Project Leader:

Stephen Russek
 

Staff-Years (FY 2001):

Funding:

Magnetic Thin Films and Devices 2002

Goals

This project develops measurements and standards for magnetic thin-film materials and devices for the magnetic-data-storage and magneto-electronics industries. These measurements and standards assist industry in the development of advanced magnetic recording systems, magnetic solid-state memory, magnetic sensors, and magnetic microwave devices. The emphasis is on the performance of nanoscale devices, consisting of multilayer and multicomponent thin-film systems, at microwave frequencies. Project members have successfully devised better methods to measure and control the dynamical properties of magnetic devices operating in the gigahertz regime. They have fabricated magnetic nanostructures to measure new spin-dependent transport phenomena and to determine the resolution of magnetic imaging systems. In addition, the project is developing new combinatorial materials techniques for magnetic thin films and new types of on-wafer magnetic metrology. Long-term goals include the development of metrology that will be required to develop quantum spin-based electronics for data storage and terahertz information processing.

Customer Needs

Our project serves the needs of U.S. industries that use and develop magnetic thin-film and magnetic-device technologies. These industries include magnetic hard-disk recording, magnetic tape recording, magnetic random-access memory (MRAM), and magneto-electronics (including sensors, isolators, and microwave devices). The data storage and magneto-electronics industries are pushing toward smaller and faster technologies that require sub-micrometer magnetic structures to operate in the gigahertz regime.

New techniques are required to measure and characterize these magnetic structures. Advances in technology are dependent on the discovery and characterization of new effects such as giant magnetoresistance and spin-dependent tunneling. A detailed understanding of spin-dependent transport is required to optimize these effects and to discover new phenomena that will lead to new device concepts.

Magnetic thin-film systems have become increasingly complicated, often containing quaternary alloys or multilayer systems with 4 to 10 elements that require atomic-level control of the layers. New techniques are required to efficiently and systematically develop and characterize the magnetic, electronic, and mechanical properties of these advanced thin-film systems. In particular, new metrological systems are required that will be capable of making on-wafer measurements on a large number of sites over a large region of parameter space.

Technical Strategy

We are developing several new techniques to address the needs of U.S. industries that require characterization of magnetic thin films and device structures on nanometer-size scales and gigahertz frequencies.

We have fabricated magnetic nanostructures that can be used to determine the resolution and relative merits of various magnetic-imaging systems. These structures include bits recorded on commercial media, small Co-Pt nanostructures fabricated by electron-beam lithography, and small structures fabricated by focused-ion-beam techniques. The magnetic structures must have stable, well characterized features on length scales down to 10 nanometers to allow the testing of commercial imaging systems.

We have fabricated test structures that allow the characterization of small magnetic devices at frequencies up to 10 gigahertz. The response of sub-micrometer magnetic devices, such as spin-valves, magnetic tunnel junctions, and giant-magnetoresistive devices with current perpendicular to the plane, have been characterized both in the linear-response and the nonlinear switching regimes. The linear-response regime is used for magnetic recording read sensors and high-speed isolators, whereas the switching regime is used for writing or storing data. Measured data have been compared to numerical simulations of the device dynamics to determine the ability of current theory and modeling to predict the behavior of magnetic devices.

We are developing new techniques to measure the high-frequency noise and effects of thermal fluctuations in small magnetic structures. Under-standing the detailed effects of thermal fluctua-tions will be critical in determining the funda-mental limit to the size of magnetic sensors, magnetic data bits, and MRAM elements.

We are developing new techniques to measure the electronic and magnetic properties of magnetic thin-film systems in situ (as they are deposited). One such technique, in-situ magnetoconductance measurements, can determine the effects of surfaces and interfaces on spin-dependent transport in a clear and unambiguous manner. The effects of sub-monolayer additions of oxygen, noble metals, and rare earths on giant magnetoresistance have been studied.

We are developing combinatorial materials techniques to assist industry in the development and characterization of complicated magnetic thin-film systems. Combinatorial materials techniques involve the fabrication of libraries of materials with a systematic variation of materials properties, such as composition and growth temperature. In addition to fabrication of libraries of materials, the combinatorial process involves the development of high-throughput on-wafer metrologies that can systematically characterize the libraries and scan for desirable materials properties.

Finally, we are exploring new physical effects to create the foundation to develop entirely new technologies relying on spin-dependent transport at the quantum level. We are investigating the use of spin-momentum transfer to induce a dynamical response for microwave and high-speed signal processing systems. We are investigating methods of measuring small numbers of spins in semiconductor devices and spin traps. Developing this metrology will be essential to the development of methods to control and manipulate small numbers of spins in a spin circuit.

Deliverables

• In 2002, we will fabricate spin valves, 100 nanometers in size, and measure switching and precession induced by spin-momentum transfer.

• During 2002, we will measure the high-frequency noise in submicrometer giant-magnetoresistive (GMR) devices to assess the fundamental size limits of GMR sensors.

• In 2002, we will design and fabricate a spin trap using GaAs heterostructures and develop measurement techniques to determine spin prop-erties in small spin packets.

• In 2003, we will develop a practical on-wafer system to measure magnetostriction.

• In 2003, we will characterize the super-paramagnetic transition in a single magnetic nanoparticle.

Accomplishments

Micrograph of multi-layer, giant-magneto-resistive (GMR) perpendicular spin-valves, about 100 nanometers in size.

Plot showing the switching of the magnetization due to the injection of a spin-polarized current.

• Rare-Earth Doping Used to Control High-Speed Dynamics of Magnetic Data Storage Components - We have explored the use of rare-earth dopants to control the high-speed dynamics in magnetic thin films used in magnetic recording heads and magnetic random access memory(MRAM).We discovered that a small amount of Tb dopant in Ni-Fe films can dramatically increase the magnetic damping without substantially changing the other magnetic properties. The films can be engineered to be underdamped, critically damped, or overdamped by varying the dopant concentration from 0 to 4 percent. High-speed measurements were made at frequencies up to 6 gigahertz by means of a pulsed inductive technique developed in the Magnetodynamics Project. Rare earths have long been known to increase magnetic damping in ferrite materials used in microwave devices. For microwave applications, damping is undesirable, and efforts have concentrated on eliminating rare-earth impurities. However, for magnetic-data-storage applications, critically damped behavior is desirable to prevent ringing and magnetic turbulence when magnetic elements are rotated or switched. For instance, a typical "spin-valve" read sensor, in response to a 250 picosecond pulsed field from a magnetic bit, will ring for approximately 2 nanoseconds after the applied bit field. Similarly, when an MRAM element is switched, the magnetic energy will cause the element to oscillate or break up into a disordered high-temperature magnetic state. The switching properties of the element will be dramatically altered until the magnetic energy is removed from the system. This can lead to undesirable switching in MRAM arrays if the clock speeds are faster than the magnetic cooling rate. Further temperature-dependent measurements and characterization of films doped with different rare earths indicate that the increased damping is due to local lattice distortions at the rare earth sites due to anisotropic orbitals that are strongly coupled to the film magnetization. The ability to engineer the high-speed dynamical properties of magnetic systems will become critical in the next few years when both magnetic recording and MRAM operation will be pushed into the gigahertz regime.

• Precessional Switching in Magnetic Memory Devices Demonstrated - A particular type of thin-film magnetic device called a "spin-valve" can be engineered to have two stable states of electrical resistance based on the relative magnetization orientation of its ferromagnetic layers. This property has motivated a strong interest in using spin-valves as recording bits in non-volatile magnetic random access memory (MRAM). Companies such as IBM, Motorola, and Honeywell are actively developing MRAM. A primary technical requirement is precise control of the switching of individual devices. We have been studying the dynamics of magnetization reversal in spin-valves. Devices have sub-micrometer dimensions and are fabricated within a test structure that includes high-bandwidth transmission lines. One line delivers ultra-fast magnetic field pulses to the device. The other line is electrically connected to the device and carries the voltage pulse generated as the device changes state. This voltage pulse serves as a probe of the magnetization dynamics of the device.
• 
  In a spin-valve, only one ferromagnetic layer, the "free layer," responds to external fields. Internal magnetic fields within the device allow only two stable magnetization directions, 180 degrees apart, along an easy axis. Current implementation of MRAM requires field pulses applied for 10 to 20 nanoseconds along either the positive or negative easy axis, depending on the desired state. We have discovered a way to switch the devices using field pulses with durations of less than 300 picoseconds directed perpendicular to the easy axis. The magnetization is reversed due to large-angle precessional motion. For pulses of longer duration, the device does not switch because the magnetization rotates back to its initial direction while the pulse is on.
• Precessional switching requires only a single polarity pulse applied perpendicular to the device easy axis, which results in a toggle operation of the magnetic state of the device. This is a simpler and more efficient bit-setting operation than using pulsed fields along the easy axis, which requires longer pulses in both directions.

Plot of the magnetization response of a Ni80Fe20 film showing underdamped behavior and of a 2 percent Tb-doped Ni80Fe20 film showing critical damping.

• Combinatorial Libraries: Phase Diagrams on a Chip - The first NIST magnetic combinatorial thin-film libraries were fabricated and characterized in collaboration with the NIST Materials Science and Engineering Laboratory (MSEL) and Veeco Instruments. We fabricated Ni-Fe-Co-Tb compositional libraries and distributed them to collaborators for measurements. The libraries, fabricated on 7.6 centimeter wafers, contained 400 sites, each 2 millimeters on a side with a different elemental composition. MSEL performed X-ray diffraction measurements to determine microstructure, Veeco characterized the libraries with scanned magneto-optic Kerr effect (MOKE)(SQUID)magnetometry, and we measured the magnetic properties using alternating-gradient-field and superconducting-quantum-interference-device magnetometers. Additional libraries were provided to MSEL to assist in the development of on-wafer magnetostriction measurements. The libraries showed a complex phase diagram with several different microstructural and magnetic regions with dramatically different properties. The library contained regions of in-plane magnetization, out-of-plane magnetization, isotropic magnetic properties, and paramagnetic behavior.

(a) Switching probability of a 0.4 micrometer x 1.1 micrometer spin valve as a function of transverse field pulse width. The decrease in switching probability as the pulse width increases is an indication of precessional switching. (b) The trajectory of the magnetization: point A is the initial position; if the pulse is turned off between points B and C, the bit will switch.

• Effect of Surfaces and Interfaces on Magnetoresistance - Using a recently developed in-situ magnetoconductance technique, we completed an analysis of the effects of surfaces and interfaces on electron scattering in spin-valve devices. The measurements demonstrated the ability to precisely characterize changes in electron transport due to atomic-level changes in surfaces and interfaces. The spin-valves were similar to those being developed for magnetic-recording read heads, and consisted of multilayers of NiO/Ni-Fe/Co/Ru/Co/Cu/Co/X, where the top layer X was varied to included noble metals, transition metals, and oxides. The magnetoconductance was measured during deposition after every 0.25 monolayer of deposition. The variation of the conductance provides information on the scattering and added conductance channels of each added monolayer, while the magnetoconductance provides information on the spin-dependent scattering. It was found that the increase in magnetoconductance, as the thickness of the free layer was increased, could not be explained by simple semiclassical transport models that predict that the saturation length should be equal to the elastic mean free path in the free layer. Measurements of the effects of nano-oxide layers (NOL) further revealed that the increase in magnetoresistance due to the NOL did not, as predicted by simple transport models, scale with free-layer thickness. These models assume the magnetoresistance change can be described semiclasically by a changing surface specularity due to NOL formation. Our work has shown that more complete quantum-mechanical models are required for quantitative description of electron transport. These measurements provide a more accurate and precise characterization of spin-dependent transport in giant magnetoresistance systems than was previously available. This type of characterization is an essential first step in the development models that can quantitatively describe and predict the performance of magnetic devices being developed for magnetic-data-storage applications.