A new layered potassium-based molybdenum–tungsten monophosphate: synthesis, crystal structure, XPS and magnetic studies

2.1 Synthesis

The crystalline compound was prepared by a two steps solid-state process. In the first step, a mixture of K₂CO₃ (Merck, 99.0 %), MoO₃ (Honeywell Fluka, 99.5 %), WO₃ (Sigma Aldrich, 99.99 %) and NH₄H₂PO₄ (Merck, 99.99 %), in the molar ratios 3:5:2:4, was finely ground in an agate mortar and then left in a ceramic crucible over 2 h inside a furnace heated at 573 K in air. After complete evaporation of H₂O, CO₂ and NH₃, 1 mole of metallic Mo was added to the powder product and mixed thoroughly for 1 h, ideally yielding nominal formula K₆Mo₆W₂P₄O₃₄. The grounded powder was loaded in a silica tube and placed under a base pressure of 10⁻⁴ mbar to be torch sealed. The ampoule was then deposited in a tubular furnace heated at 833 K for 3 days to allow the crystal growth via vapour-phase transport process. The temperature gradient was 2.5 K/cm between the hot- and the cold-zones over a 10 cm long ampoule. After the chemical reaction took place, the temperature of the furnace was decreased at 60 K/h up to 473 K and finally quenched at room temperature.

Dark blue crystals (∼20 mg with dimensions in the range 10–100 µm) formed in the cold zone of the ampoule, whereas a darker sintered powder was found in the hot zone. Moreover, the excess of the reagents was deposited as a black crust coating the internal wall of the reaction tube.

2.2 Chemical composition

Semi-quantitative chemical analysis was performed by means of Energy Dispersive X-ray Spectroscopy (EDS) on crystals dispersed on a carbon tape and covered with a 40 nm graphite layer (Figure S1-a) using a ZEISS EVO MA15 scanning electron microscope equipped with an OXFORD INCA micro-analyser, operating with an accelerating voltage of 20 KV, a current of 0.3 nA and a working distance of 10.0 mm.

The EDS analysis confirmed the presence of the intended elements (Figure S1-b). The chemical composition (average of 5 spots) of the crystal (belonging to the dark blue compounds) used for the structural study in molar ratios is: K: 0.334, P: 0.260, Mo: 0.261 and W: 0.144.

2.3 Single crystal X-ray diffraction

X-ray diffraction data were collected at room temperature (T = 293 K) on a dark blue crystal (hereafter K–MoWP) with an Oxford Diffraction Xcalibur 3 diffractometer, equipped with a graphite monochromatized Mo-Kα (λ = 0.71073 Å) radiation and a Sapphire CCD detector. Dataset integration and empirical corrections were applied using the CrysAlis PRO software [43]. A total of 8344 reflections were collected, yielding 3960 independent reflections in the mmm Laue group (R _int = 0.062). Reflections conditions (0kl: k + l = 2n + 1 and hk0: h = 2n + 1) were consistent with both the space groups Pn2₁ a and Pnma. We attempted to solve the structure in both space groups. However, the final solution was successfully achieved in the centrosymmetric Pnma space group in accordance with the statistical test on the distribution of |E| values (|E ² − 1| = 0.924).

The structure was solved by direct methods and then refined by least-squares fitting based on |F ²| using the SHELXTL program [44]. Mo(W), P and K atoms were initially located, whereas oxygen atoms were found successively in the difference Fourier maps. Crystal data, data collection and structure refinement details are summarized in Table 1.

Scattering curves for neutral atoms were taken from the International Tables of Crystallography [45]. The occupancy of all the octahedral (M), tetrahedral (T) and K sites was refined (Mo vs. W, P vs. Mo and K vs. vacancy for M, T and K sites, respectively). K sites were found to be fully occupied by potassium and their occupancy was then fixed. In the final model, with anisotropic atomic displacement parameters for all the atoms and no constraints, the residual value settled at R = 5.1 % for 2784 independent observed reflections [4σ(F) level] and 209 parameters, and at R = 8.9 % for all 3960 independent reflections.

Atomic coordinates, site occupancies and equivalent isotropic displacement parameters are given in Table 2. The anisotropic displacement parameters are listed in Table S1. Table S2 and Table 3 report the bond valence sums (calculated according to the parameters of Brown and Altermatt [46]) and bond distances (along with distortion parameters [47]), respectively. Further crystallographic data can be obtained from the CIF deposited in the Cambridge Crystallographic Data Centre, CCDC number: 2072198.

2.4 Powder X-ray diffraction

The sintered powder deposited in the hot zone was examined using powder X-ray diffraction (PXRD) in the 2θ range 5–70° with a Bruker D8 ADVANCE diffractometer operating with monochromatized Cu-Kα radiation (λ = 1.5406 Å) in θ−2θ scan, with a step size of 0.010° and a scanning rate of 0.5 s/point.

The comparison between the experimental diffraction pattern (Figure S3) and that calculated on the basis of the structural model obtained from single-crystal data (Table 2) revealed some non-matching peaks indicating the presence of modest amount of K(MoO₂)(PO₄) (PDF 01-089-1570). The unit cell-parameters, peak shape (double-Voigt approach), zero error, phase fractions and background parameters (5 terms Chebychev background polynomial) were refined using Pawley method [48] (R _p = 9.20 %, R _wp = 12.54 %, and S = 2.84).

2.5 X-ray photoelectron spectroscopy (XPS)

The XPS experiment was carried out using an ESCALAB 250Xi spectrometer (Thermo Fisher Scientific–UK) equipped with a monochromatic Al-Kα source (hν = 1486.6 eV) and a six-channeltron as detection system. The apparatus works in an Ultra-High Vacuum (UHV) system, where the base pressure of the analysis chamber during the experiments was approximately 3.4 × 10⁻⁹ mbar. An amount of ∼4 mg of dark blue crystals selected from the cold zone was ground in an agate mortar (grain size down to about 20/30 µm) and spread on a gold foil. Electric contact between the sample and the apparatus was ensured by a metallic clip. The binding energy (BE) scale was calibrated by positioning the Au 4f _7/2 peak at BE = 84.0 eV with an accuracy of ±0.1 eV, while the spectra were collected at pass energy of 50 eV. In order to investigate the electronic structure of W and Mo, spectra were acquired in the 31–45 eV and 227–240 eV energy ranges. The detected area was 900 × 900 µm², which was successively reduced down to few hundreds of squared microns, in order to maximize the signal/background ratio from the sample. The coverage was sufficient to hide the Au signal from the underlying gold foil.

The experimental data were collected and processed by the Avantage v.5 software (Thermo Fisher Scientific–UK). Shirley background subtraction and mixed Lorentzian/Gaussian peak shape (30 %) were used for the peak fitting.

2.6 Magnetic susceptibility

Variable-temperature magnetic susceptibility measurements were recorded with a Superconducting QUantum Interference Device (SQUID) susceptometer (MPMS Quantum Design) in the temperature range 10–300 K at an applied magnetic field of 1.0 T. Measurements were performed on finely ground powder sample (deposited in the hot-zone of the reaction tube): 69.2 mg were used for the investigation. Magnetic data were corrected for the sample holder and diamagnetic contribution estimated from Pascal’s constants [49]. Least-squares fitting of the experimental curves was done with the Curie–Weiss equation.

3.1 Crystal structure description

The crystal-chemical formula of K–MoWP derived from the structural refinement, in good agreement with EDS data, is K₅[(Mo_3.2W_1.8)O₁₁](PO₄)₃[(P_0.5Mo_0.5)O₄], whose charge balance requires that a small amount of the transition-metal cations is not hexavalent. This compound crystallizes in the Pnma space group with a novel crystal structure. The asymmetric unit includes three octahedral sites (M1, M2 and M3), three tetrahedral sites (T1, T2 and T3) and four high-coordinated sites (K1, K2, K3 and K4). As shown in Figure 1, corner-sharing MO₆-TO₄ chains running along [010] are arranged to form complex bundles having elliptic sections where K⁺ cations are hosted. The major axes of the elliptic sections are oriented along [201] (blue) and along [−201] (red) and appear as a herringbone packing when projected down [010]. The projection along [100] (Figure 2) shows an ordered packing of ∼14.9 Å-thick polyhedral layers related to each other through the a-glide plane perpendicular to [001]. The polyhedral connection inside the layers results in trigonal, tetragonal and two different types of hexagonal channels.

Figure 1:

Polyhedral representation of the structure of K₅[(Mo_3.2W_1.8)O₁₁](PO₄)₃[(P_0.5Mo_0.5)O₄] along the b-axis. The projection shows elliptic sections with major axes oriented along [201] (blue) and along [−201] (red), adopting a herringbone packing. Note that the extremities of the polyhedral bundles (light-green tetrahedra) are far apart along the b-axis (see Figure 2). Green: MO₆ octahedra, light-green: T1O₄ tetrahedra, pink: T2O₄ and T3O₄ tetrahedra, light blue spheres: K⁺. Oxygen atoms are omitted for clarity.

Figure 2:

Structural sketch of K₅[(Mo_3.2W_1.8)O₁₁](PO₄)₃[(P_0.5Mo_0.5)O₄] : a) view along [100] showing a layered packing along the c-axis. This view shows a repeating sequence of two types of pseudo-hexagonal channels along the b-axis. In the first type (yellow), the polyhedral connection leads to pseudo-hexagonal channels of the type potassium titanium phosphate KTiPO₅ (KTP). In the second one (blue), the polyhedra are assembled to form pseudo-hexagonal channels similar to those observed in the monophosphate tungsten bronzes with hexagonal channels (MPTBh). Fragments of the KTP b) and MPTBh c) crystal structures projected along [010]. K and O atoms are omitted for clarity.

T2 and T3 are fully occupied by P and show typical average ^IVP–O distances (1.537 and 1.531 Å, respectively). On the other hand, the refined site scattering at the T1 site yields a mean electron number of 29.5 and a bond-valence sum of 5.51 v.u. indicating the presence of a heavier, high-valent cation tentatively corresponding to either about 0.50 Mo⁶⁺ (ionic radius 0.41 Å [50]) or 0.25 W⁶⁺ (ionic radius 0.42 Å [50]) substituting for P. However, the <T1–O> distance (1.658 Å) fits better with 50 % incorporation of Mo⁶⁺ rather than 25 % W⁶⁺. Despite the difference in size between Mo⁶⁺ and P⁵⁺, no evidence of splitting of either T1 or the coordinated oxygen atoms was observed on the ΔF Fourier map. Nonetheless, some positional disorder can be inferred by the large anisotropic displacement parameters of O16 and O17, which are not shared with any octahedral cation (Table S1, Figure S2). The P ↔ Mo substitution has been observed earlier in the langbeinite-related structures K₂M³⁺ ₂(PO₄)₂(MoO₄) where P⁵⁺ and Mo⁶⁺ coexist in the 2:1 ratio in a tetrahedron exhibiting a mean bond distance of 1.58 Å [51].

M1, M2 and M3 octahedra are occupied by molybdenum and tungsten with variable proportions (Table 2). All MO₆ octahedra are rather distorted, as indicated by the wide range of variation in bond distances and distortion parameters (see Table 3). According to Domengès et al. [52], the spread of M–O bonds in W-bronzes is ascribed to the corner sharing between M-octahedra and PO₄-tetrahedra, leading to a lengthening of the P–O–M linkage. Therefore, the higher the positive charge in the octahedron, the stronger the distortion. Thus, the highest quadratic elongation (λ) value shown by M1 suggests that hexavalent cations are mainly ordered at this site. Accordingly, M1 is the smallest octahedron in the structure (<M1–O> = 1.929 Å). On the other hand, the lower λ values of M2 and M3 (see Table 3), associated with longer mean bond distances [<M2–O> = 1.949; <M3–O> = 1.948 Å], could be explained by the presence of lower-valence cations. The local MO₆ environments and bond lengths are shown in Figure 3.

Figure 3:

MO₆ octahedra environments and bond lengths.

The coordination spheres of potassium cations depicted in Figure S2 were determined on the basis of the maximum gap of K–O distances. K1 displays a 10-fold coordination with bond distances in the range 2.712(8)–3.007(6) Å. K2 is surrounded by 6 oxygen atoms with distances ranging from 2.638(8) to 2.952(6) Å; however, taking into account the threshold value (3.33 Å) suggested by Donnay and Allmann [53] for K–O bonds, two additional bonds [K2–O10 = 3.285(6) and K2–O17 = 3.316(4) Å] lead to a 6+2 coordination. K3 and K4 have 9- and 10-fold coordination, respectively, with bond lengths ranging between 2.729(6)–3.192(6) Å and 2.606(9)–3.242(7) Å.

3.2 XPS characterization

The neutrality of K₅[(Mo_3.2W_1.8)O₁₁](PO₄)₃[(P_0.5Mo_0.5)O₄] requires a total positive charge of 31.5 for 5.5 (Mo, W) atoms p.f.u., thus implying a mixed-valence state for Mo and/or W (mean oxidation number of +5.73). XPS measurements show the spin orbit splitting characteristic of Mo 3d (Mo 3d _3/2−Mo 3d _5/2 in Figure 4a) and W 4f levels (W 4f _5/2−W 4f _7/2 in Figure 4b). Mo 3d and W 4f spectrum structure can be deconvoluted by means of multi-curve fitting. Shirley background subtraction and mixed Lorentzian/Gaussian (Voigt) peak shape (30 %) were used for the peak fitting. The best fit for molybdenum was obtained by considering 6 fitting curves, i.e. three different doublets which can be connected to the presence of a different oxidation state for Mo [54, 55]: (i) 229.0 (Mo 3d _5/2) and 232.1 eV (Mo 3d _3/2) (Mo⁴⁺, green line in Figure 4a); (ii) 231.5 and 234.6 eV (Mo⁵⁺, blue); (iii) 232.7 and 235.8 eV (Mo⁶⁺, orange). The identification of the various oxidation states was accomplished by measuring the relative shift of the Mo 3d core level peaks with respect to the non-reduced (Mo⁶⁺) form, obtaining a good agreement with the values available in literature [56]. The relative abundance of the three oxidation states estimated on the basis of the proportion among the areas of the three components are 64 % (±6) for Mo⁶⁺, 30 % (±3) for Mo⁵⁺ and 6 % (±1) for Mo⁴⁺. Given the surficial character of the technique (up to a maximum of 5 nm in depth), the information on Mo oxidation states obtained from XPS might not be representative of the Mo speciation in the bulk; nonetheless, the presence of Mo⁵⁺ and Mo⁴⁺ even on the surface of the grains indicates that a significant amount of Mo must be present at its penta- and/or tetra-valent state also in the bulk. Tungsten W 4f core level spectrum was fitted by one doublet with peaks positioned at 35.3 eV (W f _7/2) and 37.4 eV (W f _5/2) (Figure 4b). The overlapped contributions arising from W 5p, Mo 4p and K 3s were properly considered. The doublet was ascribed to W⁶⁺ [57], and no reduced tungsten species was observed.

Figure 4:

High resolution XPS core level spectrum of Mo 3d a) and W 4f b) levels and their spin orbit splitting. The curve fitting of the spin orbit doublets are traced in different colours. Background subtraction is also indicated. a) Molybdenum is present in three different oxidation states (6+, 5+ and 4+), whose peak area was 19,495 CPS × eV, 9174 CPS × eV and 1938 CPS × eV, respectively. For each doublet, only the Mo 3d _5/2 component is marked by a label. b) Tungsten was detected uniquely in its (6+) oxidation state (only doublet W f _5/2−W f _7/2 is visible). The overlapped contributions arising from W 5p, Mo 4p and K 3s are also indicated. See text for the list of peak centre positions and energy difference of doublets.

3.3 Magnetic properties

The evolution of the molar susceptibility (χ _m) and its reciprocal curve (χ _m ⁻¹) versus temperature obtained upon analysis of the powder sample is shown in Figure 5. The data exhibit a conventional linear behaviour in the high-temperature regime (150–300 K), whereas the apparent deviation from the linearity observed at T < 150 K marks the occurrence of relevant antiferromagnetic (AF) interactions among paramagnetic ions.

Figure 5:

Magnetic susceptibility data as function of temperature for the molybdenum phosphate compound, Curie–Weiss fit (inset) with Curie constant C _M = 0.236 emu∙K∙mol⁻¹ (effective moment, μ _eff = 1.37 μ _B) and Weiss constant θ = −321 K.

The high-temperature regime was modelled through a linear regression with the law χ _m = C _M/(T − θ), to obtain an esteem of the Curie (C _M) and Weiss (θ) constant values. C _M and θ values were determined to be 0.236 emu∙K∙mol⁻¹ and −321 K, respectively. The effective magnetic moment, μ _eff, calculated from the Curie constant above 150 K is 1.37 μ _B. From these experimental evidences, the presence of an aliquot of low-valent molybdenum ions can be inferred. Since the second phase, K(MoO₂)(PO₄), revealed by the XRPD investigation of the bulk microcrystalline powder, is diamagnetic (Mo ions are uniquely hexavalent d ⁰) as reported by Lsnyak et al. [58], the paramagnetic ions reasonably result only from the K–MoWP phase.

The negative and relatively high value [59, 60] of the Weiss constant suggests the occurrence of relevant antiferromagnetic interactions among neighbouring paramagnetic ions. This latter indication is compatible with the occurrence of reduced Mo species (e.g., Mo⁵⁺, Mo⁴⁺ as revealed by XPS) in neighbouring M sites. Also the non-linear trend of the χ⁻¹ shown in the inset to Figure 5 is attributable to the occurrence of local AF coupling among neighbouring paramagnetic species (via exchange or super-exchange interactions), as e.g. in diluted magnetic semiconductors [61].

3.4 Relationships with other structural types

The view of the crystal structure along [100] depicted in Figure 2 shows a repeating sequence of two types of pseudo-hexagonal channels along the b-axis. The first type (yellow in Figure 2), with mean diameter of 5.00 Å, resembles those found in potassium titanium phosphate KTiPO₅ (KTP [62], see Figure 2a), although the linkage along [001] in K–MoWP is interrupted by missing tetrahedral sites, thus determining a layered packing. Thus, the stoichiometry of the KTP-like slab is [KMP_2/3O₅]. It is worth mentioning that in the KTP structure, TiO₆ octahedra are linked via common oxygen vertices forming cis-trans helical chains. Furthermore, only cis-linkage of the metal-oxygen chains is detected for KTP closely related structures, such as K₂NiWO₂(PO₄)₂ [63], K₂MgWO₂(PO₄)₂ [64] and KM^II _0.33Nb_0.67PO₅ (M^II: Mn, Co) [65]. In contrast, the framework of K–MoWP presents dissimilarity in the linkage of the MO₆ octahedral chains that follows a complex cis-cis-trans principle. In the second type of hexagonal channels (blue in Figure 2a), having a mean diameter of 4.96 Å, the polyhedral assemblage is similar to that observed in the monophosphate tungsten bronzes with hexagonal channels (MPTBh [66], as shown in Figure 2b).