1. Chemical context

The growing role of metal orthophosphates based on PO₄ and MO_n (where M is a metal cation) structural units is closely related to their ability to adopt different spatial arrangements. As has been pointed out previously, their physical and chemical properties, dynamic flexibility attributes and structural behaviour (Hadrich et al., 2001) can be correlated with the ionic radius of the metal cation (Jeżowska-Trzebiatowska et al., 1980). Furthermore, the ability of these metal cations to adopt different oxidation states as well as various coordination environments leads, in general, to open anionic three-dimensional frameworks. The structures of these classes of materials can easily accommodate a great variety of substituents, anionic and/or cationic, which can have a significant effect on the stability and on the morphology of structures and crystals, as is shown particularly in the apatite family (LeGeros & Tung, 1983) for which a considerable number of complex and versatile networks were described systematically. Open frameworks involved with various cavities such tunnels or cages, especially in phosphates containing mono, di and trivalent cations, are of particular inter­est owing to their potential applications in catalysis (Badrour et al., 2001) and as immobilizing carriers for various enzymes, e.g. CaTi₄(PO₄)₆ (Suzuki et al., 1991) as well as for their photocatalytic activities in glass-ceramics containing MgTi₄(PO₄)₆ crystals (Fu, 2014) and ionic-conductivity properties with Ca_l-xNa_2xTi₄(PO₄)₆ belonging to the Nasicon structure type (Mentre & Abraham, 1994). Much attention has been paid to compounds containing six PO₄ tetra­hedral units with different transition metal/phosphate ratios, e.g. Na₄CaFe₄(PO₄)₆ which adopts the Alluaudite structure in the ideal C2/c space group (Hidouri et al., 2004), Ba₃V4(PO₄)₆ which crystallizes as a Langbeinite-type structure (Dross & Glaum, 2004), CuTi₄(PO₄)₆ which belongs to the Nasicon family (Kasuga et al., 1999), the silver lead apatite Pb₈Ag₂(PO₄)₆ (Ternane et al., 2000), the mixed-valent iron(II/III) phosphate Fe₇(PO₄)₆ (Belik et al., 2000) and Na_2.5Y_0.5Mg₇(PO₄)₆ with a Fillowite-type structure (Jerbi et al., 2010). Through hydro­thermal processes, and as part of our systematic studies of the crystal alkaline and alkaline earth monophosphates, we have previously succeeded in elaborating a number of compounds with three-dimensional networks featuring distinctive cavities including AgMg₃(HPO₄)₂PO₄ (Assani et al., 2011), Sr₂Mn₃(HPO₄)₂(PO₄)₂ (Khmiyas et al., 2013), SrMn₂^IIMn^III(PO₄)₃ (Alhakmi et al., 2013), NaMg₃(HPO₄)₂PO₄ (Ould Saleck et al., 2015). In an extension of our investigations and structural studies of various mono-divalent transition-metal phosphates a new phosphate copper (Cu^II)-based AgSr₄Cu_4.5(PO₄)₆ was prepared and successfully characterized. Charge-distribution (CHARDI) (Nespolo et al., 2001) and bond-valence-sum (BVS) calculations were used for validating the structural model. A careful examination of the literature as well as various databases reveals that the title compound AgSr₄Cu_4.5(PO₄)₆ is original and furthermore is not related to any family of reported compounds.

2. Structural commentary

The principal building units of the crystal structure of AgSr₄Cu_4.5(PO₄)₆ are more or less distorted polyhedra (AgO₅, CuO₄, CuO₅, SrO₈, SrO₉) and nearly regular PO₄ tetra­hedra, as shown in Fig. 1. In this structure, the copper atoms adopt two different environments: CuO₄ and CuO₅. Indeed, Cu1 and Cu2 exhibit a coordination sphere of four oxygen atoms, forming a flattened parallelogram for Cu1O₄ and a distorted square plane for Cu2O₄. The other copper atoms Cu3, Cu4 and Cu5 each occupy the centers of CuO₅ square-based pyramids. A close inspection of the geometrical parameters of Cu3O₅, Cu4O₅ and Cu5O₅ polyhedra reveals that the latter exhibit significant distortion. The phospho­rus atoms are tetra­hedrally coordinated with bond lengths and angles close to those reported for P⁵⁺ for this geometry. The crystal-structure framework of AgSr₄Cu_4.5(PO₄)₆ can be viewed as a three-dimensional network of corner-sharing CuO_n (n = 4 or 5) units, thereby forming two types of [Cu₃O₁₂]¹⁸⁻ trimers. The first trimer results from the zigzag succession in the following order Cu(4)O₅ – Cu(2)O₄ – Cu(5)O₅. Similarly, the second type of trimer is built up from two-vertex-sharing of a single polyhedra, Cu1O₄, sandwiched by two neighbouring Cu3O₅ entities as shown in Fig. 2. Each oxygen atom of both trimers is connected to a nearly regular PO₄ tetra­hedron in such a way as to form two different [Cu₃P₁₀O₄₀]²⁴⁻ ribbons (see Fig. 3 and 4). These adjacent ribbons are linked together through the PO₄ tetra­hedra, thus building a layer-like [Cu_4.5(PO₄)₆]⁹⁻ arrangement perpendicular to the [100] direction as shown in Fig. 5.

Figure 1 The principal building units in the structure of the title compound. Displacement ellipsoids are drawn at the 50% probability level. Symmetry codes: (i) −x + 1, −y + 1, −z; (ii) −x, −y + 1, −z; (iii) x, y + 1, z; (iv) x − 1, y, z; (v) −x + 1, −y + 1, −z + 1; (vi) x + 1, y, z; (vii) x, y − 1, z; (viii) −x, −y, −z + 1; (ix) −x, −y + 1, −z + 1.

Figure 2 Vertex-sharing [CuO5] and CuO4 polyhedra forming two [Cu3O12] trimers.

Figure 3 A ribbon, resulting from a Cu3O5–Cu1O4–Cu3O5 trimer connection via vertices of PO4 tetra­hedra.

Figure 4 A [Cu3O12] trimer linked to PO4 through corners to build up a ribbon involving Cu4O5–Cu2O4–Cu5O5.

Figure 5 A [Cu3O12] trimer linked to PO4 through corners to form a layer parallel to the (100) plan.

Crystal cohesion and the junction between the stacked layers along the a-axis direction are ensured by ionic bonds involving the Sr²⁺ and Ag⁺ cations as shown in Fig. 6. The insertion of these mono and bivalent cations generates strong inter­actions inducing, consequently, a morphological deformation of the inter­layer space, which explains the manifestation of the distorted sites. This result is confirmed by the CHARDI analysis of the coordination polyhedra by means of the effective coordination number (ECoN; Nespolo, 2016). The distortion of the metal–oxygen polyhedron becomes stronger when the ECoN value deviates further from the habitual coordination number (CN). This structural particularity is clearly noticeable when examining the numerical values of ECoN and CN for the various SrO_n (n = 8 and 9) and AgO₅ polyhedra. The differences ECoN (Sr1)/CN(Sr1) = 7.61/8, ECoN (Sr2)/CN(Sr2) = 6.96/8 and ECoN (Sr3)/CN(Sr3) = 6.8/8, reveal an increased distortion in the SrO₈ groups ranging from the Sr1O₈ to Sr3O₈ polyhedra. The Sr2 atom is formally nine-coordinate with bond lengths varying from 2.480 (2) to 2.890 (2) Å. The site hosting Sr4 is very flexible and bulky, resulting in a greatly deformed SrO₉ polyhedron. The geometry ratio ECoN (Ag1)/CN(Ag1) = 3.93/5 of the Ag1O₅ polyhedron indicates a distorted square-pyramidal coordination environment. This behaviour can be attributed to the edge or face-sharing between these polyhedral units. This modality of linkage, as well as the ionic radius of Sr²⁺ and Ag⁺, induces a strong cation–cation electrostatic repulsion, which is reflected in the inter­atomic distances and consequently on the repetition of the ionic charge and bond-valence-sum (BVS) values.

Figure 6 Three dimensional view of AgSr4Cu4.5(PO4)6 crystal structure showing Sr2+ and Ag+ between layers stacked along the [100] direction.

The CHARDI analysis method gives the distribution of calculated ECoN numbers of a central cation among all the neighbouring anions (Hoppe, 1979). The calculation of this number is related directly to the distribution of charges in crystalline structures. The measure of the correctness of the structure (cation ratio) and of the degree of over or under bonding (anion ratio) is performed via the evaluation of the inter­nal criterion q/Q (where q is the formal oxidation number and Q the computed charge). The charge-distribution method (CD or CHARDI), developed by Hoppe et al. (1989), and the bond –valence (BVS) approach introduced to predict bond lengths in inorganic crystals (Brown, 1977, 1978) provide powerful tools for analysis of the connectivity of crystal structures and the validation of structural models. In the present study, both validation tools, BVS and CHARDI, are applied to the structural model of the title compound. Generally, for a well-refined structure, the calculated valences V(i) obtained by the BVS model and the computed charge Q(i) according to the CHARDI analysis must be in close agreement with the oxidation number of the atoms. The CHARDI computations were carried out with the CHARDI2015 program (Nespolo & Guillot, 2016), while BVS was calculated using PLATON (Spek, 2009). In the asymmetric unit, all atoms are located on general positions (Wyckoff position 2i) of space group P except for Cu1, which is located on a special position (Wyckoff position 1a). The distribution of the electric charges at the 40 crystallographic sites of the asymmetric unit shows that the Ag⁺, Sr²⁺, Cu²⁺ and P⁵⁺ cations fully occupy 16 sites. Otherwise, charge neutrality requires the location of 24 oxygen atoms in the remaining 2i sites. The first results of BVS calculations for Sr3 suggest a valence V(Sr3) = 1.900 v.u. for a coordination number CN = 7. This result can be significantly improved by widening the coordination sphere to 3.1410 Å, which allows the integration of a supplemental oxygen, thus inducing valence V(Sr3) = 1.962 v.u. The analysis of the data summarized in Table 1 reveals that the values obtained from charges Q(i) and bond-valence sums V(i) of the cations are all compatible with the weighted oxidation number q(i)·sof(i). The minor deviations reported from these parameters with respect to the formal oxidation state are closely related to the distortion level of the occupied sites. Despite these irregularities, all the values of the inter­nal criterion q(i)/Q(i) are very close to unity, which confirms the validity of the structural model obtained from the X-ray diffraction data. The convergence of the CHARDI model is evaluated by the mean absolute percentage deviation (MAPD) as shown in the equation below, which measures the agreement between q(i) and Q(i) for the whole sets of PC atoms (polyhedron-centring atoms) and of V atoms (the vertex atoms) (Eon & Nespolo, 2015). For the cationic charges in the structure, we report that the calculated value of MAPD is only 1.7%.

where N is the number of polyhedron-centring or vertex atoms in the asymmetric unit.

The calculated anionic charges Q(i) of oxygen show a lowest deviation of the order of 4.5% with respect to q(i). These values of MAPD show that the dual description as cation-centred and anion-centred is satisfactory and adequate for the studied structural model. The ratio q(i)/Q(i) is approximately equal to 1 in most cases (Table 2), with some exceptions: q(O8)/Q(O8) = 1.16, q(O12)/Q(O12) = 0.92 and q(O22)/Q(O22) = 1.15. This anomaly of negative-charge repetition could be due to the OUB effect (over–under bonding effect) (Nespolo et al., 1999), which results from the repulsive inter­actions of the cations located at the centre of the polyhedra. Therefore the anionic charges of oxygen deviate slightly from the ideal value −2. This also explains the variation of cation–anion distances in the various polyhedra in the crystal structure of AgSr₄Cu_4.5(PO₄)₆.

The plausibility of a crystal-structure model may also be tested by the global instability index (GII) (Salinas-Sanchez et al., 1992). The calculated value of the GII index measures the deviation of the bond-valence sums from the formal valence V_i averaged over all N atoms of the asymmetric unit. For an unstrained structure, GII is below 0.1 v.u. and may approach 0.2 v.u. in a structure with lattice-induced strains (Adams et al., 2004). Values larger than 0.2 v.u. are typically taken as an indication of the presence of intrinsic strains strong enough to cause instability of the crystal structure (Brown, 1992). For the crystal structure of the title compound, GII = 0.0944, which indicates high stability and rigidity of the proposed structural model.

3. Database survey

A search in the ICSD database shows that no compounds are currently known in the quaternary system AgO/SrO/CuO/P₂O₅. The same is true within the AgO/SrO/P₂O₅ ternary system. However, one compound is known in the AgO/CuO/P₂O₅ ternary system, viz. β-AgCuPO₄ which crystallizes in the Pbca space group (Quarton & Oumba, 1983). There are seven compounds known in the ternary SrO/CuO/P₂O₅ system, viz. Sr_9.1Cu_1.4(PO₄)₇, Sr₃Cu₃(PO₄)₄ (Belik et al., 2002; Effenberger, 1999), Sr_2.88Cu_3.12(PO₄)₄ (Karanović et al., 2010), Sr₅(CuO₂)_0.333(PO₄)₃ (Kazin et al., 2003), Sr₂Cu(PO₄)₂, SrCu₂(PO₄)₂ (Belik et al., 2005) and SrCu(P₂O₇) (Moqine et al., 1993). There is no apparent relation between the structures of these compounds and that of the title compound AgSr₄Cu_4.5(PO₄)₆.

4. Synthesis and crystallization

Single crystals of the title compound were obtained using the hydro­thermal method with the following mixture of reagents: silver nitrate, strontium nitrate, metallic copper and 85wt% phospho­ric acid in a proportion corresponding to the molar ratio Ag:Cu:Sr:P = 1:3:1:3. The hydro­thermal reaction was conducted in a 23 mL Teflon-lined autoclave with 12 mL of distilled water under autogenous pressure. The vessel was heated to 473 K for 4 d. After being filtered off, washed with distilled water and dried in air, the reaction product consisted of a light-blue crystals in various forms corresponding to the title compound.

5. Refinement

Crystal data, data collection and structure refinement details are summarized in Table 3. The refinement of the occupation of all atom sites shows full occupancy and leads to the stoichiometric formula AgSr₄Cu_4.5(PO₄)₆. However, the difference-Fourier map shows two electron-density peaks of intensity 4.05 and −3.87 e Å⁻³ located at 0.63 and 0.59 Å from Ag1, respectively. These rather strong peaks could not be removed using a different integration strategy or another absorption model.

Table 3 Experimental details Crystal data Chemical formula AgCu4.50O24P6Sr4 Mr 1314.08 Crystal system, space group Triclinic, P Temperature (K) 296 a, b, c (Å) 9.1070 (1), 9.1514 (1), 13.7259 (2) α, β, γ (°) 97.498 (1), 98.303 (1), 110.875 (1) V (Å3) 1036.97 (2) Z 2 Radiation type Mo Kα μ (mm−1) 16.22 Crystal size (mm) 0.30 × 0.27 × 0.21   Data collection Diffractometer Bruker X8 APEXII Absorption correction Multi-scan (SADABS; Krause et al., 2015) Tmin, Tmax 0.496, 0.747 No. of measured, independent and observed [I > 2σ(I)] reflections 32671, 8573, 7465 Rint 0.028 (sin θ/λ)max (Å−1) 0.806   Refinement R[F2 > 2σ(F2)], wR(F2), S 0.028, 0.059, 1.03 No. of reflections 8573 No. of parameters 359 Δρmax, Δρmin (e Å−3) 4.05, −3.87 Computer programs: APEX2 and SAINT (Bruker, 2009), SHELXT2014/7 (Sheldrick, 2015a), SHELXL2014/7 (Sheldrick, 2015b), ORTEP-3 for Windows (Farrugia, 2012), DIAMOND (Brandenburg, 2006) and publCIF (Westrip, 2010).