Electrochimica Acta 105 (2013) 83–91
View metadata, citation and similar papers at core.ac.uk brought to you by CORE
Contents lists available at SciVerse ScienceDirect provided by Elsevier - Publisher Connector
Electrochimica Acta
jo urnal homepage: www.elsevier.com/locate/electacta
Consequences of air exposure on the lithiated graphite SEI
a a b b
Sara Malmgren , Katarzyna Ciosek , Rebecka Lindblad , Stefan Plogmaker ,
a,b b a a,∗
Julius Kühn , Håkan Rensmo , Kristina Edström , Maria Hahlin
a
Department of Chemistry, Ångström Laboratory, Uppsala University, Box 538, 751 21 Uppsala, Sweden
b
Department of Physics and Astronomy, Uppsala University, Box 516, 751 20 Uppsala, Sweden
a r t i c l e i n f o a b s t r a c t
Article history: In the present work, consequences of air exposure on the surface composition of one of the most reactive
Received 15 February 2013
lithium-ion battery components, the lithiated graphite, was investigated using 280–835 eV soft X-ray
Received in revised form 15 April 2013
photoelectron spectroscopy (SOXPES) as well as 1486.7 eV X-ray photoelectron spectroscopy (XPS) (∼2
Accepted 16 April 2013
and ∼10 nm probing depth, respectively). Different depth regions of the solid electrolyte interphase
Available online 29 April 2013
(SEI) of graphite cycled vs. LiFePO4 were thereby examined. Furthermore, the air sensitivity of samples
subject to four different combinations of pre-treatments (washed/unwashed and exposed to air before
Keywords:
or after vacuum treatment) was explored. The samples showed important changes after exposure to
XPS
SEI air, which were found to be largely dependent on sample pre-treatment. Changes after exposure of
Graphite unwashed samples exposed before vacuum treatment were attributed to reactions involving volatile
Air species. On washed, air exposed samples, as well as unwashed samples exposed after vacuum treatment,
LiPF6 effects attributed to lithium hydroxide formation in the innermost SEI were observed and suggested to
be associated with partial delithiation of the surface region of the lithiated graphite electrode. Moreover,
effects that can be attributed to LiPF6 decomposition were observed. However, these effects were less
pronounced than those attributed to reactions involving solvent species and the lithiated graphite.
© 2013 The Authors. Published by Elsevier Ltd. Open access under CC BY-NC-ND license.
1. Introduction are still exposed to air prior to analysis. To gain improved insight
into how to interpret such results, further studies on how different
Lithium-ion batteries are widely used in portable electronics components are affected by exposure to air would be beneficial.
equipment. They are also highly interesting for larger scale energy The solid electrolyte interphase (SEI) [3] is a vital component in
storage in, e.g., electric vehicles, and may hence play an important lithium-ion batteries [4]. The SEI is situated at the interface between
role in decreasing the use of fossil fuels to mitigate global warm- the anode and the electrolyte. In lithiated state, the anode is highly
ing. The higher energy densities of lithium-ion batteries compared reductive and commonly used electrolytes are stable only if a sta-
to many other rechargeable battery technologies present a major ble passivating layer (SEI) can be formed at the anode/electrolyte
advantage in mobile applications. These higher energy densities are interface. The SEI not only lowers the rate of further electrolyte
associated with, e.g., the relatively higher voltage (∼4 V) of lithium- reductive decomposition, but is also known to decrease graphite
ion batteries compared to many alternative battery chemistries. exfoliation [4]. Anode ageing has been largely attributed to changes
At these voltages, non-aqueous electrolytes are required to avoid, at the anode/electrolyte interface [5], and insufficient battery life-
e.g., hydrogen evolution at the anode, and even low levels of water time is considered as a major issue impeding the large scale use
contamination are well-known to be associated with accelerated of lithium-ion batteries in, e.g., electric vehicles. It is thus highly
ageing. Also on post mortem characterization, exposure to air has important to improve the anode/electrolyte interfaces.
been shown to be associated with significant changes to, e.g., the Due to the thinness of SEI layers, which as an example on lithi-
anode/electrolyte interfaces [1,2]. However, in some cases, samples ated graphite electrodes has been estimated to ∼20 nm [6], post
mortem measurements using surface sensitive techniques are fre-
quently used to characterize the SEI. However, the experimental
challenges in post mortem characterization of the SEI are signif-
icant [7]. The air sensitive nature of the SEI presents one out of
many obstacles, casting doubt upon results of measurements where
air exposure is not avoided – for example during sample transfer
∗ from argon atmosphere to the measurement equipment. To facili-
Corresponding author. Tel.: +46 184713736; fax: +46 18513548.
E-mail address: [email protected] (M. Hahlin). tate experimental design and interpretation, improved insight into
0013-4686 © 2013 The Authors. Published by Elsevier Ltd. Open access under CC BY-NC-ND license. http://dx.doi.org/10.1016/j.electacta.2013.04.118
84 S. Malmgren et al. / Electrochimica Acta 105 (2013) 83–91
the consequences of exposing the lithiated graphite SEI to air under
different conditions is helpful and the motivation for this study.
However, the results are more general than this. There is also a
need to understand how detrimental short-time exposure to air
is to the interiors of a lithium-ion battery, especially for the most
reactive component of the battery – the lithiated graphite.
The lithiated graphite SEI contains a range of different compo-
nents that could react with air on exposure. The LiPF6 salt may
hydrolyze and form toxic HF gas. HF has been proposed to convert
commonly suggested SEI species like lithium carbonate, lithium
alkyl carbonate, lithium oxide and lithium alkoxide into lithium
fluoride [8], and also to attack ether linkages to form fluorinated
organic species [2]. POF3 gas formed during such hydrolysis reac-
tions has been suggested to further hydrolyze to form solid OPF2OH
and OPF(OH)2 species [9]. However, the kinetics of LiPF6 hydrolysis
is reported to be relatively slow; the reaction can take days to equili-
brate in a mixed carbonate solution at ambient temperature [9–13].
Water and oxygen contamination in ethylene carbonate (EC) and
diethylene carbonate (DEC) electrolyte solvents subjected to low Fig. 1. Potato-shaped graphite from Toyo Tanso.
potentials has been found to be associated with lithium hydrox-
ide and lithium oxide formation in the DEC based electrolyte, and
The different probing depths of these measurements (∼2 and
with conversion of lithium alkyl carbonate into lithium carbonate
∼10 nm, respectively) enabled comparison of how air exposure
in the EC based electrolyte [14]. Also several different SEI compo-
affects the outer and the inner parts of the SEI. Furthermore, the
nents formed during cell cycling could react with air. For example,
consequences of air exposure on samples subject to different
spontaneous combustion of lithium alkoxides reacting with oxy-
pre-treatments (washed/unwashed and exposed before or after
gen gas has been proposed [2]. Lithium oxide conversion to lithium
vacuum treatment) were compared. Thereby, improved insight
hydroxide and/or lithium carbonate has also been suggested [15].
into how different electrode/electrolyte interface components
Finally, the lithium in the lithiated graphite surface is also highly
react on exposure to air could be obtained.
reactive on exposure to air.
X-ray photoelectron spectroscopy (XPS) is a surface sensi-
2. Experimental
tive technique with a probing depth (∼10 nm) suitable for the
characterization of Li-ion battery SEI layers. Previously, XPS
2.1. Sample preparation
measurements have been performed on air-exposed and cycled
delithiated graphite electrodes from cells with LiPF6-containing
Graphite/LiFePO pouch cell (polymer coated aluminium) bat-
electrolytes that had been subjected to air for 3.5 h [1]. These studies 4
teries were prepared inside an argon filled glovebox (≤1 ppm H O)
showed an increased oxygen signal, an increased P–O compound 2
as described in detail elsewhere [6]. Anodes (85 wt% graphite,
contribution and an increased graphite signal after exposure to
3 wt% KS6, 2 wt% Super P amorphous carbon and 10 wt% binder
air. The latter effect was interpreted as formation of volatile com-
(Kynarflex)) and cathodes (75 wt% carbon coated LiFePO , 10 wt%
pounds during exposure consuming the SEI layer and making it 4
carbon black and 15 wt% binder (Kynarflex 2801-00)) were
thinner. Some binding energy shifts were observed in the oxy-
matched to obtain a graphite overcapacity of 20%. 1 M LiPF (Ferro)
gen spectrum, which were interpreted as formation of lithium 6
in EC:DEC mixed in volume ratio 2:1 (Novolyte technologies) and
carbonate upon exposure. Considering the increased reactivity
Solupor separator were used. Cells were galvanostatically cycled
of lithiated compared to delithiated graphite, and the possible
between 2.7 and 4.2 V at C/10 for 3.5 cycles.
kinetic limitations of different processes associated with expo-
It is known that the SEI is influenced by a number of different
sure to air, it is also interesting to study the consequences of
properties of the graphite active material [7]. The graphite particles
exposing lithiated graphite samples to air for shorter periods of
selected here have smooth surface and the particle size is about
time more similar to practical air exposure times during, e.g.,
20 m, as illustrated in Fig. 1. The anode film thickness was in the
transfer of samples between different lab equipments. Shorter
2
range 25–40 mm. The BET surface area was 3.2 m /g, with majority
air exposure time (10 min) has previously been used in a study
of pores above 50 nm pore width.
of washed and soft-baked lithiated silicon SEI layers [2]. The
Batteries were opened inside an argon filled glovebox (≤1 ppm
exposed silicon samples showed decreased carbon and phospho-
H O). The battery anode was cut in four pieces subject to four dif-
rous intensities, while the lithium signal increased. It was shown 2
ferent combinations of sample pre-treatments (washed/unwashed
that the changes in SEI composition, seen on exposure to air, were
and exposed to air before or after vacuum treatment (uAV, uVAV,
highly dependent on the electrochemical histories of the sam-
wAV, and wVAV)), see schematic representation in Fig. 2. Washed
ples, which were prepared through, e.g., cyclic voltammetry and
chronoamperometry. samples were obtained by letting 1 ml of DMC (>99.9%) run over the
sample, leaving the sample to dry and then repeating the process a
In the present work, consequences of short-time air exposure
second time. Samples exposed to air before insertion into vacuum
(in this case 4 min, estimated to be in the order of time for sample
(uAV and wAV) were measured first to decrease solvent evapo-
transfers between different instrumentations used for investi-
ration prior to air exposure. Meanwhile, the electrode pieces to be
gations of such battery surfaces) on the surface composition of
vacuum treated prior to air exposure (uVAV and wVAV) were stored
one of the most reactive components of a lithium-ion battery,
for a few hours inside the glovebox, still within the opened battery
the lithiated graphite, were examined using photoelectron spec-
and with the separator remaining on the sample. Vacuum treat-
troscopy. The examined cells were prepared through galvanostatic
−8
ment was performed through storage in high vacuum (∼10 mbar)
cycling at C/10. Both synchrotron based soft X-ray photoelectron
while reference measurements on the sample in unexposed state
spectroscopy (SOXPES) and more bulk sensitive in-house X-ray
were performed. The air exposure lasted for 4 min.
photoelectron spectroscopy (XPS) measurements were performed.
S. Malmgren et al. / Electrochimica Acta 105 (2013) 83–91 85
Fig. 2. The four different combinations of sample pre-treatments selected for the air exposure study. One electrode was cut into four pieces where two pieces were washed
and two pieces left unwashed. The electrodes were then exposed to air either after vacuum treatment (during which a reference spectrum of the sample in unexposed state
was recorded) or before insertion into vacuum. The samples are in the text referred to as uRef, uVAV, uAV, wRef, wVAV, and wAV, according to the figure.
2.2. Transfer from glovebox to spectrometer – a new transfer different instruments without having to find a new sample transfer
system solution for each instrument, versatile equipment for sample trans-
fer is desirable. In the present work, we have developed a flexible
Samples to be exposed before insertion into vacuum, uAV and transfer system and used it for transfer of samples to the SOXPES
wAV, were transferred through air (1–25 mbar partial pressure of system. The transfer system can be used for moving samples from
H2O) from the glovebox to the XPS/SOXPES instrument. Samples an argon filled glovebox to any vacuum instrument via a C35 flange
to be exposed to air after vacuum treatment, uVAV and wVAV, without exposing the samples to air. It is also flexible in the sense
were transferred from the argon filled glovebox to the measure- that it either can incorporate the commonly used omicron sam-
ment system using air-tight transfer units attached directly to the ple transfer system or be used with an instrument specific sample
XPS/SOXPES instrument. holder.
To enable sample transfer from a glovebox to a spectrometer The transfer system developed in the present work is built
without exposure to air, we have previously used several different up by two parts (see Fig. 3). The first part is an o-ring sealed
specially built or commercially available systems (e.g., the system adapter that can be plugged into the load lock of the glovebox.
used in the present work to transfer samples to XPS measure- The adapter has a C35 flange, onto which the second, portable
ments). However, these systems have not been easily adaptable to part of the transfer system can be mounted. The second part is a
fit many different instruments. To enable easy access to a range of portable unit and consists of a magnetic rod mounted on a small
Fig. 3. Photograph of the transfer chamber used in this study. The adapter (a) can be plugged into the load lock of a glovebox and either sealed using a flange (b) or attached
to the portable unit (c–g). The portable unit consist of either an omicron fork (c) or an instrument specific sample transfer magazine (d) mounted onto a magnetic rod (g)
long enough to allow easy access to the fork or sample magazine inside the glovebox. Also a small vacuum chamber (f) with a valve (e) is attached to the magnetic rod.
86 S. Malmgren et al. / Electrochimica Acta 105 (2013) 83–91
vacuum chamber with a valve. On the magnetic rod, either a sample 3. Results and discussion
magazine or the commonly used omicron sample transfer system
can be mounted to match the existing experimental instrumen- 3.1. Synchrotron based SOXPES measurements
tation. The long magnetic rod allows easy access to the sample
magazine/omicron sample-carrying fork inside the glovebox. The Fig. 4a and b shows the results of the most surface sensitive
∼
portable unit can be attached to any C35 flange, and can thus apart air exposure experiments ( 2 nm probing depth) on unwashed
from being mounted onto the adapter also be attached to the load (abbreviated uRef, uAV, and uVAV, see Fig. 2), and washed (abbrevi-
lock of a range of different experimental instruments. The versatile ated wRef, wAV, and wVAV, see Fig. 2), lithiated graphite electrodes,
system thus provides easy access to a range of different experimen- respectively. For unwashed samples, the figure shows both spec-
tal techniques without exposing samples to air during the transfer tra of electrodes exposed after vacuum treatment (uVAV) (red) and
from the glovebox to the measurement equipment. of electrodes exposed before insertion into vacuum (uAV) (light
green). The results are compared to spectra of unexposed unwashed
(uRef) (Fig. 4a) and washed (wRef) (Fig. 4b) reference electrodes
2.3. Photoelectron spectroscopy
(black, dotted), which were recorded during the vacuum treat-
ment of the uVAV and wVAV samples, respectively. In general,
Soft X-ray photoelectron spectroscopy (SOXPES) was performed
only smaller differences were observed between unexposed and
at the I411 beamline [16] at the MAX IV laboratory (Lund, Sweden).
air-exposed sample spectra. Therefore, these spectra were plotted
The excitation energies (P2p: 280 eV, C1s: 430 eV, O1s: 680 eV, F1s:
on top of each other to better illustrate the changes seen upon expo-
835 eV) were chosen so that the kinetic energies of emitted elec-
sure to air. The core level spectra were normalized with respect to
trons were ∼140 eV. Since the probing depth depends on the kinetic
the background.
energy of emitted electrons, the probing depths of the different core
The most important differences between the unexposed refer-
level measurements were thus very similar. The emission angle was
◦
60 . ence samples and the exposed samples can be found in the O1s and
C1s spectra of unwashed samples exposed to air before insertion
More depth sensitive X-ray photoelectron spectroscopy (XPS)
into vacuum (uAV) shown in Fig. 4a. In these spectra, much more
measurements utilizing 1486.7 eV AlK␣ radiation were performed
pronounced 531.8 and 532.9 eV O1s features as well as 286.2 and
at an in house Perkin Elmer PHI 5500 system. The emission angle
◦
289.8 eV C1s features are observed. These features are attributed
was 45 . The probing depth of the SOXPES and XPS measurements
to ether and carbonate environments. Since no similar changes are
is estimated to ∼2 and ∼10 nm, respectively, defining the probing
observed on exposure of vacuum treated (Figs. 4a and b) (uVAV)
depth as the emission depth including 95% of the elastically emitted
and washed (Fig. 4b) samples (wVAV and wAV) these changes
electrons and approximating the electron inelastic mean free path
are attributed to reactions involving air and volatile components
(IMFP) by the IMFP in polyethylene [17].
present in the sample. On exposure, volatile species, such as, e.g.,
XPS spectra were energy scale calibrated with respect to the
electrolyte solvent residues, can react to form less volatile compo-
C–H/C–C (hydrocarbon) feature set to 284.4 eV based on results
nents, which to a higher degree remain on the surface of the sample
of comparisons of anode and cathode spectra [6]. SOXPES spec-
at the low pressures inside the SOXPES system.
tra were energy calibrated by setting the peak maxima of the most
All exposed sample spectra (Fig. 4) show some decrease in the
intense feature in the respective core level spectrum to 284.4 eV
P2p signals. Furthermore, a more pronounced lower binding energy
(C1s), 687 eV (F1s), 531.8 eV (O1s) and 136.5 eV (P2p). These bind-
F1s component, attributed to LiF, with respect to the higher bind-
ing energies were obtained from previous results on similarly
ing energy component, assigned to P–F containing compounds, is
prepared unexposed, unwashed samples energy calibrated with
observed. These changes may be attributed to LiPF decomposition
respect to the C–H/C–C feature set to 284.4 eV [6]. In the present 6
to form P–F containing gas (e.g., PF , POF , or both) and LiF. More-
work, the number of measurement scans was kept to a minimum 5 3
over, a slightly more pronounced lower binding energy feature in
in order to reduce possible radiation damage.
the P2p spectra was generally also seen in air exposed sample mea-
Curve fits were performed using 70% Gaussian 30% Lorentzian
surements. This feature is assigned to P–O containing compounds,
Voigt peak shapes. The lithiated graphite peak was also convoluted
which have been suggested to form on, e.g., LiPF hydrolysis [9].
with an asymmetric Gelius peak with peak shape parame- 6
However, the amounts are not large and one reason could be the
ters in agreement with the CASA reference library [22], fitting
slow kinetics reported for the LiPF hydrolysis in carbonate solu-
well to XPS spectra of pure graphite powder acquired in our 6
lab. tions at ambient temperature [9–13].
In general, the fact that only smaller changes are seen in the
XPS spectra were normalized to show the relative amounts of
outermost region of the SEI on exposing unwashed and washed
the different elements, i.e., spectra were divided by spectrum area
samples after vacuum treatment (uVAV and wVAV) indicate that
and multiplied by the detected relative amount of the correspond-
volatile species play an important role in the chemical reactions
ing element. The relative amounts are here defined as the relative
occurring in the outermost SEI during exposure to air.
Scofield cross section [18], , and polyethylene mean free path [17],