Lewis Chu,Marketing Director,GUC

15
May 9, 2016 Zero- Defect Methodology for Automotive and Other Key Applications March, 2016 Authors: Ray Hou (GUC Automotive BD) Jason Lin (GUC Operations Manager) Igor Elkanovich (Verisense CTO) Lewis Chu (GUC Marketing Director) - presenter

Transcript of Lewis Chu,Marketing Director,GUC

Page 1: Lewis Chu,Marketing Director,GUC

• May 9, 2016

Zero- Defect Methodology

for Automotive and Other Key

Applications

March, 2016 Authors:

Ray Hou (GUC Automotive BD)

Jason Lin (GUC Operations Manager)

Igor Elkanovich (Verisense CTO)

Lewis Chu (GUC Marketing Director) - presenter

Page 2: Lewis Chu,Marketing Director,GUC

GUC Property – Security CCopyright © 2016 GUC Uncompromising Performance

• May 9, 2016

• Public company, traded in Taipei Stock Exchange

• Partially owned by TSMC

• Services : ASIC, Chip design service, IPs

• Worldwide offices: US, China, Europe, Japan and Korea

• Verisense, GUC Israeli partner, offers leading ASIC & FPGA design services

GUC Corporation Information

0

10

20

30

2010 2011 2012 2013 2014 2015

16nm

20nm

28nm

40nm

65nm

90nm

0.13um

0.18um above

GUC Tape Outs per year

P2

Page 3: Lewis Chu,Marketing Director,GUC

GUC Property – Security CCopyright © 2016 GUC Uncompromising Performance

• May 9, 2016

Automotive IC Reliability Requirements

• Reliability: < 5 FIT

– FIT (failure in time) is defined as a failure rate of 1 per billion hours

– 1 FIT is equivalent to having an MTBF of 1 billion hours

• Defects < 5-10 DPPM

– DPPM: defective parts per million ppm

• High max. ambient temperature

Min. Max.

0 - 40 °C 150 °C

1 - 40 °C 125 °C

2 - 40 °C 105 °C

3 - 40 °C 85 °C

GradeOperation Ambient Temperature

P3

Page 4: Lewis Chu,Marketing Director,GUC

GUC Property – Security CCopyright © 2016 GUC Uncompromising Performance

• May 9, 2016

GUC Comprehensive Reliability Management

• Reliability control at all stages thought the flow

DFR DFM DFT Reliability Manufacturing Testing

Risk/Robustnesschecking

ZD testingmethod

Outlierprogram

TightenedControl

P4

Page 5: Lewis Chu,Marketing Director,GUC

GUC Property – Security CCopyright © 2016 GUC Uncompromising Performance

• May 9, 2016

Failures Distribution in Time

Early failures Random failures Wear-out failuresFailures category

Main cause Defect Process Baseline Product design

Mechanism •High Test coverage•Outlier screening•3-temp. testing•Visual spec. tighten•Burn-in testing

•SPC/Cpk control•Equipment monitor•Reliability monitor•Continuous improvement

•Circuit design robustness•Design for Manufacturing•Package/material robustness•Drift analysis for spec margin

Screen weak chip

Reduce process variation

Enhance design robustness

P5

EM void in a Cu line

under a via

Brittle failure (Si fracture)

due to temperature cycling

Page 6: Lewis Chu,Marketing Director,GUC

GUC Property – Security CCopyright © 2016 GUC Uncompromising Performance

• May 9, 2016

DFT Approach for Automotive Product

DFT Item Automotive product

Digital • Stack at faults coverage > 99%• Transition faults coverage > 85%• Pseudo Stuck-at IDDQ faults coverage > 80%• Small Delay (Optional)• Bridge (Optional)

SRAM • March 68N• Checkerboard & Inverse background data• Physical mapping

ROM • ReadOnly 2N

Analog • Functionality and Spec check• Additional signal to increase analog testability

internal reference voltage monitoranalog bias tuning setting to touch performance boundary

• Analog test modes built in

P6

Page 7: Lewis Chu,Marketing Director,GUC

GUC Property – Security CCopyright © 2016 GUC Uncompromising Performance

• May 9, 2016

Silicon Process Variation

• Variation measured by Process Monitor (PM)

7,000

9,200

11,600

6,800

8,800

11,000

6500

7500

8500

9500

10500

11500

12500

13500

14500

15500

16500

0 400 800 1200 1600 2000 2400 2800 3200 3600 4000 4400 4800 5200 5600 6000 6400 6800

Co

un

t

Chip Sorting

PM Silicon vs. SPICE Correlation

BUFD4-7T35P140-HVT (Silicon_0.9V_25C)

BUFD4-7T35P140-SVT (Silicon_0.9V_25C)

BUFD4-7T35P140-HVT (SS_0.9V_25C)

BUFD4-7T35P140-HVT (TT_0.9V_25C)

BUFD4-7T35P140-HVT (FF_0.9V_25C)

BUFD4-7T35P140-SVT (SS_0.9V_25C)

BUFD4-7T35P140-SVT (TT_0.9V_25C)

BUFD4-7T35P140-SVT (FF_0.9V_25C)

7,000

9,200

11,600

6,800

8,800

11,000

6500

7000

7500

8000

8500

9000

9500

10000

10500

11000

11500

12000

0 400 800 1200 1600 2000 2400 2800 3200 3600 4000 4400 4800 5200 5600 6000 6400 6800

Co

un

t

Chip Sorting

PM Silicon vs. SPICE Correlation

BUFD4-7T35P140-HVT (Silicon_0.9V_25C)

BUFD4-7T35P140-SVT (Silicon_0.9V_25C)

BUFD4-7T35P140-HVT (SS_0.9V_25C)

BUFD4-7T35P140-HVT (TT_0.9V_25C)

BUFD4-7T35P140-HVT (FF_0.9V_25C)

BUFD4-7T35P140-SVT (SS_0.9V_25C)

BUFD4-7T35P140-SVT (TT_0.9V_25C)

BUFD4-7T35P140-SVT (FF_0.9V_25C)

7,000

9,200

11,600

6,800

8,800

11,000

6500

7000

7500

8000

8500

9000

9500

10000

10500

11000

11500

12000

0 400 800 1200 1600 2000 2400 2800 3200 3600 4000 4400 4800 5200 5600 6000 6400 6800

Co

un

t

Chip Sorting

PM Silicon vs. SPICE Correlation

BUFD4-7T35P140-HVT (Silicon_0.9V_25C)

BUFD4-7T35P140-SVT (Silicon_0.9V_25C)

BUFD4-7T35P140-HVT (SS_0.9V_25C)

BUFD4-7T35P140-HVT (TT_0.9V_25C)

BUFD4-7T35P140-HVT (FF_0.9V_25C)

BUFD4-7T35P140-SVT (SS_0.9V_25C)

BUFD4-7T35P140-SVT (TT_0.9V_25C)

BUFD4-7T35P140-SVT (FF_0.9V_25C)

7,000

9,200

11,600

6,800

8,800

11,000

6500

7000

7500

8000

8500

9000

9500

10000

10500

11000

11500

12000

0 400 800 1200 1600 2000 2400 2800 3200 3600 4000 4400 4800 5200 5600 6000 6400 6800

Co

un

t

Chip Sorting

PM Silicon vs. SPICE Correlation

BUFD4-7T35P140-HVT (Silicon_0.9V_25C)

BUFD4-7T35P140-SVT (Silicon_0.9V_25C)

BUFD4-7T35P140-HVT (SS_0.9V_25C)

BUFD4-7T35P140-HVT (TT_0.9V_25C)

BUFD4-7T35P140-HVT (FF_0.9V_25C)

BUFD4-7T35P140-SVT (SS_0.9V_25C)

BUFD4-7T35P140-SVT (TT_0.9V_25C)

BUFD4-7T35P140-SVT (FF_0.9V_25C)

7,000

9,200

11,600

6,800

8,800

11,000

6500

7000

7500

8000

8500

9000

9500

10000

10500

11000

11500

12000

0 400 800 1200 1600 2000 2400 2800 3200 3600 4000 4400 4800 5200 5600 6000 6400 6800

Co

un

t

Chip Sorting

PM Silicon vs. SPICE Correlation

BUFD4-7T35P140-HVT (Silicon_0.9V_25C)

BUFD4-7T35P140-SVT (Silicon_0.9V_25C)

BUFD4-7T35P140-HVT (SS_0.9V_25C)

BUFD4-7T35P140-HVT (TT_0.9V_25C)

BUFD4-7T35P140-HVT (FF_0.9V_25C)

BUFD4-7T35P140-SVT (SS_0.9V_25C)

BUFD4-7T35P140-SVT (TT_0.9V_25C)

BUFD4-7T35P140-SVT (FF_0.9V_25C)

Slow PM

Fast PM

P7

Page 8: Lewis Chu,Marketing Director,GUC

GUC Property – Security CCopyright © 2016 GUC Uncompromising Performance

• May 9, 2016

VMK: Variation Measurement Kit

• PMs inserted on regular grid

• PMs represent all used VTs, channel

lengths, gates types

• Interface protocols: APB, JTAG, I2C

• Automatic test pattern generation

• Spice simulated in all corners for

reference

• Statistic is collected through all wafers

production

P8

Page 9: Lewis Chu,Marketing Director,GUC

GUC Property – Security CCopyright © 2016 GUC Uncompromising Performance

• May 9, 2016

Soft Error Rate and Solution

• Soft errors are induced by external radiation (e.g. alpha

particles)

• Solutions

– Design: filp-flop

– Memory: ECC

– Package: low-alpha materials

P9

Page 10: Lewis Chu,Marketing Director,GUC

GUC Property – Security CCopyright © 2016 GUC Uncompromising Performance

• May 9, 2016

DFM: Extended Physical Rules Check

P10

Page 11: Lewis Chu,Marketing Director,GUC

GUC Property – Security CCopyright © 2016 GUC Uncompromising Performance

• May 9, 2016

DFM: Tightening Electromigration Rules

• EM rating for automotive:

• Additional Poly/High-R/Metal-gate R current density rule

Figure : Poly EM requirement for regular products

Figure : Rating factor for automotive products

Figure : Rating factor for regular products

P11

Page 12: Lewis Chu,Marketing Director,GUC

GUC Property – Security CCopyright © 2016 GUC Uncompromising Performance

• May 9, 2016

Potential Defect Identification

• Production test tightening for Outliers segregation

Upper/Lower Spec Limits

Tightening Tightening

P12

Page 13: Lewis Chu,Marketing Director,GUC

GUC Property – Security CCopyright © 2016 GUC Uncompromising Performance

• May 9, 2016

Outlier Program at Production Test

• ELFR (Early Life Failure Rate)

– more than 2000 samples or production ramp-up, with HTOL method to

identify the early failure mode/rate for CIP

• Dynamic Voltage Stress (DVS) testing to screen weaker parts

– 1.2x ~ 1.8x stress voltage during testing

– Failure Analysis of failed parts to seek the process improvement or spec

refining

• Statistical Yield/Bin Limit (SBL/SYL) in testing

– to separate the maverick parts for risk assessment

– Maverick mechanism (yield limit by step/machine) on assembly process

• Wafer selection rule with specific WAT risk parameters

– to separate marginal/risk wafer for assessment/tightened screening

P13

Page 14: Lewis Chu,Marketing Director,GUC

GUC Property – Security CCopyright © 2016 GUC Uncompromising Performance

• May 9, 2016

Summary

• Reliability is handled throughout all stages

– DFT, physical implementation and production

• GUC flow is based on volume production experience -

allows minimal defects and very low FIT

– Silicon-proven comprehensive DFT and testing methodology

– Overcome challenging of 28nm/16nm DFM rules

– Proprietary Solution for Process Variation Monitoring

– Intelligent Mechanisms for Testing: ELFR, DVS testing, SBL/SYL

and Maverick mechanism

P14

Page 15: Lewis Chu,Marketing Director,GUC

GUC Property – Security CCopyright © 2016 GUC Uncompromising Performance

• May 9, 2016

Thank you !

www.guc-asic.com

P15