Professional Documents
Culture Documents
36
36
36
The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elseviers archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy
SIMS characterization of segregation in InAs/GaAs heterostructures
S. Gallardo
a
, Y. Kudriatsev
a,
*, A. Villegas
a
, G. Ramrez
a
, R. Asomoza
a
, E. Cruz-Herna ndez
b
,
J.S. Rojas-Ramirez
b
, M. Lo pez-Lo pez
b
a
Electrical Engineer Department, Centro de Investigacion y de Estudios Avanzados del IPN, Apartado postal 14470, Mexico, D.F.
b
Physics Department, Centro de Investigacion y de Estudios Avanzados del IPN, Apartado postal 14470, Mexico, D.F.
1. Introduction
Epitaxial growth of solid solutions is accompanied in some cases
by a segregation of the solution components which can strongly
affect thin layers like quantumwells. InGaAs based heterostructures
represent typical example of an undesired segregation process.
However, these structures are extensively used in microelectronics
for production of special devices. So, effective analysis of segregation
process in such structures is strongly desired. Ideally a segregation
free growth method should be developed.
Ion beam sputtering techniques like SIMS are commonly
employed for study of segregation over semiconductor hetero-
structures [13]. Unfortunately, there is a strong smearing of the
initial distribution of elements in heterostructures due to some
physical processes inherent to SIMS, and caused by ion bombard-
ment like ion induced mixing and Sputtering Induced Roughness
formation over analyzed surface. These ballistic processes and
the segregation during growth process are independent from each
other. In our study we separated them using a recently developed
SIMS Depth Resolution Function (DRF), which includes ion
sputtering induced processes. We convolved the DRF with
exponentially increasing function, which described the segrega-
tion and acquired the segregation parameters for different regimes
of sample preparation.
2. Experimental
Experimental samples were grown on GaAs(1 0 0) undoped
semi-insulating substrates by molecular beam epitaxy (MBE)
employing a Riber C21 system. First, native oxide on the
substrates surface was desorbed by heating substrate at 580 8C
under As
4
ux. Next, a 300 nm thick buffer layer was grown at
600 8C for smoothing out surface roughness caused by the oxide
desorption process. Then, ve InAs d-layers (thickness of $1
ML = 3 A
2p
p
s
mix
exp
z z
0
2
2s
2
mix
_ _
(1)
Surface roughness induced by sputtering:
Iz
rg
1
2p
p
s
rg
exp
z z
0
2
2s
2
rg
_ _
(2)
Recoil Implantation:
Iz
rc
1
l
rc
exp
z z
0
l
rc
_ _
(3)
where s
mix
and l
rc
are the average dispersion of the d-layer due to the
Cascade Mixing and Recoil Implantation, respectively; and s
rg
is the
average deviation of surface peaks height, nally z
0
is the depth of d-
layer marker inside the structure.
So, the analytical DRF is dened under the RMR model by a
convolution of two Gaussians with exponentially decreasing
functions; the result of the convolution is given by following
expression [4,5]:
where s is the standard deviation of a Gaussian obtained by
convolution of two Gaussians describing the Cascade Mixing and
Surface Roughness induced by sputtering respectively.
We follow the IUPAC convention for depth resolution: depth
resolution corresponds to the distance over which a 1684% (or
vice versa) change in signal is measured. So we obtained depth
resolution for leading and trailing edges of d-layer depth prole as:
Dz
ld
Dz
2
mx
Dz
2
rg
_
(5)
Dz
tr
Dz
2
rc
Dz
2
mx
Dz
2
rg
_
(6)
where Dz
rc
= 1.668l
rc
, Dz
mix
= 2s
mix
, Dz
rg
= 2s
rg
.
Eqs. (4)(6) are the core of the RMR model let us possible to
perform precisely study the segregation process over epitaxial
MBE growths.
Previous studies demonstrated an exponential-like penetration
of the surface-segregated indiumatoms into upper GaAs caps [6,7].
So, we assumed that segregation process can be introduced into
SIMS DRF by a raising exponential function with the segregation
parameter l
sg
:
Iz
sg
1
l
sg
exp
z z
0
l
sg
_ _
(7)
We added this process to our DRF (RMR model) as a convolution of
Eqs. (4) and (7) to describe the broadening of a segregated d-layer
after a SIMS analysis. The following normalized expression was
obtained (note that a similar equation was developed by Dowsett
[8]):
Quantication of segregation process can be done if we add an
additional term (Dz
sg
) into depth resolution expression for the
leading edge of experimental prole of d-layer (Eq. (5)):
Dz
ld
Dz
2
sg
Dz
2
mx
Dz
2
rg
_
(9)
Then the segregation parameter can be dened from experi-
mental depth prole as:
l
sg
1
1:668
Dz
2
ld
Dz
2
mx
Dz
2
rg
_
(10)
3.2. Indium segregation length
We proceed nowto the quanticationof the indiumsegregation
process occurring during MBE growth. In order to do this the InAs
d-layer structures were analyzed.
Fig. 1 shows indiumdepth distributions for the rst InAs d-layer
in epitaxial GaAs layers grown at 400, 450 and 500 8C. Indium
intensity was normalized on the matrix ion intensity (As). We can
observe a decrease of the indium peak intensity as the substrate
temperature increases, accompanied with a broadening of the
leading edge. Both effects are the result of indium segregation
increased as the growth temperature increases.
It is important to notice here that we excluded initial surface
roughness of the samples (as grown) from our consideration. We
believe that the GaAs surface roughness coincides in the rst
Iz
1
2l
sg
l
rc
exp
z z
0
l
sg
s
2
2l
2
sg
_ _
1 er f
1
2
p
z z
0
s
s
l
sg
_ _ _ _ _ _
exp
z z
0
l
rc
s
2
2l
2
rc
_ _
1 er f
1
2
p
z z
0
s
s
l
rc
_ _ _ _ _ _
_ _
(8)
Iz
1
2l
rc
exp
z z
0
l
rc
s
2
2l
2
rc
_ _
1 er f
1
2
p
z z
0
s
s
l
rc
_ _ _ _ _ _
_ _
(4)
S. Gallardo et al. / Applied Surface Science 255 (2008) 13411344 1342
Author's personal copy
approximation with the roughness of the indium d-layer
formed during the epitaxial growth process, so it does not affect
on the experimental DRF [4]. Moreover we tested surface
roughness of analyzed samples and found it less than 1 nm for
all the samples.
We obtain the DRF parameters by tting the analytical DRF
model to experimental data. So, we use our DRF model (Eq. (8)) to
t the indium peak in each InAs d-layer. In this way we have
obtained the parameters for the ion sputtering induced processes
and the segregation parameter. These ttings were done for each of
the three different growth temperatures. The ttings were realized
using a LevenbergMarquardt algorithm (Origin 6.0, Origin Labs.).
Fig. 2 shows, as an example, the experimental prole for the
sample grown at 400 8C. The gure shows the nearest surface d-
layer for clarity. We observed a good tting by the DRF as
convolution of the ballistic processes added to the segregation
process (Eq. (8)).
The obtained indium segregation parameters are shown in
Fig. 3 as a function of growth temperature. We can see that the
segregation parameter exponentially increases up to 500 8C; this
suggests that indium segregation is a thermally activated
process.
Taking those data, we have drawn in the inset of Fig. 3 the
experimental indium segregation parameter in an Arrhenius
plot.
The solid line in the inset shows the tting of data points via the
expression:
logl
sg
E
a
k
B
T
(11)
where E
a
is the activation energy, k
B
the Boltzmann constant and T
the substrate temperature in Kelvin.
Fitting of experimental points gave us the activation energy of
indium segregation process in GaAs.
We found a value of 270 meV that is in a good agreement with
previously reported values of 0.22 [6] and 0.23 eV [7]. From the
above analysis the minimal segregation parameter was 8 A
.
3.3. Segregation free growth process
In order to decrease the indium segregation we prepared the
same structure using a special technique where growth interrup-
tion before and after indium layer deposition was applied. The
interruptions time was 4 min.
The idea of the interruption is to increase the probability of
arsenic atoms (produced from the decomposition of As
4
molecule)
to bonding with indiumatoms fromthe so-called oating zone [9].
Thus the amount of segregated indium atoms should decrease in
the case of time interruption.
Fig. 2. Depth distribution of indium in InAs d-layer tted by exponential-Gauss-
exponential DRF (Eq. (3)).
Fig. 3. Segregation parameter vs. substrate temperature during growth process. The
inset shows an Arrhenius plot of the segregation parameter.
Fig. 4. Depth distribution of indium in InAs d-layer of a set samples without
segregation.
Fig. 1. Depth distribution of indium in InAs d-layer measured for 3 different
structures.
S. Gallardo et al. / Applied Surface Science 255 (2008) 13411344 1343
Author's personal copy
We performed SIMS depth proling analysis of the samples
grown in this special mode using the same experimental
conditions. Fig. 4 shows the depth distribution of indium in the
InAs d-layer. Comparison between the samples grown at different
temperatures using the interruption technique shows that there
are no differences in the normalized indium intensity (normalized
on the matrix ion intensity (As)) neither broadening of the leading
edge was found.
Obtained proles were tted by our modied DRF (Eq. (8)) in
order to nd segregation parameters. The indium segregation
parameters that we obtained for all samples were equal each to
other and were 1 A