Towards anti-perovskite nitrides as potential nitrogen storage materials for chemical looping ammonia production: reduction of Co3ZnN, Ni3ZnN, Co3InN and Ni3InN under hydrogen

The ammonia production properties upon reduction in hydrogen of the antiperovskite nitrides Co3ZnN, Ni3ZnN, Co3InN, and Ni3InN have been investigated. Single phases with ideal anti-perovskite structures (Space group: Pm-3m) were prepared for all the nitrides by the ammonolysis of the corresponding precursor oxides and all the nitrides were observed to produce ammonia in high yields when reacted with H2/Ar. The cumulative ammonia production values at 400 °C were 3069, 2925, 289, and 1029 μmol-NH3 g −1 for Co3ZnN, Ni3ZnN, Co3InN, and Ni3InN, respectively and the order of the release rates was Ni3ZnN > Co3ZnN > Ni3InN > Co3InN. X-ray diffraction studies revealed that Co3ZnN and Co3InN were decomposed upon the loss of lattice N, whereas Ni3ZnN and Ni3InN were transformed into Ni3Zn and Ni3In via the intermediate phases Ni3ZnNx and Ni3InNy. The crystal structures of these intermediate phases are related to their initial structures, indicating that the loss of lattice N in Ni3ZnN and Ni3InN was topotactic. Keyword: anti-perovskite nitride, nitrogen storage material, chemical looping ammonia production, topotactic reaction


Introduction
Ammonia production, currently accomplished on the industrial scale via the Haber-Bosch Process, is a reaction of pivotal societal importance [1]. It can be credited with the sustenance of around 40% of the global population through the provision of an accessible route to synthetic J o u r n a l P r e -p r o o f fertilizer. Around 174 million tonnes of ammonia are produced on the industrial scale annually [1] and production continues to grow.
When considered in its entirety, including the production of the feedstream reagents, the Haber-Bosch Process currently accounts for around 1-2% of global energy demand [1]. The hydrogen required is generated currently from fossil sources and the process has been reported to be responsible for the production of about 2.5% of all fossil fuel based CO2 emissions worldwide [1]. Accordingly, in view of the paramount importance of this reaction, recent studies have sought to address issues of sustainability. One topic of focus has been the possibility of the development of ammonia synthesis technology which would be suitable for operation on a local scale, being much smaller than the traditional Haber-Bosch plants which are operated on a large scale, where, for example, the local generation of fertilizer on a farm could be accomplished using hydrogen generated via electrolysis employing renewably derived energy such as that from wind power. In order to accomplish this, using heterogeneous catalysis, a step change in catalyst performance is necessary and, correspondingly, the development of new catalysts of enhanced activity with respect to the industrially employed iron-based [2] and ruthenium based [3] catalysts is an area of contemporary interest. It is widely believed that in order to achieve this, the limiting scaling relationship which has been reported for ammonia synthesis will need to be broken. Amongst catalysts investigated to date have been electride supported metals in which novel mechanistic pathways are reported to be operative [4], the combination of lithium hydride and metals or metal nitrides which are proposed to circumvent scaling limitations via the transfer of reaction intermediates between phases [5], ternary [6,7] and quaternary [8] metal nitrides which might operate by nitrogen based Mars-van Krevelen mechanisms [9,10] and hydride and metal hydride based compositions [11,12] for which hydrogen based Mars-van Krevelen mechanisms may be J o u r n a l P r e -p r o o f operational. In addition to the development of novel, more active, heterogeneous catalysts, photocatalytic [13] and electrocatalytic [14,15] processes have also been investigated.
A further approach which has been of interest in the context of localised ammonia synthesis involves chemical looping in which production is accomplished via the reaction of a nitrogen containing reagent followed by its regeneration in a separate reaction step. A number of these studies have employed the hydrolysis of a nitride followed by its regeneration at high temperature employing N2, e.g. [16,17]. It has been reported that the high temperatures required for renitridation could be accomplished from solar energy. By such routes, ammonia production from nitrogen, water and sunlight becomes possible making it entirely sustainable with a dramatically reduced carbon footprint. Related to this has been the report of a system for ammonia synthesis based upon hydrolysis of lithium nitride and its subsequent regeneration by electrochemical means [18]. Whilst it is less direct, the possible availability of sustainable hydrogen derived from water, means that looping approaches based upon hydrogenation of regenerable metal nitrides is also an area of interest. Accordingly, reports have been made of ammonia production via the hydrogenation of systems based upon binary metal nitrides including Re3N [19], Ni3N [20], Cu3N [20], Zn3N2 [20] Ta3N5 [20,21] and manganese nitride [22] have been reported. In the current study, we extend these studies towards anti-perovskite nitrides (general formula A3BN, see Figure 1 for their structure) which offer the possibility of tuning the reactivity of lattice nitrogen and regeneration via the controlled variation of the metal (i.e. A and B site) composition. The aim of this work is to establish composition -performance relationships which might prove to be of value for for the computationally aided design of more active heterogeneous catalysts [23].
In the current study, we report the lattice nitrogen hydrogenation characteristics of cobalt, nickel, zinc and indium based anti-perovskite nitrides. It is of interest to note that the removal and replenishment of lattice nitrogen from anti-perovskite J o u r n a l P r e -p r o o f nitrides is in some way analogous to the exsolution of metal nanoparticles from perovskites which is an area of topical interest [24]. Also of relevance to the approach reported within the current manuscript are reports of the nitrogenation and hydrogenation characteristics of transition metal -iron intermetallic compounds [25] and the nitrogen absorption behaviour of zirconium vanadium iron and related alloys [26].

Sample preparation
Polycrystalline samples of Co3ZnN, Ni3ZnN, Co3InN, and Ni3InN were synthesized by ammonolysis of precursor oxides which in turn were prepared from nitrate solutions.

Characterization
Powder X-ray diffraction (XRD) measurements were conducted using an X'Pert PRO MPD (PANalytical) employing Cu Kα radiation (λ = 1.54056 Å). XRD patterns were collected in the 2θ 20−85° range with a step size of 0.0167°. Le Bail refinements for the XRD patterns obtained were performed using the Jana2006 program [27] with a pseudo-Voigt function in order to estimate lattice parameters. Energy dispersive X-ray spectroscopy (EDX) was collected using a

Evaluation of ammonia production
Samples (0.2 g) were placed inside quartz microreactor tubes and were held in place centrally between quartz wool plugs and housed in a Carbolite tube furnace. Samples were heated under a flow of H2/Ar (75% H2, BOC, 60 mL min −1 ) from room temperature to the target reaction temperature (400 or 500 ºC) employing a ramp rate of 10 ºC min −1 . Samples were held for 6 h at the target temperature. The vent gas was flowed through 210 mL of H2SO4 solution (0.00108 M) and the change in conductivity was related to the production of NH3.

Results and Discussion
The XRD patterns of the precursor oxides for Co3ZnN (Co-Zn-O), Ni3ZnN (Ni-Zn-O), Co3InN (Co-In-O), and Ni3InN (Ni-In-O) are presented in Figure 2. Co-Zn-O exhibited a cubic phase related to Co3O4, although the lattice parameter based on the Fd-3m space group was 8.0971 (7) Å, which is larger than the 8.065 Å of Co3O4 [28] and which is indicative of Co-Zn-O being J o u r n a l P r e -p r o o f Co3−δZnδO4 since the increase of lattice parameter with respect to Co3O4 is consistent with Zn substitution into the lattice [29].  [30][31][32] The N contents of Co3ZnN, Ni3ZnN, and Ni3InN as determined by elemental analysis (Table 1) were also consistent with the expected values based on the A3BN composition (A = Co Ni, B = Zn, In), although the N content of Co3InN was approximately 10% higher than that of theoretical value. Hydrogen, which could potentially be incorporated into the samples during the ammonolysis procedure, was not detected for any of the nitrides by combustion analysis. A/B molar ratios of Co3ZnN, Ni3ZnN, and Ni3InN estimated by EDX analysis were close the synthesis ratio of 3 for all nitrides although, as for the N analysis, that of Co3InN was higher than expected.
One possibility is that In in Co3InN might have been partially lost during synthesis since the melting point of In metal (157 ºC) is much lower than the temperature of the ammonolysis reaction (600 ºC) and this requires further investigation. The surface areas of Co3ZnN, Ni3ZnN, Co3InN, and Ni3InN were found to be 3, 5, 4, and 3 m 2 g −1 , respectively (Table 1).
Ammonia production under 75% H2/Ar was investigated in order to determine the reactivity of the lattice N in these nitrides. The ammonia production profiles are presented in Figure 4. All of J o u r n a l P r e -p r o o f the nitrides were found to produce ammonia and the production amounts at 400 ºC were 3069, 2925, 289, and 1029 μmol-NH3 g −1 for Co3ZnN, Ni3ZnN, Co3InN, and Ni3InN, respectively ( Figure 4a and Table 2). The nitrides did not complete ammonia production within 6 h at 400 ºC.
On the other hand, all of the nitrides except for Co3InN completed ammonia production in 6 h at 500 ºC (Figure 4b) which is consistent with their post-reaction N analysis ( Table 1). The production amounts at 500 ºC were 3491, 3461, 1523, and 2804 μmol-NH3 g −1 for Co3ZnN, Ni3ZnN, Co3InN, and Ni3InN, respectively ( Figure 4b and Table 2). These values are generally higher than for Li-Mn-N (2072 μmol-NH3 g −1 for 5 h at 500 ºC) [22]. Furthermore, these production amounts correspond to the consumption of 93, 92, 42, and 85 % of the total available N content of the samples respectively. These values are ca. 1.7−3.9 times higher than the reported values for Li-Mn-N (24% for 5 h at 500 ºC) [22].
Significant differences amongst the nitrides were observed in terms of their ammonia production rates which, estimated from the production amount in 30 min after the temperature reached 400 ºC, were 752, 2082, 61, and 220 μmol-NH3 g −1 h −1 for Co3ZnN, Ni3ZnN, Co3InN, and Ni3InN, respectively. These values do not correspond to initial production since small amounts of ammonia (1-6 % against theoretical maximum) were produced from these nitrides during the temperature increase. The ammonia production rates of the Ni-containing nitrides were ca.
2.8−3.6 times higher than the corresponding Co-containing nitrides and the Zn-containing nitrides were 9.5−12.3 times higher than the corresponding In-containing nitrides. The highest ammonia production rate observed for Ni3ZnN (2082 μmol-NH3 g −1 h −1 ) was ca. 3.2 times higher than the catalytic ammonia production under 75% H2/N2 Co3Mo3N (652 μmol-NH3 g −1 h −1 at 400 ºC under 0.1 MPa) [33]. N balance calculations based upon the elemental analyses reported in Table 1 and the total ammonia production reported in Table 2  air even at room temperature [34]. Decomposition into Zn3N2 followed by subsequent oxidation to ZnO seems unrealistic because the ca. 93% and 94% of the N lost was related to ammonia production at 400 and 500 ºC, respectively. In contrast, the XRD pattern of Ni3ZnN-400 exhibited Ni3Zn (Fm-3m) [35] and an unknown phase (denoted by α in Figure 5a) A similar explanation for the ammonia production rate can be applied for the relation between Ni3InN and Co3InN. The XRD pattern of Co3InN-400 evidenced Co metal and unreacted Co3InN ( Figure 5b). In addition to these phases, CoIn2 [36] was also detected in the XRD pattern of Co3InN-500. These observations mean that Co3InN decomposed into Co metal and CoIn2 upon J o u r n a l P r e -p r o o f the loss of lattice N. In contrast, the XRD pattern of Ni3InN-400 showed unreacted Ni3InN, a small amount of Ni metal, and an unknown phase (denoted as β in Figure 5b) (Table 1) and SEM investigation of all the materials investigated ( Figure S1 in the Supplementary Information) demonstrates them to have quite irregular morphology.
Ammonia production by nitrides under the H2/Ar atmosphere is assumed to proceed through a Mars-van Krevelen like mechanism [21,22], which comprises surface reactions and bulk diffusion of N atoms. Regarding the surface reactions, adsorption and dissociation of hydrogen molecules to produce H species, the reaction of the H species with lattice N atoms to form ammonia, and the desorption of the ammonia molecules are involved in ammonia production.
Differences in the chemical affinities between Zn and In can be anticipated to affect these reactions resulting in the difference in the ammonia production rates between Ni3ZnN and  An obvious next step is to investigate the nitrogen cycling properties of these materials in which regeneration and further nitrogen discharge will be investigated. In addition, it is notable that addition of dopants including lithium in the case of manganese nitride [22], cobalt in the case of tantalum nitride [21,40] and ruthenium/alumina in the case of titanium-iron intermetallic compounds [25] have been reported to influence beneficially the lattice nitrogen conversion and/or uptake characteristics.

Conclusions
Anti-perovskite nitrides Co3ZnN, Ni3ZnN, Co3InN, and Ni3InN were prepared via ammonolysis using corresponding precursor oxides, and the reactivity of N atoms in these nitrides were investigated using ammonia synthesis as a model reaction. All of these nitrides released N atoms as ammonia under a flow of H2/Ar. Ni-nitrides Ni3ZnN and Ni3InN showed higher ammonia production rate than Co-nitrides Co3ZnN and Co3InN, respectively. XRD measurement revealed that the existence of intermediate phases Ni3ZnNx and Ni3InNy, which should be the main factor for the higher ammonia production rates. TGA measurements suggested the regenerabilities for Ni3ZnN and Ni3InN but further experiments are needed.

J o u r n a l P r e -p r o o f
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Author credit statement
YG and AD undertook the experimental work in JSJH's laboratory and under his supervision.   Evaluated by BET analysis. Table 2. Ammonia production yields of the anti-perovskite nitrides compound Ammonia production (μmol-NH3 g −1 ) Theoretical a 400 °C b 500 °C b