TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol. 15, No. 2, August 201
5, pp. 249 ~
258
DOI: 10.115
9
1
/telkomni
ka.
v
15i2.809
8
249
Re
cei
v
ed Ma
y 14, 201
5; Revi
sed
Jun
e
29, 2015; Accepted July 1
4
,
2015
Side Effects in a HEMT Performance with InAlN/GaN
Z. Kourdi, B. Boua
zza*, A. Guen-Boua
zza & M. Kha
ouani
.
Electron
ic Dep
a
rtment, Unive
r
sit
y
of T
l
emcen, Re
searc
h
U
n
it of Materials
and R
e
n
e
w
a
bl
e Energ
i
es,
T
l
emcen, Alger
ia
*Corres
p
o
ndi
n
ag auth
o
r, e-mail: zkour
di@
h
otmail.fr
A
b
st
r
a
ct
We pres
ent
a
simulati
on
of
a
HEMT (h
igh
e
l
ectron
mob
ility
transistor)
structure. We
ext
r
act th
e
devic
e ch
aract
e
ristics thro
ug
h
the
ana
lysis
of DC, AC
an
d
hi
gh fre
que
ncy r
egi
mes, as s
h
o
w
n in th
is p
a
p
e
r
.
T
h
is w
o
rk d
e
monstrates th
e
o
p
timal
dev
ice
w
i
th a g
a
te l
e
n
g
th of
30
n
m
, and InAlN/GaN
heterostructure for
mi
ni
mi
z
i
ng si
d
e
effects. T
he
simulat
ed w
i
th Silvac
o so
ftw
are of the HEMT
devices w
i
th the materi
als In
AlN
show
very goo
d scala
bil
i
ty in
different a
ppl
ic
ation.
W
e
hav
e
de
monstr
ated
an exc
e
ll
ent cu
rrent de
nsity, as
hig
h
as
6
44
mA/mm, a
pe
ak
extrinsi
c tr
ansc
ond
uctanc
e of
710
mS/mm at
V
DS
=2
V
, an
d
cu
tti
n
g
freq
ue
ncy
cutoffs of 38
5
GHZ
,
maxi
mu
m fre
que
ncy
of 810
GH
z
,
ma
ximu
m effici
en
cy of 23
% for
x-Band,
maxi
mu
m
break
dow
n vo
ltage
of 36
5 V
, and
an ON/OF
F
current de
nsi
t
y ratio hi
gher t
han
8 x 1
0
8
. These v
a
lu
es w
e
r
e
deter
mi
ned
thr
oug
h th
e si
mul
a
tion
by
hydr
o
d
yna
m
ics
mo
d
e
ls, w
h
ich
mak
e
s that
opti
m
i
z
e the
d
e
sig
n
is
the
future of this technology.
Ke
y
w
ord:
HEMT, InAlN/GaN, silvaco, side
effect, power
electronics d
e
vic
e
s, semic
o
n
d
u
c
tors devices
Copy
right
©
2015 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
Devices
HEM
T
is
con
s
id
ered no
wa
days as
bein
g
excellent mo
dels is all
o
win
g
studyin
g
fundame
n
tal physi
cs,
te
ch
nology and
u
nderstan
ding
function
ele
c
t
r
ical
p
e
rfo
r
m
ance
re
sultin
g to
sup
port th
e d
e
velopme
n
t o
f
simulato
rs t
hat would
al
low
us to
c
h
art the future of
this
tec
h
nology.
With high m
o
bility and hig
h
satu
ration v
e
locity they are u
s
eful for
high freq
uen
cy and low po
wer
con
s
umi
ng a
pplication
s
[1].
And InAlN i
s
a ne
w mate
rial used toda
y for potentia
l su
ccesso
r to AlGaN [2].
It take
s
advantage of
the intere
sting
material properties of
the III-ni
trides,
such
as a wi
de
band gap, high
brea
kd
own el
ectri
c
field an
d excelle
nt t
hermal
con
d
u
c
tivity, condu
ction interfa
c
e.
Con
s
e
quently
, there is
still room for improvem
e
n
t of InAlN/GaN
HEMT
powe
r
perfo
rman
ce.
Indeed, InAlN/Ga
N tr
an
si
stors can hav
e a sh
eet ca
rr
ier d
e
n
s
ity twice high
er t
han
that of conv
entional AlG
a
N/Ga
N
HEMTs, en
ablin
g the use of
a thinne
r b
a
rri
er laye
r
while
kee
p
ing a hi
g
h
she
e
t carrie
r con
c
e
n
tratio
n at the heterointerfa
ce [3].
More
re
ce
ntly, HEMTs o
n
an
InAlN/
GaN mate
rial
structu
r
e to
demo
n
st
rate
su
peri
o
r
perfo
rman
ce
and
co
nstitut
e
an
alternati
v
e to co
mme
rcially
availab
l
e AlGa
N/Ga
N d
e
vice
s,
si
nce
it particul
a
rly
for the tre
nd i
n
the semico
ndu
ctor in
du
stry is to re
du
ce the
si
ze of
the device for
better
spe
ed
perfo
rman
ce.
But as the
g
a
te si
ze
re
du
ced,
we
have
either in th
e
sho
r
t chan
nel
or
for hot electron effects an
d poor
carrie
r transp
o
rt
efficien
cy creep
are contain
e
d
. As a result
it
become
s
utm
o
st im
porta
nt to find
sol
u
tions for the
s
e
problem
s fo
r more
spe
ed
and l
e
ss
po
wer
con
s
um
ption [4].
Ho
wever, a
s
i
de from
kee
p
i
ng the pa
ra
si
tic low in d
e
vice, it is impe
rative to impro
v
e the
intrinsi
c
gm i
n
ultra
scale
d
devices wit
h
the b
e
st p
o
ssible
ele
c
trostatic co
ntro
l, which
can
be
potentially re
alize
d
in qua
ntum-well InAlN/Ga
N/AlN structu
r
e
s
[5].
In this wo
rk,
we have d
e
mon
s
trate
d
a HEMT InA
l
N / GaN i
s
possible to
prod
uce
excelle
nt pro
pertie
s
with
the b
e
st
p
o
ssible
whil
e
minimi
zing
sid
e
ef
f
e
ct
s.
Th
rou
gh
t
h
e
optimizatio
n
of the device
desig
n an
d
quality cont
ro
l of doping i
m
plant, even
if the adopti
on of
several analy
t
ical model
s to simulate the device
s
h
i
ghly scal
ed
analysi
s
ch
aracteri
stics [6-7],
deep
level do
pant su
ch as AlN
i
s
al
so
in
tentiona
lly u
s
ed in
the
buff
e
r [8
-9], to
re
duce the
effe
ct
of buffer t
r
ap
s o
n
the
2-DEG tran
sp
ort
beh
avior
[1
0], in orde
r t
o
intentio
nall
y
unde
re
stim
ate
phen
omen
a u
n
wa
nted until
they becom
e
almost n
egli
g
ible to supp
ress buffer l
e
a
k
ag
e an
d the
s
e
ef
f
e
ct
s.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 15, No. 2, August 2015 : 249 –
258
250
A hydrodyna
mic mod
e
l a
ppro
a
ch mu
st be used to
deliver accu
rate re
sult
s for such
s
t
r
u
c
t
u
r
e
s
[1
1-
1
2
].
W
e
pr
opo
s
e
th
is mo
de
l th
a
t
ac
cou
n
ts fo
r the
pe
culia
rities of t
he G
a
N mate
rial
system. Th
e
model
s are
implemente
d
in ou
r S
ilvaco simul
a
tio
n
and ca
refu
lly
calibrated.
A
device
from
a
re
cent
ge
neration
of
t
r
an
si
stors wa
s si
mulated usi
n
g
the
AT
LAS calib
rated
setup
[1315]. A high accuracy fo
r all releva
nt cha
r
a
c
teri
stics wa
s a
c
hiev
ed.
2.
Structure o
f
Dev
i
ce
As illustrated in Figure 1(a), we
see a
cross
se
ction of the
struct
ure,
it i
s
located over
the laye
r of
substrate
(4
H-SiC),
we
find
that the
col
o
r blue
regio
n
correspon
ds to the
ele
c
tro
d
e
s
(i.e. the
sou
r
ce, d
r
ain
an
d
gate), th
e
co
lor b
r
o
w
n
co
rrespon
ds to l
a
yers of the
cha
nnel
and
cap
layer, the
col
o
r
red
corre
s
pond
s to
laye
rs
of a
don
or
and S
c
hottky,
the b
u
ffer a
n
d
spa
c
er laye
rs
corre
s
p
ond
s to gre
en colo
r and finally yellow
colo
r re
gion
s co
rresp
ond to the
s
u
b
strate. An
d the
perfo
rman
ce
simulatio
n
of
this d
e
vice
is
reali
z
ed
by Si
lvaco.
We
use two
ste
p
s for
simul
a
ted t
h
is
device a
nd th
e pro
c
e
ss
se
quen
ce i
s
as
follows:
a)
The first ste
p
focu
sed to create a
structu
r
e in the fram
ewo
r
k
DevEd
i
t.
b)
The se
co
nd step focu
se
s to analyze this
stru
cture in t
he frame
w
o
r
k Atlas system
.
The
device
contact S
c
h
o
ttky u
s
ed
Gol
d
"Au" for
T-g
a
te ele
c
tro
d
e
.
Then,
so
urce/drain
electrode
s were formed b
y
"Au" (250 nm) is
cho
s
e
n
for ohmi
c
conta
c
ts, the
device de
si
gn
feature
s
a h
e
tero
stru
cture InAlN/
G
a
N, where the
periph
e
ry o
x
ide Al
2
O
3
o
f
t
h
e
G
a
t
e
i
s
a
differen
c
e
with co
nvention
a
l desi
g
n
s
[1
3], and the
SiN pa
ssivati
on diele
c
tri
c
that minimize
s
surfa
c
e le
aka
ge and
cre
a
te
s a high d
e
n
s
ity of
shallow
traps at the
surface [16].
As a
re
sult, a
fter a
dopin
g
cap
layer is e
liminated l
e
a
k
ag
e
curre
n
t
den
sity in the
device,
and Thi
s
giv
e
s ri
se to a
con
d
u
c
tion b
and
shap
e for the b
a
rrie
r that in an
InAlN barrie
r
is
undo
ped [1
7], the sa
me
sheet carrie
r concentrati
o
n
based o
n
the
model
Fujitsu [14], the
Hall
mobility and sheet carri
er
concentration were
1300 cm
2
V
-1
s
-1
and 1
× 1
0
13
cm
-2
. The
heteroj
un
ctio
n featu
r
e
s
a
she
e
t charge
den
sity of 1.8
5
x 10
13
cm
2
.
Dimen
s
io
ns a
r
e
also
a
critical
para
m
eter for device p
e
rfo
r
man
c
e,
whil
e we find
the
differe
nt dim
ensi
o
n
s
of th
e devi
c
e
und
e
r
study is in
clu
ded in Tabl
e 1.
Table 1. Parameter
HEM
T
device
Name
S
y
mbole
Valeur [nm]
Thickness Cap La
y
e
r
E
OH
3
Thickness Lay
e
r
Schottky
E
S
7
Thickness Lay
e
r
Donor
E
D
3
Thickness Lay
e
r
Spacer
E
E
1
Thickness Canal La
y
e
r
E
C
37
Thickness Tampon La
y
e
r
E
T
150
T
h
ickness of Bul
k
E
B
100 10
3
Length D
r
ain Gat
e
L
DG
1.47 10
3
Length Gate
Sou
r
ce
L
GS
0.50 10
3
Length Gate
L
G
30
Length D
r
ain & Source
L
D
0.5 10
3
Figure 1. Dev
i
ce an
d sche
matic di
ag
ra
m of a HEMT device sim
u
l
a
ted.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Side Effects in a HEMT Pe
rform
ance wit
h
InAlN/GaN
(Z. Kourdi
)
251
3. Simulation
Results
Simulate the stru
cture with
the same di
m
ensi
o
n
s
an
d para
m
eters has bee
n sel
e
cted, or
whe
n
a device (geo
metrie
s, mesh
es, re
gion
s, el
ectro
des a
nd dopi
ng) con
s
iste
n
t
distribution
are
descri
bed
wit
h
the sa
me
seq
uen
ce of
states, al
so
in the mate
rial pa
ram
e
te
rs a
nd mo
de
ls
physi
cal a
r
e
applie
d in ov
erall at
cases. But in t
he case of
spe
c
if
ying the effect of the avalanch
e
or KINK with the impact io
nizatio
n
mod
e
l gene
ra
tion,
will be demo
n
strate
d these effects lo
ca
ted
in the next of
this wo
rk.
3.1. DC
Resul
t
s
After the definition of the HEMT stru
ctur
e and mat
e
rial
s used for model, th
e initial
solutio
n
is
ob
tained fro
m
the gate volta
ge set to
ze
ro, the structu
r
e un
de
r zero
bias
(the i
n
itial
ca
se
rep
o
rte
d
for supply v
o
ltage
s),
and
the
soluti
o
n
s are o
b
tained
from
the
ch
o
i
ce
s
spe
c
ified
in
the alg
o
rithm
,
to obtai
ning
a
depth
cha
r
acte
ri
zation
from
the
sim
u
lation.
Accu
rate
simul
a
tio
n
results
ca
n b
e
obtain
ed b
y
solving
Hydrodyn
a
mi
cs
model p
h
ysi
c
s for
simul
a
tion the a
naly
z
e
device
perfo
rmance [18], T
he devi
c
e te
mperature i
s
not
at all con
s
tant e
s
pe
ci
al
ly at the gate
exit
in the re
ality for that the si
mulation h
a
s
been b
egin
3
00K and thi
s
value is hi
gh
er with time t
h
e
simulatio
n
for that Giga sh
ould be u
s
e
d
to simulate th
e heat-flo
w in
the device [1
9].
In this device simul
a
tion, the electri
c
al t
r
ansfer characte
ri
stics are
illustrated in
Figure
2(a
)
. Fo
r th
e
HEMT
s
with
a gate
len
g
th
of 3
0
nm,
short-ch
ann
el
effect
s are a
pparent
not o
n
ly
from the la
rg
e thre
shol
d voltage
shifts
but also
from
the high
out
put co
ndu
cta
n
ce i
n
both
hi
ghly
scaled G
a
-p
o
l
ar HEMT
s [2
0].
The in
crea
si
ngly po
sitive drai
n voltag
e led to th
e
elect
r
ic fiel
d
acro
ss th
e
cha
nnel
increa
sing th
e spe
ed of the elect
r
on. T
he voltage
di
stributio
n across the ca
nal
led to a voltage
differen
c
e be
tween the g
a
te and the cha
nnel alo
n
g
it, with the
transi
s
tor d
e
mon
s
tratin
g
a
variable re
si
stance beh
avior contro
lled
by the gate v
o
ltage. Thi
s
i
ndicates excellent
gate co
ntrol
of the 2
D
EG
cha
nnel
[21],
and th
e maxi
mum d
r
ain
cu
rre
nt availa
bl
e re
ached
64
4 mA/mm
wh
en
V
GS
was bi
ased at 0.5 V &
V
DS
=2.0 V. The pin
c
h
-
off voltage
wa
s fo
und to b
e
−
1.0 V, as sho
w
n
in Figure 2(a
)
.
Figure 2(b
)
p
r
esents the e
x
tr
insic tran
scon
du
ctan
ce
(g
m
) cha
r
a
c
te
ristics of the
device,
whe
r
e th
e
si
mulation i
s
e
x
tracted
di
spl
a
ys a
maxim
u
m p
eak g
m
as
710
mS/m
m at V
DS
=2.
0
V
.
This pe
ak a
p
pears in the curve of the trans
co
ndu
cta
n
ce a
s
a dep
ende
nce on the gate bia
s
V
GS
.
This o
b
viou
sl
y reflect
s
the
DC
beh
avior
of the
simul
a
ted HEMT,
wh
ich
corre
s
p
o
n
d
to the 2
D
EG
cha
nnel
s mo
dulated by di
fferent gate
voltages. Th
es
e
pro
p
e
r
ties are supe
ri
or to the val
u
e
s
previou
s
ly re
ported fo
r sim
ilar structu
r
e
s
base
d
on AlGaN/G
a
N h
e
tero
stru
ctures
[22].
Here, a bette
r DC charact
e
risti
c
is
reali
z
ed
on
a sa
mple
with sli
ghtly
inferior electri
c
al
prop
ertie
s
in
comp
ari
s
o
n
with tho
s
e
repo
rted
earl
i
er [23]. The
total para
s
it
ic re
si
stan
ce
is
gene
rally
d
o
m
inated by
a
low Ohmi
c co
ntact re
sist
a
n
ce,
whi
c
h i
s
h
i
ghly de
sirabl
e, and
could
b
e
attributed to the increa
se
d carrie
r co
nce
n
trat
ion o
r
/an
d
an increa
se
d carrie
r mobi
lity [24].
The devi
c
e is delivere
d
to extract an
O
N
/OFF
curre
n
t density rati
o highe
r than
8 x 10
8
(with ra
nge
V
GS
betwee
n
-10V
to 1.0V),
lea
k
ag
e
current de
nsity of
f-state I
F
=9 x
10
-26
A/m, and
we have inv
e
stigate
d
the
cond
uctio
n
band p
r
ofile
s
to calcul
ate that the drai
n-ind
u
ced ba
rrie
r
lowe
ring
(DIBL) is mo
re
explicit in a
highly scal
ed
device
at th
e gate l
engt
h 30
nm wh
en
DIBL=1
68.38
mV/V with V
DS
fixed between 0.1 V and V
DD
. The effects that we
re observed d
ue
to
this
gate
l
ength are cal
l
ed sho
r
t-cha
nnel e
ffects (SCE). Effect
s occu
rri
ng
at
a la
rge
r
V
DS
are
termed
drain
indu
ced
ba
rri
er lo
we
ring
(DIBL)
effe
cts [25], as
sh
o
w
n in
Figu
re
2(d
)
, we
are
not
the firs
t to obs
e
rve this
in
s
i
mulations
for HEMTs
,
a
s
they were inv
e
stigate
d
as
early a
s
19
89
by
Awano,
et al. [26]. We
ha
ve the po
ssib
ility to
achiev
e the redu
cti
on of the
s
e
p
henom
ena
wi
th
the engin
eeri
ng of dynami
c
ally ac
tive in
terface
state
s
[27].
We
cha
nge
d
the accele
ration of the
drai
n volta
g
e
betwe
en 0
V to 3 V, when the
simulatio
n
was first
con
d
u
cted to
obta
i
n the I-V
ch
ara
c
teri
stic i
n
the DC m
o
d
e
to chan
ge
the
state of the
g
a
te voltage
b
y
5 differe
nt b
i
as valu
es, V
GS
= 0.2 V to
– 0.8 V
with
a ste
p
of
-0.
2
V.
The devi
c
e R
ON
extracted
at V
GS
= 0 V
and V
DS
in the ran
ge bet
ween 0 a
nd 1.
0 V is 0.354
Ω
·mm, the I-V cha
r
a
c
teri
stics sho
w
good
pinch-
off characte
ristic a
nd the
curre
n
t coll
apse
phen
omen
on wa
s might ultimately limit th
e scal
abilit
y of the device and it is too low com
pared to
usu
a
l value
s
obtaine
d if we com
pared
with othe
rs
work
simul
a
tio
n
becau
se th
e barrier l
a
ye
r is
very thin for
minimizi
ng pi
nch
-
off the
s
e
very impo
rt
ant in lo
gical
elect
r
oni
cs
appli
c
ation, a
n
d
becau
se mod
e
l use
d
in this work is hyd
r
odynami
cs.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 15, No. 2, August 2015 : 249 –
258
252
(a)
(b)
(c
)
(d)
Figure 2. (a)
Tran
sfe
r
cu
rve simulate
d with V
DS
stepped from 0.5 V
to 2.0 V in steps of 0.5 V and
V
GS
sweepi
ng
from 1 V to
−
10 V, (b) Tra
n
scon
du
ctan
ce characte
ri
stic
s of HEM
T
InAlN/GaN
with
a gate length
of 30 nm with
V
DS
stepped from 1.0 V to 2.0 V in steps
of 0.5 V. (c) Tran
sfe
r
cu
rve
s
i
mulated wit
h
V
DS
steppe
d from 0.05 V
and 0.5 V an
d (d)
DC O
u
tput cha
r
a
c
teri
stics of HEM
T
InAlN/GaN
wi
th a gate leng
th of 30 nm with V
GS
stepped from 0.2 V
to
−
0.8 V in step
s of
−
0.2 V
3.2. Side
Effec
t
The o
peratio
n of the
po
we
r field effe
ct trans
i
s
tors
i
s
substa
ntially
limited
on one
hand by
the con
d
u
c
tio
n
cu
rre
nt of the diod
e and
the
other g
a
te voltage by the avalan
ch
e phen
ome
n
on.
In field effect
device
s
, two
types of b
r
e
a
kd
ow
n volta
ge can
be hi
ghlighte
d
: breakdo
wn by
Kink
effect and
im
pact io
nizatio
n
[28-30], this type of
si
m
u
lation i
s
very difficult for t
he effect
of KINK
therefo
r
e def
ining of
the impact
such
Selbe
r
he
rr i
m
pact
ioni
za
tion mo
dels
[19] ha
s b
e
e
n
measured an
d model
ed e
x
tensively at room temp
eratu
r
e. Re
cently, Valdinoci an
d al. ha
s
extended
the
sim
u
lated
te
mperature
ra
nge to
4
00K
and
ha
s
dev
elope
d a
co
mpact
mod
e
l
for
both ele
c
tron
and hol
es mo
bility impact ionization.
In ord
e
r to
simulate
aval
anche b
r
ea
kdown,
the i
m
pact i
oniza
tion-ge
ne
ratio
n
model
sho
u
ld
be tu
rned
on.
Thi
s
i
s
d
one
u
s
i
ng the
imp
a
ct Selb
state
m
ent in
which the S
e
lbe
r
herr
impact ioni
za
tion model is activated, Here the
bea
m statement is use
d
to sp
ecify an optica
l
sou
r
ce of
ca
rrier p
a
ir
gen
eration in
additi
on to
the
the
r
mal ge
neratio
n provided
by
re
com
b
inatio
n
SRH.
Most
d
e
v
i
ce
simulat
o
r
s
of
t
he lat
t
i
ce
t
e
mper
ature
va
riation
s
withi
n
a stru
cture must
be
taken into a
c
cou
n
t. It is base
d
on this work on the
model of Wach
utka [31] and incl
ude
s all
thermal
sou
r
ce
s and
sinks (Joule h
e
a
t, Thomson
t
e
rm, etc.). Succe
ssful the
r
mal mod
e
lin
g
requi
re
s ap
propriate b
oun
d
a
ry con
d
ition
s
to be sp
ecifi
ed.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Side Effects in a HEMT Pe
rform
ance wit
h
InAlN/GaN
(Z. Kourdi
)
253
(a)
(b)
(c
)
(d)
Figure 3. Output cha
r
a
c
teri
stics
of the 30 nm length g
a
te InAlN/Ga
N HEMT
with
(a) Kin
k
effect,
(b) Brea
kdo
w
n effect & (c)
Self-heatin
g & (d) Voltage
V
GS
= consta
nt = -4V with
a pulse width
of
0.1
μ
s an
d a p
u
lse p
e
rio
d
of 0.2
μ
s
Figure 3(a
)
shows the out
put ch
a
r
a
c
teristics of an HEMT with LG
of 30 nm wi
th Kink
effect for temperature a
m
bient. The
device exhi
bi
ts the variati
on of drai
n current with d
r
ain
voltage sho
w
s a
ri
se afte
r stre
ss
with
abru
p
t growt
h
in
cu
rre
nt
at different
o
f
a V
GS
voltage
,
indicating exi
s
ten
c
e
of the
trap
s in
dev
ice. Th
er
efore, the en
han
ceme
nt of th
e ki
nk
effect
is
prob
ably du
e
to the traps
activated in t
he InAlN ba
rr
ier lay
e
r,
whi
c
h is
a
sharp
increa
se in t
h
e
drain
cu
rre
nt at a certai
n volt
age, thus
causi
ng an in
crea
se of
the
output co
ndu
ctan
ce an
d the
disp
ersion be
tween DC an
d other ch
ar
a
c
teri
stics. Th
e kink effe
ct also in
crea
se
s noise in excess
of the low frequen
cy com
p
onent [32]. Indeed, as t
he
output con
d
u
c
tan
c
e is fluctuating over a
rang
e of V
DS
, the condu
cti
v
ity will
fluctuate unde
r the
influence of
de-trappi
ng, and therefore
the
noise at low freque
nci
e
s
1 / f will
tend to incre
a
se
. This effect has often b
e
en attributed
to
impact io
niza
tion [33-3
4
], resultin
g in an
accumul
a
tio
n
of holes
a
m
endin
g
su
rf
ace p
o
tential
s
or
cha
nnel / su
b
s
trate inte
rface.
The open
squares
indi
cate
the critical drai
n
vo
ltage (V
kink
)
at whi
c
h
th
e outp
u
t
con
d
u
c
tan
c
e
is the maxim
u
m in the kin
k
regi
on.
Vkin
k increa
se
s regula
r
espe
ci
ally with the gate
bias,
su
gge
sti
ng that th
e d
e
-
trap
ping
process is di
rectly related to V
GD
(i. e. the elec
tric
field)
and
is field-assi
sted in natu
r
e
and it is sug
geste
d t
hat this ki
nk
coul
d be indu
ce
d
by hot elect
r
on
trappin
g
and f
i
eld-a
s
siste
d
de-trappi
ng via dono
r-l
i
k
e traps in the G
a
N buffer laye
r [35].
The off-state
I
DS
–V
DS
characteri
stics
or bre
a
kdo
w
n
voltage of
co
nventional In
AlN/Ga
N
HEMT
s with a wide d
r
ain
bias regio
n
of the
gate voltage bet
ween
–0.8 V and 0.2V are sh
own in
Figure 3(b
)
.
The conventi
onal
HEMTs
demon
strate the o
ff-state b
r
ea
kdo
w
n
vol
t
ages of
250 and
3
7
5
V, respe
c
tively, which indi
cate that the brea
kd
o
w
n
cha
r
a
c
teri
stic of the HEMT device with
an
AlN b
u
ffer lay
e
r
ha
s b
een
signifi
cantly i
m
prove
d
in
this
device. It
is b
e
lieved
th
at enh
an
cem
ent
of the
off-st
ate b
r
ea
kdo
w
n volta
ge
of the
HE
M
T
devi
c
e i
s
attribute
d
t
o
a
better carri
er
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 15, No. 2, August 2015 : 249 –
258
254
confin
ement
and the in
cre
a
se
d ba
ck-ba
rrie
r
heig
h
t of the AlN buffer layer u
s
e
d
supp
re
sses t
he
spillover of the 2DE
G
into t
he bu
ffer lay
e
r and postpone
s the
punch through of
the buffer layer,
thus re
du
cin
g
the subth
r
esh
o
ld drain leak
age current and increasi
ng the b
r
ea
kdo
w
n vol
t
age
rema
rk
ably
.
The p
unch th
roug
h of the
electron
s into
the
buffer l
a
yer cau
s
es a
rapid i
n
cre
a
se of the
drain l
e
a
k
ag
e cu
rre
nt an
d the device
brea
kd
own occurs
whe
n
the
drai
n leakage
cu
rrent
excee
d
s a
ce
rtain value [3
6], such a
s
4
00 mA/mm in V
GS
=-0.8 for example.
As mentio
ned
above, the el
ectro
n
s spillin
g over
from t
he chan
nel to
the buffer l
a
yer at a
highe
r d
r
ain
sup
p
ly voltag
e ca
n fro
m
th
e buffer l
e
a
k
age
cu
rre
nt. In many
ca
se
s, brea
kdo
w
n
in
GaN-ba
se
d
HEMT
s is ini
t
iated by the
electron
cu
rrent un
dernea
th the d
epl
eti
on regio
n
of
the
gate throu
gh the insulatin
g
buffer layer and is kno
w
n
as the buffer-laye
r
pun
ch
through effe
ct
[37], the drai
n cu
rrent cha
nge in I
DS
for
the rem
a
rkabl
e kin
k
i
s
maxi
mum ne
ar th
e pin
c
hoff of the
device a
nd re
duces
with the decli
ne in the gate volta
ge.
This ph
eno
m
enon le
ad
s to sub
s
tantiall
y reduc
ed ca
rrie
r
mobility, incre
a
sed th
reshold
voltage and
a
drop i
n
cond
uctan
c
e fo
r th
e gate voltag
es a
nd d
r
ain i
m
porta
nt. The gain
wa
s o
ne
of the top priorities, be
ca
use g
a
in affects t
he efficiency an
d the con
s
um
ption po
wer of
the
device.
The Self-he
a
ting effect is related to ma
ny
different phenom
ena li
ke to the presence of
surfa
c
e
trap
s in the
semi
con
d
u
c
tor, th
e incre
a
si
ng
passivation
layer thi
c
kne
s
s and
type
of
passivation material stro
ngly
influen
ce
on hot sp
ot
tempe
r
atu
r
e
[38] and The
tun
nelin
g
and
leakage
cu
rrent mechani
sms can si
gni
f
i
cantly co
ntri
bute to su
rfa
c
e tra
ppe
d charg
e
mod
u
l
a
tion
[39], espe
cial
ly at high drain-g
a
te bia
s
[40-
42], se
e
Figure 3
©
. The ele
c
tro
n
s
flowin
g in the
cha
nnel
are
accele
rated
b
y
the ele
c
tri
c
field. If t
he lat
t
er i
s
sufficie
n
tly high, the
electron
s in
t
he
atoms
of the
cry
s
tal pe
rcu
ssive f
r
ee
pai
r of ele
c
tro
n
hole
s
. The
h
o
les
are
colle
cted by the
g
a
te
electrode a
n
d
electro
n
s by
the drain el
ectrode. Th
i
s
ki
nd of avalan
che ca
n ca
use
light emissi
o
n
.
At low tem
p
e
r
atures, th
e e
l
ectro
n
m
obili
ty incre
a
ses
due to
a
red
u
c
tion i
n
the
di
spe
r
si
on
of pola
r
opti
c
al pho
non
s [
34]. Pitfalls in
que
stion
lo
cated un
de
r th
e gri
d
, espe
ci
ally in the d
r
ain
regio
n
si
de d
ue to dissip
ation Jo
ule el
e
c
tri
c
po
we
r. The term "d
rai
n
Lag" i
s
use
d
to describe
the
transi
ent drai
n cu
rre
nt wh
en the dr
ain
voltage is p
u
lsed from OF
F (V
DS
= 0V)
to ON (V
DS
> 0V
)
for a con
s
tant
gate voltage
[43]. There i
s
then a de
cre
a
se
of the cu
rrent I duri
ng t
h
is pul
se if it is
sufficie
n
tly long. The occu
pan
cy rates o
f
the traps onl
y depend of a
V
DS
.
We n
o
te in F
i
gure
3(d) th
e tran
sient b
ehavio
r
of the drai
n curre
n
t thus in
dicating th
e
pre
s
en
ce
of t
he p
heno
me
non
of drain
Lag. As Zh
a
ng [44]
sug
g
e
sts that the
red
u
ctio
n of
the
output cu
rre
n
t
in the GaN tran
sist
o
r
s during appli
c
atio
n of a pulse
voltage at the drain is du
e to
injectio
n of electro
n
s into t
he buffer laye
r whe
r
e they
are tra
ppe
d.
Whe
n
the drain voltage chang
es fro
m
the OFF
stat
e to the ON
state, ie for a
positive
cha
nge in V
DS
, electron
s are accele
rated by the electri
c
field gene
rated by
V
DS
. They
are
captu
r
ed
by trap
s d
eep l
o
calize
d
en
ergy
levels
wi
thin
the buffer
an
d / or th
e sub
s
trate, p
r
ovid
ed
that the
pul
se du
ratio
n
i
s
greate
r
th
an
the time
co
n
s
tant of
capt
ure,
and
sma
ller th
an th
e t
i
me
con
s
tant of e
m
issi
on. The
s
e el
e
c
tron
s captu
r
ed
by the trap
s do
not parti
cipat
e in the cu
rrent
cha
nnel. Th
e
dire
ct re
sult i
s
the
redu
ctio
n of the drain
curre
n
t until
it reache
s its
st
eady state,
as
and when the
traps a
r
e fille
d.
∗
100
(
1
)
The device h
a
s excellent chara
c
te
risti
c
s for
calculatin
g the drain o
b
tained by si
mulation
of 3.33%, we can say so d
e
vice is mo
re
stabl
e and b
e
tter, howeve
r
, after analyzing the current
delay is more pronounce
d when the t
r
ailing edge
of the pul
se
drain current
es
tablishment a
t
high an
d low
field drain.
3.3. AC
Resul
t
s
Shows in Fi
gure 4, gai
n
the current
(H
21
), maxim
u
m tran
sdu
c
er gain p
o
wer (G
MT),
stable
maxim
u
m gai
n po
wer
(GMS),
available
maxi
mum g
a
in p
o
wer (GMA)
a
nd unil
a
teral
gain
power of the HEMT with L
G
of 30 nm a
t
V
DS
= 5 V
and V
GS
= 0.0 V versus freque
ncy [1KHz-
1THZ].
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Side Effects in a HEMT Pe
rform
ance wit
h
InAlN/GaN
(Z. Kourdi
)
255
Figure 4
.
Simulated o
f
t
h
e
G
a
i
n
s
current
g
a
in
(H
21
), Unilateral
power
g
a
in (U
)
a
nd ma
x
t
r
a
n
s
d
u
c
e
r
p
o
w
e
r
gain (
M
T
G
)
v
e
rsu
s
fre
que
nc
y
for th
e
30n
m
In
AlN/GaN
H
E
MT
s. The
b
i
as
condi
tion
we
re V
DS
=
5.0 V
a
n
d
V
GS
= 0.0
V
fr
om S-
par
amete
r
s
i
mu
lat
i
ons fr
om a sm
all-
si
gn
al
mo
del
In this simula
tion, the maximum gain sh
own for the
current H
21
is 55 dB, the maximum
transdu
ce
r ga
in of powe
r
is 60 dB and th
e maximu
m stable gain of
power is 4
0
d
B
at 1 GHz.
These
result values we
re extracted
f
r
o
m
t
he extrinsi
c S-p
a
ra
mete
rs a
nd
were then u
s
ed
to verify the intrinsi
c value
s
of this devi
c
e by si
mulati
on. The ele
c
t
r
oni
c tran
sfe
r
in the chan
n
e
l is
optimize
d
du
e to the effe
ct of the ca
pacita
n
ce
s
o
f
the high v
a
lue
s
of the
gate to sou
r
ce
cap
a
cit
a
nc
e (C
GS
), which
result from the extended
effe
ctive gate length [45]
, the electro
n
i
c
transfe
r in th
e cha
nnel i
s
optimize
d
. We have extra
c
ted the cutoff freque
ncy i
s
385 G
H
z a
n
d
the
value of th
e
maximum fre
quen
cy i
s
8
1
0
G
H
z extra
c
t in sl
ope
0
dB/Dec.
Fo
r
comp
ari
s
o
n
, the
highe
st ft rep
o
rted
so fa
r i
n
nitri
de t
r
an
sisto
r
s
wa
s 3
70 G
H
z i
n
4
-
nm b
a
rrier In
AlN/GaN
HEMTs
with 30-nm g
a
te length [46
].
3.4. RF
Re
sults
In sm
all
sign
al RF me
asurements the
HEMT with
gat
e wi
dth of
0.0
3
× 12
5
μ
m
(30 n
m
×
100
μ
m
)
are
use
d
.
Thi
s
si
mulate
typica
l
10 GHz
l
a
rg
e si
gnal
sim
u
lated p
e
rfo
r
m
ance q
uantiti
e
s
related
to po
wer outp
u
t, we have
gen
erated plot
s
the
output
sign
al
sh
ape ve
rsu
s
time fo
r e
a
c
h
of the ten input signal le
vels simul
a
te
d. Ten
large
sign
al input amplitude
s a
r
e define
d
using
WAVEFORM s
t
atement
s
.
Eac
h
of these waveforms
is
applied to t
he gate in
order of increasing
amplitude [47
]
.
Show the output power (Pout
),
power-added effici
ency (
PAE) and
gain power
(GP)
versu
s
the broadb
and RF perfo
rman
ce over
the
9
-
1
1
GHz freq
uen
cy ran
ge at a
drain volta
g
e
of
30 V, gate v
o
ltage -2 V a
nd inp
u
t po
wer of 1
6
dBm
.
The devi
c
e
wa
s optimi
z
e
d
for the PA
E-
matche
d co
n
d
ition X-Ban
d
; the simula
ted output
p
o
we
r re
ach
e
d
32 dBm wi
th 15.1 dB linear
gain and 22%
of maximum PAE of as
s
o
ciated gain at
9 GHz
.
Figure 5
.
Power
Output, Gains power
and PAE for HEMT InAlN/
GaN
9
,
0x
10
9
9,
5x
1
0
9
1,
0x
10
10
1
,
1x
10
10
1
,
1x
10
10
1
,
1x
10
10
12
15
18
21
24
27
30
33
36
39
42
45
Pin=1
6
dBm
Output
powe
r
(dbm
)
Ga
i
n
Po
we
r(dB)
P
AE
(%
)
Fr
eq
ue
nc
y
(Hz
)
Out
p
ut
po
w
e
r
(d
B
m
)
+
Ga
i
n
pow
e
r
(d
B
)
10
15
20
25
30
PA
E
(%
)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 15, No. 2, August 2015 : 249 –
258
256
Comp
ared
wi
th rep
o
rted
result
s of X-b
and
HEMT p
o
we
r delive
r
s a maximum
output
power of 45.5 dBm (35.8W) with 39.5% of PAE and 1
3
.5 dB of associated
gain
at 9.5 GHz [48].
Contin
uou
s-wave power
measurement
s in
cla
s
s “A
” ope
ration a
t
10 GHz wi
th V
DS
= 15 V
reveale
d
a 19-dB linea
r g
a
in, a maximum output
p
o
we
r den
sity of 2.5 W/mm with an ~23%
power-added
efficiency (PAE), and a
9 dB large-signal
gain [49].
4. Conclu
sion
An InAlN/Ga
N HEMT d
e
v
ice with 30
nm length
gate de
sign
on SiC sub
s
trate is
develop
ed. Show that the
perfo
rman
ce
s of t
he devi
c
e a
r
e
stron
g
ly depen
de
nt on minimi
zin
g
side effe
ct, indicatin
g
that improvin
g op
erati
ng diffe
rent adverse
phen
omen
a is key to achi
eve
highe
r re
sult
s, we have
use
d
the pe
riphe
ry
oxide
of the gate and opt
imal
stru
cture
s
f
o
r
minimized th
at.
These d
e
vices exhi
bited
cu
rre
nt de
n
s
ity as
high
as 6
44 mA/
mm, a pe
ak extrinsi
c
transco
ndu
ct
ance of 7
10
mS/mm at V
DS
=2 V, an
d a
cutoff of 3
85
GHz. The
ma
ximum freq
ue
ncy
wa
s 81
0 G
H
z,
with a
maximum eff
i
cien
cy of 2
3
%, maximu
m bre
a
kdo
w
n voltage
36
5 V,
DIBL=1
68.38
mV/V and O
N
/OFF
cu
rre
n
t ratio hi
ghe
r than
8x 10
8
.
This pap
er d
e
mon
s
trate
s
t
h
e
great inte
re
st of GaN tech
nology for diff
erent ap
p
lications
with adv
erse effect th
at accompa
n
i
e
s
run of the dev
ice.
Referen
ces
[1]
Kharch
e N, Kli
m
eck G, Dae-H
y
un Kim, de
l
Alamo JA, Lu
is
ier M.
Performance a
n
a
l
ysis
of ultra-scal
e
d
InAs HEMTs
.
Electron Devic
e
s Meeting (IEDM),
IEEE Inte
rnational. 2009: 1-4.
[2]
J
Kuzm
ı
´k. IEEE Electron Dev
i
ce Lett. 2001; 22: 510.
[3]
N Sarazi
n, E Morvan, MA d
i
Forte Poiss
on,
M
Oualli, C Ga
qui
ère, O Jard
el, O Drisse, M
T
o
rdjman, M
Magis, S Del
a
ge. HEMT
techno
log
y
on Si
C
w
i
t
h
10-W
/
m
m
and 5
0
% P
AE at 10 GHz.
IEEE Electron
Device Lett.
20
10; 31(1): 1
1
-1
3.
[4]
A Ur-Rashi
d, T
Rahman, A
Hossai
n
, F
M
Mohamme
d
y
.
Analytica
l
mode
lin
g an
d simulati
on
o
f
subthres
hol
d b
ehav
ior of dua
l
materi
al gate (
D
MG).
IEEE
3
rd
Internationa
l Confer
ence o
n
Informatics,
Electron
ics & Vision. 2
0
1
4
.
[5]
Li, R W
ang, J
Guo, J Verma, Z
Hu, Y Yue,
F
F
a
ria,
Y Cao, M Kell
y
,
T
Kosel, HG Xin
g
, D Jena. Ultra-
thin b
o
d
y
GaN
-
on-ins
u
l
a
tor q
uantum
w
e
l
l
F
E
T
s
w
i
th re
gro
w
n
ohm
ic cont
acts.
IEEE Electron Devic
e
Lett.
2012; 33(
5): 661-6
63.
[6]
MJ Uren,
et al
. Punch-thr
oug
h in
sh
ort-cha
nne
l AlGa
N/GaN HF
ET
s.
IE
EE Trans. Electron Devices.
200
6; 53(2): 39
5-39
8.
[7]
S Kumar, A Agra
w
a
l, R Chaujar, S Kabr
a, M
Gupta, R Gupta. T
h
reshol
d vo
ltag
e mode
l for smal
l
geom
etr
y
A
l
Ga
N/GaN HEMT
s base
d
o
n
a
n
a
l
ytic
a
l
so
lutio
n
of 3-D P
o
isso
n
’
s eq
uati
on.
Mi
croel
ectron.
J.
2007; 38(
10/
11: 101
3-1
020.
[8]
S He
ikman, S
Kel
l
er, SP
D
enBa
a
rs, UK
Mishra. Gro
w
t
h
of F
e
d
ope
d sem
i
-ins
ulati
ng Ga
N
b
y
metalor
gan
ic chemic
al vap
o
r dep
ositio
n.
Ap
pl. Phys. Lett
.
200
2; 81(3): 43
9-44
1.
[9]
X W
a
ng, S
Hu
ang, Y Z
h
en
g,
K W
e
i,
X C
hen
, H Z
han
g,
X L
i
u. Effect of GaN Ch
an
nel
La
yer T
h
icknes
s
on DC a
nd RF
Performance
of GaN HEMT
s W
i
th Compo
s
ite AlGaN/Ga
N Buffer.
IEEE T
r
ansactions
On Electron Devices.
20
14; 6
1
(5).
[10]
S F
a
ramehr, K
Kalna, P Igi
ć
.
Drift-diffusion
and h
y
d
r
o
d
y
n
a
m
ic mode
lin
g
of current col
l
a
p
se in Ga
N
HEMT
s for
RF
po
w
e
r
app
lic
ati
on.
Se
mico
nd
u
c
tor Science a
nd T
e
chn
o
l
ogy
.
2014; 29(
2).
[11]
A Brannick, NA Zakh
leniuk, BK Ridley
,
Lester F Eastman, Ja
mes R Shealy
, WJ Sc
haff. Hy
drody
n
amic
simulati
on of s
u
rface traps in
the AlGaN/Ga
N HEMT
.
Microel
ectronics Jo
urna
l.
200
9; 40
(3): 410-4
12.
[12]
Don
g
Setu
p L
e
e
et a
l
. RF
L
L
C
hav
e us
ed
an
alum
i
num
ga
lli
um nitri
de (A
lGaN) b
a
ck-b
a
rri
er to im
pro
v
e
the freq
ue
nc
y perform
anc
e
of i
n
d
i
um
al
umin
um n
i
trid
e (InAl
N
) h
i
g
h
-el
e
ctron-mo
b
ilit
y tra
n
sistors
(HEMT
s
).
IEE
E
Electron Device Letters
. US
A. 2011.
[13]
F
u
jitsu
Lab
ora
t
ories. F
u
j
i
tsu
Dev
e
lo
ps W
o
rl
d’s
First Gallium-Nitride
HEMT
for Pow
e
r S
upply
.
Ka
w
a
s
a
ki, Jap
an. 200
9.
[14]
Ultrafast S
y
ste
m
s Group. http://
w
eb.e
ng.g
l
a.
ac.uk/grou
ps/n
ano/
UF
S/hemt
s
.html. 2007.
[15]
N R
a
man
an, B
Le
e, V M
i
sra.
Device
Mo
del
i
ng for
U
nderst
and
ing
AlGa
N/GaN HEMT
Gate-La
g
.
IE
EE
T
r
ansactio
n
s On Electron D
e
vi
ces
. 2014; 6
1
(
6
): 2012-
20
18.
[16]
P Sung, S Raj
an. Simul
a
tio
n
of Short-Cha
n
nel Effects in N- and Ga-Po
l
ar AlGaN/GaN
HEMT
s.
IEEE
T
r
ansactio
n
s On Electron D
e
vi
ce
s. 2011; 5
8
(
3
): 705.
[17]
Atlas User’s M
anu
al. DEVICE
SIMULAT
ION SOFT
W
A
RE. Silvac
o
Interna
t
iona
l.
201
3.
[18]
T
Grasser,
T
W
T
ang, H Kosin
a
, S Selb
erherr.
A Rev
i
ew
of Hydrody
na
mic a
nd E
n
ergy-T
rans
port
Mode
ls for Semic
o
n
ducto
r Device Simulation.
Proceedings
Of
T
he
IEEE.
2003; 91(2): 251-274.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Side Effects in a HEMT Pe
rform
ance wit
h
InAlN/GaN
(Z. Kourdi
)
257
[19]
Krokidis,
JP
Xa
nthakis,
NK Uz
uno
glu. A
ful
l
y
2-dime
nsi
ona
l, qua
ntum
me
c
han
ical
calc
ul
a
t
ion
of sh
ort-
chan
nel
a
nd
dr
ain
in
duce
d
ba
rrier l
o
w
e
r
i
n
g
e
ffects in HEMT
s.
Soli
d-State Electron
ics.
20
08; 5
2
: 6
25-
631.
[20]
JS Xu
e, JC Z
hang, YW
H
ou, H Z
hou,
JF
Z
hang, Y Hao. F
abr
ic
ati
on an
d char
a
c
terizatio
n
of
InAlN/GaN-b
a
s
ed d
oub
le-c
ha
nne
l hi
gh e
l
ect
r
on mob
ilit
y tra
n
sistors for el
e
c
tronic a
ppl
icat
ions.
J. Appl.
Phys
. 2012: 1
1
1
-11
4
.
[21]
B T
i
an, C Ch
e
n
, J Z
han
g, et
al. Structure
a
nd e
l
ectric
al c
h
aracteristics
of AlGaN/GaN M
I
SHF
E
T
w
i
t
h
Al
2
O
3
thin film
as both surfac
e passiv
a
tio
n
a
nd gate
die
l
ect
r
ic.
Sem
i
c
o
nd Sci Technol
. 2011: 26.
[22]
N Du
buc. E
l
ec
tro therma
l mo
deli
n
g
SiC
ME
SF
ET
tr
ansisto
rs an
d o
p
timiz
a
tion
of
an
arc
h
itecture
for
the Doh
e
rt
y
p
o
w
e
r
ampl
ifier
w
i
th hig
h
efficie
n
c
y
.
Ph.D. T
hesis. F
r
ance: Univ
ersit
y
of L
i
mog
e
s. 2003.
[23]
Y A
w
a
n
o
, M Kosug
i
, K Kose
mura
, T
Mimura, M Abe. Shor
t chann
el
effects in su
b qu
arte
r-micrometer-
gate HEMT
s: simulati
on a
nd e
x
p
e
rime
nt.
IEEE Trans Electron Dev
. 19
89;
36: 226
0-2
266.
[24]
Z
T
ang, S Hu
a
ng, Q Jia
ng, S
Liu, C
Li
u, KJ
Che
n
.
60
0 V H
i
gh-P
e
rfor
man
c
e AlGaN/GaN
HEMT
s w
i
th
AlN/SiNx Pass
i
v
ation
. CS MA
NT
ECH Confer
ence. Ne
w
Orl
eans, Lo
uis
i
an
a, USA. 2013.
[25]
JW
Chun
g, W
E
Hoke, EM
Chum
bes, T
Palac
i
os. AlGa
N/GaN HEMT
w
i
t
h
30
0-GH
z
f
max
.
IEEE
Electron D
e
vic
e
Lett.
2010; 3
1
(3): 195-
19
7.
[26]
Q Z
hou, S
Hua
ng, H
C
hen,
C Z
h
o
u
,
Z
F
eng, S
Cai,
Ch
en
KJ. Schottk
y Sourc
e
/Drai
n
Al2O3/InAlN/G
a
N MIS-HEMT
w
i
th Ste
ep S
ub-thre
s
hol
d S
w
i
n
g
a
nd H
i
g
h
ON/OF
F
Current Rati
o
.
Electron D
e
vic
e
s Meetin
g (IEDM)
. 2011: 33
41-3
344.
[27]
C Lun
gli, T
M
Barton, RE Mi
l
e
s. Avalanc
he
break
do
w
n
an
d dee
p leve
l trap are
a
GaAs MESF
ET
'
s
in
effect.
IEEE
Tr
an. Elec Dev
. 1
992; 39(
2).
[28]
LALOUE. N
onl
ine
a
r mo
del
in
g
of di
stri
bute
d
field effect
tran
sistors:
ap
plic
a
t
ion to
the
an
a
l
y
s
is of th
e
stabilit
y
of tran
sistors an
d fea
t
ures no
ise M
M
IC circ
uits. P
h
.D. T
hesis. France: U
n
iv
ersity
of Limo
ge
s
.
200
1.
[29]
R Cu
erdo, Y
Pei, Z
Ch
en,
S
Kell
er, S De
nBaars, F
C
a
ll
e, U Mishr
a
. T
he ki
nk effect
at cr
yoge
n
i
c
temperatur
es i
n
d
eep
su
bmic
ron AlG
a
N / G
a
N HEMT
s.
IEEE Electron Device
Letters
. 200
9;
30(
3):
209-
212.
[30]
M Valdi
noci, D
Ventura, MC
Vecchi, M Rudan, G Baccarani, F I
llien, A Stricker, L Z
u
llino. Impact-
ioniz
a
tio
n
in sil
i
c
on at larg
e op
eratin
g temper
ature.
SISPAD '99
. K
y
oto, Ja
p
an. 199
9.
[31]
K W
a
chutka.
Rigor
ous T
hermo
d
y
nam
ic
T
r
eatm
ent
of Heat Gen
e
ratio
n
an
d
Con
ductio
n
i
n
Semico
nductor
Device Mo
del
i
ng.
IEEE Trans
.
CAD-9
. 1990:
1141-
11
49.
[32]
RM Ch
u, YG Z
hou, J L
i
u, D
W
ang, KJ C
h
en, KM L
au.
A
l
GaN-GaN dou
ble-c
han
ne
l
H
E
MT
s.
IEEE
T
r
ans. Electron
Devices.
20
05
; 52: 438.
[33]
Zeghbr
eock. Principal of
Semiconductor
Devices. Boulder.
2004. http://ecee.co
lorado.edu/ ~
bart/book/b
ook
/chapter
3/ch
3_
6.htm.
[34]
S Nuttinck, E
Gebara, J
L
a
s
k
ar, M Harr
is.
F
l
oatin
g-bo
dy
effect in A
l
-Ga
N
/ GaN
HEM
T
pow
er.
In
Procee
din
g
of GaAs S
y
mp
osi
u
m. 2002: 3
3
-3
6.
[35]
Maoj
un W
ang, Kevin J Che
n
. Kink Effect in
AlGaN/GaN H
E
MT
s Induced b
y
Dr
ain a
nd Gate Pumpi
n
g
.
IEEE Electron Device
Letters
. 2011; 3
2
(4): 4
83.
[36]
Bahat-T
reidel
E, Hilt O, Brun
ner F, et al. P
u
nch thro
ug
h-vo
lt
age
en
ha
nce
m
ent of AlGa
N
/GaN HEMT
s
usin
g AlGaN d
oub
le h
e
teroj
u
nction co
nfin
e
m
ent.
IEEE Tr
ans Electron Devices
. 20
08; 5
5
: 12.
[37]
Uren MJ, Nash
KJ, Balmer RS, et al. P
unch-throug
h in sho
r
t-chann
el AlG
a
N/GaN HF
ET
s.
IEEE Trans
Electron D
e
vic
e
s
. 2006; 5
3
: 395.
[38]
Hag
h
she
nas, M F
a
thipour, A Mojab. Dep
end
enc
e of Self-He
a
ting Effect on Passiv
a
tion L
a
y
er
i
n
AlGaN/GaN HEMT Devices.
ISDRS
. Coll
eg
e
Park, MD, USA. 2011.
[39]
G. Meneg
hess
o
, G. Verzel
les
i
, R. Pi
erob
on,
F
.
Ra
mp
azzo,
A. Ch
ini,
U. K
.
Mishra,
C. C
ana
li
an
d E.
Z
anon
i "Surfa
ce-Rel
ated Dr
ain Curr
ent Di
spersi
on Effects in AlGaN–
GaN HEMT
s",
IEEE Trans
Electron D
e
vic
e
s
, Vol. 51, No
. 10, pp. 1554-
156
1. (Oct 2004).
[40]
R Vetur
y
, NQ
Zhan
g, S K
e
ller,
UK Mis
h
ra.
T
he imp
a
ct of surfac
e
st
ates o
n
th
e dc
an
d RF
character
i
stics of AlGaN–GaN HFE
T
s.
IEEE
Trans. Electron Devices
. 20
01
; 48: 560-5
66.
[41]
E
Kohn,
I Dau
m
iller, M
Ku
nz
e, M N
e
u
burg
e
r
, M Se
yb
oth,
T
J
Jenkins, JS
Se
w
e
ll,
J Va
n
Norsta
nd,
Y
Smorchkov
a
, UK
Mishr
a
. T
r
ansient
cha
r
acteristi
cs
of GaN-b
a
sed
heter
ostructu
re fie
l
d-effec
t
transistors.
IEEE Trans. Microw
ave Theory Tech
. 200
3; 51: 634-
642.
[42]
G Kole
y, V T
ilak, LF
Eastma
n, MG Spenc
e
r
.
Slo
w
trans
ie
nts observ
ed
i
n
AlGaN
HF
ET
s: Effects of
SiN pass
i
vatio
n
and UV i
llum
i
natio
n.
IEEE Tr
ans. Electron
Devices
. 2
003;
50: 886-8
93.
[43]
D Sirie
x
. Mod
é
lisati
on n
on-
li
néa
ire des M
ESF
ET
s sur carbur
e de sil
i
c
i
um po
ur l’am
p
lificati
on d
e
puiss
ance micr
o-on
des. T
hèse de doct
o
rat. Univers
i
té de L
i
mog
e
s. Jan 20
00.
[44]
L Zhang, LF L
e
ster, AG Baca
, RJ Shul, PC Cha
ng, CG W
illison, UK Mis
h
r
a
, SP Denb
aar
s, JC Zolper.
Epita
x
i
a
ll
y-
gro
w
n
Ga
N j
uncti
on fi
eld
effect t
r
ansistors.
IEE
E
T
r
ans.
El
ectron Devic
e
s
. 2
0
00; 4
7
: 5
07-
511.
[45]
W
Chun
g, E H
o
ke, M Ch
umb
e
s, T
Pala
cios.
AlGaN/GaN HEMT
With 300-GHz fmax
.
IE
EE Electron
Device Letters
. 2010; 3
1
(3): 1
97.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 15, No. 2, August 2015 : 249 –
258
258
[46]
Y Yue, H Z
ong
ya
ng,
J Guo, B Sensal
e-Ro
dri
guez, L Guo
w
a
ng
Li, R W
ang,
F
F
a
ria,
T
F
ang, B Song,
X
Gao, S Guo, T
Kosel, G Snider, F Patrick,
D Je
na, H
Xi
ng. A
lN/AlN/
GaN HEMT
s w
i
t
h
R
egr
o
w
n
Ohmic Contact
s
and f
T
of 370 GHz.
IEEE Electron Devic
e
Letters.
2012; 3
3
(7): 988.
[47]
http://
w
w
w
.
silv
aco.com/e
x
am
pl
es/tcad/secti
on2
1/e
x
am
ple
11/in
de
x.html
,
# (c) Silvaco In
c. 2013.
[48]
S Piotro
w
i
cz,
Z
Ouarch, E C
hartier, R
Aubr
y,
G Callet, D
Floriot, JC Jac
quet, O Jarde,
E Morvan, T
Reve
yr
and, N Sarazi
n,
SL
D
e
la
ge.
43W, 52% PAE X-B
a
nd AlG
a
N/GaN HEMTs MM
IC Amplifiers
.
Micro
w
av
e S
y
mposi
u
m Dig
e
s
t (MTT),
IEEE MTT
-
S
Interna
t
iona
l. 201
0: 505-5
08.
[49]
Sun H, et al. 102-GHz AlInN/
GaN HEMT
s o
n
s
ilico
n
w
i
th 2
.
5-W
/
mm output po
w
e
r at 10
GHz.”
IEEE
Electron D
e
vic
e
Lett
. 2009; 3
0
(8): 796-
79
8.
Evaluation Warning : The document was created with Spire.PDF for Python.