Synthesis, crystal structure determination of a novel phosphate Ag1.64Zn1.64Fe1.36(PO4)3 with an alluaudite-like structure

The orthophosphate, Ag1.64Zn1.64Fe1.36(PO4)3 crystallizes in an alluaudite-type structure. The chains characterizing the alluaudite structure are then built up from edge-sharing [Fe/Zn6] and [ZnO6] octahedra linked together by PO4 tetrahedra. The Ag+ are located in channels parallel to the c axis.


Chemical context
The first crystal structure of natural alluaudite was determined by Fisher (1955) using a specimen of pegmatite from Buranga-Rwanda. The metallic monophosphates belonging to this large alluaudite family form an important class of materials whose numerous phases present rich chemistry and great structural originality. Moore (1971) proposed the following general formulation for alluaudites: A(2)A(1)M(1)M(2) 2 (PO 4 ) 3 with A and M being cationic sites classified in decreasing order of size (r M(2) <r M(1) <r A(1) <r A(2) ). In this structure, the first site A(1) can host a mono-or divalent cation and a vacancy (&), while the second site, A(2) contains a vacancy (&) as well as a monovalent cation (Moore & Ito, 1979). The other sites, M(1) and M(2), display octahedral geometries, which may contain a distribution of di-and trivalent cations. The natural alluaudite studied by Moore exhibits the following chemical formula: Na 2.5 Li 0.1 Ca 0.5 Mn 4.5 2+ Mg 0.2 Fe 7.9 3+ (PO 4 ) 12 and crystallizes in the monoclinic system, space group C2/c. In the structure of this compound, the cations are distributed over the four types of site as follows: A(1): 2.5Na + + 0.7Mn 2+ + 0.5Ca 2+ + 0.3&, A(2): 4&, M(1): 3.8Mn 2+ + 0.1Mg 2+ + 0.1Li + , M(2): 7.9Fe 3+ + 0.1Mg 2+ . Later, Hatert et al. (2000) proposed a complex and more accurate general formula for the alluaudite structure in ISSN 2056-9890 order to take into account the different cationic sites available within the channels in the structure.
The main characteristic of the alluaudite structure is the remarkable flexibility of its anionic framework, which is amenable to various cationic substitutions in the A and M sites (Chaalia et al., 2012). As a result, a large number of alluaudite compounds with interesting physical properties have been synthesized and systematically characterized. Indeed, the existence of transition metals in the structure is often the origin of interesting properties viz. magnetic , heterogeneous catalysis [e.g., the role of AgCaCdMg 2 (PO 4 ) 3 and AgCd 2 Mg 2 (PO 4 ) 3 in the conversion of butan-2-ol] (Kacimi et al., 2005), electronic conductivity and significant ionic mobility (Richardson, 2003).

Structural commentary
The isolated phosphate, Ag    A layer perpendicular to the b axis, resulting from the connection of vertices between chains and the PO 4 tetrahedra.

Structural model validation
In order to support the current crystal structure determination, CHARDI (CHARge-DIstribution) and BVS (Bond-Valence-Sum) analyses were performed using CHARDI2015 (Nespolo & Guillot, 2016) and EXPO2014 (Altomare et al., 2013) programs, respectively. The results are summarized in Tables 1 and 2. For the proposed structural model, BVS were calculated for all constituent atoms using the dual concept: bond lengths/bond strengths. This robust validation method estimates the oxidation states of atoms [valence: V(i)], evaluates effectively the quality of the crystal structure elucidation and predicts the level of structural strains. In this model, all the nearest ion-counter ion distances less than 3 Å are considered as bonds and taken into account. The CHARDI method is a modern generalization of Pauling's concept of bond strength (Pauling, 1929). This approach introduces directly the interatomic bond distances in a self-consistent computation to assign a geometrically defined bond strength to each bond. This method adopts a Madelung-type approximation of the crystal structures by attributing point charges to the atoms (the formal charge is equal to the oxidation number; Eon & Nespolo, 2015). The CHARDI analysis also involves the distribution of computed ECoN of a central atom among all the neighbouring ligands (Hoppe, 1979). The determination of non-integer ECoN is directly interpreted in terms of atomic charge distribution in crystalline structures. For a well refined structure, the calculated valences V(i) and the Q(i) charges according to BVS and CHARDI concepts must converge towards the weighted oxidation number q(i)Ásof(i) of each atom [where q(i) = formal oxidation number and sof(i) = site occupancy]. The resulting values from both conceptions confirm the expected formal ionic charges of Ag + , Zn 2+ , Fe 3+ , P 5+ and O 2À . In the thirteen independent atomic sites within the asymmetric unit, the cationic charges are located at seven sites, while in the remaining sites the oxygen atoms balance the charges. For all cations, the internal criterion q(i)/Q(i) $ 1, where Q(i) represents the computed charge, imply the correctness of the structure determination (Nespolo et al., 1999). In the structure, all oxygen atoms exhibit a lower over or under bonding (OUB) effect with the exception of atoms O2 and O5, which deviate slightly from the formal value of À2 (Table 1). To estimate the convergence of the (CHARDI) Table 1 CHARDI and BVS analysis for the cations in the title compound.

Figure 4
Perspective view of the crystal structure of Ag model, the mean absolute percentage deviation (MAPD) was computed. MAPD measures the agreement between the q(i) and Q(i) charges for the whole sets of PC (polyhedroncentring) atoms and of V (vertex) atoms (Nespolo, 2016), qðiÞ À QðiÞ qðiÞ where N is the number of polyhedron-centring or vertex atoms in the asymmetric unit. Respecting this experimental distribution scheme, the resulting values of MAPD for the cationic and anion charges are only 1.1% and 2.4%, respectively. This result supports the applicability and adequacy of the current model. In order to prove the chemical plausibility of the crystal structure we have also calculated the Global Instability Index (GII; Salinas-Sanchez et al., 1992). The GII index estimates the coherence of the structure and measures the deviation of the bond-valence sums from the formal valence V(i) averaged over all N atoms of the asymmetric unit. In our case, we found a very good GII index of 0.087 v.u., indicating the stability and the rigidity of the proposed structural model.

Synthesis and crystallization
Single crystals of Ag 1.64 Zn 1.64 Fe 1.36 (PO 4 ) 6 were synthesized by means of a classical solid-state reaction in air. Appropriate amounts of the starting reagents: AgNO 3 , Zn(NO 3 ) 2 Á6H 2 O, Fe(NO 3 ) 3 Á9H 2 O, H 3 PO 4 (85%) were taken in the following molar ratios Ag:Zn:Fe:P = 2:2:1:3. The mixture was dissolved in concentrated nitric acid, stirred at room temperature for 24 h and subsequently evaporated to dryness. The obtained solid was carefully milled in an agate mortar, placed in a platinum crucible and heated up to the melting point of 1223 K. The molten product was maintained at this temperature for 1 h then cooled down slowly to 920 K at rate of 5 K h À1 and then rapidly to room temperature by turning off the oven. The title compound was isolated as yellow parallelepiped-shaped crystals.

Refinement
Crystal data, data collection and structure refinement details are summarized in Table 3. The refinement of all the variable parameters leads to well-defined displacement ellipsoids. In   (Brandenburg, 2006); software used to prepare material for publication: publCIF (Westrip, 2010). Special details Geometry. All esds (except the esd in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s. planes.