1. phys. stat. sol. (a) 188, No. 1, 187–190 (2001)
AlGaN Resonant Tunneling Diodes Grown by rf-MBE
A. Kikuchi1), R. Bannai, and K. Kishino
Department of Electrical and Electronics Engineering, Sophia University, 7-1, Kioi-cho,
Chiyoda-ku, Tokyo 102-8554, Japan
(Received July 4, 2001; accepted July 12, 2001)
Subject classification: 73.40.Kp; S7.14
AlGaN resonant tunneling diodes (RTD) have been successfully grown by molecular beam epi-
taxy. Two kinds of RTD samples were fabricated; one is a double barrier type RTD which has
1 nm AlN barriers and a 0.75 nm GaN well, and the other is a superlattice barrier type RTD which
has six 1 nm AlN barriers and five 1 nm GaN wells. The negative differential resistance (NDR)
effect in the current–voltage characteristics was clearly observed at room temperature. For the
double barrier RTD sample, the NDR was observed at 2.4 V with a peak current density of
930 mA/cm2 and a peak-to-valley ratio of 3.1. For the superlattice barrier RTD sample, the NDR
was observed at 1.6 V. The peak current density and peak-to-valley ratio were 142 A/cm2 and 9.7,
respectively.
Introduction The III-nitride material system has many attractive properties such as a
large bandgap energy, large bandgap discontinuity, high peak electron velocity, high
saturation electron velocity and higher thermal stability. Many studies on AlGaN based
electrical devices have been reported on HEMT, FET, HBT, etc. From the point of
view of large bandgap discontinuity of ~1.95 eV (for AlN/GaN system), various quan-
tum effect device applications are expected; such as resonant tunneling diodes (RTD)
[1], intersubband transition near infrared optical devices [2] and so on. The RTD is
attractive for high-frequency functional device applications. The large bandgap disconti-
nuity requires monolayer-order thickness controllability and a smooth interface for the
epitaxial growth. In this respect, molecular beam epitaxy (MBE) seems to be a suitable
growth technique to realize a fine control of the nitride heterostructures. In fact, re-
cently, very high two-dimensional electron gas mobility over 50000 cm2/Vs [3, 4] and
near infrared intersubband transition [5] have been demonstrated by MBE grown
AlGaN/GaN heterostructure.
In this paper, we will descrive the first successful growth of AlN/GaN RTD by MBE
using rf-plasma nitrogen source (rf-MBE). Two kinds of RTDs with different AlN bar-
rier structures were adopted. One is conventional double barrier (DB) type, and the
other is superlattice barrier (SLB) type. The current–voltage characteristics showed
clear negative differential resistance (NDR) in both samples at room temperature. The
structural dependence of peak current density and peak-to-vally current ratio of NDR
will be described.
Epitaxial Growth and Device Processing The AlN/GaN RTD structures were grown
by rf-MBE on 4 mm thick n-type GaN templates which were grown on (0001) sapphire
substrates by MOCVD. The active nitrogen was supplied through an RF-plasma source
1
) Corresponding author; Phone: +81 3 3238 3323; Fax: +81 3 3238 3321;
e-mail: kikuchi@katsumi.ee.sophia.ac.jp
# WILEY-VCH Verlag Berlin GmbH, 13086 Berlin, 2001 0031-8965/01/18811-0187 $ 17.50þ.50/0
2. 188 A. Kikuchi et al.: AlGaN Resonant Tunneling Diodes Grown by rf-MBE
with high purity nitrogen gas (99.9999%). Prior to the rf-MBE growth, the backsides of
the GaN templates were coated with Ti to enhance heat absorption and mounted on
In-free Mo blocks. Prior to the MBE growth, the GaN templates were exposed to Ga
beam at 300 oC, then the substrate temperature was ramped to 700 oC and the active
nitrogen was irradiated to the substrate surface. In the growth process, the substrate
temperature was set to be 720 oC. The AlN layers were grown by migration enhanced
epitaxy (Al and nitrogen were alternatively supplied) [6] and the GaN layers were
grown by the shutter control method (Ga was continuously supplied and nitrogen was
periodically interrupted) [7]. The detailed structure of the two samples, DB-RTD and
SLB-RTD, will be shown in the following sections.
For the electrical measurements, the samples were processed into mesa-type device
structure. 80 mm square Ti (35 nm)/Al(130 nm) electrodes were formed through elec-
tron beam evaporation and conventional lift-off technique. Then 700 nm height mesa
structures were formed by atom beam etching using the electrodes as masks. The etch-
ing was carried out under the following conditions: Ar/CH4/H2 ¼ 10/10/80 mixed gas
flow with a total pressure of 4.0 Â 10 – Torr, bias voltage of 0.8 kV and a substrate
–3
o
temperature of 400 C. The etching rate was 15.5 nm/min. Finally, Ti (35 nm)/Al
(130 nm) lower electrodes in the width of 80 mm were formed with the same process as
top electrode formation.
Double Barrier RTD The first RTD structure is of conventional double barrier type
as shown in Fig. 1. In this sample, five layers of 8 nm thick high-temperature grown
AlN multiple interlayers (HT-AlN-MIL) separated by 40 nm thick GaN layers were
inserted between the MOCVD-GaN template and RTD structure to reduce threading
dislocations [6, 8, 9]. As a result of dislocation reduction by HT-AlN-MIL, we have
realized fine step flow GaN growth by rf-MBE on MOCVD-GaN template. On the HT-
AlN-MIL, DB-RTD structure was grown as a sequence of 800 nm thick Si-doped GaN
layer, 2 nm thick undoped GaN, 1 nm (4 ML) thick AlN barrier, 0.75 nm (3 ML) thick
Fig. 1 Fig. 2
Fig. 1 Schematic diagram of AlN/GaN double barrier RTD grown on MOCVD-GaN template
Fig. 2. Current–voltage characteristics of AlN/GaN double barrier RTD measured in 12.5 mV
steps at room temperature
3. phys. stat. sol. (a) 188, No. 1 (2001) 189
undoped GaN, 1 nm (4 ML) thick AlN barrier, 2 nm thick undoped GaN and 400 nm
thick Si-doped GaN layer. Here, the Si-doping level was estimated to be 8 Â 1017 cm – .–3
The current–voltage characteristics were measured with a single sweep mode from
0 V to forward and reverse direction in 12.5 mV steps at room temperature. Figure 2
shows the typical current–voltage characterristics of the DB-RTD. The NDR was ob-
served around 2.4 V with a peak current of 60 mA. The corresponding current density
was estimated as 930 mA/cm2. In this sample, the peak-to-valley current ratio was 3.1.
By solving the Schrodinger equation using square potential profile, the resonant energy
¨
level E1 was estimated to be at 800 meV from the GaN conduction band edge which
corresponds to the NDR voltage (%2E1/e) of 1.6 V. The difference of NDR voltage
between experiment and calculation may be caused by relatively high series resistance
and rectification property of the contact due to the damage induced by the atom beam
etching process or thickness reduction of GaN well layer to 2 ML. The hysteretic prop-
erty originated from high series resistance was not confirmed here, because of the sin-
gle sweep mode measurement. Almost all samples showed asymmetrical current–volt-
age characteristics, that is, NDR was observed only for the single side in the
measurement range. The asymmetric I–V characteristics may be originated from two
possible reasons, one is a piezoelectric field induced asymmetrical potential profile and
another is a rectification property in the contact. The later is not an essential problem,
and it can be avoided by optimization of the device processes. The origin of the asym-
metric property will be clarified by the precise theoretical simulation or optimization of
device processing.
Superlattice Barrier RTD Figure 3 shows the schematic diagram of the superlattice
barrier RTD, which consists of six AlN barrier layers. This structure was directly grown
on the MOCVD-GaN template, a sequence of 800 nm thick Si-doped GaN, six layers of
1 nm (4 ML) thick AlN barrier separated by 1 nm (4 ML) thick Si-doped GaN well
layers and 400 nm thick Si-doped GaN top layer. Here, the doping level of the n-GaN
was estimated to be 2 Â 1018 cm – , which was 2.5 times higher than for DB-RTD case.
–3
Fig. 3 Fig. 4
Fig. 3. Schematic diagram of AlN/GaN superlattice barrier RTD grown on MOCVD-GaN template
Fig. 4. Current–voltage characteristics of AlN/GaN superlattice barrier RTD measured in 26.7 mV
steps at room temperature
4. 190 A. Kikuchi et al.: AlGaN Resonant Tunneling Diodes Grown by rf-MBE
The typical current–voltage characteristics at room temperature is shown in Fig. 4.
The measurement was carried out in 26.7 mV steps. NDR was clearly observed at
about 1.6 V. This smaller NDR voltage compared to DB-RTD may arise from the low-
er resonant energy level estimated to be 0.55 eV. A maximum peak-to-valley current
ratio of 9.7 was obtained. The peak current was 9 mA, which corresponds to 142 A/cm2
in the current density. This superior characteristics may be brought about by the in-
crease of doping level and introduction of superlattice barriers. The transport mecha-
nism [10] in the superlattice barrier is still under investigation and will be described in
further studies.
Conclusions We have demonstrated room temperature NDR effects in AlN/GaN reso-
nant tunneling diodes grown by rf-MBE for the first time. Two kinds of RTD structure,
double barrier type RTD and superlattice barrier type RTD, were adopted. For the
DB-RTD, the NDR was observed at 2.4 V with a peak current density of 930 mA /cm2
and a peak-to-valley ratio of 3.1. For the superlattice barrier RTD, the NDR was ob-
served at 1.6 V. The peak current density and a peak-to-valley ratio were 142 A/cm2
and 9.7, respectively. These results may open a new application field for AlGaN materi-
al systems such as high-speed functional electronic devices and near infrared quantum
cascade devices.
Acknowledgements This work was supported by the ‘‘Research for the Future” pro-
gram of the Japan Society for the Promotion of Science No. JSPS-RFTF97P00102, and
partly by science research grant-in-aid No. 13750315 from the Ministry of Education,
Culture, Sports, Science and Technology. The authors would like to acknowledge Prof.
T. Waho for his valuable discussion, Associate Prof. K. Shimomura for his permission to
utilize the experimental equipments, and Mr. K. Serizawa for his valuable help on de-
vice fabrication and measurement.
References
[1] L. L. Chang, L. Esaki, and R. Tsu, Appl. Phys. Lett. 24, 593 (1974).
[2] J. Faist, F. Capasso, D. L. Sivco, C. Sirtori, A. L. Hutchinson, and A. Y. Cho, Science 264,
553 (1994).
[3] I. P. Smorchkova, C. R. Elsass, J. P. Ibbetson, R. Vetury, B. Heying, P. Fini, E. Haus, S. P.
DenBaars, J. S. Speck, and U. K. Mishra, J. Appl. Phys. 86, 4520 (1999).
[4] M. J. Manfra, L. N. Pfeiffer, K. W. West, H. L. Stormer, K. W. Baldwin, J. W. P. Hsu, D. V.
Lang, and R. J. Molnar, Appl. Phys. Lett. 77, 2889 (2000).
[5] C. Gmachl, H. M. Ng, S.-N. G. Chu, and A. Y. Cho, Appl. Phys. Lett. 77, 3722 (2000).
[6] D. Sugihara, A. Kikuchi, K. Kusakabe, S. Nakamura, Y. Toyoura, T. Yamada, and K. Kishino,
Jpn. J. Appl. Phys. 39, L197 (2000).
[7] A. Kikuchi, M. Yoshizawa, M. Mori, N. Fujita, K. Kushi, H. Sasamoto, and K. Kishino, J.
Cryst. Growth 189/190, 109 (1998).
[8] A. Kikuchi, T. Yamada, S. Nakamura, K. Kusakabe, D. Sugihara, and K. Kishino, Jpn. J. Appl.
Phys. 39, L330 (2000).
[9] A. Kikuchi, T. Yamada, S. Nakamura, K. Kusakabe, D. Sugihara, and K. Kishino, Mater. Sci.
Eng. B 82, 12 (2001).
[10] F. Capasso, K. Mohammed, and A. Y. Cho, IEEE J. Quantum Electron. 22, 1853 (1986).