Thin films of Cu2Sb and Cu9Sb2 as anode materials in Li-ion batteries

Article abstract of DOI:10.1016/j.electacta.2008.05.005

Thin Cu₂Sb films have been prepared by heat-treating Sb films, electrodeposited on Cu substrates. The influence of the electrodeposition conditions and the heat-treatment period on composition and morphology of the films were investigated (SEM and XRD) and the obtained films were tested as anode materials for Li-ion batteries. The Cu₂Sb material showed a stable capacity of 290 mAh g^-1 (close to the theoretical capacity of 323 mAh g^-1) during more than 60 cycles. The presence of 9-11% (w/w) Sb₂O₃ in the electrodeposited films resulted in smaller particles but also slowed down formation of Cu₂Sb during the heat-treatment step. The presence of Sb₂O₃ was found to decrease the cycling stability although structural reversibility of Cu₂Sb was obtained both with and without Sb₂O₃. Longer heat-treatments of pure Sb films resulted in the formation of Cu₉Sb₂ which was shown to be reduced at a lower potential than Cu₂Sb. The Cu₉Sb₂ was converted to Cu₂Sb during repeated cycling and the capacity of the latter Cu₂Sb material was found to be 230 mAh g^-1. While reduction of the materials was complicated by simultaneous formation of an SEI layer, three plateaus could be identified during the oxidation of Li₃Sb, indicating the presence of three separate one-electron oxidation reactions.

Full text of DOI:10.1016/j.electacta.2008.05.005

Electrochimica Acta 53 (2008) 7226–7234
Contents lists available at ScienceDirect
Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta
Thin films of Cu Sb and Cu Sb as anode materials in Li-ion batteries
2
9
2
1
∗,1
Hanna Bryngelsson, Jonas Eskhult, Leif Nyholm , Kristina Edstr o¨ m
Department of Materials Chemistry, The A˚ ngstr o¨ m Laboratory, Uppsala University, Box 538, SE-751 21 Uppsala, Sweden
a r t i c l e i n f o
a b s t r a c t
Article history:
Thin Cu Sb films have been prepared by heat-treating Sb films, electrodeposited on Cu substrates. The
influence of the electrodeposition conditions and the heat-treatment period on composition and morphol-
2
Received 26 February 2008
Received in revised form 2 May 2008
Accepted 5 May 2008
ogy of the films were investigated (SEM and XRD) and the obtained films were tested as anode materials
−1
for Li-ion batteries. The Cu2Sb material showed a stable capacity of 290 mAh g (close to the theoret-
Available online 9 May 2008
−1
ical capacity of 323 mAh g ) during more than 60 cycles. The presence of 9–11% (w/w) Sb2O3 in the
electrodeposited films resulted in smaller particles but also slowed down formation of Cu2Sb during the
heat-treatment step. The presence of Sb2O3 was found to decrease the cycling stability although structural
reversibility of Cu2Sb was obtained both with and without Sb2O3. Longer heat-treatments of pure Sb films
resulted in the formation of Cu9Sb2 which was shown to be reduced at a lower potential than Cu2Sb. The
Cu9Sb2 was converted to Cu2Sb during repeated cycling and the capacity of the latter Cu2Sb material was
Keywords:
Li-ion batteries
Anode material
Cu2Sb
Electrodeposition
Cu9Sb
−1
found to be 230 mAh g . While reduction of the materials was complicated by simultaneous formation
of an SEI layer, three plateaus could be identified during the oxidation of Li3Sb, indicating the presence of
three separate one-electron oxidation reactions.
©
2008 Elsevier Ltd. All rights reserved.
1
. Introduction
out to study how to circumvent this problem [3] using for instance,
metal oxides [3–10] where the oxide serves as a buffering matrix
during the lithiation process, or by utilising nanoparticles to mini-
mize the effects of the volume changes [11–13]. Other approaches
involve embedding the active electrode material in a composite
matrix [14–16], the use of amorphous alloys [17–19], or intermetal-
lic systems (AxBy) which have a strong structural relationship to
that of their lithiated products [20–22]. In these materials, both
metals can either be active with respect to lithium (see e.g. AlSb
[23,24], SnSb [25], and InSb [22,24,26]) or one can be active and one
inactive (see e.g. Cu Sb [27], MnSb, Mn Sb [28,29], Cu Sn5 [20,21]).
In the latter case, it has been shown that element A is extruded
from the structure during the lithiation and readmitted during the
delithiation process.
Sb-compounds exhibit high capabilities of storing lithium and
therefore constitute potential anode materials for Li-ion batteries.
A maximum of three Li-ions can be stored per antimony atom,
Li-ion battery technology is today widely and successfully used
in small portable electronics and a broad knowledge of the charac-
teristics of this battery-type is currently emerging. However for an
upscale of the battery for, for instance, the hybrid electrical vehi-
cle market there is still a need for cheaper, safer and more durable
electrode materials. This requires that additional attention is paid
to parameters that are known to influence the Li-ion battery per-
formance, such as the morphology, particle size, and composition
of the electrode materials. The design of new safe anode mate-
rials with good electrochemical performance, i.e. high capacities,
low irreversible capacities, high power densities and stable cycling,
require a deeper understanding of how the latter parameters affect
the lithium insertion/extraction mechanisms. This is not an easy
task as the chemical reactions involved in the lithiation process, as
well as in the electrode/electrolyte interface reactions, clearly are
complex.
2
2
6
−
1
which leads to a theoretical capacity of 660 mAh g
(based on
−
+
Compounds alloying with lithium [1,2] such as Al, Sn, Sb, and Si
are intensively studied as anode material for Li-ion batteries due
to their high capability of storing lithium. Due to the significant
changes in the volume of these materials associated with the lithi-
ation and delithiation processes, much work is presently carried
the reaction Sb + 3e + 3Li = Li Sb) which should be compared to
3
−
1
372 mAh g for graphite used in today’s commercial batteries. The
expected volume expansion (ca. 150% [16]) during the lithiation
process, unfortunately generally leads to a dramatic capacity loss
during cycling of this type of materials. As was recently shown
[
4,5] this problem can, however, be circumvented with materials
composed of a mixture of Sb and Sb O3 nanoparticles obtained by
2
electrodeposition from a solution containing antimony tartrate. As
∗ Corresponding author. Tel.: +46 18 471 3713; fax: +46 18 51 35 48.
E-mail address: kristina.edstrom@mkem.uu.se (K. Edstr o¨ m).
has been shown by several groups [27,30–34], Cu Sb can likewise be
2
1
ISE member.
used as anode material in Li-ion batteries. According to Fransson et
0
013-4686/$ – see front matter © 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2008.05.005

H. Bryngelsson et al. / Electrochimica Acta 53 (2008) 7226–7234
7227
al. [27] Cu Sb alloys with lithium in a displacement reaction during
The present study focuses on the electrochemical properties
of Cu Sb and Cu Sb films obtained by heat-treating electrode-
2
the reduction of Cu Sb as outlined in reactions (1) and (2) below.
2
2
9
2
As is evident from these reactions, copper is re-admitted into the
structure again during subsequent charging, thus giving rise to a
posited Sb films on copper substrates in an inert atmosphere.
Sb and Sb/Sb O3 films are electrodeposited employing the same
2
reformation of Cu Sb. Cu is extruded from the structure via a solid
solution extrusion to the formation of an intermediate phase (Eq.
conditions previously used [4,5] for the deposition on Ni sub-
strates and the influence of the electrodeposition conditions on
2
(
1)) and again a solid solution extrusion to the end product (Eq. (2))
the morphology and Sb O3 content of the films are discussed.
2
during discharge,
The Cu Sb films, which are characterized using scanning electron
2
microscopy (SEM) and X-ray diffraction (XRD), are also used as
anode materials in Li-ion batteries to study the cycling stability
Cu Sb + 2Li ↔ Li CuSb + Cu
(1)
(2)
2
2
Li CuSb + Li ↔ Li Sb + Cu
and structural reversibility as a function of the particle size, Sb O3
2
3
2
content of the Sb films and the duration of the heat-treatment
step. Electrochemical quartz crystal microbalance (EQCM) exper-
iments and electrochemical testing of the batteries are employed
The gravimetric- and volumetric capacities for Cu Sb are
2
−1
−3
3
23 mAh g
and 2749 mAh cm , respectively, and it has been
shown that Cu Sb can deliver stable gravimetric capacities of
2
to investigate the influence of the substrate on the Sb O3 con-
−1
2
2
90 mAh g for more than 25 cycles [27]. The reversibility of reac-
tent of the electrodeposited Sb film. The Cu Sb lithiation and
2
tion (1), however, depends strongly on the composition and particle
size of the material as was shown by Morcrette et al. [32] who stud-
ied the influence of the electrode microstructure on the reactivity
delithiation steps are discussed based on Li-ion battery measure-
ments in the absence of binders and additives such as carbon
black.
of Cu Sb with lithium and the overall reversibility of the reactions.
2
By comparing the reactivity of Cu Sb synthesized with two dif-
2
ferent techniques, i.e. ball-milling and a ceramic method, it was
shown that structural reversibility was obtained when the crys-
tallite size was sufficiently small (approximately 30 nm) and the
powders were ball-milled with carbon. It was, however, not possi-
ble to distinguish between the influence of the particle size and that
2. Experimental
2.1. Electrodeposition of Sb and Sb/Sb2O3 and synthesis of Cu2Sb
The electrodeposition of Sb and Sb/Sb2O3 was carried out
employing solutions containing 0.075 M potassium antimony
tartrate trihydrate (K2(Sb2(C4H4O6)2)·3H2O) at pH 1.3 and 4.1,
respectively. The pH 1.3 electrolyte also contained 0.090 M KCl
and 0.110 M HCl (reagent grade chemicals from Sigma Aldrich
or Merck) while the pH 4.1 electrolyte was obtained merely by
dissolving potassium antimony tartrate trihydrate in deionised
water. The electrodepositions were made with pieces of Cu-foil
(Goodfellow, thickness 0.0185 mm, purity level 99.9%) as work-
ing electrodes. A platinum wire served as the counter electrode
while a saturated calomel electrode (SCE) was used as the refer-
ence electrode. The Cu-foil was generally etched with 50% HNO3
prior to the depositions. At pH 4.1, depositions were, however,
also carried out on Cu-foils simply cleaned in acetone in an
ultrasonic bath. A laboratory designed combined sample holder
and electrochemical cell consisting of a polyvinyl chloride (PVC)
cylinder, in which the Cu-foil served as the bottom section, was
used during the electrodepositions. The Cu-foil was placed on
a 3 mm thick copper plate, used as the current collector, on
top of a 10 mm thick PVC plate. The area of the working elec-
due to the presence of carbon in the materials. The Cu Sb synthe-
2
sized with the ceramic method which resulted in crystallite sizes
larger than 200 nm did, on the other hand, not show corresponding
structural reversibility. In addition, several reports have been pub-
lished [30–34] in which Cu Sb capacities significantly lower than
2
the capacity reported by Fransson et al., have been found, most
likely due to poor reversibility in reaction (1). The Cu Sb lithiation
2
and delithiation reaction mechanism has recently been discussed
by several groups [30,32,35] and a three-step process has been pro-
posed, in which the reaction (1) has been divided into two separate
steps. The lithiation mechanisms proposed by Morcrette et al. [32]
(
see reactions (3) and (4)) and Matsuno et al. [30] (see reactions (5)
and (6)) differ with respect to the extent of the extrusion of Cu in
the two steps;
Cu Sb + (2 − x)Li ↔ Li Cu1+xSb + (1 − x)Cu
(3)
(4)
(5)
(6)
2
2−x
Li2 Cu Sb + xLi ↔ Li CuSb + xCu
−x
1+x
2
Cu Sb + (2 − x)Li ↔ Li2−xCuSb + Cu
2
2
Li2 CuSb + xLi ↔ Li CuSb
trode, and hence the electrodeposited material, was 31.2 cm . From
−x
2
this original sample, smaller samples with an area of 3.14 cm2
were punched out for electrochemical characterization. The elec-
To better understand the factors influencing the lithium inser-
tion/extraction reactions in this type of materials, model systems
should be used in which there are as few interfering reactions as
possible. This means that the electrode materials should be stud-
ied in the absence of additives such as binders and carbon black. As
we have recently demonstrated [4,5], such electrode materials can
be obtained by electrodepositing, e.g. Sb directly on a conducting
substrate serving as the current collector. As electrochemical depo-
sition also offers precise control of the reactions involved in the
deposition, the structure, phase composition and thickness of the
films can likewise readily be controlled. We have thus shown [4,5]
that it is possible to electrochemically deposit Sb-films with and
◦
trodepositions were carried out at 23 C in quiescent solutions
−
2
using a constant cathodic current of 4.0 mA cm
for 100 s. The
experiments were performed with an EG&G model 273 potentio-
stat/galvanostat (EG&G Princeton Applied Research, Princeton, New
Jersey, USA).
The electrodeposited Sb or Sb/Sb O coatings were then heat-
2
3
◦
treated at 300 C in a vacuum furnace within an argon-filled glove
box (O , H O < 2 ppm) to facilitate the formation of Cu Sb by diffu-
2
2
2
sion of copper into the Sb or Sb/Sb O3 films.
2
2.2. Characterization of the coatings
without co-deposited of Sb O3 on Ni substrates merely by adjust-
2
ing the pH of the electrodeposition solution. It was also shown that
the cycling stability of the electrodes was dramatically improved
The morphology of the coatings was studied with a scanning
electron microscope (LEO 1550) while the composition and struc-
ture of the deposits after heat-treatment were investigated using
X-ray diffraction employing a T2T Siemens D5000 diffractometer
when Sb O was co-deposited in the films. Films with an oxide
2
3
content of around 24% exhibited stable cycling performance dur-
ing more than 50 cycles with a reversible specific capacity of about
◦
with Cu K␣ radiation (grazing incidence 3 ).
−1
6
40 mAh g
.

7228
H. Bryngelsson et al. / Electrochimica Acta 53 (2008) 7226–7234
2
.3. Electrochemical characterization in Li-ion batteries
−0.70 V, respectively. It is therefore reasonable to conclude that the
electrodeposition of antimony on copper involves the same reac-
tions as the corresponding electrodeposition on nickel substrates
[4,5]. It was also possible to deposit films on copper substrates
merely cleaned in acetone from the pH 4.1 electrolyte, while the
corresponding depositions from the pH 1.3 solution were less
straightforward due to problems with poor coverage.
The electrochemical test cells, which were manufactured in the
argon filled glove box, consisted of a stack of the working elec-
trode, a 38 ␮m thick glass-wool separator soaked in electrolyte and
a lithium foil counter electrode which was vacuum-packed in a
polymer-coated aluminium package with attached nickel contacts.
The electrolyte used in the characterization experiments was 1 M
LiPF6 dissolved in a 2:1 solution of ethyl carbonate (EC): diethyl
carbonate (DEC) (Merck, battery grade). The LiPF6 salt was dried at
8
the electrolyte was found to be <10 ppm by Karl Fischer titration.
The electrochemical performance of the electrodes was studied by
cycling the cells between 1.5 and 0.01 V vs. Li /Li for 65 cycles at a
As discussed by Bryngelsson et al. [4], the Sb (H-2L)22 dimer
−
2
should be the dominating species [36] in the pH 4.1 antimony tar-
4−
trate solution, where H
L
denotes the fully deprotonated form of
−2
◦
0 C under vacuum over night prior to use. The water content of
tartrate (i.e. tartrate having lost both the two carboxyl protons and
the two hydroxyl protons). The electrodeposition reactions involv-
ing electrodeposition of antimony and a simultaneous release of
tartrate can then be written as shown in reaction (7)–(9) below:
+
C/10 rate (implying ten hours for a complete lithiation of the elec-
trode material) using a Digatron BTS-600 system. Additionally two
electrodes were cycled one cycle at a current rate of C/10 for XRD
measurements.
In the electrochemical testing, the theoretical electrodeposited
mass was used (i.e. 0.51 mg of Sb, which corresponds to 1.1 mg
Sb (H L)22 + 6e ↔ 2Sb + 2H
−
−
L
4−
(7)
(8)
(9)
2
−2
−
−2
2H L4 + 4H ↔ 2L2−
+
−
2
Sb (H L)22 + 4H+ + 6e− ↔ 2Sb + 2L2−
−
2
−2
From reaction (9), which is obtained by combining reactions
(7) and (8), it is evident that the local pH should increase in the
vicinity of the working electrode during the reduction of the anti-
mony tartrate dimer in the absence of sufficient buffer capacity. As
is seen from reaction (8), this pH increase stems from the rapid
restoration of at least the two hydroxyl tartrate protons which
should take place at the pH values of interest in this work (since
the pKa values of the carboxyl groups are 2.98 and 4.3 [37], a
partial protonation of the carboxyl protons should in fact also
be taken into account). In the pH 4.1 solution, reactions (8) and
Cu Sb assuming that all Sb has reacted) to estimate the specific
2
capacity of the material. The theoretical mass of Cu Sb has been
2
used for all films independent of heat-treatment time. For the
Sb O containing coatings the theoretical mass was adjusted with
2
3
the mass of the codeposited oxide (obtained from EQCM measure-
ments).
3
. Result and discussion
4
−
2−
2−
−
(
9) are actually better written as 2H 2L + 4H O ↔ 2L + 4OH
−
2
2
3
.1. Electrodeposition of Sb and Sb/Sb O3
2
−
−
−
and Sb (H L)2 + 4H O + 6e ↔ 2Sb + 2L + 4OH , respectively,
2
−2
2
due to the rapid local pH increase in this solution. As previously
described [4], this local pH increase can give rise to incorporation
Fig. 1 shows chronopotentiograms obtained for the electrode-
position of antimony on etched copper substrates in quiescent
solutions containing 0.075 M antimony tartrate at pH 1.3 (Fig. 1a)
and pH 4.1 (Fig. 1b), respectively, using a current density of
of Sb O3 in the deposited films as a result of the precipitation of
Sb O in the vicinity of the working electrode according to reaction
2
2
3
−2
(10).
4
.0 mA cm . It is clearly seen that these depositions, which
resulted in film thicknesses of about 300 nm, took place at sig-
nificantly different potentials in the two solutions (i.e. at −0.45 V
and −1.35 V, respectively). These findings, which are in good agree-
ment with previous results obtained by Bryngelsson et al. [5] for the
corresponding deposition on nickel substrates, can be explained
by the smaller iR drop in the pH 1.3 solution also containing
Sb (H L)2 + 2OH− + H O ↔ Sb O + 2L
2−
2−
(10)
2
−2
2
2
3
Under controlled current conditions, the co-deposited Sb O3
rapidly becomes coated with a protecting layer of Sb [4] as a result
of the continuous deposition of Sb and the reduction of the surface
2
+
−
of the Sb O3 layer (i.e. Sb O + 6H + 6e = 2Sb + 3H O).
2
2
3
2
A corresponding increase in the local pH is, however, not
expected during electrodeposition in the pH 1.3 solution due to its
significantly higher buffer capacity. As indicated by the different
deposition potentials (after correction for the iR drop), the speci-
ation is most likely also different in the pH 1.3 and 4.1 solutions.
In the pH 1.3 solution, the antimony is most likely not present as
0
.090 M KCl and 0.110 M HCl, and differences in the antimony spe-
ciation in the two solutions [5]. The iR drop corrected deposition
potentials were incidentally found to be approximately −0.24 and
2−
Sb (H L)2 dimers but as chloride complexes [5] for which a local
2
−2
pH change should be absent upon reduction. While deposition in
the pH 4.1 solution should give rise to films containing Sb and co-
deposited Sb O , pure antimony films should thus be obtained in
2
3
the pH 1.3 solution in analogy with our previous depositions on
nickel substrates.
As is seen in Fig. 2, electrochemical quartz crystal microbalance
(EQCM) measurements were carried to determine the concentra-
tion of Sb O in the films deposited at pH 4.1 using the same current
2
3
density as in Fig. 1. The difference in the deposition potential com-
pared to in Fig. 1 (curve b) is due to the smaller iR drop due to the
smaller current in the EQCM experiment. More importantly, the
frequency vs. time plot indicates that the total mass increase dur-
ing the deposition was about 1.09 times too large compared to that
−
expected for a 3e reduction of Sb(III) to Sb (assuming 100% cur-
Fig. 1. Chronopotentiograms recorded during depositions on copper using a current
−2
rent efficiency). As this value, which thus indicates that the films
contained about 9% Sb2O3 (w/w), is significantly smaller than the
density of 4.0 mA cm in quiescent solutions of 0.075 M antimony tartrate with pH
.3 (a) and pH 4.1 (b), respectively.
1

H. Bryngelsson et al. / Electrochimica Acta 53 (2008) 7226–7234
7229
Fig. 2. Chronopotentiogram and corresponding EQCM frequency vs. time curve
obtained in a quiescent solution of 0.075 M antimony tartrate at pH 4.1 during the
−2
deposition on copper with a current density of 4.0 mA cm
.
factor 1.24 previously obtained for the corresponding deposition
on nickel, it is evident that deposition on copper gives rise to films
containing less Sb O than those deposited on nickel. The reason
2
3
for this becomes apparent upon a closer examination of the EQCM
data in Fig. 2 as the slope of the frequency vs. time plot decreases
significantly during the deposition although the initial slope was
analogous to that obtained during the deposition on nickel (see Fig.
2
in ref [5]. As we have previously shown [5] that the dissolution rate
of Sb O increases as the local concentration of tartrate increases
2
3
in the vicinity of the working electrode, it is reasonable to assume
that this effect is more pronounced for the films on copper than
those on nickel as a result of microstructure differences. As will be
shown below, this hypothesis is supported by SEM investigations
of the films, which indicate that a larger particle size was obtained
on copper than on nickel. The larger dissolution rate in the present
case can then be explained by the larger surface area exposed to
Fig. 3. XRD patterns for the coatings deposited at pH 1.3 and pH 4.1; after 1 h of
heat-treatment (a) and (b) and after 48 h of heat-treatment (c) and (d), respectively.
the tartrate containing electrolyte for the Sb/Sb O films deposited
on copper.
2
3
previously been studied by Halimi et al. [41] by annealing thermally
◦
evaporated Cu–Sb thin films at temperatures from 150 to 600 C.
◦
While the latter study showed that Cu Sb was formed up to 300 C
2
3
.2. Synthesis and characterization of Cu Sb and Cu Sb formed
and that Cu Sb2 was only formed at higher annealing tempera-
2
9
2
9
by heat-treatment
tures, our results show that heat-treatment of thin films of Sb and
◦
mixtures of Sb and Sb O3 at 300 C can give rise to both Cu Sb and
2
2
To synthesize Cu Sb, the electrodeposited Sb and Sb/Sb O films
Cu Sb depending on the heat-treatment time. The finding that less
Cu Sb was present in the Sb O3 containing sample suggests that
9 2 2
2
2
3
9 2
◦
on the copper substrates were heat-treated at 300 C in a vac-
uum furnace within an argon-filled glove box (O , H O < 2 ppm)
this oxide hinders Cu diffusion in the material. It is also clear that
2
2
◦
to increase the rate of Cu diffusion into the deposit. The resulting
material was then analyzed by powder X-ray diffraction to study
the phase composition of the films. Fig. 3 depicts XRD patterns for
films obtained after heat-treatment for 1 and 48 h for the films
deposited at pH 1.3 and 4.1, respectively. It can be seen that the
phase composition depends on the heat-treatment time as well
a heat-treatment step of 1 h at 300 C is sufficient to fully convert
the Sb and Sb/Sb O3 films into Cu Sb.
2
2
To study the morphologies of the deposits and the influence of
the heat-treatment step on the morphology, SEM micrographs were
also taken before and after the heat-treatments of the samples (see
Fig. 4). This study showed that the particle size was smaller in the
as the concentration of Sb O3 in the deposits. While Cu Sb was
presence than in the absence of Sb O in the electrodeposited films,
2
2
2 3
the only phase found after heat-treatment for 1 h of both deposits,
another phase dominated after 48 h for the pH 1.3 sample (see
in analogy with the results previously reported results [4] for the
corresponding depositions on nickel. The depositions on copper,
however, gave rise to significantly larger particles compared to the
Fig. 3c). Two phases (i.e. Cu Sb and the new phase) were, on the
2
other hand, present after heat-treating the Sb O3 containing film
agglomerated 2–10 nm sized nanoparticles of Sb and Sb O3 found
2
2
for 48 h (see Fig. 3d). As there were no Sb O3 peaks in the XRD
after deposition on nickel [4]. As is seen in Fig. 4a, the size of the
Sb particles deposited without co-deposition of oxide was in the
2
patterns obtained with the pH 4.1 films, either before or after
the heat-treatments, it is also reasonable to assume that the co-
deposited Sb O was amorphous. According to the phase diagram
range between 200 and 300 nm. When Sb O3 was co-deposited in
2
the film the length of the particles was 100–200 nm, but the thick-
ness was only around 10 nm (Fig. 4b). The fact that the morphology
2
3
for antimony and copper [38], there are two stable phases at room
temperature, i.e. Cu Sb and Cu Sb . The new phase can hence be
of the Sb/Sb O3 films deposited on copper was different from that
2
9
2
2
ascribed to the hexagonal incomensurate structure Cu Sb [39,40],
which can be expected to form in the presence of an excess of cop-
of the deposits on nickel indicates that the nucleation process was
different on copper and nickel. One reason for this could be the rapid
formation of oxide on the nickel substrates. The latter hypothesis
9
2
per in the system [41]. The formation of Cu Sb2 [39,40], has been
9

7230
H. Bryngelsson et al. / Electrochimica Acta 53 (2008) 7226–7234
Fig. 4. SEM micrographs of coatings deposited at pH 1.3 and pH 4.1; before heat-treatment (a) and (b), after 1 h of heat-treatment (c) and (d) and after 48 h of heat-treatment
e) and (f).
(
is supported by the fact that deposition on a copper surface merely
cleaned with acetone resulted in a morphology more similar to that
found after deposition on Ni (see Fig. 5). As will be shown below, the
Sb O concentration was also slightly higher in the films deposited
phology was noted between 1 h and 48 h of heat-treatment. This
means that the formation of Cu Sb and Cu Sb did not give rise to
any significant changes in the particle size of the films.
2
9
2
2
3
on acetone cleaned copper substrates than on etched copper sub-
strates. For the films deposited on acetone cleaned samples, it was,
3.3. Electrochemical characterization in a Li-ion battery
however, difficult to obtain Cu Sb as the only phase present after
the heat-treatment most likely as the presence of the copper oxide
To study the influence of the morphology and composition
on the structural reversibility and the cycling performance of the
materials the heat-treated samples were tested as anode materials
in a Li-ion cell. Fig. 6 shows chronopotentiograms for the first five
charge and discharge cycles for the coatings deposited at pH 1.3 and
pH 4.1 before heat-treatment, after 1 h of heat-treatment and after
48 h of heat-treatment, respectively. For the electrode deposited at
2
on the surface in combination with the codeposited Sb O3 slow
2
down the formation of Cu Sb by hindering the diffusion of cop-
2
per in the films. In Fig. 4 it is likewise seen that the particle size
was unaffected by the heat-treatment step although the shape of
the particles changed slightly and no significant difference in mor-

H. Bryngelsson et al. / Electrochimica Acta 53 (2008) 7226–7234
7231
and f) stemming from reaction (11)
+
−
6
Li + Sb O + 6e ↔ 3Li O + 2Sb
(11)
2
3
2
A comparison of the latter curves also shows that the oxide was
◦
stable during the heat-treatment at 300 C. As previously described
[
4,5], the amount of co-deposited oxide can be estimated from
the charge associated with the plateau at 1.6 V. This approach sug-
gests that the amount of co-deposited oxide was about 5% (Fig. 6d)
rather than 9% as indicated by the EQCM data. The value of 5% is,
however, an underestimation since the total charge of the first dis-
charge includes a contribution from the formation of an SEI layer
at potentials close to 0.8 V [4] where the lithiation of Sb starts. The
capacity of the first discharge for the electrodes heat-treated for 1 h
−
1
is about 430 mAh g . This is significantly larger than the theoreti-
−
1
−1
cal capacity of Cu Sb (323 mAh g for pure Cu Sb and 386 Ah g
2
2
for an electrode with 9% of oxide) and indicates a considerable SEI
formation. The result confirms that electrodeposition on copper
substrates gives rise to films with a significantly lower content of
Fig. 5. SEM micrograph taken before heat-treatment for a coating deposited at pH
Sb O than the 20–25% previously obtained after the correspond-
−2
2
3
4
.1 on an unetched Cu-surface using a current density of 4.0 mA cm
.
ing deposition on nickel [4]. It was also found that the irreversible
capacity was smaller (i.e. 20% of the total first cycle capacity) for
the pH 4.1 samples than for the pH 1.3 material (for which the cor-
responding values was 30%). The latter can be explained by the
difference in the particle size giving rise to different electroactive
surface areas.
pH 4.1 and heat-treated for 48 h the first, third and forth cycle are
shown. The results obtained for the Sb and Sb/Sb O3 samples that
2
had not undergone any heat-treatment show an electrochemical
performance more similar to Cu Sb than pure Sb [4,5]. This indi-
2
The first cycle discharge curves for the Cu Sb samples (see Fig. 6c
2
cates that Cu Sb is formed already at room temperature, although
2
and d) exhibit two plateaus at about 0.85 and 0.66 V which most
likely can be ascribed to a combination of the formation of an SEI
layer and stepwise lithiation of Sb. As three discrete plateaus are
seen in the corresponding second cycle discharge curves it is likely
that the pronounced SEI formation reactions during the first cycle
mask the lithiation plateaus and that the lithiation in fact involves
three different reactions. These reactions should then correspond
the process is slower than at higher temperatures. The presence of
Sb O in the pH 4.1 films (including the heat-treated samples) is
clearly seen from the voltage plateau at about 1.6 V (see Fig. 6b, d
2
3
to the step-wise lithiation of Sb to yield Li Sb via three subsequent
3
one-electron steps along the lines suggested by Morcrette et al. [32]
and Matsuno et al. [30]. This hypothesis is further supported by the
three plateaus observed during the delithiation on the first cycles
in Fig. 6c and d and the fact that XRD data confirm the reformation
of Cu Sb during the delithiation step (see below). The charges asso-
2
ciated with the first two of these three delithiation plateaus involve
approximately the same charge as would be expected for three one-
electron delithiation reactions in which Li Sb is step-wise oxidised
3
to Cu Sb. At this stage, it can also be concluded that the identifica-
2
tion of these three plateaus was significantly facilitated by the fact
that no additives such as binders and carbon black were needed
for the electrodeposited films used in this study. The charge asso-
ciated last delithiation step at 1.05 V, i.e. the reformation of Cu Sb
2
is, however, relatively lower for the electrode deposited at pH 1.3,
in which the particles are larger and no oxide is present.
Lithiation of Cu Sb occurs in the same voltage range as for pure
2
Sb [4,5]. This is an indication that the lithiation of Cu Sb involves
2
reduction of Sb, which according to the chronopotentiograms is a
three-electron process. A reduction of elemental Sb is also compat-
ible with the formation of Cu Sb during the heat-treatment steps as
2
a result of the diffusion of elemental copper into the films contain-
ing either Sb or a mixture of Sb and Sb O . This means that reactions
2
3
(1) and (2) proposed by Fransson et al. [27] can be written as in
reactions (12) and (13) to show that the lithium ions are included
to compensate the charge of the Sb(-III) species formed during the
reduction of Sb. The formation of Li Sb is hence analogous to the
3
reduction of Sb yielding stibine H Sb in aqueous solutions.
3
Cu Sb + 2Li + 2e− ↔ Li CuSb + Cu
+
(12)
(13)
Fig. 6. The first chronopotentiometric cycle, recorded at a rate of C/10, of coatings
deposited at pH 1.3 and pH 4.1; before heat-treatment (a) and (b), after 1 h of heat-
treatment (c) and (d) and after 48 h of heat-treatment (e) and (f).
2
2
Li CuSb + Li + e− ↔ Li Sb + Cu
+
2
3

7232
H. Bryngelsson et al. / Electrochimica Acta 53 (2008) 7226–7234
As already mentioned reaction (12) should then be divided into
two steps, yielding either:
Cu Sb + Li + e− ↔ LiCu Sb
+
(14)
(15)
2
2
LiCu Sb + Li + e− ↔ Li CuSb + Cu
+
2
2
According to the suggestion by Morcrette et al. [32] or:
Cu Sb + Li + e− ↔ LiCuSb + Cu
+
(16)
(17)
2
LiCuSb + Li+ + e− ↔ Li CuSb
2
based on the reactions proposed by Matsuno et al. [30]. These reac-
tions and reaction (13) clearly demonstrate that the fully lithiated
material should contain Li Sb and elemental copper.
3
The reversibility of reaction (14) (or (16)), has recently been
shown [32] to depend strongly on the composition and particle
size of the material. Our results show that both the films deposited
at pH 1.3 and those deposited at pH 4.1 exhibit a plateau at 1.05 V
where reformation of Cu Sb is expected. As already indicated, the
2
capacities associated with the plateaus, however, differed some-
−
1
what as a capacity of about 30 mAh g was found for the pH 1.3
material while the corresponding value for the pH 4.1 material
−1
Fig. 8. The first and second chronopotentiometric cycle, recorded at a rate of C/10,
of coatings deposited at pH 1.3 after 1 and 48 h of heat-treatment (a) and the corre-
sponding chronopotentiometric cycles after 260 h cycling of the same cells (b).
was 80 mAh g . The latter could suggest that the reformation of
Cu Sb benefited from the presence of smaller particles in the pH
2
4
.1 sample. This is further supported with the fact that the oxide
containing sample has a lower irreversible capacity in the first cycle.
The larger capacity at 1.05 V for the pH 4.1 material may, how-
nificantly lower potential was needed for the first cycle lithiation of
the Cu Sb2 sample and that the delithiation process took place in
9
ever, also be due to the reformation of Sb O3 (see reaction (11))
2
the same voltage range as for Cu Sb. This and the shape of the sub-
2
as discussed by Bryngelsson et al. [4]. A particle size effect was
recently described by Morcrette et al. [32] who showed that the
sequent charging and discharging curves indicate that Cu Sb rather
2
than Cu Sb is formed during the delithiation in the first cycle. The
9
2
reformation of Cu Sb only took place within materials containing
2
effect is even more clearly seen in Fig. 8 depicting the first two
charge and discharge curves and the corresponding curves for the
sufficiently small particles. The ball-milled material described by
Fransson et al. [27] also showed structural reversibility with respect
Cu Sb material after repeated cycling for 260 h. It can therefore be
9
2
to Cu Sb whereas the Cu Sb synthesized by Morcrette et al. [32]
2
2
concluded that only a fraction of the copper extruded during the
reduction of Cu Sb2 was used to form Cu Sb during the delithia-
using a ceramic method did not. Our results also show that struc-
tural reversibility was obtained in both the presence and absence
of Li O formed by reduction of Sb O according to reaction (11).
9
2
tion process. This means that the fully delithiated material must
contain Cu Sb in a matrix of elemental copper. Since Cu Sb has
2
2
3
2
2
As is seen in Fig. 7, the reformation of Cu Sb in the pH 1.3 and pH
2
a strong structural relationship with Li Sb which is not the case
3
4.1 samples heat-treated for 1 h was also confirmed by XRD mea-
for the incomensurate structure Cu Sb2 [39,40]: a reformation of
9
surements on powder that was scratched off from the delithiated
coatings after the first charge–discharge cycle.
Cu Sb would require significant structural changes. It is therefore
9
2
not surprising that Cu Sb rather than Cu Sb is formed during the
2
9
2
As is seen in Fig. 6e, the electrochemical behaviour of the coat-
delithiation process and that a lower potential is required for the
reduction of Cu Sb2 compared to for Cu Sb. The plateau observed
ing containing mainly Cu Sb2 was significantly different from that
9
9
2
seen for the Cu Sb containing materials. It is clearly seen that a sig-
2
at about 0.7 V in Fig. 6e most likely corresponds to the formation of
the SEI layer.
The charge and discharge curves for the material initially
composed of Cu Sb, Cu Sb and Sb O3 (obtained by heat-treating
2
9
2
2
the pH 4.1 deposit for 48 h) are generally similar to those obtained
in the absence of Cu Sb2 (compare Fig. 6d and f) apart from the
9
fact that the third and fourth cycle capacities are smaller in the
presence of Cu Sb2 and that the Cu Sb2 curves shift towards lower
9
9
capacities during repeated cycling. The latter effect is clearly a
result of the fact that there is a mismatch between the reduction
and oxidation charges for the Cu Sb2 containing material. On the
9
third cycle the discharge capacity is thus smaller than the charge
capacity (Fig. 6f). This can be explained by assuming a complex
microstructure containing both Li Sb and some unreacted Cu Sb
3
9
2
at the first fully discharged state. Since it is known that the refor-
mation of Cu Sb is highly dependent of the preparation procedure
2
of the material it could be expected that not all Li Sb will be
3
reformed to Cu Sb during the second charge. During subsequent
2
cycling the particle size most likely decreases, favouring the
reformation of Li Sb to Cu Sb. The delithiation capacity can hence
3
2
exceed the lithiation capacity for some cycles. For the reduction of
Cu9Sb2 much lower potentials are required compared to reduction
Fig. 7. XRD patterns obtained after the first chronopotentiometric cycle, recorded
at a rate of C/10, of coatings deposited at pH 1.3 (a) and pH 4.1 (b).

H. Bryngelsson et al. / Electrochimica Acta 53 (2008) 7226–7234
7233
Fig. 9. Specific capacity as function of cycle number plots for coatings deposited at
pH 1.3 (a) and pH 4.1 (b); before heat-treatment (curve 1), after 1 h of heat-treatment
curve 2) and after 48 h of heat-treatment (curve 3). The coatings were cycled at a
(
rate of C/10.
Fig. 10. The first chronopotentiometric cycle recorded (at a rate of C/10) after 72 h
of heat-treatment for a coating deposited at pH 4.1 on an unetched Cu-surface using
−
2
a current density of 4.0 mA cm
.
of Cu Sb which will cause Cu Sb to be preferentially reduced. The
2
2
reduction of the remaining Cu Sb2 is most likely further impeded
capacities compared to that for the material initially containing
Cu Sb and Sb O come as no surprise. In Fig. 9b it is, however, seen
9
by the fact that the reduction Cu Sb gives rise to extrusion of
2
2
2
3
elemental copper, the presence of which should slow down the
rate of Li+ transport in the material. It can thus be concluded that
this mixed material should become more and more similar to the
that the capacity loss with cycle number was less for the Cu Sb2
containing sample. As the extrusion of Cu during the lithiation pro-
9
cess reduces the volume expansion to 42% for Cu Sb compared to
2
pure Cu Sb material upon repeated cycling.
150% for pure Sb, it is reasonable to assume that the influence of
2
Fig. 9 shows the cycle stabilities of the pH 1.3 and pH 4.1 materi-
als before heat-treatment and after 1 and 48 h of heat-treatment
at 300 C, respectively. The electrodes cycled without any heat-
any size buffering effect due to the presence of Sb O3 should be
2
much smaller for Cu Sb than for Sb films. The present results clearly
2
◦
show that stable cycling can be obtained with Cu Sb films with
2
treatment show a relatively poor cycling stability (see curve 1 in
Fig. 9a and b). A stable capacity of approximately 300 mAh g
particles sizes in the range between 200 and 300 nm although the
−
1
structural reversibility of Cu Sb seems to depend on the particle
2
was, however, found during at least 65 cycles for the Cu Sb mate-
size.
2
rial based on the electrodeposition of Sb at pH 1.3 (see curve 2
in Fig. 9a). This value, which is close to the theoretical value of
Cycling experiments were also carried out on the films deposited
at pH 4.1 on copper substrates merely cleaned in acetone although
Cu Sb was not the only phase present. The coatings had accord-
2
−1
3
23 mAh g for Cu Sb, is comparable to the results obtained for
2
ball-milled Cu Sb materials published by Fransson et al. [27]. It
ing to the SEM micrographs, a morphology similar to that of films
deposited at pH 4.1 on Ni [4] and the XRD pattern also showed
an amorphous region similar to that seen for the corresponding
films on nickel. Fig. 10 depicts the electrochemical performance of
a film containing Sb O , Cu Sb, Cu Sb and according to XRD also
2
should, however, be pointed out that the results obtained by Frans-
son et al. [27] were based on the use of a composite electrode also
containing carbon black which may have contributed to the total
capacity. The slightly higher and stable capacity obtained in the
2
3
2
9
2
present work is therefore encouraging. The Cu Sb2 material (see
curve 3 in Fig. 9a) also exhibited a stable cycling performance but
traces of Sb (after heat-treatment of 72 h). The reduction charge
associated with the plateau at about 1.5 V in the chronopoten-
9
with a lower capacity than for the Cu Sb material. As the total
tiogram corresponds to a Sb O3 concentration of about 11%, which
2
2
amount of antimony was the same for both the 1 h and the 48 h
heat-treated samples, the slightly lower capacity can be explained
by the larger amount of extruded copper preventing the migra-
is only slightly higher than for the films deposited on etched copper.
The plateau at 0.4 V can most likely be ascribed to the reduc-
tion of Cu Sb2 while the results of additional EQCM experiments
9
tion of Li+ for the Cu Sb2 material. From Fig. 9b it is evident that
[42] indicate that the reduction plateau at about 1.7 V (which in
fact also can be seen in Fig. 6) is due to SEI formation. In agree-
ment with the results obtained for the films on etched copper (see
Fig. 6d and f), three plateaus are seen in the cycling curve upon
delithiation in addition to a small plateau at about 1.2 V which
9
the capacities for the Cu Sb material based on the Sb/Sb O films
2
2
3
−
1
deposited at pH 4.1 were about 370 mAh g
during the first 25
cycles. As this capacity is higher than the theoretical capacity of
−
25 mAh g , it is reasonable to assume that the Sb O3 (9% (w/w)
2
1
3
according to the EQCM data) contributed significantly to the capac-
ity. After 25 cycles a dramatic decrease in the capacity was, however,
found. This result, which clearly shows the importance of evaluat-
ing the cycling stability of materials based on a large number of
cycles, can most likely be explained by a loss of contact with an
increasing fraction of the material. The latter is supported by the
results of dissections of the cycled batteries which revealed that
can be attributed to reformation of Sb O . As is seen from the
2 3
inset in Fig. 10, showing the capacity as a function of the cycle
number, the stability of the material was, however, poor since a
dramatic decrease in the capacity was found during the first 50
cycles despite the fact that the films had a morphology similar to
that of the Sb/Sb O3 films on nickel which exhibited excellent sta-
2
bilities. This poor cycling behaviour, which is similar to that seen
parts of the initially Sb O3 containing material had lost contact
for the Sb O3 containing films in Fig. 9, is in good agreement with
2
2
with the Cu electrode whereas no such effect was seen for the pure
the hypothesis that the presence of Sb O3 decreases the stability
2
Cu Sb material. In accordance with the previous conclusion that
2
of Cu Sb containing films. The present results further show that
2
the Cu Sb did not contribute to the capacity of the material ini-
Sb O does not have the same stabilising effect on Cu Sb as for
9
2
2
3
2
tially containing a mixture of Cu Sb, Cu Sb and Sb O3 the lower
Sb.
2
9
2
2

7234
H. Bryngelsson et al. / Electrochimica Acta 53 (2008) 7226–7234
4
. Conclusions
additional support. We would likewise like to thank Prof. Josh
Thomas for many valuable scientific discussions.
It has been shown that Cu Sb films can be synthesized by
2
heat-treatments of electrodeposited films of Sb or mixtures of
References
◦
Sb and Sb O on copper substrates at 300 C, and that these
2
3
−
1
[1] J.M. Tarascon, M. Armand, Nature 414 (2001) 359.
Cu Sb films exhibit stable cycling capacities of about 290 mAh g
2
[
2] J.O. Besenhard, Handbook of Battery Materials, Wiley-VCH, Weinheim, 1999.
when tested as anode material in Li-ion batteries. Since the cur-
rent collector is used as the substrate in the synthesis step, the
present synthesis method is a versatile method particularly promis-
ing for the preparation of anode materials for microbatteries. In
analogy with electrodepositions on nickel, Sb films containing
co-deposited Sb O could be deposited from pH 4.1 solutions of
[3] J.L. Tirado, Mater. Sci. Eng. R: Rep. R40 (2003) 103.
[4] H. Bryngelsson, J. Eskhult, L. Nyholm, M. Herranen, O. Alm, K. Edstr o¨ m, Chem.
Mater. 19 (2007) 1170.
[
5] H. Bryngelsson, J. Eskhult, K. Edstr o¨ m, L. Nyholm, Electrochim. Acta 53 (2007)
1062.
[6] I.A. Courtney, J.R. Dahn, J. Electrochem. Soc. 144 (1997) 2943.
[7] I.A. Courtney, J.R. Dahn, J. Electrochem. Soc. 144 (1997) 2045.
2
3
[
8] V. Pralong, J.-B. Leriche, B. Beaudoin, E. Naudin, M. Morcrette, J.-M. Tarascon,
antimony tartrate as a result of the local pH increase at the elec-
trode surface associated with the protonation of the liberated
Solid State Ionics 166 (2004) 295.
[
9] P. Poizot, S. Laruelle, S. Grugeon, L. Dupont, J.M. Tarascon, Nature 407 (2000)
496.
tartrate. The concentrations of Sb O3 in the films was, how-
2
[
10] M. Armand, F. Dalard, D. Deroo, C. Mouliom, Solid State Ionics 15 (1985)
ever, found to be significantly lower (i.e. 9–11% as compared to
205.
2
0–25%) than in the corresponding films deposited on nickel most
[
11] D.G. Kim, H. Kim, H.J. Sohn, T. Kang, J. Power Sources 104 (2002) 221.
likely due to differences in the nucleation step on copper and
on nickel. These differences can also explain the larger particle
sizes obtained with copper substrates. It can also be concluded
[12] J. Yang, Y. Takeda, N. Imanishi, T. Ichikawa, O. Yamamoto, Solid State Ionics 135
2000) 175.
(
[
[
13] J. Yang, M. Winter, J.O. Besenhard, Solid State Ionics 90 (1996) 281.
14] O. Mao, R.B. Turner, I.A. Courtney, B.D. Fredericksen, M.I. Buckett, J. Krause, J.R.
Dahn, Electrochem. Solid-State Lett. 2 (1999) 3.
that the formation of Cu Sb during the heat-treatment step was
2
[
[
15] B.A. Boukamp, G.C. Lesh, R.A. Huggins, J. Electrochem. Soc. 128 (1981) 725.
16] J.O. Besenhard, M. Hess, P. Komenda, Solid State Ionics 40–41 (1990) 525.
slowed down when Sb O was present in the films and that Cu Sb
2
3
9
2
rather than Cu Sb was formed after heat-treatment for 48 h at
2
[
17] L.Y. Beaulieu, K.C. Hewitt, R.L. Turner, A. Bonakdarpour, A.A. Abdo, L. Chris-
tensen, K.W. Eberman, L.J. Krause, J.R. Dahn, J Electrochem Soc 150 (2003)
A149.
◦
3
00 C.
The results of Li-ion battery tests clearly show that the reduc-
[
18] L.Y. Beaulieu, T.D. Hatchard, A. Bonakdarpour, M.D. Fleischauer, J.R. Dahn, J.
Electrochem Soc. 150 (2003) A1457.
tion of Cu Sb involves three separate one-electron steps in which
2
+
Sb is step-wise reduced to Sb(-III) and that Li ions serve as charge
[19] Z. Chen, V. Chevrier, L. Christensen, J.R. Dahn, Electrochem. Solid-State Lett. 7
(2004) A310.
compensating ions. These studies were significantly facilitated by
the absence of additives, such as binders and carbon black in the
electrodes and comparisons of the first cycles clearly revealing the
influence of SEI formation on the first cycle. Since stable cycling
[20] K.D. Kepler, J.T. Vaughey, M.M. Thackeray, Electrochem. Solid-StateLett. 2(1999)
307.
[
21] D. Larcher, L.Y. Beaulieu, D.D. MacNiel, J.R. Dahn, J. Electrochem. Soc. 147 (2000)
1658.
[
[
[
22] J.T. Vaughey, J. O’Hara, M.M. Thackeray, Electrochem. Solid-State Lett. 3 (2000)
capacities were obtained for Cu Sb films with particles sizes in the
2
1
3.
range between 200 and 300 nm the present results indicate that a
23] M. Stjerndahl, H. Bryngelsson, T. Gustafsson, J.T. Vaughey, M.M. Thackeray, K.
Edstr o¨ m, Electrochim. Acta 52 (2007) 4947.
24] J.T. Vaughey, C.S. Johnson, A.J. Kropf, R. Benedek, M.M. Thackeray, H. Tostmann,
T. Sarakonsri, S. Hackney, L. Fransson, K. Edstr o¨ m, J.O. Thomas, J. Power Sources
97–98 (2001) 194.
small particle size is less important for the Cu Sb films than for the
2
corresponding Sb films. The stabilities of the Cu Sb films contain-
2
ing Sb O were, on the other hand, poor due to a gradual loss of the
2
3
contact between the material and the copper substrate. Structural
[25] M. Wachtler, M. Winter, J.O. Besenhard, J. Power Sources 105 (2002) 151.
[
26] C.S. Johnson, J.T. Vaughey, M.M. Thackeray, T. Sarakonsri, S.A. Hackney, L. Frans-
son, K. Edstr o¨ m, J.O. Thomas, Electrochem. Commun. 2 (2000) 595.
reversibility with respect to Cu Sb was, nevertheless, observed for
2
films with and without Sb O3 although the reformation of Cu Sb
2
2
[
27] L.M.L. Fransson, J.T. Vaughey, R. Benedek, K. Edstr o¨ m, J.O. Thomas, M.M. Thack-
appeared to be more pronounced for the Sb O3 containing films.
eray, Electrochem. Commun. 3 (2001) 317.
2
[
[
28] L.M.L. Fransson, J.T. Vaughey, K. Edstr o¨ m, M.M. Thackeray, J. Electrochem. Soc.
50 (2003) A86.
29] L. Haeggstroem, C.M. Ionica, J.C. Jumas, L. Aldon, P.E. Lippens, K. Edstr o¨ m, Hyper-
Lower but stable cycling capacities were found for films containing
1
Cu Sb (formed during heat-treatment for 48 h) for which a trans-
9
2
formation of Cu Sb into Cu Sb took place during the cycling. The
fine Interact. 167 (2007) 759.
9
2
2
lower capacities for these films compared to those for the Cu Sb
films obtained after heat-treatment for 1 h can be explained by the
[30] S. Matsuno, M. Noji, T. Kashiwagi, M. Nakayama, M. Wakihara, J. Phys. Chem. C
11 (2007) 7548.
31] J. Ren, X. He, W. Pu, C. Jiang, C. Wan, Electrochim. Acta 52 (2006) 1538.
2
1
[
lower potentials required for the reduction of Cu Sb and the pres-
9
2
[32] M. Morcrette, D. Larcher, J.M. Tarascon, K. Edstr o¨ m, J.T. Vaughey, M.M. Thack-
ence of a matrix of extruded copper that slows down the migration
of Li .
eray, Electrochim. Acta 52 (2007) 5339.
[33] S.-W. Song, R.P. Reade, E.J. Cairns, J.T. Vaughey, M.M. Thackeray, K.A. Striebel, J.
Electrochem. Soc. 151 (2004) A1012.
+
[
[
[
34] G. Zhang, K. Huang, S. Liu, W. Zhang, B. Gong, J. Alloys Compd. 426 (2006)
Acknowledgements
432.
35] S. Sharma, J.K. Dewhurst, C. Ambrosch-Draxl, Phys. Rev. B: Condens. Matter 70
(
2004) 104110/1.
The authors would like to thank The Swedish Research Coun-
cil (VR) (grant 621-2005-5493 and grant 2005-3356), The Swedish
Energy Agency (STEM), The G o¨ ran Gustafsson Foundation and The
Carl Trygger Foundation for financial support. The authors also
36] J. Zheng, A. Takeda, N. Furuta, J. Anal. At. Spectrom. 16 (2001) 62.
[37] G. Aylward, T. Findlay, SI Chemical Data, 5th ed., Wiley, New York, 2002.
[38] T.B. Massalski, Binary Alloy Phase Diagrams, ASM International, Ohio, 1986.
[39] E. Gunzel, K. Schubert, Zeitschrift fuer Metallkunde 49 (1958) 124.
[40] K. Motai, Y. Watanabe, S. Hashimoto, Acta Crystallogr. B B49 (1993) 655.
express their sincere gratitude to the European Network of Excel-
lence, EU FP6, “ALISTORE” HB as well as A˚ ngpannef o¨ reningen for
[41] R. Halimi, D. Hamana, E.M. Chpilevski, Thin Solid Films 139 (1986) 147.
42] J. Eskhult, H. Bryngelsson, L. Nyholm, K. Edstr o¨ m, in preparation.
[