April 6, 2005Page 1
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Application of insitu Plasma Measurement Techniques on Etch Hardware Design and Process
Development for 60nm Structures
Andreas Steinbach, Sven BarthPilot Fab 300Infineon Technologies Dresden, Germany, [email protected]
Marco ReinickeSemiconductor and Microsystems Technology LaboratoryDresden University of Technology, Germany, [email protected]
April 6, 2005Page 2
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Acknowledgement
The authors are very thankful for their support from:– Axel Henke
Stephan WegeJörg SobeInfineon Technologies Dresden, Germany
– Michael KijowskiRuhr University Bochum, Germany
– Ralf NäkeUniversity of Applied Sciences, Dresden, Germany
The work for this paper was partly supported by the EFRE fund of the European Community and by funding of the State of Saxony of the Federal Republic of Germany, project SENSOR, project number 7700 and 7701. The authors are responsible for the content of the paper.
April 6, 2005Page 3
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Outline
Challenges in Dry Etch Technology todayParadigm Shift in Process DevelopmentInsitu Plasma Measurement Techniques for Hardware Design and Process Development Application on Chamber and Process Development
– Power dissipation and RF frequencies– Chamber conditioning– Process development
Conclusions and Summary
April 6, 2005Page 4
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Demands of Dry Etch Technology Development
Feature Size
4nm5nm6nmDeep Trench CD control
12010080Deep Trench Aspect Ratio
403530Deep Trench Mask Open Aspect Ratio
4x nm5x nm6x nm
DRAM Dry Etch Challenge
Main challenges for future dry etch technology development:– Increasing aspect ratio– Tight CD control– Mask selectivity– Uniformity improvement on 300mm wafers
April 6, 2005Page 5
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Example of Technology Challenge:Deep Trench Aspect Ratio at DRAM
Aspect ratio increases exponentially !
20
30
40
50
60
70
80
90
100
110
00,050,10,150,20,250,3
Aspect ratio
Aspect Ratio Challenge
Technology
April 6, 2005Page 6
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Conventional Methodology at Plasma Process Development
~~~
Mask
R +R
RR++
++
++
+
+
+
+ +
+ ++ +
Chamber = ToolboxPlasma = Tool:
in non-equilibrium and non-linearTransport processes:
non-linearReactions on wafer:
non-equilibrium chemistry
Present Methodology:– Offline measurements on blank and patterned wafers after process– Experience and statistical analysis (DOE)– Insitu measurement methods are rarely used
Needed:– Sufficient characterization and control of wafer surface conditions– Significantly improved “tool” = plasma characterization
April 6, 2005Page 7
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Outline
Challenges in Dry Etch Technology todayParadigm Shift at Process DevelopmentInsitu Plasma Measurement Techniques for Hardware Design and Process Development Application on Chamber and Process Development
– Power dissipation and RF frequencies– Chamber conditioning– Process development
Conclusions and Summary
April 6, 2005Page 8
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Paradigm Shift at Plasma Process Development
Paradigm shift: – Joint development of chamber hardware and process by tool
vendors and customers – Joint application of insitu and exsitu measurement methods for
chamber hardware design and process developmentGoal: Process development excellence
– Increase of process development efficiency– Reliable tool hardware and process as basic requirement of AEC/APC
The following measurement methods have been used:– Quadrupol Mass Spectroscopy (QMS)– Optical Emission Spectroscopy (OES) with high and low resolution– Self-Excited Electron Resonance Spectroscopy (SEERS)– RF voltage, current and phase measurement (V,I Probe)– RF Network Analysis
April 6, 2005Page 9
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Measurement Equipment Setup for Chamber Hardware and Process Development at 300mm Etch Chamber
Anode (GDP)
RF Matchbox
RF Matchbox
Z-Scan
VI Probe
~~~
~~~
2 MHz
60 MHz
SEERS
Pump System
QuadrupolMS
OMA
Cathode (CESC)
Wafer
Plasma
Vacuum PumpN2
N2
Final Valve
Slit Valve
Anode (GDP)
RF Matchbox
RF Matchbox
V,I Probe
V,I Probe
~~~~~~
~~~~~~
2 MHz
60 MHz
Gas Inlet
SEERS
Pump System
QMS
OES
Throttle Valve
Cathode (CESC)
Wafer
Plasma
Vacuum PumpN2
N2
Slit Valve
Gas 1…..Gas n
All experiments performed in production environment
April 6, 2005Page 10
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Outline
Challenges in Dry Etch Technology todayParadigm Shift in Process DevelopmentInsitu Plasma Measurement Techniques for Hardware Design and Process DevelopmentApplication on Chamber and Process Development
– Power dissipation and RF frequencies– Chamber conditioning– Process development
Conclusions and Summary
April 6, 2005Page 11
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Quadrupol Mass Spectroscopy (QMS)
Current measurement position above throttle valve:
– Gas composition analysis– Detection of stable gas components
(no ions or neutrals)
Advantages of planned QMS installation at chamber wall:
– Detection of ions and neutrals– Measurement of plasma potential– Ion energy analysis at chamber wall
+
+
R
+
+
R+
++
+
+
GDP
CESC
Wafer
Throttle GateValve
HART TS™ Chamber
Plasma
Quadrupol MS
April 6, 2005Page 12
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Optical Emission Spectroscopy (OES)
Spectral range: 300nm – 900nmSpectral Resolution: app. 0.2nmSampling rate: app. 90s per full
spectrum
Spectral range: 200nm – 950nmResolution: > 1nmSampling rate: min. 20ms per full
spectrumFocus on detailed analysis of
molecular speciesFocus on complex DoE
parameter scans
Acton Research SpectrometerFocal length: 500 mm
Hamamatsu Optical MultichannelAnalyzer (OMA)
High resolution OESLow resolution OES
April 6, 2005Page 13
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Self Excited Electron Resonance Spectroscopy (SEERS)
RF current
ei
n
eff ν(ve-1)p
pkTp
⋅⋅+≈ ∑ σν νstoch
≈ n -1 dV∼ne V1( )-1
V∫
Electrons are key in electrically excited plasmaElectron collision rate νeff reflects: – Excitation – Ionization rates
Electron density ne is:– Plasma density– total number of positive - negative ions
April 6, 2005Page 14
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
RF Voltage and Power Measurement by V,I Probe
Ugenerator Umatchbox Uplasma
~~~50 Ω Matchbox Plasma
Lfeedthrough R
CfeedthroughV,I Probe
Measurement of RF voltage, current and phase at RF matchbox output
– RF power dissipation in feedthrough and plasma can not be separated– Above chamber resonance frequency feedthrough impedance
dominates load impedance significant voltage difference between RF matchbox output and plasma
April 6, 2005Page 15
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Chamber Impedance Measurement by Network Analyzer
Measurement of chamber and feedthrough impedance without plasma by Network Analyzer
Chamber = DUT
Device Under Test
1
111 a
bS =02 =a
a1 a2
b1 b2
April 6, 2005Page 16
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Outline
Challenges in Dry Etch Technology todayParadigm Shift in Process DevelopmentInsitu Plasma Measurement Techniques for Hardware Design and Process Development Application on Chamber and Process Development
– Power dissipation and RF frequencies– Chamber conditioning– Process development
Conclusions and Summary
April 6, 2005Page 17
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Power Dissipation and RF Frequencies: 2MHz Bias Power Impact on 60MHz Source Power
60MHz Delivered vs. Setpoint Power
250
500
750
1000
1250
1500
1750
2000
2250
2500
250 500 750 1000 1250 1500 1750 2000 2250 250060MHz Setpoint Source Power [W] @ 1750W BP
60M
Hz D
eliv
ered
Sou
rce
Pow
er [W
]
500W Bias1000W Bias1500W Bias2000W Bias
Interaction of 2MHz and 60MHz RF power sources was detected by V,I probe measurements60MHz source power loss increases with:
– Higher nominal source power
– Decreasing 2MHz Bias Power
Possible root causes: RF matching limits, temperature drift effects
60MHz Delivered Source Power vs. RF Power
700
800
900
1000
1100
1200
250 500 750 1000 1250 1500 1750 2000 2250 2500Setpoint Source Power [W] @ 1750W BP
60M
Hz D
eliv
ered
Sou
rce
Pow
er [W
]
500W Bias
1000W Bias1500W Bias
2000W Bias
Delivered Power = Setpoint Power
HBr Plasma
April 6, 2005Page 18
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Phase of Load Impedance Z vs. Frequency
-90
-65
-40
-15
10
35
60
85
30 35 40 45
Frequency [MHz]
Phas
e [°]
200mm chamber300mm chamber
1 10 10010
0
10Voltage ratio vs. ferquency
10
10−
U plasma 2 π⋅ f⋅( )
U matchbox
1001 f
MHz
calculatedmeasured
U plasma ω( ) U matchbox1
1 ω2
C feedthrough⋅ L feedthrough⋅−:=U plasma ω( ) U matchbox
1
1 ω2
C feedthrough⋅ L feedthrough⋅−:=
Power Dissipation and RF Frequencies: Chamber Resonance Frequency Analysis
Resonance frequency at about 37 …38 MHzResonance frequency depends mainly on RF feedthrough
Nearly no impact of chamber geometry (200 mm or 300 mm)60MHz excitation is above chamber resonance
considerable RF voltage loss at feedthrough
See Presentation 1.3 at 5th European AEC&APC Conference, Dresden, 2004
April 6, 2005Page 19
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Chamber Conditioning: Monitoring and Baseline by Electron Collision Rate
Electron collision rate vs. process time
1,0E+06
1,0E+07
1,0E+08
1,0E+09
0 20 40 60 80 100 120time [s]
El. C
oll.
rate
[s-1
]Exp_1a Exp_1bExp_1c Exp_2Exp_3 Exp_4
RF Match Issues
stable
unstable
Chamber conditioning was instable due to many different productive lots running in parallel !Electron collision rate directly indicates:
– Chamber hardware issues– Stability and reproducibility of chamber conditioning
April 6, 2005Page 20
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Chamber Conditioning: Chemical Impact of Fluorine Clean and HBr Plasma
79Br Partial Pressure vs. Process Time
1,0E-08
1,0E-07
1,0E-06
1,0E-05
1,0E-04
0 602 1204 1806 2408 3010Time [s]
Par
tial P
ress
ure
[mTo
rr] N2
PlasmaN2
PlasmaN2
PlasmaN2
PlasmaHBr NF3
HBr plasma NF3 plasma
QMS application to optimize chamber clean procedure:– Goal: Stable and
reproducible conditioning
– Effectively decrease fluorine (bromine) contamination inside chamber by HBr (NF3) plasma or purgeF + HBr → HF + Br
– Mass spectrometer useful to monitor behavior of Br/HF species
19F Partial Pressure vs. Process Time
1,0E-08
1,0E-07
1,0E-06
1,0E-05
1,0E-04
1,0E-03
1,0E-02
0 420 840 1260 1680 2100 2520 2940Time [s]Pa
rtial
Pre
ssur
e [m
Torr
] N2plasma
N2plasma
N2plasma
N2plasma
NF3plasma
HBr
HBr purge
HBr plasma
April 6, 2005Page 21
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
3000 4000 5000 6000 7000 8000 90000
10000
20000
30000
40000
inte
nsity
[a.u
.]
λ [Å]
Process Development: Why Application of High and Low Resolution OES ?
High resolution OES: Detailed identification of plasma species for process developmentOptical Multichannel Analyzer with low resolution and high speed: Parameter field measurement and DoE experiments, see next page
6800 6850 6900 6950 7000 7050 7100 7150 72000
4000
8000
12000
16000
20000
24000
28000
F (6
834.
260)
F (6
856.
030)
F (6
870.
220)
F (6
902.
470)
F (6
909.
810)
F (7
037.
460)
F (7
127.
890)
F (7
202.
360)
inte
nsity
[a.u
.]
λ [Å]
0
5000
10000
15000
20000
25000
30000
35000
680 685 690 695 700 705 710 715 720wavelength [nm]
inte
nsity
[a. u
.]
Lowresolution
NF3 Plasma
High resolution
April 6, 2005Page 22
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Process Development: B- Field and RF Power Impact on Plasma Excitation
Simple model gas used: N2
Electron collision rate indicates nonlinear energy transfer from RF power to gas by electronsOptical emission indicates ionic and neutral species in excited states
Electron Collision RateN2 Emission 337.1nm N2+ Emission 423.7nm
April 6, 2005Page 23
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Process Development: Source Power Impact on NF3 Etch Gas Fragmentation
Partial Pressure vs. RF Power
1,0E-07
1,0E-06
1,0E-05
1,0E-04
1,0E-03
1,0E-02
250 500 750 1000 1250 1500 1750 2000 2250 2500Setpoint Source Power [W] @ 1750W BP
Par
tial P
ress
ure
[mTo
rr]
NF(+)F2(+)NF2(+)NF3(+)O2(+)SiF3(+)
SiF3(+)
O2(+)
NF3 fragmentation increases with higher source power
Attention: Decrease of O2 and increase of SiF3indicate complete removal of SiO2hardmask only
Partial Pressure vs. RF Power
6,7E-04
6,8E-04
6,9E-04
7,0E-04
7,1E-04
7,2E-04
7,3E-04
250 500 750 1000 1250 1500Setpoint Source Power [W] @ 1750W BP
Parti
al P
ress
ure
[mTo
rr]
1,0E-07
6,0E-07
1,1E-06
1,6E-06
2,1E-06
2,6E-06
3,1E-06
3,6E-06
4,1E-06
Parti
al P
ress
ure
[mTo
rr]
F(+)NF(+)NF2(+)NF3(+)
F(+)
NF(+)
NF2(+)
NF3(+)
NF3 Plasma
April 6, 2005Page 24
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Process Development: B-Field and RF Power Impact on Electron Collision Rate
200 mTorr70 mTorr
Electron collision rate depends on relation between mean free path of gas molecules and electron Larmor Radius.Larmor Radius is an nonlinear function of 60MHz Source Power, 2MHz Bias Power and B- Field !
April 6, 2005Page 25
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Process Development: RF Source and Bias Power Impact on Plasma Density and Gas Excitation in NF3
Electron densityincreases with sourcepower and changesgas excitation and fragmentationOutlier indicates biaspower dominance at low source powerAll measuredparametersdemonstrate drift of process conditionsduring themeasurements.
5000
10000
15000
20000
25000
30000
35000
40000
250 500 750 1000 1250 1500 1750 2000 2250 2500 275060MHz Setpoint Source Power [W] @ 2MHz 1750W BP
Emis
sion
Inte
nsity
[cts
]
2,0E+09
4,0E+09
6,0E+09
8,0E+09
1,0E+10
1,2E+10
1,4E+10
1,6E+10
Elec
tron
Dens
ity [c
m-3
]O(844.6nm)
F(703.8nm)
Electron Density
1,0E-06
1,0E-05
1,0E-04
1,0E-03
1,0E-02
250 500 750 1000 1250 1500 1750 2000 2250 2500 275060MHz Setpoint Source Power [W] @ 2MHz 1750W BP
Part
ial P
ress
ure
[mTo
rr]
2,0E+09
4,0E+09
6,0E+09
8,0E+09
1,0E+10
1,2E+10
1,4E+10
1,6E+10
Elec
tron
Den
sity
[cm
-3]O(+)
F(+)
NF(+)
NF2(+)
Electron Density
Outlier indicates dominance of 2MHz Bias Power in range of 60MHz Source Power (SEERS model range)
April 6, 2005Page 26
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
Summary:The Paradigm Change in Plasma Process Development
Joint development work by tool vendors and customersJoint application of insitu and exsitu measurement methods
Enable to
Improve knowledge of tool characteristics and plasma process conditions significantlyAccelerate plasma etch tool and process developmentDevelop reliable tools and processes as a basic precondition of AEC/APC in production
April 6, 2005Page 27
Andreas SteinbachSven BarthInfineonTechnologies DresdenMarco ReinickeDresden University of Technology
… Finally a possible Back Up Solution, if the Paradigm Shift should not work …