Spin-Dependent Tunneling: A New Magnetic Sensing Technology
Quantum mechanical tunneling, when combined with magnetic scattering due to the spin of electrons, results in an exciting new phenomenon called spin-dependent tunneling. Sensors based on this property will doubtless enter the commercial arena within the next two years.
Carl H. Smith, Robert W. Schneider, and Dexin Wang
A new technology based on magnetic random access memory research is finding fertile ground in magnetic field sensing for applications that require very small size, low cost, and high sensitivity to small changes in the magnetic field. Many of today’s solid-state sensor technologies rely on changes in resistance to provide a measure of change in the property to be measured, such as with RTDs, piezoresistive-based pressure sensors and accelerometers, and Hall and magnetoresistive sensors. Other sensors rely on resistance change due to the external stimulations of a semiconductor, e.g., the photoelectric effect. The mechanism described in this article is based on the conduction of electrons through an insulator.
When is an insulator not an insulator? When it is thin enough to allow the quantum mechanical wave function of conduction electrons in a conductor on one side of the insulator to extend through the insulator to a conductor on the other side. This phenomenon is called quantum mechanical tunneling. Quantum mechanical tunneling, when combined with magnetic scattering due to the spin of electrons, results in an exciting new phenomenon: spin-dependent tunneling (SDT). Spin-dependent tunneling materials are also referred to as magnetic tunnel junctions (MTJs).
SDT materials are a recent addition to those materials that exhibit a large change in resistance with a magnetic field. In contrast to giant magnetoresistance (GMR) structures, which use conductive layers to separate magnetic layers, SDT structures use a thin insulating layer to separate two magnetic layers. Changes of resistance with a magnetic field of 10% to >70% have been observed in SDT structures (reported in EETimes, January 15, 2004). The field required for maximum change in resistance depends on the composition of the magnetic layers and the coupling between them. An applied magnetic field of a few gauss (hundreds of microTeslas) can change the direction of the mµgnetization of one of the layers, offering the possibility of extremely sensitive magnetic sensors. Complete prototype sensors with a barrier thickness of 1.5 nm aluminum oxide (Al2O3) have demonstrated up to 40% tunnel magnetoresistance (TMR) and sensitivities of
30 mV/V/G (3 mV/µT at 10 V).
The cross section of an SDT structure is shown in Figure 1.
Figure 1. The layered structure of a simple spin-dependent tunneling film stack has three essential components: a free layer shown here as nickel-iron (NiFe); the aluminum oxide (Al2O3) barrier; and a pinned ferromagnetic structure made of iron-cobalt/chromium-platinum-magnesium (FeCo/CrPtMn). The full stack thickness may be on the order of 50 nm.
The film stack contains a “free layer” and a “pinned layer” separated by a thin-film tunnel barrier. The free layer is a magnetic thin film that is able to freely rotate in the presence of a magnetic field. The pinned layer is a structure in which the ferromagnetic orientation is fixed. The relative magnetization orientation between the films at the two interfaces to the tunneling barrier layer determines the resistance of the device. It is a sensitive function of magnetic fields acting on the free layer.
The resistance change in an SDT structure can be 5–20 × greater than resistance changes in GMR devices, and up to 200 × greater than those in anisotropic magnetoresistance (AMR) devices. Because the sensitivity of a device nominally scales with change in resistance vs. change in field, SDT technology has excellent potential for high-performance, low-field magnetic sensors.
The thin insulating barrier can withstand only a few tenths of a volt. For sensors to operate at a few volts, they must be constructed from several MTJs connected in series. Each consists of a photolithographically patterned SDT stack (see Figure 2).
Figure 2. The layers and structure of a pair of shape-biased spin-dependent tunneling junctions are shown here. The magnetization in the lower cobalt-iron (CoFe) layer is pinned as part of the structure consisting of a chromium-platinum-magnesium (CrPtMn) antiferromagnetic and the antiferromagnetically coupled cobalt-iron/ruthenium/cobalt-iron (CoFe/Ru/CoFe) sandwich. The NiFeCo free layer on the bottom responds to the applied field. The shape factor biases the free layer along its long dimension.
Junctions in sensors are usually designed in series-connected pairs so that contact can be made to the top layer of the junction by subsequent metal layers. The magnetization in the lower cobalt-iron (CoFe) layer is pinned as part of a multilayer structure consisting of a chromium-platinum-magnesium (CrPtMn) antiferromagnetic and the antiferromagnetically coupled cobalt-iron/ruthenium/cobalt-iron (CoFe/Ru/CoFe) sandwich. The nickel-iron-cobalt (NiFeCo) free layer on the bottom responds to the applied field. The narrow shape of the free layer causes the magnetization of this layer to orient along its long dimension in the absence of an external field. Therefore, the directions of magnetization are perpendicular at zero field.
Figure 3. Plotting the response of a spin-dependent tunneling junction illustrates the dependence of resistance relative to the directions of the pinned and free layers.
The field to be measured is applied parallel to the axis of the antiferromagnetic structure and across the narrow dimension of the free layer. Depending on field direction, the resistance increases as the moments become more antiparallel and decreases as they become more parallel (see Figure 3). Shape-biasing the free layer obviates the requirement for field coils to bias the free layer perpendicular to the fixed layer, thus reducing the power required by the sensor.
The individual SDT junctions are then connected in four series chains to make an SDT Wheatstone bridge sensor. Several junction pairs are connected in series for each arm of the bridge (see Figure 4). Plated NiFe bodies serve to concentrate the magnetic field on two of the bridge arms and to give the device directional sensitivity. The junctions forming the other two bridge arms are placed under the
Figure 4. A Wheatstone bridge sensor can be constructed from series-connected SDT junction pairs and flux concentrators (A). The shielded resistors are indicated in the schematic of the bridge connection in (B).
NiFe flux concentrators, effectively shielding them from the applied field and allowing them to act as reference resistors.
SDT sensors can be fabricated with standard silicon microprocessing methods and are suitable for use in complex magnetometer systems as well as simple event detectors, eddy current detectors, and other devices. They can be used discretely or deployed in arrays for perimeter security, vehicle detection, and unattended sensor systems. Figure 5 shows two types of 3-axis magnetometers, each incorporating three discrete SDT transducers.
SDT sensors are attractive for low-frequency nondestructive evaluation applications such as the detection of hidden cracks or corrosion and other deeply buried flaws. By way of contrast, inductive probes do not work well at low frequencies because they are sensitive to the time derivative of the magnetic field rather than to the magnitude of the field created by eddy currents around the flaw. Detection of deep cracks requires the use of large-diameter excitation coils to increase the penetration depth of the eddy currents into the
Figure 5. These two 3-axis magnetometer systems have separate SDT transducers for each of the three orthogonal axes. The system in (A) uses field feedback to maintain each of the three transducers in an optimally sensitive regime. The ultra-low-power version in (B) can be powered by the RTS (request to send) line of a standard computer serial port.
material of interest. Tested on edge cracks at the bottom of both a single plate and a stack of thick aluminum plates, SDT eddy current probes detected an edge crack 3 mm long, 3 mm high, and 18 mm below the surface. A crack 15 mm long, 3 mm high, and 23 mm below the surface was detected at the bottom of a 2-layer aluminum structure.
Although the concept underlying spin-dependent tunneling or magnetic tunnel junctions is approximately 30 years old, its commercial potential was not appreciated until the mid 1990s. Since then, many companies have been developing the technology for use in MRAM and read-heads. It is now being exploited also as low-field magnetic sensors. Its inherent properties of high sensitivity, small size, and modest power requirements are moving solid-state magnetic field sensing into new markets.
Sensors based on spin-dependent tunneling will doubtless enter the commercial arena within the next two years. At first, they will simply replace older technologies whose applications are well defined, but shortly thereafter they will go where only the need but not the specifications have been identified. These new applications could include physiological monitoring, advanced magnetic imaging, and areas yet to be discovered.
In addition to the other researchers at NVE, the authors wish to express gratitude for support from the Air Force Research Laboratory, the Army Research Laboratory, the Defense Advanced Research Projects Agency, the Missile Defense Agency, NASA, the Navy, and the National Science Foundation.
Carl H. Smith, Ph.D., is Senior Physicist, Robert W. Schneider is Director of Marketing–Advanced Technology, and Dexin Wang, Ph.D., is Senior Physicist, NVE Corp., Eden Prairie, MN; 952-996-1603, firstname.lastname@example.org.
For further reading on this and related topics, see these Sensors articles.