398 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1960 



In a sample which is not damped, there are dimensional changes 

 associated with the magnetization along some crystal direction. If 

 the sample were clamped so that its surfaces could not move, magnetiz- 

 ing it would introduce large stresses within the material. This 

 coupling between the elastic properties of a magnetic crystal and its 

 magnetic properties is termed "magnetostrictive coupling," or mag- 

 netostriction. If a sample is stressed, extra terms are added to the mag- 

 netic energy. It is as if additions were made to the magnetocrystal- 

 line anisotropy energy. However, whereas in these crystals the aniso- 

 tropy energy is cubic, a stress might act along a single direction or be 

 uniform within a plane; thus the stress-induced extra anisotropy 

 might make a certain direction the preferred direction, or might make 

 the magnetization prefer to lie perpendicular to a direction. The 

 dimensional changes associated with magnetization are only one or 

 two parts per million in yttrium iron garnet. 



This is all very pertinent to the domain structures we see here be- 

 cause the samples are easily strained. In fact it is very difficult to 

 obtain samples which are not strained in such a way as to strongly 

 affect the magnetization. The mechanical polishing procedure de- 

 scril>ed briefly above seems to leave the surface of the crystal strained. 

 The result is that there is a stress-induced easy direction of magnetiza- 

 tion normal to the surface. If there are areas of greater strain than 

 others such as might be expected below a place where a scratch from 

 a coarse abrasive was polished out, there will be special forces on the 

 magnetization in that region. If the stress is relativelj^ uniform over 

 the surface, the domain structure will consist of fairly simple pat- 

 terns — long sections of straight parallel walls are typical, though very 

 interesting patterns can be obtained. Plate 6 is an example of such a 

 structure. If, on the other hand, the effects of local strains are large, 

 the pattern will be irregular. It might well be that there are regions 

 in which the domain walls, starting from the bottom surface, are 

 out of register with those near the top surface. Plate 7 shows the 

 domain structure seen in a crystal which is badly scratched. In addi- 

 tion to the main scratcli, it will be seen tliat there are several otlier 

 distinguishable lines which seem to prefer to have a domain wall 

 along them. These represent the stressed volume of material below a 

 scratch which has been polislied out at least until the surface is 

 smooth. 



Some of the strain with which we must reckon arises in the growth 

 process of the crystal. As far as we know, no one has yet been able 

 to produce a specimen in which the domain structure was not in some 

 degree determined by the state of stress of the sample. However, by 

 etching away the mechanically polished surface, it is possible to 

 achieve patterns which are not completely dominated by this surface 



