Analytical and Bioanalytical Chemistry

, Volume 400, Issue 3, pp 691–696

XPS investigations of electrolyte/electrode interactions for various Li-ion battery materials

Authors

    • IFW Dresden
  • D. Mikhailova
    • IFW Dresden
  • F. Scheiba
    • IFW Dresden
  • P. Reichel
    • IFW Dresden
  • A. Fiedler
    • IFW Dresden
  • H. Ehrenberg
    • IFW Dresden
Technical Note

DOI: 10.1007/s00216-010-4646-z

Cite this article as:
Oswald, S., Mikhailova, D., Scheiba, F. et al. Anal Bioanal Chem (2011) 400: 691. doi:10.1007/s00216-010-4646-z

Abstract

For future Li-ion battery applications the search for both new design concepts and materials is necessary. The electrodes of the batteries are always in contact with electrolytes, which are responsible for the transport of Li ions during the charging and discharging process. A broad range of materials is considered for both electrolytes and electrodes so that very different chemical interactions between them can occur, while good cycling behavior can only be obtained for stable solid-electrolyte interfaces. X-ray photoelectron spectroscopy (XPS) was used to study the most relevant interactions between various electrode materials in contact with different electrolyte solutions. It is shown how XPS can provide useful information on reactivities and thus preselect suitable electrode/electrolyte combinations, prior to electrochemical performance tests.

https://static-content.springer.com/image/art%3A10.1007%2Fs00216-010-4646-z/MediaObjects/216_2010_4646_Figa_HTML.gif
Figure

Characteristic changes of the Li1s XP-spectra at Li2O2 powder after storage in LiPF6 for various time point to a LiF formation

Keywords

Li ion batteryElectrolyteSurface interactionXPS

Introduction

Li-ion batteries are of growing interest, beside their application in many mobile devices, especially for future use in electric and hybrid vehicles. The increase of the specific energy density is a major challenge. Therefore, new materials and new working principles are under investigation. Basically all batteries consist of two electrodes (cathode and anode) coupled by an ionically conductive electrolyte medium. The electrolyte enables the charge carrier transport (Li+ ions) during charging and discharging, but must guarantee high stability and low self-discharge. Therefore, a key requirement for an electrode/electrolyte combination is its chemical and electrochemical stability.

X-ray photoelectron spectroscopy (XPS) is a well-accepted method for the investigation of the chemical behavior of thin films and surfaces. For example, surface-near changes of the electronic structure of LixCrMnO4 as the cathode material in a Li-ion battery were investigated. Therefore, a “quasi in situ” method was established by the use of a dedicated transfer chamber with Ar atmosphere to protect the samples in different states of charge and fatigue against changes/contamination due to contact with air during the transfer from glove box into spectrometer [1]. It was revealed that the poor cycle stability of LixCrMnO4 in the highly charged state obtained during Li-extraction up to a cut-off voltage around 5.2 V is connected with the irreversible formation of Cr(6+)-like species, which are electrochemically inactive in the following reduction process [2].

The same method was now applied to investigate the interaction between several electrolytes and various electrode materials. This work was motivated by the observation that the capacity of some materials drops already in the first cycles, depending on the specific electrolyte used. With the XPS measurements, reaction products of both the electrode and (dried) electrolyte residuals can be observed. In this paper, results are presented for LiCrMnO4, LiMoO2, and Li2O2 electrode materials after contact with electrolytes based on LiPF6, LiBOB (lithium bis(oxalato)borate) or LiClO4 dissolved in ethylene carbonate/dimethyl carbonate (EC/DMC).

Experiments

Materials synthesis

All the active or reference materials were used in powder form either prepared by ourselves or commercially available. LiCrMnO4 was synthesized following a modified Pechini process described already in [1, 2] leading to a single-phase material as confirmed by X-ray diffraction. LiMoO2 was prepared in an evacuated sealed silica tube at 950 °C from stoichiometric amounts of Li2MoO4 (Alfa Aesar, 99%) and Mo (Alfa Aesar, 99.99%). X-ray powder diffraction revealed a presence of two different modifications of LiMoO2: the rhombohedral one R-3 m with a = 2.801 Å, c = 15.507 Å, and the monoclinic one C2/m with a = 10.589 Å, b = 2.867 Å, c = 4.984 Å, β = 99.04°, which have very similar crystal structures. Li2O2 was a commercial product (Acros Organics); the considered reference materials LiF (Alfa Aesar), Li2CO3 (Fluka), and LiBOB (Chemetall), too.

Sample preparation

The materials were either investigated as pure powders (Li2O2, reference materials) or as electrode mixtures (LiCrMnO4, LiMoO2). Such mixtures are necessary to ensure reasonable mechanical stability and sufficient electrical conductivity and typically consist of 80% active material, 10% carbon black, and 10% PVDF (polyvinylidene fluoride). After intensive mixing, the materials were dried in vacuum at 100 °C. The materials (the pure ones and the cathode mixtures) were then pressed on an aluminum grid, which is also used in electrochemical experiments as current collector. The reference materials were also measured as pure powder filled in a special sample holder with molds.

Electrolyte–electrode interaction experiments

All experiments to study the electrolyte-materials interaction were performed in an argon filled glove box. The oxygen and moisture contents in the glove box were always maintained below 1 ppm. The electrolytes consist of a solution of the active conducting salt dissolved in EC and DMC in a typical weight ratio of EC:DMC = 2:1. The concentration of the active components (LiPF6, LiBOB, LiClO4) was 1 M in all cases. The cathode materials, fixed on Al grids, were soaked in the electrolyte solutions for specified periods, washed with DMC and dried in dynamic vacuum after that.

XPS measurements

The XPS measurements were carried out at a PHI 5600 CI (Physical Electronics) spectrometer which is equipped with a hemispherical analyzer operated with a typical pass energy of 29 eV and an analysis area of 800 μm in diameter. Monochromatic Al-Kα excitation (350 W) was used, additionally applying a low-energy electron charge neutralizer. To avoid any contact of the samples with air and moisture, a dedicated transfer chamber (Physical Electronics) was used to transfer the samples from the glove box to the XPS spectrometer. Unless the use of the charge neutralizer source different peak positions and/or peak shape changes induced by residual surface charging can be observed. To minimize these effects for all the shown spectra, the binding energy scale was corrected based on the C1s peak from contaminations (around 284.8 eV) or from the amorphous carbon (around 284.3 eV) as internal binding energy standard. However, this is only a first approximation, since differential charging [3] in the samples cannot be completely avoided. Otherwise, following the complex chemistry and the delicate sample handling (glove box, transfer chamber, ultra-high vacuum) products from side reaction are often found. Because of these reasons, the quantification of exact peak positions and chemical shifts is also difficult. Nevertheless, for the main peaks under discussion, peak fitting using the PHI-MULTIPAK software [4] was used to estimate the changes in chemical portions. Concentrations were estimated with the same software using standard single element sensitivity factors and assuming a homogeneous distribution of the elements in the surface-near region.

Sputter depth profiling was done using Ar+ ions (3.5 keV, scan size 2 mm × 2 mm) with conditions leading to a 1 nm/min sputter rate, determined for a Ta-oxide reference.

Results and discussion

LiCrMnO4

LiCrMnO4 as a candidate for high-voltage (>5 V against Li/Li+) cathode material was already successfully tested using the standard LiPF6-containing electrolyte. The cycle stability during charging up to 4.8 V was sufficiently good, however, only half of the expected specific capacity of the material can be reached with this low end-of-charge voltage. For complete charging up to 5.2 V, a pronounced degradation was found, which is probably caused by irreversible structural changes of the material due to the different oxidation states of the electrochemically active Cr-ions. At such high voltages, electrolyte degradation cannot be ruled out completely, but a LixCrMnO4 electrode should be stable in contact with the LiPF6-containing electrolyte.

In fact, after contact of the LiCrMnO4 material with the electrolyte for 30 min no significant changes in the spectra of Mn and Cr were found. Li at the surface converts partially to LiF, as some residual humidity in the chemicals always leads to enough HF for such a surface reaction. However, the ratio of the metal concentrations Li:Cr:Mn remained nearly constant, confirming that the LiF formation is only a very minor and neglectable side reaction.

LiMoO2

In Fig. 1, XP spectra are shown for the as prepared LiMoO2 material and after 2 h treatments with LiPF6, LiClO4 and LiBOB electrolytes, respectively. All spectra are normalized in intensity to allow for a better comparison of peak positions and shapes thus no conclusions from relative intensity variations can be drawn. In Table 1, an estimation of the peak portions Mo3d(4+/3+) and Mo3d(6+) of the electrochemical active Mo derived from a peak fit procedure is additionally presented. The Mo3d spectrum of pristine LiMoO2 without any contact with electrolyte solutions (Fig. 1b—bottom spectra) shows the presence of Mo in oxidation states between Mo(3+) and Mo(4+), according to the dependence of Mo3d5/2 binding energies on oxidation state [5]. A slight oxidation of Mo on the surface was often detected by XPS for MoO2 leading to a coexistence of both Mo(4+) and Mo(6+) species [6]. Otherwise an excess of Li in oxide environment in relation to Mo is found already in the untreated sample, which points to a Li-surface segregation during the LiMoO2 preparation.
https://static-content.springer.com/image/art%3A10.1007%2Fs00216-010-4646-z/MediaObjects/216_2010_4646_Fig1_HTML.gif
Fig. 1

XP spectra for LiMoO2 as prepared and after 2 h storage in LiPF6, LiClO4, and LiBOB (from bottom to top). Changes in the Mo valence states are visible, also LiF, Li–Cl–O and lithium carbonate formation

Table 1

Results of peak fit of the Mo(4+/3+) and Mo(6+) species at the Mo3d5/2 peak for the LiMoO2 samples after storage in different electrolytes

 

Mo (4+/3+)

Portion/%

Mo(6+)

Portion/%

BE/eV

BE/eV

Reference

229.7

38

232.8

62

LiPF6

229.8

56

232.9

44

LiClO4

LiBOB

229.7

80

232.8

20

Binding energy (BE) reference was C1s at 284.3 eV (amorphous carbon)

Preliminary XPS investigations of Mo electronic states in LixMoO2 upon lithium deinsertion in electrochemical test cells with LiPF6 electrolyte solution have shown a side reaction between lithium molybdate and the electrolyte resulting in a formation of a thick solid-electrolyte-interface(SEI)-layer, mainly containing LiF and Li carbonate, while almost no Mo can be detected anymore within the escape depths of the photo electrons. The targeted studies of the electrolyte/electrode contact were therefore performed for three different electrolyte salts. The contact with LiPF6 electrolyte (Fig. 1 second spectra from bottom) revealed an enrichment of the surface with lithium and fluorine: a LiF layer is formed fast; Li is further enriched relative to Mo at the surface. The average oxidation state of Mo was slightly shifted to Mo(6+). A further experiment with LiClO4 as electrolyte salt (Fig. 1—third spectra from bottom) revealed a similar behavior: a thick Li–Cl–O layer was formed, and not any Mo (Fig. 1b) could be observed at the interface. Also the carbonate and C–O species are enhanced (Fig. 1c).

XRD patterns of pristine LiMoO2 material and LiMoO2 after storage in LiClO4 electrolyte during 24 h shown in Fig. 2 revealed a significant shift in reflection positions toward larger 2 theta values and change of their intensities. An oxidation of LiMoO2 to isostructural Li2MoO3 by LiClO4 can be assumed.
https://static-content.springer.com/image/art%3A10.1007%2Fs00216-010-4646-z/MediaObjects/216_2010_4646_Fig2_HTML.gif
Fig. 2

Powder diffraction patterns of pristine LiMoO2 (solid line) and LiMoO2 after storage in LiClO4 electrolyte during 24 h (broken line; Cu-Kα-radiation, λ = 1.54059 Å). The reflection positions are significantly shifted toward larger 2 theta values corresponding smaller lattice parameters, which point to an oxidation of the sample

Only the electrolyte containing LiBOB showed no formation of a Li-containing overlayer (Fig. 1—top spectra). The oxidation state of Mo after contact with LiBOB was reduced to Mo(4+/3+), because the Mo3d7/2 peak at 236.5 eV corresponding to Mo(6+) has disappeared (Fig. 1b). Moreover, the Li excess on the surface was also reduced, approaching the expected LiMoO2 stoichiometry, and only traces of remaining boron were found.

Li2O2

Li2O2 is not an intercalation-type electrode material like LiCrMnO4 or LiMoO2, but formed during discharge on the cathode of a Li-air battery. In such a battery, oxygen is reduced on discharge at the cathode and reacts with lithium ions to lithium peroxide [7, 8]. Depending on the electrode kinetics of the oxygen reduction either lithium peroxide Li2O2 or lithium oxide Li2O can be formed [9]. However, on carbon-based electrodes it has been shown by ex-situ Raman [7] and X-ray diffraction [8] that the main discharge product is Li2O2. On charge lithium peroxide is decomposed again into molecular oxygen and lithium ions.

Unlike classical insertion type materials, where Li ions are extracted or inserted from or into a bulk crystal, the electrochemical reaction in a Li-air battery is an electrocatalytic process which takes place only at the electrode interface. Therefore, the interface formed between the electrocatalytically active electrode surface and lithium peroxide might strongly influence the electrode kinetics for the peroxide oxidation leading to increased overpotentials for charging the battery. Hence, the chemical stability of bulk Li2O2 was tested against different electrolytes and the resulting interfaces were analyzed using XPS.

For Li2O2 exposed to a LiPF6-containing electrolyte, an almost immediate formation of LiF was observed. Fig. 3 shows XP spectra normalized in intensity for Li2O2 compared with a LiF reference sample and Li2O2 powder exposed for 2 min, 2 h and 18 h to the LiPF6-containing electrolyte (spectra from bottom to top). In Table 2, peak fit results for Li–F and Li–O bonding parts in the Li1s spectra confirm this clearly. Prolonged exposure led to an almost complete transformation of the Li2O2 surface into LiF; after 18 h in the electrolyte the Li1s spectrum of Li2O2 is identical with the LiF reference sample. As discussed previously for LiCrMnO4, the LiF formation is most likely due to the presence of HF in the electrolyte, which originates from the decomposition of the PF6 anion by traces of water in the electrolyte. However, depth profiles obtained by argon sputtering indicate a rather thick LiF surface layer (see Fig. 4). Considering that the water concentration and hence HF concentration of the electrolyte is below 20 ppm, an attack of Li2O2 by HF is very probably not the only source of LiF formation. A direct reaction of Li2O2 with LiPF6 appears feasible and may lead to additional LiF formation.
https://static-content.springer.com/image/art%3A10.1007%2Fs00216-010-4646-z/MediaObjects/216_2010_4646_Fig3_HTML.gif
Fig. 3

XP spectra for Li2O2 as prepared, a LiF reference sample and Li2O2 after 2 min, 2 h and 18 h storage in LiPF6-electrolyte (from bottom to top). A LiF formation in contact with the electrolyte is clearly found, C–O bondings as side reaction with the electrolyte are also occurring

Table 2

Results of peak fit of the Li1s peak with respect to the portions attributed to Li–F and Li–O for Li2O2 after storage in LiPF6

 

Li–O

Portion/%

Li–F

Portion/%

BE/eV

BE/eV

Li2O2

54.3

76

55.7

24

LiF

54.4

33

55.8

67

2 min LiPF6

54.4

68

55.8

32

2 h LiPF6

54.2

44

55.7

56

18 h LiPF6

54.2

32

55.6

68

Compared with Li2O2 and LiF reference samples. Binding energy (BE) reference was C1s at 284.8 eV (hydrocarbon contamination)

https://static-content.springer.com/image/art%3A10.1007%2Fs00216-010-4646-z/MediaObjects/216_2010_4646_Fig4_HTML.gif
Fig. 4

Depth profiles for Li2O2 after 2 min, 2 h and 18 h storage in LiPF6 point to a LiF surface layer growing with time

XPS analysis of the Li2O2 surface after exposure to a LiBOB electrolyte solution showed an enrichment of boron oxide and carbonate species. This indicates an oxidative decomposition of the LiBOB salt by Li2O2 leading to the formation of boron oxide and Li2CO3.

In contrast, the XPS signature of the perchlorate anion is largely preserved even after an exposure time of 18 h, when only a weak chloride signal at 208.5 eV can be observed. While this may indicate a slow decomposition of LiClO4, it is most likely that the chloride signal originates from LiCl, already contained in the electrolyte solution as an impurity. Despite the rather good stability of the perchlorate salt, significant changes can be observed in the Li1s, O1s, and C1s spectra. This can be attributed to the formation of carbonates, which are clearly visible in the C1s spectrum. Unlike in the case of LiBOB, these carbonates cannot be a decomposition product of the electrolyte salt, but must be due to an oxidative decomposition of the electrolyte solvent.

The XPS results showed that LiPF6 and LiBOB react with Li2O2 and revealed that decomposition products of Li2O2 and the electrolyte salt can be found on the surface of the Li2O2 particles. This reaction does not only lead to a loss of lithium and electrolyte salt by forming electrochemically inactive decomposition products, but may also impact the electrocatalytic oxidation of Li2O2 by altering the interface between Li2O2 and the electrocatalytic active electrode surface. Therefore, LiPF6 and LiBOB are not suitable for the use in Li-air batteries. LiClO4 shows only weak or no decomposition and may be used as electrolyte salt. However, carbonate formation of the LiClO4 electrolyte indicates that the commonly used electrolyte solvents may also be chemically unstable in the presence of Li2O2.

The different stability of the electrolyte salt agrees well with electrochemical measurements, which showed almost no oxidation current in the LiPF6-based electrolyte, whereas in the LiClO4 electrolyte oxidation was observed above 3 V vs. Li/Li+.

Conclusions and outlook

It is shown that XPS as a standard method for examining chemical states at surfaces is also a very suitable tool for the study of electrode–electrolyte interactions in Li-ion battery research. Nevertheless, the application on real complex electrode composites is challenging by side-effects such as: surface roughness, differential charging/energy referencing, residuals of chemicals at the surfaces despite washing, residual surface alteration despite protection in a glove box and Ar-filled transfer chamber. Thus, quantitative results such as calculations of absolute surface concentrations are often not meaningful, however, the identification of the surface species resulting from the chemical reactions gives strong hints to elucidate the underlying reaction mechanisms. It can be concluded that the results for this class of materials are not predictable; the interactions between the conducting salts, the organic solvents and the oxide materials are too complex. However, conclusions about an appropriate electrolyte selection can be drawn based on XPS data in combination with results from electrochemical experiments on the same electrode/electrolyte combination.

Systematic studies on variations of electrode materials in combination with studies on suitable reference materials with XPS are planned and will be supported by spatially resolved surface investigations with high-resolution Auger-electron spectroscopy. However, stronger charging effects and electron-radiation-induced damages will be new challenges.

Acknowledgements

This work is financially supported by the BMBF (grant no. 03KP801 “Elektrochemie für Elektromobilität - Verbund Süd”)

Copyright information

© Springer-Verlag 2011