The alluaudite-like arsenate NaCaMg3(AsO4)3

The title compound, sodium calcium trimagnesium tris(arsenate), an alluaudite-like arsenate, was prepared by solid-state reaction at high temperature. The structure is built up from edge-sharing MgO6 octahedra in chains associated with the AsO4 arsenate groups. The three-dimensional network leads to two different tunnels occupied statistically by Na+ and Ca2+. One As and one Mg atom lie on twofold rotation axes; one Na and one Ca are disordered over two sites with occupancies of 0.7 and 0.3 and these sites lie on a twofold rotation axis and an inversion centre, respectively.

Supplementary data and figures for this paper are available from the IUCr electronic archives (Reference: BR2069).

Comment
The crystal structure of NaCaMg 3 (AsO 4 ) 3 is closely related to the common structure type of the well known mineral Alluaudite with the general formula X(1)X(2)M(1)M(2) 2 (PO 4 ) 3 (Moore, 1971;Yakubovitch et al., 1994). It can be described by a Mg 3 (AsO 4 ) 3 framework built up by a complex arrangement of distorted MgO 6 octahedra, and AsO 4 tetrahedra. The projection of the structure in a polyhedral representation is presented in Fig. 1. It consists of Mg1O 6 and Mg2O 6 octahedra that share edges to form staggered chains stacked parallel to the [10-1] direction. Equivalent chains are linked together through the AsO 4 tetrahedra corners. As1O 4 connects two chains and thus two of its O atoms belong of the same chain.
The As2O 4 tetrahedron shares his four oxygen summits with four different MgO 6 octahedra belonging to three adjacent chains, two belong to the same chain and the two others from two different chains. The arrangement of magnesium octahedra Mg1O  1.691Å and As2-O: 1.694 Å. This structural arrangement delimits two types of hexagonal tunnels, parallel to the c axis and located at (1/2, 0, z) and (0,0,z) respectively. Sodium Na1, Na2 and calcium Ca1, Ca2 cations are located in those channels.
Whereas the X(2) site at (0, 0, 0) contains 0.30 Na2 and 0.70 Ca2. The site in the tunnel at (0, 0, z) shifted from the X(2) site by ± 0.25 along z is occupied by Na1 and Ca1 with respectively the occupation number 0.70 and 0.30. There are a number of possible models for the cationic distribution and it's not possible to decide which is the best solution. We retain the solution with same amount of sodium and calcium. First, the occupancies of Na1 and Ca1 in the site shifted from X(2) were refined.
Second, for the site X(2), the occupancies of Na2 and Ca2 were fixed to obtain the electroneutrality. For each two cations in the same site the atomic displacement parameters were maintained the same with the instruction EADP. The bond valence sum of the Na1, Na2, Ca2, Mg1, Mg2, As1 and As2 atoms are in a good agreement with their oxidation states (Brown & Altermatt, 1985). For the calcium Ca1 which occupy partialy the tunnel the bond valence sums is different (1.33).

Experimental
Single crystals of NaCaMg 3 (AsO 4 ) 3 were prepared by a mixture of NaNO 3 , CaCO 3 , MgN 2 O 6 (H 2 O) 6 and NH 4 H 2 AsO 4 with molar ratio of (1:1:2:3). The powder was ground, then heated in a porcelain crucible progressively until 1223 K. This temperature was held for 3 days. Then the mixture was cooled slowly to room temperature at 10 K/h. The product was washed with hot water. Prismatic and colorless crystals of the title compound were extracted. Their qualitative analysis by electron microscope probe revealed that it contains sodium, calcium, oxygen, arsenic and magnesium atoms.

Data collection
Enraf-Nonius CAD-4 diffractometer R int = 0.021 Radiation source: fine-focus sealed tube θ max = 28.0º Monochromator: graphite θ min = 2.4º T = 293 (2) K h = −15→14 ω/2θ scans k = −1→16 Absorption correction: ψ scan (North et al., 1968) l = 0→8  Refinement. Refinement of F 2 against ALL reflections. The weighted R-factor wR and goodness of fit S are based on F 2 , conventional R-factors R are based on F, with F set to zero for negative F 2 . The threshold expression of F 2 > σ(F 2 ) is used only for calculating Rfactors(gt) etc. and is not relevant to the choice of reflections for refinement. R-factors based on F 2 are statistically about twice as large as those based on F, and R-factors based on ALL data will be even larger.