Tin nanoparticle loaded graphite anodes for Li-ion battery applications

Article abstract of DOI:10.1149/1.1799491

Nearly monosized Sn nanoparticles were produced by an in situ prepared single-source molecular precursor approach. The experimental conditions in the NaBH₄ reduction of (phen)SnCl₄ (phen = 1, 10 phenanthroline) in water were carefully controlled to produce two different particle size ranges, 2-5 nm (mean: 3.5 nm, standard deviation: 0.8 nm) and 7-13 nm (mean: 10.0 nm, standard deviation: 1.7 nm). The Sn nanoparticles were subsequently dispersed in graphite (KS6) and the application of the resulting nanocomposites as an active anode material for Li-ion batteries was explored. The graphite-Sn nanocomposites showed significant improvement in the cyclability of Sn over previously reported results. The cyclability improvement is believed to be due to the smallness of the Sn particles and their uniform distribution in a soft matrix (graphite) which, in addition to being a capable Li⁺ host, could also effectively buffer the specific volume changes in Sn-based Li storage compounds during charging (Li⁺ insertion) and discharging (Li⁺ extraction) reactions.

Full text of DOI:10.1149/1.1799491

A1804
Journal of The Electrochemical Society, 151 ͑11͒ A1804-A1809 ͑2004͒
0
013-4651/2004/151͑11͒/A1804/6/$7.00 © The Electrochemical Society, Inc.
Tin Nanoparticle Loaded Graphite Anodes for Li-Ion Battery
Applications
a,
a,b, ,z
b,
Yong Wang, * Jim Y. Lee, ** and Theivanayagam C. Deivaraj **
aDepartment of Chemical and Biomolecular Engineering, National University of Singapore 119260,
Singapore
b
Singapore-MIT Alliance, National University of Singapore 117576, Singapore
Nearly monosized Sn nanoparticles were produced by an in situ prepared single-source molecular precursor approach. The
experimental conditions in the NaBH reduction of (phen͒SnCl (phen ϭ 1, 10 phenanthroline͒ in water were carefully controlled
4
4
to produce two different particle size ranges, 2-5 nm ͑mean: 3.5 nm, standard deviation: 0.8 nm͒ and 7-13 nm ͑mean: 10.0 nm,
standard deviation: 1.7 nm͒. The Sn nanoparticles were subsequently dispersed in graphite ͑KS6͒ and the application of the
resulting nanocomposites as an active anode material for Li-ion batteries was explored. The graphite-Sn nanocomposites showed
significant improvement in the cyclability of Sn over previously reported results. The cyclability improvement is believed to be
due to the smallness of the Sn particles and their uniform distribution in a soft matrix ͑graphite͒ which, in addition to being a
ϩ
capable Li host, could also effectively buffer the specific volume changes in Sn-based Li storage compounds during charging
ϩ
ϩ
(
©
Li insertion͒ and discharging (Li extraction͒ reactions.
2004 The Electrochemical Society. ͓DOI: 10.1149/1.1799491͔ All rights reserved.
Manuscript submitted November 7, 2003; revised manuscript received April 6, 2004. Available electronically October 4, 2004.
The interest in Sn-based anodes for lithium ion batteries has not
sign of agglomeration throughout the experiments. The nanopar-
ticles were characterized by transmission electron microscopy
͑TEM͒, powder X-ray diffraction ͑XRD͒, UV-visible and Fourier
transform infrared ͑FTIR͒ spectroscopies, and X-ray photoelectron
spectroscopy ͑XPS͒. Sn-graphite nanocomposites were prepared by
introducing graphitic carbon during the nanoparticle synthesis. The
performance of the Sn-graphite nanocomposites as anodes in re-
chargeable Li batteries was evaluated by laboratory test cells. To the
best of our knowledge, this is the best effort in preparing the most
uniform and stable Sn nanoparticles. The electrochemical cyclability
of the Sn-graphite nanocomposites is also superior to most Sn-C or
been dampened by the often cited disadvantages of these
mateirals,1 namely, the stress-induced pulverization of the elec-
-5
ϩ
trode caused by the large specific volume changes in charging (Li
ϩ
insertion͒ and discharging (Li extraction͒ reactions. Recently Li
and Martin reported excellent cyclability and unprecedented rate
capability ͑Ͼ700 mAh/g at the 8 C rate͒ in a restricted potential
range ͑0.2-0.9 V͒ for SnO2 films with nanometer size micro-
6
structure. Such outstanding work aside, the prevailing view in the
battery community prefers to combine a moderate use of the high
capacity Li storage compounds such as Sn with graphite, the ubiq-
uitous anode material for Li-ion batteries. Such measures not only
address the material problems of the high capacity anodes, but also
alleviate the need to develop new processes for the conversion to a
completely new anode material. The dispersion of small ͑particu-
SnX-C composite materials (SnX ϭ SnO, SnO , or SnSb͒ reported
2
to date.
Experimental
larly nanoscale͒ particles of Sn, SnO, SnO , and SnSb in a carbon-
Chemicals.—Commercially available SnCl4 ͑Riedel-de Haen,
2
aceous matrix also allows operation in a more extended potential
99%͒, NaBH ͑Fluka͒, 1,10-phenanthroline ͑Nacalai Tesque͒, syn-
thetic graphite ͑Timcal KS6, 19 m /g͒, and methanol ͑Merck͒ were
4
range ͑ϳ5 mV-2.0 V͒.7 The particles must be stabilized or particle
-17
2
agglomeration would occur during applications to nullify the advan-
used as received without further purification. Deionized water was
purified by double distillation.
tages of small particles.1
5-17
The use of elemental Sn nanoparticles is
preferred over SnO nanoparticles as they can, in principle, elimi-
2
Synthesis of 10 nm Sn nanoparticles and Sn-KS6 nanocom-
nate the irreversible capacity loss associated with the decomposition
of tin compounds in the first lithiation cycle. However, the prepara-
tion of nanosized Sn particles is difficult, and there have been only
posites.—5 mL of 0.1 M SnCl aqueous solution was added to a
4
suspension of 0.1 g of 1,10-phenanthroline ͑phen͒ in 5 mL of water
͑
phen: SnCl ϭ 1:1). A clear solution was obtained after stirring
4
a few reports. The direct chemical reduction of SnCl often results
2
in particles bigger than 1 ␮m.18 Chaudret and co-workers have de-
scribed a method by which 15-25 nm tin particles were obtained
for a few minutes, indicating the formation of the coordination com-
2
2
pound (phen͒SnCl4 . 20 mL of 0.1 M NaBH4 aqueous solution
19
was then introduced dropwise (NaBH : (phen͒SnCl ϭ 4:1). The
from the thermolysis of ͓Sn͑NMe ) ] . The primary interest of
4
4
2
2 2
mixture turned red, indicating the formation of a sol of phen-capped
Sn nanoparticles. The hydrosol was further stirred for 2 h before the
Sn nanoparticles were spun down in an ultracentrifuge ͑15,000 rpm
for 1 h͒. The solid product recovered as such was washed with water
and methanol, and vacuum dried at 400 K. For the preparation of
Sn-graphite composites, a calculated amount of synthetic graphite
that work lay, however, with the oxidation of the Sn nanoparticles to
SnO nanoparticles for gas sensor applications. The physical vapor
2
deposition based techniques, while relatively more common, are
nevertheless an expensive alternative to the chemical preparation of
nanoscale Sn.20,21
This paper reports a very simple procedure for preparing Sn
nanoparticles of average particle sizes of 3.5 and 10.0 nm through
an in situ prepared single-source molecular precursor. 1,10-
phenanthroline ͑phen͒ was used to form a coordinating compound
KS6 was added to the (phen͒SnCl solution right at the start. The
4
suspension was sonicated for 1 h before the addition of the reducing
agent NaBH4 .
with SnCl which was subsequently reduced at room temperature to
4
Synthesis of 3.5 nm Sn nanoparticles and Sn-KS6 nanocom-
posites.—Smaller Sn nanoparticles could be obtained by adding 10
mL of 0.05 M SnCl instead of 5 mL of 0.1 M SnCl to the suspen-
sion of 0.1 g of phen in 10 mL of water. The color of the resulting
Sn hydrosol was pale yellow.
produce the nanoparticles. The chelating properties of phen are also
instrumental for controlling particle growth and agglomeration pro-
cesses. The Sn nanoparticles so prepared are stable and show no
4
4
Electrochemical measurements.—A N-methyl pyrrolidinone
NMP͒ slurry containing 80 wt % of the Sn-graphite powder ͑vari-
ous samples͒, 10 wt % each of conductive carbon black ͑Super-P͒
*
Electrochemical Society Student Member.
͑
*
* Electrochemical Society Active Member.
z
E-mail: cheleejy@nus.edu.sg
Downloaded on 2015-04-09 to IP 138.251.14.35 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, 151 ͑11͒ A1804-A1809 ͑2004͒
A1805
Figure 2. TEM images of 2-5 nm Sn nanoparticles at different magnification
scales.
Figure 1. TEM image of as-prepared 7-13 nm Sn nanoparticles ͑a͒ dispersed
in a single phen matrix and ͑b͒ after the removal of the phen matrix.
reducing agent (NaBH ) rapidly changes the reaction environment,
4
and polyvinylidene fluoride ͑Aldrich͒ was uniformly applied to a
copper foil ͑0.7 mm͒. The coated foil was cut into disk electrodes 11
mm in diameter. The electrodes were vacuum dried overnight at 400
K followed by compression at 2.0 ϫ 10 Pa. They were then as-
sembled into Li test cells using 0.75 mm lithium foil as the negative
electrode, microporous polypropylene separators, and an electrolyte
making it difficult to control the nucleation and growth of the par-
ticles. However, in the presence of a complexing agent or a surface
passivating ligand ͑e.g., citrate ions which coordinate to the metal
6
Ϫ
surface through the COO group͒, 100-300 nm Sn particle could be
1
8
formed by this simple reductve process. A brief survey of the
literature shows that polyvinylpyrrolidone ͑PVP͒ is one of the most
popular surface passivating agent used in metal nanoparticle
of 1 M LiPF in a 50:50 w/w mixture of ethylene carbonate ͑EC͒
6
2
3,24
and diethyl carbonate ͑DEC͒. Cell assembly was carried out in a
recirculating argon glove box where both the moisture and oxygen
contents were below 1 ppm each. All cells were tested at a constant
preparations.
Our attempt to prepare Sn nanoparticles protected
by PVP was unsuccessful as the resulting nanoparticles showed
heavy agglomeration. Hence the use of 1,10-phenanthroline ͑phen͒,
a Lewis base, to inhibit the agglomeration of nanoparticles was ex-
plored. Unlike polymers, such as PVP which stabilizes through mul-
tisite adsorption of the polymer on the nanoparticles after the latter
2
ϩ
current density of 0.4 mA/cm and were charged (Li insertion͒ and
ϩ
discharged (Li extraction͒ between fixed voltage limits ͑5 mV-2 V͒
on a Maccor series 2000 battery tester.
are formed, phen is able to bond to Sn in the SnCl precursor stage
4
Materials characterizations.—Transmission electron microscope
22
as a coordinating complex. This should in principle reduce the
͑
TEM͒ images were obtained from a JEOL JEM-2010 microscope.
reactivity of Sn in chemical reactions, and a more homogenous re-
action environment for chemical reduction may then be established.
In addition, once the Sn particles are formed, they are instantly
coordinated to phen by the pair of chelating nitrogen donor sites
Powder XRD was carried out on a Siemens D6000 diffractometer.
X-ray photoelectron spectra ͑XPS͒ were collected by a KRATOS
AXIS His. UV-visible spectra were recorded by a Shimadzu UV-
3101 PC double-beam spectrophotometer. FTIR spectra were ob-
25
adjoining the two fused heterocyclic aromatic rings. The capping
tained on a BIO-RAD FTS 135 FTIR spectrophotometer using the
KBr pellet method.
of the Sn nanoparticles prevents further growth of the nanoparticles
and is expected to control the growth rate much more effectively
than PVP.
Figure 1a shows a typical TEM image of phen-capped Sn nano-
particles ͑average particle size of 10 nm͒ on a TEM grid, which had
been washed in deionized water and dried at 400 K in vacuum. A
large number of Sn nanoparticles of almost the same size were
Results and Discussion
Tin particles ͑larger than 1 ␮m͒ could be easily obtained by
18
adding NaBH to a SnCl aqueous solution. This process is not
4
2
suitable for producing nanoparticles because the addition of a strong
Downloaded on 2015-04-09 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A1806
Journal of The Electrochemical Society, 151 ͑11͒ A1804-A1809 ͑2004͒
Figure 3. Particle size distribution from TEM measurements for ͑a͒ 10 nm
Sn nanoparticles and ͑b͒ 3.5 nm Sn nanoparticles.
found embedded in a web of phen molecules, which coordinated
strongly with one other and with the Sn nanoparticles possibly
through surface adsorption. After the TEM grid was washed with
methanol, in which phen is readily soluble, the phen domains dis-
appeared and only the nanoparticles were left behind ͑Fig. 1b͒. Mea-
surements based on these TEM images revealed that the Sn nano-
particles were nearly monosized with an average particle size of
Figure 5. XPS of phen-capped 10 nm Sn particles in the ͑a, top͒ N 1s region
and ͑b, bottom͒ the Sn 3d5/2 region.
10.0 nm and a standard deviation of 1.7 nm, and were well separated
from one another.
Sn nanoparticles of even smaller sizes and a narrower size dis-
tribution were obtained by using a more dilute solution of
(
phen͒SnCl . The TEM images in Fig. 2a and b show nanoparticles
4
with a mean particle size of 3.5 nm and a standard deviation of 0.8
nm. The nanoparticles were uniformly distributed without any no-
ticeable agglomeration both inside and outside of the phen domains.
A histogram of particle size distribution based on TEM measure-
ments is given in Fig. 3 for the two batches of Sn nanoparticles of
average particle sizes of 10 and 3.5 nm, respectively. The statistics
were generated by counting approximately 250 particles. The ex-
perimental observation of a lower (phen͒SnCl concentration aiding
4
in the formation of smaller and more uniform Sn nanoparticles cor-
roborates the slow kinetics in the particle formation process de-
scribed earlier.
In contrast to the procedure in Ref. 19, the approach mentioned
in this study does not require the handling of air-sensitive and un-
stable compounds such as ͓Sn͑NMe ) ] . The (phen͒SnCl precur-
2
2
2
4
sor can be easily synthesized in situ. Its stability in air makes it
possible to produce Sn nanoparticles under the ambient conditions
in an aqueous system. It should be emphasized that compared to the
Figure 4. UV/visible absorption spectra of ͑a͒ SnCl4 , ͑b͒ phen, ͑c͒
(
phen͒SnCl , ͑d͒ phen-capped Sn nanoparticles, ͑e͒ 10 nm Sn particles, and
4
1
9
͑f͒ standard ͑100 nm͒ Sn particles.
previous effort, smaller Sn nanoparticles with a narrower size dis-
Downloaded on 2015-04-09 to IP 138.251.14.35 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, 151 ͑11͒ A1804-A1809 ͑2004͒
A1807
tribution are obtained here, and with almost no particle agglomera-
tion.
Figure 4 shows the absorption spectra of ͑a͒ SnCl , ͑b͒ phen, ͑c͒
4
(
phen͒SnCl , ͑d͒ phen stabilized Sn nanoparticles, ͑e͒ pristine Sn
4
nanoparticles, and ͑f͒ commercially available Sn nanoparticles ͑100
2
2
nm, Aldrich͒. The absorption maxima of (phen͒SnCl precursor,
4
was located at a higher wavelength compared to SnCl and phen.
4
Upon the addition of NaBH , the (phen͒SnCl4 peak at 270 nm
4
4
ϩ
0
disappeared signalling the complete reduction from Sn to Sn . In
addition, UV-visible spectra of obtained Sn nanoparticles showed
mostly the UV absorption features of phen, which is taken to indi-
cate the presence of remnants of phen on the Sn nanoparticle sur-
face. UV-visible spectra were also collected after washing the phen-
stabilized Sn nanoparticles with methanol and the results were
indistinguishable from the commercially available Sn nanoparticles,
and without any UV-visible signature of phen. FTIR measurements
also agreed well with these results. The coordination compound
(
1
phen͒SnCl exhibited a number of sharp peaks ͑1590, 1539, 1428,
4
Ϫ1
150, 856, 717 cm ͒ which could be attributed to the phen ligand.
Ϫ1
Peaks at 1559, 1536, 1424, 852 cm remained visible for the
phen-Sn nanoparticles before washing, indicating again the presence
of phen on the nanoparticle surface. Comparable FTIR in the spectra
of (phen͒SnCl4 and phen-stabilized Sn were shifted to slightly
higher wavenumbers relative to phen, possibly as a result of the
2
5
coordination of phen to SnCl or to the Sn nanoparticles.
4
The presence of the N 1s signal in the XPS spectrum of the
prepared Sn nanoparticles further confirmed the adsorption of phen
on the nanoparticle surface ͑Fig. 5a͒. Combining the results from
TEM imaging, UV-visible spectroscopy and FTIR spectroscopy, it
could be postulated that the phen molecules coordinated strongly
with Sn to provide an extensive surface coverage on the Sn nano-
particles. A similar effect of strongly coordinating ligands was pre-
2
6
viously observed for phen-stabilized Pd nanoparticles.
Based on these observations, we propose that the role of 1,10-
phenanthroline as follows ͑Scheme I͒: phen promotes the formation
of the coordination compound (phen͒SnCl4 ,22 The formation of a
coordination compound between the ligand and the metal at the
precursor level ͑phen and SnCl in this case͒ could potentially offer
4
better control in the nucleation and growth of the nanoparticles. This
advantage was largely overlooked in the current procedures for pre-
paring ligand-stabilized nanoparticles. For example, PVP-protected
2
4
Ag nanoparticles were prepared by the reduction of AgClO4 . As
there is no affinity between PVP and AgClO at the precursor level,
4
there is minimal control in nucleation besides the steric effects of
PVP. Particle growth control that depends on the interactions be-
tween PVP and the nanoparticles only comes in after the nanopar-
ticles are formed.
The Sn 3d peak in the XPS spectrum of the Sn nanoparticles
͑
Fig. 5b͒ appeared as a doublet with a separation of 8.5 eV. Decon-
Figure 6. TEM images of ͑a͒ pristine KS6, ͑b͒ 18.7 wt % Sn ͑10 nm͒-KS6
nanocomposites, and ͑c͒ 19.8 wt % Sn ͑3.5 nm͒-KS6 nanocomposites.
volution of the Sn 3d5/2 signal indicated the presence of two species
with binding energies of 484.7 and 486.4 eV in the ratio of 1:6.7,
most likely due to Sn and SnO or SnO , respectively ͑SnO and
2
19
SnO2 have experimentally indistinguishable binding energies͒.
This is an indication of the surface oxidation of the Sn nanoparticles.
It is indeed common for metal nanoparticles to form a metal oxide
shell under ambient conditions. A similar observation was also re-
Scheme 1. Proposed reaction scheme for the preparation of phen-capped Sn
nanoparticles.
Downloaded on 2015-04-09 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A1808
Journal of The Electrochemical Society, 151 ͑11͒ A1804-A1809 ͑2004͒
Figure 8. Cyclability of various Sn-modified graphite composite anodes ͑0.4
2
mA/cm , 5 mV-2V͒.
Figure 7. The first cycle lithiation and delithiation of graphite and 9.5 wt %
Sn ͑10 nm͒-graphite.
ϩ
and 2V vs. Li/Li . Discharge and recharge capacities of 618 and 402
mAh/g, respectively, were obtained in the first cycle, resulting in a
ported for the Sn nanoparticles obtained in Ref. 19. The surface
6
5% coulombic efficiency. This value is an improvement over our
oxide is likely to be SnO or SnO , which hinders further oxidation.
16,17
2
previous results ͑ϳ60-62%͒ of SnO -KS6 nanocomposites.
2
SnO nanoparticles, a wide bandgap n-type semiconductor material,
2
When SnO is involved, there is an additional mechanism for irre-
2
could be formed after prolonged heating of the Sn nanoparticles.
versible capacity loss due to the decomposition of SnO into Li O
2
2
Selected area electron diffraction ͑SAED͒ of the Sn nanoaprticles
and Sn at 1 V and above in the first lithiation cycle as shown in Eq.
below
͑
inset of Fig. 1b͒ showed the characteristic ring pattern of an amor-
1
phous material. XRD also confirmed that the Sn nanoparticles were
X-ray amorphous even after heating to 450 K in vacuum ͑the melt-
ing point of Sn is 505 K͒. Amorphous Sn has been found to have
Ϫ
SnO2 ϩ 4Li ϩ 4e → 2Li2O ϩ Sn
͓1͔
2
7
ϩ
better cyclability compared to the crystalline polymorph.
Subsequent to the metallic Sn formation reaction, Li insertion
Figure 6a shows the TEM image of pristine KS6. Synthetic
graphite KS6 with a flake size of 1-5 ␮m is a reasonably good
electron conductor and a good Li host ͑only 6% volume change in
Li-ion intercalation and deintercalation reactions͒, and hence its ad-
dition to Sn would not diminish the overall capacity in the way an
active-inactive domain system would.7 The TEM images in Fig. 6b
and c show, respectively, the uniform distribution of 18.7 wt % Sn
and extraction reactions would then proceed reversibly via Eq. 2
Ϫ
Sn ϩ xLi ϩ xe ꢀ Li Sn ͑x
ϭ 4.4͒
max
͓2͔
x
If Sn is used as the Li storage compound, the first reaction is not
expected to occur and the coulombic efficiency should be, in prin-
ciple, increased. The Sn nanoparticles in the Sn-graphite composites
actually had a core shell structure of the type Sn-SnO or Sn-SnO2
because of the facile oxidation of the Sn surface. Hence the irrevers-
ible capacity loss due to decomposition of tin oxides could not be
completely eliminated. The core-shell structure, on the other hand,
could be of special interest and advantageous to lithium-ion battery
applications. This is because the Li2O formed in the decomposition
,8
͑
10 nm͒ and 19.8 wt % Sn ͑3.5 nm͒ nanoparticles on the graphite
surface for the Sn-graphite nanocomposites prepared in the way de-
scribed in the experimental section. A large number of Sn nanopar-
ticles were located on the surface of KS6, the edges of the graphite
flakes could be seen clearly and no nanoparticles were found outside
of graphite. The uniform dispersion of Sn without aggregation led us
to hypothesize that there are active sites on the graphite surface,
which help to bind and stabilize the nanoparticles.
of SnO or SnO in the first lithiation cycle is known to help restrain
2
5
the pulverization of Sn particles during applications. Hence a mod-
Elemental analysis by inductively coupled plasma spectroscopy
ICPS͒ showed Sn ͑3.5 nm͒ loadings of 10.3 wt %, 19.8 wt %, and
Sn ͑10 nm͒ loadings of 9.5 wt % and 18.7 wt % in four samples of
erate presence of SnO or SnO would benefit from the Li O exis-
tence without an unduly large irreversible capacity loss.
2
2
͑
The cycling performance of a series of Sn-KS6 composites ͑10.3
and 19.8 wt % of 3.5 nm Sn, and 9.5 and 18.7 wt % of 10 nm Sn͒
is compared in Fig. 8. Generally, an increase in the amount or the
size of the Sn nanoparticles led to poorer electrode cyclability. For
the sample with 10.3 wt % of Sn ͑3.5 nm͒, discharge capacities of
415 and 379 mAh/g were obtained in the first and the 60th cycle,
(
͑
phen͒SnCl -derived Sn-KS6 composites. The first cycle charging
4
Li insertion͒ and discharging ͑Li extraction͒ curves of 9.5 wt %
Sn͑10 nm͒-KS6 composite and pristine graphite ͑KS6͒ are shown in
Fig. 7. Cycling was performed at a constant current density of 0.4
2
mA/cm ͑approximating the C/6 rate͒ with cutoff potentials at 5 mV
Table I. Cyclability of various Sn-graphite and SnO -graphite nanocomposites.
2
Capacity retained at cycles ͑%͒
Initial specific capacity
Initial efficiency
Samples
͑mAh/g͒
͑%͒
10
30
40
60
1
6
9
1
9
1
1
1
9
1
.8 wt % SnO ͑17 nm͒
384
428
404
462
415
442
402
434
300
61.5
60.5
62.5
59.7
65.5
63.8
65.0
64.2
74.1
99.4
99.2
98.4
97.2
99.7
97.7
98.7
97.9
96.3
90.1
90.7
83.6
97.3
90.5
93.8
89.2
99.1
-
94.7
82.5
82.3
72.3
95.1
85.9
90.0
76.0
98.3
-
-
-
-
2
1
6
6.5 wt % SnO ͑17 nm͒
2
1
7
.2 wt % SnO ͑5 nm͒
2
1
7
4.5 wt % SnO ͑5 nm͒
-
2
0.3 wt % Sn ͑3.5 nm͒
9.8 wt % Sn ͑3.5 nm͒
.5 wt % Sn ͑10 nm͒
8.7 wt % Sn ͑10 nm͒
91.3
76.9
85.8
64.9
93.8
-
Pristine KS6
Pristine Sn or SnO2
100
-
ϳ600-700
ϳ40-60
Downloaded on 2015-04-09 to IP 138.251.14.35 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, 151 ͑11͒ A1804-A1809 ͑2004͒
respectively. In other words, 91.3% of the initial capacity could be Conclusion
A1809
retained after 60 cycles, which was only marginally lower than the
retention value of pristine KS6 ͑93.8%͒ in the same number of
cycles. We are not aware of any other report of Sn-C composites
showing such high cyclability. The good cyclability of Sn-KS6
nanocomposites prepared by the current method is believed to be a
combination of the uniform distribution of nonagglomerating small
Sn particles on the graphite surface and the successful control of the
volume changes in Li-Sn reactions.
In conclusion, Sn nanoparticles about 10 and 3.5 nm in size and
Sn-graphite nanocomposites were prepared from an in situ formed
(
phen͒SnCl precursor. The phen ligand stabilizes the nanoparticles
4
against agglomeration during preparation and the dispersion of the
nanoparticles in graphitic carbon inhibits particle agglomeration dur-
ing applications. In principle, phen could also be useful as a protec-
tive agent in the preparation of other metal nanoparticles. The 10.3
wt % 3.5 nm Sn-graphite nanocomposite prepared as such displayed
The Sn-KS6 composites showed distinctively better performance
than what was expected from a physical mixture of two independent
components. For example, the discharge capacity ͑415 mAh/g͒ of
ϩ
a high reversible Li storage capacity of 415 mAh/g, of which
91.3% could be retained after 60 charge and discharge cycles. This
is a significant improvement over previous efforts on Sn-based
graphite composites reported in the literature. Particle size and par-
ticle size distribution are both very important factors determining
the applicability of Sn-based nanoparticles in Li-ion batteries.
1
0.3 wt % Sn-KS6 is higher than the weighted sum of the discharge
16
capacities of graphite ͑300 mAh/g͒ and theoretical capacity of Sn
1
5
͑
990 mAh/g͒ ͑usually the realizable discharge capacities of Sn or
SnO2 are around 600-700 mAh/g͒.9 It is, however, not clear
whether one could ͑or should͒ partition the observed nanocomposite
capacities between their constituents. As a first approximation, the
small capacity loss after 60 cycles could be attributed to mostly Sn
nanoparticles. The agglomeration of Sn nanoparticles appears to be
,18
The Singapore-MIT alliance, National University of Singapore, assisted
in meeting the publication costs of this article.
References
2
8,29
an inevitable event after prolonged cycling of the electrode.
1
2
. Y. Idota, A. Matsufuji, Y. Mackawa, and T. Miyasaki, Science, 276, 1395 ͑1997͒.
. O. Mao, R. L. Turner, I. A. Courtney, B. D. Fredericksen, M. I. Buckett, L. J.
Krause, and J. R. Dahn, Electrochem. Solid-State Lett., 2, 3 ͑1999͒.
Relative to the 10.3 wt % Sn ͑3.5 nm͒ sample, the 19.8 wt % Sn
3.5 nm͒-KS6 composite had an expectedly higher capacity ͑442
͑
mAh/g͒ due to the increase in the tin content, but capacity fading
was also more pronounced with 23.1% of the initial capacity lost in
3
. I. A. Courtney, W. R. McKinnon, and J. R. Dahn, J. Electrochem. Soc., 146, 59
͑
1999͒.
6
0 cycles. This was most likely due to the increase in the rate and
4. A. Ulus, Y. Rosenberg, L. Burstein, and E. Peled, J. Electrochem. Soc., 149, A635
͑2002͒.
the extent of Sn particle agglomeration in composites with higher tin
contents. For the composites containing 9.5 and 18.7 wt % of 10 nm
Sn, the initial capacities were 402 and 434 mAh/g, respectively, 85.8
and 64.9% of which could be retained after 60 cycles. The stronger
capacity fading in 10 nm Sn nanocomposites clearly indicates the
advantage of using smaller nanoparticles. In addition, the 3.5 nm Sn
nanoparticles were also more monodispersed than the 10 nm Sn
nanoparticles. Therefore a uniform distribution of smaller Sn nano-
particles would lead to a more homogenous volume change during
5
. J. Yang, Y. Takeda, N. Imanishi, and O. Yamamoto, J. Electrochem. Soc., 146, 4009
1999͒.
͑
6
7
8
. N. Li and C. Martin, J. Electrochem. Soc., 148, A164 ͑2001͒.
. J. Y. Lee, R. F. Zhang, and Z. L. Liu, Electrochem. Solid-State Lett., 3, 167 ͑2000͒.
. J. Y. Lee, R. F. Zhang, and Z. L. Liu, J. Power Sources, 90, 70 ͑2000͒.
9. J. Read, D. Foster, J. Wolfenstine, and W. Behl, J. Power Sources, 96, 277 ͑2001͒.
10. G. X. Wang, J. H. Ahn, M. J. Lindsay, L. Sun, D. H. Bradhurst, S. X. Dou, and H.
K. Liu, J. Power Sources, 97-98, 211 ͑2001͒.
1
1. M. Egashira, H. Takatsuji, S. Okada, and J. Yamaki, J. Power Sources, 107, 56
2002͒.
͑
ϩ
Li insertion and extraction reactions, which is easier for graphite to
buffer.
1
2. J. K. Lee, D. H. Ryu, J. B. Ju, Y. G. Shul, B. W. Cho, and D. Park, J. Power
Sources, 107, 90 ͑2002͒.
13. B. Veeraraghavan, A. Durairajan, B. Haran, B. Popov, and R. Guidotti, J. Electro-
chem. Soc., 149, A675 ͑2002͒.
4. Y. Liu, J. Y. Xie, Y. Takeda, and J. Yang, J. Appl. Electrochem., 32, 687 ͑2002͒.
5. H. Li, Q. Wang, L. H. Shi, L. Q. Chen, and X. J. Huang, Chem. Mater., 14, 103
Table I compares the electrochemical properties of the Sn-KS6
16,17
prepared here with our previous work
on SnO -KS6 nanocom-
2
1
1
posites prepared by a variety of different methods. The following
conclusions can be drawn from such a comparison: ͑i͒ Sn-graphite
nanocomposites generally show the expected higher coulombic ef-
ficiencies than SnO -graphite nanocomposites, (ii) all samples with
lower Sn or SnO contents show improved cyclabilities than those
with higher contents. Capacity fading becomes noticeable when the
Sn or SnO content is increased from 9 to around 16 wt %, indicat-
ing that the range for optimal Sn or SnO loading is rather limited.
Even when Sn or SnO particles are small and uniformly distributed,
͑
2002͒.
16. Y. Wang, J. Y. Lee, and B. H. Chen, Electrochem. Solid-State Lett., 6, A19 ͑2003͒.
1
1
7. Y. Wang and J. Y. Lee, Electrochem. Commun., 5, 292 ͑2003͒.
8. J. Yang, M. Wachtler, M. Winter, and J. O. Besenhard, Electrochem. Solid-State
Lett., 2, 161 ͑1999͒.
9. C. Nayral, T. Ould-Ely, A. Maisonnat, B. Chaudret, P. Fau, L. Lescouzere, and A.
Peyre-Lavigne, Adv. Mater. (Weinheim, Ger.), 11, 61 ͑1999͒.
2
2
1
2
20. A. Stella, M. Nisoli, S. De Silvestri, O. Svelto, G. Lanzani, P. Cheyssac, and R.
2
Kofman, Phys. Rev. B, 53, 15497 ͑1996͒.
1. T. Bachels and H. J. Guntherodt, Phys. Rev. Lett., 85, 1250 ͑2000͒.
2. V. J. Hall and E. R. T. Tiekink, Z. Kristallogr., 211, 47 ͑1996͒.
2
2
2
a large number these particles would not be able to deliver high
capacity concurrently with good cyclability. (iii) The table also
23. J. S. Bradley, J. M. Millar, and E. W. Hill, J. Am. Chem. Soc., 113, 4106 ͑1991͒.
24. I. P. Santos and L. M. L. Marzan, Langmuir, 18, 2888 ͑2002͒.
shows that the 5 nm SnO nanoparticles actually displays more se-
2
2
2
5. T. Trindade, P. O’Brien, and X. M. Zhang, Chem. Mater., 9, 523 ͑1997͒.
6. N. Toshima, Y. Shiraishi, T. Teranishi, M. Miyake, T. Tominaga, H. Watanabe, W.
Brijoux, H. Bonneemann, and G. Schmid, Appl. Organomet. Chem., 15, 178
vere capacity fading than samples containing larger SnO ͑17 nm͒
2
and Sn ͑10 nm͒ particles. The particles size alone is therefore a
necessary but insufficient condition to assure good cyclability. It is
rather the combination of small particle size, uniform particle size
distribution, and low particle concentration that is important to cy-
clability preservation. The 10.3 wt % Sn ͑3.5 nm͒ nanocomposite
excelled because it possessed all of these desirable attributes. This
should not come as a surprise as it is common knowledge that the
properties of nanoparticle systems are strongly dependent on both
particle size and particle size distribution.30
͑
2001͒.
27. J. J. Zhu, Z. H. Lu, S. T. Aruna, D. Aurbach, and A. Gedanken, Chem. Mater., 12,
2557 ͑2000͒.
2
8. H. Li, X. J. Huang, L. Q. Chen, Z. G. Wu, and Y. Liang, Electrochem. Solid-State
Lett., 2, 547 ͑1999͒.
2
9. H. Li, L. H. Shi, W. Lu, X. J. Huang, and L. Q. Chen, J. Electrochem. Soc., 148,
A915 ͑2001͒.
30. I. S. Armadi, Z. L. Wang, T. C. Green, A. Henglein, and M. A. El-Sayed, Science,
272, 1924 ͑1996͒.
Downloaded on 2015-04-09 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).