Stacking Untested Wafers to Improve Yield. The 3D Enigma

Similar documents
EECS 598: Integrating Emerging Technologies with Computer Architecture. Lecture 10: Three-Dimensional (3D) Integration

Advancing high performance heterogeneous integration through die stacking

TechSearch International, Inc.

Hybrid Memory Cube (HMC)

Interconnect Challenges in a Many Core Compute Environment. Jerry Bautista, PhD Gen Mgr, New Business Initiatives Intel, Tech and Manuf Grp

3D SYSTEM INTEGRATION TECHNOLOGY CHOICES AND CHALLENGE ERIC BEYNE, ANTONIO LA MANNA

Emerging IC Packaging Platforms for ICT Systems - MEPTEC, IMAPS and SEMI Bay Area Luncheon Presentation

Monolithic 3D Integration using Standard Fab & Standard Transistors. Zvi Or-Bach CEO MonolithIC 3D Inc.

All Programmable: from Silicon to System

SRC 3D Summit. Bob Patti, CTO

Physical Design Implementation for 3D IC Methodology and Tools. Dave Noice Vassilios Gerousis

Heterogeneous Integration and the Photonics Packaging Roadmap

DRAM Main Memory. Dual Inline Memory Module (DIMM)

Problem 2 If the cost of a 12 inch wafer (actually 300mm) is $3500, what is the cost/die for the circuit in Problem 1.

High Volume Manufacturing Supply Chain Ecosystem for 2.5D HBM2 ASIC SiPs

3D systems-on-chip. A clever partitioning of circuits to improve area, cost, power and performance. The 3D technology landscape

The Memory Hierarchy 1

Future Memories. Jim Handy OBJECTIVE ANALYSIS

Test and Measurement Challenges for 3D IC Development. R. Robertazzi IBM Research

TSV Test. Marc Loranger Director of Test Technologies Nov 11 th 2009, Seoul Korea

3D Integration: New Opportunities for Speed, Power and Performance. Robert Patti, CTO

Bringing 3D Integration to Packaging Mainstream

Xilinx SSI Technology Concept to Silicon Development Overview

NoC Round Table / ESA Sep Asynchronous Three Dimensional Networks on. on Chip. Abbas Sheibanyrad

3D TECHNOLOGIES: SOME PERSPECTIVES FOR MEMORY INTERCONNECT AND CONTROLLER

Interposer Technology: Past, Now, and Future

Opportunities & Challenges: 28nm & 2.5/3-D IC Design and Manufacturing

Introduction 1. GENERAL TRENDS. 1. The technology scale down DEEP SUBMICRON CMOS DESIGN

Advanced CSP & Turnkey Solutions. Fumio Ohyama Tera Probe, Inc.

Non-contact Test at Advanced Process Nodes

Stacked Silicon Interconnect Technology (SSIT)

TechSearch International, Inc.

PicoServer : Using 3D Stacking Technology To Enable A Compact Energy Efficient Chip Multiprocessor

High performance HBM Known Good Stack Testing

From 3D Toolbox to 3D Integration: Examples of Successful 3D Applicative Demonstrators N.Sillon. CEA. All rights reserved

3D technology for Advanced Medical Devices Applications

3-D Package Integration Enabling Technologies

PUSHING THE LIMITS, A PERSPECTIVE ON ROUTER ARCHITECTURE CHALLENGES

Comparison & highlight on the last 3D TSV technologies trends Romain Fraux

Centip3De: A 64-Core, 3D Stacked, Near-Threshold System

edram to the Rescue Why edram 1/3 Area 1/5 Power SER 2-3 Fit/Mbit vs 2k-5k for SRAM Smaller is faster What s Next?

ECE7995 (4) Basics of Memory Hierarchy. [Adapted from Mary Jane Irwin s slides (PSU)]

Photonics & 3D, Convergence Towards a New Market Segment Eric Mounier Thibault Buisson IRT Nanoelec, Grenoble, 21 mars 2016

DFT-3D: What it means to Design For 3DIC Test? Sanjiv Taneja Vice President, R&D Silicon Realization Group

There is a paradigm shift in semiconductor industry towards 2.5D and 3D integration of heterogeneous parts to build complex systems.

Packaging Technology for Image-Processing LSI

CSE502: Computer Architecture CSE 502: Computer Architecture

Beyond Moore. Beyond Programmable Logic.

A 400Gbps Multi-Core Network Processor

Paradigms Shifts in CMP

Brief Background in Fiber Optics

An integrated solution for KGD: At-speed wafer-level testing and full-contact wafer-level burn-in after flip chip bumping

3D technology evolution to smart interposer and high density 3D ICs

Topic 21: Memory Technology

Topic 21: Memory Technology

AIM Photonics: Manufacturing Challenges for Photonic Integrated Circuits

Lecture 2: Performance

ARCHIVE Françoise von Trapp Editorial Director 3D InCites ABSTRACT

From Majorca with love

BREAKING THE MEMORY WALL

Multi-Die Packaging How Ready Are We?

OVERCOMING THE MEMORY WALL FINAL REPORT. By Jennifer Inouye Paul Molloy Matt Wisler

On GPU Bus Power Reduction with 3D IC Technologies

Chapter 2 Three-Dimensional Integration: A More Than Moore Technology

Memory as We Approach a New Horizon

Near-Threshold Computing: Reclaiming Moore s Law

IMEC CORE CMOS P. MARCHAL

SEMI 大半导体产业网 MEMS Packaging Technology Trend

UNIT 4 INTEGRATED CIRCUIT DESIGN METHODOLOGY E5163

VISUALIZING THE PACKAGING ROADMAP

Some material adapted from Mohamed Younis, UMBC CMSC 611 Spr 2003 course slides Some material adapted from David Culler, UC Berkeley CS252, Spr 2002

Probing 25µm-diameter micro-bumps for Wide-I/O 3D SICs

High Performance Memory Opportunities in 2.5D Network Flow Processors

HPC Technology Trends

Z-RAM Ultra-Dense Memory for 90nm and Below. Hot Chips David E. Fisch, Anant Singh, Greg Popov Innovative Silicon Inc.

Driving Leading Edge Microprocessor Technology

Introduction. SK hynix

Towards Performance Modeling of 3D Memory Integrated FPGA Architectures

Silicon Based Packaging for 400/800/1600 Gb/s Optical Interconnects

Monolithic Integration of Energy-efficient CMOS Silicon Photonic Interconnects

Industry Trends in 3D and Advanced Packaging

A Design Tradeoff Study with Monolithic 3D Integration

Introduction. Summary. Why computer architecture? Technology trends Cost issues

SOLVING MANUFACTURING CHALLENGES AND BRINGING SPIN TORQUE MRAM TO THE MAINSTREAM

Paul Franzon. Department of Electrical and Computer Engineering

PSMC Roadmap For Integrated Photonics Manufacturing

Multilevel Memories. Joel Emer Computer Science and Artificial Intelligence Laboratory Massachusetts Institute of Technology

ECE 571 Advanced Microprocessor-Based Design Lecture 17

Memories: Memory Technology

Scaling the Compute and High Speed Networking Needs of the Data Center with Silicon Photonics ECOC 2017

Gigascale Integration Design Challenges & Opportunities. Shekhar Borkar Circuit Research, Intel Labs October 24, 2004

ZEISS Launches New High-resolution 3D X-ray Imaging Solutions for Advanced Semiconductor Packaging Failure Analysis

CMOS Photonic Processor-Memory Networks

Lecture: Memory, Multiprocessors. Topics: wrap-up of memory systems, intro to multiprocessors and multi-threaded programming models

Technology, Manufacturing and Markets of Magnetoresistive Random Access Memory (MRAM) Brad Engel, VP- Product Development & Quality

Intel: Driving the Future of IT Technologies. Kevin C. Kahn Senior Fellow, Intel Labs Intel Corporation

Chapter 0 Introduction

Moore s s Law, 40 years and Counting

CHAPTER 1 INTRODUCTION

When it comes to double-density Flash memory, some pairs are just better.

Transcription:

Stacking Untested Wafers to Improve Yield or 3D: Where the Timid Go to Die The 3D Enigma The Promise High Performance Low Power Improved Density More than Moore or at least as much as Moore The Reality A decade of next year High costs Low yields Non-existent ecosystem Loss of credibility www.cpmt.org/scv/ 1

Historic Precedent Poor Science www.cpmt.org/scv/ 2

Excessive Complexity Fear www.cpmt.org/scv/ 3

But Flying is EASY Heroes focus on Landing US Airways Flight 1549 in the Hudson River Capt. Chesley Sullenberger in command. www.cpmt.org/scv/ 4

Step 1: Commit to 3D DiRAM True 3D RAM Architecture Tezzaron Confidential 10 www.cpmt.org/scv/ 5

Dis-Integrated 3D RAM Architecture DiRAM Architecture Memory Cells and Access Transistors Sense Amps, etc. I/O Layer Tezzaron Confidential 11 256 Independent Channels Each Channel 256 Mb Storage 64 Gb/s Bandwidth 9ns Latency 15ns trc 16 Paired Banks Tezzaron Confidential 12 www.cpmt.org/scv/ 6

DiRAM4 Stack Performance 64 Gigabits Storage 16.4 Terabit/s Data Bandwidth 4096 Open Pages 256 > 500 Billion Independent Transactions Per Channels Second Tezzaron Confidential 13 DiRAM4 Scale* Drawing 14 mm Top View Side View 175 mm 2 0.5 mm 12.5 mm Isometric View * Almost to scale Tezzaron Confidential 14 www.cpmt.org/scv/ 7

Via-Free Wafer Stacking Copper TSV 10µ x 50µ Tezzaron Tungsten SuperContact 1µ x 5.5µ 2x Tezzaron Confidential 15 Tiny, Common, Cheap, Fast and One 100 10µ Diameter 1µ Diameter Copper TSV Tungsten SuperContact www.cpmt.org/scv/ 8

Radically Different Manufacturing Conventional Flow Fabricate Wafer Probe Test Die Thin Wafer Singulate Die Stack Good Die Package Stack Burn-In & Test Stack Tezzaron Flow Fabricate Wafer Stack Stack & Wafers Thin Wafers Thin Top Wafer Probe Test Wafer Repeat Stacks Probe Test Stacks Singulate Singulate Stacks Stacks Package Stacks Burn-In & Test Stack Tezzaron Confidential 17 Never Handle A Thin Wafer The Tezzaron Mantra Bond Two Grind One to make the world s thinnest RAM wafers Tezzaron Confidential 18 www.cpmt.org/scv/ 9

That can t work bonding un-tested die will produce near zero yields, poor reliability and high costs. Translation You Tezzaron people are crazy! Tezzaron Confidential 19 Novati Technologies - Austin Tezzaron subsidiary Manufacturing volume Services available 3D Assembly Options Cu-Cu DBI, Oxide Bonding Intermetallic Gold-Indium, Gold-Gold Silicon Interposers Passive Passive Plus Active Tezzaron Confidential 20 www.cpmt.org/scv/ 10

Super dense interconnect allows Bi-STAR Built-in Self Test And Repair Controlled by embedded ARM processor Enabled by per-cell control interconnect Super-fine grained test and repair Continuous, in-the-system hard and soft error repair Tezzaron Confidential 21 Bi-STAR Does More, Works Better Bi-STAR Repairs Bad memory cells Bad line drivers Bad sense amps Shorted word lines Shorted bit-lines Leaky bits Bad secondary bus drivers Bi-STAR Tests Tests > 300,000 000 nodes per clock cycle Tests > 1,000x faster than external memory tester Via SPI port, works with Host to allow continuous scrub / repair Tezzaron Confidential 22 www.cpmt.org/scv/ 11

Bi-STAR Repair Improves Yield 100% Yield Stack Height Tezzaron Confidential 23 Solve 3D Problems with 3D Bonding Untested Wafers Enables Super Thin Wafers Enables Effective Repair Enables Sub-micron Vias www.cpmt.org/scv/ 12

DiRAM: Efficiency for the Future Less aggressive wafers Higher array efficiency Much lower test cost Higher yield Longer product life cycles Tezzaron Confidential 25 DiRAM: Efficiency for the Future Less aggressive wafers Higher array efficiency Much lower test cost Higher yield Longer product life cycles Tezzaron Confidential 26 www.cpmt.org/scv/ 13

High End Routing Tasks Packet Buffer (Burst Read/Write) Tables (Read Dominated) Stats (Read-Mod-Write) Performance Metrics Density Line Rate / Seconds Bandwidth Line Rate x 2.5 Transaction Rate Transactions x Line Rate / Min Packet Size Tezzaron Confidential 27 400Gb Routing with DiRAM4 Network Processor 22 mm x 19 mm DiRAM4 14 mm x 12.5 mm 26 mm x 32 mm Interposer Tezzaron Confidential 28 www.cpmt.org/scv/ 14

1Tb Routing with DiRAM4 Network Processor 14 mm m x 12.5 mm DiRAM4 22 mm x 19 mm 14 mm x 12 2.5 mm DiRAM M4 27 mm x 35 mm Interposer Tezzaron Confidential 29 The Right I/O for Each Market Photonic I/O Lowest Power Highest Performance Pico-SerDes DiRAM4 I/O Lower Power Launches Moderate Performance SerDes with I/O Highest Power 0.7 V CMOS I/O Moderate Performance 2.5D Si or Organic Low Power High Performance Tezzaron Confidential 30 www.cpmt.org/scv/ 15

Tezzaron DiRAM Roadmap 64 Tb/s 32 Tb/s 16 Tb/s 2014 2016 2018 Tezzaron Confidential 31 David Chapman VP Marketing & Technical Sales 512-356-2534 dchapman@tezzaron.com Tezzaron Confidential 32 www.cpmt.org/scv/ 16