Post on 23-Jul-2020
transcript
Etch Product Business Group
External Use
Minimizing Plasma-Induced Charging Damage during Multi-Step Etching of Dual-Damascene Trench and Via Structures with Process Optimization
M. Kutney, S. Ma, D. Buchberger, A. Zhao, G. Delgadino, D. Hoffman, and K. Horioka*Etch Product Business Group, Dielectric Etch Div.*Applied Materials, Inc., Japan
Etch Product Business Group 2
External Use
Introduction and Objectives
Plasma-induced charging effects are– Influenced by plasma stability and uniformity – Strong functions of chamber design and process conditions
The role of plasma stability can be examined using multiple charging-sensitive electrical test structures– EEPROM-based sensors – Metal-gate and poly-gate antenna structures
Identify the risk factors that contribute to charging effects in a very-high-frequency capacitively coupled (VHF CCP) dielectric etcher
Factors related to plasma uniformity and stability were studied in the VHF CCP Centura® Enabler® dielectric-etch chamber – Designed for all-in-one processing of sub-65 nm dual-
damascene structures
Etch Product Business Group 3
External Use
Plasma Charging Damage – Responsible Groups
Four different groups influence the plasma-charging-damage riskUnderstanding all of the group details is challenging
Process•profile•uniformity•selectivity•sequence
Hardware•chamber•power source•cathode•process kit
Device•scaling•integration•reliability•testing method
Layout•design rule•die stepping•alignment mark•edge exclusion
chiller
RF power supply
Throttle valveGate valveTurbo pump
wafer
ESC
Magnet liner
gas distribution plate
Etch Product Business Group 4
External Use
Available Plasma-Damage Characterization ToolsDevice Characterization– 200 mm CHARM®-2 wafers: reprogrammable EEPROM wafer
Measures threshold-voltage shifts following plasma exposure on various antenna EEPROM structuresLocal peak voltages and currents are then determined
– 200 & 300 mm MOS antenna capacitor & transistor wafersWidely accepted in industry for plasma-damage characterizationDifficult and expensive for equipment vendors to acquire, especially 300 mm!!
Surface-Charge Characterization– 200 & 300 mm Contact Potential Difference (CPD)
Measures residue charge before and after plasma exposure on blanket 1kÅ thermal oxide wafersIncompatible for oxide-etching chemistries
Chamber Characterization– Langmuir probe: measures plasma density
Plasma density does not necessarily correlate to real plasma damageCannot be used for every process condition
– VDC cathode: special cathode for VDC uniformity measurementMeasure Vdc at different location and use ∆VDC to relate to damage
Etch Product Business Group 5
External Use
300 mm Enabler® Dielectric Etch Chamber OverviewVery high frequency source (>100MHz) allows very low DC bias etching and ashing capabilities
Controllable plasma density and neutral dissociation for different etching applications
Two additional knobs –CSTU (magnetic field) and NSTU (gas distribution) to tune etch rate and CD uniformities
Very wide, easily tunable and usable operating windows
Ion Energy Distribution control with dual bias RF
~
~
Polymer/plasma confinement
2.0 MHz RF bias power option
Charged species tuning unit (CSTU)
Neutral species tuning unit (NSTU)
Wafer temperature control
>100 MHz VHF source
~
13.56 MHz RF bias power
Etch Product Business Group 6
External Use
Process Risk #1 – Excessive Magnetic-Field Strength
As the CSTU value exceeds the optimum setpoint for a process, the plasma and the etch rate become non-uniform
Too little CSTU Too much CSTU
Center slow ERCenter fast ER
VDC diagnostic cathode
Normalized ratio optimum CSTU
CSTU3.0%3.5%4.0%4.5%5.0%5.5%6.0%6.5%
Nor
mal
ized
VD
C
σ/mean
PuckPedestal
34 VDC probe pins
RF
VDC
Voltagedivider
Low passfilter
0% 500% 1000% 1500% 100%
%
%
%
%
%
%
%
%
%
%
%
-150 -100 -50 0 50 100 150
Radial position (mm)
Too little CSTUNearly optimum CSTU Too much CSTU
Ash
rat
eM
ax a
sh r
ate
Etch Product Business Group 7
External Use
0 2 4 6 8 10 12 14 16.01
.1
1
51020305070809095
99
99.9
99.99
Pre100% CSTU270% CSTU670% CSTU
EEPROM-based voltage response
Perc
entil
e
Process Risk #1 – Excessive Magnetic-Field Strength
As the CSTU value exceeds the optimum setpoint for a process, the risk of plasma-induced charging damage grows
200 mm EEPROM-based voltageversus normalized CSTU
Pre 100% 270% 670%
Logic customer’s poly-gate MOS capacitor
voltage-breakdown yield
100% 100% 100% 82%
82%100%100%
Ref: www.charm-2.com
Etch Product Business Group 8
External Use
2
6
10
14
18
0% 500% 1000% 1500%
Normalized ratio
Vol
tage
(V)
0.0
0.5
1.0
1.5
2.0 Current (m
A/cm
2)
Mean & 2σ voltageMaximum current
Fail Fail
optimum CSTUCSTU
100%
3.0%3.5%4.0%4.5%5.0%5.5%6.0%6.5%
Nor
mal
ized
VD
C
σ/mean
2σco
nfiden
ce
band
Process Risk #1 – Excessive Magnetic-Field Strength
As the CSTU value exceeds the optimum setpoint for a process, the risk of plasma-induced charging damage grows
200 mm EEPROM voltageversus normalized CSTU
Logic customer’s poly-gate MOS capacitor
voltage-breakdown yield
Pre 100% 270% 670%
100% 100% 100% 82%
Etch Product Business Group 9
External Use
Simplified Antenna MOS Capacitor Structure: SPIDER(SPIDER Systems, Inc)
SPIDER structure
Pre-leakage should be <0.1 nAσ = ±5% accuracyPass criteria = ≥95% at 1 nA
Field thermal oxide 5000 Å
Gate oxide 50 Å
Aluminum antenna electrode 2 µm2
Silicon
Bare metal-gate antenna MOS capacitor structure for plasma-charging-damage evaluation (Ref: IEEE TED, 45(3), p.722, 1998)
Short-flow plasma charging damage test: gate leakage current
– Two terminals: plasma-exposed Al pad & Si substrate backside
– Metal gate is directly on top of gate oxide and has various areas
– Sensitive to plasma charging damage to every part of the process
– Measure leakage before & immediately after plasma exposure
– Cannot detect electron-shading effect
Etch Product Business Group 10
External Use
Logic customer’s poly-gate MOS capacitor
voltage-breakdown yield(200k:1 antenna ratio)
Process Risk #1 – Excessive Magnetic-Field Strength
As the CSTU value exceeds the optimum setpoint for a process, the risk of plasma-induced charging damage grows
Pre 100% 270% 670%
100% 100% 100% 82%
300 mm metal-gate leakage currentversus normalized CSTU
N/A
10-13 10-11 10-9 10-7 10-5.01
.1
1
51020305070809095
99
99.9
99.99
Pre 100:1Pre 200k:1100% 100:1100% 200k:1670% 100:1670% 200k:1
Gate-leakage current (A)
Perc
entil
e
Pass Fail
82%100%
Etch Product Business Group 11
External Use
Process Risk #2 – Unstable Plasma during Transitions
Often, many process parameters are changed between steps– For example, bias & source powers, pressure, CSTU, & chemistry– Linearly ramped to new set points or without any control– Simultaneously changed at the beginning of each step with
varying rates Unoptimized transitions increase the plasma-damage risk – Plasma is unstable as it undergoes significant distribution,
density, and energy changesPlasma changes can be represented by plasma conductance– Characterizes the energy allowed to flow through the plasma
Etch Product Business Group 12
External Use
Process Risk #2 – Unstable Plasma during Transitions
Unoptimized transitions have– Significant plasma conductance changes during transitions to
and from the steady-state step-2 etching condition– Plasma conductance before and after step 2 also deviate
Optimized transitions have– Significantly less plasma conductance change– Plasma conductance before and after step 2 deviate less than the
steady-state step-2 value
Etch Product Business Group 13
External Use
10-13 10-11 10-9 10-7 10-5.01
.1
1
51020305070809095
99
99.9
99.99
Pre 200:1Pre 100000:1Post 200:1Post 100000:1
Gate-leakage current (A)
Perc
entil
e
Low-k trench etch with the optimized single-step process
Pass Fail
MOS Gate-Leakage Yields for Unoptimized & Optimized Low-κ Single-Step Etching
300 mm MOS metal-gate-leakage currents verify risk:Unoptimized does not passOptimized does pass
Logic customer’s MOS poly-gate-breakdown-voltage yield also validate:Unoptimized: 88% & 37%Optimized: 100% & 100%
1k:1 100k:1
.01
.1
1
51020305070809095
99
99.9
99.99
Pre 200:1Pre 100000:1Post 200:1Post 100000:1
Perc
entil
e
Low-k trench etch with the unoptimized single-step process
Pass Fail
Optim
ized
pro
cess
Unoptim
ized
pro
cess
Etch Product Business Group 14
External Use
Plasma-Induced Charging-Damage Results for Single-and Multi-Step Processes before and after Optimization
100%100%+0.9 & −0.2−2 & −54 & 5N/AN/AOptimized multiple step
N/AN/A+11.0 & −2.0−6 & −1010 & 14N/AN/AUnoptimized multiple step
100%100%+0.6 & −0.3−2 & −43 & 599.5%97.1%Optimized single step
37%88%+2.0 & −0.9−2 & −66 & 1032.1%79.1%Unoptimized single step
100%100%+0.0 & −0.0~−2 &−2~2 & ~2100%100%Pre-measure control
100%100%|I | < 1 mA/cm2
|Vmean| < 5 V & |V95% | < 10 V
≥95% @1 nA(σ±5%)
Passing specification:
100k:11k:1+Imax &
−Imin
−Vmean & −V95%
+Vmean & +V95%
100k:1200:1
200 & 300 mm poly-gate MOS cap. voltage-
breakdown yield
200 mm EEPROM-based
voltage and current sensors
300 mm metal-gate MOS cap. leakage-current yieldMeasurement:
Data that meet the threshold specifications and customer criteria are bolded,while data that do not meet are not
By optimizing the process, the plasma-damage risk is significantly reduced
Etch Product Business Group 15
External Use
Conclusions
Two risk factors that contribute to plasma-induced-charging sensitivity in a VHF CCP have been identified– Excessive magnetic fields generate a larger charge
distribution that is conducive to charging damageCan be easily avoided with simple process ground rules
– Plasma instability can occur during transitions from one plasma state to another
By continuously controlling and stabilizing the plasma during a transition, charging effects can be reduced
Continuous damage-free etch processes can now be developed– Etching and ashing of complex multi-layer stacks– For example, all-in-one via and trench etching
Etch Product Business Group 16
External Use