electronics

Article
Effect of the High-Temperature Off-State Stresses on
the Degradation of AlGaN/GaN HEMTs
Jinfu Lin, Hongxia Liu * , Shulong Wang , Chang Liu, Mengyu Li and Lei Wu
Key Laboratory for Wide Band Gap Semiconductor Materials and Devices of Education,
The School of Microelectronics, Xidian University, Xi’an 710071, China; JinfuLin9@163.com (J.L.);
slwang@xidian.edu.cn (S.W.); lizzy123456@163.com (C.L.); ankh9633@gmail.com (M.L.);
wulei12@foxmail.com (L.W.)
* Correspondence: hxliu@mail.xidian.edu.cn; Tel.: +86-130-8756-8718




Received: 21 October 2019; Accepted: 11 November 2019; Published: 13 November 2019


Abstract: GaN-based high electron mobility transistors offer high carrier density combined with
high electron mobility and often require operation at high frequencies, voltages, and temperatures.
The device may be under high temperature and high voltage at the same time in actual operation.
In this work, the impact of separate off-state stresses, separate high-temperature stresses, and off-state
stresses at high temperatures on AlGaN/GaN high electron mobility transistors (HEMTs) grown on
Si substrates was investigated. The output current and gate leakage of the device degenerated to
different degrees under either isolated off-state or high-temperature stress. The threshold voltage of
the device only exhibited obvious negative drift under the action of high-temperature and off-state
stresses. The parameter at high temperature (or room temperature) before stress application was the
reference. We found that there was no significant difference in the degradation rate of drain current
and transconductance peak when the same off-state stress was applied to the device at different
temperatures. It was concluded that, under the high-temperature off-state electric field pressure,
there were two degradation mechanisms: one was the inverse piezoelectric polarization mechanism
only related to the electric field, and the other was the degradation mechanism of the simultaneous
action of temperature and electric field.

Keywords: AlGaN/GaN HEMT; off-state stresses; electrothermal coupling; device degradation

1. Introduction
GaN-based high electron mobility transistors (HEMTs) are excellent candidates for high-frequency,
high-field, and high-temperature applications in the microwave range, due to their large band gap,
high breakdown electric field, and high mobility two-dimensional electron gas (2DEG) [1–3]. On the
one hand, high-field degradation characteristics refer to the possible degradation effect of AlGaN/GaN
HEMT devices after the application of the enhanced electric field. At present, the primary explanation
for the degradation of AlGaN/GaN HEMT devices under strong electric field stress is the thermionic
electron effect. The thermionic electron will lead to the capture and generation of surface state traps
between the gate and drain, thus affecting the performance of devices and causing device degradation.
Some studies believe that the inherent traps [4,5] will capture the thermal electrons, while others
believe that the thermal electrons themselves will form new defects [6–9]. Actually, the thermionic
electron effect could not be used to explain all of the high-field degradation phenomena. The device
lacks channel thermal electrons under the off-state stress of the large electric field. The experimental
results show that the device will also degrade under off-state stress [10,11]. It is suggested that there
may be other mechanisms for degradation of the device under off-state stress.

Electronics 2019, 8, 1339; doi:10.3390/electronics8111339 www.mdpi.com/journal/electronics

Electronics 2019, 8, 1339 2 of 8

John et al. [12] proposed a new degradation mechanism hypothesis for this phenomenon—the
inverse piezoelectric polarization effect. At a high electric field, the expansion of the barrier layer caused
by the inverse piezoelectric effect leads to lattice relaxation and defects. However, it is undeniable that
the same stress conditions will have different effects on different process technologies and structural
devices, and the internal mechanisms may be different. Therefore, it still requires a lot of research to
grasp the mechanism of high-field degradation fully.
On the other hand, the high temperature resistance of GaN materials determines that AlGaN/GaN
HEMT devices are often used in high-temperature environments [13–15]. Therefore, it is necessary
to study the reliability of devices during high-temperature operation. In past research on device
reliability, the experiments of temperature change characteristics [15–19] and high-field stress [4–12]
were separately performed, which is different from the current situation where the device is operated in a
high-temperature environment. Therefore, this paper studied the electrothermal coupling degradation
characteristics of AlGaN/GaN HEMT devices.
We studied the degradation of the sample device under separate off-state stresses, separate
high-temperature stresses, and off-state stresses at high temperatures, in order to compare and study
the parameter changes of the device when the temperature stress and the electric stress existed at the
same time, and analyze the cause of the change.

2. Materials and Methods

The AlGaN/GaN HEMT device used in this experiment adopted a standard process.
The AlGaN/GaN heterojunction structure was grown by metal–organic chemical vapor deposition
(MOCVD) on SiC substrates. Both the GaN buffer layer (2 µm) and the AlGaN layer (25 nm) were
unintentionally doped. The Al composition in the AlGaN barrier layer was 30%. Standard Ti/Al/Ni/Au
(20 nm/140 nm/55 nm/45 nm) source/drain ohmic metallization was annealed at 850 ◦ C for 30 s and a
rectangular-shaped Ni/Au gate contact was used. This was followed by the deposition of 60 nm and
200 nm plasma-enhanced chemical vapor deposition (PECVD) Si3 N4 passivation layer, respectively.
For the off-state high-power stress, the source was held at ground level while the applied stress
conditions consisted of a VGstress = −3 V and VDstress = 30 V. Following this, the device was turned
off, the channel was left unopened, and the stress time was 6000 s. After stress, the device was left
stationary for 72 h to restore. We measured the characteristics by using an Agilent B1500A device
analyzer and a cascade probe station. Then, in the high-temperature experiment, the device was heated
by a DHG-9033BS-III electrothermal constant temperature blast drying oven. The test sample was first
placed at room temperature for testing then placed in the thermostatic chamber. The temperatures
were set to 50, 75, and 100 ◦ C, in sequence. In order to make sure that the temperature of the device was
uniform before measurement, the sample was stored for two minutes after the oven reached the preset
temperature. Finally, the most important study was the degradation characteristics of AlGaN/GaN
HEMT devices after applying off-state stress at different ambient temperatures. The experimental
samples, test methods, and stress conditions used were the same as before. During the experiment,
the three sets of devices were placed at ambient temperatures of 50, 75, and 100 ◦ C.

3. Results and Discussion

3.1. Off-State Stress

A characteristic curve comparing reference, off-state stress condition, and recovery condition is
shown in Figure 1. It can be seen from the transfer characteristic curve of Figure 1a that, after stress,
the maximum transconductance and drain of the device current decreased. Among them, the threshold
voltage was unchanged, the maximum transconductance Gmmax was reduced by 24.9%, and the
maximum drain current was reduced by 31%. Nevertheless, the transfer characteristic curve of the
device was restored to the pre-stress level after the device was static for 72 h. This phenomenon is
generally attributed to the fact that, after stress, the electrons of the device—under the action of a
Electronics 2019, 8, 1339 3 of 8

strong electric
Electronics field—obtain
2019, 8, x FOR PEER REVIEWenough energy and enter the AlGaN barrier layer or the device surface 3 of 8
to be captured by traps [7]. Thus, the concentration of 2DEG in the channel decreased, which was
macroscopically
the characteristic showncurve as the
wasdecrease
due to oftheoutput leakage
inherent current
trapped of thein
carriers device. The degradation
the device. After storingof thethe
characteristic
sample for acurve whilewas due to the
to recover, theinherent trapped
trap released thecarriers
trappedincarrier
the device.
and theAfter storing
device the samplewere
characteristics for
a restored
while to recover, the trap released
to the pre-stress level. the trapped carrier and the device characteristics were restored to
the pre-stress level.shows the variation of the gate leakage current curve of the device. It can be seen
Figure 1(b)
Figure 1b
that, after stress,shows the variation
the gate of the gate
leakage current leakage
increased by current
nearly twocurve of the
orders device. It can
of magnitude. Thisbemay
seenbe
that, after the
because stress, the gate
electric fieldleakage
directioncurrent increased
generated by theby nearly two
negative gate orders
voltageofacting
magnitude. This may
on the AlGaN be
barrier
because
layer isthetheelectric
same as field
thatdirection generated
of the AlGaN by the negative
polarization electricgate voltage
field, acting
and the twoon the AlGaN
electric fieldsbarrier
interact
layer
withiseach
the same
otheras that of
[6,20]. Asthe AlGaNthe
a result, polarization
barrier layer electric field, and
of AlGaN was the two electric
subjected fieldssignificant
to a more interact
with each
tensile otherwhich
stress, [6,20].stretched
As a result, the barrier
the lattice layer
structure andofproduced
AlGaN was newsubjected
defects intothea more significant
material. The new
tensile stress, which
trap-assisted carrier stretched
tunnelingthe lattice
passedstructure
throughand theproduced new defects
AlGaN barrier layer inand
the formed
material.a The
newnew gate
trap-assisted carrier tunneling passed through the AlGaN barrier layer and
current leakage channel—that is, the gate leakage current of AlGaN/GaN HEMT devices increased formed a new gate current
leakage channel—that
after stress. Moreover, is,the
thegate
gateleakage
leakagecurrent
currentwasof AlGaN/GaN
unrecovered, HEMT
whichdevices increasedwith
was consistent afterthe
stress.
post-
Moreover, the gate leakage current was unrecovered, which was consistent with
stress curve. It is further indicated that new traps introduced under off-state stress caused the gate the post-stress curve.
It leakage
is further indicated
channel andthatthesenewnewtraps introduced
defects under
could not off-state stress
be recovered withcaused the gate
the removal of leakage
stress. channel
and these new defects could not be recovered with the removal of stress.

0.4 0.4
-1
VGstress= -3V, VDstress= 30V 10 VGstress= -3V, VDstress= 30V (b)
VD= 1V VD=VS= 0V
-3
0.3 0.3 10
Drain Current[A]

Gate Current[A]

Fresh -5
After off-state stress 10
Gm[S]

0.2 72h to recovery 0.2 -7

10

10
-9 Fresh
0.1 0.1 After off-state stress
-11
10 72h to recovery
(a)
0.0 0.0
-5 -4 -3 -2 -1 0 1 -6 -5 -4 -3 -2 -1 0 1
Gate Voltage[V] Gate Voltage[V]

(a) (b)
Figure1. 1.Different
Figure Differentconditions
conditionsofoftypical
typicalDCDCcharacteristic
characteristiccurves.
curves.(a)
(a)Transfer
Transfercharacteristic
characteristicand
and
transconductance
transconductance characteristics;
characteristics; (b)(b) gate
gate leakage
leakage current.
current.

3.2.
3.2.High-Temperature
High-TemperatureStress
Stress
Figure
Figure 2 shows
2 shows a representative
a representative curve of the
curve DCDC
of the characteristics of AlGaN/GaN
characteristics of AlGaN/GaN HEMT HEMTdevices at
devices
different ambient temperatures. It can be seen intuitively from Figure 2a that,
at different ambient temperatures. It can be seen intuitively from Figure 2a that, as the ambientas the ambient temperature
increased,
temperature the increased,
transconductance peak and drain
the transconductance peakcurrent of thecurrent
and drain deviceofcontinuously decreased.
the device continuously
The threshold voltage had slightly negatively drifted but the change was not
decreased. The threshold voltage had slightly negatively drifted but the change was not obvious. obvious. The main reason
The
was the decrease of the carrier saturation velocity in the high-temperature environment
main reason was the decrease of the carrier saturation velocity in the high-temperature environment [21].
Compared with the data at room temperature, the peak value of the transconductance of the
[21].
device decreased by 3.52% at 50 ◦ C, 8.60% at 75 ◦ C, and 13.47% at 100 ◦ C. In terms of drain current,
the drain current dropped by 3.32% at 50 ◦ C, 8.34% at 75 ◦ C, and 12.34% at 100 ◦ C. In summary,
taking the0.6transconductance peak and drain current of the10device -1 as the ordinate, and the ambient
VD= 5V 0.4 room temperature (b)
temperature0.5 as the abscissa, the curve of the DC parameter of-3 the device as a function of temperature
50?
10
Drain Current[A]

Gate Current[A]

was obtained
0.4 (Figure 3). The inset was the degradation
room temperature 0.3 75?
rate of the transconductance peak and drain
50? 10-5 100?
◦
Gm[S]

current relative
0.3 to room
75? temperature at 50–100 C. It can be seen that the transconductance peak and
100? 0.2 10-7
drain current of the device decreased linearly with the increase of the ambient temperature, and the
0.2
degradation rate was also approximately linear 10-9
with the ambient temperature. It shows that the DC
0.1
0.1
characteristics(a)of the device degraded uniformly with increasing 10-11 temperature. From the gate leakage
0.0
current characteristics of the device in Figure0.02b at different ambient temperatures, it can be seen that
-5 -4 -3 -2 -1 0 -6 -5 -4 -3 -2 -1 0 1
the gate leakage current of the device did not change significantly with
Gate Voltage[V]
the increase of the ambient
Gate Voltage[V]

(a) (b)
Figure 2. Typical DC characteristic curves at different temperatures. (a) Transfer and
transconductance characteristics; (b) gate leakage current.
 Figure 1. Different conditions of typical DC characteristic curves. (a) Transfer characteristic and
transconductance characteristics; (b) gate leakage current.

3.2. High-Temperature Stress

Electronics
Electronics2019,
2019,8,8,x1339
FOR PEER REVIEW 4 4ofof8 8
Figure 2 shows a representative curve of the DC characteristics of AlGaN/GaN HEMT devices
at Compared
different ambient with thetemperatures.
data at roomIttemperature,
can be seen the intuitively
peak value from of Figure 2a that, as the of
the transconductance ambient
the
temperature.
temperature
device decreased It can be
increased, considered
by 3.52%the at 50 that the
transconductanceSchottky
°C, 8.60% at 75 °C, junction
peak
andand of the
drain
13.47% experimental
at current
100 °C. In of termssample
the device used
of drain is highly
continuously
current,
reliable,
thedecreased.andThe
drain current remains
dropped substantially
threshold byvoltage
3.32% at stable
had °C,over
50slightly
8.34% this
attemperature
negatively
75 °C, drifted
and 12.34%range.
but theatAfter
change
100 °C.the device
was
In not returned
summary, obvious.
takingto
The
room
themain temperature,
reason was the
transconductance the peak
transfer
decrease andcharacteristic
ofdrain
the carrier
currentcurve ofreturned
saturation to itsas
the velocity
device original
in the level. Theand
the high-temperature
ordinate, carrier
the saturation
environment
ambient
velocity
[21].
temperature of the
as the abscissa, the curve of the DC parameter of the device as a function of temperatureof
device was constant at the same temperature. This further indicates that the decrease
drain current
was obtained (Figure at high 3). temperature
The inset was is mainly due to the decrease
the degradation rate of the of transconductance
carrier saturation peak velocity.
and drain
current relative to room temperature at 50–100 °C. It can be seen that the transconductance peak and
0.6
drain current ofVDthe = 5V
device decreased linearly 0.4with the increase
10-1 of the ambient temperature,
room temperature (b) and the
0.5
degradation rate was also approximately linear with the ambient 10-3 temperature.
50? It shows that the DC
Drain Current[A]

Gate Current[A]
0.4 room temperature 0.3 75?
characteristics of the 50?
device degraded uniformly with increasing 10-5
temperature.
100? From the gate leakage

Gm[S]
current characteristics
0.3 75? of the device in Figure 2b at different ambient temperatures, it can be seen that
0.2
100? 10-7
the gate leakage
0.2 current of the device did not change significantly with the increase of the ambient
10-9
temperature. 0.1
It can be considered that the Schottky 0.1 junction of the experimental sample used is highly
(a) 10-11
reliable, and remains substantially stable over this temperature range. After the device returned to
0.0 0.0
room temperature,
-5 -4 the -3transfer -2 characteristic
-1 0 curve returned -6 to
-5 its -4 original
-3 -2 level.
-1 0The 1 carrier
saturation velocity of the Gatedevice
Voltage[V]was constant at the same temperature. This Gate further
Voltage[V] indicates that the
decrease of drain current at high temperature is mainly due to the decrease of carrier saturation
velocity. (a) (b)
Figure2. Typical
Figure 2. Typical DC characteristic
DC characteristic curves at
curves at different different temperatures.
temperatures. (a) Transfer and (a) Transfer and
transconductance
transconductance
characteristics; characteristics;
(b) gate (b) gate leakage current.
leakage current.

Transconductance peak 0.54

0.37 Drain current
0.53
Transconductance peak[S]

0.52

Drain Current[A]
0.36

0.51
0.35 14

12 0.50
Degrada tion ra te[% ]

10
0.34 0.49
8

6
0.48
0.33 4

2 0.47
50℃ 75℃ 100℃
0.32 Environment temperature
0.46
room 50? 75? 100?
temperature
Environment temperature
Figure
Figure3.3.The
Thecurve
curveofofthe
thepeak
peaktransconductance
transconductanceand
anddrain
draincurrent (VG==0 0VVand
current(VG VD==5 5V)
andVD V)totothe
the
ambient
ambienttemperature.
temperature.

3.3. Off-State Stress at High Temperature

3.3. Off-State Stress at High Temperature
In this section, the experimental results at 75 ◦ C are taken as an example. The direct current
In this section, the experimental results at 75 °C are taken as an example. The direct current
characteristics of the device at room temperature, the high-temperature environment before stress
characteristics of the device at room temperature, the high-temperature environment before stress
application, the high-temperature environment after stress application, the return to room temperature,
application, the high-temperature environment after stress application, the return to room
and the static condition for 72 h were compared. After applying the off-state stress at 75 ◦ C, one reduced
temperature, and the static condition for 72 hours were compared. After applying the off-state stress
the temperature to room temperature without any stress and stored the sample. These were considered
at 75 °C, one reduced the temperature to room temperature without any stress and stored the sample.
the static conditions. Then, the experimental results at different ambient temperatures were statistically
These were considered the static conditions. Then, the experimental results at different ambient
analyzed to investigate the effect of temperature on the off-state stress degradation characteristics.
temperatures were statistically analyzed to investigate the effect of temperature on the off-state stress
As can be seen in Figure 4a, the device transfer and transconductance characteristics after 75 ◦ C
degradation characteristics.
application of off-state stress were significantly degraded compared to the initial measurement at
As can be seen in Figure 4a, the device transfer and transconductance characteristics after 75 °C
room temperature. Individually, when the ambient temperature rose to 75 ◦ C, the drain current and
application of off-state stress were significantly degraded compared to the initial measurement at
transconductance peak of the device decreased but the threshold voltage did not change significantly.
room temperature. Individually, when the ambient temperature rose to 75 °C, the drain current and
This was consistent with the experimental results of the previous high-temperature characteristics
transconductance peak of the device decreased but the threshold voltage did not change significantly.
This was consistent with the experimental results of the previous high-temperature characteristics
 Electronics 2019, 8, x FOR PEER REVIEW 5 of 8

Electronics 2019, 8, 1339 5 of 8

mainly because the temperature caused a decrease in the two-dimensional electron gas mobility at
the channel of the device. After applying the off-state stress at 75 °C, the drain current and
transconductance
mainly peak of the
because the temperature device
caused decreased
a decrease in thefurther, and the threshold
two-dimensional electron gas voltage showed
mobility at the a
significant negative drift. The specific data is shown in Table
◦ 1, where the degradation
channel of the device. After applying the off-state stress at 75 C, the drain current and transconductance rate of each
parameter
peak relative
of the device to the initial
decreased measured
further, and thevalue is compared.
threshold It can beaseen
voltage showed from the
significant data that
negative the
drift.
main degradation occurred after application of the off-state stress. It can
The specific data is shown in Table 1, where the degradation rate of each parameter relative to the also be seen that the device
had measured
initial a significant recovery
value of drain
is compared. current
It can be seenand transconductance
from the data that thepeak mainafter returning
degradation to room
occurred
temperature. However, the threshold voltage recovery was not apparent, and
after application of the off-state stress. It can also be seen that the device had a significant recovery no more replies could
of
be achieved after 72 hours of standing. In combination with Figure 2a, it can
drain current and transconductance peak after returning to room temperature. However, the threshold be seen that the threshold
voltagerecovery
voltage of the device
was not wasapparent,
only slightly
and negative
no more at high temperature,
replies could be achieved whichaftercan be72attributed
h of standing.to the
increase of the density of 2DEG surface at high temperature. It led to a rise in
In combination with Figure 2a, it can be seen that the threshold voltage of the device was only slightly the number of electrons
consumed
negative in the
at high channel that
temperature, whichrequired a higher negative
can be attributed bias toofbe
to the increase theexhausted. It shows
density of 2DEG that at
surface the
current
high technology
temperature. level
It led to can ensure
a rise in thethe stability
number of metal contact
of electrons consumed of the device
in the at this
channel temperature.
that required
a higher negative bias to be exhausted. It shows that the current technology level can ensureunder
The degradation mechanism [18] of metal contact driven by ambient temperature is accelerated the
the high
stability ofelectric field. Inofother
metal contact words,atthe
the device thisSchottky contactThe
temperature. of the gate electrode
degradation begins [18]
mechanism to irreversibly
of metal
degenerate
contact drivenunder the double
by ambient action is
temperature of accelerated
high temperature
under the and high
high electric
electric field
field. In [22].
otherAccording
words, theto
∆𝐸 𝑑 𝑞𝑁 𝑑2
Schottky
𝐵
− 𝐶 of
𝑉𝑡ℎ = 𝜙contact − the𝑑 gate electrode
, when begins
the height of to
theirreversibly (𝜙𝐵 ) under
degenerate
Schottky barrier the double
decreases, action ofvoltage
the threshold high
𝑞 2ɛ qNd d2d
∆E
temperature and high electric field [22]. According to Vth = φB − q C −
(𝑉𝑡ℎ ) decreases. 2ε , when the height of the
Schottky barrier (φB ) decreases, the threshold voltage (Vth ) decreases.

0.40 (a) 0.40 10-1 VGstress= -3V, VDstress= 30V

(b)
0.35 0.35 VD= 1V
VGstress= -3V, VDstress= 30V -3
10
Drain Current[A]

Gate Current[A]

0.30 VD= 1V 0.30

0.25 0.25 10-5
0.20 0.20
Gm[S]

10-7
0.15 0.15 Fresh
75?
0.10 0.10 10-9
Stress under 75?
0.05 0.05 return to room temperature
10-11
72 hours to recovery
0.00 0.00
-5 -4 -3 -2 -1 0 1 -6 -5 -4 -3 -2 -1 0 1
Gate Voltage[V] Gate Voltage[V]

(a) (b)
Figure
Figure 4. 4.Typical
TypicalDCDCcharacteristic
characteristiccurves
curvesat at different
different conditions.(a)(a)Transfer
conditions. Transfercharacteristic
characteristicand
and
transconductance characteristics; (b) gate leakage current.
transconductance characteristics; (b) gate leakage current.

Figure
Figure4b4bshowsshowsa acomparison
comparisonofofgate gateleakage
leakagecurrent
currentcharacteristics
characteristicsofofAlGaN/GaN
AlGaN/GaNHEMT HEMT
devices.
devices.ItItcan
canbebeseen
seenthat
thatthere
therewas
wasnonosignificant
significantdifference
differencebetween
between the initial
the initialtest and
test andthe test
the atat
test
◦ C but the gate leakage current of the device after application of the off-state stress had increased
7575 °C but the gate leakage current of the device after application of the off-state stress had increased
significantly
significantlybybymore morethanthantwotwoorders
orders ofof
magnitude.
magnitude.ItItisisindicated
indicatedthatthatthe theoff-state
off-statestress
stresshad
hada a
significant
significanteffect
effectononthethe
Schottky
Schottkycontact
contactof of
thethe
gate.
gate.
InIn
order
orderto analyze the effects
to analyze of different
the effects ambient temperatures
of different on the off-state
ambient temperatures on stress degradation,
the off-state stress
the parameter degradation ofdegradation
the device atofroom temperature, ◦
degradation, the parameter the device at room50, 75, and 10050,C75,
temperature, was
and statistically
100 °C was
analyzed. Based
statistically on the parameters
analyzed. Based on at thehigh temperature
parameters (or room
at high temperature)
temperature (or roombeforetemperature)
stress application,
before
the degradation rate of the device parameters after applying the off-state stress at different
stress application, the degradation rate of the device parameters after applying the off-state stress at temperatures
different
was temperatures
calculated. was are
The results calculated.
shown in The results
Figure 5.are shown
It can be in Figure
seen that 5.there
It canwasbe seen that therelaw
no apparent was
no apparent
between law betweenrate
the degradation theofdegradation
the devicerate drainof current
the device
anddrain current and the transconductance
the transconductance peak and the
peak and the
temperature. Thistemperature.
is because the This is because of
degradation thethe
degradation
device under of off-state
the device under
stress is a off-state stress is a
high-field-driven
high-field-driven
inverse piezoelectricinverse piezoelectric
polarization mechanism. polarization mechanism.
This mechanism is drivenThis
bymechanism
an electric fieldis driven by an
rather than
electricand
current, fieldtherather thantemperature
ambient current, anddoesthe ambient
not havetemperature
a significantdoes
effectnot
onhave a significant
the applied effect
electric on
field.
the applied
Therefore, electric
it does not field. Therefore,
contribute it does notunder
to degradation contribute to degradation
the off-state stress of under the off-state stress of
the device.
the device.
 Electronics 2019, 8, x FOR PEER REVIEW 6 of 8
Electronics 2019, 8, 1339 6 of 8

38
36 Transconductance peak
Drain Current

Degradation rate[%]
34
32
30
28
26
24
22
20
room 50? 75? 100?
temperature
Environment temperature
Figure5.5.Degradation
Figure Degradationrate
rateofofdevice
deviceparameters
parameterscaused
causedbybyoff-state
off-statestress
stress
atat different
different temperatures.
temperatures.

AtAtthe
thesame
sametime,
time,ititcan
canbebeseen
seenfrom
fromTable
Table1 1thatthatthe
therecovery
recoverycharacteristics
characteristicsofofthetheexperimental
experimental
samples
samples subjected
subjected to to
thethe
off-state stress
off-state at high
stress temperature
at high temperature werewere
poor.poor.Although the drain
Although current
the drain and
current
transconductance of the device had recovered, there was still a large
and transconductance of the device had recovered, there was still a large gap between the values gap between the values before
the experiment.
before The gate leakage
the experiment. The gate current andcurrent
leakage the threshold
and the voltage were substantially
threshold voltage wereunrecovered.
substantially
Byunrecovered.
comparing the Bydevice recovery
comparing thecharacteristics
device recovery at room temperature,
characteristics at itroom
can be seen that the it
temperature, degradation
can be seen
ofthat
the device was not
the degradation of only the reverse piezoelectric polarization mechanism mentioned
device was not only the reverse piezoelectric polarization mechanism above but also
a mentioned
mechanismabove in which
but both
also atemperature
mechanism and voltage
in which bothproduced
temperature effects.andInvoltage
addition to the influence
produced effects. In
onaddition
the Schottky contact described above, there may have also been an influence
to the influence on the Schottky contact described above, there may have also been an on the AlGaN/GaN
interface
influence layer. It has
on the been proved
AlGaN/GaN that inlayer.
interface the process
It has ofbeenreverse
proved gatethatbias,
in the
the AlGaN/GaN interface
process of reverse gate
trap captures
bias, electrons in
the AlGaN/GaN the channel
interface and causes
trap captures the negative
electrons shift of threshold
in the channel and causesvoltage [9]. Atshift
the negative highof
temperature,
threshold voltagethe interface
[9]. Atstructure of AlGaN/GaN
high temperature, is unstable,
the interface which of
structure is easy to generate
AlGaN/GaN more interface
is unstable, which
traps.
is easyTherefore,
to generate themore
severe negative
interface driftTherefore,
traps. of the device threshold
the severe voltage
negative at the
drift high
of the electric
device field
threshold
and high at
voltage temperature may be
the high electric theand
field result
highof temperature
the degradation mayof beboth the Schottky
the result and AlGaN/GaN
of the degradation of both
interface.
the SchottkyMoreover, the source–drain
and AlGaN/GaN current
interface. was reduced,
Moreover, and the deterioration
the source–drain of both
current was mechanisms
reduced, and the
was irreversible.
deterioration of both mechanisms was irreversible.
Table 1. Negative deviation rate of threshold voltage and degradation rate of transconductance peak
Table 1. Negative deviation rate of threshold voltage and degradation rate of transconductance peak
and drain current.
and drain current.
Temperature
Temperature Off-state Stress Stress
of Off-state Return toReturn
Roomto Room 72 Hours to
Degradation Rate [%] 72 h to Recovery
Degradation rate [%] of 75 ◦ C Under 75 ◦ C Temperature
75 ℃ Under 75 ℃ Temperature Recovery
Threshold
Thresholdvoltage
voltage 1.71 1.71 33.15 33.15 26.51 26.51 27.62 27.62
Transconductance
Transconductancepeakpeak 11.52 11.52 32.20 32.20 13.61 13.61 11.26 11.26
Drain current (VG =1V
Drain current (VG =1V /VD = 1 15.00 40.26 22.11 18.68 18.68
/VD = 1 V) 15.00 40.26 22.11
V)

4.4.Conclusions
Conclusions
The
Theoff-state
off-statestress
stresswas
wasapplied
appliedtotothe
theAlGaN/GaN
AlGaN/GaNHEMT HEMTatatdifferent
differentambient
ambienttemperatures
temperaturesinin
order
ordertotostudy
studyelectrothermal
electrothermalcoupling
couplingreliability.
reliability.The
Theexperimental
experimentalresults
resultsshow
showthat
thatthe
theincrease
increaseofof
ambient
ambient temperature lead to the decrease of the drain current and the transconductance peakofofthe
temperature lead to the decrease of the drain current and the transconductance peak the
device.
device.TheThedecrease
decreaserateratevaried
varieduniformly
uniformlywithwiththe
thetemperature
temperatureand andthis
thiscould
couldbeberestored
restoredwhenwhenthethe
device
devicecame
cameback
backtotoroom
roomtemperature.
temperature.The Themain
mainreason
reasonwaswasthe
thedecrease
decreaseofof2DEG2DEGmobility
mobilityininthe
the
high-temperature
high-temperature environment. The degradation mechanism of the device consisted of twoparts,
environment. The degradation mechanism of the device consisted of two parts,one
one
ofofwhich
which was
wasthethe
inverse piezoelectric
inverse polarization
piezoelectric polarizationeffect onlyonly
effect related to the
related to electric field.field.
the electric There was no
There was
obvious relationship between the degradation rate of drain current and the peak
no obvious relationship between the degradation rate of drain current and the peak transconductance transconductance
and
andtemperature.
temperature.ItItwaswasbecause
becausethethedrain
draincurrent
currentdegradation
degradationofofthethedevice
deviceunder
underthethein-state
in-statestress
stress
asasthe
the inverse piezoelectric polarization mechanism driven by the electric field. This mechanismasas
inverse piezoelectric polarization mechanism driven by the electric field. This mechanism
Electronics 2019, 8, 1339 7 of 8

driven by electric field rather than current, and the ambient temperature had no visible effect on
the applied electric field, so it did not affect the degradation of devices under the off-state stress.
It led to recoverable leakage current and unrecoverable gate leakage degradation. The other was the
degradation mechanism of the simultaneous action of temperature and electric field. It did not only
cause the unrecoverable degradation of the leakage current but also made the unrecoverable threshold
voltage shift negatively.

Author Contributions: Conceptualization, J.L.; validation, M.L.; formal analysis, J.L.; investigation, C.L. and J.L.;
resources, L.W.; writing—original draft preparation, J.L.; writing—review and editing, J.L.; visualization, J.L.;
supervision, S.W.; project administration, S.W.; funding acquisition, H.L.
Funding: This research is supported by the National Natural Science Foundation of China (Grant Nos.61376099,
61434007, and 61504100) and the Major Fundamental Research Program of Shaanxi (Grant No.2017ZDJC-26).
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Pengelly, R.S.; Wood, S.M.; Milligan, J.W. A review of GaN on SiC high electron-mobility power transistors
and MMICs. IEEE Trans. Microw. Theory Tech. 2012, 60, 1764–1783. [CrossRef]
2. Akita, M.; Kishimoto, S.; Mizutani, T. High-frequency measurements of AlGaN/GaN HEMTs at high
temperatures. IEEE Electron Device Lett. 2001, 22, 376–377. [CrossRef]
3. Palacios, T.; Chakraborty, A.; Rajan, S. High-power AlGaN/GaN HEMTs for Ka-band applications.
IEEE Electron Device Lett. 2005, 26, 781–783. [CrossRef]
4. Kim, H.; Tilak, V.; Green, B.M. Reliability Evaluation of High Power AlGaN/GaN HEMTs on SiC Substrate.
Phys. Status Solidi 2001, 188, 203–206. [CrossRef]
5. Valizadeh, P.; Pavlidis, D. Effects of RF and DC stress on AlGaN/GaN MODFETs: A low-frequency noise-based
investigation. IEEE Device Mater. Reliab. 2005, 5, 555–563. [CrossRef]
6. Makaram, P.; Joh, J.; Jesús, A.D.A. Evolution of structural defects associated with electrical degradation in
AlGaN/GaN high electron mobility transistors. Appl. Phys. Lett. 2010, 96, 756–758. [CrossRef]
7. Joh, J.; Alamo, J.A.D. A Current-Transient Methodology for Trap Analysis for GaN High Electron Mobility
Transistors. IEEE Trans. Electron Devices 2011, 58, 132–140. [CrossRef]
8. Omar, C.; Loic, T.; Alain, B. Effect of HTRB lifetest on AlGaN/GaN HEMTs under different voltages and
temperatures stresses. Microelectron. Reliab. 2018, 88–90, 402–405.
9. Malik, A.; Sharma, C.; Laishram, R. Role of AlGaN/GaN interface traps on negative threshold voltage shift in
AlGaN/GaN HEMT. Solid-State Electron. 2018, 142, 8–13. [CrossRef]
10. Douglas, E.A.; Chang, C.Y.; Cheney, D.J. AlGaN/GaN High Electron Mobility Transistor degradation under
on- and off-state stress. Microelectron. Reliab. 2011, 51, 207–211. [CrossRef]
11. Rong, J.; Xiao, S.; Jingtian, F. Multiple Defects Cause Degradation After High Field Stress in AlGaN/GaN
HEMTs. IEEE Trans. Device Mater. Reliab. 2018, 18, 364–376.
12. Joh, J.; Del, A. A model for the critical voltage for electrical degradation of GaN high electron mobility
transistors. Microelectron. Reliab. 2010, 50, 767–773. [CrossRef]
13. Darwish, A.M.; Bayba, A.J.; Hung, H.A. Thermal resistance calculation of AlGaN-GaN devices. IEEE Trans.
Microw. Theory Tech. 2004, 52, 2611–2620. [CrossRef]
14. Oxley, C.H.; Uren, M.J.; Coates, A. On the temperature and carrier density dependence of electron saturation
velocity in an AlGaN/GaN HEMT. IEEE Trans. Electron Devices 2006, 53, 565–567. [CrossRef]
15. Lee, J.M.; Min, B.G.; Ju, C.W. High temperature storage test and its effect on the thermal stability and
electrical characteristics of AlGaN/GaN high electron mobility transistors. Curr. Appl. Phys. 2017, 17, 157–161.
[CrossRef]
16. Husna, F.; Lachab, M.; Sultana, M. High-temperature performance of AlGaN/GaN MOSHEMT with SiO2
gate insulator fabricated on Si (111) substrate. IEEE Trans. Electron Devices 2012, 59, 2424–2429. [CrossRef]
17. Alim, M.A.; Rezazadeh, A.A.; Gaquiere, C. Temperature Effect on DC and Equivalent Circuit Parameters of
0.15-µm Gate Length GaN/SiC HEMT for Microwave Applications. IEEE Trans. Microw. Theory Tech. 2016,
64, 3483–3491. [CrossRef]
Electronics 2019, 8, 1339 8 of 8

18. Zhao, M.; Wang, X.; Liu, X. Thermal storage of AlGaN/GaN high-electron-mobility transistors. IEEE Trans.
Device Mater. Reliab. 2010, 10, 360–365. [CrossRef]
19. Omar, C.; Loic, T.; Eric, W. Temperature and voltage effects on HTRB and HTGB stresses for AlGaN/GaN
HEMTs. In Proceedings of the 2018 IEEE International Reliability Physics Symposium, Burlingame, CA,
USA, 11–15 March 2018.
20. Meneghesso, G.; Meneghini, M.; Stocco, A. Degradation of AlGaN/GaN HEMT devices: Role of reverse-bias
and hot electron stress. Microelectron. Eng. 2013, 109, 257–261. [CrossRef]
21. Chang, Y.; Tong, K.Y.; Surya, C. Numerical simulation of current–voltage characteristics of AlGaN/GaN
HEMTs at high temperatures. Semicond. Sci. Technol. 2015, 20, 188–192. [CrossRef]
22. Sun, H.; Bajo, M.M.; Uren, M.J. Progressive failure site generation in AlGaN/GaN high electron mobility
transistors under OFF-state stress: Weibull statistics and temperature dependence. Appl. Phys. Lett. 2015,
106, 7261. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).