Silicon nanowire fabric as a lithium ion battery electrode material

Article abstract of DOI:10.1021/ja208232h

A nonwoven fabric with paperlike qualities composed of silicon nanowires is reported. The nanowires, made by the supercritical-fluid-liquid-solid process, are crystalline, range in diameter from 10 to 50 nm with an average length of >100 μm, and are coated with a thin chemisorbed polyphenylsilane shell. About 90% of the nanowire fabric volume is void space. Thermal annealing of the nanowire fabric in a reducing environment converts the polyphenylsilane coating to a carbonaceous layer that significantly increases the electrical conductivity of the material. This makes the nanowire fabric useful as a self-supporting, mechanically flexible, high-energy-storage anode material in a lithium ion battery. Anode capacities of more than 800 mA h g^-1 were achieved without the addition of conductive carbon or binder.

Full text of DOI:10.1021/ja208232h

ARTICLE
pubs.acs.org/JACS
Silicon Nanowire Fabric as a Lithium Ion Battery Electrode Material
†
†
†
†
†
Aaron M. Chockla, Justin T. Harris, Vahid A. Akhavan, Timothy D. Bogart, Vincent C. Holmberg,
†
†,‡
‡
†,
Chet Steinhagen, C. Buddie Mullins, Keith J. Stevenson, and Brian A. Korgel *
†
‡
Department of Chemical Engineering and Department of Chemistry and Biochemistry, Texas Materials Institute, Center for Nano-
and Molecular Science and Technology, The University of Texas at Austin, Austin, Texas 78712-1062, United States
S Supporting Information
b
ABSTRACT: A nonwoven fabric with paperlike qualities composed
of silicon nanowires is reported. The nanowires, made by the super-
critical-fluidÀliquidÀsolid process, are crystalline, range in diameter
from 10 to 50 nm with an average length of >100 μm, and are coated
with a thin chemisorbed polyphenylsilane shell. About 90% of the
nanowire fabric volume is void space. Thermal annealing of the
nanowire fabric in a reducing environment converts the polyphenyl-
silane coating to a carbonaceous layer that significantly increases the
electrical conductivity of the material. This makes the nanowire fabric
useful as a self-supporting, mechanically flexible, high-energy-storage
À1
anode material in a lithium ion battery. Anode capacities of more than 800 mA h g were achieved without the addition of
conductive carbon or binder.
7
,14À17,20,21
’
INTRODUCTION
promising.
For example, Cui and co-workers achieved
À1
a capacity of 2725 mA h g with very good stability, retaining
more than 1400 mA h g after 700 cycles using an intercon-
nected amorphous Si hollow nanosphere thin-film electrode.
Unfortunately, these electrode materials are very expensive at
present and cannot be produced in the significant quantities
needed for commercial LIB applications.
Solvent-based processes for nanowire synthesis, such as super-
The demand for portable electronic devices has been steadily
À1
rising, in part because of the increasing global dependence on
mobile information, entertainment, and communication. These
devices require portable power sources. Lithium ion (Li )
batteries (LIBs) have become widely used because of their high
energy density, but there is a strong desire to improve many
aspects of LIBs, including cost, safety, energy density, capacity,
cycle life, and rate capability. The recent interest in using LIBs in
automobiles has heightened the search for new materials that can
achieve higher energy density and prolonged cycle life.
2
2
+
12,13,19
1
critical-fluidÀliquidÀsolid (SFLS) and solutionÀliquidÀsolid
(
SLS) growth, can produce large amounts of Si nanowires
2
3À28
2
(SiNWs) at relatively low cost.
In a 10 mL laboratory-scale
+
reactor, hundreds of milligrams of nanowires with lengths of
100 μm can be produced in a few hours,
to create nanowire-based nonwoven fabrics several inches on a
side. These nanowires exhibit excellent mechanical properties,
are highly flexible, and have strength-to-weight ratios greater
than those of both carbon fiber and Kevlar.
material is thin (thickness of 50À150 μm) and has a paperlike
appearance and feel. The first fabrics consisting entirely of one-
dimensional nanomaterials were made from carbon nanotubes
(CNTs) because of their mechanical strength, flexibility, and
large aspect ratio and the ability to produce the material in
macroscopic quantities.
melane-M (K2 Mn O ) that could be synthesized in large
quantities by hydrothermal methods have been used to form
macroscopic free-standing membranes. Recently, SFLS
synthesis has been used to produce Ge nanowire fabric.
Fabric-like materials consisting of CNTs and graphene
sheets have recently been explored as electrodes for flexible
A LIB functions by shuttling Li between the anode and
23,26
>
which is sufficient
cathode. The typical cathode in a commercially available LIB is
composed of lithium metal oxide (i.e., LiCoO ) and the anode
of graphite. The LIB capacity is limited in part by the inter-
calation of Li by the graphitic anode material; thus, higher
capacity batteries require anode materials that can accommodate
more Li . The theoretical capacity of the graphite anode is
2
27
+
27,29
This fabric
+
À1 3,4
3
72 mA h g
.
Lithium-alloying materials have the potential
for significantly higher storage capacities. For instance, Si alloys
+
with Li at room temperature to form Li Si , which corresponds to
a substantially higher theoretical capacity of 3579 mA h g .
However, Si undergoes an enormous volume expansion of nearly
1
5 4
À1 5À7
30,31
Metal oxide nanowires of crypto-
8
Àx
8
16
3
00% when fully lithiated. Bulk crystalline Si cannot tolerate the
stresses associated with these lithation/delithiation cycles and
3
2
9
crumbles, resulting in battery failure. On the other hand, Si
nanostructures have been found to tolerate extreme changes in
volume with cycling.
above 2000 mA h g
27,33
7,10À21
Thin Si films have realized capacities
À1 12,13,19
.
Si nanoparticles have also been
explored, but with overall specific capacities that have been
limited by the need for a conductive carbon matrix to ensure
Received: September 2, 2011
Published: November 09, 2011
11,18
electrical contact with the electrode.
Si nanowires have been
r 2011 American Chemical Society
20914
dx.doi.org/10.1021/ja208232h
|J. Am. Chem. Soc. 2011, 133, 20914–20921

Journal of the American Chemical Society
ARTICLE
1
1,21,34À37
LIB applications
and as electrode supports for other
The XPS results were analyzed using CasaXPS software. Background
subtraction was done using a Shirley background model. Peak areas were
deconvoluted as Voigt functions (30% Gaussian character). The Si 2p
1
1,37
21,34
materials, as in grapheneÀsilicon,
CNTÀSiNW,
and
2
1,35
CNTÀLi Ti O
electrodes. Here we demonstrate the
4
6 12
0
0
+
2+
3+
4+
creation of a Si nanowire fabric and show that it can function
as a standalone anode material without the need for additional
conductive fillers (activated carbon) or polymeric binders. For
good battery performance, it was necessary to convert a thin
region was deconvoluted as Si 2p3/2, Si 2p1/2, Si , Si , Si , and Si
peaks placed at 99.3, 99.9, 100.1, 101.1, 102.1, and 103.2 eV, respectively.
The zero-valent Si spin splits were assumed to have the same full width at
half-maximum (FWHM), and the area of the 2p1/2 peak was set as half
the area of the 2p3/2 peak. Similarly, one FWHM value was used for all of
the other oxide peaks; however, for simplicity, the other oxidative states
were not divided into their respective spin splits. Values for the FWHM
of the oxidized states were constrained by placing the FWHM between
3
8
polyphenylsilane coating on the Si nanowires to carbon by
annealing under forming gas (7% H in N ) at 900 ꢀC. This thin
layer provides the electrical conductivity needed for efficient Li
charging and discharging of the layer. The LIB performance of
the Si nanowire fabric and the role of the shell on the cycling
behavior are reported.
2
2
+
1.8 and 2.2 times the zero-valent peaks.
The C1s region was deconvoluted into three peaks placed at 284,
84.7, and 285.5 eV. These binding energies correspond to SiÀC, CÀC,
2
and CÀO bonding, respectively. The O 1s region was deconvoluted into
two peaks. The first peak, placed at 532.5 eV, corresponds to SiO and
2
’
EXPERIMENTAL SECTION
CÀO bonding. The second peak, placed at 533.9 eV, relates to O2
adsorbed on the surface of the wires. Only one FWHM was used to
generate the peaks for each species of C and O.
Materials. All chemicals were used as received without further
purification. Dodecanethiol (DDT, g98%), hydrogen tetrachloroaurate
trihydrate (g99.9%), sodium borohydride (g98%), toluene (anhydrous,
ACS grade, 99.8%), ethanol (99.9%), tetraoctylammonium bromide
TOAB, 98%), and chloroform (99.8%) were purchased from Sigma-
Aldrich. Monophenylsilane (MPS, >95%) was obtained from Gelest, and
LiPF [1.0 M in a 1:1 ethylcarbonate (EC, >99.95%)/ diethyl carbonate
DEC, >99.98%) mixture] was purchased from Novolyte. For battery
assembly, Celgard 2400 membranes (25 μm, purchased from Celgard)
were used as separators, and Li metal (99.9%) was obtained from Alfa Aesar.
Silicon Nanowire Synthesis. Si nanowires were synthesized by
gold nanocrystal-seeded SFLS growth in toluene using the synthetic
Raman spectroscopy was performed using a Renishaw inVia micro-
scope equipped with a 514.5 nm argon ion laser in backscattering con-
À1
figuration. The Stokes Raman signal at 521 cm for single-crystalline
(
bulk Si oriented with the [110] direction normal to the laser was used
to calibrate the instrument. The beam was focused using an optical
microscope with a 50Â objective lens. Spectra were taken on free-
standing sheets of nanowire fabric lying on glass slides by single scans
6
(
À1
from 100 to 2000 cm at 5% laser power (0.2 mW).
LIB Anode Processing and Battery Testing. Si nanowire fabric
was prepared by sonicating nanowire dispersions (5À10 mg/mL Si
nanowires in chloroform) for 1 h. The dispersions were drop-cast into a
Teflon trough and left overnight to dry. After drying, the fabric was
peeled from the trough using tweezers (and a thin razor blade wedged
between the fabric and trough bed when necessary; see movie S1 in the
Supporting Information) and cut into 1 cm  1 cm squares that were
loaded into quartz boats and heated under a moderate flow of forming
gas in a tube furnace.
38
procedures detailed in previous reports. Briefly, MPS and 2 nm Au
nanoparticles were mixed together in a 290:1 Si:Au mole ratio and
diluted with anhydrous toluene to achieve an MPS concentration of
39
1
56 mM. Nanowires were then grown in a flow-through sealed titanium
reactor at 490 ꢀC and 10.3 MPa. After the reaction, the nanowires were
collected from the reactor walls and washed with chloroform (10 mL)
and ethanol (5 mL) several times via centrifugation at 8000 rpm. The
nanowires were then dispersed in chloroform to a concentration of
mg/mL and stored in a vial under ambient conditions prior to use.
Material Characterization. Scanning electron microscopy (SEM)
images were acquired using a Zeiss Supra 40 scanning electron microscope
with an in-lens arrangement, a working voltage of 5 keV, and a working
distance of 5 mm. SEM samples were imaged on silicon wafers obtained
from SEH. Transmission electron microscopy (TEM) images were
digitally acquired using either a FEI Tecnai Spirit BioTwin transmission
electron microscope operated at 80 kV or a JEOL 2010F field-emission
microscope operated at 200 kV. TEM samples were prepared by drop-
casting from chloroform dispersions onto 200 mesh lacey carbon copper
TEM grids (Electron Microscopy Sciences).
X-ray diffraction (XRD) was carried out using a Rigaku R-Axis Spider
diffractometer with an image-plate detector and Cu Kα radiation (λ =
.5418 Å) operated at 40 kV and 40 mA. Measurements were taken on
samples on a 0.5 mm nylon loop. Samples were scanned for 10 min while
rotating at 1 deg/s under ambient conditions. The radial data were
integrated from 2θ = 20 to 80ꢀ and plotted. Background scattering from
the nylon loop was subtracted from the sample measurement.
Batteries were prepared within an argon-filled glovebox using a Cu
foil current collector. The Si nanowire fabric was placed on the Cu
current collector to serve as the anode material. The electrolyte was a 1.0
5
6
M solution of LiPF in a 1:1 EC (>99.95%):DEC (>99.98%) mixture
(Novolyte). A few drops of electrolyte were deposited over the nanowire
fabric. A Celgard 2400 separator membrane (25 μm thick) was then
added, followed by Li metal foil (99.9%, Alfa Aesar) as the counter
electrode. Galvanostatic measurements were made using an Arbin
+
BT-2143 test unit that was cycled between 3 V and 10 mV vs Li/Li at
À1 5À7
a rate of C/20 (C = 3579 mA h g
).
After cycling, the batteries were disassembled, and the anode was
cleaned by soaking the fabric in acetonitrile overnight and then rinsing
with 0.5 M sulfuric acid for 1 min (i.e., until bubbling subsided). The
anode was then placed in a vial and soaked in isopropyl alcohol (IPA) for
several minutes. The IPA was decanted, and the anode was soaked two
more times in the same way and then dried in air before SEM and TEM
imaging.
1
’ RESULTS AND DISCUSSION
X-ray photoelectron spectroscopy (XPS) was performed on a Kratos
photoelectron spectrophotometer with monochromatic Al Kα radiation
Si Nanowire Fabric. Figure 1 shows a photograph and SEM
images of Si nanowire fabric made by casting concentrated
dispersions of SFLS-grown nanowires in toluene into a Teflon
trough (See movie S1 in the Supporting Information). The fabric
is composed of highly entangled nanowires. It is ∼50 μm thick
and relatively porous, containing ∼90% void volume. The Si
nanowire fabric is pliable with good mechanical integrity (see
movie S2 in the Supporting Information). Even though the
(1487 eV). The Si nanowire fabric was secured on the experimental tray
using double-sided Cu tape. Spectra were collected at 0.1 eV intervals
using an integration time of 800 ms through a tungsten coil set at 4.8 V
bias with respect to the sample. Data were collected continuously under
À9
high vacuum (10 Torr). The effect of sample charging on the XPS
0
data was corrected by shifting the Si 2p3/2 peak to a binding energy
of 99.3 eV.
2
0915
dx.doi.org/10.1021/ja208232h |J. Am. Chem. Soc. 2011, 133, 20914–20921

Journal of the American Chemical Society
ARTICLE
Figure 2. (a) Galvanostatic cycling tests of Si nanowire fabric before
Figure 1. (a) Photograph of mechanically flexible Si nanowire fabric.
and after annealing at 700, 800, 900, 1000, and 1100 ꢀC at a cycle rate of
(b, c) SEM images of the fabric. (d) High-resolution TEM (HRTEM)
À1 5À7
C/20 (C = 3579 mA h g
). The inset shows discharge capacities
image of a Si nanowire showing its crystallinity. (e) Fast Fourier
transform (FFT) pattern from the HRTEM image in (d) showing a
Æ110æ growth direction. (f) Photographs of Si nanowire fabrics annealed
at the indicated temperatures under a reducing atmosphere.
after 20 cycles for batteries with nanowire fabric anodes annealed at
various temperatures. (b) Galvanostatic cycling of a Si nanowire fabric
anode annealed at 900 ꢀC.
nanowires are single crystals of a material (i.e., Si) that is brittle in
bulk form, they are mechanically flexible and strong because of
27
their narrow diameter.
Electrochemical Performance of Si Nanowire Fabric An-
odes in LIBs. 1 cm  1 cm squares of Si nanowire fabric were
tested as standalone Si-based anode materials without added
conductive carbon or binder for a LIB. The as-made Si nanowire
fabric performed poorly. As shown in Figure 2a, the capacity in
À1
the first cycle was just under 600 mA h g but then dropped to
nearly zero after only a couple of cycles. Tests of the electrical
conductivity showed that the nanowire fabric is electrically
insulating with a conductivity of ∼0.2 nS/m. The nanowire
fabric must be sufficiently conductive to facilitate efficient lithium
4
0
alloying with Si.
The electrical conductivity of the Si nanowire fabric was
enhanced significantly by annealing under a reducing atmo-
sphere at temperatures greater than 700 ꢀC. Figure 3 shows
currentÀpotential curves for nanowire fabric annealed at various
temperatures. Annealing at temperatures between 700 and
Figure 3. CurrentÀpotential measurements of Si nanowire fabric
placed between two indium tin oxide (ITO) electrodes (inset). The
nanowire fabric was annealed under a reducing atmosphere at the
indicated temperatures. Measurements were made using ∼150 μm
2
thick fabric samples with a contact area of 1 cm .
1
000 ꢀC led to conductivities ranging from 70 to 150 nS/m.
Fabric annealed at 1100 ꢀC had a conductivity of 1400 nS/m.
Figure 1f shows photographs of Si nanowire fabric after
annealing under forming gas at various temperatures. The fabric
turns increasingly black with increasing annealing temperature.
The Si nanowires produced by the SFLS process using phenylsi-
lane as a reactant have a thin polyphenylsilane shell on their
enhanced electrical conductivity. When the nanowires are an-
nealed in the presence of oxygen and water, the shell material
oxidizes to a defective and insulating SiO layer.
2
The annealed Si nanowire fabric performed well in galvano-
static cycling tests. Figure 2a shows the capacity (with cycling at a
3
8
À1
5À7
4
surface. Annealing under a reducing environment transforms
C/20 rate, where C = 3579 mA h g , corresponding to Li Si
)
15
this shell material to carbon, which provides the significantly
for Si nanowire fabric annealed under a reducing atmosphere at
2
0916
dx.doi.org/10.1021/ja208232h |J. Am. Chem. Soc. 2011, 133, 20914–20921

Journal of the American Chemical Society
ARTICLE
Table 1. Summary of the Capacity Retention Data for Si
Nanowire Fabric Anodes Annealed at the Indicated
Temperatures
À1
capacity (mA h g
)
retention (%)
a
b
temperature (ꢀC) 1st cycle 2nd cycle 20th cycle
C
ret1
C
ret2
no anneal
534
1180
1341
1485
821
142
257
738
898
439
341
8
111
579
804
433
363
2
9
6
43
78
89
99
7
8
9
1
1
00
00
43
00
54.0
57
000
100
608
60
106
a
Amount of charge capacity retained after 20 cycles relative to the first
b
charging cycle. Amount of charge capacity retained after 20 cycles
relative to the second charging cycle.
Figure 5. Raman spectra of Si nanowire fabric before and after anneal-
ing under a reducing atmosphere at various temperatures. The Si (TO)
À1
peak at 521 cm and carbon-related D and G bands at 1380 and
1
À1
560 cm are labeled.
Figure 4 shows the constant-current voltage profiles and
differential capacity (dQ/dV) curves for the first and 20th cycles
of batteries with Si nanowires annealed at 800, 900, 1000, and
1
100 ꢀC. Each constant-current voltage profile shows two
plateaus during the charge cycle, corresponding to the lithiation
of crystalline silicon and the formation of the Li Si phase. The
6
15 4
discharge cycles have a steeper slope than the charge cycles
corresponding to a broader peak in the dQ/dV curves),
(
indicating that delithiation occurs over a relatively broad range
of potentials, which is consistent with amorphous silicon. The
dQ/dV plots provide a better illustration of the potentials at
which the lithiation and delithiation events occur. Each of the
first-cycle charge curves in the dQ/dV plots shows a peak
between 150 and 200 mV, corresponding to lithiation of crystal-
line Si. This peak is shifted slightly in the 900 ꢀC sample
(
Figure 4d) to ∼250 mV. A peak near 50 mV is also observed
6
for each sample and is attributed to the formation of Li Si . By
1
5 4
the 20th cycle, the first lithiation event has shifted to ∼300 mV,
which falls in the range expected for lithiation of amorphous
41
Si (a-Si). The second lithiation event has a significantly reduced
intensity, indicating a marked consumption of crystalline silicon.
The first discharge cycle shows a prominent peak at ∼450 mV,
which corresponds to the delithiation potentials of amorphous
Figure 4. Constant-current voltage profiles for Si nanowire fabric
anodes annealed at (a) 800, (c) 900, (e) 1000, and (g) 1100 ꢀC with
corresponding differential capacity curves (b, d, f, and h, respectively).
41
various temperatures. All of the nanowire samples exhibited a large
initial irreversible capacity loss during the first cycle, which is typical
for Si anode materials and results from formation of so-
lidÀelectrolyte interface (SEI) layers upon cycling to low
potentials. The nanowire fabric annealed to 700 ꢀC per-
formed better than the as-made nanowires, but the capacity
was still very limited. On the other hand, nanowires annealed
at 900 ꢀC performed quite well and retained a cycling capacity
silicon and Li Si . Lithium extraction during the 20th cycle
15 4
41
occurs at the a-Si delithiation potential (300 mV).
Composition of the Carbonaceous Shell on the Si Nano-
wires. Raman spectroscopy, XPS, and XRD were used to deter-
mine the composition of the shell material formed after anneal-
ing the nanowires under forming gas. Figure 5 shows Raman
spectra of nanowires as-made and after annealing at various
temperatures. The as-made nanowires exhibit the characteristic
À1
À1
À1
of 800 mA h g after 20 cycles. As shown in Figure 2b, the
Si (TO) band at 521 cm along with a smaller peak at 495 cm
À1
42À45
capacity was still higher than 500 mA h g after 100 cycles.
that is most likely related to stacking faults.
After annealing
When the nanowires were annealed at even higher tempera-
ture (>900 ꢀC), they did not perform as well. The Figure 2a
inset shows a plot of the discharge capacity after 20 cycles for
Si nanowire fabric electrodes annealed at various tempera-
tures. The highest performance was achieved with nanowires
annealed at 900 ꢀC. Table 1 summarizes the capacity reten-
tion data of the Si nanowire fabric annealed at various
temperatures.
of the nanowires at 700 ꢀC, the nanowire spectra still have the
characteristic Si-related peaks, but there is a significant sloping
baseline due to fluorescence from the sample. This fluores-
cence is due to a chemical change in the shell material that is
probably related to the phenyl species. With increasing annealing
temperature, the fluorescence signal diminishes in intensity.
Nanowires annealed at 900 ꢀC and above have almost no
fluorescent background. The nanowires annealed at 800 ꢀC
2
0917
dx.doi.org/10.1021/ja208232h |J. Am. Chem. Soc. 2011, 133, 20914–20921

Journal of the American Chemical Society
ARTICLE
Figure 6. XPS data for the O 1s, C 1s, and Si 2p states of Si nanowire fabric annealed at the indicated temperatures. The intensities of all of the peaks are
normalized, but the carbon signal decreases significantly relative to the Si signal as the annealing temperature increases. Figure 7 plots the relative
amounts of O, C, and Si obtained from integration of the XPS peaks. The peak-fitting procedures are described in the Experimental Section.
and higher exhibit two prominent bands at ∼1350 and
À1
∼
1600 cm , corresponding to disordered (D) and graphitic
4
6À49
(
G) carbon.
These peaks are not present in the Raman
spectra of the as-made nanowires nor in the spectra of the
nanowires annealed at 700 ꢀC. It appears that the fluorescent
species formed as a result of annealing at 700 ꢀC is converted to
the carbon that forms at higher annealing temperatures.
The much improved battery performance of the Si nanowire
fabric after annealing results from the formation of the electrically
conductive carbon layer on the nanowires. It is obvious from the
change in color of the nanowire fabric from its typical pale-yellow
to black that the shell material has been converted to carbon, as
confirmed by the Raman spectra. But why does the battery
performance diminish when the nanowires are heated above
9
00 ꢀC? The Raman spectra show that the same carbon species
that helped to achieve good battery response are still present on
the nanowires annealed at 1000 and 1100 ꢀC, and the electri-
cal conductivity of the nanowires is also high (Figure 3). How-
ever, XPS (Figures 6 and 7) and XRD (Figure 8) reveal that a
significant amount of Si oxidation and some SiC formation occur
when the nanowires are annealed at these higher temperatures.
The XRD patterns in Figure 8 show that the nanowires are still
composed predominantly of crystalline Si after annealing, in-
dicating that the oxidation and carbide formation occurs only on
the nanowire surface. It appears that residual oxygen or water,
most likely contained initially in the polyphenylsilane shell,
Figure 7. Summary of compositional analyses extracted from the XPS
data in Figure 6: (a) relative Si, C, and O content determined by
integrating the O 1s, C 1s, and Si 2p peaks; (b) relative SiO, SiO
SiO content determined from the Si 2p peak structure.
x
, and
oxidizes the surfaces of the nanowires. Nanowires annealed at
0
2
1
100 ꢀC no longer have a measurable Si XPS peak, and all of the
4+
2+
Si signal corresponds to either Si or Si . XRD also indicates
that some of the surface carbon reacts with the underlying Si to form
SiC. The SiC diffraction peaks are very broad, indicating that the
crystalline SiC domains are extremely small, which is consistent with
the lack of a SiC-related Raman signal at 790 cm . The Si signal
does not show up in the XPS or Raman spectra because the surface
layer becomes too thick to penetrate. Additionally, there is a notice-
able decrease in the relative carbon signal in the XPS spectra, along
with the appearance of a relatively strong XPS signal at higher
binding energy of 285.5 eV, indicating that the carbonaceous layer
has also oxidized to some extent (Figures 6 and 7).
À1 50
0
2
0918
dx.doi.org/10.1021/ja208232h |J. Am. Chem. Soc. 2011, 133, 20914–20921

Journal of the American Chemical Society
ARTICLE
Figure 8. XRD patterns of Si nanowire fabric before and after annealing
at the indicated temperatures under forming gas. The peak positions
corresponding to diamond cubic Si (JCPDS no. 00-027-1402) and
silicon carbide (SiC, labeled with *; JCPDS no. 00-029-1131) are shown.
Figure 10. TEM and scanning TEM (STEM) images of Si nanowires
annealed at 900 ꢀC under forming gas after 20 electrochemical cycles in a
lithium cell. (a, c) Two types of nanowires were observed in nearly equal
proportions: (a) amorphous wires that appear to have a hollow core and
(c) crystalline nanowires with a thick amorphous shell. (b, d) FFTs of
the TEM images in (a) and (c). (e, f) STEM images and (g, h) corre-
sponding EDS line scans at the indicated positions of (e, g) amorphous
wires that appear to have a hollow core and (f, h) crystalline nanowires
with a thick amorphous shell.
shows SEM images of the Si nanowire fabric that had been
annealed at 900 ꢀC under forming gas before and after 20
galvanostatic cycles. The nanowires removed from the coin cell
housing were covered with an organic surface coating, presum-
ably comprised of ROCO Li and ROLi species associated with
2
14
reduction of the EC/DEC solvent. Rinsing the nanowires with
acetonitrile, H SO , and IPA removed the organic coating (SEI
2
4
layer). In comparison with the nanowires prior to electrochem-
ical cycling, the nanowires appear to be larger in diameter.
TEM images of the nanowires after galvanostatic cycling
showed two different populations of nanowires in nearly equal
proportions. As shown in Figure 10, some nanowires were
amorphous and appeared to have hollowed cores. Others had a
crystalline core with a thick amorphous shell. Nanoprobe
energy-dispersive spectroscopy (EDS) maps of Si, O, and C
composition across the nanowire diameters showed that
nanowires with crystalline cores had a higher concentration
of Si in the center, with a measurable amount of carbon and
oxygen mostly at the nanowire surface. The amorphous
nanowires with the hollow cores had a high concentration
of Si in the shell, along with C and O. It appears that the
structural integrity of these Si nanowires was lost during
lithium cycling, with a rupture of the shell. The presence of
the nanowires with crystalline cores may indicate that these
Figure 9. SEM images of Si nanowire fabric annealed at 900 ꢀC under
forming gas (a) prior to electrochemical cycling, (b) after 20 galvano-
static cycles without washing, and (c) after 20 galvanostatic cycles with
washing using acetonitrile, H SO , and IPA.
2
4
The oxide and SiC that forms at the nanowire surface after
annealing at 1000 and 1100 ꢀC is most likely responsible for the
reduced lithiation/delithiation of the nanowire fabric. SiC is known
51
to be electrochemically inactive for lithiathion, and the oxide layer
creates an electrically insulating barrier between the Si core and the
electrolyte that is not conducive to battery performance.
Si Nanowire Structure after Electrochemical Cycling. The
coin cell housing was disassembled after cycling to examine the
structural integrity of the nanowires by SEM and TEM. Figure 9
2
0919
dx.doi.org/10.1021/ja208232h |J. Am. Chem. Soc. 2011, 133, 20914–20921

Journal of the American Chemical Society
ARTICLE
nanowires were not involved in the electrochemical cycling.
These issues require further study.
Research Laboratory (FA-8650-07-2-5061). C.B.M. and K.J.S.
acknowledge financial support from the Welch Foundation (F-1436
to C.B.M. and F-1529 to K.J.S.). V.C.H. acknowledges the Fannie
and John Hertz Foundation and the National Science Foundation
Graduate Research Fellowship Program for financial support.
Raman spectroscopic characterization was supported as part
of the program “Understanding Charge Separation and Trans-
fer at Interfaces in Energy Materials (EFRC: CST)”, an Energy
Frontier Research Center funded by the U.S. Department
of Energy, Office of Science, Office of Basic Energy Sciences,
under Award DE-SC0001091.
’
CONCLUSION
A nonwoven fabric of Si nanowires has been reported. The
fabric has the look and feel of tissue paper yet is composed
entirely of crystalline Si. Annealing the nanowire fabric at 900 ꢀC
under reducing atmosphere (forming gas) leads to good perfor-
+
mance as an anode in a Li battery without addition of carbon
binder. The key to their good performance is the presence of a
thin electrically conductive carbon layer on the nanowires. When
the nanowires were annealed above 900 ꢀC, the battery perfor-
mance diminishes because of the formation of a surface barrier
’ REFERENCES
(1) Goodenough, J. B.; Kim, Y. Chem. Mater. 2010, 22, 587.
layer of SiO and SiC under the conducting carbonaceous coating
2
(
(
(
2) Armand, M.; Tarascon, J. M. Nature 2008, 451, 652.
3) Dahn, J. R.; Zheng, T.; Liu, Y.; Xue, J. S. Science 1995, 270, 90.
4) Winter, M.; Besenhard, J. O.; Spahr, M. E.; Nov ꢀa k, P. Adv. Mater.
that prevents efficient lithitation and delithiation.
This study has shown that the surface composition of Si
nanowires plays a critical role in their effectiveness as an anode
material in an LIB. Si nanowires can alloy with lithium, but their
capacity and cycling stability all rely on the quality of the surfaces.
In the present case of Si nanowires grown by the SFLS process
with monophenylsilane as a reactant, a residual polyphenylsilane
shell can be converted from an inactive layer to a coating that
enables Si nanowire lithiation/delithiation. The coating is con-
formal and uniform and does not require a separate materials
deposition step. Carbon binder, which would lower the specific
capacity of the material, need not be added. This Si nanowire
fabric represents a mechanically flexible silicon anode material
with the potential to be applied in lightweight paper-based or
plastic LIBs. Future research needs to focus on understanding the
cycling durability and the factors that limit performance, includ-
ing the formation and stability of the solidÀelectrolyte interface
layer. Another issue that requires further consideration is the
influence of the nanowire density on performance. Since the
nanowire fabric has 90% void volume, the volumetric capacity of
the nanowire fabric is relatively low. The specific capacity of 800
1
998, 10, 725.
5) Weydanz, W. J.; Wohlfahrt-Mehrens, M.; Huggins, R. A. J. Power
Sources 1999, 81, 237.
6) Obrovac, M. N.; Christensen, L. Electrochem. Solid-State Lett.
004, 7, A93.
7) Chan, C. K.; Peng, H.; Liu, G.; McIlwrath, K.; Zhang, X. F.;
(
(
2
(
Huggins, R. A.; Cui, Y. Nat. Nanotechnol. 2008, 3, 31.
(8) Xu, Y. H.; Yin, G. P.; Zuo, P. J. Electrochim. Acta 2008, 54, 341.
(9) Boukamp, B. A.; Lesh, G. C.; Huggins, R. A. J. Electrochem. Soc.
1981, 128, 725.
(10) Chan, C. K.; Patel, R. N.; O’Connell, M. J.; Korgel, B. A.; Cui, Y.
ACS Nano 2010, 4, 1443.
(
11) Lee, J. K.; Smith, K. B.; Hayner, C. M.; Kung, H. H. Chem.
Commun. 2010, 46, 2025.
12) Graetz, J.; Ahn, C. C.; Yazami, R.; Fultz, B. Electrochem. Solid-
State Lett. 2003, 6, A194.
13) Kasavajjula, U.; Wang, C.; Appleby, A. J. J. Power Sources 2007,
63, 1003.
14) Chan, C. K.; Ruffo, R.; Hong, S. S.; Huggins, R. A.; Cui, Y.
(
(
1
(
J. Power Sources 2009, 189, 34.
(15) Hu, L.; Wu, H.; Hong, S. S.; Cui, L.; McDonough, J. R.; Bohy,
S.; Cui, Y. Chem. Commun. 2010, 47, 367.
À1
À3
mA h g corresponds to 186 mA h cm , which is significantly
À3
lower than the volumetric capacity of 777À867 mA h cm for
graphite. However, it should be possible to increase the volu-
metric capacities of the nanowire fabric by densifying the films
with pressure. An estimation of the maximum volumetric capac-
ity of a densified Si nanowire fabric looks compelling, with a
(16) Kang, K.; Lee, H.-S.; Han, D.-W.; Kim, G.-S.; Lee, D.; Lee, G.;
Kang, Y.-M.; Jo, M.-H. Appl. Phys. Lett. 2010, 96, No. 053110.
(
17) La €i k, B.; Eude, L.; Pereira-Ramos, J.-P.; Cojocaru, C. S.; Pribat,
D.; Rouvi ꢁe re, E. Electrochim. Acta 2008, 53, 5528.
(18) Lee, J.-K.; Kung, M. C.; Trahey, L.; Missaghi, M. N.; Kung,
H. H. Chem. Mater. 2009, 21, 6.
À3
maximum of 1864 mA h cm , which is more than a factor of 2
higher than that of graphite. The influence of volumetric expan-
sion and contraction upon lithiation and delithiation on cycling
stability, however, would still need to be understood.
(
19) Maranchi, J. P.; Hepp, A. F.; Kumta, P. N. Electrochem. Solid-
State Lett. 2003, 6, A198.
20) Peng, K.; Jie, J.; Zhang, W.; Lee, S.-T. Appl. Phys. Lett. 2008,
(
93, No. 033105.
’
ASSOCIATED CONTENT
(21) Huang, R.; Fan, X.; Shen, W.; Zhu, J. Appl. Phys. Lett. 2009,
5, No. 133119.
9
S
Supporting Information. Movies (MPG) showing the fab-
(22) Yao, Y.; McDowell, M. T.; Ryu, I.; Wu, H.; Liu, N.; Hu, L.; Nix,
b
W. D.; Cui, Y. Nano Lett. 2011, 11, 2949.
23) Heitsch, A. T.; Fanfair, D. D.; Tuan, H.-Y.; Korgel, B. A. J. Am.
Chem. Soc. 2008, 130, 5436.
24) Lee, D. C.; Hanrath, T.; Korgel, B. A. Angew. Chem., Int. Ed.
005, 44, 3573.
25) Holmes, J. D.; Johnston, K. P.; Doty, R. C.; Korgel, B. A. Science
000, 287, 1471.
rication and handling of Si nanowire fabric. This material is avail-
able free of charge via the Internet at http://pubs.acs.org.
(
(
’
AUTHOR INFORMATION
2
2
(
Corresponding Author
korgel@che.utexas.edu
(26) Hanrath, T.; Korgel, B. A. Adv. Mater. 2003, 15, 437.
(27) Smith, D. A.; Holmberg, V. C.; Korgel, B. A. ACS Nano 2010,
4
2
, 2356.
’
ACKNOWLEDGMENT
(28) Heitsch, A. T.; Akhavan, V. A.; Korgel, B. A. Chem. Mater. 2011,
We thank Kate Collier for HRTEM images of Si nanowires.
3, 2697.
Financial support of this research was provided in part by
the Robert A. Welch Foundation (F-1464) and the Air Force
(29) Smith, D. A.; Holmberg, V. C.; Lee, D. C.; Korgel, B. A. J. Phys.
Chem. C 2008, 112, 10725.
2
0920
dx.doi.org/10.1021/ja208232h |J. Am. Chem. Soc. 2011, 133, 20914–20921

Journal of the American Chemical Society
ARTICLE
(
30) de Heer, W. A.; Bacsa, W. S.; Ch ^a telain, A.; Gerfin, T.;
Humphrey-Baker, R.; Forro, L.; Ugarte, D. Science 1995, 268, 845.
31) Endo, M.; Muramatsu, H.; Hayashi, T.; Kim, Y. A.; Terrones,
M.; Dresselhaus, M. S. Nature 2005, 433, 476.
32) Yuan, J.; Liu, X.; Akbulut, O.; Hu, J.; Suib, S. L.; Kong, J.;
Stellacci, F. Nat. Nanotechnol. 2008, 3, 332.
33) Holmberg, V. C.; Patel, R. N.; Korgel, B. A. J. Mater. Res. 2011,
6, 2305.
34) Choi, J. W.; Hu, L.; Cui, L.; McDonough, J. R.; Cui, Y. J. Power
Sources 2010, 195, 8311.
35) Hu, L.; Wu, H.; La Mantia, F.; Yang, Y.; Cui, Y. ACS Nano 2010,
, 5843.
36) Ng, S. H.; Wang, J.; Guo, Z. P.; Chen, J.; Wang, G. X.; Liu, H. K.
Electrochim. Acta 2005, 51, 23.
37) Wang, J.-Z.; Zhong, C.; Chou, S.-L.; Liu, H.-K. Electrochem.
Commun. 2010, 12, 1467.
(
(
(
2
(
(
4
(
(
(
(
38) Tuan, H.-Y.; Korgel, B. A. Chem. Mater. 2008, 20, 1239.
39) Saunders, A. E.; Sigman, M. B.; Korgel, B. A. J. Phys. Chem. B
2
004, 108, 193.
40) Choi, H. S.; Park, C. R. Towards High Performance Anodes
(
with Fast Charge/Discharge Rate for LIB Based Electrical Vehicles.
http://www.intechopen.com/articles/show/title/towards-high-perfor-
mance-anodes-with-fast-charge-discharge-rate-for-lib-based-electrical-
vehicles (accessed Sept 2, 2011).
(41) Obrovac, M. N.; Krause, L. J. J. Electrochem. Soc. 2007,
154, A103.
(42) Lopez, F. J.; Hemesath, E. R.; Lauhon, L. J. Nano Lett. 2009,
9, 2774.
(
43) Davidson, F. M.; Lee, D. C.; Fanfair, D. D.; Korgel, B. A. J. Phys.
Chem. C 2007, 111, 2929.
44) Wang, R.-p.; Zhou, G.-w.; Liu, Y.-l.; Pan, S.-h.; Zhang, H.-z.; Yu,
D.-p.; Zhang, Z. Phys. Rev. B 2000, 61, 16827.
45) Hessel, C. M.; Rasch, M. R.; Hueso, J. L.; Goodfellow, B. W.;
Akhavan, V. A.; Puvanakrishnan, P.; Tunnel, J. W.; Korgel, B. A. Small
(
(
2
010, 6, 2026.
(
(
46) Tuinstra, F.; Koenig, J. L. J. Chem. Phys. 1970, 53, 1126.
47) Lespade, P.; Al-Jishi, R.; Dresselhaus, M. S. Carbon 1982,
20, 427.
(
(
(
48) Ferrari, A. C.; Robertson, J. Phys. Rev. B 2000, 61, 14095.
49) Ferrari, A. C. Solid State Commun. 2007, 143, 47.
50) Monika, W.; Wang, Y.; Zerda, T. W. J. Phys.: Condens. Matter
2
005, 17, 2387.
51) Timmons, A.; Todd, A. D. W.; Mead, S. D.; Carey, G. H.;
Sanderson, R. J.; Mar, R. E.; Dahn, J. R. J. Electrochem. Soc. 2007,
54, A865.
(
1
2
0921
dx.doi.org/10.1021/ja208232h |J. Am. Chem. Soc. 2011, 133, 20914–20921