High radiation tolerance of InAs/AlSb high-electron-mobility transistors

Abstract

InAs/AlSb-based high-electron-mobility transistors (HEMTs) were irradiated with 2 MeV protons. Radiation damage caused the source-drain current Ids to decrease nearly linearly with fluence Φ at a rate of Δ[Ids(Φ)/Ids(0)]/ΔΦ ≈ 7×10−16 cm2. Radiation-induced decreases in Ids have been observed for other HEMT material systems, and have been attributed to high-efficiency defect-induced scattering of carriers out of the two-dimensional electron gas. However, in the InAs/AlSb system the rate of decrease of Ids is about 140 times less than that for typical GaAs/AlGaAs HEMTs. An explanation is presented in which the high radiation tolerance of InAs/AlSb HEMTs is related to carrier reinjection and the unusually large energy offset between the AlSb barriers and the InAs quantum well.

Full text

High radiation tolerance of InAs/AlSb high-electron-mobility transistors
B. D. Weaver,a兲J. B. Boos,b兲N. A. Papanicolaou,b兲B. R. Bennett,c兲D. Park,b兲and
R. Bassb兲
Naval Research Laboratory, Code 6818, Washington, D.C. 20375
共Received 13 April 2005; accepted 27 August 2005; published online 18 October 2005兲
InAs/AlSb-based high-electron-mobility transistors 共HEMTs兲were irradiated with 2 MeV protons.
Radiation damage caused the source-drain current Ids to decrease nearly linearly with fluence ⌽at
a rate of ⌬关Ids共⌽兲/Ids共0兲兴/⌬⌽⬇7⫻10−16 cm2. Radiation-induced decreases in Ids have been
observed for other HEMT material systems, and have been attributed to high-efficiency
defect-induced scattering of carriers out of the two-dimensional electron gas. However, in the
InAs/AlSb system the rate of decrease of Ids is about 140 times less than that for typical
GaAs/AlGaAs HEMTs. An explanation is presented in which the high radiation tolerance of
InAs/AlSb HEMTs is related to carrier reinjection and the unusually large energy offset between the
AlSb barriers and the InAs quantum well. 关DOI: 10.1063/1.2115071兴
High-electron-mobility transistors 共HEMTs兲fabricated
from InAs/AlSb are potentially excellent candidates for sat-
ellite applications due to their low operating voltages, low
power consumption, and high speed. However, in order to be
useful in space these devices must tolerate the radiation en-
vironment of Earth’s Van Allen belts. HEMTs, in general, are
attractive for space applications because, being majority car-
rier devices, they tend to be less sensitive to displacement
damage effects than minority carrier transistors.1Radiation-
effects studies have been performed on HEMTs fabricated
from the GaAs/AlGaAs, InGaAs / AlGaAs, InGaAs/InGaP,
and InGaAs/InAlAs materials systems.2–9 We report here the
results of a displacement damage experiment on InAs/AlSb
HEMTs, and show that they are the most radiation-tolerant
HEMTs studied to date.
Details on the material heterostructure and device fabri-
cation have been presented elsewhere10,11 so only a brief
summary is given here. Heterostructures were grown by mo-
lecular beam epitaxy on a semi-insulating GaAs substrate. A
2.4
␮
m thick AlSb layer was grown to accomodate the 8%
lattice mismatch. Additional layers were, respectively, a
150 Å thick InAs channel, a 125 Å thick AlSb barrier, a
50 Å thick In0.5Al0.5As layer, and a 20 Å thick InAs cap. As
determined by Hall measurements, the room-temperature
mobility and sheet density of the starting material were
27 000 cm2/V s and 1.2⫻1012 cm−2, respectively. The het-
erostructures were fabricated into devices by using electron-
beam lithography, heat treatment, and wet etching, and re-
sulted in HEMTs having a 0.15
␮
m gate length. The HEMTs
exhibit characteristic frequencies fTand fmax of 125 GHz and
100 GHz, respectively, at source-drain voltages Vds of 0.4 V.
By fitting the measured Sparameters to a commonly-used
circuit topology, an intrinsic fTof 200 GHz was obtained
after removal of the gate bonding pad capacitance from the
equivalent circuit. The value of fmax is primarily limited by
gate metal resistance.
Before irradiation, current-voltage 共I-V兲curves were
measured for twenty devices on a wafer as Vds was swept
from 0 to 0.5 V and the gate voltage Vgs was varied in steps
from 0 to −0.75 V. Typical results, shown in the main body
of Fig. 1, exhibit the familiar on-off switching behavior of
transistors as Vgs→−0.75 V. The maximum power drain
for the HEMT of Fig. 1 was 6–7 mW at Vds =0.5 V and
Vgs=0. Gate leakage currents were also measured, but were
generally small and unaffected by irradiation so are not dis-
cussed further here.
Following the initial measurements, samples were irra-
diated at room temperature with incremental fluences of
2 MeV protons incident 7° from the surface normal to pre-
vent channeling effects. The range of 2 MeV protons in
InAs/AlSb / GaAs is much greater than the device thickness,
so the protons traversed the devices without significant en-
ergy loss and thus created a nearly uniform profile of dis-
placed atoms, mainly vacancies, and interstitials. At the
maximum fluence, ⌽=7⫻1014 H+/cm2, the initial induced
defect concentration cwas c⬃1.4⫻10−4 displacements per
target material atom. Generally, ⌽and care linearly
related:12
c=a⌽N,共1兲
where the parameter adepends on the material density and
the energy threshold for atomic displacement, and N, the
a兲Author to whom correspondence should be addressed; electronic mail:
weaver1@ccf.nrl.navy.mil
b兲Also at: NRL, Code 6853, Washington, D.C.
c兲Also at: NRL, Code 6876, Washington, D.C.
FIG. 1. Main body: I-Vcurves of a typical unirradiated 2⫻25
␮
m2
InAs/ AlSb HEMT. Inset: I-Vcurves after irradiation to a fluence of 7
⫻1014 2 MeV H+/cm2.
APPLIED PHYSICS LETTERS 87, 173501 共2005兲
0003-6951/2005/87共17兲/173501/3/$22.50 87, 173501-1
Downloaded 18 Oct 2005 to 132.250.135.221. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp

nonionizing energy loss, represents the density-normalized
rate at which incident particles lose their energy to displace-
ment damage. For 2 MeV protons in InAs/AlSb/ GaAs,
N⬇3.1⫻10−2 MeV cm2/g.
I-Vcurves for a typical irradiated HEMT are shown in
the inset of Fig. 1 for ⌽=7⫻1014 H+/cm2. As can be seen,
irradiation causes the drain current Ids to decrease for all
measured values of Vds and Vgs but does not alter the general
shape of the I-Vcurves or affect the switching behavior. In
HEMTs in general, radiation damage decreases the mobility,
the carrier concentration, the conductance, the characteristic
frequencies and the gain, and increases the power drain.
In Fig. 2, the normalized drain current Inorm
⬅Ids共⌽兲/Ids共0兲at Vgs=0 is plotted versus Vds for various
fluences. Normalizing the current in this way eliminates
variations in the preirradiation values of Ids that might occur
between devices, and allows the fluence dependence of Ids to
be examined. An example is shown in Fig. 3, where values
of Inorm are plotted versus ⌽for Vgs=0 and Vds =0.5 V. It can
be seen that Inorm decreases nearly linearly with ⌽at a rate
⌬Inorm/⌬⌽⬇7⫻10−16 cm2. Identical values of ⌬Inorm /⌬⌽
were obtained for all applied voltages 共as long as Ids⫽0兲and
so the value of ⌬Inorm/⌬⌽ is a general result of the
experiment.
Near-linear particle-induced decreases in Inorm have also
been observed for HEMTs fabricated from GaAs/AlGaAs,
InGaAs/AlGaAs, InGaAs / InGaP, and InGaAs/InAlAs.2–9
One simple explanation for this finding is that since carriers
in a two-dimensional electron gas 共2DEG兲are dimensionally
constrained, and because scattering from a defect alters a
carrier’s momentum in three dimensions, radiation-induced
scattering in the 2DEG region of HEMTs removes carriers
from the conduction state with high efficiency.12 As a result,
the drain current has been shown to decrease in direct pro-
portion to the induced defect concentration, and hence in
proportion to ⌽via Eq. 共1兲. However, if carrier removal
were the only factor determining the fluence dependence of
Ids all 2DEG-based transistors would exhibit the same radia-
tion response regardless of the material system. Such is not
the case. An additional phenomenon—carrier reinjection—
also apparently affects Ids.
Due to the energy difference ⌬Ebetween the
conduction-band edges in the barrier and the 2DEG in the
channel of a HEMT, it is energetically favorable for carriers
that have been scattered out of the 2DEG to be reinjected.
Provided that the reinjection probability increases with in-
creasing well depth ⌬E, the normalized drain current is ex-
pected to obey the following relationship:12
Inorm =1−p共c兲P共⌬E兲,共2兲
where p共c兲is the probability that a given carrier is scattered
out of the 2DEG by an interaction with a radiation-induced
defect and P共⌬E兲is the probability that the same carrier is
subsequently reinjected. For comparatively small defect con-
centrations, p共c兲⬀c. Then, Eq. 共1兲can be used to rewrite Eq.
共2兲as
Inorm =1−abP共⌬E兲N⌽,共3兲
where bis the ratio of proportionality between p共c兲and c.
关Although a,b, and P共⌬E兲cannot be individually deter-
mined the product a⫻b⫻P共⌬E兲can be determined from the
rate of decrease of Inorm with ⌽.兴Thus, the scattering of
carriers out of the 2DEG results in a linear decrease of Ids
and Inorm with fluence, while the values of ⌬Eand P共⌬E兲
determine the rate of decrease.
Equation 共3兲makes it possible to compare displacement
damage effects in HEMTs fabricated from different material
systems and irradiated with different particles. This has been
done for GaAs/AlGaAs HEMTs irradiated with 2 MeV H+,
3–20 MeV H+, 220 MeV C+, 14.5 MeV Si+, and 1 MeV
neutrons.12 For InGaAs/AlGaAs HEMTs, the incident par-
ticles were 1 MeV neutrons, 20 MeV He+, and 1 MeV e−.
For InGaAs/InGaP HEMTs, the incident particles were
20 MeV He+and 220 MeV C+; and for InGaAs / InAlAs
HEMTs the incident particles were 3 MeV helium ions. In
each case, Inorm decreased nearly linearly with fluence, while
the rate of decrease correlated with the magnitude of ⌬E.
This observation led to the prediction that InAs/AlSb
HEMTs, for which ⌬E=1.26 eV, would be significantly
more radiation tolerant than GaAs/AlGaAs HEMTs, for
which ⌬E=0.28– 0.38 eV, depending on the fractional con-
centration of aluminum.
In Fig. 4, values of ⌬Inorm/⌬⌽ are shown plotted for
equivalent fluences of 2 MeV protons for the various mate-
rials systems. In cases where samples were irradiated with
particles other than 2 MeV protons, a standard scaling
method was used to convert relative damage rates to a single-
particle equivalent by comparing the ratios of the nonioniz-
ing energy losses.12 The results show that the drain current in
InGaAs/InGaP HEMTs is about five times less sensitive to
radiation damage than it is in GaAs/AlGaAs and
InGaAs/AlGaAs HEMTs. Also, InGaAs/ InAlAs HEMTs are
about 40 times more radiation tolerant and InAs/AlSb are
FIG. 2. Normalized drain current Inorm =Ids共⌽兲/Ids共0兲plotted vs Vds for vari-
ous fluences, and at Vgs=0.
FIG. 3. Normalized drain current versus proton fluence at Vds= 0.5 V and
Vgs= 0. A near-linear decrease in current is observed for all operating gate
and drain voltages measured.
173501-2 Weaver et al. Appl. Phys. Lett. 87, 173501 共2005兲
Downloaded 18 Oct 2005 to 132.250.135.221. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp

about 143 times more tolerant than GaAs/AlGaAs and
InGaAs/AlGaAs HEMTs. Thus, we can confirm the predic-
tion that InAs/AlSb HEMTs are highly tolerant of displace-
ment damage effects, as a probable result of the large energy
offset between the conduction-band edges in the barriers and
quantum well.
This work was supported in part by the Office of Naval
Research and the Defense Advanced Research Projects
Agency.
1A. Holmes-Siedle and L. Adama, Handbook of Radiation Effects 共Oxford,
England, 1993兲, pp. 52–54.
2M. J. O’Loughlin, IEEE Trans. Nucl. Sci. 34, 1808 共1987兲.
3W. T. Anderson, A. R. Knudson, A. Meulenberg, H.-L. Hung, J. A. Rous-
sos, and G. Kiriakidis, IEEE Trans. Nucl. Sci. 37, 2065 共1990兲.
4G. J. Papaioannou, M. J. Papastamatiou, and A. Christou, J. Appl. Phys.
78, 3066 共1995兲.
5H. Ohyama, E. Simeon, S. Kuroda, C. Claeys, Y. Takami, T. Hakata, and
H. Sunaga, IEEE Trans. Nucl. Sci. 45, 2861 共1998兲.
6H. Ohyama, K. Yajima, E. Simoen, T. Katoh, C. Claeys, Y. Takami, K.
Kobayashi, M. Yoneoka, M. Nakabayashi, T. Hakata, and H. Takizawa,
IEEE Trans. Nucl. Sci. 47,2546共2000兲.
7H. Ohyama, E. Simoen, S. Kuroda, C. Claeys, Y. Takami, T. Hakata, K.
Kobayashi, M. Nakabayashi, and H. Sunaga, Microelectron. Reliab. 41,
79 共2001兲.
8B. Jun, S. Subramanian, and A. Peczalski, IEEE Trans. Nucl. Sci. 48,
2250 共2001兲.
9E. M. Jackson, B. D. Weaver, S. Ardala, R. Wilkins, A. C. Seabaugh, and
B. Brar, Appl. Phys. Lett. 79, 2791 共2001兲.
10J. B. Boos, W. Kruppa, B. R. Bennett, D. Park, S. W. Kirchofer, R. Bass,
and H. B. Dietrich, IEEE Trans. Electron Devices 45, 1869 共1998兲.
11B. R. Bennett, B. P. Tinkham, J. B. Boos, M. D. Lange, and R. Tsai, J. Vac.
Sci. Technol. B 22, 688 共2004兲.
12B. D. Weaver and E. M. Jackson, Appl. Phys. Lett. 80,2791共2001兲.
FIG. 4. Values of ⌬Inorm /⌬⌽ plotted for equivalent fluences of 2 MeV
protons for five different HEMT materials systems. If GaAs/ AlGaAs
HEMTs are arbitrarily assigned a radiation tolerance of 1 then the tolerances
of InGaAs/ AlGaAs, InGaAs/InGaP, InGaAs/ InAlAs, and InAs/ AlSb
HEMTs are, respectively, about 1, 5, 40, and 140.
173501-3 Weaver et al. Appl. Phys. Lett. 87, 173501 共2005兲
Downloaded 18 Oct 2005 to 132.250.135.221. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp