Magnetization dynamics lasts till the end of the pulse.

1. Precessional Reversal in Orthogonal Spin Transfer Magnetic Random Access Memory Devices H. Liu,1 D. Bedau,1 D. Backes,1 J. A. Katine,2 and A. D. Kent1,3 1. Department of Physics, New York University, New York, New York 10003, USA 2. HGST Research, San Jose, California 95135, USA 3. Spin Transfer Technologies, 33 Arch Street, Boston, Massachusetts 02110, USA Introduction Motivation Experimental Setup Time-Resolved Measurements Spin – Transfer Switching: Reversal of the direction of magnetization with a spin-transfer torque (STT) • Study spin-transfer driven magnetization dynamics in orthogonal spin-transfer (OST) devices, devices with a spin-polarizing layer magnetized perpendicularly to a free layer. • Control and observe magnetization dynamics on nanosecond time scales. Single-Shot Switching Traces Spin-transfer torque (STT): g(θ) : angular strength of STT γ : gyromagnetic ratio  : reduced Planck constant P : spin polarization of the current I : amplitude of the current e : charge of an electron. the transverse component of the angular moment transferred from a spin-polarized current to a magnetization in unit time • In Collinear Devices • No initial torque, if no thermal fluctuation. • Waiting for large thermal fluctuation. • Incubation delay (~ ns). • Unpredictable switching process. • In Orthogonal Devices • Large initial torque from perpendicular polarizer. • Fast switching through precession about Bdemag. • Deterministic switching as described in Ref [1]. • Low power consumption. An arbitrary waveform generator (AWG) is used to generate pulses. A real-time oscilloscope (DPO) is connected in series with the device to measure the current through the device during the pulses. DC resistance is measured before and after each pulse to determine the state of the device. Devices were saturated by either applied magnetic field or current pulses. Both statistical and time-resolved measurements were conducted. Single-shot traces from the oscilloscope indicating the magnetization dynamics during a 0.78 V, 2 ns pulse from the AWG (a) for an AP to P switching event, (b) a precessional switching event from P to AP to P then to AP, and (c) an oscillation from P to AP then back to P state. The switching start time (τstart) and the switching time (τswitch) can be obtained from these single-shot traces, as shown by the vertical dashed lines in (a). Switching Behavior Observed in OST-Devices: Statistical Measurements Single-Shot Switching Statistics • Bipolar switching. • Unipolar switching. • Precessional switching. Magnetic Random Access Memory (MRAM) Fast Switching Statistics is obtained through more than 10,000 AP to P and P to AP switching events. Most AP to P transitions are fast, less than 200 ps. Most P to AP transitions are precessional. Therefore, switching mechanisms for AP to P switching and P to AP switching are different for the same pulse configuration. This can be explained by the strength of the perpendicular spin torques acting on the free layer: Larger current in the P state than the AP state, therefore greater perpendicular spin torques for device starting at the P state. Typical AP→P switching: direct and fast without precession • Use magnetization orientation of a nanomagnet to store information. • Non-volatile memory. • Low energy cost during operation. • High density • Use GMR / TMR to read. • Use spin-transfer effect to write. • Fast read / write time • Fast switching • 100 % (1000 / 1000) under 500 ps • No nanosecond incubation delay • Low energy cost • -0.6 V, 500 ps • 400 Ω < R < 800 Ω • 225 fJ < E < 450 fJ 60 x 180 nm2, Hexagon MR = 107%, Rp = 400 Ω Hc = 14 mT has potential of becoming a “universal memory” Typical P→AP switching: Precessional switching No-switching zone No-switching zone Switching Probabilities Correlation Between Start Time and Switching Time Sample Characterization 50 x 100 nm2, Rect. MR = 112%, Rp = 2.2 kΩ Hc = 8 mT • Characterized: tens of thousands of samples. • Size: 40 nm x 80 nm ~ 105 nm x 240 nm. • Shape: rectangles, ellipses and hexagons. • TMR 100 %, RA 5 ~ 10 Ωμm2. PAP PAP • Layer stack. • Perpendicular polarizer: a Co/Pd multilayer exchange coupled to a Co/Ni multilayer. • MTJ: 3 CoFeB|0.8 MgO|2.3 Co0.4Fe0.4B0.2|0.6 Ru|2 Co0.7Fe0.3|16 PtMn (layer thicknesses in nanometers). • (b) Vibrating sample magnetometry (VSM) measurements of the magnetization of the layer stack. • (c) Device resistance versus in-plane field. • MR = 107%. • Coercive field: 7 mT. • Loop shift: -2 mT. Magnetization dynamics lasts till the end of the pulse. Thermal fluctuation. 2 ns < τstart+ τswitch < 2.4 ns Most switching events have 1.5 oscillations Bi-polar switching, i.e. both pulse polarities can switch the device for either P->AP or AP->P switching. Switching probability is non-monotonic with pulse duration 1 ns < τstart < 2.4 ns τswitch < 200 ps, less than half oscillation. Direct transition Summary Fabricated OST-MRAM devices that incorporate magnetic tunnel junctions. Achieved high TMR (>100%) with low RA~ 5 W mm2 and perpendicular polarizer. Demonstrated 100% switching probability in thermally stable elements with 500 ps duration pulses (0.7 V), requiring just 450 fJ. Time-resolvedindividual switching events showing both direct and precessional switching. Related different switching mechanisms to the strength of the perpendicular spin torques. References • A. D. Kent, B. Ozyilmaz, and E. del BarcoAppl. Phys. Lett. 84, 3897 (2004) • H. Liu, D. Bedau, D. Backes, J. A. Katine, J. Langer, and A. D. Kent Appl. Phys. Lett.97, 242510 (2010). • H. Liu, D. Bedau, D. Backes, J. A. Katine, and A. D. Kent Appl. Phys. Lett.101, 032403 (2012) http://www.physics.nyu.edu/kentlab Presenting author’s email: hl757@nyu.edu This work is supported by Spin Transfer Technologies