Implementing FDC in the Wafer Dicing

Process to Improve Product Quality

Advanced Process Control Conference XXIX 2017
Austin Airport Hilton Hotel, Austin, Texas

Henry Hsu, Joe Hung, Sophia Lin, Vincent DiFilippo

Rudolph Technologies
 Agenda
• Introduction
• Wafer dicing issues and impact
• Kerf solutions approach
• FDC as a tool for monitoring the dicing process
• Correlation of FDC and kerf metrology
• Typical Use Case
• Identifying critical parameters
• Closing the loop with FDC monitoring
• Conclusion

2
 Introduction

A tool-based FDC system can collect extensive tool sensor data

from a dicer saw, generate meaningful statistical data, and store
these in a database.

This FDC data can then be correlated by lot and wafer to measured
wafer metrology data.

Advanced statistical techniques can then be used to identify which

tool signals most influence die chipping.

These signals can then be monitored by FDC models, completing

the circle to improve product quality.

3
 Wafer Dicing Issues and Impact
Typical Process Issues…
Side Wall Cracking Backside Chipping

Impact: Yields, Increased escape rates Impact: Yield, Escape rates

Front Side Delamination Bridged Dies Layer Delamination

Impact: Yield, Increased escape rates Impact: Process issues, Yield Impacts Impact: Yield, Performance

4
 Rudolph Kerf Solutions Overview
Multiple levels of SAW Analysis and Process Control

Die Seal Ring KERF Metrology KERF Metrology

Inspection Inspection Inspection & SAW Control

DIE DIE

Inspection Area
DSR Inspection Area
Street Street

KERF Area KERF Area

All of the Kerf Metrology & …
 Defect present verification  Adv. metrology measurements  Dicer process monitoring
 Die disposition  Cut quality feedback  Adv. FDC / Metrology correlations
 Reticle-based setup  Adv. die disposition  Ideal for process development
 Same pass as 2D inspection  Process monitoring  OEE improvements
 Increased process understanding  Excursion root cause
understanding

5
 Rudolph KERF Solutions

KERF Metrology & SAW Process Control

Enabling SAW process control, yield improvements, and escape reduction !!

SAW PROCESS INSPECTION /

Scribe / Inspection METROLOGY
 Process Monitoring
Saw
 Faster Process  100% Inspection Control
improvement  KERF Metrology
 Predictive Yields  Die Classification / Binning
Dicing tool NSX320
 Escape Reduction

Closed Loop
Optimization

FDC PROCESS CONTROL DISCOVER ANALYSIS

 High speed data collection  Complete SAW Visibility
 100+ signals recorded  Metrology to FDC
 Fleet Process Control Equipment Sentinel Discover Kerf correlation
FDC Analysis Metrology  Process Improvements
 Closed Loop Correlation

6
Wafer Saw Process for Advanced Packaging
 Automated manufacturing reports
 Dynamic metrology and defect sampling
 Fleet management: tool matching and performance rating

Equipment Discover
Sentinel ® Enterprise™
Grooving Saw Post Saw Post Saw
Defect Metrology

 Stop saw mid-wafer during groove or saw  Focus on reliability defects

 Optimize recipe settings for quality and throughput  Monitor Pass/Fail and quality
 Reduce scrap and rework, detect issues early in process metrics for metrology

Rudolph provides process control at advanced nodes.

7
What is FDC (Fault Detection and Classification)?
• Detects an abnormal status of Collection
the equipment or the process
running on it.
• Monitors key parameters as
defined by fab engineering (such
as low chamber pressure or low
RF reflected power) Monitoring

• Identifies the detected failure,

such as a leak in the chamber or a
faulty RF power supply
• Knowledge-based
• Statistical
• Acts to alarm or shut the tool Detection
down prior to misprocessing
Fault

8
 Higher Data Sampling = More Sensitivity
Current: 1 sec Frequency (SECS/GEM)

Spindle Current

Same
process
High Speed Data (custom data interface)

Spindle Current

9
 FDC and Inspection Data Alignment

Kerf Chipping Width

FDC Parameters

Raw data of FDC parameters and inspection tool measurements of kerf width.
Collected across a wafer, these data were collected separately on a dicer and an inspection
tool then automatically aligned in the Discover database.

10
 Typical Use Case…

Discover Genesis

NSX330 Categorize wafers based on the metrology data

Dicer Saw

Process Multiple Inspection Metrology

Wafer runs

Analytic Flow
Process Flow
Good Bad
Wafers Wafers

FDC Split the FDC trace signals by the good/bad

FDC wafer data based on the metrology
FDC FDC

FDC Data HVM Data

Recorded

Real time model now

monitors for excursions. Identify individual or combination of signals
Alarms go notify the host, that drive chipping, then model it.
e-mail, stop the tool, etc.

Monitor for future excursions

11
Identifying Critical Parameters – Visualization

12
Identifying Critical Parameters – Analysis
Align FDC and metrology
data to
generate statistics

THEN
correlate summary data
against kerf metrology data
using multivariate methods,
such as PLS in the Genesis
analysis package.

13
 Identifying Critical Parameters - Bad Cutlines

Relates to bad cut-line

Relates to
bad cut-line

By using summarized data and our

visualization tool, we can identify signal
peaks correlated with a bad cutline.

14
 FDC Monitoring

Key Parameters
Equipment Sentinel
Dicer  CT and ST vacuum
 Spindle speed
 Blade exposure
 Water pressure
 Feed speed
 Air pressure

Key parameters can be monitored

by user-created models:
• Real time monitoring
• SPC monitors
• Golden trace
• Multivariate monitors

15
 Conclusion

TO IMPROVE DIE QUALITY…

Align FDC and kerf metrology data to allow correlation of tool

signals to kerf parameters.

Create FDC heuristic models to monitor newly identified process

parameters for future excursions.

Monitor key dicer signals with FDC to reduce kerf chipping.

16