Chapter 10 Memories Boonchuay Supmonchai Integrated Design Application Research (IDAR) Laboratory August 7, 2005.

Slides:



Advertisements
Similar presentations
Semiconductor Memory Design. Organization of Memory Systems Driven only from outside Data flow in and out A cell is accessed for reading by selecting.
Advertisements

+ CS 325: CS Hardware and Software Organization and Architecture Internal Memory.
Sistemi Elettronici Programmabili1 Progettazione di circuiti e sistemi VLSI Anno Accademico Lezione Memorie (vedi anche i file pcs1_memorie.pdf.
COEN 180 DRAM. Dynamic Random Access Memory Dynamic: Periodically refresh information in a bit cell. Else it is lost. Small footprint: transistor + capacitor.
Digital Integrated Circuits A Design Perspective
Budapest University of Technology and Economics Department of Electron Devices Microelectronics, BSc course MOS circuits: basic construction.
Elettronica T AA Digital Integrated Circuits © Prentice Hall 2003 SRAM & DRAM.
Digital Integrated Circuits A Design Perspective
Introduction to CMOS VLSI Design Lecture 13: SRAM
Prof. John Nestor ECE Department Lafayette College Easton, Pennsylvania ECE VLSI Circuit Design Lecture 25 - Subsystem.
11/29/2004EE 42 fall 2004 lecture 371 Lecture #37: Memory Last lecture: –Transmission line equations –Reflections and termination –High frequency measurements.
Digital Integrated Circuits© Prentice Hall 1995 Memory SEMICONDUCTOR MEMORIES.
S. Reda EN160 SP’07 Design and Implementation of VLSI Systems (EN0160) Lecture 32: Array Subsystems (DRAM/ROM) Prof. Sherief Reda Division of Engineering,
Introduction to CMOS VLSI Design SRAM/DRAM
Digital Integrated Circuits© Prentice Hall 1995 Memory SEMICONDUCTOR MEMORIES.
Registers  Flip-flops are available in a variety of configurations. A simple one with two independent D flip-flops with clear and preset signals is illustrated.
Static Memory Outline –Types of Static Memory –Static RAM –Battery Backup –EPROM –Flash Memory –EEPROM Goal –Understand types of static memory –Understand.
Lecture 19: SRAM.
Parts from Lecture 9: SRAM Parts from
Digital Integrated Circuits A Design Perspective
Digital Integrated Circuits for Communication
55:035 Computer Architecture and Organization
Semiconductor Memories Lecture 1: May 10, 2006 EE Summer Camp Abhinav Agarwal.
12/1/2004EE 42 fall 2004 lecture 381 Lecture #38: Memory (2) Last lecture: –Memory Architecture –Static Ram This lecture –Dynamic Ram –E 2 memory.
COMP3221: Microprocessors and Embedded Systems
Review: Basic Building Blocks  Datapath l Execution units -Adder, multiplier, divider, shifter, etc. l Register file and pipeline registers l Multiplexers,
Semiconductor Memories.  Semiconductor memory is an electronic data storage device, often used as computer memory, implemented on a semiconductor-based.
© Digital Integrated Circuits 2nd Memories Digital Integrated Circuits A Design Perspective SemiconductorMemories Jan M. Rabaey Anantha Chandrakasan Borivoje.
Semiconductor Memories Mohammad Sharifkhani. Outline Introduction Non-volatile memories.
Digital Integrated Circuits© Prentice Hall 1995 Memory SEMICONDUCTOR MEMORIES Adapted from Jan Rabaey's IC Design. Copyright 1996 UCB.
© Digital Integrated Circuits 2nd Memories Digital Integrated Circuits A Design Perspective SemiconductorMemories Jan M. Rabaey Anantha Chandrakasan Borivoje.
FPGA-Based System Design: Chapter 3 Copyright  2004 Prentice Hall PTR Topics n Latches and flip-flops. n RAMs and ROMs.
Modern VLSI Design 4e: Chapter 6 Copyright  2008 Wayne Wolf Topics Memories: –ROM; –SRAM; –DRAM; –Flash. Image sensors. FPGAs. PLAs.
Memory and Storage Dr. Rebhi S. Baraka
Z. Feng MTU EE4800 CMOS Digital IC Design & Analysis 12.1 EE4800 CMOS Digital IC Design & Analysis Lecture 12 SRAM Zhuo Feng.
Memory Semiconductor Memory Classification ETEG 431 SG Size: Bits, Bytes, Words. Timing Parameter: Read, Write Cycle… Function: ROM, RWM, Volatile, Static,
Digital Design: Principles and Practices
Digital Logic Design Instructor: Kasım Sinan YILDIRIM
Advanced VLSI Design Unit 06: SRAM
CSE477 L24 RAM Cores.1Irwin&Vijay, PSU, 2002 CSE477 VLSI Digital Circuits Fall 2002 Lecture 24: RAM Cores Mary Jane Irwin ( )
ECE 300 Advanced VLSI Design Fall 2006 Lecture 19: Memories
CSE477 L23 Memories.1Irwin&Vijay, PSU, 2002 CSE477 VLSI Digital Circuits Fall 2002 Lecture 23: Semiconductor Memories Mary Jane Irwin (
Reading Assignment: Chapter 10 of Rabaey Chapter 8.3 of Weste
Washington State University
Computer Memory Storage Decoding Addressing 1. Memories We've Seen SIMM = Single Inline Memory Module DIMM = Dual IMM SODIMM = Small Outline DIMM RAM.
© Digital Integrated Circuits 2nd Memories Digital Integrated Circuits A Design Perspective SemiconductorMemories Jan M. Rabaey Anantha Chandrakasan Borivoje.
Washington State University
CMPEN 411 VLSI Digital Circuits Spring 2009 Lecture 22: Memery, ROM
Budapest University of Technology and Economics Department of Electron Devices Microelectronics, BSc course MOS circuits: basic construction.
1 Semiconductor Memories. 2 Semiconductor Memory Classification Read-Write Memory Non-Volatile Read-Write Memory Read-Only Memory EPROM E 2 PROM FLASH.
CMPEN 411 VLSI Digital Circuits Spring 2009 Lecture 24: Peripheral Memory Circuits [Adapted from Rabaey’s Digital Integrated Circuits, Second Edition,
Introduction to Computer Organization and Architecture Lecture 7 By Juthawut Chantharamalee wut_cha/home.htm.
EE 466/586 VLSI Design Partha Pande School of EECS Washington State University
Sp09 CMPEN 411 L21 S.1 CMPEN 411 VLSI Digital Circuits Spring 2009 Lecture 21: Shifters, Decoders, Muxes [Adapted from Rabaey’s Digital Integrated Circuits,
EE 534 summer 2004 University of South Alabama EE534 VLSI Design System summer 2004 Lecture 14:Chapter 10 Semiconductors memories.
Memory (Contd..) Memory Timing: Definitions ETEG 431 SG.
Computer Architecture Chapter (5): Internal Memory
EE586 VLSI Design Partha Pande School of EECS Washington State University
Norhayati Soin 06 KEEE 4426 WEEK 15/1 6/04/2006 CHAPTER 6 Semiconductor Memories.
Chapter 5 Internal Memory
William Stallings Computer Organization and Architecture 8th Edition
Digital Integrated Circuits A Design Perspective
William Stallings Computer Organization and Architecture 8th Edition
Memory.
Semiconductor Memories
William Stallings Computer Organization and Architecture 8th Edition
Semiconductor memories are classified in different ways. A distinction is made between read-only (ROM) and read-write (RWM) memories. The contents RWMs.
Presentation transcript:

Chapter 10 Memories Boonchuay Supmonchai Integrated Design Application Research (IDAR) Laboratory August 7, 2005

B.Supmonchai Digital ICs Memories 2 Outlines  Memory Classification  ROM, SRAM, DRAM, CAM  Memory Architecture  Memory Core Circuits  Periphery Circuits  Reliability  Case Studies

B.Supmonchai Digital ICs Memories 3 Review: Basic Building Blocks  Datapath  Execution units  Adder, multiplier, divider, shifter, etc.  Register file and pipeline registers  Multiplexers, decoders  Control  Finite state machines (PLA, ROM, random logic)  Interconnect  Switches, arbiters, buses  Memory  Caches (SRAMs), TLBs, DRAMs, buffers

B.Supmonchai Digital ICs Memories 4 Main Memory (DRAM) eDRAM Main Board Second Level Cache (SRAM) CPU Board Secondary Memory (Disk) Control Datapath On-Chip Components RegFile DataCache Instr Cache ITLB DTLB Speed (ns) Speed (ns).1’s 1’s 10’s 100’s 1,000’s Size (bytes) Size (bytes) 100’s K’s 10K’s M’s T’s A Typical Memory Hierarchy principle of locality  By taking advantage of the principle of locality:  Present the user with as much memory as is available in the cheapest technology.  Provide access at the speed offered by the fastest technology. highest lowest highest lowestCost

B.Supmonchai Digital ICs Memories 5 Semiconductor Memories Read-Write Memory Non-Volatile Read-Write Memory Read-Only Memory Random Access Non-Random Access SRAM (cache, register file) DRAM FIFO LIFO Shift Register CAM EPROM E 2 PROM FLASH Mask- programmed Electrically- programmed (PROM) 50% of the transistors in a microprocessor are in the memory (reg file, caches, etc.)

B.Supmonchai Digital ICs Memories 6 Memory Timing Definitions Read Read Cycle Read Access Write Cycle WriteSetup Data Valid Write Hold Read Access Data Written Write Data

B.Supmonchai Digital ICs Memories 7 1D-Memory Architecture: Decoders Too many select signals: N words  N select signals Decoder reduces #s of select signals N words  K (= log 2 N) select signals Word 0 Word 1 Word 2 WordN - 2 WordN - 1 Storage cell M bits S0S0 Input-Output (M bits) S1S1 S2S2 S N-2 S N-1 N words M bits A 0 A 1 A K-1 K = log 2 N Word 0 Word 1 Word 2 WordN - 2 WordN - 1 Storage cell Input-Output (M bits) S0S0 Decoder

B.Supmonchai Digital ICs Memories 8 2D-Memory Architecture: Array-Based 1D-architecture suffers from the ASPECT RATIO problem (HEIGHT >> WIDTH) Amplify swing to rail-to-rail amplitude rail-to-rail amplitude Amplify swing to rail-to-rail amplitude rail-to-rail amplitude Select appropriate word Still too slow for bigger memories (> 256K) because the word and bit lines are too long.

B.Supmonchai Digital ICs Memories 9 3D-Memory Architecture: Hierarchical Advantages: 1. Shorter word and/or bit lines 2. Block address activates only 1 block saving power 2. Block address activates only 1 block saving power -

B.Supmonchai Digital ICs Memories 10 Contents-Addressable Memory (CAM) Data (64 bits) Comparand CAM Array Mask Control Logic R/W Address (9 bits) I/O Buffers Priority Encoder 2 9 Validity Bits Address Decoder 2 9 word x 64 bits Commands

B.Supmonchai Digital ICs Memories 11 CAM Operations ReadWriteMatch  3 Modes of CAM Operation: Read, Write, and Match Match  In Match mode, the data are compared to the content in the memories to locate where the data are kept, i.e. its address.  The Comparand keeps the data pattern to match and the Mask indicates which bits are significant.  All 512 rows of the CAM array then simultaneously compare the significant bits of the comparand with the data contained in that row.  Validity bits are set for the rows that contain matched pattern.  If there are more than one matched row, the Row address of the CAM array is used to break the tie.  “Match Found” bit is set if there is a match.

B.Supmonchai Digital ICs Memories 12 Read-Only Memory (ROM) Cells WL BL“1” WL BL “0” Diode ROM WL BL WL BL V DD MOS ROM 1 WL BL WL BL GND MOS ROM 2 No isolation More area and higher drive strength

B.Supmonchai Digital ICs Memories 13 MOS OR ROM Implementation Shared to reduce overhead Mirror Cell

B.Supmonchai Digital ICs Memories 14 MOS NOR ROM Implementation Missing NMOS means storing a “1”

B.Supmonchai Digital ICs Memories 15 MOS NOR ROM Layout Polysilicon Metal1 Diffusion Metal1 on Diffusion BL[0]BL[1]BL[2]BL[3] WL[0] WL[1] WL[2] WL[3] GND ACTIVE Layer (9.5 x 7 ) BL[0]BL[1]BL[2]BL[3] WL[0] WL[1] WL[2] WL[3] GND CONTACT Layer (11 x 7 )

B.Supmonchai Digital ICs Memories 16 WL[0] WL[1] WL[2] WL[3] V DD Pull-up devices BL[3]BL[2]BL[1]BL[0] MOS NAND ROM All word lines high by default with exception of selected row No Supply or GND needed in the cell

B.Supmonchai Digital ICs Memories 17 MOS NAND ROM Layout Using Implantation (5 x 6 ) BL[0]BL[1]BL[2]BL[3] WL[0] WL[1] WL[2] WL[3] Polysilicon Threshold-Altering Implant Diffusion Metal1 on Diffusion METAL-1 Layer (8 x 7 ) BL[0]BL[1]BL[2]BL[3] WL[0] WL[1] WL[2] WL[3] Drastically reduced cell size but Loss in Performance compared to NOR ROM

B.Supmonchai Digital ICs Memories 18 Transient Model for NOR ROM Transient Response  time from word line activation to bit line traversing voltage swing (typically 10% of V dd ). Word line parasitics (polysilicon)  Wire resistance  Wire capacitance  Gate capacitance Word line parasitics (polysilicon)  Wire resistance  Wire capacitance  Gate capacitance Bit line parasitics (Metal1)  Wire resistance (negligible)  Wire capacitance  Drain capacitance Bit line parasitics (Metal1)  Wire resistance (negligible)  Wire capacitance  Drain capacitanceVDD C bit r word c word WL BL Metal1 Poly

B.Supmonchai Digital ICs Memories 19 Transient Model for NAND ROM VDD C L r word c word WL BL c bit bit r Word line parasitics (polysilicon)  Wire resistance  Wire capacitance  Gate capacitance Word line parasitics (polysilicon)  Wire resistance  Wire capacitance  Gate capacitance Bit line parasitics (Metal1)  Resistance of cascaded transistors dominates  Drain/Source and complete gate capacitance Bit line parasitics (Metal1)  Resistance of cascaded transistors dominates  Drain/Source and complete gate capacitance Worst case scenario: when the cascade chain is populated  distributed RC model

B.Supmonchai Digital ICs Memories 20 Example: 512  512 NOR ROM rcM  Word line delay can be computed using the distributed rc-line model where the line consists of M sections/cells.  Assume a (0.5/0.25) pull-up device and a (1.3125/0.25) pull-down transistor t word = 0.38(r word c word )M 2 = 0.38(17.5 Ω * ( )fF) * = 1.4 ns C bit = 512*( )fF = 0.46 pF C bit = 512*( )fF = 0.46 pF t bit, HL = 0.69(13 kΩ /2 || 31 kΩ /5.25)*0.46 pF = 0.98 ns t bit, LH = 0.69(31 kΩ /5.25)*0.46 pF = 1.87 ns  Bit line response time depends on the transition direction.

B.Supmonchai Digital ICs Memories 21 Example: 512  512 NAND ROM  Using similar technique, word line delay of the NAND ROM can be computed as  For the bit line, the worst case high-to-low delay occurs when the complete column is populated with 0s except the bottommost transistor. By the distributed rc-line model (which is a fair approximation) t word = 0.38(15 Ω * ( )fF) * = 1.3 ns t bit, LH = 0.69(31 kΩ /0.0077) * (511*0.85 fF) = 1.2 µs - slow! t bit, HL = 0.38(8.7 kΩ * 0.85 fF) * = 0.73 µs  Using Elmore delay for the low-to-high worst case delay which occurs when the bottommost transistor is off.

B.Supmonchai Digital ICs Memories 22 NOR and NAND ROM Disadvantages  NOR and NAND ROM inherit all disadvantages of the pseudo-NMOS gate:  Ratioed Logic  Ratioed Logic: VOL is determined by the ratio of the pull-up and pull-down devices -> unacceptable transistor ratios.  Static Power Consumption  Static Power Consumption: static current exists between the supply rails when the output is low -> severe power dissipation  Solution: use fully complimentary NAND and NOR gate (i.e. CMOS)  large number of transistor -> large area

B.Supmonchai Digital ICs Memories 23 Precharged MOS NOR ROM pre f WL[0] GND BL[0] WL[1] WL[2] WL[3] BL[1] Precharge devices BL[2]BL[3] GND DD V Can be made as large as necessary, but clock driver becomes harder to design.

B.Supmonchai Digital ICs Memories 24 Precharged ROM Advantages  Eliminates the static power dissipation and the ratioed logic requirements with the same cell complexity.  Eliminate the enabling NMOS transistor at the bottom of the pull-down network.  Enables independent control of the pull-up and pull-down timing. dynamic precharging  Almost all of the large memories currently designed, including NVRWM and RAMS, use dynamic precharging.

B.Supmonchai Digital ICs Memories 25  Drive the word line from both sides  Use a metal bypass  Use Silicides polysilicon word line metal word line driverdriverWL Reduce delay by 4 polysilicon word line metal bypass WL metal word line K cells Decreasing Word Line Delay

B.Supmonchai Digital ICs Memories 26 Decreasing Bit Line Delay (Energy)  Reduce the bit line voltage swing (typical value ~ 0.5 V)  need sense amp for each column to sense/restore signal pulsed word line  Isolate memory cells from the bit lines after sensing (to prevent the cells from changing the bit line voltage further) - pulsed word line  generation of word line pulses very critical  too short  too short - sense amp operation may fail  too long  too long - power efficiency degraded (because bit line swing size depends on duration of the word line pulse)  use feedback signal from bit lines bit line isolation  Isolate sense amps from bit lines after sensing (to prevent bit lines from having large voltage swings) - bit line isolation

B.Supmonchai Digital ICs Memories 27 ROM Perspectives  Application-Specific ROMs  Application-Specific ROMs - custom designed for a particular application  Designer has freedom to choose any mask layer (or combination thereof) to program the device.  Commodity ROMs  Commodity ROMs - mass-produced memories that later customized according to customer specifications.  Mask-programmable  Mask-programmable: use contact/metal mask to program the memory; becoming unpopular  Electrically programmable  Electrically programmable: use fuses/anti-fuses to program; “write once” (cannot be reprogrammed).

B.Supmonchai Digital ICs Memories 28 Non-Volatile Memories (NVM) Basics Schematic symbol G S D The Floating-gate Avalanche-injection MOS transistor (FAMOS) (FAMOS) Device cross-section Floating gate Source Substrate Gate Drain n + n +_ p t ox t Allow threshold voltage to be altered electrically and retained indefinitely.

B.Supmonchai Digital ICs Memories V 10 V5 V 20 V DS Avalanche injection Floating-Gate Transistor Programming 0 V - 5 V 0 V DS Removing programming voltage leaves charge trapped 5 V V 5 V DS Programming results in higherV T. V T V T ~ 7V

B.Supmonchai Digital ICs Memories 30 Note on NVMs  NVM structure is identical to the ROM except that it uses FAMOS transistors or similar devices at the cell level instead. method of erasing  The method of erasing is the main differentiating factor between the classes of NVM  The programming of the NVM slower  Is typically an order of magnitude slower than the reading operation. no less than twice  Requires high voltage power supply - no less than twice the supply voltage need in the reading operation

B.Supmonchai Digital ICs Memories 31 EPROM Summary  Based on ROM structure and FAMOS devices. simple and dense  Extremely simple and dense: suitable for large low-cost memories that do not require regular programming.  Erased by shining UV lights through a transparent window in the package.  UV makes oxide slightly conductive, leaking charges very slowly from the floating gate.

B.Supmonchai Digital ICs Memories 32 EPROM Drawbacks slow  Erasure process is slow from 10’s to 1000’s sec and must be done “Off System”.  Limited Endurance  Limited Endurance: Number of erase/program cycles is generally limited to at most  Reliability Problems  Reliability Problems: device threshold might vary with repeated programming.  Use on-chip circuitry to control the threshold voltage to within an acceptable range during programming.  High power dissipation  High power dissipation during programming.

B.Supmonchai Digital ICs Memories 33 EEPROM: FLOTOX Device FLOTOX cross section Fowler-Nordheim I-V characteristic Floating gate Source Substrate p Gate Drain n 1 n 1 20–30 nm 10 nm -10 V 10 V I V GD The FLOating-gate Tunneling OXide (FLOTOX) transistor Injecting electrons onto the gate raises V T. Reverse operation (voltage) lower V T BidirectionalProgramming

B.Supmonchai Digital ICs Memories 34 EEPROM Cell WL BL V DD  hard.  Absolute threshold control for FLOTOX transistor is hard.   Over-erasing programmed transistor might end up with a depletion device.  FLOTOX always ON  2 transistor cell  Solution: 2 transistor cell  Write FLOTOX  Read NMOS FLOTOX NMOS

B.Supmonchai Digital ICs Memories 35 EEPROM Summary  Larger Area  Larger Area than EPROM of the same capacity  FLOTOX is inherently larger than FAMOS  2T cells as to 1T cell for the EPROM.  Thin oxide is hard to fabricate and expensive. higher  Cost/bit is higher than the EPROMs  Higher versatility:  Higher versatility: in situ programming/erasure  Last longer V T  Last longer (up to 10 5 erase/program cycles) but repeated programming can cause V T to drift due to permanently trapped charges in SiO 2

B.Supmonchai Digital ICs Memories 36 Flash EEPROM: ETOX Device Control gate erasure p-substrate Floating gate Thin tunneling oxide n 1 source n 1 drain programming Many other options …

B.Supmonchai Digital ICs Memories 37 EPROM Flash Courtesy Intel Cross-sections of NVM cells

B.Supmonchai Digital ICs Memories 38 Erase in a NOR Flash Memory

B.Supmonchai Digital ICs Memories 39 Write in a NOR Flash Memory

B.Supmonchai Digital ICs Memories 40 Read in a NOR Flash Memory

B.Supmonchai Digital ICs Memories 41 NAND Flash Memory Unit Cell Word line(poly) Source line (Diff. Layer)

B.Supmonchai Digital ICs Memories 42 Word linesSelect transistor Bit line contactSource line contact Active area STI Courtesy Toshiba NAND Flash Memory

B.Supmonchai Digital ICs Memories 43 State-of-the-art NVM

B.Supmonchai Digital ICs Memories 44  STATIC (SRAM)  DYNAMIC (DRAM) Read-Write Memories (RAM)  Data stored as long as supply is applied  Large (6 transistors/cell)  Fast  Differential  Periodic refresh required  Small (1-3 transistors/cell)  Slower  Single Ended

B.Supmonchai Digital ICs Memories 45 6-Transistor CMOS SRAM Cell WL BL V DD M 5 M 6 M 4 M 1 M 2 M 3 BL Q Q

B.Supmonchai Digital ICs Memories 46 CMOS SRAM Analysis (Read) WL BL V DD M 5 M 6 M 4 M 1 V V V BL Q = 1 Q = 0 C bit C

B.Supmonchai Digital ICs Memories 47 CMOS SRAM Analysis (Read) Cell Ratio (CR) 2.53 Voltage Rise (V)

B.Supmonchai Digital ICs Memories 48 CMOS SRAM Analysis (Write) BL = 1 = 0 Q = 0 Q = 1 M 1 M 4 M 5 M 6 V DD V WL

B.Supmonchai Digital ICs Memories 49 CMOS SRAM Analysis (Write)

B.Supmonchai Digital ICs Memories 50 V DD GND Q Q WL BL M1 M3 M4M2 M5M6 6T-SRAM Layout

B.Supmonchai Digital ICs Memories 51 Resistance-load SRAM Cell Static power dissipation -- Want R L large Bit lines precharged to V DD to address t p problem M 3 R L R L V DD WL QQ M 1 M 2 M 4 BL

B.Supmonchai Digital ICs Memories 52 SRAM Characteristics

B.Supmonchai Digital ICs Memories 53 3-Transistor DRAM Cell No constraints on device ratios Reads are non-destructive Value stored at node X when writing a “1” = V WWL -V Tn WWL BL1 M 1 X M 3 M 2 C S 2 RWL V DD V 2 V T D V V 2 V T BL2 1 X RWL WWL

B.Supmonchai Digital ICs Memories 54 BL2BL1GND RWL WWL M3 M2 M1 3T-DRAM Layout

B.Supmonchai Digital ICs Memories 55 1-Transistor DRAM Cell Write: C S is charged or discharged by asserting WL and BL. Read: Charge redistribution takes places between bit line and storage capacitance Voltage swing is small; typically around 250 mV.  V BL V PRE –V BIT V PRE – C S C S C BL == V

B.Supmonchai Digital ICs Memories 56 DRAM Cell Observations  1T DRAM requires a sense amplifier for each bit line, due to charge redistribution read-out.  DRAM memory cells are single ended in contrast to SRAM cells.  The read-out of the 1T DRAM cell is destructive  read and refresh operations are necessary for correct operation.  Unlike 3T cell, 1T cell requires presence of an extra capacitance that must be explicitly included in the design.

B.Supmonchai Digital ICs Memories 57 DRAM Cell Observations II  When writing a “1” into a DRAM cell, a threshold voltage is lost. This charge loss can be circumvented by bootstrapping the word lines to a higher value than VDD

B.Supmonchai Digital ICs Memories 58 Sense Amp Operation D V(1) V V(0) t V PRE V BL Sense amp activated Word line activated

B.Supmonchai Digital ICs Memories 59 1-T DRAM Cell Uses Polysilicon-Diffusion Capacitance Expensive in Area Cross-section Layout M 1 word line Diffused bit line Polysilicon gate Polysilicon plate Capacitor Metal word line Poly SiO 2 Field Oxide n + n + Inversion layer induced by plate bias Poly

B.Supmonchai Digital ICs Memories 60 SEM of poly-diffusion capacitor 1T-DRAM 1T-DRAM SEM

B.Supmonchai Digital ICs Memories 61 Cell Plate Si Capacitor Insulator Storage Node Poly 2nd Field Oxide Refilling Poly Si Substrate Trench Cell Stacked-capacitor Cell Capacitor dielectric layer Cell plate Word line Insulating Layer IsolationTransfer gate Storage electrode Advanced 1T DRAM Cells

B.Supmonchai Digital ICs Memories 62 Static CAM Memory Cell

B.Supmonchai Digital ICs Memories 63 CAM in Cache Memory

B.Supmonchai Digital ICs Memories 64 Periphery  Decoders  Sense Amplifiers  Input/Output Buffers  Control/Timing Circuitry

B.Supmonchai Digital ICs Memories 65 Collection of 2 M complex logic gates Organized in regular and dense fashion (N)AND Decoder NOR Decoder Row Decoders

B.Supmonchai Digital ICs Memories 66 Hierarchical Decoders A 2 A 2 A 2 A 3 WL 0 A 2 A 3 A 2 A 3 A 2 A 3 A 3 A 3 A 0 A 0 A 0 A 1 A 0 A 1 A 0 A 1 A 0 A 1 A 1 A 1 1 Multi-stage implementation improves performance NAND decoder using 2-input pre-decoders

B.Supmonchai Digital ICs Memories 67 2-input NOR decoder2-input NAND decoder Dynamic Decoders

B.Supmonchai Digital ICs Memories 68 Advantages: speed (t pd does not add to overall memory access time) Only one extra transistor in signal path Only one extra transistor in signal path Disadvantage: Large transistor count 2-input NOR decoder A 0 S 0 BL A 1 S 1 S 2 S 3 D 4-Input PT Based Column Decoder

B.Supmonchai Digital ICs Memories 69 4-to-1 Tree Based Column Decoder Number of devices drastically reduced Delay increases quadratically with # of sections; prohibitive for large decoders buffers progressive sizing combination of tree and pass transistor approaches Solutions: BL D A 0 A 0 A 1 A 1

B.Supmonchai Digital ICs Memories 70 t p C  V  I av = make  V as small as possible smalllarge Idea: Use Sense Amplifer output input s.a. small transition Sense Amplifiers

B.Supmonchai Digital ICs Memories 71 Differential Sense Amplifier M 4 M 1 M 5 M 3 M 2 V DD bit SE Out y Directly applicable to SRAMs

B.Supmonchai Digital ICs Memories 72 Differential Sensing - SRAM

B.Supmonchai Digital ICs Memories 73 Latch-Based Sense Amplifier (DRAM) EQ V DD BL SE  Initialized in its meta- stable point with EQ  Once adequate voltage gap created, sense amp enabled with SE  Positive feedback quickly forces output to a stable operating point.

B.Supmonchai Digital ICs Memories 74 Case Studies  Programmable Logic Array  SRAM  Flash Memory

B.Supmonchai Digital ICs Memories 75 Programmable Logic Array (PLA)  Structured approach to random logic  “Two level logic implementation”  NOR-NOR (Product of Sums)  NAND-NAND (Sum of Products)  Circuit structure similar to ROM with main difference:  ROM: fully populated  PLA: one element per min term  Importance of PLA has drastically reduced  Slow  Better optimization techniques (Multi-level logic synthesis)

B.Supmonchai Digital ICs Memories 76 Pseudo-NMOS PLA AND-planeOR-plane

B.Supmonchai Digital ICs Memories 77 Dynamic PLA AND-planeOR-plane GND V DD V X 0 X 0 X 1 f 0 f 1 X 1 X 2 X 2 AND f f OR f f

B.Supmonchai Digital ICs Memories 78 (a) Clock signals(b) Timing generation circuitry f t pre t eval f AND f f f f OR f Dummy AND row Clock Signal Generation for self-timed dynamic PLA Dynamic PLA: Clock Signal Generation

B.Supmonchai Digital ICs Memories 79 PLA Layout

B.Supmonchai Digital ICs Memories 80 4 Megabit SRAM Hierarchical Word-line Architecture

B.Supmonchai Digital ICs Memories 81 Bit-line Circuitry

B.Supmonchai Digital ICs Memories 82 Sense Amplifier (and Waveforms) I/O Lines Address Data-cut ATD BEQ SEQ DATA Vdd GND SA, SA Vdd GND

B.Supmonchai Digital ICs Memories 83 1 Gigabit Flash Memory From [Nakamura02]

B.Supmonchai Digital ICs Memories 84 Writing Flash Memory Read level (4.5 V) Number of cells V1V2V Vt of memory cells 3V4V Evolution of thresholdsFinal Distribution From [Nakamura02]

B.Supmonchai Digital ICs Memories mm 2 1Gbit NAND Flash Memory 10.7mm 11.7mm 2kB Page buffer & cacheCharge pump bit lines 32 word lines x 1024 blocks From [Nakamura02]

B.Supmonchai Digital ICs Memories mm 2 1Gbit NAND Flash Memory Technology: Technology: 0.13  m p-sub CMOS triple-well 1poly, 1polycide, 1W, 2Al Cell size: Cell size:  m2 Chip size: Chip size: 125.2mm2 Organization: Organization: 2112 x 8b x 64 page x 1k block Power supply: Power supply: 2.7V-3.6V Cycle time: Cycle time: 50ns Read time: Read time:25  s Program time: Program time:200  s / page Erase time: Erase time: 2ms / block From [Nakamura02]

B.Supmonchai Digital ICs Memories 87 Semiconductor Memory Trends From [Itoh01] Quadruple every three year

B.Supmonchai Digital ICs Memories 88 Trends in Memory Cell Area From [Itoh01]

B.Supmonchai Digital ICs Memories 89 Technology feature size for different SRAM generations Technology Scaling Trends