ARTICLE IN PRESS

Physica B 403 (2008) 2634–2638

www.elsevier.com/locate/physb

Characterization of thermal, optical and carrier transport properties of

porous silicon using the photoacoustic technique
Chan Kok Shenga, W. Mahmood Mat Yunusa,, Wan Md. Zin Wan Yunusb,
Zainal Abidin Taliba, Anuar Kassimb
a
Department of Physics, Faculty of Science, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia
b
Department of Chemistry, Faculty of Science, Universiti Putra Malaysia, 43400 UPM Serdang, Malaysia
Received 9 October 2007; received in revised form 31 December 2007; accepted 22 January 2008

Abstract

In this work, the porous silicon layer was prepared by the electrochemical anodization etching process on n-type and p-type silicon
wafers. The formation of the porous layer has been identified by photoluminescence and SEM measurements. The optical absorption,
energy gap, carrier transport and thermal properties of n-type and p-type porous silicon layers were investigated by analyzing the
experimental data from photoacoustic measurements. The values of thermal diffusivity, energy gap and carrier transport properties have
been found to be porosity-dependent. The energy band gap of n-type and p-type porous silicon layers was higher than the energy band
gap obtained for silicon substrate (1.11 eV). In the range of porosity (50–76%) of the studies, our results found that the optical band-gap
energy of p-type porous silicon (1.80–2.00 eV) was higher than that of the n-type porous silicon layer (1.70–1.86 eV). The thermal
diffusivity value of the n-type porous layer was found to be higher than that of the p-type and both were observed to increase linearly
with increasing layer porosity.
r 2008 Elsevier B.V. All rights reserved.

PACS: 65.80.+a; 78.55.m; 78.70.+m

Keywords: Porous silicon; Photoacoustic; Energy gap; Thermal properties

1. Introduction Photoacoustic (PA) and photothermal (PT) techniques

Nowadays, porous photonic materials have found
various potential applications in display technology,
optical data storage and chemical sensors. Porous silicon
(PSi) photonic material has generated a great deal of
attention because of its efficient visible photoluminescence
at room temperature and the compatibility with standard
silicon (Si) processes in fabricating integrated optoelec-
tronic devices, such as light-emitting diode and solar
cell. PSi is formed by electrochemical anodization of
a Si wafer in hydrofluoric acid (HF) at room tempera-
ture [1,2].
Corresponding author. Tel.: +603 89466684; fax: +603 89454454.
E-mail address: mahmood@science.upm.edu.my
(W. Mahmood Mat Yunus).
have been established as diagnostic methods with good
sensitivity to the dynamics of photo-excited carriers. This
technique has significant advantages due to its great
versatility as a sensitive, non-destructive and non-invasive
character for the evaluation of material parameters, as well
as in the semiconductor industry for characterizing
processes in the manufacturing of microelectronic devices.
On the other hand, PA spectra are rather insensitive to
light-scattering effects in porous materials, such as porous
TiO2, which has been reported by Toyoda [3]. Since the
generation of excess carriers will produce thermal waves
due to the carrier thermalization and recombination
processes, the carrier transport properties of semiconduc-
tor materials can be quantified from PA phase signal
experimental data [4–9]. In this paper, we report the PA
measurement of thermal diffusivity, energy band gap and

0921-4526/$ - see front matter r 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.physb.2008.01.029
 ARTICLE IN PRESS
C.K. Sheng et al. / Physica B 403 (2008) 2634–2638 2635

carrier transport properties of the PSi layers prepared at
various surface porosities.
2. Experimental details
PSi samples were prepared on p-type and n-type Si
wafers in an electrolyte mixture of HF containing ethanol
in the proportion 1:1. The etching duration was fixed at
20 min and the etching process was done at room
tempterature. A direct current (dc) bias voltage was applied
between the Pt electrode that served as the cathode and the
Al–Si wafer as the anode. The generation process of PSi is
termed as anodization since Si acts as the anode during the
etching process. Five different current densities were used
in the preparation of PSi, i.e. 16.98, 22.64, 28.29, 33.96 and
39.62 mA/cm2, respectively. In this work, the PSi layers on
p-type and n-type Si substrates are referred to as p-PSi and
n-PSi, respectively. After anodization, the PSi samples were
dried in air and stored in the dark for 24 h prior to PA
Xenon arc Lamp
Chopper
Monochromator
Lens
Personal Lock-in
Amplifier
Computer
measurements. Porosity was measured using a gravimetric
technique. The formation of the PSi layer on Si substrate in
the present work was verified by measuring photolumines-
cence intensity using a fiber optics spectrophotometer
(Ocean Optics USA). A scanning electron micrograph
(SEM) and EDX measurements were also carried out to
study the surface morphology and atomic information of
the PSi samples.
The PA measurements were carried out by the gas-
microphone detection technique. The xenon light beam,
after being mechanically modulated by an optical chopper
(SR540), was focused onto the sample kept inside a non-
resonant PA cell. The sound wave generated from the
sample could be subsequently detected and converted into
PA signal by a highly sensitive electret microphone fixed in
the PA cell. The PA signal was then amplified by a low-
noise preamplifier (SR560) and further processed using a
lock-in amplifier (SR530). A personal computer was
interfaced to the system for automatic data collection and
analysis. Fig. 1 shows the schematic diagram of the PA
experimental set-up for optical absorption and energy-gap
measurement of the PSi layer. For carrier transport
properties measurements the same set-up was used where
the xenon light source was replaced by 75 mW (632.8 nm)
He–Ne laser (Melles Griot, 05LHR991). The PA phase
signal was measured by changing the chopping frequency
from 30 to 100 Hz. A vibration-free experimental set-up
with a good normalization using high-quality carbon black
has been used throughout the measurements to enhance the
signal–noise ratio. In addition, the measurement was
repeated several times by keeping the beam intensity and
the sample at the same position.

Preamplifier 3. Results

PA Cell For optical absorption and energy gap, the measure-

Fig. 1. Schematic diagram of the PA experimental set-up for spectro- ments were carried out by recording the PA signal as a
scopic study. function of photon energy. The absorption peaks shown in

0.16 1.6

0.14 45 minutes 1.4 15 Hz

PA Intensity (Arb. Units)

40 minutes
0.12 1.2
35 minutes
0.10 30 minutes 1.0 23 Hz
(Ihv)1/2

25 minutes
0.08 20 minutes 0.8

0.06 0.6 33 Hz

0.04 0.4

0.02 0.2
Silicon Substrate 1.11 eV
2.00 eV
0.00 0.0
1.5 1.8 2.1 2.4 2.7 3.0 3.3 3.6 1.0 1.5 2.0 2.5 3.0
Photon Energy (eV) Photon Energy (eV)

Fig. 2. Normalized PA signal-intensity spectra for (a) p-PSi samples prepared at six different anodization durations (modulation frequency of 33 Hz),
(b) typical plot of (Ihn)1/2 versus photon energy for p-PSi samples.
 ARTICLE IN PRESS
2636 C.K. Sheng et al. / Physica B 403 (2008) 2634–2638

Fig. 2(a) are obviously shifted to the higher photon energy,
which implies that the energy gap increased with the
increase of porosity. Fig. 2(b) demonstrates the plots of PA
absorption versus photon energy for evaluating the band-
gap energy of the PSi layer. The relationship between
sample porosity and the current density of PSi layers
prepared on p-type and n-type Si substrates is shown in
Fig. 3. The porosity is linearly dependent on the current
density used in the electrochemical etching process. It was
also observed that for the same current density, p-type Si
substrates produced higher porosity than the n-type
80
76
72
p-PSi
68
Porosity (%)
substrate. Fig. 4 shows typical SEM and EDX spectra
for porous layer on different substrates and prepared at
different current densities. The SEM images show the
existence of small micropores surrounded by Si crystalline
boundaries as confirmed by the presence of oxygen
atom beside the Si peak of the EDX spectrum as shown
in Fig. 4(e) and (f).
Plots of PA phase signal as a function of modulation
frequency for p-PSi and n-PSi samples are presented in
Fig. 5. Since the measurements were carried out in the
frequency range of 30–100 Hz, the PA phase was indepen-
dent of the thickness of the porous layer due to the shorter
thermal diffusion length phenomenon. This has been
confirmed by the (Ihn)1/2 plot as a function of photon
energy as shown in Fig. 2(b), where at a chopping
frequency 433 Hz the excitation beam was totally on the
porous layer. Thus, it is confirmed that the PA technique
has an advantage compared with the convention optical

n-PSi technique. The decrease of PA signal phase is well

64 understood as the decrease of thermal velocity due to the
60 reduction of thermal diffusion length [5,10]. The thermal
and carrier transport parameters of these PSi samples were
56 obtained by fitting the experimental data of the PA phase
52
signal to the theoretical curve calculated using the equation
phase-chopping relationship, which is given by,
48  
18 21 24 27 30 33 36 39 p 1 ðaD=vÞðoteff þ 1Þ
f ¼ þ tan , (1)
Current Density (mA/cm2) 2 ðaD=vÞð1  oteff Þ  1  ðoteff Þ
Fig. 3. Relationship between current density and porosity for the PSi where teff=t[(D/a)1] and a=(pf/as)1/2 with D, v, a, teff
layer on p type and n-type Si substrates. and t corresponding to carrier diffusion coefficient, surface

Fig. 4. Typical scanning electron micrographs (SEM) at magnification of 2000  for (a) p-PSi (J ¼ 33.96 mA/cm2), (b) p-PSi (J ¼ 22.64 mA/cm2) and
(c) n-PSi (J ¼ 22.64 mA/cm2), and typical EDX spectra for: (d) Si substrate, (e) p-PSi and (f) n-PSi samples with J ¼ 28.29 mA/cm2.
 ARTICLE IN PRESS
C.K. Sheng et al. / Physica B 403 (2008) 2634–2638 2637

1.05 1.00
39.62 mA/cm2 39.62 mA/cm2
1.00 33.96 mA/cm2 0.96
33.96 mA/cm2
28.29 mA/cm2 28.29 mA/cm2

Phase (Rad.)

Phase (Rad.)
0.95 22.64 mA/cm2 0.92
22.64 mA/cm2
16.98 mA/cm2 16.98 mA/cm2
0.90 0.88

0.85 0.84

0.80 0.80

0.75 0.76
20 40 60 80 100 120 20 40 60 80 100 120
Modulation Frequency (Hz) Frequency (Hz)

Fig. 5. PA phase signal versus modulation frequency for (a) p-PSi and (b) n-PSi samples prepared at five different anodization current densities. The solid
lines represent the theoretical curves.

Surface Recombination Velocity (cm/s)

0.063 4.4 380 100
n-PSi

Recombination Lifetime (µs)

0.060
Diffusion Coefficient (cm2/s)
Thermal Diffusivity (cm2/s)

4.2 370 90

0.057 n-PSi p-PSi

360 80
4.0
0.054 n-PSi
p-PSi 350 70
3.8
0.051 340 p-PSi 60
p-PSi 3.6
0.048 330 50
0.045 3.4
320 n-PSi 40
0.042 3.2
310 30
50 55 60 65 70 75
50 55 60 65 70 75
Porosity (%)
Porosity (%)
Fig. 6. Dependence of thermal diffusivity and diffusion coefficient on
Fig. 7. Dependence of surface recombination velocity and recombination
current density of p-PSi samples.
lifetime on sample porosity.

recombination velocity, thermal diffusivity, effective re- 2.00

combination lifetime and bulk recombination lifetime,
respectively [7]. The solid lines in Fig. 5 show the best p-PSi
1.95
curves of Eq. (1) to the experimental data of PA phase
Energy Gap (eV)

signal. The corresponding thermal diffusivity and diffusion 1.90

coefficient as a function of layer porosity are shown in 1.85
Figs. 6 and 7. Both figures show that the thermal diffusivity
and diffusion coefficient decrease with the increasing 1.80
porosity of PSi layers. It is interesting to note that through
the present work the value of thermal diffusivity and 1.75 n-PSi
diffusion coefficients of n-PSi are always higher than that 1.70
of p-PSi samples. However, the surface recombination
velocity for p-PSi is found to be higher than the values 1.65
obtained for n-PSi in porosity ranging from 50% to 76%. 50 55 60 65 70 75 80
The same analysis was also applied for Si substrate. Porosity (%)
Regardless of the n-type or p-type of samples, the thermal Fig. 8. Dependence of energy gap on porosity for the p-PSi and n-PSi
diffusivity and diffusion coefficient of the Si substrate samples prepared at different anodization durations.
were 0.920 and 18.60 cm2/s1, respectively. In addition, the
surface recombination velocity (852 cm/s) is higher than the
PSi layer. For evaluating the band-gap energy of the PSi plot. Fig. 8 depicts the dependence of energy band gap on
layer, the optical absorption was plotted against the the porosity of p-PSi and n-PSi layers, where in both cases
photon energy. The value of the band-gap energy the band-gap energy increases with the increase of layer
was determined by extrapolating the linear part of the porosity.
 ARTICLE IN PRESS
2638 C.K. Sheng et al. / Physica B 403 (2008) 2634–2638
4. Discussion
The reduction of thermal diffusivity and diffusion
coefficient may be due to the thermal insulating effect of
the PSi layer. This behavior may come from the presence of
air inside the micropores. The sponge-like spherical or
columnar morphology of micropores may also reduce the
thermal transport and carrier diffusion process [11]. The
consistent reduction of thermal diffusivity and diffusion
coefficient values against sample porosity could also be due
to the enhancement of air and oxygen molecules impreg-
nated inside the enlarged micropores that possess higher
thermal resistance as compared to the Si substrate.
Low thermal transport feature of PSi allows this material
to be used as a new Si-based thermal insulator, which is
compatible with the recent semiconductor technology as
suggested by Bernini [11]. On the other hand, reducing air
and oxygen impregnated inside the smaller micropores at
lower porosity decreases the thermal diffusivity and
diffusion coefficient. This explains that the thermal
diffusivity and diffusion coefficient values of p-PSi samples
are lower than n-PSi samples in the porosity ranging from
50% to 76%. The rise of recombination lifetime as
compared to Si substrate indicates that a slower recombi-
nation process occurred in the p-PSi sample. This could be
due to the formation of a spherical crystallite boundary
surrounding the micropores, which also lead to the
widening of the band gap. Typical carrier recombination
lifetimes of PSi were estimated in the order of 10–100 ms, as
reported by Bernini [11]. The increase in the surface
recombination velocity as porosity increases might be
attributed to the reduction of crystallite boundary and
formation of thinner boundary within the micropores,
which resulted in a faster recombination process of p-PSi
samples [1,11]. However, in p-PSi samples the recombina-
tion rate is faster than that in the n-PSi samples, which
could be understood to be due to the effect of the increased
injection of electrons into the p-type material during the
anodization process, leading to a faster and efficient carrier
recombination process. The value of band-gap energy
determined from optical absorption of the PA signal shows
that the band-gap energy of p-PSi is always higher than
that of the n-PSi sample. In addition to that, the band-gap
energy for both increases linearly with the increasing layer
porosity.
5. Conclusion
The PA technique has been used to investigate the thermal
and carrier transport properties of PSi prepared on p-type
and n-type Si substrates within the porosity range of 50–76%.
The study shows that the thermal diffusivity and diffusion
coefficient decrease uniformly; the recombination process was
enhanced with higher surface recombination velocity and
shorter recombination lifetime with the increase of sample
porosity. The measured thermal diffusivity and diffusion
coefficient values of p-PSi samples were much lower than that
of n-PSi samples, this may be due to the thermal insulation of
air and oxygen molecules impregnated inside the enlarged
micropores of p-PSi. However, the p-PSi shows a faster
recombination rate when compared to n-PSi as indicated by a
reduction of recombination lifetime and the increase of
surface recombination velocity. For the same porosity value,
the energy band-gap values for p-PSi are always higher than
the energy-gap values obtained for the n-PSi layer.
Acknowledgments
The financial support by the Malaysian Government
through IRPA and NSF funding are acknowledged. The
authors would also like to acknowledge the Department of
Physics, Universiti Putra Malaysia, for providing the research
facilities for this research to be carried out successfully.
References
[1] L.T. Canham, Appl. Phys. Lett. 57 (1990) 1046.
[2] H.A. Lopez, P.M. Fauchet, Mater. Sci. Eng. B 81 (2001) 91.
[3] T. Toyoda, T. Takahashi, Q. Shen, J. Appl. Phys. 88 (2000) 6444.
[4] A. Rosencwaig, G. Gersho, J. Appl. Phys. 47 (1976) 64.
[5] A. Rosencwaig, Photoacoustic and Photoacoustic spectroscopy,
New York, 1980.
[6] A. Mandelis, Photoacoustic and Thermal Wave Phenomena in
Semiconductors, North-Holland, Amsterdam, 1987.
[7] A. Pinto Neto, H. Vargas, N.F. Leite, L.C.M. Miranda, Phys. Rev. B
40 (1990) 3924.
[8] C.A.S. Lima, M.B.S. Lima, L.C.M. Miranda, J. Baeza, J. Freer,
N. Reyes, J. Ruiz, M.D. Silva, Meas. Sci. Technol. 11 (2000) 504.
[9] J.B. Alvarado, M.L. Vargas, Anal. Sci. 17 (2001) 309.
[10] T. Toyoda, T. Hayakawa, K. Abe, T. Shigenari, Q. Shen, J. Lumin.
87–89 (2000) 1237.
[11] U. Bernini, S. Lettieri, P. Maddalena, R. Vitiello, G. Di Francia,
J. Phys.: Condens. Matter 13 (2001) 1141.