Random macropore formation in p-type silicon in HF-containing organic solutions: Host matrix for metal deposition

Article abstract of DOI:10.1149/1.1864292

Random macropore formation in variously doped p-type silicon substrates (0.01-20 Ω cm) by electrochemical anodization in aqueous HF-containing dimethylsulfoxide, dimethylformamide, or acetonitrile was investigated under different operating conditions. A systematic study of the effect of the main parameters controlling the macropore formation is reported. The results revealed that the nature of organic solvent, the concentration of HF, and the substrate doping level play major roles in determining pore formation and morphology. Also, the behavior of these macropores in contact with an aqueous solution of Ag₂SO₄ in the dark and at room temperature is investigated and a possible reaction mechanism for the deposition process is proposed.

Full text of DOI:10.1149/1.1864292

Journal of The Electrochemical Society, 152 ͑4͒ C213-C220 ͑2005͒
C213
0013-4651/2005/152͑4͒/C213/8/$7.00 © The Electrochemical Society, Inc.
Random Macropore Formation in p-Type Silicon
in HF-Containing Organic Solutions
Host Matrix for Metal Deposition
Farid A. Harraz,a,b Kentaro Kamada,a Katsutoshi Kobayashi,a Tetsuo Sakka,a,
*
,z
and Yukio H. Ogataa,
*
aInstitute of Advanced Energy, Kyoto University, Uji, Kyoto 611-0011, Japan
bAdvanced Materials Technology Department, Central Metallurgical Research and Development Institute,
Helwan, Cairo 11421, Egypt
Random macropore formation in variously doped p-type silicon substrates ͑0.01-20 ⍀ cm͒ by electrochemical anodization in
aqueous HF-containing dimethylsulfoxide, dimethylformamide, or acetonitrile was investigated under different operating condi-
tions. A systematic study of the effect of the main parameters controlling the macropore formation is reported. The results revealed
that the nature of organic solvent, the concentration of HF, and the substrate doping level play major roles in determining pore
formation and morphology. Also, the behavior of these macropores in contact with an aqueous solution of Ag2SO4 in the dark and
at room temperature is investigated and a possible reaction mechanism for the deposition process is proposed.
© 2005 The Electrochemical Society. ͓DOI: 10.1149/1.1864292͔ All rights reserved.
Manuscript submitted August 5, 2004; revised manuscript received September 6, 2004. Available electronically March 9, 2005.
The electrochemical etching of silicon in HF-based solutions is
known to form a porous layer on the surface.1,2 The dimensions of
these pores are tiny and mostly located in the nanometer-scale
range.3,4 Recently, the formation of macropores ͑pores in the
micrometer-size diameter range͒ has attracted increasing interest be-
cause of their promising features for a wide scope of applications
such as microelectromechanical systems ͑MEMS͒,5-7 sensors,8,9 and
biotechnology.10 Furthermore, Lehmann and Rönnebeck reported in
a recent work11 that macropores filled with lead can produce X-ray
filters. Lévy-Clément and co-workers12,13 extensively studied the
use of macropores prepared on p-type Si in organic solutions as
antireflection layers on polycrystalline solar cells. One of the promi-
nent applications of macropores is in the field of photonic bandgap
materials, so-called photonic crystals14,15 in which a periodic varia-
tion of the refractive index in the submicrometer range ͑a scale
matched to the wavelength of light͒ is fabricated. In the current
study, we show that p-type macropores can be also used as templates
for introduction of noble metal deep into the pores. This process can
produce novel composite structures from macroporous silicon.
The first macroporous structure prepared on n-type Si substrates
was observed by Theunissen in 197216 in aqueous HF solution and
was investigated in detail in the 1990s.17,18 However, Propst and
Kohl reported the first experiments of macropore formation on
p-type Si only in 1994.19 They used anhydrous HF in acetonitrile
and moderately doped p-type substrates. Later on, further studies
showed the possibility to obtain macropores on p-type Si either in
aqueous HF20 or aqueous HF-containing organic solutions.21
While macropore formation in n-type Si is well explained in
terms of the space charge region ͑SCR͒, which leads to a collection
of the photogenerated carriers at the pore tips,22 pore formation in
p-type electrodes is controversial. In contrast to n-type Si, p-type
substrate ͑in the anodic regime͒ is now under the forward bias con-
dition. Propst and Kohl,19 who first observed the macropores on
p-type Si, attributed their results to specific chemical processes.
Later, several mechanisms have been invoked to explain the
macropore formation in p-type Si. Of particular importance are the
depletion and field effects model,23 the chemical passivation
model,21 the linear stability model,24 the current burst model,25 and
the pure chemical model.26 In the depletion and field effects model,
charge-transfer mechanisms across the Schottky barrier are pro-
posed. The chemical passivation model explains macropore forma-
tion as a result of adsorption of organic molecules on the silicon
electrodes, which in turn changes the way of silicon dissolution. The
main idea of the linear stability model is that the formation of
macropores occurs only if a certain ratio of electrolyte and substrate
resistivities is achieved; however, the model is further modified with
new assumptions.27 The current burst model postulates that the
charge transfer at the Si/electrolyte interface is spatially and tempo-
rally inhomogeneous. Finally, the pure chemical model gives an
explanation based on chemical kinetics for pore initiation during the
anodic dissolution of n-type Si in HF solution, which can be applied
to p-type Si as well. According to this model, the silicon surface is
dissolved anodically to the Si͑I͒ intermediate which is very mobile
and acts as a catalyst for the dissolution reaction. However, none of
the above-mentioned models can explain all the experimental data
available in the literatures. Furthermore, a definite understanding of
the limits of pore formation and morphologies are not available for
p-type substrates.
Ordered pore arrays on Si are usually fabricated by applying a
multistage process including the photolithographic patterning and
subsequent alkaline etching.18,28 If the preparation of such well-
ordered pore alignments without pre-etching could be achieved, it
would be fascinating. Hence, basic understanding of random pore
formation is important.
Many organic solvents have been usually mixed with HF during
the macropore formation in p-type silicon. Of these, dimethylsulfox-
ide ͑DMSO͒, dimethylformamide ͑DMF͒, and acetonitrile ͑MeCN͒
are the most frequently investigated and are used in this study. Fur-
ther, this series of solvents in addition to the EtOH have different
physicochemical properties and hence, different effects during pore
formation are expected.
In this paper, a systematic study of the effect of the main param-
eters controlling the macropore formation, such as substrate doping
level, HF concentration, type of organic solvent, applied current
density, and anodization time, is reported. Then, the use of such
macropores as a “host matrix” for the impregnation of Ag metal by
a simple immersion plating technique in the dark and at room tem-
perature is presented.
Experimental
The Si͑100͒ single crystals used in this study were p-type boron-
doped with different resistivities: 0.01-0.02, 0.1-0.2, 3.2-4.8, and
10-20 ⍀ cm, corresponding to doping densities of 7 ϫ 1018, 1
ϫ 1017, 8 ϫ 1015, and 1 ϫ 1015 cm−3, respectively. All wafers were
mirror-polished ͑without etch pits͒ with 0.6 mm thickness, pre-
treated in acetone, and afterward dipped in 5 wt% HF to remove the
native oxide layer.
*
Electrochemical Society Active Member.
z E-mail: y-ogata@iae.kyoto-u.ac.jp
The electrochemical anodization was performed galvanostati-
Downloaded on 2014-06-21 to IP 157.182.150.22 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C214
Journal of The Electrochemical Society, 152 ͑4͒ C213-C220 ͑2005͒
cally in a cell made of trifluoroethylene resin using a two-electrode
setup with a Pt rod serving as a counter electrode. The exposed
surface area of the Si electrode was 0.785 cm2, adjusted by using an
O-ring at the bottom of the cell. The electrolyte was a mixture of
aqueous HF ͑47 wt%͒ and the following organic solvents: DMSO,
DMF, or MeCN. Different concentrations of HF were used from 2 to
10 M ͑3-17 wt%͒. Tetra-n-butylammonium perchlorate ͑TBAP͒ as
a supporting electrolyte was added in appropriate concentrations.
The current density was varied from 2 to 50 mA/cm2 for anodiza-
tion time from 4 min to 3 h.
Current-potential curves were measured using an automatic po-
larization system HZ-3000, Hokuto Denko. A three-electrode cell
was used with an Ag/AgCl electrode as a reference and a Pt wire as
a counter electrode. The ohmic contact on the rear side between Si
electrode and a Cu current collector was achieved by rubbing
Ga-In eutectic alloy.
Surface topography using a pin-type Tencor P-10 profilometer
was performed for some macroporous samples. Scan speed of
0.4 mm/s, horizontal resolution of 8 ␮m, vertical resolution of
0.2 nm, maximum vertical full scale of 150 ␮m, and stylus force of
5 mg were the main line scan parameters.
The macroporous layer formed in aqueous HF/DMSO solution
was immersed for 60 min in an aqueous solution of 5 ϫ 10−3
M
Ag2SO4 in the dark at room temperature. After the immersion, the
samples were rinsed in pure water and dried by blowing Ar gas. An
Elionix model ERA-800 field emission-type scanning electron mi-
croscope ͑FE-SEM͒ was used for structure evaluation. X-ray diffrac-
tion patterns ͑XRD͒ were measured on a Rigaku RAD 2C system
with an attachment for thin-film measurement. The radiation source
was Cu K␣͑1.54 Å͒ with a data acquisition rate of 1 deg/min.
Results and Discussion
Current-potential behavior.—Figure 1A shows the current-
potential behavior of a 10-20 ⍀ cm p-type Si at scan rate of
10 mV/s in solutions containing 4 M HF + 0.1 M TBAP in MeCN,
water, DMF, and DMSO. A typical current density-potential ͑i-E͒
response with a current peak is obtained in aqueous HF solution
͑curve b͒. The region at the initial rising part is related to porous
silicon layer formation, followed by a passivating region due to the
anodic oxidation of the substrate. At a potential higher than 6 V ͑vs.
Ag/AgCl͒, current oscillation is observed which is attributed to the
breakage and repair of the oxide layer in the presence of HF
solution.29 In the solution containing DMSO ͑curve d͒, the current
peak and oscillation are also observed but less pronounced and oc-
cur at a lower current value compared to the pure aqueous HF so-
lution. In HF/DMF solution ͑curve c͒ the current peak is shifted to a
higher potential and is reduced to a broad shoulder, which has a
higher current than the case of DMSO solution. We think that the
current peak is there but may be hidden due to either the high scan
rate or the high electrolyte resistivity. This is further clarified by
performing the same experiments at a lower scan rate of 1 mV/s,
and the results are depicted in Fig. 1B. The current peak in DMF
solution ͑curve c͒ is now obvious and the oscillation is also detected.
On the contrary, the i-E behavior in the MeCN-containing solu-
tion ͑Fig. 1A, curve a͒ is quite different. The current rises almost
linearly with the applied potential, exhibits a small hump at around
2 V ͑vs. Ag/AgCl͒, and then increases again up to 200 mA/cm2 ͑the
maximum current scale is not shown͒. The appearance of such a
hump is clear in curve a of Fig. 1B. At the beginning of the initial
part before the hump, a porous layer is formed. After the hump,
electropolishing of Si occurs. The hump position, current density
and potential, is determined by the kinetics at the interface. Diffu-
sion of reactants affects the current density at high potential but
within the porous formation region, and therefore a fast scan makes
the rate higher. Furthermore, the morphology may affect the trans-
port phenomena. Hydroxide and/or oxide formation is considered to
take part in the electropolishing process, although the formation
appears as an intermediate state. The balance of these above pro-
Figure 1. Current density-potential curves: ͑A͒ for a 10-20 ⍀ cm p-type Si
at scan rate of 10 mV/s in solutions containing 4 M HF + 0.1 M TBAP
mixed with ͑a͒ MeCN, ͑b͒ H2O, ͑c͒ DMF, and ͑d͒ DMSO. ͑B͒ is the same as
͑A͒, except that the scan rate was 1 mV/s and ͑curve b͒ is for EtOH solution.
cesses determines the hump position. Here a fast scan favors the
increase of the critical current density, as can be seen in curve a of
Fig. 1A compared to curve a of Fig. 1B.
In using HF in EtOH instead of water ͑curve b in Fig. 1B͒, the
current rises with the electrode potential up to 33 mA/cm2 and a
current peak typical for aqueous solution is not observed within this
potential range. This means that the porous silicon is still formed
and no passivation is detected under this condition in this ethanol-
based solution.
Macropore formation in different types of organic sol-
vents.—Macroporous structures in all samples under various condi-
tions ͑2-50 mA/cm2 and 2-10 M HF͒ were not observed either in
pure aqueous or ethanolic solutions containing HF, however, nano-
porous layers were formed instead. The presence of an organic sol-
vent in the electrolyte during the macropore formation in p-type Si
is essential in this work. In all electrolytes we observed the evolu-
tion of hydrogen gas during the anodization process. Top and cross-
sectional views of FE-SEM images after the galvanostatic treatment
Downloaded on 2014-06-21 to IP 157.182.150.22 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑4͒ C213-C220 ͑2005͒
C215
Figure 3. Topographic profiles of the macroporous silicon prepared in a 4 M
HF solution with different organic solvents. The current density and the
anodization time were 2 mA/cm2 and 60 min, respectively.
ing in the FE-SEM image of Fig. 2f. However in DMSO or DMF
solutions, an around 1 or 2 ␮m deep profile is observed, respec-
tively. This result implies that a nanoporous layer might also be
formed in DMSO or DMF, then chemically attacked and subse-
quently dissolved in both solutions; rough surfaces are observed in
FE-SEM images of Fig. 2b and d. It has also been demonstrated
experimentally that pure DMSO31 and both DMSO and DMF32
could react chemically with the nanoporous silicon. This assumption
is supported by the observation of progressive transition from a
homogeneous nanoporous layer to macropores during pore forma-
tion in organic solutions.24,27
Figure 2. Top ͑left͒ and cross-sectional view ͑right͒ FE-SEM micrographs of
macropores formed by an electrochemical etching in a 10-20 ⍀ cm p-type Si
at 2 mA/cm2 for 60 min in 4 M HF containing different organic solvents:
͑a, b͒ DMSO, ͑c, d͒ DMF, and ͑e, f͒ MeCN.
of the 10-20 ⍀ cm samples in a 4 M HF with DMSO, DMF, and
MeCN are shown in Fig. 2. The applied current density and the
anodization time were 2 mA/cm2 and 60 min, respectively. In the
HF/DMSO and HF/DMF solutions, round macropores approxi-
mately 1 ␮m in diameter and 7 ␮m in depth are formed ͑Fig. 2a-d͒.
The pores show almost the same depth and grow parallel to each
other. The parallel growth of pores is related to the passivating be-
havior of the pore walls.27 The images shown in Fig. 2a-d reveal
also that most of the pore volume obtained in both electrolytes is
empty. However, we show that a thin nanoporous layer is also there.
In contrast, a bilayer structure is formed in the HF/MeCN solu-
tion, which is considerably different from that obtained in HF/
DMSO or HF/DMF solutions. A thin nanoporous layer, about 1 ␮m
thick, is formed at the top of the surface with a macroporous layer
beneath it ͑see Fig. 2e and f͒. This nanoporous layer cracks upon
drying, revealing macropores below it, as can be seen in the top
view image. A similar kind of nanoporous layer formation during
the macropore formation on a moderately doped p-type substrate
using HF/MeCN21 or HF/MeCN containing diethyleneglycol
solutions30 has been reported.
Based on this observation, the solvent type and effect can be
classified into three categories: (i) DMSO and DMF favor
macropore formation, (ii) MeCN gives
a bilayer of nano/
macropores, and (iii) aqueous and ethanolic solutions lead to the
formation of nanopores. Now a question arises to understand the
different behavior of these solvents during the pore formation. Ac-
cording to Lust and Levy-Clement,33,34 the different effects are at-
tributed to the different solution chemistry, which in turn affects and
controls the rate of chemical and electrochemical reactions taking
place during the anodic dissolution of silicon in HF solution. They
considered the different protic character ͑acid-base properties͒ of the
solvents to account for the different behavior of water and ethanol
on one side and DMSO, DMF, and MeCN on the other side. Water
and ethanol are protic, i.e., strong hydrogen-bond donors, and hence,
anions are much more stable there. DMSO, DMF, and MeCN are
aprotic solvents as they exchange protons very slowly. This means
that the anions, e.g., F− and HF2−, are very mobile in these solvents
and consequently enhance the silicon dissolution rate. The different
effects between DMSO and DMF as one group and MeCN as an-
other group are related to their different donor or solvating proper-
ties. As it has been also proposed in Ref. 33 and 34, the oxidizing
properties, the redox potential, the build-in potential, and the solu-
bility of the reaction products may have effects on the different
behaviors of the solvents.
In the cross-sectional view of Fig. 2f, the macroporous layer
depth obtained in MeCN solution is approximately 11 ␮m and is
mainly filled with nanopores. Above the dotted line in the image,
empty macropores are observed, whereas filled macropores are still
there under the line. In addition, the pores are intermittent, probably
because the pore growth direction is not perfectly perpendicular to
the surface. This tortuous way of growth might enhance the ten-
dency to branching in this solution.
The topographic profiles of the macropores formed in these or-
ganic solvents are shown in Fig. 3. The surface profiles were found
to depend on the solvent used. In the MeCN solution, the surface of
the macroporous layer has almost the same level as that of the un-
treated wafer. This is in consistent with the smooth surface appear-
As mentioned previously, we could not obtain macropores in HF
solutions with either water or ethanol. However, Lehmann and
Rönnebeck23 reported that the presence of an organic solvent in the
electrolyte is not essential for macropore formation on p-type Si.
They observed macropores in aqueous HF solution for substrate
Downloaded on 2014-06-21 to IP 157.182.150.22 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C216
Journal of The Electrochemical Society, 152 ͑4͒ C213-C220 ͑2005͒
Figure 5. Top-view FE-SEM micrographs of macropores obtained in 4 M
HF/DMSO at 2 mA/cm2 for 60 min using p-type Si of different resistivities:
͑a͒ 10-20, ͑b͒ 3.2-4.8, ͑c͒ 0.1-0.2, and ͑d͒ 0.01-0.02 ⍀ cm.
current density and hence the electropolishing regime is never
reached. Under identical conditions, better pore formation and mor-
phology are observed in DMSO- and DMF-based solutions in com-
parison to the MeCN-containing HF.
Substrate doping density.—The influence of the doping level
͑0.01-20 ⍀ cm,7 ϫ 1018 to 1015 cm−3͒ on the size and density of
the pores in a 4 M HF/DMSO solution at 2 mA/cm2 for 60 min was
investigated. The resulting morphologies are shown in Fig. 5. It can
be seen that the pore diameter and the pore spacing decrease,
whereas the pore number density increases with decreasing the sub-
strate resistivity from 10-20 to 3.2-4.8 ⍀ cm ͑Fig. 5a and b͒. Fur-
thermore, the thickness of the macropore walls is greatly reduced.
The average pore wall thickness is about 800 nm in case of
10-20 ⍀ cm, reduced to around 150 nm on 3.2-4.8 ⍀ cm substrate.
In general, the wall thickness is found to be smaller than the pore
diameter at a specific resisitivity. Square root dependence for the
pore wall thickness with substrate doping density is reported.23 Ran-
dom pore distribution and dimensions and poorly defined pore mor-
phology are obtained when the resistivity is reduced to 0.1
-0.2 ⍀ cm, as shown in Fig. 5c. Meanwhile, if the substrate resis-
tivity is reduced further to 0.01-0.02 ⍀ cm, only a homogeneous
nanoporous layer is formed, as shown in Fig. 5d.
From these observations, it appears that there is a critical value
of substrate resistivity ͑0.1-0.2 ⍀ cm in this study͒ below which no
macropore can be formed. This phenomenon can be explained as a
result of diminution of the passivating properties of the pore walls
when the doping density exceeds a critical level, consistent with the
improved version of the linear stability model.27 Similar limiting
value of resistivity was reported by another group in electrolytes
containing organic solvents.33 However, a limiting resistivity of
around 5 ⍀ cm was found in aqueous solution.23 The disappearance
of macroporous structures on polished samples at a limiting resis-
tivity depends not only on the electrolyte composition but also on
the nature of the organic solvent.34
Figure 4. Top-view FE-SEM images of macropores formed in
10-20 ⍀ cm p-type Si at 10 mA/cm2 for 20 min in a 4 M HF/DMSO solu-
tion: ͑a͒ as-anodized sample and ͑b͒ after immersion in 0.1 M KOH for
30 min.
resistivity above 5 ⍀ cm. They used prepatterned substrates with
etch pits; however, our samples are mirror-polished without etch
pits. The presence of such etch pits certainly favors Si dissolution
and pore formation.
If the macropores are immersed in an alkaline solution of KOH,
the nanoporous body is dissolved with the evolution of hydrogen
gas. Figure 4a and b, for example, shows top-view FE-SEM micro-
graphs of macropores prepared in HF/DMSO solution before and
after immersion in 0.1 M KOH for 30 min, respectively. After the
immersion, the round macropores are found to convert to square
shape and the average pore sizes widened nearly twice. In this so-
lution, chemical etching occurs, which is an orientation-dependent
process.35 The overall etch rate of the sample under this condition is
0.077 ␮m/min. The ͑111͒ plane is the most resistant to etching com-
pared to the other crystal orientations.35 It seems that during the
etching the ͑100͒ plane starts to disappear gradually, leaving the
͑111͒ crystal faces behind, and this is the reason for the morphology
changes.
Based on these results, it is clear that high doping density sub-
strates tend to form nanoporous layers rather than macropores.
The influence of increasing the doping density on pore formation is
similar to some extent to the influence of increasing the applied
potential.
Based on this investigation, we conclude that the pore morphol-
ogy and orientation are strongly dependent on the type of organic
solvent added to the electrolyte. The presence of the nanoporous
body with the macroporous structure in the MeCN solution indicates
that the current density at the macropore tip is below the critical
Current density and anodization time.—The effect of applied
current density and anodization time on macropore formation and
pore size were investigated. At constant current density the pore size
increases with anodization time ͑time varied from 20 min up to 3 h͒.
Downloaded on 2014-06-21 to IP 157.182.150.22 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑4͒ C213-C220 ͑2005͒
C217
Figure 6. FE-SEM micrographs of macropores formed in a 10-20 ⍀ cm
p-type Si in 4 M HF/DMSO at different current densities and various times:
͑a͒ 2 mA/cm2 for 100 min, ͑b͒ 5 mA/cm2 for 40 min, ͑c͒ 10 mA/cm2 for
20 min, and ͑d͒ 50 mA/cm2 for 4 min. The number of coulombs was
12 C/cm2 in all cases.
As the anodization time increases, the interpore spacing decreases
until the neighboring pores joined together, forming wider pores.
The phenomenon of pore joining has been proposed by Lehmann
and Föll.36
Figure 6 shows FE-SEM micrographs of the substrate anodized
in a 4 M HF/DMSO at different current densities and various times.
The number of coulombs was kept constant at 12 C/cm2 in all cases.
It can be observed that the pore size has no significant dependence
on current density ͑it decreases slightly with increasing current den-
sity͒. The variation of pore growth rate ͑pore depth per time͒ with
current density at the same quantity of charge was evaluated and the
result is depicted in Fig. 7. The growth rate shows a linear depen-
dence on current density. This result is in contrast with the case of
Figure 8. Top ͑left͒ and cross-sectional view ͑right͒ FE-SEM micrographs of
macropores formed in 10-20 ⍀ cm p-type Si at 10 mA/cm2 for 20 min in a
DMSO solution with different HF concentrations: ͑a, b͒ 2 M, ͑c, d͒ 4 M, and
͑e, f͒ 10 M HF.
macropores prepared on n-type substrates, in which the growth rate
is independent of current density.23 Such increase with current den-
sity, however, is similar to the behavior of microporous silicon. At
the same current density the growth rate is about two times greater
than the growth rate of microporous structure, in agreement with the
data reported in Ref. 23.
However, Vyatkin and co-workers37 observed saturation in the
growth rate of p-type macropores after a linear dependence with the
current density. They used two different electrolytes of HF/DMF and
HF/H2O/͑CH3͒2CHCO ͑10-20 ⍀ cm substrates͒ with a constant an-
odization time. They considered the leveling in growth rate as
evidence that the charge-transfer process at the electrolyte/
semiconductor interface is not anymore a limiting process for pore
formation after exceeding a specific value of current density. Fur-
ther, in this range of current density the etching rate should be con-
trolled by a chemical reaction. Their observation is not contradictory
to ours. Both results confirm that the growth rate is now not only
dependent on the current density but also on the chemical composi-
tion of the electrolyte. This is further emphasized in the next section.
Variation of HF concentration.—The effect of changing HF
concentration on pore morphology and depth was examined. Be-
cause the HF is prepared from an original 47 wt% HF solution, the
electrolytes always include a certain amount of water ͑1 M HF so-
lution contains about 1.1 M H2O͒. Top and cross-sectional view
FE-SEM micrographs of macropores obtained at constant current
density and time in a DMSO solution with 2, 4, and 10 M HF are
shown in Fig. 8. At first sight it may seem that the resulting
macropore of top-view images in Fig. 8a, c, and e have quite similar
pore morphologies, and one may conclude that variation of HF con-
centration has no effect on macropore formation. On the contrary,
drastic changes in both pore morphology and pore depth are clearly
observed from cross-sectional views in Fig. 8b, d, and f. Unstable
pore growth with the smallest growth rate is obtained in the solution
Figure 7. Pore growth rate as a function of applied current density for
macropores obtained in 10-20 ⍀ cm p-type Si substrates in a 4 M HF/
DMSO solution.
Downloaded on 2014-06-21 to IP 157.182.150.22 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C218
Journal of The Electrochemical Society, 152 ͑4͒ C213-C220 ͑2005͒
Figure 9. Topographic profiles of the macroporous structures prepared at
10 mA/cm2 for 20 min in a DMSO solution mixed with different HF con-
centrations.
with the 2 M HF. The observed pore depth after 20 min anodization
at 10 mA/cm2 is only 3.5 ␮m. At this low concentration of HF, the
etching of pore walls is probably enhanced, which in turn gives a
small aspect ratio ͑pore depth per diameter͒. Increasing the concen-
tration of HF to 4 M improves the quality and stability of pore
growth; the pore growth rate is greatly enhanced ͑a pore depth of
about 13 ␮m is obtained, Fig. 8d͒. When the HF concentration is
further increased to 10 M, the reaction rate is a bit increased; the
pore depth is developed to around 15 ␮m ͑Fig. 8f͒. In both 4 and
10 M HF solutions, most of the pores are now empty.
Surface profiles of macropores prepared in different HF concen-
trations were performed. The profiles depicted in Fig. 9 depend on
the HF concentration. In solution containing 2 M HF, the profile
shows an average of a 3 ␮m layer that etched away from the sur-
face. The curvature of the profile is likely induced by stress.38 The
dissolution is greatly enhanced at the border of the macroporous
area with the wafer; a similar result was observed for mesoporous
structures in an earlier report.39 However, when the HF concentra-
tion is increased to 4 or 10 M, mild dissolution occurs; around
0.5-0.9 ␮m deep profiles are observed in both solutions. This means
that 2 M HF content is the most aggressive solution.
Based on these results, it seems that there is a threshold concen-
tration of HF below which unstable pore growth and very small
aspect ratio are obtained. Under our own experimental conditions,
the optimum concentration for stable macropore formation should
not be below 4 M. The observed dependence of pore growth rate on
HF concentration can be ascribed to different solution aggressive-
ness. There is general agreement that the HF concentration in the
electrolyte determines the critical current density of porous layer
formation ͑JPSL͒. This means that the effects of changing HF and
current density on reaction rate resemble each other. This confirms
again that the electrochemical reaction associated with the pore for-
mation is not only dependent on current density and semiconductor
properties but also on the electrolyte composition.
Figure 10. FE-SEM images of macropores after immersion plating in 5
ϫ 10−3 M Ag2SO4 for 60 min in the dark and at room temperature: ͑a͒
top-view, and ͑b͒ cross-sectional view, and ͑c͒ cross-sectional view at a
higher magnification. The macropores are formed in a 10-20 ⍀ cm p-type Si
at 10 mA/cm2 for 10 min in a solution composed of 4 M HF/DMSO.
The behavior of macropores
Impregnation with Ag metal.—
formed using 10-20 ⍀ cm substrate in a DMSO solution in contact
with Ag+ ions ͑at OCP, 60 min, in the dark, room temperature͒ is
investigated. Figure 10 shows top and cross-sectional view FE-SEM
micrographs obtained after immersion in a 5 ϫ 10−3 M Ag2SO4
aqueous solution. As shown in the images, Ag particles with a size
in the range of 50 nm are clearly observed on the entire surface of
macropores ͑both on the top surface and deep inside the pores͒. This
result indicates that there is no great problem of mass transport
Downloaded on 2014-06-21 to IP 157.182.150.22 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑4͒ C213-C220 ͑2005͒
C219
Figure 11. Schematic representation of energy band diagram for a p-type Si
͑in the dark͒ in contact with 5 ϫ 10−3 M Ag2SO4 solution, illustrating the
energetic position of the bandedges with respect to the equilibrium potential
of the redox couple.
Figure 12. XRD patterns for macropores prepared in 10-20 ⍀ cm p-type Si
substrates after immersion in 5 ϫ 10−3 M Ag2SO4 in the dark at different
times: ͑a͒ 0, ͑b͒ 1, ͑c͒ 2, ͑d͒ 5, ͑e͒ 20, and ͑f͒ 60 min.
inside the pores, which may limit the deposition process. In other
words, Ag+ ions easily diffuse into the pores and thereby the reduc-
tion immediately takes place. According to this observation, it is
concluded that the macropore prepared in p-type Si can be used as a
host matrix to introduce noble metals deep into the pores by impreg-
nating the pores in an appropriate immersion plating solution.
Now a question arises to understand the reaction mechanism in-
volved during the deposition process. We used a p-type electrode
and the deposition is observed at the open-circuit potential ͑OCP͒ in
the dark. This means that the reaction mechanism most likely pro-
ceeds by hole injection from Ag+ ion solution to the valence band of
the silicon. To clarify the possibility of this mechanism, the energy
band diagram for a p-type Si in contact with the Ag+-containing
solution is determined. Our calculations are based on the following
assumptions: (i) the redox potential of Ag+/Ag is calculated as
0.47 V vs. Ag/AgCl for 5 ϫ 10−3 M metal ions in solution; (ii)
under equilibrium conditions, the Fermi level in the semiconductor
is represented by the electrode potential ͑OCP͒,40 which was mea-
sured in the dark, and the value was taken during the initial stage of
immersion ͑0.17 V vs. Ag/AgCl͒; (iii) for a p-Si ͑doping density
1015 cm−3͒ the Fermi level ͑EF͒ is located 0.24 V above the position
of the valence bandedge ͑EV͒; i.e., EF − EV = 0.24 V;41 (iv) the Si
bandedges are fixed.
Oxidation:
Si + 2H2O + 4h+ → SiO2 + 4H+
As a last comment, the crystallinity of Ag deposits on the
macroporous surface is analyzed with XRD at different deposition
times. The result is shown in Fig. 12. The patterns exhibit two peaks
at 38.2° and 44.4°, corresponding, respectively, to Ag͑111͒ and
Ag͑200͒. The peak intensity for both phases increases with immer-
sion time ͑Fig. 10b-f͒. Assuming that the Ag particles are stress-free,
the average crystallite size can be estimated by applying the Scher-
rer equation42
D = C␭/␤ cos ␪
where ␭ is the X-ray wavelength ͑nm͒, ␤ is the full width at half
maximum ͑fwhm͒ of the peak ͑radians͒, ␪ is Bragg angle, C is a
factor ͑typically from 0.9 to 1.0͒ depending on crystallite shape, and
D is the crystallite size ͑nm͒. For the Ag͑111͒ peak of Fig. 12f,
average particle size was found to be 42 nm and for the ͑200͒ peak
the value was 43 nm. These values are close and in good agreement
with the average size detected from FE-SEM images in Fig. 10.
From this result, it is concluded that macroporous structure can
be used as a host matrix for metal deposition into the pores. For any
metal/metal ion redox couple with a more positive equilibrium po-
tential than the valence bandedge, the deposition can take place in
the dark by a simple immersion plating method.
The resulting simplified band model for the reduction of Ag+
ions at the Si surface is illustrated in Fig. 11 at its flatband condition.
As can be seen, the Ag redox level has a good overlap with the
valence bandedge of Si; it has a more positive potential than the
position of the valence bandedge. Consequently, hole injection from
the ions in solution to the valence band of Si is energetically favor-
able, and deposition does occur at the OCP, as is already observed
experimentally. Hole injection leads to oxidation of the macroporous
substrate. Surface oxidation is indeed observed and detected by Fou-
rier transform infrared ͑FTIR͒ spectra ͑not shown here͒. Accord-
ingly, the deposition of Ag from Ag2SO4 can be represented by the
following redox couple
Conclusions
To better understand the role of different factors affecting the
macropore formation in p-type Si substrates, we have performed a
series of experimental investigations in HF-containing organic elec-
trolytes. The results led to some important conclusions:
1. The presence of an organic solvent plays an essential role for
better pore formation.
Reduction:
Ag+ → Ag + h+
Downloaded on 2014-06-21 to IP 157.182.150.22 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C220
Journal of The Electrochemical Society, 152 ͑4͒ C213-C220 ͑2005͒
Mariasaldi, and F. Cavani, Sens. Actuators, A, 74, 1 ͑1999͒.
9. C. M. A. Ashruf, P. J. French, P. M. Sarro, R. Kazinczi, X. H. Xia, and J. J. Kelly,
J. Micromech. Microeng., 10, 505 ͑2000͒.
2. A nanoporous layer is observed in the MeCN-containing so-
lution. In case of DMSO or DMF solutions, this nanoporous layer is
most likely formed and consequently etched away.
3. Filled macroporous structures obtained with nanoporous body
in the MeCN solution indicate that the current density at the pore tip
is far from the electropolishing regime.
4. Macropores are observed for a limiting substrate resistivity of
0.1-0.2 ⍀ cm. For higher doping density, only nanoporous layers
are formed due to a loss of the passivating character of the pore
walls.
5. The content of HF in solution is critical for the dissolution
process. An appropriate concentration should be added to control the
pore morphology and the growth velocity.
6. The pores grow perpendicularly on the substrate surface and
follow the ͗100͘ direction. The macropore growth rate has a linear
dependence on current density, and the value is found to be about
two times greater than that of the micoporous silicon.
7. The pore size and spacing decrease with decreasing the sub-
strate resistivity, however they have insignificant dependence on
current density.
8. Finally, as-prepared macroporous structure has been success-
fully used as a host matrix for Ag deposition at the OCP without
using any catalytic treatment or external bias. The crystallinity of Ag
deposits is confirmed and the reaction mechanism is clarified. Gen-
erally, for any metal/metal ion couples with energy levels that over-
lap with the valence bandedge, open-circuit deposition is expected
to occur in the dark via injection of holes from the metal ion solu-
tion to the valence band.
10. V. Lehmann, Electrochemistry of Silicon, p. 223, Wiley-VCH, Weinheim ͑2002͒.
11. V. Lehmann and S. Rönnebeck, Sens. Actuators, A, 95, 202 ͑2001͒.
12. E. A. Ponomarev and C. Lévy-Clément, J. Porous Mater., 7, 51 ͑2000͒.
13. C. Lévy-Clément, S. Lust, S. Bastide, Q. N. Lê, and D. Sarti, Phys. Status Solidi A,
197, 27 ͑2003͒.
14. U. Grüning, S. Ottow, and V. Lehmann, Appl. Phys. Lett., 68, 747 ͑1996͒.
15. S. R. Nicewarner-Pena, R. G. Freeman, B. D. Reiss, L. He, D. J. Pena, I. D. Walton,
R. Cromer, C. D. Keating, and M. J. Natan, Science, 294, 137 ͑2001͒.
16. M. J. J. Theunissen, J. Electrochem. Soc., 119, 351 ͑1972͒.
17. V. Lehmann, J. Electrochem. Soc., 140, 2836 ͑1993͒.
18. V. Lehmann and U. Grüning, Thin Solid Films, 297, 13 ͑1997͒.
19. E. K. Propst and P. A. Kohl, J. Electrochem. Soc., 141, 1006 ͑1994͒.
20. R. B. Wehrspohn, J.-N. Chazalviel, F. Ozanam, and I. Solomon, Thin Solid Films,
297, 5 ͑1997͒.
21. E. A. Ponomarev and C. Lévy-Clément, Electrochem. Solid-State Lett., 1, 42
͑1998͒.
22. P. C. Searson, J. M. Macaulay, and F. M. Ross, J. Appl. Phys., 72, 253 ͑1992͒.
23. V. Lehmann and S. Rönnebeck, J. Electrochem. Soc., 146, 2968 ͑1999͒.
24. R. B. Wehrspohn, J.-N. Chazalviel, and F. Ozanam, J. Electrochem. Soc., 145, 2958
͑1998͒.
25. J. Carstensen, M. Christophersen, and H. Föll, Mater. Sci. Eng., B, 69-70, 23
͑2000͒.
26. E. S. Kooij and D. Vanmaekelbergh, J. Electrochem. Soc., 144, 1296 ͑1997͒.
27. J.-N. Chazalviel, F. Ozanam, N. Gabouze, S. Fellah, and R. B. Wehrspohn, J.
Electrochem. Soc., 149, C511 ͑2002͒.
28. F. A. Harraz, K. Kamada, J. Sasano, S. Izuo, T. Sakka, and Y. H. Ogata, Phys. Status
Solidi A, In Press.
Acknowledgments
29. J. Carstensen, R. Prange, and H. Föll, J. Electrochem. Soc., 146, 1134 ͑1999͒.
30. M. Christophersen, J. Carstensen, S. Rönnebeck, C. Jäger, W. Jäger, and H. Föll, J.
Electrochem. Soc., 148, E267 ͑2001͒.
The authors thank D. Hamm of Kyoto University for valuable
discussion and Y. Mizutani and T. Nagayama of Kyoto Municipal
Institute for Industrial Research for their support during FE-SEM
observation. This research was partly supported by a Grant-in-Aid
from the Ministry of Education, Culture, Sports, Science and Tech-
nology. F.A.H. acknowledges support from the Japan Society for the
Promotion of Science in the form of a JSPS postdoctoral fellowship.
31. J. H. Song and M. J. Sailor, Inorg. Chem., 37, 3355 ͑1998͒.
32. F. A. Harraz, T. Sakka, and Y. H. Ogata, Electrochim. Acta, 46, 2805 ͑2001͒.
33. S. Lust and C. Lévy-Clément, J. Electrochem. Soc., 149, C338 ͑2002͒.
34. S. Lust and C. Lévy-Clément, Phys. Status Solidi A, 182, 17 ͑2000͒.
35. S. M. Sze, Semiconductor Devices: Physics and Technology, 2nd ed., p. 428, Wiley
& Sons, Inc., New York ͑2002͒.
Kyoto University assisted in meeting the publication costs of this article.
36. V. Lehmann and H. Föll, J. Electrochem. Soc., 137, 653 ͑1990͒.
37. A. Vyatkin, V. Starkov, V. Tzeitlin, H. Presting, J. Konle, and U. König, J.
Electrochem. Soc., 149, G70 ͑2002͒.
References
1. A. Uhlir, Bell Syst. Tech. J., 35, 333 ͑1956͒.
38. J. M. Lopez-Villegas, M. Navarro, D. Papadimitriou, J. Basses, and J. Samitier,
Thin Solid Films, 276, 238 ͑1996͒.
2. D. R. Turner, J. Electrochem. Soc., 105, 402 ͑1958͒.
3. R. L. Smith and S. D. Collins, J. Appl. Phys., 71, R1 ͑1992͒.
4. A. G. Cullis, L. T. Canham, and P. D. J. Calcott, J. Appl. Phys., 82, 909 ͑1997͒.
5. V. Lehmann, W. Honlein, R. Reisinger, A. Spitzer, H. Wendt, and J. Willer, Thin
Solid Films, 276, 138 ͑1996͒.
39. D. Hamm, J. Sasano, T. Sakka, and Y. H. Ogata, J. Electrochem. Soc., 149, C331
͑2002͒.
40. R. Memming, Semiconductor Electrochemistry, p. 104, Wiley, Weinheim ͑2001͒.
41. S. M. Sze, Physics of Semiconductor Devices, 2nd ed., p. 26, Wiley, New York
͑1981͒.
6. S. Ottow, V. Lehmann, and H. Föll, J. Electrochem. Soc., 143, 385 ͑1996͒.
7. H. Ohji, P. T. J. Gennissen, P. J. French, and K. Tsutsumi, J. Micromech. Microeng.,
10, 440 ͑2000͒.
42. H. P. Klug and L. E. Alexander, X-Ray Diffraction Procedures for Polycrystalline
and Amorphous Materials, p. 540, Wiley-Interscience, New York ͑1974͒.
8. R. Angelucci, A. Poggi, L. Dori, G. C. Cadinali, A. Parisini, A. Tagliani, M.
Downloaded on 2014-06-21 to IP 157.182.150.22 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).