Acta Cryst. (1992). B48, 747-752 From Crystal Structure Data Towards Reaction Paths in the Solid State? BY A. VEGAS AND M. MARTiNEZ-RIPOLL UEI de Cristalograf[a, Instituto Rocasolano, CSIC, Serrano 119, E-28006 Madrid, Spain (Received 17 January 1992; accepted 15 April 1992) 747 Abstract The structures of FeB, GdSi, Re3B, Ni2In and 6-Ni2Si are analyzed on the basis of their parent metal structures which become modified during com- pound formation. The observed metallic fragments maintain the original topology and distances, and the small distortions observed in them do not occur randomly, but are correlated as in molecular com- pounds. This analysis is not only used to describe the structures in a simpler way, but also to infer plausi- ble reaction mechanisms in the solid state. Introduction Metal clusters have been widely studied and much of the work, aimed at correlating physical properties with structural and bonding aspects, has been col- lected in reviews (Franzen, 1978; Jansen, 1987; Simon, 1981, 1988; Hughbanks, 1989). Until recently, the existence of metal clusters was seen as a logical feature in interstitial compounds, which were considered as bulky metals, slightly dis- torted by the inclusion of small amounts of other atoms (see Wells, 1975). This was also true of the so-called metal-rich compounds, in which the unsaturated valences of the metal atoms are used to form metal-metal bonds. As examples of the latter we may mention Fe3C (Lipson & Petch, 1940) and Cs~O3 (Simon, 1988), both clearly related to Fe and Cs metals, respectively. These bonding aspects led Franzen (1978) to con- sider the metal-rich compounds as modified metals in which the metallic interactions are partially modified as the result of metal-nonmetal bonding. Other examples of relating the metal substructure to the parent metal can be found in the articles of Simon (1981, 1988), dealing with structure and bonding in clusters of metal-rich compounds and clusters of valence-electron-poor metals. More explicit referen- ces to such a relationship can be found in the articles dealing with d ~° metal aggregates (see Jansen, 1987) and in an article devoted to the bonding in clusters and condensed clusters of early transition and rare- earth metals (Hughbanks, 1989). However, in an independent way, and looking for an alternative to the model of O'Keeffe & Hyde 0108-7681/92/060747-06506.00 (1981, 1985) for describing crystal structures, we have shown that the relationship between the struc- ture of the parent metal and the metal substructure in a given compound is not restricted to compounds with unsaturated bonding. It also appears in com- pounds with saturated valences [apatite: Cas(POa)3(OH,F), olivine: Mg2SiO4, etc], even in compounds traditionally considered as ionic (KC103) or those such as TiO2, which is far from being considered a metal-rich compound (Vegas, Romero & Martinez-Ripoll, 1990, 1991). For some of the structures discussed there, a possible reaction mechanism was advanced in the sense that metallic fragments observed in many structures could be seen to be the result of a process of either dynamic deformation or breaking up, initiated in the parent metal. In the present work we apply this model to simple binary compounds such as FeB, GdSi, Re3B, Ni2In and &-Ni2Si in order to deduce a plausible reaction path from the examination of the metallic fragments appearing in them. Discussion The FeB structure and elemental Fe Iron presents the phases collected in Table 1 (Landolt-B/brnstein, 1971). The FeB structure (Hendricks & Kosting, 1930) is orthorhombic, Pnma, with a = 5.495, b = 2.946 and c = 4.053 ~, and is represented in Fig. l(a). The classical description consists of columns of face-sharing Fe6 trigonal prisms running parallel to the b axis. The columns are linked by additional edge-sharing. The B atoms are off-center of these trigonal prisms. Parth+ (1981) has described the structure as a twinned h.c.p. arrangement of Fe atoms with boron in the trigonal prisms generated at the twin plane. O'Keeffe & Hyde (1985) have related this structure to the cation arrays of many sulfates, selenates, chromates, etc. Our alternative description is based on the space between trigonal prisms (shaded in Fig. l a). These apparently empty holes are really fragments of a body centered cubic net, the Fe---Fe distances being within a few percent of those of the high-temperature b.c.c, phase of metallic iron. The fragments are easily © 1992 International Union of Crystallography