Magnetoresistance In Magnetic Tunnel Junctions Research Articles

On the calculation of the magnetoresistance of tunnel junctions with parallel paths of conduction

At the interfaces between the metallic electrodes and barrier in magnetic tunnel junctions it is possible for localized states to form which are orthogonal to the itinerant states for the junction, as well as resonant states that can form at the interfaces. These states form highly conducting paths across the barrier when their orbitals point directly into the barrier; these paths are in addition to those formed by the itinerant states across the entire junction. Most calculations of transport in magnetic tunnel junctions are made with the assumptions that the transverse momentum of the tunnelling electrons is conserved, in which case the itinerant electron states remain orthogonal to localized states. In principle it is possible to include diffuse scattering in both the bulk of the electrodes and the barrier so that the transverse momentum is not conserved, as well as the processes that couple localized states at the electrode-barrier interface to the itinerant states in the bulk of the electrodes. However, including these effects leads to lengthy calculations. Therefore, to assess the conduction across the barrier through the localized states that exist in parallel to the itinerant states we propose an approximate scheme in which we calculate the conductance of only the barrier region. While we do not take explicit account of either of the effects mentioned above, we do calculate the tunnelling through all the states that exist at the electrode-barrier interfaces by placing reservoirs directly across the barriers. To calculate the current and magnetoresistance for magnetic tunnel junctions (the junction magnetoresistance (JMR)) we have used the lattice model developed by Caroli et al. The propagators, density of states and hopping integrals entering the expressions for the current are determined by using the spin polarized scalar-relativistic screened Korringa-Kohn-Rostoker method that has been adapted to layered structures. By using vacuum as the insulating barrier we have determined with no adjustable parameters the JMR in the linear response region of tunnel junctions with fcc Co(100), fccNi(100) and bcc Fe(100) as electrodes. The JMR ratios that we find for these metal/vacuum/metal junctions are comparable with those measured with alumina as the insulating barrier. For vacuum barriers we find that tunnelling currents have minority- spin polarization whereas the tunnelling currents for th se electrodes have been observed to be positively (majority) polarized for alumina barriers and minority polarized for SrTiO 3 barriers. In addition to determining the JMR ratios in linear response we have also determined how the magnetoresistance of magnetic tunnel junctions varies with a finite voltage bias applied across the junction. In particular we have found how the shape of the potential barrier is altered by the applied bias and how this affects the current. Comparisons with data as they become available will eventually determine whether our approximate scheme or the ballistic Landauer-Büttiker approach is better able to represent the features of the electronic structure that control tunnelling in magnetic tunnel junctions.

Bulk contributions to tunnel magnetoresistance in magnetic tunnel junctions

We have carefully designed an experiment to elucidate the interface and bulk contributions to spin-polarized transport in CoFe-based magnetic tunnel junctions (MTJ's). We demonstrated that the bulklike contribution to tunneling magnetoresistance (TMR) exists in the magnetic tunneling process, which is determined by the tunneling characteristic length of the ferromagnetic electrodes. This characteristic length in MTJ's is different from the spin diffusion length in metallic based spin valves or magnetic multilayers due to the fundamental different transport behaviors in two structures. The experiment designed with a wedge-shaped CoFe layer inserted at the interface between the insulating barrier and reference electrode offers a unique and simple way to determine this characteristic length, which is about 0.8 nm for the CoFe electrode. For interfacial contributions to dominate TMR, the magnetic electrodes should have a very short tunneling characteristic length of atomic scale.

Reduced temperature and bias-voltage dependence of the magnetoresistance in magnetic tunnel junctions with Hf-inserted Al2O3 barrier

A modified tunnel barrier structure for the magnetic tunnel junction (MTJ) was fabricated by inserting a Hf layer in the middle of the Al2O3 tunnel barrier. MTJs with the Hf-inserted barrier show a higher tunnel mangnetoresistance (TMR) ratio and weaker temperature and bias-voltage dependence of TMR compared to the MTJs with a conventional Al2O3 barrier. The enhancement of the TMR ratio and the reduction of the temperature and bias-voltage dependence were attributed to the reduction of defects in the barrier.

Voltage dependence of magnetoresistance in magnetic tunnel junctions with AlN tunnel barrier

The voltage dependence of magnetoresistance (MR) and the low-frequency voltage fluctuation were studied in magnetic tunnel junctions with an AlN or an Al2O3 barrier formed by plasma nitridation or oxidation. The junctions consist of free and pinned NiFe layers by a FeMn antiferromagnetic layer. For the Al2O3 junction, the variation of half-reduction voltage of MR is about 50 mV from 401.3 mV in maximum to 351.4 mV in minimum depending upon the oxidation time. For the AlN junction, the less nitrided junction with a nitridation time of 120 s shows the higher-half-reduction voltage by about 100 mV than for optimally nitrided or oxidized ones. From the low-frequency voltage noise data, the less nitrided AlN junction shows larger 1/f noise magnitude compared with other junctions with an optimal condition, implying more defect states in the barrier due to unnitrided Al metals. In order to investigate the influence of nitrogen on ferromagnetic layer, another junction with an Al2O3 barrier was made after the pinned NiFe layer was exposed to N2 plasma for 10 s. This junction reveals a lower MR and worse bias voltage dependence than any junctions studied here. The voltage dependence of MR, therefore, is likely influenced by the interface state rather than the defects in the barrier, although the two factors play an important role in determining MR.

Do ballistic channels contribute to the magnetoresistance in magnetic tunnel junctions?

The barrier quality and conduction mechanisms in ferromagnet–insulator–ferromagnet (FIF) junctions are studied by fabricating adjacent ferromagnet–insulator–superconductor (FIS) junctions, using the same barrier and top electrode in both junctions. Transport measurements on the FIS junctions provide information about the transport processes in the FIF junctions. The FIS junctions have a large zero-bias conductance (ZBC) that varies with barrier thickness, in part due to ballistic channels in parallel with the tunnel barrier. As the ZBC increases, the magnetoresistance (MR) in the FIF junctions decreases, implying that these ballistic channels are not magnetoresistive and that their presence reduces the total MR in the FIF junctions.

Tunneling criteria for magnetic-insulator-magnetic structures

The bias and temperature dependent resistance and magnetoresistance of magnetic tunnel junctions with and without intentional shorts through the insulating barrier were studied. Based on the experimental results, a set of quality criteria was formulated that enables the identification of barrier shorts. While the temperature and bias dependencies of the junction resistance and of the fitted barrier parameters are very sensitive to the presence of such shorts, the same dependencies of the magnetoresistance are surprisingly insensitive. Finally, junctions with a shorted barrier exhibit a dramatic increase in noise level and junction instability.

Criteria for ferromagnetic–insulator–ferromagnetic tunneling

The Rowell criteria, commonly used to identify tunneling in magnetic tunnel junctions (MTJ), are scrutinized. While neither the exponential-thickness dependence of the conductivity nor fits of non-linear transport data are found to be reliable tunneling criteria, the temperature-dependent conductivity does remain a solid criterion. Based on experimental studies of the bias and temperature-dependent resistance and magnetoresistance of MTJs, with and without shorted barriers, a new set of criteria is formulated.

Tunnel magnetoresistance in magnetic tunnel junctions with ZnS barrier

A first experimental evidence of a significant tunneling magnetoresistance signal of about 5% at 300 K for a magnetic tunnel junction consisting of hard and soft magnetic layers separated by a 2 nm ZnS semiconducting barrier is reported. The samples have been grown by sputtering on Si(111) substrate at room temperature and have the following structure: Fe6 nmCu30 nmCoFe1.8 nmRu0.8 nmCoFe3 nmZnSxCoFe1 nmFe4 nmCu10 nmRu3 nm. The hard magnetic bottom electrode consists of the artificial antiferromagnetic structure in which the rigidity is ensured by the antiferromagnetic exchange coupling between two FeCo layers through a Ru spacer layer. Barrier impedance scanning microscope (BISM) measurements reveal a good homogeneity of the barrier thickness. Electric transport measurements over square tunnel elements with lateral sizes between 3 and 100 μm, exhibit a typical tunnel current–voltage variations and tunnel resistance of 2–3 kΩ μm2 with small variations which never exceed a factor of 2, which is in good agreement with the BISM results. This good reproducibility of the junctions is very promising for MRAMs and transistors applications.

Interface composition in magnetic tunnel junctions

The magnetoresistance of magnetic tunnel junctions critically depends on the exact composition at the interfaces. As such the completeness of the oxidation process of the Al layer (used to produce Al2O3-based tunnel junctions) plays an essential role in the magnetoresistance. We studied the chemical properties of ferromagnet/Al2O3 interfaces as a function of original Al layer thickness. We have studied the concentrations of elementary and oxidized Al, Co, Ni, and Fe for varying roughness of the ferromagnetic layer. The oxidation process critically depends on the roughness of the underlying ferromagnetic (FM) layer. Al layers grown onto smooth FM layers oxidize homogeneously whereas Al layers grown on rough FM layers show a complicated oxidation behavior. Within the sensitivity of the analysis technique, we did not observe oxidation of the ferromagnetic layers, even for the overoxidized part of the samples.

Tunnel magnetoresistance in magnetic tunnel junctions with a ZnS barrier

We report on the junction magnetoresistance in magnetic tunnel junctions of the hard–soft type with magnetic layers separated by a ZnS barrier. The hard magnetic bottom electrode consists of an artificial antiferromagnetic structure in which the rigidity is ensured by the antiferromagnetic exchange coupling between two FeCo layers through a Ru spacer layer. The samples were grown by sputtering on Si (111) wafers at room temperature and have the following structure: Fe6 nmCu30 nm(CoFe)1.8 nmRu0.8 nm(CoFe)3 nmZnSx(CoFe)1 nmFe4 nmCu10 nmRu3 nm. The square tunnel elements, with lateral sizes of 10, 20, 50, and 100μm, exhibit typical tunnel resistance of 2–3 kΩ μm2 and nonlinear zero field current–voltage (J–V) variation. The most interesting result is the observation of junction magnetoresistance of about 5% at room temperature with a 2 nm thick ZnS barrier.

A study on tunneling magnetoresistance in magnetic tunnel junctions oxidized by ozone

Magnetic tunnel junctions of Ta/NiFe/FeMn/NiFe/CoFe/Al 2O 3/CoFe/NiFe structure were studied. Two different types of oxidation atmosphere were used to form the barrier layer; plasma oxidation and ozone oxidation. The maximum tunneling magnetoresistance ratio of 32% was observed at room temperature for 50 min ozone oxidation. The junction resistance was several hundred ohms at 0.1 mm×0.2 mm junction. This is about one order lower than that of the plasma oxidized junctions. The plasma free oxidation process would eliminate the possible structural damage by an oxygen ion bombardment and give improved control over the tunnel barrier growth.

Temperature-dependent magnetoresistance of magnetic tunnel junctions with ultraviolet light-assisted oxidized barriers

Co(10 nm)/AlO x ( nominally 2 nm)/Co(20 nm) tunnel junctions have been prepared under ultrahigh vacuum conditions applying a shadow mask technique. An ultraviolet light-assisted oxidation process of the AlOx barrier has been optimized by in situ x-ray photoelectron spectroscopy, in conjunction with temperature-dependent tunneling magnetoresistance measurements. Optimum-oxidized tunnel junctions show a magnetoresistance of 20% at 285 K, and up to 38% at 100 K. For under-oxidized samples, with a remaining Al layer between the Co bottom electrode and the AlOx barrier, the tunneling magnetoresistance decreases more rapidly with increasing temperature than observed for the over-oxidized samples. The resistance × area product of optimum-oxidized tunneling junctions exhibits a minimum, and increases for under- and over-oxidized samples.

Reduced magnetoresistance in magnetic tunnel junctions caused by geometrical artifacts

Spuriously reduced magnetoresistance (MR) ratios have been observed in magnetic tunnel junctions in which a square contact portion with dimensions smaller than the width of the lead electrodes connects both the top and bottom lead electrodes. The phenomenon becomes apparent by measuring the magnetoresistance of the junctions with various sizes systematically varied under a fixed line width of the electrodes. Observed junction size dependence of resistance (R)×area(A) products and MR ratios were analyzed through finite difference calculation, and it was found that there exist junction sizes for which R×A products and MR ratios are larger and smaller, respectively, than the intrinsic ones.

Magnetoresistance of magnetic tunnel junctions in the presence of a nonmagnetic layer

Magnetoresistance of magnetic tunnel junctions in the presence of a nonmagnetic layer

Correlation between micromagnetism and tunnel magnetoresistance in magnetic tunnel junctions using artificial ferrimagnets

The impact of the micromagnetic configuration on transport properties of magnetic tunnel junctions is presented. An artificial ferrimagnet system (AFi), made of two ferromagnetic layers with different magnetic moments antiferromagnetically coupled by exchange interactions through a Ru spacer layer, is used as a hard magnetic subsystem of the magnetic junction. Magnetic-force microscopy studies have shown that the magnetization reversal in artificial ferrimagnet systems occurs through the formation of N\'eel-type $360\ifmmode^\circ\else\textdegree\fi{}$ domain walls. The stability of these walls is demonstrated to be different in the two magnetic layers of the AFi. These aspects have major effects in magnetic tunnel junction devices, due to the extreme sensitivity of spin tunneling to spatial variations in local magnetic order. The large tunnel magnetoresistance (TMR) of our samples, in the range of 20--30 % at room temperature, makes the tunnel junctions highly sensitive to magnetic fluctuations. A sharp switching of the soft magnetic layer upon field reversal prevents a domain structure from occurring in the soft magnetic layer, in the field window used for our micromagnetic studies. The tunnel device can thus be used as a sensitive probe for measuring small magnetic fluctuations associated with micromagnetic defects, domains, and walls in the AFi system. These fluctuations modulate the resistance of the tunnel junction and are fully reflected in the shape and the amplitude of the TMR signal.

Models for magnetoresistance in tunnel junctions

We show there are putative pitfalls when one predicts the magnetoresistance of magnetic tunnel junctions (JMR) based on different toy models. Amongst them are the sensitivity of the MR to the details of the profile of the potential barrier between the metallic electrodes and the insulating barrier, and the common assumption of only one band of electrons. We indicate the ingredients that are necessary to obtain a more complete description of the JMR of magnetic tunnel junctions.

Interface structures and magnetoresistance in magnetic tunnel junctions

The interface structures of magnetic tunnel junctions were studied using x-ray photoelectron spectroscopy (XPS). The structures were correlated with magnetoresistance (MR) characteristics. For MR measurements, Fe(50 nm)/AlOx/CoFe(30 nm) junctions with an in situ naturally oxidized Al tunnel barrier were fabricated. The thickness of the Al layer, an important parameter in MR characteristics, was varied from 0 to 5 nm. MR curves showed that the largest MR ratio occurred when the Al layers were 2–3 nm in thickness. XPS analysis showed that an Al layer greater than 1 nm thick covers the entire surface of the Fe underlayer. However, if the Al layer is more than 1 nm thick, the unoxidized Al remaining after the oxidation process increases as the thickness is increased. For Al layers that are greater than 3 nm thick, the MR ratio is strongly affected by unoxidized Al, probably due to the decrease in spin polarization at the surface of an Fe/Al electrode. On the other hand, the hysteresis loops indicate that the difference in coercive force between Fe and CoFe layers reduces with decreasing Al thickness for Al layers less than 2.5 nm thick. This means that the antiparallel direction of magnetization in the two layers becomes incomplete due to the gradual increase of the ferromagnetic coupling between them. As a result, the MR ratio decreases, although a 1-nm-thick Al layer seems to be enough to cover the Fe surface.

Magnetoresistance in Magnetic Tunnel Junctions with Nonmagnetic Metallic Spacers and Cap Layers

In a free-electron approximation, transport properties of magnetic tunnel junctions with nonmagnetic metallic spacers and cap layers were studied. The magnetoresistance is oscillatory functions of the thicknesses of nonmagnetic metallic spacers and ferromagnets, their period is determined by the Fermi surface properties of the metals, and magnetoresistances (up to about 103%) much greater than those predicted by the Julliere's model are obtained. The results indicate that this structure has potential in designing spin-polarized tunneling devices.

Magnetoresistance of Magnetic Tunnel Junctions in the Presence of a Nonmagnetic Layer

We find that electron propagation across a uniform nonmagnetic metallic layer ${d}_{i}$ in a magnetically layered structure is coherent up to a critical length ${d}_{c}$. This provides a possible explanation for the slow decay of the magnetoresistance of magnetic tunnel junctions when a layer of Cu or Ag is inserted between the magnetic electrode and the barrier. However, for many (most) structures the roughness of the interfaces breaks the coherence for ${d}_{i}\ensuremath{\ll}{d}_{c}$. While the loss of coherence is not fatal for the magnetoresistance of metallic multilayers, it is for tunnel junctions. The quantum well states arising from the insertion of a nonmagnetic layer in a magnetic tunnel junction do not promote magnetoresistance.

Low-field magnetoresistance in magnetic tunnel junctions prepared by contact masks and lithography: 25% magnetoresistance at 295 K in mega-ohm micron-sized junctions (abstract)

Interest in magnetic tunnel junctions (MTJ) has been increased by the recent observation of large room-temperature magnetoresistance (MR) in structures comprised of sandwiches of two ferromagnetic (FM) layers with different coercivities separated by a thin Al2O3 tunnel barrier. Using an ultrahigh vacuum dc magnetron sputtering system, a variety of MTJ structures have been explored. Junctions were fabricated directly using computer-controlled placement of successive contact masks, and by patterning using optical lithography techniques and processes, compatible with MR head structures. Structures were prepared in which one of the FM layers is exchange biased with an antiferromagnet MnFe layer. Thus, the junctions exhibit two well-defined magnetic states in which the FM layers are either parallel or antiparallel to one another. The tunnel barrier was prepared by in situ plasma oxidation of thin Al layers sputter deposited at room temperature. Using FM layers comprised of Co or permalloy (Ni81Fe19), MR values exceeding 30% were obtained at room temperature in low fields (5–10 Oe). Similar magnetic and MR properties are found in junctions ranging in size from 80×80 μm2 (contact masks) to 2×2 μm2 (lithography). The smallest junctions, of area 2×2 μm2, have resistance values of ∼5 MΩ and 25% MR. Thus, changes in resistance of ∼1 MΩ are found. Such structures may form an excellent magnetic switch. However, for many device applications, much lower resistances are required. The junction resistance can readily be varied by several orders of magnitude by varying the thickness and oxidation conditions of the Al layer as well as by varying the thicknesses and nature of the layers within the electrodes themselves. Note that it is important that the voltage drop laterally across the electrodes must be small compared to that across the tunnel barrier itself. This requires measurement of junctions with low sheet resistances in lithographically patterned devices.