J.
Phys.
D:
Appl.
Phys.
27
(1994)
13+1336.
Printed
in
the
UK
I
RAPID COMMUNICATION
I
Porosity superlattices: a new class
of
Si
I
1
heterostructures
M
G
Bergert,
C
Diekert,
M
Thonissent, L Vescant,
H
Lutht,
H
Miindert,
W
TheiOt,
M
Wernket and
P
Grosset
t
lnstitut
ftr
Schicht-
und
lonentechnik
(ISI).
Forschungszentrum Jirlich
GmbH,
PO
Box
1913, D-52425
JClich, Germany
Received
24
February
1994
I
Physikalisches
Institut,
RWTH
Aachen,
D-52056
Aachen, Germany
Abstract.
Porosity superlattices have
been
investigated
by
transmission electron
microscopy, photoluminescence and reflectance spectroscopy.
The
superlattices
were
formed on p-type doped Si
using
two different
techniques.
Firstly, for
homogeneously doped substrates
we
have periodically varied
the
formation
current density and thereby
the
porosity. Secondly,
the
current density was kept
constant while etching was performed on periodically doped Si layers. For the
first
type
of superlattices the layer thicknesses
were
determined
by
transmission
electron microscopy. The results are
in
good agreement with the values calculated
from the etching
rate
and time. For both types
of
superlattices, reflectance and
photoluminescence spectra show strong modulation
due
to
the
periodicity
of
the
superlattice.
Since the middle
of
the fifties it has been well known
how porous Si layers can be formed
[l],
but interest
in porous Si-its microscopic structure and optical
properties-has dramatically increased due to reports
of photoluminescence
(PL)
[Z]
and electroluminescence
(EL)
[3,4].
Quantum efficiencies of more
than
10%
for
PL
and
0.1%
for
EL
stimulated again the idea
of
Si
based optoelectronics and have been the driving force
for a lot
of
research activities over the last
two
years.
Typical luminescence spectra show a broad band with
a peak maximum between
550
and
850
nm and a full
width
at
half maximum
(FWHM)
of
more
than
100 nm
[2].
For the
EL
it seems nearly impossible
to
tune the
wavelength
of
the
emission peak with the exception
of
using electrolytical front contacts
[5,6].
In this paper
we show for the first time that porosity superlattices can
act
as
filters which will reduce the
FWHM
and shift the
maximum
of
the emitted light in a reproducible way.
Porosity superlattices
are
a new type of Si based
heterostructures which exhibit a periodic variation
in
depth of the porosity. Despite the continuing debate
on
the mechanisms of the luminescences-how must the
basic quantum size model be modified
to
explain
the
luminescences behaviours-porosity superlattices open
a wide field
of
possible applications especially coloured
flatscreens made
of
Si. In addition, porosity superlattices
can lead to a better understanding
of
the reaction kinetics
during the formation process of porous Si layers.
Porosity superlattices can be formed in different
ways. First, etch parameters, such as the current density
0022-3727/94/061333+04$19.50
0
1994
IOP
Publishing
Lid
or the light power, can periodically be changed during
the electrochemical etch process. Superlattices formed
in
this
way will be denoted
as
type I superlattices.
Secondly, using periodically doped substrates while
keeping the etch parameters constant will also result in
the formation of a porosity superlattice (type
U).
In
this
communication we report results obtained
on
both types
of
porosity superlattices.
For the formation
of
type
I
porosity superlattices p-
type boron doped Si(100) substrates with resistivities of
0.01
and
0.2
Qcm were used. The anodization was
performed in the dark using a mixture of
50%
HF
with
ethanol 1:l. Here, the type
I
superlattices were formed
by
varying the
current density periodically during the
etch process.
As
current source a Keithley
238
was used
which allows
a
computer controlled etch process.
For the formation of type
n
porosity superlattices
periodically doped films were grown
on
Si (100)
substrates (p-type,
0.01
Qccm). The doping levels
were
1
x
10’’
and
1
x
Two types
of
periodically doped samples were investigated with single
layer thicknesses of
75
and 150
nm.
The number of
periods
was
10
and
5,
respectively, resulting in an
epitaxial layer thickness of
1.5
pm
for
both types of
sample.
The reflectance
of
the superlauices was measured
under normal incidence using a Perkin-Elmer Lambda
2
spectrometer.
For
the
PL
measurements the
samples were excited with the
457
nm line of
an
Ar+
ion laser
at
a power density
of
100
1333
Rapid communication
Figure
1.
TEM
cross section of a porosity superlattice
(type
I)
formed
on
p-doped substrate
(0.2
Qcm).
Layers
I:
64%
porosity.
20
nm,
Layers
II:
84%
porosity,
200
nm.
mW
cm-*.
To avoid photostimulated oxidation [7]
PL
layer thicknesses calculated from these parameters are
measurements were performed under ultrahigh vacuum 16 and
218
nm, respectively. These values are in good
(UHV)
conditions. The spectra were taken with a
DILOR
agreement with those estimated from
TEM
which are
XY
monochromator using a
GaAs
photomultiplier
20
nm for the low porosity layers and
200
nm for the
tube.
The
PL
spectra were corrected for the spectral high porosity layers.
response of the monochromator. Transmission electron For single layer
porous
films the above current
microscopy
(EM)
pictures were taken taken with
a
Jeol densities correspond to porosities of
64
and
84%.
4000FX
using an electron energy of
400
keV. respectively. Taking the calculated single layer
For
a
type
I
superlattice formed by variation of the
thicknesses into account the mean porosity of the
current density the layer thickness
of
a single layer is
sample has been calculated to
83%.
From gravimetrical
given by the etch rate for a given current density and the
measurements, however, only a mean porosity
of
76%
particular etch time. For the
0.01
S2
cm substrate material is found. This deviation is most likely due to porosity
the porosities
of
a single layer can be vaned from about
gradients at the interface between layers with different
25
to
75%
corresponding
to
current densities from
10
porosities, because for
a
superlattice with
3
times thicker
to
240
mA cm-*. However, on
0.2
S2
cm substrates
only
single layers the measured mean porosity was 79%
.
The
porous Si layers with porosities between
55
and 75% can
reason for such porosity gradients might be recharging
be formed, corresponding to current densities between
10
effects of capacities in the etching cell, e.g. due to the
and
120
mAcm-2. For both materials the upper porosity
Helmholtz layer. The effect of
a
porosity gradient
on
limit is given by the mechanical stability of the layers.
the mean porosity of the sample will of course be larger
Porous
Si layers with porosities higher than
75%
peel
for thinner single layers and can therefore explain the
off
from the substrate due to high strain values. Porosity obtained results.
superlattices are a very promising approach to overcome In addition, the porosity of the highly porous single
this problem. Within the superlattice low porosity layers
layers of the superlattices has been calculated from the
can be used to stabilize high porosity layers which then
mean porosity obtained by gravimetrical measurements,
may have porosities higher than 75%. assuming that the porosity of the low porosity layers
Figure
1
shows
a
TEM
cross section of
a
type
I
are known. Values higher than
82%
are found,
porosity superlattice with the electron beam parallel to
indicating that it is indeed possible to embed otherwise
a (110) direction of the sample.
In
the
?EM
the lower
unstable
porous
layers into mechanically stable porosity
porosity layers appear dark due to the higher densitiy superlattices.
of the material. The superlattice was formed on p-type
The most obvious difference
of
porosity superlattices
doped substrate
(0.2
Qcm). Etching was performed by
as
compared to single porous layers is the bright
a current sequence of
19
mAcm-’ for
1
s
and 175
colourful appearance of the superlattice structures. The
mAcm-z for
2
s
which was repeated
60
times.
The reason therefore is the very strong modulation
of
the
1334
Rapid communication
in
I
,~
""I
0.8
W
U
9
0.4-
0.2
-
I
0
10000
20000
30000
40000
50000
0.0'
WAVENUMBER
(cm-')
Figure
2.
Reflectance spectrum
of
a type
I
porosity
superlattice
20x
(64% 52
nm,
89%
774
nm)
(full
line)
compared
with
bulk
single
crystalline Si (broken
line).
reflectance
in
the visible spectral range. Figure
2
shows
the
reflectance spectrum of a type
I
porosity superlattice
measured under normal incidence. The superlattice has
been formed
on
0.2
Qcm p-type doped substrate by
repeating a sequence
of
19
mAc&
for
3
s
and
207
mAcm-2 for
6
s
20
times. According to
our
results
obtained on single
porous
films these etch parameters
correspond to the formation of a superlattice
with
single
layer thicknesses of 48 and 719 nm and porosities
of
64
and
89%,
respectively. Summing up the layer
thicknesses an overall layer thickness of
15.3
pm is
expected. Cleaving the sample an overall layer thickness
of about
16.5
pm
was found using an optical microscope.
This again shows, that in superlattice structures the usual
relationship between current density and etch rate is not
valid.
In comparison with bulk single crystalline silicon,
whose reflectance is shown as a dashed line in figure
2,
the
reflectance of the superlattice is strongly modulated.
For example, the reflectance of a superlattice can be
higher by nearly a factor of
3
for energies at around
10000
cm-I. This strong modulation of the reflectance
is caused by multiple reflections and interference effects
due to the periodicity
of
the superlattice. The reason for
multiple reflections
are
the
different refractive indices
of
layers with different porosities.
In order to simulate the reflectance spectra the
effective dielectric function of the single layers has
been calculated within the Bruggeman effective medium
theory
[SI.
This is of course only a
first
approach
because in this simple theory topology effects are not
represented adequately, as is the case for the more
sophisticated Bergman theory
191.
However, in the
Bruggeman theory the only free parameter is the volume
fraction of silicon assuming air
as
the matrix material,
which makes it easy
to
calculate the dielectric function
for different porosities. Figure 3 shows a comparison
between a measured reflectance spectrum (figure 3(c))
and the corresponding simulations.
For
the simulation
shown
in
figure
3(a)
the porosities
are
estimated from
the etch parameters but the layer thicknesses have been
rescaled according to the observed increase of the
full
layer thickness.
0
10000
20000
300QO
40000
50000
WAVENUMBER
(cm-')
Figure
3.
Simulation
of
the
reflectance spectrum
of
a
porosity superlattice (type
I)
using
the Bruggeman effective
medium
theory. (a)
20x (64% 52
nm,
89%
774
nm).
(b)
20x (64% 52
nm,
85%
774
nm).
(c) Measurement.
WAVELENGTH
(nm)
Figure
4.
Photoluminescence spectrum
of
a type
I
porosity
superlattice
60x
(34%
25
nm,
78%
264
nm)
formed on
p+-doped substrate
(0.01
Qcm).
The
main p6ak has
a
FWHM
of
only
17
nm.
Concerning the general lineshape, fairly good
agreement between
the
simulation and the measured
spectrum is obtained. However, the peak positions of
the maxima in the reflectance do not fit very well. The
reason for this might be a deviation of the porosity
of the highly porous layers from the value calculated
from
the
etch parameters which might
be
due
to
porosity
gradients. Decreasing
the
porosity of the highly porous
layers in the simulation from 89 to
85%
the agreement
can considerably be improved (figure
3(b)).
But it
must also be kept in mind that the effective dielectric
function calculated within the Bruggeman theory using
the dielectric function of bulk crystalline Si is only a
rough assumption.
Due to the strong modulation of the optical properties
also the
PL
lineshape is changed (figure
4).
The porosity
superlattice acts
as
a filter which can be used to narrow
the broad luminescence spectrum of porous Si. This
is because the emitted light must partially pass through
the superlattice and therefore it also undergoes multiple
reflections and interferences. The
WHM
of the
PL
peak
shown
in
figure
4
is
only
17
nm compared
to
typical
values of more than
100
nm
for single porous layers
[Z].
The modulation of the optical properties can be tuned
in a well controlled way. Here, the reflectance spectra
1
335
Rapid communication
z
0.2
c
4
014i
WAVENUMBER
(c~')
Figure
5.
Reflectance spectra
of
type
II
porosity super-
lattices.
The
single layers have a thickness
of
75
nm
(a)
and 150
nm
(b).
(Formation current density:
47
mAcm-*).
of
type
II
porosity superlattices formed with
the
same
formation current density are shown (figure
5).
These
reflectance spectra must be compared to the reflectance
spectra of the
non-porous
epitaxial layers which
are
identical to the one of bulk single crystalline Si shown
in figure
2
(dashed line).
In figure
5
structures with short modulation period
are due to interferences caused by the total layer
thickness of about
1.5
pm.
The dominating structures
are again due to multiple interferences at the different
single layer interfaces. The different values of the
reflectance at around
14000
cm-' are due to the different
number
of
the
single layers.
For
the sample with
10
periods again a reflectance higher than
95%
is
obtained.
Because the thicknesses of the single layers differ by
a factor of
2
structures occurring at the
same
energetic
position are caused by different orders
of
interferences.
Therefore, the structure at about
14000
cm-' is a
first
order interference in figure 5(a) but a second
order
interference in
figure
5@).
The same explanation holds
also
for
the second and
fourth
order occurring at about
23000
cm-'.
In conclusion for the first time porosity superlattices
have been studied.
In
this paper
results
obtained
on
homogeneously
as
well
as
on periodically p-type doped
substrates are shown. The layer thicknesses
of
type
I
superlattices are in good agreement with the calculated
values.
For
thin
and extremely high porous layers a
deviation of the porosiry
from
the expected values is
observed. The reason for this
is
probably a porosity
gradient at the interface between two layers. The
reflectance spectra
of
superlattices show very sharp
structures. The origin
of
these structures are multiple
reflections and interferences due
to
the superlattice
shucture. Simulations
of
the reflectance spectra show
a rather good agreement with the experiment and can
be used to design further superlattice structures
for
applications. In addition, we have demonstrated the
strong influence of the superlattice on
the
PL
lineshape.
This offers the possiblity
to
narrow and shift the
luminescence of light emitting devices made of porous
silicon,
e.g.,
for
use
in flat coloured displays.
References
[I]
Uhlir
A
1956
Bell
Sysr.
Tech.
J.
35
333
[21
Canham
L
T
1990
APPL
Phvs.
Leu.
57
1046
i31
Halimaoui A,
Oules
C.
Bomchil
0,
Bsiesy
A,
Gaspard
F,
Herino R, Ligeon M and Muller
F
1991
Appl. Phys.
Len.
59
304
141
Richter
A,
Steiner
P,
Kozlowski
F
and
Lang W
1991
IEEE Electron.
Device
Lelr.
12 691
(51
Bressers
P
M
M
C,
Knappen
J
W
I,
Meulenkamp
E
A
and
Kelly
J
J
1992
Appl. Phys.
Lert.
61
108
[61
Canham L
T,
Leong
W
Y,
Beale M
I
I,
Cox
T
I
and
Taylor L
1992
Appl. Phys.
Lett.
61
2563
[71
Mauckner
G,
Walter
T,
Baier
T,
Tho&
K
and Sauer R
1993
Mar.
Res.
Soc.
Symp. Proc.
283 109
[SI
Bmgga"
D
A
G
1935
Ann.
Phys.
24
636
191
Thein W, Grosse
P,
Munder
H,
Liith
H.
Herino R
and
Ligeon
M
1993
Mar.
Res.
Soc.
Symp. Proc.
283 215
1336