Integration-Level Testing of Sub-Nanosecond Microchannel ...
Nanosecond Risetime Pulse Characterization of SiC p n ... · Nanosecond Risetime Pulse...
Transcript of Nanosecond Risetime Pulse Characterization of SiC p n ... · Nanosecond Risetime Pulse...
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
Nanosecond Risetime Pulse Characterization of SiC p+n Junction Diode Breakdown and Switching Properties
Philip G. Neudeck
NASA Lewis Research CenterCleveland, OH USA
International Conference on Silicon Carbide,III-Nitrides and Related Materials
September 1-5, 1997Stockholm, Sweden
Chris FaziU.S. Army Research Laboratory
Adelphi, MD USA
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
Acknowledgments
NASA LewisDavid Larkin
J. Anthony PowellLuann Keys
Andrew TrunekJohn Heisler
Bruce ViergutzGene Schwarze
Jan NiedraGlenn Beheim
W. Dan Williams
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
Outline
Pulse testing reveals very important SiC device behaviors not observedby conventional DC and RF testing.
Reverse bias diode pulse testingStable and unstable SiC reverse breakdown.
Forward bias diode pulse testing
These behaviors directly impact SiC power device performance & reliability.
Rectifier reverse recovery switching transients.
Perimeter-governed device minority carrier lifetimes.
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
HVDCPowerSupply
1/2" SemirigidCoax TransmissionLine (~150 ft)
1 MΩHg GasSwitch
3.8X atten.
RG-58Coax(50 Ω) Digital
Oscilloscope
Channel 1V(t), Trigger
Channel 2I(t)
5X atten.
TektronixCT-2/P6041 CurrentProbeDevice
UnderTest
10 µF(1 kV)200 Ω
or 400 Ω
2X atten.
DCSupply
(250W 0 µH)
+
+
-
-
Pulse Test Circuit
Bias pulse is formed by discharge of semirigid coaxwhen Hg switch is momentarily triggered.
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
V = 0 V V = 0 V
A Tale of Two Diodes(Part 1: DC Testing)
Wafer A* Wafer B**
Epitaxial Small-Area 4H-SiC p+n Diodes
VDC BKDN = 140 V VDC BKDN = 142 V
* NASA Lewis Run #1841J. Appl. Phys. 80, p. 1219
** NASA Lewis Run #1905IEEE EDL 18, p. 96
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
A Tale of Two Diodes(Part 2: Reverse Bias Pulse Testing)
Wafer A(VDC BKDN = 140 V)
Wafer B(VDC BKDN = 142 V)
0
100
200
300
Vol
tage
(V
)
0
1
0 100 200 300 400 500Time (ns)
Cur
rent
(A
)
Shot #2
Input PulseAmplitude = 116 V
Experiment: Subject devices to single-shot reverse-bias pulses of increasing amplitude until catastrophic breakdown failure occurs.
0
50
100
Vol
tage
(V
)
0
1
-100 0 100 200 300 400 500 600
Cur
rent
(A
)
Input PulseAmplitude = 83 V
(b) Shot #2
Time (ns)
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
A Tale of Two Diodes(Part 2: Reverse Bias Pulse Testing)
Wafer A(VDC BKDN = 140 V)
Wafer B(VDC BKDN = 142 V)
Time (ns)
Catastrophic Device Failure,Device Physically Destroyed!
0
1
-100 0 100 200 300 400 500 600
Cur
rent
(A
)
0
50
100
Vol
tage
(V
) Input PulseAmplitude = 94 V
(c) Shot #3
Device Still Good,Positive Temp. Coeff. Breakdown!
0
100
200
300
Vol
tage
(V
)0
1
2
3
4
0 100 200 300 400 500C
urre
nt (
A)
Time (ns)
Shot #22
Input PulseAmplitude = 322 V
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
Pulse Breakdown Discussion
Behavior of devices on Wafer A is unacceptable for many power applications.• Extremely high reliability, immunity to “glitches” required for most
aerospace applications.
Differences between “unstable” Wafer A and “stable” Wafer B:
• Single epi-growth (Wafer B) vs. two-step epi growth (Wafer A).
• n-substrate (Wafer B) vs. p-substrate (Wafer A).
• SIMS revealed excess Al, N near Wafer A junction not present in Wafer B.
Exact physical mechanism still uncertain.
• Bulk failure mechanism - no evidence of surface breakdown.
Positive temperature coefficient breakdown observed only on very small-area (A < 1 x 10-4 cm2) Wafer B devices.
• Elementary (1c) screw dislocations affecting breakdown???
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
PN Diode Reverse Recovery*
Idealized Test Circuit Diode Reverse Recovery Current Transient
(zero inductance)
ts = Storage Time
Minority carrier (hole) lifetime τp
related to storage time ts by:
* G. Neudeck, The PN Junction Diode, 2nd Ed., Addison-Wesley Publishing, p. 111.
ts = τ p erf– 1 1 + 1
IR / IF
2
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
Reverse Recovery Current TransientsDevice Area = 8.1 x 10-3 cm2, Rs = 200 Ω
IF varied for approximately fixed IR IR varied for fixed IF
ts increases as IF increases. ts decreases as IR increases.
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
-20 0 20 40 60 80 100
Cur
rent
(A
)
Time (ns)
0.7 A0.6 A0.5 A0.4 A0.3 A0.2 A
IF(t=0-)V
R = 30 V
-1.5
-1.0
-0.5
0.0
-20 0 20 40 60 80 100C
urre
nt (
A)
Time (ns)
VR = 40 V
VR = 30 V
IF(t=0 -) = 0.65 A
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
Storage Time (ts) Dependence on IR/IF
10-9
10-8
10-7
10-6
0.1 1 10
Sto
rage
Tim
e t s
(sec
onds
)
IR/I
F
where τ p = 300 ns
ExperimentallyMeasured t
s Data
ts = τ p erf – 1 1 + 1
IR / IF
2
Experimentally measuredstorage time behaviorfollows predicted physicaltheory.
Effective minority carrier lifetime for this device is300 ns (A = 8.1 x 10-3 cm2)
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
Storage Times (ts) of Larger vs. Smaller Devices
Effective minority carrier lifetime decrease with decreasing area suggestspresence of significant perimeter surface recombination effects.
10-9
10-8
10-7
1 10
A = 3.1 x 10-4 cm2
p = 80 nsτ
A = 8.1 x 10-3 cm2
p = 300 nsτ
t s Sto
rage
Tim
e (s
ec)
IR / I
F
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
Device Hole Recombination =
Top View of Diode
n- Layer
Side View of Diode
Perimeter HoleRecombination
Bulk HoleRecombination
p+ Layer
p+n Diode Effective Lifetime
τp Eff. = τp extracted fromreverse recovery switchingmeasurement ts vs. IR/IF data.
Can estimate τp Bulk and sp Perim. from linear plot of 1/τp Eff. vs. P/A.
REff .A = RBulkA + RPerim.P∆pn
τ p Eff .A ≈ ∆pn
τ p BulkA + s p Perim.∆pnP
1τ p Eff .
≈ 1τ p Bulk
+ s p Perim.PA
y = b + mx
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
The bulk minority carrierlifetime inherent to this SiCepilayer is much longerthan the average lifetimemeasured on asmall-area device. This is due to largeperimeter surfacerecombination.
0
5
10
15
20
0 50 100 150 200 250 300
1/ p
(µse
c-1)
Perimeter/Area (cm-1)
τ
133160200266400800∞
Device Diameter (µm)
τp Bulk ≈ 0.7 µs
Bulk Minority Carrier Lifetime Extraction
1τ p Eff .
≈ 1τ p Bulk
+ s p Perim.PA
y = b + mx
(4H-SiC, ND = 2 - 4 x 1016 cm-3)
τp Bulk ≈ 0.7 µs
1τ p Eff.
≈ 1τ p B u l k
+ sp Perim.PA
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 50 100 150 200 250 300
1/t
s (
nsec
-1)
P/A (cm-1)
JF ≈ 1 kA/cm2
JR ≈ 2 kA/cm2
133160200266400800∞
Device Diameter (µm)
Storage Times at Constant Current Density
Indicates bulk Auger recombination insignificant compared to perimeter-governed SRH recombination.
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
Discussion
This work demonstrates by example that perimeter surface recombination cansignificantly impact SiC bipolar device electrical characteristics viareduced effective minority carrier lifetimes.
• Greater impact on smaller (IC) devices than larger (power) devices.
• Lifetime reduction likely to be exacerbated by “multi-finger” or “multi-cell”geometries that increase effective perimeter-to-area ratio.
• Possible contributing factor to experimental observations of:
- Low current gains (< 20) in SiC BJT’s produced to date.
- SiC pn diode current densities below theoretical predictions.
- Fast switching response of SiC pn diodes and thyristors.
Development and optimization of appropriate SiC surface passivation andjunction termination technologies could reduce or eliminate lifetime-limitingrole of surface recombination in SiC bipolar devices.
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
• Potential impact on n- or p-type 4H- and 6H-SiC at all doping densities (?).
• Effect present in ion implanted or heavily compensated SiC junctions?
Discussion (cont.)
Figure fromJanzen & Kordina,ICSCRM-95 p. 657.
τp Bulk = 0.7 µsLinkoping U.6H-SiCPL Decay Data
A = 8.1 x 10-3 cm2
A = 3.1 x 10-4 cm2
NASA4H-SiC
National Aeronautics andSpace AdministrationLewis Research Center
INSTRUMENTATION & CONTROLTECHNOLOGY DIVISION
PGN9/97
Summary
Pulse testing reveals very important device behaviors not observed by conventional DC and RF testing.
• reliability
• switching speed
• current (density) rating
Observed behaviors directly impact SiC power device
Pulse testing should play an important role in SiC power device development and qualification.