1.2 Surface Properties of the α and β Phases

Extending the earlier work by Chen and coworkers on β-Mo₂C (61), Illas and coworkers investigated the projected density of states (PDOS) of MoC_x phases (Figure 3) (97). The study highlighted the difference in the character of the fcc α-MoC_1–x and hcp β-Mo₂C phases, with the former having more covalent character due to the presence of localised states in the DOS, and the latter being more metallic with a broad region of states around the Fermi level. A combination of charge density difference (CDD) iso-surfaces and electron localisation function (ELF) plot analysis spatially confirmed the electron distribution being more localised on the carbon for the α-fcc phase, while being spread between the atoms in the β phase. Analysis of the low index planes, including non-polar and polar surfaces, of these phases showed significant differences work function with the β-Mo₂C (011) being the most stable with a ~3.4 eV work function.

Since this initial work, significant effort has been expended on theoretically determining the properties and stability of specific surface terminations of a range molybdenum carbide phases. Surface free energies of different stoichiometric and nonstoichiometric surfaces (i.e. molybdenum, carbon, mixed terminations) have been determined at different temperatures and carbon chemical potential (μ_C) of a range of molybdenum carbide phases. Computed Wulff constructions then allow for determination of particle morphology at different conditions. Saidi showed that the prevalence of the (011) surface increases in β-Mo₂C as the chemical potential is lowered to ca. –10 eV below which (021) and carbon terminated (100) begin to dominate (98). In comparable studies, Jin and coworkers (99) and recently Hao and coworkers (100) determined the surface stability and morphology of α-MoC_1–x. Generally, molybdenum-rich (carbon deficient) surfaces decreased in stability with increasing μ_C, the inverse being observed for carbon-rich surfaces, and no change being observed for stoichiometric surfaces. The result was the domination of the (311) surface at ‘conventional’ synthesis conditions, while proposed syngas conditions result in the emergence of (100) and (211) surfaces.

In addition to corroborating the Wulff constructions suggested above (Figure 4(a)), Paolucci and coworkers theoretically considered the influence of particle size and predicted crystallisation pathways of all major molybdenum carbide phases (79). A clear observation was that small particle size is a predominant factor in the thermodynamic stability of the α-MoC_1–x phase (Figure 4(b)). Further, the phase boundary between carbides and molybdenum metal changes with size and μ_C, which suggests that carbon vacancies may stabilise larger α-MoC_1–x particles, as supported by work from Hugosson et al. and Zaoui et al. (72, 101). While these theoretical studies provide insight into the thermodynamic stability of certain structures, kinetic aspects must also be considered, particularly for the synthesis of the α-MoC_1–x that frequently is formed from topotactic rearrangement of a suitable precursor phase (discussed in detail below). Finally, in a comprehensive theoretical study, Quesne and colleagues showed that significant surface reconstruction or rumpling was observed for low index planes, with higher relaxation being seen for less stable surfaces (102). The lowest energy surface was the (100) and the lowest work function (an indication of high reactivity during reductions) was the (110).

1.3 An Overview of the Catalytic Application of α-MoC_1–x

In the following section we will briefly discuss the different catalytic properties of α-MoC_1–x and β-Mo₂C phases within the context of key catalytic applications (53, 103). Specifically, reforming (56, 104, 105) and water-gas shift reactions for sustainable hydrogen production (45, 46, 57, 88, 106), CO₂ hydrogenation (31, 89) and the hydrogen evolution reaction (63, 107, 108). Given the focus of the article is α-MoC_1–x synthesis, the discussion below is not exhaustive and the reader is directed to specific reviews regarding the catalytic applications highlighted above.

1.4 Thermal Catalytic Applications of α-MoC_1–x

As Ma and coworkers have succinctly described in their recent review, the β-Mo₂C phase is historically the most studied for high-temperature reactions, such as hydrocarbon isomerisations, dehydrogenations and ammonia syntheses (57, 109). The thermodynamic stability, superior C–H bond activation and strong dissociative hydrogen adsorption are key parameters for these reactions. On the other hand, α-MoC_1–x has recently been shown to have superior catalytic performance at lower temperatures for reactions such as methanol reforming, water-gas shift and biomass utilisation (alcoholises of lignin (110), furfural and succinic acid hydrogenation (111, 112) and Kraft lignin decomposition (113)). The improved activity for these water-containing reactions is attributed to high O₂ and O–H bond activation in the α-MoC_1–x phase. A summary of reaction conditions and key findings from select studies regarding reforming and water-gas shift are given in Tables II and III respectively.

Several aspects of the α-MoC_1–x, (70) phase and how it differs from β-Mo₂C account for this improved O–H activation at low reaction temperatures. Firstly, it is important to note that α-MoC_1–x is suitable only for processes at low temperatures (below 300°C), specifically because it is a metastable phase that at elevated temperatures is likely to transform into the thermodynamically stable β-Mo₂C (57). The high surface area, compared to β-Mo₂C, clearly increases the proportion of active sites available during a reaction. While clearly advantageous, this does not account for the improved reactivity in water activation. An intrinsic increase in catalytic activity of the α phase in these reactions is due to the fact that α-MoC_1–x (0.25 < x < 0.35), with Mo:C ratio between 1.3–1.5, has a lower percentage of molybdenum, compared to β-Mo₂C, which has a Mo:C ratio of 2 (70). According to Rodriguez et al. (88) this reduced molybdenum content is beneficial in limiting the reactivity of active oxygen species to produce oxycarbides (89, 119). A reduction of the metal:carbon ratio results in a lower reactivity of the surface because fewer metal atoms are present on the surface and with more positive charge. As noted by Palma et al. this will prevent catalyst deactivation from activated water species as it facilitates the desorption of OH (and O) species (120). The excellent low-temperature water-gas shift reactivity of this catalyst is explained by this reduced water activation vs. the β-Mo₂C.

It has also been noted that α-MoC_1–x has reduced hydrogen adsorption strengths, C–H and C–O bond activation in comparison to β-Mo₂C. These changes in adsorption strengths are credited with the reasonable catalytic performance in the reverse water-gas shift reaction (CO₂ reduction to carbon monoxide) (34, 121). Gao et al. noted the high carbon monoxide selectivity in α-MoC_1–x and attributed it to its weak adsorption (121). Given the reverse water-gas shift reactivity and low methanation, CO₂ hydrogenation to methanol has become an area of great interest regarding α-MoC_1–x, as shown in Table IV.

The interplay between the reverse water-gas shift reaction, methanation and CO₂ hydrogenation was studied by Rodriguez and Illas using model polycrystalline catalysts under ultrahigh vacuum conditions (33, 89). It is reported that with δ-MoC (α-MoC_1–x) a reaction mechanism goes via a formate intermediate (COOH) with minimal C–O cleavage and results in reduced methane formation. It was observed that on β-Mo₂C (001) surfaces methane selectivity was much higher due to facile CO₂ (and carbon monoxide) activation and dissociation.

Theoretical and single studies have allowed for correlation of specific surface terminations with key elementary reaction steps in reforming and hydrogenation reactions. Taking Quesne’s observed stability of the (001) α-MoC_1–x surface (102), Illas and coworkers investigated a range of different pathways for hydrogen adsorption on α-MoC_1–x and other Group IV transition metal carbides. They showed that this surface is particularly favourable towards hydrogenation reactions with strong associated H* hydrogen adsorption energies vs. non-dissociative adsorption (i.e. a Kubas type interaction between hydrogen and surface). Moreover, α-MoC_1–x is particularly suitable in this sense since it presents the highest adsorption energies among the transition metal carbides, although these values are still moderate and comparable to late transition metals (127). Ge and coworkers considered methane adsorption, dissociation and coupling on the (111) cubic α-MoC_1–x and noted comparable properties to platinum (128).

Another useful property of α-MoC_1–x is the strong interaction with (noble) metals which allows for the production of a stable supported catalyst with multiple functionalities in one material. In this context, the MoC phase and its O–H bond activation can be combined with noble metals with complementary functions (like C–H or C–C cleavages) to produce high reactivity (56, 57, 129, 130). Several papers have noted that this is specific to α-MoC_1–x and that there is a significantly reduced interaction between metals, such as platinum or gold, and β-Mo₂C, resulting in no significant improvement in activity for the latter carbide phase (45, 88) (Figure 5).

1.5 Hydrogen Evolution Reaction

As a critical component of water splitting the hydrogen evolution reaction (HER) remains a topic of great interest and requires significant improvement in catalytic performance, ideally with little or no pgm content. Reactivity can be correlated to the hydrogen adsorption free energy (taking into account the catalysts dependence of the rate constant) (131, 132). Molybdenum carbides have been extensively studied for the acid/alkaline hydrogen evolution reaction as a potential replacement to platinum (129, 133). Numerous molybdenum carbide phases have been explored, with many studies finding conflicting or unclear preference for one phase over another (Table V). However, through excellent synthetic control of molybdenum carbide phase (vide infra) Leonard and coworkers have provided compelling evidence for β-Mo₂C being the most active (85). This observation that β-Mo₂C is the best phase for electrochemical hydrogen production has been further verified (59, 134). Saidi and coworkers have, using density functional theory (DFT) approaches (vide supra), shown that the best HER activity is observed on the (011) surface followed by the C terminated (001). The much higher H* adsorption strength on β-Mo₂C provides a clear explanation as to why the phase is superior for acidic based media. An open question remains regarding the potential of α-MoC_1–x in alkaline media, where water dissociation becomes a key reaction step (135). Interestingly some reports, such as that by Wang and coworkers on HER in acidic media, have also proposed a synergistic effect through the formation of α-MoC_1–x/β-Mo₂C heterostructures (136). Therefore, significant interest in α-MoC_1–x catalysts for this reaction remains.

In summary, a broad picture emerges in which the more metallic β-Mo₂C is considered superior for C–H bond activation and strong dissociative hydrogen adsorption, but with moderate surface area values. The phase is also considered thermally stable for use in high-temperature applications. In contrast α-MoC_1–x is considered to have notably better O–H bond activation, forming high hydroxyl group coverage in aqueous phase reactions or alkoxy intermediates in alcoholic media. In addition, α-MoC_1–x has been applied as a non-benign support for both platinum and first row transition metals, with very strong metal support interactions being observed leading to formation of single-atom catalysts. These properties, coupled with higher surface areas, present an opportunity to develop novel and interesting catalyst technologies that operate at relatively low temperatures and require water tolerance or activation.

The scientific and technical challenges that arise from pursuing such metastable phase are twofold. Firstly, given the metastable nature of the phase, possible bulk phase changes cannot be discounted during catalytic reaction or handling under ambient conditions. An extension of this is the possibility, as highlighted by Hargreaves and Alexander (53, 144), that the presence of oxygen could influence the surface or subsurface structure of α-MoC_1–x. Thus, the nature of the active catalyst may be different to that originally envisaged, complicating rational design. The second challenge is the successful synthesis of the α-MoC_1–x without forming significant quantities of more thermodynamically favourable phases, such as β-Mo₂C. The remainder of this review will focus on reported synthetic strategies, the associated challenges with such methodologies and their effectiveness.

2.1 Temperature-Programmed Synthesis of α-MoC_1–x

The method developed by Volpe and Boudart is reported in a pair of seminal papers (147, 148) and involves the initial synthesis of γ-Mo₂N, from the ca. 700°C heat treatment of MoO₃ in pure ammonia, referred to as ammonolysis. Following this, the γ-Mo₂N undergoes carburisation using CH₄/H₂ between 400–700°C to form highly pure α-MoC_1–x. The importance of the intermediate γ-Mo₂N for α-MoC_1–x synthesis has been discussed in detail in the following papers (147, 149, 150). The overall process from MoO₃ through the cubic γ-Mo₂N intermediate to the fcc α-MoC_1–x, is considered topotactic. The initial changes from orthorhombic MoO₃ to the cubic γ-Mo₂N structure involves only a ‘soft’ variation of the metal planes distance, and the final fcc cubic lattice of γ-Mo₂N is retained in the formation of cubic α-MoC_1–x (Figures 6 and 7). A lack of molybdenum rearrangement during this topotactic process is what is hypothesised to prevent the formation of the thermodynamically stable β phase (147, 149, 150). Interestingly the morphology of α-MoC_1–x is dictated by that of the original precursor phase (pseudomorphism), i.e. the morphology of orthorhombic MoO₃ is retained after synthesis (see microscopy in Figure 6). It has been observed that these platelet structures are secondary morphologies of very small (sub-10 nm) primary particles which provides α-MoC_1–x with dramatically greater porosity and surface area than MoO₃ (surface area of the oxide is often reported to be below 1 m² g^–1). These observations are corroborated by the much more recent theoretical studies by Paolucci and coworkers (vide supra) that highlight the thermodynamic stability of α-MoC_1–x at low particle size (79). What is not clear is the morphology of these primary particles in these original temperature programmed reactions.

While the process is scientifically elegant and conceptually simple, it has several practical challenges that make synthesis of α-MoC_1–x via this route non-trivial. Specifically, this concerns the ammonolysis step, which requires careful control of a number of parameters to facilitate the topotactic process, such as low heating rates and high space velocity of gases (151). The reader is directed to the review by Alexander and Hargreaves for a more detailed discussion on metal nitrides synthesis and characterisation (144). The challenging conditions of ammonolysis clearly present logistic issues, both at the laboratory scale (although this can be overcome with sufficient knowledge and investment) and in large-scale application. In addition to the more obvious issue around handling ammonia at elevated temperatures, control of flow dynamics and heat transfer issues are of concern in large-scale application. In the context of the synthesis of metastable carbides like α-MoC_1–x, different strategies can be employed to overcome these issues. These vary from alternative routes to synthesise γ-Mo₂N to avoiding nitride phases altogether.

Progress towards ammonolysis-free synthesis was made directly by Levi and Boudart, who found that adding transition metals like platinum to the starting molybdenum precursor was a successful way to produce α-MoC_1–x (70). Lee and coworkers further hypothesised that specific transition metals, such as platinum, palladium and nickel, facilitate, via dissociative hydrogen adsorption and spillover, the formation of a hydrogen bronze H_xMoO₃ (70, 149). This phase lowered the reduction temperature of Mo(VI) from 600°C, seen for pure MoO₃, to ca. 175°C and in the process produced a cubic oxycarbide MoO_xC_y phase, which provided the conditions for the topotactic transition to α-MoC_1–x (152, 153). Despite the observation that pseudomorphism from MoO₃ to α-MoC_1–x was limited when using transition metals, the surface areas reported were comparable (200 m² g^–1) to those from the Boudart γ-Mo₂N route (225 m² g^–1), making the process appear desirable for use in catalyst synthesis. Yet, it is possible that the presence of the transition metal impurities in the final product could be problematic, becoming sources of catalytic poisons or promoters of undesired catalytic side reactions. However, given that α-MoC_1–x is being actively used as a non-benign support for these very transition metals, as seen in work from Ma and coworkers on aqueous phase reforming of methanol (56), such problems may appear unfounded in certain circumstances. Indeed, Sun et al. recently showed that metal loaded MoO₃ transformed into metal/α-MoC_1–x with higher activity for water-gas shift than an unloaded carburised MoO₃ sample (154). Although the influence of the transition metal on activity cannot be deconvoluted from the molybdenum carbide phase produced, as the absence of the metal produced Mo₂C as opposed to α-MoC_1–x. Lastly, even if the transition metal is beneficial to catalytic performance, it is likely that most of said metal is ‘wasted’, from a catalytic perspective, in the bulk of the synthesised material and unusable for surface catalytic reactions (70, 149).

Later Bouchy et al. (152, 155–157) discovered that a prereduction of MoO₃ in pure hydrogen at 350°C prior to carburisation could produce α-MoC_1–x without requiring ammonolysis or transition metals. Through the same mechanism as shown with transition metals, hydrogen insertion into the oxide bulk to form a hydrogen bronze MoO_xH_y then facilitates the formation of the desired α-MoC_1–x (155). However, it was observed that reduction of MoO₃ under pure hydrogen resulted in a mixture of MoO_xH_y and MoO₂, which subsequently on carburisation produced a mixture of α-MoC_1–x and β-Mo₂C. In addition, the procedure initially used pure methane as the carburisation gas, which resulted in significant carbon deposition and a relatively low surface area of 90 m² g^–1. Replacement of pure methane with 9:1 H₂:CH₄ mixture produced a carbon deposit free catalyst with a surface area of 179 m² g^–1. Unfortunately, no evidence for the MoC_1–x:Mo₂C ratio was provided for this higher surface area material and only a statement to the effect that Mo₂C was a minor phase was provided. While the Mo₂C was the minor phase it is unproven that this specific methodology can produce a pure single phase MoC_1–x catalyst.

Several researches have studied the influence of using different n -C_nH_{2n +2x} carburising agents (152). The group of the late Malcolm Green reported that α-MoC_1–x could be synthesised directly from MoO₃ when the carbon chain is increased from C₁ to C₄ (158, 159). Through the use of extended X-ray absorption fine structure (EXAFS) and ¹³C nuclear magnetic resonance (NMR) they showed that ethane resulted in a mixed MoC_1–x:Mo₂C material while butane produced a poorly crystalline material comprised of only α-MoC_1–x. Although intermediate phases were not reported by Green, it was shown in analogous experiments that an MoO_xH_yC_z phase was formed during heat treatment of H₂/C₄H₁₀ (157). A summary of the proposed topotactic synthesis via MoO_xH_yC_z precursors and XRD evidence is shown in Figure 8.

Unfortunately, the reported surface area of α-MoC_1–x produced via Green’s route was only 35 m² g^–1, although it should be noted that no attempt was made to identify the presence of carbon deposits or, if present, remove them. Following this work a host of different carbon sources (151), such as C₂H₂ (160), C₂H₆ (159, 161, 162), C₃H₈ (163), C₄H₁₀ (35), CO (150), C₇H₈ and C₇H₁₆ (164) have been used to produce different carbide phases. Bouchy and coworkers showed that this technique always produced MoO₂ in addition to the oxycarbide (152, 156, 165). Perret and coworkers showed in TiO₂/ZrO₂ supported MoC_1–x for succinic acid hydrogenation, that changing the carburisation temperature and the hydrocarbon percentage in the hydrocarbon/hydrogen mixture altered the carbon content in the α-MoC_1–x, with a consequent influence on the catalytic activity (112).

2.2 Temperature-Programmed Synthesis of α-MoC_1–x Using Carbon Supports and Amine Precursors

Temperature-programmed reactions to produce molybdenum carbides, including α-MoC_1–x, are (as stated above) dependent on molybdenum precursor structure and the source of carbon used. Recently, as a natural progression of this statement, the technique has been successfully adapted to make use of carbon and nitrogen sources already present in the molybdenum precursor.

One concept is the addition of an excess of solid carbon phases which then facilitate carburisation in a range of atmospheres, including those excluding a carburisation gas or hydrogen. Through impregnation of a molybdenum precursor onto a carbon, a final catalyst of MoC_x/C can be produced where the carbon acts like a conventional support structure. Several different support structures have been employed, such as activated carbons (166) graphene carbon nanofibres (41), carbon nanotubes (167, 168), nanowire (121), doped graphene oxide nanosheet (169), carbon nitride g-C₃N₄ (141) and glucose (133). Li and coworkers (110, 113) demonstrated that activated carbon impregnated with ammonium molybdate could be converted into either α-MoC_1–x or Mo₂C under methane/hydrogen, pure hydrogen or under nitrogen only (XRD evidence provided in Figure 9). A general trend on heating of ammonium molybdate → MoO_x/MoO₂ → α-MoC_1–x → Mo₂C was seen under all gas atmospheres, with methane/hydrogen producing pure α-MoC_1–x at the lowest temperature. Interestingly, under nitrogen, mixed phases of α-MoC_1–x and Mo₂C were produced and found to be the most active catalysts for the dehydrogenation of lignin. It is interesting to note that no intermediate oxycarbide or hydrogen bronze phases are observed or discussed in the work, making it unclear as to how α-MoC_1–x was formed.

Sun and coworkers performed an interesting analogous experiment where ammonium molybdate was impregnated onto C₃N₄. After heat treatment under an inert atmosphere γ-Mo₂N was formed prior to α-MoC_1–x (141). Under the conditions used by this group an analogous temperature programmed reaction using an activated carbon only produced β-Mo₂C.

Another emerging strategy for bulk α-MoC_1–x synthesis has been the utilisation of polymeric or composite molybdenum compounds as precursors during temperature programmed synthesis. Leonard and coworkers pioneered the controlled synthesis of a host of molybdenum carbide phases, including α-MoC_1–x, through thermal decomposition of an amide-molybdenum oxide composite (85, 170). These composites were synthesised from the simple precipitation of an aqueous ammonium molybdate and amine solution by addition of hydrochloric acid to pH 3. The composite materials were then heat treated under argon up to 675°C to form α-MoC_1–x with excellent reported crystallinity, but no reported surface area. However, their successful employment in HER suggests that the active surface areas must be reasonable. The choice of amine, amine:molybdate ratio and heat treatment temperature dictated the morphology and phase of carbide produced (α,β,η,γ). As far as we are aware this is the only reported synthetic method of controlling α-MoC_1–x morphology and therefore has significant potential to experimentally explore surface specific reactivity as hypothesised from theoretical studies. Thermogravimetric analysis (TGA) of the composites indicated two main weight losses, the first being the decomposition of the amine and MoO₂ formation and the second carburisation to form α-MoC_1–x. Interestingly no evidence was sought to identify oxycarbide or γ-Mo₂N and its requirement to produce the α-MoC_1–x phase. A number of variations on the amine synthesis route have been reported using ammonium molybdate as precursor with different amino-containing compounds such as dopamine and aniline (121, 136, 171). Tang and Asefa et al. (86) proposed a similar synthesis method which is claimed to produce highly crystalline nitrogen-doped α-MoC_1–x or η-MoC nanosheets by drying a solution in deionised water of melamine, cyanuric acid and ammonium and pyrolysing this under a nitrogen atmosphere at temperatures between 450–850°C. α-MoC_1–x formed at lower temperature pyrolysis (550–650°C) while η-MoC formed at higher temperature (750°C or 850°C).

The use of alternative molybdenum precursors has also been reported, usually in combination with an amine but not exclusively. Ruddy and coworkers used a gel made from the evaporation of ethanol from a solution containing MoCl₅ and 4-chloro-ortho -phenylenediamine, without any pH adjustment, which on heat treatment under nitrogen at 850°C produced α-MoC_1–x (172). Inspired by the work of Hashimoto, Nakanishi and coworker who synthesised molybdenum carbonitride from a molybdate-polydiaminopyridine compound (173), Bayati et al. synthesised α-MoC_1–x via an analogous polypyrole/phosphomolybdate/graphene oxide composite (169). The authors produced high surface area catalysts of 150–560 m² g^–1, comprised of a mixture of α-MoC_1–x and graphitic carbon, after carburisation under an inert atmosphere at 900°C. Interestingly, a slight shift in the reflections of the observed fcc phase were taken to indicate residual molybdenum nitride, hypothesised to have been produced from the decomposition of the pyrrole. Bayati and coworkers noted, as with many other carbide preparation methods, that carbon deposits form during synthesis (169). Interestingly, they noted that these carbonaceous deposits encapsulated α-MoC_1–x particles and considered this to be advantageous as a method of preventing α-MoC_1–x sintering during preparation and use as an ammonia electro-oxidation catalyst. We provide a note of caution regarding such opportunistic C@α‐MoC_1–x structures, as control over carbon layer thickness and porosity, which would be essential for practical catalytic use, requires controlling.

2.3 Temperature-Programmed Synthesis of α-MoC_1–x With Non-Carbon Supports and Additives

The use of the heteropolymetalate, phosphomolybdate (H₃Mo₁₂PO₄₀·12H₂O), as a precursor to α-MoC_1–x has also been reported in the elegant hard templating synthesis of a mesoporous hydrogen evolution catalyst (129). A summary of the preparation is shown in Figure 10. H₃Mo₁₂PO₄₀.12H₂O was impregnated into the mesoporous silica KIT-6 and then carburised under 2:1 CH₄:H₂ at 650°C to form α-MoC_1–x, subsequently the KIT-6 was removed by hydrogen fluoride. Again a fine carbon overlay was formed and considered advantageous, even to the point where further etching of the α-MoC_1–x produced an exciting ordered carbon framework (129, 174). Besides the elegant mesoporous structures, the work suggested that an intermediate molybdenum nitride phase was not required to access α-MoC_1–x when using the phosphomolybdate precursor. Further, no pretreatment in hydrogen was required. It is apparent that use of comparable carburisation conditions with the conventional MoO₃ precursor would exclusively produce the thermodynamically stable β-Mo₂C phase. Unfortunately, neither publication performed a simple control synthesis of phosphomolybdate carburisation without the additional complexity of confined pore architectures.

As noted in the preceding paragraph, strategies for the formation of mesoporous or nanostructured α-MoC_1–x make use of composite structures that include silica and α-MoC_1–x. Work at the start of the millennium by Hamid, coworkers and Haldor Topsøe showed that α-MoC_1–x could be formed within ZSM‐5 pores via Green’s carburisation with H₂/C₄H₁₀, but not via a pure hydrogen treatment. These confined α-MoC_1–x/ZSM-5 catalysts were shown to be superior catalysts, compared to β-Mo₂C/ZSM-5, for methane to aromatic reactions (157). Ruddy and coworkers identified the challenge of incorporating molybdenum-amine precursors into mesoporous structures, which they overcame through the use of surface modified SBA-15 to produce 2 nm α-MoC_1–x from an inert gas heat treatment. It was noted that the acid sites of SBA‐15 in close proximity to α-MoC_1–x produced a highly active catalyst for acetic acid deoxygenation (172).

Specific influence on carburisation of the confinement of molybdenum precursors in mesoporous structures is limited. However, one interesting study is provided by Wang and coworkers, who used an amine synthesis route with varying weight percent silica nanospheres being present during polymerisation (during pH adjustment) (136). These spheres were suggested to form packed structures with the molybdenum-polydopamine within the voids (Figure 11). Using the same 800°C argon heat treatment the fraction of α-MoC_1–x:Mo₂C changed with the weight percent silica. The absence of silica resulted in α-MoC_1–x being the only observable phase by XRD, X-ray photoelectron spectroscopy (XPS) and Raman spectroscopy, while increasing the weight percent silica resulted in significant β-Mo₂C content being observed (Figure 12). The authors hypothesise that this phenomenon was caused by confinement reducing the concentration of carbonaceous deposits during the heat treatment which then favours β-Mo₂C formation. Despite the lack of evidence to support the hypothesis, the observation is interesting and warrants further study. The same group boarded the concept of additives influencing carbide phase formation by studying the doping of the molybdenum-polydopamine with boric acid and found that its inclusion had a modest effect on suppressing Mo₂C formation (171).

Sun and coworkers report an alternative study, with the addition of zinc oxide to a ammonium molybdate and platinic chloride precursor (104). In this instance, traditional carburisation gases hydrogen and methane were employed and the addition of platinum during the carburisation will, as discussed above, influence carbide phase formation (115). It was observed that weight percent zinc up to ~30 wt% increased α-MoC_1–x formation, while above this threshold XRD reflections of Mo₂C began to dominate. In all instances the reported surface areas were low (maximum of 40 m² g^–1) compared to α-MoC_1–x produced by conventional temperature programmed synthesis (>200 m² g^–1) and carbon deposits were not checked for.

Roy et al. (175) have proposed a synthesis of the α-MoC_1–x by starting from Mg(MoO₄) and Zn(MoO₄), compounds prepared from the molybdic acid with MgO and ZnO. Avoiding the ammonolysis, a mixture of 20% carbon monoxide in hydrogen at 700°C was used in a so called ‘sacrificial support method’, in which the formation of magnesium/zinc oxide-phase blocks the formation of the MoO₂ intermediate and consequently inhibits β-Mo₂C formation. The residual zinc oxide and magnesium oxide were then leached from the catalyst using hydrochloric acid.

2.4 Alternative Preparation Routes to α-MoC_1–x

Drawbacks of the Levi Boudart method of temperature-programmed reduction (TPR), based around challenges associated with scale-up of ammonolysis, carbon deposits from using amine polymer precursors and the issues of slow-ramping high-temperature procedures have led to a number of alternative synthesis methods being developed. These processes have attempted to eliminate, or minimise phase sensitivity to, high-temperature carburisation in the synthesis of ammonolysis α-MoC_1–x.

Through an electrical wire-explosion technique, Kim et al. published a room temperature route to synthesise ultrafine α-MoC NPs (139). The process used molybdenum wire, which was electrically superheated, resulting in a proposed evaporative explosive process, the products of which were dispersed in oleic acid (chosen for its high carbon content). Nanoparticles of α-MoC were formed by this process alone, but according to transmission electron microscopy (TEM) and XPS, these sub-20 nm particles were embedded within a carbon shell. Heat treatment under a hydrogen environment at 600°C was then required to remove carbonyl functionality and reduce carbon content from 28 wt% to 17 wt%. However, the α-MoC_1–x remained encapsulated within a graphitic shell. Despite this the catalysts were active for the HER. Platinum was then deposited on these MoC_1–x@C core-shell catalysts to further improve their performance through the ethanol oxidation method, which utilises localised electrons, formed from ethanol oxidation to acetic acid, to reduce the platinic chloride to metal (176).

Separately, Cao et al. (177) and Du and coworkers (178) have grown α-MoC_1–x with notable high surface area through an electrospinning technique followed by a pyrolysis of the MoC precursor. The former work by Cao et al. focused primarily on electrospinning as a method of producing a zinc-doped precursor that facilitated the formation of the α-MoC_1–x phase during pyrolysis at 800°C. Analogous samples without zinc doping formed the Mo₂C phase at the equivalent temperature, leading to the authors to hypothesise that zinc doping supresses the phase transition. The precursor was synthesised by passing a solution of zinc acetate, molybdenum acetylacetonate, polyvinylpyrrolidone in dimethylformamide through a needle with an applied potential of 15 kV. The pyrolysis process in hydrogen/argon at 800°C was proposed to cause zinc oxide reduction and sublimation, which resulted in exciting changes in final surface area depending on Zn:Mo molar ratio used, with a Zn:Mo of 8 producing the highest specific surface area of 418 m² g^–1. Lastly, the authors demonstrated that the synthesised α-MoC_1–x was stable up to 1000°C that allowed for the preparation of remarkably high crystalline material. Du and coworkers utilised a similar electrospinning process but used a phosphomolybdic precursor and co-sprayed platinum acetylacetonate to synthesise single atom Pt/α-MoC_1–x nanorods.

Malmstadt and coworkers (34) have recently proposed, in a highly cited paper, a continuous microfluidic (mF) synthesis of α-MoC_1–x, which is particularly notable for not requiring any high-temperature (maximum temperature used <475°C) carburisation or pyrolysis. The reported method exploits the thermolytic decomposition of the molybdenum precursor Mo(CO)₆, which has been reported previously to produce α-MoC_1–x when heated under vacuum (179, 180) with partial pressures of carbon monoxide and hydrogen. The process was first adapted by Malmstadt and coworkers by thermolytic decomposition of Mo(CO)₆ in a solution with oleylamine and 1-octadecene. The formation of a Mo(CO)₄(OAm)₂ species was reported to form at ca. 200°C followed by decomposition at >240°C into α-MoC_1–x. This process was then further adapted into a continuous flow millifluidic reactor (Figure 13) consisting of the Mo(CO)₆ solution being pumped through a borosilicate reactor, prepressurised with nitrogen and immersed in sand bath at 320°C. The product of the reaction was quenched in hexane before the MoC_1–x was collected by centrifugation. The produced MoC_1–x were characterised by molybdenum K-edge X-ray absorption fine structure (XAFS), XRD, XPS and X-ray pair distribution function (PDF, shown in Figure 14) to demonstrate that partially disordered nanostructured α-MoC_1–x capped with oleylamine were formed. These particles were found to be stable under air without passivation, a hereto undiscussed point. The particles were then supported on activated carbon and shown to be highly active and selective for CO₂ Fischer-Tropsch chemistry. The reported procedure represents, in our opinion, the greatest departure from the conventional process originally set out by Boudart, with α-MoC_1–x being produced at significantly lower temperature and not requiring a temperature-controlled ramping process.