Surface chemistry of intermetallic AlSb-anodes for Li-ion batteries

Article abstract of DOI:10.1016/j.electacta.2007.01.064

The solid electrolyte interphase (SEI) layer on AlSb electrodes has been studied in Li/AlSb cells containing a LiPF₆ EC/DEC electrolyte using X-ray photoelectron spectroscopy (XPS). Data were collected before SEI-formation, during formation, and after formation at 0.01 V versus Li⁰/Li⁺, and at full delithiation in cycled cells at 1.20 V. The thickness of the SEI layer increases during lithiation and decreases during delithiation. This dynamic behaviour occurs continuously on cycling the cells. The growth of the SEI layer can be attributed predominantly to the deposition of carbonaceous species below 0.50 V versus Li⁰/Li⁺; these species disappear almost completely during delithiation. The extra surface-layer formation is a consequence of the additional charge that is needed to lithiate the remaining Sb component of the micrometer-sized AlSb particles at low potentials as seen by synchrotron-based X-ray diffraction. Aluminium is not reactive to lithium alloying in this electrolyte. Relatively small amounts of LiF were detected in the AlSb SEI layers compared to that commonly found in the SEI layers on graphite electrodes.

Full text of DOI:10.1016/j.electacta.2007.01.064

Electrochimica Acta 52 (2007) 4947–4955
Surface chemistry of intermetallic AlSb-anodes for Li-ion batteries
M. Stjerndahla, H. Bryngelssona, T. Gustafssona, J.T. Vaugheyb,
b
a,∗
¨
M.M. Thackeray , K. Edstrom
a
˚
Department of Materials Chemistry, Angstro¨m Laboratory, Uppsala University Box 538, SE-751 21 Uppsala, Sweden
b Electrochemical Technology Program, Chemical Technology Division, Argonne National Laboratory, Argonne, IL 60439, USA
Received 25 September 2006; received in revised form 1 December 2006; accepted 27 January 2007
Available online 6 February 2007
Abstract
The solid electrolyte interphase (SEI) layer on AlSb electrodes has been studied in Li/AlSb cells containing a LiPF6 EC/DEC electrolyte using
X-ray photoelectron spectroscopy (XPS). Data were collected before SEI-formation, during formation, and after formation at 0.01 V versus Li0/Li+,
and at full delithiation in cycled cells at 1.20 V. The thickness of the SEI layer increases during lithiation and decreases during delithiation. This
dynamic behaviour occurs continuously on cycling the cells. The growth of the SEI layer can be attributed predominantly to the deposition of
carbonaceous species below 0.50 V versus Li0/Li+; these species disappear almost completely during delithiation. The extra surface-layer formation
is a consequence of the additional charge that is needed to lithiate the remaining Sb component of the micrometer-sized AlSb particles at low
potentials as seen by synchrotron-based X-ray diffraction. Aluminium is not reactive to lithium alloying in this electrolyte. Relatively small amounts
of LiF were detected in the AlSb SEI layers compared to that commonly found in the SEI layers on graphite electrodes.
© 2007 Elsevier Ltd. All rights reserved.
Keywords: SEI; AlSb; Intermetallic; Anode; Li-ion battery
1. Introduction
ion batteries. There are several metals (M) that alloy with
lithium, e.g. Al, In, Sb, Si, Sn and Pb. These binary LixM systems
provide high theoretical gravimetric and volumetric capacities.
A problem, however, is that large volume changes occur dur-
ing the insertion of lithium into the host metal structure, so the
metal particles crack and become electrochemically pulverized;
it is therefore difficult to cycle LixM electrodes more than a
few times. There are several ways to overcome this problem.
One way is to use intermetallic systems. There are two main
groups of intermetallic compounds that have been evaluated for
Li-ion battery purposes. The first group consists of materials in
which one metal is active to lithium alloying while the second
is inactive, for example, Cu6Sn5 [12–16], Cu2Sb [17], Ni3Sn4
[18], NiSb2 [19], MnSb [20,21], Mn2Sb [21], SnxFe [22,23],
Mo1−xSnx [24]. The second group consists of compounds in
which both metals are active to lithium alloying, for example
InSb [25–28], AlSb [26,28–30], SnSb [31–38,] and Ag3Sb [39].
As on graphite, the SEI layer on intermetallic anodes forms
below 0.8 V versus Li0/Li+. However, SEI layers on intermetal-
lic surfaces have not been studied in as much detail as those on
graphite. Differences between the SEI layers on intermetallic-
and graphite electrodes can be expected because the large vol-
ume changes that occur during cycling of the intermetallic
The solid electrolyte interphase (SEI) layer formed on
graphite anodes has been extensively studied in the past by
a range of different methods [1–11]. This SEI layer, which
is formed below 0.8 V versus Li0/Li+ due to the reduction
of thermodynamically unstable organic solvents in the elec-
trolyte, leads to an irreversible loss of capacity on the initial
charge/discharge cycle of lithium-ion cells. At the same time,
however, the SEI layer plays a vitally important role in protect-
ing the graphite electrode from co-intercalation of electrolyte
solvents and consequent exfoliation. Although several models
forthemorphological structureofSEIlayershavebeenproposed
[1,3,4,7], a more detailed understanding of these complex lay-
ers is still required; in this respect, a greater appreciation for the
need to use extremely pure electrolytes, and dry solvents and
electrodes has developed over recent years.
Intermetallic compounds have been studied as a group of
anode materials that could possibly replace graphite in lithium-
∗
Corresponding author.
E-mail address: kristina.edstrom@mkem.uu.se (K. Edstro¨m).
0013-4686/$ – see front matter © 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2007.01.064

4948
M. Stjerndahl et al. / Electrochimica Acta 52 (2007) 4947–4955
Table 1
electrodes lead to repeated SEI formation, disintegration and
reformation during charge and discharge. In this respect, it has
already been shown that the SEI layers on Sn-containing anodes
are different to those on graphite [40]. Such repetitive SEI for-
mation/disintegration processes can be expected to yield aged
SEI layers on the active particles onto which freshly formed SEI
layers are deposited on successive cycles. This implies that the
SEI thickness will increase on long term cycling [35].
Electrochemical conditions for samples A–F prior to XPS analysis
Sample
Electrochemical conditions
A
B
C
D
E
Potentiostatically lithiated and equilibrated at 0.90 V
Potentiostatically lithiated and equilibrated at 0.75 V
Potentiostatically lithiated and equilibrated at 0.50 V
Potentiostatically lithiated and equilibrated at 0.01 V
Galvanostatically cycled three times between 1.20 and 0.50 V,
XPS data recorded at 1.20 V
In this study, we have chosen AlSb as a model system to
investigate SEI layers on intermetallic electrode surfaces by
undertaking detailed X-ray photoelectron spectroscopy (XPS)
analyses of AlSb electrodes in lithium cells at various points on
the charge/discharge curve. AlSb was selected because it con-
tains two metals that are active to lithium alloying and because
there is a clear difference in potential in alloying lithium with
Sb and Al. For example, lithium reacts with antimony at room
temperature to form Li3Sb at approximately 0.90 V, whereas
lithium alloys with aluminium to form LiAl at approximately
0.3 V. With this potential difference, it is possible to study the
two alloying phenomena separately. We used XPS as the pri-
mary characterization technique to study the SEI layer, because
this technique is sensitive to layers that are typically tenths of
nanometres thick. Disadvantages of the XPS technique are: (1)
the sample must be studied ex situ and (2) there is a risk of
pumping off volatile components or species that make up the
SEI layer. Despite these disadvantages, XPS is one of the few
experimental techniques that can be used effectively to probe
extremely thin layers, thereby providing valuable information
about the surface character of intermetallic electrodes such as
AlSb.
F
Galvanostatically cycled three times between 1.20 and 0.01 V,
XPS data recorded at 1.20 V
3.14 cm2 circular discs. The electrodes were dried at 120 ◦C in
vacuum prior to use. The typical loading of active material was
in the range of 6–7 mg cm−2. Electrochemical (coffee-bag type)
cells were assembled in the glove box by laminating the AlSb
electrodes with a glass–wool separator soaked in electrolyte,
which consisted of 1 M LiPF6 dissolved in ethylene carbonate
(EC, Merck battery grade) and diethyl carbonate (DEC, Merck
battery grade) in a volumetric ratio of 2 to 1. The LiPF6 salt
was vacuum dried at 80 ◦C in the glove box prior to use; the sol-
vents were used as-received. The laminate was vacuum sealed
in a polymer coated aluminium foil with attached nickel con-
tacts. Lithium-foil was used as the counter electrode. The cells
were potentiostatically equilibrated during the initial lithiation
at 0.90 V (sample A), 0.75 V (sample B), 0.50 V (sample C) or
0.01 V (sample D), or galvanostatically cycled, with a Digatron
BTS-600 Battery Tester, at room temperature at a current density
of 3.0 mA g−1, which corresponded to a charge or discharge time
of 10 h. For the constant current cycling experiments, the delithi-
ation cut-off potential was 1.20 V; the lithiation cut-off potential
was either 0.50 V (sample E) or 0.01 V (sample F). Electrochem-
ical conditions used to charge and/or discharge samples A–F are
summarized in Table 1. For the potentiostatic relaxation experi-
ments, aMacPileIITM instrumentwasused. Allquotedpotentials
are versus Li0/Li+.
2. Experimental
2.1. Synthesis of AlSb
A high-frequency furnace was used for the drop synthesis of
AlSb. An ingot made from aluminium was heated to its melt-
ing point (660 ◦C) under an argon atmosphere and a pressure
of 300 mbar. Bits of antimony were dropped into the melt until
the proportion 1:1 with aluminium was reached. The AlSb ingot
was cooled and ground by hand before ball milling. The pow-
der was mixed with 3 wt.% graphite as a solid lubricant in a
glove box with an argon atmosphere (<3 ppm H2O and O2) and
placed in a SPEX (CertiPrep) stainless steel vessel; the vessel
was sealed with an o-ring to maintain the inert atmosphere. High
energy ball milling was performed for 16 h at room tempera-
ture. XRD patterns of the AlSb sample were collected before
and after ball milling on a Siemens D5000 diffractometer using
Cu K␣1-radiation.
2.3. XRD characterisation of electrodes
Powder synchrotron X-ray diffraction measurements were
performed at the I711 beamline at MAX Lab, Lund, Sweden,
using a Huber scanning plate image detector and Si for cali-
bration. Diffractograms of powder samples, measured in glass
capillaries, and of separate electrodes were recorded with a
˚
wavelength of 1.39258 A. For the electrodes, a custom made
battery holder for transmission geometry was used. The holder
was oscillated during 6× 60 s long exposures. Diffractograms
were recorded from electrodes made of non-ball milled AlSb,
in order to avoid the amorphisation of the material. Except
for that, the electrodes were made as described above. The
batteries were potentiostatically equilibrated before measure-
ment on a Bio-Logic VMP2 to 0.50 and 0.01 V, respectively,
using a 10 ␮A current cut-off. The batteries were opened in the
glove-box and the electrodes were put in a plastic bag which
was then vacuum sealed and transported to the synchrotron
facilitation. Calibrated wavelengths and detector positions were
used.
2.2. Electrode assembly
AlSb electrodes were manufactured by preparing a mixture
of 80 wt.% ball milled AlSb powder, 10 wt.% acetylene black
(Chevron) carbon powder and 10 wt.% binder (polyvinylidene
difluoride, PVDF) dissolved in N-methyl pyrrolidone (NMP).
The resulting slurry was coated on a copper foil and cut into

M. Stjerndahl et al. / Electrochimica Acta 52 (2007) 4947–4955
4949
2.4. XPS characterisation
organic compounds studied here are extremely sensitive, even
to low-energy electron bombardment [41,47,48].
For surface analysis, electrodes were immersed in DEC for
30 min before they were vacuum dried at room temperature
for 1 h. The electrodes were mounted on a sample holder in
the glove box and transported to the analysis chamber using
a specially designed transport vessel to avoid contamination
from air. Measurements were performed on a PHI 5500 system,
using monochromatized Al K␣1, at 1486.6 eV. Six electrodes,
samples A–F, were analysed. Some charging effects due to SEI-
layer products and electrolyte salt decomposition products were
observed during the XPSmeasurements. Therefore, peakassign-
ments (Table 2) were made after background correction and
binding energy calibration with reference compounds, not using
absolute binding energies. For the C 1s and O 1s spectra, cal-
ibration was made with respect to the combined carbon black
and adventitious carbon peak that was set to 284.4 eV. For the
F 1s spectra, calibration was made with respect to the LiF peak
that after careful consideration of previous measurements and
reference data was set to 686.5 eV. Since no LiF was present in
the spectra of the pristine electrode, PVDF at 687.8 eV was used
as a reference. It is worth noting that although some differential
charging problems were expected (due to carbon black having a
higher conductivity than LiF or PVDF) the shifts applied to the
C 1s, O 1s and F 1s spectra of the pristine electrode were rather
small, less than 0.1 eV. This procedure has also been described in
detail in previous reports [4,6,44]. An electron flood gun charge
neutraliser was not used, since there is evidence that the type of
3. Results and discussion
3.1. XRD characterisation of AlSb
XRD measurements of the synthesised AlSb powder before
and after ball milling, Fig. 1, show the product to be a single
cubic phase with a zinc-blende type structure. There was no
evidence of any unreacted starting material. The X-ray diffrac-
tion pattern of ball-milled AlSb samples exhibited significantly
broader peaks than unmilled samples, indicating that a reduc-
tion of particle size and a decrease in crystallinity had occurred
during the milling.
3.2. Electrochemical characterisation
Fig. 2 shows the initial cycle of lithium cells with AlSb elec-
trodes lithiated to a lower cut-off potential of 0.50 or 0.01 V,
and delithiated to an upper cut-off potential of 1.20 V. In the
figure, letters A–F represent the state of charge for each sample,
see Table 1. On lithiation, the potential drops immediately from
the original open circuit potential at 2.75–1.55 V. Thereafter, the
sloping voltage response between 1.55 and 0.90 V indicates the
electrochemical reduction of an antimony oxide surface layer
formed during the synthesis of the AlSb particles. This process
is observed only during the initial lithiation. The oxide layer is
Table 2
Summary of XPS peak assignments
Assignments
Measured binding energy (eV)
C 1s
O 1s
533b
F 1s
Sb 3d3/2
Sb 3d5/2
Carbon black
Hydrocarbon
–(CH2CH2O)n–
R–CH2–OH
Li2CO3
R–CH2OCO2Li
R–CH2OCO2Li
R–CH2OLi
–(CH2–CF2)n–
–(CH2–CF2)n–
R–CF2–R
AlSb
284.4a
285.0b
286.5b
290b
288d
532c
533.5d
532c
290–291d
288e
532c
286.2f
687.8f
687.8f
688.2g
290.7f
290.7–292b,f
537.7h
528.6h
526.9
Li3Sb
LiF
LiPF6
LixPFy
686.5i
688b
687–688b
LixPFyOz
∼535b
a
Ref. [44].
Ref. [6].
b
c
With careful consideration of experimental data and Refs. [6,41,42,49] we have assigned this peak to 532 eV.
d
e
f
Ref. [41].
Ref. [42].
Ref. [46].
Ref. [45].
Ref. [43].
Ref. [4].
g
h
i

4950
M. Stjerndahl et al. / Electrochimica Acta 52 (2007) 4947–4955
towards the end of lithiation (see next section). It is believed that
the surfaces of the finely divided and highly reactive Al particles
extruded during the reaction are quickly passivated by oxidation
in the electrolyte, thus rendering the particles electrochemically
inactive. The inset in Fig. 2 shows the lithiation capacity for the
first three cycles of cells E and F. As expected, both cells show
a significant irreversible capacity loss on the initial cycle, cell
F providing higher reversible capacity (∼400 mA h g−1) than
cell E (∼300 mA h g−1) on account of its lower cut-off potential
(0.01 V).
3.3. X-ray diffraction of electrodes
The diffractograms of electrodes equilibrated to 0.50 and
0.01 V are shown in Fig. 3, where the 2–θ scale has been con-
verted to correspond to the Cu K␣1-wavelength. At 0.50 V
several sharp peaks corresponding to “AlSb” and the packag-
ing material for the electrode can be seen. Reflections from the
packaging material are marked with (*). It is up to this point not
clear whether these main phase peaks should be indexed as AlSb
or LixAlSb. However, considering the results from the electro-
chemistry discussed earlier, this is an indication that the main
phase at 0.50 V is indeed LixAlSb. Two low-angle peaks are also
detected at 14.05◦ and 20.61◦. These peaks could not be indexed
to any known phase in the electrode or packaging material. But,
while the structure of LixAlSb is not fully understood, it could
not be ruled out that the peaks belong to this phase. Finally,
at 22.36◦, there is a small peak that can be indexed as Li2Sb.
This is taken as an indication that the extrusion process, dis-
cussed above, has partially been activated. This process would
extrude a small amount of aluminium while the same amount of
Li2Sb forms. However, the fact that no Li3Sb could be detected
at 0.50 V makes it clear that the main extrusion process, where
all aluminium is extruded and Li3Sb forms as the main phase,
Fig. 1. XRD diffractogram of synthesised AlSb taken before and after ball
milling.
believed to have formed during the ball milling process despite
the care that was taken to ensure that the atmosphere within the
SPEX-mortar reaction vessel was as inert as possible.
Between 0.90 and 0.50 V there is a gently sloping plateau
which accounts for most of the reversible capacity. This is sug-
gested to be the insertion of lithium into the interstitial sites
of the AlSb zinc-blende framework, thus forming a metastable
LixAlSb phase:
xLi + AlSb → LixAlSb
At slightly lower potentials, this process is coupled with the
concomitant extrusion of aluminium, which in the end yields
a Li3Sb product [28]. Furthermore, the sloping nature of this
plateau indicates that the lithium forms a solid-solution in AlSb.
Further reaction between 0.50 and 0.01 V is not attributed to
lithiation of the extruded aluminium to form LiAl, which is
suppressed in carbonate-based electrolytes, but rather to the
overpotential required to continue the AlSb to Li3Sb reaction
Fig. 2. Initial lithiation/delithiation curves of two Li/AlSb cells cycled between
1.20–0.50 V (circles) and 1.20–0.01 V (squares). Inset shows the specific lithi-
ation capacity for these cells for the first three 3 cycles.
Fig. 3. XRD diffractograms of lithium inserted AlSb electrodes at 0.50 and
0.01 V. (*) Marks reflections from the packaging material.

M. Stjerndahl et al. / Electrochimica Acta 52 (2007) 4947–4955
4951
Fig. 4. (a) XPS Sb 3d spectra for the AlSb reference electrode. (b) XPS Sb 3d
spectra for sample A.
Fig. 5. XPS C 1s spectra (a) and F 1s spectra (b) for the AlSb reference electrode.
binding energies ranging from 284.4 to 292.4 eV. The 284.4 eV
peak is attributed predominantly to the conductive carbon black
additive in the electrode, but also to adventitious carbon from
the vacuum chamber. The peak at 286.0 eV was used for the
intensity normalisation since it is the most dominant species
in the spectrum [41]. The peak at 285.0 eV is attributed to
hydrocarbons that likely originate from both adventitious car-
bon and the SEI-layer. The assignment of various SEI-products
at higher binding energies includes: alcohols (R–CH2–OH) and
polymeric ethers (–(CH2CH2O)n–) at 286.5–286.9 eV, Li-alkyl
have at this potential not yet begun. At 0.01 V the situation has
changed, the extrusion process discussed above is now complete
and the main phase, seen as three broad peaks at 23.3◦, 38.8◦
and 45.8◦, is thus indexed as Li3Sb. No peaks corresponding to
AlSb or LixAlSb could be detected. It is noteworthy that the two
unindexed low-angle reflections have also disappeared.
3.4. XPS Sb 3d
As mentioned earlier, the sloping voltage profile between
1.55 and 0.90 V in Fig. 2 originates predominantly from the
electrochemical reduction of an antimony oxide surface layer.
This assignment is supported by Fig. 4 which shows the Sb 3d
spectra of the reference AlSb (parent) sample prior to cell assem-
bly (Fig. 4a) and electrode sample A that has been lithiated to
0.90 V (Fig. 4b). In Fig. 4a, the Sb 3d (5/2) peak is clearly visible
at 528.2 eV and the Sb 3d (3/2) peak at 537.6 eV. The differ-
ence between these two antimony peaks is the expected 9.35 eV
[46]. At slightly higher binding energy from the antimony peaks
are the equivalent peaks for antimony oxide (SbxOy) at 530.4
and 539.7 eV, respectively, which have intensities significantly
greater than those of the AlSb peaks. This is best seen for the Sb
3d (3/2) peaks where there is no O 1s overlap. The Sb 3d (5/2)
SbxOy peak, however, overlaps with the O 1s peak at approxi-
mately 533 eV. In Fig. 4b, the same antimony peaks are shown,
namely, Sb 3d (5/2) at 528.6 eV, Sb 3d (3/2) at 537.7 eV. How-
ever, in this case, the antimony oxide peak, expected at higher
binding energy than the Sb 3d (3/2) peak at 537.7 eV, is hardly
visible.
3.5. XPS C 1s
The C 1s spectrum of a reference sample of AlSb is shown
in Fig. 5a. The spectrum consists of contributions from carbon
black (284.4 eV) and the PVDF binder. All assigned binding
energies for these peaks are given in Table 2. This spectrum
can be compared to the intensity-normalised C 1s spectra for
samples A–D, shown in Fig. 6. The C 1s spectra for A–D are
similar; the peaks could be assigned to several species with
Fig. 6. XPS C 1s spectra for samples A–D (a). Spectra are background sub-
tracted, normalised to the 286 eV peak and plotted on the same intensity scale.
The peak at 284.4 eV attributed to carbon black (b) shown for samples A–D.

4952
M. Stjerndahl et al. / Electrochimica Acta 52 (2007) 4947–4955
carbonates at approximately 287.9 and 289.7 eV, Li-alcoxides
(R–CH2OLi) also at 287.9 eV and Li-carbonate (Li2CO3) at
290 eV. The peaks at 292 eV (for A–D in Fig. 6) are assigned to
fluorinated carbons (–CF2–). These fluorinated carbons are not
present in the pristine electrode (Fig. 5a). They seem to result
from reactions between the PVDF binder and the electrolyte dur-
ing electrochemical cycling. In Fig. 6b, the carbon black peak
at 284.4 eV is emphasized to show the increase in thickness of
the layer as more and more of the carbonaceous species of the
SEI covers the electrode surface (leading to smaller and smaller
carbon black signals). In the fully lithiated sample at 0.01 V, the
SEI layer completely covers the electrode surface.
Fig. 7 compares the C 1s binding energies of a graphite elec-
trode cycled three times with AlSb electrode samples E and F.
There are no significant peak shifts in binding energy between
the three electrodes, but the relative intensity for the peaks in
sample E compared to those for graphite and for sample F shows
the carbonaceous layer to be thinner on sample E. The differ-
ence between the graphite electrode and the two AlSb electrode
samples lies mainly in the relation between the different species
present in the SEI. There are more of the Li-alkyl carbonates
or Li-alcoxides (R–CH2OLi) on graphite than on any of the
AlSb surfaces. The amount of –CF2-containing compounds is
comparable for graphite and sample F but larger than that for
sample E. A comparison of the spectra of samples E and F with
those of A–D (Fig. 6a and b) shows that the thickness of the
carbon-containing SEI components on the electrodes increases
during lithiation (A and D) and then decreases again when the
electrodes are delithiated (E and F).
the delithiated state (1.20 V), has a similar SEI thickness as that
of sample D (0.01 V, lithiated state), as depicted in Fig. 7. This
result provides strong evidence of a dynamic process in which
componentsoftheSEIlayercomeandgoduringelectrochemical
cycling. This trend is also apparent in Fig. 7 in which the peaks
in the C 1s spectrum of sample E, that has been cycled three
times between 1.20 and 0.50 V and equilibrated in the delithiated
state (1.20 V), are less intense than those of sample C (0.50 V,
lithiated state and Fig. 6a). Moreover, there are no significant
differences in the spectra of sample E and sample F (Fig. 7)
either in composition or in intensity, suggesting that the SEI
layer on AlSb electrodes grows predominantly in the potential
range 0.50–0.01 V. Because the peak intensities in the spectra
of samples E and F (in Fig. 7) are so similar, it appears that the
thickness of the SEI layer at 1.20 V is independent of the lower
cut-off potential used during the first three cycles of the cell. This
can also be compared to graphite treated in a similar way (Fig. 7).
3.6. XPS O 1s
The O 1s spectra for samples A–F are plotted in Fig. 8. The Sb
3d peaks are present in the same energy range as the O 1s spectra.
The Sb 3d (5/2) peak is clearly seen in sample A at 528.6 eV and
in samples E and F at 528.4 eV. This peak is not strongly evident
in samples B and D, but it appears in sample C at a significantly
lower binding energy than in sample A, namely at 526.9 eV. In
sample A, this peak correspond to unreacted AlSb; it disappears
on lithiation to 0.01 V (sample D), consistent with the formation
of an increasingly thick SEI layer; the weak Sb 3d (5/2) peak that
occurs at lower binding energy (526.9 eV) in sample C that was
lithiated to 0.50 V is attributed to a small amount of the lithiated
product Li2Sb. The re-emergence of the Sb 3d (5/2) peak in the
cycled electrode samples E and F at 528.4 eV is consistent with
a decreasing thickness of the SEI layer during delithiation to
1.20 V, as described earlier.
It is evident from the decrease in intensity of the carbon black
peaks (Fig. 6b) that the thickness of the SEI layer increases as
the equilibrium potential of the cell is lowered. For example, the
C 1s peaks of sample D lithiated to 0.01 V give a significantly
thicker SEI than those of sample C that has been lithiated to
0.50 V. Another particularly striking feature of the XPS data is
that the peaks in the C 1s spectrum of sample F, that had been
cycled three times between 1.20 and 0.01 V and equilibrated in
The strongest O 1s peak originates from Li-alkyl car-
bonates (R–CH2OCO2Li) at approximately 533.5 eV. At the
Fig. 7. XPS C 1s spectra for graphite, samples E and F plotted on the same intensity scale.

M. Stjerndahl et al. / Electrochimica Acta 52 (2007) 4947–4955
4953
Li-alkyl carbonates in the SEI layer are indeed dynamic. By con-
trast, the O 1s peaks associated with salt degradation products
appear essentially in the same position, suggesting that the for-
mation/deposition processes for such products in the SEI layer
are irreversible. It is also important to note that the major O 1s
peakanditsshouldertotherightinsampleEthathadbeencycled
between 1.20 and 0.50 V appear at slightly lower binding energy
compared to sample F that had been cycled between 1.20 and
0.01 V. This observation implies that lithiating AlSb electrodes
to 0.01 V will generate a SEI layer richer in Li-alkyl carbonates
and salt degradation products than the SEI layer formed when
using a 0.50 V cut-off potential.
3.7. XPS F 1s
Fig. 9 shows the F 1s spectra for samples A–F. These can be
compared to the F 1s spectrum of the pristine electrode shown
in Fig. 5b. In these spectra, the background has been subtracted
Fig. 8. XPS O 1s spectra for samples A–F.
lower binding energy side of this peak, there is a small
shoulder at approximately 532.5 eV originating from Li-alkyl
carbonates (R–CH2OCO2Li) and Li-alcoxides (R–CH2OLi) as
well as lithium carbonate. The more pronounced shoulder at
535–536 eV, i.e., at a higher binding energy to the main peak,
is attributed to the LiPF6 electrolyte salt and its degradation
products. It is evident that there is a small continuous shift
of the peaks from 533.3 eV in sample A to higher binding
energies in this series. At the same time, the intensity for the
electrolyte salt peak increases continuously from samples B to
D. These features are attributed to increasing amounts of elec-
trolyte salt degradation products as the potential is decreased to
0.01 V.
Interestingly, in the O 1s spectra of the cycled electrodes
(samples E and F) in Fig. 8, the main O 1s peak and the shoulder
on the lower binding energy side revert back to their original
positions, indicating that the formation/deposition processes of
Fig. 9. XPS F 1s spectra for samples A–F.

4954
M. Stjerndahl et al. / Electrochimica Acta 52 (2007) 4947–4955
and the binding energy scale shifted so that the peak correspond-
ing to LiF at the lower binding energy is set to 686.5 eV. The
broad peak characterising sample A could be resolved into two
peaks by curve-fitting (indicated by the dotted line in Fig. 9),
one centred at 686.5 eV corresponding to LiF, and the second at
688 eV corresponding to the PVDF binder and LiPF6 salt. By
contrast, two peaks at 686.5 and 689 eV in the spectra of samples
B–F are clearly resolved; they are attributed to LiF and LiPF6
salt degradation products, respectively. The shoulder at 692 eV
seen for samples E and F is attributed to R–CF2 bonds. From
a quantitative standpoint, Sample F that has been cycled three
times to 0.01 V shows the highest concentration of LiPF6 salt
degradation products.
A combination of the X-ray diffraction data, the electro-
chemical results and the surface characterisation, shows that the
overpotential needed to transform AlSb completely to Li3Sb
(at potentials below 0.50 V) leads to a formation of an extra
thick SEI layer. It stands to reason, therefore, that AlSb elec-
trodes that are lithiated to slightly higher potentials, such as
sample E (0.50 V) with thinner SEI layers and less electrolyte
salt degradation products, should provide superior long-term
cycling behaviour to electrodes that are lithiated to 0.01 V.
[2] E. Peled, D. Golodnitsky, G. Ardel, J. Electrochem. Soc. 144 (1997) L208.
[3] D. Aurbach, O. Chusid, J. Power Sources 68 (1997) 463.
[4] A.M. Andersson, K. Edstro¨m, J. Electrochem. Soc. 148 (2001) A1100.
[5] C.S. Wang, A.J. Appleby, F.E. Little, J. Electroanal. Chem. 519 (2002) 9.
[6] A.M. Andersson, M. Herstedt, A.G. Bishop, K. Edstro¨m, Electrochim. Acta
47 (2002) 1885.
[7] A.M. Andersson, A. Henningson, H. Siegbahn, U. Jansson, K. Edstro¨m, J.
Power Sources 119–121 (2003) 522.
[8] M. Inaba, H. Tomiyasu, A. Tasaka, S.-K. Jeong, Z. Ogumi, Langmuir 20
(2004) 1348.
[9] A. Augustsson, M. Herstedt, J.-H. Guo, K. Edstro¨m, G.V. Zhuang, P.N.
Ross, J.-E. Rubensson, J. Nordgren, Phys. Chem. Chem. Phys. 6 (2004)
4185.
[10] H. Ota, T. Akai, H. Namita, S. Yamaguchi, M. Nomura, J. Power Sources
119–121 (2003) 567.
[11] G.V. Zhuang, P.N. Ross, Electrochem. Solid-State Lett. 6 (2003) A136.
[12] W. Choi, J.Y. Lee, H.S. Lim, Electrochem. Commun. 6 (2004) 816.
[13] S. Sharma, L. Fransson, E. Sjo¨stedt, L. Nordstro¨m, B. Johansson, K.
Edstro¨m, J. Electrochem. Soc. 150 (2003) A330.
[14] L. Fransson, E. Nordstro¨m, K. Edstro¨m, L. Ha¨ggstro¨m, J.T. Vaughey, M.M.
Thackeray, J. Electrochem. Soc. 149 (2002) A736.
[15] J. Wolfenstine, S. Campos, D. Foster, J. Read, W.K. Behl, J. Power Sources
109 (2002) 230.
[16] D. Larcher, L.Y. Beaulieu, D.D. MacNeil, J.R. Dahn, J. Electrochem. Soc.
147 (2000) 1658.
[17] L.M.L. Fransson, J.T. Vaughey, R. Benedek, K. Edstro¨m, J.O. Thomas,
M.M. Thackeray, Electrochem. Commun. 3 (2001) 317.
[18] I. Amadei, S. Panero, B. Scrosati, G. Cocco, L. Schiffini, J. Power Sources
143 (2005) 227.
4. Conclusion
[19] J. Xie, X.B. Zhao, G.S. Cao, M.J. Zhao, S.F. Su, J. Alloys Compd. 393
(2005) 283.
[20] C.M. Ionica, P.E. Lippens, J.O. Fourcade, J.-C. Jumas, J. Power Sources
146 (2005) 478.
[21] L.M.L. Fransson, J.T. Vaughey, K. Edstro¨m, M.M. Thackeray, J. Elec-
trochem. Soc. 150 (2003) A86.
[22] O. Mao, R.A. Dunlap, J.R. Dahn, J. Electrochem. Soc. 146 (1999) 405.
[23] O. Mao, R.A. Dunlap, J.R. Dahn, J. Electrochem. Soc. 146 (1999) 414.
[24] A. Bonakdarpour, K.C. Hewitt, T.D. Hatchard, M.D. Fleischauer, J.R.
Dahn, Thin Solid Films 440 (2003) 11.
The SEI-layer on ball milled AlSb electrodes exhibits a
dynamic behaviour in the sense that it grows significantly during
the first lithiation, reaching a maximum thickness at 0.01 V with
most of the SEI layer being deposited between 0.50 and 0.01 V.
At 0.01 V, the SEI is comprised predominantly of Li-alkyl car-
bonatesandelectrolytesaltdegradationproducts. Onsubsequent
cycling, the SEI-layer on delithiated electrodes decreases almost
to the same thickness as before lithiation. Li-alkyl carbonate
formation in the SEI layer is found to be dynamic, whereas the
formation of salt degradation products is not. These results are
linked to the overpotential needed at low potentials to transform
remaining AlSb to Li3Sb. Our conclusion is that the accumu-
lation of salt degradation products on cycling AlSb electrodes
to 0.01 V and, subsequently, the formation of an increasingly
thick SEI layer are amongst the reasons why AlSb electrodes
demonstrate superior cycle life if only lithiated to 0.50 V versus
Li0/Li+ [28].
[25] S. Sharma, J.K. Dewhurst, C. Ambrosch-Draxl, Phys. Rev. B: Condens.
Matter Mater. Phys. 70 (2004) 104110/1.
[26] H. Tostmann, A.J. Kropf, C.S. Johnson, J.T. Vaughey, M.M. Thackeray,
Phys. Rev. B: Condens. Matter Mater. Phys. 66 (2002) 014106/1.
[27] C.S. Johnson, J.T. Vaughey, M.M. Thackeray, T. Sarakonsri, S.A. Hackney,
L. Fransson, K. Edstro¨m, J.O. Thomas, Electrochem. Commun. 2 (2000)
595.
[28] J.T. Vaughey, C.S. Johnson, A.J. Kropf, R. Benedek, M.M. Thackeray,
H. Tostmann, T. Sarakonsri, S. Hackney, L. Fransson, K. Edstro¨m, J.O.
Thomas, J. Power Sources 97 (2001) 194.
[29] Z.-F. Hou, Z.-Z. Zhu, M.-C. Huang, Y. Yong, Phys. Chem. Commun. (2003)
47–50.
[30] H. Honda, H. Sakaguchi, Y. Fukuda, T. Esaka, MRS Bull. 38 (2003) 647.
[31] A. Trifonova, M. Wachtler, M.R. Wagner, H. Schroettner, C. Mitterbauer,
F. Hofer, K.-C. Mo¨ller, M. Winter, J.O. Besenhard, Solid State Ionics 168
(2004) 51.
[32] A. Trifonova, M. Wachtler, M. Winter, J.O. Besenhard, Ionics 8 (2002)
321.
[33] F.J. Fernandez-Madrigal, P. Lavela, C. Perez Vicente, J.L. Tirado, J.C.
Jumas, J. Olivier-Fourcade, Chem. Mater. 14 (2002) 2962.
[34] H. Li, L. Shi, W.X. Lu, Huang, L. Chen, J. Electrochem. Soc. 148 (2001)
A915.
[35] M. Wachtler, J.O. Besenhard, M. Winter, J. Power Sources 94 (2001) 189.
[36] J. Yang, Y. Takeda, N. Imanishi, T. Ichikawa, O. Yamamoto, Solid State
Ionics 135 (2000) 175.
Acknowledgements
This work has been supported by the Swedish Research
Council (VR) and the Goran Gustafsson Foundation. We also
¨
thank Prof. John O. Thomas at the Department of Materials
Chemistry at Uppsala University for many fruitful discus-
sions. Financial support for JTV and MMT from the Office of
FreedomCar and Vehicle Technologies of the US Department
of Energy under Contract No. W31-109-Eng-38 is gratefully
acknowledged.
[37] H. Yang, Y. Takeda, Q. Li, N. Imanishi, O. Yamamoto, J. Power Sources
90 (2000) 64.
[38] J.O. Besenhard, M. Wachtler, M. Winter, R. Andreaus, I. Rom, W. Sitte, J.
Power Sources 81–82 (1999) 268.
References
[1] E. Peled, J. Electrochem. Soc. 126 (1979) 2047.

M. Stjerndahl et al. / Electrochimica Acta 52 (2007) 4947–4955
4955
[39] J.T. Vaughey, L. Fransson, H.A. Swinger, K. Edstro¨m, M.M. Thackeray, J.
Power Sources 119–121 (2003) 64.
[45] G. Beamson, D. Briggs, High Resolution XPS of Organic Polymers, Wiley,
Chichester, 1992.
[40] M.R. Wagner, P.R. Raimann, A. Trifonova, K.-C. Mo¨ller, J.O. Besenhard,
M. Winter, Electrochem. Solid-State Lett. 7 (2004) A201.
[41] G. Zhuang, Y. Chen, P.N. Ross, Langmuir 15 (1999) 1470.
[42] L.J.J. Rendek, G.S. Chottiner, D.A. Scherson, J. Electrochem. Soc. 149
(2002) E408.
[43] W.E. Morgan, W.J. Stec, J.R. Van Wazer, Inorg. Chem. 12 (1973) 953.
[44] M. Herstedt, M. Stjerndahl, A. Nyte´n, T. Gustafsson, H. Rensmo, H. Sieg-
bahn, N. Ravet, M. Armand, J.O. Thomas, K. Edstro¨m, Electrochem.
Solid-State Lett. 6 (2003) A202.
[46] A. Tressaud, E. Papirer, F. Moguet, G. Nanse, P. Fioux, Carbon 35 (1997)
175.
[47] K. Wang, Surface Science Studies of Electrochemical Energy Storage
Devices. PhD Thesis, Case Western Reserve University, 1992.
[48] C. Wagner, W. Riggs, L. Davis, J. Moulder, G. Muilenbery, Handbook of X-
ray Photoelectron Spectroscopy, Perkin-Elmer Corp., Physical Electronics
Division, Eden Prairie, MN, 1979.
[49] R. Dedryve´re, L. Gireud, S. Grugeon, S. Laruelle, J.-M. Tarascon, D.
Gonbeau, J. Phys. Chem. B 109 (2005) 15868.