Atomic force microscopy and photoluminescence study of Ge layers and self-organized Ge quantum dots on Si(100)

Article abstract of DOI:10.1063/1.116669

In this letter we present an atomic force microscopy and photoluminescence investigation of two- and three-dimensional Ge structures deposited on Si(100) substrates. The photoluminescence spectra demonstrated the presence of quantum confinement effects in flat and uncapped Ge layers and in Ge islands even for the relatively large size of the structures. Moreover, a correlation between the sample morphology and the features of the photoluminescence is demonstrated.

Full text of DOI:10.1063/1.116669

Atomic force microscopy and photoluminescence study of Ge layers and selforganized
Ge quantum dots on Si(100)
E. Palange, G. Capellini, L. Di Gaspare, and F. Evangelisti
Citation: Applied Physics Letters 68, 2982 (1996); doi: 10.1063/1.116669
View online: http://dx.doi.org/10.1063/1.116669
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/68/21?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Selfassembled quantum dots of InSb grown on InP by atomic layer molecular beam epitaxy: Morphology and
strain relaxation
Appl. Phys. Lett. 69, 3887 (1996); 10.1063/1.117559
Raman, opticalabsorption, and transmission electron microscopy study of size effects in germanium quantum
dots
Appl. Phys. Lett. 69, 2409 (1996); 10.1063/1.117653
Size effects on the phonon spectra of quantum dots in CdTedoped glasses
Appl. Phys. Lett. 69, 357 (1996); 10.1063/1.118059
Emission from discrete levels in selfformed InGaAs/GaAs quantum dots by electric carrier injection: Influence of
phonon bottleneck
Appl. Phys. Lett. 68, 3013 (1996); 10.1063/1.116681
Observation of lateral confinement effect in Ge quantum wires selfaligned at step edges on Si(100)
Appl. Phys. Lett. 68, 1847 (1996); 10.1063/1.116033
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 130.160.4.77
On: Sat, 20 Dec 2014 05:05:28

Atomic force microscopy and photoluminescence study of Ge layers
and self-organized Ge quantum dots on Si(100)
E. Palangea)
`
Dipartimento di Fisica, Universita di Roma ‘‘La Sapienza,’’ Piazzale A. Moro 2, 00185 Roma, Italy
G. Capellini, L. Di Gaspare, and F. Evangelisti
`
Dipartimento di Fisica ‘‘E. Amaldi’’ and Istituto Nazionale di Fisica della Materia, Universita di Roma III,
Via Vasca Navale 84, 00146 Roma, Italy
͑Received 7 December 1995; accepted for publication 15 March 1996͒
In this letter we present an atomic force microscopy and photoluminescence investigation of two-
and three-dimensional Ge structures deposited on Si͑100͒ substrates. The photoluminescence
spectra demonstrated the presence of quantum confinement effects in flat and uncapped Ge layers
and in Ge islands even for the relatively large size of the structures. Moreover, a correlation between
the sample morphology and the features of the photoluminescence is demonstrated. © 1996
American Institute of Physics. ͓S0003-6951͑96͒02221-8͔
In the past few years the physical properties of semicon-
ductor quantum confined systems have attracted wide
attention.1 In particular, due to the confinement of free car-
riers and excitons, quantum dots ͑QDs͒ are expected to give
an increase of the quantum efficiency by the enhancement of
the optical electron-hole recombination strength.2 This ap-
pears particularly important for indirect-gap materials, in
view of their possible applications in Si-integrated optoelec-
tronic devices operating in the spectral range of interest for
telecommunications.
In the literature, a wide list of recipes for realizing QDs
have been reported.3–7 In the case of self-aggregated dots
deposited by molecular beam epitaxy or chemical vapor
deposition ͑CVD͒ techniques, the reported photolumines-
cence ͑PL͒ spectral features are mainly attributed to three-
dimensional ͑3D͒ island formation even though there is not
as yet, a clear understanding of the quantum confinement
mechanism.
In this letter we report a comparative study of Ge struc-
tures on Si͑100͒ substrates by means of atomic force micros-
copy ͑AFM͒ and PL. The samples were prepared by low
pressure CVD. By changing the deposition parameters we
reproduced the three known modes for such heterostructures.
We will show that it is possible to identify confinement ef-
fects for samples without cap layers even for Ge structures
much larger than the Bohr exciton radius in this material.
We performed AFM and lateral force microscopy ͑LFM͒
topographies in a contact mode with an atomic resolution
Park Scientific Instruments CP equipped with high aspect
ratio conical ͑80° sidewall angle͒ Ultralever™ tips with a 5
nm minimum radius. The absence of tip imaging was always
verified by rotating the samples with respect to the scan di-
rection. The PL spectra were acquired in a standard lock-in
configuration by using an unfocused cw Arϩ ion laser, a 1-m
spectrometer and a LN2 cooled Ge photodetector.
X-ray photoemission spectroscopy ͑XPS͒ and RHEED
analysis showed that the surfaces were atomically clean and
2ϫ1 reconstructed. Moreover, the surfaces were flat with no
radiative dislocations in the Si substrate as checked by AFM
and PL analysis.
The samples were grown using high purity germane
without carrier gas. The deposition temperatures ranged from
350 to 850 °C and the gas pressure from 0.5 to 3 mTorr. As
verified by AFM and TEM analysis, under these growth con-
ditions we prepared three sample series in which we obtained
the layer-by-layer Frank–Van de Merwe ͑FVdM͒, the
Stranski–Krastanov ͑SK͒, and the Volmer–Weber ͑VW͒
growth modes. Generally we did not deposit a cap layer
since our main purpose was to study the relationship between
the optical and the morphological properties of the grown
structures.
In Fig. 1͑a͒ we present an AFM topography of a sample
grown according to the layer-by-layer mode at 350 °C and a
germane pressure of 3 mTorr. The Ge film is relaxed and
covers uniformly the Si substrate showing a planar surface
with a rms roughness of 1 monolayer ͑ML͒. Terraces are
present typically 1–2 ML thick on the average and ϳ150 nm
wide, with a density of ϳ1 island/ ␮m2. XPS measurements
indicate a thickness of ϳ3 nm, i.e., several times larger than
the SK critical thickness, that we measured to be ϳ3 ML by
cross sectional TEM, in agreement with the values reported
in the literature. As already noted,8 this large planar growth
results from the surface passivation and the surfactant role of
the hydrogen carried by the germane molecules when the
growth is at a temperature lower than the hydrogen desorp-
tion temperature ͓ϳ370 °C for Ge͑100͔͒. The surface hydro-
gen inhibits surface diffusion and prevents the 3D cluster
formation. The PL data reported in Fig. 1͑b͒ show that, be-
side the Si peaks on the high energy side, two major features
are present, the first centered at ϳ1015 meV and the second
at ϳ960 meV.
FZ Si͑100͒ substrates were cleaned with a procedure
consisting of an ex situ chemical etching followed by an
in situ thermal annealing in a stainless steel preparation
chamber with the base pressure in the 10Ϫ10 Torr range.
As for the 1015 meV peak, its energy position agrees
with the energy calculated for the no-phonon ͑NP͒ recombi-
nation of the Ge heavy-hole exciton, with the hole state lo-
calized in an 1.5-nm-thick Si/Ge/GeOx asymmetric well. The
energy broadening ͑ϳ15 meV at FWHM͒ is compatible with
a͒
Electronic mail: Palange@axcasp.caspur.it
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 130.160.4.77
2982 Appl. Phys. Lett. 68 (21), 20 May 1996 0003-6951/96/68(21)/2982/3/$10.00 © 1996 American Institute of Physics
On: Sat, 20 Dec 2014 05:05:28

FIG. 2. ͑a͒ AFM perspective view with artificial shadowing of a sample of
the SK type. In the top left-hand side corner is displayed a typical island
with the characteristic growth directions. The island side is 200 nm and the
height is 55 nm; ͑b͒ PL spectra of a capped ͑upper spectrum͒ and uncapped
͑lower spectrum͒ SK sample measured at 10 K.
FIG. 1. ͑a͒ AFM topography of a nominally 3-nm-thick Ge layer on
Si͑100͒; ͑b͒ PL spectrum of the same sample measured at 10 K.
the thickness fluctuations indicated by the AFM. It is worth
noting that the band offset at the Si/Ge interface was
measured directly on similar samples by using the photo-
electric yield spectroscopy technique and was found to be
⌬EVϭ0.36Ϯ0.02 eV.9 The fact that the Ge layer thickness
as evaluated by XPS and as obtained from the calculation of
the well binding energy do not agree better than within a
factor of two is not surprising, if the possible sources of
systematic errors are considered. These are: the uncertainty
in the exact value of the escape depth for the photoemitted
electrons in the XPS measurements; the simple effective
mass scheme we have employed for the evaluation of the
energy levels in the well; the reduction of the well thickness
due to the growth of the Ge oxide layer upon exposure of the
samples to air.
On the low energy side of the 1015 meV peak a shoulder
is present, which can be attributed to the TASi ͑18 meV͒
phonon replica of the NP line. The lower energy peak at 960
meV can be attributed to the TOSi phonon replica of the NP
line.10 This peak is also replicated by a TASi phonon line. All
these PL peaks are superimposed to a broad band that was
attributed to macroscopic defects.11
The second sample series of the SK type was obtained
for deposition temperatures in the range 400–700 °C. A typi-
cal topography of one of these samples is displayed in Fig.
2͑a͒. 3D islands are present, that nucleate above a critical
thickness to relieve the elastic strain due to the film-substrate
lattice mismatch. Their growth begins as tetragonal pyramids
with lateral facets oriented in the ͑11n͒ direction ͑n being
large͒. Upon increasing the deposited material, the value of n
decreases down to nϭ3 corresponding to the orientation of
minimum surface energy.12 After this limit we observed a
morphological transition toward truncated tetragonal pyra-
mids with a ͑100͒ top facet. A further increase of the depos-
ited material causes island coalescence with a poor sample
morphology.
In Fig. 2͑b͒ we report a typical PL spectrum for samples
belonging to this series together with the spectrum of a
nominally identical Si-capped sample. Beside the Si sub-
strate emission, the spectra present a broad band, which, in
the case of the uncapped sample, is centered at ϳ800 meV
and has a FWHM of ϳ80 meV. This PL band exhibits a
one-to-one correspondence with the appearance of 3D
islands.6,7,10,13 Its energy position is compatible with the
emission of Ge QDs having a mean diameter of ϳ10 nm.2
We also observed this feature in samples having pyramidal
shaped islands with ϳ300 nm base diameter and ϳ50 nm
height, i.e., several times larger than the Ge exciton Bohr
radius. Upon increasing the deposited material, the peak po-
sition of this band exhibited a slight red shift. In order to
explain these experimental evidences we suggest that the ex-
citonic recombination does occur in regions close to the top
of the islands, where the stronger confinement effects can
increase the recombination probability by weakening the mo-
mentum conservation rule. The substantial bandwidth can
result from the dimension fluctuations of the recombination
regions. A similar argument has been recently put forward to
explain PL spectra in SiGe quantum wires.14 The 2D emis-
sion taking place in the thin Ge layer uniformly covering the
substrate present in the SK growth mode is absent in the
spectra. This behavior is due to the absence of bound states
in the asymmetric well described above for thicknesses
smaller than 6 ML, i.e., twice the expected thickness of the
uniform layer. As a result, a ‘‘quantum depletion’’ takes
place that drives the charges toward the islands. The PL data
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 130.160.4.77
Appl. Phys. Lett., Vol. 68, No. 21, 20 May 1996 Palange et al. 2983
On: Sat, 20 Dec 2014 05:05:28

in single quantum dots similar to what was reported for the
luminescence in the InAs systems.16 However, we believe
that this is extremely unlikely in our case, since in our
samples these peaks always appear in the same position. As
a consequence, a very narrow distribution of islands with a
preferential size would be necessary to sustain such a hy-
pothesis, a narrow distribution which we do not observe.
In summary, we have shown the possibility to realize
quantum confinement conditions in uncapped 2D and 3D
Ge/Si structures. In the latter case the PL measurements evi-
dence quantum confinement effects even for clusters much
larger than the Bohr exciton radius. Our results suggest that
the 3D cluster shape plays an important role to determine the
quantum confinement conditions.
FIG. 3. PL spectrum of a VW type sample measured at 10 K.
of the capped sample support the explanation of the PL band
in terms of quantum confined systems. Indeed, in the capped
sample the band is centered at lower energy ͑780 meV͒ with
a FWHM equal to ϳ65 meV. This red shift can be accounted
for by the smaller confining potential of the capping Si as
compared to that of the oxide. Moreover, since in a shal-
lower confining potential the eigenvalues are less sensitive to
the well dimensions, the dimension fluctuations of the re-
combination regions on the island top are less effective
hence, the narrowing of the band in the capped sample.
The third sample series was obtained by increasing the
deposition temperature above 700 °C. In this regime the is-
lands were distributed according to two coexisting distribu-
tions: the first consisting of large and steep islands and the
second formed by very small islands having typical sizes of
10 nm. These small islands are regarded as the growth
precursors.15 As inferred by a close scrutiny of the LFM
images, these small clusters do not seem to be connected by
a Ge wetting layer to the larger islands. Therefore, we con-
cluded that the samples were of the Volmer–Weber ͑VW͒
type. The PL spectra of this series show the two bands re-
ported in Fig. 3. The first PL band is centered around 800
meV and is practically identical to the emission observed in
the SK type samples. Similarly, to the SK case it can be
attributed to the emission from the top regions of the large
islands. The second PL band is centered at higher energy
around 970 meV. It would be tempting to relate it to the PL
emission originated in the small islands acting as growth
precursors. However, further work is needed in order to
check this hypothesis. Finally, we want to indicate that in
some samples we observed the appearance of narrow lines
͑FWHMϷ2 meV, see Fig. 3͒ probably due to transitions of
excitons bounded to impurities. An alternative explanation
could be that these lines are related to transitions occurring
We are grateful to D. Zintu for her assistance during
the experiment and A. Miriametro and R. Moretto for tech-
nical assistance. This work was partially supported by
the Consiglio Nazionale delle Ricerche Contract No.
93.01269.CT02. L.D.G. is a recipient of a Postdoctoral Fel-
`
lowship of the Universita di Roma III and G.C. is a recipient
`
of a Doctoral Fellowship of the Universita di Tor Vergata,
Roma.
1 F. Capasso and S. Datta, Phys. Today 74 ͑1990͒.
2 T. Takahagara and K. Takeda, Phys. Rev. B 46, 15578 ͑1992͒.
3 J. R. Heath, J. J. Shiang, and A. P. Alivisatos, J. Chem. Phys. 101, 1607
͑1994͒.
4 J. Brunner, T. S. Rupp, H. Gossner, R. Ritter, I. Eisele, and G. Abstreiter,
Appl. Phys. Lett. 64, 994 ͑1994͒.
5 N. Usami, T. Mine, S. Fukatsu, and Y. Shiraki, Appl. Phys. Lett. 64, 1126
͑1994͒.
6 H. Sunamura, N. Usami, S. Fukatsu, and Y. Shiraki, Appl. Phys. Lett. 66,
3025 ͑1995͒.
7
¨
R. Apetz, L. Vescan, A. Hartmann, C. Dieker, and H. Luth, Appl. Phys.
Lett. 66, 445 ͑1995͒.
8 B. Cunningham, J. O. Chu, and S. Akbar, Appl. Phys. Lett. 59, 3574
͑1991͒; A. Sakai and T. Tatsumi, Appl. Phys. Lett. 64, 52 ͑1994͒.
9 L. Di Gaspare, G. Capellini, C. Chudoba, M. Sebastiani, and F. Evange-
listi, Appl. Surf. Sci. ͑to be published͒.
10 P. Schittenhelm, M. Gail, J. Brunner, J. F. Nutzel, and G. Abstreiter, Appl.
Phys. Lett. 67, 1292 ͑1995͒.
L. Vescan, K. Schmidt, C. Dieker, H. P. Tang, T. Vescan, and H. Luth,
11
¨
Thin Solid Films 222, 5 ͑1992͒.
12 D. J. Eaglesham, F. C. Unterwald, and D. C. Jacobson, Phys. Rev. Lett.
70, 966 ͑1993͒.
13 J. Brunner, J. Nutzel, M. Gail, U. Menczingar, and G. Abstreiter, J. Vac.
Sci. Technol. B 11, 1097 ͑1993͒.
A. Hartmann, C. Dieker, R. Loo, L. Vescan, H. Luth, and U. Bangert,
14
¨
Appl. Phys. Lett. 67, 1888 ͑1995͒.
15 Y. W. Mo, D. E. Savage, B. S. Swartzentruber, and M. G. Lagally, Phys.
Rev. Lett. 65, 1020 ͑1990͒.
16
´
¨
J.-Y. Marzin, J.-M. Gerard, A. Izrael, D. Barrier, and G. Bastard, Phys.
Rev. Lett. 73, 716 ͑1994͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 130.160.4.77
2984 Appl. Phys. Lett., Vol. 68, No. 21, 20 May 1996 Palange et al.
On: Sat, 20 Dec 2014 05:05:28