A NEW MODEL FOR MAGNETISM IN AMORPHOUS METALS

Abstract

The structure of amorphous rare-earth transition-metal compounds is discussed on the basis of the random packing of atomic spheres. This model is shown to account for several features of the experimental data on amorphous TbFez, and, in particular, is shown to provide a justification for a recent model of the magnetic properties of such materials. 1. Introduction. - Recent experiments by Rhyne et al. (I) show that the intermetallic compound TbFe, has strong magnetic properties even when prepared by sputtering (2) in a non-crystalline state. Measu- rements of the sputtered material show that the saturation~magnetization is reduced from 4.7 pB/atom in the crystalline cubic Laves phase material (3) to 2.8 pB/atom, and that the Curie temperature is reducedifrom 720 K (3) to 388 K. Analysis of the elastic neutron scattering data shows no trace of structure characteristic of the crystalline material. In a recent publication the present authors (4) proposed a model which seems to account for the magnetic properties of the sputtered material. We suggested that other rare-earth transition-metal com- pounds williexhitit similar properties when prepared inzthe non-cryetalline state. Our model depends upon the assumption that the non-crystalline state of rare- earth transition-metal compounds has a topologically disordered or (( amorphous )) structure, of the type given by the random close packing (RCP) of atomic spheres. This structure has been shown by Cargill (5) to be consistent with experimental radial distribution functions (RDF's) for all (( amorphous \> metallic systems so far studied. In the present paper we aiscuss the RCP model in more detail and show that it is consistent with the assumptions of the model for the magnetic properties (4). Some further discussion is then given of the possibility of induced-moment systems in the amorphous state. sizes have been based in an empirical way on the ball- bearing packings of Finney (6) or the computer simulations of Bennett (7! which are both concerned with spheres of one size. We have followed Bennett's technique but have generalized it to RCP's with spheres of two different sizes with a relative concen- tration of 2 : 1. The ratio of radii for the spheres is taken as J? : Ji which is the ideal ratio for the constituent atoms in the ideal close-packed cubic Laves phase structure (8). The RDF for the larger (terbium) atoms resulting from a calculation with 1 700 atoms is shown in figure 1, where it is compared with the RDF obtained by Rhyne (9) from theneutron diffraction data for TbFe,. At first sight agreement is' disappointing, since even the first peak, at 3.24 A, does not agree well with data. Agreement is not improved by mixing in