Optical detection of magnetisation dynamics induced by spin-orbit torques

Lead Research Organisation: UNIVERSITY OF EXETER
Department Name: Physics and Astronomy

Abstract

The modern world is completely dependent upon electronic devices that operate through the flow of charged particles called electrons i.e. electric current. However the electron also carries 'spin' angular momentum, and has an associated magnetic moment, like a tiny bar magnet. The aim of Spintronics is to use the spin of an electron to control its motion and how it interacts with magnetic materials. The most celebrated spintronic device is the 'spin-valve', a trilayer structure in which two ferromagnetic (FM) layers are separated by a non-magnetic spacer layer. The spin-valve is engineered so that the magnetic moment of one FM layer is fixed, while that of the other is free to align with an applied magnetic field, like a compass needle. As the relative orientation of the two magnetic moments varies, a large change in electrical resistance of the trilayer is observed. Since the resistance is easily measured, the spin-valve can act as a magnetic field sensor. In fact a spin-valve sensor is used to read back information in every hard disk that is sold today.

When current is passed between the fixed and free FM layers an inverse effect can be observed. The flow of electrons transfers angular momentum from one FM to the other, and, by Newton's 2nd Law, exerts a spin transfer torque (STT). This torque can act upon the magnetic moment of the free layer, causing it to change its orientation. The spin-valve can also be designed to have two stables states, with different electrical resistance, that can be used to store digital information. Arrays of such devices are used in magnetic random access memory (MRAM). Alternatively, in a spin transfer oscillator (STO), the free layer magnetization oscillates at microwave frequency when DC current is applied. Since the resistance also oscillates, microwave voltage oscillations are generated. The STO is unusual in that its frequency can be tuned through multiple octaves by varying the DC current. Multiple STOs can be defined at chip level, as circuit components, or in arrays for increased power output.

In recent years it has been realized and demonstrated that the spin-orbit interaction, a relativistic effect, may also be used to manipulate the electron spin. The spin can in turn be used to generate a STT, which has been termed spin-orbit torque (SOT) in light of its origin. SOTs are generated by the spin Hall effect (SHE) and the Rashba effect, but the separation of these torques from each other, and from the torque generated by the flow of charge (Oersted torque), is still being debated. The optimization of SOT for use in MRAM has attracted enormous interest because it removes the need to pass large electric currents through fragile insulating layers that conduct electricity by quantum mechanical tunneling.

In this project we will use time resolved scanning Kerr microscopy (TRSKM) to explore, understand and optimize SOTs in device structures of the highest quality supplied by HGST, Brown University and the University of Gothenburg, all of whom are leaders in their respective fields. Crucially we will modify our TRSKM so that a magnetic field can be applied with any orientation in 3 dimensional space, while high frequency electrical probes are connected to the device, and a focused optical probe is used to determine the instantaneous orientation of the magnetization vector. This internationally unique instrument will allow us to determine the SOTs from the static and dynamic response of the magnetization, rather than the electrical resistance, as different electrical stimuli are applied. Furthermore the sub-micron spatial resolution of TRSKM will allow us to separate different torques through their spatial variation, and understand how SOTs interact with dynamic magnetic modes in a confined geometry. Finally, we will use this same instrument to understand how SOTs induce magnetic precession in STOs and switching in candidate MRAM devices.

Planned Impact

The Big Data revolution will allow us to record and analyze more of our lives, leading to societal benefits such as preemptive healthcare. Its success depends upon the underlying hardware having the necessary speed, capacity and energy efficiency. The basic science to be studied within this project will lead to improved technology for data storage and wireless communication. Specifically, our work will assist development of magnetic random access memory (MRAM) that is fast, non-volatile, and energy efficient, and spin transfer oscillators (STOs) that are chip-level microwave sources for signal processing and communication.

We will develop improved understanding of the spin-orbit torques (SOTs) that are highly promising for use in driving MRAM and STO devices. We have enlisted project partners capable of both supplying the highest quality samples and exploiting the knowledge that we obtain in the development of new technology. We will work with HGST, who recently acquired Sandisk, on the characterization and use of SOTs in candidate MRAM structures, while Professor Johan Åkerman at the University of Gothenburg will supply STO devices and develop commercial applications with Nanosc AB. Professor Gang Xiao at Brown University is also chief technology officer of Micro Magnetics Inc. He will exploit improved understanding of SOT materials in the design of the vacuum deposition systems and analytical instruments that his company manufactures. In Exeter we will develop a time resolved optical microscope for characterization of SOTs and study of MRAM and STO operation. Our project partners, and others, will benefit from direct detection of the magnetic state, which provides a more reliable determination of the SOTs, and because fewer electrical connections to the device will be needed, simplifying the fabrication of test devices. Optical measurements could also be used for metrology of partially built wafers during manufacturing, checking their properties so that bad wafers may be stopped, with consequent cost-savings.

While there are no immediate plans for MRAM manufacture within the UK, mainstream integrated circuit manufacturers (e.g. Plessey in Plymouth) may wish to embed MRAM into other products as the benefits become clearer. Alternatively, Seagate in Londonderry already possess know-how and capability relevant to MRAM manufacturer and could enter this market in future. A reservoir of university expertise in the underlying science will be important in attracting inward investment, and we will maintain a dialogue with potential players by visiting their premises to give presentations. Similarly, by developing university expertise in STO technology, UK companies working in wireless communications, such as British Telecom after its merger with EE, will be better placed to enter the field as wider opportunities for their business emerge.

High frequency measurement is of growing importance across a wide range of information technologies as bandwidth and bit rates increase. The use of state of the art electrical and optical measurement apparatus during the project will equip the postdoctoral worker and PhD student for a career in academia or high-tech industry. Former group members have readily found employment in industry, national facilities, or academic positions, demonstrating that their skills are valued and in demand. The project and its staff will also provide high frequency expertise to the Exeter Centre for Doctoral Training (CDT) in Electromagnetic Metamaterials.

Finally, it is important that members of the general public understand the broad principles by which device technology works so that they engage with it and experience its benefit within the economy and at home. We will therefore use our involvement with the science and technology to excite the interest of a general audience, encourage young people to want work in this field, and stimulate thinking about how new technology may be used.
 
Description 1. A projected field electromagnet was procured and a method developed for setting a 3 dimensional vector field. In essence it is necessary to perform test runs of a sweep with a particular field history and create a look-up table for future use. However it was found that the water cooling for the high current coils also produced significant vibration such that the system needs to be used advisedly for measurements where high spatial resolution is required. Minimum vibration and thermal drift can be achieved through accurate positioning of a permanent bias magnet using a stepper motor.

2. Spatially and time resolved scanning Kerr microscopy measurements were made on high quality samples produced by our project partners at Brown University, University of Gothenburg, and Western Digital. The in-plane (anti-damping) torque was quantified and found to dominate in the samples tested, while the out-of-plane (field-like) torque was generally difficult to separate from the Oersted field generated by the injected current.

3. Studies of small amplitude precession in spin Hall nano-oscillators revealed a surprising spatial variation of the spin-orbit torque. This is understood in terms of spreading of the injected microwave current, presumably due to the reactance associated with the sample geometry, that is not observed for an injected DC current. DC current injection led to the first reported images of the so-called bullet mode. However, the size of the injection-locked bullet mode appeared to increase with the size of the DC current, suggesting that spreading of the infected RF current destabilises its position, leading to a reduced locking range.
Exploitation Route Understanding of the spatial variation of spin-orbit toque will be useful in future device design.
Sectors Digital/Communication/Information Technologies (including Software)