Upload
others
View
14
Download
0
Embed Size (px)
Citation preview
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
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
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
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®
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™
Sematech 854pattern wafer
Sematech 854pattern wafer
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
ount IC1000™IC1000™
Politex®Politex®
EcoVision™EcoVision™
Sematech 854pattern wafer
Sematech 854pattern wafer
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
ount IC1000™IC1000™
Politex®Politex®
EcoVision™EcoVision™
Sematech 854pattern wafer
Sematech 854pattern wafer
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
ount IC1000™IC1000™
Politex®Politex®
EcoVision™EcoVision™
Sematech 854pattern wafer
Sematech 854pattern wafer
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®
Mirra® Tool – Platen 2Platen Speed = 33 rpmSlurry Flow = 80 ml/min
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
Backflow No backflow
Mixing
Backmixing No Backmixing
Process Condition
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
Backflow No backflow
Mixing
Backmixing No Backmixing
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
Mirra® Tool – Platen 2Platen Speed = 33 rpmSlurry Flow = 80 ml/min
IC1010TM on Suba®
Mirra® Tool – Platen 2Platen Speed = 33 rpmSlurry Flow = 80 ml/min
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
0.060” Pitch0.060” Pitch
0.120” Pitch0.120” 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