1 Introduction

Metal nitrides constitute an interesting class of heterogeneous catalyst [1–4]. In some cases their activities reportedly rival, or even exceed, those of commercial catalysts and comparisons between the efficacy of certain metal nitrides with that of noble metals have frequently been drawn in the literature. With the aim of develo** novel nitrogen transfer pathways, it is of interest to explore the reactivity of the lattice nitrogen within nitrides. This idea has its origins in the Mars-van Krevelen mechanism wherein oxidation over metal oxide catalysts is accomplished by the direct reaction of lattice oxygen with the substrate, and its subsequent replenishment from the gas phase oxidant [5]. This mechanism can also be developed into a two step process where the oxidation of substrate and the re-oxidation of reduced catalyst are performed in separate steps which can deliver selectivity and heat transfer advantages when the target product is, e.g., susceptible to further oxidation in the gas-phase. The selective oxidation of butane to yield maleic anhydride provides an example of a catalytic process for which such a two-stage procedure has been investigated [6]. In addition to oxidation reactions catalysed by metal oxides, Mars-van Krevelen-like reaction mechanisms are documented for other types of reaction [7], for example those involving sulfide [8] and carbide [9, 28] and ammonia [29] decomposition catalysts and in SiO2 supported form active catalysts for the amination of ethylene with NH3 [30]. In the preparation of sponge-like Fe3N, calcination under N2 was reported to be crucial to successfully generate the nitride phase [29]. The bulk structure of industrial ammonia synthesis catalysts is reportedly nitrided [31]. In the current study, the iron nitride was prepared by nitridation of iron metal with ammonia as described in the experimental section. The resultant nitrogen content, as determined by combustion analysis, is reported in Table 1. The stoichiometric N contents expected for Fe2N and Fe3N, two commonly reported phases, are expected to be 11.13 and 7.71 wt% respectively. It can be seen that the value reported in the Table lies between these two values implying that a material with an intermediate stoichiometry is formed. This is not unexpected, since a number of different phases of iron nitride have been documented and a range of stoichiometries are known to exist, e.g. [32, 33]. Powder X-ray analysis of the sample was performed and the diffractogram is reported in Fig. 1. Reasonable matches can be made to a number of the listed iron nitride patterns in the JCPDS index including 03-0910 Fe2N, 01-1236 Fe3N, 03-1174 zeta-FeN, 03-0983 Fe2N and 06-0656 Fe2N. The relative intensity of the reflections is most consistent with 01-1236 Fe3N pattern. The positions of the reflections corresponding to this phase are shown in Fig. 1 where it can be seen that the reflections are shifted to lower 2θ values. This could be consistent with the formation of an ε-Fe3N1+x phase. It should also be noted that the overall intensity of the reflections is fairly low implying the possibility of a relatively high content of x-ray amorphous phase(s). Figure 2 shows the production of ammonia from this sample, as monitored by the decrease in conductivity of the standard H2SO4 solution at the reactor exit, for the reaction of 3/1 H2/N2 and 3/1 H2/Ar at 400 °C in the presence of this sample. In both cases, there is a relatively sharp decrease in conductivity occurring in the initial 1 h on stream, which possibly corresponds to hydrogenation of surface NH x groups formed on pretreatment. At longer times on stream an apparently steady state rate corresponding to ca. 50 μmol h−1 g−1 is attained in the case of the H2/N2 reaction, as might be expected for a catalytic reaction. In the case of H2/Ar the diminution in conductivity is seen to be lower, which is consistent with the consumption of the sample’s nitrogen by H2. Post-reaction N analyses were undertaken and the results are reported in Table 1 in which it can be seen that the majority of the nitrogen has been lost from both samples, with the loss for H2/N2 being lower as might be anticipated. The recovery of lost N as NH3 from the H2/Ar treated sample corresponds to <10 % of the pre-activated sample N content. Post-reaction XRD analyses were undertaken and these are also shown in Fig. 1. It can be seen that there is, in general, an overall loss of crystallinity upon reaction. A number of other differences are apparent. The formation of Fe metal, as evidenced by reflections occurring at ca. 45° and ca. 65° 2θ is apparent in the pattern of the H2/Ar treated sample. Surprisingly, in view of its similar N content this phase is not evident in the pattern of the H2/N2 treated sample. However, there are additional reflections which may correspond to lower binary nitrides such as Fe4N. Given the drastic reduction, more significant changes in the post-reaction XRD patterns may have been anticipated. However, in this respect, as mentioned above, it is important to remember that there could be a significant fraction of XRD invisible phase present.

Table 1 Pre- and post-reaction N content of binary iron nitride
Full size table
Fig. 1
figure 1

XRD patterns of pre- and post-reaction iron nitride samples

Full size image
Fig. 2
figure 2

Conductivity profiles for the iron nitride samples reacted with 3/1 H2/N2 and 3/1 H2/Ar at 400 °C

Full size image

A very recent study has shown the generation of ammonia by hydrogenation of small cobalt nitride clusters of the form ComN (where m = 7,8 and 9) [34]. Accordingly, it is of interest to determine the reduction behaviour of binary cobalt nitrides. Within the literature, there have been only a few studies of the catalytic behaviour of cobalt nitrides. Amongst those reported, has been the application of supported Co4N for preferential CO oxidation [35], NO decomposition [36] and hydrazine decomposition [37]. Fang et al. [38] have reported the stepwise decomposition of Co4N via Co3N to Co2N upon thermal annealing thin films of cobalt nitride. The preparation of cobalt nitride was found to be challenging with formation possibly only being achieved within a narrow temperature window. The method adopted was ammonolysis of Co3O4 at 700 °C for 2 hours followed by cooling in NH3. The XRD pattern of the resultant phase is shown in Fig. 3. As has been discussed elsewhere [35–37], the strong similarity in the patterns for Co and Co4N makes unambiguous assignment extremely challenging. The pre-reaction N content (Table 2) demonstrates the presence of some nitrogen associated with the samples prepared in this manner, but the quantities are significantly lower than would be expected for stoichiometric Co4N and there is variation between different batches. Therefore, in view of these uncertainties, it can not be definitively concluded that a bulk phase binary nitride, as opposed to Co metal containing sorbed NH x species, was formed. Nevertheless, it was still deemed of interest to determine the reduction characteristics of samples since the reaction of sorbed NH x species may still be useful in the development of two stage amination processes whereby ammonia is partly decomposed upon a surface in the first step and subsequently reacted by a target molecule in the second step. This type of process could be of value in those reactions where direct amination by NH3 is thermodynamically limited by dehydrogenation, for example in the direct amination of benzene to yield aniline [39, 40]. To determine the overall stability of the resultant Co–N phase, temperature programmed reaction was undertaken as presented in Fig. 4. From this figure, it can be seen that the loss of NH3 occurs within ca 1 h on stream at 250 °C and this corresponds to total loss of nitrogen from the sample as reported in Table 2. Hence, subsequent isothermal reactions were conducted using 3/1 H2/N2 and 3/1 H2/Ar at this temperature. The results are presented in Fig. 5; Table 2 in which it can be seen that there may be a marginal effect of reaction atmosphere with the presence of N2 possibly suppressing the loss of nitrogen. The quantity of ammonia produced as a proportion of N lost is around 13 % in the case of the H2/Ar reaction.

Fig. 3
figure 3

XRD patterns of pre- and post-reaction cobalt nitride samples

Full size image
Table 2 Pre- and post-reaction N content of cobalt samples
Full size table
Fig. 4
figure 4

Temperature programmed conductivity profile for the cobalt nitride sample reacted with 3/1 H2/Ar

Full size image
Fig. 5
figure 5

Conductivity profiles for the cobalt nitride samples reacted with 3/1 H2/N2 and 3/1 H2/Ar at 250 °C

Full size image

Rhenium nitride has been investigated for its ammonia synthesis activity and partial decomposition at 350 °C to yield a mixture of Re metal and Re3N was reported to occur resulting in a catalyst of higher activity than Re metal alone [41]. Denitridation of rhenium nitride was also observed during hydrodenitrogenation [42]. A number of studies have reported that Re based systems are active catalysts for ammonia synthesis [43–48]. Therefore, in view of the ability of Re to activate nitrogen and the apparent reactivity of lattice nitrogen in rhenium nitride, it was of interest to investigate this system further within the context of the current study. Rhenium nitride was prepared according to the method documented literature wherein ammonolysis of NH4ReO4 was undertaken at 350 °C for 2 h [41]. The XRD pattern of the resultant material is presented in Fig. 6. Consistent with the literature [41, 42], the pattern consists of very broad reflections. Given that the surface area of this material was determined to be 2 m2g−1, which indicates the presence of large non-porous crystallites, it is likely that this effect is related to the occurrence of disorder. The nitrogen content of pre-reaction samples is reported in Table 3. It is generally found that rhenium nitride with a stoichiometry of around Re3N is formed when this preparation route is employed [41, 42]. Given that the theoretical nitrogen content of this phase equates to 2.44 wt%, it can be seen that the value determined is close to that which would be expected. Rhenium nitride reportedly decomposes to the metal above ca 370 °C [42]. Consequently, reduction of the sample employing 3/1 Ar/H2 was undertaken using a temperature programme between 300 and 400 °C as reported in Fig. 7. From this figure, it can be seen that there is further loss of the sample’s nitrogen at 350 °C, which is consistent with the ammonia synthesis study reported earlier [41]. The post-reaction XRD pattern of this material is consistent with the presence of both reflections corresponding to Re metal and the broader reflections corresponding to the pre-reaction rhenium nitride phase (Fig. 6). Subsequently, the sample was further investigated at 350 °C where a comparison between the ammonia production efficacy under H2/Ar and H2/N2 was undertaken as shown in Fig. 8. From this figure, the greater degree of ammonia production in the presence of H2/N2 can be seen, which is consistent with the known catalytic behaviour of this material. The rate of ammonia production in this sample corresponds to ca. 130 μmol h−1 g−1 (although steady state is not attained) which compares with the value of 179 μmol h−1 g−1 reported previously [41]. Both samples possess similar post-reaction N analyses with the H2/N2 reacted sample, as may be expected, containing slightly more N—the greater degree of N loss in these samples as compared to the temperature programmed reacted samples may be attributed to the fact that they have spent longer at 350 °C or above (6 h for the isothermal runs vs. ca 4 h for the temperature programmed run.) They also possess similar XRD patterns which correspond to Re metal. The proportion of ammonia formed in the isothermal H2/Ar treated sample corresponds to close to 30 % of the nitrogen content of the pre-reaction sample.

Fig. 6
figure 6

XRD patterns of the pre- and post-reaction rhenium nitride samples

Full size image
Table 3 Pre- and post-reaction N content of the rhenium nitride samples
Full size table
Fig. 7
figure 7

Temperature programmed conductivity profile for the rhenium nitride sample reacted with 3/1 H2/Ar

Full size image
Fig. 8
figure 8

Conductivity profiles for the rhenium nitride samples reacted with 3/1 H2/N2 and 3/1 H2/Ar at 350 °C

Full size image

The inclusion of cobalt in nitrided rhenium catalysts has been reported to strongly improve ammonia synthesis activity, particularly when a material of composition Co-Re4 is prepared [41, 46]. Ammonolysis was conducted at much higher temperature (700 °C) and it was proposed that, despite the relatively low thermal stability of Re3N, a rhenium nitride phase was formed. We have repeated the preparation and the powder XRD pattern of the resultant material is presented in Fig. 9. The features are consistent with those published previously [41] with the sharper reflections corresponding to Re and Co. Comparison of ammonia production at 400 °C for H2/N2 and H2/Ar feeds was undertaken and the results are presented in Fig. 10. A low level of ammonia, corresponding to ca.75 μmol g−1 was produced in the case of the H2/Ar feed. In contrast, the H2/N2 reaction generated ammonia corresponding to a steady state rate of 472 μmol h−1 g−1 (which compares with the 600 μmol h−1 g−1 reported at 350 °C reported elsewhere [41, 46]). Post-reaction XRD patterns indicate little change, aside from a possible increase in reflection intensities arsing from increased crystallinity, occurring upon reaction.

Fig. 9
figure 9

XRD patterns of pre- and post-reaction Co-Re4 samples (Co , Re )

Full size image
Fig. 10
figure 10

Conductivity profiles for the Co-Re4 samples reacted with 3/1 H2/N2 and 3/1 H2/Ar at 400 °C

Full size image

4 Conclusions

The denitridation behaviour of nitrides of iron, cobalt and rhenium under an Ar/H2 atmosphere has been probed to determine the reactivity of lattice nitrogen as an initial screen for their potential application as nitrogen transfer reagents. In all cases, ammonia was produced as a minor product, accounting for 10 % of the nitrogen content of the iron nitride studied at 400 °C, 13 % of that of cobalt nitride at 250 °C and 30 % of that of rhenium nitride at 350 °C. In addition, the ambient pressure ammonia synthesis activity of the materials using N2/H2 has been investigated at the temperature of denitridation and the catalytic activity of the iron and rhenium systems, following an initial relatively sharp decline, possibly originating from hydrogenation of NH x species. The behaviour of nitrided Co-Re4 was also investigated and whilst only very low levels of NH3 formation under Ar/H2 at 400 °C were apparent, its ambient pressure NH3 synthesis was found to be significantly higher than all the other materials investigated.