Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Usage of Hercules inside
commercial FDC System MaestriaDennis Föh
Advanced Process Control
Micronas
Freiburg
2
Why are Plasma Parameters ideal for FDC in Etch ?
Electron Density and Collision Rate are sensitive to
Process parameters (e.g. gas flow, pressure, power)
Process / chamber drift (conditioning, clean, WAC)
Product differences (e.g. open area)
Tool failure (e.g. baratron, rf, MFC)
Plasma Parameters provided by HERCULES® help to
optimize conditioning, WAC, MTBC, cleaning procedures
understand process / tool issues
3
Influence of Electron Density on Etch Process
Source : Plasmetrex
n+ion
density
ji+ion energy
current density
ne νcollision per volume element
productionof
etch species
etch rate
selectivity
homogeneity
neelectron density
electron energy distribution
physical influence
chemical impact
RF peak voltage
chemical etching
desorptionof etch
products
physical sputtering
4
Influence of Tool / Process Parameters on Collision Rate
Source : Plasmetrex
Collisionrate ν
Gas temperature
Wall temperature Wafer temperature
Relative concentration of gases
Wafer
Crosssection
Arcing
Gas flow
PressureBaratron
MFCsThrottle valve
B field
Density of neutrals
RF power in plasma
Generator
MatchboxRF feedthrough
chuck
8th European AEC/APC Conference 2007
5
Basic Setup of Hercules®
01020304050
1 14 27 40 53 66 79 92 105118131144
Etch Time [s]
Collision Rate [106 s-1]Plasmachamber
Plasma parameterse.g., electron collision rate.
Measurementsignal
-0,6
-0,4
-0,2
0
0,2
0,4
0,6
0,8
1
1,2
0 5 10 15 20
ACI
Time
13,56 MHz
1
bias / peak voltage
Hercules® APC xM
Matchbox
Generator
Source : Plasmetrex
SensorTCP 9400/9600
8th European AEC/APC Conference 2007
6
Influence of Poly TCP Process Parameters on Collision Rate
Std Std StdHBr+10% HBr-10%
p +10% p -10% Std StdStd
Ptop +10% Ptop -10% Std StdStd
±10% TCP-Power = ca. 15-25% change in collision-rate
±10% HBr = ca. ±3% change in collision-rate ±10% pressure = ca. ±11% change in collision-rate
7
Gate-Poly MSPH 0.45µ - VPCD 0.5µ (Poly-OE-Step)
2 Lots MSPH 0.45µ (61% OA)
2 Lots VPCD 0.5µ (74% OA)
Plasma parameters are very sensitive
to the etched open area (OA)
Collision Rate vs. Open Area & Endpoint Sensitivity
Collision Rate
Electron Density
8th European AEC/APC Conference 2007
Possibility of EOP detection.
8
Plasma Clean Cycles before Wet Clean
Large change inside chamber during 1st cycle only. 3-5 cycles are sufficient forstable chamber before chamber opening. Already implemented in production
without any negative impact on cleaning efficiency. (time saving : 1 hr.)
Collision Rate
9
Effect of Nitride to Poly Switch w/o Seasoning
Switching between fluorine-based chemistry and chlorine-based chemistry could cause
yield loss due to particles fallingfrom top plate !
1st PLRAEZ after FLDAEZ
707580859095
100
1 5 9 13 17 21 25
Wafer #
Norm
aliz
ed
Yie
ld [%
]
2nd PLRAEZ after FLDAEZ
707580859095
100
1 5 9 13 17 21 25
Wafer #
Norm
aliz
edY
ield
[%]
10
Comparison Continuous / Random Switchrandom switch continuous switch
Poly-WAC Poly-WAC
Nitride-WAC Nitride-WAC
Electron density WAC
Collision RateEtch Processes
Poly Poly
Nitride
Nitride
Poly
Collision Rate WAC
11
FDC-Software Maestria (PDF Solutions / Si Automation)
8th European AEC/APC Conference 2007
12
Maestria Deployment Status @ MicronasEtch : Lam Alliance + Lam Rainbow + FSI Mercury + Mattson Aspen IIDiffusion : ASM 400 Implant : Varian EHP 500Litho : Canon I5+ StepperThinfilm : AMAT Endura
32 Tools incl. 65 modules / chambers850 strategies active, 130 with OCAPs, 420 with limitsPCA (Process Control Analyzer) training for process & maintenance (30 engineers)PCA implemented in daily work
but : sometimes internal tool data is not enough, additional sensors are necessary...
13
Source : PDF Solutions modified for Hercules
Hercules & Maestria Deployment
14
Source : PDF Solutions modified for Hercules
8th European AEC/APC Conference 2007
15
DC Plan : Integration of Hercules Variables
16
Definition of Indicators
17
18
19
Easy Correlation Tool Parameters - Sensor Data
Implementation of external sensor data into Maestria system allowseasy comparison / correlation of tool and sensor data
Collision Rate Bottom-Temp
8th European AEC/APC Conference 2007
20
FDC Example : Defect Baratron (1)
Because of sensitivity to process / tool parameters, collision rate and electron densityare ideal for detecting tool failures resulting in process parameter drifts. In this examplea baratron had a defect causing a shift in real pressure. The process was not aborted
by the tool because the defect baratron fooled the tool with the "correct" pressure. Additional parameters like plasma parameters are helpful for FDC.
Col
lisio
nR
ate
different pressure caused bydefect baratron !
21
FDC Example : Defect Baratron (2)
8th European AEC/APC Conference 2007
Problem is indicated only by collision rate & VATvalve angle. Collision rate is a convenient in situ parameter for the real process status inside the chamber !
22
Process Control by Wafer-to-Wafer-Difference
Wafer-to-wafer-difference of plasma parameters is an easy recipe independent process/tool control ;
easy identification of excursions, tool & process problems !
Collision Rate Electron Density
8th European AEC/APC Conference 2007
Dennis Föh
Section Head APC
Micronas
Hans-Bunte-Str. 19
79108 Freiburg
Germany
www.micronas.com