HAL Id: hal-03314703
https://hal.archives-ouvertes.fr/hal-03314703
Submitted on 5 Aug 2021
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entic research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diusion de documents
scientiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Electronic States and Luminescence in Porous Silicon
Quantum Dots: The Role of Oxygen
M. V Wolkin, J. Jorne, P. M Fauchet, G. Allan, Christophe Delerue
To cite this version:
M. V Wolkin, J. Jorne, P. M Fauchet, G. Allan, Christophe Delerue. Electronic States and Lumi-
nescence in Porous Silicon Quantum Dots: The Role of Oxygen. Physical Review Letters, American
Physical Society, 1999, 82 (1), pp.197-200. �10.1103/PhysRevLett.82.197�. �hal-03314703�
VOLUME
82, NUMBER 1 PHYSICAL REVIEW LETTERS 4J
ANUARY
1999
Electronic States and Luminescence in Porous Silicon Quantum Dots: The Role of Oxygen
M. V. Wolkin, J. Jorne,* and P. M. Fauchet
†
Materials Science Program, University of Rochester, Rochester, New York 14627
G. Allan and C. Delerue
Département Institut Supérieur d’Electronique du Nord, Institut d’Electronique et de Microélectronique du Nord,
41 boulevard Vauban 59046 Lille cédex, France
(
Received 25 September 1998)
Depending on the size, the photoluminescence (PL) of silicon quantum dots present in porous silicon
can be tuned from the near infrared to the ultraviolet when the surface is passivated with Si-H bonds.
After exposure to oxygen, the PL shifts to the red by as much as 1 eV. This shift and the changes in
PL intensity and decay time, show that both quantum confinement and surface passivation determine
the electronic states of silicon quantum dots. A theoretical model in which new electronic states appear
in the band gap of the smaller quantum dots when a Si
—
—
O bond is formed, is in good agreement
with experiments. This result clarifies the controversy regarding the PL mechanisms in porous silicon.
[S0031-9007(98)08118-6]
PACS numbers: 78.55.Mb, 71.24.+q
The study of silicon quantum dots is a very active field
of research, because of the interesting fundamental physi-
cal properties of these mesoscale objects and of promising
applications in advanced electronic devices [1] and opto-
electronic devices [2]. Of all the forms of silicon contain-
ing quantum dots, porous silicon (PSi) is the one that has
attracted the most attention to date [3–5], mostly because
of its intense visible photoluminescence (PL). Numerous
models have been proposed to explain its PL, includ-
ing quantum confinement [6–8], surface states [9], de-
fects in the oxide [10], and even specific chemical species
(e.g., siloxenes [11]). Presently, although a detailed un-
derstanding of the PL has yet to be achieved [4], it is
usually accepted that the band gap opens as a result of
quantum confinement, which pushes the PL in the visible
for crystallite sizes below 5 nm [8]. It has even been pro-
posed that quantum confinement provides an explanation
for all the slow PL from PSi [3]. However, many groups
have reported that when the crystallite size decreases to
a few nanometers, the PL in air does not increase much
beyond 2.1 eV even when the crystallite size drops well
below 3 nm [12,13]. This observation does not coincide
with theory, which predicts a much larger opening of the
band gap, in excess of 3 eV for sizes below 2 nm [8,14].
The goal in this Letter is to solve that apparent
contradiction and to give a global explanation for the
visible PL and the electronic states in PSi. We examine
oxygen-free PSi samples with different porosities and
emitting throughout the visible spectrum (red to blue,
following Ref. [15]). After exposure to air, a redshift of
the PL is observed, which can be as large as 1 eV for blue
luminescent samples that contain crystallites smaller than
2 nm. After a full characterization of the PL properties
and the chemical composition, we propose a model that
describes the recombination of carriers over a wide range
of nanocrystal sizes and different environments. In this
model, the redshift of the PL immediately after exposure
to oxygen is related to the trapping of an electron (or even
an exciton) by Si
—
—
O bonds that produce localized levels
in the bandgap of nanocrystals smaller than ,3 nm.
The PSi samples were formed by electrochemical
etching followed by photoassisted stain etching of 6 Vcm
p-type Si wafers at current densities of 8
50 mAycm
2
using 10%–25% HF:ethanol solutions. Stain etching,
accomplished under illumination with a 500 W halogen
lamp, was used to further increase the porosity [16].
All samples were rinsed in ethanol and immediately
transferred into an Ar environment. Special care was
taken that the samples never be exposed to air. The blue
sample was the only one measured under vacuum. A
second set of identical samples were rinsed in ethanol and
then exposed to air.
The PL spectra were obtained at room temperature by
using either a pulsed excitation nitrogen laser (337 nm) or
a continuous excitation HeCd laser (325 nm). Figure 1a
shows the PL spectra of five types of oxygen-free PSi
samples with different porosities. Stable red, orange,
yellow, green, and blue spectra, in increasing order of
porosity, were obtained and measured when the samples
were kept under the Ar environment. The PL intensity
increased by several orders of magnitude as the PL
wavelength changed from the red to the yellow, consistent
with the quantum confinement model. However, as
the porosities were further increased, the PL intensity
dropped, at least in part due to the fact that the penetration
depth became larger than the PSi thickness. Figure 1b
shows how the spectra were modified after the samples
had been exposed to air for 24 h. Two major trends were
observed. The PL from the samples emitted in the blue
to orange region was redshifted, and the PL intensities
decreased. The magnitude of the redshift increased with
increasing porosities, and the PL peak energy eventually
0031-9007y99y82(1)y197(4)$15.00 © 1998 The American Physical Society 197
VOLUME
82, NUMBER 1 PHYSICAL REVIEW LETTERS 4J
ANUARY
1999
saturated near 2.1 eV (590 nm) for the green and blue
samples. For the samples that initially emitted in the red,
no redshift was observed, and the PL remained the same
whether the samples were stored in Ar or aged in air.
We measured the PL decay time in Ar and in air at
the maximum PL energy of each sample, and found it be
in the microsecond range. The lifetimes were obtained by
fitting a stretched exponential function to the experimental
decay transients. In Ar, as the porosity increased, the
lifetime decreased monotonically, from 32 ms (in the red)
to 0.07 ms (in the blue). After exposure to air, the
lifetimes also decreased with increasing porosities, until
they reached 2 ms (initially green PL), after which they
stayed constant. In addition, the difference between the
lifetime measured in air and Ar increased with increasing
porosity.
In order to understand the origin of the redshift after
24 h of exposure to air, we investigated the PL in different
gas environments. A large redshift was observed as soon
as the samples were transferred from Ar to a pure oxygen
atmosphere. In contrast, no redshift at all was detected
when the samples were kept in pure hydrogen or in
vacuum. We hypothesized that the large redshift was
related to surface passivation, and probably the presence
of oxygen.
To test this hypothesis, we investigated the evolution
of the chemical coverage of an Ar-stored sample as
it was exposed to air. Figure 2a shows the evolution
of the Fourier transform infrared spectroscopy (FTIR)
transmission spectra of a blue-green sample before st
0d and after exposure to air st . 0d. The spectrum of
the fresh sample showed strong absorption bands near
2100 and 664 cm
21
, associated with the stretching and
deformation of Si—H
n
sn 1 3d, and no sign of an
oxygen peak, which confirmed that the samples stored in
Ar were well passivated by hydrogen and free of oxygen.
FIG. 1. Room temperature photoluminescence spectra from
PSi samples with different porosities kept under Ar atmosphere
(a) and after exposure to air (b).
As fast as 3 min after exposure to air, a Si—O—Si feature
at 1070 cm
21
appeared and gradually became dominant.
In addition, after 100 min, a new peak was observed
at 850 cm
21
related to Si—O—H. The Si H
n
peaks
at 2100 cm
21
decreased progressively with time and
disappeared after 24 h. When the samples were exposed
to air for longer than 200 min, no significant change in the
Si—O—Si and Si—O—H peaks was observed, indicating
stabilization of the surface chemical coverage.
As the surface passivation was gradually changing,
the PL was redshifted. Figure 2b shows the progressive
redshift of the PL with time. Most of it was obtained in
the first few minutes of exposure, and stabilization was
achieved after aging for 200 min. This result correlates
well with the change of the surface passivation. Based
on the previous results, we argue that both porosity (or
size) and chemical coverage dictate the recombination
mechanism.
All these results suggest that the electron-hole recom-
bination in samples exposed to oxygen occurs via carriers
trapped in oxygen-related localized states that are stabi-
lized by the widening of the gap induced by quantum
confinement. Thus we have applied electronic structure
calculations to various situations involving oxygen atoms
at the surface of Si clusters. As expected because of the
large offset between bulk SiO
2
and Si s,4 eVd, for nor-
mal Si—O—Si bonds, we do not find any localized gap
state. Similar results are obtained for Si—O—H bonds.
But, when nanocrystalline Si is oxidized and a Si—O—Si
FIG. 2. Evolution of the FTIR transmission spectra (a) and
the PL redshift (b), for a blue-green sample as a function of air
exposure time.
198
VOLUME
82, NUMBER 1 PHYSICAL REVIEW LETTERS 4J
ANUARY
1999
layer is formed on the surface, the Si—Si or Si—O—Si
bonds are likely to weaken or break in many places
because of the large stress at the SiySiO
2
interface [17].
Some mechanisms can act to passivate the dangling bonds
[18]. A Si
—
—
O double bond is more likely to be formed
and stabilize the interface, since it requires neither a large
deformation energy nor an excess element. It would also
terminate two dangling bonds. Such bonds have been
suggested at the SiySiO
2
interface [18].
The electronic structure of Si clusters with one Si
—
—
O
bond (the other dangling bonds being saturated by hydro-
gen atoms) is calculated as a function of the cluster size.
We use a self-consistent tight-binding method closely fol-
lowing Ref. [19]. To correctly reproduce the valence and
conduction bands of bulk silicon, we use the parame-
ters of Ref. [20], which include three-center integrals up
to the third nearest neighbors. The interactions between
Si and O are obtained from Ref. [21]. The Hamilton-
ian matrix is diagonalized using an inverted Lanczos it-
eration procedure which allows us to treat clusters up
to ,5 nm. The optical matrix elements are obtained as
in Ref. [8], i.e., without the assistance of phonons. For
small clusters we have verified the consistency of our
results with those of ab initio local density calculations
performed with the code of Ref. [22]: Similar electronic
structures are obtained even if the positions of the states in
the gap cannot be compared exactly because of the well-
known underestimation of the band gap by local density
calculations.
The calculated electronic states in Si nanocrystals are
presented in Fig. 3. The model suggests that when a Si
cluster is passivated by hydrogen, recombination is via
free excitons states for all sizes. The PL energy is equal
to the free exciton band gap and follows the quantum con-
finement model. However, if the Si nanocrystallite is pas-
sivated by oxygen, a stabilized electronic state (or even a
trapped exciton) may be formed on the Si
—
—
O covalent
bond. The electron state is a p state, localized on the Si
atom, and the hole size is a p state, localized on the oxy-
gen atom. For oxygen-passivated clusters, three different
recombination mechanisms are suggested, depending on
the size of the cluster. Each zone in Fig. 3 corresponds
to a different mechanism. In zone I, recombination is via
free excitons. As the cluster size decreases, the PL energy
increases, exactly as predicted by quantum confinement.
There is no redshift whether the surface termination is hy-
drogen or oxygen, since the band gap is not wide enough
to stabilize the Si
—
—
O surface state. In zone II, recom-
bination involves a trapped electron and a free hole. As
the size decreases, the PL emission energy still increases,
but not as fast as predicted by quantum confinement, since
the trapped electron state energy is size independent. In
zone III, recombination is via trapped excitons. As the
size decreases, the PL energy stays constant, and there is
a large PL redshift when the nanocrystallite surface be-
comes exposed to oxygen.
FIG. 3. Electronic states in Si nanocrystals as a function of
cluster size and surface passivation. The trapped electron state
is a p-state localized on the Si atom of the Si
—
—
O bond and the
trapped hole state is a p-state localized on the oxygen atom.
In order to compare the calculations with experimental
results, it is necessary to evaluate the nanocrystallite size.
In ultrahigh porosity samples, the crystallites are very
small s#2 nmd, and there is no obvious way to measure
their size reliably. If we accept that the PL in PSi
stored under an Ar atmosphere is due to recombination
via free excitonic states, the PL energy itself can be
used to deduce the average size. Therefore, we have
equated the calculated excitonic band gap and the peak
PL energy to obtain the size of the nanocrystals in
each PSi sample. Figure 4 presents the experimental
PL energy (measured in Ar and air), and the calculated
PL energy (free exciton energy and lowest transition
energy for a nanocrystal with a Si
—
—
O bond) as a
function of nanocrystal sizes. The agreement between
experiments and theory is good, despite the simplicity
of the model. As expected, three zones are observed;
zone I for free excitonic recombination, independent of
the surface bonds; zone II where, in the presence of a
Si
—
—
O bond, the electron is localized on that bond but
the hole is free; and zone III where the electron and the
hole are trapped on the Si
—
—
O bond (“trapped exciton”).
The magnitude of the measured redshift is as calculated
in the model. In addition, we note that the experimental
and theoretical PL decay lifetimes are in agreement.
Therefore, we conclude that the model proposed for the
electronic states and the luminescence of porous silicon
quantum dots explains the experimental data.
Taken together, the results in this paper indicate that:
(a) The band gap of Si nanocrystals opens by quan-
tum confinement. The emission energy increases with
decreasing sizes, covering the entire visible spectrum;
(b) surface passivation plays an important role, especially
199
VOLUME
82, NUMBER 1 PHYSICAL REVIEW LETTERS 4J
ANUARY
1999
FIG. 4. Comparison between experimental and theoretical PL
energies as a function of crystallite size. The upper line is
the free exciton band gap and the lower line is the lowest
transition energy in the presence of a Si
—
—
O bond. The d
and s are the peak PL energies obtained from Figs. 1a and
1b, respectively. In zone I the PL peak energies are identical,
whether the samples have been exposed to oxygen or not.
at small sizes sd , 3 nmd. The recombination mecha-
nism in oxidized nanocrystallites of this size is different
from that in hydrogen passivated crystallites; (c) the red-
shift upon oxidation is related to recombination involv-
ing a trapped electron or exciton. It explains the huge
Stokes shift observed for the smallest nanocrystallites and
the upper limit of the emission energy (2.1 eV), which is
independent of size. The agreement between the theoreti-
cal model and the experimental results indicates that it is
likely that the carriers are trapped in the Si
—
—
O bond. The
fact that a redshift was observed in a pure oxygen envi-
ronment shows that other bonds (such as Si—O—H) are
less likely to be responsible for the PL; (d) the nanocrystal
sizes in the PSi samples investigated by most researchers
are larger than 3 nm and display red or orange emission.
In those, quantum confinement alone can explain most of
the results, and surface passivation has a negligible im-
pact on the radiative recombination mechanism; (e) many
researchers define “fresh samples” as samples which have
been exposed to air from a few minutes up to half an
hour. However, our results show that oxidation occurs
in seconds, changing the sample’s recombination mecha-
nism and optical properties. Our results also suggest that
an oxide may not provide a good passivation in small Si
crystallites, in contrast to bulk Si. This observation may
be very important for future nanoelectronics applications.
Support from the U.S. Army Research Office and
the Electric Power Research Institute is gratefully ac-
knowledged. We thank L. Tsybeskov, C. Wamsley,
and G. Wicks for technical assistance. The “Institut
d’Electronique et de Microélectronique du Nord” is
“Unite mixte 9929 du Centre National de la Recherche
Scientifique.”
*Also with Department of Chemical Engineering.
†
Also with Department of Electrical and Computer
Engineering.
[1] A. Fowler, Phys. Today 50, No. 10, 50 (1997).
[2] K. D. Hirschman, L. Tsybeskov, S. P. Duttagupta, and
P.M. Fauchet, Nature (London) 384, 338 (1996).
[3] A. G. Cullis, L. T. Canham, and P. D. J. Calcott, J. Appl.
Phys. 82, 909 (1997).
[4] P. M. Fauchet, J. Lumin. 70, 294 (1996).
[5] R. T. Collins, P. M. Fauchet, and M. A. Tischler, Phys.
Today 50, No. 1, 24 (1997).
[6] L. T. Canham, Appl. Phys. Lett. 57, 1046 (1990).
[7] V. Lehman and U. Gösele, Appl. Phys. Lett. 58, 865
(1991).
[8] J. P. Proot, C. Delerue, and G. Allan, Appl. Phys. Lett. 61,
1948 (1992).
[9] F. Koch, V. Petrova-Koch, and T. Muschik, J. Lumin. 57,
271 (1993).
[10] S. M. Prokes, Appl. Phys. Lett. 62, 3244 (1993).
[11] M. S. Brandt, H. D. Fuchs, M. Stutzmann, J. Weber, and
M. Cardona, Solid State Commun. 81, 307 (1992).
[12] J. von Behren, T. Van Buuren, M. Zacharias, E. H.
Chimowitz, and P. M. Fauchet, Solid State Commun. 105,
317 (1998).
[13] S. Schuppler et al., Phys. Rev. B 52, 4910 (1995).
[14] L. Wang and A. Zunger, Phys. Chem. 98, 2158
(1994).
[15] H. Mizuno, H. Koyama, and N. Koshida, Appl. Phys. Lett.
69, 3779 (1996).
[16] H. Koyama and N. Koshida, J. Appl. Phys. 74, 6365
(1993).
[17] A. Ourmazd, D. W. Taylor, J.A. Rentschler, and J. Bevk,
Phys. Rev. Lett. 59, 213 (1987).
[18] F. Herman and R. V. Kasowski, J. Vac. Sci. Technol. 19,
395 (1981).
[19] M. Lannoo and J. Bourgoin, in Point Defects in Semicon-
ductors I, edited by M. Cardona (Springer-Verlag, New
York, 1981).
[20] C. Tserbak, H. M. Polatoglou, and G. Theodorou, Phys.
Rev. B 47, 7104 (1993).
[21] M. O’Reilly and J. Robertson, Phys. Rev. B 27, 3780
(1983).
[22] DMol User Guide (Molecular Simulations, San Diego,
1996).
200
