4H-SiC defects
Nicolò Piluso
Outline 2 - Introduction
Overview 4H-SiC (growth, properties, application) - Defects
Bulk Epitaxial layer
- Characterization methods
Photoluminescence Raman Spectroscopy DLTS µ-PCD KOH molten Optical microscopy, TEM, SEM Scattered Light
4H-SiC lattice (Wurtzite unit cell) Siatom
Catom
Introduction
at 300K Si GaAs 4H-SiC GaN
Eg(eV) 1.12 1.4 3.2 3.4
Ec (MV/cm) 0.25 0.3 2.2 3
µn (cm2/Vs) 1350 8500 1000 1000
εr 11.9 13 10 9.5
Vsat (cm/s) 1x107 1x107 2x107 3x107
λ (W/cmK) 1.5 0.5 3-5 1.3
Semiconductor properties
3
Some devices are pushing Si boundary SiC and GaN offer promise for improvement
Slide from Prof. S. E. Saddow
Applications of Power Devices
Introduction 4
Growth Methods: PVT - Bulk (thick, high doping) CVD - Epitaxy (thin film, mid-low doping)
PVT
CVD
cutting
Ingot - Bulk
Epitaxy
5
From Ch3, Kimoto-Cooper Book
Micropipe defects are indeed located at the center of a large spiral on the surface of the SiC boule, and
the diameters of the pinholes range from 0.5 µm to several micrometers
Micropipe
TSD TED-BPD
Bulk Defects
6
> C/Si Ratio > Polarity (Si or C face) > Hexagonality of the polytype - 4H-SiC (0.5 Hex), more stable in C-rich ambient - 6H-SiC (0.33 Hex)
Polytype mixing during boule growth ! switching, coalescence, nucleation !
Occurrence of spiral growth around TSD that is generated to remedy to
mismatch of polytype
7
BPD generation during boule growth
A doping variation affects the BPD generation due to the lattice
parameters variation.
Most TEDs are formed by conversion from BPDs
along the growth direction during growth.
A stress reduction could reduce the BPD generation
8
From Ch5,p144 Kimoto-Cooper Book
TSD generation and removal TSD
BPD TED
BPD
Vendor Value EPD = TSD+TED+BPD
9 Defects reduction (Bulk)
10
From Ch5,p161 Kimoto-Cooper Book
Epi Defects
From Ch4,p97 Kimoto-Cooper Book
11
CVD process Si-rich ! Carrot
C-rich ! Triangle
Epi Defects
Extended epi defect Current Density (cm-2) Down Fall 0.1
Triangular defects / Comets 0.2
Carrots / Pit 0.5-1
Stacking Faults 1
Dislocations Vs Epi defects 12
0
10
20
30
40
50
60
70
80
90
100
0 5000 10000 15000 20000 25000 30000
Carrots
0 5000 10000 15000 20000 25000 30000
0 20 40 60 80
100 120 140 160 180 200
3500 3700 3900 4100 4300 4500 4700 4900 5100
Stacking Faults
0 4000 8000 12000 16000 20000 24000 28000 32000
PVT growth direction
ingo
t
seed
EPD
EPD
High density
Low density
The spread observed is mainly due to uncorrect EPD value declared. EPD decrease along the bulk growth. The EPD value is, typically, detected by etching a wafer within or on the edge of the ingot. This method produce a reference value for all the ingot, but, truly speaking, it is only an approximate value for all the wafers.
EPD=TSD+TED+BPD
Epi defect: Step bunching and roughness 13
-Step Flow -Epitaxial process parameters -No relevant effects on device performances - Crucial growth parameters: Temperature, growth rate, Si/C
AFM
Rq < 1nm
Stacking Faults 14
4H- SiC
SF
Band gap diagramme
4H- SiC
15 Impacts of epi defects on devices Current
Density (cm-2) Defect SBD MOSFET, JFET Pin, BJT, Thristor,
IGBT
~ 0 Micropipe VB reduction (by 50-80%)
500 TSD (without pit) NO NO NO, but can causes local reduction of carrier lifetime
3000
TED (without pit) NO NO NO, but can causes local reduction of carrier
lifetime
1000
BPD (including interface
dislocation, half loop array)
NO, but can cause degradation of MPS Diode
NO, but can cause degradation of body
diode
Bipolar degradation (increase of on-
resistence and leakage current)
0.01 - 1 In-grown SF VB reduction (by 20-50%) V
B reduction (by 20-50%)
VB
reduction (by 20-50%)
0.1 - 1 Carrot, triangular defect V
B reduction (by 30-70%)
V
B reduction (by 30-70%)
VB
reduction (by 30-70%)
0.1 - 1 Down fall VB reduction (by 50-90%)
V
B reduction (by 50-90%)
VB
reduction (by 50-90%)
From Kimoto-Cooper book, Fundamentals of Silicon Carbide Technology, 2014
Spatially resolved micro-photoluminescence and micro-Raman setup
• Confocal apparatus • Grating monochromator
• XY motorized stage
• CCD detector
PL or Raman map
325 nm UV laser 633 nm Vis laser
PL or Raman signals
x
y
XY motorized stage
Sample
Objective
Evaluations of: - Crystallographic defect (PL and Raman) - Doping (PL and Raman) - Stress (Raman) - Polytype inclusion (PL and Raman)
17
Stacking Faults Doping
-40000 -20000 0 20000 40000
775,6
776,0
776,4
776,8
777,2
777,6
Stress free value
TO R
aman
Shi
ft (c
m-1
)
Distance (µm)
Vendor B Vendor C BridgestoneVendor A
Stress
-30 000
-20 000
-10 000
0
10 000
20 000
30 000
Y (µ
m)
x103-20 0 20X (µm)
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
4000 µm
0
200
400
600
800
1 000
1 200
1 400
1 600
1 800
2 000
2 200
2 400
2 600
Inte
nsity
(cnt
)
750 760 770 780 790 800 810 820Raman Shift (cm-1)
798
.0
767
.6
789
.5
0
50
100
150
200
250
300
350
400
450
500
550
600
650
Inte
nsity
(cnt
)
740 750 760 770 780 790 800 810Raman Shift (cm-1)
776
.8
0
200
400
600
800
1 000
1 200
1 400
1 600
1 800
2 000
2 200
2 400
2 600
Inte
nsity
(cnt
)
730 740 750 760 770 780 790 800 810 820Raman Shift (cm-1)
797
.6
770
.4
777
.3
786
.4
6H
4H 4H+6H
(+15R ?)
Polytype inclusion analysis (HeNe)
Transverse optical mode (TO)
19
Im
Vc
Sii
Ci
VSi
" Intrinsic (interstitial, vacancy, antisite) Sii,Ci, Vsi, Vc, Csi, etc
" Extrinsic (impurity) " Complex
carbon
silicon
[000
1]
Point Defects
Perfect lattice Point defects
As
Energy levels of major deep levels observed in as grown n-type and p-type 4H-SiC epitaxial layers
EH6/7 is a carbon vacancy
Also, the origin of the Z1⁄2 center was identified as the acceptor
levels of a carbon vacancy
After Annealing at 1600°C, Z1/2 and EH6/7 remain important levels
Z1/2 = Ec-0.63eV EH6/7 ~ Ec/2 eV
µ-PCD and DLTS methods 20
Reducing of the on-resistance in bipolar devices
21
Implantation process produces a damage on the lattice crystal. By performing an annealing at high temperature (T > 1600 °C) the crystal
is partially recovered. However, an agglomeration of defects is still present and
observed by PL characterization.
An optimization of Ion dose and Ann. temperature is on going
Defects from implantation
0,00E+00
5,00E-01
1,00E+00
1,50E+00
2,00E+00
2,50E+00
3,00E+00
350 400 450 500 550 600 650
Inte
nsity
Wavelength (nm)
Defect @ 470nm (2.6eV) Gap
Not implanted
Implanted + HT Decreasing the source dose we observe a large decrease of the
point defects peak observed by PL
1.5 2.0 2.5 3.0
0.000
0.001
0.002
0.003
0.004
0.005
0.006
Inten
sity (
a.u.)
Energy (eV)
body (1650°C) body (1700°C) body (1750°C)
Increasing the annealing temperature the peak at 1.6 eV (EH6/7) decreases and the peak at 2.6 eV (Z1/2) increases
22 Candela tool (scattered light)
23
100 µm
Triangles
Carrots
Stacking Faults Macrostep Particle
Particle
EpiDefects
EPI 4H-SiC defects (Candela CS920 Images)
Conclusions 24
-T. Kimoto, Japanese Journal of Applied Physics 54, 040103 (2015)
-T. Kimoto, J.A. Cooper, FUNDAMENTALS OF SILICON CARBIDE TECHNOLOGY, 2014 John Wiley & Sons Singapore Pte. Ltd.
-G. Feng, et al. Physica B 404 (2009) 4745–4748
-M.B.J. Wijesundara and R. Azevedo, Silicon Carbide Microsystems for Harsh Environments
-F. La Via, Silicon Carbide Epitaxy, Research Signpost 2012
-Owner studies
References
" Continuous quality improvement
" Wide choice in characterization techniques
" Different quality from different vendors
" Non destructive VS destructive characterization methods (dislocation density)
Non destructive (PL) Destructive (KOH etch)