Cell - ACS Publications - American Chemical Society

Subscriber access provided by University of Rochester | River Campus & Miner Libraries

Article

Noticeable role of TFSI- anion in the carbon cathode degradation of Li-O2 cells Daniela Giacco, Marco Carboni, Sergio Brutti, and Andrea G. Marrani ACS Appl. Mater. Interfaces, Just Accepted Manuscript • DOI: 10.1021/acsami.7b05153 • Publication Date (Web): 30 Aug 2017 Downloaded from http://pubs.acs.org on August 30, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

ACS Applied Materials & Interfaces is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

Noticeable role of TFSI- anion in the carbon cathode degradation of Li-O2 cells Daniela Giacco,§ Marco Carboni, § Sergio Brutti,* ‡,# Andrea G. Marrani*§ §

Department of Chemistry, Sapienza University of Rome, P.le Aldo Moro 5, 00185 Rome, Italy

‡

Department of Science, University of Basilicata, V.le Ateneo Lucano 10, 85100 Potenza, Italy

#

Istituto dei Sistemi Complessi, Consiglio Nazionale delle Ricerche (ISC-CNR), Via dei Taurini, 00185 Rome, Italy

KEYWORDS: Li-O2 battery, carbon cathode surface reactivity, LiTFSI/TEGDME electrolyte, X-ray Photoelectron Spectroscopy, Fourier Transform Infrared Spectroscopy, Transmission Electron Microscopy.

Abstract In this work we address the phenomena at the basis of the performance loss in a Li-O2 cell operating in the presence of a LiTFSI/TEGDME salt/solvent couple and a porous carbonaceous cathode. The cell was discharged/charged applying both voltage and capacity limits, and the effects of repeated galvanostatic cycling were addressed. The ex-situ characterization of carbonaceous cathodes in correspondence of different cut-off voltages was based on vibrational spectroscopies, transmission electron microscopy and X-ray photoelectron spectroscopy. The reversible precipitation/decomposition of undesired products deriving from degradation of both carbon cathode and ethereal solvent is pointed out within a single voltage limited (2.0 – 4.6 V) discharge/charge cycle, whereas their irreversible accumulation on the surface of the electrode 1 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 39

results after 100 capacity limited cycles. At the same time, the presence of polar degradation products (carbonates, carboxylates) at the cathode surface is accompanied by the build-up of a surface electric potential gradient, as revealed by differential binding energy shifts resulting from C 1s photoelectron spectra. This effect, seldom reported for Li-ion batteries, is for the first time put in evidence for a Li-O2 cell. Furthermore, the use of TFSI- anion is shown to lead to carbonate based degradation products not involving the formation of Li2CO3. The peculiar occurrence of such degradation phenomena are attributed to the intrinsic low-donor number characteristic of the TFSI- anion.

Introduction In the last decades, various scientific research efforts have been devoted to modern electronic devices for innovative applications (mobile phones, tablets, laptop computers, smartphones and iPhones). Great interest has also been directed towards the electric transportation: in fact, a largescale diffusion of e-vehicles could limit environmental issues such as polluting gas emissions from internal combustion engines and greenhouse gas accumulation in the atmosphere. To this end, rechargeable energy storage and conversion devices with high power and energy density can play a pivotal role.1,2 Li-ion batteries are probably the most popular energy storage systems but their current performance is not fully adequate for all electric vehicles: their energy density (~ 2000 Wh kg-1) is smaller than that of internal combustion engines (~ 12000 Wh kg-1).3 On the other hand, Li-air cells are at the cutting edge among the studied innovative devices for energy storage thanks to the absence of polluting emissions, the possibility of assembling them with relatively cheap materials and their high theoretical performances. The theoretical specific energy density of non-aqueous Li-O2 batteries is ~3500 Wh kg-1,3–5 including the mass of stored O2. 2 ACS Paragon Plus Environment

Page 3 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Currently, non-aqueous Li-air cells suffer many limitations. First of all, their configuration is not really “air-open” due to the detrimental role of CO2 and moisture on the reversibility of the redox process.3,4,6–9 For this reason, R&D efforts focus on model devices using ultra-pure O2 in closed or flow configurations, the so-called “Li-O2” cells. Li-O2 cell most common structure couples an electropositive carbon electrode with an electronegative Li foil and a non-aqueous electrolyte. During discharges and charges, besides the reversible Li-O2 redox activity to form and decompose lithium peroxide, i.e. Li2O2, also additional remarkable degradation processes occur due to apparently unavoidable parasitic reactivity of intermediates (e.g. LiO2, 1O2) and products (Li2O2) towards the carbon electrode and the electrolyte.10–13 Understanding the electrochemical and chemical processes competing with the Li2O2 reversible formation in Li-O2 batteries is fundamental to identify mitigation strategies to improve cell efficiency upon cycling. According to literature, two possible mechanisms lead to the formation of Li2O2 in Li-O2 cells. The surface mechanism implies a mono-electronic O2 reduction to soluble LiO2, the adsorption of this intermediate on the cathode surface and a second mono-electronic reduction leading to the formation of Li2O2. This latter quickly grows in a compact layer covering completely the cathode surface. Due to the Li2O2 insulating character, this is a self-limiting process. Moreover, on charge the oxygen evolution reaction (OER) from Li2O2 results limited and the cell suffers from reversibility loss. Instead, via the solution-based mechanism, the soluble lithium superoxide intermediate diffuses away from the cathode surface and it slowly disproportionates to insoluble Li2O2. In this way, lithium peroxide grows in large particles and the surface is not passivated.5,14,15 In order to promote the second mechanism and to achieve discharge capacities close to the theoretical values and large reversibility, the key factors are (a) the adoption of small discharge rates, (b) the use of high donor number (DN) lithium salt anions and organic solvents as electrolytes, and (c) the exploitation of cathode materials with low binding energy with respect to 3 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 39

LiO2.16–20 Nevertheless, solvents with intermediate DN (such as ethers) constitute a valid tradeoff, since they have been reported to ensure limited degradation issues (dramatically affecting the performance of high DN solvents), together with acceptable Coulombic efficiency.21 In this work, we illustrate the results of a study on model Li-O2 cells assembled with a porous carbon cathode and lithium bis(trifluoromethane-sulfonyl)imide dissolved in tetraethylene glycol dimethyl ether) as electrolyte (i.e. LiTFSI/TEGDME). LiTFSI is a highly dissociated salt in ethereal solvents, and its anion has a low DN, while TEGDME is a moderately low-DN solvent.18 The goal is to describe the reactivity at the triple interface O2/cathode/electrolyte where the chemical and electrochemical reactions occur. Great attention is given to the TFSIinfluence on the chemical nature of the degradation products. Post-mortem cathodes have been analyzed by means of a multi-techniques approach. X-ray photoelectron spectroscopy (XPS) and vibrational spectroscopies (Raman and FTIR) have been used to characterize the products grown on discharge and charge. Cathodes morphological changes have been investigated by means of transmission electron microscopy (TEM). These exsitu analyses have been performed on electrodes recuperated from cells stopped at various stages of their galvanostatic discharge and charge.

Experimental Li-O2 cells and electrolyte solutions were prepared in a Iteco Engineering argon-filled glovebox with moisture concentration below 0.1 ppm. The batteries were assembled by using a MTI Corp. stainless steel lithium-air test cell consisting of a stainless steel current collector, a metallic lithium foil as anode, a glass-fiber separator (Whatman) impregnated with 250 µL of a non-aqueous electrolyte, and a commercial porous carbon foil (MTI Corp., 14 mm in diameter) as cathode. The cathode foil is constituted of a mixture of Super P carbon and a fluorinated polymeric binder (Kynar, Arkema) dispersed on an inert carbonaceous gas diffusion layer (the 4 ACS Paragon Plus Environment

Page 5 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


ratio between carbon Super P and the binder is 38:62 w/w). The electrolyte is a 1 molal solution of LiTFSI dissolved in TEGDME (Sigma-Aldrich, moisture controlled grade). The salt and the porous cathodes were dried at 110 °C overnight under vacuum before use. The solvent was used after drying/storage on regenerated 3 Å molecular sieves (Sigma-Aldrich) and lithium chips for at least 15 days in a glovebox. The Li-O2 cells were filled with pure O2 (5.0 purity spilled from a high-pressure cylinder through a stainless steel gas lines, preliminarily evacuated, equipped with a molecular sieve-filled moisture trap) setting a static final pressure of 2.2 bar in the dead volume (~ 3 cm3) above the porous electrode for all tests. Galvanostatic cycling of the cells were carried out connecting them to a MTI Corp. battery cycler imposing capacity limits of 0.2 and 0.4 mAh cm-2 and current density values of 0.2 and 0.4 mA cm-2, while the number of cycles was set to 100. In another set of experiments, the cell was set to a single cycle with a current density of 0.2 mA cm-2 and stopped at cut-off voltages of 2.6, 2.5 and 2.0 V in discharge, and 4.0 and 4.6 V in charge. The measured capacity was normalized dividing by the geometrical area of Super P carbon electrode (1.54 cm2). After the electrochemical measurements, the cells were disassembled in an Iteco Engineering argon-filled glovebox with a moisture concentration below 0.1 ppm. Their cathodes were recuperated and washed in TEGDME and in tetrahydrofuran (THF) to remove the excess of electrolyte, and dried under vacuum at room temperature. The benchmark samples were a pristine carbon foil and a carbon foil wet with electrolyte. Recuperated cathodes and benchmark samples were characterized by X-ray photoelectron spectroscopy (XPS), Fourier transform infrared spectroscopy (FTIR), Raman spectroscopy and transmission electron microscopy (TEM). XP spectra were recorded using a modified Omicron NanoTechnology MXPS system equipped with a monochromatic X-ray source (Omicron XM-1000) and an Omicron EA-127 energy analyzer. The exciting radiation was Al Kα (hυ = 1486.7 eV), generated operating the anode at 14-15 kV and 10-20 mA. All of the photoionization regions were acquired using an 5 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 39

analyzer pass energy of 20 eV, except for the survey scan, taken at 50 eV pass energy. Take-off angles of 11° with respect to the sample surface normal were adopted. The measurements were performed at room temperature, and the base pressure in the analyzer chamber was about 2 × 109

mbar. The C 1s binding energy (BE) of the -CF2- group at 292.0 eV belonging to the

fluorinated polymer binder22 of the pristine carbon electrode was used as an internal standard reference for the BE scale (accuracy of ±0.05 eV). The motivation for this choice is explained in the Results and Discussion section. Samples were transferred into the spectrometer through an argon-filled glove-bag connected to the fast-entry lock chamber of the instrument in order to avoid contact with air. The experimental spectra were reconstructed by fitting the secondary electrons’ background to a Shirley function and the elastic peaks to pseudo-Voigt functions described by a common set of parameters (position, FWHM, Gaussian-Lorentzian ratio) free to vary within narrow limits. During the fitting procedures the Gaussian-Lorentzian ratio was left free to vary between 0.6 and 0.9. The asymmetry was fixed to 0.0 for all the peaks except for the graphitic-like carbon peak in the C 1s region, for which it was set to 0.15.23 XPS atomic ratios between relevant element components were estimated from experimentally determined area ratios (with ±10% associated error), which were corrected for the corresponding photoelectron cross sections according to Scofield calculations,24 and for the square root dependence of the photoelectron kinetic energy. Sample degradation due to X-ray exposure was not evident within the duration time of each experimental observation. In particular, the lineshape of the Li 1s XPS signal was not found to evolve during acquisition of the spectrum.25 FTIR spectra were acquired by a Jasco FTIR-460 Plus apparatus. All spectra were recorded in the wavenumber range between 2000 and 400 cm-1 at room temperature in transmission mode. Materials removed from the cathodes as well as from a pristine carbon foil, as fine powder, were mixed in a Ar-filled glovebox with CsI and then pressed in pellets by a Pike die set and hand-

6 ACS Paragon Plus Environment

Page 7 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


press. FTIR spectra were also recorded for pure Li2O2, Li2O and Li2CO3 powders as benchmark materials. Raman spectra were carried out by using a micro-Raman spectrophotometer LabRam HR HORIBA Jobin Yvon equipped with a He-Ne (632.8 nm) laser source (20 mW) and a CCD detector. A sealed cell with a sapphire window has been adopted in order to protect samples for contact with moisture. TEM micrographs were recorded by using a FEI G2 20 HR-TEM instrument equipped with a LaB6 electron beam source and two 2D flat cameras (low resolution and high resolution) at 200 kV e-beam acceleration. Samples were suspended in THF in sealed vials by ultrasonic treatment (5 cycles of 15 minutes of ultrasonic treatment followed by 45 minutes of rest to cool down the sample and thus avoid thermal heating), and dispersed on copper holey carbon film grids for observation.

Results and discussion Electrochemical tests The electrochemical performances of our cell configuration is shown in Figure 1 in terms of specific capacity and charge reversibility values versus cycle number.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 39

Figure 1. Cell performances upon cycling: (a,c) specific capacity and (b,d) charge reversibility (see text for more details).

Generally speaking, our cell performances are comparable with those reported in the available literature for similar configurations,26 in terms of specific capacity and cycling life. At a current rate as large as 0.2 mA cm-2 the target capacity limitation (i.e. 0.2 mAh cm-2, see Figure 1a) is successfully reached for more than 70 cycles, both in discharge and charge. On the contrary, at higher rates, i.e. 0.4 mA cm-2, the cell fails to supply the programmed maximum capacity. This trend is due to the occurrence of larger charge and discharge overvoltages at increasing current rates (see Figure S1 in the Supporting Information).27 A similar trend is observed if the discharge/charge capacity limitation is doubled. In fact, at the same current rate, i.e. 0.2 mA cm2

, identical cells reach the capacity limitation of 0.2 and 0.4 mAh cm-2 (Figure 1c), both in

discharge and charge for 70 and 25 cycles, respectively. Beyond these cycle number the discharge capacity always exceeds the charge for both capacity limitations, and a remarkable drop in capacity is observed for the 0.4 mAh cm-2 limited cell. 8 ACS Paragon Plus Environment

Page 9 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Turning to the trend of the charge reversibility (i.e. the ratio between the charge and discharge capacities in the same cycle), at 0.2 mA cm-2 it drops below 1 after cycle 71 whereas at 0.4 mA cm-2 it is always smaller than unity. In the former case, the high charge reversibility recorded up to the 71st cycle is explained considering that the imposed capacity limit (0.2 mAh cm-2) amounts to less than the 10% of the total discharge capacity of the cell when neither capacity nor voltage limits are imposed (see Figure 2). For what concerns the cell cycled at 0.2 mA cm-2 but with a doubled capacity limitation of 0.4 mAh cm-2, the charge reversibility falls below 1 after cycle 25. In our cell configuration a charge reversibility smaller than 1 originates from (a) the excess of lithium metal in the negative electrode and (b) the increase of charge overvoltages upon cycling (see Figure S2 in the Supporting Information). However, the increase of the voltage hysteresis occurs monotonically at 0.2 mA cm-2 cycle-by-cycle even for charge reversibility approaching 1. This trend can be related with the accumulation over the positive electrode surface of insoluble and not fully redox-reversible reaction products, that can be either Li2O2, Li2O or degradation byproducts, thus leading to an increase of the overall electrode resistance and to larger overvoltages.3,11,28 In order to decouple the Li-O2 redox chemistry occurring on the surface of the cathode (e.g. formation and decomposition of Li2O2) from the parasitic reactivity of the carbon support or the electrolyte, we performed a post-mortem characterization of electrodes at various stages of discharge/charge in the first cycle and after prolonged cycling (see the Table 1 below). The electrochemical discharge and discharge/charge tests were carried out under galvanostatic control at a constant current rate of 0.2 mA cm-2 with fixed voltage cut-offs. An example of the measured cell voltage profile is shown in Figure 2 for a cell fully cycled between 2.0 and 4.6 V. The corresponding performances in terms of specific capacity and charge reversibility are summarized in the Table 1.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 39

Figure 2. Galvanostatic cycle between 2.0 and 4.6 V. Cathodic and anodic cut-off voltages for partially discharged and charged cells are marked with black dots. Applied current density was 0.2 mA cm-2.

Upon discharge (Figure 2, blue curve), the voltage profile shows the expected stable plateau at about 2.6 V.3 This voltage plateau is below the thermodynamically estimated potential for the expected oxygen reduction reactions (ORR)4 involving the formation of lithium compounds such as lithium peroxide and oxide, thus suggesting the occurrence of moderate overvoltages.16–20 As expected, upon charge the voltage profile shows a short stable plateau at about 3.3 V and a long stable plateau at about 4.4 V. These charge plateau are above the thermodynamically estimated potential for the expected oxygen evolution reactions (OER).3 Compared to discharge the measured overvoltage in charge is, as expected, much larger. This is a common feature of LiO2 cells, mostly due to the presence of large amounts of lithium oxides and degradation products grown within the cathode pores. Due to their total/partial insulating character, these species partially limit the charge transfer and the OER, thus enhancing the charge overvoltage.10,29 In our system, the choice of both TFSI- anion and TEGDME solvent is expected to promote a surface-based growth mechanism of Li2O2 which leads to the formation of a compact layer of 10 ACS Paragon Plus Environment

Page 11 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


material in a self-limiting process.16–20 The growth of such compact layer vs. the formation of toroidal shaped crystal aggregates of Li2O2 is associated to the low values of DN for both components of the electrolyte under study, which limits the lifetime of freshly generated O2anions at the electrode surface, promoting the 2-electron reduction of oxygen and the fast growth of Li2O2 compact layers.16–20 As expected, upon discharge the specific capacity increases for deeper reductions limited by smaller cathodic voltage cut-offs. Similarly, on charge, the cells supply an increasing total capacity with the increase of the anodic voltage cut-off, reaching an overall charge reversibility of 1 for the cell fully cycled between 2.0 e 4.6 V.

Table 1. Summary of the studied samples and corresponding performances in galvanostatic tests. Sample

Galvanostatic test conditions j=0.2 mA cm-2

Discharge Capacity (mAh cm-2)

Charge Capacity (mAh cm-2)

Charge reversibilitya

Disch. 2.6V

Discharge to 2.6 V

0.99 ± 0.15

-

-

Disch. 2.5V

Discharge to 2.5 V

2.21 ± 0.33

-

-

Disch. 2.0V

Discharge to 2.0 V

2.99 ± 0.45

-

-

Ch. 4.0V

Discharge to 2.0 V and charge to 4.0 V

2.99 ± 0.45

1.37 ± 0.21

0.46 ± 0.07

Ch. 4.6V

Discharge to 2.0 V and charge to 4.6 V

2.99 ± 0.45

2.97 ± 0.44

0.99 ± 0.07

1 cycle

1 cycle at 0.4 mAh cm-2 limited capacity; end state in charge

10 cycles

10 cycles at 0.4 mAh cm-2 limited capacity; end state in charge

100 cycles

100 cycles at 0.4 mAh cm-2 limited capacity; end state in charge

a

Charge reversibility has been calculated as the ratio between the charge and the corresponding discharge specific capacity

TEM analysis The evolution of the active material onto the positive electrode upon the first discharge-charge galvanostatic cycle in the Li-O2 cell is shown in Figure 3. 11 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 39

Figure 3. TEM micrographs of the electrode active material: (A) pristine carbon cathode; (B) Sample discharged to 2.6 V; (C) Sample discharged to 2.0 V; (D) Sample discharged to 2.0 V and charged to 4.6 V. In the inset of panel (B) the calculated fast-Fourier transform (FFT) of the selected area is shown. The carbon cathode is constituted by the expected round shaped nanoparticles (20-30 nm in diameter). Upon discharge, at 2.6 V, a discontinuous polymer-like matrix apparently surrounds the carbon particles as well as additional irregular and smaller nanoparticles (5-10 nm in


Page 13 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


diameter). These elusive morphologies may possibly be constituted by Li2O2, as reported by the tentative indexing of the partial FFT shown in the inset of panel (B) in Figure 3. In fact, the interplanar distance found at 8 Å may be due to the reflection of the (001) plane of the Li2O2 lattice, to be compared to d(002)= 3.55 Å30 for the layer stacking in graphitic carbons and d(001)= 7.2-7.4 Å31 for the graphene oxide (possible oxidative degradation product of graphite, as observed by us in ref. 29). At the end of the discharge at 2.0 V the morphology of the material is constituted by an amorphous matrix. Within this matrix, shadows of round shaped particles can be seen, having dimensions close to the original carbon particles. At the end of charge at 4.6 V the original morphology is nicely recuperated and only the round shaped carbon particles can be observed throughout the entire sample.

FTIR Spectroscopy In Figure 4 the ex-situ FTIR spectra of the discharged and charged cathodes are shown. The intensity of the spectra recorded on post-mortem samples has been normalized to the intensity recorded at 1100 cm-1, corresponding to the most intense absorption peak of TEGDME.32 The reference spectra of a pristine carbon foil and pure Li2O, Li2O2 and Li2CO3 powders are also shown to facilitate the analysis. Briefly, both the reference lithium oxide and lithium peroxide show a broad band in the 400 – 600 cm-1 region (Figure 4a and b), related to the Li-O bond stretching.11,33,34–38 Li2CO3 also gives a minor spectral contribution in this region (Figure 4c), but its FTIR fingerprint bands are centered at about 860, 1440 and 1507 cm-1. The first band is for the O=C-O bending mode, while the others are related to the stretching vibrations of the same group.35–37,39–42 In spectrum 4d (Pristine), an intense and broad band with maximum at 1020 cm-1 can be observed, possibly associated to the stretching mode of superficial oxidized groups of the 13 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 39

graphitic-like carbon support, such as C-O-C and C=O.43,44 At lower wavenumbers there are minor contributions attributable to the fluorinated binder vibrations.45,46

Figure 4. FTIR spectra of Li2O, Li2O2 and Li2CO3 pure powders (spectra a - c), and of materials from pristine (spectrum d), discharged (spectra e - g) and cycled (spectra h - i) carbonaceous electrodes of the LiTFSI/ TEGDME Li-O2 cells addressed in this work. In the 400 – 600 cm-1 region, where the stretching of Li-O bond is expected, a broad band increases upon discharge, remains almost equal on charge up to 4.0 V and disappears charging the cell to 4.6 V (Figure 4 e-i). Although our FTIR spectra are apparently unable to clearly discriminate the formation of Li2O2 or Li2O, the evolution trend of the band below 600 cm-1 may


Page 15 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


confirm the overall accumulation of oxidized lithium compounds upon discharge and their removal at high overvoltage. Upon discharge (Figure 4 e-g), even at the high voltage of 2.6 V, also other signals increase in the spectral ranges attributable to the C-O and C=O stretching mode. These signals are likely due to the accumulation of carbonates and alkylcarbonates (C-O and C=O symmetric stretching at 1020 – 1300 cm-1, C-O and C=O asymmetric stretching at 1400 – 1660 cm-1)11,33,35–38,40,41 and carboxylates (C=O symmetric stretching at 1500 – 1660 cm-1).11,33 However, there is a lack of bands in the 1700 – 1800 cm-1 region, where the asymmetrical stretching of C=O carboxylates groups is expected,38,42,47 suggesting the presence of these species in the outermost layers of our samples. On passing, it is important to recall that, unlike XPS, the transmission FTIR technique allows to detect compounds, such as inorganic Li2CO3, grown in inner regions of the samples. Thus, it is worth noting the apparent lack of any fingerprint for this last compound both in discharge and charge. This evidence suggests a remarkable altered degradation reactivity in comparison with our previous study on similar cathodes with lithium triflate (LiTFO)/TEGDME electrolyte.28 On charge (Figure 4 h-i), the FTIR spectra show the permanence of carbonates and alkylcarbonates on the cathode surface of the sample charged to 4.0 V, whereas the same spectral features are completely absent when the cut-off voltage is pushed up to 4.6 V. These results are compatible with the recovery of the pristine morphology shown by TEM images (Figure 3d) likely associated to oxidative removal of carbonaceous deposits at high voltages by the expected CO2 release.3 As a final point, we would like to underline that the precipitation of LiOH can be excluded according to the lack of any sharp peak at 3678 or 3574 cm−1 in all samples, as already discussed by us in ref. 29.

X-ray photoelectron spectroscopy – General 15 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 39

As already mentioned, XP spectra were acquired from cathode samples extracted from the cell stopped at different cut-off voltages of the first discharge-charge cycle of the cell, as summarized in Table 1. In order to facilitate the interpretation of these spectra, a bare carbon electrode (hereafter called “Pristine”) and a carbon electrode impregnated with the electrolyte (hereafter called “Uncycled”) were analysed as benchmark materials. The binding energy shifts (∆BE) of the signals corresponding to the species identified and analyzed in the C XP core region with respect to one reference peak are summarized in Table 2.

Table 2. Binding Energy shifts (∆BE) relative to graphitic carbon as determined after curve fitting of the experimental C 1s XP spectra of the materials analyzed in this work.a Peak

∆BE (eV)

Assignments

C 1s

0

Graphitic-like carbon

1.0 - 1.3

Graphite defects

2.1 - 2.4

Epoxy, C=O superficial groups (oxidized graphite)

3.0

C-O (TEGDME)

5.0

Graphite shake-up

4.8 - 5.1

-COO-, -COOH

5.8 - 6.1

-O(C=O)O- carbonate

7.7

-CF2- in binder

9.1

-CF3 in LiTFSI anion

a

The inner BE shift due to surface electric potential build-up, as discussed in the

text, is not considered in this table.

X-ray photoelectron spectroscopy – C 1s Figure 5 shows the C 1s photoionization regions of the cathode samples upon the first galvanostatic cycle. Figures 5a and b show the pristine and uncycled samples chosen as control samples to facilitate the interpretation of spectra c-g, associated to cathodes discharged to 2.6, 2.5 and 2.0 V and then charged to 4.0 and 4.6 V (after discharging to 2.0 V). 16 ACS Paragon Plus Environment

Page 17 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 5. C 1s XP spectra of pristine and uncycled (black lines, spectra a and b), discharged (blue lines, spectra c - e) and cycled (red lines, spectra f - g) carbon cathodes of the LiTFSI/TEGDME Li-O2 cells addressed in this work. Fitting results are reported as continuous lines (−).

The C 1s spectrum of the pristine carbon support (Figure 5a) displays the typical contributions from a lowly ordered graphitic carbon material, whose main feature is the graphitic sp2 hybridized C signal at 284.3 eV, slightly asymmetric at its high BE side.23,48 The subsequent peak is attributable to defects that, for the pristine sample, are mainly sp3-hybridized C sites.49 Its BE shift (∆BE, see Table 2) with respect to graphitic carbon was found to vary within the range 17 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 39

+1.0 ÷ +1.3 eV. At 286.7 eV a faint contribution from oxidized graphite (i.e. superficial epoxy and C=O groups)49,50 can be found. A further weak component, typical of extended π-delocalized carbonaceous systems, is the broad shake-up transition at 289.3 eV, which roughly accounts for 10% of the main feature intensity.49 At 292.0 eV, a component due to the fluorinated binder of the cathode can be found, associated with -CF2- groups.22 This signal was chosen as an internal reference for the BE scale of the other samples. In fact, as evident in Figure 5, a BE position shift is experienced by some components of the spectra (see green and black lines), as in the presence of a differential surface charging, apparently dependent on the working conditions of the cell. The -CF2- component was adopted as a reference for BE scale because it is associated to one of the constituents of the carbon cathode material. This aspect will be commented further in the text. In Figure 5b (Uncycled), the components associated to the carbon cathode are hardly detectable, since the cathode surface is probably covered by a layer of electrolyte. Indeed, the additional features at 287.3 eV and 293.4 eV are associated with ethereal C-O bonds in physisorbed TEGDME33,50 and with –CF3 groups in LiTFSI anion,51,52 respectively. After thoroughly rinsing the electrodes with pure solvents after cycling, the TEGDME and Li salt signals are significantly reduced compared to those from carbon Super P, as shown in the spectra of discharged and cycled cathodes (Figure 5c-g). Figure 5c shows the effects of discharging the cell to 2.6 V. Signals similar to those in the reference spectrum a can be identified, although with different relative intensities. The most striking difference with the pristine cathode is the inversion of the relative intensities of the graphite-like carbon signal (see the black line in Figure 5) and the defect-related peak (see the magenta line in Figure 5). The increase of the signal due to the defects in the sp2 framework of the carbon material of the cathode has already been evidenced previously by us,28 but in this case a stronger enhancement is observed. One may speculate that due to the low DN of the solvent and the TFSI- anion,3,19,53 a thick layer of various reactive compounds, such as Li2O2, is 18 ACS Paragon Plus Environment

Page 19 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


deposited in discharge onto the surface of the carbon foil. This thick precipitate may lead to the attenuation of outcoming photoelectrons and the corresponding 2D peroxide precipitation may contribute to the structural damage of the graphite C=C network. It is worth noting that in the presence of a different Li salt, i.e. LiSO3CF3 (LiTFO), we reported that the enhancement of such component increases remarkably only when the discharge potential drops to 2.0 V. Therefore, the use of LiTFSI has apparently a more dramatic impact on the degradation of the carbon cathode. Following the evolution of the two components so far addressed (see black and magenta lines in Figure 5), one can see that the predominance of the defects-related feature over the graphitic one is constant throughout the discharged samples (spectra c – e). On the other hand, such feature is still present even in the sample charged to 4.0 V (spectrum f), whereas it is completely recovered to its pristine intensity in the sample charged to 4.6 V (spectrum g). We hypothesize that this behaviour, similar that found in the presence of LiTFO,28 can be explained based on partial decomposition, upon deep charging (vide infra), of (i) lithium reactive compounds, and of (ii) oxidized carbon by-products (such as carbonates), both contributing to the elimination of the defects-related component by CO2 release.3 On passing, it may be of interest to underline that the Raman spectra of samples before cycling, at the end of discharge and after 1 full discharge/charge show very similar ratios between the D and G peaks (see Figure S3 in the Supporting Information). This may suggest that the formation/removal of defective carbons occurs on carbon catalyst particles only within 1 nm depth, which is the thickness investigated with XPS in the present experimental conditions. Signals associated to carboxylate and carbonate groups33,50–52,54,55 are present at 288.5 and 289.5 eV, respectively, in spectra c-f. Their presence calls for the occurrence of degradation phenomena related to the carbon based components of the cell, such as the carbon support of the cathode and the TEGDME solvent. At variance with what we found for a LiTFO based cell, where carboxylates were found to behave as an intermediate step to the formation of carbonates,28 the variable relative intensity ratio of the two contributions throughout the spectra 19 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 39

from discharged samples probably reveals that the two oxidized carbon moieties, in this case, are not part of the same reaction path, but have different origins. Some authors56 already reported the formation of carbonates and carboxylates as possibly due to exposure of TEGDME to Li2O2 particles formed during the discharge phase of the cell operation. As to the presence of Li2CO3, it is not possible to discern whether the peak at 289.4 eV can be assigned to lithium carbonate or to variably substituted organic carbonates. According to literature, Li2CO3 mainly nucleates on the cathode surface and in its pores due to the oxidant attacks from Li2O2 particles to the underlying carbon support.3,57 As reported in the next subsection, in our case, the Li2CO3 fingerprint is also missing in Li 1s spectra, and its absence is in agreement with the corresponding FTIR spectra (see the FTIR spectroscopy section above). Charging the cell to 4.0 V (Figure 5f) apparently introduces just few modifications to the surface of discharged samples. In fact, oxidative degradation products (carboxylates, carbonates) are still present to a significant extent, while defectiveness of the graphitic network is only partly recovered (see the magenta line in Figure 5). As evidenced also in our previous work using LiTFO as the electrolyte salt,28 in the first discharge/charge cycle of the cell, extreme charging up to 4.6 V restores the main spectral feature of the carbon cathode. In fact, defectiveness is recovered parallel to the abatement of oxidized carbon features (spectrum g). As mentioned above, one can notice that, with reference to the –CF2– position (red line), a negative BE shift of 0.8 eV is experienced by the graphite-like (black line), defects (magenta line) and TEGDME (green line) peaks of spectra from the discharged and 4.0 V-charged samples. On the other hand, according to the proposed energy calibration, the position of carbonates peak (blue line) is coherent with the literature. In the literature, similar BE shifts have been reported in one case for signals associated to the solid electrolyte interphase (SEI) and bulk materials of Li-ion cells anodes.58 The nature of such shift is not apparently chemical, and, since it is found to be dependent on the presence of carboxylates/carbonates species (in our case parallel to the giant enhancement of defect-related signal), it may be related to changes to the 20 ACS Paragon Plus Environment

Page 21 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


electrode’s surface induced by its state of charge. Edström et al.58 explained the occurrence of a similar shift in a Li-ion battery as due to the build-up of an electric potential gradient at the interface between the SEI and the anode during lithiation. Such gradient was attributed not much to the insertion of lithium, but to the deposition of polar components, such as adsorbed solvent molecules and/or degradation products. In our case, upon referencing to the binder –CF2– groups, one can see that the down-shifted components are those related to the graphitic and defective carbon and the TEGDME solvent, which are likely localized in a portion of the electrode where a fractional negative charge is concentrated.58 On the other hand, the degradation products around 289 eV seem to maintain their chemical shift from the –CF2– group. In Figure S4 we report a similar set of spectra as that of Figure 5, but recorded from a carbon electrode used in a LiTFO/TEGDME system.28 In these spectra, the same BE referencing with respect to the binder –CF2– group was operated, and it is apparent how, although smaller (-0.44 eV), a similar BE shift trend can be evidenced, especially for the C=C and TEGDME components. Similar to the case of LiTFSI, also in the presence of LiTFO this shift disappears upon charge to 4.6 eV, in parallel to the complete removal of carbonates. In this context, we propose that a mechanism similar to that described for Li-ion cells58 is in operation in Li-O2 cells. Therefore, we attribute the negative BE shift for the C=C and TEGDME components of the carbon electrode used in our LiTFSI/TEGDME system to an electric potential gradient induced by accumulation of oxidized carbon by-products formed during discharge, such as carbonates, which directly involve degradation of both Super P carbon and TEGDME solvent. On the other hand, we exclude the possible role of oxidised lithium compounds in contributing to the BE shift,58 since the Li/C area ratio (see the FTIR section above and the Table S1) was not found to be directly proportional to the extent of the BE shift itself. The possible layering of polar materials or ionic couples onto the carbonaceous catalysts might be further investigated by electrochemical impedance spectroscopy measurements.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 39

X-ray photoelectron spectroscopy – Li 1s The Li 1s region in the XP spectra is shown in Figure 6 for all the cathodes. Spectrum a shows a single contribution around 56.2 eV, attributable to the Li+ ion in the LiTFSI salt.59 Such signal is absent in the other spectra of the discharged and cycled cathodes, likely due to the rinsing procedure applied after extraction of the cathodes from the cell.

Figure 6. Li 1s XP spectra of uncycled (black line, spectrum a), discharged (blue lines, spectra b - d) and cycled (red and black lines, spectra e - f) carbon cathodes of the LiTFSI/TEGDME Li-O2 cells addressed in this work.

In spectra b - d, a broad signal can be detected in the range 54 ÷ 55.5 eV, which we attribute to the ORR products accumulated on the carbon cathodes upon discharge to 2.6, 2.5 and 2.0 V, 22 ACS Paragon Plus Environment

Page 23 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


in particular Li2O2. Due to the inherent low photoionization cross-section and large peak width, the possible attribution of a portion of the Li 1s signal to Li2O could not be ascertained, while it seems more reasonable the presence of Li2O2.33,57,60–62 The presence of a well-pronounced Li 1s signal at the surface of a 2.6 V-discharged cathode suggests, in line with the literature, that O2 reduction to can take place already at 2.6 V.11 The presence of Li2O2 is detected up to the 4.0 V-charged sample (spectrum e). On the other hand, when the charge is pushed up to 4.6 V (spectrum f) a dramatic abatement of Li 1s signal is detected, suggesting an almost complete decomposition of lithiated compounds followed by O2 evolution. Apparently, in all the spectra, no other signals can be observed, nor related to residual physisorbed LiTFSI salt molecules, neither from other species formed as reactions by-products, such as lithium carbonate or lithium fluoride.3 As already discussed above, the possible role of the LixOy compounds on the establishment of the surface electric potential gradient responsible for the BE shift of some peaks in the C 1s region can be ruled out. In fact, as reported in Table S1, the Li/C ratio, as determined form the corrected areas of XPS signals, tends to vary with no apparent correlation with the constant BE shift observed.

X-ray photoelectron spectroscopy – Valence Region In Figure 7 the photoionization regions within the BE range 0 – 21.0 eV are shown for the discharged and charged cathodes (spectra c – g), as well as for pristine and uncycled carbon foil (spectra a – b).



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 39

Figure 7. XP spectra of valence ionization region of pristine (spectrum a), uncycled (spectrum b), discharged (spectra c – e), and cycled (spectra f – g) carbon cathodes of the LiTFSI/TEGDME Li-O2 cells addressed in this work. Blue, red and green dashed lines indicate the bands arising from ionization of CO32- MOs, as discussed in the text. It is important to recall that, when X-ray exciting sources are used, the valence region suffers from low signal intensity, due to inherently low photoionization cross section. On the other hand, the kinetic energy of the emitted photoelectrons is high, close to the photon energy used, which results in a rather significant thickness of the sampled layer, due to a correspondingly long photoelectron inelastic mean free path. Although theoretical modelling would be necessary for a


Page 25 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


detailed interpretation of the valence photoionization features, fingerprint signals can usually be detected, which help and support the interpretation of core level regions. The spectrum of the pristine sample (a) is dominated by the density-of-states (DOS) at the surface of the carbonaceous cathode, mostly due to lowly ordered graphitic material, which gradually rises only after ~5 eV, whereas is close to zero for BE < 5 eV.63,64 In spectrum b (Uncycled), a sequence of at least five bands showing up roughly at 6.6, 10.4, 15.0, 17.0 and 18.8 eV can be found, likely attributable to ionization of molecular orbitals of the LiTFSI/TEGDME couple. Although not coupled to the TEGDME solvent, the TFSI- anion has been the object of few studies, both theoretical (DFT) and spectroscopical (UPS/XPS),65–69 which substantially support our attribution of the valence features in the Uncycled sample. In the spectra for the discharged and charged cathodes (Figure 7c-g), the bands due to the LiTFSI/TEGDME couple can be considered absent because of the washing of the cathodes. Three bands at 5.0 (green line), 10.0 (red line) and 12.0 (blue line) eV are visible in all the spectra, with a maximum of intensity in spectrum d (cathode discharged to 2.5 V). These bands constitute a fingerprint of carbonate species on the cathodes surface. Assuming the D3h geometry for carbonate anion, the band indicated with the green line accounts for ionization of the 1a2’, le”, and 4e’ nonbonding MOs with contributions almost exclusively from O 2p AOs.70–74 On the other hand, the band crossed by the red line represent ionization of 3e’ and 1a2” MOs, while the blue line indicates a 4a1’ bonding MO. The composition of such MOs is an admixture of O 2p and 2s and C 2p and 2s AOs.70,73,75 As already commented in the previous paragraphs, we attribute the presence of carbonate species to oxidation of the carbon cathode surface. Indeed, the occurrence of photoionization features due to carbonate species in the valence region parallels the appearance of corresponding peaks at 290 eV in the C 1s spectra of the same samples (see Figure 5).

X-ray photoelectron spectroscopy – Cathodes after multiple cycling at limited capacity 25 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 39

Figure 8 reports the C 1s spectra acquired from the carbon electrode samples cycled within the voltage range 2.0 V (discharge) and 4.6 V (charge), but limited at the capacity value of 0.4 mAh cm-2 (see also Figure 1). A comparison is reported among the pristine sample (spectrum a) and samples subject to 1, 10 and 100 discharge/charge cycles (spectra b, c, d, respectively). Both samples cycled once (b) and 10 times (c) are substantially “clean”, except for an unexpectedly intense contribution from the epoxyl- group (violet peak), which remains unexplained, at present. Other signals are those from the solvent (green peak), Super P/binder (shaded and red peaks) and ‒CF3 from the salt (~293 eV). As to the spectrum of 100-times cycled sample (d), new features arise, e.g. the defects-related peak (284.8 eV), the carboxylate (288.5 eV) and carbonate (289.5 eV) signals. Furthermore, the TEGDME contribution appears much more intense. Similar to the previous two sets of spectra (Figures 5 and 6), the BE scale was referenced to the binder –CF2– group value at 292.0 eV. It turns out, then, that although the final voltage was 4.6 V, after 100 cycles partial degradation of the electrode material with accumulation of oxidized carbon byproducts occurred. Furthermore, the presence of carboxylates/carbonates is accompanied by a 0.65 eV BE shift for the C=C, epoxyl- and defects-related components, which supports the above mentioned role of such degradation species in this phenomenon. As to the possible accumulation of oxidized lithium compounds, after 100 discharge/charge cycles the presence of Li2O2 could be inferred (Figure S8). Overall, XP spectra of carbon electrodes used in capacity-limited galvanostatic cycles show that 100 of such cycles, at least, lead to an accumulation of inert insoluble and not fully redoxreversible reaction products at the surface, which contribute to the build-up of considerable overvoltages, as discussed previously (see the electrochemical results section above). Further evidences of this materials accumulation is shown in the Supporting Information (see Figures S3 and S11) by Raman spectroscopy and TEM. In particular, in the TEM images the presence of a diffuse smooth amorphous-like layers (thickness

Cell - ACS Publications - American Chemical Society

Recommend Documents