Russian Journal of Coordination Chemistry, Vol. 21, No. 3, 1995, pp. 153-161. Translated from Koordinatsionnaya Khimiya, Vol. 21, No. 3, 1995, pp. 163-171. Original Russian Text Copyright 1995 by Serezhkin, Blalov,Shevchenko
Voronoi-Dirichlet Polyhedra for Uranium(VI) Atoms in Oxygen-Containing Compounds
V. N. Serezhkin, V. A. Blatov, and A. P. Shevchenko
Samara State University, Samara, Russia
Received March 14,1994
Abstract - The geometric characteristics of Voronoi-Dirichlet polyhedra (VDP) for 354 U(VI) atoms in the structures of oxygen-containing compounds are established. It is shown that the volumes of VDPs, as distinct from the volumes of the corresponding coordination polyhedra, are virtually independent of the coordination numbers of uranium atoms and are on average 9.18(25) µ3. The results obtained are in agreement with the previously suggested model of deformable spheres describing the packing of structural units in uranium(VI) compounds.
It is known that the data on crystal structures obtained from X-ray or neutron diffraction investigations represent the totality of point coordinates corresponding to the equilibrium positions of atoms in the unit cell. On the basis of this information, what is known as the description of the crystal structure is created. For this purpose, the points corresponding to atomic positions are linked by segments symbolizing chemical bonds between pairs of atoms; in this way, an infinite unoriented graph is obtained. In crystal chemistry, it is conventional to use the local topological characteristics of this graph, in particular, atomic coordination numbers and coordination polyhedra (CP) for its qualitative and quantitative description. In this case, the coordination number of a certain atom A (that is, the order of a certain vertex of the graph) equals the number of segments (that is, A-X bonds) that converge to the point corresponding to the position of the A atom. The question as to precisely which X atoms are linked to the A atom under consideration (i.e., enter into the first coordination sphere) in the structure of a particular crystal is settled in classical crystal chemistry by analyzing interatomic distances, that is, the lengths of the A-Xi segments. All of the atoms (A, X, etc.) are regarded as rigid spheres [1, p. 22] of a particular radius. In the case when the experimentally found r(A-Xi) distance does not exceed the standard r(A-X) value, which equals the sum of the radii of these atoms, the corresponding pair of atoms is considered to be chemically bonded, and the Xi atom is to be included in the coordination sphere of the A atom.
The crystal structure of a particular compound is considered in this approach as a specific three-dimensional periodic packing of an infinite set of spheres. The atoms of each chemical element in this packing correspond to the spheres of a particular radius; the proportion of spheres of different radii is defined by the stoichiometric composition of the compound, and the coordination number of any atom equals the total number of adjacent spheres touching the sphere corresponding to this atom.
Because the determination of interatomic distances by the data of X-ray diffraction analysis is a trivial problem, it is not surprising that the essentially one-dimensional considered method of determination of atomic coordination numbers in the crystal structure has been widely accepted in spite of a number of disadvantages.
On the one hand, the interaction between two atoms (for example, A and X) in the crystal structure may be generally described in terms of strong (ionic, covalent, and metallic) or weak (van der Waals) chemical bonding. Therefore, not a single but at least several values correspond to different types of bonds for atoms of the same chemical nature in the available systems of crystalline radii. Moreover, for a given type of chemical bond (in particular, ionic), a certain atom is characterized by several more values of radii that depend on the oxidation state and spin state of the atom and even on the value of the coordination number desired. Therefore, it is not surprising that, for a particular pair of atoms A and X, the distance r(A-X) may vary within a large range, even when a strong chemical bond occurs between these atoms. For example, the distances between the U(VI) and oxygen atoms in the structures of known uranyl complexes fall in the range from 1.60 to 2.65 µ [2]. Thus, because a given atom can generally simultaneously participate in the formation of chemical bonds of several types, it is theoretically impossible to represent this atom as a sphere of a fixed radius.
On the other hand, the representation of atoms by rigid undeformable spheres, which has played a great role in the formation of crystal chemistry, is now in contradiction to the modem concept of the nature of chemical bonds. Even on the assumption that the electron-density distribution of the isolated A and X atoms has a spheric shape, the formation of a strong chemical bond between these atoms should be considered a result of the overlapping of their outer valence orbitals, which may be considered as an "interpenetration" of atoms [1, p. 87]. Because it is impossible to estimate the extent of this overlapping (interpenetration) on the grounds of crystal structure data, even the simplest and most reliable atomic radii (covalent or metallic), equal by definition to half the shortest interatomic distance in the structures of simple compounds and true only for the model of touching rigid spheres, lose their physical meaning.
Even without additional arguments, the above-mentioned reasoning evidences the conventionality of the representation of atoms in the crystal as rigid spheres of a particular radius; moreover, the electron density distribution at a certain atom loses its spheric shape under the action of the crystal field. In this connection, the concept of atomic radii in the crystal structure is thought to be, to a great extent, arbitrary [1, p. 82].
An alternative "three-dimensional" method based on the use of the Voronoi-Dirichlet polyhedra (VDP) or atomic domains is known [3] in addition to the considered classical "one-dimensional" method of the determination of atomic coordination numbers.
It is known that the VDP of a certain atom is the polyhedron of the smallest volume formed by the planes passing through the middles of and normal to the segments joining this atom with each of its possible neighbors. For example, in the case when an A atom is surrounded in a crystal by six X atoms forming a CP in the form of a regular octahedron or trigonal prism, the VDP of this atom has the form of a cube or a trigonal bipyramid, respectively. Thus, the coordination number of the A atom is generally equal to the number of faces of its VDP, because each face of the VDP corresponds to one vertex of the coordination polyhedron, and the VDP is dual to the CP. Below, the polyhedra obtained by the procedure used for constructing VDPs at an arbitrary value of Kd {Kd is the coefficient of division of the A-X bond, with a perpendicular plane passing through a point Z on the AX segment; this coefficient equals the ratio r(A-Z)/r(A-X)] are called dual CPs. Thus, the VDP is a special case of the dual CP at Kd = 0.5. A specific VDP corresponds to each crystallographic type of atom in the structure of the compound, and the crystal structure as a whole may be considered as a totality of VDPs touching each other with congruent faces, completely filling the entire space of the crystal, because any point of the crystal belongs to at least one VDP. As noted above [4], in the structure of simple compounds, one can regard the VDP as a geometric image of an atom in a certain crystal field and consider it the region of action of this atom, because any point of this area is closer to "its own" atom than to any adjacent one of the same chemical nature.
Although the method of determination of the atomic coordination number in crystals by means of VDPs has been known for a rather long time, as far as we know it has very rarely been used in the crystal chemistry of inorganic and coordination compounds. On the one hand, this is due to the fact that calculation of the geometric characteristics of VDPs is very tedious, especially for crystals of low symmetry containing a large number of crystallographically different atoms. On the other hand, the application of VDPs in the structural analysis of inorganic and coordination compounds is hindered by the fact that the problem of taking into account the distinctions in the chemical nature of atoms arises in the construction of the regions of atomic action, because the problem of the isolation of the spatial region of an atom of each chemical sort has not been solved even for the case when two sorts of atoms are present. It was shown recently [4] that, for structures with several chemical sorts of atoms, the VDP describes the region of action of a certain atom, correct to similarity, under conditions of uniformity of the nearest environment of this atom, that is, the occurrence of only atoms of a single chemical sort in the first coordination sphere of the basis atom. In accordance with [5], the regions of atomic action should be thought of as the basins of nuclei-attractors in the vector field of the electron density gradient.
This work was undertaken with the aim of studying more thoroughly the equality of the VDP volumes of uranium(VI) atoms found in [4] for crystals of 30 compounds. Simultaneously, we planned to verify the possibility of the application of VDPs for the determination of the coordination numbers of U(VI) atoms in the structures of oxygen-containing compounds using a wider variety of objects.
GEOMETRIC CHARACTERISTICS OF THE VORONOI-DIRICHLET POLYHEDRA OF U(VI) ATOMS
The calculation of characteristics of the VDPs of U(VI) atoms in the structures of oxygen-containing compounds, including uranyl complexes and simple and mixed oxides (uranates), was carried out by the structurally topological TOPOS software [6]. These programs make it possible to calculate atomic VDPs in crystal structures of any degree of complexity. The original information on crystal structures contained in the Cambridge crystallographic databank and the databank for uranyl compounds [7] was used in the calculations, provided that three conditions were fulfilled: (1) the crystal structure was determined with Rf