www.afm-journal.

de

FULL PAPER
Silicon Inverse-Opal-Based Macroporous Materials as
Negative Electrodes for Lithium Ion Batteries
By Alexei Esmanski and Geoffrey A. Ozin*

materials based on colloidal crystal tem-

Several types of silicon-based inverse-opal films are synthesized, plating have been synthesized.[2–6]
In recent years it has become widely
characterized by a range of experimental techniques, and studied in terms of
recognized that inverse-opal architectures
electrochemical performance. Amorphous silicon inverse opals are fabricated may benefit a wide range of applications
via chemical vapor deposition. Galvanostatic cycling demonstrates that these beyond the field of photonic crystals. The
materials possess high capacities and reasonable capacity retentions. nanoscale structure of the inverse-opal
Amorphous silicon inverse opals perform unsatisfactorily at high rates due to skeleton provides a large surface area,
the low conductivity of silicon. The conductivity of silicon inverse opals can be short diffusion lengths, easy access for
and fast transport of reagents inside the
improved by their crystallization. Nanocrystalline silicon inverse opals
porous structure. The ‘‘non-photonic’’
demonstrate much better rate capabilities but the capacities fade to zero after applications of inverse colloidal crystals
several cycles. Silicon–carbon composite inverse-opal materials are include such diverse fields as chemical
synthesized by depositing a thin layer of carbon via pyrolysis of a sucrose- and biological sensing,[7–9] simultaneous
based precursor onto the silicon inverse opals. The amount of carbon chromatographic separation and optical
sensing,[10,11] biomedical research,[12,13]
deposited proves to be insufficient to stabilize the structures and silicon–
catalysis,[14,15] and power storage and
carbon composites demonstrate unsatisfactory electrochemical behavior. generation.[16,17]
Carbon inverse opals are coated with amorphous silicon producing another Nowadays the demand for light-weight
type of macroporous composite. These electrodes demonstrate significant electric power sources that can provide
improvement both in capacity retentions and in rate capabilities. The inner high energy and high power density is
carbon matrix not only increases the material conductivity but also results in greater than ever before, due to the
exponential growth in the number of
lower silicon pulverization during cycling.
portable electronic devices and cordless
tools, and to increasing interest in electric
and hybrid powered vehicles. Lithium ion cells have become the
1. Introduction dominant battery type for many of these applications,[18] as they
offer specific energies higher than other electrochemical power
Three-dimensional ordered arrays of monodisperse colloidal sources. Lithium ion batteries employ lithium intercalation or
particles have generated intense interest, mainly due to the lithium insertion compounds both as the positive and negative
special optical properties of such materials and their possible electrode materials. As a battery is cycled, lithium ions exchange
applications in photonics. These materials are usually referred to between anode and cathode. The technology of lithium ion
as (artificial) opals because they mimic the structure and batteries is still a relatively young yet fast growing field, and we
iridescence of natural opals. A new synthetic paradigm appeared believe that improvements in the battery performance are on the
in the late 1990s when it was recognized that colloidal crystals near horizon. These improvements will probably be achieved
could be utilized as templates for the preparation of three- through the introduction of new battery materials and through
dimensionally ordered porous replicas, so called ‘inverse opals’. the employment of new material architectures.
After opal assembly, the voids between spheres are filled with the In the early 1990s, Sony Corporation introduced to the market
desired material, and the template is then removed. Following the lithium ion batteries with carbon-based negative electrodes,[19]
original work by Velev et al.[1] a wide variety of macroporous and nowadays carbons are still ahead of other negative electrode
materials in terms of their practical utility. However, a new type of
negative electrode, based on lithium alloying with various metals
[*] Prof. G. A. Ozin, Dr. A. Esmanski
Materials Chemistry Research Group
and metalloids, is emerging.[20] The major advantage of such
Chemistry Department materials in comparison with carbons is that alloys provide
University of Toronto significantly higher specific capacities. Silicon negative electrodes
80 St. George St. show by far the highest capacities towards lithium of all materials
Toronto, Ontario, M5S 3H6 (Canada) that have been studied as negative electrodes for lithium ion cells.
E-mail: gozin@chem.utoronto.ca
High capacity, attractive voltage profile, abundance, low cost and
DOI: 10.1002/adfm.200900306 environmental friendliness makes silicon a very promising

Adv. Funct. Mater. 2009, 19, 1999–2010 ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1999

negative electrode material. The use of silicon electrodes could
FULL PAPER
potentially allow for simple incorporation of microbattery
production into the industrial semiconductor manufacturing
sector. The main challenge of silicon implementation in
commercial batteries is that the material experiences up to 4 fold
volume changes during lithium alloying–dealloying, which leads
to its mechanical disintegration during repeated cycling and loss
of electronic contact with the current collector or conductive
additives.
Powder X-ray diffraction (PXRD) studies of Li–Si alloys at
different degrees of lithiation demonstrate that the crystal
structure of silicon is destroyed during lithium insertion and
an amorphous lithium silicide structure is formed.[21,22] At
potentials below 50–70 mV vs. Li/Liþ a crystalline Li15Si4
compound is formed. This composition corresponds to a
theoretical capacity of 3 579 mA h g
1. During extraction, Li15Si4
is converted to amorphous lithiated silicon, which in turn
converts to amorphous silicon upon complete lithium extraction.
Limiting the lithium insertion voltage to above 50 mV vs. Li/Liþ
should improve cycling performance of silicon-based materials by
preventing the formation of two phase regions, which induce
high internal stresses, pulverization of electrode material, and
capacity fading.
Periodic macroporous nanostructured materials fabricated in
the form of inverse-opal films provide several advantages when
used as lithium ion battery electrodes. Firstly, lithium solid-state
diffusion distances are kept short since the wall thicknesses are
on the order of nanometers to tens of nanometers, which
improves rate capabilities of the electrodes. Secondly, the large
surface area decreases the current density, again increasing rate
performance. Thirdly, macropores with a size range of several
hundred nanometers enable easy infiltration of electrolyte and
fast liquid-phase lithium diffusion, reducing the concentration
polarization and increasing rate performance and capacity of the
cell. Fourthly, the continuous network of electrode material may
provide better electrical conductivity than aggregates of loosely
connected particles. Fifthly, the use of film electrodes eliminates
the need for binders and conductive additives that constitute up to
20 wt% of ‘‘normal’’ electrodes, and so provides increased specific
capacity. Finally, and probably most importantly, macroporosity
should help to accommodate volume swings during cycling
without losing the structural integrity of the electrode. Periodic
macroporous silicon structures seem to be a promising material
architecture for fundamental scientific explorations of a new
genre of battery electrode materials.
Electrochemical behavior of inverse-opal lithium ion battery
electrode materials has been studied by several research groups in
recent years. In 2001 Take et al. reported the use of carbon
inverse-opal as a negative electrode; however, only very limited
electrochemical data was provided at the time.[23] Shortly after, the
synthesis and electrochemical behavior of vanadia gel inverse-
opal positive electrodes was demonstrated.[17] The study showed
excellent behavior of the ordered macroporous structures at high
current rates. Since then several more electroactive materials have
been synthesized in the inverse-opal form and explored as both
positive[24–26] and negative electrodes.[27–42] In most cases the use
of inverse opals led to an improvement in rate capabilities and
often to better cycling behavior of the materials. Composite
materials based on inverse-opal structures have been also
2000 ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.afm-journal.de
suggested to be used as lithium ion battery electrodes.[26–29,43]
Carbon inverse opals have been employed as a basis for most
composite electrodes, as they can serve as a light matrix with high
electrical conductivity and good structural stability that is
maintained during cycling. Deposition of active electrode
materials onto the carbon inverse opals allows for the combina-
tion of the benefits of thin film electrodes (short lithium diffusion
lengths and electron paths, better accommodation of structural
changes during lithium insertion-extraction) with the advantages
of macroporous electrodes. However, this becomes possible at the
expense of specific capacity, which is somewhat lowered by the
carbon matrix.
Despite all these pioneering investigations, until recently
silicon inverse opals, with the highest gravimetric specific
capacities towards lithium, have not been studied as potential
electrodes for lithium ion batteries. While this paper was in
preparation Kim et al. published a report[42] in which electro-
chemical performance of powdered silicon inverse opals was
investigated. Their observations seem to confirm our original ideas,
and are in qualitative agreement with the results presented herein.
We report an investigation which includes syntheses and
detailed studies of the electrochemical behavior of amorphous
hydrogenated silicon inverse-opal films {(a–Si:H)–io}, nanocrys-
talline silicon inverse-opal films {(nc–Si)–io}, nanocrystalline
silicon inverse-opal films with carbon coating {(nc–Si)–io–(C)},
and of carbon inverse-opal films with amorphous silicon coating
{(C)–io–(a–Si:H)}. The dependence of the cycling stability and
rate performance of these materials on the template sphere size
and the film thickness is discussed.
2. Results and Discussion
2.1. Hydrogenated Amorphous Silicon Inverse Opals
2.1.1. Preparation and Characterization
Preparation of (a–Si:H)–io films via chemical vapor deposition
(CVD) results in materials of high uniformity and surface
smoothness (Fig. 1). The amorphous nature of silicon was confirmed
by PXRD (Supporting information, Fig. S1a) and by Raman micro-
spectroscopy (m-Raman; Supporting information, Fig. S2).
The oxygen content in (a–Si:H)–io was estimated by thermo-
gravimetric analysis (TGA) to be 4.9 wt%. The calculations were
based on the assumption that the material contained only silicon
and oxygen; see the Supporting information for calculation
details. It should be noted that in practice the oxygen content
could be somewhat lower due to the presence of other impurities.
If the oxidation of these impurities occurred in the same
temperature range as formation of silicon dioxide, then the
weight loss caused by formation of their volatile oxidation
products was wrongly attributed to the presence of oxygen in the
initial material.
The conductivities of (a–Si:H)–io measured by two-probe
method were found to lay in the range of 6  10
7 S cm
1 to 4 
10
6 S cm
1. These values are relatively high, considering that
most of the reported conductivities of hydrogenated amorphous
silicon are 10
8 S cm
1 or less.[44] An increase in conductivity can
be explained by combination of factors, including high deposition
temperature and hence lower hydrogen content, and relatively
Adv. Funct. Mater. 2009, 19, 1999–2010
www.afm-journal.de
Figure 1. Scanning electron microscopy (SEM) images of (a–Si:H)–io
templated from 890-nm pre-sintered silica microspheres.
high level of oxygen, which is known to increase the room
temperature conductivity of amorphous silicon.[45]
2.1.2. Cyclic Voltammetry
The shapes of cyclic voltammetry (CV) curves of 5-mm (a–Si:H)–
io films templated from 250-nm, 480-nm and 890-nm spheres
Figure 2. CV curves and SEM images after 5 CV cycles of (a–Si:H)–io 5-mm-thick films. Materials are templated from: a, d) 250-nm spheres; b, e) 480-nm
spheres; c, f) 890-nm spheres.
Adv. Funct. Mater. 2009, 19, 1999–2010 ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
(Fig. 2a–c) are consistent with the results previously reported in
FULL PAPER
literature for nanostructured silicon anodes.[46–48]
The shape of the curve corresponding to the first cycle differs
essentially from the subsequent ones. The onset of lithiation
during the first cycle occurs later and the cathodic peak at 0.20 V is
much less pronounced than for the subsequent cycles. Interest-
ingly, the first anodic peak has a lower intensity and is
significantly shifted towards higher voltages during the first
cycle, while the second peak is similar in shape and position to the
later cycles. Similar effects have been observed previously.[46,48,49]
The reason for such behavior is not exactly clear; however, it can
be conjectured that some sort of local restructuring or ‘activation’
takes place in silicon during the initial lithium alloying.
After the first cycle, scans assume remarkably repeatable
shapes suggesting high reversibility of the lithiation–de-lithiation
reaction. Anodic peaks are better separated from each other in the
case of electrodes templated from smaller spheres. The position
of the first anodic peak shifts slightly towards higher voltages with
increasing sphere size (from 0.33 V for 250-nm spheres to 0.36 V
for 890-nm spheres), while the position of the second de-lithiation
peak is almost intact. This suggests that the rate of the reaction in
the early stages of de-lithiation was to some extent controlled by
the rate of solid-state lithium diffusion in the electrode material,
so that thicker walls caused larger polarization.
Scanning electron microscopy (SEM) imaging of materials
after five CV scans revealed significant morphological changes
in the structure of (a–Si:H)–io electrodes (Fig. 2d–f). The
cycled materials had an interesting ‘sponge-like’ structure.
Importantly, even though the morphology was very different from
the un-cycled material the structure was still porous and open to
electrolyte.
2.1.3. Galvanostatic Cycling
The cycling behavior of 10-mm (a–Si:H)–io film electrode
templated from 890-nm spheres is demonstrated in Figure 3a
and b. Most electrodes demonstrated initial de-lithiation capacities
2001
 www.afm-journal.de

be caused by several side processes, namely by formation of solid-

FULL PAPER

electrolyte interphase (SEI), by the irreversible reaction of lithium

with the dangling bonds in silicon and by the reaction of lithium
with surface oxides and fluorides. Initial coulombic efficiencies
increased with the increase in sphere diameter. This, together
with the fact that the surface to volume ratio of an inverse-opal
material is inversely proportional to the diameter of the
templating spheres, is an indication that irreversible capacities
during the first galvanostatic cycles arise from processes on the
electrode surfaces. Coulombic efficiencies increased during the
first 3–5 cycles, eventually reaching levels over 97%.
Electrodes prepared from 890-nm spheres demonstrated better
cycling stability, with their capacity retentions being around 70%
after 145 cycles, compared to 50% or lower in the case of 250-nm
spheres. It can be conjectured that the smaller pores of the latter
materials are more easily blocked or ‘‘clogged’’ by the swelling
electrode material during lithiation, preventing penetration of the
electrolyte and thus gradually excluding more and more material
from the electrochemical reaction. A similar effect has been
observed for tin oxide inverse-opal electrodes.[38]
The rate performance of (a–Si:H)–io electrodes is outlined in
Figure 3c. Capacities decreased exponentially with the increase of
current, reaching negligible values before or at 1C rate. No
apparent correlation of rate performance with the sphere size was
observed, suggesting that lithium diffusion is not a rate-limiting
process. On the other hand, thinner films demonstrated better
performance at high currents, which indicates that the rate of
lithium alloying–dealloying is controlled by electrical conductivity
of the electrode material.

2.1.4. Material Characterization at Various Lithium Contents

Figure 3. a, b) Galvanostatic cycling of 10-mm-thick (a–Si:H)–io films
templated from 890-nm spheres; c) rate performance of (a–Si:H)–io
electrodes: de-lithiation capacities at different current rates.
over 2500 mAh g
1 (Table 1), which is comparable to the values for
nanostructured silicon anodes reported in literature.[46–49]
Low coulombic efficiencies, lying in the range of 65 to 87%,
and hence high irreversible capacities during the first cycle, could
Morphology changes occurring at different degrees of lithiation
were visualized by SEM. For this purpose electrodes were cycled
galvanostatically at C/10 rate for twenty cycles. Cycling was then
stopped during the lithiation half-cycle of cycle 21 at 2 V (at the
end of the 20th cycle), 100 mV and 50 mV (at the end of lithiation),
cells were disassembled in an argon-filled glove box and electrode
materials were transferred under argon atmosphere in electron
microscope.
SEM images of 10-mm thick electrodes obtained from 890-nm
microspheres at different degrees of lithiation are shown in
Figure 4a–c. De-lithiated structures closely resembled pristine
(a–Si:H)–io, indicating high reversibility of electrochemical
lithium insertion. Partially lithiated electrodes had thicker walls
and smaller pores between the air spheres; the structures,
however, remained highly open and accessible to electrolyte. The

Table 1. Cyclic performance of (a–Si:H)–io electrodes: initial delithiation capacities and coulombic efficiencies; capacity retentions and coulombic
efficiencies for cycle 145.

Cycle # Sphere size [nm] 250 480 890

Film thickness [mm] 5 10 5 10 15 5 10 15

1
1 Capacity [mAh g ] 2829 2469 2069 2569 2394 2496 2698 2569
Coulombic efficiency [%] 76.0 70.3 65.2 73.2 72.2 81.9 87.2 82.7
145 Capacity retention [%] 50.5 16.2 56.7[a] 54.7 54.1 69.9 73.9 67.0

Coulombic efficiency [%] 98.1 93.6 97.3 97.8 97.5 99.2 99.3 98.7

[a] Data for cycle 139.

2002 ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Funct. Mater. 2009, 19, 1999–2010
www.afm-journal.de
Figure 4. a–c) (a–Si:H)–io electrodes templated from 890-nm spheres at
different degrees of lithiation; d) (a–Si:H)–io templated from 890-nm
spheres after complete loss of cycling ability.
walls of electrodes fully loaded with lithium were very thick, and
virtually no macroporosity was observed. The structure of an
electrode that had completely lost its capacity towards lithium as a
result of prolonged galvanostatic cycling is shown in Figure 4d.
Similarly to the materials at maximum degree of lithiation, it
possesses very thick walls and zero porosity. It is probable that the
cycling performance of (a–Si:H)–io electrodes degraded mostly
due to progressive blockage of macropores by lithiated silicon,
which then prevented access of electrolyte to parts of the material
and excluded those parts from the electrochemical reaction.
The m-Raman spectra of de-lithiated amorphous (a–Si:H)–io
electrodes after cycling were similar to the spectra of pristine
(a–Si:H)–io (Supporting information, Fig. S2). However, the
peaks of cycled materials were shifted by 7–10 cm
1 towards
higher frequencies, probably due to residual mechanical stress
originally induced in the material by volume swings during
electrochemical cycling. Compressive stresses in silicon are
known to result in the shift the Raman peak to higher
wavenumbers.[50]
2.2. Nanocrystalline Silicon Inverse Opals
2.2.1. Preparation and Characterization
The rate performance of (a–Si:H)–io film electrodes is limited by
the low conductivity of amorphous silicon. In order to increase
the conductivity (a–Si:H)–io were annealed at 800 8C in inert
atmosphere resulting in films of macroporous (nc–Si)–io.
In contrast to the smooth surface of amorphous silicon (Fig. 1)
the crystalline samples had a rough ‘grainy’ structure (Fig. 5). The
shape and position of the peak on m-Raman spectra indicated
formation of a fully crystalline silicon (Supporting information,
Fig. S2). PXRD data of (nc–Si)–io also confirmed that annealing
resulted in crystalline silicon (Supporting information, Fig. S1b).
The particle size was estimated from broadening of diffraction
peaks to be in the range of 34–42 nm. The conductivity of (nc–Si)–
io films measured by a two-probe method varied in the range of
7  10
4 to 2  10
3 S cm
1.
Adv. Funct. Mater. 2009, 19, 1999–2010 ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
FULL PAPER
Figure 5. SEM images of (nc–Si)–io templated from 480-nm silica
spheres.
2.2.2. Cyclic Voltammetry
The shapes of (nc–Si)–io cyclic voltammograms (Fig. 6a)
resemble the CV curves of (a–Si:H)–io electrodes, especially
during later cycles, which is to be expected since the electrode
material becomes amorphous upon lithiation.[46] However, there
are some striking differences between two sets of data.
Firstly, in case of crystalline silicon, the cathodic process was
very intensive at low voltages during the first and, to some extent,
the second cycle, with the currents at 50 mV being 10–20 times
higher than the currents at 200 mV, compared to the ratio of about
2 for amorphous electrodes.
Secondly, an additional anodic peak appeared around
0.42–0.43 V during the first two cycles. This peak is quite
unusual for nanostructured silicon electrodes, but a similarly
positioned peak, or the corresponding plateau in galvanostatic
process, was observed during de-lithiation of large silicon
particles or thick silicon films. The peak has been attributed to
de-lithiation of Li4Si15,[21,51,52] the phase that usually forms only if
the material is cycled at voltages below 50 mV. None of the
conditions previously considered necessary for observation of the
peak at 0.42 V are met in the case of crystalline silicon inverse
opals: XRD results suggest that the silicon crystallites were small,
below 50 nm in diameter, and the material was not exposed to
voltages below 50 mV vs. Li/Liþ. Clearly, there is an unknown
factor that promotes formation of crystalline Li4Si15, or some
other undefined phase responsible for the peak, at relatively high
voltages from nanocrystalline silicon, since the characteristic
2003
 www.afm-journal.de

2.2.3. Galvanostatic Cycling

FULL PAPER

10-mm-thick (nc–Si)–io film electrodes templated from 480-nm

silica spheres were cycled galvanostatically at various current rates
(Fig. 6b). The electrodes demonstrated high initial capacities
combined with relatively high coulombic efficiencies even at 10C
rate, probably due to much higher electronic conductivity of
crystalline silicon in comparison with the amorphous one.
However, capacities faded fast, reaching negligible values by the
15th cycle at best.
SEM imaging of (nc–Si)–io after complete loss of their
capacities revealed complete degradation of the material into
separate visually unstructured islands, with no traces of inverse-
opal structure (Fig. 6c). Such a striking difference from the
amorphous electrodes can be explained by the fact that the silicon
nanocrystallites were only weakly attached to each other and easily
separated from the macroporous network under the mechanical
stress caused by lithiation–de-lithiation.

2.3. Nanocrystalline Silicon–Carbon Inverse-Opal Composites

2.3.1. Preparation and Characterization

Silicon–carbon composite inverse-opal materials were synthe-
sized by depositing a thin layer of hard carbon onto the silicon
inverse opals in an attempt to achieve mechanical stabilization of
crystalline silicon structures and further increase conductivity of
the material. Initial polymerization of sucrose infiltrated into
the (a–Si:H)–io lead to non-conductive coating that covered the

Figure 6. a) CV of 10-mm-thick (nc–Si)–io film templated from 250-nm

spheres; b) galvanostatic cycling of 10-mm-thick (nc–Si)–io films templated
from 480-nm spheres: de-lithiation capacities and coulombic efficiencies at
C/10 and 10C rates; c) (nc–Si)–io material after 20 galvanostatic cycles.

peak disappeared after two CV cycles and no such peaks were

observed during cycling of amorphous silicon inverse opals.
Thirdly, unlike the case of amorphous silicon, the positions of
anodic peaks did not correlate with the template sphere size,
suggesting either higher lithium mobility in the electrode
material or shorter diffusion lengths. We speculate that shorter
solid state diffusion lengths can arise from the fact that electrolyte
has access to the bulk of the electrode material due to the nano-
particulate nature of the latter (Fig. 5).
Lastly, and most importantly from a practical point of view, the
current intensity decreased sharply with each cycle, suggesting
that some irreversible structural changes occurred during cycling Figure 7. a) Composite of (a–Si:H)–io templated from 890-nm spheres
which lead to diminution of electrode material available for with a polymer after sucrose infiltration and initial polymerization at 150 8C;
electrochemical reaction. b) (nc–Si)–io–(C) templated from 890 nm silica spheres.

2004 ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Funct. Mater. 2009, 19, 1999–2010
www.afm-journal.de
silicon surface (Fig. 7a) After pyrolysis the materials had a ‘grainy’
structure and looked very similar to (nc–Si)–io (Fig. 7b). The
carbon coating could not be observed by SEM in most cases after
pyrolysis.
m-Raman spectra confirmed full conversion of silicon to the
crystalline form. The peak area ratios of the D and G carbon bands
for different (nc–Si)–io–(C) samples varied from 0.55 to 0.63,
being in good agreement with literature data for glassy carbons
(Supporting information, Fig. S3).[27,53] It is worth noting that the
carbon peak intensities in Raman spectra taken at different spots
of the same sample varied strongly, indicating inhomogeneity of
the carbon coating.
The conductivities of (nc–Si)–io–(C) measured by a two-probe
method varied widely from 4  10
3 to 9  10
2 S cm
1 for
Figure 8. a) CV of 10-mm-thick (nc–Si)–io–(C) film templated from
890-nm spheres; b) galvanostatic cycling of 10-mm-thick (nc–Si)–io–(C)
films templated from 890-nm spheres: de-lithiation capacities and cou-
lombic efficiencies at C/10 and 10C rates; c) (nc–Si)–io–(C) material after
20 galvanostatic cycles.
Adv. Funct. Mater. 2009, 19, 1999–2010 ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
different samples. Large discrepancies in conductivities between
FULL PAPER
samples can be caused by poor continuity of the carbon coatings.
It is possible that the hydrophobic nature of the silicon surface
prevented complete infiltration of sucrose solutions or caused
formation of islands of polymer, rather than continuous films,
during the drying and polymerization steps.
2.3.2. Electrochemical Cycling
CV and galvanostatic cycling of 10-mm-thick (nc–Si)–io–(C)
electrode, templated from 890-nm silica spheres, produced
results that strikingly resembled the electrochemical behavior of
(nc–Si)–io electrodes.
The signal intensity decreased sharply with cycling during CV,
and an additional anodic peak located at 0.43 V was observed in
the first and second cycles (Fig. 8a).
Figure 9. SEM images of (C)–io–(a–Si:H) templated from a) 250-nm,
b) 480-nm, and c) 890-nm silica spheres.
2005
 www.afm-journal.de

Table 2. Elemental composition of carbon–silicon composite inverse opals inverse opals (Supporting information, Fig. S4). The carbon
FULL PAPER

obtained from TGA data.

Raman bands were very weak, probably due to the presence of a
Content [at%]
silicon coating. TGA of carbon inverse opals and of (C)–io–(a–
Si:H) in air allowed for the estimation of elemental composition
Sphere size [nm] 250 480 890 of the (C)–io–(a–Si:H) synthesized from opals with different
sphere sizes (Table 2). Again, an assumption that the materials
Element Si 86.3 90.2 88.1
contained only carbon, silicon, and oxygen was made. A detailed
C 8.2 7.1 7.4
description of the calculation method is given in the Supporting
O 5.5 2.7 4.6
information. Interestingly, there is no correlation between sphere
size and Si:C ratio. This suggests that the carbon inverse opals
Galvanostatic cycling demonstrated high capacities during the were very porous and carbon occupied only small fraction of
first several cycles, even at 10C rate, and fast degradation with the interstices between air spheres.
complete loss of capacity after 8–11 cycles (Fig. 8b). The (nc–Si)– The conductivities of the (C)–io–(a–Si:H) materials varied from
io–(C) material, after galvanostatic cycling at C/10 rate, appeared 0.1 to 0.6 S cm
1 with no apparent dependence on the template
to have completely disintegrated and no macroporous structure sphere size. These values are about an order of magnitude lower
was observed by SEM (Fig. 8c). than the conductivities of pristine carbon inverse opals and six
orders of magnitude higher than the conductivities of (a–Si:H)–io
films.
2.4. Carbon–Amorphous Silicon Inverse-Opal Composites

2.4.1. Preparation and Characterization 2.4.2. Cyclic Voltammetry

A different approach to increase cyclability and rate performance
of silicon-based macroporous electrodes was also taken. Carbon
inverse opals were coated with a layer of amorphous silicon via
CVD. SEM imaging of (C)–io–(a–Si:H) films templated from
various sphere sizes confirmed that structures remained
macroporous, although the pore size is significantly diminished
for the 250 nm sphere opal (Fig. 9). Raman spectroscopy
confirmed formation of an amorphous silicon coating on carbon
The CV curves of (C)–io–(a–Si:H) films are very similar to the case of
(a–Si:H)–io (Fig. 10a). The first cycle was quite distinct from the
following ones, particularly during lithiation. From cycle 2 and
beyond the curves had very similar shapes and intensities,
suggesting a highly reversible reaction. In contrast to (a–Si:H)–io
the position of the first anodic peak did not change with the sphere
size, probably due to the fact that thicknesses of silicon layer and
hence lithium diffusion lengths were similar for all the sphere sizes.

2.4.3. Galvanostatic Cycling

The cycling behavior of 5- and
10-mm-thick films templated
from 250-nm spheres, 5-, 10-
and 15-mm-thick films from 480-
nm spheres, and 5-, 10- and
15-mm-thick films from 890-nm
spheres of (C)–io–(a–Si:H) elec-
trodes was studied by galvano-
static cycling at current rates
ranging from C/10 to 10C
(Fig. 10b and c).
The (C)–io–(a–Si:H) electro-
des demonstrated high de-
lithiation capacities around
2000 mAh g
1 in the first cycle
(Table 3). Coulombic efficiencies
during the first cycle lay in the
range of 70 to 80%. They
increased with an increase in
sphere diameter, but the trend
was less pronounced than in the
case of (a–Si:H)–io electrodes. It
is probable that (C)–io–(a–Si:H)
had higher surface areas than (a–
Figure 10. a) CV of 10-mm-thick (C)–io–(a–Si:H) film templated from 250-nm spheres; b, c) galvanostatic cycling Si:H)–io due to increased surface
of 5-mm-thick (C)–io–(a–Si:H) film templated from 250-nm spheres; d) rate performance of (C)–io–(a–Si:H) roughness. On the other hand,
electrodes: de-lithiation capacities at different current rates. unlike (a–Si:H)–io, the composite

2006 ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Funct. Mater. 2009, 19, 1999–2010
www.afm-journal.de

Table 3. Cyclic performance of (C)–io(a–Si:H): initial de-lithiation capacities and coulombic efficiencies; capacity retentions and coulombic efficiencies for

FULL PAPER
cycle 145.

Sphere size [nm] 250 480 890

Cycle # Film thickness [mm] 5 10 5 10 15 5 10 15

1 Capacity [mAh g
1] 2406 1890 1905 2557 1736 2165 2195 2374
Coulombic efficiency [%] 71.2 67.3 75.5 76.5 75.7 78.1 79.5 77.5
145 Capacity retention [%] 88.6 56.2 82.7 86.9 72.4 87.6 84.1 91.0
Coulombic efficiency [%] 97.6 95.6 98.2 98.8 98.4 98.2 96.8 97.8

materials were not treated with HF after silicon deposition, and so tures and studied their cycling performance and rate capabilities
had less impurity on the silicon surface. These two factors should as electrodes for lithium ion batteries. Amorphous silicon inverse
have opposite effects on the initial irreversible capacity of (C)–io– opals were synthesized by CVD into the interstices of silica opal
(a–Si:H) electrodes, leading to somewhat lower efficiencies of templates and subsequent removal of the template. This material
materials with large macropores in comparison with (a–Si:H)–io, demonstrated high capacities at low currents and decent capacity
and similar or even slightly higher efficiencies for materials with
smaller pores.
After 3–5 galvanostatic cycles coulombic efficiencies increased
to 97–99%. (C)–io–(a–Si:H) electrodes demonstrated very good
cycling stability with capacity retentions over 80% after 145 cycles
for the majority of samples. Similar to (a–Si:H)–io, (C)–io–(a–
Si:H) electrodes templated from larger spheres cycled better.
However, the difference in cycling stability was less pronounced
than in (a–Si:H)–io electrodes. It seems that the carbon
‘‘backbone’’ prevented complete clogging of the pores by lithiated
silicon.
Rate performance of (C)–io–(a–Si:H) electrodes (Fig. 10d)
remarkably improved in comparison to (a–Si:H)–io, with the
capacities delivered at a 1C rate being 35–55% of the capacities at
a C/10 rate. The capacities of most electrodes decreased to
negligible levels by the 10C rate. Rate performances did not
correlate with electrode thicknesses. This suggests that, unlike (a–
Si:H)–io, conductivity of the electrode material is not the rate-
controlling step of the electrochemical reaction.
Raman spectra of (C)–io–(a–Si:H) after 145 cycles revealed a
shift of the silicon peak by 5 cm
1 towards higher wavenumbers
in comparison with pristine (C)–io–(a–Si:H) films (Supporting
information, Fig. S5). A similar, but larger shift was observed for
cycled (a–Si:H)–io, and is probably caused by mechanical stress
induced during lithiation.[50] The carbon signal on the Raman
spectra was still very weak, which indicates that silicon uniformly
covers the carbon matrix even after extensive cycling.
SEM imaging of (C)–io–(a–Si:H) electrodes galvanostatically
cycled for 145 cycles (Fig. 11) showed that the macroporous
structures remained largely intact and silicon coating was well-
preserved on the carbon matrix, indicating a high reversibility of
electrochemical lithium insertion.
Morphology changes caused by lithiation were also visualized
by SEM (Fig. 12). Lithiated electrodes had thicker walls and
smaller pores between the air spheres; the structures, however,
remained open and accessible to electrolyte even at the highest
degree of lithiation.

3. Conclusions
We have synthesized several types of silicon and carbon–silicon Figure 11. (C)–io–(a–Si:H) electrodes templated from a) 250-nm,
periodic macroporous materials based on inverse-opal architec- b) 480-nm, and c) 890-nm silica spheres after 145 galvanostatic cycles.

Adv. Funct. Mater. 2009, 19, 1999–2010 ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 2007

FULL PAPER
Figure 12. (C)–io–(a–Si:H) electrodes templated from 890-nm silica
spheres at different degrees of lithiation.
retentions, but its performance is severely restricted due to the
low electrical conductivity of silicon. Conversion of amorphous
silicon to the nanocrystalline form by heat treatment in an inert
atmosphere enhanced the conductivity of silicon inverse opals by
several orders of magnitude. This led to significant improvement
in discharge capacities at high rates. However, the crystalline
material proved to be unstable to mechanical stresses induced by
volume swings, and the inverse-opal structures collapsed during
cycling, leading to rapid loss of electrode capacity. Silicon inverse
opals were infiltrated with the sucrose carbon precursor and
pyrolyzed to produce periodic macroporous nanocrystalline
silicon coated with a thin carbon layer. It was hoped that the
carbon coating would prevent disintegration of the macroporous
structure during cycling and increase the electrical conductivity of
the material. Contrary to expectations, no improvement in cycling
behavior in comparison with the pristine crystalline silicon
inverse opal electrodes was observed, and the macroporous
structure was found to have completely collapsed after
20 galvanostatic cycles. Another approach to the synthesis of
carbon–silicon composite inverse opals was then used, whereby
thin (50 nm) layers of amorphous silicon were deposited by CVD
onto carbon inverse opals. This proved to be the ‘winning
strategy’, as such composite materials demonstrated high
discharge capacities close to the capacities of pristine amorphous
silicon opals, combined with good capacity retention and rate
performance.
Overall, amorphous silicon materials based on inverse-opal
structures demonstrated promising cycling characteristics, con-
2008 ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.afm-journal.de
firming the idea that high macroporosity coupled with the
amorphous nature of silicon might be beneficial for reducing
stresses associated with large volume changes during electro-
chemical lithium insertion–removal.
Introduction of a conductive carbon backbone into the
structure led to great improvement in rate performance. The
presence of carbon also decreased pulverization of silicon and
pore clogging, hence achieving better capacity retentions after
prolonged cycling.
4. Experimental
Growth of Silica Opal Templates: Monodisperse silica spheres of three
different diameters were used in this study: 250 nm, 480 nm and 890 nm, as
determined by SEM. These spheres were synthesized by a modified Stöber
method [54], which involved multiple re-growth cycles as described by
Giesche [55]. In brief, hydrolysis of tetraethyl orthosilicate and poly-
condensation of silica was performed in anhydrous ethanol medium in the
presence of ammonium hydroxide as a catalyst. After synthesis the spheres
were meticulously cleaned by centrifuging, decanting, and re-dispersing in
water and then in ethanol until no ammonia odor could be observed. The
pre-cleaned spheres were heated in air at 450 8C (250-nm spheres) or at
600 8C (480- and 890-nm spheres) for 8 hours in order to reduce sphere
shrinkage and hence to minimize cracking of the opal templates during the
subsequent high temperature treatment [56]. Opal films were deposited via
a low-pressure modification [57] of the evaporation-induced self-assembly
method (LP EISA) [58]. Vials of ethanolic silica sphere dispersion and a
vertically positioned substrate (soda lime glass, silicon, quartz, or sapphire
slides) were kept in a low-pressure chamber during the film deposition to
accelerate the evaporation of the solvent, so that its rate was the same or
slightly higher than the sphere sedimentation rate. Opal films with the
thicknesses of approximately 5, 10, and 15 mm were prepared by varying LP
EISA silica sphere fraction in the dispersion in the range from 0.5 vol% to
2 vol% and deposition pressure from 55 Torr to 110 Torr. The mechanical
strength of the silica opal films was then enhanced by interconnecting
the spheres, as well as connecting the initial layer of spheres to the
substrate with a thin layer of amorphous silica in order to simplify handling
of the material and to reduce its cracking and exfoliation from the substrate
caused by high temperature treatment during further processing [59].
Preparation of Silicon Inverse Opals: Hydrogenated amorphous silicon
was deposited in the interstices of silica opals via dynamic cold-wall low-
pressure CVD from disilane Si2H6 precursor. Typically, the opal was heated
to 450 8C in a custom-built CVD reactor with base pressure maintained at
10
7 Torr. Disilane gas was then flowed over the hot substrate at a rate of
0.1 sccm. The pressure in the reactor during the CVD process slightly
fluctuated around 1 Torr. These conditions allowed for silicon deposition
on a flat substrate with a rate of 40–60 nm h
1. To ensure as complete an
infiltration of silicon into the opal pores as possible the deposition was
performed for a somewhat longer period of time than was actually required
for infiltration. The deposition time was varied from 90 minutes for 250-nm
spheres to 240 minutes for 890-nm spheres. The silicon overlayer grown on
the outer surface of the silica–silicon composite had to be removed to
ensure proper access for the HF etchant to the silica spheres and, more
importantly, access of electrolyte to the interior of the inverse opal during
electrochemical cycling. This overlayer was removed by unidirectional
reactive ion etching (RIE) with CHF3 plasma. The silica–silicon composites
were then converted to silicon inverse opals. This was done by way of
chemically etching the silica in aqueous HF solution (2 vol%) for a period of
8 to 10 hours. a–Si:H could then be converted to the nanocrystalline form
by heating silicon inverse opals to 800 8C in argon or nitrogen atmosphere
for 12 hours.
Preparation of Silicon–Carbon Inverse-Opal Composites: Carbon precur-
sor solution was prepared by dissolving sucrose (35 g) in water (100 mL)
with subsequent addition of concentrated H2SO4 (1 mL). The molar ratio of
sucrose to sulfuric acid in the final mixture was 1:0.18. (a–Si:H)–io films
were then immersed in the sugar solution and kept there in vacuo for
Adv. Funct. Mater. 2009, 19, 1999–2010
www.afm-journal.de
Table 4. The sequence of C rates during galvanostatic cycling.
C rate C/10 C/5 C/2 1C 2C 5C 10C
Cycle # 1–25 26–35 46–55 66–75 86–95 106–115 126–135
36–45
56–65
76–85
96–105
116–125
136–145
2 hours to ensure complete infiltration of the precursor. After removal from
the solution the top surface of the inverse opal was gently pressed against a
flat elastic polydimethylsiloxane (PDMS) sheet and fixed there with
foldback paper clips to ensure that no polymer overlayer would be formed.
This ‘‘sandwich’’ was then kept in air at 80 8C for 6 hours to evaporate most
of the moisture, and for another 12 hours at 150 8C to obtain an in situ
cross-linked polymer. After removal of the PDMS, the silicon–polymer
composite was carbonized at 800 8C for 12 hours under nitrogen
atmosphere using a heating ramp of 2 8C min
1.
Preparation of Carbon–Silicon Inverse-Opal Composites: Mechanically
stabilized silica opal films were infiltrated with a sucrose solution and
polymerized at 150 8C in air as described above. The infiltration and
polymerization process was repeated 3 times to increase carbon content in
the material. The composites were then pyrolyzed at 1 000 8C for 12 hours
under nitrogen atmosphere with a ramping speed of 2 8C min
1. The silica
template was etched away with aqueous hydrofluoric acid (2 vol%). The
obtained carbon inverse opals were coated with a silicon layer by means of
dynamic low pressure CVD from disilane at 450 8C. The silicon deposition
was performed for 45 min resulting in uniform coating of the carbon
backbone. The thickness of silicon walls was estimated from SEM images
to be around 50 nm regardless of the sphere size.
Electrochemical Characterization of the Materials: All electrochemical
testing of electrode materials was performed using type 2325 coin cells
with macroporous film electrode placed on a stainless steel current
collector and lithium metal as a counter electrode. Solution of LiPF6 in 1:1
ethylene carbonate–dimethyl carbonate mixture (1 M) was used as
electrolyte, and Celgard 2340 material was implemented as a separator.
All CV experiments were performed at a scanning rate of 0.1 mV sec
1
between 0.05 V and 2 V vs. Li/Liþ. The cycling performance of the materials
was studied by galvanostatic cycling between 0.05 V and 2 V vs. Li/Liþ at
current rates ranging from 357.9 mA g
1 (C/10 rate) to 35 790 mA g
1 (10C
rate). Since the exact silicon to carbon ratios in the composite materials
were not known, calculations of C rates were based on the theoretical
capacity of 3 579 mA g
1, as if the electrode materials were composed of
silicon only. The sequence of current rates applied for characterization of
(a–Si:H)–io and (C)–io–(a–Si:H) is summarized in Table 4.
Acknowledgements
GAO is Government of Canada Research Chair in Materials Chemistry. He
is deeply indebted to the Natural Sciences and Engineering Research
Council of Canada for generous and sustained financial support of his
work. The authors thank Prof. Linda Nazar at the University of Waterloo
and members of her research group for kindly granting access to the
facilities in her laboratory and for technical support. Supporting
Information is available online from Wiley InterScience or from the author.
Received: February 23, 2009
Published online: May 7, 2009
[1] O. D. Velev, T. A. Jede, R. F. Lobo, A. M. Lenhoff, Nature 1997, 389, 447.
[2] M. Deutsch, Y. A. Vlasov, D. J. Norris, Adv. Mater. 2000, 12, 1176.
Adv. Funct. Mater. 2009, 19, 1999–2010 ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[3] B. T. Holland, C. F. Blanford, A. Stein, Science 1998, 281, 538.
FULL PAPER
[4] H. W. Yan, C. F. Blanford, B. T. Holland, M. Parent, W. H. Smyrl, A. Stein,
Adv. Mater. 1999, 11, 1003.
[5] A. Blanco, E. Chomski, S. Grabtchak, M. Ibisate, S. John, S. W. Leonard, C.
Lopez, F. Meseguer, H. Miguez, J. P. Mondia, G. A. Ozin, O. Toader, H. M.
van Driel, Nature 2000, 405, 437.
[6] A. A. Zakhidov, R. H. Baughman, Z. Iqbal, C. Cui, I. Khayrullin, S. O. Dantas,
J. Marti, V. G. Ralchenko, Science 1998, 282, 897.
[7] M. W. Perpall, K. P. U. Perera, J. DiMaio, J. Ballato, S. H. Foulger, D. W.
Smith, Langmuir 2003, 19, 7153.
[8] R. W. J. Scott, S. M. Yang, G. Chabanis, N. Coombs, D. E. Williams, G. A.
Ozin, Adv. Mater. 2001, 13, 1468.
[9] T. Cassagneau, F. Caruso, Adv. Mater. 2002, 14, 1837.
[10] H.-H. Chen, H. Suzuki, O. Sato, Z.-Z. Gu, Appl. Phys. A 2005, 81, 1127.
[11] U. Kamp, V. Kitaev, G. von Freymann, G. A. Ozin, S. A. Mabury, Adv. Mater.
2005, 17, 438.
[12] H. Yan, K. Zhang, C. F. Blanford, L. F. Francis, A. Stein, Chem. Mater. 2001,
13, 1374.
[13] Y. Liu, S. Wang, J. W. Lee, N. A. Kotov, Chem. Mater. 2005, 17,
4918.
[14] R. C. Schroden, C. F. Blanford, B. J. Melde, B. J. S. Johnson, A. Stein, Chem.
Mater. 2001, 13, 1074.
[15] J. I. L. Chen, G. von Freymann, S. Y. Choi, V. Kitaev, G. A. Ozin, Adv. Mater.
2006, 18, 1915.
[16] J.-S. Yu, S. Kang, S. B. Yoon, G. J. Chai, Am. Chem. Soc. 2002, 124,
9382.
[17] J. S. Sakamoto, B. Dunn, J. Mater. Chem. 2002, 12, 2859.
[18] G. M. Ehrlich, in Handbook of Batteries, 3rd ed. (Eds: D. Linden, T. B.
Reddy), McGraw-Hill, USA 2002, Ch. 35.
[19] T. Nagura, R. Tozawa, Prog. Batteries Sol. Cells 1990, 9, 209.
[20] Sony Press Release, http://www.sony.net/SonyInfo/News/Press/200502/
05-006E/index.html (last accessed February 15, 2005).
[21] M. N. Obrovac, L. Christensen, Electrochem. Solid-State Lett. 2004, 7,
A93.
[22] T. D. Hatchard, J. R. Dahn, J. Electrochem. Soc. 2004, 151, A838.
[23] H. Take, H. Kajii, K. Yoshino, Synth. Met. 2001, 121, 1313.
[24] J. Lu, Z. Tang, Z. Zhang, W. Shenet, Mater. Res. Bull. 2005, 40,
2039.
[25] N. S. Ergang, J. C. Lytle, H. Yan, A. Stein, J. Electrochem. Soc. 2005, 152,
A1989.
[26] M. Hara, H. Nakano, K. Dokko, S. Okuda, A. Kaeriyama, K. Kanamura,
J. Power Sources 2009, 189, 485.
[27] K. T. Lee, J. C. Lytle, N. S. Ergang, S. M. Oh, A. Stein, Adv. Funct. Mater.
2005, 15, 547.
[28] Z. Wang, F. Li, N. S. Ergang, A. Stein, Chem. Mater. 2006, 18, 5543.
[29] F. Su, X. S. Zhao, Y. Wang, J. Zeng, Z. Zhou, J. Y. Lee, J. Phys. Chem. B 2005,
109, 20200.
[30] H. Yamada, T. Yamato, I. Moriguchi, T. Kudo, Solid State Ionics 2004, 175,
195.
[31] H. Yamada, T. Yamato, I. Moriguchi, T. Kudo, Chem. Lett. 2004, 33, 1548.
[32] L. Kavan, M. Zukalová, M. Kalbác, M. Graetzel, J. Electrochem. Soc. 2004,
151, A1301.
[33] Z. Bing, Y. Yuan, Y. Wang, Z.-W. Fu, Electrochem. Solid-State Lett. 2006, 9,
A101.
[34] L. Fu, T. Zhang, Q. Cao, H. P. Zhang, Y. P. Wu, Electrochem. Commun. 2007,
9, 2140.
[35] E. M. Sorensen, S. J. Barry, H.-K. Jung, J. M. Rondinelli, J. T. Vaughey, K. R.
Poeppelmeier, Chem. Mater. 2006, 18, 482.
[36] S.-W. Woo, K. Dokko, K. Kanamura, Key Eng. Mater. 2006, 320, 263.
[37] S.-W. Woo, K. Dokko, K. Kanamura, Electrochim. Acta 2007, 53, 79.
[38] J. C. Lytle, H. Yan, N. S. Ergang, W. H. Smyrl, A. Stein, J. Mater. Chem. 2004,
14, 1616.
[39] F.-S. Ke, L. Huang, H.-H. Jiang, H.-B. Wei, F.-Z. Yang, S.-G. Sun, Electro-
chem. Commun. 2007, 9, 228.
2009

[40] F.-S. Ke, L. Huang, H.-B. Wei, J.-S. Cai, X.-Y. Fan, F.-Z. Yang, S.-G. Sun,
FULL PAPER
J. Power Sources 2007, 170, 450.
[41] F.-S. Ke, L. Huang, J.-S. Cai, S.-G. Sun, Electrochim. Acta 2007, 52, 6741.
[42] H. Kim, B. Han, J. Choo, J. Cho, Angew. Chem. Int. Ed. 2008, 47, 10151.
[43] H. Yamada, K. Tagawa, M. Komatsu, I. Moriguchi, T. Kudo, J. Phys. Chem. C
2007, 111, 8397.
[44] D. M. Goldie, in Properties of Amorphous Silicon and its Alloys (Ed: T. Searle,)
Institution of Engineering and Technology, UK 1998, Ch. 4.1.
[45] S. Tsuda, T. Takahama, Y. Hishikawa, H. Tarui, H. Nishiwaki, K. Wakisaka,
S. Nakano, J. Non-Cryst. Solids 1993, 164–166, 679.
[46] M. Green, E. Fielder, B. Scrosati, M. Wachtler, J. S. Moreno, Electrochem.
Solid-State Lett. 2003, 6, A75.
[47] V. Baranchugov, E. Markevich, E. Pollak, G. Salitra, D. Aurbach, Electro-
chem. Commun. 2007, 9, 796.
[48] C. K. Chan, H. Peng, G. Liu, K. McIlwrath, X. F. Zhang, R. A. Huggins, Y.
Cui, Nat. Nanotechnol. 2008, 3, 31.
2010 ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.afm-journal.de
[49] T. Jiang, S. Zhang, X. Qiu, W. Zhu, L. Chen, Electrochem. Commun. 2007, 9,
930.
[50] V. Paillard, P. Puech, R. Sirvin, S. Hamma, P. Roca i Cabarrocas, J. Appl.
Phys. 2001, 90, 3276.
[51] J. Saint, M. Morcrette, D. Larcher, L. Laffont, S. Beattie, J.-P. Pérès, D.
Talaga, M. Couzi, J.-M. Tarascon, Adv. Funct. Mater. 2007, 17, 1765.
[52] M. N. Obrovac, L. J. Krause, J. Electrochem. Soc. 2007, 154, A103.
[53] K. Ray, R. McCreery, Anal. Chem. 1997, 69, 4680.
[54] W. Stöber, A. Fink, E. Bohn, J. Colloid Interface Sci. 1968, 26, 62.
[55] H. Giesche, J. Eur. Ceram. Soc. 1994, 14, 205.
[56] A. A. Chabanov, Y. Jun, D. J. Norris, Appl. Phys. Lett. 2004, 84, 3573.
[57] V. Kitaev, G. A. Ozin, Adv. Mater. 2003, 15, 75.
[58] P. Jiang, J. F. Bertone, K. S. Hwang, V. L. Colvin, Chem. Mater. 1999, 11,
2132.
[59] H. Miguez, N. Tetreault, B. Hatton, S. M. Yang, D. Perovic, G. A. Ozin,
Chem. Commun. 2002, 2736.
Adv. Funct. Mater. 2009, 19, 1999–2010