Hot Electron Energy Relaxation In AlGaN/GaN Heterostructures

1
Hot Electron Energy Relaxation In AlGaN/GaN
Heterostructures
1 School Of Physics And Astronomy, University of
Nottingham, University Park, Nottingham, UK 2
School Of Electrical And Electronic Engineering,
University of Nottingham, University Park,
Nottingham, UK 3 Institute For Microstructural
Sciences, National Research Council Canada, M-50
Ottawa, Ontario, Canada
C.E. Martinez1, N.M. Stanton1, A.J. Kent1, M.L.
Williams2, I. Harrison2, H. Tang3, J.B. Webb3 And
J.A. Bardwell3
BACKGROUND Gallium nitride and its alloys have
received much attention in recent years due to
their applications in high power, high frequency
and high temperature electronic devices.
However, many of the fundamental physical
processes in this important material system are
not fully understood. One such process is that
of hot carrier energy relaxation, which is of
fundamental importance to the performance of
semiconductor electronic devices. In this work
we have measured the electron energy relaxation
rate in a series of AlGaN/GaN heterostructures.
The results are compared with the results of
numerical calculations, allowing the details of
the 2D electron-phonon interaction in GaN to be
determined.
THE EXPERIMENT (CONT.) Zero field measurements
were made of the device resistance as a function
of the power dissipated in the device. To obtain
Te, these measurements were compared with a
calibration of the device resistance as a
function of temperature. In these relatively high
mobility devices, a novel technique to separate
the contributions of impurity and phonon
scattering was used to obtain the temperature
calibration Ouali et. al. Physica B 263 239
(1999).
FIGURE 1 The longitudinal device resistance for
the template layer sample as a function of
magnetic field for two different lattice
temperatures. Shubnikov-de Haas (SdH)
oscillations are clearly visible from 2T
onwards. FIGURE 2 The device resistance as a
function of magnetic field at the base
temperature (template layer sample), for
different currents through the device.
THE RESULTS Figure 3 shows the energy loss rates
per electron as a function of Te for the samples
measured. To calculate the power loss we used
the same approach as for GaAs, with appropriate
material parameters Stanton et. al.
phys.stat.sol (b) 228 607 (2001). The GaN
template layer and the SiC substrate sample show
similar behaviour. The two sets of data obtained
by the different methods are in excellent
agreement in the range of overlap 2K lt Te lt 8K.
For electron temperatures Telt 10K, we observe a
Te5 dependence of the power loss on electron
temperature. Comparison with the experimental
data suggests that PE coupling is dominant at low
temperatures. When considering the sapphire
sample, we observe a Te4.5 dependence of power
loss. That the SdH measurement agrees with the
calculated curve while the zero-field measurement
shows a more efficient power loss indicates the
presence of a parallel conduction channel in the
sample. In this situation, the magnetoresistance
measurement would be sensitive only to the 2D
carriers. The zero-field transport measurement
however would include any contribution from a
parallel channel, then the measured energy loss
rate is a sum of the loss rates from the two
separate pathways.
THE SAMPLES The three samples (Table 1) studied
were undoped AlGaN/GaN heterostructures grown by
ammonia MBE on sapphire and SiC substrates and a
GaN template layer /sapphire substrate Webb et.
al. Phys. Rev. B 66 245305 (2002). For the
samples grown on sapphire and SiC, the layers
consisted of 2mm C-doped GaN, followed by a GaN
channel layer of 200nm, capped with 20nm
AlxGa1-xN (x 0.05 - 0.10). The template sample
structure differed only in the absence of the
thick GaN layer. A Hall bar geometry was defined
using standard photolithographic techniques and
the mesa formed by reactive ion etching.
Ti/Al/Ti/Au contacts were formed by thermal
evaporation and annealing.
FIGURE 3A, B, C
TABLE 1 The sample substrates and parameters.
THE EXPERIMENT All measurements were made in a
liquid helium cryostat with a base temperature of
T1.5K. Two methods of determining the electron
temperature, Te, as a function of the power
dissipated In the device were used. In the first
method the amplitude of the Shubnikov-de Haas
oscillations was used. Figures 1 and 2 show the
longitudinal device resistance as a function of
magnetic field, when at two different lattice
temperatures (Fig1) and at the base temperature
for two different currents through the device
(Fig2). Comparing these two measurements, for a
range of lattice temperatures and currents allows
us to obtain the power loss per electron for Te lt
8K.
CONCLUSIONS The hot electron energy relaxation
rate in a series of AlGaN/GaN heterostructures
grown on different substrate materials has been
measured over the electron temperature range
2K-40K. This wide range of temperatures was
achieved using a combination of magnetoresistance
and zero-field transport measurements. For the
SiC and template layer substrates, good agreement
between the two measurement techniques is
observed, and numerical calculations agree well
with the experimental results. For the sapphire
sample, the discrepancy between measurement
techniques and the calculation is explained by
the presence of a parallel conduction channel.
FIGURES 3A, 3B, 3C The hot electron energy
relaxation rates for each of the samples, showing
the rates as determined from the magnetotransport
measurements (circles), zero-field measurements
(squares) and a comparison with the calculated
curve (solid line). (A) GaN template on sapphire
(B) Silicon carbide substrate (C) Sapphire
substrate.