Content uploaded by S. Franchi
Author content
All content in this area was uploaded by S. Franchi on Feb 16, 2015
Content may be subject to copyright.
Optical spectroscopy of quasimonolayer InAs at the onset of quantum-dot nucleation
A. S. Bhatti,*M. Grassi Alessi, and M. Capizzi
Istituto Nazionale di Fisica della Materia, Dipartimento di Fisica, Universita
`di Roma ‘‘La Sapienza,’’ Piazzale A. Moro 2,
I-00185 Roma, Italy
P. Frigeri and S. Franchi
CNR–MASPEC Institute, Parco Area delle Scienze, I-43100 Parma, Italy
!Received 27 January 1999"
We have performed low-temperature photoluminescence !PL", resonant PL !RPL", and PL excitation !PLE"
measurements on different series of self-organized InAs/GaAs quantum dots !QD’s"in samples with an InAs
nominal coverage !L"varying from 1.2 to 3 monolayers !ML". Drastic changes in the PL spectra have been
observed for Lvalues spanning across the so-called critical thickness Lc(#1.7 ML". RPL has shown that both
QD’s and quasi-three-dimensional QD precursors contribute to the spectra of samples with L!Lc. PLE
measurements have allowed us to introduce an accurate determination of Lcand to verify its dependence on
sample growth conditions. Finally, the analysis of PL spectra in all investigated samples and its comparison
with spatially resolved PL measurements has suggested a different interpretation of doublet and triplet bands
usually found for L#Lcand previously ascribed to excited states or multimodal distributions of QD families.
$S0163-1829!99"05627-1%
I. INTRODUCTION
The search for zero-dimensional structures either for ap-
plication purposes or for fundamental studies is an intensely
pursued research topic.1In particular, the prediction of im-
proved laser device performances at room temperature has
given strong support to research on quantum dots !QD’s".2
Although growth mechanisms and optical properties have
been widely investigated and greatly improved, a number of
issues are still a matter of debate. These are, for instance, the
dependence on growth conditions of the critical thickness
(Lc) for the self-aggregation and evolution of QD’s, the ef-
fects of In and Ga interdiffusion for different coverages, etc.
All these features strongly affect the optical properties of the
dots and the overall device performance.
Recently, we have shown that QD’s in InAs/GaAs hetero-
structures with the same InAs nominal coverage !L"emit at a
number of quantized, discrete energies.3This has been inter-
preted in terms of different QD shapes and of their depen-
dence on growth conditions. These results hold also for other
lattice mismatched systems, e.g., to InP/GaxIn1"xP
heterostructures.4According to Ref. 3, the evolution of QD’s
with InAs coverage defines three distinct Lregions, I– III, as
shown in Fig. 1 where the emission energy of the strongest
QD band (Ep) is reported as a function of L. Solid symbols
refer to samples further investigated in the present work,
open symbols refer to samples previously prepared in a dif-
ferent laboratory under different growth conditions.5The
second, intermediate region !II", spans across the onset for
the formation of quantum dots and exhibits the most drastic
dependence of the optical properties on L. In this region, the
QD emission energy falls off quickly with L. Moreover, QD
photoluminescence !PL"bands are characterized by doublets
or triplets with roughly constant energy separation (#30
meV", as it will be shown in the following, or by singlets
broader than those of samples in regions I or III. Finally, PL
bands related to the recombination of heavy-hole !HH"and
light-hole !LH"excitons in the two-dimensional !2D"layer
underlying the dots, usually called wetting layer !WL",
dominate the PL excitation !PLE"spectra.
In the present paper, we discuss the most striking aspects
of the above-mentioned intermediate growth region. A
smooth transition takes place from region I, characterized by
structures that can be identified as QD precursors, to region
II, where both precursors and QD’s coexist. The QD precur-
sors are capable of confining carriers and giving rise to PL
bands below the WL energy. Those bands can be distin-
guished from the bands due to fully formed QD’s by per-
forming PL experiments in a crossed excitation/detection ge-
ometry. The same type of multiplets with roughly constant
energy separation (#30 meV"observed in the PL bands of
FIG. 1. Peak emission energies vs Lof the main bands observed
in the PL spectra of all the samples listed in Table I. Dashed lines
are guides to the eye and define three different regions in the plane,
I– III. Data from Ref. 5 are reported as open circles.
PHYSICAL REVIEW B 15 JULY 1999-IIVOLUME 60, NUMBER 4
PRB 60
0163-1829/99/60!4"/2592!7"/$15.00 2592 ©1999 The American Physical Society
samples in region II are present also in the PL bands of
samples with Lquite different from Lc. These features and
PL measurements as a function of exciting power density
allow us to exclude that those multiplets can be explained in
terms of excited states or different QD families. We tenta-
tively explain those multiplets in terms of carrier interaction
with InAs longitudinal-optical phonons. Finally, we study
the 2D to 3D transition and the onset for the aggregation of
fully developed QD’s. Theoretical models generally predict
that the InAs/GaAs system, under thermal equilibrium con-
ditions, follows the Stranski-Krastanow growth mode. In
particular, QD’s are expected to nucleate spontaneously over
an InAs layer only when a critical thickness of InAs has been
deposited. This thickness has been identified with a nominal
coverage of about 1.7 ML by reflection high-energy electron
diffraction !RHEED"and microscopical measurements.
However, recent theoretical calculations6,7 have shown that
the critical thickness value should depend on sample growth
conditions and on the sensitivity of the experimental tech-
nique used for its determination. In order to support these
theoretical results, we have introduced an empirical determi-
nation of Lcby means of PLE experiments whose sensitivity
to the formation of QD’s is higher than that of RHEED. We
show that the critical thickness value depends on growth
temperature. Moreover, the WL growth is subject to fluctua-
tions until it is frozen in a disordered morphology by the
overcoming nucleation of dots. These results indicate that
strong kinetic effects, usually not included in the theoretical
models, partially mask the thermodynamic equilibrium con-
ditions during the QD growth.
II. EXPERIMENT
Low-temperature photoluminescence, resonant PL !RPL",
spatially resolved PL (
&
PL), and PLE spectroscopy were
performed on InAs/GaAs samples grown by molecular-beam
epitaxy under different conditions. Sample specifications
have been summarized in Table I. Optical measurements
were performed at 10 K, unless otherwise specified, in a
closed cycle He optical cryostat by means of a He-Ne or an
ion Ar#laser and of a Ti-sapphire tunable laser. The signal
was spectrally analyzed by a double monochromator and col-
lected by a photomultiplier or an InxGa1"xAs detector. All
the spectra were corrected in order to take into account the
response of the system. The PL and PLE experiments were
carried out at various laser power densities with a laser spot
size variable between 1 and 200
&
m.
III. RESULTS AND DISCUSSION
In Fig. 2, we show the PL spectra of three samples of the
'
series with Lbetween 1.2 and 1.8 monolayer !ML"for
excitation energies (Eex) above $Eex$1.61 eV, nonresonant
excitation, Fig. 2!a"% and below $Eex$1.485 eV, resonant
excitation, Fig. 2!b"% the GaAs band gap. The nonresonant
PL spectrum of the L$1.2-ML sample exhibits a sharp peak
at 1.44 eV, which is due to the recombination of excitons
confined in the 2D InAs layer. By a careful analysis of the
spectrum, two additional bands, broader and one hundred
times weaker than that at 1.44 eV, are identified at about
1.41 and 1.3 eV; see the inset in the figure where the loga-
rithm of the PL intensity is reported. By increasing the InAs
coverage to 1.4 ML, the band related to the 2D layer disap-
pears and the 1.41 eV broadband becomes dominant, while
the 1.3 eV band is still weak. After a further increase of the
InAs coverage to L$1.8 ML, the relative weight of the two
broadbands is reversed.
The PL spectra taken for resonant excitation in the same
three samples reported in Fig. 2!a"are shown in Fig. 2!b".
These spectra exhibit two main bands of comparable weight,
separated by more than 100 meV and roughly coincident
with the two broadbands observed in the case of nonresonant
excitation. The simultaneous presence of both bands for all
coverages in the
'
series, with comparable PL intensities, is
confirmed by RPL measurements. We ascribe these two
bands to the recombination of carriers confined in two dif-
TABLE I. Specifications of the samples investigated in the
present work. The relative uncertainty on sample thickness is %5%.
All samples have a GaAs cap layer of 20 nm.
Samples L!ML"TG(° C)
(
1.8, 3.0 500
)
1.5, 1.7, 1.9, 2.1, 2,6, 3.2 500
'
a1.2, 1.25, 1.35, 1.4, 1.45 460
'
1.65, 1.7, 1.8, 2.0 460
*
1.2, 1.45, 1.7, 1.9, 2.1, 2.5, 3.1 520
aThe samples have been prepared in a single growth on a nonrotat-
ing holder.
FIG. 2. PL spectra of three samples from
'
series. The spectra
are taken at 10 K with excitation energy !a"above, 1.61 eV, and !b"
below, 1.49 eV, the GaAs band gap, 1.519 eV. The excitation den-
sity is 1 Wcm"2.
PRB 60 2593OPTICAL SPECTROSCOPY OF QUASIMONOLAYER InAs...
ferent zero-dimensional structures, i.e., two different types of
QD’s. Under resonant conditions, carriers are expected to be
directly and selectively generated in a reduced number of
QD excited states and, thus, to recombine only in the corre-
sponding dots. On the contrary, for non-resonant excitation,
free carriers are mainly generated in the GaAs barrier. Thus,
the resulting PL line shape depends on the complex mecha-
nism of carrier capture by the QD, eventually leading to
different probability of relaxation into different types of dots.
Further measurements under resonant conditions clarify
the different nature of the two types of QD’s. Two PL spec-
tra detected in the 1.7-ML sample of the
'
series for differ-
ent excitation/detection configurations are shown in Fig. 3.
In the standard backscattering geometry—both excitation
and detection along the direction perpendicular to the InAs
layer—a doublet appears at energies below 1.35 eV, while a
weaker, broad structure is observed at #1.40 eV. When de-
tecting the PL along the InAs plane—the excitation being
maintained along the direction perpendicular to the InAs
layer—a peak at 1.42 eV dominates the PL spectrum. We
ascribe the two bands emitting at lower energies and insen-
sitive to the detection geometry to ordinary, three-
dimensional QD’s shaped like pyramids, an assignment con-
firmed by transmission electron microscopy !TEM"
measurements.8We ascribe, instead, the 1.42 eV band,
which lies at an energy lower than but close to that expected
for a 2D layer, to recombination of carriers confined in struc-
tures with a lens-shaped geometry and low aspect !height to
base"ratio. In fact, PL bands with features similar to those of
the 1.42 eV band have been observed previously in a series
of samples grown in a different laboratory that are character-
ized by the presence of lens-shaped structures, as shown by
high resolution TEM measurements.5,9 In these structures,
the dependence of the PL peak energy on Lpractically coin-
cides, as shown by the open symbols in Fig. 1, with that of
the higher-energy band at #1.42 eV reported here for the
samples of the
'
series. We assume that such flat structures
with almost 2D geometry are morphologically linked to the
disordered 2D InAs layer. In that case, a sizable wave-
guiding effect can lead to a strong increase of PL intensity
for detection along the plane of the 2D InAs layer, as ob-
served here.
Some sort of InAs ‘‘islands,’’ precursors of QD’s have
been theoretically predicted10 and experimentally
determined11 on the grounds of morphological measure-
ments. These structures were referred to as platelets and
quasi-three-dimensional !Q3D"structures, respectively. The
present measurements clearly indicate that platelets and Q3D
structures are the same nanostructures, of the same kind of
those detected here and in Refs. 5 and 9. Q3D structures !or
platelets", therefore, can confine carriers and give rise to
bound states at energies below that of the so-called wetting
layer. These structures aggregate below the critical thickness
and coexist with already well-formed pyramidal QD’s, at
least for L!2 ML.
Let us introduce now a more sensitive technique for the
estimate of Lcand discuss then the critical thickness concept.
PLE spectra taken in samples of the series
)
and
*
are re-
ported in Fig. 4. The spectra were taken at 10 K, with a
detection energy equal to that of the corresponding strongest
QD peak in the PL spectrum. The two main bands observed
in Figs. 4!a"and 4!b"are due to recombination of photoex-
cited heavy-hole and light-hole excitons in the 2D InAs
layer—HHE and LHE, respectively. For increasing InAs
nominal coverage, the energy peak of both bands shifts to
lower energies due to the increase of the average thickness of
the 2D InAs layer. In the meanwhile, the linewidth increases
for both the HHE and LHE bands, whose relative weights
become comparable for L&1.7 ML. The latter features can
be attributed to an increase in the exciton localization due to
the morphological disorder induced in the 2D layer by the
onset of QD aggregation. In the 1.5 ML sample of the
)
series in Fig. 4!a", two bands can be observed at #1.37 eV
and #1.40 eV, namely, at 64 and 96 meV from the detection
energy, as already reported in literature.12 The HHE bands in
the samples of the
*
series in Fig. 4!b"show weak, low-
energy shoulders that gain strength for L&2.0 ML. The de-
pendence of HHE peak energy on the InAs coverage Lis
summarized in Fig. 5 for the two series of samples. The main
peak energies are reported as solid dots and rhombuses, the
FIG. 4. PLE spectra for !a"the
)
and !b"the
*
samples. The
heavy-hole and light-hole excitons in the wetting layer !HHE and
LHE, respectively"get broader and shift to lower energies for in-
creasing L. Multiple structures are observed in the PLE of
*
samples for L+2 ML.
FIG. 3. PL spectra of the 1.7-ML
'
sample detected along a
direction parallel !solid line"or perpendicular !dashed line"to the
InAs 2D layer. In both cases, the excitation is perpendicular to the
2D InAs layer.
2594 PRB 60
A. S. BHATTI et al.
weak shoulder energies as open circles. In the same figure,
the HHE energy evaluated in a simple effective-mass ap-
proximation for InAs ultrathin quantum well13 is also re-
ported as a solid line, while previous data from Ref. 5 are
reported for reference as solid triangles. The HHE energy
falls off sharply and coincides within the experimental un-
certainty on L!see Table I"with the data from Ref. 5, as well
as with the theoretically expected trend, at least for low-L
values. For high-Lvalues, the HHE energy begins to saturate
and asymptotically converges to a constant value. This be-
havior defines an empirical value of the critical thickness for
QD self-aggregation. In fact, the crossing between the theo-
retical curve !solid line"and the asymptotic constant value
!dashed lines"defines the value Lcof Lover which QD
nucleation starts and the thickness of the 2D layer suddenly
stops to increase. With this definition, Lcis equal to #1.4
ML for the samples of the
*
series, grown at 520 °C, and is
equal to #1.7 ML for the samples of the
)
series, grown at
500 °C. If thermal equilibrium conditions are not established
during the sample growth, as should be expected in the case
of molecular-beam epitaxy, the transition from the 2D to the
3D growth is smooth, as observed in the case of the samples
of the series
)
, grown at 500 °C. Thermal fluctuations can
also broaden the transition and, therefore, the growth of the
2D InAs layer only asymptotically reaches an equilibrium
value. In the case of samples grown at 520 °C !series
*
), the
transition from 2D to 3D seems, instead, to be sharper. These
results well agree with the conclusions recently reached on
the grounds of reflectance anisotropy spectroscopy !RAS"
measurements,14 a technique very sensitive to sample growth
and surface conditions. In fact, RAS measurements in InAs/
GaAs heterostructures indicate that a small fraction of the
incident molecular-beam fluxes are still incorporated into the
2D InAs layer for L'Lc. This effect is more pronounced,
the lower the growth temperature is, in agreement with the
results presented here. It may be worth noting that the coun-
terpart of this effect is observed in Fig. 1, where the QD
emission energy exhibits a sharper transition from region II
to region III the higher the sample growth temperature is. It
should be emphasized also that optical—in particular PL-
based techniques—are by far more sensitive than the
RHEED technique generally used to determine Lc. In fact,
the former technique detects the weak emission of the carri-
ers localized in the first formed low-density QD’s, while the
latter technique determines Lcon the grounds of a much
higher surface density of QD’s, namely, that needed to pro-
duce a sizable change in the electron diffraction pattern.
Therefore, 3D structures can be detected by PL even when
their density is expected to be quite small. This is shown in
Fig. 2 where a slight contribution to PL from QD precursors
has been measured in the case of a 1.2-ML sample.
In conclusion, the onset of QD formation definitely inhib-
its the growth of the 2D layer, for different values of Lcfor
the two sets of samples grown at different temperatures. This
provides experimental evidence that the critical thickness de-
pends on the growth temperature, in agreement with the pre-
dictions of a recent theoretical model.6That same model,
which takes into account mass transfer and In and Ga inter-
diffusion during the growth process, has also predicted an
exponential increase of the QD density and a dependence of
Lcon the sensitivity of the experimental technique used for
its determination, as suggested here. The fractional values
estimated for Lcshow that the 2D layer evolution is subject
to strong fluctuations. These give rise to a disordered final
morphology with inhomogeneous WL thicknesses, a likely
source for the weak shoulders of the HHE band in the PLE
spectra of the
*
series.15 This evidence together with the
broadening of the PLE peaks for high-Lvalues shows that
the Stranski-Krastanow picture applies to a limiting ideal
case, while a major role during the growth process is played
by kinetic effects, usually not taken into account by theoret-
ical models based on thermodynamic equilibrium conditions
!see, e.g., Refs. 10 and 16, and references therein". Some-
what similar conclusions have been recently achieved on the
grounds of scanning tunneling microscopy and PL measure-
ments in the same type of heterostructures studied here,
which have provided evidence that 2D clusters, Q3D clus-
ters, and 3D islands coexist in the intermediate region II and
have shown a peculiar reentrant nature in the formation of
Q3D clusters.17,18
A wider set of nonresonantly excited PL spectra taken in
samples of the
'
series is shown in Fig. 6. The QD bands
with emission energies lower than 1.4 eV exhibit doublet or
triplet structures, as summarized in Fig. 7 where the data for
the full set of all samples of the
'
series are reported. An
analogous behavior is generally observed in the other two
series of samples,
)
and
*
, with emission bands at energies
between 1.2 and 1.4 eV. Below 1.2 eV the QD bands re-
sume, instead, a simple, symmetrical Gaussian line shape.
Similar results have often been reported in literature, where
multiplets are usually ascribed either to the emission from
excited states or to the simultaneous presence of many QD
families of different geometry and mean size.19– 23
In the present case, none of the multiplet components can
be ascribed to emission from excited states. In fact, the in-
tensity of all the multiplet components increases linearly for
an increase in the excitation power density (P) over three
orders of magnitude, starting from P$0.001 W cm"2. More-
over, the highest-energy component of the multiplets is often
FIG. 5. Peak energy of the HHE in the wetting layer as a func-
tion of Lfor the samples of the
)
!solid rhombuses"and
*
!solid
dots"series. The energies of the additional shoulders observed in
the PLE of
*
samples are reported as open circles. Solid triangles
are taken from Ref. 13. The solid line gives the HHE energy vs Las
calculated for strained InAs in GaAs in a simple square-well model.
The two horizontal dashed lines, which coincide with the
asymptotic values of the HHE peak energy at high-Lvalues, define
the critical thickness values by their crossing with the solid line.
PRB 60 2595OPTICAL SPECTROSCOPY OF QUASIMONOLAYER InAs...
stronger than those at lower energy, independently of the
excitation power density.
The energy difference ,Ebetween each two adjacent
components of the multiplets is roughly equal to 32 meV,
within the experimental uncertainty, independently on InAs
coverage or growth conditions, namely, on QD shape and
size; see Ref. 3. This is shown in Fig. 8, where ,Ehas been
reported versus Lfor all the investigated samples of Table I
where structured QD bands have been observed for nonreso-
nant excitation, i.e., in samples of the series
)
,
'
, and
*
. It is
highly unlikely, therefore, that those PL multiplets can be
ascribed to multiple QD families with different shape or
mean size.
On the other hand, a mean value of 32 meV for ,Ecor-
responds to the estimated and measured energy of the LO
phonon in the bulk or at the interface of strained InAs, either
QD or WL.12,24– 26 Therefore, we suggest that extended pho-
non !or local"modes and their interaction with carriers play a
key role in the interpretation of the above multiplets. An
inhomogeneous broadening due to strong fluctuations in QD
sizes can sometimes hide the effects of that carrier phonon
interaction. This leads to unusually broad QD PL bands like
those observed in an energy range from 1.2 to 1.3 eV, i.e.,
where multiplets are more commonly observed. This last ef-
fect has been directly verified by spatially resolved PL mea-
surements on a sample of the
*
series with L$3 ML. This
sample exhibits fully grown pyramidal QD’s and a broad
Gaussian PL band under normal excitation conditions.9
&
PL
measurements have been performed by using a microscope
objective, which both focused the laser spot down to a diam-
eter of 1
&
m and collected the PL signal. This has allowed us
to !i"increase the exciting power density by several orders of
magnitude without heating the sample in order to make evi-
dent possible contributions of excited states to the PL spectra
and !ii"reduce the number of excited QD’s and the corre-
sponding inhomogeneous broadening. A
&
PL spectrum
taken at P$10 Wcm"2in the 3-ML
(
sample is shown in
Fig. 9 by the dotted line, together with a conventional PL
spectrum taken at P$0.05 Wcm"2in the same sample
!dashed line". In both cases, a He-Ne laser has been used for
the excitation, and the sample temperature was equal to 77
K. The conventional PL spectrum exhibits a single sym-
metrical Gaussian band, peaked at E0$1.267 eV. The
&
PL
spectrum strongly differs, instead, from the conventional PL
spectrum. First, the increased power density introduces an
additional band at higher energy (E1$1.357 eV, 90 meV
above E0), which can soundly be attributed to a QD excited
state. Second, a structured line shape can now be resolved for
both the ground and the excited states, most likely because of
the reduced number of excited QD’s. For both states the
characteristic energy spacing of the multiplets is the same,
equal to 32 meV. Therefore, two multiplets, separated by 90
meV, have been fitted to the data !solid line in the figure".
Each multiplet was made by three Gaussian bands 30 meV
broad and spaced by 32 meV. The relative intensities of the
Gaussian contributions were the only free parameters. The
agreement between the model !solid line"and the data !dot-
ted line"is quite good.
The resolution of a triplet structure in the PL of the
ground state under proper condition of excitation is a feature
peculiar, most likely intrinsic, of QD’s emitting below 1.35
eV. In particular, this feature seems to characterize the PL
spectra of the samples in the intermediate region of growth,
across the critical thickness; see Fig. 6. Multiplet structures
FIG. 6. PL spectra of different samples from the
'
series, with L
between 1.25 and 2.0 ML.
FIG. 7. Peak energies of all QD and QD precursor PL bands
observed in samples of the
'
series, plotted as a function of InAs
coverage for nonresonant excitation conditions (Eexc$1.61 eV".
The relative intensity of the emission bands is proportional to the
symbol size.
FIG. 8. The energy of each component of the multiplets ob-
served in the PL spectra of samples reported in Table I is measured
with respect to the main PL peak energy and reported vs L. The
dashed lines define constant values corresponding to integer mul-
tiples of 32 meV from the main peak position.
2596 PRB 60
A. S. BHATTI et al.
seem to be absent, instead, in the PL spectra of QD precur-
sors, emitting above 1.4 eV. This has allowed us to get a
rough estimate for the critical thickness also in the case of
the samples of the
'
series (Lc!1.4 ML; see Fig. 7", whose
PLE spectra do not provide clear evidence of the WL band.
Multiplet structures seem to be absent also in the PL spectra
of QD’s emitting below 1.2 ML, namely, large QD’s with
high aspect ratio.3In the framework of an electron-phonon
related mechanism, these features can be explained in terms
of a dot size that maximizes the carrier charge per atom and
the electron-phonon interaction strength. In fact, in QD pre-
cursors, the carrier wave function is likely spread outside the
QD; in large QD’s characterized by high aspect ratio values,
the carrier wave function is delocalized across the whole QD,
which begins to recover 3D bulk-like features. Finally, PL
measurements under strictly resonant conditions in the same
set of samples show that both extended modes !phonons"and
modes localized in a lattice perturbed by a strong electron-
phonon interaction contribute to RPL spectra.27
IV. CONCLUSIONS
We have shown that emission bands due to QD precursors
!or Q3D’s"can be observed in the PL spectra of InAs/GaAs
heterostructures both before and after the onset of QD aggre-
gation. Under suitable conditions of detection, the lumines-
cence of Q3D’s can also be discriminated from that of fully
evolved QD’s. Multiplet structures have been shown to char-
acterize the PL spectra of QD in the region close to the
self-aggregation critical thickness, or, more generally, of QD
emitting at about 1.3 eV, as shown by microphotolumines-
cence measurements. These multiplets always exhibit struc-
tures separated by #32 meV, the energy expected in such
structures for InAs-like LO phonon modes. We ascribe the
presence of these structures to effects of interaction between
phonons and carriers, at least during carrier relaxation pro-
cesses.
Finally, we have shown that PLE measurements on dif-
ferent sets of samples provide an alternative method to de-
termine the thickness for the onset of QD formation and give
an insight into the effects of kinetics during the growth. In
particular, the critical thickness depends on growth tempera-
ture, and the QD nucleation freezes the WL growth. The
resulting framework emphasizes the need to refine theoreti-
cal models, including kinetic effects, in order to correctly
reproduce the complex growth mechanism of QD’s.
ACKNOWLEDGMENTS
One of the authors !A.S.B."is grateful to ICTP-TRIL for
financial support and to the University of Punjab for leave.
This work was partially supported by CNR-MADESS and by
MURST.
*On leave from Dept. of Physics, University of the Punjab, Lahore
54590, Pakistan.
1P. M. Petroff and S. I. Denbaars, Superlattices Microstruct. 15, 15
!1994".
2Y. Arakawa and H. Sakaki, Appl. Phys. Lett. 40, 939 !1982".
3M. Grassi Alessi, M. Capizzi, A. S. Bhatti, A. Frova, F. Martelli,
P. Frigeri, A. Bosacchi, and S. Franchi, Phys. Rev. B 59, 7620
!1999".
4L. Samuelson, S. Anand, N. Carlsson, P. Castrillo, K. Georgsson,
D. Hessman, M.-E. Pistol, C. Pryor, W. Seifert, L. R. Wallen-
berg, A. Carlsson, J.-O. Bovin, S. Nomura, Y. Aoyagi, T.
Sugano, K. Uchida, and N. Miura, in Proceedings of the 23rd
International Conference on the Physics of Semiconductors,
Berlin, 1996, edited by M. Scheffler and R. Zimmermann
!World Scientific Singapore, 1997", p. 1269.
5A. Polimeni, A. Patane
´, M. Capizzi, F. Martelli, L. Nasi, and G.
Salviati, Phys. Rev. B 53, R4213 !1996".
6B. A. Joyce, J. L. Sudijono, J. G. Belk, H. Yamaguchi, X. M.
Zhang, H. T. Dobbs, A. Zangwill, D. D. Vvedensky, and T. S.
Jones, Jpn. J. Appl. Phys., Part 1 36, 4111 !1997".
7B. W. Wessels, J. Vac. Sci. Technol. B 15, 1056 !1997".
8G. Salviati !unpublished".
9A. Patane
´, M. Grassi Alessi, F. Intonti, A. Polimeni, M. Capizzi,
F. Martelli, L. Nasi, L. Lazzarini, G. Salviati, A. Bosacchi, and
S. Franchi, J. Appl. Phys. 83, 5529 !1998".
10 C. Priester and M. Lannoo, Phys. Rev. Lett. 75, 93 !1995".
11 T. R. Ramachandran, R. Heitz, P. Chen, and A. Madhukar, Appl.
Phys. Lett. 70, 640 !1997".
12 R. Heitz, M. Grundmann, N. N. Ledentsov, L. Eckey, M. Veit, D.
Bimberg, V. M. Ustinov, A. Yu. Egorov, A. E. Zhukov, P. S.
Kop’ev, and Zh. I. Alferov, Appl. Phys. Lett. 68, 361 !1996".
13 A. Patane
´, M. Grassi Alessi, F. Intonti, A. Polimeni, M. Capizzi,
F. Martelli, M. Geddo, A. Bosacchi, and S. Franchi, Phys. Status
Solidi A 164, 493 !1997".
14 D. I. Westwood, Z. Sobiesierski, E. Steimetz, T. Zettler, and W.
Richter, Appl. Surf. Sci. 123/124, 347 !1998".
15 Evidence that the WL thickness may not be constant throughout
the sample has been provided also by PL measurements in
InxGa1"xAs/GaAs heterostructures grown by low-pressure
metal-organic chemical vapor deposition; see V. Tu
¨rck, F. Hei-
FIG. 9. Emission spectra at T$77 K of the L$3-ML
(
sample.
Under conventional low excitation (P$0.05 Wcm"2) PL spec-
troscopy a single although asymmetric band is observed !dashed
line". Under 1
&
m spatially resolved, high excitation (P
$10 Wcm"2) PL, that band is resolved into a triplet !dotted line".
A second triplet, due to an excited-state contribution, appears in the
&
PL spectrum 90 meV above the ground state. The solid line gives
the best fit to the data as obtained in terms of two triplets separated
by 90 meV and each made of three equally spaced !32 meV"Gauss-
ian contributions.
PRB 60 2597OPTICAL SPECTROSCOPY OF QUASIMONOLAYER InAs...
nrichsdorff, M. Veit, R. Heitz, M. Grundmann, A. Krost, and D.
Bimberg, Appl. Surf. Sci. 123/124, 352 !1998".
16 Istavan Daruka and Albert-Laszlo Barabasi, Phys. Rev. Lett. 79,
3708 !1997".
17 R. Heitz, T. R. Ramachandran, A. Kalburge, Q. Xie, I.
Mukhametzhanov, P. Chen, and A. Madhukar, Phys. Rev. Lett.
78, 4071 !1997".
18 A. Madhukar, T. R. Ramachandran, A. Konkar, I. Mukhametzha-
nov, W. Yu, and P. Chen, Appl. Surf. Sci. 123/124, 266 !1998".
19 S. Fafard, D. Leonard, J. L. Merz, and P. M. Petroff, Appl. Phys.
Lett. 65, 1388 !1994".
20 S. Fafard, R. Leon, D. Leonard, J. L. Merz, and P. M. Petroff,
Phys. Rev. B 52, 5752 !1995".
21 K. H. Schmidt, G. Medeiros-Ribeiro, M. Oestrich, P. M. Petroff,
and G. H. Do
¨hler, Phys. Rev. B 54, 11 346 !1996".
22 M. Colocci, F. Bogani, L. Carraresi, R. Mattolini, A. Bosacchi, S.
Franchi, P. Frigeri, M. Rosa-Clot, and S. Taddei,
Appl. Phys. Lett. 70, 3140 !1997".
23 P. M. Petroff, K. H. Schmidt, G. Medeiros Ribeiro, A. Lorke, and
J. Kotthaus, Jpn. J. Appl. Phys., Part 1 36, 4068 !1997".
24 In strained nanostructures, the InAs LO phonon energy of 29.9
meV is modified by both strain and phonon confinement. A QD
LO phonon energy of 32.1 meV has been predicted from the
calculated strain distribution for pyramidal QD’s !Ref. 25",
while phonon modes with energies of 29.6 and about 32 meV
have been observed in RPL and PLE measurements on InAs QD
and ascribed to phonons in the WL and in the QD, respectively
!Ref. 12". Resonant PL and Raman study of InAs QD have
shown, as well, the existence of 30-meV InAs LO phonons !Ref.
26". Finally, an energy of 35 meV has been estimated for inter-
face phonons on the grounds of PL measurements !Ref. 12".
25 M. Grundmann, O. Stier, and D. Bimberg, Phys. Rev. B 52,
11 969 !1995".
26 J.-Y. Marzin, J.-M. Gerard, O. Cabrol, B. Jusserand, and B. Ser-
mage, Nuovo Cimento D 17, 1285 !1995".
27 M. Grassi Alessi and M. Capizzi !unpublished".
2598 PRB 60
A. S. BHATTI et al.