A new generation of electronic devices that use the spin of the electron instead of its charge as a means to manipulate information has recently emerged. These so called "spintronic" devices exploit effects such as giant magneto-resistance (GMR) in magnetic thin-film heterostructures and are already commercialized in today's high-density hard disk drives. Another potential major economic impact from the discovery of GMR is anticipated to come from nonvolatile magnetic random access memory (MRAM). The keys to next generation devices depend upon the enhancement of magneto-resistive sensitivity and stability of GMR structures, as well as the invention of novel methods to change the magnetizations of one or more ferromagnetic layers. In this dissertation, I have addressed both aspects by improving fabrication processes in various magneto-resistive thin films and developing a novel magnetic memory cell utilizing current pulse induced magnetization switch.

In the study of magnetic multilayer thin films, three advanced process issues have been addressed, these include: (i) the use of exchange coupling as a tool to estimate the critical thickness for the pinhole appearance in ultra-thin Cu films in GMR structures and Al2O3 barrier in tunneling magneto-resistance (TMR) structures; (ii) the role of aluminum oxides and metals as barriers against thermal oxidation of ferromagnetic metals in air; (iii) assessments of ballistic magneto-resistance (BMR) effects into practical devices by using electrochemical deposition to fabricate nanometer size contacts in both thin film and wire geometries. In the study of current pulse induced magnetization switch for MRAM, we have demonstrated domain wall motion in patterned ferromagnetic films for the first time and developed selective bi-stable domain configurations controlled by current pulses. Based on these discoveries, we built and successfully implemented a one-byte memory cell, which has far simpler structure than conventional MRAM.