Slide 1: DFI and other New Sources of Data for Yield, Reliability and Process Control
16th Annual PDF Solutions Users Conference S1.4 – DFI™ and other New Sources of Data for Yield, Reliability and Process Control October 15, 2019 Dennis Ciplickas, VP of Characterization Solutions
Slide 2: Forward-looking statements
This presentation contains forward-looking statements regarding PDF Solutions’ future products and business prospects that involve risk and uncertainty. Actual results could differ materially from those discussed. You should review PDF Solutions’ SEC filings, including its annual report on Form 10-K
Slide 3: Introduction
Introduction o PDF’s business has always been based on data o Deep insight from the unique data generated by our Characterization Vehicle® (CV®) Infrastructure is a key enabler for time-to-market in our Integrated Yield Ramp (IYR) business o As PDF’s analytics business has grown, so has the complexi
Slide 4: Industry Trends
Industry Trends o Automotive – Advanced Driver-Assistance Systems (ADAS) – Aggressive move into advanced technology nodes – Functional Safety – Quality o Data center – New computing architectures – CPU, GPU, AI, in-memory computing – New memory and storage architectures – SRAM, NVM, DRAM, NAND FLASH
Slide 5: Quality Driver: Automotive Electronics
Quality Driver: Automotive Electronics 1950 1% of total vehicle cost is electronic 2010 30% of total vehicle cost is electronic Est. 2030† 50% of total vehicle cost is electronic 5,000 1ppm failure semiconductor devices per rate = 5 failures every 10,000 cars manufactured per day average car 1,000 c
Slide 6: Power and Performance vs. Reliability
Need royalty free image here from DavidP Power and Performance vs. Reliability o Data Centers consume 3% of total electricity TODAY! † – Facebook: 100MW data center with 442,368 servers†† – Facebook: average server utilization ~40% (265,420 unused servers) †† o AI: Rat brain 518,400 processors, huma
Slide 7: Memory Driver: Data Volume + New Computing Architectures
Memory Driver: Data Volume + New Computing Architectures Flash Storage Persistent memories In-Memory Compute Embedded Memories o 3D NAND o X-Point Memories o Low-Power, Low-Latency o Low-Power, Low-Cost o Cost/Bit scaling o Power/Speed scaling embedded NVM code, data storage for o Multi-level, Analo
Slide 8: Key Takeaways
Key Takeaways o Safety, quality, and uptime are of paramount importance for new systems Reliability o New chip technologies demand new sources of data to ensure reliability o “Yesterday’s yield problem is tomorrow’s reliability problem” o Rapid learning cycles are essential to creating a quality pro
Slide 9: PDF Differentiated Data for Reliability, Efficiency, and Foresight
PDF Differentiated Data for Reliability, Efficiency, and Foresight Design-for-Inspection™ Characterization Vehicle® CV® Core System System System Functional SoC Test interface 2nd generation 5th generation eProbe® 250 pdFasTest® Sensor E-Beam HW E-Test HW System Software Interface Monitor PDF CV® Te
Slide 10: Emerging Memories
Emerging Memories o 62.7% of the total system energy is spent on Data movement† In-Memory Computing o Low latency Low-Power memory replacement SRAM MRAM for L3 Cache? o Emerging memories for Deep Learning On-chip High Density, Low Power solutions needed o Embedded memories for Automotive and I
Slide 11: Enabling Emerging Memory Development
Enabling Emerging Memory Development BEOL – integrated embedded memory elements :::: :::: :::: :::: :::: :::: Silicon Substrate Full flow lots fabricate both Cross-point array with pdFasTest® back-end memory element and parallel parametric pulse testing front-end selector devices slow enables effi
Slide 12: PDF Characterization System for Emerging Embedded Memories
PDF Characterization System for Emerging Embedded Memories o Unique Platform for BEOL-integrated Emerging Memories MRAM, PCM, ReRAM – Innovative test structure and test method – Cross Point Array for high Bit statistics – pdFasTest® F380+ highly parallel pulsed parametric test o Characterize milli
Slide 13: PDF Differentiated Data for Reliability, Efficiency, and Foresight
PDF Differentiated Data for Reliability, Efficiency, and Foresight Design-for-Inspection™ Characterization Vehicle® CV® Core System System System Functional SoC Test interface 2nd generation 5th generation eProbe® 250 pdFasTest® Sensor E-Beam HW E-Test HW System Software Interface Monitor PDF CV® Te
Slide 14: Inputs from our Partners
o “Detect the Undetectable” Find electrically relevant defects buried inside 3D structures o “We are not late on a program because of functional yield, what kills us is we give the customer a chip, they package it, burn it in and Inputs figure out that there is a fundamental issue in a material st
Slide 15: The Advantages of the DFI System
The Advantages of the DFI™ System o DFI is a contactless electrical test measurement using an e-beam system DFI Capability o DFI measurement speed (raw): 100M’s DUTs/hr today, Billions of DUTs/hr next o Calibrated Leakage Value: pA to nA baseline, capture outliers >10x o DFI DUT quantity: 1B to 100B
Slide 16: Why Does Leakage Matter?
Why Does Leakage Matter? Spacer, Contact/Metal It is incredibly difficult to Gate Dielectric catch such tiny leakages in Isolation 10B’s of transistors per chip > 1 µA > 1 µA > 100 nA > 100 nA Leakage Outlier Bins Identify Latent TDDB > 10 nA > 10 nA Weak Spots > 1 nA > 1 nA RELIABILITY RISK Typical
Slide 17: DFI Data: DUT Leakage Failure Rate Statistics
DFI™ Data: DUT Leakage Failure Rate Statistics > 1 uA > 100nA > 10nA > 1 nA o Green die are zero ppm: Die with no ppm failures observed o Leakage level provides unique reliability insight into latent weak spots o A composite of different fail mode estimates can indicate risk for early die failure or
Slide 18: DFI Die Control: Risk Grading
DFI™ Die Control: Risk Grading o High resolution map of every die for single fail type or composite of all fail types o Spatial signature determines good/bad/risky die o Within die perspective to assess risk for across field variability or local risk to certain fail type Detailed across-die measurem
Slide 19: What is DFI status today?
What is DFI™ status today? eProbe® System o eProbe® 150: Running in full automation in 3 R&D fabs o eProbe® 250: 1st tool running in full automation in production fab Silicon o Over 100 tape-outs to date, spanning 6 nodes o ~60 Product/MPW tape-outs with DFI content PDF Solutions, All Rights Reserve
Slide 20: DFI Accomplishments 2019
DFI™ Accomplishments 2019 eProbe® 150 o eProbe® 150 and Exensio DFI deployed at multiple sites using dense DFI test chips with throughput 10M’s cells/hr o In active use for yield learning on multiple nodes (22nm, 14nm, 7nm, 5nm) eProbe® 250 o Demonstrated fill cell measurement throughput at 500M Cel
Slide 21: Plenty of Work Remains!
Plenty of Work Remains! o Demonstrate production-worthy wafer level and die-level reliability grading app by leakage monitoring of critical fail modes o Make DFI™ an essential element for fab control by providing insight into otherwise undetectable issues o Expand applications of DFI™ into 3D NAND &
Slide 22: PDF Differentiated Data for Reliability, Efficiency, and Foresight
PDF Differentiated Data for Reliability, Efficiency, and Foresight Design-for-Inspection™ Characterization Vehicle® CV® Core System System System Functional SoC Test interface 2nd generation 5th generation eProbe® 250 pdFasTest® Sensor E-Beam HW E-Test HW System Software Interface Monitor PDF CV® Te
Slide 23: Leveraging IYR Technology: The CV Core System
Leveraging IYR Technology: The CV® Core System SoC Functional Test o The CV Core System integrates into your performance and reliability interface optimization strategy: – Capture weak die at t=0 Sensor – Monitor Degradation in the field – Detect Change/Failure, Self Diagnosis, Failure Prediction Sy
Slide 24: The CV Core System
The CV® Core System Exensio® Enterprise Analytics Reliability or Performance CV Core ✓ Functional Test SoC interface Software Agent x Sensor time System Software Interface Monitor Power x ✓ … Wafer Chip/Module/System Chip/System Design Manufacturing Assembly and Test Field Deployed Systems time Futu
Slide 25: PDF Differentiated Data Leverages Exensio End-to-end Analytics
PDF Differentiated Data Leverages Exensio® End-to-end Analytics Inline E-Test Wafer Sort Final Test / Burn-in In-System DFI™ CV® Core CV® Infrastructure o “Always available” characterization by SoC o Product relevant electrical fail o Detailed and accurate software at t>0 mode coverage, ppb resoluti
Slide 26: Smart Testing Benefits from Differentiated Data
Smart Testing Benefits from Differentiated Data CV® DFI™ Infrastructure CV® Core *Smart Testing can be applied to any test Target Inline Test Test Decision Final Final Fab WAT ASSY Station1 Station2 Node Test 1 Test 2 Data Input Data Output Data 200 Goals: Differentiated data leads to Fallout (DPP
Slide 27: Summary
Summary o PDF’s business has always been based on data o Semiconductor system complexity continues to increase o We continue to innovate new sources of data to efficiently improve reliability, performance, and yield, and reduce cost o The Exensio® end-to-end analytics platform leverages this data to
Slide 28: Thank You
Thank You pdf-solutions pdfsolutionsinc pdfs.inc pdf_solutions pdf_cn