Light olefins (C₂⁼–C₄⁼) are the most important feedstocks in the chemical industry, with a global production of > 200 million metric tons per year, and are currently produced through either the cracking of petroleum-derived hydrocarbons or a multistage coal-based methanol-to-olefins process that commonly involves extensive energy consumption and results in enormous carbon emissions [1-3].

The recent increase in the availability of worldwide shale gas and natural gas resources, which contain considerable amounts of light alkanes (C₂–C₄), has stimulated a quick technical shift to obtaining commodity light olefins through gas-based dehydrogenation routes [4, 5]. Taking propylene production as an example, catalytic dehydrogenation of propane (PDH) has supplied ~5% of global propylene, and this is anticipated to increase to 20% by 2020 [6]. Nevertheless, direct dehydrogenation protocols relying on platinum-or chromium oxide-based catalysts practically suffer from thermodynamic limitations in further enhancing the reaction efficiency and also rapid catalyst deactivation through coking and catalyst sintering, and thus require frequent regeneration of the catalysts under harsh conditions [7, 8].

Oxidative dehydrogenation (ODH) is an attractive alternative to direct dehydrogenation processes because of favorable thermodynamic and kinetic characteristics, e.g., exothermic, lower reaction temperature, faster reaction rate, and especially no coking [9-11]. In the past few decades, catalytic systems giving good performances in the ODH reaction of light alkanes have been those based on transition metal oxides (e.g., V₂O₅, MoO₃, and NiO) or alkaline-earth metal oxychlorides [12]. However, the formed electron-rich olefins easily react with the surfaces of metal oxide catalysts, resulting in the cleavage of the C-C bond through a subsequent oxygen insertion, thus forming the undesired over-oxidation product, CO₂. As summarized in Fig. 1(A) for the well-developed metal oxide catalysts, the over-oxidation product, CO₂, constitutes a very high proportion of the products, typically around 20% and in some cases even 50%, at conversion levels of ~20%, a typical value of industrial interest [13-22]. There is therefore an urgent need to develop new catalysts that can produce olefins with high selectivity under the severe conditions of alkane activation, but this is a great challenge.

Boron nitride has been extensively applied in the fields of physics, electronics, and aerospace as a sealing material, taking advantage of its excellent anti-oxidation properties and structural stability [23, 24]. In catalysis research, boron nitride is generally considered to be chemically inert, and has been used merely as a carrier for a long time [25, 26]. Also, Fu and Bao [27, 28] have confirmed that a boron nitride cover can be used as a promoter to tune molecule-metal interactions and enhance metal-catalyzed reactions. Recently, the Hermans group at University of Wisconsin-Madison, USA, and our research group at Dalian University of Technology, China, independently disclosed that boron nitride could efficiently catalyze the oxidative dehydrogenation of propane to propylene with impressive selectivity, and with only negligible CO₂ formation (Fig. 1(A)) [29, 30]. It is important to highlight that such a novel catalyst system addresses the difficulty of selectivity control for olefins that has long existed in oxidative dehydrogenation because of the over-oxidation reactions that produce a substantial amount of undesired CO₂.

In Hermans' study, hexagonal boron nitride (h-BN) exhibited high selectivity for propylene (79%) and ethylene (12%) at a 14% conversion level of propane [29]. In our case, edge-hydroxylation treatment granted boron nitride higher activity (20.6%), selectivity (80.2%), and stability for at least 300 h [30]. Our control experiment with a feed gas of propylene and molecular oxygen also showed that the boron nitride catalyst has almost no activity for propylene oxidation, and thus prevents its over-oxidation to CO₂ [30]. One of the most important requirements for commercially attractive oxidative dehydrogenation catalysts is high productivity (g_olefin g_cat^-1 h^-1) of the olefins. Hexagonal boron nitride exhibited a productivity of ~1 g_olefin g_cat^-1 h^-1 [29], while hydroxylated boron nitride (BNOH) showed a higher olefin production of 6.8 g_olefin g_cat^-1 h^-1 (Fig. 1(B)), which is far higher than those of metal oxide catalysts.

In addition to the ODH reaction of propane, follow-up reports showed that boron nitrides were active catalysts for the oxidative dehydrogenations of n-butane and isobutane [32] and ethane [33], with the latter conducted at higher temperatures. The study on ethane ODH also demonstrated the outstanding high-temperature oxidation resistance of boron nitride. For example, using a hydroxylated boron nitride catalyst, at a very high conversion level of 63%, the ethylene selectivity was 80%, which is comparable to that of the industrial steam-cracking route [33]. All of these results confirmed that boron nitride offers rather high selectivity, and only small amounts of CO₂ are formed. A recent report showed that boron nitride nanosheets have delivered a remarkable 60% selectivity at an ethane conversion of 78% and remained stable over 400 h [34]. In addition, carbon-doped boron nitride nanosheets have also shown high activity and selectivity in the oxidative dehydrogenation of ethylbenzene to styrene, as well as excellent oxidation resistance [35].

Although these studies corroborate the unprecedented catalytic properties of boron nitride in the oxidative dehydrogenation of light alkanes, the origin of the catalytic behavior of such a novel and unique catalyst system is still controversial. This is actually normal for the emergence of new catalysts. Based on DFT calculations, Hermans and coworkers [29] proposed that the B-O-O-N intermediate formed on an armchair edge of boron nitride by activating molecular oxygen was likely to be the catalytically active site for the cleavage of the initial C-H bond in propane (Fig. 2(A)). However, many studies have shown that the most frequently exposed terminals in hexagonal boron nitride materials are zig-zag edges [36-38]. It appears that the deduction based on the armchair edge and the B-O-O-N intermediate remains to be clarified.

In our earliest investigation, both hexagonal boron nitride and boron nitride nanotubes (BNNTs) were tested in the oxidative dehydrogenation reaction of propane. It appeared that these two materials exhibited very low initial activity (typically < 3%) for propane conversion, and there existed a catalyst activation period, i.e., a gradual increase in propane conversion along with the ODH reaction progress (Fig. 2(B)). This suggests that it is difficult for the intact boron nitride to activate molecular oxygen under the ODH reaction conditions. The catalytic activity most likely originates from the in situ surface functionalization of boron nitride under the reaction conditions. Thus, there is a compelling need to rationally tailor terminated edges and functional groups of boron nitride for optimal catalytic performance and to understand its catalytic origin.

To achieve these goals, we made many efforts to explore the functionalization of boron nitride. We adhered to two principles: exposing the edge as much as possible for the reactants' accessibility and maximizing the number of functional groups for abundant active sites. As exemplified in our previous study [30], a sodium-assisted high-temperature steam activation process rendered the best catalytic efficacy of boron nitride in the ODH reaction. A comparison of the aberration-corrected TEM images of boron nitride before and after functionalization in Fig. 3 reveals the generation of a large number of accessible edges after functionalization. Also, lattice relaxations and distortions at the edge of boron nitride were observed. High-sensitivity low-energy ion scattering (HS-LEIS) studies revealed that a gradual substitution of nitrogen by oxygen occured during the functionalization of pristine boron nitride. Furthermore, electron energy loss spectroscopy (EELS, Fig. 3) of boron nitride clearly affirmed that more edges other than those in the planes of boron nitride were functionalized by oxygen atoms. Considering the inert nature of pure boron nitride, these evolutions in structure and composition at the edge of boron nitride after the functionalization are likely to contribute to the unique catalytic function in the ODH reaction of propane.

These insights shed light on the catalytic origin of activated boron nitride in the ODH reaction. Initially, the explicit configuration and functional groups at the edges of active boron nitride were identified using advanced spectroscopy techniques, including K-edge EELS, X-ray photoelectron spectroscopy, two-dimensional multiple-quantum MAS NMR, and infrared spectroscopy. We precisely confirmed the dominantly exposed zig-zag boron-terminated edges and the co-existence of B-O-B and B-OH groups at these edges of the active boron nitride catalyst. When only C₃H₈ was pulsed into the boron nitride catalyst, there was no propane conversion or generation of any products, indicating that the oxygen-containing edge boron species themselves have no catalytic activity for C-H cleavage in propane [30].

A more interesting finding from in situ IR was that the intensity of B-OH groups gradually decreased only with co-feeding of propane and oxygen, accompanied by the formation of propylene, indicating that the -OH groups interacted with molecular oxygen and were incorporated into the reaction network. Isotope-labeling measurements provided further insight into this process, where the hydrogen atoms in the B-OH group were abstracted during the ODH reaction. The kinetic analysis confirmed that oxygen reaction dependence followed up the L-H model, indicating a dissociative adsorption of molecular oxygen in the reaction process. Meanwhile, ¹⁸O isotope tracer study demonstrated that an oxygen exchange between the surface oxygen atoms in the B-OH group and molecular oxygen occurred in the reaction process [30]. These objective findings allowed us to conclude that the B-OH groups at the edge of boron nitride are initially oxidized by molecular oxygen, and trigger the dehydrogenation reaction of propane. Reaction order of propane was two, and the normal KIE (kH/kD) of 1.4–1.5 was observed, indicating that the activation of propane determined the total reaction rate [30]. Based on the above observations, a relative simple reaction pathway (Scheme 1) is proposed including a redox reaction cycle. The B-OH sites initially react with molecular oxygen, leading to the production of B-O-O-B intermediates (oxidation step). The B-O-O-B species further abstract the hydrogen atoms from propane, forming C₃H₆ and H₂O (reduction step), and then recover to B-OH sites with the assistance of water.

Recently, the Su group's operando infrared spectroscopy and ¹⁸O isotope tracer study demonstrated that the active sites of B-O(H) were formed at the edges of boron nitride via the aid of ethane, and the dehydrogenation cycle could be completed with the assistance of O₂ at B-O sites [34]. They also pointed out that a detailed comparison between the roles of the B-O and B-OH groups in the alkane ODH reaction was still of value for a deep understanding of the reaction mechanism. Another recent paper from Hermans' group reported that boron-containing materials, such as boron carbide, titanium boride, nickel boride, cobalt boride, hafnium boride, tungsten boride, and elemental boron, showed the same exceptional behavior in the propane ODH reaction as boron nitride [39]. Based on the results from X-ray photoelectron and IR spectroscopy, they suggested the formation of an analogous surface-stabilized BOx active site for all tested boride catalysts [39]. These findings support our viewpoints on the catalytic origin of boron nitride in ODH.

An ODH catalyst, boron nitride, was discovered recently, and features superior selectivity toward light olefins, excellent stability, and atom economy for valuable carbon-based products by minimizing CO₂ emission. More specifically, the key points are as follows:

(1) Pure boron nitride is catalytically inert, and an activation period is required under the reaction conditions. Hence, a straightforward strategy for preparing an active boron nitride catalyst with chemical oxy-functionalization at the edges via alkali-metal-assisted steam activation to tailor and functionalize the terminal edge of boron nitride to enhance catalytic efficacy is desired.

(2) Boron nitride is highly promising for industrial implementation because of its high production of light olefins, negligible formation of CO₂, remarkable catalyst stability, superior anti-oxidation ability, and excellent thermal conductivity.

(3) A new redox reaction cycle is proposed, i.e. neighboring B-OH sites are initially oxidized by molecular oxygen (oxidation step), and then trigger dehydrogenation of light alkanes to olefins (reduction step).

The development of this novel catalyst system not only opens up a new research direction in selective cleavage of C-H bonds of alkanes by metal-free catalysts, but also enriches our fundamental understanding of the industrially important ODH process. Nevertheless, from the point of view of activity, selectivity, and stability, there is still room for improvement in catalytic performance. It should be pointed out that minimizing the cleavage of C-C bonds remains challenging under the oxidative dehydrogenation conditions. Furthermore, for such a novel catalytic system, there is an urgent need to develop new structure-tuning methods for optimizing the catalytic performance, and to clarify the roles of the B-O and B-OH groups in the alkane ODH reactions. Future studies are expected to involve theoretical simulations and the capture of reaction intermediates to disclose the reaction mechanism under the reaction conditions.

We thank Prof. De Chen (Department of Chemical Engineering, Norwegian University of Science and Technology) and Prof. Xue-Zhi Duan (State Key Laboratory of Chemical Engineering, East China University of Science and Technology) for the valuable discussions.