This paper describes objective technical results and analysis.

Any subjective views or opinions that might be expressed

in the paper do not necessarily represent the views of the U.S. Department of Energy or the United States Government.

SAND2019-12493C

Analysis of the Dependence of

Critical Electric Field on
Semiconductor Bandgap

...5)
f.... TAdheinEleidc,ttrclhtehm
enmiclail,S?tcihety

Robert Kaplar1, Oleksiy Slobodyan1,

Jack Flicker1, Jeramy Dickerson1,
Trevor Smith1, Andrew Binder1, and
Mark Hollis2 Supported by the Laboratory Directed
Research and Development program
at Sandia National Laboratories, a
multimission laboratory managed and
Sandia National Laboratories operated by National Technology and
Engineering Solutions of Sandia, LLC.,
a wholly owned subsidiary of
2 MIT Lincoln Labs Honeywell International, Inc., for the
U.S. Department of Energy's National
Nuclear Security Administration under
contract DE-NA-0003525.
 2
Empirical Fitting of Critical Electric Field on
Semiconductor Bandgap
Empirical fi e LiNeu Lk) c.k.,ILL Lapabilities of emerging power electronics
materials: Ecru—EY
4
➢ Fits by Hudgins et al. in 2003 remain the standard
Direct Bandgap Material Indirect Bandgap Material
103

cdirect—gap 1.73 x losq.s cindirect—gap =

L'crit L'crit
2.38 x 1054.nd)
10,
: — 2.5 -Y—
— 2.0 SiC(4H)
GaN

InP °C(3C) SiC(6H)

•
••••••••••••••--•••

InAs * GaSb

•6 10'
• Indirect Gap(Hudgins)
Direct Gap (Hudgins) I --- Original(Sze & Gibbons)
- Original(Sze & Gibbons) id
103 * InSb

le 10' le
10-t 10' 1011 101
Bandgap Energy(eV) Eandgeg Energy(eV)

• Empirical fits to data from the literature (not existing theory)

• Not normalized to doping, device structure, temperature
• Subject to device quality
• Hudgins et al., IEEE Trans. on Pow. Elec., Vol. 18, No. 3 (2003) NffSPIL
 3 Figures of Merit
Figures of merit often rely on parameters predicted from empirical fits

103

Specific on-resistance (mg/cm2)

20
Ecrit Eg's
c.) - UFOM = Si
AIN
10
2 15 - VBDIRon,sp —Ec
3rit Improving
-o - varies with Eg at performance
a)
Li= least to the 6th
• 10
- power
c.)
a)
Diamond
To 5 GaN .0
c.)
Direct gap
o Si SiC — — Indirect gap
0
1 2 3 4 5 6 7 103 104
Bandgap (eV) Breakdown voltage (V)

• UFOM only valid for non-punch-through device structures

• UFOM only valid for vertical devices

• Hudgins et al., IEEE Trans. on Pow. Elec. 18, 3 (2003)

• Tsao et al., Advanced Elec. Mat. 4, 1600501 (2018) WOK& AMS4
4 Impact Ionization •

Impact ionization drives avalanche breakdown in a device

Charge carriers accelerated under Electric Field

high electric fields lose energy and
ionizing more electron-hole pairs,
resulting in multiplication of
charge, leading to rapid increase in
current a avalanche breakdown

Carrier ionization rates are given by:

isri.131
(EnO,p0)
an = an ,p • exp

Multiplication factor describes increase in current under impact ionization:

Avalanche occurs when denominator approaches zero

WWI
 1 Critical Electric Field and Device Performance
The critical field is defined as the IEI
Gauss' Law: Voltage:
maximum electric field that leads EEcrit = NDWD VBD = WDEcrit/2
to avalanche breakdown in a 1D
analytical model Voltage:
,s2
crit
r _iectric Dielei_ _ VBD = 2qATD
ioating guard-ri
edge terminati
6Floating guard-rir,,_
dge termination

(1
Electric Field
n- GaN Drift region
epitaxial layer

n+ native GaN substrate

WD
Specific on-resistance: Romsi
qitn ND

3-7I
= EPnEcrEt
I UFOM: 17"
4
Ron,sp

2D Electron Gas
Unipolar Figure of Merit(UFOM)
---------- -
Electric Field
channel • Ecrit is obtained from VBD
• —on,sp
R is extracted from forward-
current slope
• Carrier mobility is often taken from
literature 0ENEEdir AMS41
 6 Procedure for Renormalization of Ecrit values
1 • UFOM assumes a given doping and field profile
• Valid comparison of critical electric field between devices requires two
major kinds of corrections:

• Doping:
➢ Critical electric field is doping-dependent
➢ Devices of different doping levels must be compared to an equivalent doping level

• Structure
➢ Electric fields in non-punch-through (NPT) and punch-through (PT) style devices are not directly comparable
➢ To equate, PT devices are transformed into an "equivalent" NPT device (integrated E-field curve is
constant)
Critical Electric Field for Breakdown (V/cm)

107

Non-punch-through Punch-through
n- n+ n-

1E1 Ecrit,NPT
1E1 Ecrit,PT
Ecrit,NPT - corrected
10'
q ND
slope= —
E
Si

10' X
10" 10,4 10,4 1016 10" WNPT L"PT
Doping Concentration (cm 3) WNPT- extra polated

• B.J. Baliga, Fundamentals of Power Semiconductor Devices, 2nd Ed. (2018) OMER& Amsa
 7
1 Correction for Doping

Analytical solution for renormalization requires two critical assumptions

so the impact ionization integrals in the multiplication factor equation
can be evaluated
1. Electron a hole ionization rates are equal: a, = ap
2. Ionization rate follows a power-law form: a = cr0E6
o with most-widely accepted : 6 = 7

Thus correction for different doping profiles follows:

xD1 X.D2
f cr(x)dx = cr(x)dx = 1

1
1
° 1 °
a(E)dE cr(E)dE
ND1f
Ecriti ND2 Ecrit2

Ecrit1 (ND1

Ecrit2 ND2

• B.J. Baliga, Fundamentals of Power Semiconductor Devices, 2nd Ed. (2018) 0ENERGY WWI
1 Correction for NPT & PT Device Structure
Pu nch-th rough
I
•
p+ n- n+

IEI Ecrit,PT
Ecrit,NPT - corrected

•
•
•
•

•
•
Mk_
14n7 x
I("PT
WATT- extrapolated
(.I
- ND 2
VBD,PT = Ecrit,PTWPT
2E WPT
qND
E(x) = Ecrit,PT E xix WPT
Equating the impact ionization integrals for different electric field profiles:
fXNPT oXPT q ND
f Ecrit PT E vv PT
cr(x)dx = a(E)dE = a(E)dE
o ECrtit,NPT

[1 (1
Ecrit,NPT — (IND,PT WPT
Ecrit,PT E Ecrit,PT
NffS41
 9
I Combined Corrections

Doping and device structure corrections for Ecrit combine into:

1/8 811/8
Ecrit,NPT (ND,NPT qND,PT WPT
1
Ecrit,PT ND,PT E Ecrit,PT

• Using this equation gives us the ability to equate Ecrit for devices with
different doping and field profileF
Reference: DRIFT- ND WPT VBD Ecrit,PT- Ecrit,NPT WNPT - -
Ecrit,NPT
LAYER (CM-3) (-11-n) (V) measured corrected extrapolated normalized to
MATERIAL (MV/cm) (MV/cm) (Pm) ND=loi6 cm-3
(MV/cm)
Armstrong et al. GaN 3x1015 30 3930 2.09 2.09 40.10 2.43
Alterman et al. A10.3Ga0.7N 5x1016 4.3 1627 5.76 5.76 6.26 4.71
Ohta et al. GaN 9x1015 22 4700 3.86 3.86 24.64 3.91
Hu et al. GaN 2.5x1015 8 1406 1.93 1.88 25.46 2.23
Nishikawa et al. GaN 2x1016 0.225 52.375 2.37 1.98 5.68 1.81
Nishikawa et al. fidild5a0.71 N 2x1016 0.225 83.75 3.76 3.00 8.16 2.75
Nishikawa et al. Ad0LI 5a0.66N 2x1016 0.225 91.125 4.14 3.23 8.72 3.00
Nishikawa et al. A10.46Ga0.54N 2x1016 0.225 112.5 5.04 3.90 10.26 3.58
Nishikawa et al. A10.52Ga0A8N 2x1016 0.225 138.75 6.21 4.69 12.20 4.30
Nishikawa et al. Alo.57G ao_43N 2x1016 0.225 181.5 8.11 5.94 15.30 5.45

• A. M. Armstrong et al., Elec. Lett. 52, 1170 (2016)

• A. A. Alterman et al., Elec. Lett. 52, 1319 (2016)
• H. Ohta et al., IEEE Elec. Dev. Lett. 36, 1180 (2015)
• Z. Hu et al., App. Phys. Lett. 107(2015)
• A. Nishikawa et al., Jap. Jour. of App. Phys. 46, 4B (2007) 0ENERGY WWI
 1° I Updated Ecrit VS. Eg Values

Semiconductor Bandgap at Type Sze Hudgins Wang Our

3oo K (eV) research
InSb 0.17 Direct 1x103 1x103 a

106 0.354 Direct 4x104 4x104 a

[
Ge 0.661 Indirect 2.5x105 1x105 1x105 2.0X105
GaSb 0.726 Direct 5x104 5x104 b

In0.53Ga0.47As 0.74 Direct 2.84x105

Well-established Si 1.12 Indirect 4.37x105 3x105 3x105 3.7x105
InP 1.344 Direct 5x105 5x105 5.0x105
G aAs 1.424 Direct 4.98x105 4x105 6x105 5.4x105
GaP 2.26 Indirect 7.59x105 1x106 1x106 b

3C-SiC 2.36 Indirect 1.3x106 1x106

Widebandgap 6H-SiC 3.0 Indirect 2.4x106 5x106 2.9x106
4H-SiC 3.23 Indirect 3.18x106 3.2x106
GaN 3.45 Direct 3x106 5x106 3.9x106
Ultra- 3.45-6 Direct d

4.7 d
Widebandgap f3-Ga203 Direct 15x106
C (diamond) 5.5 Indirect 5.7x106 21.5x106 10.1x106

a tunneling cannot be decisively excluded as a contributing factor in carrier multiplication

b no reliable data in the literature was found
C insufficient device data to perform normalization

d no experimental data confirming temperature-dependent behavior indicative of true

avalanche breakdown

• S. M. Sze and G. Gibbons, App. Phys. Lett. 8, 111 (1966)

• Hudgins et al., IEEE Trans. on Pow. Elec. 18, No. 3 (2003)
• L.-M. Wang, 25th IEEE Int. Conf. on Microelec. (2006) 0ENERGY WWI
" I Well-Established Semiconductors

• Impact ionization coefficients and E„it vs. doping is known for many of the well-
established semiconductors
• Compiled in Semiconductor Parameters Volumes 1 a, 2 edited by Levinshtein,
Rumyantsev Et Shur
Ge Si InP10-
10'

eakdown vokage V, (V)

10° 100 10° 10"
10" 10,6 10" 10" 10- 1013 10" 10" 10 o" 10-
Concentration N (cm 3)
Concentration N(cm-3) Concentration N(cm ')

103 GaA
4
> 10 106

al
Li:-
o 1• 03 105 o.)
o
c

104
c
.8 10 3: 2
10
O -o
O
cias
O
ap_ a
101
1014 1015 1016 1011 lol° 1013 1014 10
15
10
16
1017 1018
Impurity concentration N ( cm-3) Concentration N (cm-3)

• M. Levinshtein, S. Rumyantsev, M. Shur, Semiconductor Paramaters Vol. 1 (1996)

• M. Levinshtein, S. Rumyantsev, M. Shur, Semiconductor Paramaters Vol. 2(1999) Oiliadir WWI
 12 I WBG Semiconductor: SiC

ap Impact Ionization Data Critical electric field vs. doping

2x104
5x

1x1 04
2x
co
a)
E4=
4-
a) 5x103

o 5x
co
N 3x10
o
4C3 2x 105
CU 2x103
E
Decreasing a
with temperature 1x
avalanche breakdown • ping a
1x103 I ' I
2.8 3.2 3.6 4.0

1/E( x 107 cmN)

• R. Raghunathan and B.J. Baliga, Solid-State Elec. 43 (1999) 0ENERGY WWI

 13 I WBG Semiconductor: GaN

Breakdown vs. Temperature Data Best Device Reported in Literature

ri-GaN: Mg(2x10-'cm- ). 30 nm
Ti (30 nm)/AI (250 nm)
Pd (200 nm)/Ni(100 nm) c
IncrOasing Vad with
Current (IAI)

1021
mik..._ —NIL
NM\\ si.

temperature.'„ -GaNI-Mg (1.5x10.8crn-3 1020

1101/40 500 nrn E
avalanche breakdown m
1019
un-GaN (Si <2x1015 cm ). 5.5 pm E
o Mg
a - 10 gm r( 1018
b = 15 gm
c = 15 gm un-GaN n-GaN n -GaN
n -GaN : Si (9x10'5 cm-3). 22 pm d = I m 1017 ().4

n--GaN:Si (1.6)(1013 cm-3), 5.5 p.m

n-GaN:Si (2x1018 cm-3). 2 pm

10 15 20 25 30 35 40
Depth(gm)
—1500 —1000 —500 0 n-GaN Substrate
Ti(50 nm)/AI(250 nm)
Voltage (V)
1.0E 03

1.0E-04

1.0E-05

1.0E-06

T=175°C
1.0E-07

1.0E-08

c.)
1.0E-09

1.0E-10
T=25°C

1.0E 11
-800 -700 -600 -500 -400 -300 -200 -100
1000 2000 3000 4000 5000
Reverse Voltage(V)
Voltage(V)

• O. Slobodyan et al., MRS. Comm. 8, 4 (2018)

• I. Kizilyalli et al., IEEE Trans. On Elec. Dev. 60, 10, (2013)
• Ohta et al., IEEE Elec. Dev. Lett. 36, 11, (2015) 0ENERGY WWI
 14 I UWBG Semiconductors: 13-Ga203 & Al„Ga,,N •
f3-Ga203 has predicted Ecrit 7 MV/Cm AIN has predicted Ecrit-10 MV/cm and
GaN Ecrit-3.9 MV/cm with AlxGal _xN
alloys values falling in-between
NI/Au Pd/Au 0.01
SiN,(360 nm) 0
(80/420 nrn) siO7(40 ivININI cslE
111111111111111111111111111111 i-AlxGa1-x N
_2 ...••••••
Si-doped Ga203 Epi(8 pm) p-GaN
-4
46% 34%
6
Sn-doped Ga203(650 pm) . II
-6
Ti/Au (20/80 nrn)

•
a) -10
-8 n-Al„Ga1.xN
(1) -0.01
57%
52% I• 29°4 0%
:1
n-AlyGa,.),N buffer
-12 a)
n-SiC substrate 1
-14 -0.02
111111111111111111111111111111111111111111111111111I 1
0 -16 Ti/Au a)
1
-2000 -1500 -1000 -500 0
() -0.03
11
Voltage (V) (Drift layei ; I 11
225 nm -200 -150 -100 -50 0
Ecrit = 2.8 MI7lcm Reverse voltage (V)

Ent = 1.81 MVlcm Eg/

t° = 2.75 MVlcm

= 2.96 MVlcm Ezicpivto = 3.58 MVlcm

Ec5r2i73 = 4.30 MVlcm EWivt° = 5.45 MVlcm

No current temperature dependence of breakdown voltage or ionization

coefficients support avalanche breakdown in f3-Ga203 or AlxGa1,N

• J. Yang et al., ECS Jour. of Sol. State Sci. and Tech. 8, 7(2019)
• A. Nishikawa et al., Jap. Jour. of App. Phys. 46, 4B (2007) OINERdif NffSli
 15 I UWBG Semiconductor: C (diamond)

Breakdown vs. Temperature Data Best Device Reported in Literature

Electrical Field E (MV/cm) 3
73
101 (c)
p.o 0.5 1.0 1.5 2.0 2.5 423K 10-2 b)
10- I I I I
• 10.3
.73K < 10_4
AP444.
— 50 pm diameter 296K 323K (a) 4ta‘Nt
10- 523K • 10.5
z523K 150 gm
s,
4 44.0.
4731, VW •

1 1
- c1.0.13.6 µin
107
4E'
Au Schottky
---- o8 lb (100) HPHT
----- Ti/Au Ohmic
7 d,ub=500 pm
104

423K 10-1° -10 -8 -6 -4 -2

373K
\ 323K
Voltage (kV)
296K Increasing VBD with
10-1 temperature
avalanche breakdowr
But... a complicated doping profile
l e e e l l
10-14 I • I • • I
500 1000
Reverse Bias V (V)

SIMS concentrations (cm

to(c)
1016 102°
(b)

s•/
zQ 1015 4o. •
gr
H detection limit 0,..g
10'9
do (a)
vi N detection limit •
1014 1.1. 18
10
0 2 4 6 8 10 12 14

Depth (pm)

• M. Suzuki et al., Phys. Stat. Sol. A 210, 10 (2013)

• P.-N. Volpe et al., App. Phyl. Lett. 97,(2010) ENERGY NffSiai
16

1 Fitting of Updated Critical Electric Field vs.

Semiconductor Bandgap Values
• Updated power-law fit Lo !levy criticaL electric field values
• Best-fit power-law slope E„it—E3
• Direct Bandgap
;: y =1.86
• Direct Bandgap 1
1
• Indirect Bandgap --- Direct Bandgap Fit
— 1.86 Power Fit • Indirect Bandgap
Indirect Bandgap Fit

1o6-
LL
lJ

lJ

U 105—
"s7
u

02 L0
Bandgap (eV)
60
104
02 10
Bandgap (eV)
60 1
Semiconductor
Ge
1110.53Ga0AAS

Si
InP
Type
Indirect
Direct
Indirect
Direct
Bandgap (eV) Our research
0.661
0.74
1.12
1.344
2.0x105
2.84x105
3.7x105
5.0x105
Criticial Field Values
All
Direct Bandgap
Slope
1.86
1.76
R2

0.980
0.947
1
G aAs Direct 1.424 5.4x105
Indirect Bandgap 1.89 0.992
6H-SiC Indirect 3.0 2.9x106
4H-SiC Indirect 3.23 3.2x106
GaN Direct 3.45 3.9x106
C (diamond) Indirect 5.5 1.01x107 0ENERGY NffSogi
17 I Comparison of Updated Empirical Fit with
Historical Fits

• normalized data
0.75 Power Fit
R2 = 0.631
1.86 Power Fit
R2 = 0.980
2.0 Power Fit
R2 = 0.974
2.5 Power Fit
R2 = 0.864
3.0 Power Fit
R2 = 0.612

0.75 1 2 3 4 5 6
Bandgap (eV)
• For normalized data, the proposed 1.86 fit is much better than most
historical fits to date
• This fit should more accurately predict material performance
capabilities Nadir Nowt
 18
I Simulation: Ridley's Lucky-electron
• To provide fundamental basis for E„it, developed a phenomenological model
for breakdown
• Ionization integral transformed from integral over space 4 integral over E-field
• Ionization rate as a function of field is inserted into integral

• Based on the fundamental avalanche model developed by B.K. Ridley

("lucky-drift" model)
• Higher fidelity in high/low field regimes (compared to Shockley's "lucky electron" or
Chenowyth models)
• More tractable for calculation than Monte Carlo full band approximations
Ridley's Lucky-Drift Model:

1
crA.= —l-x
x
e +

x=
eEA.
(e-2rx2
1 — 2rx
) T
Eth
[-x(1-0
+ po e
(e,-2rx2(1-0 — e-x(1-)11
+

= 2rx2
PT
1 — 2rx
0

PT = 1 — e
-2rx(x-3)1
x>3
1
Lucky-drift can be described by three variables:
o Eth, thresholdenergy required to initiate avalanche
. r, ratio of energy between electron-phonon interaction and electron-electron momentum transfer
. X, mean free path of electron

Approach is to transform Eth, r, and A. into functions dependent on Eg

• B.K. Ridley., Jour. of Phys. C - Sol. State. Phys. 16, 3373 (1983) eINERdif NffSli
 19 I Determining Eg dependence of Threshold •
Energy and Phonon Collision loss for Ridley's
Lucky Electron Model
Eth
• Energy hot carrier must possess to create an electron-hole pair
• Many assume ionization can be initiated by any electron with energy > Es
• Actual E> 1.5Eg as derived by Maes et al.

• Average energy loss per phonon collision to the threshold energy

hco h is the reduced Planck constant
r= w is phonon angular frequency,
[2n(w)+ I]Eth n(w) is the quantization number.

• Only Eth is function of Eg, so r reduces to

r oc E -1

• W. Maes, K. de Meyer, R. van Overstraeten, Sol.-State. Elec. 33, 705 (1990) OMER& Amsa
1 Determining Eg dependence of Mean-free-path
for Ridley's Lucky Electron Model

• Mean free path length of a carrier before thermal relaxation

• Assume nonpolar optical phonon scattering 4 -I- is energy independent and
depends inversely on the effective mass (m*)
For a carrier at Eth
A = vg • T

/1 a
1

m*2
,\12Eth
m*
so:
with

'ac
vfl(E)

1
1
Eth 2
2E
M.*
and

1
y OC

EEZ
1
m*2
1
1
rn,* 2 m* 2 m*2 m*2
A linear relationship between m* and Eg can also be derived via Bloch Theory for a periodic potential
1
1 Eg 2 1
oc OC
3 1 3
E92 E92 E
Use Si values for r, X to calculate proportionality constants:
Asi = 7.1 nm to 11.94 nm
rsi = 0.049 to 0.063

OMER& AMS4
 21 I Parameters for Ridley's Lucky Electron Model

Parameter Value Reference

Threshold Energy, Eth 1.5 • Eg Maes et al.
Ratio of average energy loss per 0.063
Ridley
collision to the Eth energy, r Ey
12 x 10-7
Mean relaxation length, X Several
Ea 3/2

• B.K. Ridley., Jour. of Phys. C - Sol. State. Phys. 16, 3373 (1983)
• W. Maes, K. de Meyer, R. van Overstraeten, Sol.-State. Elec. 33, 705 (1990)
• W. Maes, K. de Meyer, R. van Overstraeten, Sol.-State. Elec. 33, 705 (1990)
• T. H. Ning, C. M. Osburn, and H. N. Yu, Jour. of App. Phys. 48, 286 (1977)
• T. Simon, K. Ping-Keung, and H. Chenming, IEEE Trans. on Elec. Dev. 31, 1116 (1984)
• J. W. Slotboom et al., 1987 Inter. Elec. Dev. Meeting (1987)
• Y. Z. Chen and T. W. Tang, Jour. of App. Phys. 65, 4279 (1989)
• F. M. Abou El-Ela and I. M. Hamada, Fizika A 13, 89 (2004) Oblaidir AMS4 I
22 I Flow Diagram for Simulation of Ridley's Lucky
Electron Model
Materials constants Calculate
Dielectric constant: E Threshold energy: Eth
Applied Voltage: V
Bandgap: Eg Mean free path: Ä.
[Energy loss ratio: r
l Device Parameters
Doping: ND 1
Calculate Maximum Electric Field: Šmax
1 2E,E0V
sqND
1
Calculate "Lucky Drift" Ionization rates: a, Et CCP

1
Calculate Electron Et hole multiplication coefficients Mn Et Mp:
0 r E

Mn =
Er E0

q ND
f
c
L

0
max
ap(E)•
exp[r
-TO
q ND
r 0
(an(E') — ap(E'))dEldE —4 1
Emax

ErE0
M~ = an(E) • ±q
exp[- ;:fE (ap(E — an(E'))dEldE 1
(IND tmax

YES NO I Iterate voltage:

<=M [Are either >1? = VL + AV
Avalanche Breakdown OENEEdir
1 Comparison of Empirical Fit and Simulation
Results
• Empirical and simulated values are in reasonable agreement

• I ndirect Bandgap
Semiconductor Bandgap Our Simulated
■ Direct Bandgap (eV) research
1.86 Power Fit InSb 0.17 3.41x104
-Simulated Fit InAs 0.354 8.70x104
Ge 0.661 2.0x105 1.88x105
GaSb 0.726 2.20x105
I n0.53Ga0.47As 0.74 2.84x105 2.35x105
Si 1.12 3.7x 105 4.83x105
InP 1.344 5.0x 105 6.65x105
G aAs 1.424 5.4x 105 7.36x105
GaP 2.26 1.73x106
3C-SiC 2.36 1.89x106
6H-SiC 3.0 2.9x106 2.94x106
4H-SiC 3.23 3.2x106 3.36x106
GaN 3.45 3.9x106 3.80x106
104 f3-Ga203 4.7 6.71x106
0.5 0.75 1 2 3 4 5 6
Bandgap (eV) C (diamond) 5.5 1.01x107 9.27x106

Criticial Field Values Slope 1

1
R2

Empirical 1.86 0.980

Simulation 1.67 0.990

0ENERGY NffSogi
24 I Conclusions & Precations
• Updated critical electric field vs semiconductor bandgap values
• Performed re-normalization to properly compare devices with different
dopings and structures
• Power-law slope of: 1.86 is best fit
• No difference between direct against indirect bandgap semiconductors

• Performed first-principles physics simulation to determine Ecrit for

various semiconductors
• Reasonable agreement with empirical data

• Future development of WBG Et UWBG materials is expected to change

some critical field values
• Not valid for high-electron mobility transistors (HEMTs) or lateral
devices!
• Electric field profiles are very different

0ENERGY AMS4