Feature-level Compensation & Control CMP September 15, 2005 A UC Discovery Project
Feature-level Compensation & Control
CMPSeptember 15, 2005
A UC Discovery Project
FLCC
Chemical Mechanical Planarization - Faculty Team
Mechanical Phenomena
Chemical Phenomena
Interfacial and Colloid
Phenomena
Including ‘slurryless’ planarization - E-CMP
Jan B. TalbotChemical EngineeringUCSB
David A. DornfeldMechanical EngineeringUCB
Fiona M. DoyleMaterials Science and EngineeringUCB
FLCC
Chemical Mechanical Planarization - Student Team
Mechanical Phenomena
Chemical Phenomena
Interfacial and Colloid
Phenomena
Sunghoon Lee ME-UCB
Alex DeFeo ME-UCB
Ling WangMSE UCB
Jihong Choi ME-UCB
Robin IhnfeldtChem Eng UCSB
9/15/05FLCC - CMP
4
FLCC
CMP ResearchDescription: The major objective of this work continues to be to
establish an effective linkage between capable process models for CMP and its consumables to be applied to process recipe generation and process optimization and linked to device design and other critical processes surrounding CMP. Specific issues include dishing, erosion and overpolishing in metal polishing, which have an impact on circuit performance — all pattern dependent effects at the chip level —wetability effects in polishing, and novel consumable design (pads and abrasives) for optimized performance. We develop integrated feature-level process models which drive process optimization to minimize feature, chip and wafer-level defects.
Goals: The final goals remain reliable, verifiable process control in the face of decreasing feature sizes, more complex patterns and morechallenging materials, including heterogeneous structures and process models linked to CAD tools for realizing “CMP compatible chip design.”
9/15/05FLCC - CMP
5
FLCC
Year 2 Highlights• Developed capability in integrated model for determining and
assessing pattern sensitivity.• The effects of various common additives on zeta potential for
alumina slurries used in copper CMP were determined.• The influence of common slurry additives on the colloidal behavior of
alumina suspensions used for copper CMP were characterized.• Design rules for SMART pad design and fabrication are developed
and simulation verification of pad slurry flow characteristics done with initial performance validation tests run.
• Experiments to provide crucial data for prediction of the galvanic damage sometimes seen during CMP were conducted including measurement of relative reactivity of copper and barrier materials in different slurries.
9/15/05FLCC - CMP
6
FLCC
Year 3 Plan• Mechanisms for coupling of chemical and mechanical phenomena in CMP (M22 YII.7)Use kinetic data for weight changes during passivation by peroxide to develop dynamic model for
response after abrasion of surface layers and creation of new copper surface.• Wetting and adhesion studies on two phase or multiphase surfaces (M7 [carried
forward, with modifications, from year 1] YII.8)Study the effects of wetting and adhesional behavior of metals, low-k dielectrics and other phases on
their polishing behavior, using isolated materials and for standard feature set from the cooperative photomask activity. Explore modification of the wetting and adhesional behavior through judicious selection of pad material, and optimized use of surfactants and other solvents.
• Further develop basic understanding of agglomeration/dispersion effects (M23 YII.9)Relate colloidal chemistry to surface charge and particle size distribution changes.• Develop SMART prototyping methodology (M24 YII.10)Determine best manufacturing processes for prototyping pads for use in validation testing based on
common photomask design.• Integrate SMART pad design criteria into comprehensive model (M25 YII.11)Develop capability in integrated model for determining process-based or device design-based criteria for
SMART-pad and other commercial surface and property specifications, specially for assessing pattern sensitivity.
• Basic material removal model development (Milestone Added, YII.12)Continue development of process model with attention to low down force applications/non-Prestonian
material removal as well as subsurface damage effects.
FLCC
Bulk Cu CMP Barrier polishing W CMP Oxide CMP Poly-Si CMP
Physical models of material removal mechanism in abrasive scale
Chemical reactions
Bulk Cu slurry Barrier slurry W slurry Oxide slurry Poly-Si slurry
Mechanical material removal mechanism in abrasive scale
Abrasive type, size and concentration
[oxidizer], [complexing agent], [corrosion inhibitor],
pH …
Pad asperity density/shape
Pad mechanical propertiesin abrasive scale
Pad properties in die scale
Slurry supply/ flow patternin wafer scale
Wafer scale pressure NU Models of WIWNU
Models ofWIDNU
Topography
Wafer scale velocity profile
Wafer bending with zone pressures
Better control of WIWNU
Reducing ‘Fang’
Small dishing & erosion
Ultra low-k integration
Smaller WIDNU
Reducing slurry usageUniform pad performance
thru it’s lifetimeLonger pad life time
Reducing scratch defects
Better planarization efficiency
E-CMPPad groove
Pad design
Fabrication
Test
Fabrication technique
Slurry supply/ flow pattern in die scale
Cu CMP
modeldesign goalPad development
PatternMIT model
Dornfeld modelDoyle
An overview of CMP research in FLCC
Talbot
9/15/05FLCC - CMP
8
FLCC
Today’s Presentation- see the posters for details -
SMART Pad • Model-based pad design• Initial experimental results
Corrosion studies/wettability• galvanic corrosion of Ta-Cu couple• Wetting studies on multi-phase surfaces
Comprehensive model• Linkage with model beyond end-point (dishing/erosion)• Expansion to copper CMP model• Pad development for low down force CMP / E-CMP• Study on non-uniform break thru in a die scale in copper Ecmp
9/15/05FLCC - CMP
9
FLCC
SMART Pad - Pad Characterization (1)
• Ra = 12.5µm
• Rz = 96.7µm
• Pore diameter : 30~50 µm
• Peak to Peak : 200~300µm
100µm
45µm
-45µm 100µm 300µm 500µm
(SEM, x150)
200~300µm
(White light Interferometer, x200)
9/15/05FLCC - CMP
10
FLCC
Pad Characterization (2)
Asperity: Real contact area10~50 µmPores
40~60µm
Simplified Pad Model
Peak to Peak200~300 µm
1. Reaction Region (10~15 µm)
2. Transition Region
3. Reservoir Region
Reaction region
Reservoir region
Transition region
9/15/05FLCC - CMP
11
FLCC
Defects from Conventional Pads
wafer
Pressure
Position
Pad asperity
Nominal pressure
Avg. contact pressure
Cu-CMP defects
wafer
ErosionDishingFang
wafer
50 µm
Large asperity
wafer
ILD
Rounding
10 µmSmall asperity
wafer
ILD
Over polishing
ILD-CMP defects Stress concentration
9/15/05FLCC - CMP
12
FLCC
Design Rules for a New Pad
• Compatible features to abrasive • Constant re-generation of nano
scale surface roughness
• Constant contact area
(width:10-50um)• The ratio of real contact area
(10-15%)
• Conditioning-less CMP
• High slurry efficiency
• Stacked layer
(Hard/soft)
• Slurry channel
Nano scaleMicro scaleMacro scale
Design rules for a pad
Soft Layer(i.e. low stiffness)
Hard Layer(i.e. high stiffness)
Channel Nano scale features
50-70µm
50-200µm
9/15/05FLCC - CMP
13
FLCC
Fabrication Process
1. Master 2. PDMS Casting 3. PDMS Mold 4. Hard LayerCasting
5. Soft LayerCasting
6. Demolding
New pad
9/15/05FLCC - CMP
14
FLCC
Simulation of Slurry FlowType B – With slurry guidanceType A – Without slurry guidance
• Area : 4.294^-10 m2
• Flow rate : 3.24^-10 kg/sec• Area : 4.3^-10 m2
• Flow rate : 3.93^-11 kg/sec8 times more flow rate
On contact area
9/15/05FLCC - CMP
15
FLCC
Results
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1000 1100 1200 1300 1400 1500
New In 10min In 20min-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1000 1100 1200 1300 1400 1500
New in 10mins in 40mins
Rel
ativ
e st
ep h
eigh
t (µm
)
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1000 1100 1200 1300 1400 1500
New in 3mins in 7mins in 10mins
wafer
SiO2
0.77µm
1.7µm
IC1000/SUBA400 (overpolishing:2500Å)
Type A (squares) (Not planrized) Type B (V shape) (overpolishing:800Å)
ILD pattern
9/15/05FLCC - CMP
16
FLCC
Corrosion - the Problems
Copper tends to corrode in solutions open to air, barrier materials such as Ta have similar corrosion problems
electrolyte
2H+ 2H2OCu2+
2e-
Adsorbed O2
Cu = Cu2+ +2e-(acidic)
O2 +4H+ + 4e- =2H2O
i corr, CuEcorr, Cu
log I
(-) p
oten
tial
(+)
Ecu/Cu2+
EO2/H2O
i0O2/H2O
i0Cu/Cu2+
Polarization curves for corrosion reactions, describing the change in reaction rate as each half cell reaction is polarized away from equilibrium state; both the anodic dissolution of metal and cathodic reduction of an oxidant are polarized to a common potential, the corrosion potential, Ecorr
Cu corrosion in areated solution. The nature of copper species depends on the solution chemistry. Continuous corrosion requires electrical and electrolytic conduction paths and the presence of a potential difference
9/15/05FLCC - CMP
17
FLCC
log I
(-) p
oten
tial
(+)
Ecorr,
N
Ecorr,
Micorr, M
icorr, N
i’ corr, Ni’ corr, N
log I
(-) p
oten
tial
(+)
Ecorr,
N
Ecorr,
Micorr, M
icorr, N
i’ corr, Ni corr, N
i’ corr, Ni’ corr, M
Egalvanic
log I
(-) p
oten
tial
(+)
Ecorr, N
Ecorr, MM= Mm+ + me-
N= Nn+ + ne-
Cathodic reaction on M
Cathodic reaction on N
icorr, M
icorr, N
M
N
M N
Corrosion of more noble metal, M, is suppressed due to cathodic polarization
Corrosion of more active metal, N, accelerates due to anodic polarization
A new common potential, called mixed potential, is reached at the steady state of corrosion
Galvanic effect: accelerates corrosion of the more active metal, N, because of anodic polarization and suppresses corrosion of the more noble metal, M, because of cathodic polarization
Solution chemistry might change the polarization behavior of each material; thereby affecting the preferential galvanic corrosion of the corroding material
Increasing the area of the more noble material accelerates corrosion of N
Galvanic effects lead to more severe corrosion of the more active metal. Damage sensitive to solution chemistry and area ratio effects
9/15/05FLCC - CMP
18
FLCC
The Problems
ANODE CATHODE
Bulk electrolyte
0.1 um1 um
5 um
0.3
0.2
0.1
00 0.5 1.0
Radius, r (cm)Lo
cal c
urre
nt
dens
ity
0.1 um-thick electrolyte
1 um
5 um
Bulk electrolyte
ANODE CATHODE
Radius, r (cm)0 0.5 1.0
Pote
ntial
, E (V
)
1.0
0.5
0.0
Anode Cathode
Anode Cathode
Galvanic corrosion in bulk solution
Galvanic corrosion in thin-layer of solution
expected structures
The characteristic pronounced corrosion at the phase boundary in galvanic corrosion is due to the steep potential gradient and high current density at the boundary between two phases
θ?θ
Hydrophilic surface
θ
θ: contact angle for characterizing the wetting of a surface by a specific solution; zero or small contact angles = good wetting
Simulated potential and current distribution, E. McCafferty, JES 124 (12), 1977
The accessibility and distribution of electrolyte (ions and water) on surfaces may affect galvanic corrosion. The wettability of materials may affect the distribution of the solution on surfaces
Hydrophobic surface
solution
9/15/05FLCC - CMP
19
FLCC
Experimental Study
Cu TaPt
Cu wires for electrode connection
Sample and sample carrier
Slurry film Polishing padRotating platen
Slurry deliverypressurePotentiostat
samp
le2
Ref.
electr
ode
samp
le1
Experimental setup and samples
Galvanic currents and potentials will be measured with or without polishing, to investigate the effects of surface modification by chemicals and mechanical force on the polarity and galvanic current; area ratio of Cu:Ta is to be varied using different sample assemblies with different relative areas
sample assembly capable electrochemical control and measurement
Polisher
9/15/05FLCC - CMP
20
FLCC
Results of electrochemical measurements
0 200 400 600 800 1000-2.0x10-6
0.0
2.0x10-6
4.0x10-6
6.0x10-6
8.0x10-6
1.0x10-5
1.2x10-5
1.4x10-5
1.6x10-5
H2O2
Gal
vani
c C
urre
nt,A
Time,s
Ta:Cu=2.6:6.0 (cm2) Ta:Cu=3.9:6.0 (cm2) Ta:Cu=3.9:4.5 (cm2)
in pH=9 carbonate buffer, 0.01M glycine and 0.3%H2O2
0 200 400 600 800 1000-0.300
-0.250
-0.200
-0.150
-0.100
-0.050
0.000
0.050
0.100
0.150
0.200
Pote
ntia
l, V
vs S
CE
Time, s
Ta:Cu=3.9:6.0 (cm2) Ta:Cu=2.6:6.0 (cm2) Ta:Cu=3.9:4.5 (cm2)
Area ratio effect, without polishing
DI H2OpH9
buffer glycine
H2O2
9/15/05FLCC - CMP
21
FLCC
Results of electrochemical measurement
Effects of [H2O2], pH=9, without polishing
0 300 600 900 1200
0.0
5.0x10-6
1.0x10-5
1.5x10-5
2.0x10-5
Gal
vani
c C
urre
nt, A
Time,s
0.3% 0.1% 0.6% 1.0%
0 300 600 900 1200-0.300
-0.200
-0.100
0.000
0.100
0.200H2O2, 30 wt%
stock glycine sltn
carbonate buffer
Mix
ed P
oten
tial,
V vs
SC
E
0.3% 0.1% 0.6% 1.0%
in pH=9 carbonate buffer, 0.01M glycine and varying [H2O2]
9/15/05FLCC - CMP
22
FLCC
Effects of polishing on galvanic currents and potentials
Masako Kodera, etc., J. Electrochem. Soc., 152 (6) G506-G510 (2005)
9/15/05FLCC - CMP
23
FLCC
Contact angle measurement
Sessile drop method: Angle between the baseline of the drop and the tangent at the drop boundary is measured.
Kruss Contact Angle Measuring System (Goniometry approach)The basic elements of a goniometer include a light source, sample stage, lens and image capture. Based on the image of the liquid drop, contact angle can be assessed directly by measuring the angle formed between the solid and the tangent to the drop surface
The contact angles as function of solution chemistry on Ta/TaN, Cu and dielectric material (TEOS or SiO2) will be explored
9/15/05FLCC - CMP
24
FLCC
Comprehensive CMP Model
- Statistical representation of CMP pad surface
- Interaction between wafer topography and pad asperities
- Hertzian contact model for elastic pad deformation
- Linkage with conventional chip-scale pattern density model
RHDyx
dPHDyxEVCyxMRR
wp
yxz
zppad
2/34/1
),( 4/72/3*
),(
)(),,(),(
κρ
δδδε∫ ×⋅=
pattern density effect
pad surface conditionprocess parameters
thin film propertiespad material properties
abrasive size
chemical reactions
wafer topography
9/15/05FLCC - CMP
25
FLCC
Effect of Pad Asperity Height Distribution
smooth padrough pad
σ = 3µm, α = 3µm σ = 6µm, α = 3µm
Time (x10sec)
Oxi
de th
ickn
ess
(µm
)
Time (x10sec)
Oxi
de th
ickn
ess
(µm
)Higher planarization efficiency, Lower removal rate
Lower planarization efficiency, Higher removal rate
9/15/05FLCC - CMP
26
FLCC
Effect of Nominal Down Pressure
low pressure high pressure
9/15/05FLCC - CMP
27
FLCC
Oxide Test Pattern Wafer for Model Calibration and Verification
100%50%
50%
20%
10%
10% 20%
0%
0%
0%
8 mm
8 mm
Line width : 20 µm
4 inch wafer
(11 dies across the diameter)
CVD nitride
Pattern nitride (mask #1)
CVD OxideSiO2
Etch oxide (mask #2)
SiSi3N4
20 nm
2.5 µm
0.5 µm
9/15/05FLCC - CMP
28
FLCC
Model vs. Experiment (Oxide CMP)
highest point in the die (measured)
lowest point in the die (measured)
lowest point in the die (model)
highest point in the die (model)
t=0min t=2min t=4min t=8min
Experiment :
Model :
9/15/05FLCC - CMP
29
FLCC
Comparison of Final Oxide Thickness Variation Over the Test Die
experimentmodel prediction
RMS error = 30.8 nm(100 measurement sites)