Home      Faculty      Research      Fast Dynamics Laboratory      Contact us

Project 2: Nanoscale Spin-Transport Systems

2.2. Spin Transfer in Tunnel Junctions

This research is based on studies performed on chromium oxide systems in a collaboration between Dowben, performing photoemission and inverse photoemission studies, and Doudin. We have shown that the conduction of CrO₂ granular films is governed by intergrain tunneling through a thin (1-2 nm) insulating Cr₂O₃ film. The large MR observed at low temperatures is enhanced by Coulomb blockade effects, as shown by electron-transport measurements, and by a combination of photoemission and inverse photoemission spectra showing a spectacular change of the band gap with temperature.
At temperatures where the Coulomb blockade is suppressed by the thermal energy, we observed hysteretic I-V curves. The resistance of the sample at low voltages depends on the polarity of a previously applied voltage of less than 1 V amplitude (see Figure 1). The effect disappears when a magnetic field is applied, which clearly establishes the magnetic nature of the effect. We tested the speed of the switching by applying a voltage pulse as short as 100 ns, which is the limit of our experimental set-up.
A second central component of this research, conducted in the W. M. Keck Fast Dynamics Laboratory, is to measure fast conductance in junctions, spin valves, and other structures electronically. Such magnetoresistance measurements have proved useful in spin-valve structures, and they will also be of great value in characterizing and understanding the properties of the nanojunctions proposed here. They will make use of the fast digitizing oscilloscope and require a picosecond pulse generator to provide the electrical excitation. This setup will require design and construction of a new probe for our existing high-field magneto-electronics cryostat.

Figure 1: Conductance versus applied voltage for a CrO2/Cr2O3/CrO2 junction. At low temperature (1.6 K), the strong non-linearity is due to Coulomb blockade effects. At higher temperatures, the conductance changes by 20% when reversing the sign of the applied voltage.

We plan to use our measurements of switching speeds as the starting point for investigating the use of tunnel junction systems for making new memory devices based on spin-transfer effects. The geometry of the system is depicted in Figure 2, where an “impurity” in a tunnel junction has a magnetization that can be reversed by an electric current; the magnetization direction depends on the polarity of the applied current. Experiments on metallic multilayers showed that the exchange interaction between the nanoparticle and the electrodes is ferromagnetic when the electrons flow from a large magnetic electrode, and antiferromagnetic when the electrons flow from the nanoparticle. If the two ferromagnetic electrodes have antiparallel magnetic orientation, or if one of the electrodes is non-magnetic, we expect a change of resistance when reversing the polarity of the current, related to the TMR value. The resistance value of our samples (5 - 10 kW) is of the ideal order of magnitude for applications, which is a significant advantage over metallic systems (having resistances of a few tens of W at most). However, the necessity of a large current density for switching the magnetic orientation of the island is the main difficulty for tunnel junctions. Note however that heating effects are small in our chromium oxide junctions. We are aware of similar hysteresis curves in two other tunnel-type system.

Figure 2: Schematic of a double-junction magnetic geometry, where the magnetization of the island is reversed by the spin-polarized flow of electrons. The corresponding resistances are indicated in a simple model of sequential tunneling with an asymmetrical position of the island. The shaded magnetization depicts possible non-ferromagnetic electrodes, with related equal values for RAPAP (l) and RP(l).

One essential goal of this research is to investigate and understand more clearly how the intermediate localized impurity states in the bandgap of the Cr₂O₃ barrier contribute to the magnetoresistive and hysteresis properties. We plan to change the occupation rates of these states by illuminating the junctions with photons of variable energy range. The study of exciton states in Cr₂O₃was a pioneering example of investigations of magnetic insulators and revealed a rich absorption spectroscopy in the energy range 1.6 – 2.5 eV. Evidence of Cr³⁺-Cr⁴⁺ center recombination, and absorption line energies are known.
The value of our approach is that the transition selection rules differ slightly in photoexcitation and inverse photoemission so that more insight into the symmetry and magnetic properties are possible through the comparison of the techniques. This method is identical to that used to compare inverse photoemission and X-ray absorption spectroscopies. A recent review has been published by the group of Dowben, who performed the XPS investigations on the chromium oxide films, and is already investigating the spin-dependent photoemission spectra.
Following the diagram of Fig. 2, we intend to make junctions of good dielectric properties, allowing large current densities to flow. One ideal candidate is boron carbide (B₅C), for which Dowben has unique and extensive experience. Implantation of Ni clusters has been already performed on B₅C, and we intend to test if a non-magnetic top layer will allow the spin-transfer effect to be observed. A non-magnetic electrode is an advantage to increasing the difference in the hysteresis curve (see the resistor schemes of Fig. 2), but might be detrimental to the magnitude of the effective exchange interaction with the magnetic island.