Pad Characterization Update

Caprice GrayNov. 9, 2006

Cabot MicroelectronicsAurora, IL

Outline

• Review where we were last time• Non-DELIF imaging• DELIF Contact detection feasibility and

preliminary results• Thesis Outline• DELIF Modeling• Experimental Outline• Questions/Issue that should be resolved before

proceeding forward

Last Meeting Accomplishments

• Set up beam splitter aligner

• Replace poor quality optical filters with high quality filters

• Determined the source of ratio drift by modeling intensity sources for each camera

Source of Ratio Drift

RI

I

I I

I IA t Bca l

pad

cam dn

cam dn

1 1

2 2

I I I I I I

I I I I Icam pad dn L ca l

padca lL

cam pad dn L ca l

1 1 1 1 1 1

2 2 2 2 2

, , , , ,

, , , ,

I I I I

I I IL cam L cam

pad cam ca l

, ,

, ,

. .

.1 1 2 2

1 1 2

0 0 2 0 0 1

0 0 5 0

RI

I

I I I

I IA t Bca l

pad

pad

cam dn ca lL

cam dn

0 9 3

0 9 91 1 1

2 2

.

., ,

,

Calibration for all previous data sets that show ratio drift.

All Components the cameras actually measure

Primary Source of ratio drift

Experimentally Determined Constants

New Ratio Calculation

Determined by drift

Imaging Benchmarks: SEM

• Imaged in 3 regions of a used pad

1. Edge of wafer track (most glazed region)

2. Outside Wafer track (no glazing)

3. Center of wafer track (less glazing)

Imaging Benchmarks: SEM

Outside of wafer track

Edge of wafer track

Center of wafer track

In a grooveCenter of wafer track

Imaging Benchmarks: SEM

SEM Image

Highest Quality DELIF

– depth

Imaging Benchmarks: Confocal Microscopy

Similar to DELIF Resolution(scale bar = 160 m)

Zoomed image of a pore, a cross section near the pad surface

(scale bar = 40 m)

Imaging Benchmarks: Confocal Microscopy

• 24 images are taken every 2 mm

• Video start at the deepest layer and progresses to the surface

Rohm and Haas Pad-Wafer Contact Study

• Confocal Reflectance Interference Contrast Microscopy

• Images are taken through sapphire.– No slurry– Static

• Imaging region < 1 mm2

– Asperity Contact ~ 50 m2

• Pads studied:– IC1000– VP3000– Politex– Experimental

• Tested 0-6 psi, contact 0-6%, experimental contact ~ 20%

DELIF for Contact

• 360,000+ pixels 2% ~ 7200 pixels • 50 m2 ~ 7-8 pixels (6.7 m2/pixel)

– Focus must be really good– We are at the resolution limit for our system

• At this resolution, we are seeing contact region intensity smoothing

Data for static DELIF on CMC D100

CMC D100 Imaging Issues

• Experiment: 50 consecutive static images

• Image drift correction was nearly 100% of signal

• Focus is difficult because Ra>DOF– Must carefully focus at

tops of pixels– Out of focus light bleeds

into pixels that are in focus

CMC D100 Imaging Issues

D100 Emission

FX9 Emission

Thesis Outline

• Introduction– Why are we looking for pad-wafer contact in CMP– Literature review of experimental pad-wafer contact

measurements– Project Goal = detect in-situ pad-wafer contact

• Methods– What is DELIF, and why use it– System model and calibration techniques– Hardware description– System limitations– Image Acquisition and Processing

In-Situ Detection of Pad/Wafer Contact with DELIF during CMP

Thesis Outline

• Experimental Outline– Description of variables and responses– Post acquisition image interpretation (histogram analysis, etc.)

• Results and Discussion– Comparison on contact on hard vs. soft pads– Static measurements vs. dynamic measurements– Pressure variation effect on contact %– Relate results to previously studied pad-wafer contact

measurements• Conclusions

– Summary of Experimental results– Contribution of results to current understanding of CMP– Could this technique be expanded to study multiple polishing

pad types?

DELIF Modeling

• Goals– Examine how comparable

our actual system is to current DELIF models

– Verify our linear calibration technique (Ratio Intensity Fluid layer thickness)

• Results– slurry particles quadratic

calibration– As slurry particle

concentration 0, calibration linear

Old Model

Current System

BAtI

I

f

f 1

2

EDtCtI

I

pad

dye 2

Experimental Outline

• Finish Tweaking Optics to get repeatable static data

• Identify 2 types of polishing pads that will work optically with our system.– Similar Ra ~ 4-6 m– High and Low bulk modulus

• Proceed to DOE– Possible Variables: pressure, pad/wafer velocity, pad

type, slurry dilution, pH– Responses: Contact area %, Contact area size,

friction signature?

Open Questions

• Which 2 pads should I focus my study on and will I have a steady supply of them?

• Does slurry dilution matter?– Accurate depth measurements or minimize chatter?

• Should I continue to search for a static contact measurement benchmarking technique, or just use (the limited) data from literature?

• Which experimental variables should I concentrate on?– My variable preferences: pad modulus, pad-wafer

speed, down-force