Yield & Process Improvement by Data Analysis along the ... · Keywords are Fault Detection and...
Transcript of Yield & Process Improvement by Data Analysis along the ... · Keywords are Fault Detection and...
Roberto RappDirector Manufacturing Process Integration
Semiconductor & MEMS Plant Reutlingen
Robert Bosch GmbH
Yield & Process Improvement by Data Analysis along the Supply Chain
SEMICON EUROPA MunichNov. 14th, 2018
Automotive Electronics | Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
2
SEMICON EUROPA Munich, 11/14/18Abstract
Along the semiconductor supply chain an increasing amount of production data is generated and waiting to be used for continuous process-, yield- and quality-improvements. To classic Statistical Process Control (SPC) and data analysis, done by engineers, a variety of new applications were added over the years. Keywords are Fault Detection and Classification (FDC), Automatic Defect Classification (ADC), Advanced Process Control (APC), data lakes, data mining - just to name a few.
Looking at the supply chain from wafer start, up to electrical wafer test (EWS) traceability is a given. Other than ASICs many MEMS elements can’t be traced back to, like wafer number or x/y position. So it is important to make use of data from different steps in production.
In this presentation examples for automatic analysis of inline defectivity and EWS are shown. Furthermore the successful implementation of traceability, from final product test of MEMS sensors to inline production data is discussed.
Automotive Electronics | Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
3
SEMICON EUROPA Munich, 11/14/18Agenda1. Introduction – A Sensor’s Perspective
2. Data Analysis along the Supply Chain
• Use of Defectivity Data (single process)
• Outlier Detection in Electrical Wafer Sort and Final Test
• Combination of Inline Defectiviy Data with EWS
• ASIC to MEMS Mapping
• A brief outlook
SEMICON EUROPA Munich, 11/14/18
Automotive Electronics Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
4
Megatrends of the Future
Electrified Automated Connected and IoT
Power Semiconductor
System-ICs
System-ICs IP
Automotive Sensors CE-Sensors
Micro-optical Systems
SEMICON EUROPA Munich, 11/14/18
Automotive Electronics | Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
5
MEMS Main Automotive ApplicationsAcceleration Sensors GyroscopesPressure SensorsPressure Sensors Acceleration Sensors Gyroscopes
Engine Management
Side Airbag Airbag
Vehicle Dynamics Control
Airbag
Vehicle Dynamics Control
Standard equipment
SEMICON EUROPA Munich, 11/14/18
Automotive Electronics | Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
6
Sensors for automated driving – a MEMS perspective10
Sen
sors 1 Mass flow sensor
1 Pressure sensor [Barometric air pressure]2 Pressure sensors [Manifold air pressure, oil]1 High pressure sensor [Common Rail]1 Pressure sensor [Tank pressure]1 Pressure sensor [Start/stop function]2 Acceleration sensors [Active engine mounting]1 Pressure sensor [Diesel particulate filter]
Engine Management e.g. Diesel
27 S
enso
rs 2 High-g acceleration sensors [Airbag]1 Angular rate sensor, 1 Low-g acceleration sensor [Roll-over sensing], 1 Acceleration sensor (Structure-borne sound sensor) [Airbag] 4 Acceleration sensors, 2 Pressure sensors [Peripheral airbag sensors]2 Pressure sensors [Pedestrian safety]1 Angular rate sensor, 1 Low-g acceleration sensor, 1 High pressure sensor [ESP (incl. ACC)]1 Angular rate sensor [Active steering] 1 Acceleration sensor [eCall]4 Pressure sensors, 4 acceleration sensors [TPMS]1 Pressure sensor [Occupant detection]
17 S
enso
rs 2 Pressure sensors [Automatic transmission] 5 Acceleration sensors [Active suspension] 1 Pressure sensor, 1 Humidity sensor, 2 Gas sensors [Air conditioning, air quality]1 Angular rate sensor, 1 Acceleration sensor [Navigation] 3 Microphones [telephone]1 Bolometer Array [Night vision] 1 Acceleration sensor [Car alarm](Seldom: 16 Pressure sensors (up to 8 pressure sensors per seat)
Comfort
Safety
… high-performance IMU… µMirror for head lamps... µMirror for Lidar… human-machine interface… driver monitoring… connectivity
HAD
Automotive Electronics | Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
7
SEMICON EUROPA Munich, 11/14/18Agenda1. Introduction – A Sensor’s Perspective
2. Data Analysis along the Supply Chain
• Use of Defectivity Data (single process)
• Outlier Detection in Electrical Wafer Sort and Final Test
• Combination of Inline Defectiviy Data with EWS
• ASIC to MEMS Mapping
• A brief outlook
SEMICON EUROPA Munich, 11/14/18
Automotive Electronics | Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
8
Value stream for semiconductor plant Reutlingen
Wafer-Fab
MSBVM
MEMSFab
MSBFinaltest
1 x tägl. 7 Tagewoche
Rohwafer
IEingang
Die-Bank
IEingang
IEingang
6 x tägl. 7 Tagewoche
LOP
Assembly
Chips
LOP
RBACFinaltest
MSBAssembly
Eingang
I
RBHHAssembly
Foundry
Eingang
IWafer-Fab
MEMSFab
WaferTest
Assem-bly
Final-Test
HatvanAssem-
bly
ChinaFinal-Test
LOG LOG ENGENGENG
ExternalAssembly
ExternalFoundry
RB Reutlingen
RB worldwide
Supplier worldwide
DeliveryCustomer
Automotive Electronics | Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
9
SEMICON EUROPA Munich, 11/14/18Use of data in Wafer FabFlow of Defect-Inspektion with SPACE and eCAP
Automotive Electronics | Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
10
SEMICON EUROPA Munich, 11/14/18Use of data in Wafer FabAutomated Inline Reaction on defect types by Space
Process StepInspection + review+ signature analysis SPC + eCAP
Process & Equipment
Tool 1:
Tool 2:
Classification by- Inspection tool RTB- SPR- ADC/operator
Tool 1 Chamber1
Tool 1 Chamber2
Tool 2 Chamber1
Tool 2 Chamber2
SPC Concept
eCap (electronic Control action plan):- Immediate reaction on tool and chamber level- disposition of affected material- Maintenance and requalification
SPAC
E - G
atew
ay
SPAC
E - F
ram
ewor
k
Regelkarte
Specifi-cation
Database
SPC-Violation
SPC-Violation
SPC-Violation
Request
Reply
Event
Event
Event
Regelkarte
Regelkarte
SPACE eCAP
Chamber 1
Chamber 2
Chamber 1
Chamber 2
Input: DD Data
Output: OCAP on chamber level
SEMICON EUROPA Munich, 11/14/18
Automotive Electronics | Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
11
Use of data for quality improvement in EWS and FTDetection of outliers and defects
Outlier within specification Good dies in bad neighbourhood
© Taino Palermo
© Pintail Technologies
© Olaf Engel, Wuppertal; www.muehlstein-online.de
SEMICON EUROPA Munich, 11/14/18
Automotive Electronics | Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
12
Outlier removal as a method of quality improvement General Considerations
Schematic of Outliers
Example: parametric test result wafer test
Part Average Testing (PAT)Outliers: parts with abnormal characteristic with respect to a main distribution
Outliers can potentially fail in field even though they were tested in spec
AEC-Q001 provides description for statistical calculation of PAT limits and outlier removal
Lot to lot variations require introduction of dynamic PAT limits: yield vs efficiency
Part Average Test Limits and Outliers
Outliers Main Distribution Spec Limits PAT Limits
upper spec limit
lower spec limit
outlier
Use of data in in EWS & FTStatistical BIN Limits & Part Average Test
SEMICON EUROPA Munich, 11/14/18
Automotive Electronics | Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
13
Testing
4s limit based on last 6 months of production
Outlier
Statistical BIN Limits
OCAP for shop floor
Lot dispatching, further measures
Analysis in expert team
Testing
Part Average Test / Parametric BOPP
Lot no. Fail
rate
OutlierPart no. Te
st re
sult
Dynamic PAT limits based on test results
Scrap bad dies, inform engineering if fail rate > threshold
Para
met
er a
Parameter b
OutlierBOPP Correlations
SEMICON EUROPA Munich, 11/14/18
Automotive Electronics | Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
14
Use of data in EWS: Bosch Post Processing, geometrical Application
Scratch detectionElectrical failures in grayOutlier classification:• Black: expansion of cluster• Yellow: position with reduced yield for most wafers of lot• Pink: Scratch detection• Red: electrical outliers • Orange: electrical outliers (compared to neighborhood)• Blue: small electrical outliers in several tests
Removal of good dies in bad environment
Expansion of cluster
Position with reduced yield for most wafers in lot
Statistical BIN Limits & Part Average TestExamples from Final test and Customer data
Part Average TestStatistical BIN Limits
SEMICON EUROPA Munich, 11/14/18
Automotive Electronics | Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
15
Outliers
Damaged bond wires
Root cause:damaged lead framepusher
Failure analysis revealed damaged bond wiresRoot cause found in Assembly
BIN trend increase in fail frequency detected
Failure analysis showed a crack in the MEMS elementEffectiveness of the selected limits is proven
Particle in MEMS found
Evaluation of score limits in addition to outlier limits- Acceleration sensor exceeded score limits in multiple tests- Accumulation of scores leads to scrapping of the tested device
Crack
SEMICON EUROPA Munich, 11/14/18
Automotive Electronics | Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
16
Wafer-Fab
MSBVM
MEMSFab
MSBFinaltest
1 x tägl. 7 Tagewoche
Rohwafer
IEingang
Die-Bank
IEingang
IEingang
6 x tägl. 7 Tagewoche
LOP
Assembly
Chips
LOP
RBACFinaltest
MSBAssembly
Eingang
I
RBHHAssembly
Foundry
Eingang
IWafer-Fab
MEMSFab VM
Assem-bly
Final-Test
HatvanAssem-
bly
ChinaFinal-Test
LOG LOG ENGENGENG
ASIC to MEMS Mapping (A2MM)
ASIC-2-MEMS-MappingExternal
Assembly
ExternalFoundry
CustomerDelivery
Saw
ing
SEMICON EUROPA Munich, 11/14/18
Automotive Electronics | Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
17
A2MM: Data Analytics in worldwide value streamDelivery
Customer
SEMICON EUROPA Munich, 11/14/18
Automotive Electronics | Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
18
1. Data for Volume and Traceability2. Data for Product- and Process-Quality3. Data Generation 4. Data Security
Data within the supply chain – “Outlook”
Semicon 2018 Munich, 11/14/18
Automotive Electronics | Roberto Rapp | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
19
Abbreviations ADC: Automatic Defect Classification ASIC: Application-Specific Integrated Circuit APC: Advanced Process Control BOPP: Bosch Post Processing (Software) ECAP: Electronic Control Action Plan FDC: Fault Detection and Classification HAD: Highly automated driving OCAP: Out of Control Action Plan MES: Manufacturing Execution System MEMS: Micro-Electro-Mechanical System PAT: Part Average Test RTC: Real Time Classification SPC: Statistical Process Control SPR: Spatial Pattern Recognition
1. SC Management Meeting
Dr. Dirk Hoheisel | 10/31/2018© Robert Bosch GmbH 2017. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
20
Bosch Semiconductors.We make the world safer, smarter and accessible.
Thank you
For your attention