Spin Electronics

Goals

The Spin Electronics Project focuses on developing measurements to better understand the interactions between the electron spin in current-carrying electrons and the magnetization of ferromagnetic films. These measurements will allow the investigation of magnetization dynamics at nanometer lengths scales characterized by precession angles that are orders of magnitude larger than previously accessible. They will facilitate the development and continued scaling of spintronic devices. The techniques developed as part of this program will support industrial roadmaps that have targeted magnetic disk drives with terabit-per-square-inch densities and magnetic random-access memory (MRAM) at the 65 nanometer node by the end of the decade. In addition, these measurements will help industry develop smaller and cheaper nanoscale magnetic microwave devices that may replace much larger and expensive on-chip microwave circuitry.

Customer Needs

Until recently, the only means known to control the magnetization state of ferromagnetic structures was through the use of applied magnetic fields. However, within the last several years it has been demonstrated that this can also be accomplished through the transfer of the electron spin angular momentum from current-carrying electrons to the magnetization of magnetic films, generally referred to as the spin-momentum-transfer (SMT) effect. Spin transfer represents a fundamentally new way to control and manipulate the magnetic states of devices, one that emerges only at the nanoscale. It will allow investigation of magnetization dynamics and spin-waves on length scales smaller than previously possible. It creates the ability to switch nanopatterned magnetic storage devices at speeds not previously accessible. It presents opportunities for the development of spin-switched MRAM and active microwave devices operating above frequencies of 100 gigahertz.

Time-domain data showing the relative variation of
the oscillator phase over the locking range for several
values of DC bias current.

At present, the SMT effect is too poorly understood to effectively exploit or diminish its effects in practical devices, and little is accurately known about the dynamics induced by the effect. We will develop metrology to investigate and understand how spin-based effects can be avoided or exploited in magnetic nanostructures in order to assist the magnetic data storage industry and to facilitate the development of nanoscale microwave devices operating at high frequencies. These measurements will help industry develop future generation spintronic devices and facilitate the continued scaling of magnetic data storage into the deep nanometer range.

Technical Strategy

We are performing measurements on magnetic nanostructures in order to investigate magnetization dynamics at nanometer length scales and subnanosecond time scales. The high-speed dynamics and switching events are induced through the spin-transfer effect, which utilizes the transfer of the electron spin angular momentum in nanometric magnetic heterostructures to induce magnetic excitations at length scales down to 10 nanometers and at frequencies up to 100 gigahertz. We are measuring the high frequency properties of the spin transfer induced dynamics in patterned magnetic nanostructures as well as dynamics that are induced locally in continuous films.

We are trying to determine how spin-based effects can be avoided and exploited in magnetic nanostructures in order to assist the magnetic data storage industry and facilitate the development of nanoscale microwave devices. We will quantify the intrinsic device properties that determine the high frequency dynamics of spin-transfer nanoscale oscillators — such as oscillation frequency, linewidth, and power, which currently only qualitatively agree with theory — and develop measurements to relate the stochastic (statistical) switching characteristics of spin-switched MRAM to the component materials properties. For example, we are presently developing techniques to measure the magnetic precessional damping parameter in single, active devices with dimensions below 100 nanometers.

Accomplishments

Micromagnetic simulations showing the spin-wave interaction between two local spin transfer oscillators. Each device emits spin-waves towards the other, causing the devices to synchronize.

Electron micrograph of two nanocontacts (dark circles in light rectangles) with FIB cut between them. Scale bar is 200 nm. Inset: Micromagnetic simulation showing spin-waves emitted by two contacts, and reflections from FIB cut.

Injection Locking in Spin-Transfer Microwave Oscillators — Synchronization of weakly coupled oscillators, generally referred to as injection locking or frequency entrainment, has numerous examples in nature and generally occurs in oscillator systems having at least weak nonlinear interactions. Examples range from biological systems, such as the synchronized flashing of fireflies and singing of certain crickets, to those in the physical sciences, such as Josephson junction arrays and the synchronization of the Moon’s s and Electrical Engineering Laboratory rotation with respect to its orbit about the Earth. This feature is exploited in many modern technologies such as wireless communications, the American power grid, various power combining architectures, and phased array antenna networks. One of the simplest methods of synchronizing electronic based oscillators is through the application of an AC signal close to the oscillator’s natural frequency, inducing the device to oscillate sympathetically at the drive frequency.

We have directly measured phase locking of spin transfer oscillators to an injected AC current. In this scenario, the devices are forced to oscillate at the same frequency as the injected signal. The oscillators lock to signals up to several hundred megahertz away from their natural oscillation frequencies, depending on the relative strength of the input. As the DC current passing through the devices varies over the locking range, time-domain measurements show that the phase of the spin-transfer oscillations varies over a range of approximately plus or minus 90 degrees relative to the input. This is in good agreement with general theoretical analysis of injection locking of nonlinear oscillators.

Mutual Synchronization of Spin-Transfer Oscillators — Our demonstration that spin-transfer oscillators interact with input signals nonlinearly opens the possibility of creating phase-coherent device arrays. The power emitted from a single spin-torque nano-oscillator (STNO) is at present typically less than 1 nanowatt. To achieve a more useful power level, on the order of microwatts, a device could consist of an array of phase coherent STNOs, as has been done with arrays of Josephson junctions and larger semiconductor oscillators. We showed that two STNOs in close proximity mutually phase-lock (that is, synchronize), exhibiting a general tendency of interacting nonlinear oscillator systems.

The phase-locked state is distinct, characterized by a sudden narrowing of signal linewidth and an increase in power due to the coherence of the individual oscillators. For instance, the device oscillator linewidth decreases by a factor of 10 as compared to the individual oscillators, a significant improvement with respect to thermal fluctuations. Furthermore, we showed that the combined power output represents a fully coherent combination of the two signals, a requirement to demonstrate true phase-locking between the devices.

Determination of Synchronization Mechanism in Spin-Transfer Nanocontacts — The fact that two spin-transfer nanocontact oscillators fabricated in close proximity to each other on the same magnetic film can phase-lock provides a means for coherently combining the powers of the devices. Furthermore, the power level increases as the square of the number of oscillators. While a promising result, an important question facing further development is the origin of the phaselocking. Spin-transfer nano-oscillators have at least two possible means of interacting. Since the devices produce magnetic fields, the oscillators could simply interact like two bar magnets, with the field from one affecting the oscillations of the other. Alternatively, since the oscillators are connected via a magnetic film, they could interact via propagating “spin-waves.” Spin-waves are analogous to water waves that propagate from the splash produced by a rock thrown into a pool.

To separate these two possible mechanisms, we devised a set of experiments. First, we fabricated two oscillators on a magnetic film and showed that they interact and phase-lock as in our previous work. Then we physically cut the magnetic fi lm between the oscillators with a focused-ion beam (FIB) nanoscale cutting tool. Cutting the magnetic film would still allow the two oscillators to see the magnetic fields produced by each other but would stop any spin-waves from propagating, just as a barrier dividing a pool of water stops ripples from propagating. We found that phase-locking no longer occurred, showing that the oscillators interact predominately via spin-waves.

These results set the stage for arranging larger numbers of nanocontact oscillators into coherent arrays for use in new types of electronic devices. In addition, this electrical method provides a new way to measure spin-waves themselves on length scales smaller than previously imaginable. This new metrological tool may reveal new magnetodynamic phenomena at nanometer length scales.

Influence of Thermal Effects in the Current- Induced Switching of Magnetic Nanostructures —

Plot of the differential resistance versus current and
magnetic field for an elongated hexagon sample,
50 nanometers by 100 nanometers in size. The
negative- going current sweep was subtracted
from the positive-going current sweep to highlight
hysteretic regions of the device.

The spin-torque effect is an efficient way to change the orientation of the free layer in multilayer nanopillars known as a “spin valves.” This effect has potential applications in commercial MRAM and has recently been demonstrated as a viable alternative to the cross-point writing scheme of conventional MRAM, which is just now coming to market. One of the reasons for pursuing alternative switching schemes is the scaling limitations of conventional MRAM. Spin-torque switching of MRAM has the advantage that, as the size of the devices is reduced, the current needed to switch the free layer orientation is decreased. However, one difficulty with implementing MRAM nanopillars in large-scale commercial products is device-todevice variation of the magnetic properties and the impact of these variations on the critical switching current.

To gauge device-to-device variations, we measured how thermal effects influence the free layer reversal of spin-valve nanopillars via the spin-torque effect. We found that room temperature pulsed switching probability measurements can be used to accurately predict the critical current for switching at “zero temperature.” Low temperature measurements of several different devices provided experimental values of the critical switching current. Device-todevice variations in the critical switching current are drastically reduced at low temperature, with good agreement between experiment and theory.

Detailed Comparison of Spin-Transfer Precessional Dynamics with Theoretical Predictions — Since the initial predictions that a spin polarized current can exert a torque on a nanoscale magnet, much progress has been made in understanding the spin-transfer effect and its manifestations. A number of researchers have successfully demonstrated both current-induced switching and steady-state magnetic precession in patterned magnetic nanostructures, nanowires, and nanocontacts as well as current-induced motion of magnetic domain walls. Theoretical efforts to better understand these effects have included analytical approaches, numerical single-domain modeling, and micromagnetic simulations; a rough general qualitative agreement between experiment and theory regarding the most basic results has been achieved. However, the measured precessional dynamics in magnetic nanocontacts do not always agree with the predictions of both micromagnetic and single-domain modeling. For instance, measurements show much more complicated evolution of the precessional frequency with current and field than predicted by single-domain simulations and spin-wave theories.

We have measured the detailed dependence of the oscillation frequencies, linewidths, and output powers of spin-transfer nanocontact oscillators as functions of applied fi eld strength, bias current, and angle of the applied magnetic fi eld. For fi elds applied only moderately out of the plane of the fi lm, the evolution of these properties is continuous. However, for fi elds applied more strongly out of plane they exhibit discontinuities in both current and applied fi eld. These discontinuities typically correlate with changes in the device resistance, changes in device output power, and a broadening of their spectral linewidths. However, away from these discontinuities, the oscillator output powers are larger and the linewidths narrower when

Measured data from a spin-transfer nanoscale
oscillator comparing the measured device
linewidth and the evolution of the precession
frequency with current and magnetic field.
The linewidth is shown as shading mapped
directly onto the frequency surface so that
the two parameters can be compared.

compared to geometries having the fi elds applied at lower angles. Our measurements suggest that the discontinuous evolution of the frequency with current and applied fi eld results from an abrupt change in precessional trajectories of the magnetization in the free layer. Comparisons with present theories give only qualitative agreement with our experimental results.

Award

U.S. Department of Commerce Silver Medal (Bill Rippard, Stephen Russek, and Tom Silva) and EEEL Distinguished Associate Award (Matt Pufall and Shehzaad Kaka) for the discovery of mutual phase-locking, external frequency-locking, and frequency modulation of spin-transfer nano-oscillators, 2006.