A b s t r a c t --W h e a t s t o n e bridge magnetic field sensors using Giant Magnetoresistive Ratio (GMR) m u l t i l a y e r s w e r e designed, f a b r i c a t e d , a n d evaluated. The GMR ranged from 10% to 20% with saturation fields of 60 Oe to 300 Oe. The multilater resistances decreased linearly with magnetic field and showed little hysteresis.I n one sensor configuration, a permanent magnet bias was placed between two pairs of magnetoresistors, each pair representing opposite legs of the bridge. This sensor gave a bipolar bridge output whose output range was approximately GMR times the bridge source voltage.The second sensor configuration used shielding on one resistor pair, and it gave a bridge output dependent on the magnetic field magnitude, but not polarity, and the output range was approximately one half GMR times the bridge source voltage.Field amplifications of 3 to 6 were accomplished by creating a gap in a low reluctance magnetic path, thus providing the full range of outputs with 1/3 to 1/6 of the intrinsic saturation fields of the GMR multilayers.

Newly developed materials that exhibit large changes in effective resistance with applied fields are being put to practical use. Magnetic multilayers with giant magnetoresistance (GMR) and spin dependent tunnelling (SDT) structures are being used in magnetic field sensors. Spin valves are being sold in read heads for hard drives and galvanic isolators. Both spin valves and SDT structures are being used in non-volatile random access memory development. After a brief introduction to these materials, the development of their uses in sensors, read heads, isolators and non-volatile memory are summarized. GMR magnetic field sensors represent a small, but growing market. SDT sensors have the potential to sense very small fields (to 1 pT). Spin valve read heads have enabled very high aerial packing densities for hard drives, up to 24 Gbits per square inch. GMR isolators can be used to duplicate the function of opto-isolators, but at much higher speeds and packing densities. Application of these materials to non-volatile random access memory could result in speeds and densities of semiconductor memory with the non-volatility of hard disk drives. Future directions in this field indicate a merging of semiconductor and these new magnetic materials.

Memory cells have been fabricated and tested to demonstrate storage in the pinned layer of a giant magnetoresistance (GMR) spin valve film. The spin valve was top pinned with a FeMn film and gave about 4% GMR ratio. The memory cell consisted of an oblong, 0.6 μm×7.0 μm GMR bit with first metal contacts at each end and a perpendicular first and second metal word line passing over the bit. Joule heating due to current pulses through both the memory cell and the word line raised the temperature of the FeMn pinning layer above its Néel point. The magnetic field generated by the word line current switched the pinning direction, depending on the polarity of the word line current. Sense line currents up to 5 mA provided a half select without disturbing the bit. In combination with a 5 mA sense current, the bit was written with a word current pulse of 190 mA. The improved thermal stability of the pinned storage layer memory cell is shown to become necessary as the size of a magnetoresistive memory cell drops below about 0.1 μm×0.4 μm.

Three chopping techniques to address 1/f noise in spin-dependent tunneling magnetoresistive sensors are investigated. These include modulation of the sensitivity using orthogonal fields, modulation, and second-harmonic generation using the nonlinear response of the magnetoresistive element and modulation of the flux concentrator permeability. Of these, only the second technique resulted in a slight reduction in low-frequency noise. In order to achieve significant noise reduction by chopping, domain noise will have to be reduced.

Very high density magnetoresistance random access memory (MRAM) cells may be subject to thermal upset. This article describes designs that enhance thermal stability and increase ultimate density by using the combination of heat and magnetic field for writing data. The basic storage mechanism can be shape anisotropy, the coupling between an antiferromagnetic layer and a ferromagnetic layer, or a combination of the two. Two designs are described in this article. The first uses a low Curie point material with high shape anisotropy at room temperature. These cells use active semiconductor devices to restrict heating current to only one cell in an array. The second approach employs the interface coupling between a ferromagnetic film and an antiferromagnetic film as the storage mechansism. A cell may be written by heating above the Néel temperature and cooling the interface in a magnetic field by using orthogonal lines for heating and magnetic field. Heating and cooling times are a few nanoseconds. These design approaches could lead to stable MRAM cells with diameters less than 0.1 μm and requiring lower drive currents.