wide bandgap semiconductor research at mississippi state ... · bharat krishnan, hrishikesh das,...
Post on 05-Aug-2020
0 Views
Preview:
TRANSCRIPT
Wide Bandgap Semiconductor Research
at Mississippi State University
Dr. Yaroslav Koshka
Associate Professor
Department of Electrical and Computer EngineeringMississippi State University
Presentation Outline
1. Why SiC
2. History of SiC program at MSU
3. Advancements in SiC epitaxial growth
4. Recombination Induced Defect Reactions involving hydrogen in SiC.
1. Why SiC?
2. History of SiC program at MSU
3. Advancements in SiC epitaxial growth
4. Recombination Induced Defect Reactions involving hydrogen in SiC.
SiC Material Properties(comparing to other semiconductors)
(peak) (peak)2
(pick)12.5(pick)22Saturation drift velocity [105m/s]
8”12”n/a3”4”Commercially available substrates
DIDIIDirect/Indirect Bandgap0.40.3~ 3> 2.4~ 3Critical electric field [MV/cm]12.811.89.0109.7Dielectric constant
65551224~ 90081 (Π)
405 (⊥)980 (Π)865 (⊥ )
Low field electron mobility (300ºK, 1e16cm-3) [cm2/V*s]
1.41.13.393.03.26Bandgap [eV]
0.551.31.34.94.9Thermal conductivity [W/cm*K]
GaAsSiGaN6H-SiC4H-SiCProperty\Material
Drift region resistance
Heavily doped n-type substrate
Drift region
Schottky contact
Cathode contact
Tdrift
Drift region
Heavily doped p-type substrate
DrainSource
P-type epitaxial layer
Sourcecontact
Gatecontact Drain
contact
Body contact
Tdriftdrift
drift spn drift
TR
q Nµ=
P-type anode
Heavily doped n-type substrate
Drift region
Anode contact
Cathode contact
Tdrift ( )drift
drift spn p F a
n driftdrift
TR J
q NT
µ µ τµ
=+
+
Drift region blocking capability(1-D approach)
1 1D
dE x q N xdxϕ ρ
ε ε= = − =
1crit Drift DriftE q N T
ε=
01DriftT 2DriftT
1DriftN 2DriftNdriftT
assuming free carrier concentration to be negligible (no leakage current)for non-”punch-through” layer:
0
( )DriftT
BlockingU E x dx= ∫Voltage drop on drift region is given by:
( ) ,div ε ϕ ρ∇ = −solve the Poisson equationTo find ( ),E x
Drift region blocking capability(1-D approach)
1 1D
dE x q N xdxϕ ρ
ε ε= = − =
01DriftT
driftT
assuming free carrier concentration to be negligible (no leakage current)for non-”punch-through” layer:
0
( )DriftT
BlockingU E x dx= ∫Voltage drop on drift region is given by:
( ) ,div ε ϕ ρ∇ = −solve the Poisson equationTo find ( ),E x
Ecrit 2
Ecrit 1
Ndrift 2 << Ndrift 1
Tdrift 2 >> Tdrift 1
1crit Drift DriftE q N T
ε=
Specific on-resistance of SiC
400x
Advantages of wide bandgap
100
101
102
103
102 103 104
Spe
cific
On-
Res
ista
nce
(mž
cm2 )
Blocking Voltage (V)
Silicon
4H-SiC
(mΩ
cm2
) VB2/ RON = constantVB
2/ RON = constant
• Resistance-Area productof 4H-SiC is 400x lowerthan silicon at any givenblocking voltage
Advantages of wide bandgapIntrinsic carrier concentration
Example:n-type, ND = 1015 cm-3
Advantages in Switching Characteristics
Si pin diode – 1.32 µs SiC Schottky diode – 72 ns
600 V diode recovery comparison
Presentation Outline
1. Why SiC
2. History of SiC program at MSU
3. Advancements in SiC epitaxial growth
4. Recombination Induced Defect Reactions involving hydrogen in SiC.
1. Why SiC?
2. History of SiC program at MSU
3. Advancements in SiC epitaxial growth
4. Recombination Induced Defect Reactions involving hydrogen in SiC.
History of Wide Band Gap Semiconductors at Mississippi State University
Materials - Devices- Circuits - Systems
MESFET Fab at GE
0 5 10 15 200
200
400
600
800SiC161, Wafer A1x100 µmVg=0V, Vg step=-5V
I DS (m
A/m
m)
VDS (V)
Emerging Materials Research Laboratory
EMRL
Emerging Materials Research Laboratory
EMRL
0.3 µm n channel
Semi-insulating substrate
0.55 µm p- buffer
ONR N00014-01-C-0167
C. Wood
ONR N00014-01-C-0167
C. Wood
SiC Epi at MSU
Testing at GE & MSU ADPC
SMDC DASG60-00-C-0074
L. Riley
SMDC DASG60-00-C-0074
L. Riley
Power Electronics Prototyping / Power
Component Test BedPEP/PCT
Power Electronics Prototyping / Power
Component Test BedPEP/PCT
Prototype Power
Electronics at MSU
Modify HMMWV at
MSU
Sentinel Demo by
MSU/Guard/SMDC
Commercial Packaging
Miss. Center for Advanced Semiconductor Prototyping
MCASP
Miss. Center for Advanced Semiconductor Prototyping
MCASP
AFRL/MDA F33615-01-D-2103
J. Scofield
AFRL/MDA F33615-01-D-2103
J. Scofield
SiC SBD Design
SBD Fab at MSU
Leveraging Basic Research into Production -Rapid, economical transfer of government sponsored technology development into systems
LAM 9400 EtcherAIXTRON SiC Reactor
• Multi Wafer SiC Epi• Sub-micron lithography• Multi Wafer Plasma Etching• Multi Wafer PECVD• Multi Wafer Metal Deposition
MCASPMississippi Center for Advanced
Semiconductor Prototyping
EMRL• SiC CVD Epi Research
• Defect Engineering
• Materials Characterization
• Device Design
Component Production• Mississippi Small Business
• SBIR’s
www.ece.msstate.edu/mcasp
Epitaxial Process Simulation
SiH4
Si(g)
Material Science Program at MSU
Novel epitaxial growth techniques
Defects and Impurities. Defect Engineering in WBG Semiconductors
SiC for Nanotechnologies
PL DLTS
SEM FTIR
…
Traditional Material Characterization
ONR, AFRL, SBIR/STTR…
Novel Non-ContactMaterial Characterization
Techniques
Presentation Outline
1. Why SiC
2. History of SiC program at MSU
3. Advancements in SiC epitaxial growth
4. Recombination Induced Defect Reactions involving hydrogen in SiC.
1. Why SiC?
2. History of SiC program at MSU
3. Advancements in SiC epitaxial growth
4. Recombination Induced Defect Reactions involving hydrogen in SiC.
Lower-temperature epitaxial growth of 4H-SiC
using CH3Cl carbon gas precursor.
Huang-De Lin, Galyna Melnychuk, Yaroslav Koshka
1 Mississippi State University, Box 9571, Mississippi State, MS 39762, USA
The hot-wall CVD reactors.
Traditional Precursors:SiH4 and C3H8
• Growth temperatures >15000C• Lower temperatures => morphology degradation => polycrystalline
Silane decomposition
SiH4
SiH2
Si<g>
CH3Cl
CH4
C2H4
Cl
Chloromethane decomposition
Simulated kinetics of chemical reactions of SiC epitaxial growth
in the cross-section of the CVD reactor
Our new model – vapor-phase formation of Si droplets (clusters)
Dense silicon-droplet
cloud
Growth at Lower Temperatures
13000C
View of the susceptor from the rear-port window
• Si vapor condensation (cloud) is detrimental for epiquality and growth rate.
High Si/C ratioLow Si/C ratio
50 µm
Optimized
Surface morphology of the 13000C growth
2 µm
>2 µm/hr
J. of Crystal Growth Patent Pending
0
0.5
1
1.5
2
2.5
0 10 20 30
Gro
wth
rate
, 1/h
rCH3Cl flow
TGrowth = 1300 0C
SiH4 flow, sccm
Exponentialfitting
Degraded morphology
Growth rate dependence on silane flow @13000C
The saturation of the growth rate at higher SiH4 flows is due to
silicon vapor condensation in clustersreducing the amount of Si available for epitaxial
growth
⎟⎟⎠
⎞⎜⎜⎝
⎛ ><−−∝><
4
44 exp1)(
SiH
SiHSiHRττ SiH4
J. of Crystal Growth
Arhenius temperature dependences: (a) exponential rate coefficient of the SiH4 flow dependence
compared to (b) the growth rate temperature dependence.
⎟⎟⎠
⎞⎜⎜⎝
⎛ ><−−∝><
4
44 exp1)(
SiH
SiHSiHRττ SiH4 (T)
The value of EA is the same for R and τ SiH4 => the growth rate is determined by silicon vapor condensation.
1
10
0.57 0.59 0.61 0.63
1000/T, K-1G
row
th ra
te, u
m/h
r
EA = 0.573 eV
(a)
Gro
wth
rate
R (µ
m/h
r)1000/T, K-1
1
10
100
0.57 0.59 0.61 0.63
1000/T, K-1
Exp
onen
tial r
ate
coef
ficie
nt
(scc
m) EA = 0.576 eV
(b)
τ SiH
4(s
ccm
)
1000/T, K-1
Dense cloud of Si clusters Strongly reduced cloud
(a) No HCl (b) With HCl added
HCl experiment:Rear view of the glowing susceptor
during 13000C epitaxial growth
Selective epitaxial growth of 4H-SiC with SiO2 mask.
Bharat Krishnan, Hrishikesh Das, Huang-De Lin, Galyna Melnychuk, Yaroslav Koshka.
1 Mississippi State University, Box 9571, Mississippi State, MS 39762, USA
Patent Pending
Plasma Plasma
Semiconductor
Mask
Semiconductor
Pattern formation by Etching
CVD CVD
Semiconductor
Mask
Semiconductor
Pattern formation by Selective Epitaxial Growth
SiO2 mask for low-temperature selective epitaxial growth of 4H-SiC (LTSEG)
SiO2mask
• SiO2 Severely degrade at regular growth temperatures (15000C)• Survives at 13000C of our novel low-temperature growth method
SiCsubstrate
SiO2 mask4H-SiC mesas
2 µm
(a) with SiO2 mask
Low-temperature SEG at 13000C:
4H-SiC mesas4H-SiC substrate
2 µm
(b) SiO2 mask removed
>3 µm / hr
Patent Pending
5 µm
[ 0211 ] (off-axis direction)
Edge defects
SEM image of the mesa-lines selectively grown in SiO2window at 13000C.
Lines oriented in different directions reveal the orientation-dependent defect generation.
Patent Pending
Cross-sectional SEM of 30-µm-wide mesa line selectively grown in SiO2 window at 13000C.
3 µm
SiO2defective region
[ ]0211
1 µmEdge facet
SiO2
Patent Pending
Presentation Outline
1. Why SiC?
2. History of SiC program at MSU
3. Advancements in SiC epitaxial growth
4. Recombination Induced Defect Reactions involving hydrogen in SiC.
1. Why SiC?
2. History of SiC program at MSU
3. Advancements in SiC epitaxial growth
4. Recombination Induced Defect Reactions involving hydrogen in SiC.
Plasma hydrogenation.
• Processing in new Inductively Coupled Plasma (ICP) system enabled further improvement of hydrogen
incorporation in comparison to RIE hydrogenation.
• Experimental Reactive Ion Etching (RIE) system → Hydrogen Plasma.
Hydrogen plasma
SiC subjected to Plasma Hydrogenation.Incorporated hydrogen forms stable complexes with defects and impurities. Hydrogen concentration profiles repeat profiles of acceptor concentration.
[1] S. Janson, A. Hallén, M. K. Linnarsson, and B. G. Svensson, Phys. Rev. B 64, 195202 (2001)
→ Stable hydrogen-defect complex in SiC→ Mobile hydrogen atom→ Defect or impurity
Effect of plasma hydrogenation on the concentration of active acceptors (Al).
• Passivation of Al acceptors down to 1015cm-3 and to the depth in access of 2 µm was achieved after 2 hrs of hydrogenation.
1E+15
1E+16
1E+17
1E+18
1E+19
0 0.5 1 1.5 2 2.5
depth, µm
Con
cent
ratio
n, c
m-3
Al (SIMS)
Acceptors before hydrogenation (CV)
Acceptors after hydrogenation (CV)
1019
1018
1017
1016
1015
Ec
Ev Al
H
Al-H
Simple Diffusion
⎟⎠⎞
⎜⎝⎛
∂∂
∂∂
=∂
∂xHD
xtH
x][][
H – hydrogen concentration
tHT
xHD
xtH
x ∂∂
−⎟⎠⎞
⎜⎝⎛
∂∂
∂∂
=∂
∂ ][][][][]][[][ HTTHK
tHT ν−=∂
∂
T – concentration of empty traps.HT – concentration of traps filled with hydrogen.
K – trapping constant.ν – dissociation frequency
Trap-Limited Diffusion
H at interstitial sites H trapped by defect or impurity
-13
-8
-3
2
7
4100 4120 4140 4160 4180 4200 4220
PL in
tens
ity, a
rb.u
n.
S0 R0
H3
Wavelength, Å
4Al 15 K
Hydrogenated
Initial
PL spectra before and after plasma hydrogenation of epilayer moderately doped with Al.
Y. Koshka, M. S. Mazzola, Appl. Phys. Lett, 79(6), 752 (2001)
• Efficient trapping of hydrogen by Al acceptors is confirmed by reduction of Al-BE PL;• Al acceptors are not the only trapping centers for hydrogen. H-related PL lines indicate
simultaneous formation of H complex with Si vacancy (VSi-H ).
• A 4B0 peak previously associated with a bound exciton at the neutral boron is in fat related to a hydrogen-defect complex (possibly, B-H).
PL spectra of B-doped 4H-SiC epilayer:affect of plasma hydrogenation:
p = 8-9 x 1015 cm-3
tepi = 5 µm
Y. Koshka, Appl. Phys. Lett., 82, 3260 (2003).
Optically-stimulated passivation of defects with hydrogen.
Phenomenon of Recombination-Induced Passivation
(RIP)
→ Stable hydrogen-defect complex in SiC
→ Mobile hydrogen atom→ Defect or impurity
Above band-gap light
Al-HB-H4B0
Schematic illustration of Recombination-Induced Passivation.
Certain amount of the incorporated hydrogen does not form stable complexes with defects (so called “free hydrogen”). Recombination-induced defect migration causes formation of various kinds of stable defect complexes.
T = 15 KT < ~ 250 K
VSi-H
Photoluminescence spectra after Regular Hydrogenation and after Recombination Induced Passivation.
• Al-BE PL additionally quenches and disappears under optical excitation at low T in hydrogenated samples.
-2
0
2
4
6
8
10
12
14
16
3800 3825 3850 3875 3900 3925
Al-BE
Q0
4H-SiC15 K
Wavelength, Å
hydrogenated before excitation
after excitation with above band-gap
initial
Contribution from thesubstrate Contribution from
the substratep = 6-8 x 1016 cm-3
tepi = 2 µ m
after excitation with above bandgap light
Changes in 4B0 PL caused by optical excitation at 15K
Recombination-Induced formation of defects responsible for 4B0 PL (possibly Boron-Hydrogen complex).
-0.6
1.4
3.4
5.4
7.4
9.4
3800 3825 3850 3875 3900 3925
PL in
tens
ity, a
rb.u
n. Q0
Al
B-HH1
15K
Wavelength, Å
(2) 10 min of Optical Excitation
(3) 2 hrs of Optical Excitation
(1) Hydrogenated before optical excitation
-20
30
80
130
180
230
3800 3850 3900 3950
PL in
tens
ity (a
rb.u
nits
)
-20
30
80
130
180
230
3800 3850 3900 3950
PL in
tens
ity (a
rb.u
nits
)
-20
30
80
130
180
230
3800 3850 3900 3950
PL in
tens
ity (a
rb.u
nits
)
Wavelength (Å)
as-hydrogenated4 hrs of optical excitation
H1
recovered @ room T
3916 Å
H1
H1
0
50
100
0 5000 10000 15000
Time of excitation (s)H
1 PL
inte
nsity
Optically induced changes of VSi-H (H1 line). The optically induced growth of H1 can not be masked by the concurren
metastable quenching.
Before Optical Excitation
After Optical Excitation
Recombination-induced formation of a strong radiative recombination channel
(VSi-H emission)
1E+16
1E+17
1E+18
1E+19
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
Depth, um
Con
cent
ratio
n, c
m-3
--- No optical excitation
--- After short excitation…
--- After prolonged excitation with above-bandgap light at 15 K
Deuterium concentration in plasma-deuterated B-doped SiC epitaxial layer.
The spot subjected to an optical excitation at 15K shows much deeper deuterium penetration => recombination-induced athermal migration.
~ NB
SIMS profiles
p+
Inversion of the conductivity type under optical excitation in hydrogenated p-type epilayers: strong recombination-induced passivation of acceptors.
0
2E-11
4E-11
6E-11
8E-11
1E-10
1.2E-10
-20 0 20bias, V
Cap
acita
nce,
pF
120
100
80
60
40
20
0
as-hydrogenatedp = 1-2 x 1017 cm -3
n = 3-4 x 1015 cm -3after excitation n
H+
H0
TC TSiHex
Hypothetical paths for H migration via Bourgoin-Corbett athermal migration mechanism.
Capture e- Capture h+
H0
H+
Recombination-induced formation of Hydrogen-Defect Complexes:
Recombination energy => release of hydrogen from the trapping cites near the surface, athermal migration and trapping by more stable sites (e.g., Al and B acceptors, silicon
vacancies, etc.)
Hydrogen trapped at the surface
after plasma
hydrogenation
Al-H B-H4B0
VSi-H
RIP effect (Above-bandgap light at 4 - 250K)
Remaining questions …The exact path for H migration? The energy position of the hydrogen state responsible for the recombination-induced migration?Device applications?
top related