Advances in CMP Fundamental Understanding and …...EcoVision™EcoVision™ Sematech 854 pattern wafer Sematech 854 pattern wafer 120 140 180 024 68 Scratch Defect Count 0 20 40 60

Advances in CMP Fundamental Understanding and Advances in CMP Fundamental Understanding and Applications to Consumable and Process DesignApplications to Consumable and Process Design

Gregory P. Muldowney, Carolina L. Elmufdi& A. Scott Lawing*

Introduction

Contact Area Measurement and ModelingFundamental understanding

Ex: Sub-pad effectRational consumable design

Ex: Low Stress Polishing

Process Fundamentals: BackmixingBackmixing criteriaBackmixing process response

UniformityDefectivity

Pattern transferPad & Conditioning Process Design: VisionPad™5000 Polishing Pad for STI

Contact Area Measurement and ModelingContact Area Measurement and Modeling

Muldowney & Elmufdi

PApplied

PC1 PC2 PC3

Wafer

CMP Pad

AppliedC PP >>∴

ACAWafer

WaferC AA <<

Elimination of defects is more effectively achieved by elimination of high contact pressures

Point Contact Area and PressurePoint Contact Area and Pressure

C

AppliedC A

PP =

Very large PC

damages surface

Breakthrough increases in pad-wafer contact area will reduce defects even with no change in applied downforce or coefficient of friction

Cross-Sectional AreaCross-Sectional Area

Contact Shape vs. CrossContact Shape vs. Cross--SectionSection

Simulated Contact Shape:Top Pad Only

Simulated Contact Shape:Top Pad Only

10 μm waferdisplacement10 μm waferdisplacement

Difference between contact area and cross-section is observed at all scales

Difference between contact area and cross-section is strikingly larger when a sub-pad absorbs most of the displacement

Simulated Contact Shape:Top Pad + Sub-Pad

Simulated Contact Shape:Top Pad + Sub-Pad

New materials to planarizeUnforgiving error budgetsFragile device structuresLow defect tolerance

LowLow--Stress CMP for Emerging DevicesStress CMP for Emerging Devices

Bulk SiAlternatives

Bulk SiAlternatives

TransistorExtensionsTransistorExtensions

InterconnectMaterials

InterconnectMaterials

New Memory Materials& Architectures

New Memory Materials& Architectures

Novel slurry chemistriesWafer-scale pad flatnessLow-stress planarizationLow-stress planarization!

Contact State Determines DefectivityContact State Determines Defectivity

Defects approach zero as contact pressure drops to same order ofmagnitude as applied pressure

There is no general trend of lower defects at lower COF across pad types

Scratch Defect Counts in Cu Barrier CMP at 1.5 psiAMAT Mirra® Polisher, 93/87 rpm, LK393C4 slurry

Scratch Defect Counts in Cu Barrier CMP at 1.5 psiAMAT Mirra® Polisher, 93/87 rpm, LK393C4 slurry

0

20

40

60

80

100

120

140

160

0 0.2 0.4 0.6

Wafer-Average Coefficient of Friction

Scra

tch

Def

ect C

ount TaN COFs

Cu COFs

Sematech 854pattern wafer

Sematech 854pattern waferIC FamilyIC Family

Politex®Politex®

EcoVisionTMEcoVisionTM

0

20

40

60

80

100

120

140

160

0 0.2 0.4 0.6


Scra

tch

Def

ect C

ount TaN COFs

Cu COFs



Politex®Politex®


0

20

40

60

80

100

120

140

160

0 0.2 0.4 0.6


Scra

tch

Def

ect C

ount TaN COFs

Cu COFs



Politex®Politex®


0

20

40

60

80

100

120

140

160

0 0.2 0.4 0.6


Scra

tch

Def

ect C

ount TaN COFs

Cu COFs



Politex®Politex®

EcoVisionTMEcoVisionTM020406080

100120140160180

0 2 4 6 8

Contact Area - %

Scra

tch

Def

ect C

ount

020406080

100120140160180

0 2 4 6 8

Contact Area - %

Scr

atch

Def

ect C

ount IC1000™IC1000™

Politex®Politex®

EcoVision™EcoVision™



020406080

100120140160180

0 2 4 6 8

Contact Area - %

Scra

tch

Def

ect C

ount

020406080

100120140160180

0 2 4 6 8

Contact Area - %

Scr

atch

Def

ect C


Politex®Politex®




020406080

100120140160180

0 2 4 6 8

Contact Area - %

Scra

tch

Def

ect C

ount

020406080

100120140160180

0 2 4 6 8

Contact Area - %

Scr

atch

Def

ect C


Politex®Politex®




020406080

100120140160180

0 2 4 6 8

Contact Area - %

Scra

tch

Def

ect C

ount

020406080

100120140160180

0 2 4 6 8

Contact Area - %

Scr

atch

Def

ect C


Politex®Politex®




CMP Defects vs. Contact PressureCMP Defects vs. Contact Pressure

0

20

40

60

80

100

120

140

160

180

0 200 400 600 800 1000 1200

Contact Pressure - psi

Tota

l Scr

atch

es

IC1000TMIC1000TM

Politex®Politex®

EcoVisionTMEcoVisionTM LK393C4 Barrier SlurrySematech 854 pattern wafer

Polish Pressure: 1.5 psiPlaten/Carrier Speed: 93/87

rpmSlurry Flow: 200 ml/min

LK393C4 Barrier SlurrySematech 854 pattern wafer

Polish Pressure: 1.5 psiPlaten/Carrier Speed: 93/87

rpmSlurry Flow: 200 ml/min

Point contact pressure has emerged as a key driver of scratch defectsHigh-contact pads are an ongoing R&D focus to provide low-stress polish

Process Fundamentals: Process Fundamentals: BackmixingBackmixing

Muldowney(Additional data by Lawing)

BackmixingBackmixing Effect in CMP Tracer ExperimentsEffect in CMP Tracer Experiments

Backmixed Area Resists Tracer

Tracer Media

Glass Substrate

Pad

BowWave

FluidInlet

Non-Backmixed Area Conveys Tracer

Tracer Media

Glass Substrate

Pad

BowWave

FluidInlet

Results for IC1000Pad with Circular Grooves

Backmixed region resists fresh slurry influx for many wafer rotations Non-backmixed condition conveys slurry across full wafer track

Results for IC1000Pad with Circular Grooves

Backmixing affects wafer over width w* at edge:

For wafer to be unaffected by backmixing (w* = 0):

P

W

RRw WΩΩ+

−=1 *

⎟⎟⎠

⎞⎜⎜⎝

⎛ −Ω≤Ω

W

WPW R

RR

Slurry Backmixing Criteria

R

RW

ΩW

ΩP

w*

For RW /R of 0.7-0.8, need ΩP/ΩW > 2.5 to avoid backmixing

Larger wafers (higher RW/R) make backmixing more likely and affected width w* greater

Effect of Effect of BackmixingBackmixing on Steadyon Steady--State Concentration FieldState Concentration Field

Mass FractionSpent Slurry

0 rpm 4 rpm 12 rpm

Case:Pad 33 rpmPolish Pressure 3 psiMedium ConditioningFlat 43 μm Gap 25 rpm

Backmixingconcentrates spent slurry near wafer edge

Un-Backmixed

33 rpm 61 rpmWafer Surface

Backmixed

Area ofDetail

Slurry Temperature

(°C)

0 rpm

61 rpm

Effect of Effect of BackmixingBackmixing on Steadyon Steady--State Temperature FieldState Temperature Field

4 rpm 12 rpm

33 rpmWafer Surface

Case:Pad 33 rpmPolish Pressure 3 psiMedium ConditioningFlat 43 μm Gap 25 rpm

Backmixingconcentrates heat near wafer edge

Un-Backmixed

Backmixed

Area ofDetail

Effect of Effect of BackmixingBackmixing on Wafer Edge Profileson Wafer Edge Profiles

3000

3500

4000

4500

5000

5500

6000

6500

7000

7500

8000

80 82 84 86 88 90 92 94

Distance from Wafer Center, mm

Film

Thi

ckne

ss, Å

CS=61, P=3, OSCCS=61, P=3, NOSCCS=12, P=2, NOSCCS=12, P=3, NOSCCS=5, P=3, NOSCCS=61, P=2, NOSCCS=33, P=3, NOSC

CS = Carrier Speed, rpm P = Polish Pressure, psi OSC = Oscillating CarrierNOSC = Non-Oscillating Carrier

No Backmixing

Backmixing

3000

3500

4000

4500

5000

5500

6000

6500

7000

7500

8000

80 82 84 86 88 90 92 94

Distance from Wafer Center, mm

Film

Thi

ckne

ss, Å

CS=61, P=3, OSCCS=61, P=3, NOSCCS=12, P=2, NOSCCS=12, P=3, NOSCCS=5, P=3, NOSCCS=61, P=2, NOSCCS=33, P=3, NOSC

CS = Carrier Speed, rpm P = Polish Pressure, psi OSC = Oscillating CarrierNOSC = Non-Oscillating Carrier

No Backmixing

Backmixing

Data from CDE ResMap16810-mm Edge Exclusion

No-backmixing improved edge profiles regardless of polish pressureReflects more consistent transport in absence of reverse slurry flow

Cu

Film

Thi

ckne

ss, Å

Mirra® Tool – Platen 2Platen Speed = 33 rpmSlurry Flow = 80 ml/min

IC1010TM on Suba®


IC1010TM on Suba®

Rate

3646.22

807.873

2519.656±98.515

Unif

17.1119

-5.4296

3.023114±3.3881

CFF

(75-

95) 22.0268

-44.75

-11.0861±7.2673

Desi

rabi

lity 1

0

0.920639

Pm

1.5 3

2.49919

PS

75 125

125

RRsp/Pm

0.6

1.5

1.2121

PS/CS

1.5 4

2.79676

Desirability0 1

Effect of Effect of BackmixingBackmixing on Uniformity Optimizationon Uniformity Optimization

Desirability functions indicate one way to optimize the process based on the results of the DOE. In this case targets for rate and profile (Center Fast Factor) were input and uniformity (%σ) was minimized

Note that increasing PS/CS drives CFF down (i.e. more center slow!)

Effect of Effect of BackmixingBackmixing on Uniformity Optimizationon Uniformity Optimization

More control is obtained over center-to-edge profile at high platen speed/carrier speed ratio

10001200140016001800200022002400260028003000

0 20 40 60 80 100

Wafer Radius

Polis

h R

ate

(A/m

in)

4.1 Prr3.7 Prr3.3 PrrPOR

New Processes (No Backmixing):Pm = 2.7, Pit = PmPS = 123, CS = 44

POR (Backmixing):Pm = 2.9, Pit = Pm, Prr = 3.4PS = 75, CS = 50

Area ofDetail

Slurry renewal is slow near pad center due to reverse motion of wafer, allowing debris to collect and recycle

Agglomerated debris and particles ejected from groove may inflict scratch defects

Spiral grooves renew slurry more effectively at pad center to arrest this defect mechanism

Steady-StateFraction of

Spent Slurry

Transient Fraction of

Fresh Slurry

Fresh Slurry

Spent Slurry

100 % Spent Slurry& Polish Debris

CMP Wafer

DelayedMixing Wakes

SampleParticle Tracks inPad-Wafer Gap

Polishing Particles

Unused Particles

Scratching Particle

AgglomerationRegion

Wafer Centerof Rotation

Pad Centerof Rotation

Origins of Scratch DefectsOrigins of Scratch Defects

Effect of Effect of BackmixingBackmixing on Cu on Cu DefectivityDefectivity

Tota

l Def

ects

- S

crat

ch

120

130

140

150

160

170

180

Backflow No backflow

Mixing

Backmixing No Backmixing

Process Condition

Tota

l Def

ects

- S

crat

ch

120

130

140

150

160

170

180


Mixing


Process Condition

Tota

l Def

ects

- S

crat

ch

120

130

140

150

160

170

180


Mixing


Process Condition

Tota

l Def

ects

- S

crat

ch

120

130

140

150

160

170

180


Mixing


Process Condition

Data from OrbotTM WF-720

No-backmixing improved defect counts by 25% at 3 psi pressurePad rotation conveys away spent particles and polish debris faster than wafer rotation can recapture them


IC1010TM on Suba®


IC1010TM on Suba®

⎟⎟⎠

⎞⎜⎜⎝

⎛−≤

ΩΩ 1

WP

W

RR

Pad

Circular Grooves

Wafer

FinishedWafer

Dual-AxisRotary

Polisher

Table Rotation

Wafer Rotation

Wafer Oscillation

Groove Pattern TransferGroove Pattern Transfer

Rotation/oscillation exposes point on wafer to many points on padOff-center position exposes wafer to all angles with respect to pad surface and groovesDespite motions, wafer is imprinted with rings matching groove pitch

With Carrier OscillationWith Carrier Oscillation Without Carrier OscillationWithout Carrier Oscillation

0.030” Pitch0.030” Pitch



Porous IC1000Porous IC1000™™ PadPad

Effect of Effect of BackmixingBackmixing on Pattern Transferon Pattern Transfer

PS/CS = 4PS/CS = 4 PS/CS = 2PS/CS = 2 PS/CS = 1PS/CS = 1

PS/CS = 0.67PS/CS = 0.67200-mm Wafer

View

r*

100 mm 50 mm 25 mm

21 mm 17 mm

PS/CS = 0.85PS/CS = 0.85

Pad Conditioning & Process Design:Pad Conditioning & Process Design:VisionPadVisionPad™™ 5000 for STI5000 for STI

Lawing

Advanced Pad Roadmap

Logic DRAM

>65

65,7045

57,50

32

Standard

Mfg.

‘04 ’05 ’06 ’07 ’08,’09

Proce

ss

Node

VisionPad™3000

VisionPad™3000

VisionPads:VP3200, VP3500,VP5000, EV4000

VisionPads:VP3200, VP3500,VP5000, EV4000

IC1000™ATIC1000™AT

IC1000™IC1000™

Advance

d Cas

ting (A

CT)

- Impro

ved c

onsis

tency

ACT & Te

chno

logy

Impr

ovem

ents

-Flat

ter, s

moothe

r pad

s

-Groo

ve im

prove

ments

- Proc

ess C

ontro

l

Pad Formulation Platforms

Process Technology

Softer Formulation to Reduce Defects

VisionPads for STI and Cu

New PlatformsCu, STI, ILD

New PlatformsCu, STI, ILDNext Generation

IC1000™ pads

Hsinch

u

Mfg. 200

7

VisionPad™3100

VisionPad™3100

AdvancedPads

AdvancedPads

New M

fg.

Proce

sses

New VP Formulations for ILD, Cu

00.010.020.030.040.050.060.070.08

Pad Height Pr

obab

ility

Den

sity

(μ m-1

)

High Cut Roughness

Low Cut Roughness

Conditioner Design, Natural Porosity & Pad Texture Evolution

0

0.02

0.04

0.06

0.08

0.1

0.12

Pad Height

Prob

abili

ty D

ensi

ty (μm

-1)

High Level of Porosity

Lower Level of Porosity

Zero Level of PorosityNatural porosity of pad +

Final pad surface statistics

Cutting behavior of conditioner =

Changes in porosity drive these statistics

Changes in conditioner design drive these statistics

Manipulating the size, shape and density of diamonds can change cutting characteristics and cut rate > 10x

Pad Surface Height Distribution

Pad Surface Profile

Final pad surface is the product of the inherent pad texture (porosity) and the conditioner cutting characteristic (near surface roughness)

Each pad-conditioner combination will have a unique (intrinsic) surface structureCut rate, cutting characteristics and the resulting near surfaceroughness can be driven over a large range through conditioner design

Pad Surface

Pad Bulk Material

Conditioning Held Constant

Porosity Held Constant

Conditioning for Textural Compatibility

Materials with different pore structures require conditioning tailored to their specific porosityIncompatible conditioning results in a disruption of the naturalporosity of the material

Too DeepVisionPad 5000 + 181060

(Cutting depth > porosity depth)

CompatibleVisionPad 5000 + SPD01

(Cutting depth ≤ porosity depth)

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18VP5000/SPD01

VP5000/181060

Pad Height

Pad

Hei

ght P

roba

bilit

y D

ensi

ty (µ

m-1

)

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09IC/NSPD01IC/181060

Pad Height

Pad

Hei

ght P

roba

bilit

y D

ensi

ty (µ

m-1

)

Example: IC1000, VisionPad 5000, NSPD01 & 181060

The natural porosity of IC1000 is compatible with both 181060 and SPD01Note that the negative tail of the distribution is unaffected by conditioning indicating that the natural porosity is intact in both cases

The natural porosity of VisionPad 5000 is incompatible with 181060 and compatible with SPD01

Note that the combination of VisionPad 5000 & 181060 results in a widening of the negative tail of the distribution indicating that the natural porosity of the material has been disrupted

VisionPad 5000 IC1000

0

0.2

0.4

0.6

0.8

1

VisionPad5000/SPD01 VisionPad5000/CG181060

IC1010/SPD01 IC1010/CG181060

Performance Comparison

Data from STI Polishing with Celexis™ CX97S slurry

Dishing and polish rate are significantly improved with the VisionPad 5000/SPD01 combination

Dis

hing

on

500

µm fe

atur

e

Pol

ish

Rat

e

Pad Type (Conditioner Type)

0.70

0.75

0.80

0.85

0.90

0.95

1.00

VisionPad5000/SPD01 VisionPad5000/CG181060

IC1010/SPD01 IC1010/CG181060

Pad Type (Conditioner Type)

Conclusions

Rohm and Haas Electronic Materials has a comprehensive CMP fundamental research program in placeThis research program is designed to provide rational guidance for product and process designSome highlights reviewed here include:

Fundamental aspects of pad-wafer contact mechanicsProcess interactionsNext generation pad design

Documents

Advances in CMP Fundamental Understanding and …...EcoVision™EcoVision™ Sematech 854 pattern wafer Sematech 854 pattern wafer 120 140 180 024 68 Scratch Defect Count 0 20 40 60