New transition metal oxide fluorides with ReO₃-type structure

1 Introduction

Nearly all transition metal oxide fluorides are synthetic products, and many are structurally related to well-known oxide structure types such as spinel, cassiterite, ReO₃, or garnet [1]. Some compounds are structurally derived from fluorides crystallizing in the chiolite, K₂NiF₄, or fluorite type [1]. Transition metal oxide fluorides with rutile- or cassiterite-type structures are known for the compositions MOF (M = Ti, V, Fe) and M₂O₃F (M = Ti, V). In addition, FeVO₃F, FeTiO₃F, and NbTi₂O₅F have been reported [1]. Oxide fluorides based on the fluorite type are observed for lanthanide and actinide compounds [2]. Oxide fluorides of the MO_3–xF_x type exhibiting the rhenium trioxide structure are known for M = Ti, Nb, Ta, Mo, or W [3–6]. In MO₂F-type phases (x = 1), the M atom possesses the highest formal oxidation state, M⁵⁺. If the M cation is small (ionic radius ~0.6 Å) the products often crystallize in the ReO₃-type structure or one closely related to it. In contrast, if M is larger, as in the case of the lanthanides or actinides (ionic radii ~1.0 Å), different structure types arise, for example the tysonite type known from LaF₃ [7]. In the case of a cubic ReO₃-type structure the metal atoms are octahedrally coordinated by randomly distributed oxygen and fluorine atoms. The M(O,F)₆ octahedra share all six corners with adjacent octahedra. MOF₃-type phases (M = Nb, Ta) have also been described, crystallizing in structures related to the SnF₄ type [8].

It should be mentioned that transition metal oxide fluorides are of interest as cathode materials for lithium ion batteries (e.g. [9]). The voltage of batteries using transition metal oxide fluoride- or transition metal fluoride-based cathode masses is expected to be higher compared to that of the corresponding oxides for the same redox pair [10]. In addition, oxide fluorides have also been found to be promising laser hosts and fluorescent materials for optical applications [11].

Recently a series of new mixed metal oxide fluorides MNb₂O₃F₆ (M = Mg, Zn), MgM₂OF₈ (M = Ti, Zr), Fe₂TiO₈F, and TiNb₂O₅F₄ has been reported [12]. These compounds crystallize in the ReO₃-type (MNb₂O₃F₆, M = Mg, Zn) and related structures, for example the NaSbF₆ type (MgTi₂OF₈ and MgZr₂OF₈) or the VF₃ type (Fe₂TiOF₈). In the present contribution we report on the syntheses and crystal structures of previously unknown mixed metal oxide fluorides MNbO₂F₄ (M = Cr, Fe), CrNb₂O₄F₅, and Fe₂Nb₃O₆F₉.

2 Results and discussion

All oxide fluorides presented here were prepared by the method reported by Permer and Lundberg [13], originally used for the synthesis of TiOF₂ and NbO₂F. In a first step, 0.5 g niobium oxide Nb₂O₅ (Fluka 99%) was added to 3 mL HF (aq.) (Merck, 48%). In a second step, a stoichiometric amount of chromium nitrate Cr(NO₃)₃·9H₂O (Aldrich 99%) or iron nitrate Fe(NO₃)₃·9H₂O (Aldrich 99.99%) was added to the niobium fluoride suspension and the mixture stirred for 15 h in a Teflon beaker. Slow evaporation (80°C, 2 days) of the resulting mixture to dryness using a hot plate yielded the respective crystalline oxide fluorides. The remaining powders were washed several times with ethanol and afterwards dried at temperatures between 150 and 200°C in an inert atmosphere. Phase-pure MNbO₂F₄ (M = Fe, Cr), CrNb₂O₄F₅, and Fe₂Nb₃O₆F₉ were synthesized. The CrNbO₂F₄ powder still contains ~3 wt% CrNbO₄ as a second phase. For all prepared samples the analytically determined compositions (see Experimental section) fit well with the given formulas. No indication for significant amounts of hydrogen was found for any sample.

The crystal structures of the new phases were refined using conventional X-ray powder techniques (for details see Experimental part). In general, due to their positions in the periodic table, the elements oxygen and fluorine cannot be distinguished using X-ray methods. Unfortunately, also for neutron diffraction the coherent scattering lengths are very similar: b_O = 5.803 fm, b_F = 5.654 fm. Consequently, we will not give experimental results concerning O/F order and treat oxygen and fluorine as statistically distributed. It is evident that the real crystal structures may be of lower symmetry (non-cubic) than that presented. However, it should be mentioned that no indications for symmetry reduction could be derived from the observed metrics. The X-ray powder diffraction patterns of the finally obtained oxide fluorides are presented in Figs. 1–4. All crystal structures could easily be refined in the cubic ReO₃-type (space group Pm3̅m, cations on 1a (0, 0, 0), anions on 3d (1/2, 0, 0) positions) also known from TiOF₂, MO₂F (M = Nb, Ta) [4, 5], (MNb₂O₃F₆ (M = Mg, Zn)), MgTiOF₈, and TiNb₂O₅F₄ [12]. As commonly known, the ReO₃-type structure can be described as a continuous 3D array of vertex-sharing octahedra (Fig. 5). As explained above, all oxide fluorides with this structure were refined assuming a random distribution of fluorine and oxygen on the anion sites. The final structural parameters are presented in Tables 1–3, the average bond lengths between metal cations and the surrounding anions in Table 4. In order to give an integer value for Z (number of formula units in the unit cell), the chemical compositions presented in Tables 1 and 2 are normalized to one cation and three anions (ReO₃-type).

Fig. 1:

Powder X-ray diffraction diagram of FeNbO₂F₄ with the results of the Rietveld refinement.

Fig. 2:

Powder X-ray diffraction diagram of Fe₂Nb₃O₆F₉ with the results of the Rietveld refinement.

Fig. 3:

Powder X-ray diffraction diagram of CrNbO₂F₄ with the results of the Rietveld refinement (markers: CrNbO₂F₄, top; ~3 wt% CrNbO₄, bottom).

Fig. 4:

Powder X-ray diffraction diagram of CrNb₂O₄F₅ with the results of the Rietveld refinement.

Fig. 5:

Unit cell of the cubic ReO₃-type (e.g. FeNbO₂F₄), outlined by black lines, showing the (Nb/Fe)(O/F)₆ polyhedra.

The refined site occupation factors are in good agreement with the chemical formulas (Table 3). In addition, comparing the determined bond lengths with reported values from the literature for pure oxides and fluorides it can be stated that the results for the oxide fluorides are in good agreement with the average values for the oxides/fluorides in the respective compositions (Table 4). As an example, for FeNbO₂F₄ the refined bond length of 192.8 pm fits well to the average value calculated from NbF_5, FeF₃, and FeNbO₄: 193.8 pm. Fe₂Nb₃O₆F₉ exhibits a refined bond length of 193.3 pm. The value calculated from the fluorides and the oxide is 194.1 pm. Also for chromium niobium oxide fluorides the refined bond lengths fit well to the average values calculated from NbF₅, CrF₃, and CrNbO₄. A bond length of 191.4 pm has experimentally been determined for CrNbO₂F₄, 191.6 pm has been calculated. Finally, CrNb₂O₄F₅ shows 192.7 pm (refined) as compared to 193.5 pm (calculated). As already mentioned in the Introduction, small M cations (ionic radii ~0.6 Å) lead to products crystallizing in the ReO₃ structure or one closely related to it. As described, all metal oxide fluorides presented in this contribution crystallize in the ReO₃-type structure. This is no surprise because the effective ionic radii for Nb⁺⁵, Cr⁺³, and Fe⁺³ (coordination number 6) are 64, 61.5, and 64.5 pm, respectively [19].

3 Experimental section

4 Conclusion

The simple synthesis route described in our contribution allows the preparation of various mixed metal oxide fluorides and appears to be a promising pathway to yet unknown compounds. This may also be of interest for the preparation of new cathode materials for lithium ion batteries with improved electrochemical properties (higher voltage of the battery etc.). All transition metal oxide fluorides presented in this work crystallize in the ReO₃-type. In the near future we will report on oxide fluorides exhibiting the rutile- or fluorite-type structure.