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

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

Author : Fabrice BERNARD-GRANGER September 18 th, 2014

Lecture 6 NEW TYPES OF MEMORY

Flash Memory Cell Compact Modeling Using PSP Model

Digital Integrated Circuits A Design Perspective

Advanced Flash and Nano-Floating Gate Memories

Perpendicular MTJ stack development for STT MRAM on Endura PVD platform

Moores Law for DRAM. 2x increase in capacity every 18 months 2006: 4GB

Sequential Logic. Handouts: Lecture Slides Spring /27/01. L06 Sequential Logic 1

12. Memories / Bipolar transistors

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

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

Digital Integrated Circuits A Design Perspective. Semiconductor. Memories. Memories

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

Memory, Latches, & Registers

Mechanism of Switching and Related Challenges in Transition Metal Oxide Based RRAM Devices

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

Semiconductor Memories

ECE 546 Lecture 10 MOS Transistors

CMOS Inverter. Performance Scaling

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

Magnetic core memory (1951) cm 2 ( bit)

MM74C150 MM82C19 16-Line to 1-Line Multiplexer 3-STATE 16-Line to 1-Line Multiplexer

Non Volatile Memories Compact Models for Variability Evaluation

Spintronics. Seminar report SUBMITTED TO: SUBMITTED BY:

Floating Gate Devices: Operation and Compact Modeling

CMOS Digital Integrated Circuits Lec 13 Semiconductor Memories

Single Event Effects: SRAM

SEMICONDUCTOR MEMORIES

GMU, ECE 680 Physical VLSI Design 1

Lecture 25. Semiconductor Memories. Issues in Memory

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

Hw 6 and 7 Graded and available Project Phase 2 Graded Project Phase 3 Launch Today

Supplementary Information for. Non-volatile memory based on ferroelectric photovoltaic effect

Nomenclature, Advantages, Applications. Logic States, Read Ops, Write Ops

EE141- Fall 2002 Lecture 27. Memory EE141. Announcements. We finished all the labs No homework this week Projects are due next Tuesday 9am EE141

Topics. Dynamic CMOS Sequential Design Memory and Control. John A. Chandy Dept. of Electrical and Computer Engineering University of Connecticut

An Overview of Spin-based Integrated Circuits

Thin Film Transistors (TFT)

A Review of Spintronics based Data Storage. M.Tech Student Professor

A 68 Parallel Row Access Neuromorphic Core with 22K Multi-Level Synapses Based on Logic- Compatible Embedded Flash Memory Technology

NONVOLATILE SPINTRONICS: PERSPECTIVES ON INSTANT-ON NONVOLATILE NANOELECTRONIC SYSTEMS

NEM Relay Design for Compact, Ultra-Low-Power Digital Logic Circuits

Schmitt-Trigger Inverter/ CMOS Logic Level Shifter

Access from the University of Nottingham repository:

Multiple Gate CMOS and Beyond

A final review session will be offered on Thursday, May 10 from 10AM to 12noon in 521 Cory (the Hogan Room).

ECE 342 Electronic Circuits. 3. MOS Transistors

Semiconductor Memory Classification

KINGS COLLEGE OF ENGINEERING DEPARTMENT OF ELECTRONICS AND COMMUNICATION ENGINEERING QUESTION BANK

NTE74HC173 Integrated Circuit TTL High Speed CMOS, 4 Bit D Type Flip Flop with 3 State Outputs

VLSI Design Verification and Test Simulation CMPE 646. Specification. Design(netlist) True-value Simulator

University of Toronto. Final Exam

CHAPTER I. Introduction. 1.1 State of the art for non-volatile memory

Evaluation of Non-Quasi-Static Effects during SEU in Deep-Submicron MOS Devices and Circuits

The Pennsylvania State University. The Graduate School. College of Engineering NON-VOLATILE FERROELECTRIC TRANSISTOR BASED MEMORY DESIGN:

CSE140L: Components and Design Techniques for Digital Systems Lab. FSMs. Instructor: Mohsen Imani. Slides from Tajana Simunic Rosing

N ano scale l S il ii lco i n B ased N o nvo lat l i atl ie l M em ory r Chungwoo Kim, Ph.D.

Electronic Devices and Circuits Lecture 15 - Digital Circuits: Inverter Basics - Outline Announcements. = total current; I D

74HC86. Quad 2 Input Exclusive OR Gate. High Performance Silicon Gate CMOS

MM74C906 Hex Open Drain N-Channel Buffers

Wouldn t it be great if

ESE570 Spring University of Pennsylvania Department of Electrical and System Engineering Digital Integrated Cicruits AND VLSI Fundamentals

Digital Electronics Part II - Circuits

Memory Trend. Memory Architectures The Memory Core Periphery

EE241 - Spring 2000 Advanced Digital Integrated Circuits. References

Section 4. Nonlinear Circuits


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

Chapter 3 Basics Semiconductor Devices and Processing

5. Sequential Logic x Computation Structures Part 1 Digital Circuits. Copyright 2015 MIT EECS

Lecture 16: Circuit Pitfalls

Administrative Stuff

Overview ECE 553: TESTING AND TESTABLE DESIGN OF. Memory Density. Test Time in Seconds (Memory Size n Bits) 10/28/2014

Solid-State Electronics

2 Input NAND Gate L74VHC1G00

ESE570 Spring University of Pennsylvania Department of Electrical and System Engineering Digital Integrated Cicruits AND VLSI Fundamentals

EE155/255 Green Electronics

LH5P8128. CMOS 1M (128K 8) Pseudo-Static RAM PIN CONNECTIONS

ESE 570: Digital Integrated Circuits and VLSI Fundamentals

74VHCT138ATTR 3 TO 8 LINE DECODER (INVERTING)

P-channel enhancement mode MOS transistor

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

DS0026 Dual High-Speed MOS Driver

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

MM74C90 MM74C93 4-Bit Decade Counter 4-Bit Binary Counter

SRAM & FLASH Mixed Module

Reliability Testing. Process-Related Reliability Tests. Quality and Reliability Report. Non-Volatile Memory Cycling Endurance

INCREASING power density and static leakage currents

ECE 342 Electronic Circuits. Lecture 6 MOS Transistors

MRAM: Device Basics and Emerging Technologies

! Charge Leakage/Charge Sharing. " Domino Logic Design Considerations. ! Logic Comparisons. ! Memory. " Classification. " ROM Memories.

ESE570 Spring University of Pennsylvania Department of Electrical and System Engineering Digital Integrated Cicruits AND VLSI Fundamentals

5 V 64K X 16 CMOS SRAM

INTEGRATED CIRCUITS. For a complete data sheet, please also download:

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

Lecture 11: MOSFET Modeling

MM74C00 MM74C02 MM74C04 Quad 2-Input NAND Gate Quad 2-Input NOR Gate Hex Inverter

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

Embedded MRAM Technology For logic VLSI Application

Transcription:

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

Universal Memory Modeling because there is no universal memory device! Modeling needs in light of scaling A top-down view: Finite-State Machine (FSM) Type I and Type II memories Modeling examples: Flash, PCM, and MRAM Summary -2-

Scaling Trend: Logic Device Bulk/SOI MOSFET Strained MOSFET HKMG MOSFET MG MOSFET Dominated by silicon based CMOS Dedicated compact model, such as BSIM and PSP, to describe the physics underneath Bottom-up modules: electro-static and transport Future abstraction, using I on and I off, for fast design evaluation Common behavior to cover post-si FET devices -3-

Scaling Trend: Memory Device SRAM Flash PCM STT DRAM FRAM RRAM Tremendous variety in memory physics, materials, structures, and devices! SRAM and DRAM used to be the mainstream 3D technology, such as TSV, offers future opportunity of heterogeneous integration -4-

Design Perspective Speed Power Density, Date-Retention, Speed (R/W), Power, Voltage, Compatibility, Reliability Diverse design needs drive the development and selection of different types of memory -5-

Headline News Much more technological choices for memory design Still, there is no universal solution -6-

Modeling of Memory Devices From discrete, bottom-up approaches to a universal model Essential to efficiently bridge technology evaluation and design choice -7-

A Top-Down View General Behavior: A Finite-State Machine (FSM) 1 0 0 1 1 0 The diagram of finite states is the formal and general method to describe a sequential unit The state may be defined as the output data (0 and 1), which can be a electrical signal or other physical values The transition/retention mechanisms distinguish one type of memory from another type -8-

Equivalent Circuit Diagram Input Control Output Control Static, Monotonic A common modeling structure for SPICE simulation Cell: performs the state transition and data retention Input control: translates the input signals for Write Output control: translates the output signals for Read Physical modules to cover various types of memories -9-

Two Types of Memory Type I Type II Two types based on the forces of cell operation Forces for 1 0 or 0 1 can be different Type I: external force only to stabilize/distinguish states Type II: both internal and external forces to control the states self-feedback mechanisms for meta-stability -10-

Modeling Examples DRAM SRAM DRAM: Type I Read and Write through the same resistor, but different capacitances SRAM: Type II Feedback between the cross-coupled inverters for data retention -11-

More Classifications Category Program (t) Erase (t) States Retention DRAM Charging Discharging Q in Cap (VM) Refresh Q Type I Type II Flash Charging (Hot carrier/ Tunneling) Discharging (Hot carrier/ Tunneling) Q in/out FG (NVM) PCM High voltage/current Reverse reaction Resistance (Amorphous/crystal states) (NVM) RRAM Conducting path (low R) Removal of the path Resistance change (NVM) SRAM Charging Discharging Q + positive feedback (VM) MRAM- MTJ FeRAM Spin alignment (currentinduced magnetic field) E-Field to change the polarization Reverse magnetic field Reverse E-Field * Magneto resistance (NVM) Polarizations (NVM) Device structure/material material material positive feedback material material Different operation modes may involve different physical forces -12-

Flash (Type I) V CG V FG C CG C FG V CG V FG C FG I G =f(v FG -V CH ) V FG is changed by carrier injection and regarded as the state value The program/erase is through the tunneling current, I G When V CG is applied, I G is activated Voltage difference between V FG and V CH determines I G -13-

SPICE Simulation vs. TCAD 4.0 DC simulation 1.05 1.04 Transient V FG (V) 3.0 2.0 V FG (V) 1.03 1.02 1.01 1 1.0 0.0 Symbol: TCAD Line: Model 3 5 6 7 8 9 10 11 12 13 14 15 16 0.99 0.98 0.97 Symbol: TCAD Line: Model V CG (V) Time (s) The model is implemented through sub-circuit and/or Verilog-A SPICE simulation converges well and matches TCAD data, with appropriate tuning of model parameters Read and Write in Flash may have different speeds, depending on the specific tunneling mechanism -14-

PCM (Type I) Top electrode R thermal Chalcogenide Resistor (heater) C thermal Bottom electrode R thermal Chalcogenide exhibits reversible transition between the amorphous and crystalline state, controlled by the temperature profile -15-

Phase Change Behavior 1.2 Resistance (a.u.) 1 0.8 0.6 0.4 0.2 0 Symbol: measurement (A. L. Lacaita, et al., IEDM 04) Line: simulation -0.2 0.0 21.5 45.8 61.4 85.7 92.8 138.4 202.0 259.2 299.1 338.9 375.2 413.3 441.1 497.6 542.0 569.3 593.4 612.2 641.6 665.7 684.7 708.9 731.4 759.1 778.2 795.5 Programming current (µa) The success of phase change (state transition) depends on the magnitude and the width of the input current pulse If not successful, the original phase (state) remains -16-

STT MRAM (Type II) θ dm dt dm 0Ms H M K1 uea M M dt s 2 Zeeman (external) Demagnetization (internal) Anisotropic (internal) Write in STT MRAM requires the magnetization change of one ferromagnetic layer, involves three energies d M dt d H cos M dt 0 s K s θ V sin cos -17-

Dynamic Simulation Magnetization Angle (degree) 180 150 120 90 60 30 0 I < I c I (top left to bottom right) 1000 500 100 50 15 10 Pulse width: 10ns 0.0 1.0x10-7 2.0x10-7 Time (s) The dynamic change is well captured by the model The balance between external/internal (feedthrough/feedback) forces are calculated by SPICE, reducing computation risk and cost -18-

Further Abstraction Out DRAM SRAM STT The transfer characteristic is essential to determine noise margin (stability) and drive strength (speed) Finite points to extract key static/dynamic properties Statistical variability and reliability embedded into finite points In -19-

Summary 1 0 0 1 1 0 A universal SPICE model for various memory devices The generic interface between circuit and device research Easy customization to different physical mechanisms Fully compatible with the SPICE engine -20-