This project develops instruments, techniques, and theory for the understanding of the high-speed response of commercially important magnetic materials. Techniques used include linear and nonlinear magneto-optics, and pulsed inductive microwave magnetometry. Emphasis is on broadband (above 1 gigahertz), time-resolved measurements for the study of magnetization dynamics under large-field excitation. Research concentrates on the nature of coherence and damping in ferromagnetic systems and on the fundamental limits of magnetic data storage. Exploratory research on spin-electronic systems and physics is underway. The project provides results of interest to the magnetic-disk-drive industry, developers of magnetic random-access memory, and the growing spintronics community. Project members have measured deleterious magnetic turbulence during the magnetic switching process, evanescent flux-pulse propagation in metallic films, and anisotropic coupling (damping) between uniform excitations and the crystal lattice. Coherent-control methods have been used to switch magnetization without unwanted precessional ringing. An inductive current probe was developed to assess trace-suspension interconnects for disk-drive recording heads.
Our primary customers are the magneto-electronics industries. These include the magnetic-disk-drive industry, the magnetic-sensor industry, and those companies currently developing magnetic random-access memory (MRAM). As commercial disk drives approach data-transfer rates of 1 gigabit per second, there is increased need for an understanding of magnetization dynamics. In addition, measurement techniques are needed that can quantify the switching speeds of commercial materials. Once the response of a material has been benchmarked, the engineer can develop electronic components (e.g., heads, disks, or MRAM) that can fully exploit the bandwidth potential of the material.
We are providing novel metrology for the burgeoning spintronics industry. The spin precession of charge carriers in semiconductor hosts has significant potential for telecommunications applications. Unlike the case of conventional semiconductor switching, the frequency of spin precession is not fundamentally limited by the physical thickness of dielectric spacers. We plan to investigate novel magnetic/semiconductor heterostructures of interest to the telecommunications industry.
The focus of this project is the measurement of switching time of magnetic materials for applications in data storage. This has led to the development of instrumentation and experiments using magneto-optics and microwave circuits. Microwave coplanar waveguides are used to deliver magnetic-field pulses to materials under test. In response, the specimen's magnetization switches, but not smoothly. Rather, the magnetization vector undergoes precession. Sometimes, the magnetization can precess nonuniformly, resulting in the generation of spin-waves or, in the case of small devices, incoherent rotation.
Our technical strategy is to identify future needs in the data-storage and other important industries, develop new metrology tools, and do the experiments and modeling to provide data and theoretical underpinnings.
We concentrate on two major problems in the magnetic-data-storage industry: (1) data-transfer rate, the problem of gyromagnetic effects, and the need for large damping without resorting to high fields; and (2) storage density and the problem of thermally activated reversal of magnetization.
Data-transfer rates are increasing at 40 percent per year (30 percent from improved linear bit density, and 10 percent from greater disk rotational speed). The maximum data-transfer rate is currently 100 megabytes per second, with data channel performance approaching 1 gigahertz. In two years, frequencies for writing and reading will be well into the microwave region, which raises the question, "How fast can magnetic materials switch?"
The current laboratory demonstration record for storage density is over 16 gigabits per square centimeter (100 gigabits per square inch). How much farther can longitudinal media (with in-plane magnetization) be pushed? Can perpendicular recording, patterned media with discrete data bits, or heat-assisted magnetic recording extend magnetic recording beyond the superparamagnetic limit at which magnetization becomes thermally unstable? As the data-storage industry seeks its own answers to these pressing questions, we must strive to provide the necessary metrology to benchmark the temporal performance of new methods of magnetic data storage.
We have sought to extend magneto-optics for the quantitative measurement of magnetization dynamics in practical ferromagnetic films. Methods include time-resolved generalized magneto-optic ellipsometry (TRe-GME), time-resolved second-harmonic magneto-optic Kerr effect (TRe-SHMOKE), and quantitative wide-field Kerr microscopy. All these systems rely upon RF waveguide technology for the delivery of fast magnetic field pulses to excite magnetization switching in specimens. We use several methods to detect the state of magnetization as a function of time. These include the following:
While the aforementioned instruments have immediate use for the characterization of magnetic data-storage materials, they are also powerful tools for the elucidation of magnetodynamic theory. The primary mathematical tools for the analysis of magnetic switching data are essentially phenomenological. As such, they have limited utility in aiding industry in its goal to control the high-speed switching properties of heads and media. We have sought to provide firm theoretical foundations for the analysis of time-resolved data, with special emphasis on those theories that provide clear and unambiguous predictions that can be tested with our instruments.
Deliverables: Magnetodynamic Characterization
Deliverables: Magnetodynamic Instrumentation
To enable future applications in spin electronics, such as ultra-high-frequency oscillators, our goal is to obtain and measure coherent spin dynamics in metal/semiconductor heterostructures. We are investigating both spin-momentum transfer (SMT) and optically generated spin populations in semiconductors.
We are using mechanical point contact spectroscopy to investigate current-induced excitations in multilayer films. It is known that, for sufficiently high current densities and applied magnetic fields, there is an abrupt increase in the resistance of a point-contact junction. The resistance step is attributed to the generation of magnons (spin waves) by the SMT effect. We find that SMT is a generic effect, occurring for a wide range of experimental conditions: for both in-plane and out-of-plane fields, for multilayers grown at the both the first and second maxima in giant magnetoresistance (GMR), and for ferromagnetically coupled multilayers. We found that SMT occurs in a number of different and previously unexplored alloys of Co, Fe and Ni.
In addition to SMT, we are working on a pulsed-laser technique to pump and probe spin populations in semiconductors at cryogenic temperatures. The spin population is measured using the rotation of linear polarized light that is transmitted through a bulk sample. The project's work in spin electronics is funded by the Defense Advanced Research Projects Agency (DARPA).
Deliverables: Spin-Momentum Transfer
Deliverables: Spins in Semiconductors
Magnetodynamics Lecture Series — In January 2002, the division hosted a series of lectures by Robert Stamps, University of Western Australia, on the underpinnings of magnetodynamic theory. The lectures focused on the fundamental principles that lead to the ferromagnetic ground state and methods for analyzing excitations above the ground state. The lectures concluded with a timely discussion on the validity of conventional spin-relaxation theory in a regime of large-angle precessional motion, such as might occur when switching the magnetization orientation with a large field pulse.
Photo and schematic of the CryoPIMM. The photo shows the longitudinal and transverse magnets surrounding the coplanar waveguide structure, which is underneath the brass fixture in the center. Measurement samples are placed between the waveguide and a cold finger beneath it. Data obtained with the CryoPIMM will shed light on the physical mechanism of damping in materials used for data-storage applications.
Cryogenic Capability Added to Pulsed Inductive Microwave Magnetometer — As part of our program in high-speed magnetics, we developed an automated pulsed inductive microwave magnetometer (PIMM) to characterize magnetic thin films. The PIMM is designed to measure the magnetodynamical properties of materials used in recording heads for magnetic data storage. The data-storage industry is developing new magnetic alloys with high saturation magnetization to use in write heads. The magnetic damping behavior of these new alloys will determine their usefulness for high-speed recording.
The PIMM has now been enhanced with variable temperature capability. The new instrument can measure the magnetodynamic response of magnetically soft thin-film materials at temperatures from 25 kelvins to 325 kelvins. In addition, the CryoPIMM has been augmented with high-field magnets that can apply DC bias fields as high as 45 milliteslas, permitting the study of materials with high anisotropy, such as single-crystal films of iron and nickel.
The CryoPIMM will be a powerful new tool to investigate the fundamental origins of precessional damping in thin metallic films. Most magnetic materials with a high permeability also exhibit underdamped response when driven with RF fields. The origin of the oscillatory response stems from the gyromagnetic properties inherent in all ferromagnets. The magnetic moment of the electron is fundamentally coupled to the quantum-mechanical spin angular momentum: When a torque is applied to the magnetization, the intrinsic response of the electron moment is precession, much as a gyroscope precesses under the influence of the Earth's gravitational field. However, in sharp contrast to a mechanical gyroscope, the angular momentum of the electron spin precesses at megahertz to gigahertz frequencies.
In the absence of any coupling between the electron spins and the rest of the crystal environment, the precession would continue indefinitely. In reality, the spins are coupled to the atomic lattice such that the precession is eventually damped. Nevertheless, the resulting oscillations of the magnetic moment can be deleterious in practical applications such as magnetic data storage. For example, the data transfer rate in commercial disk drives is now approaching 1 gigabit per second. Disk-drive engineers must be careful to avoid effects stemming from gyromagnetic precession at these frequencies. Most importantly, there is a need to determine sources of damping, with the goal of controlling the damping as a material design parameter.
There are multiple conflicting theories for damping in metallic thin films. One is "magnon-electron scattering" or "sd-exchange." This theory predicts a strong temperature dependence in the range of 4 kelvins to 100 kelvins. The coupling between conduction electrons and fundamental magnetic excitations ("magnons") is enhanced when the time interval between inelastic scattering events for the conduction electrons is longer than the inverse of the magnon-electron coupling energy. At this point, energy and angular momentum are efficiently transferred from the magnetization to the electrons at the Fermi surface. Observation of a temperature dependence in the damping would be confirming evidence for the sd-exchange theory.
Signal-to-Noise Ratio Improved for Pulsed Inductive Microwave Magnetometer — We discovered a method for enhancing the signal-to-noise ratio for the PIMM by a factor of 4. The method makes use of eddy currents generated in a thick conductive sheet placed over the sample to be measured. The eddy
Differing Dynamic and Static Magnetic Anisotropy in Thin Permalloy Films — We found that the values of dynamic and static uniaxial anisotropy in thin polycrystalline Permalloy films differ by as much as a factor of 1.5. The dynamic anisotropy has an additional isotropic component not observed in static measurements. The time-resolved precessional response was measured as a function of an in-plane applied bias field. The frequency dependence on bias field was fitted with high precision to the Kittel formula for ferromagnetic resonance, thereby extracting anisotropy field. We interpret the constant offset field as a transient component of the magnetic anisotropy that affects dynamical response at time scales only below 10 nanoseconds.
Damping as a Function of Pulse Amplitude and Bias Field in Thin-Film Permalloy — We found that the damping parameter in thin-film Permalloy is independent of transverse pulse field amplitudes but decreases monotonically with increasing longitudinal bias fields. Even though the magnetization is rotated in response to the transverse field by angles well in excess of the ferromagnetic resonance (FMR) spinwave instability threshold, there is no evidence for any nonlinear dependence of damping on pulse amplitude. We surmise that the intrinsic damping in Permalloy is sufficiently large to damp any precessional motion before any spinwave instabilities have a chance to grow to measurable levels.
Stability of Nanoparticles for Magnetic Data Storage — We have found a simple formula that describes the effects of exchange interactions on the thermal stability of nanoparticles in magnetic recording media. Since the production of the first hard-disk drive in the early 1950s — the IBM RAMAC, with an areal density around 150 bits per square centimeter — data storage density has improved immensely. The capacity of computer hard disks has increased by more than seven orders of magnitude in 45 years. At the same time, the price per megabyte has dropped from hundreds of dollars in the 1980s to a few cents in 2002.
The current growth rate in areal density of 100 percent per year raises questions regarding the stability of small magnetic bits. As the bit size approaches the nanometer scale, thermal fluctuations compromise the stability of bits over time: the so-called "superparamagnetic limit." One of the new approaches that have been proposed to improve stability in this regime is the use of exchange-coupled media with bits oriented perpendicular to the substrate.
The magnetic exchange interaction causes individual particles to become correlated. That is, thermal fluctuations sensed by each particle are shared with the others, making the collection of particles more thermally stable. Our formula relates stability parameters, such as the "blocking" temperature, to the magnitude of the exchange interactions, which is measured as a mean magnetic field. This formalism will facilitate the engineering of media for data storage by quantifying the effect of interparticle exchange interactions on thermal stability.
Spin-Momentum Transfer Effects Seen in Multilayer Structures — Initial work on SMT consisted of obtaining Andreev reflection spectra using a superconducting point contact and a magnetic film. The spectroscopic structure of the Andreev reflection measurement is determined by the formation of Cooper pairs when electrons leave a normal metal and enter a superconductor. The net spin of the Cooper pairs must be zero, forcing every electron that enters the superconductor to accompany another electron of opposite spin. However, in a ferromagnet, there is an asymmetric spin distribution that reduces the statistical odds that every "up" spin entering the superconductor will be correlated with an available "down" spin. The successful observation of the Andreev reflection signature was proof-of-concept for the ability to establish ballistic point contacts.
The next step was to observe magnon scattering peaks in point contact spectra of magnetic multilayers. The peaks indicate dynamic excitations generated by an electron current flowing from a silver tip into a magnetic exchange-coupled multilayer. Such peaks have been observed before in antiferromagnetic exchange-coupled multilayers. The peaks are apparent even in zero applied magnetic fields, which had never before been seen.
Spin-Momentum Transfer Efficiency Estimated at 30 Percent — An estimate of the SMT efficiency from a polarized conduction current was obtained from point-contact data for Cu/Co multilayers. The analysis uses the theory of IBM researcher John Slonczewski, who first predicted the SMT effect in 1996. From this theory, the critical current at which the point-contact resistance experiences a sudden jump can be used to determine the SMT efficiency in an experimental geometry with an applied field oriented perpendicular to the magnetic multilayer. These first estimates of SMT efficiency for point-contact data give values from 25 to 35 percent, close to the maximum expected values calculated by Slonczewski for Co-based multilayers.
Spin-Momentum Transfer Ubiquitous in Multilayer Magnetic Films — We were able to measure SMT effects in numerous multilayer structures, including Cu/Co-Fe, Cu/Fe and Cu/Ni-Fe. These results indicate that SMT, which gives rise to resistance steps in point contact measurements, is a general property of magnetic multilayers and is not specific to Cu/Co multilayers. The multilayer films that exhibit SMT features need not even exhibit a measurable GMR signal: no GMR was measured for the Cu/Fe multilayers.
Pulsed Laser Technique Pumps and Probes Spin Populations in Semiconductors — We obtained the first spin-detection data for optically oriented GaAs using a pulsed-laser Faraday detection mechanism. All measurements were conducted at 5 kelvins. Using pump lasers at 780 nanometers and 670 nanometers, we found that a nondegenerate system can be used to pump and probe the spin population injected into moderately doped GaAs. The spin population is measured using the rotation of linear polarized light that is transmitted through a bulk sample. The degree of rotation is directly proportional to the spin polarization. The spin-coherence times were measured by sweeping an applied magnetic field and measuring the dependence of the spin polarization. In what is effectively a zero-frequency measurement of electron spin resonance, the full-width-at-half-maximum is proportional to the spin-dephasing rate. Initial measurements of the spin-dephasing times yielded only a weak dependence on the pump power used to inject the polarized spins.