Substitution of W⁵⁺ in monophosphate tungsten bronzes by combinations Mⁿ⁺/W⁶⁺

2.1 Synthesis and crystallization

2.2 X-ray powder diffraction (IP Guinier technique)

Powder diffraction patterns were recorded at ambient temperature using an imaging plate Guinier camera with an integrated read out system (HUBER G670, CuK_α1 radiation, λ = 1.54059 Å for details see [26]). Lattice parameters for the mm-MPTBs (see Table 1) were determined using the program SOS [27] and SiO₂ (Merck, Darmstadt, p. A.) as internal standard. The assignment of the reflections was checked by comparing the observed XRPD pattern to simulations based on the single crystal structure refinements.

2.3 Single-crystal X-ray diffraction

Crystal structure analyses were carried out for W^VOPO₄, (Cr^III_1/3W^VI_2/3)OPO₄, and (Mo^III_1/3W^VI_2/3)OPO₄ from single-crystal data. Details on data collections and reductions are summarized in Table 2. The starting parameters for the least-squares refinements were obtained by Direct Methods with Shelx-97 [31]. Full-matrix least-squares refinement in space group P2₁/m with anisotropic displacement parameters was carried out by using Shelx-97 in the WinGX [32] framework of programs. Mixed occupancy of the metal sites by tungsten and molybdenum or chromium was accounted for. The final atomic coordinates and anisotropic displacement parameters as well as selected inter-atomic distances are provided as supporting information in Tables S3 to S9.

Further details of the three crystal structure investigations may be obtained from Fachinformationszentrum Karlsruhe, 76344 Eggenstein-Leopoldshafen, Germany (fax: +49-7247-808-666; e-mail: crysdata@fiz-karlsruhe.de, http://www.fiz-karlsruhe.de/request_for_deposited_data.html) on quoting the deposition numbers CSD-430823 for WOPO₄, CSD-430824 for (Mo_1/3W_2/3)OPO₄, and CSD-430825 for (Cr_1/3W_2/3)OPO₄.

2.4 Magnetic measurements

The magnetic susceptibility of crushed crystals of WOPO₄ and of microcrystalline powders of (Cr_1/3W_2/3)OPO₄, and (Fe_1/3W_2/3)OPO₄ was measured as a function of temperature (2 ≤ T ≤ 300 K, Figure 11) using a vibrating sample magnetometer (Quantum Design QD-PPMS-VSM). The measurement on WOPO₄ was performed by first cooling the sample from room temperature to 2 K under applied magnetic field and then rising the temperature using the same temperature sweep rate. A diamagnetic correction was applied to the observed susceptibilities.

2.5 Electron microscopy

TEM studies were conducted on transmission electron microscopes (i) FEI-Philips CM300 UT/FEG operated at 300 kV and (ii) FEI-Philips CM30 T/LaB₆, operated at 300 kV, both equipped with a Gatan CCD for image recording and with a Thermo NSS system for EDS analysis using HPGe and Si(Li) Nanotrace detectors, respectively. The powder samples were prepared both dispersed in cyclo-hexane by ultrasonication and without dispersion in a solvent on holey carbon films supported on a copper grid. Electron diffraction pattern and HRTEM image simulations were carried out using the Jems software package [33].

SEM EDS [34, 35] analyses of Cr^III_4/3W^VI_8/3O₁₂(PO₂)₄ and Cr^III_4/3W^VI_20/3O₂₄(PO₂)₄ were carried out by REM, DSM-940, Zeiss (operated in 20 kV, equipped with PV 9800 energy dispersive detector [36]). Prior to the EDS analyses crystals were soaked in 5% HF for 2 days, washed with distilled water and sputtered with gold. No evidence for impurity elements was found. The observed ratio n(Cr): n(W): n(P) was always close to the expected one.

3.1 Synthesis, crystallization, and thermal decomposition

Several hitherto unknown mixed transition metal tungsten phosphates of the MPTB structure family have been synthesized (Tables S1 and S2) and characterized by their XRPD pattern (Table 1, and Figs. 1, 2 and 4). Phosphates (Ti^IV_1/2W^VI_1/2)OPO₄, (M^III_1/3W^VI_2/3)OPO₄ (M: V, Cr, Fe, Mo) and (V^III_1−xW^V,VI_x)OPO₄ (x ≤ 0.8) are isotypic to WOPO₄ (MPTB with m = 2 [17, 23]). The phosphates M^III_4/3W^VI_20/3O₂₄(PO₂)₄ (M: V, Cr, Fe) with higher content of WO₃ are isotypic to W₈O₂₄(PO₂)₄, the MPTB with m = 4 [2].

Fig. 1:

Comparison of the XRPD pattern (IP Guinier technique, CuK_α1 radiation) of phosphates (M^III_1/3W^VI_2/3)OPO₄ with M: V, Cr, Fe, Mo. Asterisks indicate a small impurity of V_4/3W_20/3O₂₄(PO₂)₄ [23, 37]. The Guinier diagram of (Mo^III_1/3W^VI_2/3)OPO₄ is compared to the simulated diffraction pattern based on the crystal structure refinements [this work].

Fig. 2:

Comparison of the XRPD pattern (IP Guinier technique, CuK_α1 radiation) of phosphates M^III_4/3W_20/3O₂₄(PO₂)₄ with M: V, Cr, Fe. Asterisks indicate a small impurity of monoclinic WO₃ [38, 39]. The pattern of Fe^III_4/3W_20/3O₂₄(PO₂)₄ is compared to the simulated diffraction pattern based on the crystal structure of W₈O₂₄(PO₂)₄ [2].

Synthesis of (Ti^IV_1/2W^VI_1/2)OPO₄, (M^III_1/3W^VI_2/3)OPO₄, and M^III_4/3W^VI_20/3O₂₄(PO₂)₄ (M: Cr, Fe) was achieved by SCS followed by annealing in air with stepwise rising of the temperature to 900°C. Synthesis of the “reduced” phases (V^III_1/3W^VI_2/3)OPO₄ and V^III_4/3W^VI_20/3O₂₄(PO₂)₄ containing vanadium(III) besides tungsten(VI) started also from precursors obtained by SCS (Tables S1, S2). Yet, subsequent heat treatment (550°C) of the reaction intermediate after ignition was carried out in an argon flow (p(O₂) ≈ 2·10⁻⁵ bar) to set the oxidation state of vanadium to +3. Formation of the solid solution (V^III_1−xW^V,VI_x)OPO₄ (e.g. x = 0.8) containing vanadium(III) and tungsten(V) besides tungsten(VI) required reduction of the precursors from SCS in moist hydrogen. Hydrogen saturated at ambient temperature with water vapor (∑p = 1 bar, p(H₂O) = 0.023 bar [40]) corresponds at ϑ = 550°C to an oxygen pressure p(O₂) ≈ 10⁻²⁹ bar, which is sufficient to reduce W⁶⁺ to W⁵⁺. Subsequent heat treatment of thus reduced precursors in sealed silica tubes at ϑ = 800°C for several days improved the crystallinity of the products. Annealing in air of precursors from SCS for synthesis of (V^III_0.333W^VI_0.667)OPO₄ led to oxidation and (according to XRPD) the formation of W^VI₂O₃(PO₄)₂ (monoclinic modification [19]), a second phase very similar to V^III_4/3W^VI_20/3O₂₄(PO₂)₄, yet probably with a higher oxidation state of vanadium and slightly lower tungsten content. In addition, the XRPD pattern showed an increased background, possibly due to the formation of vitreous vanadium(V) phosphate.

The general procedures for the synthesis of the mm-MPTBs derived from W₈O₂₄(PO₂)₄ (MPTB with m = 4) from precursors obtained via SCS are very similar to those leading to the phases with m = 2. Therefore, it is quite remarkable that phases with higher WO₃ content (m = 4) could also be synthesized via vapor phase moderated solid state reactions in sealed silica tubes using small amounts of NH₄Cl or chlorine (from in situ decomposition of PtCl₂) as mineralizer. Surprisingly, this procedure never led to the formation of the mm-MPTBs derived from WOPO₄. Instead, formation of the mixed-metal tungsten ortho-pyrophosphates M^III(W^VIO₂)₂(P₂O₇)(PO₄) (M: V, Cr, Fe, Mo) [41] with the same overall chemical composition was observed. These phases with M: V, Cr, and Fe are also formed upon heating (M^III_1/3W^VI_2/3)OPO₄ (from SCS and subsequent annealing) to temperatures above 950°C, while the reverse reaction, the transformation of phases M^III(W^VIO₂)₂(P₂O₇)(PO₄) into the corresponding phase (M^III_1/3W^VI_2/3)OPO₄, did never occur in our experiments. All these observations suggest that mm-MPTBs (M^III_1/3W^VI_2/3)OPO₄ are thermodynamically metastable with respect to the ortho-pyrophosphates M^III(W^VIO₂)₂(P₂O₇)(PO₄), while the mm-MPTBs M^III_4/3W^VI_20/3O₂₄(PO₂)₄ (M: V, Cr, Fe) with a higher content of WO₃ are indeed thermodynamically stable phases. Attempts to synthesize mm-MPTBs with even higher content of WO₃ (m > 4) have failed so far. Such experiments led always to WO₃ besides the mm-MPTBs with m = 2 and 4.

Chemical vapor transport of WOPO₄ using water and iodine as transport agents is expected to proceed according to equations (2) and (3) in agreement with literature on CVT of WO₂ [42], WO₃ [43, 44], WP₂O₇ [45] and further anhydrous phosphates [24]. By analogy to anhydrous molybdenum phophates [46] and MoO₂ [47], for the mixed phosphate (Mo^III_1/3W^VI_2/3)OPO₄ heterogeneous equilibria similar to (2) and (3) are expected.

In agreement with literature on chemical vapor transport of Cr₂O₃ [48], WO₃ and the phosphates CrPO₄ and W₂O₃(PO₄)₂ [24] in a temperature gradient 1050 → 900°C using chlorine as transport agent, CVT of (Cr_1/3W_2/3)OPO₄ is expected to proceed via equilibrium (4).

(Ti^IV_1/2W^VI_1/2)OPO₄ was never obtained as single-phase product, which hampered the determination of its lattice parameters (Table 1). Monoclinic W₂O₃(PO₄)₂ [19], monoclinic WO₃ [38, 39] and TiP₂O₇ [49] were observed as by-products. Upon heating at temperatures above 850°C transformation into bi-phasic equilibrium mixtures consisting of WO₃ and TiP₂O₇ was observed.

Upon heating at temperatures above 900°C (Cr^III_1/3W^VI_2/3)OPO₄ transforms to Cr^III(W^VIO₂)₂(P₂O₇)(PO₄) [41] (eq. 5a), which eventually decomposes at ϑ ≥ 1000°C to W₂O₃(PO₄)₂ [19] and α-CrPO₄ [50] (eq. 5b). Decomposition of Cr^III_4/3W^VI_20/3O₂₄(PO₂)₄ into W₂O₃(PO₄)₂ [19], WO₃ [38, 39], and α-CrPO₄ is observed at above 1000°C (eq. 6).

3.2 Crystal structures

For a survey of the crystal structures of monophosphate tungsten bronzes (WO₃)_2m(PO₂)₄ we refer to [1, 5]. WOPO₄ can be considered a second member (m = 2) of this series (Fig. 3a). Its crystal structure has been solved more than two decades ago (Pna2₁, Z = 4, a = 11.174(3) Å, b = 6.550(2) Å, c = 5.228(1) Å [17]). It is closely related to those of monoclinic and orthorhombic NbOPO₄ [51–53] and to TaOPO₄ [54, 55].

Fig. 3:

Crystal structures of WOPO₄ [this work] (a) and of W₈O₂₄(PO₂)₄ [2] (b). Mixed occupancy (M^III_1/3W^VI_2/3) for metal sites in the isotypic mm-MPTBs (M^III_1/3W^VI_2/3)OPO₄ and M^III_4/3W^VI_20/3O₂₄(PO₂)₄ is indicated by light green octahedra.

The XRPD pattern (IP Guinier photograph) of WOPO₄ observed in the course of our study (Fig. 4) shows splitting of several reflections compared to the simulation based on the reported orthorhombic structure model WOPO₄ [17]. Closer inspection of this discrepancy led to a monoclinic unit cell with β = 90.33°. The monoclinic symmetry is confirmed by the re-determination of the crystal structure in space group P2₁/m from single-crystal data (Tables 2, S4, S7, and S8). Despite the rather low residuals slightly enlarged anisotropic displacement parameters for oxygen atoms O5 and O6 (Fig. 5) suggested even lower symmetry. However, refinements in space groups P2₁ and Pm (including assumed pseudomerohedral twinning) did not yield an improvement. Evaluation of the XRPD pattern of the series of mm-MPTB given in Table 1 led for these phases to monoclinic unit cells similar to WOPO₄. Single-crystal structure refinements of (Mo^III_1/3W^VI_2/3)OPO₄ and (Cr^III_1/3W^VI_2/3)OPO₄ (Tables 2, S4, S5, S6, S8, S9) confirmed the statistical distribution of Mo/W and Cr/W over the two chemically very similar metal sites in the structures. Selected interatomic distances for the mm-MPTBs are given in Table S5 and Fig. 5 and show no peculiarities. As for WOPO₄ the structure refinement for (Mo^III_1/3W^VI_2/3)OPO₄ led to enlarged ADPs for O5 and O6 (Fig. 5). For (Cr^III_1/3W^VI_2/3)OPO₄ these anisotropies became even larger and led to the introduction of only half occupied split sites (O5a, O5b, O6a, O6b; Fig. 5). Selected area electron diffraction pattern of (Cr^III_1/3W^VI_2/3)OPO₄ (Figs. 6 and 7) are in agreement with the chosen monoclinic unit cell and provide no hint on an enlarged cell as it is observed for NbOPO₄ and TaOPO₄ [51, 55].

Fig. 4:

WOPO₄. Experimental (IP Guinier technique, CuK_α1 radiation, middle) and simulated diffraction pattern based on the revised single crystal structure refinement (bottom). Magnification of characteristically split reflections (top).

Fig. 5:

ORTEP representation of [MO₆] and [PO₄] polyhedra in WOPO₄ (a), (Mo_1/3W_2/3)OPO₄ (b), and (Cr_1/3W_2/3)OPO₄ with split positions (c). Displacement ellipsoids are drawn at the 50% probability level, software Diamond 4.0 [56].

Fig. 6:

(Cr^III_1/3W^VI_2/3)OPO₄. Contrast-inverted selected area electron diffraction (SAED) pattern in [100] zone axis orientation; d_obs(0 0 1) = 11.18 Å, d_obs(0 1 0) = 5.10 Å (see also Fig. S1).

Fig. 7:

(Cr^III_1/3W^VI_2/3)OPO₄. Contrast-inverted SAED pattern in [001] zone axis orientation. d_obs(1 0 0) = 6.472 Å, d_obs(0 1 0) = 5.194 Å (see also Fig. S2).

Substitution of W^V by M^III_1/3W^VI_2/3 is also possible in the MPTB with m = 4, W₈O₂₄(PO₂)₄, which is described in literature [2] by an orthorhombic structure model (P2₁2₁2₁, Z = 1, a = 5.285(2) Å, b = 6.569(1) Å, c = 17.351(3) Å). The powder diffraction pattern of the mm-MPTBs (M^III_4/3W^VI_20/3)O₂₄(PO₂)₄ (M^III: V, Cr, Fe) clearly exhibit a small but significant monoclinic distortion (Table 1) with the a axis of the orthorhombic structure model becoming the monoclinic b axis. Despite several attemps we did not succeed in growing single crystals of the mm-MPTBs suitable for single-crystal structure analysis. The simulated powder diffraction pattern for (Fe^III_4/3W^VI_20/3)O₂₄(PO₂)₄ (Fig. 2) is based on the structure model of W₈O₂₄(PO₂)₄ (Fig. 3b) after its transformation to space group P2₁ (online tool TRANSTRU at the Bilbao Crystallographic Server [57–60]). For the simulation refined lattice parameters were used. The orthorhombic structure model of W₈O₂₄(PO₂)₄ consists of two chemically different metal sites W1 and W2. Octahedra [W1O₆] are sharing five vertices with [WO₆] octahedra and only one with an adjacent [PO₄] tetrahedron. In contrast, octahedra [W2O₆] are sharing three vertices with [PO₄] tetrahedra and three with neighboring [WO₆] octahedra. Based on valence sum considerations [61] we assumed in the simulation preferred occupancy of the sites originating from W2 by trivalent cations. Intensities calculated for random distribution of iron and tungsten over the metal sites are, however, not significantly different. The unit cell dimensions obtained from the XRPD pattern of mm-MPTBs (M^III_4/3W^VI_20/3)O₂₄(PO₂)₄ are in agreement with the SAED pattern of (Cr^III_4/3W^VI_20/3)O₂₄(PO₂)₄ (Figs. 8 and 9). Further support for the structure model of (Cr^III_4/3W^VI_20/3)O₂₄(PO₂)₄ is drawn from the HRTEM image overlayed in Fig. 10 with a structural image based on coordination polyhedra. The characteristic herring bone pattern of the HRTEM image is nicely matched.

Fig. 8:

(Cr^III_4/3W^VI_20/3)O₂₄(PO₂)₄. Contrast-inverted SAED pattern in [100] zone axis orientation; d_obs(0 1 0) = 6.64 Å, d_obs(0 0 1) = 17.23 Å (see also Fig. S5).

Fig. 9:

(Cr^III_4/3W^VI_20/3)O₂₄(PO₂)₄. Contrast-inverted SAED pattern in [010] zone axis orientation; d_obs(1 0 0) = 5.30 Å, d_obs(0 0 1) = 17.27 Å (see also Fig. S6).

Fig. 10:

(Cr^III_4/3W^VI_20/3)O₂₄(PO₂)₄. Fourier-filtered HRTEM image (CM30 T) with overlaid polyhedral representation of the structure; view along [100].

3.3 Magnetic behavior

The temperature dependent magnetic behavior of W^VOPO₄, (Cr^III_1/3W^VI_2/3O)PO₄ and (Fe^III_1/3W^VI_2/3O)PO₄ is shown in Fig. 11. For the chromium and iron compounds over the whole temperature range Curie–Weiss behavior [62] with constant magnetic moments of μ_exp = 3.85 μ_B per Cr³⁺ (θ_p = –6.3 K) and 5.85 μ_B per Fe³⁺ (θ_p = –14.8 K) is observed. These moments are close to the spin-only values of 3.87 μ_B (d³) and 5.92 μ_B (d⁵, high spin) expected for Cr³⁺ and Fe³⁺, respectively. This agreement can be taken as further confirmation for the assumed chemical composition of these mm-MPTBs. W^VOPO₄ shows Curie–Weiss behavior (μ_exp = 1.06 μ_B, θ_P = –78 K) above 100 K and antiferromagnetic ordering (T_N = 15.6 K) at low temperatures. These observations are in full agreement with information in the literature [9].

Fig. 11:

Temperature-dependent molar susceptibility, χ_mol and its reciprocal χ_mol⁻¹ of W^VOPO₄, and phosphates (M^III_1/3W^VI_2/3)OPO₄ (M: Cr, Fe) in the range 2 ≤ T ≤ 300 K.

4.1 Supporting information

Tables S1 to S9 and Figures S1 to S6 are given as Supporting Information available online (DOI: 10.1515/znb-2016-0036).