Modeling of Electron Transport in GaN-Based
Materials and Devices
S. Vitanov*, V. Palankovski*, R. Quay1" and E. Langer*
* Institute for Microelectronics, TUWien, Gufihausstrafie 27-29/E360, 1040 Wien, Austria
^Fraunhofer Inst, for Solid-State Physics (IAF), Tullastr. 72, Freiburg, Germany
Abstract. Material models which incorporate the basic characteristics of the underlying physics in a given semiconductor
material are the core of device modeling. We employ a Monte Carlo (MC) technique to investigate stationary electron transport
in GaN and AlGaN [1]. We obtain a set of model parameters which gives agreement with experimental data available for
different physical conditions (doping, temperature, electric field, etc.). Such a calibrated set of models and model parameters
delivers valuable data for low-field mobility, velocity saturation, energy relaxation times, etc. We use these data as a basis for
the development of analytical models for the numerical simulation of GaN-based electron devices. As a particular example
we analyze an AlGaN/GaN HEMT with lg=300 nm from IAF using the two-dimensional device simulator Minimos-NT [2].
We study the impact of different models and efects (polarization charge, thermionic field emission, self-heating effects).
Keywords: Modeling, GaN, HEMT
PACS: 72.80.Ey, 73.61.Ey
INTRODUCTION
AlGaN/GaN based high electron mobility transistors
(HEMTs) have been subject of extensive investigations
in the last years. Their performance makes them suitable
for power amplifiers in infrastructure base station applications. In order to fully develop the potential of the device, an accurate simulation model is needed.
SIMULATION RESULTS
We employ a single-particle MC technique to investigate
stationary electron transport in GaN. Our model includes
the three lowest valleys of the conduction band. Several
stochastic mechanisms such as acoustic phonon, polar
optical phonon, inter-valley phonon, ionized impurity
scattering, and piezoelectric scattering are considered
[1]. Fig. 1 compares experimental [3, 4, 5, 6, 7] and
MC simulation data for the low-field electron mobility
in GaN as a function of the free carrier concentration.
The figure also includes an analytical model fitted to
our MC simulation results. This model is incorporated
in the two-dimensional device simulator MINIMOS-NT
[2]. For device simulation, we consider electron energy
relaxation times which depend on electron energy and
lattice temperature. Fig. 2 compares our analytical model
to MC simulation data for GaN.
Fig. 3 gives an example of a typical fully planar
GaN/AlGaN structure. All layers are non-intentionally
doped except the 5-doping which is introduced to provide additional carriers.
The crucial factor building the channel in HEMTs is
the polarization charges at the AlGaN/GaN heterointerfaces. The positive charge at the channel/spacer interface
is compensated by a negative surface charge at the barrier/cap interface. An optimum value of 1.1 x 1013 cm~2
is found. As can be seen in Fig. 4 (the electron concentration for VDS=7 V, VGS=0 V is shown) the device is a
normally on transistor. Another unknown is penetration
depth of the drain/source metal contacts which may build
an alloy with the AlGaN supply layer. We assume a metal
diffusion to the 5-doping in our simulations.
We further assess the impact of thermionic emission
and field emission (tunneling) effects which critically determine the current transport across the heterojunctions.
An optimal tunnel length of 7.5 nm is found.
Since the longitudinal electric field in the channel
reaches peak values of above 500 kV/cm, a hydrodynamic approach is used to properly model the electron transport and energy relaxation. We further account
for self-heating (SH) effects using a global temperature
model. Fig. 5 shows simulated and measured data for the
transfer curves. Fig. 6 compares measured and simulated
output characteristics.
CONCLUSION
We incorporate a new material model in our twodimensional device simulator. Our results allow not only
to get a good agreement between simulation and measured electrical data of AlGaN/GaN HEMTs, but to gain
understanding and insight in the effects taking place in
the devices.
CP893, Physics ofSemiconductors, 28' International Conference
edited by W. Jantsch and F. Schaffler
© 2007 American Institute of Physics 978-0-7354-0397-0/07/$23.00
1399
1750
0.10
m
^*^
1500 1250 .
^ \
* \,
\
fl •
Texp. coll. Chin 1994 :
• exp. Gaskill 1995 J
Oexp. coll. Schwierz
*exp. Koehler2003
+exp. Zanato 2004
J
— MC simulation
— new model
^\
\
1000 $>
0.09 - • - - • exp. data
0.07 -
O
500
^
(y
A)
0.04 -
\
250 -
----m
''' ,w~~
0.01
io1J
10'" 1 0 "
10'° 1 0 "
0.00
10"
- V ==1V
DS
<i-4
-5
-3
-2
-1
FIGURE 5. Transfer characteristics
0.30
0.12
g 0.25
0.10
I 0.20
-1
Y~
V G S [V]
FIGURE 1. Mobility vs. concentration (Monte Carlo results
and models used in the device simulator
^
: - (
y
Carrier concentration [cm ]
^
-
0.03
\
0.02
0
10'"
-
VDS=7V
2 0.05
\
750
<>•
-
0.06 -
0
9
/
- simulation
0.08
- exp data
sirm
v ==ov "
*GS
\
V =-IV
0.08
r^^
1
^
= -2V'_
V
2 0.06
1 0.15
*GS
IS
2 0.10
><
|
J '
«MCdata300K
TMCdata400K
• MCdata500K
model
Jr
Jr
0.05 f
0.00
10
20
30
40
50
60
:
0.04
;
0.02
W Gate
source J G a N C a p
1 AlGaN Barrier
V = -4V
ft/
ft
* GS
""""""TT
0.00
70
0
2
5
8
10
.
i
12
.
.
15
V D S [V]
FIGURE 2. Electron energy relaxation times as a function of
electron temperature for different lattice temperatures
|
* GS
/*
T /300K
^ ^ ^ ^ H
V =- 3 V fit
f"
FIGURE 6. Output characteristics
ACKNOWLEDGMENTS
B f SiN passivation J 1
^rain^J
The authors acknowledge support from Austrian Science
Fund (FWF), Project START Y247-N13, and by the
TARGET European Network of Excellence.
8 -doping
AlGaN Spacer
GaN Channel
REFERENCES
GaN Buffer
FIGURE 3. Layer structure of high electron mobility transistors considered in this work
u
FIGURE 4. Electron concentration [cm ] in the device
1. V. Palankovski, A. Marchlewski, E. Ungersbock, and
S. Selberherr, Proc. 5th MATHMOD, CDROM, Vienna,
pp. 14-1-14-9, 2006.
2. http://www.iue.tuwien.ac.at/software/minimos-nt.
3. V. Chin, T. Tansley, and T. Osotachn, J.Appl.Phys., vol. 75,
no. 11, pp. 7365-7372, 1994.
4. D. Gaskill, L. Rowland, and K. Doverspike, Properties of
Group III Nitrides, no. 11 in EMIS Datareviews Series, sec.
3.2, pp. 101-116, IEEINSPEC, 1994.
5. K. Kohler, S. Miiller, N. Rollbiihler, R. Kiefer, R. Quay, and
G. Weimann, Proc. Intl. Symp. Compound Semiconductors,
pp. 235-238, 2003.
6. D. Zanato, N. Balkan, G. Hill, and W. J. Schaff,
Superlattices & Microstructures, vol. 36, no. 4-6, pp.
455-463, 2004.
7. F. Schwierz, Solide-State Electron., vol. 49, no. 6, pp.
889-895, 2005.
1400