Crystal structures and luminescence properties of Li₆MN₄:Ce³⁺ (M = Mo, W)

2.1 Crystal structure

The lithonitridotungstate crystallizes in the tetragonal crystal system in the space group P4₂/nmc (no. 137). For the crystal structure of Li₆WN₄, the lattice parameters a = 6.6844(1) and c = 4.9294(1) Å and a cell volume of V = 220.252(8) ų were obtained from single-crystal X-ray diffraction data. The details of the structure refinement are given in Table 1 and the atomic coordinates, displacement parameters, interatomic distances and angles are given in Tables 2–5.

The tungsten and lithium atoms are tetrahedrally coordinated by nitrogen atoms (Figure 1). The W–N distances of 1.882(2) Å are in the range of the values known in the literature. For example, in the compound Ba₃[WN₄] of Rabenau et al., in which the tungsten atom is also tetrahedrally coordinated by nitrogen atoms, the W–N distances have a mean value of 1.902 Å [2, 29, 30]. Likewise, the Li–N distances with a range of 2.041(2)–2.227(3) Å (see Figure 1) agree with published values for other lithium nitridometallates of different transition metals. E.g., Ma et al. give a Li–N distance of 2.252 Å for the compound LiSiON, in which lithium is tetrahedrally coordinated by three oxygen atoms and one nitrogen atom [2, 23, 31].

Figure 1:

Coordination polyhedra of the tungsten site and the two lithium sites in Li₆WN₄ with the interatomic distances in Å (standard deviations in parentheses). Nitrogen atoms are drawn as red spheres.

Regarding the coordination polyhedra, the structure is built up from a three-dimensional network of edge- and corner-linked LiN₄ tetrahedra. Along [001], channels are formed by an alternating arrangement of eight edge-linked LiN₄ tetrahedra of the Li1 site, with the Li2 site forming a square. The channels consist of alternating WN₄ tetrahedra and vacancies (see Figure 2).

Figure 2:

Crystal structure of Li₆WN₄ in the viewing directions [001] (top left), [100] (top middle) and [010] (bottom left). The unit cell is illustrated with black edges. Nitrogen atoms are shown in red and tungsten atoms in turquoise. The differently colored LiN₄ tetrahedra (Li1: green and Li2: orange) represent the two different Li positions. The squarish channels are marked in yellow.

In the crystal structure of Li₆WN₄, the positions of the nitrogen atoms are close to a ccp (cubic closest packing) arrangement where seven out of eight tetrahedral voids are occupied by W and Li (see Figure 3).

Figure 3:

Left: edge-linked Li1 tetrahedra (green) and a W tetrahedron (turquoise) linked by corners to the Li1 channel. Middle: strands of edge-linked Li2 tetrahedra (orange) which are linked by corners. Right: the resulting distorted ccp of nitrogen atoms with the seven out of eight occupied tetrahedra voids. The distorted ccp arrangement (black lines) of the N atoms (red spheres) is drawn in all three representations for better understanding.

2.2 MAPLE, CHARDI and BLBS calculations

Madelung Part of Lattice Energy (MAPLE) [32, 33], bond-length/bond-strength (BLBS) [34, 35] and Charge Distribution (CHARDI) [36, 37] calculations were performed to prove the electrostatic consistency of the crystal structure model. The MAPLE sums of Li₆WN₄ are listed in Table 6. The results of the CHARDI and BLBS calculations are shown in Table 7.

Comparing the MAPLE sum based on WO₃, Li₃N and Si₃N₄ with the calculated MAPLE sum of Li₆WN₄ and SiO₂, it was found that the values differ by only 0.26 %. The results of the BLBS method show the expected charges for the atoms, which are in good agreement with the oxidation states according to an electro-neutral sum formula. The result obtained using the BLBS method is supported by the CHARDI calculations, which is also in good agreement.

The validity of the structural model was confirmed for the powder phases by Rietveld refinement (a phase-pure sample obtained via the described synthesis pathway for the tungsten variant, and a high-quality powder sample with a minor amount of an unknown side phase for the molybdenum variant), as shown in Figures 4 and 5. The results of the Rietveld analysis for the W and Mo phases are shown in Table 8.

Figure 4:

Rietveld fit of the measured powder diffraction data (black), the calculated pattern based on single-crystal data of Li₆WN₄ (red), and the difference plot (blue). Tick marks illustrate the theoretical reflection positions of Li₆WN₄ (green).

Figure 5:

Rietveld fit of the measured powder diffraction data (black), the calculated pattern (developed on Li₆WN₄ single-crystal data by substituting W for Mo (red)), and the difference plot (blue). Tick marks illustrate the theoretical reflection positions of Li₆MoN₄ (green). The asterisks denote reflections belonging to an unknown side phase.

2.3 Luminescence

The two title compounds could be successfully doped with the activator ion Ce³⁺ and the luminescence properties of Li₆WN₄:Ce³⁺ single crystals and Li₆MoN₄:Ce³⁺ single particles were investigated. The compound Li₆WN₄:Ce³⁺ with a nominal activator concentration of 1 mol% Ce³⁺ can be excited with UV-near to blue light. The single crystals, also taken for the structure determination, excited at λ = 448 nm show an emission maximum at λ_max = 693 nm with a full width at half maximum (fwhm) of 97 nm (0.269 eV, 2166 cm⁻¹). In comparison, single particles taken from a powder sample of the isotypic compound Li₆MoN₄:Ce³⁺ with a nominal activator concentration of 1 mol% Ce³⁺ show an emission maximum at λ_max = 653 nm with a full width at half maximum (fwhm) of 133 nm (0.385 eV, 3108 cm⁻¹). In the CIE-xy color space, this corresponds to x = 0.707(1) and y = 0.293(1) for Li₆WN₄:Ce³⁺ and x = 0.598(1) and y = 0.400(1) for Li₆MoN₄:Ce³⁺ (see Figure 6). The single crystals of the compound Li₆WN₄:Ce³⁺ have a slightly reddish body color in daylight and are stable for several months under atmospheric conditions.

Figure 6:

Left: the blue line represents the luminescence spectrum of single particles taken from a powder sample of Li₆MoN₄:Ce³⁺ recorded at an excitation wavelength of 448 nm. The red line represents the luminescence spectrum of a Li₆WN₄:Ce³⁺ crystal recorded at an excitation wavelength of 448 nm. The three gray dashed lines signal detector artefacts that occur at wavelengths 688, 703 and 733 nm. Right: color points of the emission spectra of Li₆MoN₄:Ce³⁺ particles and Li₆WN₄:Ce³⁺ crystals in the CIE-xy color space.

In the compounds Li₆MN₄:Ce³⁺ (M = Mo, W), two possible positions occur that can be occupied by the activator ion Ce³⁺, although a precise assignment is not possible (see Figure 7). Comparing the distances in the MN₄ tetrahedra of the two Rietveld refinements (Table 5) and the ionic radii of tungsten (0.56 Å) [42] and molybdenum (0.55 Å) [42], the molybdenum-centered tetrahedra have a larger distance between the nitrogen and the metal atoms than the tungsten-centered tetrahedra (the ionic radii applied in each case are those of the oxidation states 6+ and the coordination numbers 4). This difference leads to a smaller splitting of the d orbitals and thus results in a shift of the bands (see Figure 6 left).

Figure 7:

Left: the two possible coordination environments of the activator ion Ce³⁺ in the tungsten variant including the distances to the N³⁻ anions (red spheres) (distances taken from the powder data for both compounds). Right: the two possible coordination environments of the activator ion Ce³⁺ in the molybdenum variant including the distances to the N³⁻ anions (red spheres).

For the Ce³⁺ dopant, an octahedral or even larger coordination is more likely than the tetrahedral coordination (see Figure 7) owing to the comparatively large ionic radius of cerium (r = 1.15 Å) [42]. Two different possible positions of the octahedrally coordinated activator ion Eu²⁺ were described for the compound Li₂CaSi₂N₄:Eu²⁺ [43]. Therein, the octahedral coordination polyhedra are occupied by Ca and N atoms, with Ca–N distances ranging from 2.489(3) to 2.509(3) Å for Ca1 and 2.586(3) Å for Ca2. Compared to the distances between the center of the octahedron to its nitrogen atoms in Li₆MN₄:Ce³⁺ (M = Mo, W) (Mo: Ø center–N = 2.41(2) Å and W: Ø center–N = 2.42(2) Å), the distances found in Li₂CaSi₂N₄:Eu²⁺ are slightly larger. Considering that Eu²⁺ (r = 1.31 Å) has a slightly larger ionic radius than Ce³⁺ (r = 1.15 Å), the occupation of an octahedrally coordinated center in the structure of the title compounds by Ce³⁺ would also be possible [42].

A spectrum unfolding was performed for the emission spectrum of the single-crystal Li₆WN₄:Ce³⁺ (see Figure 8). This unfolding shows two bands, of which one band has a maximum at λ_maxP1 = 1.799 eV (14,507 cm⁻¹) with a full width at half maximum (fwhm) of 0.218 eV (1757 cm⁻¹) and the second band shows a maximum at λ_maxP2 = 1.718 eV (13,855 cm⁻¹) with a full width at half maximum (fwhm) of 0.301 eV (2432 cm⁻¹). This double band is a result of spin–orbit coupling of the two sublevels ²F_5/2 and ²F_7/2 of the 4f¹ ground state configuration of Ce³⁺ [44].

Figure 8:

Left: the black line represents the luminescence spectrum of a Li₆WN₄:Ce³⁺ crystal recorded at an excitation wavelength of 448 nm. The green dashed line represents an ideal Gaussian fit of the experimental emission band. The bimodal Gaussian fit is represented by the blue and red curves. Right: optical appearance of the 1 mol% Ce³⁺-doped compound Li₆WN₄ under UV light (365 nm) at the bottom of a tungsten crucible and black particles due to corrosion of the tungsten crucible by lithium.

In order to exclude defect luminescence and host structure emission, several experiments were carried out. For this purpose, the compounds Li₆MN₄ (M = Mo, W) were prepared without the activator ion Ce³⁺, for which no luminescence could be detected. Likewise, the possibility of falsification of the emissions due to impurities in the crucibles was eliminated.

4.1 Synthesis

Because the syntheses involve reactants that are unstable or sensitive to hydrolysis in air, the preparations were carried out in a glovebox filled with inert gas (Ar 5.0, Messer Austria GmbH). In an agate mortar, the starting materials M (M: W – Plansee, 99.9 %; Mo – Plansee, 99.9 %) and Li₃N (Alfa Aesar, 99.4 %) were mixed in a stoichiometric ratio of 1:4 together with 1 mol% Ce (Sigma Aldrich, 99.9 %) based on the metal contents. The ratio 1:4 has been found to be the best from empirical studies, although traces of corrosion on the tungsten crucible and sublimation of reactant and subsequent deposition on the cooler reaction apparatus were always observed during the syntheses. The reactants were then poured into a tungsten crucible and transferred to the radio-frequency furnace system (type TrueHeat HF 5010; Hüttinger Elektronik, Freiburg, Germany) under an inert gas atmosphere. To prevent a reaction of the lithium nitride melt with the tungsten crucible during the production of the Mo variant, the reactants were packed in a capsule made of Mo foil (Alfa Aesar, 99.95 %, 0.025 mm thick) and then placed in the tungsten crucible. During the entire annealing process, forming gas (forming gas H10 5.0, Messner Austria GmbH), which was regulated to 10 mL min⁻¹, flowed into the reaction apparatus to ensure a reducing reaction atmosphere. The annealing process consisted of a heating ramp of 200 K h⁻¹ to the final temperature of 1000 °C, a holding time of 12 h at this temperature, a cooling ramp of 0.15 K min⁻¹ to 600 °C and finally quenching to room temperature. The products obtained, which had a slightly reddish body color, could be further investigated and characterized under atmospheric conditions and showed no decomposition due to humidity or atmospheric conditions for several days.

4.2 Single-crystal X-ray diffraction

Single-crystals were isolated from the obtained products using a needle selected under a polarization microscope and analyzed with the Bruker D8 Quest single-crystal diffractometer (Bruker, Billerica, USA) equipped with a PHOTON III C14 detector and an Incoatec Microfocus diffraction source (Incoatec, Geesthacht, Germany) with MoKα radiation (λ = 0.71073 Å). The programs Saint [45] and Sadabs [46] were used for data processing and multi-scan absorption correction. Olex2. solve 1.5 [47] using Direct Methods and Shelxl [48] using the least-squares method, both implemented in the program Olex2 1.5 [49], were used for structure solution and parameter refinement, respectively. The structural data were standardized using Structure Tidy [50] implemented in Platon (version 170613) [51].

Further details of the crystal structure investigation on the single-crystal of Li₆WN₄ can be obtained from the Cambridge Crystallographic Data Center at www.ccdc.cam.ac.uk/structures/ using the deposition number 2282797 (Li₆WN₄).

4.3 Powder X-ray diffraction

Powder X-ray diffraction (PXRD) was used for phase analysis, using a STOE Stadi P powder diffractometer (STOE & Cie., Darmstadt, Germany) with Ge(111)-monochromatized MoKα₁ (λ = 0.7093 Å) radiation and a Mythen 1K detector in modified Debye-Scherrer geometry over a 2θ range of 2–50°. The experimental powder pattern was compared with the calculated powder pattern derived from single-crystal data of Li₆WN₄ on the basis of Rietveld analysis. Secondary-phase identification was performed with the ICDD PDF-2 database and the Topas 4.2 program [52] was used for Rietveld refinements.

4.4 Luminescence

The emission spectra of the compounds Li₆MN₄:Ce³⁺ (M = Mo, W) were recorded with a setup consisting of a blue laser diode (λ = 448 nm, THORLABS, Ostfildern, Germany) in combination with a CCD-based spectrometer (AVA AvaSpec 2048, AVANTES, Apeldoorn, The Netherlands). A tungsten-halogen calibration lamp (HL3-PLUS CAL, Ocean Insight, Ostfildern, Germany) was used to calibrate the spectral radiance of the system for the first time. Ava Avasoft software (version 7) was used for data preparation.