Full text of DOI:10.1007/BF02665867

7
Jo
4
ur
8
nal of ELECTRONIC MATERIALS, Vol. 30, No. 6, 2001
Möck, Topuria, Browning, Titova, Dobrowolska, Lee, aSp
n
ec
d
ia
F
l I
u
ss
r
ue
d
P
y
a
n
pe
a
r
Self-Ordered CdSe Quantum Dots in
ZnSe and (Zn,Mn)Se Matrices Assessed
by Transmission Electron Microscopy and
Photoluminescence Spectroscopy
1
1
1
2
P. MÖCK, T. TOPURIA, N.D. BROWNING, L. TITOVA,
2
2
2
M. DOBROWOLSKA, S. LEE, and J.K. FURDYNA
1
.—University of Illinois at Chicago, Department of Physics (MC 273), 845 W. Taylor Street,
Chicago, IL, 60607-7059. 2.—University of Notre Dame, Department of Physics,
Notre Dame, IN, 46556
Single and multilayer sheets of self-assembled CdSe [quantum dots (QDs)] were
grown by means of molecular beam epitaxy in both ZnSe and (Zn Mn )Se
0.9
0.1
matrices. Both types of structures were assessed by means of transmission
electron microscopy in the scanning, high-resolution, and diffraction-contrast
modes. Complementary results from wider sample areas were obtained by
means of photoluminescence spectroscopy. In one of the samples analyzed, a
fractional monolayer of MnSe was deposited immediately before the CdSe
deposition. Asecondstructuregrownunderidenticalconditions, butwithoutthe
MnSe fractional monolayer, was also analyzed. This comparison provides direct
evidence for an enhanced size and shape homogeneity of 3D QDs caused by the
presence of a tiny amount of MnSe at the interface. In the multilayer structure,
we observed the co-existence of highly strained quasi-2D QDs and CdSe rich
aggregates with compositional modulations on certain crystallographic planes
in close proximity.
Key words: Self-assembled, self-ordered, CdSe, quantum dots, Z-contrast
STEM, TEM, PL
propertiesofself-assembledquantumdotsarethought
to be achievable by means of self-ordering processes3
during heteroepitaxial growth in the Stranski-
Krastanov mode (i.e., where according to the ideal-
ized classical model, three-dimensional (3D) islands
are formed on top of a two-dimensional (2D) wetting
layer(WL)afteracertaincriticalthicknessisreached).
A topic of current debate2 is whether the observed
QD configurations might be at thermodynamic equi-
INTRODUCTION
–5
Semiconductor quantum dot (QD) structures are
expectedtoleadto“paradigmchangesinsemiconduc-
1
tor physics” and have, therefore, been attracting
considerable interest in recent years. For QDs to
become useful in opto-electronic devices, they should
simultaneously fulfill the following conditions: di-
mensionsoftheorderofmagnitudeoftheBohrradius
of the exciton (leading to quantized energy levels
inside a QDs), smaller bandgap than the surrounding
matrix, highest possible size and composition homo-
geneity, highest possible number density, and no
–5
6
libriumormaybepreventedby(possiblysynergistic )
kinetics from reaching the equilibrium state.7 Fur-
thermore, at this point in time, the nature of the
ongoing self-ordering processes has not been pre-
–9
2
10
defects such as misfit dislocations. These desired
cisely clarified.
Self-assembled CdSe QDs in ZnSe based matrices
on GaAs are considered to be a promising material for
(
Received November 21, 2000; accepted March 1, 2001)
48
7

Self-Ordered CdSe Quantum Dots in ZnSe and (Zn,Mn)Se Matrices Assessed
by Transmission Electron Microscopy and Photoluminescence Spectroscopy
749
novel opto-electronic devices that may work in the
green to blue range of the electromagnetic spectrum.
EXPERIMENTAL
Since 1996, several groups have been growing such
MBE Growth
QD structures by means of molecular beam epitaxy
(
MBE), migration enhanced epitaxy, metal-organic
The three samples analyzed here were all grown by
means of MBE in a Riber 32 R&D machine equipped
with elemental sources. In the following, the samples
are denoted as A, B, and C. (001) GaAs substrates
were turned into pseudo-substrates by growing ap-
proximately 2 mm (samples A and B) and approxi-
mately 1 mm (sample C) thick ZnSe buffer layers at
300∞C. During the growth of the ZnSe buffer layers,
thereflectionhighenergyelectrondiffractionpattern
showedastreaky2¥ 1reconstructionandoscillations
of the intensity of the specular reflection.
Samples A and B form a pair and were grown under
nominally the same growth conditions, except for the
initial deposition of a fractional monolayer of MnSe
immediately before the deposition of the 2.6 MLs
(0.79 nm) of CdSe in sample B. Only the Mn shutter
wasopenedduringthisprocess,whichwewilldubMn
“sprinkling”, but since MBE proceeds in an excess of
the column VI element, effectively, a fractional mono-
layer of MnSe must have been deposited. The Mn flux
was chosen to be equal to that needed for the deposi-
tion of 0.1 ML of MnSe under the given growth
conditions. Since there was no Se flux during the
Mn deposition, we can assume that there is a nominal
MnSe layer thickness of less than 0.1 ML. This frac-
tional monolayer of MnSe must have an in-plane
lattice constant that is intermediate between ZnSe
and CdSe (since bulk MnSe has a lattice constant of
0.592 nm) and may be considered as constituting
2D islands of reduced strain with respect to the
surrounding ZnSe surface. In an idealized scenario,
molecular beam epitaxy (MOMBE), and metal-or-
ganic vapor phase epitaxy (MOVPE), a review of
3
which is given in the textbook by Bimberg et al. and
the proceedings of the 9th International Conference
on II-VI Compounds.11 (Zn,Mn)Se rather than pure
ZnSematriceshavebeenfoundtobebeneficialforthe
1
2,13
growth of the QDs,
leading to dilute magnetic
1
4
semiconductors whose luminescence properties can
be controlled by a magnetic field. Recently, fractional
monolayers(ML)ofBeSehavebeenusedtoinducethe
growth of 3D CdSe islands/QDs at a thickness region
of the WL that usually leads only to the formation of
a 2D quantum well.15
In this paper we will present results of transmis-
sion electron microscopy (TEM) studies in both the
scanning and parallel beam mode and photolumines-
cence (PL) assessments of single-layer CdSe QDs in a
ZnSe matrix. The main analysis will focus on the
comparison of two samples, one with and one without
a fractional monolayer of MnSe deposited immedi-
ately before the CdSe deposition commenced. In addi-
tion to this analysis of the influence of a tiny amount
of MnSe on the size and shape distribution of the 3D
QDs, we will present preliminary results from a
multilayer structure of Cd0.03Mn0.097Zn0.873Se. In par-
ticular, Z-contrast scanning transmission electron
microscopy (STEM), high-resolution TEM (HRTEM),
and transmission high energy electron diffraction
(
THEED) indicate the existence of compositional
modulations to the structure which are as yet unex-
plained.
a
b
Fig. 1. Histograms of the convolution of the size and shape distribution of 3D QDs: (a) 78 QDs of sample A, and (b) 90 QDs of sample B.

7
50
Möck, Topuria, Browning, Titova, Dobrowolska, Lee, and Furdyna
a
b
c
Fig. 2. (a) CTEM and (b) bright field STEM, and (c) low-resolution Z-contrast STEM image of sample A in plan view. Images (b) and (c) have been
takensimultaneouslyfromthesamearea. ThecontrastreversalinconnectionwithanearlycompleteretentionofthesizeofindividualQDsindicates
that under the employed imaging conditions strain field effects are negligible in the bright field STEM image.
ences17 from the interpretation of our data. Low-
resolution Z-contrast images of these samples can,
therefore, be considered in a first approximation as a
distribution map of the local Cd concentration. Em-
ploying Z-contrast imaging one can derive reliable
dataonthesizeandshapedistributionoftypicalQDs.
Since uncovered CdSe islands are subject to a fast
Ostwald ripening process,19 such data could not be
obtained before by the commonly applied atomic force
microscopy techniques. Furthermore, bright features
inatomicresolutionZ-contrastimagescanbedirectly
interpreted as atomic columns. The relative bright-
ness of individual atomic columns is, to a first ap-
proximation,directlyproportionaltothesquareofthe
average atomic number of the constituting atoms. By
comparingintensityprofilesandatomiccolumnspac-
ing in parallel directions within CdSe rich aggregates
and the surrounding matrix, it was possible to quan-
tify elemental compositions and distributions on the
atomic level with an accuracy of some 20%.18
less than 10% of the ZnSe surface is covered by these
D MnSe islands. The CdSe layers of samples A and
B were grown at 350∞C and a rate of 0.071 ML s to a
nominalthickness(2.6ML),whichiswithintherange
of the commonly accepted critical thickness for the
2
–1
2
D to 3D island transition of the wetting layer. After
the deposition of the CdSe layers, the growth was
interrupted for 2 sec. The final growth stage involved
capping the CdSe by the deposition of approximately
5
0 nm of ZnSe at 350∞C.
For sample C, nominally 82.5 nm of ZnSe were first
depositedhomoepitaxiallyontotheZnSe/GaAspseudo-
substrate at 300∞C. A multilayer structure of 8 se-
quences of 10 ML (2.83 nm) of Zn Mn Se cladding
0.9
0.1
layer and 0.3 ML (0.09 nm) CdSe sheet was then
deposited at 350∞C. While the Zn Mn Se cladding
0
.9
0.1
–1
layers were grown at a deposition rate of 1 ML s ,
the CdSe sheets were grown at a deposition rate of
.038 ML s . As the final (9th) cladding layer of the
multilayer structure, 10 ML of Zn Mn Se were
–
1
0
0.9
0.1
deposited, and finally a ZnSe capping layer of 50 nm
nominal thickness was grown at 350∞C. Within the
multilayerstructure,thereare,thus,nominally87.3%
of the cation sites occupied by Zn, 9.7% by Mn, and 3%
by Cd (as the Zn to Mn ratio is 9, we can consider the
matrix surrounding the CdSe layers of sample C as
Zn Mn Se).
In these experiments with a 200 kV JEOL JEM
2010F Schottky field emission STEM, we typically
used a sub 0.2 nm electron probe. Energy dispersive
x-ray spectroscopy (EDXS) was performed on this
microscope employing a Noran Voyager system. Con-
ventionaldiffractioncontrastTEM(CTEM),THEED,
and HRTEM were performed on a JEOL JEM 3010 at
0.9
0.1
3
00 kV, giving a point resolution of about 0.17 nm at
Transmission Electron Microscopy Studies
and PL Assessment
Scherzer focus in the latter case. Although HRTEM
and CTEM are sensitive to strain fields around QDs,
this can, to some extent, be reduced under kinemati-
cal diffraction conditions close to a low indexed zone
axis. THEED can easily be performed in the selected
areamodetoderivecrystallographicinformationsuch
as the self-ordering of atoms onto certain sites within
the unit cell or the co-existence of different crystal
phases. For all experiments, plan-view and [110]
cross-section specimens were prepared by the stan-
dard TEM techniques of mechanical polishing fol-
lowed by ion-milling.
The primary method of analysis of these samples
was the Z-contrast imaging technique in the STEM.
The Z-contrast technique produces an essentially
16
incoherent image which shows a high sensitivity to
2
2
2
2
thesquareoftheatomicnumbers(Z =48 ,Z =30 ,
Cd
Zn
2
2
ZMn = 25 ) of the constituent atoms of the imaged
material and is fairly insensitive to short- and long-
range strain fields.1 By comparing Z-contrast im-
ages that had been recorded at different inner angles
of the annular dark field detector (also called Howie
detector), we could safely exclude strain field influ-
7,18
The PL spectra were recorded using a 1m long

Self-Ordered CdSe Quantum Dots in ZnSe and (Zn,Mn)Se Matrices Assessed
by Transmission Electron Microscopy and Photoluminescence Spectroscopy
751
grating monochromator. The excitation source was a
multi-lineArlaser. Thesampleswereplacedonacold
finger in a closed cycle He refrigerator.
geneous distribution of the shape and size of the QDs
was observed. More or less circular to oval shapes
dominated (Fig. 3a and b). The smaller QDs of up to
approximately 10 nm diameter were usually more
circular. The larger QDs were usually more oval with
an average ratio of the longer to the shorter axis of on
average about 1.2 to 1.4. These observations can be
summarized as indications of QDs shape/size order,
i.e., the 2nd/3rd levels of the phenomenological hier-
RESULTS AND DISCUSSIONS
Controlling the Shape and Size Distribution of
3
D QDs by Mn “Sprinkling”
Pronounced inhomogeneities in the shape and size
3
of the QDs were observed for sample A (no Mn “sprin-
kling”), as shown by the histogram in Fig. 1a. This
histogram contains a convolution of the size and
shape of the QDs that was obtained by plotting the
largest and smallest lateral dimension as seen in low-
resolutionZ-contrastSTEMimages.Statisticalanaly-
sis of this data gives an average QD size (expectation
archyofself-orderingprocesses. Thisiswellreflected
in Fig. 1b, where the histogram is again a convolution
of the size and shape distribution. The statistical
analysisofthedatarepresentedgraphicallyinFig. 1b
gives an average QD size of 25 nm and a standard
deviation of 6 nm. Expressed as a relative spread of
the QDs sizes along the lines of Ref. 21, we obtain a
shape/size “homogeneity” of 24%. This is still more
than twice the value that has been achieved for the
size “homogeneity” of InAs QDs in GaAs matrices,21
but a remarkable improvement over sample A. The
average QDs size has also increased slightly due to
the influence of the deposition of a fractional ML of
MnSe.
The structural observations from both samples can
be correlated to the PL spectra, Fig. 4. The PL peak of
sampleBisnarrowerandatanenergythatisapproxi-
mately 0.027 eV lower than the PL peak of sample A.
OurstandardinterpretationofanarrowerPLpeakat
a slightly lower emission energy around 2.3 eV is that
the shape/size distribution of 3D QDs is more homog-
enous and that the QDs are slightly larger and/or
contain more Cd.22
2
0
value ) of 22 nm and a standard deviation of 7 nm.
Expressed as a relative spread of the QD base widths
(
sizes) along the lines of Ref. 21, we obtain a shape/
size “homogeneity” of 32%. The QDs, seen in both low
resolution Z-contrast, Fig. 2b, and bright field images
(
where the influence of the strain field surroundings
can be suppressed by a suitable choice of the imaging
conditions) in STEM, possessed a wide variety of
shapes. On average, the length-to-width ratio of the
elongated QDs was approximately 1.4. There were, in
addition, QDs with shapes of more or less regular
rings, pairs of triangles, and dumbbells. About 20% of
the QDs were significantly elongated with a ratio of
the longer to smaller axis of about 4:1 and were
considered as representing coalesced QDs. Further-
more, orientation order, i.e., the lowest level in the
phenomenological hierarchy of self-ordered QDs ar-
An improved shape and size homogeneity, i.e.,
higherlevelsofself-ordering,inotherwords,hasbeen
observed before in 3D InP islands on (Ga,In)P, which
was lattice matched to GaAs, as a result of the depo-
3
rangements, wasobservedatspecialnucleationsites,
such as terrace edges.
For sample B (with Mn “sprinkling”), a more homo-
a
b
Fig. 3. (a) CTEM and (b) low-resolution Z-contrast STEM images of sample B in plan view.

7
52
Möck, Topuria, Browning, Titova, Dobrowolska, Lee, and Furdyna
10
–2
about (1.8±0.5) ¥ 10 cm . The number density of 3D
QDs was, thus, not only slightly higher as the a result
of the deposition of less than 0.1 ML of MnSe prior to
the deposition of the CdSe, but also more homog-
enous. This may be partly explained by the nominal
thickness of the CdSe layer, which may at certain
partsofthesamplebeabovethecriticalthicknessand
at certain other parts below it. In addition, it is
conceivable that a fractional ML of MnSe has similar
influences on the self-ordering processes to a frac-
tional ML of BeTe,15 although the static strain rela-
tions are reversed since bulk BeTe has a 9.4% smaller
lattice constant than ZnSe. As observed in other
semiconductor QDs systems before,2 a higher num-
ber density leads for 3D CdSe QDs in ZnSe matrix to
higher levels in the self-ordering hierarchy.
5,26
Further Structural Differences
Due to Mn “Sprinkling”
Fig. 4. PL spectroscopy results of samples A and B. The PL peak of
sample A is at 2.307 and has a full width at half maximum (FWHM) of
approximately 97 meV. The PL peak of sample B (Mn “sprinkling”) is
at 2.28 eV and has a FWHM of approximately 50 meV.
SincetheBohrradiusoftheexcitoninCdSeisabout
5
nm,15 all of the QDs described above cannot be
considered as “true” QDs in the sense of the definition
given in the introduction. There is, however, some
spatial confinement due to their size which justifies
the description as 3D QDs. The strained state of the
vast majority of the 3D QDs in both samples could be
inferred from the existence of strain contrast in the
CTEM images shown in Figs. 2a and 3a.
2
3
sition of a GaP layer in the ML range. As in our case,
the formation of smaller and less isotropically shaped
3
Dislandswassuppressedinfavoroftheformationof
larger and more isotropically shaped 3D islands. The
effect was explained by static strain considerations24
and a smoothing effect that may have delayed the
formation of the 3D QDs. We suspect that both of
these explanations are insufficient and are currently
working on alternatives. Another conclusion to be
drawn from our comparison of samples A and B is: as
in the case of InSb islands on GaSb, GaSb islands on
GaAs, and InSb QDs in GaSb matrices before,25 the
phenomenological model of the hierarchy of self-or-
For both samples, we observed smaller quasi-2D
QDsoflateraldimensionsoftheorderofmagnitudeof
about5nmindark-fieldCTEMimageswithanumber
11
–2
density in the 10 cm range for sample A (no Mn
3
dering processes has been proven to be applicable to
self-assembled semiconductor QDs.
Controlling the Number Density Value and
Homogeneity of 3D QDs by Mn “Sprinkling”
CTEM and low-resolution Z-contrast STEM shown
in Fig. 2a and b, reveal a rather inhomogeneous
distribution of the number density of the 3D QDs for
sample A (no Mn “sprinkling”). Areas with number
10
–2
densities of up to about 2 ¥ 10 cm co-existed with
areas where the 3D QDs number density was only
1
0
–2
about 10 cm . Based on the observation that there
are approximately two times more low number den-
sity areas than high-density areas, we estimate an
10
–2
average number density of about 1.3 ¥ 10 cm . On
another TEM specimen of sample A, there were,
however, hardly any 3D QDs visible. The above given
figures for the 3D QDs number density may, there-
fore, not be representative for the whole of sample A.
However, what is clear from the results is that there
is an inhomogeneous distribution of 3D QDs on this
sample.
Fig. 5. Low resolution Z-contrast image of sample B, [110] cross-
section. Arrows indicate 3D QDs with a vertical extension of approxi-
mately 8 nm and a lateral extension of approximately 16 nm. Note that
the QDs are, as we have observed before on similar samples,22
extended in the third dimension to an approximately equal amount
aboveandbelowtheleveloftheWL,possessingshapesthatresemble
“convex lenses”.
For sample B, areas with number densities of up to
1
0
–2
about 2.2 ¥ 10 cm coexisted with areas where the
3
1
D QDs number density was down to about 1.3 ¥
10 –2
0
cm , yielding an average number density of

Self-Ordered CdSe Quantum Dots in ZnSe and (Zn,Mn)Se Matrices Assessed
by Transmission Electron Microscopy and Photoluminescence Spectroscopy
753
10
–2
“
sprinkling”) and the 10 cm range for sample B.
QDs and Compositional Modulations in the
The high number density of these tiny QDs in sample
A explains the somewhat speckled contrast in Fig. 2a.
This type of QD, however, does not discernibly con-
tributetothePLspectrashowninFig.4,asithasbeen
shown earlier from photoluminesce at energies of
around 2.4–2.7 eV.2 The lower number density of
quasi-2D QDs in sample B may also be explained as
being a result of the deposition of the fractional ML of
MnSe. As the formation of 3D QDs probably proceed
in competition with the formation of quasi-2D QDs, a
higher number density of 3D QDs may have been
achieved by kinetic processes at the expense of the
number density of the quasi-2D QDs.
Nominal (Cd,Mn,Zn)Se Multilayer Structure
For the Cd0.03Mn0.097Zn0.873Se multi-layer sample
(C), the co-existence of different types of CdSe rich
aggregates in close proximity was observed. The first
of these aggregates were easily recognizable by their
pronounced strain fields in cross-section HRTEM
images. The lateral dimension of these aggregates
was of the order of magnitude 5 nm and this type of
QD may be considered as being quasi-2D. Much
larger CdSe rich aggregates of predominantly the
order of magnitude 100–200 nm lateral and 50–100
nm vertical represent the second type. Some of these
aggregates showed more or less well developed facets
in low- and atomic-resolution Z-contrast images and
were, despite their large size, usually free of struc-
tural defects (Fig. 6). Intriguingly, the contrast in the
imageindicatesthatthereisacompositionalordering
taking place in every second lattice plane of the
structure.
7,28
The 3D Nature of the QD Structures
Employing a nominal electron probe size of 0.5 nm,
EDXS spectra were taken for sample B from two
typical QDs and their surrounding areas. In order to
improve x-ray quantum counting statistics, analysis
times of several hours were employed. An enhanced
Cd concentration with respect to the surroundings of
about 20% was observed at the positions of the QDs.
This was also consistent with estimations of the local
Cd-concentration differences from high resolution
Z-contrast images. We interpret these results mainly
as projection effects and, thus, as proof of the 3D
nature of the typical QDs shown in Fig. 2a and b
Based on Fourier transforms of areas of the order of
2
magnitude 100 nm from the aggregate and its sur-
rounding matrix in Fig. 6b, we interpret the atomic
arrangement as compositional modulation on succes-
sive ± (110) planes. Self-ordering of such and other
kindshasbeenobservedbeforeinternaryandquater-
nary alloys of various III-V2 and II-VI compound
semiconductors which belong in their binary forms to
thesphaleritestructuretype.WeconsiderheretheZn
and Mn atoms as distributed spatially randomly, but
in a numerical relation of 9 to 1 over their respective
9,30
31
(
although there may also be differences in the Cd
composition of the 3D QDs and the remains of the
WL). For sample B, a low resolution Z-contrast as-
sessment of a [110] cross-section specimen also con-
firms the 3D nature of the QDs (Fig. 5).
±
(110)cation site super-lattice planes and Cd as a
second type of atomic species which occupies only its
a
b
Fig. 6. (a) Low resolution and (b) atomic resolution Z-contrast STEM image of a large 3D aggregate that shows self-ordering on an atomic level,
sample C, [110] cross-section. This aggregate was the only one of four of its kind that showed a structural defect in the form of a low-angle grain
boundary.

7
54
Möck, Topuria, Browning, Titova, Dobrowolska, Lee, and Furdyna
proper positions in alternating ± (110)cation site su-
per-lattice planes. This implies an aggregate compo-
sition of about Cd Mn0.056Zn0.444Se surrounded by a
magnitude 5 nm) aggregates with crystallographi-
cally different compositional modulation. We con-
clude that self-ordering processes of CdSe QDs show
a much richer phenomenology that the classical
Stranski-Krastanow growth model accounts for and
may, therefore, be of a dissipative nature.
0
.5
matrix of Mn Zn Se. Our interpretation is con-
0
.1
0.9
firmed within an accuracy of approximately 20% by a
quantitative analysis of the atomic resolution Z-con-
trast images along the lines of Ref. 18. Selected area
THEED patterns also confirm this interpretation
qualitatively.
ACKNOWLEDGEMENTS
The authors are grateful to Dr. Alan Nicholls from
the Electron Microscopy Service, Research Resources
Center of the University of Illinois at Chicago, for
experimentalsupportduringtheelectronmicroscopi-
cal investigations. PM and TT were supported by the
NationalSciencefoundationprojectnoDMR-9733895.
A rather isotropically shaped smaller aggregate of
theorderofmagnitude25nmwasobservedinHRTEM
to be free of structural defects and did not possess a
noticeable strain field. The analysis of the Fourier
transform of this aggregate and its surrounding ma-
trixindicatedanorderinginsuccessive± (111) planes,
thealsopresent±1/2 (113) extraspotsprobablybeing
due to double diffraction effects. Its composition is
likely to be Cd Mn0.056Zn0.444Se as well, effectively
REFERENCES
1
.
.
D. Bimberg, Semiconductors 33, 951 (1999).
2
N.N. Ledentsov, V.M Ustinov, V.A. Shchukin, V.A. Kop’ev,
Zh.I.Alferov,andD.Bimberg,Semiconductors32,343(1998).
D. Bimberg, M. Grundmann, and N.N. Ledentsov, Quantum
Dot Heterostructures (New York: John Wiley & Sons, 1999).
0
.5
constituting a 3D QD with a compositional modula-
tion inside. Aggregates of different compositional
modulation and with sizes of the order of magnitude
3.
4. V.A. Shchukin and D. Bimberg, Rev. Mod. Phys. 71, 1125
1999).
(
5
nm were also observed. The related selected area
1
1
1
1
1
1
5. R.A. BudimanandH.E. Ruda, J. Appl. Phys. 88, 4586(2000).
THEED patterns showed ±(
–
), ±(–
),
2
2
2
2
2
2
6
.
R. Heitz, T.R. Ramachandran, A. Kalburge, Q. Xie,
I. Mukhametzhanov, P. Chen, and A. Madhukar, Phys. Rev.
Lett. 78, 4071 (1997).
A.-L. Barabási, Appl. Phys. Lett. 70, 2565 (1997).
A. Madhukar, Q. Xie, P. Chen, and A. Konkar, Appl. Phys.
Lett. 64, 2727 (1994).
1
1
3
3
1
1
3
3
2
±(
–
–
), ±(
3
–
), ±
, ±(0 –
), ±(
2
2
2
2
2
2
(110)
2
2
3
2
1
1
0), ±(
–
1) reflections indicating that there
3
3
7
.
.
are, in addition to the well known cation ordering on
8
{
111} and {110} planes, atomic arrangements in close
spatial proximity that have not been reported before
in the literature.
9.
D.E. Jesson, G. Chen, K.M. Chan, and S.J. Pennycook, Phys.
Rev. Lett. 5156 (1998).
It is noticeable that this (Cd,Mn,Zn)Se multilayer
structure was grown with the intention of producing
vertically ordered arrays of quasi-2D CdSe QDs. At
present, we cannot explain why aggregates with dif-
ferenttypesofself-orderedatomicarrangementshave
been formed in addition to the quasi-2D QDs we
intended to grow. If the self-ordering processes dur-
ing Stranski-Krastanov growth were, however, of a
10. Self-orderingprocessescanbeeitherconservativeordissipa-
tive. An example for conservative self-ordering is the forma-
tion of crystalline structures with low energy content at
moderatelylowtemperatures.Itseemstobearguablewhether
ornottheprocessesthatleadtotheformationofsemiconduc-
tor QDs in certain II-VI and III-V compound semiconductor
systems may be classified as belonging to the conservative
self-ordering type since they show a very rich phenomenol-
ogy, e.g., Ref. 6, which depends critically on the growth
conditions.
1
0
dissipative nature, the co-existence of aggregates
withdifferentatomicarrangementsandstrainstatus
would be a quite common phenomenon. Further ex-
perimental data from III-V and II-VI compound semi-
conductor QDs systems appears to support this and is
presented elswhere.3
11. I. Suemune and A. Ishibashi, J. Cryst. Growth 214/215,
1 (2000).
1
2. S.H. Xin, P.D. Wang, A. Yin, C. Kim, M. Dobrowolska, J.M.
Merz, and J.K. Furdyna, Appl. Phys. Lett. 69, 3884 (1996).
3. C.S. Kim, M. Kim, S. Lee, J. Kossut, J.K. Furdyna, and
M. Dobrowolska, J. Cryst. Growth 214/215, 395 (2000).
1
2,33
14. J.K. Furdyna, J. Vac. Sci. Technol. A4, 2002 (1986).
1
5. M. Keim, M. Korn, F. Bensing, A. Waag, G. Landwehr,
S.V. Ivanov, S.V. Sorokin, A.A. Sitnikova, T.V. Shubina, and
A.A. Toropov, J. Cryst. Growth 214/215, 684 (2000).
6. P.D. Nellist and S.J. Pennycook, Ultramicroscopy 78, 111
(1999).
SUMMARY AND CONCLUSION
Summarizing the results on the single sheet CdSe
structures, we have shown that depositing a tiny
amount of MnSe before the deposition of the CdSe
layers (which order themselves into 3D QDs approxi-
mately 25 nm in size) leads to a significant improve-
ment of the shape and size homogeneity and higher
hierarchylevelsoftheself-orderedarrangement.This
is reflected by a narrower PL peak from these QDs.
Summarizing the preliminary results on the multi-
layer (Cd,Mn,Zn)Se structure, we observed the co-
existenceofthefollowingCdSerichaggregates:small
1
17. S.J. Pennycook, M.F. Chrisholm, Y. Yan, G. Duscher, and
S.T. Pantelides, Phys. B, 273 (1999).
1
1
2
2
8. S.J. Pennycook, S.D. Berger, and R.J. Culbertson, J. Micros-
copy 144, 229 (1986).
9. S. Lee, I. Daruka, C.S. Kim, A.-L. Barabási, J.L. Merz, and
J.K. Furdyna, Phys. Rev. Lett 81, 3279 (1998).
0. D.J. Hudson, Statistics, Lectures on Elementary Statistics
and Probabilities (1964).
1. D. Bimberg, M. Grundmann, and N.N. Ledentsov, MRS
Bulletin 23, 31 (1998).
22. C.S. Kim, M. Kim, J.K. Furdyna, M. Dobrowolska, S. Lee,
H. Rho, L.M. Smith, H.E. Jackson, E.M. James, Y. Xin, and
N.D. Browning, Phys. Rev. Lett. 85, 1124 (2000).
(
order of magnitude 5 nm, quasi 2D), highly strained
QDs, very large (order of magnitude 100 nm) compo-
sitionally modulated aggregates, negligibly strained
compositional modulated aggregates of the order of
magnitude of typical 3D QDs, and small (order of
2
3. N. Carlsson, K. Georgsson, L. Montelius, L. Samuelson,
W. Seifert, and R. Wallenberg, J. Cryst. Growth 156, 23
(
1995).
24. In order to explain the observed effects, it was suggested in

Self-Ordered CdSe Quantum Dots in ZnSe and (Zn,Mn)Se Matrices Assessed
by Transmission Electron Microscopy and Photoluminescence Spectroscopy
755
Ref. 23 that the very thin GaP layer may, although pseudo-
morphologically strained, lead to partial strain compensa-
tion since bulk GaP has a 3.6% smaller lattice constant than
and D. Bimberg, Appl. Phys. Lett. 76, 418 (2000).
28. M. Strassburg, Th. Deniozou, A. Hoffmann, R. Heitz,
U.W. Pohl, D. Bimberg, D. Litvinov, A. Rosenauer, D.
Gerthsen, S. Schwedhelm, K. Lischka, and D. Schikora,
Appl. Phys. Lett. 76, 685 (2000).
(
Ga,In)P when it is lattice matched to GaAs. Bulk InP,
finally, has a 3.8% larger lattice constant than GaAs. We
suspect, however, that static strain considerations may not
29. J.Y. Tsao, Materials Fundamentals of Molecular Beam Epi-
taxy (New York: Academic Press, Inc., 1993), pp. 93–150.
30. A. Zunger and S. Mahajan, Handbook on Semiconductors,
completely revised edition, ed. T.S. Moss, vol 3B, ed
S. Mahajan (Amsterdam: Elsevier Science B.V., 1994), pp.
1399–1514.
31. K. Park, L. Salamaca-Riba, and B.T. Jonker, Appl. Phys.
Lett. 61, 2302 (1992).
32. P. Möck, T. Topuria, N.D. Browning, G.R. Booker,
N.J. Mason, R.J. Nicholas. L.V. Titova, M. Dobrowolska,
S. Lee, and J.K. Furdyna, Mater. Res. Soc. Symp. 640, 6.3.1
(2000).
33. P. Möck, T. Topuria, N.D. Browning, G.R. Booker,
N.J. Mason, R.J. Nicholas. L.V. Titova, M. Dobrowolska,
S. Lee, and J.K. Furdyna, Proc. 6th Int. Symp. on Adv. Phys.
Fields, “Growth of Well-defined Nanostructures” (Tsukuba,
Japan: National Institute for Materials Science, 2001).
2
3
be sufficient in the above mentioned case nor in the pre-
sented case of the influence of a fractional monolayer of
MnSe on the formation of 3D CdSe QDs, where the strain
relations are qualitatively different. Alternatively, we sug-
gestthattheself-orderingprocessesthatleadtoanimproved
size and shape homogeneity of 3D CdSe and InP QDs may be
1
0
of a dissipative nature and, therefore, be more successfully
described by a dynamic synergistic approach (as already
6
called for by other authors in 1997 )—which we are going to
outline in forthcoming papers.
2
2
2
5. P. Möck, G.R. Booker, N.J. Mason, E. Alphandéry, and
R.J. Nicholas, IEE Proc. Optoelectron. 147, 209 (2000).
6. M. Schmidbauer, Th. Wiebach, H. Raidt, M. Hanke,
R. Köhler, and H. Wawra, Phys. Rev. B 58, 10523 (1998).
7. D.Schikora,S.Schwedhelm,D.J.As,K.Lischka,D.Litvinov,
A. Rosenauer, D. Gerthsen, M. Strassburg, A. Hoffmann,