Ni₂Mo₃N: crystal structure, thermal properties, and catalytic activity for ammonia decomposition

2.1 Chemical characterization

Ni₂Mo₃N was prepared as phase-pure powder of dark gray color with metallic sheen. The expected cation ratio for nickel molybdenum nitride of Ni:Mo = 2:3 was confirmed by energy dispersive X-ray spectroscopy (EDX) with 39.7 at% for nickel and 60.3 at% for molybdenum, normalized to the cations. The anion contents were determined by means of hot gas extraction: 3.41 wt% nitrogen and 0.49 wt% oxygen was measured (calculated: 3.34 wt% nitrogen, 0 wt% oxygen). The standard deviations for EDX and hot gas extraction are estimated to be 2 %. In our previous work on Fe₃Mo₃N [19] we showed that the oxygen content is caused by an oxygen-rich surface layer (scanning transmission electron microscopy and electron energy loss spectroscopy (STEM-EELS)). No oxygen was found in the bulk of the crystals. Consequently, the oxygen content was neglected for the Rietveld [20] refinement. We assume that the composition is comparable for the nickel molybdenum nitride presented here and formulated as Ni₂Mo₃N.

2.2 Crystal structure

The powder X-ray diffraction (XRD) pattern with the results of the Rietveld refinement is presented in Figure 1. The structural model reported by Weil et al. [17] was used as starting input.

Figure 1:

Powder X-ray diffraction pattern of Ni₂Mo₃N with the results of the Rietveld refinement. The measured pattern is given in red, the calculated one in black, while the reflection positions are presented in green and the difference plot in blue color.

The background was fitted by linear interpolation between 27 points. Table 1 lists the Wyckoff positions, atomic coordinates, and Debye-Waller factors for Ni₂Mo₃N. Table 2 sums up the results of the Rietveld refinement.

Ni₂Mo₃N crystallizes in the filled β-Mn type (space group P4₁32). The cubic unit cell is shown in Figure 2. In the β-Mn type, the atoms occupy the 8c and 12d sites, the filled β-Mn type contains additional atoms on the 4a site (here: nitrogen). The lattice parameter of β-Mn is a = 6.315 Å, which is slightly smaller than that of Ni₂Mo₃N (a = 6.634 Å) [21]. The (NMo₆) octahedra, depicted in Figure 2, are corner-shared, forming a three-dimensional network similar to that in Fe₃Mo₃N. The screw axis provides rotation of the polyhedra. Weil et al. [17] described the relation between the filled β-Mn type and the η-carbide type (e.g., adopted by Fe₃Mo₃N). Besides the obvious fact that the unit cell of the η-carbide type is much larger (for Fe₃Mo₃N: a = 11.0777 Å [19]) than that of the filled β-Mn type, the main difference seems to be the space available for the metal atoms that are located between the octahedra. In the η-carbide type, the octahedra form rings with a hexagonal shaped void, but in the β-Mn type the void inside the ring has a pentagonal geometry (see Figure 3).

Figure 2:

Crystal structure of Ni₂Mo₃N (Ni green, Mo violet, N blue, and (NMo₆) octahedra light blue colored), the unit cell is presented in black.

Figure 3:

Rings of (NMo₆) octahedra in (a) Fe₃Mo₃N with a hexagonal shaped void and (b) in Ni₂Mo₃N with a pentagonal shaped void.

The coordination spheres of all atom types in Ni₂Mo₃N are illustrated in Figure 4. The polyhedra around the nickel atoms may be described as strongly distorted icosahedra consisting of three nickel and nine molybdenum atoms. In the crystal structure of Fe₃Mo₃N [19], distorted icosahedra are also present – but with a different composition of (Fe[Fe₆Mo₆]). As described above, (NMo₆) octahedra can be found in both compounds. Interestingly, the coordination spheres around the molybdenum atoms are different: An icosahedron-like environment was found for Fe₃Mo₃N, while Ni₂Mo₃N exhibits a complex environment with a coordination number of 14. A list of selected interatomic distances is given in Table 3. The lengths are in accordance with values in the literature [17] with exception of the Mo–Mo distances. In Ni₂Mo₃N, there are four Mo–Mo distances of 2.775 Å and two of 2.818 Å – which are reversed in the reference.

Figure 4:

Coordination spheres for Ni₂Mo₃N with nickel in green, molybdenum in violet, and nitrogen in blue. The faces of the polyhedra have the colors of the central atoms.

Using group-theoretical concepts [22], [23], [24], the crystal structure of Ni₂Mo₃N can be derived from the rock salt type. Figure 5 shows the group-subgroup relations (Bärnighausen formalism) for one possible pathway: we begin with space group F m 3 ‾ m for the rock salt type, where the atoms occupy the Wyckoff positions 4a and 4b. The first step is a translationengleiche transition of index 2 (t2), resulting in space group F432. The following decentering of the lattice by a klassengleiche transition of index 4 (k4) leads to space group P4₂32. In the latter step a screw axis is created and the 4a and 4b site become 4c and 4b, respectively. Next, space group I4₁32 is reached by doubling the lattice parameters and a klassengleiche transition of index 4 (k4) takes place again. The 4b site splits to 8a and a 24h, 4c splits to an 8b and a 24g site. The final symmetry reduction leads to space group P4₁32, adopted by Ni₂Mo₃N, by a klassengleiche transition of index 2 (k2), splitting the 8b site to 4a, occupied by nitrogen, and 4b. The 24g site splits to two non-equivalent 12d sites, one of them occupied by molybdenum. The 8a site becomes an 8c site, occupied by nickel, and the 24h site is transformed to a 24e site. Interestingly, in a previous work by Stegemann et al. [25], an alternative Bärnighausen tree starting from the alloy NaAu₂, adopting the MgCu₂ type, and ending with Na₂Au₃Al has been shown.

Figure 5:

Group-subgroup relations (Bärnighausen formalism) between the rock salt type and the crystal structure of Ni₂Mo₃N (filled β-Mn type).

The high-temperature behavior of Ni₂Mo₃N was investigated by in-situ XRD measurements by heating the powder in flowing ammonia in steps of 50 K to 875 °C. Interestingly, and in clear contrast to Fe₃Mo₃N [19], the material was stable over the whole temperature range.

2.3 Catalytic measurements

The results of the decomposition of NH₃ to N₂ and H₂ in the range of 400–600 °C for the newly prepared Ni₂Mo₃N are given in Figure 6 and Table 4. For comparison, an industrial Ni-based reference, and the inert material (SiC) used as reactor filler are included, the latter being chosen in order to prove that it is practically inactive. Although the industrial reference is more active, Ni₂Mo₃N shows a significant conversion and might be an interesting starting point for further investigations. Especially a comparison of the conversion between the heating and cooling cycle is indicating a slight activation for both catalysts. A direct comparison of the conversion for a similar nitride catalyst reported in the literature is not possible due to the different test conditions applied, and missing information [15]. There appears to be a typing error in the given units as the hydrogen productivities presented are utopian high and do not fit to the gas hourly space velocity (GHSV) applied. A comparison of the hydrogen productivities reveals that with 129 mmol H₂ g_cat⁻¹ h⁻¹ a similar productivity at 500 °C for the Ni₂Mo₃N catalyst investigated in this work and with 122 mmol H₂ g_cat⁻¹ h⁻¹ reported by Zaman et al. [15] there is a similar performance. At higher temperatures the values obtained in this work are far higher than the ones reported by Zaman et al. (335 vs. 170 mmol H₂ g_cat⁻¹ h⁻¹ at 550 °C and 684 vs. 206 mmol H₂ g_cat⁻¹ h⁻¹ at 600 °C), most likely due to thermodynamic or transport limitations set under the different test conditions applied. The difficulties in comparing catalysts from both works highlight the need to apply standardized test conditions and to report the data obtained appropriately to enable a reliable comparison of literature studies and identify suitable catalysts for industrial application [26, 27].

Figure 6:

Conversion of the Ni₂Mo₃N and an industrial Ni-based catalyst as reference for the decomposition of ammonia in a temperature range between 400 and 600 °C. Reaction conditions: p = 1 atm, flow rate = 0.6 mL mg_cat⁻¹ min⁻¹, rate of temperature changes between the steps = 2 K min⁻¹.

4.1 Synthesis

In this work, Ni₂Mo₃N was synthesized by ammonolysis of an oxidic precursor (NiMoO₄) which was prepared via a modified Pechini [28] sol-gel route. The modified Pechini route, in contrast to the original method, starts from aqueous solutions to reduce the amount of organics [29]. Nickel(II) chloride (3.1 g) and a 12-fold molar excess of citric acid (55.2 g) were dissolved in 250 mL of a mixture of equal parts absolute ethanol and distilled water to receive a nickel citrate solution. For the molybdenum citrate solution, molybdenum(V)-chloride (4.8 g) and a 12-fold molar excess of citric acid (40.6 g) were employed, and absolute ethanol (250 mL) was added dropwise. The exact metal ion concentrations of the citrate solutions were determined gravimetrically. Both citrate solutions were mixed in the desired metal ratio (Ni:Mo = 2:3) and an at least 17-fold molar excess of ethylene glycol was added. Heating in steps of 50 K to 350 °C resulted in a gel that was transformed through overnight calcination at 450 °C to a green-brown powder: the oxidic precursor NiMoO₄. The solvents and additives evaporated completely during the heating process. The last step was the nitridation via ammonolysis: the sample was annealed in a tube furnace at 875 °C for 20 h with a heating rate of 400 K h⁻¹ and an ammonia flow rate of 9 L h⁻¹.

4.2 Chemical and structural characterization

For energy-dispersive X-ray spectroscopy (EDX) a DSM 982 GEMINI spectrometer (Carl Zeiss AG, Oberkochen, Germany) equipped with a XFlash 6 | 60 detector (Bruker, Billerica, USA) was used. The standard deviation of this method is estimated to be 2 %. A LECO TC-300/EF-300 N/O analyzer was used for hot gas extraction (ZrO₂ and steel as standard materials). Hereby, the standard deviation is estimated to be 2 %. XRD was performed using a PANalytical X’Pert Pro powder diffractometer operating in a Bragg-Brentano set-up. Nickel-filtered CuKα radiation with wavelengths of 1.54056 and 1.54439 Å was used for the experiments. The 2θ measuring range was set to 10°–120° with a step size of 0.026°. The program package FullProf Suite 2021 [30] was used to execute the Rietveld refinements. In-situ powder X-ray diffraction measurements were performed with a Rigaku SmartLab 3 kW system (CuKα radiation) equipped with Reaktor X.

Further details of the crystal structure investigation may be obtained from FIZ Karlsruhe, 76344 Eggenstein-Leopoldshafen, Germany (Fax: +49 7247 808 666; E-mail: crysdata@fiz-karlsruhe.de) on quoting the deposition number CSD-2309672.

4.3 Activity test on ammonia decomposition

The catalytic activity for ammonia decomposition of Ni₂Mo₃N was tested in a fixed-bed reactor. Two hundred milligrams of the catalyst (particle size 200–300 µm) was diluted in a mass ratio of 1:2 with SiC (particle size 300–400 µm). The use of different particle sizes enables a later separation of the spent catalyst from the dilution material. The test procedure started with heating the sample with a rate of 1 K min⁻¹ in nitrogen (quality 5.0) to 400 °C applying a flow rate of 1 mL mg_cat⁻¹ min⁻¹. After reaching the desired temperature, the gas supply was switched to ammonia (quality 5.0) with a flow rate of 0.6 mL mg_cat⁻¹ min⁻¹ and the temperature was kept constant for 60 min. Afterwards, the temperature was increased in steps of 50 K to 600 °C and back to the initial temperature applying a heating rate of 2 K min⁻¹. For each step the temperature was kept constant for 60 min. For comparison, an industrial Ni-based catalyst was tested under similar conditions, but with a reductive pretreatment with 10 % H₂ in N₂. In addition, a blank test was made filling the reactor only with SiC using the same procedure as described above. Product gas analysis was performed via IR detectors for quantification of NH₃ and H₂O as well as a TCD for H₂ (Emerson XStream XEGP). A detailed description of the experimental setup and test conditions can be found in ref. [26].