GMR Multilayers And Head Design For Ultra High Density Recording

Abstract

Recently reported work on application of giant magnetoresistance (GMR) technology toward building ultra high density (> 1 Gbithnch') magnetic reproduce heads appears to favor the spin-valve design as a replacement for the anisotropic magnetoresistive (AMR) sensor in an otherwise conventional shielded-MR configuration.' One potential problem of trying to similarly substitute in an antiferromagnetically-coupled GMR multilayer, is that unlike a single active layer spin-valve, the high sense current densities (J, > lo7 A/cm*) required for both biasing and adequate signal levels will produce large magnetic fields internal to the multilayer which can induce a graded, opposite polarity transverse bias magnetization in each half of the multilayer.2 A picturization of this effect in an 2Nmagnetic multilayer is shown in Fig. 1. However, such a bipolar bias magnetization distribution is a 2N-layer GMR analogue to the previously described AMR dual-magnetoresistive (DMR) head.3 Like a DMR with an effective layer thickness teff = N t , and reproduce gap geff = 4/3 N(t+t'), such a selfbiased "GMR-DMR" head can achieve extremely high linear resolution without shields. In addition to the obvious potential for much increased AR/R ratio, a GMR-DMR should be intrinsically selfstabilized by the multilayer's antiferromagnetic interlayer exchange coupling without the fabrication complexity of additional exchange pinning layers. With N >> 1, the cross-track response should be approximately symmetric. It can be shown2 that the characteristic flux propagation length A of a practical GMR-DMR will be roughly a few tenths micron, and so this head design is naturally most suited for the submicron active sensor dimensions useful for ultra high density recording.