DEPARTMENT OF INFORMATION ENGINEERING GAN POWER HEMT RELIABILITY RESEARCH WITHIN THE POWERBASE ECSEL PROJECT : FROM FAILURE PHYSICS TO INDUSTRIAL EVALUATION Enrico Zanoni , Matteo Meneghini, Gaudenzio Meneghesso, Carlo De Santi, Alessandro Barbato, Matteo Borga, Maria Ruzzarin University of Padova, Department of Information Engineering
40
Embed
G N POWER HEMT RELIABILITY RESEARCH WITHIN THE … · GaN power HEMT reliability research within the POWERBASE project–[email protected] GaN reliability beyond initial state
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
DEPARTMENT OF
INFORMATION
ENGINEERING
GAN POWER HEMT RELIABILITY RESEARCH WITHIN
THE POWERBASE ECSEL PROJECT : FROM FAILURE
PHYSICS TO INDUSTRIAL EVALUATION
Enrico Zanoni, Matteo Meneghini, Gaudenzio Meneghesso, Carlo De Santi, Alessandro Barbato, Matteo Borga, Maria Ruzzarin
University of Padova, Department of Information Engineering
GaN power HEMT reliability research within the POWERBASE project– [email protected]
OUTLINE• The POWERBASE consortium and research project
• GaN strengths and issues (dynamic-Ron, breakdown effects)
• Failure physics: where we were and what we learned
• drain-to-substrate conduction mechanisms and time-
dependent breakdown
• forward bias p-gate degradation
• hot-electron effects
• Industrial reliability
• dynamic testing approaches
• JEDEC-like accelerated tests
• need for standardization
• Conclusions
OUTLINE
GaN power HEMT reliability research within the POWERBASE project– [email protected]
The POWERBASE project consortium
PowerBase, an ECSEL project with 39 partners from 9 European countries
• demonstrate and assess GaN on Si e-mode (normally off) devices
• Pilot production line for fully processed GaN HEMT compatible with high volume CMOS power manufacturing facilities
• Pilot production line for advanced packaging of GaN-based power products
• Prove target achievements in key demonstrators applications in the “smart energy” domain
• Demonstrate reliability and quality as key enabler to gain market acceptance for these new technologies.
Funded by the Electronic Component and Systems
for European Leadership Joint Undertaking (JU-
ECSEL), grant 662133. This JU receives support
from the European Union HORIZON 2020 research
and innovation programme and from Austria,
Belgium, Germany, Italy, The Netherlands, Norway
Slovakia, Spain and United Kingdom
GaN power HEMT reliability research within the POWERBASE project– [email protected]
POWERBASE WP6 „Reliability“ partners
University of Bristol, UoB
KompetenzzentrumAutomobil- undIndustrieelektronik, KAI
InteruniversitairMicro-ElectronicaCentrum, IMEC
Institute for Microstructure of Materials and Systems (FhG-IMWS)
Universityof Padova, UNIPD
InfineonTechnologiesAustria, IFAT
Slovak University of Technology in Bratislava, STUBA
NanoDesign,NANO
University of Graz, UNIGRAZ
GaN power HEMT reliability research within the POWERBASE project– [email protected]
OUTLINE• The POWERBASE consortium and research project
• GaN strengths and issues (dynamic-Ron, breakdown effects)
• Failure physics: where we were and what we learned
• drain-to-substrate conduction mechanisms and time-
dependent breakdown
• forward bias p-gate degradation
• hot-electron effects
• Industrial reliability
• dynamic testing approaches
• JEDEC-like accelerated tests
• need for standardization
• Conclusions
OUTLINE
GaN power HEMT reliability research within the POWERBASE project– [email protected]
K. Chen, O. Häberlen et al. GaN-on-Si Power Technology: Devices and ApplicationsIEEE TED 64, 779 (2017)
Comparison on devices with similar RDSON
QOSS is the output charge, required to load the drain-source capacitance (higher for Si SJ)
QRR is the reverse recovery charge (no body diode in GaN devices)
QG is the total gate charge
Robust VTH
Reliability?
GaN transistors significantly minimize resistive and switching losses, comparedto Si superjunction
GaN on Si power technology
GaN power HEMT reliability research within the POWERBASE project– [email protected]
GAN HEMTS VS COOLMOS IN BOOST APPLICATION
• GaN HEMT (cascoded) and SiCoolMOS are compared• At 500 kHz/1.5 kW (HS) overall loss reduced from 42 W
to 26 W (-41 %)• 13.5 W of total loss due to inductor and diode• Performance enhancement mainly due to better
RDSONxQOSS
(GaN: 5 mWµC, Si SJ: 23.5 mWµC)[QOSS=charge for charging DS capacitance]
Hard switched, 500 kHz, 230:400 V, CCM
Valley switching DCM, 100 kHz, 230:400 V
K. Chen, O. Häberlen et al. GaN-on-Si Power Technology: Devices and ApplicationsIEEE TED 64, 779 (2017)
GaN on Si power technology
GaN power HEMT reliability research within the POWERBASE project– [email protected]
due to compensation or unwanted impurities, e.g. C
deep levels induce carrier trapping effects during switching: dynamic Ron
GaN power HEMT reliability research within the POWERBASE project– [email protected]
DYNAMIC RON: A LIMITATION?
VDD
(50 V)
VGS
(-8 V)
99 µs
1 µs
OFF
ON
OFF
ON
OFF
ON
CurrentCollapse
Measured by pulsing from restconditions (VG_QB=0, VD_QB=0)
Measured by pulsing from trappingconditions (VG_QB=-8, VD_QB=50)
Currentcollapse
Dynamic-Ron recoverable increase in on-resistance induced by charge trapping in off-state On-resistance is higher when measured after a switching event (OFFON)
May significantly limit the dynamic performance of the devices, the maximum switching frequency, and increase switching losses
Dynamic on-resistance effects
GaN power HEMT reliability research within the POWERBASE project– [email protected]
HT1: EC - ET = 0.557 eV, s = 2.02E-18 cm2, Si-Ga vacancy or N
vacancy
HT2: EC - ET = 0.878 eV, s = 1.63e-15 cm2, GaN dislocation
ET1: EC - ET = 0.377 eV, s = 1.75e-14 cm2 AlGaN/GaN interface
ET2: EC - ET = 0.230 eV, s = 6.73E-17 cm2, N vacancy
Advancements in defect
analysis: Identification of
the deep levels
responsible for gate
leakage based on C-
DLTS
High gate leakage Low gate leakage
analysis of deep levels and dynamic Ron data
demonstrate epi and device improvement over
POWERBASE timeline
Objective 6.2:
Parasitic effects characterization and modeling
Slovak University of Technology in Bratislava, STUBA
NanoDesign,NANO
GaN power HEMT reliability research within the POWERBASE project– [email protected]
Recessed and regrowth gate p-GaN Technology
Fully recessed 1st barrier
Regrowth of 2nd barrier and p-GaN layer
Excellent Vth stability
p-GaN ring at drain for dynamic RDSON optimization
Device concept
Hole injection at drain side
Key features
Infineon, Proceedings of ISPSD 2015 (Hongkong) and ISPSD 2016 (Prague)20
By inserting a p-GaN gate layer enhancement-mode operation is achieved
GaN power HEMT reliability research within the POWERBASE project– [email protected]
WP6 reliability methodology
GaN-related testingtools and methods
Identification of failure modes and
mechanisms
Derivation of acceleration laws
statistically relevantindustrial reliability
testing
Trusted extrapolationof lifetime data
GaN power HEMT reliability research within the POWERBASE project– [email protected]
OUTLINE• The POWERBASE consortium and research project
• GaN strengths and issues (dynamic-Ron, breakdown effects)
• Failure physics: where we were and what we learned
• drain-to-substrate conduction mechanisms and time-
dependent breakdown
• forward bias p-gate degradation
• hot-electron effects
• Industrial reliability
• dynamic testing approaches
• JEDEC-like accelerated tests
• need for standardization
• Conclusions
OUTLINE
GaN power HEMT reliability research within the POWERBASE project– [email protected]
Main failure mechanisms are related with chargetrapping and time-dependent breakdown
Knowledge on failure modes and mechanisms of p-gate enhancement-
mode power GaN HEMT and their acceleration laws
hot-electron trapping
(hard switching)
time dependent
breakdown of p-gate
stack (positive bias)
time dependent
gate-drain breakdown
drain-substrate
leakage and time
dependent
breakdown
positive and negative threshold
voltage instability
at the beginning of
POWERBASE,
knowledge on failure
mechanisms referred to
GaN HEMT for GaN-on-
SiC microwave
applications only
GaN power HEMT reliability research within the POWERBASE project– [email protected]
GaN reliability beyond initial state of the art
Knowledge on failure modes and mechanisms of p-gate enhancement-
mode power GaN HEMT and on the related acceleration laws
• conduction mechanisms of drain-to-substrate leakage
• impact of substrate and buffer design
on the performance and breakdown of
GaN-on-Si power HEMT
• - substrate orientation
• - p doping level
• time-dependent breakdown of drain-to-substrate vertical structure
• degradation mechanisms of p-gate adopting Schottky and ohmic
contact to p
• hot-electron effects
VD=800 V
GaN power HEMT reliability research within the POWERBASE project– [email protected]
Influence of p-Si substrate doping on drain-substrate breakdown and reliability
When high voltage is applied to the drain with
respect to p-Si substrate a trade-off is established
GaN power HEMT reliability research within the POWERBASE project– [email protected]
Extended end-of-life accelerated testing
33
500 V / 15 kW power supplies & active loads Climate chamber with 24 stress test modules
GaN power HEMT reliability research within the POWERBASE project– [email protected]
Long-term reliability package test
Long-term reliability test-plan has been implemented by extending the JEDEC
guidelines for electronic devices
GaN power HEMT reliability research within the POWERBASE project– [email protected]
Pre-qualification summary: 600V & 70mW,
Standard Stress Tests based on JESD47
StressRequired
Duration
Sample
SizeStatus/Results
Risk
Rate
High Temperature Reverse Bias
(HTRB)
JESD22 A-108 (Q101)
1000 hr 1 x 28pcs
• 168hr: 0 / 28
• 500hr: 0 / 28
• 1000hr: 0 / 28
*. Minor Idss current drift but no noticeable change in
all the other parametersPass
Positive High Temperature Gate
Stress (P-HTGF)
JESD22 A-108 (Q101)
1000 hr 1 x 30
• 168hr: 0 / 30
• 500hr: 0 / 30
• 1000hr: 0 / 30
• Ig & Id leakage current drift but within the
specification limitPass
Negative High Temperature Gate
Stress (N-HTGS)
JESD22 A-108 (Q101)
1000 hr 1 x 30
• 168hr: 0 / 30
• 500hr: 0 / 30
• 1000hr: 0 /30
• No noticeable parametric drift
Pass
High Humidity High Temperature
Reverse Bias (H3TRB)
JESD22 A-101
1000 hr 1 x 30
• 168hr: 0 / 30
• 500hr: 0 / 30
• 1000hr:0 / 30
Pass
High Temperature Storage Life
(HTSL) JESD22 A-1031000 hr 1 x 40
• 168hr: 0 / 40
• 500hr: 0 / 40
• 1000hr: 0 / 40
Pass
Intermittent Operational Life Test
(IOL)
MIL-STD 750/Meth.1037
15,000x 1 x 30• 7500x: 0 / 30
• 15,000hx: 0 /30Pass
ESD-HBM
JS-001Target: 2kV
3 pcs per
voltage
PASS
(Target: 2 kV, Class 2)
• No failure up to 2kV
• Fail at 3kVPass
ESD-CDM
JS-002Target: 500V 3 x 3
Pass
(Target: 500 V, Class C3)• No failure up to 1250V Pass
Preliminary data of “Temperature Cycle (TC, 55 °C / +125 °C)” showed no failures up to 2000 Cycles and regular TC lots will run for the qualification.
Highly Accelerated Stress Test (HAST, 130 °C / 85 % r.h. / 480 V, JESD22 A-110) is under evaluation and already passed 96hr without failures (June 2018).
GaN power HEMT reliability research within the POWERBASE project– [email protected]
Parametric drift:
PowerBase 600V GaN NormOff HEMT
1.E-02
1.E-01
1.E+00
1.E+01
1.E+02
1.E+03
0hr 168hr 500hr 100hr
Id (
uA
) at
Vd
= 6
00
V
Stress Tme (Hr)
IDSS at Vd = +600VPost-HTRB
1.E-02
1.E-01
1.E+00
1.E+01
1.E+02
0hr 168hr 500hr 1000hrIGSS
(u
A)
at V
g =
+1.5
V
Stress Tme (Hr)
IGSS at Vg = +1.5VPost-HTRB
0.9
1
1.1
1.2
1.3
1.4
1.5
0hr 168hr 500hr 1000hr
Vth
(V
) at
Id =
2.6
mA
Stress Tme (Hr)
Vth at Id=2.6mAPost- HTRB
40
50
60
70
80
0hr 168hr 500hr 1000hrRD
Son
(mW
) at
Id =
8A
Stress Tme (Hr)
RDSon(mW) at Id=8APost-HTRB
✓ No parametric drift except for a small increase in IDSS
GaN power HEMT reliability research within the POWERBASE project– [email protected]
Application-related reliability test
How to test GaN HEMT devices
for reliability & robustness under
application conditions?
❖ Standard JEDEC qualification –
PTC, HTOL, HTRB, IOL …
(dynamic testing not considered)
❖ Lab characterization – SOA,
double pulse, UIS …
❖ Customer life test of final
product @ system level
❖ Carry out device level stress
testing in an application
related test configuration
Control module
PWM Guard Switch
Power Supply
Active Load GaN
GaN
External devices
Application board GaN test board
I IN
I OUT
I SENS V SENS
Blocking Diodes
HostPC
Ethernet
GaN power HEMT reliability research within the POWERBASE project– [email protected]
Lifetime testing
Projection at 480 V
𝐿 𝑡 = 𝐴 ∗ 𝑒−𝛾𝑉 ∗ 𝑒𝐸𝑎
𝑘𝑇
Life Time Model:
Lifetime requirement:< 1 fit for 15 years at
480V, 125 °C
➢ Very tight distribution (high ß good process control)
➢ The conservative model predicts a lifetime of ~55 years at 100
ppm, 480 V and 125 °C
➢ > 3X safety margin from Infineon’s criteria
JEDEC testing 3 x 77
parts, 480V, 1000h
GaN power HEMT reliability research within the POWERBASE project– [email protected]
Increased buffer reliabilityHigh reliability, stability on 200mm (650V epi on p+)
39
63% failure63% failure, area scaling
■ scaling to 0.9mm2
power transistor.
0.01% failure, area scaling
■ 63% failure to 0.01% failure.
10 years820V
Buffer induced Ron,dyn Buffer reliability (TDDB)
Large step in overall reliability of 650V buffer architecture on p+
▪ Buffer induced Ron,dyn is below 5% at 650V
▪ Maximum applicable drain voltage is 820V (large margin over spec)
▪ Very high β, indicating good uniformity
GaN power HEMT reliability research within the POWERBASE project– [email protected]
GaN Standardization: JEDEC JC-70.1
The power electronics industry recently reached an important milestone:
The Joint Electron Device Engineering Council (JEDEC) announcedthe formation of a new committee in Sep-2017:JC-70 Wide Bandgap Power Electronic Conversion Semiconductors
JC-70.1 subcommittee is dedicated to GaN
This committee’s charter is to standardize reliability and qualification procedures, data sheet elements and parameters, and test and characterization methods
Having a common standardwill enable the power industryto compare and contrastdifferent GaN deviceson a common base.
40
GaN power HEMT reliability research within the POWERBASE project– [email protected]
Need for GaN-specific standard testing
methodology : JEDEC committee JC-70.1
today at 15:55
Status of GaN device qualification standards effort by new JEDEC
committee JC-70.1
Tim McDonald, Senior Consulting Advisor , CoolGaNTM Technology
Development, Infineon Technologies
GaN power HEMT reliability research within the POWERBASE project– [email protected]
CONCLUSIONS: POWERBASE impact
• Significant improvement of the knowledge on failure modes and
mechanisms of conventional- and enhanced-substrate power GaN HEMTs
• Development and validation of new testing and failure analysis techniques
specifically tailored to GaN technology
• Accurate modeling tools and methods for device thermal evaluation and
control of junction temperature (not shown)
• Highly-accelerated stress testing of GaN power HEMTs now possible for
the first time in large scale under closely application-related conditions
• Before POWERBASE, no End of Life data for GaN power HEMTS was
available under real application conditions of GaN power HEMTs. These
data are now linked to specific failure modes and mechanisms
GaN power HEMT reliability research within the POWERBASE project– [email protected]