Chemical Sensing and Advanced Process Control for AlGaN/GaN HEMT Manufacturing [Lab or

advertisement
Chemical Sensing and Advanced Process Control for
AlGaN/GaN HEMT Manufacturing
[Lab or
Research Group
Logo]
Soon Cho, Michael E. Aumer, Darren B. Thomson, Dan Janiak, Gary W. Rubloff, Deborah P. Partlow
Process Variability is a Concern...
High Performance Applications
… GaN Epilayer Photoluminescence Quality Prediction
GaN-based electronic devices for high frequency, high power
applications, such as radar electronics
Band edge and deep level must be optimized concurrently
AlGaN Cap Layer Thickness Metrology
Challenges
Growth of GaN epi films and GaN-based alloys with
reproducibility and quality constraints that are much
higher relative to optoelectronic applications
SensorSensor-based metrology can predict/control AlGaN thickness to
within 3A or 1%
Band Edge
9
G379
8
G386
7
G382
G383
6
G380
5
0
-3
0.0
1.5x10
-3
2.5x10
-3
3.0x10
-
3.5x10
G386
0.03
G383
G382
0.02
G380
G379
0.01
Better Quality
0.00
0.0
Normalized H2O level during GaN epitaxy
-5
1.6x10
-5
1.8x10
-5
2.0x10
-5
2.2x10
-5
2.4x10
Normalized O2 level during GaN epitaxy
Low impurity condition must be established prior to GaN growth step
step
Pre-process Contamination Control is a key
sensor
sensor
pattern generation
metrology
etching
GaN Epilayer Crystal Quality Prediction
metrology
•
run-to-run
feedback control
real-time fault
run-to-run
detection
feedforward
control
fault classification
& response
Based on intrinsic process chemistry – two major
byproducts Methane & Ethane representing different
reaction pathways
Monitoring the Me/Et ratio predicts product crystal quality to
~ 5% precision
•
GaN-based Heterostructure Design
Multiple material parameters across the
multi-layer structure must be cooptimized Î Direct consequences in
terms of device performance
•
10
Film Thickness
XRD FWHM
G380
300
G382
G386
250
200
G379
G383
Thick film anomaly
150
Better Quality
100
0
0.01.6
1.8
2.0
2.2
2.4
Methane/Ethane ratio during GaN epitaxy
2.6
GaN PL band edge intensity (a.u.)
Post-Process
Characterization
Gan XRD FWHM @ (002) (arcsec)
350
Better Quality
PL Band Edge
•
G386
7
G382
G383
6
G380
5
0
0.0
-4
5.0x10
-3
1.0x10
-3
1.5x10
-3
2.0x10
G386
0.03
G383
G382
0.02
G380
G379
0.01
Better Quality
0.00
-5
0.0
1.0x10
1.2x10
-5
-5
1.4x10
-5
1.6x10
-5
1.8x10
-5
2.0x10
Normalized O2 level at the end of pre-deposition anneal
•
•
In-situ metrology is key to achieving real-time APC
GaN for advanced electronic application faced with serious
manufacturing hurdles
•
Chemical sensing in GaN MOCVD for Advanced Process Control
•
•
•
•
•
9
G379
8
G386
•
7
G382
G383
G380
6
5
0
0.01.6
•
1.8
2.0
2.2
2.4
Methane/Ethane ratio during GaN epitaxy
Lower Me/Et ratio produces better quality (XRD, PL band edge)
∫ (Signal) dt
8
Deep Level
0.04
Reduce impurity levels in reactor through prepre-process purge,
anneal, and R2R bakebake-out
Watch for impurity levels to drop below critical limit to guarantee
guarantee
acceptable PL quality
Same Me/Et ratio predicts PL band edge intensity to ~ 5% precision
precision
TARGET
G379
Normalized H2O level at the end of pre-deposition anneal
GaN Epilayer Photoluminescence Quality Prediction …
GaN MOCVD Process
Metrology Model
Band Edge
CONCLUSIONS
SensorSensor-based REALREAL-TIME Advanced Process Control
Real-time In-situ
Mass Spectrometry
0.05
Better Quality
9
GaN PL deep level intensity (a.u.)
GaN PL band edge intensity (a.u.)
10
Advanced Process Control (APC) already pervasive in the
Si ULSI industry is applied to the GaN-based processes to
bring reproducibility and process optimization
real-time
control
-3
2.0x10
Deep Level
0.04
Î Can predict, but little room for course correction during GaN step
step itself
Approach
deposition
0.05
Better Quality
GaN PL deep level intensity (a.u.)
GaN PL band edge intensity (a.u.)
10
Î Surface reaction (Ethane) pathway desired
2.6
Accelerated learning at R&D stage
Manufacturing reproducibility
Insight into intrinsic chemistry & process
Extensive fault detection & management in use already
Precision metrology for critical AlGaN layer thickness (20-25nm)
to ~1% in real-time demonstrated
Accurate prediction of product quality (crystal quality, PL) in
real-time demonstrated
APC benefits exploited by Northrop Grumman development
• AlGaN thickness control
• Pre-process contamination control
Download