Jang Won-Ho1
Seo Kwang-Seok2
Cha Ho-Young1*
-
(School of Electrical and Electronic Engineering, Hongik University, Seoul, 04066,
Korea)
-
(Department of Electrical and Computer Engineering, Seoul National University, Seoul
151-744, Korea)
Copyright © The Institute of Electronics and Information Engineers(IEIE)
Index Terms
AlGaN/GaN heterojunction, p-GaN gate, selective plasma etching
I. INTRODUCTION
Gallium nitride (GaN) is a wide energy bandgap material with high breakdown field
and high electron mobility, which can overcome the limitations of conventional Si-based
power semiconductor devices (1). AlGaN/GaN heterojunctions have two-dimensional electron gas (2DEG) channels with
high carrier densities enabled by the spontaneous and piezoelectric polarization effects
(2). However, these strong polarization effects cause difficulties when implementing
the enhancement-mode (E-mode) operations. Several methods have been developed for
E-mode operations, including p-GaN/AlGaN/GaN, recessed metal–insulator–semiconductor
(MIS) gate, and thin-AlGaN/GaN structures (3-5), among which the p-GaN gated AlGaN/GaN heterojunction field-effect transistors (HFETs)
have been commercialized successfully.
The most important process step in the fabrication of the p-GaN gated AlGaN/GaN HFET
is the etching process for the p-GaN layer outside the gate region; during this step,
not only must the p-GaN etch depth be controlled precisely but the plasma-induced
damage must also be minimized. Because the thickness of the AlGaN barrier layer is
very thin, even a small amount of over-etching of the AlGaN layer and/or plasma damage
would reduce the current density significantly owing to the reduced 2DEG density.
While a conventional time-controlled etching process or an atomic layer etching process
can be utilized to etch the p-GaN layer (6,7), a selective etching process between the p-GaN and AlGaN is highly recommended from
the point of view of process reproducibility. A high selectivity was previously reported
between GaN and AlGaN employing inductively coupled plasma (ICP) etching using Cl2/N2/O2 or Cl2/Ar/O2 gas mixtures where adding O2 formed an etch-resistive oxide layer on the AlGaN surface (8-10). However, a high self-bias voltage required to achieve the high selectivity would
cause plasma induced damage at the surface. Since the AlGaN barrier layer under the
p-GaN layer is very thin, the plasma damage caused during the selective etching process
plays a critical role in device characteristics. In order to minimize such surface
damage and over-etching issues, we proposed a two-step etching process to etch the
p-GaN layer, wherein the p-GaN layer was first partially etched using a low damage
BCl3/Cl2 nonselective etching process and then the remaining thin p-GaN layer was etched using
an optimized Cl2/N2/O2 selective process to reduce the over-etching time.
II. EXPERIMENTS AND DISCUSSION
1. Optimization of Selective Etching Process
The selective etching process used in this work involved an oxidation reaction mechanism.
Adding the O2 gas into the Cl2/N2 gas mixture oxidizes the GaN or AlGaN surface, and the strong binding energy of the
oxidized AlGaN then stops the etching process. The binding energy of Al-O is higher
than those of Al-N and Ga-N (9). In addition, the Ga-O bonding is chemically more reactive than the Al-O bonding
(11), which also helps the selective etching process.
Two epitaxial structures were used for the selectivity measurements: one with a p-GaN
layer on the top surface and the other with an AlGaN layer on the top surface. The
etching process was performed using an Oxford Plasmalab 100 ICP system with a Cl2/N2/O2 gas mixture. The gas flow rates of Cl2/N2 were fixed at 40/10 sccm, while the O2 flow rate, chamber pressure, and chuck temperature were varied in the ranges of 0–4
sccm, 5–25 mTorr, and 5–60°C, respectively.
Fig. 1. Etch rates and selectivities between p-GaN and Al0.24Ga0.76N as functions
of (a) chamber pressure, (b) O2 gas flow rate, (c) chamber temperature.
The etch rates and selectivities as functions of the chamber pressure are shown in
Fig. 1(a), where the O
2 flow rate, source radio frequency (RF) power, bias RF power, and chuck temperature
are 2.5 sccm, 1750 W, 25 W, and 20°C, respectively. The selectivity is defined as
the ratio of the p-GaN etch rate to the AlGaN etch rate. The etch rates for both p-GaN
and AlGaN were observed to decrease as the chamber pressure increased. A selectivity
of 23:1 (for p-GaN:AlGaN) was achieved at a chamber pressure of 20 mTorr, where the
self-bias voltage was 213 V and the etch rates for p-GaN and AlGaN were 2.3 and 0.1
Å/s, respectively.
The etch rates and selectivities as the function of the O2 flow rate are shown in Fig. 1(b), where the chamber pressure, source RF power, bias RF power, and chuck temperature
are 20 mTorr, 1750 W, 25 W, and 20°C, respectively. The etch rates for both p-GaN
and AlGaN were observed to decrease as the O2 flow rate increased up to 2 sccm, which was associated with more oxidation at the
surface. The etch rate for AlGaN did not decrease further at O2 flow rates of >2 sccm. For the O2 flow rate of 2 sccm, the etch rates for the p-GaN and AlGaN were 4 and 0.1 Å/s, respectively,
resulting in a selectivity of 40:1 (for p-GaN:AlGaN). The self-bias voltage was recorded
as 224 V.
The etch rates and selectivities as functions of the chamber temperature are shown
in Fig. 1(c), where the chamber pressure, the O2 flow rate, source RF power, and bias RF power are 20 mTorr, 2 sccm, 2000 W, and 25
W, respectively. A relatively higher source RF power was used in this case because
of the unstable matching issues of the system at elevated temperatures. It should
be noted that the etch rates for the p-GaN increased as the chamber temperature increased,
whereas the etch rates for the AlGaN decreased with increase in chamber temperature.
Owing to this contrasting behavior, the selectivities increased as the chamber temperature
increased. However, the etch rate became saturated at a chamber temperature of 60°C.
The highest selectivity recorded was 53:1 (for p-GaN:AlGaN) at a chamber temperature
of 60°C, where the etch rates for p-GaN and AlGaN were 3.6 and 0.068 Å/s, respectively,
and the self-bias voltage was 198 V.
2. Fabrication of p-GaN Gated AlGaN/GaN HFET
The epitaxial structure consisted of a 70 nm p-GaN layer, a 15 nm Al0.2Ga0.8N layer,
a 320 nm i-GaN layer, and a 3.6 µm GaN buffer layer on a Si(111) substrate. After
cleaning with solvent and acid, the p-GaN layer outside the gate region was etched
via a two-step plasma etching process. As the self-bias voltage of the optimized selective
etching process was much higher than that of the low damage BCl3/Cl2 etching process used here, the two-step etching process was designed to etch the
p-GaN layer. First, the p-GaN layer was partially etched using the low damage BCl3/Cl2 etching process with an etch depth target of 50 nm. The self-bias voltage for this
low damage process was 20 V. Then, the remaining p-GaN layer was etched using the
optimized Cl2/N¬2/O2 selective etching process to minimize the over-etching time. Because the remaining
p-GaN layer was very thin, the over-etching time could be reduced significantly in
comparison with a single-step selective etching process. Therefore, any plasma induced
damage at the etched AlGaN surface can be minimized by the two-step etching process
as well. After the selective etching process, the oxidized surface was treated by
a buffered oxide etchand solution, and the sample was annealed at 500°C for 5 min
in an N2 ambient for damage recovery. The ohmic contacts were formed by a fully recessed Au-free
ohmic process, and the ohmic metal stack of Ti/Al/TiN was annealed at 550°C for 1
min in an N2 ambient. Mesa isolation was then carried out by the BCl3/Cl2 based plasma etching. TiN metal stacks were evaporated for the gate and pad electrodes,
and the surface was passivated with a 180 nm SiN¬x film using ICP chemical vapor deposition.
A source-connected field-plate was formed using a TiN/Ti metal stack, and the source-to-gate
distance, p-GaN gate length, and gate-to-drain distance were fabricated as 2, 4, and
11 μm, respectively, where the gate length inside the p-GaN region was 2 μm. The cross-sectional
schematic of the p-GaN gated AlGaN/ GaN HFET is shown in Fig. 2.
Fig. 2. Cross-sectional schematic of the p-GaN gated AlGaN/GaN E-mode HFET.
III. RESULTS AND DISCUSSION
To validate the selective etching process, capacitance–voltage (C-V) measurements
were conducted using a C-V test-element-group pattern fabricated on the AlGaN surface
where the p-GaN layer was etched. Fig. 3(a) shows the electron distribution versus the depth derived from the C-V characteristics
(12). The thickness of the AlGaN layer extracted from the C-V characteristics is 15 nm,
which is in agreement with the initial AlGaN barrier thickness. Data from an atomic
force microscopic image in Fig. 3(b) also confirmed that the etch depth was the same as that of the p-GaN layer.
Fig. 3. (a) Carrier distribution versus AlGaN depth for the etched region as derived
from C-V measurements, (b) atomic force microscopy data for the p-GaN etched region.
The transfer current–voltage (I-V) characteristics of the fabricated device were measured
at V
ds = 10 V. As shown in
Fig. 4(a), the fabricated p-GaN gated HFET exhibited excellent normally-off E-mode operation
with negligible hysteresis characteristics. A threshold voltage of 2.45 V, an on/off
ratio of ~10
9, and a maximum drain current density (I
d.max) of 182 mA/mm at V
gs = 8 V were achieved. Further, the off-state drain leakage current was < 10
-10 A/mm.
The output current–voltage characteristics are shown in Fig. 4(b), from which the specific on-resistance extracted at Vds = 2 V was 5.55 mΩ·cm2 by taking into account the intrinsic channel region between the source and drain.
Fig. 4. (a) Transfer characteristics, (b) output current–voltage characteristics of
the fabricated p-GaN gated AlGaN/GaN HFET.
To evaluate the trapping effects, the current collapse phenomenon was investigated
by pulsed measurements with a pulse width of 200 ns and a period of 1 ms. The pulsed
characteristics with different quiescent drain bias voltages are shown in
Fig. 5, where the quiescent gate bias voltage was fixed at 0 V. Owing to the field plate,
only a little current collapse was observed to occur under the short pulse conditions.
Fig. 5. Pulsed I-V characteristics of the fabricated p-GaN gated AlGaN/GaN HFET.
Off-state breakdown voltage was measured with V
gs = 0 V, using the Keithley 2410 and 2651A sourcemeters. As shown in
Fig. 6, a slight increase in the leakage current and no breakdown were observed up to V
ds = 1100 V, which was the limit for the equipment.
Fig. 6. Off-state breakdown characteristics of the fabricated p-GaN gated AlGaN/GaN
HFET.
IV. CONCLUSIONS
A selective p-GaN etching process was developed in this study using a Cl2/N2/O2 based ICP etching scheme. A selectivity of 53:1 (for p-GaN/AlGaN) was achieved using
the following process conditions: gas flow rates of Cl2/N2/O2 = 40/10/2 sccm, source RF power of 2000 W, bias RF power of 25 W, and chuck temperature
of 60°C. To minimize the plasma induced damage at the etched surface, a two-step etching
process was utilized to fabricate the p-GaN gated AlGaN/GaN HFET, where the upper
p-GaN layer was etched using a low damage nonselective etching process and the remaining
thin p-GaN layer was etched via a selective etching process developed in this work.
The fabricated p-GaN gated AlGaN/GaN HFET exhibited excellent E-mode operation with
a high threshold voltage of 2.45 V, low specific on-resistance of 5.55 mΩ·cm2, and high off-state breakdown voltage of > 1100 V. Thus, it is suggested that the
two-step etching process proposed in this work can be utilized for p-GaN gated AlGaN/GaN
HFET fabrication to not only minimize etching damage but also improve process reproducibility.
ACKNOWLEDGMENTS
This work was supported by the Basic Science Research Program (No. 2015R1A6A1A03031833
and No. 2019R1A2C1008894) and Material Component Development Program of MOTIE/KEIT
(10080736).
We thank Tae-Hyun Kim, Jun-Hyeok Yim, and Chan-Hee Jang of Hongik University, who
provided the fabrication support for this research.
REFERENCES
Mishra U.K., 2009, AlGaN/GaN HEMTs-An Overview of Device Operation and Applications,
Proceedings of the IEEE, Vol. 90, No. 60, pp. 1022-1031
Ambacher O., 2000, Two Dimensional Electron Gases Induced by Spontaneous and Piezoelectric
Polarization in Undoped and Doped AlGaN/GaN Heterostructures, Journal of Applied Physics,
Vol. 87, No. 1, pp. 334-344
Uemoto Y., 2007, Gate Injection Transistor (GIT)—A Normally-Off AlGaN/GaN Power Transistor
Using Conductivity Modulation, Journal of Applied Physics. IEEE Transactions on Electron
Devices, Vol. 54, No. 12, pp. 3393-3399
Landford W.B., 2005, Normally Off AlGaN/GaN High Electron Mobility Transistors with
P-InGaN Cap Layer, Journal of Applied Physics, Vol. 113, No. 3, pp. 034502
Huang S., 2018, Ultrathin-Barrier AlGaN/GaN Heterostructure: A Recess-Free Technology
for Manufacturing High-Performance GaN-on-Si Power Devices, IEEE Transactions on Electron
Devices, Vol. 65, No. 1, pp. 207-214
Hwang I., 2013, p-GaN Gate HEMTs With Tungsten Gate Metal for High Threshold Voltage
and Low Gate Current, IEEE Electron Device Letters, Vol. 34, No. 2, pp. 202-204
Lin Y., ye2018ar, tiA Novel Digital Etch Technique for p-GaN Gate HEMTtle, 2018 IEEE
International Conference on Semiconductor Electronics (ICSE), pp. 121-123
Zhong Y., 2017, Self-Terminated Etching of GaN with a High Selectivity Over AlGaN
Under Inductively Coupled Cl2/N2/O2 Plasma with a Low-Energy Ion Bombardment, Applied
surface science, Vol. 420, No. 31, pp. 817-824
Han Y.J., 2003, Highly Selective Dry Etching of GaN over AlGaN Using Inductively Coupled
Cl2/N2/O2 Plasmas, Japanese Journal of Applied Physics, Vol. 42, pp. 1139-1141
Lee J.M., 2000, Highly Selective Dry Etching of III Nitrides Using an Inductively
Coupled Cl2/Ar/O2 plasma, Journal of Vacuum Science & Technology B, Vol. 42, No. 10,
pp. 1409-1411
Zhu K., 2002, Plasma Etching of AlN/AlGaInN Superlattices for Device Fabrication,
Applied Physics Letters, Vol. 81, No. 25, pp. 4688
Schroder D.K., 2006, “Semiconductor Material and Device Characterization, pp. 61-65
Author
Won-Ho Jang received the B.S. and M.S. degrees in Electronic and Electrical Engineering
from Hongik University, Seoul, Korea, in 2016 and 2018, respectively.
He is currently pursuing the Ph.D. degree at Hongik University.
His research intereswnsts include the analysis of Gallium nitride devices.
Kwang-Seok Seo received the B.S. in electrical and computer engi-neering at Seoul
National University, Seoul, Korea, in 1976, M.S. degree in electrical engineering
at Korea Advanced Institute of Science and Technology in 1978.
And Ph.D. degree in electrical engineering from the University of Michigan Ann Arbor,
USA, in 1987.
He was a senior research engineer at the Korea Institute of Electronics Technology
until 1982.
He was a Postdoctoral Research Associate at the IBM T.J. Watson Research Center until
1988.
He is currently a Professor with the department of electrical and computer engineering,Seoul
National University, Seoul, Republic of Korea.
His research interests include high speed device physics and technology, compound
semiconductor devices.
Ho-Young Cha received the B.S. and M.S. degrees in Electrical Engineering from the
Seoul National University, Seoul, Korea, in 1996 and 1999, respectively, and the Ph.D.
degree in Electrical and Computer Engineering from Cornell University, Ithaca, NY,
in 2004.
He was a Postdoctoral Research Associate with Cornell University until 2005, where
he focused on the design and fabrication of wide bandgap semiconductor devices.
He was with the General Electric Global Research Center, Niskayuna, NY, from 2005
to 2007, developing wide-bandgap semiconductor sensors and high-power devices.
Since 2007, he is currently Professor in the School of Electronic and Electrical Engineering.
His research interests include wide-bandgap semiconductor devices.
He has authored over 100 publications in his research area.