eLearning

High-resolution triple-axis x-ray diffraction was per-
formed using both (004) (in figure 2) and (224) glancing exit
(GE) ω-2θ scans to determine the degree of strain relaxation
and the composition of the top InAlAs layer. To distinguish the
strain effect from the statistically distributed lattice plane tilts
(usually refered to asmosaicity, which is related to cross-hatch
issues on the surface) [19–21], the asymmetric two-bounce Si
(220) analyser crystal was placed in front of the detector in the
high-resolution triple-axis x-ray diffraction mode. Figure 3
shows the HRXRD ω-2θ (004) and (224) GE scans of sam-
ple B. To eliminate the effect of this tilt, two sets of (004)
and (224) scans were performed at two orientations, 0◦
and
180◦
, by rotating the sample. The average peak separation of
each set of scans was calculated. By comparing the (004),
(224) GE HRXRD, the strain relaxation and composition of
the top InAlAs layer were calculated using the Bede Scien-
tific PeakSplit computer program; the degrees of relaxation
for all samples are about 98% i.e. the top layers were nearly
fully relaxed. AFM was performed on a 5×5µm2
area. The
root mean square (RMS) surface roughness is listed in table 1.
The root mean square surface roughness of the sample grown
at 380 ◦
C is 0.802 nm, which is the smallest value compared
with those of the other samples.Sample B has more smooth
surface morphology, which indicates a better crystalline
quality.
Reciprocal space-mappingmeasurementswere performed
in (004) and (224) GE directions to study the mosaicity in
the top epilayers. The asymmetric two-bounce Si (220)
analyser crystal was placed in front of the detector in the high-
resolution triple-axis x-ray diffraction mode. The reciprocal
space mapping of sample B is illustrated in figure 4; the
(224) GE XRD mapping shows that the top layer was
nearly fully relaxed. There was a broadening of the top
layer diffraction peak in the perpendicular direction, which
corresponds to the ω-scan direction. This broadening indicates
mosaicity in the top layers, which is due to defects in the
epilayer [22].
High resolution triple-axis x-ray ω scans were carried out
around the diffraction peak of the top layers to investigate the
mosaicity in the samples; thus it can characterize the quality
of the material. The full-width at half-maximum (FWHM)
of the peak for samples is listed in table 1, it is obvious that
the FWHM of the peak for sample B is smaller than those
for samples A and C. The XRD ω scan is widely used to
investigate the mosaicity of the epitaxial layer; furthermore
it also characterizes the dislocations indirectly [1]. Since the
threading dislocations affect the FWHM value of the XRD
peak, the FWHM along the scan direction also indicates the
presence of threading dislocations [23], so the x-ray rocking
curves provide non-destructive measurements of dislocation
densities [24]. This suggests that sample B grown at 380 ◦
C
has less threading dislocation density and better crystalline
quality.
Figure 5 shows the PL spectra of samples A,B and C at
room temperature and low temperature(LT) of 14K. It is clear
fromthe PL spectra that sampleBshows that the PL intensity is
largest and FWHMis the smallest. It is very interesting that the
RT PL of sample C is very broad. Due to growth temperature
much lower than the congruent sublimation temperature, about
610 ◦
C for the InAlAs layer [25], this broadening originates
from the presence of clustering in the InAlAs layer which
seems to occur because of the large difference between the
In–As and Al–As bond energies [11]. The nonuniformity of
the alloy composition caused by the clustering is due to the
nonperfect incorporation of the Al cations. Furthermore, this
could not be corrected by a higher As flux as the growth of
sample E, which showed spotty patterns of the reconstructed
surface during growth resulting from the stronger sticking
coefficient of Al. It leads first to a spatial variation of the
band gap and secondly, to local internal strain. Both effects
influence the InAlAs PL band material [11].
On low-temperature step-graded InAlAs metamorphic
buffer layers, whose grown conditions are same as those
of sample B, the grown InAlAs/ InGaAs/GaAs MM-HEMT
structure consists of , from bottom to top, a 300 nm undoped
InAlAs layer, a 18 nm undoped InGaAs active layer, a 5 nm
undoped InAlAs spacer, a Si δ- doping layer with a doping
levelof7×1012
cm−2
, a 35 nmundoped InAlAs barrier, a 10 nm
n-InGaAs contact layer doped with Si at 7×1018
cm−3
.At
RT,the carrier concentrations are more than 3×1012
cm−2
and
Hall mobilities are more than 9000 cm−2
Vs−1
. At77K,the
carrier concentrations are more than 3×1012
cm−2
and Hall
mobilities are more than 35000 cm−2
Vs−1
. These values are
high enough to make devices.
4. Conclusion
The quality of low temperature compositionally step-graded
InAlAs metamorphic buffers grown under different growth
parameters on the GaAs substrate were investigated with
HRXRD, AFM, RHEED pattern and PL. The top InAlAs
layer is nearly fully relaxed. All the results show that
the quality of low temperature compositionally step-graded
InAlAs metamorphic buffers is very sensitive to growth
parameters due to the large difference between the In–As and
Al–As bond energies. The carrier concentrations and Hall
mobilities of the InAlAs/ InGaAs/GaAsMM-HEMT structure
on low-temperature step-graded InAlAs metamorphic buffer
layers grown in optimized conditions are high enough to make
devices.