University of Cassino and southern Lazio Department of...
Transcript of University of Cassino and southern Lazio Department of...
University of Cassino and southern Lazio
Department of Electrical and
Information Engineering
Electronic Research Group
Contact: prof. Giovanni Busatto [email protected] +39 07762993699
Laboratory of Industrial Electronics
ELECTRONICS Group STAFF
Laboratory of Industrial Electronics
Department of Electrical and Information Engineering
Full Professor: Giovanni Busatto
Associate Professors: Francesco Iannuzzo
Annunziata Sanseverino
Researchers: Francesco Velardi
Carmine Abbate
Ph.D. Student: Valentina De Luca
Laboratory Technician: Tomasino Iovini
Main Activities
Experimental (NDT) characterization of Power Devices
Overvoltage, overcurrent, high and low temperature,
short circuit stresses on power devices and modules
Cosmic Ray effects on Power Devices (Burnout & Total
dose)
2D/3D Devices FEM Simulation
Devices Modeling & Simulation – Lumped-Charge approach
High-Voltage/high-current/picoampere PCB layout design
Recent Active collaborations
ANSALDOBREDA – Naples – Italy STMicroelectronics – Catania – Italy Fairchild – Munich – Germany ECPE – European Centre for Power Electronics – Nuremberg -
Germany INFN – Italian Institute for Nuclear Physics – Rome - Italy
ECPE Competence Center
…specialized in non destructive testing of discrete and power modules
The laboratory is an…
Non-Destructive Tester NDT Facility
High voltage Area
Principle Schematic
Tester control panel
Features:
• SOA Tests • Sort circuit • Unclamped • Temperature • Aging • …
Voltage: 0 - 6500V (8000V) Current: 0 - 8000A
Applications:
High Voltage NDT
High voltage Devices: 1700V ≤ Vcc ≤ 6500V Ls = 110 nH Cs = 5nF
High current Devices: Ic ≤ 8000A Vcc ≤ 1700V Ls < 50 nH Cs = 12nF
High Current NDT
Very Low Inductance NDT
Very Fast Devices:
Vcc ≤ 1700V, Ls < 30 nH, Cs < 300pF
Detail of busbar
Temperature Characterization
-50°C +200°C Special Fluid
Railway inverter under test
Static Characterization
Curve Tracer 576 1.7kV, 20A pulsed
Agilent B1500A w/ 2 SMUs: • High current and voltage – junction BV • High precision (pA) – oxides percolation
Ad hoc set-ups: high voltage and current, high precision (e.g. 3kV 1nA)
Keithley Measurement System w/2 SMUs: • High current and voltage 2410
• High precision (pA) 2601
Post Failure Analysis
NISENE decapsulation system
Inspection microscope
Development of high voltage, high performances switch Marx modulator
High voltage bridge leg For railway applications
Cosmic Ray effects on Power Devices
Single Event Burnout: Experiment
3-µm resolution Ion Impact Mapping
Operations of power devices in highly stressing electrical
conditions
C. Abbate, G. Busatto, F.Iannuzzo
DIEI – Università di Cassino e del Lazio Meridionale Via G. di Biasio, 43, Cassino, Italy
e-mail: [email protected]
Outline Introduction Failure mechanisms under extreme electrical stresses:
In the device linear operating region During the device switching In avalanche conditions External causes (eg. cosmic rays impacts)
Non destructive techniques for electrical testing of power devices The basics of ND techniques taken from the literature The approach to ND SOA characterization of power module Typical problems related with the use of Non Destructive tester The “precursors” of the instabilities
Cosmic rays impact Conclusions
Switch On Switch Off
A power device in a switching circuit
Buck converter Test circuit
Switching Waveforms
Collector Voltage
Col
lect
or V
olta
ge
SOA
Electrical limits of a power semiconductor switch
Trench Gate Structure
Double Diffused Planar Structure
Elementary cells of a Power MOSFET
Power MOSFET parasitic BJT
Drain
P +
N +
P _
GateSource
N_
Body
N+
Collector
Emitter
Base
RP+ A possible cause of failure!
IGBT (Insulated Gate Bipolar Transistor)
Trench Gate Structure Planar Structure
IGBT (Insulated Gate Bipolar Transistor)
Trench Gate Structure Planar Structure
IGBT parasitic thyristor
A further possible cause
of failure!
Outline Introduction Failure mechanisms under extreme electrical stresses:
In the device linear operating region During the device switching In avalanche conditions External causes (eg. cosmic rays impacts)
Non destructive techniques for electrical testing of power devices The basics of ND techniques taken from the literature The approach to ND SOA characterization of power module Typical problems related with the use of Non Destructive tester The “precursors” of the instabilities
Cosmic rays impact Conclusions
Failure mechanisms under extreme electrical stresses In the device linear operating region
Short circuit operations (or overload conditions) Instable DC operations (second breakdown)
During the device switching At the turn-off Reverse recovery of internal diode
In avalanche conditions UIS – Unclamped Inductive Switch
Short circuit of a Power Switch: Definitions Type 1: Short circuit takes place when the device is in
the off-state so it is turned on in short circuit
Type 1
Type 2
Type 2: Short circuit takes place when the device is in the on-state
Type 1 short circuit
Vgs
Commercial devices can sustain S.C. for 10 ms at the TJMAX and with VGS=16V
The failure of a Power MOSFET after a short circuit
Fresh Device After a short circuit Failure
Type 1 short circuit: First failure mechanism
A.Ammous, K.Ammous, H.Morel, B.Allard, D. Bergogne, F. Sellami, J.P.Chante “Electrothermal Modeling of IGBT’s: Application to Short-Circuit Conditions,” IEEE Trans. Power Electronics, Vol. 15, No. 4, JULY 2000
Device avalanching due to stray inductance
Type 1 short circuit: Second failure mechanism
Device avalanching due to stray inductance can cause device failure
H.G.Eckel, L.Sack, “Experimental Investigation on the Behaviour of IGBT at Short-Ciruit during the On-State,” Proc. IEEE PESC, 1995
Type 2 short circuit
S.C.
Avalanche limit
Failure mechanisms under extreme electrical stresses In the device linear operating region
Short circuit operations (or overload conditions) Instable DC operations (second breakdown)
During the device switching At the turn-off Reverse recovery of internal diode
In avalanche conditions UIS – Unclamped Inductive Switch
Failure of Power Devices in the active region
DC Safe Operating Area
Second Breakdown
DC electro-thermal instability of a low voltage Power MOSFET
Evolution of the temperature in the hot-spot
P.Spirito, G.Breglio, V.d'Alessandro, N.Rinaldi, “Analytical Model for Thermal Instability of Low Voltage Power MOS and S.O.A. in Pulse Operation,” Proc. ISPSD 2002
Basic mechanism in thermal run-away
Regenerative Thermo-electric Effect
Generated Power vs. Temperature
ID
ID
aT<0
Stable Unstable aT>0
Trans-characteristics of a low voltage Power MOSFET
P.Spirito, G.Breglio, V.d’Alessandro, “Modeling the Onset of Thermal Instability in Low Voltage Power MOS: an Experimental Validation,” Proc. ISPSD 2005
VGS=Const 0
TID
Ta
Possible thermal run-away
Experimental evidence of hot-spot formation
P.Spirito, G.Breglio, V.d’Alessandro, “Modeling the Onset of Thermal Instability in Low Voltage Power MOS: an Experimental Validation,” Proc. ISPSD 2005
Radiometric temperature detection for 55V Power MOSFET operated under pulsed bias conditions
Failure mechanisms under extreme electrical stresses In the device linear operating region
Short circuit operations (or overload conditions) Instable DC operations (second breakdown)
During the device switching At the turn-off Reverse recovery of internal diode
In avalanche conditions UIS – Unclamped Inductive Switch
Second Breakdown in a BJT
3D structure of a Power BJT
Typical top layout of a power BJT
Sketch of the 3D structure of
a power BJT
Crowding of emitter current
B.A.Betty, S. Krishna, M.S.Adler, “Second Breakdown in Power Transistors Due to Avalanche Injection” Trans. Electron Devices, Vol.ED-23, No.8, 1976
Reverse Base Bias: central current crowding
Second Breakdown in a BJT Avalanche injection
Impact ionizzation
At increasing Drain current
Second Breakdown in IGBT
Vcc=1800V,
Ic=4000A
RON=ROFF=0.35Ω
Lload=100µH
T=145°C
IGBT Modules rated at 3300V-1200A
Second breakdown along the voltage rise
Internal structure of a power IGBT module
IG1
IC1
Failure mechanisms under extreme electrical stresses In the device linear operating region
Short circuit operations (or overload conditions) Instable DC operations (second breakdown)
During the device switching At the turn-off Reverse recovery of internal diode
In avalanche conditions UIS – Unclamped Inductive Switch
Reverse Recovery of Power MOSFET internal diode
Test circuito Experimental waveforms
Second Breakdown
Parassitic BJT Activation
Drain
P +
N +
P _
GateSource
N_
Body
N+
Collector
Emitter
Base
RP+
Supplemental Charge
The role of gate capacitance in the activation of parasitic BJT
Displacement current
Phenomenon insight
53
Displacement current
Failure mechanisms under extreme electrical stresses In the device linear operating region
Short circuit operations (or overload conditions) Instable DC operations (second breakdown)
During the device switching At the turn-off Reverse recovery of internal diode
In avalanche conditions UIS – Unclamped Inductive Switch
UIS Typical waveforms
Safe Unclamped Inductive Switch
Test circuit Typical waveforms
Failure of Power MOSFET in Unclamped Inductive Switch
UIS – Failure Waveforms
The gate is Off
The current Is still
flowing
The voltage suddenly falls down
A. Icaza-Deckelmann, G. Wachutka, J. Krumrey, F. Hirler, “Failure Mechanism of Power DMOS Transistors under UIS Stress Conditions,” ASDAM 2002 - EDSSC ‘03
Failure of Power MOSFET in Unclamped Inductive Switch
Electric Field
Avalanche multiplication
Low Drain Current
Ih
Ie
Ih
Ie
Failure of Power MOSFET in Unclamped Inductive Switch
Maximum temperature in the device and currents at the source contact vs. time
(I=1mA)
Ih
Ie
Failure of Power MOSFET in Unclamped Inductive Switch
Ih
Ie
Avalanche multiplication
Electric Field
Kirk Effect
High Drain Current
Experimental evidence of instability in UIS
Outline Introduction Failure mechanisms under extreme electrical stresses:
In the device linear operating region During the device switching In avalanche conditions External causes (eg. cosmic rays impacts)
Non destructive techniques for electrical testing of power devices Motivation of ND techniques The basics of ND techniques taken from the literature The approach to ND SOA characterization of power module Typical problems related with the use of Non Destructive tester The “precursors” of the instabilities
Cosmic rays impact Conclusions
ND techniques for electrical testing of power devices Motivation of ND techniques
The basics of ND techniques taken from the literature
The approach to ND SOA characterization of power module
Typical problems related with the use of Non Destructive tester
The “precursors” of the instabilities
Collector Voltage
Col
lect
or V
olta
ge
RBSOA
How a Non-Destructive tester can help in robustness validation tests?
Gate Resistance
Failu
re V
olta
ge
How a Non-Destructive tester can help in robustness validation tests?
ND techniques for electrical testing of power devices Motivation of ND techniques
The basics of ND techniques taken from the literature
The approach to ND SOA characterization of power module
Typical problems related with the use of Non Destructive tester
The “precursors” of the instabilities
The first non destructive tester proposed in the literature
D.W. Berning: “Semiconductor Measurement Technology: A programmable Reverse Bias Safe Operating Area transistor tester” National Institute of Standard and Technology Special Publication 400-87, August 1990.
Basic schematic and principle of operation
The first non destructive tester proposed in the literature
Tester performances: VClamp,Max=2000V IC,Max=25,5A tCrowBar,on=30ns @ ICrowBar=40A tCrowBar,on=65ns @ ICrowBar=100A
Problems with vacuum tubes
• Current capabilities of vacuum tubes are quite poor and their use for collector currents larger than 50A is not practical
• Tubes are very expensive
A second version of ND tester
G. Carpenter, F.C.Y. Lee, D.Y. Chen: “An 1800-V 300-A N Nondestructive Tester for Bipolar Power Transistors” IEEE Transaction on Power Electronics, vol.5, n°3, pp. 314-322, 1990.
The MOSFET shunt circuit of the Carpenter non-destructive tester
Tester performances: VClamp,Max=1800V IC,Max=300A tCrowBar,d=10ns @ ICrowBar=300A tCrowBar,on=300ns @ ICrowBar=300A
Problems when using dV/dt sense to trigger the crow bar switch A delay time is observed which is sensitive to
the output current value The methods to speed up the turn on of the
power MOSFET cannot be easily extended to high power IGBTs when used as crow bar switches
The instabilities of the power devices often are not accompanied by a sudden variation of the collector voltage
Other techniques must be used for the activation of the Crow Bar switch!
ND techniques for electrical testing of power devices Motivation of ND techniques
The basics of ND techniques taken from the literature
The approach to ND SOA characterization of power module
Typical problems related with the use of Non Destructive tester
The “precursors” of the instabilities
Non Destructive Tester for High Power Modules
Q 1 , 1
V 0-8000V
I N
H.V. Power Supply
Series SW
Crow-Bar SW
V 100V
A U X
Q 1 , 2
Q 1 , 3
Q 2 , 1
D 2 D 3
D 1 L L
C
DUT
R 1
Q 2 , 2 Q 2 , 3
No dV/dt sense is used
Short circuit High temperature test
High voltage Unclamped tests
1Series SW state
DUT Collector Voltage
0
T2T3T4 TimeT1T0
Crow bar SW state
DUT Collector Current
DUT state
Crow bar Current
Tester Operations
VIN
Series SW
Crow-Bar SW
D2 D3
D1LL
C
DUT
ND techniques for electrical testing of power devices Motivation of ND techniques
The basics of ND techniques taken from the literature
The approach to ND SOA characterization of power module
Typical problems related with the use of Non Destructive tester
The “precursors” of the instabilities
Jitter of the crow bar turn-on
Stray capacitance of the crowbar switch
The effect of the stray inductance on the crow-bar turn-on time
The effect of the crow-bar Reverse Bias Voltage
Reverse Bias Voltage
Improved NDT
Other effects of stray inductances
LS=110nH 3300V – 1200A IGBT Module
600V – 300A IGBT Module
Different typologies of non-destructive set-up High voltage IGBTs
1700V ≤ Vcc ≤ 6500V Ls = 110 nH
High current IGBTs Vcc ≤ 1700V Ls < 50 nH
The stray inductance 600V – 300A IGBT Module
LS=110nH
LS=50nH
Stray capacitance of the busbar
Ls = 110 nH Ls < 50 nH
CSBB = 5nF CSBB = 12nF
Stray capacitance of the busbar
1200V – 25A IGBT
CSBB = 12nF
Reduced busbar stray capacitance
CSBB = 400pF
Reduction of the busbar stray capacitance
1200V – 25A IGBT CSBB = 12nF
CSBB = 400pF
A first conclusion about the tester characteristics
High voltage IGBT modules (1700V - 6500V)
High voltage series and CB switch High voltage dumping capacitor
Lower current capabilities Higher stray inductances
Lower crow bar stray capacitance
Lower stray inductances Higher dumping capacitors
Larger crow bar stray capacitance
High current series and CB switch High capacitance dumping capacitor
Lower voltage IGBT modules (600V - 1700V)
Discrete IGBT devices
Lower current much faster CB switch Lower voltage more performing
dumping capacitor
Lower current capabilities Much lower stray inductances
Lower stray busbar capacitance
A first conclusion about the tester characteristics
• It is not possible to have one experimental
set up good for any devices/modules
• Each phenomenon to be studied requires its specific experimental
set-up
Clamped test on 1500V-10A DAUX and VNEG=50V
VIN=1200V, Ic=19A, RGOFF =10Ω
Second breakdown
Crow Bar activation
Collector current is zeroed
80ns DUT Saved
Single chip MOS-GTO
Example of Tester Operation
UIS Failure
Unclamped test on 600V - 300A IGBT Modules by HC-UNDT
VNEG=800V, RGOFF=15Ω
Second breakdown
Crow Bar activation
Collector current is zeroed
60ns
DUT Fails
UIS Failure
Ts=25°C
ESB 0.1mJ (Very low energy after failure)
1 20ns
Unclamped Test JFET: Id=21A, Vav1800V, LL=1.5mH
tav 2.5µs Eav=51mJ
Post Failure Analysis
Very small damaged area
Melted area between gate and source
Is it possible to save DUT in UIS?
Other precursors must be identified
We cannot rely on dV/dt and dI/dt Indicators!!!!
Unclamped Turn-off: 1200V-400A Modules
Unclamped turn-off test
Vav=1300V Ic=400A ROFF=3.3Ω LLOAD=50mH
Second Breakdown
Precursor on the gate voltage
TCASE= 25°C
Vav=1300V Ic=400A ROFF=3.3Ω LLOAD=50mH
Unclamped Turn-off: 1200V-400A Modules
Unclamped turn-off test
Precursor on the gate voltage TCASE= 25°C
Vav=1300V Ic=400A ROFF= 3.3 Ω 2.0 Ω LLOAD=50mH
TCASE= 25°C
Unclamped Turn-off: 1200V-400A Modules
Second Breakdown
ROFF=2.0W
ROFF=3.3W
Unclamped turn-off test
OTHER USES OF PULSED POWER SUPPLY
Control of the energy in Avalanche Cycles (application to SiC JFETs)
Automatic Tester Operation
Effects of Avalanche Cycles on SiC JFETs
Failure during
avalanche (400 cycles)
Effects of Avalanche Cycles on SiC JFETs
Gate Leakage
Drain Leakage
Effects of Avalanche Cycles on SiC JFETs
Drain current Effect on drain leakage at fixed URS times (1.2us) and cycles (350).
Conclusions The Non Destructive Tester is a useful tool to test
power semiconductor devices at the edges of the SOA It is not possible to use the same NDT for testing
discrete devices and modules The NDT must be designed according to the
characteristics of the device/modules to be tested The pulsed apparatus can be used for other
applications In high current and in unclamped inductive turn-off a
precursor on the waveform can be recognized that evidence instabilities taking place inside the device and can be used to save the sample under test
Outline Introduction Failure mechanisms under extreme electrical stresses:
In the device linear operating region During the device switching In avalanche conditions External causes (eg. cosmic rays impacts)
Non destructive techniques for electrical testing of power devices The basics of ND techniques taken from the literature The approach to ND SOA characterization of power module Typical problems related with the use of Non Destructive tester The “precursors” of the instabilities
Conclusions Cosmic rays impact
Cosmic rays impact
Neutron flux at sea level is 105 neutrons/cm2-year
with E>2 MeV which may cause SEE in electronics
“… semiconductor failures induced by cosmic radiation are no longer [only] an aerospace problem. Such failure mechanisms must be accounted for in automotive electronics systems design.” www.automotivedesignline.com, June 2006
Particle shower
Galactic Cosmic Ray
Neutron induced Single Event Effects (SEE) A neutron interacts with a nucleus to produce a heavily ionizing secondary that then causes an anomalous macroscopic effect in a working electronic device
IRRADIATION SPECIES
• Protons • Neutrons • Heavy ions
IRRADIATION FACILITIES
• INFN LNS Catania • Tandem (Heavy ions) • Ciclotrone (Protons)
• INFN LNL Legnaro • Tandem (Heavy ions)
• ENEA Casaccia Roma • Tapiro (Neutrons)
SEE irradiation experiments
SEB in a Power Diode
Typical test circuit
DUT
Vbias C
v(t)
i(t)
1MW
50W
50W line
Ion be am
Charge Amplification (2D Simulation)
0
100
200E-
field
[kV/
cm] 300
400
0 50 100 150 200 250 300 350 400 450Depth [ m]m
1013
1014
1015
1016
1017
1018
Ele
ctro
ns-D
ensi
ty [c
m]
-3
T=0
25ps 100ps 150ps
230ps
500ps 1ns
P+ N+ N-
4kV diode
Biasing voltage:
1800V
Impacting Ion:
12C (17MeV)
G. Soelkner et al. “Charge Carrier Avalanche Multiplication …”, IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 47, NO. 6, DECEMBER 2000, pp. 2365 - 2372
-5
0
5
10
15
20
25
30C
urre
nt [
A]
-40 0 40 80 120 160 200 240Time [ns]
100ns
Diode Currente during a destructive impact
Biasing Voltage: 2200V
2D Simulation a SEB
0 50 100 150 200 250 300 350 400 450Depth [ m]m
25ps50ps 75ps 100ps125ps
150ps
300ps
1013
1014
1015
1016
1017
1018
E-D
ensi
ty [c
m]
-3
0
100
200
E-fi
eld[
kV/c
m] 300
400
0 50 100 150 200 250 300 350 400 450Depth [ m]m
25ps50ps 75ps 100ps 125ps
150ps
4kV diode
Biasing voltage:
2200V
Impacting Ion:
12C (17MeV)
A double injection like phenomenon
0
100
200
E-fi
eld[
kV/c
m] 300
400
0 50 100 150 200 250 300 350 400 450Depth [ m]m
1013
1014
1015
1016
1017
1018
E-D
ensi
ty [c
m]
-3
Impact ionization
High: Current density
Carriers concentration
Electric field
SEGR SEB
SEE in Power MOSFET
Test circuit
Computer for off-line statistical analysis
Oscilloscope
GPIB
Single Event Burn-out
3D finite element simulations are performed using the ATLAS TCAD simulation tool by Silvaco International.
The simulated elementary cell with its lumped elements and the parameters used to simulate the ionizing track for a bromine ion at 230MeV.
c
0c
t
tt
R
r
t
terfcπt
eeNtr,Q
2
c
0
32 cm
pairs
Rπ3.6
LETN
R 0.124mm
t0 4ps
tC 2ps
SEE in power mosfets Numerical simulation activity
The role of the parasitic BJT in the charge generation mechanism.
Vds=60V Vds=100V (SEB)
SEE in power mosfets 3D numerical simulation
The double injection phenomenon.
SEE in power mosfets 3D numerical simulation
Double injection in Power MOSFET
0 5 10 15 20 25 30 350
0.5
1
1.5
2
2.5x 10
17
Distance [ mm ]
Hole
Concentr
ation [
cm
-3 ]
Hole Concentration t=700ps x=9mm
Vds=100V
Vds=60V
Impact ionizzation
0 5 10 15 20 25 30 350
0.5
1
1.5
2
2.5x 10
5
Distance [ mm ]
Ele
ctr
ic F
ield
[ V
/cm
]
Electric Field t=700ps x=9mm
VDS=60V
VDS=100V
SEGR
SEE in gate oxide of Power MOSFET
Single Event Gate Rupture
A conceptual model for SEGR
J. R. Brews, et. Al. “A Conceptual model for SEGR in Power MOSFET’s,” IEEE TRANS. ON NUCLEAR SCIENCE, VOL. 40, NO. 6, DECEMBER 1993
Ion track
SEB in IGBTs
Collettore
N+
P +
P +
P _
GateEmettitore
N_
Collettore
N+
P +
P +
P _
GateEmettitore
N_
Anode
Kathode
Gate
RP+
SEB in IGBTs
2D simulation of SEB in IGBT
W. Kaindl, et. Al. “Cosmic Radiation-Induced Failure Mechanism of High Voltage IGBT,” Proc. of the 17th ISPSD, May 23-26, 2005, Santa Barbara, CA
Thank You for Your attention
DIEI – Università di Cassino e del Lazio Meridionale Via G. di Biasio, 43, Cassino, Italy
e-mail: [email protected]