The Role of Composition and Phase upon the Lattice Nitrogen Reactivity of Ternary Molybdenum Nitrides

The reactivity of the lattice nitrogen in the filled β-Mn structured Co2Mo3N and the η-carbide structured Co3Mo3N and Fe3Mo3N has been investigated under 3:1 H2/Ar at temperatures up to 900 °C. The lattice nitrogen in Co3Mo3N was found to be reactive, as reported previously, whereas Co2Mo3N was shown to be stable up to 800 °C. Upon H2/Ar treatment at 900 °C, the Co2Mo3N, Co3Mo3N and Fe3Mo3N phases decomposed. These results suggest that both metal composition and phase have an influence on the bulk lattice nitrogen reactivity of the ternary nitrides.


Introduction
Ammonia production is of high importance as it can be used as a precursor to form synthetic fertilisers. The industrial route for producing ammonia is via the Haber-Bosch process and this is stated to sustain around 40% of the world population [1]. However, when considering the entirety of the process, including the feedstream production, it accounts for 1-2% of the world energy demand [2]. One possible way to overcome this disadvantage is the development of a process that could operate under more moderate conditions and facilitate localised and sustainable production. However, this would require the development of a more active ammonia synthesis catalyst. The η-carbide structured Co 3 Mo 3 N has a high ammonia synthesis activity, especially when doped with a low percentage of Cs + and is reported to have a higher rate than the iron-based catalyst used in the Haber Bosch process [2]. One possible explanation for the high activity of this nitride has been proposed by a volcano relationship, where the combination of Co and Mo gives a material with an almost optimal nitrogen binding energy and results in a high turnover frequency for ammonia synthesis [3]. In this proposal, the lattice nitrogen is suggested to have no active role and CoMo is the active phase. The role of the lattice nitrogen is stated to be to ensure that the correct structural ordering occurs such that the (111) face containing both Co and Mo is exposed. One alternative explanation for the activity of Co 3 Mo 3 N is that it proceeds via a Mars-van Krevelen mechanism and hence, the lattice nitrogen is reactive. In this mechanism, the lattice nitrogen is directly hydrogenated to yield ammonia, generating a transient lattice vacancy, which is then replenished by gas-phase nitrogen. This proposed explanation has been supported by experimental work, where an investigation into the isostructural Co 3 Mo 3 C phase showed that the carbide required a higher temperature than Co 3 Mo 3 N before activity was exhibited and nitridation of the lattice carbon occurred during the reaction [4]. Furthermore, nitrogen isotopic exchange studies have shown that a substantial percentage of the lattice nitrogen in Co 3 Mo 3 N can be exchanged, depending on the pre-treatment conditions used [5]. In addition, computational modelling studies have suggested that an associative N 2 activation pathway on Co 3 Mo 3 N may occur at nitrogen lattice vacancies [6] and that there are significant lattice nitrogen vacancies for this nitride at temperatures relevant for ammonia synthesis [7]. This mechanism differs significantly from the widely proposed view of dinitrogen dissociation being the rate determining step in relation to promoted iron-based ammonia synthesis catalysts, although for that system Spencer has pointed out that under industrially relevant conditions ammonia synthesis cannot be described in terms of a single rate determining step and that hydrogenation of adsorbed nitrogen species and ammonia desorption play a more significant role [8].
Related to the reactivity of lattice N, Co 3 Mo 3 N is also of interest due to its ability to lose half of its lattice nitrogen to form Co 6 Mo 6 N, where the remaining lattice nitrogen relocates from the 16c Wyckoff site to the 8a site [9,10]. The original nitride can then be regenerated from Co 6 Mo 6 N under either N 2 /H 2 or pure N 2 [9,11]. Therefore, the lattice nitrogen in Co 3 Mo 3 N has been shown to be highly reactive. This is not the case for the filled β-Mn structured Ni 2 Mo 3 N, where the bulk lattice nitrogen has been shown to be unreactive, even at temperatures of 900 °C under 3:1 H 2 /Ar [12]. Furthermore, the loss of lattice nitrogen from the η-carbide structured Fe 3 Mo 3 N has been reported to be minimal under 3:1 H 2 /Ar up to 800 °C [14,15]. A comparison with the η-carbide structured Ni 3 Mo 3 N, which is isostructural with both Co 3 Mo 3 N and Fe 3 Mo 3 N, would be of further interest. However, this phase has not yet been successfully prepared even though, as documented elsewhere, there have been incorrect reports of its existence published in the literature [15]. Nitridation of a NiMoO 4 precursor, Ni 3 Mo 3 C or Ni 6 Mo 6 C instead yields a mixture of Ni metal and Ni 2 Mo 3 N [12,13]. The filled β-Mn structured Co 2 Mo 3 N is a known phase and Adamski et al. have reduced a mixture containing both Co 2 Mo 3 N and Co 3 Mo 3 N phases under pure hydrogen at 700 °C. Co 3 Mo 3 N was observed to have transformed to Co 6 Mo 6 N, as would be expected for this phase [9,10] and the lattice nitrogen in Co 2 Mo 3 N appeared to be less reactive under these conditions.
The basis of the current study is to establish the potential role and structure/composition activity dependence of lattice nitrogen reactivity in terms of ternary metal nitrides. The analysis of the lattice nitrogen reactivity of nitrides through the comparison of their ammonia synthesis rates under Ar/ H 2 would give an insight into the structure-activity and composition-activity relationships. Therefore, an investigation of the η-carbide structured Fe 3 Mo 3 N has been performed to provide a comparison with the isostructural Co 3 Mo 3 N. This information will give an insight into the role metal composition has on the lattice nitrogen reactivity and it is hoped that a more thorough understanding of this aspect may be applied to the further development of novel ammonia synthesis catalysts.

Preparation of Fe 3 Mo 3 N
Iron molybdenum oxide was prepared by a similar procedure as detailed by Bem et al. [17]. 0. 25

Preparation of Co 2 Mo 3 N
Cobalt molybdenum oxide was prepared by using a modified form of the Pechini method as described by Bion et al. [18]. The necessary amounts of ammonium molybdate tetrahydrate ((ΝΗ 4 ) 6 Μο 7 Ο 24 .4Η 2 Ο, Fluka Analytical, puriss. p.a., ACS reagent, ≥ 99.0%) and cobalt (II) nitrate hexahydrate (Co(NO 3 ) 2 .6H 2 O, Alfa Aesar, ACS, 98.0-102.0%, crystalline) were added to a 10% aqueous solution of nitric acid to give a 2:3 ratio of Co:Mo. Citric acid monohydrate (C 6 H 8 O 7 . H 2 O, Sigma Aldrich, ACS reagent, ≥ 99.0%) was then added to the mixture. The mixture was stirred at room temperature, until the starting materials had been dissolved. Subsequently, the mixture was evaporated off at 70 °C, until a red coloured gel had formed. The gel was dried in an oven overnight at 120 °C and was then calcined in air at 500 °C for 2 h. Co 2 Mo 3 N was prepared via nitridation of the mixed oxide precursor under 3:1 H 2 /N 2 gas mixture at 700 °C for 3 h with a flow rate of 60 mL/min. A heating ramp rate of 10 °C/min was used to reach 700 °C.

Preparation of Co 3 Mo 3 N
Cobalt molybdenum oxide (CoMoO 4 ) was prepared by the reaction of the necessary amounts of ammonium molybdate tetrahydrate ((ΝΗ 4 ) 6 Μο 7 Ο 24 .4Η 2 Ο, 99.98%, SigmaAldrich) with cobalt (II) nitrate hexahydrate (Co(NO 3 ) 2 .6H 2 O, > 98%, Sigma-Aldrich). The starting materials were dissolved separately in either 100 mL or 150 mL of deionised water, respectively. The solution of cobalt was then added dropwise to the molybdenum solution. The resultant mixture was heated to 80 °C until a purple solid had formed, then vacuum filtered and the purple precipitate was washed twice with deionised water and once with ethanol. The oxide powder (CoMoO 4 .nH 2 O) was dried overnight in an oven at 150 °C and was then calcined in air at 600 °C for 3 h. Co 3 Mo 3 N was prepared by nitridation of the oxide precursor under 3:1 H 2 /N 2 gas mixture (BOC, 99.98%) at either 700 °C, 800 °C or 900 °C for 4 h with a flow rate of 60 mL/min. A heating ramp rate of 10 °C/min was used. The resultant nitride was cooled to ambient temperature under 3:1 H 2 /N 2 . For clarity, the nitride prepared at the different temperatures will be referred to as either CoMoN700, CoMoN800 and Co 3 Mo 3 N900.

Lattice Nitrogen Reactivity Testing
Approximately 0.3 g of material was tested at ambient pressure under 3:1 H 2 /Ar (BOC, 99.98%) with a gas flow rate of 60 mL/min. The stability and possible lattice nitrogen reactivity of the prepared Co 2 Mo 3 N and Co 3 Mo 3 N material was investigated by testing the material under 3:1 H 2 /Ar at 700 °C, 800 °C and 900 °C. Fe 3 Mo 3 N was also examined at 900 °C under 3:1 H 2 /Ar for its lattice nitrogen reactivity. The decrease in conductivity of a 0.00108 M 200 mL solution of H 2 SO 4 , which the exit stream of gas flowed through, was used to determine the production of ammonia. At such high reaction temperatures (700-900 °C), ammonia is thermodynamically unstable with respect to decomposition into N 2 and H 2 .

Material Characterisation
Powder XRD patterns were obtained for the materials pre-and post-reaction by using a PANalytical X-Pert Pro Diffractometer (40 kV, 40 mA) with a monochromatised CuKα source (1.5418 Ǻ). The data was collected in the range of 5-85° 2θ with a total scan time of 51 min and a step size of 0.02°. Phases were identified by comparison with the JCPDS database. CHN analysis was undertaken by combustion using an Exeter Analytical Inc CE-440 elemental analyser. SEM images were taken by using a Philips XL30 environmental scanning electron microscope. The samples were first covered with a film of gold/ palladium to reduce charging. The microscope was also used to acquire Energy-Dispersive X-ray spectroscopy (EDX). The BET surface area of the pre-reaction Fe 3 Mo 3 N was measured using N 2 physisorption undertaken using a Quantachrome Quadrasorb evo Gas Sorption Surface Area and Pore Size Analyzer. Approximately 0.1 g of the material was degassed at 110 °C overnight under vacuum before the measurement.

Lattice Nitrogen Reactivity of Fe 3 Mo 3 N
As the η-carbide structured Co 3 Mo 3 N and filled β-Mn structured Ni 2 Mo 3 N have been shown to have different lattice nitrogen reactivities, it would be of interest to investigate whether the phase or metal composition has the larger impact on the reactivity. This could lead to the development and design of active nitrides for ammonia synthesis. Therefore, the lattice nitrogen reactivity of another η-carbide structured nitride, Fe 3 Mo 3 N, has been investigated in this work. As this nitride has the same crystal structure type as Co 3 Mo 3 N but a different metal composition, it may prove possible to gain an insight into the role of the metal composition-activity relationship. It has previously been reported that the lattice nitrogen of Fe 3 Mo 3 N is less reactive than in Co 3 Mo 3 N. Fe 3 Mo 3 N has been observed to be stable under 3:1 H 2 /Ar up to 800 °C, with a minimal loss of lattice nitrogen [13,14]. Furthermore, neutron diffraction analysis revealed that bulk nitrogen was not lost from Fe 3 Mo 3 N [14]. A higher temperature has been investigated in the present study to investigate whether the bulk lattice nitrogen in the Fe 3 Mo 3 N phase is reactive but requires a higher temperature than that observed for Co 3 Mo 3 N. The Fe 3 Mo 3 N appeared from XRD to be phase pure when prepared from ammonolysis of an iron molybdenum oxide precursor as evidenced in Fig. 1a. The nitrogen analysis (Table 1) showed that the nitride had a considerably higher nitrogen content than the calculated stoichiometric value for Fe 3 Mo 3 N (2.98 wt%). However, NH x species on the surface of the material, that  (Fig. 2). The material was observed to have a surface area of 13 m 2 /g, which is comparable with those reported in the literature for this nitride [19,20]. However, as the nitride was passivated prior to exposure to air, as is standard for nitride materials, the BET surface area will possibly not be representative of the area under the testing conditions. As expected, passivation of nitrides changes surface composition [21] and can lead to changes in surface area [22].
The reactivity of the lattice nitrogen in Fe 3 Mo 3 N was examined by treating the nitride at 900 °C under 3:1 H 2 /Ar at ambient pressure for 7 h. If the material operated via a Marsvan Krevelen mechanism for ammonia synthesis, a reduction in the nitrogen content might occur and additionally, there may be a transformation of the structure as was observed for the Co 3 Mo 3 N counterpart. The conductivity profile (Online Resource Fig. S1) implies that only a minimal amount of ammonia was produced under these conditions. A total of 35 μmol of ammonia was produced over the 7 h of the reaction. However, it must be noted that ammonia is thermodynamically unstable at this temperature and will decompose [23] and, therefore, the loss of lattice nitrogen in the final form of N 2 will be more favourable. The post-reaction XRD pattern shows that the nitride had decomposed to Mo metal and an iron nitride phase under these reaction conditions (Fig. 1b). It appears that trace amounts of Fe 3 Mo 3 N were still present in the material as reflections were observed at 42° and 72° 2θ. The nitrogen analysis confirms the loss of lattice nitrogen from this material as there was a significant decrease post-reaction ( Table 1). The percentage of the overall nitrogen that was removed from Fe 3 Mo 3 N and was converted to ammonia was ca. 6%, with the rest of the nitrogen presumably being lost in the form of N 2 . SEM analysis showed that the morphology was retained after the decomposition of the material compared to pre-reaction (Fig. 2).
The lattice nitrogen in Fe 3 Mo 3 N was much more reactive at 900 °C than for the filled β-Mn structured Ni 2 Mo 3 N, which was shown to be stable at this temperature [12]. Therefore, this possibly suggests that metal composition has a major impact on the activity.

Lattice Nitrogen Reactivity of Co 2 Mo 3 N
The ability of the Co 3 Mo 3 N phase to lose bulk lattice nitrogen at 700 °C under H 2 /Ar [24], contrasted with Fe 3 Mo 3 N as discussed above, suggests that this is possibly due to the CoMo composition. Therefore, it was of interest to examine the filled β-Mn structured Co 2 Mo 3 N phase, which has the same metal composition but different crystal structure type. Comparing the lattice nitrogen reactivity of Co 2 Mo 3 N with the η-carbide structured Co 3 Mo 3 N might provide an insight into the crystal structure-activity relationship. To investigate this potential relationship, attempts were made to prepare the pure phase Co 2 Mo 3 N. The modified Pechini method was chosen for the synthesis process as it has previously been used to synthesis pure phase filled β-Mn structured Ni 2 Mo 3 N [18]. The filled β-Mn structured Co 2 Mo 3 N was prepared via nitridation of the oxide precursor at 700 °C for three hours under 3:1 H 2 /N 2 . This was found to be the optimum preparation temperature and duration to give the highest amount of the desired nitride. The preparation of the pure phase nitride was unsuccessful, with instead a mixture of Co 2 Mo 3 N, Co 3 Mo 3 N and β-Mo 2 N 0.78 being formed (Fig. 3). This material contained 3.82 wt% nitrogen from elemental analysis as presented in Table 2 (the expected stoichiometric nitrogen content of Co 2 Mo 3 N is 3.34 wt%). The SEM images of the mixed Co 2 Mo 3 N show that the material had a porous and smooth surface (Fig. 4a). Although the material  Fig. S2). A total of 37 μmol of ammonia was formed over the 7 h of reaction. The conductivity profiles for the 800 °C and 900 °C reactions (Online Resource Figure S2) show that the rates were non-steady state and the production of ammonia was minimal. The total amount of ammonia formed was 32 μmol and 22 μmol over the 7 h at 800 °C and 900 °C, respectively. These results agree with the limited stability of ammonia under these conditions and therefore, it would be expected that most of the lattice nitrogen would be lost in the final form of N 2 . The Co 3 Mo 3 N reflections were shown to have shifted to higher 2θ angles post 700 °C reaction (Fig. 5a), suggesting that there was a decrease in the lattice nitrogen content of this phase, which is to be expected when it transforms to the Co 6 Mo 6 N phase [13]. Due to the degree of overlap between the reflections of Co 6 Mo 6 N and Co 2 Mo 3 N this is somewhat difficult to observe. However, this was most clearly noticed when comparing the pre-and post-reaction reflection at approximately 35.5° 2θ, where there is no overlap of reflections. The shift of Co 3 Mo 3 N reflections to higher  2θ values was also observed post 800 °C reaction (Fig. 5b). The lattice nitrogen in Co 2 Mo 3 N appears to be unreactive at 700 °C and 800 °C under H 2 /Ar as there was no shift in the peak positions for this phase compared to pre-reaction (Fig. 5a, b) as clearly seen by the peak at approximately 45° 2θ. These results agree with the previously published work of Adamski et al. who noticed that under pure hydrogen at 700 °C Co 3 Mo 3 N transitioned to Co 6 Mo 6 N, while Co 2 Mo 3 N remained stable [16]. β-Mo 2 N 0.78 was believed to have been reduced to molybdenum metal during the reaction at both 700 °C and 800 °C (Fig. 5a, b). As there are not any unique reflections for Mo, the presence of this phase cannot be completely established. However, it has been previously reported that molybdenum nitride is reduced to Mo under 3:1 H 2 /Ar at 700 °C [14] and therefore, it would be expected to occur. The lack of reactivity of the lattice nitrogen in the filled β-Mn structured Co 2 Mo 3 N under H 2 /Ar up to a temperature of 800 °C has also been observed for the isostructural Ni 2 Mo 3 N and, as discussed above, η-carbide structured Fe 3 Mo 3 N [12,14]. Although Fe 3 Mo 3 N and Co 3 Mo 3 N have the same crystal structure type, the lattice nitrogen reactivity is different between these two nitrides under H 2 /Ar at 700 °C and 800 °C, suggesting that metal composition is more important than the structure type. However, Co 3 Mo 3 N and Co 2 Mo 3 N, have different stabilities suggesting that the phase also has an impact. Therefore, there appears to be a complex relationship between metal composition and phase which controls the lattice nitrogen reactivity of the ternary nitrides. In both the filled β-Mn structure and η-carbide structure, the local nitrogen environment is similar with the lattice nitrogen being coordinated to six Mo species in each structure. When Co 3 Mo 3 N transforms into Co 6 Mo 6 N, the residual lattice nitrogen relocates from the 16c site to the 8a site, an apparently less reactive site from which it is not lost. The XRD pattern of the "Co 2 Mo 3 N" material post 900 °C reaction shows it has decomposed to α-Co, Co 0.08 Mo 0.92 and either a cobalt nitride phase or a cobalt molybdenum alloy (Fig. 5c). Fe 3 Mo 3 N was shown to have a similar decomposition as Co 2 Mo 3 N and Co 3 Mo 3 N under these conditions. This is again in contrast to the filled β-Mn structured Ni 2 Mo 3 N, which was stable at 900 °C [12]. As expected, the nitrogen analysis showed that the nitrogen content decreased significantly post H 2 /Ar reaction for all three temperatures compared to pre-reaction as seen in Table 2. As the post 700 °C reaction material consisted of Co 2 Mo 3 N, Co 6 Mo 6 N and Mo, the stoichiometric percentage in this material is expected to be 1.92 wt% nitrogen, if it is assumed that the phases have a 1:1:1 ratio. The obtained value of 2.07 wt% agrees quite well with this value. For the post 800 °C reaction material, based upon the XRD pattern, it would be expected that the loss of nitrogen from this material is predominantly from the Co 3 Mo 3 N and β-Mo 2 N 0.78 phases. The significant loss of nitrogen from the post 900 °C material agrees with the observed decomposition during the reaction. The percentage of the overall nitrogen that was removed from the material and was converted to ammonia was ca. 25%, 24% and 13% for the 700 °C, 800 °C and 900 °C reactions, respectively. The SEM images in Fig. 4 show that the morphology for all three post H 2 /Ar reaction materials were similar to the pre-reaction material. Therefore, the decomposition of the material during the 900 °C reaction did not have a major effect on the morphology. However, the material appeared to be more porous following 900 °C reaction. Other materials have been reported to have a porous nature post-reaction when they decomposed under 3:1 H 2 /Ar [25].

Co 3 Mo 3 N Prepared by Nitridation with N 2 /H 2
The η-carbide structured Co 3 Mo 3 N is traditionally prepared in a pure form by temperature programmed ammonolysis of a CoMoO 4 precursor [13,24,26]. However, the formation of nitrides via temperature programmed ammonolysis has several disadvantages, including reduction in efficiency in the process due to large amounts of heat transfer and problems arising due to using ammonia at a large scale [27]. Therefore, although the bulk lattice nitrogen in Ni 2 Mo 3 N appears to be relatively unreactive, this nitride has an advantage in that it is able to be prepared under N 2 /H 2 at 700 °C [18]. Co 3 Mo 3 N appears to be uniquely active with regards to its lattice nitrogen reactivity which could have interesting applications as a nitrogen transfer material. It would therefore be of interest to prepare Co 3 Mo 3 N under N 2 /H 2 . In this work, attempts were made to prepare Co 3 Mo 3 N directly by nitridation of the oxide with N 2 /H 2 at 700 °C, 800 °C or 900 °C. The direct N 2 /H 2 nitridation process at 700 °C resulted in a mixed phased material of Co 2 Mo 3 N with minor amounts of Co 3 Mo 3 N and β-Mo 2 N 0.78 as can be observed in Fig. 6(a). As stoichiometric amounts of cobalt and molybdenum were used in the preparation, it appears that there is unidentified Co present in the material. Cobalt metal may be present but undetected due to peak overlap at 44° 2θ, as previously suggested by Adamski et al. [16]. The preparation at 800 °C results in a mixture of Co 3 Mo 3 N and Co 2 Mo 3 N, with Co 3 Mo 3 N being the majority phase (Fig. 6b). A small reflection at 44° 2θ is observed, suggesting that the excess cobalt is present in its elemental form. Preparation at the highest temperature of 900 °C produces the pure phase highly crystalline Co 3 Mo 3 N (Fig. 6c). Adamski et al. have proposed that the oxide precursor is first transformed to Co 2 Mo 3 N and a molybdenum nitride phase, before forming the Co 3 Mo 3 N phase [28]. Therefore, this may be an explanation for why the pure phased Co 3 Mo 3 N is only formed at high temperatures under N 2 /H 2 . This result shows promise for forming Co 3 Mo 3 N under nitridation conditions. It appears that there is a significant difference in morphology between the cobalt molybdenum nitrides as observed in Fig. 7. The material prepared at 700 °C (CoMoN700) consists of small solid particles. The morphology of the material prepared at 900 °C (Co 3 Mo 3 N900) was similar to the expected needle structure for Co 3 Mo 3 N that is observed when it is prepared under ammonolysis [4,29]. The nitrogen analysis (Table 3) mainly agrees with the assignment for the nitrides prepared at the different temperatures. CoMoN700 has the highest nitrogen content. The nitrogen content of the material prepared at 800 °C (CoMoN800) is below that expected for stoichiometric Co 3 Mo 3 N and the material was found from XRD to comprise Co 2 Mo 3 N, Co 3 Mo 3 N and Co. The nitrogen content of Co 3 Mo 3 N900 is in very close agreement with the expected stoichiometric nitrogen weight percentage for Co 3 Mo 3 N (2.93 wt%), further confirming the successful synthesis of the pure phase Co 3 Mo 3 N.

Conclusions
The bulk lattice nitrogen reactivity of the η-carbide structured Fe 3 Mo 3 N and Co 3 Mo 3 N and filled β-Mn structured Co 2 Mo 3 N has been investigated by testing the materials under Ar/H 2 . The metal composition-activity relationship was examined by comparison of the isostructural Fe 3 Mo 3 N and Co 3 Mo 3 N. Fe 3 Mo 3 N was shown to decompose to Mo and an iron nitride phase under 3:1 H 2 /Ar at 900 °C. In the literature, the bulk lattice nitrogen of Fe 3 Mo 3 N has been reported to be unreactive up to a temperature of 800 °C [13,   14]. Therefore, it appears that the Fe 3 Mo 3 N phase has a different reactivity to Co 3 Mo 3 N, even at high temperatures, possibly suggesting that metal composition is the dominant influence on the bulk lattice nitrogen reactivity. Additionally, with this in mind, an investigation of the filled β-Mn structured Co 2 Mo 3 N and the η-carbide structured Co 3 Mo 3 N has also been performed to provide an insight into a possible structure-activity relationship. The filled β-Mn structured Co 2 Mo 3 N could not be synthesised in the pure form, with instead the material being a mixture of Co 2 Mo 3 N and Co 3 Mo 3 N. The mixed phase Co 2 Mo 3 N and Co 3 Mo 3 N were examined for their lattice nitrogen reactivities between 700 °C and 900 °C under 3:1 H 2 /Ar. At 700 °C and 800 °C, the filled β-Mn structured Co 2 Mo 3 N was shown to be stable and η-carbide structured Co 3 Mo 3 N was observed to be more reactive by losing the majority of its lattice nitrogen. When reacted at 900 °C, the material decomposed to its metal constituents and the majority of the lattice nitrogen was removed, as was observed for Fe 3 Mo 3 N. It has previously been observed that the filled β-Mn structured Ni 2 Mo 3 N was stable at this temperature [12]. Overall, the lattice nitrogen reactivity of the ternary nitrides has been found to be complex. Additionally, pure phase η-carbide structured Co 3 Mo 3 N with needle morphology was synthesised under 3:1 H 2 /N 2 at 900 °C.