influences the intrinsic activity of Zr_cus⁴⁺ sites.

Binary CrZrO_x should be especially mentioned due to their unexpectedly high activity.¹¹⁶ The space time yield of isobutene formation at 550 °C over the most active Cr₁₀Zr₉₀O_x catalyst was two times higher than that over an analogue of commercial Cr–K/Al₂O₃. This catalyst also showed excellent stability in a series of 30 dehydrogenation–regeneration cycles at 550 °C and 600 °C without any changes in its initial performance from cycle to cycle.

Binary ZnZrO_x materials are also active and selective for isobutene production.¹¹⁷ It was found that promoting ZrO₂ with ZnO reduces surface acidity of ZrO₂ and generates new Lewis acid–base pairs ( ( stands for oxygen vacancy), and Zn²⁺–O²⁻), which could efficiently hinder side reactions. In agreement with previous studies of BDH^98,116 and PDH^26,93 over ZrO₂-based catalysts, coordinatively unsaturated Zr⁴⁺ sites were also suggested to be the active sites on the surface of ZnZrO_x. Isobutene selectivity of about 95% at 32.5% isobutane conversion was obtained at 580 °C.

3.2.2 Al₂O₃-Based catalyst. Bare Al₂O₃, which is typically used as a support, can also dehydrogenate isobutane.¹³⁴ Coordinatively unsaturated Al³⁺ cations play an important role in this reaction. Such sites are generated upon removal of surface OH groups during catalyst treatment at high temperatures and/or in situ under reaction conditions. Some treatment parameters including temperature, duration and atmosphere were confirmed to influence the number of catalytically active sites.

3.3 Benchmarking and demonstrating progress in the development of BDH catalysts

Similar to the PDH reaction (Section 2.4), we compare the most efficient catalysts developed for the BDH reaction in the last 5 years with each other and with those described in a review of Weckhuysen and co-workers in 2014.¹⁸ The catalysts are compared in terms of their productivity (space time yield of isobutene formation; STY(isobutene)) and the selectivity to isobutene at industrially relevant degrees of isobutane conversion. If the STY(isobutene) values were not provided in literature, we tried to calculate them when the necessary data (contact time, GHSV or WHSH, feed composition, isobutane conversion and isobutene selectivity) were given. Only studies using reaction feeds with isobutane content of at least 20 vol% were considered. The space time yield of isobutene formation STY(iso-C₄H₈) at isobutane conversion of at least 20% is shown in Fig. 16(a).

Fig. 16 (a) Space time yield of isobutene formation (STY(i-C₄H₈)) obtained over selected catalysts reported within the last 5 years (circles) and over selected best-performing catalysts from ref. 18 (stars); (b) selectivity to isobutene (S(i-C₄H₈)) at certain degrees of isobutane conversion (X(C₄H₁₀)) at 550 °C (closed symbols) and 600 °C (open symbols) for selected catalysts. Requirements for the selected catalysts: reaction mixture with at least 20 vol% isobutane. Its conversion was at least 20%. Catalyst composition and reaction conditions are given in Table 4.

Two Pt-containing and one Cr-containing catalysts from the previous review¹⁸ fulfilled the above-mentioned criteria, whereas 31 catalysts have been developed since 2016 (Fig. 16(a) and Table 4). Among them, there are four Cr-containing, eight ZrO₂-based, one binary CrZrO_x, fourteen V-containing, two bare Al₂O₃, and two Mo-containing catalysts. Noticeably, the Pt-based catalysts from ref. 18 outperform the developed metal oxide catalysts. The second most active catalyst is Cr₁₀Zr₉₀O_x. The corresponding STY(isobutene) value is (2.85 kg(isobutene) kg_cat⁻¹ h⁻¹) at 550 °C. The productivity of ZrO₂-based catalysts (not containing CrO_x) at 550 °C only slightly exceeds 1 kg(isobutene) kg_cat⁻¹ h⁻¹. VO_x-based catalysts can achieve high STY(isobutene) values (up to 1.5 kg(i-C₄H₈) kg_cat⁻¹ h⁻¹) at 610 °C. Other catalysts are less productive (lower than 1 kg(i-C₄H₈) kg_cat⁻¹ h⁻¹).

Table 4 Catalyst composition and catalytic data related to Fig. 16

Catalyst	T/°C	Feed composition	WHSV of (i-C4H10)/h^−1	X(i-C4H10)/%	S(i-C4H8)/%	Y(i-C4H8)/%	STY (i-C4H8)/kg(i-C4H8) kgcat^−1 h^−1	Ref.
Cr–K/Al2O3	550	40 vol% i-C4H10–60 vol% N2	6.3	31.0	94.4	29.3	2.34	98
C–Cr/γ-Al2O3	550	100 vol% i-C4H10		52.0	68.0	35.4	0.42	123
Cr/γ-Al2O3	550	100 vol% i-C4H10		56.5	72.6	41.0	0.49	123
2Ca–Cr/γ-Al2O3	550	100 vol% i-C4H10		56.0	77.5	43.4	0.52	123
V5–K2O/γ-Al2O3	590	50 vol% i-C4H10–50 vol% H2		54.0	68.2	36.8	1.04	106
V10–K2O/γ-Al2O3	590	50 vol% i-C4H10–50 vol% H2		62.5	68.0	42.5	1.20	106
V15–K2O/γ-Al2O3	590	50 vol% i-C4H10–50 vol% H2		62.9	68.9	43.3	1.22	106
V20–K2O/γ-Al2O3	590	50 vol% i-C4H10–50 vol% H2		59.4	74.3	44.1	1.24	106
VOx–K2O/γ-Al2O3	590	50 vol% i-C4H10–50 vol% H2		55.0	74.3	40.9	1.15	135
VOx/Al2O3	550	40 vol% i-C4H10–60 vol% N2		22.0	95.0	20.9		129
VOx/Al2O3	550	40 vol% i-C4H10–60 vol% N2		28.0	91.0	25.5	0.31	129
VOx/Al2O3	550	40 vol% i-C4H10–60 vol% N2		39.0	90.0	35.1		129
VOx/Al2O3	550	40 vol% i-C4H10–60 vol% N2		42.0	89.0	37.4		129
VOx/Al2O3	550	40 vol% i-C4H10–60 vol% N2		50.0	85.0	42.5		129
VOx/Si	550	40 vol% i-C4H10–60 vol% N2		28.0	73.0	20.4		129
VOx/Si	550	40 vol% i-C4H10–60 vol% N2		47.0	53.0	24.9		129
VOx/Si	550	40 vol% i-C4H10–60 vol% N2		52.0	47.0	24.4	0.29	129
VOx/S10	550	40 vol% i-C4H10–60 vol% N2		27.0	72.0	19.4		129
VOx/S10	550	40 vol% i-C4H10–60 vol% N2		31.0	70.0	21.7		129
VOx/S10	550	40 vol% i-C4H10–60 vol% N2		42.0	50.0	21.0	0.25	129
0.3V/MCM	550	40 vol% i-C4H10–60 vol% N2		28.4	58.0	16.5		36
1.4V/MCM	550	40 vol% i-C4H10–60 vol% N2		21.3	75.0	16.0		36
2.4V/MCM	550	40 vol% i-C4H10–60 vol% N2		21.4	77.0	16.5		36
4.1V/MCM	550	40 vol% i-C4H10–60 vol% N2		21.0	81.0	17.0		36
V-1.5K/γ-Al2O3	610	50 vol% i-C4H10–50 vol% H2		40.1	73.8	29.6	0.83	109
V-1.5K-S/γ-Al2O3	610	50 vol% i-C4H10–50 vol% H2		56.8	87.7	49.8	1.4	109
V-3K/γ-Al2O3	610	50 vol% i-C4H10–50 vol% H2		27.0	62.0	16.7	0.47	109
V-3K-S/γ-Al2O3	610	50 vol% i-C4H10–50 vol% H2		50.9	78.5	40.0	1.12	109
V-0B-MCM-41	600	20 vol% i-C4H10–80 vol% N2		39.0	76.5	29.8	0.45	110
V-1.5B-MCM-41	600	20 vol% i-C4H10–80 vol% N2		46.5	83.5	38.8	0.58	110
ZnZrO-1	580	50 vol% i-C4H10–50 vol% H2	1.8	33.0	95.5	32.0	0.53	117
ZnZrO-5	580	50 vol% i-C4H10–50 vol% H2	1.8	39.0	96.6	38.0	0.64	117
ZnZrO-9	580	50 vol% i-C4H10–50 vol% H2	1.8	28.0	96.0	27.0	0.45	117
Zn1/S-1	550	50 vol% i-C4H10–50 vol% N2		27.5	83.8	23.0		130
Zn3/S-1	550	50 vol% i-C4H10–50 vol% N2		60.2	68.2	41.0		130
Zn6/S-1	550	50 vol% i-C4H10–50 vol% N2		61.3	64.8	40.0		130
Zn8/S-1	550	50 vol% i-C4H10–50 vol% N2		66.7	59.8	40.0		130
Zn12/S-1	550	50 vol% i-C4H10–50 vol% N2		65.5	61.7	40.0		130
10Zn/Al2O3	600	100 vol% i-C4H10		54.0	72.0	38.0		131
5% Fe-SBA-15	600	100 vol% i-C4H10		22.0	47.5	10.0		120
10% Fe-SBA-15	600	100 vol% i-C4H10		27.0	47.5	13.0		120
15% Fe-SBA-15	600	100 vol% i-C4H10		25.0	51.0	13.0		120
MoAl(C)	600	100 vol% i-C4H10		25.0	44.0	11.0	0.33	115
MoAl	600	100 vol% i-C4H10		37.0	33.0	12.2	0.37	115
ZrO2	550	40 vol% i-C4H10–60 vol% N2	0.8	21.0	75.0	15.8	0.16	98
La8Zr92Ox	550	40 vol% i-C4H10–60 vol% N2	1.2	42.9	93.7	40.2	0.6	98
Y9Zr91Ox	550	40 vol% i-C4H10–60 vol% N2	1.9	46.4	93.1	43.2	1.04	98
Sm10Zr90Ox	550	40 vol% i-C4H10–60 vol% N2	1.2	37.3	92.1	34.4	0.52	98
ZrO2	580	50 vol% i-C4H10–50 vol% H2	1.8	21	95.0	20.0	0.34	117
Cr10Zr90Ox	550	40 vol% i-C4H10–60 vol% N2	5.9	44.9	84.6	38.0	2.85	116
Al2O3, 10 h_600 °C H2/N2	550	40 vol% i-C4H10–60 vol% N2		20.0	75.0	15.0	0.3	134
Al2O3, 5 h_700 °C N2	550	40 vol% i-C4H10–60 vol% N2		21.0	75.0	16.0	0.32	134
2 wt% Pt–1 wt% Sn/CeO2/C	520	33 vol% i-C4H10–7 vol%H2–60 vol% He	24.8	37.0	90.0	33.3	7.97	18
0.58 wt% Pt–Sn/K–L	600	33 vol% i-C4H10–67 vol%H2	13.2	61.0	92.0	56.1	7.15	18
40% Cr2O3/60% Al2O3	588	100 vol% i-C4H10	3.3	23.0	99.4	22.9	0.73	18

---

The selectivity to isobutene over selected catalysts tested at 550 °C or 600 °C is compared in Fig. 16(b). The black lines in this figure stand for the yield of isobutene of 36 and 53%. Noticeably, apart from the catalysts based on MoO_x, FeO_x or Al₂O₃, their counterparts with catalytically active VO_x, ZnO_x species or bulk ZrO₂-based catalysts can ensure the selectivity to isobutene above 80% at degrees of isobutane conversion higher than 20%. Moreover, industrially relevant isobutene yields were achieved over these catalysts.

4 Coke formation in PDH and BDH and characterization

In addition to activity and product selectivity, catalyst on-stream stability is another important property relevant for industrial applications. There are several origins of deactivation of dehydrogenation catalysts (Fig. 17).^92,105,136 Some of them are more specific and observed only for a certain kind of catalysts. For example, the reduction of ZnO into metallic Zn followed by its volatilization is a typical reason for the deactivation of ZnO-based catalysts.⁸⁰ The others are more general and can be observed for the most of the catalysts. One of such phenomena typical for all dehydrogenation catalysts including those applied on large scale is coke formation. Against this background, the purposes of the below section are to discuss mechanistic aspects of coke formation and methods for monitoring coke formation and characterising this undesired product.

Fig. 17 The most common reasons for deactivation of alkane dehydrogenation catalysts.

4.1 Mechanistic concepts on coke formation

Regardless of the kind of alkane dehydrogenation catalysts, coke deposits can be formed directly from the feed alkane and through consecutive transformations of the desired olefins. The latter reactions seem to dominate. Apart some general considerations/assumptions concerning coke formation, detailed mechanistic aspects of coke formation are scarcely studied on metal-oxide catalysts unlike their Pt-based counterparts. It is supposed that coke species sequentially transform from aliphatic into aromatic, and finally into graphite-like deposits with proceeding DH reaction.^135,137 It is also assumed that surface acidic sites contribute to coke formation.¹⁸ Only a few detailed studies dealing with elucidation of the mechanism of coke formation over metal-oxide catalysts have been reported up to now. They analysed supported catalysts with active VO_x species^129,135,138 or bulk ZrO₂-based catalysts.^43,69,97,98

In contrast to the generally accepted concept relating coke formation to catalyst acidity, no correlation between the concentration of acidic sites determined through NH₃-TPD and the amount of coke formed (coke selectivity) could be established in the PDH reaction over VO_x/SiO₂–Al₂O₃,¹³⁸ ZnO_x/MZrO_x⁴³ and CrZrO_x/SiO₂.⁶⁹ For the VO_x/SiO₂–Al₂O₃ catalysts, VO_x species were found to be more active for the formation of coke than acidic sites of support. The rate of coke formation determined from in situ TGA tests and the rate of catalyst deactivation in PDH correlate with the edge energies of supported VO_x species determined from UV-vis spectra.¹³⁸ On this basis, it was concluded that the degree of polymerisation of VO_x species affects coke formation. The effect of the VO_x distribution was explained as follows. To form aromatic structures followed by their further oligomerization and condensation to large graphitic structures, adsorbed propene molecules should be near to each other. Such situation is difficult to realize for isolated VO_x sites. Upon increasing the size of VO_x species, a probability of interaction between several adsorbed propene molecules increases resulting in a higher rate of carbon deposition.

A similar concept was very recently suggested for ZnO_x/MZrO_x⁴³ and CrZrO_x/SiO₂,⁶⁹ where coke formation takes place over supported ZnO_x and CrO_x species respectively. Those authors used time-resolved operando UV-vis spectroscopy for analysing coke formation. The degree of polymerisation of ZnO_x species at same Zn loading can be tuned through the kind of metal oxide dopant for ZrO₂ in MZrO_x. Cr content in CrZrO_x/SiO₂ determines CrO_x distribution.

It should be especially mentioned that in the case of isobutane dehydrogenation, coke species are formed via multiple mechanisms, e.g. directly from isobutene on the same sites as for the BDH reaction, or through isomerization of isobutene into n-butenes followed by dehydrogenation of the latter to butadiene, which undergoes oligomerization and cyclization reactions.^129,135 The contribution of different pathways to coke formation is mostly influenced by the nature of support material. Since skeletal isomerization of isobutene is catalysed by Brønsted acidic sites, the supports possessing such sites provoke coke formation through opening the route through n-butenes/butadiene.

Concerning PDH over bulk ZrO₂-based catalysts, several factors were identified to influence coke formation: (i) surface acidity;⁹⁸ (ii) phase composition and crystallite size of ZrO₂;^26,96 (iii) reducibility of ZrO₂.^96,97 In general, coke can originate directly from propane or via reactions with participation of propene. The contribution of each pathway depends on phase composition and crystallite size of ZrO₂.⁹⁶ DFT calculations predict that desorption of propene formed from propane is easier from t-ZrO₂ than from m-ZrO₂. Longer residence time of propene on catalyst surface can lead to its transformation into coke. This might be a reason that propane-aided pathway predominates over m-ZrO₂. It should be noted that there are at least two types of active sites on the surface of ZrO₂ which are able to adsorb and convert propane: (i) Zr_cus located at steps, kinks or corners of the lattice of ZrO₂ and (ii) regular surface zirconium cations. The former sites are responsible for the transformation of propane into propene and in a lesser degree into coke, while the latter sites produce coke. Accordingly, increasing the ratio between Zr_cus and regular sites is beneficial for decreasing coke selectivity. The concentration of regular sites and therefore the probability of interaction of adsorbed propene molecules with each other leading to coke formation can be decreased by decreasing the size of ZrO₂ crystallites. Another possible way for decreasing the concentration of regular sites is the removal of lattice oxygen from ZrO₂ during reductive treatment. Such idea is supported by the fact that coke formation is suppressed with increasing reducibility of ZrO₂.⁹⁷

4.2 Characterization of coke species

Since deposition of coke is the main reason for catalyst deactivation in PDH and BDH, it is highly important to understand factors affecting coke formation with the purpose to develop catalysts with suppressed ability to form coke. To help researchers aiming to elucidate coke formation, now we discuss methods suitable for characterising coke formation. Their specific characteristics are briefly introduced. There are a lot of methods which can be applied for characterization of coke (Fig. 18). They can be divided into two groups: those used directly under experimental conditions (in situ/operando) and those applied to characterize spent catalysts after the reaction (ex situ).

Fig. 18 In situ/operando and ex situ methods for coke characterization.

4.2.1 In situ/operando methods. The in situ/operando methods allow monitoring reaction-induced catalyst changes under reaction conditions as a function of time.^137,139,140 The most common methods applied for investigation of alkane dehydrogenation catalysts are Raman spectroscopy, diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy and UV-vis spectroscopy.

Raman spectroscopy has been widely used to study the chemical nature of coke species formed on catalyst surface.^139,140 Raman bands related to coke species at ∼1325 cm⁻¹ and ∼1590 cm⁻¹ are assigned to D (disorder-allowed vibration modes) and G (E_2g optical mode of graphitic carbon) bands respectively. The D band is usually deconvoluted into 4 peaks: D₁ band (∼1320 cm⁻¹) related to ring vibrations of defects and edges of graphitic lattice and attributed to the in-plane imperfection such as defects and heteroatoms. The D₂ band (∼1620 cm⁻¹) is assigned to disordered graphitic lattice vibration mode with E_2g symmetry. The D₃ band (∼1500 cm⁻¹) corresponds to amorphous carbon, while the D₄ band (∼1220 cm⁻¹) is characteristic for the disordered graphitic lattice of an A_1g symmetry.^135,139 The ratio of I_D1/I_G intensities is assumed to indicate the degree of coke graphitization. An increase in the I_D1/I_G ratio implies either increasing the number of defects in the graphitic lattice or decreasing the size of graphitic crystallites. It has also been reported that the size of graphitic crystallites influences the position of G band. Thus, with decreasing the size of crystallites, G band shifts to higher wavenumber.^139,141

DRIFT spectroscopic analysis of adsorbed propene at different temperatures allows to investigate the mechanism of coke formation from propene since different coke precursors will be observed at different temperatures. It is generally assumed that during PDH reaction aliphatic structures of coke (band at ∼2960 cm⁻¹) are firstly formed via deep dehydrogenation reactions and then gradually transform into aromatic structures.¹⁴¹

UV-vis spectroscopy can be applied for analysing the kinds of coke species and the kinetics of their formation and removal during PDH reaction and regeneration stages, respectively. It is supposed that UV-vis bands with maxima at 600 nm and 800–900 nm are related to polyaromatic species, while the band with the maximum at 420 nm can be assigned to low-condensed aromatic species.⁷⁰ The kinetics of coke formation/removal can be evaluated by analysing temporal changes of Kubelka–Munk function at certain wavelengths.

In situ/operando methods provide qualitative information about coke formation: composition of coke precursors (DRIFT spectroscopy), kinetics of coke formation (Raman and UV-vis spectroscopy) and graphitization degree (Raman and UV-vis spectroscopy). Such methods alone however cannot give quantitative information. To this end, ex situ analysis of spent catalysts is required.

4.2.2 Ex situ methods. Ex situ methods such as solid-state ¹³C NMR, SEM and TEM, temperature-programmed oxidation, XPS, thermogravimetric analysis and XRD are applied for characterization of coked catalysts. It should be taken into account that when transporting spent catalysts from the reactors, where dehydrogenation reaction was performed, into the cells or sample holders applied for ex situ characterization, the catalysts contact air and may contain adsorbed CO₂ and H₂O, that might distort the characterization results. Moreover, coke species might undergo restructuring after dehydrogenation reaction during cooling down in the reactor.

Solid-state ¹³C NMR allows to distinguish different carbon species. A resonance centred at around 31 ppm is generally assigned to aliphatic carbon species related to –CH₂ groups, while a peak at around 132 ppm is related to polyaromatic carbonaceous species (Fig. 19).¹³⁵

Fig. 19 Solid-state ¹³C NMR spectra of VO_x–K₂O/Al₂O₃ catalyst tested in isobutane dehydrogenation after different times on stream. Reproduced from ref. 135 with permission from Elsevier, copyright 2020.

Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) can be used to investigate the morphology of coke species and their location.

Temperature-programmed oxidation (TPO) is a common method for quantitative analysis of coke deposits. The amount of carbon oxides (CO_x) released during TPO experiment represents the amount of carbon deposits. The TPO profile can also give information about the kind of carbon species. The temperature of the maximum of CO_x release gives information about graphitization degree of coke deposits. Thus, the higher the temperature, the higher the graphitization degree is. When H₂O is also measured upon TPO, the ratio of H/C can be determined from the concentration of released CO_x and H₂O. The ratio is an indicator for the kind of carbon deposits, i.e. olefinic, aromatic or graphitic. However, precautions must be considered when measuring CO_x and particularly H₂O. When spent catalysts are characterized after their exposure to air and storage before starting TPO, adsorbed water and carbon dioxide from the air might worsen the correct quantification of carbon and hydrogen in surface carbon-containing deposits. Thus, the best way is to perform TPO tests in the same set-up used for alkane dehydrogenation without catalyst exposure to air.

The amount of coke can also be determined by thermogravimetric analysis (TG). Such method implies the registration of changes in catalyst weight during combustion of carbon deposits.

X-ray photoelectron spectroscopy (XPS) is typically used for characterizing and quantifying surface chemical composition of spent catalysts. Thus, carbon deposits can also be analysed. Typically, the C 1s signal at 283.4 eV is related to sp² hybridized carbon, while such signal at around 285 eV is attributed to sp³ hybridized carbon.¹³⁵ The signal at around 284 eV is assigned to adventitious carbon.

X-ray diffraction (XRD) method can give information about the existence of crystalline graphitic carbon.

5 Oxidative propane dehydrogenation

Since the beginning of nineties of the last century, the oxidative dehydrogenation of propane to propene (ODP) has been intensively investigated to establish a cost-efficient and environmentally friendly alternative to the commercially applied cracking processes and the non-oxidative dehydrogenation of propane. Despite the enormous efforts made by various research groups, the ODP reaction is still not commercialized due to too low selectivity to propene at industrially relevant degrees of propane conversion. The reason is higher reactivity of the desired olefin in comparison with the feed alkane. Thus, this reaction is also intriguing from a fundamental viewpoint, i.e. selective activation of stronger C–H bond in C₃H₈ with suppressed ability for the undesired cleavage of weaker C–C bond. Below, we briefly analyse the progress in catalyst development with respect to fulfilling industrially relevant requirements (Section 5.1) and summarize fundamentals upon which the scientific community has agreed (Section 5.2).

5.1 Is there any progress in catalyst development since 2008?

Based on the results of our search for the relevant literature from 2008 on, about 1300 different catalysts have been synthesized and tested for their activity and selectivity in the ODP reaction up to now (Fig. 20(a)). The catalysts are divided into various groups with respect to the nature of active component. A major part (in total about 700) of these catalysts contain vanadium as the active component (Fig. 20(b)). Other relatively often investigated catalysts comprise B, Co, Cr, Fe or Ni. To check if any progress in catalyst development has been achieved, we proceeded as follows. The selectivity–conversion data for different V-containing or V-free catalysts from ref. 10 were used as the state-of-the-art performance reported before 2007. These data highlighted by black circles are given in Fig. 20(c and d). This figure also shows the performance of commercial PDH technologies, which is the target for the ODP reaction. The grey lines stand for the yield of propene of 35 and 53% typical for these technologies.

Fig. 20 (a) Annual number of catalysts tested and reported in literature from 2008 up to August 2020, (b) overall number of catalysts developed since 2008 with a certain active component. Selectivity–conversion plots for propene over (c) V-based or (d) V-free catalysts tested under different reaction conditions. Black circles in (c) and (d) represent the catalysts from ref. 10. Closed and open red symbols in (c) stand for the catalysts developed after 2007 with propene selectivity above 70% at propane conversion above 10% using C₃H₈/O₂ = 2–0.5 () or C₃H₈/O₂ > 10 () feeds or upon alternating feeding of C₃H₈ and air (). Coloured circles in (d) distinguish catalysts with different active components: NiO_x (), B (), CrO_x (), Pt () or C (). Catalyst composition and reaction condition for the datapoints after 2007 are given in Tables S1 and S2 (ESI†) with experimental details related to tests with V-containing and V-free catalysts in (c) and (d) respectively. Experimental details of tests with alternating feeds of propane and air in (c) are given in Table S3 (ESI†).

To demonstrate the developments after 2007, we decided to apply a certain criterion for selecting promising catalysts. If we showed all the selectivity–conversion datapoints reported in this period of time, Fig. 20(c and d) would be reader unfriendly. Thus, the catalysts with propene selectivity above 70% at propane conversion larger than 10% have been selected and are shown as filled not-black circles in Fig. 20(c and d). Only a few catalysts among about 700 V-containing catalysts reported in literature^75,142–149 fulfilled this criterion. Moreover, none of them revealed the target propene selectivity at propane conversion above 20% (Fig. 20(c)). Thus, since 2008 no significant progress in the development of V-containing catalysts with industrially relevant performance has been achieved when performing the ODP reaction with stoichiometric O₂–C₃H₈ feeds. Reaction engineering aspects seem to be promising for improving catalyst performance. As seen in Fig. 20(c), propene selectivity of around 80% at propane conversion of above 50% can be achieved when operating under O₂-lean (strong excess of C₃H₈ over O₂)¹⁵⁰ or O₂-free (chemical looping or redox operation)^151–156 conditions. The developments in these fields are discussed in Sections 7.3 or 7.4.

In comparison with V-containing catalysts, some improvements in the performance of V-free catalysts have been achieved (Fig. 20(d)). For example, Xie et al.¹⁵⁷ reported propene selectivity of 80% at propane conversion of about 70% over a NiO-modified CeO₂ catalyst at 500 °C. The latter oxide has nanorod morphology. A stable operation over about 100 hours on stream was reported. It should, however, be mentioned that such high performance was achieved when using a reaction feed containing HCl. No propene was formed in the absence of HCl. This additive affected not only the selectivity to propene but also increased propane conversion. Based on the results of catalyst characterization and DFT calculations, Cl˙-like species were suggested to be responsible for the selective conversion of propane to propene. They may be formed upon reaction of HCl with surface O₂²⁻ species generated upon activation of gas-phase oxygen on an anion vacancy. From an application viewpoint, special requirements with respect to reactor material, safety operation and environmental aspects should be fulfilled when operating with HCl.

Another catalytic system to be mentioned is PtSn/SiO₂.¹⁵⁸ The support is a dealuminated beta zeolite. The highest propene selectivity of 79% at propane conversion of 48% was achieved at 550 °C. The catalyst slowly deactivated with time on propane stream due to conversion of the bimetallic PtSn species into the individual components. Boron-containing catalysts introduced a few years ago also show high propene selectivity above 70% but at degrees of propane conversion not higher than 25%.^159–166 Other catalysts in (Fig. 20(d)), which fulfil the above defined criterion for propene selectivity, contain Ni,^{157,167–172} Cr,^173,174 Mn,¹⁷⁵ or C¹⁷⁶ as active components.

We also compared catalysts in terms of their productivity, i.e. space time yield of propene STY(C₃H₆) formation (kg_C3H6 kg_catalyst h⁻¹). The STY(C₃H₆) values of the recently developed catalysts from Fig. 20(c and d) are shown in Fig. 21 together with such values obtained over the catalysts developed before 2007 from ref. 10. For brevity, the latter materials with STY values larger than 0.5 kg_C3H6 kg_catalyst h⁻¹ were used. Like the selectivity–conversion relationship, no improvements in the productivity of V-containing catalysts have been achieved after 2007. Contrarily, some progress could be identified for V-free catalysts. Boron-based catalysts should be especially mentioned. The very high activity reported for a NiO-modified CeO₂¹⁵⁷ is probably related to the presence of HCl.

Fig. 21 Space time yield of propene formation (STY(C₃H₆)) determined over different catalysts developed before 2007 (black) and after 2007 (coloured symbols). The black datapoints are from ref. 10. The coloured symbols stand for the catalysts in Fig. 20(c and d). Catalyst composition and reaction conditions for the datapoints after 2007 are given in Tables S1 and S2 (ESI†) for V-containing and V-free catalysts, respectively.

The above discussion can be summarized as follows. The recently introduced boron-based catalysts demonstrate the potential of developing materials with uncommon active components for improving propene selectivity. Despite a great number of studies dealing with V-based catalysts, no evident improvements in propene selectivity at industrially relevant degrees of propane conversion could be achieved through catalyst design. This statement also arises a question, whether the desired progress can be realized. Probably, we need deeper understanding of the relationships between catalyst physicochemical properties including the structure of VO_x species and the kinetics of selective and unselective reaction pathways. The state-of-the-art knowledge about the fundamentals of the ODP reaction over catalysts based on metal oxides is, therefore, described and discussed in the next section.

5.2 Fundamental aspects of ODP

In general, it is well accepted that the ODP reaction over V-based catalysts occurs through a Mars–van Krevelen redox mechanism involving lattice oxygen of VO_x species for activating two C–H bonds in propane to yield propene. In a second step, the reduced VO_x species is reoxidized by gas-phase oxygen or another oxidant. Lattice oxygen is also involved into consecutive conversion of the desired olefin to carbon oxides. These mechanistic considerations should also be valid for other catalysts based on reducible metal oxides. In addition to lattice oxygen, adsorbed oxygen species are also assumed to contribute to the undesired side reactions.^177,178 The extent of reduction of vanadium under ODP conditions is another selectivity-governing factor.^179–182 VO_x dispersion and the kind of support are also considered to be decisive descriptors for catalyst activity and selectivity as supported by the below analysis.

5.2.1 Intrinsic activity of supported catalytically active species. For highlighting the role of the kind of support, the kind of supported metal oxide and its dispersion in the ODP reaction, we analyse turn-over frequency (TOF) and primary (at degrees of propane conversion below 10%) selectivity to propene. If the TOF values have not been reported in literature, they were calculated by us under assumption that all supported species are highly dispersed. Datapoints obtained at degrees of propane conversion below 10% were used for these calculations.

As the highest number of differently supported and loaded catalysts with VO_x,^183–190 ZnO_x,¹⁹¹ B,¹⁹² CoO_x¹⁹³ or FeO_x¹⁹⁴ were tested at 450 °C, we plotted the corresponding TOF values in Fig. 22 as a function of apparent surface density of the active component. Due to the high activity of CrO_x-¹⁷⁴ or NiO_x-¹⁹⁵ containing catalysts, it was not possible to calculate the TOF values at 450 °C. The lowest possible temperature is 300 °C. Thus, the latter data are given in Fig. 22. Cr loading above 5 atoms nm⁻² has a negative effect on the TOF values due to the presence of Cr₂O₃. For NiO_x/CeO₂, a continuous decrease in the intrinsic activity of NiO_x species with rising Ni loading was established. The strength of the decrease is significantly stronger for these materials in comparison with CrO_x/nanodiamonds.

Fig. 22 Turnover frequency of propene formation (TOF(C₃H₆)) versus apparent density of V,^{183–188,190,196} Zn,¹⁹¹ Fe,¹⁹⁴ Co,¹⁹³ B,¹⁹² Ni¹⁹⁵ or Cr¹⁷⁴ in different catalysts. The colours distinguish different supports, which are Al₂O₃ (light green circles), carbon (black circles), Ca₅[OH|(PO₄)₃] (red circles), CeO₂ (blue circles), MgO (grey circles), MgAlO_x (pink circles), SiO₂ (brown circles), TiO₂ (dark green circles) or ZrO₂ (orange circles). The kind of active metal oxide and reaction temperature are given in each plot. Further details are provided in Table S4 (ESI†).

The intrinsic activity of B, FeO_x and CoO_x also decreases with rising content of the active component (Fig. 22). Although a direct comparison is not fully justified due to different kinds of supports, FeO_x species should show higher activity than CoO_x and B. ZnO_x species are intrinsically less active than FeO_x and CoO_x and their activity also decreases with rising Zn loading regardless the kind of supporting material. For achieving high activity, TiO₂ and ZrO₂ supports should be applied for catalyst preparation.

The V-related TOF values for catalysts based on Al₂O₃,^183,184 calcium hydroxyapatites (Ca₅[OH|(PO₄)₃]),¹⁸⁵ CeO₂,^186,187 MgO,¹⁸⁸ SiO₂¹⁹⁶ and TiO₂¹⁹⁰ supports are given in Fig. 22 as a function of V loading. The lowest TOF values were obtained over Ca₅[OH|(PO₄)₃]-¹⁸⁵ or MgO-supported¹⁸⁸ catalysts. In agreement with a previous review on the kinetics of the ODP reaction,¹² Al₂O₃, TiO₂, and ZrO₂ appear to be the most efficient supports in terms of intrinsic activity of supported VO_x species.

No effect of V loading up to 10 Atom nm⁻² on the TOF values of CeO₂-supported catalysts could be identified. The datapoint for 6.5 V nm⁻² was calculated from ref. 186, while other points are from ref. 187. As all the values follow same trend, the absence of any effect of V loading on the TOF values of VO_x/CeO₂ materials might be justified. This may be related to high dispersion of VO_x on the surface of CeO₂. Isolated VO_x and polyvanadate species but no V₂O₅ were detected in VO_x/CeO₂ with 6.5 V nm⁻².¹⁸⁶ It can be assumed from Fig. 22 that the activity of MgO- or Ca₅[OH|(PO₄)₃]-supported VO_x species decreases with rising V loading, although there are only two loading datapoints for each catalyst have been reported. A negative effect of the loading may exist for VO_x/Al₂O₃. The intrinsic activity of VO_x species on the surface of SiO₂ does not seem to depend on V loading below 2 V nm⁻².

To further analyse the effect of the kind of support and V loading on TOF, we prepared Table 5 containing the values reported for VO_x/SiO₂ and VO_x/SiTiO_x catalysts at 500 °C. The catalysts contain less than 3.1 V nm⁻² and should not possess three-dimensional VO_x particles. Interestingly, although the VO_x/SiO₂ catalysts were tested at different partial pressures of propane (0.05–0.63 bar) and oxygen (0.05–0.33 bar), the TOF values did not significantly depend on the pressures. No obvious dependence on V loading could also be deduced. However, when comparing the TOF values additionally related to the partial pressures of propane and oxygen, a negative effect of the loading will be obtained. Such dependence may be explained as follows: (i) any direct comparison of TOF values obtained over differently prepared but similarly loaded VO_x/SiO₂ is not possible as they were prepared from different chemicals by different researchers as well as tested in different laboratories under different reaction conditions or (ii) isolated VO_x species have higher intrinsic activity than their polymerized counterparts.

Table 5 Selected characteristics of VO_x/SiO₂ and VO_x/Si_aTi_bO₂ catalysts tested in the ODP reaction at 500 °C. The SiO₂ and Si_aTi_bO₂ are abbreviated as Si and SiTi respectively. The numbers in the brackets stand for the a/b ratio. Propene selectivity and TOF(C₃H₆) values were determined at propane conversion below 10%. ω_V stand for apparent V surface density

Support	ω V/nm^−2	p(C3H8)/bar	p(O2)/bar	TOF(C3H6)/s^−1	S(C3H6)/%	Ref.
Si	1.64 × 10^−1	0.05	0.05	2.91 × 10^−3	80.0	198
Si	2.89 × 10^−1	0.05	0.05	1.92 × 10^−3	89.0	198
Si	1.74 × 10^0	0.48	0.06	1.72 × 10^−3	63.0	184
Si	2.35 × 10^0	0.10	0.10	2.74 × 10^−3	88.2	147
Si	2.19 × 10^0	0.10	0.10	3.31 × 10^−3	91.0	147
Si	1.90 × 10^0	0.10	0.10	4.33 × 10^−3	89.3	147
Si	1.23 × 10^0	0.10	0.10	6.21 × 10^−3	86.5	147
Si	9.90 × 10^−1	0.10	0.10	5.85 × 10^−3	91.9	147
Si	6.64 × 10^−1	0.10	0.10	5.34 × 10^−3	90.3	147
Si	1.55 × 10^0	0.67	0.33	8.57 × 10^−3	80.0	196
Si	1.83 × 10^0	0.67	0.33	9.85 × 10^−3	74.0	196
Si	2.19 × 10^0	0.67	0.33	1.04 × 10^−2	72.0	196
Si	2.58 × 10^0	0.67	0.33	1.13 × 10^−2	76.0	196
Si	3.08 × 10^0	0.67	0.33	9.14 × 10^−3	58.0	196
Si	5.86 × 10^−1	0.29	0.15	1.10 × 10^−2	75.1	197
SiTi(168)	6.95 × 10^−1	0.29	0.15	1.16 × 10^−2	78.6	197
SiTi(55)	7.59 × 10^−1	0.29	0.15	6.66 × 10^−3	64.8	197
SiTi(22)	1.49 × 10^0	0.29	0.15	1.01 × 10^−2	69.7	197
SiTi(10)	1.11 × 10^0	0.29	0.15	4.55 × 10^−2	50.9	197
SiTi(8)	1.37 × 10^0	0.29	0.15	4.05 × 10^−2	38.4	197
SiTi(5.7)	1.75 × 10^0	0.29	0.15	3.02 × 10^−2	40.2	197
SiTi(8.3)	0.59 × 10^0	0.40	0.20	1.50 × 10^−2	68.0	179
SiTi(1.6)	0.66 × 10^0	0.40	0.20	1.40 × 10^−1	75.0	179
SiTi(0.7)	1.10 × 10^0	0.40	0.20	2.60 × 10^−1	90.0	179

---

The effect of Si/Ti ratio in Si_aTi_bO₂ supports on the activity of VO_x species was thoroughly investigated in two independent studies.^179,197 As a general trend, the reported TOF values increased with a decrease in the ratio at a close V loading (Table 5). The positive effect is due to improved reducibility of VO_x species through the presence of Ti.

We also tried to validate if reported positive effects of some dopants (supported metal oxides, additives or mixed phase) on the intrinsic activity of VO_x species are of general character or were simply determined in single studies. To this end, we calculated average V-related TOF values for catalysts based on a certain support (Al₂O₃,^{144,183,184,199–203} SiO₂,^{144,147,150,184,196,198,203–208} TiO₂,^184,199,209 ZrO₂,²⁰⁹ SiZrO_x²¹⁰) with V density below 7 mn⁻² and free of any promoter. The error bars were also determined as a standard deviation. So obtained TOF values are compared in Fig. 23 with the values reported for catalysts based on same support but containing a promoter for VO_x. The following promoted catalytic systems have been previously investigated: Mg–VO_x/Al₂O₃,^201,202 P–VO_x/Al₂O₃²¹¹ (vanadyl phosphate), F–VO_x/SiO₂,²⁰⁵ K–VO_x/SiO₂,²¹² Mo–VO_x/SiO₂,²⁰⁸ Nb–VO_x/SiO₂,²¹³ Sb–VO_x/SiO₂,²¹⁴ P–VO_x/TiO₂²¹¹ (vanadyl phosphate), P–VO_x/ZrO₂²¹¹ (vanadyl phosphate), Sb–VO_x/SiZrO₂(3.5 wt% SiO₂).²¹⁰ Although the uncertainty of such comparison is high, it can be concluded that all promoters probably with exception of Nb have rather negative effect on the intrinsic activity of VO_x species. According to ref. 208 the negative effect of Mo on the V-related TOF values of Mo–VO_x/SiO₂ is related to the structure of VO_x species. Addition of MoO_x prevents the formation of V–O–V bond in favour of isolated VO_x species. The latter species were concluded to be less active than their oligomerized counterparts. This statement agrees with the results in Fig. 22. An increase in the TOF values of VO_x/SiO₂ with rising V loading is seen in this figure. Interesting, such behaviour could not be identified for VO_x on other supports and for other supported metal oxides and boron (Fig. 22). Further studies are, however, required for understanding if and how the intrinsic activity of supported metal oxide species depends on their structure and the kind of support.

Fig. 23 Averaged V-related turnover frequencies of propene formation over differently supported V-containing catalysts in the absence (●) or the presence of different promoters (F – , K – , Mo – , Nb – , Mg – , P – , Sb – ). The kind of support is written in each figure. Table S5 (ESI†) contains full catalyst compositions and reaction conditions.

Although the addition of K to VO_x/SiO₂ has a negative effect on the V-related TOF value (Fig. 23), a positive effect of Na presence was very recently reported in ref. 215. Those authors investigated the ODP reaction over Na₆V₁₀O₂₈/SiO₂, α-NaVO₃/SiO₂ and VO_x/SiO₂. The two latter catalysts showed lower activity, too. If Na₆V₁₀O₂₈/SiO₂ transforms to α′-NaV₂O₅/SiO₂, the activity decreases. However, no control test with Na-promoted VO_x/SiO₂ has been carried out.

5.2.2 Primary propene selectivity. Catalyst activity is certainly an important parameter, while the selectivity to propene is the property which is crucial when discussing about large-scale applicability of the ODP reaction. Thus, it is highly important to know how this catalyst property can be tuned. Using the catalysts from Fig. 22, Table 5, Tables S4 and S5 (ESI†), we now analyse the effects of the kind of support and active component as well as its loading on propene selectivity at propane conversion below 10%. This performance is mainly determined by primary propane conversion to propene and to a lesser extent by undesired consecutive propene transformations. The collected literature data are shown in Fig. 24. We start our discussion with V-based catalysts. A negative effect of V loading is obvious for VO_x/SiO₂, VO_x/MgO and VO_x/Al₂O₃. These materials can be ordered in the same sequence with respect to the effect of the kind of support on the desired selectivity. TiO₂ and CaPO_x do not seem to be efficient supports. Interestingly, the selectivity to propene over VO_x/CeO₂ passed a maximum at about 6 V nm⁻².

Fig. 24 Primary (X(C₃H₈) < 10%) propene selectivity (S(C₃H₆)) versus apparent density of metal in active metal oxide species. The colours distinguish different supports (Al₂O₃ (light green circles), carbon (black circles), Ca₅[OH|(PO₄)₃] (red circles), CeO₂ (blue circles), MgO (grey circles), MgAlO_x (pink circles), SiO₂ (brown circles), TiO₂ (dark green circles) or ZrO₂ (orange circles)), while the kind of active metal oxide and reaction temperature are given in each figure. Further details are provided in Table S4 (ESI†).

In comparison with V-containing catalysts, the selectivity reported for Zn-based catalysts is significantly lower. Co/MgAlO_x catalysts reveal the highest (above 90%) primary propene selectivity, which does not depend on Co loading up to about 17 Co nm⁻². B/C and FeO_x/CaPO_x show significantly lower selectivity. CrO_x/C catalysts seem to be efficient for the dehydrogenation reaction and the desired selectivity does not decrease with Cr loading. This selectivity is also not affected by Ni content in NiO_x/CeO₂, although NiO_x species are intrinsically less selective.

In order to check if the conclusions made in one specific study are of general character, various VO_x/SiO₂ catalysts in Table 5 are compared in terms of the effect of V loading on the primary propene selectivity at 500 °C. Although the reported datapoints scatter, no general negative effect could be identified for the catalysts with up to 5 V nm⁻² as one could assume from Fig. 24, where the results from only one study are shown. In this view, the effect of Ti/Si in VO_x/SiTiO_x with a similar (between 0.7 and 1.7 V nm⁻²) V loading on the primary propene selectivity is also contradictive. One research group¹⁹⁷ established a negative effect, while a positive effect was reported in another study.¹⁷⁹ Such obvious discrepancies may derive from different reaction conditions including catalyst treatment procedures as well as different methods of catalyst synthesis.

Further, available literature is analysed to check whether and how the support and the kind of promoter for VO_x as well as the type of active component affect the primary selectivity to propene. Another question to be clarified is the effect of temperature on this catalyst performance. To this end, an average selectivity at a certain reaction temperature was calculated for (i) VO_x-containing (without any promoter) catalysts based on Al₂O₃,^{144,162,183,184,200–203,216–220} CeO₂,^{186,187,216,219} SiO₂,^{75,147,150,184,196,198,204,207,216,217,219,221–223} TiO₂^{144,184,190,205,206,213,216,217,219,224–227} or MgO^188,219 supports or (ii) all catalysts containing V,^{75,142,144,147,150,162,183,184,186–188,190,196,198,200–207,213,214,216–229} B,^{160,162–164,166,192,218,220,230,231} Co,^{175,193,219,227,232–237} Cr,^{173,174,238–240} Mo^{208,241–244} or Ni^{157,167,168,171,172,189,195,219,233,235,245–252} without or with one or several promoters. For the calculations, the selectivity values obtained at a propane conversion below 10% were used. Fig. 25 was constructed on this basis. The corresponding values also containing error bars for each averaged datapoint are shown in Fig. S1 (ESI†).

Fig. 25 Averaged propene selectivity calculated from literature data obtained at a propane conversion below 10% at different temperatures. Effect of support for VO_x: the colours distinguish different supports (Al₂O₃ (light green circles), CeO₂ (blue circles), SiO₂ (brown circles), TiO₂ (dark green circles) or MgO (grey circles).) for VO_x species. Effect of promoter for VO_x: supported catalysts with VO_x exclusively () or all catalysts containing V (). Effect of active component: B (●), Co (), Cr (), Mo () or Ni ().

We start our discussion with the effects of temperature and support on the selectivity to propene over VO_x species. For the catalysts based on the SiO₂, TiO₂ or Al₂O₃ supports, the desired selectivity increases with rising reaction temperature. Such effect can be due to different activation energies for propene formation and for propane/propene conversion into carbon oxides. The latter reaction should have lower activation energy. The positive effect of temperature on the selectivity to propene is also valid when considering both supported and bulk V-containing catalysts with or without any promoter. It is also obvious that any effect of promoter on the selectivity to propene could be hardly seen. This means that the improvements reported in various papers are not significant or the reference V-containing (without any promoter) catalysts used in those papers show lower performance in comparison with similarly composed catalysts tested in other studies. Concerning the effect of the kind of support on product selectivity, it is difficult to distinguish between the SiO₂, TiO₂ or Al₂O₃ supports when bearing in mind the error bars calculated for each catalyst (Fig. S1, ESI†). Opposite to these supports, the propene selectivity over the catalysts based on the CeO₂ or MgO supports decreases with an increase in temperature. A possible reason may be the formation of mixed phases at high temperature. It is, however, worth mentioning that only a few studies fulfilled the above-mentioned criterion for selecting them for our analysis. Thus, further studies are required.

Regarding the effect of active component on product selectivity, the datapoints in Fig. 25 are too strongly scattered to draw a definitive conclusion. Nevertheless, when analysing the datapoints at propene selectivity above 80%, some hints can be obtained. Regardless of the reaction temperature, an averaged propene selectivity over V-based catalysts is below this value. The selectivity above 80% achieved over Mo-containing catalysts at 200 and 480 °C were reported in single studies only, i.e.ref. 244 and 243, respectively. Reaction temperatures of 350 and 400 °C are more representative, as they were applied in several studies dealing with the ODP reaction.^241,242,244 On this basis, one can conclude that up to now tested Mo-based catalysts do not show high primary propene selectivity. B-, Co- or Ni-containing catalysts seem to be suitable for achieving the desired performance. Cr-containing catalysts are also promising when they operate at about 300 °C.

In summary, the support material, the kind of active component and its loading are decisive for the primary propene selectivity over different supported catalysts. Such conclusion is actually not novel. When considering the discussion of catalyst development in Section 5.1, one may assume that this knowledge is not enough to ensure the design and the preparation of catalysts with high propene selectivity. Some progress can be achieved when deeper fundamental correlations between the real structure of supported active species, the kinetics of their reduction/reoxidation, the nature of oxygen species formed upon activation of gas-phase oxygen and the kinetics of selective and unselective pathways of propane conversion can be established. In this regards, density functional theory calculations (DFT) have the potential for providing molecular-level details relevant for catalyst design as discussed in the next section.

5.2.3 Molecular-level insights into ODP over VO_x-based catalysts. In general, there are three specific purposes of DFT calculations: (i) elucidating the kind of active oxygen species and the role of metal oxide dispersion, (ii) providing insights into the structure of supported metal/metal oxide species and (iii) clarifying elementary pathways of product formation in the course of the ODP reaction. The mechanism of the oxidative propane dehydrogenation to propene over bulk V₂O₅ and supported VO_x species has been investigated by various research groups.^253–260 We do not refer here to the papers before 2010. It is accepted that the dehydrogenation reaction starts with the homolytic cleavage of the methylene C–H bond. So-formed iso-propyl radical bounds to a HOV(IV)O_x site. Propene is formed through abstraction of second hydrogen either by this OH group or by another oxygen species. Various oxygen species can participate in propane activation. Lo et al.²⁵⁶ summarized possible reaction mechanisms over V₂O₅ as schematically shown in Fig. 26. Researchers agreed upon that the vanadyl group (V=O) is responsible for the first activation step, with the vanadyl-O mediated mechanism being dominant. With respect to product selectivity, both the desired pathway and side reactions leading to iso-propanol or acetone are equivalently possible over V₂O₅.²⁵⁵ This was concluded to be the main reason for low propene selectivity over catalysts containing even nanosized V₂O₅. In contrast to the latter material, the dehydrogenation reaction seems to dominate over supported catalysts as calculated for one monolayer VO_x/TiO₂.²⁵⁵ Supported catalysts also reveal higher reactivity due to reducing the activation barrier for the rate-limiting step that is breaking C–H bond in propane.

Fig. 26 (a) Possible mechanisms for the oxidative dehydrogenation of propane to propene over V₂O₅ with participation of lattice oxygen. (b) Oxygen species and their participation in propane/propene oxidation. Adapted from ref. 174 and 256 respectively.

The important role of the vanadyl group for the first propane activation step was also confirmed for the ODP reaction over supported catalysts. In addition to different lattice oxygen species (vanadyl or bridged), peroxo-oxygen species can be formed upon activation of gas-phase O₂ over reduced VO_x species.^178,259 Such species reveal high activity for undesired propene oxidation to carbon oxides. Thus, to control this pathway, the catalyst should be able to quickly convert this unselective oxygen species into lattice oxygen. This conclusion is indirectly supported by the fact that the selectivity to propene increases when O₂ oxidant is replaced by N₂O. The latter can generate lattice oxygen exclusively.

The kind of support or supported VO_x species (isolated or dimeric) is decisive for the second C–H cleavage. It is predicted that a HOV(IV)O_x site formed upon propane activation over VO_x/SiO₂ with isolated sites does not subtract another hydrogen from the generated iso-propyl radical. The latter should desorb and react with V=O of another VO_x species yielding propene and a second HOV(IV)O_x site. The situation changes when analysing the reaction pathways of propane dehydrogenation over dimeric VO_x species on SiO₂.¹⁷⁸ Bridged oxygen (V–O–V) or neighbouring vanadyl oxygen can participate in the second hydrogen abstraction. It was also concluded that a V(V)/V(IV) cycle should prevail over a V(V)/V(III) cycle. These calculations also predict independence of the V-related TOF on V loading upon propane dehydrogenation over highly dispersed VO_x sites (from isolated VO_x to those with a fewer number of V atoms). This conclusion agrees with our analysis of experimental literature data (Fig. 22 and Table 5).

The effect of anatase TiO₂ crystalline face on the reactivity of isolated and dimeric VO_x species in the ODP reaction was investigated in ref. 261. VO_x/TiO₂(100) and VO_x/TiO₂(001) models were applied. Bridged (V–O–Ti) of isolated VO_x on the surface of TiO₂(100) is predicted to show higher activity than the vanadyl oxygen, while these both species in isolated VO_x on the surface of TiO₂(001) should have similar reactivity. In comparison with the isolated species, propane dehydrogenation over dimeric VO_x species on the surface of TiO₂(100) or TiO₂(001) is initiated through breaking the methylene C–H bond by the vanadyl oxygen. In terms of intrinsic activity, VO_x/TiO₂(100) should perform superior to VO_x/TiO₂(001). This difference was related to higher ability of the former system to bond H atoms and to lower stability of supported VO_x species on the surface of TiO₂(100).

All the above discussed studies with supported catalysts considered one structure of VO_x species. Such species is terminated by vanadyl oxygen and possesses connections with the support through three bridged oxygen. Cheng et al.²⁶² constructed possible monomeric and dimeric structures after H₃VO₄ grafting on the surface of anatase TiO₂(001). The hydroxylated surface was used for calculations. Six monomeric and two dimeric VO_x species were constructed (Fig. 27). They were used for calculating reaction pathways and energies for propanol formation from propane. No difference in the overall reaction mechanism between these differently structured VO_x species could be established. The first step is the activation of propane C–H bond by the vanadyl group in agreement with several previous studies discussed above. In a second step, an O–C bond is formed. The reactivity was not found to depend on the size of the active species. However, its molecular structure (coordination environment) seems to play an important role for catalyst activity. VO_x species with square pyramidal coordination environment (M-molecular or M-tridentate structures in Fig. 27) should possess higher reactivity in comparison with their tetrahedrally coordinated counterparts (M-bidentate and M-bidentate_2 in Fig. 27). Such difference was explained by more efficient stabilization of reduced square pyramidal coordinated VO_x species in the reaction intermediate structures. These theoretical results demonstrate the potential in catalyst development when controlling the structure of supported VO_x species and may also explain different experimental TOF values determined for V-containing catalysts prepared with a certain support in different studies.

Fig. 27 Possible structures of monomeric and dimeric VO_x species on the TiO₂ anatase (001) surface. Reproduced from ref. 262 with permission from Elsevier, copyright 2020.

Du et al.²⁶¹ have carried out a systematic study of possible structures of supported VO_x species formed upon reaction of H₃VO₄ with the TiO₂ anatase (001), (100) and (101) surfaces. Surprisingly, five-coordinated VO_x species were identified to be the most stable structures on the (001) surface. Surface stress upon adsorption of H₃VO₄ determines the final structures. The stress does not seem to play an important role for the (100) and (101) surfaces. Consequently, tetrahedrally coordinated VO_x species on the latter faces are the dominant structures. It was also concluded that VO_x species on the (001) surface are more stable than those on the (100) and (101) surfaces. Thus, the latter surfaces are less suitable for stabilizing isolated VO_x species and will favour their polymerization.

The surface structures of isolated, diatomic and polymerized VO_x species on the (101) and (001) tetragonal ZrO₂ surfaces were examined.²⁶³ The effect of water was also analysed. Differently structured species were formed upon adsorption of V₂O₅ and H₂O on these single surfaces. For the lowest number of V atoms (0.25 per eight Zr cations), a dimer is the most stable structure on the (101) surface. Each V cation has a terminal oxygen (vanadyl) and is connected via a bridge oxygen with another V cation. Contrarily, monoatomic species is the dominant structure on the (001) tetragonal ZrO₂ surface at same V loading. A dimeric species is formed at higher V loading. No stable structures, where V⁵⁺ replaces Zr⁴⁺ could be identified. However, a structure with one vanadyl oxygen can be stabilized. In addition, a monolayer structure is more preferable than V₂O₅ crystallites. For all analysed structures, vanadium atoms are connected with zirconium atoms through lattice oxygen. Water was concluded to be a decisive parameter for stabilizing a certain VO_x structure and its reducibility. The diatomic structures on the (101) surface are transformed into the isolated species in the presence of about 0.01 bar water. The latter species show lower reducibility than the dimers.

Different isolated VO_x·Ce₁₂O₂₄ (x = 0–4) structures were modelled as a function of temperature (130–630 °C) at very low oxygen partial pressure (about 10⁻⁹ bar).²⁶⁴ The (111) CeO₂ surface was used for calculations. The most stable structure was concluded to be VO₂·Ce₁₂O₂₄. Ce⁴⁺ is reduced to Ce³⁺, while the oxidation state of vanadium is +5. The most stable structure is composed of one vanadyl group and three V–O–Ce bonds. It was also concluded that the reducibility of ceria increases in the presence of supported VO_x species. Any consequences for the ODP reaction were not considered.

Summarising this part, the real structure of supported VO_x species on the surface of a certain support, their stability and reactivity strongly depend on the kind of support face, temperature, and water presence. As different faces can be exposed for adsorption of VO_x species upon catalyst preparation, the resulting catalysts will probably contain differently structured VO_x species. Thus, it is practically impossible to determine the activity and selectivity of a certain structure in the ODP reaction. Such information may become available if supports with a preferential orientation of exposed faces will be prepared and applied for adsorption of VO_x species.

6 Oxidative isobutane dehydrogenation

In comparison with the ODP reaction, the oxidative dehydrogenation of isobutane to isobutene (ODB) is less investigated. Our search of literature on the latter topic resulted in only 52 articles published since 2000. About 190 different catalysts have been developed and tested. In contrast to the ODP reaction, the largest number of the prepared catalysts contain chromium as active component.^265–278 In total, about 49 such catalysts were synthesized in comparison with 45 catalysts based on vanadium.^{269,272,279–288} Surprisingly, carbon-based catalysts (in total 21 materials) build the third group of mostly investigated catalysts.^289–296 Another important class of catalysts is based on molybdenum typically in form of molybdates.^297–301 Phosphates and pyrophosphates of different metals were also tested for the ODB reaction.^302–305 Against this background, we initially discuss developments within each group of the catalysts and then compare their application potential.

6.1 Active components and mechanistic aspects of ODB

6.1.1 Cr-Containing catalysts. Cr-Containing catalysts can operate at temperatures below 300 °C^{267,270,276,278} producing mainly CO_x as side products.²⁷⁸ Isobutene selectivity higher than 95% at isobutane conversion of 12.9% was reported for 10 mol% CrO_x/La₂(CO₃)₃ tested in a pulse reactor at 240–250 °C.²⁶⁸ These properties probably motivated the researchers to develop such catalysts, although industrially relevant performance was achieved at significantly higher temperatures. From fundamental viewpoints, the effects of support, promoter, Cr precursor and reaction conditions on catalyst activity and selectivity were investigated. Application of different chromium precursors for catalyst preparation have shown that Cr(NO₃)₃·9H₂O is the most suitable precursor in comparison with K₂Cr₂O₇, CaCr₂O₇·H₂O, CrO₃ and CrK(SO₄)₂·12H₂O²⁷⁰ according to ranking based on isobutene yield value at 250 °C. An optimal chromium loading for supported catalysts should be about 10 wt%.^266,267,277

With respect to the effect of support, Jibril et al.²⁷⁰ concluded that γ-Al₂O₃ is the best choice among MgO, TiO₂, SiO₂ and γ-Al₂O₃. Calcination and pre-treatment temperatures have a minor effect on the performance of VO_x/Al₂O₃ catalysts.²⁶⁷ However, the surface area of the support seems to have a positive effect on the selectivity to isobutene. The highest selectivity of ca. 68% at isobutane conversion of about 10% was obtained at 250 °C over the catalysts with Puralox 150/170 and high-surface-area aluminium oxide support.²⁶⁷ Therefore, not only the surface area but also the type of support material should be considered for preparation of active and selective catalysts. A Cr-containing catalyst prepared using high surface area (787 m² g⁻¹) SiO₂ support (SBA-15) converted 14% isobutane with 79% selectivity at 540 °C.²⁷⁷ Since formation of highly dispersed mono and polychromate domains was mentioned to be an important requirement for effective catalyst,²⁷⁷ high surface area of the support could be a factor facilitating formation of such domains upon catalyst synthesis.

Different dopants were investigated for their effect on catalytic performance of Al₂O₃-supported catalysts. Using promoters on the basis of oxides of Ni, Co, Mo, Na, W, V, Li, La or Bi for synthesis of Cr–M–O/Al₂O₃ catalysts with a total content of metal of 10 wt% did not result in any significant improvement of catalytic performance.²⁷⁶ Contrarily, promoting CrO_x/γ-Al₂O₃ with CaO was established to positively affect the selectivity to isobutene when the content of the promoter was below 2 wt%.^274,275 The promoter facilitated the formation of highly dispersed Cr⁶⁺ species, which are the active sites. In addition, catalyst acidity decreased, thus facilitating isobutene desorption and hindering its readsorption. A bulk Ca-chromate phase was formed with an increase in Ca content that was detrimental for catalyst performance. Although promoting CrO_x/Al₂O₃ with K₂O was also established to reduce the overall acidity, the selectivity to isobutene decreased.²⁷⁸ The same study demonstrated that this promoter improved the selectivity over CrO_x/TiO₂ from 34.8 to 55.5%.²⁷⁸ An opposite effect was observed for CrO_x/Al₂O₃. Such a contradictory effect was related to the changes in the rate of oxygen chemisorption and coverage by oxygen species. These parameters increased for CrO_x/TiO₂ and decreased for CrO_x/Al₂O₃ when these catalysts were doped with K, while acid–base properties changed in the same way upon potassium addition. Thus, the effect of the acidity on isobutene desorption/readsorption and accordingly on the selectivity to isobutene should be further elucidated.

In comparison with standard metal oxides, mixed metal oxide supports seems to be more suitable for achieving high isobutene selectivity. A catalyst containing chromium oxide (8 wt%) on the surface of the fluorite-type Ce_0.60Zr_0.35Y_0.02O₂ support resulted in 93% selectivity at isobutane conversion of 11% at 540 °C.²⁶⁶ Very auspicious results (isobutane conversion of 52.4% and isobutene selectivity of 84%) were obtained over CrO_x(18 wt%)/ZnAlLaO_x at even higher temperature of 580 °C.²⁶⁵ It is still to be proven if the reaction temperature or/and the support are important for the high catalyst performance.

6.1.2 V-Containing catalysts. Both supported and bulk catalysts based on vanadium have been developed. The focus of investigations dealing with these catalysts was put on elucidating the effects of the kind of support for VO_x species, the method for their deposition and their structure on product selectivity and catalyst activity. Vanadium can be incorporated into the structure of support or located on its surface in form of supported VO_x species.

To incorporate vanadium into the structure of SiO₂ (MCM-41), two different methods were used: (i) direct hydrothermal synthesis using VOSO₄²⁸² or VOC₂O₄²⁷⁹ as vanadium source or (ii) template-ion exchange using MCM-41 containing ca. 50 wt% template and VOC₂O₄. The catalysts prepared according to the former method selectively converted isobutane to isobutene when the vanadium content exceeded 1 wt%. Contrarily, the template-ion exchange method resulted in non-selective catalysts. The effect of the preparation method was related to the location/structure of vanadium species. Isolated tetrahedrally coordinated VO_x species sites dispersed on the wall of MCM-41 were produced by the template-ion exchange method, while the selective sites formed through the direct hydrothermal synthesis are incorporated inside the walls of MCM-41. The conclusion about the low selectivity of supported isolated VO_x species seems to be contradictive. Actually, selective catalysts could be synthesized by simple impregnation of MCM-41 with NH₄VO₃.²⁸¹ The selectivity was further increased when using V(t-BuO)₃O as a source of vanadium. The highest selectivity value of about 84% at isobutane conversion of 48% was achieved at 560 °C. Such high performance was explained by high dispersion of surface VO_x species.

Another interesting approach for enhancing dispersion of VO_x species and their resistance against sintering is a grafting technique and the usage of promoters. Using this method, two series of VO_x–TiO₂–SiO₂ materials were prepared: (i) titanium and vanadyl alkoxides were simultaneously grafted on SiO₂ and (ii) vanadyl tri-isopropoxide was grafted on a TiO₂–SiO₂ support.²⁸⁰ The former materials showed superior selectivity. The improvement was related to the kind of catalytically active species. Isolated bridged oxygen species (V–O–Ti) formed on the silica surface via method i are less active but more selective than polyvanadylic V–O–V formed when vanadyl tri-isopropoxide was grafted on the TiO₂–SiO₂ support.

Depending on the kind of support, its shape can be decisive for the performance of supported VO_x species. For example, the same species were obtained through deposition of vanadium on nano-shaped TiO₂ with different surface facets.²⁸⁵ Accordingly, these catalysts revealed similar activity and selectivity. Contrarily, when CeO₂ rods or octahedra were used as a support for VO_x,²⁸⁷ the octahedra-based catalyst showed higher selectivity and lower activation energy for isobutane conversion. The observed effects were explained by the difference in surface oxygen vacancy formation energy, number of defect sites and surface O–O distance of different surface CeO₂ planes.

Another group of V-based catalysts intensively tested in the ODB reaction are vanadium-containing mixed oxides. For example, the performance of Cr–V–Nb mixed oxides depends strongly on the content of components.²⁷² Isobutane conversion increases from 20% to 50% upon increasing the ratio of Cr/V from 2.3 to 4.7. The selectivity to isobutene also increases. The highest selectivity of 90% was achieved at isobutane conversion of 45% at 573 °C using a feed mixture with O₂/i-C₄H₁₀ ratio of 1.06. When Cr content is further increased the selectivity decreased. Such effect of Cr loading was assumed to be due to the changes of active surface areas and redox properties of the catalysts upon variation of the catalyst composition.

Bulk V–Sb mixed oxides show rather low selectivity to isobutene, i.e. about 11% at isobutane conversion of ca. 13%. Their performance could be slightly increased upon promoting with niobium oxide (22% selectivity to isobutene at isobutane conversion of 15%).²⁸³ Further improvements were achieved when depositing VSbO_x on microspheric γ-Al₂O₃.²⁸³ The selectivity of 68% at isobutane conversion of 36.5% was reported. The usage of other supports like α-Al₂O₃, Si–Al–O, SiO₂ or MgO for preparation of VSbO_x-containing catalysts worsened the performance.^284,306 Promoting VSbO_x/γ-Al₂O₃ with Ni resulted in an increase in the conversion to 42–44% and keeping the selectivity on the same level of ca. 70%.²⁸³ The positive effect of Ni was explained by the formation of the nickel vanadate NiV₂O₆ phase. The catalyst containing this phase is easier reducible and possesses higher amount of mobile oxygen species. Thus, more facile redox cycle of active vanadium species was considered to be responsible for the enhancement of the catalyst performance.

For supported VMoO_x, Al₂O₃ was also found to be a more suitable support than CeO₂ or TiO₂.²⁸⁸ Since Al₂O₃ possesses significantly higher surface area, better performance of Al₂O₃-supported catalysts was related to the higher dispersion of the active sites on the catalyst surface. VMoO_x/Al₂O₃ with V₂O₅–MoO₃ content of 5 wt% converted 22.2% of isobutane with isobutene selectivity of 72.3% at 500 °C. The selectivity decreased upon increasing the content of the active components or reaction temperature. As isobutane conversion also increased, it is not clear if the decrease in the selectivity was really related to the content of the active component or temperature or simply to higher conversion. It is well known that the selectivity decreases with rising isobutane conversion over same catalyst under isothermal conditions due to higher formation of consecutive reaction products (CO_x).

Bulk Mo–V–Sb mixed oxides also produce methacrolein with a selectivity of 9.3% at 440 °C. At isobutane conversion of 7.6% the corresponding isobutene selectivity amounts to 25.9%. The isobutene selectivity decreases upon an increase in Mo content, while selectivity towards methacrolein increases.²⁸⁶

6.1.3 Carbon-based catalysts. Carbon-based materials are not only inexpensive and sustainable but also possess special chemical and physical properties (large specific surface areas, relatively high oxidation stability, good electron conductivity, variable surface functionalities, etc.), which can be useful for purposeful tuning catalyst performance. Graphitic carbon,^289,290,294 activated carbon,^291–293 carbon xerogel²⁹⁶ and few-layered graphenes²⁹⁵ were tested for the ODB reaction. Although the up to now developed catalysts show lower activity and selectivity than VO_x- or CrO_x-containing catalysts, such alternative materials are interesting from a fundamental viewpoint related to the nature of active oxygen species. Therefore, the below discussion is aimed at factors affecting the performance of carbon-based (metal free) catalysts.

Carbon catalysts possess many different surface functional groups like quinonic, ether, carbonylic, phenolic, carboxylic, lactone, etc. (Fig. 28). Some of these groups negatively influence catalyst performance. So carboxylic anhydride groups formed on the surface of carbon xerogel during the ODB reaction lowered the TOF values.²⁹⁶ Acidic groups on the surface of activated carbon fibre reduced catalytic activity.²⁹² Therefore, modifying carbon surface is an attractive way to control the nature of surface functional groups and respectively catalyst performance. Such modifications could, for example, be achieved through incorporation of nitrogen into the structure of carbon xerogel.²⁹⁶ Treatment of synthesized materials under different conditions is an useful tool for controlling the type and the amount of surface functional groups or improving their accessibility.^{289,292–296} So, closed fullerene-like cavities of graphitized mesoporous carbon are uniformly opened through air oxidation at 500 °C without affecting the graphitic structure but enhancing catalyst activity. These cavities can be closed during oxidative treatment at 600 °C.

Fig. 28 Oxygen-containing functional groups on the surface of carbon catalysts.

Regardless of the type of carbon material, the ODB reaction is catalysed by carbonyl-quinone groups.^291–296 DFT calculations²⁹⁵ predict that dicarbonyls at zigzag edges and quinones at arm chair edges provide the main contribution to the isobutene formation. Although some experimental results indicate that total oxidation occurs on other type of surface sites,²⁸⁹ these sites were not still reliably identified. There are some debates about mechanistic aspects of regeneration of selective sites. The regeneration of active sites occurs by reaction with gas-phase O₂. Similar to the mechanistic concept suggested for the metal oxides this reaction yields water.²⁹³ However, DFT calculations predicted a very high activation barrier (>150 kJ mol⁻¹) for regeneration step via water removal.²⁹⁵ So, it was suggested that hydrogenated functionality reacted with O₂ forming H₂O₂.

Not only the temperature but also the duration of thermal treatment and the kind of treating atmosphere can affect catalyst activity.^{289,291–293} Carbon catalysts exposed to N₂O, O₂ or H₂ at the same temperature for the same time performed differently.²⁹³ The O₂-treated catalyst was more active, but less selective than its counterparts treated in N₂O or H₂. An increase in treatment time in air at 500 °C from 24 to 48 h led to ca. two fold increase in the rate of isobutane conversion over graphitic carbon.²⁸⁹ This enhancement was related to the generation of active quinone-type functional groups. Additional active carbonyl-quinone groups can be also incorporated upon exposure of carbon materials to supercritical CO₂ or water.²⁹² Contrarily, treatment of activated carbon catalysts in peroxyacetic or nitric acid or in ammonium peroxydisulfate decreased isobutene yield.²⁹¹ The negative effect of nitric acid treatment was explained by destroying of micropores and generation of oxygen-containing functionalities, especially carboxylic acid groups.²⁹² An increase in isobutene selectivity can also be achieved through blocking active sites responsible for the formation of CO_x. Treatment of graphitic mesoporous carbon by phosphoric acid caused such changes.²⁹⁴ However, the positive effect was observed just until a certain coverage by phosphorous surface groups was achieved. This coverage corresponds to P content of 0.19 at%.

6.1.4 Molybdenum-based catalysts. Metal molybdates are another type of inorganic compounds used for hydrocarbon partial oxidation and dehydrogenation reactions.^297–301 These materials possess two dissimilar structures differing in molybdenum cation coordination. For example, in the nickel or cobalt molybdates, molybdenum is octahedrally coordinated in the α-phase but tetrahedrally coordinated in the β-phase (ref. 299 and references herein). It is assumed that the tetrahedrally coordinated sites are more advantageous for selective olefin synthesis because these structures bind olefin weaker.²⁹⁹ The desired β-phase can be formed from the α-phase upon thermal treatment above 675 °C. The stabilization can also be achieved at room temperature when using SiO₂ or TiO₂ as supports for NiMoO₄.^297,301 Nevertheless, these catalysts showed rather low isobutene selectivity even at low degree of isobutane conversion. For example, at 475 °C 39 wt% NiMoO₄/SiO₂ converted 20.3% of isobutane to isobutene with selectivity of 11.7%. At the same temperature isobutane conversion over 6.9 wt% NiMoO₄/TiO₂ achieved 21.8%; the respective selectivity to isobutene was ca. 37%.

An ODB study over magnesium molybdates possessing different crystalline phases has shown that MgMoO₄ is the most selective due to decreased ability for cracking of isobutane/isobutene to propene and methane.²⁹⁸ So, at 530 °C selectivity to isobutene over 1.0MgO/1.0MoO₃ was amounted to 84% at isobutene conversion of 6% using reaction mixture containing 2.7 vol% O₂ and 8.2% of isobutane balanced by He.

Similar to the ODB over V- or Cr-based catalysts, the formation of isobutene occurred via a Mars–van Krevelen mechanism.³⁰⁷ Chemisorbed oxygen species are responsible for CO_x formation. This mechanistic scheme provided the best fit of experimental kinetic data obtained for CoMoO₄ and NiMoO₄³⁰⁷ as well as for Co_0.95MoO₄, NiMoO₄ and MnMoO₄.³⁰⁰ The highest rates of isobutene and propene formation were achieved over Co_0.95MoO₄. It was explained by an optimal strength of olefin-catalyst bond and surface concentration of strong acidic sites.

6.1.5 Metal phosphate/pyrophosphate as catalysts for ODB. Inspired by the high selectivity and activity of vanadyl pyrophosphate for the oxidation of n-butane to maleic anhydride,³⁰⁸ various metal (Ni, Cr, Ce, Mn, Ba, Zr, Mg, V³⁰² and Ti³⁰³) pyrophosphates or phosphates of cerium or lanthanum^304,305 were also tested as catalysts in the ODB reaction. They operate between 400 and 600 °C. V₄(P₂O₇)₃ showed the highest activity, while CeP₂O₇ was the most selective catalyst at degrees of isobutane conversion higher than 20%. The highest yield of isobutene of 20.5% was, however, obtained over Mn₂P₂O₇ at 500 °C. The corresponding value for CeP₂O₇ was only slightly lower, i.e. 20.4%. However, the selectivity to isobutene over the latter catalyst was 51.2%. It is significantly higher than 45.8% achieved over Mn₂P₂O₇. The interplay between Ce and P seems to be important for the selective production of isobutene. CePO₄ showed isobutene selectivity of about 86% but at about only 7% isobutane conversion.³⁰⁴ Even simple phosphating of CeO₂ resulted in an improvement in the selectivity. As shown in ref. 305 the selectivity to isobutene increased from 35 to 67% and 87% with increase in P content to 1.1 and 2.2 wt%, respectively. Importantly, isobutane conversion over bare and phosphated CeO₂ was about 11%. Unfortunately, no tests at higher conversion degrees were performed to check how strongly the selectivity to isobutene will decrease.

6.2 Catalyst comparison in terms of their industrial relevance

To demonstrate an industrial relevance of different catalysts, the catalysts with isobutene selectivity above 70% at isobutane conversion higher than 10% have been selected. The corresponding selectivity–conversion data are shown in Fig. 29.

Fig. 29 Selectivity–conversion plots for V- (),^{272,281,283,288} Cr- ()^265,266,272 or Ce-based ()³⁰² catalysts with isobutene selectivity above 70% at isobutane conversion above 10%. Catalyst composition and reaction conditions are given in Table 6.

Different symbols correspond to catalysts with different active components. The grey lines stand for the yield of propene of 35 and 53%, which are typical for commercial BDH process. In total, 14 catalysts fulfilled the above requirement. Six catalysts contain vanadium as active component. Seven catalysts are based on chromium. Cerium phosphate also showed relatively high performance. Their exact composition and reaction conditions are given in Table 6. The catalysts Cr_0.67V_0.24Nb_0.09, Cr_0.74V_0.19Nb_0.07 and Cr_0.78V_0.16Nb_0.06 from ref. 272 are not shown in this table because the reaction feed applied for ODHB tests was not clearly defined. The reported conversion/selectivity values are 22.8/77.8, 44.4/81.2 and 49.4/70.4%, respectively. They were tested at 300 °C. In comparison with the ODP reaction, several catalysts show the selectivity–conversion data, which are close to those of commercial BDH catalysts (isobutane conversion of 64–65%, isobutene selectivity of 88–89%, 580 °C¹¹²). Nevertheless, the ODB is still not commercialized.

Table 6 Catalysts and reaction conditions for the ODB reaction with co-feeding isobutane and oxygen. The catalysts with isobutene selectivity above 70% at isobutane conversion larger than 10% have been selected from literature since 2000

Catalysts	T/°C	p(i-C4H10)/bar	p(O2)/bar	X(i-C4H10)/%	S(i-C4H8)/%	STY(i-C4H8)/kg kg^−1 h^−1	Ref.
a Prepared using V(t-BuO)3O. b Prepared using NH4VO3. c V/Sb/Ni = 8.8/1/3.6. d V/Sb/Ni = 8.8/1/7.0.
18 wt% Cr/ZnAlLaOx	580	0.79	0.23	52.4	84.1	1.71 × 10^−1	265
6 wt% CrOx/Ce0.60Zr0.35Y0.05O2	540	0.04	0.08	10.5	91.7	2.89 × 10^−4	266
8 wt% CrOx/Ce0.60Zr0.35Y0.05O2	540	0.04	0.08	11.0	93.2	3.08 × 10^−4	266
10 wt% CrOx/Ce0.60Zr0.35Y0.05O2	540	0.04	0.08	11.1	90.6	3.02 × 10^−4	266
CeP2O7	500	0.27	0.07	20.5	70.8	7.21 × 10^−3	302
10.7 wt% V2O5/MCM-41^a	560	0.12	0.06	48.2	87.3	1.51 × 10^0	281
10.7 wt% V2O5/MCM-41^b	560	0.12	0.06	50.1	78.1	1.40 × 10^0	281
V–Sb–Ni–O/Al2O3^c	550	0.2	0.12	43.6	69.6	1.37 × 10^−1	283
V–Sb–Ni–O/Al2O3^d	550	0.2	0.12	46.6	66.3	1.39 × 10^−1	283
5 wt% V2O5–MoO3/Al2O3	500	0.18	0.09	22.2	72.3	1.02 × 10^−1	288
10 wt% V2O5–MoO3/Al2O3	500	0.18	0.09	25.5	70.5	1.14 × 10^−1	288

---

7 Alternative oxidising agents and reaction engineering aspects for improving product selectivity

The main drawback of the oxidative dehydrogenation of C₃–C₄ alkanes to the corresponding olefins is low selectivity to the desired products due to their conversion into carbon oxides. In addition to catalyst, the selectivity can be controlled through reaction engineering aspects or the usage of alternative oxidants. Among the latter, the soft oxidants CO₂ and N₂O were applied. The corresponding results are presented and discussed in Sections 7.1 and 7.2. Another opportunity to improve overall process economy and the selectivity to the olefins is to carry out the dehydrogenation reaction under O₂-lean conditions (alkane/O₂ > 10) or using alternating feeds of alkane and air. The developments in this area are discussed in Sections 7.3 and 7.4.

7.1 General knowledge about alkane dehydrogenation with CO₂

CO₂-DH of propane (eqn (1)) or isobutane can be considered as a combination of non-oxidative alkane dehydrogenation (eqn (2) for propane) and reversed water gas shift reaction (RWGS, eqn (3)). Besides these two reactions, dry reforming of alkane/olefin can occur as a side reaction (eqn (4) for propane). This reaction is thermodynamically favoured at 600 °C³⁰⁹ but seems to be kinetically hindered over most of the catalysts studied.

Different positive effects of CO₂ on alkane dehydrogenation are possible and discussed in literature.

i. Enhancing equilibrium conversion in the non-oxidative dehydrogenation due to H₂ removal through RWGS.

ii. Improving the selectivity to olefins through site isolation caused by partial blockage of the active sites by adsorbed CO₂ (competition with propane adsorption and consequently lowering catalyst activity).

iii. Increasing on-stream stability due to removal of coke through Boudouard reaction.

For the PDH reaction, an increase in thermodynamic limit of propane conversion by about 10% can be reached at 600 °C or the reaction temperature can be lowered by about 50 °C to reach the same conversion when performing the dehydrogenation reaction in CO₂ presence at same partial pressure of propane.³¹⁰ Mechanistically, the CO₂-DH reaction can proceed either through oxidative dehydrogenation with participation of lattice oxygen of metal oxide species followed by reoxidation of the reduced sites by CO₂, or by simple combination of non-oxidative hydrogenation and RWGS. Literature data are controversial in this respect.

C₃H₈ + CO₂ ⇆ C₃H₆ + CO + H₂O ΔH_298K = +165 kJ mol⁻¹ CO₂-PDH (1)

C₃H₈ ⇆ C₃H₆ + H₂ ΔH_298K = +124 kJ mol⁻¹ PDH (2)

CO₂ + H₂ ⇆ CO + H₂O ΔH_298K = +41 kJ mol⁻¹ RWGS (3)

C₃H₈ + 3CO₂ ⇆ 6CO + 4H₂ ΔH_298K = +620 kJ mol⁻¹ dry reforming (4)

The literature on alkane dehydrogenation with CO₂ has been reviewed previously;^9,11,311 the last review¹¹ on CO₂-PDH (CO₂-mediated propane dehydrogenation) has been published in 2018 and covers the literature until 2015. Therefore, the present review comprises mainly the literature published from 2014 on. However, for discussing general trends and extracting rules, older publications are considered too. The CO₂-mediated isobutane dehydrogenation (CO₂-BDH) is less investigated and has not been reviewed after 2004. The developments in this field are described in a separate section after discussing the CO₂-PDH literature.

7.1.1 CO₂-Mediated propane dehydrogenation. To illustrate if the CO₂-PDH is advantageous over the conventional PDH reaction, we prepared Fig. 30, where the yield of propene obtained in each of these reactions is compared. The studies directly comparing the CO₂-PDH and PDH reactions under identical (same temperature and partial pressure of propane) reaction conditions have been selected. Detailed catalytic data are summarized in Table 7. It is seen in Fig. 30 that CO₂ has rather negative effect on the yield over bulk Ga₂O₃ or supported GaO_x-containing catalysts with exception of GaO_x/TiO₂, where a small positive effect was reported. The negative effect for Al₂O₃-supported catalysts seems to be validated as it has been reported in several independent studies while other supported catalysts have been investigated in single studies only. A similar negative effect of CO₂ was established for CrO_x/Al₂O₃. Contrarily, the yield of propene over InO_x/Al₂O₃ is improved in the presence of CO₂. Interestingly, when using SiO₂ as a support for CrO_x or VO_x, CO₂ has a positive effect on the yield. As not many independent studies have been carried out up to now, further studies are required to draw a definitive conclusion about the effects of support and active metal oxide species with respect to the influence of CO₂ on propene production. After this general discussion, recent progress in CO₂-PDH over catalysts with different catalytically active metal oxide species is presented.

Fig. 30 Comparison of propene yields obtained in traditional PDH vs. CO₂-PDH over metal-oxide based catalysts. For clarity reasons in the catalyst names only the element symbols of active metals and supports are given, i.e. Ga/Zr means GaO_x species supported on ZrO₂ and Ga80–Al20 means mixed-metal oxide containing 80 wt% Ga₂O₃ and 20 wt% Al₂O₃. Data are from literature, for references see Table 7.

Table 7 Direct comparison of PDH and CO₂-PDH over identical catalysts using identical reaction conditions

Active material	Support	T/°C	PDH				CO2-PDH					Ref.
			p(C3H8)/bar	X(C3H8)/%	S(C3H6)/%	Y(C3H6)/%	p(C3H8)/bar	p(CO2)/bar	X(C3H8)/%	S(C3H6)/%	Y(C3H6)/%
CrOx	SiO2	550	0.2	12.7	92.1	11.7	0.2	0.8	17.2	94.4	16.3	334
5CrOx	SBA-1	550	0.067	33.0	86.0	28.4	0.067	0.333	37.2	85.4	31.8	314
CrOx	Al2O3	600	0.2	22.1	93.1	20.5	0.2	0.8	12.7	84.4	10.7	310
CrOx	Al2O3	550	0.2	14.0	95.8	13.4	0.2	0.8	12.9	88.3	11.4	334
4Cr	Al2O3	550	0.125	9.6	93.6	9.0	0.125	0.375	5.2	89.8	4.7	309
CrOx	ZrO2	550	0.025	69.0	74.0	51.1	0.025	0.065	41.0	62.	25.4	335
CrOx	ZrO2	550	0.025	69.0	67.0	46.2	0.025	0.065	58.0	52.0	30.2	335
CrOx	ZrO2	550	0.025	74.0	76.0	56.2	0.025	0.065	59.0	55.0	32.5	335
CrOx	ZrO2	550	0.025	69.0	65.0	44.9	0.025	0.065	58.0	57.0	33.1	335
6.8V	MCM-41	600	0.11	23.9	76.4	18.2	0.11	0.55	53.5	84.8	45.4	322
6.8V	MCM-41	600	0.11	23.9	76.4	18.2	0.11	0.33	43.2	89.2	38.5	322
6.8V	MCM-41	600	0.11	23.9	76.4	18.2	0.11	0.11	40.6	88.6	36.0	322
VOx	SiO2	600	n.r.	47.0	74.8	35.1	0.111	0.444	59.1	81.9	48.4	323
7V	Al2O3	550	0.125	7.2	96.3	6.9	0.125	0.375	6.7	96.7	6.5	309
7V4Cr	Al2O3	550	0.125	9.2	95.4	8.8	0.125	0.375	10.0	99.6	9.9	309
7V5Mo	Al2O3	550	0.125	10.3	96.7	9.9	0.125	0.375	10.1	96.6	9.7	309
7V7W	Al2O3	550	0.125	6.9	97.3	6.7	0.125	0.375	9.9	97.1	9.6	309
5Mo	Al2O3	550	0.125	2.6	96.9	2.5	0.125	0.375	2.2	96.3	2.1	309
7W	Al2O3	550	0.125	0.6	90.6	0.6	0.125	0.375	0.6	90.9	0.5	309
Ga2O3		500	0.025	41.3	93.3	38.5	0.025	0.05	35.9	97.2	34.9	325
Ga2O3	TiO2	600	0.025	23.0	85.0	19.6	0.025	0.05	32.0	73.0	23.4	326
Ga2O3	Al2O3	600	0.2	9.6	89.3	8.6	0.2	0.8	5.4	90.8	5.3	310
Ga2O3	Al2O3	600	0.025	33.0	92.0	30.4	0.025	0.05	26.0	94.0	24.4	326
Ga2O3	ZrO2	600	0.025	39.0	74.0	28.9	0.025	0.05	30.0	65.0	19.5	326
Ga2O3	SiO2	600	0.025	7.2	92.0	6.6	0.025	0.05	6.4	92.0	5.9	326
Ga2O3	MgO	600	0.025	5.3	34.0	1.8	0.025	0.05	4.3	29.0	1.2	326
Ga8Al2O15		500	0.025	51.7	91.6	47.4	0.025	0.05	49.7	91.7	45.6	325
Ga5Al5O15		500	0.025	38.4	92.3	35.4	0.025	0.05	33.7	92.9	31.3	325
Ga2Al8O15		500	0.025	22.8	94.9	21.6	0.025	0.05	19.3	92.9	17.9	325
Ga–Al–O ht		550	0.032	40.5	92.1	37.3	0.032	0.097	35.2	95.1	33.4	327
Ga–Al–O prec.		550	0.032	33.6	91.2	30.6	0.032	0.097	26.2	95.2	24.9	327
Ga–Al–O grind.		550	0.032	11.4	90.8	10.4	0.032	0.097	8.7	95.8	8.3	327
In2O3		600	0.025	3.0	69.0	2.1	0.025	0.1	1.0	55.0	0.6	329
In-Al-40		600	0.025	12.0	80.0	9.6	0.025	0.1	22.5	72.0	16.2	329
In-Al-20		600	0.025	16.0	84.0	13.4	0.025	0.1	27.5	78.0	21.4	329
In-Al-10		600	0.025	14.0	86.0	12.0	0.025	0.1	26.0	78.0	20.3	329

---

Cr-Containing catalysts. CrO_x-Based catalysts have been most frequently applied for the CO₂-PDH. In the first review published in 2004 only such catalysts have been analysed.⁹ This may be because Cr₂O₃ supported on alumina is one of the two types of commercially applied catalysts for the PDH^15–18 and water gas shift³¹² reactions. Meanwhile, catalysts containing CrO_x on different supports (mesoporous silica, alumina, zirconia) have been studied.

Owing to the neutral (neither strong acidic nor strong basic sites) character of SiO₂, CrO_x/SiO₂ catalysts have been investigated in several studies focused on various mechanistic aspects.^313–318 Michorczyk et al.³¹⁹ applied operando UV-vis spectroscopy for determining the oxidation state of chromium in CrO_x/SiO₂(SBA-1) under reaction conditions. It was shown that the initial Cr⁶⁺ species are reduced to Cr³⁺/Cr²⁺ ones already within the first 10 minutes of PDH or CO₂-PDH. Such changes indicate that the oxidative DH over Cr⁶⁺ plays only a minor role. Similar results were obtained for CrO_x/SiO₂ (TUD-1 mesoporous silica matrix) prepared by microwave-assisted method.³¹⁵ Propene selectivity was above 90% (propane conversion between 5 and 24%) up to 550 °C but decreased to 75% (propane conversion of 45%) at 650 °C. The catalysts initially deactivate rapidly due to coke formation but became more stable after about 2 hours for additional 6 hours. The deactivation rate is lower for the CO₂-assisted reaction what is explained by the lower rate of coke formation.

Wang et al.³¹⁶ studied the influence of preparation method (Cr addition at different stages of synthesis of mesoporous silica from TEOS) on the catalytic performance in CO₂-PDH. The method affects dispersion of CrO_x species as well as surface physicochemical properties. The sample, where Cr was added after TEOS, showed the highest activity, and possesses the lowest surface acidity and the highest Cr⁶⁺/Cr³⁺ ratio at the surface. For this sample, propene selectivity of about 90% at propane conversion of 40% was obtained. The catalysts deactivate but reactivation with O₂ restored the initial activity over 5 cycles. Agafonov et al.³¹⁷ compared the catalytic performance of silica-supported CrO_x catalyst with CrO_x/Al₂O₃ and GaO_x supported on both silica and alumina. The highest propene yield of about 40% was obtained over CrO_x/SiO₂, however the propene selectivity was only 50%. This is due to a high activity of the fresh catalyst towards methane and coke formation. CrO_x/Al₂O₃ was even less active and selective. CO₂ was concluded to actively participate in the RWGS reaction and in coke removal through the Boudouart reaction. Very recently, Jin et al.³¹⁸ studied Ru-promoted CrO_x/SiO₂ catalysts for CO₂-PDH. The activity showed a volcano-type curve with respect to Ru loading. Small amounts of Ru promote the activation of both propane and CO₂ as well as help to shift the equilibrium to higher conversions through removal of adsorbed H₂ by RWGS reaction. At higher Ru loading (3 wt% Ru), however, CO₂ adsorption increases and hinders propane adsorption. Moreover, the high Ru loaded catalyst shows high activity for dry reforming of propene as a side reaction that decreases propene selectivity.

Very recently, Michorczyk et al.³²⁰ investigated Cr-containing aluminium-free (Si/Al = 1000) beta zeolite prepared by dealumination followed by its incipient wetness impregnation. Low acidity of the dealuminated zeolite is responsible for high propene selectivity (>80%) at propane conversion of up to 33%. When switching between CO₂-free and CO₂-containing feeds it could be shown that higher propene formation rate was obtained over the dealuminated catalysts with CO₂ whereas the rate over the Al-containing catalysts decreased in CO₂ presence. Actually, as seen in Fig. 30, other SiO₂-supported Cr-containing catalysts revealed higher propene yield in CO₂-PDH than in PDH, while an opposite effect is valid for their Al₂O₃-supported counterparts. Thus, the kind of support seems to be decisive for the effect of CO₂ on propene production.

De Oliveira et al.³²¹ studied PDH and CO₂-PDH over CrO_x/ZrO₂ prepared by conventional and microwave-assisted hydrothermal preparation methods. The yield of propene is lower for all catalysts when adding CO₂ to the feed (Fig. 30). This is due to a decrease in both propane conversion and propene selectivity (see Table 7). The lower activity is caused by blocking active sites due to strong CO₂ adsorption whereas the decrease in the selectivity is due to blocking basic sites by CO₂ that favours acidity and leads to increased methane formation. Additional experiments showed that reduced CrO_x can partly be re-oxidized by CO₂. Some of the catalysts deactivate slower when CO₂ is present in the feed. Regeneration of deactivated catalysts with CO₂ is not as effective as with O₂ because only part of coke can be removed.

CO₂-Assisted PDH over CrO_x/Al₂O₃ was very recently studied by Sandupatla et al.³⁰⁹ The propene yield is lower for the CO₂-PDH in comparison with the conventional PDH (Fig. 30) what is mainly due to decreased activity (Table 7). Reduced CrO_x sites present under reaction conditions catalyse the PDH reaction but can also strongly adsorb CO₂ what explains the reduced activity in CO₂ presence.

V-Containing catalysts. Supported V-containing materials are the most intensively studied catalysts both in the PDH (Section 2) and ODP (Section 5) reactions. Therefore, it is not surprising that this type of catalysts has also been applied for CO₂-PDH. Different reaction mechanisms are proposed for the latter reaction; either oxidative propane activation followed by re-oxidation of reduced VO_x species by CO₂ or the PDH reaction assisted by RWGS that can shift the equilibrium to higher conversion.

The direct comparison of PDH and CO₂-PDH under identical reaction conditions (Fig. 30 and Table 7) shows that the yield of propene over VO_x supported on mesoporous silica^322,323 is strongly increased when adding CO₂ to the feed. However, no increase in the yield was found over VO_x/Al₂O₃.³⁰⁹ To analyse the positive effect of CO₂ on propene production in detail we plotted propene selectivity versus propane conversion over VO_x/SiO₂ in conventional PDH and CO₂-PDH^322,323 (Fig. 31). Both the selectivity and the conversion are increased when adding CO₂ to the feed. The propane conversion could be further improved upon increasing CO₂/propane ratio at a constant partial pressure of propane in the feed. Thus, CO₂ enhances both the turnover frequency and the selectivity. Unfortunately, the corresponding selectivity–conversion relationships were not provided to understand which undesired pathways are suppressed in CO₂ presence.

Fig. 31 Selectivity–conversion plot for propene in PDH and CO₂-PDH over VO_x/SiO₂. Data are from ref. 322 and 323.

The influence of V loading on activity was very similar for VO_x either incorporated in MCM-41³²² or in three-dimensional dendritic mesoporous silica nanospheres.³²³ The propane conversion passes a maximum with increasing V loading. This was explained by the higher intrinsic activity of lower aggregated 2D VO_x species in comparison with 3D V₂O₅ aggregates. The catalysts deactivated partly within 2 hours on stream due to coke formation. They could restore their activity after regeneration in O₂ at the reaction temperature. In these both studies, the oxidative mechanism of CO₂-PDH was suggested, although RWGS was not excluded. However, no solid experimental proof was presented in ref. 322 and 323. Such information was provided in a separate study, where WO_x–VO_x/SiO₂ catalysts (0–8.8 wt% W and 3–4 wt% V) prepared by incipient wetness impregnation were tested for CO₂-PDH over the samples treated with O₂, reduced by H₂ or reoxidized by CO₂ after the dehydrogenation reaction.³²⁴ Moreover, pure RWGS reaction and CO₂-PDH with a feed mixture D₂:C₃H₈:CO₂ were analysed over one catalyst sample. It was concluded that the CO₂-PDH reaction proceeds through a Mars–van Krevelen mechanism. Besides, RWGS happens in combination with the PDH reaction, too. The experimental findings were supported by DFT calculations which show that the C–H activation over V⁵⁺O_x is the rate limiting step, while reoxidation of the reduced VO_x by CO₂ is fast.

Sandupatla et al.³⁰⁹ studied the CO₂-PDH over VO_x(7 wt%)/Al₂O₃ prepared by incipient wetness impregnation. The propene yield could not be increased in comparison with the PDH. However, when doping VO_x/Al₂O₃ by CrO_x or WO_x, higher yield was obtained in the CO₂-PDH (Fig. 30 and Table 7). This result is surprising because VO_x/Al₂O₃ CrO_x/Al₂O₃ and WO_x/Al₂O₃ becomes less active in CO₂ presence (Table 7).

Ga-Containing catalysts. Ga-Containing catalysts have been frequently studied for CO₂-PDH.¹¹ Bulk Ga₂O₃ deactivates fast,³²⁵ while its supported counterparts or Ga-containing mixed-metal oxides reveal higher on-stream stability.¹¹ Tetrahedrally coordinated Ga³⁺ Lewis sites are assumed to be responsible for propane activation. As mentioned above, CO₂ does not improve the yield of propene over SiO₂-, Al₂O₃- or ZrO₂-supported catalysts (Fig. 30 and Table 7). A slight increase in the yield was reported only for GaO_x/TiO₂.³²⁶ This was explained by the high resistance of this catalysts against blocking sites by CO₂ and its high activity in RWGS. Nevertheless, CO₂ can have a positive influence on the selectivity to propene and on-stream stability as seen from the below discussion.

The influence of different supports (SiO₂, Al₂O₃, TiO₂, ZrO₂, MgO) on catalytic performance of supported GaO_x species was studied in detail by Xu et al.³²⁶ The usage of SiO₂ and MgO as supports resulted in low activity. Agafonov et al.³¹⁷ confirmed low activity of GaO_x/SiO₂. The catalysts based on other supports were active what was related to their acidic properties.

Despite the fact that propene yield over Ga–Al-based catalysts (supported GaO_x and mixed-metal oxide) is lower in CO₂ presence (Fig. 30 and Table 7), the selectivity to propene is increased for several catalysts and the lower yield is only due to lower activity (Fig. 32). It is explained by blockage of active sites by CO₂ (see below). Ga₂O₃–Al₂O₃ solid solutions were studied by Chen et al.³²⁵ With increasing Ga₂O₃ content (from 20 over 50 to 80 wt%) higher propene yields were obtained (see Fig. 30 and Table 7). The lowest deactivation rate over 50 hours at high conversion was found for 50 wt% Ga₂O₃–50 wt% Al₂O₃. In comparison with the PDH reaction, higher propene selectivity was achieved in CO₂ presence (Fig. 32 and Table 7). This conclusion, however, needs validation through, for example, analysis of selectivity–conversion profiles obtained for individual catalysts at same temperature and using a feed mixture with same propane partial pressure.

Fig. 32 Selectivity–conversion plot for propene in PDH and CO₂-PDH over GaO_x/Al₂O₃ and GaAlO_x. Data are from ref. 310 and 325–327.

The effect of preparation method (hydrothermal, co-precipitation and grind-mixing) on catalytic performance of Ga–Al mixed oxides was investigated by Xiao et al.³²⁷ The hydrothermally prepared sample showed higher and more stable propane conversion whereas the propene selectivity was initially 95% for all samples. Importantly, the selectivity did not decrease with time on stream over this catalyst only. Its superior performance was attributed to higher concentration of medium strong Lewis-acidic tetrahedral Ga³⁺ on the surface. In addition, a small fraction of reduced Ga^δ+ (δ < 2) sites was identified by XPS on the surface of this catalyst. It is, however, not clear how and if these species besides Ga³⁺ contribute to the high catalyst activity and selectivity. XPS spectra of gallium after both PDH and CO₂-PDH are similar indicating the presence of Ga³⁺ on the surface what let the authors conclude that no redox mechanism occurs over their catalysts.

In-Containing catalysts. Bulk In₂O₃ shows only low activity and selectivity to propene.^328,329 However, the catalytic performance is enhanced when InO_x species are highly dispersed either on the surface of an oxidic support material^328,330,331 or within a mixed-metal oxide.^328,329,331 This is due to easier reducibility of dispersed InO_x species in comparison with bulk In₂O₃.^328,329 For binary mixed metal oxides the following activity order was determined: InAlO_x > InZnO_x > InZrO_x and InTiO_x ≫ InMeO_x (Me = Fe, Mg, Si, or Ce). The latter materials show as low propene yield as bulk In₂O₃.³²⁸ Metallic In⁰ surface species are identified as catalytically active species. This was also concluded in a study published quite recently dealing with CO₂-PDH over In/HZSM-5 catalysts prepared by incipient wetness impregnation.³³² In general, the following four experimental findings support the important role of In⁰ for propene formation:

(i) In³⁺ is reduced to In⁰ under reaction conditions and In⁰ cannot be re-oxidized by CO₂.^328,329

(ii) In⁰ was identified by XPS on the surface of catalysts after CO₂-PDH or reduction by H₂ and treated afterwards by CO₂.^329,330

(iii) The catalytic activity and product selectivity depend on the concentration of surface In⁰ species.^328–330

(iv) Fresh calcined mixed In₂O₃–Al₂O₃ catalysts show an induction period of about three hours where activity increases slightly, and selectivity is enhanced drastically (from 25 to more than 80%). High activity and selectivity are achieved over reduced catalysts from the beginning of the reaction.³²⁸

The catalytic performance of In-based catalysts is strongly influenced by acidic and basic properties of the catalyst that can be adjusted by the ratio of In₂O₃ to the second metal oxide and the nature of the second metal oxide itself. In contrast to bulk In₂O₃, the number of acidic sites in In–Al–O catalysts increases due to acidic sites of alumina. The total amount of basic sites (mmol g_cat⁻¹) increases also but the specific basicity (μmol g_cat⁻¹) is lower what is consistent with the higher specific basicity of the low-surface area In₂O₃ compared to alumina.³²⁵ The activity of In–Al–O catalysts increases with temperature but at temperatures above 600 °C the selectivity to propene decreases due to higher cracking activity. On-stream stability also decreased at high temperatures.³²⁸

Very recently, In/HZSM-5 catalysts prepared by incipient wetness impregnation were studied for CO₂-PDH.³³² Small In₂O₃ crystallites (20–30 nm) are supported on the outer surface of HZSM-5. As already described for supported and mixed-metal oxide catalysts, a part of In³⁺ is reduced at low temperature (150–300 °C) to In⁰ and the reduction temperature increases with In₂O₃ loading. Thus, In⁰ species are present under reaction conditions as confirmed by XPS of spent catalysts. Catalytic results are similar compared to the catalysts described above. Studying different propane/CO₂ ratios showed that activity and selectivity were not much influenced.

Other catalysts. Besides catalysts based on Cr, V, Ga or In, other active elements have been described in literature, too. MoO_x or WO_x supported on alumina show much lower propane conversion compared to VO_x or CrO_x deposited on the same support. Propene yield was lower for CO₂-PDH compared to conventional PDH (ref. 309, Fig. 30 and Table 7).

ZnO supported on HZSM-5 with the Si/Al ratio of 160 was studied for CO₂-PDH.³³³ Initial propane conversion of 68% decreases to 41% after 30 hours. However, propene selectivity was low (47–62%) due to formation of aromatics and cracking products. A promoting effect of CO₂ on propene yield was found and attributed to RWGS reaction. Moreover, catalyst stability was enhanced by CO₂.

In summary, several catalysts based on oxides of Cr, V, Ga, In, Mo, W or Zn have been studied for CO₂-PDH. Except for In, that is metallic, all other active sites are present as metal-oxide species on the catalyst surface under reaction conditions. Depending on the kind of support and active component, CO₂ may have either positive or negative effects on propene yield. This is due to different catalyst resistance against blockage of catalytically active sites by CO₂. For some catalysts, CO₂ positively affects the selectivity and on-stream stability. With this respect, supported or mixed metal oxides on Ga basis perform superior to other metal oxides.

7.1.2 CO₂-Mediated isobutane dehydrogenation. Compared with the CO₂-PDH reaction, less literature is available for CO₂-mediated isobutane dehydrogenation (CO₂-BDH). Besides dehydrogenation and cracking reactions, isobutane/isobutene can also undergo side isomerization reactions. However, in several studies n-butenes are not mentioned as reaction products and, hence, it is not clear if the presented isobutene selectivity values are too high because they also contain n-butenes.

Several studies published catalytic data obtained under the same reaction conditions (temperature and partial pressure of isobutane) both in the presence and the absence of CO₂. Fig. 33 shows the corresponding yields to isobutene to illustrate if the activity or the selectivity is changed by the presence of CO₂ in the feed. Additional experimental data are given in Table 8. As seen in Fig. 33, the isobutene yield over several catalysts is higher in the presence of CO₂. This is in most cases due to higher activity (Table 8). However, the presented initial data are measured at the beginning of the reaction and very often the catalysts deactivate faster when CO₂ is in the feed. Isobutene selectivity in CO₂-BDH is in most studies equal or lower (due to the higher conversion) than in BDH. This selectivity is improved by CO₂ only in CO₂-BDH over VO_x supported on silica (SBA-15 or silicalite-1) (Table 8).

Fig. 33 Comparison of isobutene yields for BDH vs. CO₂-BDH over metal-oxide based catalysts. For clarity reasons only the element symbols of active metals and supports in the catalyst names are given, i.e. V/Si means VO_x species supported on SiO₂ and 10V–Mg means V₂O₅–MgO mixed-metal oxide with 10 wt% V₂O₅. Data are from literature, for references see Table 8.

Table 8 Direct comparison of BDH and CO₂-BDH over identical catalysts using identical reaction conditions

Active material	Support	T/°C	BDH				CO2-BDH					Ref.
			p(i-C4H10)/bar	X(i-C4H10)/%	S(i-C4H8)/%	Y(i-C4H8)/%	p(i-C4H10)/bar	p(CO2)/bar	X(i-C4H10) 7%	S(i-C4H8)/%	Y(i-C4H8)/%
a Measured after 4 h on stream. b Measured after 6 h on stream. All other values were measured at initial time on stream after 0.1–0.2 hours.
20% VOx	CeO2	600	0.14	2.9	73.9	2.1	0.14	0.86	3.9	64.4	2.5	337
20% VOx	La2O3	600	0.14	4.2	65.4	2.7	0.14	0.86	4.2	63.2	2.7	337
20% VOx	SiO2	600	0.14	12.1	90.4	10.9	0.14	0.86	11.6	86.0	10.0	337
20% VOx	TiO2	600	0.14	2.2	68.5	1.5	0.14	0.86	2.8	63.6	1.8	337
20% VOx	Al2O3	600	0.14	26.2	83.9	22.0	0.14	0.86	13.2	78.1	10.3	337
20% VOx	ZrO2	600	0.14	2.4	69.7	1.7	0.14	0.86	2.7	65.7	1.8	337
20% VOx	ZnO	600	0.14	11.5	88.3	10.2	0.14	0.86	7.9	59.3	4.7	337
20% VOx	Activated C	600	0.14	45.5	85.0	38.7	0.14	0.86	54.8	79.9	43.8	337
12VOx	MSU-1	600	0.25	40.5	82.8	33.5	0.25	0.75	58.8	78.5	46.2	336
7VOx	SBA-15	570	n.r.	32.5	76.0	24.7	0.17	0.83	41.0	84.5	34.6	338
7VOx	SBA-15	570	n.r.	31.5^a	67.5^a	21.3^a	0.17	0.83	23.0^a	88^a	20.2^a	338
2.5V–Mg–Ox		600	0.17	5.8	72.0	4.2	0.17	0.83	9.8	68.0	6.7	339
5V–Mg–Ox		600	0.17	7.8	76.0	5.9	0.17	0.83	12.0	71.0	8.5	339
10V–Mg–Ox		600	0.17	8.5	82.0	7.0	0.17	0.83	13.7	77.0	10.5	339
20V–Mg–Ox		600	0.17	12.4	80.0	9.9	0.17	0.83	16.0	75.0	12.0	339
30V–Mg–Ox		600	0.17	17.5	82.0	14.4	0.17	0.83	13.5	75.0	10.1	339
40V–Mg–Ox		600	0.17	18.5	86.0	15.9	0.17	0.83	11.9	73.0	8.7	339
50V–Mg–Ox		600	0.17	17.1	84.0	14.4	0.17	0.83	10.0	69.0	6.9	339
60V–Mg–Ox		600	0.17	12.0	80.0	9.6	0.17	0.83	6.3	70.0	4.4	339
69V–Mg–Ox		600	0.17	8.0	75.0	6.0	0.17	0.83	6.4	80.0	5.1	339
82V–Mg–Ox		600	0.17	9.7	80.0	7.8	0.17	0.83	7.7	83.0	6.4	339
3Cr	Silicalite-1	570	0.5	31.0	67.5	20.9	0.5	0.5	37.0	71.5	26.5	343
3Cr	Silicalite-1	570	0.5	19.5^b	71.0^b	13.8^b	0.5	0.5	27.5^b	75.5^b	20.8^b	343
NiO	Activated C	550	0.14	40.5	86.1	34.9	0.14	0.86	48.0	86.8	41.7	345

---

The results recently obtained in CO₂-BDH over VO_x/SiO₂ are not consistent. Whereas the isobutane conversion over VO_x/MSU-1³³⁶ is much higher with CO₂, it is not changed over VO_x/SiO₂.³³⁷ Over VO_x/SBA-15³³⁸ it is only higher for the beginning of the reaction and then decreases rapidly and becomes much lower than without CO₂. The same behaviour was determined over VMgO_x mixed-metal oxides with V loading up to 20 wt% but not for their higher loaded counterparts where the conversion is always lower for CO₂-BDH in comparison with BDH.³³⁹ The isobutene selectivity is increased in the presence of CO₂ in ref. 336 but decreased in ref. 337 and 338. The catalyst stability is improved in ref. 336 but worsen in ref. 338.

The influence of support on the activity and the selectivity to isobutene in the CO₂-BDH was studied by Ogonowski and Skrzynska.³³⁷ The activity decreased in the order: VO_x/activated carbon ≫ VO_x/Al₂O₃ > VO_x/SiO₂ > VO_x/ZnO. VO_x supported on CeO₂, TiO₂ or ZrO₂ showed very low activity. Among these catalysts only VO_x/activated carbon revealed an increase in isobutene selectivity in CO₂-BDH in comparison with BDH. An increase in isobutane conversion was found over VO_x supported on or incorporated into CeO₂–ZrO₂ as was shown by lower reaction temperature needed in TPRS experiment.³⁴⁰ The selectivity to C₄ olefins was above 85% at isobutane conversion up to 20%.

CO₂-BDH over VMgO_x mixed metal oxides with V₂O₅ content between 2.5 and 82 wt% was studied in ref. 339 and 341. High isobutene selectivity of 75–85% over these catalysts was obtained at isobutane conversion below 20% for both CO₂-BDH and BDH. The selectivity was slightly higher without CO₂ (Table 8). For lower V₂O₅ loading (<30 wt%) the isobutane conversion is only initially higher with CO₂ but then decreases with a higher rate than in BDH. The higher deactivation rate is due to higher rate of coke formation in CO₂-BDH caused by the higher surface acidity of the catalyst in presence of CO₂. The acidity was determined by means of isopropanol decomposition measurements. Although some coke is removed by CO₂ through the Boudouard reaction, the rate of its removal is lower than the formation rate. Mechanistic study over the same catalyst (VMgO_x)³⁴¹ showed that two-step dehydrogenation (BDH and RWGS) occurs whereas the redox mechanism was excluded because oxidizing properties of CO₂ are insufficient.

For CrO_x supported on SBA-15 (partly doped by Ce)³⁴² or silicalite-1,³⁴³ it was shown that the activity in CO₂-BDH is lower over the catalysts pre-reduced with H₂ in comparison with the oxidized ones. The conversion is higher in the presence than in the absence of CO₂.³⁴² This indicates that Cr⁶⁺ is more active than reduced Cr species and the latter can be partly reoxidized in situ by CO₂ as shown by XPS.³⁴³ A direct correlation was established between isobutane conversion and content of Cr⁶⁺.³⁴³ From these results, it was concluded that CO₂-BDH follows a redox mechanism. However, because H₂ was also detected among the reaction products, the two-step reaction pathway (non-oxidative BDH and RWGS) also occurs. The isobutene selectivity is between 90 and 92% at isobutane conversion up to 35% over Cr–Ce/SBA-15 catalyst³⁴² but between 70 and 80% at 21–36% conversion over Cr/silicalite-1 catalysts.³⁴³

An increase in both activity and selectivity over NiO/Al₂O₃ in CO₂-BDH (CO₂/isobutane ratio of 15) in comparison with the BDH reaction was reported in ref. 344. However, it was not indicated if the isobutane was diluted to the same degree in both reaction feeds. NiO supported on activated carbon was studied for the CO₂-BDH by the same group.³⁴⁵ NiO is partly reduced to Ni by the carbon support during calcination in Ar atmosphere at higher temperature and, thus, by variation of calcination temperature the different catalytic performance of NiO and Ni could be studied. Whereas the former catalyses BDH and RWGS, the latter is only active for BDH.

MoO₃ catalysts either undoped or doped with Fe, Ce or Sn were prepared by decomposition of heteropoly molybdates and tested for CO₂-BDH.³⁴⁶ With increasing CO₂/isobutane ratio the conversion increased. The selectivity to isobutene over these catalysts was very low (below 25%) because of high activity for cracking what results in selectivity to cracking products up to more than 90%. Experiments without CO₂ were not reported.

Summarising the attempts of CO₂-mediated dehydrogenation of propane and isobutane, one can state that promising results were obtained over some catalysts. For different mesoporous silica with supported VO_x species, olefin yield in the presence of CO₂ is higher than in the CO₂-free dehydrogenation and reaches the industrially relevant level of 36–53% (see Fig. 13 and 16). However, co-feeding of CO₂ does not help to reduce catalyst deactivation. Although it was shown that regeneration in air is possible for some cycles, long-time studies are missing up to now. Further studies in this respect are necessary before drawing a definitive conclusion on industrial applicability of such approach. Moreover, economic evaluation of a CO₂-assisted process is necessary that compares the benefit from higher olefin yield with the additional effort for removing co-formed CO from the product mixture. Since to date only a few studies on CO₂-assisted dehydrogenation of propane or isobutane over the promising VO_x/SiO₂ catalysts have been published, even better-performing catalysts and deeper understanding of the function of CO₂ in the reaction mechanism including coke formation can be achieved in the nearer future.

7.2 Dehydrogenation with N₂O

A few studies have used N₂O as an oxidising agent in the ODP reaction using V-based catalysts,^179,211,347 Cr-based²³⁹ catalysts or Fe-containing zeolites.^348–352 The main purpose of the studies with V-based catalysts was to elucidate factors affecting selective and non-selective reaction pathways. When using N₂O instead of O₂ for the ODP reaction, the selectivity to propene increases due to an increase in the reduction degree of bulk and surface VO_x.^179,181,182 In addition, non-selective biatomic (peroxo) adsorbed species cannot be formed from N₂O that also has a positive effect on the desired selectivity.^177,178

Fe-Containing zeolites of the MFI type have been often applied for the ODP reaction in the past. Such studies have been stimulated by the fact that a certain kind (so-called alpha oxygen) of oxygen species is formed from N₂O over FeO_x sites and can selectively oxidize benzene to phenol.³⁵³ We now discuss the results reported after 2010. Similar to the ODP reaction with O₂, we plot available selectivity–conversion data obtained at 450 and 500 °C over differently prepared catalysts. The results are shown in Fig. 34.

Fig. 34 Selectivity–conversion data obtained in the ODP reaction with N₂O over FeO_x/ZSM-5 catalysts prepared through grafting³⁵² (▲), solid³⁴⁸ (■) or liquid ion³⁵⁰ (●) exchange methods. The lines show the yield of propene.

Propene selectivity above 70% was achieved at degrees of propane conversion between 30 and 40%. This performance is superior to that achieved with O₂ (Fig. 20). The results in Fig. 34 also provide some hints about the importance of method of deposition of FeO_x species, iron loading and zeolite structure. The highest yield was obtained over the catalysts prepared through a grafting method,³⁵² while those synthesised through solid³⁴⁸ or liquid³⁵⁰ ion exchange methods were less selective. In the grafting method, ferrocene (iron precursor) sublimated at high temperatures and then reacted with the Brønsted acidic sites of zeolite supports, which were H-ZSM-5, H-mordenite, H-USY, H-ZSM-35 and H-beta. In contrast to all other catalysts, FeO_x/ZSM-35 showed very low activity. The highest activity and propene selectivity were achieved over FeO_x/ZSM-5. For the latter catalyst, the effect of calcination temperature on the ODP performance was investigated. As a general trend, the conversion of propane decreased with increasing calcination temperature from 600 to 900 °C, while the selectivity to propene increased. Importantly, when these catalysts were used in the ODP reaction with O₂, carbon oxides were the main reaction products. It was concluded that N₂O provides an atomic oxygen species that facilitate the desired dehydrogenation reaction. Extra-framework Fe–O–Al species were suggested to be responsible for generation of such species from N₂O. FeO_x/ZSM-5 catalysts prepared through a solid-state exchange technique using FeCl₂·4H₂O as the iron precursor possess differently structured FeO_x species ranging from isolated to large Fe₂O₃ particles.³⁴⁸ The presence of the latter is probably a reason for low propene selectivity obtained over such catalysts.

A simple impregnation method with FeCl₂ dissolved in acetylacetone was used to prepare two FeO_x/ZSM-5 catalysts. They were tested in the ODP reaction at 400 °C using a diluted feed (1.5 vol% N₂O and 1.5 vol% C₃H₈). Propene selectivity between 40–50% was achieved at about 30% propane conversion. Under such reaction conditions, coke formation, which is a major undesired reaction, is strongly suppressed. Nevertheless, the catalysts lose their activity with rising time on stream. Protonic zeolite sites were concluded to contribute to coke formation and accordingly affect catalyst deactivation.

Although the usage of N₂O has a positive effect on propene selectivity, this oxidant is not attractive for large-scale applications. First of all, it is very expensive in comparison with O₂ and not available in large amounts as required for propene production. Moreover, N₂O decomposition is an exothermic process and thus precautions should be taken into account for heat management. In case of propane recycling (typically propane conversion is below 50%), non-reacted propane must be separated from N₂ formed from N₂O.

7.3 O₂-Lean dehydrogenation

The oxidative dehydrogenation of propane and isobutane is typically carried out using feeds with the ratio of C₃H₈(C₄H₁₀)/O₂ between 2 and 0.5 to achieve high propane conversion. However, the reported conversion degrees are significantly lower than the expected stoichiometric values due to formation of carbon oxides requiring between 7 and 13 oxygen atoms in comparisons to 0.5 for the desired selective reaction. The oxidative dehydrogenation of propane and isobutane was also investigated under O₂-lean conditions (C₃H₈(C₄H₁₀)/O₂ > 10).^150,354 For the oxidative dehydrogenation of isobutane, an overall C₃–C₄ olefin selectivity of 97% (85% to isobutene, 9% to n-butenes and 2% to propene) at isobutane conversion of 54% was achieved at 560 °C over VO_x(5 wt%)/MCM-41.³⁵⁴ When performing the ODP reaction under O₂-lean conditions at same temperature and over same catalyst, the selectivity to propene of 80% at 50% propane conversion was reported.¹⁵⁰ From a mechanistic viewpoint, it was suggested that the oxidative alkane dehydrogenation takes place in the up-stream catalyst layer when gas-phase O₂ is still present. After the oxidant was completely consumed, olefin is formed through the non-oxidative dehydrogenation in the bottom part of catalyst layer. In comparison with the pure PDH reaction, heat required for this reaction can be provided by the ODP reaction in the upper catalyst layer. Another advantage of O₂-lean operation is the low concentration of gas-phase oxygen. Thus, safer and cheaper operation can be ensured than in the case of oxidative dehydrogenation with O₂-rich feeds.

Similar to ref. 354, Fan et al.²⁸¹ also investigated the ODB reaction over VO_x/MCM-41 with different V loading under O₂-lean conditions. The catalyst with V content of 6 wt% showed the highest yield of isobutene of 42.1% at the corresponding selectivity of 87.3%. Isolated VO_x species were concluded to reveal the highest activity, selectivity and on-stream stability.

7.4 Propane oxidative dehydrogenation upon alternating propane and air feeds

Another opportunity to improve overall economy of the ODP reaction and the selectivity to propene is the usage of air instead of pure oxygen for eliminating expensive air separation. Such operation can be achieved through usage of oxygen-conducting membranes or upon alternating feeding of air and propane. The latter operation mode is also called chemical looping. Propane and air are separately fed to the reactor, where the alkane is oxidized by lattice oxygen of reducible metal oxide followed by catalyst reoxidation with air. Using a membrane (BaBi_0.05Co_0.8Nb_0.15O_3−δ perovskite) reactor with CoO_x/SiO₂ catalyst, propane yield of about 50% at propene selectivity of 74% was obtained over 50 h on stream at 650 °C.³⁵⁵ Unfortunately, such high temperatures are required to ensure reasonable flux of lattice oxygen through the membrane. No data about catalyst productivity was provided.

In comparison with membrane reactors, chemical looping approach can be applied at low temperatures as typically surface or near surface lattice oxygen is consumed for propane oxidation. As lattice oxygen diffusion is suppressed under such conditions, catalysts treated in propane must be reoxidized after a few minutes on stream. Several studies elucidated the application potential of such approach with V-based,^{151–156,356–359} Pb-based,³⁶⁰ Ce-based,³⁶¹ Co-based²³⁴ or Ni-based catalysts.^{227,362–364} As seen in Fig. 20, an attractive performance, i.e. propene selectivity above 80% at propane conversion above 60%, can be achieved when using a V-based catalyst.

It is, however, worth mentioning that the published studies on this topic do not always exactly report how product selectivity was calculated. For example, no reports about formation of carbon oxides upon catalyst reoxidation are provided. It is well known that reduced V-based catalysts can form coke. As this product was formed from fed propane it must also be taken into consideration. Thus, it is important to determine overall amounts of propane consumed and products formed both in propane and air cycles through integration of the corresponding temporal profiles. On this basis, the selectivity to different products should be calculated. In some papers, it is not reported if such approach was used. If the selectivity values are determined after a certain reaction time or without considering coke formation, they may be overestimated. For example, based on propane converted into gaseous carbon-containing products only, Rostom and de Lasa¹⁵³ obtained 93% propene selectivity at 25% propane conversion. This high selectivity value drops to 85% when considering coke formation.

A recent study¹⁵⁶ should be especially mentioned as it provides some mechanistic insights into the factors governing product selectivity. Those authors investigated MoVO_x binary materials. In agreement with various previous studies, lattice oxygen is responsible for the oxidative propane dehydrogenation to propene. Importantly, the binding energy of V–O bonds increased upon addition of Mo promoter. Consequently, non-selective propane conversion into carbon oxides could be slightly hindered. The most selective MoVO_x (V/Mo = 6) catalysts resulted in 89% propene selectivity at 36% propane conversion at 500 °C. This catalyst showed stable operation over 100 dehydrogenation/reoxidation cycles. Unfortunately, it is not clear if coke formation was considered for calculating the selectivity.

Despite the above approaches are attractive they suffer from low productivity, which is limited by the ability of the materials to provide their lattice oxygen. Thus, novel materials with high oxygen-storage capacity and oxygen mobility are needed to ensure high productivity.

7.5 Alkane dehydrogenation with selective hydrogen combustion

Selective hydrogen combustion (SHC) was proposed to be used in combination with non-oxidative alkane dehydrogenation in order to increase olefin yield.^365,366 Since the non-oxidative dehydrogenation reaction is endothermic and requires a lot of energy, combusting a part of hydrogen obtained during dehydrogenation process produces the required energy making the overall process heat-balanced or even exothermic. Moreover, consumption of hydrogen in SHC reaction shifts the reaction equilibrium towards the desired olefin. As a result, higher olefin yields can be obtained. The SHC catalysts typically described in literature consist of oxides of metals of group IIIA, IVA and VA (Bi, In, Pb, Sb, etc.), mixed oxides based on Mo, W, Mn as well as Pt-containing materials.^{365,367–369}

There are two operation modes for the performing DH reaction coupled with SHC: (i) sequential co-fed DH→SHC→DH mode and (ii) redox DH + SHC mode.^365,366 The first configuration is continuous and implies spatial separation of DH and SHC catalysts. The DH and SHC catalysts are arranged in series, high oxidation state of SHC catalyst is achieved by continuous co-feeding of oxygen or air into the SHC reactor. Hazard arising from mixing oxygen with hydrogen and hydrocarbons at high temperature as well as low selectivity and stability of hydrogen combustion catalysts are the main disadvantages of such mode. In the second mode, the DH and SHC catalysts are physically mixed. Alkane dehydrogenation and hydrogen combustion proceeds in the same reactor in the absence of gaseous oxygen. Hydrogen combustion occurs with participation of lattice oxygen of the SHC catalyst. Regeneration of the catalysts is performed in oxygen or air flow in the oxidation step. Dehydrogenation and oxidation steps are separated by purge cycles with inert gas to avoid contact of hydrocarbons and hydrogen with oxygen.³⁶⁷

8 Synthesis and characterization of supported and bulk catalysts

As demonstrated in Sections 2–7, the performance (activity, selectivity and on-stream stability) of differently composed catalysts in both non-oxidative, oxidative and CO₂-mediated dehydrogenation reactions is strongly influenced by their various physicochemical properties and the structure of supported metal oxide species or surface defects (Fig. 35). All these catalyst characteristics are affected by preparation method, loading of active species, the kind of support, the presence and the kind of promoter as well as catalyst treatment conditions. All these factors could have different influence on the properties of different catalysts. Nevertheless, some general features could be identified within one preparation method. Below we will discuss the most common methods for the preparation of supported and bulk catalysts with the purpose to highlight their applicability for controlling specific solid material properties.

Fig. 35 Key factors and properties affecting catalyst performance as well as characterization methods for analysing such properties. The most common methods applied for analysing the structure of supported species and for detection of defect sites are written in red and will be discussed in detail in Sections 8.3 and 8.4.

8.1 Methods for preparation of supported catalysts

Impregnation is one of the easiest and frequently used preparation methods.³⁷⁰ It implies adding the solution (typically aqueous solution) of metal precursor(s) to a support material followed by drying (and calcination if necessary). Two different impregnation strategies are applied, i.e. (i) wet impregnation performed with an excess of a solvent and (ii) incipient wetness impregnation (or dry impregnation), which implies the usage of a certain amount of solvent required only to fill the pore volume of a support material. Although each impregnation method is simple and produces low waste amounts, it offers only little control over the structure of supported metal oxide species and their dispersion. The structure of the final catalyst can be tuned upon adjusting the pH and ionic strength.³⁷¹

Deposition–precipitation (DP) method was initially developed for the synthesis of catalysts with high loading of active components which is not possible to achieve via the impregnation method due to the limited solubility of precursor components.^370,372 The DP method ensures the formation of metal oxide compound with low solubility (often metal hydroxides) on the surface of a support material from a solution of precursor usually by a change of pH. The surface of the support acts as a seed for the nucleation. Typically, this method cannot guarantee uniform size distribution of deposited metal oxide species. However, catalysts with highly dispersed metal oxide species can be prepared through varying conditions of the synthesis. It is also worth mentioning that the interaction between the precursor and the support plays a crucial role in the production of the final catalyst. Typically, when the interaction is too weak, large particles are formed, while too strong interacting can result in the formation of metal–support phases.³⁷⁰

Chemical vapor deposition (CVD) method is carried out by reaction/decomposition of a gaseous metal precursor stream onto the support.³⁷¹ An obvious advantage of this method in comparison with the impregnation and DP methods is the ability to control coverage, nucleation rate and particle size. Usually, the catalysts prepared by CVD method are characterized by higher dispersion and better catalytic activity, however, particle size is not always uniform.

Organometallic grafting is applied to produce catalysts possessing well-defined supported species with a certain structure and specific, tailored surface coordination.^371,373,374 It is a molecular-level technique which implies the reaction of organometallic precursors with the surface of a support material followed by removal of the organic ligands. So prepared catalysts are often used for investigating the correlation between the supported structures and spectroscopic characteristics as well as for studying reaction mechanisms and understanding structure–performance relationships. This method does not seem to be suitable for large-scale applications.

Atomic layer deposition method (ALD) is a method, where two self-limiting reactions occur sequentially between gas precursors and a substrate.^371,375,376 Typically, two types of gaseous precursors are pulsed alternately into the preparation chamber. In each new pulse, precursor molecules react with the surface functional groups generated in a previous pulse. Unreacted precursor and gaseous by-products are removed by purging with an inert gas. At the end of each cycle, functional groups are generated on the surface of support material. They can again react with the gas-phase precursor. Thus, the desired precursor is deposited in a layer-by-layer fashion. By varying the number and type of ALD cycles, desired loading and surface structure can be obtained. The method allows the production of highly dispersed catalysts with uniform structure of supported species with near atomic precision. The usage of ALD method is, however, limited by the availability of suitable gaseous precursors, which can react with support functional groups but are thermally stable at growth temperature. Moreover, the limited size of conventional ALD chamber does not allow the production of high catalyst amount. For addressing this question, spatial ALD with continuous operation and/or ALD based on fluidized bed reactor have been proposed.^377,378

In summary, Table 9 was prepared and provides a brief description of each of the above-mentioned methods including their advantages and drawbacks.

Table 9 Synthesis of supported catalysts: preparation methods, their advantages and drawbacks

	Impregnation	Deposition–precipitation	Chemical vapor deposition	Organometallic grafting	Atomic layer deposition (ALD)
Description	Impregnation of a support material with a solution of active component precursor	Catalyst precursor is deposited on the surface of support material by precipitation	Gaseous precursor is decomposed on the surface of support material	Molecular organometallic precursor reacts with the surface of support material followed by removal of the organic ligands	Deposition of gaseous precursor on the surface of support material in a layer-by-layer fashion
Advantages	Simple execution, low waste streams	Simple execution, allows to produce highly loaded catalysts	Ability to control coverage, nucleation rate and particle size; usually high dispersion of supported species	Well-defined isolated surface species with uniform structure	Ability to control loading and surface structures resistant to aggregation. High dispersion
Disadvantages	Little control of the kind of surface species and their dispersion	Usually nonuniform distribution of particle size and structures	Special equipment is needed; particle size is not always uniform	Expensive, difficult to execute	Expensive; limited by availability of suitable precursors; high amount of catalyst is difficult to be produced via conventional ALD

---

8.2 Methods for preparation of bulk catalysts

Precipitation is one of the most frequently applied methods to prepare bulk catalysts and supports.^379,380 It can be used for the preparation of either single metal oxide or mixed metal oxide catalysts. Usually hydroxides or carbonates are the most preferable substances to be precipitated due to their low solubility, simplicity of their decomposition into oxides and minimal toxicity.³⁷² Depending on precipitation conditions, it is possible to vary crystallinity and crystallite size of the resulting precipitant. The most significant factors that affect the properties of the precipitant include pH, residence time, temperature, agitation, nature and concentration of precursors as well as the usage of various additives.³⁸⁰ Usually, high concentration of precursors, low temperature and short ageing time lead to the formation of finely crystalline or amorphous substances, which are difficult to filtrate. Low concentration of precursors, high temperature, and extended ageing result in coarse crystalline materials, which are easier to filtrate. The formation of large amounts of salt-containing waste solutions is the main drawback of this preparation method.

Sol–gel method implies the formation of sol (liquid suspension of solid particles smaller than 1 μm) followed by its transformation into gel.^380–382 The sol is formed as a result of hydrolysis of an inorganic salt or a metal alkoxide. Further polymerization through condensation of hydroxyl and/or alkoxy groups results in the formation of gel. It should be mentioned that such method allows control of the texture, composition, homogeneity and structural properties of the final materials. It offers better control over surface area, pore volume and pore size distribution in comparison with the precipitation method.³⁷² Generally, if the initial gel contains polymeric branched and cross-linked chains, it has large void regions and is structurally rigid.³⁸¹ Macroporous and mesoporous oxides are usually formed after calcination of such gel. The gel with a low amount of branched and cross-linked chains has smaller void regions and is structurally weak. Such gel collapses readily during calcination and results in the formation of microporous oxide with low surface area. pH value during gel formation and ageing of gel strongly influence structural properties of the final metal oxide. The rate of condensation process at low pH values is slow in comparison with the hydrolysis, that results in the formation of weakly cross-linked gel and final metal oxide with low surface area. The rate increases with rising pH. This results in the formation of high amount of cross-linkages in the gel and in an increase in pore volume and surface area of resulting metal oxides. In most cases, ageing also enhances cross-linking of the gel.

Hydrothermal method is a technology for crystallizing materials from an aqueous solution in an autoclave above 100 °C.^383,384 The procedure includes three basic steps: achievement of supersaturation, nucleation and crystal growth.³⁸⁰ The properties of the resulting material are affected by pH, temperature, pressure, time and concentration of precursors.³⁷²

Templating method implies the usage of a porous polymeric matrix as a template.^385,386 The preparation includes impregnation of a template with an aqueous solution of metal-oxide precursor, followed by its combustion. The pore structure of the resulting oxide is strongly influenced by template, concentration of precursor, heating rate, and combustion temperature. Typically, so prepared oxides are characterized by high surface area, presence of mesopores and good thermal stability.

8.3 Characterization of supported species

Characterization of supported species participating in catalytic reaction is of a great importance for identifying the kind of active sites, elucidating the fundamentals for their formation and transformation as well as for understanding reaction mechanism and structure–activity–selectivity relationships. The most common methods applied for analysing the structure of supported species are X-ray absorption spectroscopy (XAS), electronic paramagnetic resonance (EPR), UV-vis spectroscopy, IR spectroscopy, Raman spectroscopy and NMR. A brief description of the methods is given below.

XAS is a powerful tool for determining the local geometric and electronic structure of a matter on an atomic scale.^387,388 The method is element specific and does not require a long range order. An XAS spectrum can be divided into two regions: X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS). XANES data are related to the electronic transitions from core electronic states of the metal to the higher energy excited electronic states (lowest unoccupied molecular orbital). Thus, the information about oxidation state and geometry of the metal site can be obtained. As EXAFS signal is observed because of an interaction of the outgoing photoelectrons and backscattered electrons from the near neighbour atoms, the numbers and types of neighbouring atoms as well as interatomic distances can be determined. Importantly, XAS spectra can be collected under in situ or even operando conditions.

EPR is a technique for detection of paramagnetic species with unpaired electrons and analysing their coordination.³⁸⁹ The concentration of such species can be determined by comparing their EPR signal with that of a standard with known spin concentration. In situ or operando EPR allows investigation of the changes occurring with the species upon their treatment in various atmospheres at different temperatures as well as kinetics of such transformations.³⁹⁰

UV-vis spectroscopy is widely applied for analysing oxidation state of metals and molecular structure of MO_x (M is a metal) species.^391,392 However, some UV-vis bands can overlap with each other and other can be too weak, therefore, distinguishing between multiple species can be complicated.³⁹³ The UV-vis edge energy (E_g) can be calculated from the spectra to estimate oligomerization degree of MO_x species: the higher the E_g value is, the higher is the dispersion of the species. The UV-vis technique can be applied under in situ and operando conditions.

IR spectroscopy provides molecular vibrational information and is more sensitive to asymmetric vibrations. It allows the detection of anchoring OH sites for MO_x species on the surface of support and provides some information about molecular structure of MO_x species. The method can be used in situ or operando for deriving insights into surface intermediates and reaction products formed on the surface of catalysts tested.³⁹³ The usage of probe molecules in IR spectroscopy is a powerful tool to investigate the surface properties of the catalysts (acidity, basicity, dispersion of supported component, structure of supported species, etc.) as well as the nature of adsorbates under reaction conditions.^394,395 Some probe molecules frequently applied in IR studies include pyridine, NH₃, CO₂, CO, H₂, NO, C₂H₄. It should be mentioned that probe molecules must satisfy several requirements: they should be detectable by IR, have relatively small size (with some exceptions) and high extinction coefficients, the probe–surface interaction should be ideally weak. Moreover, the selection of a probe molecule is based on assigned task and the features of the sample under the study. Some specific examples of the usage of probe molecules in IR spectroscopy can be found in ref. 394 and 395.

Similar to IR, Raman spectroscopy also provides molecular vibrational information, but is more sensitive to symmetric vibrations. It is typically applied for determining molecular structure of supported MO_x species and their transformations under different conditions.^393,396,397

Solid-state NMR is used to study the substances possessing nuclei with a non-zero nuclear magnetic moment.³⁹⁸ The method is often combined with DFT calculations and gives information about the coordination and chemical environment of specific species.³⁹⁹

In summary, the described methods have been often used for the characterization of supported MO_x species. However, no single technique is fully enough for obtaining a complete picture about the nature of supported MO_x species. Each method has its own requirements and limitations. Moreover, the interpretation of the results obtained with most of these methods is based on the comparison of the obtained data with those of reference materials of known structure. The divergence of the measured parameter from that of a standard can be interpreted in a different way leading to different conclusions. Accordingly, only the usage of multiple characterization techniques can give true insights into the structure of MO_x species.

8.4 Methods for detection of surface defects of bulk catalysts

In general, various defects on the surface of metal oxide catalysts have higher reactivity in comparison with non-defective sites and thus play an important role in heterogeneous catalysts.^400–402 For the PDH (Section 2.3) and BDH (Section 3.2) reactions, coordinatively unsaturated metal cations were concluded to be the active sites on the surface of alternative-type bulk catalysts (Fig. 4). They are located near to oxygen vacancy. Thus, for further development of such catalysts, it is important to identify and quantify such defects. Several characterization methods can be applied for such purposes. They are pulse titration tests with probe molecules, e.g. O₂ or N₂O, ionic conductivity measurements, NH₃-TPD, CO-TPD, UV-vis spectroscopy, HAADF-STEM, XPS and EPR. Some of them can provide quantitative information, while others can only identify the presence of such defect sites. Below we provide a brief description of each method.

Titration pulse experiments with an oxidant are based on the phenomenon that anion vacancies can react with the pulsed probe molecule. O₂ or N₂O are typically used for such purposes. The experiments can be carried out with any set-up enabling to pulse a known amount of such molecules and their quantification at the reactor outlet. When the number of anion vacancies is very low, that is typically the case, the temporal analysis of products (TAP) reactor is an ideal equipment for such purposes.^403–405 The lowest pulse size of this technique is about 0.2 nmol. Regardless of the used technique, an on-line mass spectrometer is typically applied for detection of pulsed gases. A certain number of pulses is sequentially introduced until all oxygen vacancies are filled due to their reaction with the pulsed oxidant. This situation is realized when the number of pulsed molecules is equal to the number of these molecules at the reactor outlet. Some precautions should be considered when performing such tests. Reaction temperature should be identified, at which surface vacancies can react with pulsed probe molecules, while diffusion of surface lattice oxygen into metal oxide bulk is hindered. When N₂O is used as a probe molecule, it is important to check if oxygen species formed from N₂O can also decompose N₂O. The latter process must be avoided. The O₂-pulse titration can be considered as semiquantitative method for the detection of active sites in ZrO₂-based catalysts as it is difficult to suppress lattice oxygen diffusion.⁹³

Ionic (lattice oxygen) conductivity measurements of metal oxides can also provide information about the presence of oxygen vacancies in these materials. Such type of conductivity correlates with the number of oxygen vacancies.^26,96 The higher the conductivity, the more oxygen vacancies are present in the sample. This method is, however, not quantitative. Typically, the measurements of overall electrical conductivity of oxides are performed with two inert electrodes upon variation of partial pressure of oxygen. The conductivity of a pure ionic conductor does not depend on the pressure, while the conductivity of n- or p-conductors decreases or increases. A more detailed description of such experiments and the main concepts related to the electronic properties of metal oxides are provided in ref. 406 and 407.

The number of coordinatively unsaturated metal cations can be determined from temperature-programmed desorption tests using probe molecules selectively adsorbing on such sites. It was demonstrated that the formation of Lewis sites in bare alumina is mainly due to the defective Al_cus³⁺.^28,134 Such sites can bind basic probe molecules such as pyridine and NH₃.^28,94 Accordingly, the amount of Al_cus³⁺ can be determined by calculating the amount of desorbed NH₃ from the surface of pre-treated alumina during NH₃-TPD experiment. It should be however mentioned that the presence of Brønsted acidic sites in the sample results in overestimating the amount of Al_cus³⁺ determined by NH₃-TPD. To ensure the absence of Brønsted acidic sites a Pyridine-adsorbed Fourier Transform Infrared Spectroscopy (Py-FTIR) can be applied. Such technique distinguishes Brønsted (the band at around 1540 cm⁻¹) and Lewis (the band at around 1450 cm⁻¹) acidic sites. Similarly, CO-TPD technique was successfully applied for the titration of Zr_cus⁴⁺ sites in ZrO₂-based catalysts.^26,99 Being a Lewis acidic site, Zr_cus⁴⁺ adsorbs CO molecule. The amount of desorbed CO should be equal to the amount of Zr_cus⁴⁺ sites.

High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) can provide limited information about the amount and distribution of oxygen vacancies.²⁶ The X-ray photoelectron spectroscopy (XPS) and electron paramagnetic resonance (EPR) provide information about the presence of oxygen vacancies.²⁷ The XPS spectra of samples possessing oxygen vacancies are usually characterized by a shift in the binding energy of O 1s peak and/or of metal cation, while the EPR spectra of such samples possess the corresponding signal related to the oxygen vacancies (g = 2.001–2.005).

8.5 General experimental requirements

After analysing the literature in the preceding sections, we have noticed that sometimes not all required information is provided in the published articles for a proper comparison of different catalysts. For example, degree of alkane conversion, contact time, WHSV or GHSV are not provided. As a consequence, it is impossible to estimate catalyst activity when such values are not reported. In some studies, WHSV and GHSV are mixed up.

In addition to some missing information, conditions applied for catalytic tests are not industrially relevant and so-obtained data may result in wrong conclusions. Particularly for the non-oxidative or CO₂-mediated dehydrogenation of propane and isobutane, a lot of studies have been carried out under conditions with very diluted (10 vol% of alkane and lower) reaction feeds. As the equilibrium conversion strongly depends on alkane partial pressures (Fig. 3), this catalyst characteristics cannot be used for catalyst comparison. Such conditions are also industrially irrelevant and not favourable for coke formation. Thus, catalysts tested with diluted feeds may artificially show higher selectivity in comparison to those tested with alkane-rich feeds. Moreover, many researchers calculate product selectivity on the basis of gas-phase components without considering coke formation. Such calculation way overestimates the selectivity values. Against this background, we would like to highlight some general recommendations for investigating the non-oxidative or CO₂-mediated dehydrogenation of propane and isobutane as well as for evaluation of experimental data. Some experimental requirements for carrying out the oxidative propane dehydrogenation to propene are published in a review of Carrero et al.¹² Our recommendations are as follows:

i. Ensuring that the tests are carried out under conditions free of any mass-transport limitations and strong contributions of gas-phase reactions. The latter play a role at high temperatures and long contact times.

ii. When determining reaction rates and other parameters related to the rate (e.g. turn over frequency or activation energy), the degree of alkane conversion should be below 10% to ensure a pseudo differential reactor operation regime.

iii. For demonstrating application potential of the developed catalysts, it is highly recommended to use industrially relevant feeds (alkane partial pressure higher than 0.3 bar). Comparison with the state-of-the-art catalysts in terms of their selectivity and productivity (space time yield) should be made under similar experimental conditions and degrees of alkane conversion, which should be at least larger than 20%.

iv. Coke cannot be ignored as a reaction product when alkane conversion is higher than 10%. Otherwise, the selectivity to gas-phase products is overestimated. Thus, carbon balance is highly important for ensuring precision in the obtained data. Eqn (3) can be used for calculating coke selectivity.

v. For analysing coke formation, diluted feeds (alkane partial pressure below 0.3 bar) should be avoided. As olefin is the main precursor of coke, which is actually a minor product, high olefin partial pressures are required for precise determining the selectivity to this side product.

vi. Reaction induced changes in the number of moles (two molecules, i.e. H₂ and C₃H₆, are formed from one C₃H₈) must be taken into account when particularly operating with alkane-rich feeds. The changes can be easily considered when using reaction feeds with an inert standard, e.g. N₂. Alkane conversion and selectivity to gas-phase products and coke can be calculated as following:

where “feed” or “inert” represent alkane or inert gas, e.g. N₂, used an internal standard, S_i is the selectivity to a gas-phase product i, β stands for the number of C atoms in the product i or in the fed alkane, and x with superscripts “inlet” and “outlet” stands for the concentration of gas-phase components i or the fed alkane at the reactor inlet and outlet, respectively.

9 Conclusions

The present review has provided a critical analysis of the current status on the conversion of propane and isobutane into the corresponding olefins through non-oxidative, oxidative and CO₂-mediated routes. A particular focus was put on comparing the developed catalysts in terms of their application potential. Catalyst design concepts and mechanistic aspects of product formation including the kind of active sites and ways for their purposeful creation have also been thoroughly reviewed with the purpose to identify fundamentals required for tailored catalyst design. Some reaction engineering concepts have been briefly discussed.

From a mechanistic viewpoint, some general similarities, and differences between non-oxidative, oxidative, and CO₂-mediated dehydrogenation of propane/isobutane have been identified. For catalysts with supported metal oxide species, it is undisputable that all these reactions are initiated through breaking strong C–H bonds in these alkanes with participation of M-O²⁻ pairs (H₂ is formed and the metal does not change its oxidation state) or lattice/adsorbed oxygen (H₂O is formed, the metal changes its oxidation state). The structure of supported MO_x species and the kind of support material are decisive for catalyst activity and olefin selectivity. Although there are still some controversies, isolated metal oxide species seem to reveal higher selectivity, activity, and on-stream stability in all these dehydrogenation reactions in comparison with polymerized species. According to DFT calculations, the structure and consequently the reactivity of MO_x species depend on the face of support material, where they are located. The knowledge about the exact structure of supported species and their catalytic performance provides some hints for catalyst design.

In comparison with supported catalysts, recently introduced bulk catalysts on the basis of non-reducible metal oxides, such as ZrO₂, TiO₂, Al₂O₃, Gd₂O₃ and Eu₂O₃ have other active sites. Alkane activation occurs with participation of coordinatively unsaturated metal cations at anion vacancies. For TiO₂ and Al₂O₃, the corresponding cation and one neighbouring lattice oxygen participate in breaking C–H bonds in propane. Owing to combining experimental studies with DFT calculations, an alternative kind of active sites was discovered for ZrO₂-based catalysts; two coordinatively unsaturated Zr cations form the active site. Fundamentals affecting intrinsic activity of coordinatively unsaturated metal cations and ways for their creation have been elucidated and build the basis for purposeful catalyst preparation.

On the basis of literature analysis in Fig. 13 and 16, we can safely assert that there is a significant progress in the development of catalysts for the PDH and BDH reactions. Both industrially relevant olefin selectivity and productivity were achieved using different catalysts. For some catalyst systems, long-term stability performance has been validated over several dehydrogenation/regeneration cycles lasting in total for a few days. Among the developed catalysts, those based on bulk ZrO₂ should be especially mentioned. In comparison with V- or Cr-containing catalysts often used in these dehydrogenation reactions, these novel catalysts are environmentally friendly. Moreover, they have been introduced for the first time just five years ago. Thus, further improvements in the performance of such catalysts are expected.

Despite continuous studies on the ODP reaction using stoichiometric mixtures of propane and oxygen, no visible progress in the development of catalysts with propene selectivity above 80% at propane conversion higher than 30% has been achieved over the last 20 years. The low selectivity to propene is related to the interplay between the kinetics of selective (formation of propene) and non-selective consecutive reactions of the target olefin to carbon oxides. In comparison with the ODP reaction, some progress has been achieved in the ODB reaction. The isobutene yield values reported are comparable with those typical for BDH. Nevertheless, the necessity to separate air for obtaining pure oxygen is a negative cost factor limiting the commercialization of ODP and ODB. This drawback can be overcome and the undesired oxidation of propene/isobutene to carbon oxides can be partially hindered, when alkane dehydrogenation is carried out over reducible metal oxides using alternating feeds containing alkane or air. Such process operation, however, suffers from the limited ability of the catalysts to provide their lattice oxygen for the dehydrogenation reaction and from the necessity to interrupt the desired reaction after a relatively short period of time for catalyst reoxidation. Carrying out the ODP/ODB reactions under O₂-lean (strong excess of alkane over oxygen) seems to be an attractive approach from cost (lower amounts of expensive oxygen are required), productivity (no often process interruption for catalyst regeneration), on-stream stability and selectivity viewpoints. Under such conditions, the oxidative dehydrogenation occurs in the upstream part of catalyst bed till complete consumption of oxygen and provides heat for the non-oxidative reaction taking place in the downstream catalyst bed.

Generally, materials with supported metal oxides used for the non-oxidative and oxidative dehydrogenation reactions can also catalyse the CO₂-assisted alkane dehydrogenation. Studies on the latter subject are also focused on understanding the role of the kind of supported metal oxide species and support material for product selectivity, catalyst activity and on-stream stability. The derived conclusions are, however, controversial. In terms of application potential, some promising results have been obtained in terms of olefin yield.

10 Outlook

Our personal view on possible developments related to improving catalysts performance in the dehydrogenation of propane/isobutane to the corresponding olefins is provided below. To overcome the main drawback (low selectivity at industrially relevant degrees of alkane conversion) of the ODP and ODB reactions through catalyst design, new preparation methods enabling creation of certain structures of supported metal oxides on specific faces of support materials are required. So-designed catalysts should be tested for product selectivity in a broad range of alkane conversion to understand their ability/activity for consecutive oxidation of the target olefins. On this basis, common structure–selectivity–catalyst property relationships can be established. Such experimental approach should be supported by DFT calculations. The latter should be focused on understanding factors affecting the undesired oxidation reactions. We expect, however, that the most potential improvements can be achieved through reaction engineering aspects (Sections 7.3 and 7.4). O₂-Lean or O₂-free (chemical looping, redox process) dehydrogenation seems to be the most promising options. For the latter mode, catalysts with high capacity to release high amounts of lattice oxygen and to quickly refill over several thousand dehydrogenation/regeneration cycles have to be developed. Oxygen-conducting membranes can also be applied for carrying out ODP/ODB with air without or with an additional catalyst. They should, however, ensure high enough flux of lattice oxygen to meet industrial productivity requirements.

Although there are various catalytic materials showing industrially relevant performance with respect to activity and olefin selectivity in the PDH and BDH reactions, their industrial potential has not been validated in long-term tests. In addition, tests with reaction feeds representative for the commercial products are scarcely reported. Concerning development of novel materials, bulk catalysts based on oxides of non-reducible metal oxides seem to be attractive, as only a few such materials have been prepared and tested up to now. Their performance can be tuned when controlling the coordination number of surface cations, their location on certain facets and/or the shape of crystallites. Regardless of catalyst type, further deeper understanding of catalyst property–selectivity–activity relationships is still required. With this respect, our knowledge about the mechanism of coke formation on a molecular level and factors affecting this process is really superficial, as a major part of papers dealing with catalyst development is focused on mechanistic and kinetic aspects of alkane activation and olefin formation. DFT calculations related to coke formation, which are missing in open literature up to now, could support catalyst development. The above recommendations are also valid for CO₂-mediated alkane dehydrogenation.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

Financial support by Deutsche Forschungsgemeinschaft (KO 2261/8-1), the National Natural Science Foundation of China (Grant No. 21961132026, 21878331, 91645108), Science Foundation of China University of Petroleum, Beijing (C201604), Ministry of Science and Technology of PRC (National Key Research and Development Program Nanotechnology Specific Project (No. 2020YFA0210900)) and the State of Mecklenburg-Vorpommern is gratefully acknowledged. The authors are also thankful to A. Skrypnik, Z. Aydin, N. Ortner, A. Perechodjuk, Q. Yang, K. Wu, Y. M. Li and D. Zhao for assistance with analysing literature data.

References

1. J. Hagen, in Industrial Catalysis, ed. J. Hagen, Wiley-VCH Verlag GmbH & Co. KGaA, 2015, pp. 261–298 DOI:10.1002/3527607684.ch8.
2. J. Hagen, Industrial Catalysis, Wiley-VCH Verlag GmbH & Co. KGaA, 2015, pp. 47–80 DOI:10.1002/3527607684.ch3.
3. F. J. Keil, Microporous Mesoporous Mater., 1999, 29, 49–66.
4. P. Tian, Y. X. Wei, M. Ye and Z. M. Liu, ACS Catal., 2015, 5, 1922–1938.
5. I. Yarulina, A. D. Chowdhury, F. Meirer, B. M. Weckhuysen and J. Gascon, Nat. Catal., 2018, 1, 398–411.
6. https://www.engineering-airliquide.com/de/lurgi-mtp-methanol-zu-propylen, Lurgi MTP™ – Methanol-zu-Propylen.
7. J. C. Mol and P. W. N. M. van Leeuwen, in Handbook of Heterogeneous Catalysis, ed. G. Ertl, H. Knözinger, F. Schüth and J. Weitkamp, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, 2008, vol. 6, pp. 3420–3456.
8. S. Lwin and I. E. Wachs, ACS Catal., 2014, 4, 2505–2520.
9. S. Wang and Z. H. Zhu, Energy Fuels, 2004, 18, 1126–1139.
10. F. Cavani, N. Ballarini and A. Cericola, Catal. Today, 2007, 127, 113–131.
11. M. A. Atanga, F. Rezaei, A. Jawad, M. Fitch and A. A. Rownaghi, Appl. Catal., B, 2018, 220, 429–445.
12. C. A. Carrero, R. Schloegl, I. E. Wachs and R. Schomaecker, ACS Catal., 2014, 4, 3357–3380.
13. J. T. Grant, J. M. Venegas, W. P. McDermott and I. Hermans, Chem. Rev., 2018, 118, 2769–2815.
14. C. Boswell, On-purpose Technologies ready to fill Propylene gap, http://www.icis.com/Articles/2012/04/16/9549968/on-purpose-technologies-ready-to-fill-propylene-gap.html.
15. B. V. Vora, Top. Catal., 2012, 55, 1297–1308.
16. J. C. Bricker, Top. Catal., 2012, 55, 1309–1314.
17. K. J. Caspary, H. Gehrke, M. Heinritz-Adrian and M. Schwefer, in Handbook of Heterogeneous Catalysis, ed. G. Ertl, H. Knözinger, F. Schüth and J. Weitkamp, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, 2008, ch. 14.6, vol. 7, pp. 3206–3228.
18. J. J. H. B. Sattler, J. Ruiz-Martinez, E. Santillan-Jimenez and B. M. Weckhuysen, Chem. Rev., 2014, 114, 10613–10653.
19. https://www.osha.gov/SLTC/hexavalentchromium/index.html, accessed October 20, 2020.
20. G. Wang, X. Zhu and C. Li, Chem. Rec., 2020, 20, 604–616.
21. Z.-P. Hu, D. Yang, Z. Wang and Z.-Y. Yuan, Chin. J. Catal., 2019, 40, 1233–1254.
22. C. Sim-Yee, A. Hisyam and H. Prasetiawan, Int. J. Chem. React. Eng., 2015, 14.
23. K. J. Caspary, H. Gehrke, M. Heinritz-Adrian and M. Schwefer, Handbook of Heterogeneous Catalysis, Wiley-VCH Verlag GmbH & Co. KGaA, 2008 DOI:10.1002/9783527610044.hetcat0162.
24. Z. Nawaz, Rev. Chem. Eng., 2015, 31, 413–436.
25. T. Otroshchenko, S. Sokolov, M. Stoyanova, V. A. Kondratenko, U. Rodemerck, D. Linke and E. V. Kondratenko, Angew. Chem., Int. Ed., 2015, 54, 15880–15883.
26. Y. Zhang, Y. Zhao, T. Otroshchenko, H. Lund, M.-M. Pohl, U. Rodemerck, D. Linke, H. Jiao, G. Jiang and E. V. Kondratenko, Nat. Commun., 2018, 9.
27. C.-F. Li, X. Guo, Q.-H. Shang, X. Yan, C. Ren, W.-Z. Lang and Y.-J. Guo, Ind. Eng. Chem. Res., 2020, 59, 4377–4387.
28. H. Zhu, P. Wang, Z. Xu, T. Wang, Y. Yue and X. Bao, Catal. Sci. Technol., 2020, 10, 3537–3541.
29. A. Perechodjuk, V. A. Kondratenko, H. Lund, N. Rockstroh and E. V. Kondratenko, Chem. Commun., 2020, 56, 13021–13024.
30. P.-P. Li, W.-Z. Lang, K. Xia, L. Luan, X. Yan and Y.-J. Guo, Appl. Catal., A, 2016, 522, 172–179.
31. M. F. Delley, M.-C. Silaghi, F. Nunez-Zarur, K. V. Kovtunov, O. G. Salnikov, D. P. Estes, I. V. Koptyug, A. Comas-Vives and C. Coperet, Organometallics, 2017, 36, 234–244.
32. A. Wegrzyniak, S. Jarczewski, A. Wegrzynowicz, B. Michorczyk, P. Kustrowski and P. Michorczyk, Nanomaterials, 2017, 7.
33. A. Węgrzyniak, A. Rokicińska, E. Hędrzak, B. Michorczyk, K. Zeńczak-Tomera, P. Kuśtrowski and P. Michorczyk, Catal. Sci. Technol., 2017, 7, 6059–6068.
34. G. Liu, Z.-J. Zhao, T. Wu, L. Zeng and J. Gong, ACS Catal., 2016, 6, 5207–5214.
35. V. V. Kaichev, Y. A. Chesalov, A. A. Saraev and A. M. Tsapina, J. Phys. Chem. C, 2019, 123, 19668–19680.
36. U. Rodemerck, M. Stoyanova, E. V. Kondratenko and D. Linke, J. Catal., 2017, 352, 256–263.
37. C. Chen, M. Sun, Z. Hu, Y. Liu, S. Zhang and Z.-Y. Yuan, Chin. J. Catal., 2020, 41, 276–285.
38. P. Bai, Z. Ma, T. Li, Y. Tian, Z. Zhang, Z. Zhong, W. Xing, P. Wu, X. Liu and Z. Yan, ACS Appl. Mater. Interfaces, 2016, 8, 25979–25990.
39. P. Hu, Y. Chen, X. Yan, W.-Z. Lang and Y.-J. Guo, Ind. Eng. Chem. Res., 2019, 58, 4065–4073.
40. P. Hu, W.-Z. Lang, X. Yan, X.-F. Chen and Y.-J. Guo, Appl. Catal., A, 2018, 553, 65–73.
41. C. Xiong, S. Chen, P. Yang, S. Zha, Z.-J. Zhao and J. Gong, ACS Catal., 2019, 9, 5816–5827.
42. G. Liu, L. Zeng, Z.-J. Zhao, H. Tian, T. Wu and J. Gong, ACS Catal., 2016, 6, 2158–2162.
43. S. Han, D. Zhao, T. Otroshchenko, H. Lund, U. Bentrup, V. A. Kondratenko, N. Rockstroh, S. Bartling, D. E. Doronkin, J.-D. Grunwaldt, U. Rodemerck, D. Linke, M. Gao, G. Jiang and E. V. Kondratenko, ACS Catal., 2020, 10, 8933–8949.
44. T. Gong, L. Qin, J. Lu and H. Feng, Phys. Chem. Chem. Phys., 2016, 18, 601–614.
45. V. J. Cybulskis, S. U. Pradhan, J. J. Lovon-Quintana, A. S. Hock, B. Hu, G. Zhang, W. N. Delgass, F. H. Ribeiro and J. T. Miller, Catal. Lett., 2017, 147, 1252–1262.
46. M. W. Schreiber, C. P. Plaisance, M. Baumgärtl, K. Reuter, A. Jentys, R. Bermejo-Deval and J. A. Lercher, J. Am. Chem. Soc., 2018, 140, 4849–4859.
47. N. M. Phadke, E. Mansoor, M. Bondil, M. Head-Gordon and A. T. Bell, J. Am. Chem. Soc., 2019, 141, 1614–1627.
48. E. Mansoor, M. Head-Gordon and A. T. Bell, ACS Catal., 2018, 8, 6146–6162.
49. G. Wang, H. Zhang, H. Wang, Q. Zhu, C. Li and H. Shan, J. Catal., 2016, 344, 606–608.
50. H. Wang, H. Huang, K. Bashir and C. Li, Appl. Catal., A, 2020, 590, 117291.
51. H. Wang, H. Wang, X. Li and C. Li, Appl. Surf. Sci., 2017, 407, 456–462.
52. G. Wang, H. Zhang, Q. Zhu, X. Zhu, X. Li, H. Wang, C. Li and H. Shan, J. Catal., 2017, 351, 90–94.
53. Q. Liu, M. Luo, Z. Zhao and Q. Zhao, Chem. Eng. J., 2020, 380, 122423.
54. X. Li, P. Wang, H. Wang and C. Li, Appl. Surf. Sci., 2018, 441, 688–693.
55. C. Chen, S. Zhang, Z. Wang and Z.-Y. Yuan, J. Catal., 2020, 383, 77–87.
56. D. P. Estes, G. Siddiqi, F. Allouche, K. V. Kovtunov, O. V. Safonova, A. L. Trigub, I. V. Koptyug and C. Coperet, J. Am. Chem. Soc., 2016, 138, 14987–14997.
57. B. Hu, W.-G. Kim, T. P. Sulmonetti, M. L. Sarazen, S. Tan, J. So, Y. Liu, R. S. Dixit, S. Nair and C. W. Jones, ChemCatChem, 2017, 9, 3330–3337.
58. Y. Zhao, H. Sohn, B. Hu, J. Niklas, O. G. Poluektov, J. Tian, M. Delferro and A. S. Hock, ACS Omega, 2018, 3, 11117–11127.
59. C.-W. Zhang, J. Wen, L. Wang, X.-G. Wang and L. Shi, New J. Chem., 2020, 44, 7450–7459.
60. J. Liu, Y. Liu, Y. Ni, H. Liu, W. Zhu and Z. Liu, Catal. Sci. Technol., 2020, 10, 1739–1746.
61. Y. Zhang, S. Yang, J. Lu, Y. Mei, D. He and Y. Luo, Ind. Eng. Chem. Res., 2019, 58, 19818–19824.
62. Y. Gu, H. Liu, M. Yang, Z. Ma, L. Zhao, W. Xing, P. Wu, X. Liu, S. Mintova, P. Bai and Z. Yan, Appl. Catal., B, 2020, 274, 119089.
63. T. Wu, G. Liu, L. Zeng, G. Sun, S. Chen, R. Mu, S. Agbotse Gbonfoun, Z.-J. Zhao and J. Gong, AIChE J., 2017, 63, 4911–4919.
64. O. O. James, S. Mandal, N. Alele, B. Chowdhury and S. Maity, Fuel Process. Technol., 2016, 149, 239–255.
65. X.-Q. Gao, W.-D. Lu, S.-Z. Hu, W.-C. Li and A.-H. Lu, Chin. J. Catal., 2019, 40, 184–191.
66. Z.-P. Hu, Y. Wang, D. Yang and Z.-Y. Yuan, J. Energy Chem., 2020, 47, 225–233.
67. D. He, Y. Zhang, S. Yang, Y. Mei and Y. Luo, ChemCatChem, 2018, 10, 5434–5440.
68. T. P. Otroshchenko, U. Rodemerck, D. Linke and E. V. Kondratenko, J. Catal., 2017, 356, 197–205.
69. S. Han, Y. Zhao, T. Otroshchenko, Y. Zhang, D. Zhao, H. Lund, T. H. Vuong, J. Rabeah, U. Bentrup, V. A. Kondratenko, U. Rodemerck, D. Linke, M. Gao, H. Jiao, G. Jiang and E. V. Kondratenko, ACS Catal., 2020, 10, 1575–1590.
70. S. Han, T. Otroshchenko, D. Zhao, H. Lund, N. Rockstroh, T. H. Vuong, J. Rabeah, U. Rodemerck, D. Linke, M. Gao, G. Jiang and E. V. Kondratenko, Appl. Catal., A, 2020, 590, 117350.
71. S. Han, T. Otroshchenko, D. Zhao, H. Lund, U. Rodemerck, D. Linke, M. Gao, G. Jiang and E. V. Kondratenko, Catal. Commun., 2020, 138, 105956.
72. S. Sokolov, M. Stoyanova, U. Rodemerck, D. Linke and E. V. Kondratenko, J. Catal., 2012, 293, 67–75.
73. Z.-J. Zhao, T. Wu, C. Xiong, G. Sun, R. Mu, L. Zeng and J. Gong, Angew. Chem., Int. Ed., 2018, 57, 6791–6795.
74. P. Hu, W.-Z. Lang, X. Yan, L.-F. Chu and Y.-J. Guo, J. Catal., 2018, 358, 108–117.
75. Q. Liu, Z. Yang, M. Luo, Z. Zhao, J. Wang, Z. Xie and L. Guo, Microporous Mesoporous Mater., 2019, 282, 133–145.
76. Y. Xie, R. Luo, G. Sun, S. Chen, Z.-J. Zhao, R. Mu and J. Gong, Chem. Sci., 2020, 11, 3845–3851.
77. S. Sokolov, M. Stoyanova, U. Rodemerck, D. Linke and E. V. Kondratenko, Catal. Sci. Technol., 2014, 4, 1323–1332.
78. C. Chen, Z. Hu, J. Ren, S. Zhang, Z. Wang and Z.-Y. Yuan, ChemCatChem, 2019, 11, 868–877.
79. C. Chen, Z.-P. Hu, J.-T. Ren, S. Zhang, Z. Wang and Z.-Y. Yuan, Mol. Catal., 2019, 476, 110508.
80. D. Zhao, Y. Li, S. Han, Y. Zhang, G. Jiang, Y. Wang, K. Guo, Z. Zhao, C. Xu, R. Li, C. Yu, J. Zhang, B. Ge and E. V. Kondratenko, iScience, 2019, 13, 269–276.
81. C. Chen, M. Sun, Z. Hu, J. Ren, S. Zhang and Z.-Y. Yuan, Catal. Sci. Technol., 2019, 9, 1979–1988.
82. J. Camacho-Bunquin, P. Aich, M. Ferrandon, A. “Bean” Getsoian, U. Das, F. Dogan, L. A. Curtiss, J. T. Miller, C. L. Marshall, A. S. Hock and P. C. Stair, J. Catal., 2017, 345, 170–182.
83. S. Han, D. Zhao, H. Lund, N. Rockstroh, S. Bartling, D. E. Doronkin, J.-D. Grunwaldt, M. Gao, G. Jiang and E. V. Kondratenko, Catal. Sci. Technol., 2020, 10, 7046–7055.
84. C.-T. Shao, W.-Z. Lang, X. Yan and Y.-J. Guo, RSC Adv., 2017, 7, 4710–4723.
85. Y. Xiang, H. Wang, J. Cheng and J. Matsubu, Catal. Sci. Technol., 2018, 8, 1500–1516.
86. A. Bhan and W. Nicholas Delgass, Catal. Rev., 2008, 50, 19–151.
87. S.-W. Choi, W.-G. Kim, J.-S. So, J. S. Moore, Y. Liu, R. S. Dixit, J. G. Pendergast, C. Sievers, D. S. Sholl, S. Nair and C. W. Jones, J. Catal., 2017, 345, 113–123.
88. W.-g. Kim, J. So, S.-W. Choi, Y. Liu, R. S. Dixit, C. Sievers, D. S. Sholl, S. Nair and C. W. Jones, Chem. Mater., 2017, 29, 7213–7222.
89. M. Nakai, K. Miyake, R. Inoue, K. Ono, H. Al Jabri, Y. Hirota, Y. Uchida, S. Tanaka, M. Miyamoto, Y. Oumi, C. Y. Kong and N. Nishiyama, Catal. Sci. Technol., 2019, 9, 6234–6239.
90. Y. Dai, J. Gu, S. Tian, Y. Wu, J. Chen, F. Li, Y. Du, L. Peng, W. Ding and Y. Yang, J. Catal., 2020, 381, 482–492.
91. D. P. Estes, Chimia, 2017, 71, 177–180.
92. N. Dewangan, J. Ashok, M. Sethia, S. Das, S. Pati, H. Kus and S. Kawi, ChemCatChem, 2019, 11, 4923–4934.
93. T. Otroshchenko, V. A. Kondratenko, U. Rodemerck, D. Linke and E. V. Kondratenko, J. Catal., 2017, 348, 282–290.
94. U. Rodemerck, E. V. Kondratenko, T. Otroshchenko and D. Linke, Chem. Commun., 2016, 52, 12222–12225.
95. M. Dixit, P. Kostetskyy and G. Mpourmpakis, ACS Catal., 2018, 8, 11570–11578.
96. Y. Zhang, Y. Zhao, T. Otroshchenko, S. Han, H. Lund, U. Rodemerck, D. Linke, H. Jiao, G. Jiang and E. V. Kondratenko, J. Catal., 2019, 371, 313–324.
97. T. Otroshchenko, O. Bulavchenko, H. V. Thanh, J. Rabeah, U. Bentrup, A. Matvienko, U. Rodemerck, B. Paul, R. Kraehnert, D. Linke and E. V. Kondratenko, Appl. Catal., A, 2019, 585.
98. T. P. Otroshchenko, V. A. Kondratenko, U. Rodemerck, D. Linke and E. V. Kondratenko, Catal. Sci. Technol., 2017, 7, 4499–4510.
99. Y. Zhang, Y. Zhao, T. Otroshchenko, A. Perechodjuk, V. A. Kondratenko, S. Bartling, U. Rodemerck, D. Linke, H. Jiao, G. Jiang and E. V. Kondratenko, ACS Catal., 2020, 10, 6377–6388.
100. D. Hoang and H. Lieske, Catal. Lett., 1994, 27, 33–42.
101. J. A. Wang, T. López, X. Bokhimi and O. Novaro, J. Mol. Catal. A: Chem., 2005, 239, 249–256.
102. C. Berger-Karin, J. Radnik and E. V. Kondratenko, J. Catal., 2011, 280, 116–124.
103. A. Perechodjuk, Y. Zhang, V. A. Kondratenko, U. Rodemerck, D. Linke, S. Bartling, C. R. Kreyenschulte, G. Jiang and E. V. Kondratenko, Appl. Catal., A, 2020, 602, 117731.
104. T. Otroshchenko and E. V. Kondratenko, Catal. Commun., 2020, 144, 106068.
105. S. Sim, S. Gong, J. Bae, Y.-K. Park, J. Kim, W. C. Choi, U. G. Hong, D. S. Park, I. K. Song, H. Seo, N. Y. Kang and S. Park, Mol. Catal., 2017, 436, 164–173.
106. Y.-P. Tian, P. Bai, S.-M. Liu, X.-M. Liu and Z.-F. Yan, Fuel Process. Technol., 2016, 151, 31–39.
107. A. Rodriguez-Gomez, A. D. Chowdhury, M. Caglayan, J. A. Bau, E. Abou-Hamad and J. Gascon, Catal. Sci. Technol., 2020, 10, 6138–6150.
108. R. R. Langeslay, D. M. Kaphan, C. L. Marshall, P. C. Stair, A. P. Sattelberger and M. Delferro, Chem. Rev., 2019, 119, 2128–2191.
109. Y.-P. Tian, Y.-A. Liu, X.-M. Liu and Z.-F. Yan, Catal. Sci. Technol., 2018, 8, 5473–5481.
110. X.-s. Wang, G.-l. Zhou, Z.-w. Chen, Q. Li, H.-j. Zhou and C.-m. Xu, Appl. Catal., A, 2018, 555, 171–177.
111. G. Wang, C. Li and H. Shan, Catal. Sci. Technol., 2016, 6, 3128–3136.
112. A. N. Matveyeva, N. A. Zaitseva, P. Mäki-Arvela, A. Aho, A. K. Bachina, S. P. Fedorov, D. Y. Murzin and N. A. Pakhomov, Ind. Eng. Chem. Res., 2018, 57, 927–938.
113. A. N. Matveyeva, S. O. Omarov, D. A. Sladkovskiy and D. Y. Murzin, Chem. Eng. J., 2019, 372, 1194–1204.
114. A. N. Matveyeva, J. Wärnå, N. A. Pakhomov and D. Yu. Murzin, Chem. Eng. J., 2020, 381.
115. H. Zhao, H. Song, L. Chou, J. Zhao, J. Yang and L. Yan, Catal. Sci. Technol., 2017, 7, 3258–3267.
116. T. Otroshchenko, J. Radnik, M. Schneider, U. Rodemerck, D. Linke and E. V. Kondratenko, Chem. Commun., 2016, 52, 8164–8167.
117. Y. Liu, C. Xia, Q. Wang, L. Zhang, A. Huang, M. Ke and Z. Song, Catal. Sci. Technol., 2018, 8, 4916–4924.
118. B. Liu, H. Zhao, J. Yang, J. Zhao, L. Yan, H. Song and L. Chou, Microporous Mesoporous Mater., 2020, 293.
119. M. Cheng, H. Zhao, J. Yang, J. Zhao, L. Yan, H. Song and L. Chou, Catal. Lett., 2019, 149, 1326–1336.
120. M. Cheng, H. Zhao, J. Yang, J. Zhao, L. Yan, H. Song and L. Chou, Microporous Mesoporous Mater., 2018, 266, 117–125.
121. V. Z. Fridman, R. Xing and M. Severance, Appl. Catal., A, 2016, 523, 39–53.
122. V. Z. Fridman and R. Xing, Appl. Catal., A, 2017, 530, 154–165.
123. G. Wang, N. Song, K. Lu, W. Wang, L. Bing, Q. Zhang, H. Fu, F. Wang and D. Han, Catalysts, 2019, 9.
124. D. A. Nazimov, O. V. Klimov, A. V. Saiko, S. N. Trukhan, T. S. Glazneva, I. P. Prosvirin, S. V. Cherepanova and A. S. Noskov, Catal. Today, 2020 DOI:10.1016/j.cattod.2020.03.005.
125. A. A. Salaeva, M. A. Salaev, O. V. Vodyankina and G. V. Mamontov, Appl. Catal., A, 2019, 581, 82–90.
126. A. A. Salaeva, M. A. Salaev and G. V. Mamontov, Chem. Eng. Sci., 2020, 215.
127. T. A. Bugrova and G. V. Mamontov, Kinet. Catal., 2018, 59, 143–149.
128. A. A. Shutilov and G. A. Zenkovets, Mater. Today: Proc., 2020, 25, 483–486.
129. U. Rodemerck, S. Sokolov, M. Stoyanova, U. Bentrup, D. Linke and E. V. Kondratenko, J. Catal., 2016, 338, 174–183.
130. G. Liu, J. Liu, N. He, C. Miao, J. Wang, Q. Xin and H. Guo, RSC Adv., 2018, 8, 18663–18671.
131. M. Cheng, H. Zhao, J. Yang, J. Zhao, L. Yan, H. Song and L. Chou, RSC Adv., 2019, 9, 9828–9837.
132. J. Mu, J. Shi, L. J. France, Y. Wu, Q. Zeng, B. Liu, L. Jiang, J. Long and X. Li, ACS Appl. Mater. Interfaces, 2018, 10, 23112–23121.
133. E. Cheng, L. McCullough, H. Noh, O. Farha, J. Hupp and J. Notestein, Ind. Eng. Chem. Res., 2020, 59, 1113–1122.
134. U. Rodemerck, E. V. Kondratenko, T. Otroshchenko and D. Linke, Chem. Commun., 2016, 52, 12222–12225.
135. Y.-P. Tian, X.-M. Liu, M. J. Rood and Z.-F. Yan, Appl. Catal., A, 2017, 545, 1–9.
136. R. L. Puurunen, B. G. Beheydt and B. M. Weckhuysen, J. Catal., 2001, 204, 253–257.
137. S. M. K. Airaksinen, M. A. Bañares and A. O. I. Krause, J. Catal., 2005, 230, 507–513.
138. S. Sokolov, V. Y. Bychkov, M. Stoyanova, U. Rodemerck, U. Bentrup, D. Linke, Y. P. Tyulenin, V. N. Korchak and E. V. Kondratenko, ChemCatChem, 2015, 7, 1691–1700.
139. J. J. H. B. Sattler, A. M. Beale and B. M. Weckhuysen, Phys. Chem. Chem. Phys., 2013, 15, 12095–12103.
140. J. J. H. B. Sattler, A. M. Mens and B. M. Weckhuysen, ChemCatChem, 2014, 6, 3139–3145.
141. H.-Z. Wang, L.-L. Sun, Z.-J. Sui, Y.-A. Zhu, G.-H. Ye, D. Chen, X.-G. Zhou and W.-K. Yuan, Ind. Eng. Chem. Res., 2018, 57, 8647–8654.
142. B. R. Kumar and R. Kumar, J. Nat. Gas Chem., 2008, 17, 256–263.
143. K. Scheurell, G. Scholz, A. Pawlik and E. Kemnitz, Solid State Sci., 2008, 10, 873–883.
144. C. Oliva, S. Cappelli, I. Rossetti, N. Ballarini, F. Cavani and L. Forni, Chem. Eng. J., 2009, 154, 131–136.
145. J. Xu, M. Chen, Y. M. Liu, Y. Cao, H. Y. He and K. N. Fan, Microporous Mesoporous Mater., 2009, 118, 354–360.
146. M. D. Putra, S. M. Al-Zahrani and A. E. Abasaeed, Catal. Commun., 2011, 14, 107–110.
147. H. Zhu, S. Ould-Chikh, H. Dong, I. Llorens, Y. Saih, D. H. Anjum, J. L. Hazemann and J. M. Basset, ChemCatChem, 2015, 7, 3332–3339.
148. Q. X. Luo, X. K. Zhang, B. L. Hou, J. G. Chen, C. Zhu, Z. W. Liu, Z. T. Liu and J. Lu, Catal.: Sci. Technol., 2018, 8, 4864–4876.
149. J. Guo, X. Li, Y. Tang and J. Zhang, ChemistrySelect, 2019, 4, 13576–13581.
150. O. Ovsitser, R. Schomaecker, E. V. Kondratenko, T. Wolfram and A. Trunschke, Catal. Today, 2012, 192, 16–19.
151. S. A. Al-Ghamdi and H. I. De, Lasa, Fuel, 2014, 128, 120–140.
152. A. A. Ayandiran, I. A. Bakare, H. Binous, S. Al-Ghamdi, S. A. Razzak and M. M. Hossain, Catal.: Sci. Technol., 2016, 6, 5154–5167.
153. S. Rostom and H. I. De Lasa, Ind. Eng. Chem. Res., 2017, 56, 13109–13124.
154. M. M. Hossain, Ind. Eng. Chem. Res., 2017, 56, 4309–4318.
155. S. Rostom and H. De Lasa, Ind. Eng. Chem. Res., 2018, 57, 10251–10260.
156. S. Chen, L. Zeng, R. Mu, C. Xiong, Z. J. Zhao, C. Zhao, C. Pei, L. Peng, J. Luo, L. S. Fan and J. Gong, J. Am. Chem. Soc., 2019, 141, 18653–18657.
157. Q. Xie, H. Zhang, J. Kang, J. Cheng, Q. Zhang and Y. Wang, ACS Catal., 2018, 8, 4902–4916.
158. L. Sun, Y. Chai, W. Dai, G. Wu, N. Guan and L. Li, Catal.: Sci. Technol., 2018, 8, 3044–3051.
159. J. T. Grant, C. A. Carrero, F. Goeltl, J. Venegas, P. Mueller, S. P. Burt, S. E. Specht, W. P. McDermott, A. Chieregato and I. Hermans, Science, 2016, 354, 1570–1573.
160. J. T. Grant, W. P. McDermott, J. M. Venegas, S. P. Burt, J. Micka, S. P. Phivilay, C. A. Carrero and I. Hermans, ChemCatChem, 2017, 9, 3623–3626.
161. L. Shi, D. Wang, W. Song, D. Shao, W. P. Zhang and A. H. Lu, ChemCatChem, 2017, 9, 1788–1793.
162. J. Tian, J. Tan, M. Xu, Z. Zhang, S. Wan, S. Wang, J. Lin and Y. Wang, Sci. Adv., 2019, 5.
163. A. M. Love, M. C. Cendejas, B. Thomas, W. P. McDermott, P. Uchupalanun, C. Kruszynski, S. P. Burt, T. Agbi, A. J. Rossini and I. Hermans, J. Phys. Chem. C, 2019, 123, 27000–27011.
164. Q. Liu, Y. Wu, F. Xing, Q. Liu, X. Guo and C. Huang, J. Catal., 2020, 381, 599–607.
165. B. Qiu, F. Jiang, W. D. Lu, B. Yan, W. C. Li, Z. C. Zhao and A. H. Lu, J. Catal., 2020, 385, 176–182.
166. Y. Wang, W. C. Li, Y. X. Zhou, R. Lu and A. H. Lu, Catal. Today, 2020, 339, 62–66.
167. Q. Zhang, C. Cao, T. Xu, M. Sun, J. Zhang, Y. Wang and H. Wan, Chem. Commun., 2009, 2376–2378,  10.1039/b823369a.
168. M. Chen, J. L. Wu, Y. M. Liu, Y. Cao, L. Guo, H. Y. He and K. N. Fan, J. Solid State Chem., 2011, 184, 3357–3363.
169. L. Wang, W. Chu, C. Jiang, Y. Liu, J. Wen and Z. Xie, J. Nat. Gas Chem., 2012, 21, 43–48.
170. B. Farin, P. Eloy, C. Poleunis, M. Devillers and E. M. Gaigneaux, Catal.: Sci. Technol., 2016, 6, 6046–6056.
171. S. Baoyi, X. Aiju and W. Jiang, Integr. Ferroelectr., 2016, 171, 16–22.
172. B. Farin, M. Devillers and E. M. Gaigneaux, Microporous Mesoporous Mater., 2017, 242, 200–207.
173. F. Ma, S. Chen, Y. Li, H. Zhou, A. Xu and W. Lu, Appl. Surf. Sci., 2014, 313, 654–659.
174. F. Ma, S. Chen, H. Zhou, Y. Li and W. Lu, RSC Adv., 2014, 4, 40776–40781.
175. S. Tanasoi, G. Mitran, N. Tanchoux, T. Cacciaguerra, F. Fajula, I. Sndulescu, D. Tichit and I. C. Marcu, Appl. Catal., A, 2011, 395, 78–86.
176. L. Cao, P. Dai, L. Zhu, L. Yan, R. Chen, D. Liu, X. Gu, L. Li, Q. Xue and X. Zhao, Appl. Catal., B, 2020, 262.
177. E. V. Kondratenko and A. Brückner, J. Catal., 2010, 274, 111–116.
178. X. Rozanska, E. V. Kondratenko and J. Sauer, J. Catal., 2008, 256, 84–94.
179. O. Ovsitser, M. Cherian, A. Brückner and E. V. Kondratenko, J. Catal., 2009, 265, 8–18.
180. O. Ovsitser, M. Cherian and E. V. Kondratenko, J. Phys. Chem. C, 2007, 111, 8594–8602.
181. E. V. Kondratenko, O. Ovsitser, J. Radnik, M. Schneider, R. Kraehnert and U. U. Dingerdissen, Appl. Catal., A, 2007, 319, 98–110.
182. E. V. Kondratenko, M. Cherian, M. Baerns, D. S. Su, R. Schlogl, X. Wang and I. E. Wachs, J. Catal., 2005, 234, 131–142.
183. O. Schwarz, B. Frank, C. Hess and R. Schomäcker, Catal. Commun., 2008, 9, 229–233.
184. A. Dinse, C. Carrero, A. Ozarowski, R. Schomäcker, R. Schlögl and K. P. Dinse, ChemCatChem, 2012, 4, 641–652.
185. S. Sugiyama, T. Osaka, Y. Ueno and K. I. Sotowa, J. Jpn. Pet. Inst., 2008, 51, 50–57.
186. M. N. Taylor, A. F. Carley, T. E. Davies and S. H. Taylor, Top. Catal., 2009, 52, 1660–1668.
187. X. Gao, C. Chen, S. Ren, J. Zhang and D. Su, Cuihua Xuebao, 2012, 33, 1069–1074.
188. I. V. Mishakov, A. A. Vedyagin, A. F. Bedilo, V. I. Zaikovskii and K. J. Klabunde, Catal. Today, 2009, 144, 278–284.
189. G. Mitran, A. Urda, N. Tanchoux, F. Fajula and I. C. Marcu, Catal. Lett., 2009, 131, 250–257.
190. A. H. S. Kootenaei, J. Towfighi, A. Khodadadi and Y. Mortazavi, Appl. Surf. Sci., 2014, 298, 26–35.
191. S. Arndt, B. Uysal, A. Berthold, T. Otrebma, Y. Aksu, M. Driess and R. Schomäcker, J. Nat. Gas Chem., 2012, 21, 581–594.
192. X. Sun, Y. Ding, B. Zhang, R. Huang and D. S. Su, Chem. Commun., 2015, 51, 9145–9148.
193. G. Mitran, T. Cacciaguerra, S. Loridant, D. Tichit and I. C. Marcu, Appl. Catal., A, 2012, 417-418, 153–162.
194. M. Khachani, M. Kacimi, A. Ensuque, J. Y. Piquemal, C. Connan, F. Bozon-Verduraz and M. Ziyad, Appl. Catal., A, 2010, 388, 113–123.
195. X. Zhang, R. You, D. Li, T. Cao and W. Huang, ACS Appl. Mater. Interfaces, 2017, 9, 35897–35907.
196. G. Mitran, R. Ahmed, E. Iro, S. Hajimirzaee, S. Hodgson, A. Urdă, M. Olea and I. C. Marcu, Catal. Today, 2018, 306, 260–267.
197. N. Hamilton, T. Wolfram, G. Tzolova Müller, M. Hävecker, J. Kröhnert, C. Carrero, R. Schomäcker, A. Trunschke and R. Schlögl, Catal.: Sci. Technol., 2012, 2, 1346–1359.
198. S. A. Karakoulia, K. S. Triantafyllidis, G. Tsilomelekis, S. Boghosian and A. A. Lemonidou, Catal. Today, 2009, 141, 245–253.
199. C. A. Carrero, C. J. Keturakis, A. Orrego, R. Schomäcker and I. E. Wachs, Dalton Trans., 2013, 42, 12644–12653.
200. O. Schwarz, D. Habel, O. Ovsitser, E. V. Kondratenko, C. Hess, R. Schomäcker and H. Schubert, J. Mol. Catal. A: Chem., 2008, 293, 45–52.
201. T. Kharlamova, E. Sushchenko, T. Izaak and O. Vodyankina, Catal. Today, 2016, 278, 174–184.
202. E. D. Sushchenko, T. S. Kharlamova, T. I. Izaak and O. V. Vodyankina, Kinet. Catal., 2017, 58, 630–641.
203. I. Rossetti, E. Bahadori, A. Tripodi and G. Ramis, Materials, 2019, 16.
204. S. A. Karakoulia, K. S. Triantafyllidis and A. A. Lemonidou, Microporous Mesoporous Mater., 2008, 110, 157–166.
205. Y. Liu, C. Jiang, W. Chu, W. Sun and Z. Xie, React. Kinet., Mech. Catal., 2010, 101, 141–151.
206. B. Schimmoeller, Y. Jiang, S. E. Pratsinis and A. Baiker, J. Catal., 2010, 274, 64–75.
207. K. Chalupka, C. Thomas, Y. Millot, F. Averseng and S. Dzwigaj, J. Catal., 2013, 305, 46–55.
208. T. Fu, Y. Wang, A. Wernbacher, R. Schlögl and A. Trunschke, ACS Catal., 2019, 9, 4875–4886.
209. P. R. Shah, I. Baldychev, J. M. Vohs and R. J. Gorte, Appl. Catal., A, 2009, 361, 13–17.
210. S. A. D'Ippolito, M. A. Bañares, J. L. G. Fierro and C. L. Pieck, Catal. Lett., 2008, 122, 252–258.
211. M. P. Casaletto, G. Landi, L. Lisi, P. Patrono and F. Pinzari, J. Mol. Catal. A: Chem., 2010, 329, 50–56.
212. Q. Liu, M. Luo, Z. Zhao and L. Guo, Catal. Lett., 2019, 149, 1345–1358.
213. W. Ruettinger, A. Benderly, S. Han, X. Shen, Y. Ding and S. L. Suib, Catal. Lett., 2011, 141, 15–21.
214. H. Zhang, S. Cao, Y. Zou, Y. M. Wang, X. Zhou, Y. Shen and X. Zheng, Catal. Commun., 2014, 45, 158–161.
215. M. Nadjafi, P. M. Abdala, R. Verel, D. Hosseini, O. V. Safonova, A. Fedorov and C. R. Müller, ACS Catal., 2020, 10, 2314–2321.
216. A. Dinse, B. Frank, C. Hess, D. Habel and R. Schomäcker, J. Mol. Catal. A: Chem., 2008, 289, 28–37.
217. A. Dinse, A. Ozarowski, C. Hess, R. Schomäcker and K. P. Dinse, J. Phys. Chem. C, 2008, 112, 17664–17671.
218. B. Frank, J. Zhang, R. Blume, R. Schlögl and D. S. Su, Angew. Chem., Int. Ed., 2009, 48, 6913–6917.
219. K. Fukudome and T. Suzuki, Catal. Surv. Asia, 2015, 19, 172–187.
220. J. Tian, J. Lin, M. Xu, S. Wan, J. Lin and Y. Wang, Chem. Eng. Sci., 2018, 186, 142–151.
221. S. Barman, N. Maity, K. Bhatte, S. Ould-Chikh, O. Dachwald, C. Haeßner, Y. Saih, E. Abou-Hamad, I. Llorens, J. L. Hazemann, K. Köhler, V. D'Elia and J. M. Basset, ACS Catal., 2016, 6, 5908–5921.
222. J. T. Grant, A. M. Love, C. A. Carrero, F. Huang, J. Panger, R. Verel and I. Hermans, Top. Catal., 2016, 59, 1545–1553.
223. S. Zhang and H. Liu, Appl. Catal., A, 2019, 573, 41–48.
224. A. Löfberg, T. Giornelli, S. Paul and E. Bordes-Richard, Appl. Catal., A, 2011, 391, 43–51.
225. A. Essakhi, A. Löfberg, S. Paul, P. Supiot, B. Mutel, V. Le Courtois and E. Bordes-Richard, Top. Catal., 2011, 54, 698–707.
226. H. Kazerooni, J. Towfighi Darian, Y. Mortazavi, A. A. Khadadadi and R. Asadi, Catal. Lett., 2020, 150, 2807–2822.
227. K. Fukudome, N. O. Ikenaga, T. Miyake and T. Suzuki, Catal.: Sci. Technol., 2011, 1, 987–998.
228. K. C. Szeto, B. Loges, N. Merle, N. Popoff, A. Quadrelli, H. Jia, E. Berrier, A. De Mallmann, L. Delevoye, R. M. Gauvin and M. Taoufik, Organometallics, 2013, 32, 6452–6460.
229. C. Carrero, M. Kauer, A. Dinse, T. Wolfram, N. Hamilton, A. Trunschke, R. Schlögl and R. Schomäcker, Catal.: Sci. Technol., 2014, 4, 786–794.
230. Y. Marco, L. Roldán, E. Muñoz and E. García-Bordejé, ChemSusChem, 2014, 7, 2496–2504.
231. W. D. Lu, D. Wang, Z. Zhao, W. Song, W. C. Li and A. H. Lu, ACS Catal., 2019, 9, 8263–8270.
232. S. Pei, B. Zhang, K. Jiao, R. Bao, B. Yue and H. He, Acta Phys.-Chim. Sin., 2008, 24, 561–564.
233. M. Salamanca, Y. E. Licea, A. Echavarría, A. C. Faro Jr and L. A. Palacio, Phys. Chem. Chem. Phys., 2009, 11, 9583–9591.
234. S. Crapanzano, I. V. Babich and L. Lefferts, Appl. Catal., A, 2011, 391, 70–77.
235. J. Velasquez, A. Echavarria, A. Faro and L. A. Palacio, Ind. Eng. Chem. Res., 2013, 52, 5582–5586.
236. Z. Li, A. W. Peters, V. Bernales, M. A. Ortuño, N. M. Schweitzer, M. R. Destefano, L. C. Gallington, A. E. Platero-Prats, K. W. Chapman, C. J. Cramer, L. Gagliardi, J. T. Hupp and O. K. Farha, ACS Cent. Sci., 2017, 3, 31–38.
237. M. Salamanca-Guzmán, Y. E. Licea-Fonseca, A. Echavarría-Isaza, A. Faro and L. A. Palacio-Santos, Rev. Fac. Ing., 2017, 2017, 97–104.
238. C. Boucetta, M. Kacimi, A. Ensuque, J. Y. Piquemal, F. Bozon-Verduraz and M. Ziyad, Appl. Catal., A, 2009, 356, 201–210.
239. J. Janas, J. Gurgul, R. P. Socha, J. Kowalska, K. Nowinska, T. Shishido, M. Che and S. Dzwigaj, J. Phys. Chem. C, 2009, 113, 13273–13281.
240. E. Rombi, D. Gazzoli, M. G. Cutrufello, S. De Rossi and I. Ferino, Appl. Surf. Sci., 2010, 256, 5576–5580.
241. Z. Bai, P. Li, L. Liu and G. Xiong, ChemCatChem, 2012, 4, 260–264.
242. G. Xiong and J. Sang, J. Mol. Catal. A: Chem., 2014, 392, 315–320.
243. Z. Zhai, X. Wang, R. Licht and A. T. Bell, J. Catal., 2015, 325, 87–100.
244. E. Goudarzi, R. Asadi, J. T. Darian and A. Shahbazi Kootenaei, RSC Adv., 2019, 9, 11797–11809.
245. D. Shee and G. Deo, J. Mol. Catal. A: Chem., 2009, 308, 46–55.
246. J. Li, C. Wang, C. Huang, W. Weng and H. Wan, Catal. Lett., 2010, 137, 81–87.
247. J. H. Li, C. C. Wang, C. J. Huang, Y. F. Sun, W. Z. Weng and H. L. Wan, Appl. Catal., A, 2010, 382, 99–105.
248. K. Fukudome, A. Kanno, N. O. Ikenaga, T. Miyake and T. Suzuki, Catal. Lett., 2011, 141, 68–77.
249. W. Fang, Q. J. Ge, J. F. Yu and H. Y. Xu, Ind. Eng. Chem. Res., 2011, 50, 1962–1967.
250. T. Li, J. Wang, B. Zhaorigetu and A. Xu, React. Kinet., Mech. Catal., 2013, 110, 421–435.
251. X. Fan, J. Li, Z. Zhao, Y. Wei, J. Liu, A. Duan and G. Jiang, J. Energy Chem., 2014, 23, 171–178.
252. B. Farin, C. Swalus, M. Devillers and E. M. Gaigneaux, Catal. Today, 2013, 203, 24–31.
253. S. Wannakao, B. Boekfa, P. Khongpracha, M. Probst and J. Limtrakul, ChemPhysChem, 2010, 11, 3432–3438.
254. N. H. Nguyen, T. H. Tran, M. T. Nguyen and M. C. Le, Int. J. Quantum Chem., 2010, 110, 2653–2670.
255. M.-F. R. Konstantinos Alexopoulos and G. B. Marin, J. Catal., 2012, 289, 127–139.
256. J. M. H. Lo, Z. A. Premji, T. Ziegler and P. D. Clark, J. Phys. Chem. C, 2013, 117, 11258–11274.
257. Y. J. Du, Z. H. Li and K. N. Fan, J. Mol. Catal. A: Chem., 2013, 379, 122–138.
258. X. Rozanska, R. Fortrie and J. Sauer, J. Am. Chem. Soc., 2014, 136, 7751–7761.
259. J. Liu, F. Mohamed and J. Sauer, J. Catal., 2014, 317, 75–82.
260. V. I. Avdeev and A. F. Bedilo, Res. Chem. Intermed., 2016, 42, 5237–5252.
261. Y.-J. Du, Z. H. Li and K.-N. Fan, Surf. Sci., 2012, 606, 956–964.
262. L. Cheng, G. A. Ferguson, S. A. Zygmunt and L. A. Curtiss, J. Catal., 2013, 302, 31–36.
263. A. Hofmann, M. V. Ganduglia-Pirováno and J. Sauer, J. Phys. Chem. C, 2009, 113, 18191–18203.
264. C. Popa, M. V. Ganduglia-Pirovano and J. Sauer, J. Phys. Chem. C, 2011, 115, 7399–7410.
265. L. He, L. Fu and Y. Tang, Catal.: Sci. Technol., 2015, 5, 1115–1125.
266. G. Wang, H. Dai, L. Zhang, J. Deng, C. Liu, H. He and C. T. Au, Appl. Catal., A, 2010, 375, 272–278.
267. S. M. Al-Zahrani, N. O. Elbashir, A. E. Abasaeed and M. Abdulwahed, Ind. Eng. Chem. Res., 2001, 40, 781–784.
268. M. Hoang, J. F. Mathews, K. C. Pratt and Z. Xie, Kinet. Catal., 2010, 51, 398–403.
269. S. M. Al-Zahrani, N. O. Elbashir, A. E. Abasaeed and M. Abdulwahed, J. Mol. Catal. A: Chem., 2004, 218, 179–186.
270. B. Y. Jibril, N. O. Elbashir, S. M. Al-Zahrani and A. E. Abasaeed, Chem. Eng. Process., 2005, 44, 835–840.
271. K. Samson, B. Grzybowska, R. Grabowski, M. A. Banares and E. Lozano Diz, Pol. J. Chem., 2009, 83, 1977–1992.
272. G. Karamullaoglu, S. Onen and T. Dogu, Chem. Eng. Process., 2002, 41, 337–347.
273. Y. Kato, S. Nitta, S. Shimazu, M. Kurashina, M. Katoh, W. Ninomiya and S. Sugiyama, J. Chem. Eng. Jpn., 2019, 52, 99–105.
274. G. Neri, A. Pistone, S. De Rossi, E. Rombi, C. Milone and S. Galvagno, Appl. Catal., A, 2004, 260, 75–86.
275. P. Moriceau, B. Grzybowska, L. Gengembre and Y. Barbaux, Appl. Catal., A, 2000, 199, 73–82.
276. N. O. Elbashir, S. M. Al-Zahrani, A. E. Abasaeed and M. Abdulwahed, Chem. Eng. Process., 2003, 42, 817–823.
277. G. Wang, L. Zhang, J. Deng, H. Dai, H. He and C. T. Au, Appl. Catal., A, 2009, 355, 192–201.
278. J. Słoczyński, B. Grzybowska, A. Kozłowska, K. Samson, R. Grabowski, A. Kotarba and M. Hermanowska, Catal. Today, 2011, 169, 29–35.
279. Q. Zhang, Y. Wang, Y. Ohishi, T. Shishido and K. Takehira, J. Catal., 2001, 202, 308–318.
280. V. Iannazzo, G. Neri, S. Galvagno, M. Di Serio, R. Tesser and E. Santacesaria, Appl. Catal., A, 2003, 246, 49–68.
281. Z. Fan, T. Zeng, W. Wu, D. Jiang and C. Miao, Ind. Eng. Chem. Res., 2019, 58, 10249–10254.
282. B. Sulikowski, Z. Olejniczak, E. Wloch, J. Rakoczy, R. X. Valenzuela and V. Cortés Corberán, Appl. Catal., A, 2002, 232, 189–202.
283. V. P. Vislovskiy, N. T. Shamilov, A. M. Sardarly, R. M. Talyshinskii, V. Y. Bychkov, P. Ruiz, V. Cortés Corberán, Z. Schay and Z. Koppany, Appl. Catal., A, 2003, 250, 143–150.
284. N. T. Shamilov and V. P. Vislovskiy, J. Korean Chem. Soc., 2011, 55, 812–818.
285. S. Kraemer, A. J. Rondinone, Y. T. Tsai, V. Schwartz, S. H. Overbury, J. C. Idrobo and Z. Wu, Catal. Today, 2016, 263, 84–90.
286. T. Shishido, A. Inoue, T. Konishi, I. Matsuura and K. Takehira, Catal. Lett., 2000, 68, 215–221.
287. Z. Wu, V. Schwartz, M. Li, A. J. Rondinone and S. H. Overbury, J. Phys. Chem. Lett., 2012, 3, 1517–1522.
288. G. Mitran, I. C. Marcu, A. Urdă and I. Săndulescu, J. Serb. Chem. Soc., 2010, 75, 1115–1124.
289. H. Xie, Z. Wu, S. H. Overbury, C. Liang and V. Schwartz, J. Catal., 2009, 267, 158–166.
290. C. Liang, H. Xie, V. Schwartz, J. Howe, S. Dai and S. H. Overbury, J. Am. Chem. Soc., 2009, 131, 7735–7741.
291. I. Gniot, P. Kirszensztejn and M. Kozłowski, Appl. Catal., A, 2009, 362, 67–74.
292. N. Martin-Sanchez, O. S. G. P. Soares, M. F. R. Pereira, M. J. Sanchez-Montero, J. L. Figueiredo and F. Salvador, Appl. Catal., A, 2015, 502, 71–77.
293. J. de Jesús Díaz Velásquez, L. M. C. Suárez and J. L. Figueiredo, Appl. Catal., A, 2006, 311, 51–57.
294. V. Schwartz, H. Xie, H. M. Meyer Iii, S. H. Overbury and C. Liang, Carbon, 2011, 49, 659–668.
295. G. K. P. Dathar, Y. T. Tsai, K. Gierszal, Y. Xu, C. Liang, A. J. Rondinone, S. H. Overbury and V. Schwartz, ChemSusChem, 2014, 7, 483–491.
296. I. Pelech, O. S. G. P. Soares, M. F. R. Pereira and J. L. Figueiredo, Catal. Today, 2015, 249, 176–183.
297. R. Zãvoianu, C. R. Dias and M. F. Portela, Catal. Commun., 2001, 2, 37–42.
298. Y. J. Zhang, I. Rodríguez-Ramos and A. Guerrero-Ruiz, Catal. Today, 2000, 61, 377–382.
299. Y. A. Agafonov, N. V. Nekrasov, N. A. Gaidai, M. A. Botavina, P. E. Davydov and A. L. Lapidus, Kinet. Catal., 2009, 50, 577–582.
300. Y. A. Agafonov, N. V. Nekrasov, N. A. Gaidai and A. L. Lapidus, Kinet. Catal., 2007, 48, 255–264.
301. C. R. Dias, R. Zavoianu and M. F. Portela, Catal. Commun., 2002, 3, 85–90.
302. S. M. Al-Zahrani, N. O. Elbashir, A. E. Abasaeed and M. Abdulwahed, Catal. Lett., 2000, 69, 65–70.
303. I. C. Marcu, J. M. M. Millet and I. Săndulescu, J. Serb. Chem. Soc., 2005, 70, 791–798.
304. Y. Takita, X. Qing, A. Takami, H. Nishiguchi and K. Nagaoka, Appl. Catal., A, 2005, 296, 63–69.
305. I. C. Marcu, M. N. Urlan, A. Rédey and I. Săndulescu, C. R. Chim., 2010, 13, 365–371.
306. N. T. Shamilov and V. P. Vislovskiy, J. Korean Chem. Soc., 2011, 55, 81–85.
307. Y. A. Agafonov, N. V. Nekrasov and N. A. Gaidai, Kinet. Catal., 2001, 42, 821–827.
308. K. Lohbeck, H. Haferkorn, W. Fuhrmann and N. Fedtke, Ullmann's Encyclopedia of Industrial Chemistry, 2000 DOI:10.1002/14356007.a16_053.
309. A. S. Sandupatla, K. Ray, P. Thaosen, C. Sivananda and G. Deo, Catal. Today, 2020, 354, 176–182.
310. P. Michorczyk and J. Ogonowski, React. Kinet. Catal. Lett., 2003, 78, 41–47.
311. F. Cavani, N. Ballarini and A. Cericola, Catal. Today, 2007, 127, 113–131.
312. K.-O. Hinrichsen, K. Kochloefl and M. Muhler, in Handbook of Heterogeneous Catalysis, ed. G. Ertl, H. Knözinger, F. Schüth and J. Weitkamp, 2008, pp. 2905–2920 DOI:10.1002/9783527610044.hetcat0147.
313. Y. A. Agafonov, N. A. Gaidai and A. L. Lapidus, Russ. Chem. Bull., 2014, 63, 381–388.
314. P. Michorczyk, P. Pietrzyk and J. Ogonowski, Microporous Mesoporous Mater., 2012, 161, 56–66.
315. A. Burri, M. A. Hasib, Y. H. Mo, B. M. Reddy and S. E. Park, Catal. Lett., 2018, 148, 576–585.
316. H. M. Wang, Y. Chen, X. Yan, W. Z. Lang and Y. J. Guo, Microporous Mesoporous Mater., 2019, 284, 69–77.
317. Y. A. Agafonov, N. A. Gaidai and A. L. Lapidus, Kinet. Catal., 2018, 59, 744–753.
318. R. Jin, J. Easa, D. T. Tran and C. P. O'Brien, Catal.: Sci. Technol., 2020, 10, 1769–1777.
319. P. Michorczyk, P. Pietrzyk and J. Ogonowski, Microporous Mesoporous Mater., 2012, 161, 56–66.
320. P. Michorczyk, K. Zenczak-Tomera, B. Michorczyk, A. Wegrzyniak, M. Basta, Y. Millot, L. Valentin and S. Dzwigaj, J. CO₂ Util., 2020, 36, 54–63.
321. J. F. S. de Oliveira, D. P. Volanti, J. M. C. Bueno and A. P. Ferreira, Appl. Catal., A, 2018, 558, 55–66.
322. Z. F. Han, X. L. Xue, J. M. Wu, W. Z. Lang and Y. J. Guo, Chin. J. Catal., 2018, 39, 1099–1109.
323. X. L. Xue, W. Z. Lang, X. Yan and Y. J. Guo, ACS Appl. Mater. Interfaces, 2017, 9, 15408–15423.
324. I. Ascoop, V. V. Galvita, K. Alexopoulos, M. F. Reyniers, P. Van Der Voort, V. Bliznuk and G. B. Marin, J. Catal., 2016, 335, 1–10.
325. M. Chen, J. Xu, F.-Z. Su, Y.-M. Liu, Y. Cao, H.-Y. He and K.-N. Fan, J. Catal., 2008, 256, 293–300.
326. B. Xu, B. Zheng, W. Hua, Y. Yue and Z. Gao, J. Catal., 2006, 239, 470–477.
327. H. Xiao, J. F. Zhang, P. Wang, X. X. Wang, F. Pang, Z. Z. Zhang and Y. S. Tan, Catal. Sci. Technol., 2016, 6, 5183–5195.
328. M. Chen, J. Xu, Y. M. Liu, Y. Cao, H. Y. He and J. H. Zhuang, Appl. Catal., A, 2010, 377, 35–41.
329. M. Chen, J. Xu, Y. Cao, H.-Y. He, K.-N. Fan and J.-H. Zhuang, J. Catal., 2010, 272, 101–108.
330. M. Chen, J. L. Wu, Y. M. Liu, Y. Cao, L. Guo, H. Y. He and K. N. Fan, Appl. Catal., A, 2011, 407, 20–28.
331. M. R. Kantserova, N. V. Vlasenko, S. M. Orlyk, K. Veltruska and I. Matolinova, Theor. Exp. Chem., 2019, 55, 207–214.
332. H. F. Tian, J. K. Liao, F. Zha, X. J. Guo, X. H. Tang, Y. Chang and X. X. Ma, ChemistrySelect, 2020, 5, 3626–3637.
333. Y. Ren, J. Wang, W. Hua, Y. Yue and Z. Gao, J. Ind. Eng. Chem., 2012, 18, 731–736.
334. P. Michorczyk, K. Zeńczak, R. Niekurzak and J. Ogonowski, Pol. J. Chem. Technol., 2012, 14, 77.
335. J. F. S. de Oliveira, D. P. Volanti, J. M. C. Bueno and A. P. Ferreira, Appl. Catal., A, 2018, 558, 55–66.
336. G. Sun, Q. Huang, S. Huang, Q. Wang, H. Li, H. Liu, S. Wan, X. Zhang and J. Wang, Catalysts, 2016, 6.
337. J. Ogonowski and E. Skrzyńska, React. Kinet. Catal. Lett., 2006, 88, 293–300.
338. C. L. Wei, F. Q. Xue, C. X. Miao, Y. H. Yue, W. M. Yang, W. M. Hua and Z. Gao, Catalysts, 2016, 6.
339. J. Ogonowski and E. Skrzyńska, Catal. Lett., 2008, 121, 234–240.
340. R. Yuan, Y. Li, H. Yan, H. Wang, J. Song, Z. Zhang, W. Fan, J. Chen, Z. Liu, Z. Liu and Z. Hao, Cuihua Xuebao, 2014, 35, 1329–1336.
341. J. Ogonowski and E. Skrzynska, Catal. Commun., 2009, 11, 132–136.
342. C. Wei, F. Xue, C. Miao, Y. Yue, W. Yang, W. Hua and Z. Gao, Chin. J. Chem., 2017, 35, 1619–1626.
343. Y. J. Luo, C. X. Miao, Y. H. Yue, W. M. Yang, W. M. Hua and Z. Gao, Catalysts, 2019, 9.
344. J. F. Ding, R. Shao, J. Wu, Z. F. Qin and J. G. Wang, React. Kinet., Mech. Catal., 2010, 101, 173–181.
345. J. F. Ding, Z. F. Qin, X. K. Li, G. F. Wang and J. G. Wang, J. Mol. Catal. A: Chem., 2010, 315, 221–225.
346. A. Aouissi, D. Aldhayan and S. Alkahtani, Chin. J. Catal., 2012, 33, 1474–1479.
347. I. V. Mishakov, E. V. Ilyina, A. F. Bedilo and A. A. Vedyagin, React. Kinet. Catal. Lett., 2009, 97, 355–361.
348. A. Ates, C. Hardacre and A. Goguet, Appl. Catal., A, 2012, 441–442, 30–41.
349. P. Sazama, N. K. Sathu, E. Tabor, B. Wichterlová, Š. Sklenák and Z. Sobalík, J. Catal., 2013, 299, 188–203.
350. G. Wu, F. Hei, N. Guan and L. Li, Catal.: Sci. Technol., 2013, 3, 1333–1342.
351. G. Wu, Y. Hao, N. Zhang, N. Guan, L. Li and W. Grünert, Microporous Mesoporous Mater., 2014, 198, 82–91.
352. G. Wu, F. Hei, N. Zhang, N. Guan, L. Li and W. Grünert, Appl. Catal., A, 2013, 468, 230–239.
353. G. I. Panov, CATTECH, 2000, 4, 18–32.
354. O. Ovsitser and E. V. Kondratenko, Chem. Commun., 2010, 46, 4974–4976.
355. Z. Wang, Z. Bian, N. Dewangan, J. Xu and S. Kawi, J. Membr. Sci., 2019, 578, 36–42.
356. I. Rossetti, L. Fabbrini, N. Ballarini, C. Oliva, F. Cavani, A. Cericola, B. Bonelli, M. Piumetti, E. Garrone, H. Dyrbeck, E. A. Blekkan and L. Forni, J. Catal., 2008, 256, 45–61.
357. V. Balcaen, I. Sack, M. Olea and G. B. Marin, Appl. Catal., A, 2009, 371, 31–42.
358. S. Al-Ghamdi, J. Moreira and H. De Lasa, Ind. Eng. Chem. Res., 2014, 53, 15317–15332.
359. S. Rostom and H. de Lasa, Chem. Eng. Process., 2019, 137, 87–99.
360. J. Beckers and G. Rothenberg, Green Chem., 2009, 11, 1550–1554.
361. J. Beckers and G. Rothenberg, Dalton Trans., 2009, 5673–5682,  10.1039/b904681j.
362. S. Crapanzano, I. V. Babich and L. Lefferts, Catal. Today, 2013, 203, 17–23.
363. S. Crapanzano, I. V. Babich and L. Lefferts, Appl. Catal., A, 2010, 385, 14–21.
364. S. Crapanzano, I. V. Babich and L. Lefferts, Appl. Catal., A, 2010, 378, 144–150.
365. R. K. Grasselli, D. L. Stern and J. G. Tsikoyiannis, Appl. Catal., A, 1999, 189, 1–8.
366. R. K. Grasselli, D. L. Stern and J. G. Tsikoyiannis, Appl. Catal., A, 1999, 189, 9–14.
367. L. M. van der Zande, E. A. de Graaf and G. Rothenberg, Adv. Synth. Catal., 2002, 344, 884–889.
368. M.-L. Yang, C. Fan, Y.-A. Zhu, Z.-J. Sui, X.-G. Zhou and D. Chen, J. Phys. Chem. C, 2015, 119, 21386–21394.
369. R. B. Dudek, Y. Gao, J. Zhang and F. Li, AIChE J., 2018, 64, 3141–3150.
370. P. Munnik, P. E. de Jongh and K. P. de Jong, Chem. Rev., 2015, 115, 6687–6718.
371. S. L. Wegener, T. J. Marks and P. C. Stair, Acc. Chem. Res., 2012, 45, 206–214.
372. C. Perego and P. Villa, Catal. Today, 1997, 34, 281–305.
373. S. Barman, N. Maity, K. Bhatte, S. Ould-Chikh, O. Dachwald, C. Haeßner, Y. Saih, E. Abou-Hamad, I. Llorens, J.-L. Hazemann, K. Köhler, V. D’Elia and J.-M. Basset, ACS Catal., 2016, 6, 5908–5921.
374. C. Copéret, Acc. Chem. Res., 2019, 52, 1697–1708.
375. J. Lu, J. W. Elam and P. C. Stair, Surf. Sci. Rep., 2016, 71, 410–472.
376. A. M. Johnson, B. R. Quezada, L. D. Marks and P. C. Stair, Top. Catal., 2014, 57, 177–187.
377. J. R. v. Ommen, D. Kooijman, M. d. Niet, M. Talebi and A. Goulas, J. Vac. Sci. Technol., A, 2015, 33, 021513.
378. W. Hengwei and L. Junling, Acta Phys.-Chim. Sin., 2018, 34, 1334–1357.
379. M. Campanati, G. Fornasari and A. Vaccari, Catal. Today, 2003, 77, 299–314.
380. H. K. Olaf Deutschmann, K. Kochloefl and T. Turek, Ullmann's Encyclopedia of Industrial Chemistry, 2009 DOI:10.1002/14356007.a05_313.pub2.
381. H. H. Kung and E. I. Ko, Chem. Eng. J. Biochem., 1996, 64, 203–214.
382. J. A. Schwarz, C. Contescu and A. Contescu, Chem. Rev., 1995, 95, 477–510.
383. M. Shandilya, R. Rai and J. Singh, Adv. Appl. Ceram., 2016, 115, 354–376.
384. A. Rabenau, Angew. Chem., Int. Ed. Engl., 1985, 24, 1026–1040.
385. A. N. Shigapov, G. W. Graham, R. W. McCabe and H. K. Plummer, Appl. Catal., A, 2001, 210, 287–300.
386. R. Liu and C.-a. Wang, Microporous Mesoporous Mater., 2014, 186, 1–6.
387. J. Yano and V. K. Yachandra, Photosynth. Res., 2009, 102, 241.
388. E. E. Alp, S. M. Mini and M. Ramanathan, X-ray absorption spectroscopy: EXAFS and XANES - A versatile tool to study the atomic and electronic structure of materials, United States, 1990.
389. M. Che and E. Giamello, in Stud. Surf. Sci. Catal., ed. J. L. G. Fierro, Elsevier, 1990, vol. 57, pp. B265–B332.
390. A. Brückner, Chem. Soc. Rev., 2010, 39, 4673–4684.
391. X. Gao, S. R. Bare, B. M. Weckhuysen and I. E. Wachs, J. Phys. Chem. B, 1998, 102, 10842–10852.
392. E. L. Lee and I. E. Wachs, J. Phys. Chem. C, 2007, 111, 14410–14425.
393. A. Chakrabarti and I. E. Wachs, Catal. Lett., 2015, 145, 985–994.
394. J. Ryczkowski, Catal. Today, 2001, 68, 263–381.
395. C. Lamberti, A. Zecchina, E. Groppo and S. Bordiga, Chem. Soc. Rev., 2010, 39, 4951–5001.
396. I. E. Wachs and C. A. Roberts, Chem. Soc. Rev., 2010, 39, 5002–5017.
397. M. A. Bañares and I. E. Wachs, J. Raman Spectrosc., 2002, 33, 359–380.
398. J. J. Fitzgerald and S. M. DePaul, in Solid-State NMR Spectroscopy of Inorganic Materials, American Chemical Society, 1999, ch. 1, vol. 717, pp. 2–133.
399. M. de Oliveira, D. Seeburg, J. Weiß, S. Wohlrab, G. Buntkowsky, U. Bentrup and T. Gutmann, Catal. Sci. Technol., 2019, 9, 6180–6190.
400. J. A. Dumesic, G. W. Huber and M. Boudart, in Handbook of Heterogeneous Catalysis, ed. G. Ertl, H. Knözinger, F. Schüth and J. Weitkamp, 2008, vol. 1, pp. 1–15.
401. J. C. Védrine, Catalysts, 2017, 7, 341.
402. Y. Bai, H. Huang, C. Wang, R. Long and Y. Xiong, Mater. Chem. Front., 2017, 1, 1951–1964.
403. J. T. Gleaves, G. S. Yablonskii, P. Phanawadee and Y. Schuurman, Appl. Catal., A, 1997, 160, 55–88.
404. J. Pérez-Ramírez and E. V. Kondratenko, Catal. Today, 2007, 121, 160–169.
405. K. Morgan, N. Maguire, R. Fushimi, J. T. Gleaves, A. Goguet, M. P. Harold, E. V. Kondratenko, U. Menon, Y. Schuurman and G. S. Yablonsky, Catal. Sci. Technol., 2017, 7, 2416–2439.
406. Z. Zhang, X. E. Verykios and M. Baerns, Catal. Rev., 1994, 36, 507–556.
407. C. Ahamer, A. K. Opitz, G. M. Rupp and J. Fleig, J. Electrochem. Soc., 2017, 164, F790–F803.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/d0cs01140a

---