Evidence of x-ray absorption-edge shift as a function of luminescence wavelength in porous silicon

Article abstract of DOI:10.1103/PhysRevB.62.9911

X-ray absorption fine structure (XAFS) at the Si K edge in porous silicon has been measured by monitoring the photoluminescence yield (PLY) in different parts of the luminescence band: at increasing luminescence energy, a continuous positive shift of the x-ray absorption edge has been observed. The peculiar selectivity of the partial PLY-XAFS technique to the luminescent sites allows us to perform a size selection of the different nanostructures distributed in a single porous-silicon sample. The recombination of carriers localized in quantum confined nanocrystals is confirmed to be the main cause of the optical emission.

Full text of DOI:10.1103/PhysRevB.62.9911

PHYSICAL REVIEW B
VOLUME 62, NUMBER 15
15 OCTOBER 2000-I
Evidence of x-ray absorption-edge shift as a function of luminescence wavelength
in porous silicon
G. Dalba, N. Daldosso, P. Fornasini, M. Grimaldi, and R. Grisenti
`
Dipartimento di Fisica dell’Universita di Trento and Istituto Nazionale per la Fisica della Materia, I-38050 Povo, Trento, Italy
*
F. Rocca
CeFsa-Centro di Fisica degli Stati Aggregati del Consiglio Nazionale delle Richerche e Istituto Trentino di Cultura, I-38050 Povo,
Trento, Italy
͑Received 13 July 2000͒
X-ray absorption fine structure ͑XAFS͒ at the Si K edge in porous silicon has been measured by monitoring
the photoluminescence yield ͑PLY͒ in different parts of the luminescence band: at increasing luminescence
energy, a continuous positive shift of the x-ray absorption edge has been observed. The peculiar selectivity of
the partial PLY-XAFS technique to the luminescent sites allows us to perform a size selection of the different
nanostructures distributed in a single porous-silicon sample. The recombination of carriers localized in quan-
tum confined nanocrystals is confirmed to be the main cause of the optical emission.
In recent years, a great amount of scientific effort has
been made to understand the mechanism of visible light
emission in porous silicon.1,2 Photoluminescence in porous
silicon is due to the reduced dimensions of the nanostruc-
tures ͑wires or dots͒ obtained by a partial electrochemical
dissolution of bulk crystalline silicon. Shape and dimensions
of the porous structures are determined by the physical prop-
erties of the starting Si wafer and by the electrochemical
process parameters ͑HF content, current density, etching
time, light exposure time͒.3 By varying one of these quanti-
ties, it is possible to change the porosity and the nanostruc-
ture dimensions, obtaining a shift of the emitted light wave-
length, typically from the infrared to the green-blue region.
At present, a strong agreement has been reached on the
ability of the quantum confinement ͑QC͒ model to explain
the light emission properties of porous silicon.4,5 This model
predicts an enlargement of the band gap when the porous
structures are reduced to nanometer sizes: owing to the mo-
mentum uncertainty, radiative recombination of carriers lo-
calized in the quantum-confined nanocrystals is highly fa-
vored with respect to bulk Si. According to the QC model,
the emission energy should correspond to the width of the
energy gap Egap , in case of a band-to-band transition. The
observed broadening of the luminescence band is attributed
to the presence of a distribution of nanostructure sizes, which
corresponds to a distribution of Egap values inside the porous
layer.
Up to now, the effects of QC in porous silicon have been
tested only by comparing different samples, without consid-
ering the size distribution of luminescent sites inside each
single specimen.6,7 In this paper we exploit the peculiar site
sensitivity of x-ray absorption fine structure ͑XAFS͒ mea-
sured in photoluminescence yield ͑PLY͒ mode,8 making a
selection of nanostructure sizes within each single sample.
The work is based on the simultaneous measurement of sev-
eral x-ray absorption near edge structure ͑XANES͒ spectra in
PLY mode, each spectrum corresponding to a different emis-
sion energy range within the same optical band. This tech-
nique will be referred to as ‘‘partial PLY-XANES.’’ Actu-
ally, following the hypothesis that each nanoparticle of the
porous-silicon layer yields a sharp luminescence spectrum
͑whose emission energy is defined by its size͒, we may ex-
pect to obtain different partial PLY-XANES spectra by sam-
pling different energy ranges of the broad optical lumines-
cence band.
An essential role in supporting the QC model has already
been played by XAFS, by considering both XANES and ex-
tended x-ray absorption fine structure ͑EXAFS͒. XAFS has
been implemented in three different detection modes at the K
edge of Si: transmission,9 total electron yield ͑TEY͒,10–12
and PLY,8,13 also called XEOL ͑x-ray excited optical
luminescence͒.14 These detection modes are characterized by
different sensitivities. Transmission probes all the Si atoms
on the surface and in the bulk of the sample. TEY, measuring
secondary electrons, is sensitive to all Si atoms of the porous
layer; PLY measures the intensity of visible radiation emerg-
ing from luminescent Si sites. TEY-EXAFS experiments
have shown10,12 that porous-silicon samples with lumines-
cence bands centered at higher energies have in general
smaller site dimensions. In previous papers,8,15 the present
authors have experimentally shown that PLY, contrary to
TEY, monitors the local structure only of the Si atoms in the
luminescent sites. This site sensitivity appears to be specific
for quantum-confined systems and cannot be considered a
general property of all light emitting systems.16,17 This pecu-
liarity has been attributed to the presence of a very efficient
radiative recombination channel in the vicinity of a subset of
the absorbing Si atoms. For porous silicon, the localization
of carriers has been proved by different experiments:18 it is
proposed to occur via local size modulation of wire
structures1,19 or surface-confined states.20
The porous Si samples analyzed in this work were pre-
pared using a standard procedure,21 starting from p-type
boron-doped ͑100͒ Si wafers, with resistivity 20 ⍀ cm. In the
following, two types of samples, denoted as red and green,
will be considered. The red sample was obtained in an ethyl
alcohol solution of HF (15%) by applying a current density
of 40 mA/cm2 for 7.5 min; the green one by a current den-
0163-1829/2000/62͑15͒/9911͑4͒/$15.00
PRB 62
9911
©2000 The American Physical Society

9912
BRIEF REPORTS
PRB 62
FIG. 2. Experimental partial PLY-XANES spectra of the red
porous-silicon sample obtained by integrating the luminescence
band shown in Fig. 1 over energy intervals of 0.1 eV width.
FIG. 1. Normalized XEOL spectra of two porous-silicon
samples obtained from a 20 ⍀ cm crystal silicon wafer. The lumi-
nescence intensity has been integrated over intervals of 0.1 eV
width ͑vertical dashed lines͒ to obtain partial PLY-XAFS spectra.
O-Si-O bonds, at the expected energies of 1844 and 1847
eV, respectively.13 By normalizing the spectra at their maxi-
mum value, a neat shift of the x-ray absorption edge position
can be easily observed when the photoluminescence energy
increases, as shown in Fig. 3 for some selected spectra. Since
each set of PLY spectra was obtained from the same x-ray
energy scan, relative shifts of the absorption edge smaller
than the beamline energy resolution could be reliably appre-
ciated. For each sample, the relative shifts of the partial
PLY-XANES spectra have been evaluated with respect to the
most intense spectrum ͑corresponding to the maximum of
the XEOL band at 1.7 eV and 2.2 eV for red and green
samples, respectively͒, after a normalization to the absorp-
tion maximum at about 1841 eV. The average relative en-
ergy shifts are reported in Fig. 4. They were evaluated in two
ways: ͑a͒ the energy differences were measured in various
points of the absorption edge around the inflection point and
averaged; ͑b͒ the difference between spectra was minimized
by shifting each spectrum in a best-fitting procedure. The
error bars in Fig. 4 indicate the maximum spread of values
obtained by the different data-analysis procedures. It is easy
to recognize a common linear increase of the x-ray
absorption-edge energy monitored by partial PLY as a func-
tion of optical emission energy.
sity of 100 mA/cm2 for 6 min and a post-etching treatment
͑5 min under a 500 W halogen lamp͒. After the etching pro-
cedure, the samples were rinsed ͑the red one first in metha-
nol, then in pentane; the green one in deionized water͒ and
transferred, still wet, inside the experimental chamber on the
x-ray beamline, which was immediately evacuated to avoid
exposure to the air. Both samples presented homogeneous
porous layers, with thickness 8 and 11 ␮m, respectively.
Their luminescence bands were centered at about 730 and
570 nm, respectively. The normalized x-ray excited emission
spectra of red and green samples are shown in Fig. 1: they
exhibit a FWHM of about 0.3 eV and are about 1 eV wide at
the bottom. The shape difference shown by the two lumines-
cence bands is due to the different preparation parameters
and post-etching treatment. The bands shown in Fig. 1 were
obtained at the excitation energy of 1850 eV, just beyond the
Si absorption K edge ͑1839 eV͒. When exciting at different
energies, below and above the edge, no large changes of the
shape and position of the bands were observed. Slight varia-
tions of the relative intensities were, however, produced by
varying the x-ray energy near the absorption edge, and are
considered in the following analysis.
XAFS measurements were carried out with synchrotron
radiation at the Super ACO storage ring of the LURE Labo-
ratories in Orsay ͑France͒, using the SA 32 beamline. Elec-
tron energy and average current were 800 MeV and 400 mA,
respectively. The x-ray energy resolution was estimated 0.7
eV at 2000 eV. A system of lenses and optical fibers col-
lected the light emitted from the sample focusing it on a
dispersive spectrometer ͑CP200 Jobin-Yvon͒. An optimized
back-illuminated CCD ͑1100PB Princeton Instrument͒,
cooled with liquid nitrogen, was used as a detector to collect
in a few seconds the whole luminescence band at each x-ray
energy. After the x-ray energy scan, each band was corrected
for the spectral function of the XEOL apparatus, and subdi-
vided in stripes of 0.1 eV width ͑Fig. 1͒. Considering both
the samples, a total interval of luminescence energy ranging
from 1.6 eV up to 2.4 eV was explored. Unsmoothed partial
PLY-XANES spectra, obtained by integrating the lumines-
cence intensity from each stripe, are shown in Fig. 2 for the
red sample. These spectra do not reveal any presence of
oxygen-related absorption structures, such as Si-O-H or
Relative up-shifts of the x-ray absorption edge of porous
silicon with respect to c-Si were measured by different au-
thors both at L and K edges using TEY.6,22 These experimen-
FIG. 3. Partial PLY-XANES spectra of the red porous-silicon
sample normalized at the maximum: a neat positive shift of the
edge position with increasing the photoluminescence energy is ob-
served.

PRB 62
BRIEF REPORTS
9913
also monitor changes of the local environments or chemical
coverages of the nanostructures and their influence on the
photoluminescence. The study of the XANES region ͑0-40
eV͒ as a function of photoluminescence energy can provide
information on the variations of the CB electron density in
luminescent sites with different sizes. Improving the statis-
tics ͑i.e., using high-intensity synchrotron sources͒, the tech-
nique could also be exploited to study the EXAFS oscilla-
tions, in order to determine a quantitative correlation
between the CB shift and the average dimensions of lumi-
nescent sites.
A quantitative evaluation of the nanostructure’s size is out
of the possibilities of the present work, because we have only
measured relative shifts of the CB as a function of light
emission energy. It is well known that the absolute value of
the emission energy is influenced not only by the dimensions
of porous nanostructures, but also by their chemical
coverage:25 a different chemical passivation induces a shift
of the energy position of the photoluminescence band. How-
ever, the present results strongly support the importance of
QC in the photoluminescence properties of porous silicon,
because of the homogeneous surface properties inside the
measured area of each sample. An interesting comparison
can be done with the results presented in Ref. 25: our expe-
rience confirms that samples with yellow-green photolumi-
nescence ͑in the native solution͒ show a redshift as soon as
they are dried ͑in air, but also in vacuum͒. However, the
relative up-shift of the CB energy shown in Fig. 4 does not
present evidence of saturation: this result confirms that the
surface of emitting nanostructures in both red and green
samples is without any detectable oxygen coverage. As a
matter of fact, as Allan and co-workers have shown,25 the
presence of oxygen would produce a much lower CB shift or
even no shift. We note that their model for a free exciton
recombination in Si clusters totally passivated by hydrogen
foresees a ratio lower than 2:1 for the VB and CB shifts,
indicating a more complicated situation for real samples than
for model environments.
In conclusion, in this paper we have presented a tech-
nique, the partial PLY-XAFS, which allows us to correlate
the x-ray absorption-edge position to the emission energy of
luminescent nanocrystals, i.e., in an indirect way, the shift of
the bottom of the CB to the nanoparticles sizes. Thanks to
this technique, it is possible to follow the enlargement of the
energy gap, due to the presence of different nanoparticle
sizes in the same sample of porous silicon, without resorting
to a comparison with different calibrated samples. The quan-
titative analysis of the positive x-ray edge shift at increasing
photoemission energy in nonoxidized porous-silicon samples
shows that the luminescence mechanism agrees with the
quantum confinement theory.
FIG. 4. The relative shifts of all the edges for red and green
samples are plotted versus the photoluminescence energy. A
straight line of 1/3 slope is superimposed to the data.
tal results were interpreted as strong evidence of QC effects,
because the edge energy corresponds to a direct transition
from inner core levels ͑considered insensitive to the local
structure͒ to the unoccupied electronic states in the bottom of
the conduction band ͑CB͒. On such a basis, we can interpret
the relative shifts of the partial PLY-XANES edges as due to
the energy raise of the bottom of the CB in different lumi-
nescence sites characterized by increasing light emission en-
ergy and decreasing size.
Figure 4 shows that the correlation between the photolu-
minescence energy and the x-ray absorption-edge position is
linear. The present result can be compared with that found in
an experimental work on the electronic properties of Si
nanocrystals by van Buuren et al.23 These authors, measur-
ing different samples with selected sizes, correlated the ob-
served shifts in both CB and valence band ͑VB͒ with the
average dimensions of the nanocrystals, showing clear evi-
dence of quantum size effects. From the Data in Ref. 23 one
can deduce that the CB bottom is linearly correlated to the
band gap, with 1/3 slope. We have reported this behavior in
Fig. 4 ͑dashed lines͒: the agreement between the present par-
tial PLY-XANES results and those of Ref. 23 is quite good,
confirming that QC in nonoxidized porous silicon and in Si
nanodots produces a similar effect on the electronic states, in
particular on the enlargement of Egap . Very recently, the
same 2:1 ratio between VB and CB shifts has been measured
by photoemission and XANES spectroscopies for a porous-
silicon sample whose dimensions were reduced by succes-
sive wet-etching procedures.24
The main advantage of the present experiment consists in
the possibility to study the effects of QC on a continuous size
distribution within the same sample, rather than on a few
distinct samples, each one characterized by a different aver-
age size. Besides, since a PLY-XANES experiment is done
on a single sample with a homogeneous surface, we get rid
of the influence of different surface-coverage effects and of
the experimental uncertainties due to successive, indepen-
dent measurements of different samples.
The experimental differences between partial PLY-
XANES confirm once more8 the site selectivity of this tech-
nique: in fact, for porous silicon by PLY-XAFS it is possible
to study the local structure only of the emitting nanostruc-
tures, and in particular to distinguish those characterized by
different dimensions. In principle, the partial PLY-XAFS can
The authors are grateful to C. Armellini for the prepara-
tion and basic characterization of the samples, to R. Graziola
for his collaboration on the development of the new XEOL
apparatus, and to L. Pavesi, I. Mihalcescu, and R. Romestain
for stimulating discussions. We thank P. Lagarde and A. M.
Flank for scientific collaboration on the EXAFS measure-
ments. We acknowledge the support of the Training and Mo-
bility of Researchers ͑TMR͒ Program of the European Com-
munity for measurements at the LURE Laboratories ͑Orsay,
France͒.

9914
*
BRIEF REPORTS
PRB 62
Email address: rocca@science.unitn.it
15 G. Dalba, N. Daldosso, P. Fornasini, R. Graziola, R. Grisenti, and
F. Rocca, J. Non-Cryst. Solids 232, 370 ͑1998͒.
16 L. Soderholm, G.K. Liu, M.R. Antonio, and F.W. Lytle, J. Chem.
Phys. 109, 6745 ͑1998͒.
17 Y.L. Soo, S.W. Huang, Z.H. Ming, Y.H. Kao, G.C. Smith, E.
Goldburt, R. Hodel, B. Kulkarni, J.V.D. Veliadis, and R.N.
Bhargava, J. Appl. Phys. 83, 5404 ͑1998͒.
18 I. Mihalcescu, J.C. Vial, and R. Romestain, Phys. Rev. Lett. 80,
3392 ͑1998͒.
19 D. Ninno, G. Iadonisi, and F. Buonocore, Solid State Commun.
112, 521 ͑1999͒.
20 F. Koch, V. Petrova-Koch, T. Muschik, A. Nikolov, and V.
Gavrilenko, in Microcrystalline Semiconductors: Materials Sci-
ence & Devices, edited by P.M. Fauchet, C.C. Tsai, L.T. Can-
ham, I. Shimizu, and Y. Aoyagi, MRS Symposia Proceedings
No. 283 ͑Materials Research Society, Pittsburgh, 1993͒, p. 197.
21 L. Pavesi, M. Ceschini, G. Mariotto, E. Zanghellini, O. Bisi, M.
Anderle, L. Calliari, M. Fedrizzi, and L. Fedrizzi, J. Appl. Phys.
75, 1118 ͑1994͒.
1 A.G. Cullis, L.T. Canham, and P.D.J. Calcott, J. Appl. Phys. 82,
909 ͑1997͒.
2 Silicon-Based Microphotonics: From Basics to Applications, Pro-
ceeding of the International School of Physics ‘‘Enrico Fermi,’’
Course CXLI, edited by O. Bisi, S.U. Campisano, L. Pavesi, and
F. Priolo ͑IOS Press, Amsterdam, 1999͒.
3 R.L. Smith and S.C. Collins, J. Appl. Phys. 71, R1 ͑1992͒.
4 L.T. Canham, Appl. Phys. Lett. 57, 1046 ͑1990͒.
5
¨
V. Lehmann and U. Gosele, Appl. Phys. Lett. 58, 856 ͑1991͒.
6 T. van Buuren, Y. Gao, T. Tiedje, J.R. Dahn, and B.M. Way,
Appl. Phys. Lett. 60, 3013 ͑1992͒.
7 J. von Behren, T. van Buuren, M. Zacharias, E.H. Chimowitz,
and P.M. Fauchet, Solid State Commun. 105, 317 ͑1998͒.
8 G. Dalba, P. Fornasini, R. Grisenti, N. Daldosso, and F. Rocca,
Appl. Phys. Lett. 74, 1454 ͑1999͒.
9 G. Dalba, P. Fornasini, M. Grazioli, R. Grisenti, Y. Soldo, and F.
Rocca, Nucl. Instrum. Methods Phys. Res. B 97, 322 ͑1995͒.
10 S. Schuppler, S.L. Friedman, M.A. Marcus, D.L. Adler, Y.H. Xie,
F.M. Ross, T.D. Harris, W.L. Brown, Y.J. Chabal, L.E. Brus,
and P.H. Citrin, Phys. Rev. Lett. 72, 2648 ͑1994͒.
11 J.R. Dahn, B.M. Way, E.W. Fuller, W.J. Weydanz, J.S. Tse, D.D.
Klug, T. van Buuren, and T. Tiedje, J. Appl. Phys. 75, 1946
͑1994͒.
12 Qi Zhang and S.C. Bayliss, J. Appl. Phys. 79, 1351 ͑1996͒.
13 D.T. Jiang, I. Coultard, T.K. Sham, J.W. Lorimer, S.P. Frigo,
X.H. Feng, and R.A. Rosenberg, J. Appl. Phys. 74, 6335 ͑1993͒.
14 R.F. Pettifer, A. Glanfield, S. Gardelis, B. Hamilton, P. Dawson,
and A.D. Smith, Physica B 208&209, 484 ͑1995͒.
22 T.K. Sham, X.H. Feng, D.T. Jiang, B.X. Yang, J.Z. Xiong, A.
Bzowski, D.C. Houghton, B. Bryskiewicz, and E. Wang, Can. J.
Phys. 70, 813 ͑1992͒.
23 T. van Buuren, L.N. Dinh, L.L. Chase, W.J. Siekhaus, and L.J.
Terminello, Phys. Rev. Lett. 80, 3803 ͑1998͒.
24 T. Monguchi, H. Fujioka, K. Ono, Y. Baba, and M. Oshima, J.
Electrochem. Soc. 147, 602 ͑2000͒.
25 M.V. Wolkin, J. Jorne, P.M. Fauchet, G. Allan, and C. Delerue,
Phys. Rev. Lett. 82, 197 ͑1999͒.