From Spin Torque Random Access Memory to Spintronic Memristor. Xiaobin Wang Seagate Technology

Similar documents
Wouldn t it be great if

MRAM: Device Basics and Emerging Technologies

Magnetic tunnel junction beyond memory from logic to neuromorphic computing WANJUN PARK DEPT. OF ELECTRONIC ENGINEERING, HANYANG UNIVERSITY

A Technology-Agnostic MTJ SPICE Model with User-Defined Dimensions for STT-MRAM Scalability Studies

A Perpendicular Spin Torque Switching based MRAM for the 28 nm Technology Node

Perpendicular MTJ stack development for STT MRAM on Endura PVD platform

Author : Fabrice BERNARD-GRANGER September 18 th, 2014

Window Function Analysis of Nonlinear Behaviour of Fourth Fundamental Passive Circuit Element: Memristor

Lecture 6 NEW TYPES OF MEMORY

Mon., Feb. 04 & Wed., Feb. 06, A few more instructive slides related to GMR and GMR sensors

SPICE Modeling of STT-RAM for Resilient Design. Zihan Xu, Ketul Sutaria, Chengen Yang, Chaitali Chakrabarti, Yu (Kevin) Cao School of ECEE, ASU

New Approaches to Reducing Energy Consumption of MRAM write cycles, Ultra-high efficient writing (Voltage-Control) Spintronics Memory (VoCSM)

Advanced Lab Course. Tunneling Magneto Resistance

Low Energy Spin Transfer Torque RAM (STT-RAM / SPRAM) Zach Foresta April 23, 2009

Pinched hysteresis loops are a fingerprint of square law capacitors

arxiv: v1 [cond-mat.mtrl-sci] 28 Jul 2008

Nanoelectronics 12. Atsufumi Hirohata Department of Electronics. Quick Review over the Last Lecture

Time resolved transport studies of magnetization reversal in orthogonal spin transfer magnetic tunnel junction devices

MOSFET: Introduction

Influence of exchange bias on magnetic losses in CoFeB/MgO/CoFeB tunnel junctions

Current-driven Magnetization Reversal in a Ferromagnetic Semiconductor. (Ga,Mn)As/GaAs/(Ga,Mn)As Tunnel Junction

Page 1. A portion of this study was supported by NEDO.

Spintronics. Seminar report SUBMITTED TO: SUBMITTED BY:

MTJ-Based Nonvolatile Logic-in-Memory Architecture and Its Application

Spin orbit torque driven magnetic switching and memory. Debanjan Bhowmik

Saroj P. Dash. Chalmers University of Technology. Göteborg, Sweden. Microtechnology and Nanoscience-MC2

MOS Transistor Theory

A Universal Memory Model for Design Exploration. Ketul Sutaria, Chi-Chao Wang, Yu (Kevin) Cao School of ECEE, ASU

MSE 7025 Magnetic Materials (and Spintronics)

ECE-343 Test 1: Feb 10, :00-8:00pm, Closed Book. Name : SOLUTION

Enhanced spin orbit torques by oxygen incorporation in tungsten films

From Hall Effect to TMR

Low Energy SPRAM. Figure 1 Spin valve GMR device hysteresis curve showing states of parallel (P)/anti-parallel (AP) poles,

Lecture 15: MOS Transistor models: Body effects, SPICE models. Context. In the last lecture, we discussed the modes of operation of a MOS FET:

Designing Information Devices and Systems II Fall 2017 Miki Lustig and Michel Maharbiz Homework 1. This homework is due September 5, 2017, at 11:59AM.

Implementing Memristor Based Chaotic Circuits

Complex Dynamics of a Memristor Based Chua s Canonical Circuit

Chapter 4 Field-Effect Transistors

NEUROMORPHIC COMPUTING WITH MAGNETO-METALLIC NEURONS & SYNAPSES: PROSPECTS AND PERSPECTIVES

An Overview of Spin-based Integrated Circuits

Magnetic oscillations driven by the spin Hall effect in 3-terminal magnetic tunnel junction. devices. Cornell University, Ithaca, NY 14853

Mesoscopic Spintronics

Advanced Topics In Solid State Devices EE290B. Will a New Milli-Volt Switch Replace the Transistor for Digital Applications?

3. Basic building blocks. Analog Design for CMOS VLSI Systems Franco Maloberti

Nonvolatile CMOS Circuits Using Magnetic Tunnel Junction

Spin Torque and Magnetic Tunnel Junctions

During such a time interval, the MOS is said to be in "deep depletion" and the only charge present in the semiconductor is the depletion charge.

MOS Transistor I-V Characteristics and Parasitics

Quantum Dot Structures Measuring Hamming Distance for Associative Memories

Magnetic Race- Track Memory: Current Induced Domain Wall Motion!

V High frequency magnetic measurements

GMR Read head. Eric Fullerton ECE, CMRR. Introduction to recording Basic GMR sensor Next generation heads TMR, CPP-GMR UCT) Challenges ATE

Announcements. EE141- Fall 2002 Lecture 7. MOS Capacitances Inverter Delay Power

A short history of memristor development. R. Stanley Williams HP Labs

ECE 438: Digital Integrated Circuits Assignment #4 Solution The Inverter

Homework Assignment 09

Mathematical analysis of a third-order memristor-based Chua oscillators

Single Electron Devices and Circuits

9. Spin Torque Majority Gate

Emerging spintronics-based logic technologies

CMOS Digital Integrated Circuits Lec 13 Semiconductor Memories

Magnetic core memory (1951) cm 2 ( bit)

This document is an author-formatted work. The definitive version for citation appears as:

The Devices: MOS Transistors

Center for Spintronic Materials, Interfaces, and Novel Architectures. Voltage Controlled Antiferromagnetics and Future Spin Memory

Finding the Missing Memristor

ECE520 VLSI Design. Lecture 23: SRAM & DRAM Memories. Payman Zarkesh-Ha

Lecture 4: CMOS Transistor Theory

Gate voltage modulation of spin-hall-torque-driven magnetic switching. Cornell University, Ithaca, NY 14853

Memristive behavior in magnetoelectric devices

MOS Transistor Theory

ECE-343 Test 2: Mar 21, :00-8:00, Closed Book. Name : SOLUTION

EEC 118 Lecture #2: MOSFET Structure and Basic Operation. Rajeevan Amirtharajah University of California, Davis Jeff Parkhurst Intel Corporation

Most matter is electrically neutral; its atoms and molecules have the same number of electrons as protons.

01 Development of Hard Disk Drives

The Physical Structure (NMOS)

EE115C Winter 2017 Digital Electronic Circuits. Lecture 3: MOS RC Model, CMOS Manufacturing

The Devices. Digital Integrated Circuits A Design Perspective. Jan M. Rabaey Anantha Chandrakasan Borivoje Nikolic. July 30, 2002

Fabrication and Measurement of Spin Devices. Purdue Birck Presentation

Electronic Circuits Summary

Solid-State Electronics

ECE 497 JS Lecture - 12 Device Technologies

Principles and Applications of Superconducting Quantum Interference Devices (SQUIDs)

Recent developments in spintronic

Analysis and design of a new SRAM memory cell based on vertical lambda bipolar transistor

Administrative Stuff

Device Models (PN Diode, MOSFET )

Lecture 5: CMOS Transistor Theory

Thermal Magnetic Random Access Memory

The Devices. Jan M. Rabaey

Lecture 6 Power Zhuo Feng. Z. Feng MTU EE4800 CMOS Digital IC Design & Analysis 2010

Magneto-Seebeck effect in spin-valve with in-plane thermal gradient

Mod. Sim. Dyn. Sys. Amplifiers page 1

E40M Review - Part 1

L ECE 4211 UConn F. Jain Scaling Laws for NanoFETs Chapter 10 Logic Gate Scaling

Spin electronics at the nanoscale. Michel Viret Service de Physique de l Etat Condensé CEA Saclay France

Mod. Sim. Dyn. Sys. Amplifiers page 1

Compact Modeling of STT-RAM and MeRAM A Verilog-A model of Magnetic Tunnel Junction Behavioral Dynamics

Digital Electronics Part II - Circuits

EE105 Fall 2014 Microelectronic Devices and Circuits. NMOS Transistor Capacitances: Saturation Region

Transcription:

From Spin Torque Random Access Memory to Spintronic Memristor Xiaobin Wang Seagate Technology

Contents Spin Torque Random Access Memory: dynamics characterization, device scale down challenges and opportunities Spin Torque Memristor: concept, device and application examples X. Wang 2

Spin Torque Random Access Memory (SPRAM) Working Principle Top electrode MTJ Bottom electrode Select Transistor. (1) Magnetic tunneling junction (MTJ) Reference layer Tunneling insulating barrier Free layer Integrate magnetic tunneling junction (MTJ) with CMOS Reading through magneto-resistance principle Writing through spin torque excitation X. Wang 3

SPRAM Switching Behavior 6 I 1 R H 2 R AP MTJ coupled with CMOS 5 4 R L 3-0 + V MTJ static picture Ref: Y. Chen, X. Wang, H. Li, D. Dimitar, H. Liu, International Symposium on Quality Electronic Design, 2008. Resistance (Ohm) MTJ switching asymmetry Spin torque switching asymmetry for a field balanced MTJ stack Reference: X. Wang, W. Zhu, D. Dimitrov, Phys. Rev. B, 79, 104408, 2009. Magnetic field (Oe) X. Wang 4

SPRAM System Dynamical Approach: Quantum-Micromagnetic-SPICE Dynamical approach Macroscopic magnetic (size, Ms, etc) 25 Dynamic Magnetization 20 CMOS I-V curve, equivalent capacitors,etc MTJ electronic spin transport (band structure, tunneling) 15 10 5 5 15 10 Ref: X. Wang, Y. Chen, H. Li, D. Dimitar, H. Liu, IEEE Tran. On Magn., 44(11), 2470, 2008 Dynamic CMOS circuit WL V DD C GD C GS BL SL C DB C SB C GB I I I I I M GD DB M M + I GD = C = I GD DB dv = CDB dt R( t) = V DD R( t) = V + I T dv dt V ( BL SL) ( SL BL) or Reference: X. Wang, W. Zhu, D. Dimitrov, Phys. Rev. B, 79, 104408, 2009. Quantum electronic spin transport X. Wang 5

Challenge for SPRAM to Scale Down Field driving: switching field scales up with 1/volume or 1/surface dimension CMOS driving current (ua) 150 140 130 120 110 100 90 80 22nm x 100nm CMOS 22 nm x 45 nm MTJ, 17.4F 2 22nm x 75nm CMOS 22nm x 45nm MTJ, 13.2F 2 22nm x 45nm CMOS 22nm x 22nm MTJ, 9F 2 70 1500 2000 2500 3000 Spin torque driving: switching current density scales up with 1/volume or 1/surface dimension MTJ Resistance (Ohm) As magnetic device scales down, with CMOS driving stength decreasing, achieving fast nanosecond time scale magnetization switching and maintaining thermal stability at years time scale become a challenge. X. Wang 6

Challenge for SPRAM to Scale Down Variation of single device different switching Spin torque MRAM device to device variations reduce sensing and writing margins As device scales down, increased variability is another challenge. X. Wang 7

Current Reduction Through Magnetization Dynamics critical switching current D, D x y 2 MsVα η : demag factors in - plane, D z ( D z D 2 Stable energy in equilibrium : M V ( D y D s y )( D : demag factor out x ) z D x ) - plane D z >> D x, D y, different scaling behavior of MTJ writing current magnitude and thermal stability energy Geometric wise: thin film for TMR etc. Magnetization dynamics wise: Symmetric magnetization motion: D z =D y Road leads to same scaling of switching current and thermal stability energy X. Wang 8

Current Reduction Through Magnetization Dynamics RF frequency time varying polarization angle Composite material stack Constant polarization direction solution Optimal time varying polarization direction solution Normalized magnetization M z /M s Hard magnetic layer to keep thermal stability 1 0.8 0.6 0.4 0.2 0-0.2-0.4-0.6-0.8-1 Soft magnetic layer to decrease switching field/current magnitude 0 0.2 0.4 0.6 0.8 1 1.2 1.4 x 10-8 Time (sec) X. Wang 9

Current Reduction Through Electronic and Spin Transport Across Interface Increase insulating layer thickness only, RA increases 5 times, critical switching voltage only increases 2.5 times Ref: X. Wang, W. Zhu, Y. Zheng, Z. Gao, H. Xi, INTERMAG 2009 MTJ with spin moment conservation layer between free layer and top cap layer More opaque barrier, less magnetization precessing induced spin pumping, effective damping reduction MR(%) 150 130 110 90 70 50 30 10-10 -2-30 -1.5-1 -0.5 0 0.5 1 1.5 2-50 J(MA/cm^2) Single0.0 Double -0.5 Ref: Y. Zheng, W. Zhu, X. Wang, Z. Gao, D. Wang, D. Dimitrov et. al. submitted to Appli. Phys. Letters 1.0 0.5-1.0 0.000001 0.0001 0.01 1 Dual tunneling barrier, critical switching current down to 0.6MA/cm2, quantum confinement effects? Jc(MA/cm^2) tp(s) Double AP->P Double P->AP X. Wang 10

Variation Controlling at Device Level MTJ Resistance (Ω) 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 40ns pulse DC extrapolation R H -1000-500 0 500 1000 R L I R Sensing Current (µa) destructive Unique Device characters leads to new sensing scheme Self-reference sensing non-destructive Ref. Y. Chen, H. Li, X. Wang, W. Zhu, W. Xu and T. Zhang, Design, Automation & Test in Europe Conference and Exhibition, 2010. X. Wang 11

Variation Controlling at System Level Reduce switching current requirement by intentionally move away from worst case scenario design a smaller-than-worst-case transistor sizing approach. For 256Mb SPRAM design at 45nm node, under a normalized write current threshold deviation of 20%, the overall memory die size can be reduced by more than 20% compared with the conventional worst-case transistor sizing design. Ref: W. Xu, Y. Chen, X. Wang, and T. Zhang, 46th Design Automation Conference, 2009. X. Wang 12

Heat Asssited Solutions Surface anti-ferromagnetic coupled (AFC) magnetic layer with relatively low Curier temperature: At room temperature, the AFC coupling provides surface anisotropy to maintain thermal stability of the square magnetic element. During the writing process, the joule heating of spin torque current raises temperature of AFC surface magnetic layer above Curier temperature and the AFC induced surface anisotropy disappears. The element can be switched at low threshold current. 500 1.00 6 nsec 64Mb 400 10 nsec I (µa) 300 200 100 Required heating current Cell size (um 2 ) 0.10 1Gb 256Mb 65nm, 6nsec 65nm, 10nsec 40nm, 6nsec 0 0 20 40 60 80 100 120 140 D (nm) R A = 12.5 Ωµm 2 0.01 0 20 40 60 80 100 120 140 D (nm) 40nm, 10nsec Write current decreases with decreasing memory element dimension D. Meanwhile, the MTJ resistance increases with decreasing memory element dimension. For a given technology node, an optimal memory cell size can be found due to the tradeoff between critical switching current and MTJ resistance Ref: H. Xi, J. Sreicklin, H. Li, Y. Chen, X. Wang, and Y. Zheng, Z. Gao, M. Tang, IEEE Trans. Magn, 46,no. 3, 860, 2010 X. Wang 13

MTJ Memristor and Mutibit Data Storage and Logic 00 demonstrated multi-bit MTJ devices 10 01 Ref: X. Lou, Z. Gao, D. Dimitrov, and M. Tang, Appli. Phys. Letters, 93, 242502, 2008 11 From nonlinear resistor point of view: reversible branch: 11 to 10, 00 to 01 Irreversible branch: 00 to 10

MTJ Memristor and Mutibit Data Storage and Logic 1400 Switching current (ua) 1200 1000 800 600 400 200 switch 00->01 switch 00->10 more elongated shape and less surface area From memristor point of view: Switching from 00 to 10 and from 00 to 01 are all possible by adjusting pulse width Different slope, switching depends upon integration of current, not current magnitude! 0 0 0.5 1 1.5 2 2.5 3 x 10 8 Inverse of pulse width (sec -1 ) Ref. X. Wang, Y. Chen, Design, Automation & Test in Europe Conference and Exhibition, 2010. X. Wang 15

Memristor as Fourth Circuit Element Resistor Memristor What makes memristor different from an ordinary constant resistor or even a curren or voltage dependent nonlinear resistor: memristance is a function of charge, which depends upon the hysteretic behavior of the current (or voltage) profile. V capacitor ϕ = Vdt Leon O. Chua, Memristor-the missing circuit element, IEEE Trans. Circuit Theory, vol. 18, no. 5, 1971 Chua (1971) proposed memristor for logic completeness of circuit element q q = Idt What makes memristor different from a capacitor or inductor is the ability to accumulate current or voltage information at X. Wang 16 constant voltage or current.

Memristance in Resistive Memory Stack low resistance state high resistance state measured simulated 1) Nano-scale two terminal device as two connecting resistors (doped region and undoped region) 2) Bias electric current drives the front of the doped region 1) J. M. Tour and T. He, Nature 453, 42, (2008). 2) D. B. Strukov; G. S. Snider; D. R. Stewart and R. S. Williams, Nature, 453, 80, (2008) X. Wang 17

Spintronic Memristor Through Spin Torque Induced Magnetization Motion GMR/TMR device: resistance depends upon magnetization state Spin Torque device: current electronic spin changes the magnetization state of the device. The magnetization state of the device depends upon the cumulative effects of electron spin excitations. Spintronic memristor: resistance depends upon the integral effects of its current profile. X. Wang, Y. Chen. H. Xi, H. Li, D. Dimitrov: IEEE electronic device letters vol. 30, no. 3, pp.294, 2009. X. Wang 18

Domain Wall Memristor and Temperature Sensor interactions between macroscopic coherent magnetic structures and random fluctuations Nano-scale feature size, low cost, and mature integration technology with the CMOS process. Targeting the highly integrated on-chip thermal detection applications X. Wang, Y. Chen, Y. Gu, H. Li: IEEE electronic device letters vol. 31, no. 1, pp.20, 2010. X. Wang 19

Memristor for Power Management The ability to accumulate current/voltage through constant current and/or voltage driving strength: power monitor device E = VIdt = V Idt Negative feedback: power control device Ref. X. Wang, Y. Chen, Design, Automation & Test in Europe Conference and Exhibition, 2010. X. Wang 20

Memristor for Data Security Question: Administrator and user. Design a scheme to fight against following action: after reading the stored data information, the user tries to restore the device state to the same state as before reading, so that administrator could not find whether the data has been accessed. Solution depends upon whether the user is given the limited or the same authority on device writing and reading compared to that of the administrator. For the case of user with limited reading and/or writing authority, data information security task can be achieved more easily. For limited writing authority case, user may not be able to restore the device to the state set by the administrator before. For reading limited case, the user may not know the state of the device set by the administrator. A solution giving user and administrator the same authority on reading and writing: 1) Writing process: administrator sets high resistance state (0) fully saturated and low resistance state (1) partially saturated. 2) Reading process: two constant voltage pulses excite device a few times (order of 10). One pulse pushes domain wall toward high resistance end and the other pulse tries to push domain wall toward low resistance end. 3) The device reports two values for reading: 1) final state of the device close to high or low resistance state (High or Low) and 2) whether the device resistance has been significantly changed during reading (Yes or No). Final domain wall settled position 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 Administrator reads the data using pulses with a particular second pulse driving strength. 0 0.8 1 1.2 1.4 1.6 1.8 2 2.2 Second pulse magnitude Ref. X. Wang, Y. Chen, Design, Automation & Test in Europe Conference and Exhibition, 2010. X. Wang 21

Path to Future: Potentials GMR with spin valve structure TMR with tunneling structure Physics process is clear and scaling behavior is well understood: controllable and tunable device Easily integration on top of CMOS New geometry structures New material Y. V. Pershin and M. Di Ventra, Phys. Rev. B, 78, 113309, 2008 Spin blockade, half-metallic material X. Wang 22

Path to Future. nano-scale, ultra-fast. variability, memory effects. analogue/digit mixture fuzzy, neural system deterministic, precise digital control.. X. Wang 23