Crystal structure of Ba12Co11O33. Reinvestigation using the superspace group approach of orthorhombic oxides A1+x(A'xB1-x)O3 based on [A8O24] and [A8A'2O18] layers

We report the structure determination on a single crystal of Ba12Co11O33, which belongs to the family of 1-dimensional structures related to the 2H hexagonal perovskite. The structure can be derived from the hcp stacking of [Ba8Co2O18] and [Ba8O24] layers. Assuming a similitude in the building principle with that present in the homologue rhombohedral series based on [A3A'O6] and [A3O9] layers, a generic model within the superspace formalism of the whole family is proposed and used successfully for Ba12Co11O33. Two different strategies are presented. The first one considers the structure as a commensurate modulated composite containing two subsystems [CoO3] and [Ba]. With the [CoO3] subsystem chosen as the reference, the superspace group is Fddd(00γC)0s0 with a = 11.4129(2) Å, b = 19.7677(2) Å, c1 = 2.4722(1) Å, and q = (6/11)c1* (Z = 8 for the formula Ba12/11CoO3). The final global R value is 4.46% for 4625 independent reflections (at a I/σ(I) > 3 level) and only 123 refineable parameters. The second model follows the natural description of the structure as formed by the uniform hcp stacking of the two types of layers [Ba8Co2O18] and [Ba8O24]. The structure is formulated as BaCo11/12O33/12 (Z = 16) and considered to be a commensurate modulated phase with crenel occupational modulations. The superspace group can be denoted as Xdcd(00γL)qq0 with a = 11.4129(2) Å, b = 19.7677(2) Å, c = 4.5324(1) Å, and q = (1/6)c*. The final global R value is 4.36% for the same number of independent reflections and 151 refineable parameters. Both approaches yield similar results, demonstrating for the first time the feasibility of refining these structures as layerlike modulated structures, instead of composites. The chains consist of polyhedra sharing faces formed by 10 consecutive CoO6 octahedra followed by one trigonal prism. Previous models based on simulated and high-resolution electron microscopy images are refuted