Structural features of the oxidonitridophosphates K3 M III(PO3)3N (M III = Al, Ga)

Cubic K3 M III(PO3)3N (M III = Al, Ga) is isostructural with (M III = Al, V, Ti). In the potassium compounds, the (PO3)3N6– anion coordinates in a tetradentate manner to two potassium cations.


Figure 1
A view of the asymmetric units of K 3 Al(PO 3 ) 3 N (I) and K 3 Ga(PO 3 ) 3 N (II), with displacement ellipsoids drawn at the 50% probability level.

Figure 3
The coordination environment of potassium cations in (I) and (II).
K-O distances ranging from 2.623 (3) to 3.261 (3) Å , which includes three mono-and three bidentately coordinating (PO 3 ) 3 N 6anions. K2O 9 N and K3O 9 N polyhedra are formed as a result of one tetra-and three bidentately coordinating oxidonitridophosphate anions. In the latter case, the upper boundary for K-O distances is 3.412 (8) Å ; K2-N distances are 2.904 (6) and 2.977 (18) Å and K3-N contacts are 3.340 (6) and 3.310 (18) Å for (I) and (II), respectively. The coordination environments around the alkali metal for Kcontaining oxidonitridophosphates (K2O 9 N and K3O 9 N polyhedra) differ from those of the Na-containing compounds (Na2O 6 N and Na3O 6 N polyhedra). In addition, the (PO 3 ) 3 N 6anions coordinate the two potassium cations in a tetradentate manner (Fig. 4). As is shown schematically in Fig. 4 and in Table 3 for the isotypic M I 3 M III (PO 3 ) 3 N compounds, the M I , M III and N atoms are arranged along the [111] direction in the sequence -M1 I -M2 I -N-M3 I -M III -M1 I -whereby the M2 I -N-M3 I distances change in a different manner. In case of (I) and (II), the shape of the (PO 3 ) 3 N 6anion is similar (the P-N distances are about 1.70 Å and the P-N-K3 angles are within 83-84 ; Tables 1 and 2). For the Na-containing analogues, the P-N bond is slightly larger (1.71-1.74 Å ) and the P-N-Na3 angles are smaller within a wider range from 75 to 78 (Conanec et al., 1994Zatovsky et al., 2006;Kim & Kim, 2013).
To clarify some points regarding the structural changes we have calculated the bond-valence sums (BVS) for the K, P and Al atoms (I) and Ga atoms (II), respectively. Parameters were taken from Brown & Altermatt (1985) and for the K-N bond from Brese & O'Keeffe (1991). The BVS for positively charged atoms in (I) is 21.64 v.u. and 22.28 v.u. for (II) while the sum of the charges of nine O and one N atom is equal to 21 v.u. (Table 4). The higher values for the Ga-containing compound might be explained as follows. The BVS for Al in [AlO 6 ] was found to be 3.00 v.u. while for Ga in [GaO 6 ] it is 3.20 v.u. The remaining atoms also show a slight overbonding (Table 4). We suppose that the anionic part (PO 3 ) 3 N 6is rigid enough and cannot be stretched to larger sizes relative to the larger [GaO 6 ] octahedron into a more expanded framework. This is the reason why shorter interatomic K-O and P-O distances are observed in the structure of (II) compared to that of (I) (Tables 1 and 2). As expected, the unit-cell parameters of (I) are smaller than for (II), in good agreement with the ionic radii of Al 3+ and Ga 3+ (Shannon, 1976). In other words, the rigid and almost flat 'three-blade propeller' anions combine with [M III O 6 ] octahedra to form the framework in which the cavities for the alkali cations become smaller as greater octahedra are involved. Moreover, the greater [M III O 6 ] octahedra strongly influence the 'three-blade propeller' anion, resulting in slightly shorter P-O and P-N bonds. On the other hand, the local environments of the K cations should also be mentioned. In the K2O 9 N polyhedron, the BVS for K2 is 1.12 v.u. in (I) and 1.07 in (II). The contribution to the K2-N1 bond to the valence sum is 0.12 v.u. for (I) and 0.10 for (II). These high values indicate a strong interaction between the two atoms. The cavities in which K1 and K2 are located become larger with longer or almost the same KÁ Á ÁO and KÁ Á ÁN contacts, while the cavities in which K3 is situated become smaller with shortened KÁ Á ÁO contacts in (II) in comparison with (I). BVS calculations for the phosphate tetrahedra show similar results (Table 4).
It should also be noted that further theoretical calculations of the electronic structure can bring final clarity to the principles of bonding of alkali cations with (PO 3 ) 3 N 6anions in K 3 M III (PO 3 ) 3 N compounds. This could be a way for the creation of new materials with desired properties based on K-containing oxidonitridophosphates. Coordination of K2 and K3 cations by the (PO 3 ) 3 N 6anion for (I) and (II).  Table 4 Distances (Å ) between atoms for (I), (II) and isotypic M I 3 M III (PO 3 ) 3 N compounds along [111].

Synthesis and crystallization
For the synthesis of (I) and (II), KH 2 PO 4 , K 4 P 2 O 7 Á10H 2 O, urea, Al 2 O 3 or Ga 2 O 3 (all analytically or extra pure grade) were used as initial reagents. The sequence of preparation procedure was as follows: (1) phosphates KPO 3 and K 4 P 2 O 7 were each prepared by calcining KH 2 PO 4 and K 4 P 2 O 7 Á10H 2 O at 873 K; (2) a mixture of 20.07 g of KPO 3 , 13.21 g of K 4 P 2 O 7 and 30.03 g of urea (molar ratio K:P = 1:1.32, urea:P = 2:1) ground in an agate mortar, was heated to complete degassing at 623 K in a porcelain dish. The resulting solid was reground and heated to become a homogeneous liquid at 1023 K and then quenched by pouring the melt onto a copper sheet to form a glass. The glass was crushed using a mill, and a powder with a particle size of less than 125 mm was separated. According to chemical analysis, the prepared glass had the composition K 1.32 PO 2.43 N 0.50 ; (3) a mixture of 10 g of glass (K 1.32 PO 2.43 N 0.50 ) and 0.3 g of Al 2 O 3 or 0.7 g of Ga 2 O 3 powders were placed into porcelain crucibles and heated up to 1043 K and then cooled to 923 K at a rate of 25 K h À1 . After cooling to room temperature, colorless tetrahedral crystals of (I) or (II) were washed with deionized water. Elemental analysis indicated the presence of K, Al (or Ga), P and N in the atomic ratio 3:1:3:1. The growing of well-shaped crystals of (I) and (II) suitable for single crystal X-ray diffraction analysis was one of main tasks during the present study. Hence, a similar way for the preparation of the potassium-containing phosphates to that for the previously reported sodium-containing compounds was applied, following the self-flux method for the preparation of crystalline nitrogen-containing phosphates (Zatovsky et al., 2006). Thermal decomposition of urea is a multistage process and leads to the formation of C 3 N 4 (Wang et al., 2017). The initial M I -P-O-N (M I = alkali metal) melt can be obtained by the reaction of urea with alkali metal phosphates, when a mixture of phosphates and C 3 N 4 interact. The change of the phosphate:C 3 N 4 ratio (or phosphate:urea) and the nature of the alkali metal affects the composition of the resulting melt. In our case, the molar ratio of K:P was chosen to be 1:1.32 because a mixture of KPO 3 and K 4 P 2 O 7 in this ratio has the lowest melting point close to 886 K, and the urea:P ratio was set to 2:1. As a result, a glass of composition K 1.32 PO 2.43 N 0.50 was obtained.
The solubilities of Al 2 O 3 and Ga 2 O 3 in the K 1.32 PO 2.43 N 0.50 self-fluxes differ significantly. Crystallization of compound (I) occurs as a result of the interaction of self-fluxes and 2-4%wt. Al 2 O 3 . The formation of a mixture of (I) and Al 2 O 3 was observed when the initial amount of aluminum oxide was higher than 5%wt. The solubility of Ga 2 O 3 in the self-flux is about 7%wt. at 1043 K, and subsequent cooling of the homogeneous melt led to the crystallization of compound (II). In all cases, the amount of nitrogen in the self-fluxes rapidly decreases above 1063 K. This process occurs due to the thermal instability of P-N bonds at high temperatures, and leads to a redox reaction with the release of nitrogen and phosphorus. The latter vaporizes from the phosphate melts and starts to burn, which can be observed by periodical sparks on the melt surface (Zatovsky et al., 2000). As a result, K-M III -P-O (M III = Al, Ga) melts prone to vitrifying are formed.
The obtained compounds (I) and (II) were further characterized using FTIR spectroscopy; FTIR spectra were collected at room temperature on KBr discs using a Thermo NICOLET Nexus 470 spectrophotometer. As can be seen in Fig. 5, the spectra of (I) and (II) are similar with respect to intensities and band positions (the difference in the band positions does not exceed 27 cm À1 ). The characteristic bands are in good agreement with the presence of the N(PO 3 ) 3 6anion with C 3v symmetry, which provides for a set of vibration modes: 6A1 + 5A2 + 11E (3A1 + A2 + 4E belong to stretching vibrations, and 3A1 + 4A2 + 7E are due to deformation vibrations). As shown in Fig. 5, the following regions can be distinguished in the FTIR spectra: (1) the bands in the region between 980 and 1220 cm À1 are assigned to as and s (PO 3 ) stretching vibrations [four absorption bands in the frequency range between 1070 and 1220 cm À1 belong to as (P-O), and two bands of between 980 and 1060 cm À1 correspond to s (P-O)]; (2) as (P-N-P) and s (P-N-P) bands can be observed in the regions between 920 and 950 cm À1 and around 920 cm À1 , respectively; (3) the range between 400 and Acta Cryst. FTIR spectra of (I) and (II). sitions, we had expected that both structures should be isostructural with the previously reported Na 3 Al(PO 3 ) 3 N and Na 3 Ti(PO 3 ) 3 N structures. In fact, analysis of the single-crystal data showed that both compounds crystallize in the same space group type (P2 1 3) as the Na-containing oxidonitridophosphates. Originally, the crystal structures were solved by direct methods but we also performed refinements using the atomic coordinates of Na 3 Ti(PO 3 ) 3 N as a starting model. The results were the same, confirming that both structures are isostructural with Na 3 Ti(PO 3 ) 3 N (as well as with all previously reported cubic oxionitridophosphates with the same formula type).  DIAMOND (Brandenburg, 2006); software used to prepare material for publication: WinGX (Farrugia, 2012), enCIFer (Allen et al., 2004) and 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.

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.
Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Å 2 )