Characterization of LTO//NMC Batteries Containing Ionic Liquid or Carbonate Electrolytes after Cycling and Overcharge

The batteries were galvanostatically cycled for 100 cycles. The program consisted of 5 cycles at C/20-rate (1 for pouch cells) followed by cycles at C/10-rate (charged and discharged in 10 h). The cycling performances of LTO // NMC coin cells were investigated with the three electrolytes [EC:DEC][LiPF₆], [C₁C₄Im][Li][NTf₂] and [PYR₁₄][Li][NTf₂] in coin cells (∼2 mAh) (Figure 2 and 3) and pouch cells (∼10 mAh) (Figure 4). Pouch cell volumes were measured after 30 and 50 cycles, which could explain some discontinuities in the capacity values.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

At 25°C the three systems showed very high stability and didn't show any decomposition, Figure 2. [EC:DEC][LiPF₆] electrolyte demonstrated the highest initial discharge capacity (188 mAh.g⁻¹) as compared to [C₁C₄Im][Li][NTf₂] (165 mAh.g⁻¹), and [PYR₁₄][Li] [NTf₂] (83 mAh.g^-1).

At 60°C, for each electrolyte the performances in both configurations of batteries (coin and pouch cells) were similar. [EC:DEC][LiPF₆] electrolyte demonstrated almost the same initial discharge capacity than [C₁C₄Im][Li][NTf₂] (200 vs 191 mAh.g⁻¹), and higher than for [PYR₁₄][Li][NTf₂] (167 mAh.g^–1). For all the electrolytes the capacity continuously faded. During the first 30 cycles the capacity loss was 0.9 mAh.g⁻¹ per cycle in the case of [EC:DEC][LiPF₆], 1.4 mAh.g⁻¹ per cycle for [C₁C₄Im][Li][NTf₂] and 0.1 mAh.g⁻¹ per cycle for [PYR₁₄][Li][NTf₂]. Unexpectedly, the capacity of cells containing [C₁C₄Im][Li][NTf₂] showed a sharp capacity drop, inducing a decrease of 2.4 mAh.g^–1 per cycle after ∼ 50 cycles at C/10.

The volumes of these gaseous emissions were measured by weighing the pouch cells in air and in ethanol and using Archimedes' principle, after assembling (initial volume), after the first cycle, and after 100 cycles, Table I. For all the electrolytes, no significant volume evolution was observed after the first cycle. After 100 cycles, the cell volumes containing [C₁C₄Im][Li][NTf₂] and [EC:DEC][LiPF₆] simultaneously changed. A volume increase of approximately 1 mL was measured for the batteries containing [C₁C₄Im][Li][NTf₂] and [EC:DEC][LiPF₆] electrolytes, and of 0.1 mL for the battery containing [PYR₁₄][Li][NTf₂] electrolyte.

Emitted gases, analyzed by GC-IR, Figure 5, were mainly carbon dioxide CO₂, diethylether O(C₂H₅)₂ and difluoroethane for the [EC:DEC][LiPF₆] electrolyte, in line with the literature.^5,6 From [C₁C₄Im][Li][NTf₂] electrolyte, carbon monoxide, alkanes (methane, propane and butane) and fluorinated products derived from NTf₂ anion decomposition including CF₃H were identified. Note that during the thermal decomposition of the [C₁C₄Im][Li][NTf₂] electrolyte, the evolved gas phase was mainly constituted of butene isomers, resulting from the cleavage of the butyl-N bond of the imidazolium ring.⁴

Zoom In Zoom Out Reset image size

The LTO // NMC pouch cells with [EC:DEC][LiPF₆], [C₁C₄Im][Li][NTf₂] and [PYR₁₄][Li][NTf₂] electrolytes were overcharged up to 4.5 V (6 V vs Li⁺/Li) at C/10-rate, and a float voltage was set for 20 h at 60°C. The current used to maintain the cell voltage at 4.5 V during this step was 0.4 mA in the case of [C₁C₄Im][Li][NTf₂], three times higher than for the two other electrolytes, Figure 6 and Table II. In parallel, after overcharge the cell volumes increased by 2.81 mL for the [C₁C₄Im][Li][NTf₂] electrolyte, by 0.30 mL for [EC:DEC][LiPF₆], and by 0.12 mL for [PYR₁₄][Li][NTf₂], Table II. Similarly to cycling experiments, the gases generated during overcharge were identified by GC-IR as carbon monoxide CO, dihydrogen H₂, fluorinated species, and alkanes such as methane, propane and butane, Figure 7.⁴ At 60°C, in both cycling and overcharge experiments, the [C₁C₄Im][Li][NTf₂] electrolyte exhibited the lowest stability. Its observed instability in this LTO // NMC system was unexpected since imidazolium-based electrolytes exhibited a higher cycling stability with others electrode systems such as Li₄Ti₅O₁₂ // LiFePO₄ or Li // LiCoO₂.^7,8 This observation could be explained by the nature of the positive electrode which can be affected by metal dissolution at high voltages.⁹ This can induce a modification of the extreme surface of the positive electrode,^10,11 modifying its potential, and generating new surface reactions with the electrolyte. To highlight this electrode impact, the XPS analyses of the electrode surfaces before and after overcharge were realized.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

As the battery overcharge induces overheating of its different elements, we firstly studied the thermal stability of each element of the battery. IL-based electrolytes were found stable up to 350°C, and imidazolium-based IL were found more stable than pyrrolidinium-based ones.⁴ The LTO and NMC electrodes heated up to 350°C for 2 h remained unchanged as proved by Scanning Electron Microscopy (SEM) and X-Ray Diffraction (XRD) analyses.

After overcharge at 60°C, post mortem analyses of the electrode surfaces with [C₁C₄Im][Li][NTf₂] as electrolyte were carried out by SEM, XRD and XPS. XRD diffractograms revealed no change for the LTO structure, a spinel of cell parameter 8.352 Å. Contrarily, the cell parameters a and c of NMC structure (rhombohedric α-NaFeO₂, Figure 8)^9,12–16 varied from 2.860 and 14.225 Å to 2.831 and 14.322 Å, respectively. The modification of these parameters was related to the delithiation of NMC during overcharge. Note that SEM analyses of LTO and NMC electrodes did not show cracks or any other surface damage.

Zoom In Zoom Out Reset image size

The X-ray Photoelectron Spectroscopy (XPS) survey spectra of pristine LTO and NMC electrodes showed the presence of different elements on the surface, Figure 9, as reported in the literature.¹⁷ A comparative analysis of the high resolution XPS spectra for each atomic component C 1s, N 1s, O 1s, F 1s, and S 2p obtained for both post mortem electrodes and pure [C₁C₄ImNTf₂]^18,19 is reported in Figure 10 and Table III.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

On washed post mortem NMC electrode, bulk oxygen and manganese were still visible, meaning that the IL film on the electrode was thinner than 10 nm (XPS depth of analysis).²⁰ The surface layer was mainly constituted of physisorbed IL as confirmed by the presence of C 1s, N 1s, O 1s, F 1s, and S 2p peaks, assigned to [C₁C₄ImNTf₂].^19–21

In contrast, in the XPS spectrum of washed post mortem LTO electrode most of the initial elements were hidden, revealing that the surface was covered by a residual electrolyte film thicker than 10 nm. Note that the comparison of N 1s peak intensities (between 400.9 eV peak for N_cation and 399.0 eV peak for N_anion) in the spectra of the pure IL and of the LTO surface showed a strong decrease of the cation contribution. Besides the S 2p_3/2 and F 1s peaks corresponding to the anion, new peaks were observed, Figure 9. The new F 1s peaks at 684.3 and 686.1 eV could be assigned to LiF and Li₂NSO₂CF₃, the last one associated with the new N 1s peak at 403.4 eV. The additional S 2p_3/2 peaks at 161.4 and 167.3 eV could be related to the presence of Li₂S and Li₂S₂O₄ species. The new O 1s peaks at 533.2 and 534.0 eV may be due to the presence of lithium carbonates, alkyl lithium carbonates or the remainder of dimethylcarbonate used to wash the electrodes. Most of the new peaks observed in the XPS spectra of LTO surface result from the decomposition and reduction of the NTf₂ anion. Both the formation of a larger gas volume in the case of [C₁C₄Im][Li][NTf₂] compared to [PYR₁₄][Li][NTf₂] electrolyte and the easy decomposition of NTf₂ anion proved by XPS results could be corelated to the easy deprotonation of imidazolium cycle, Figure 11.²² This could lead to the formation of free HNTf₂ and carbene derivatives that could be washed out from the surface.

Zoom In Zoom Out Reset image size

IL and their electrolytes were synthesized and purified as reported in the literature.⁴ These electrolytes were prepared under inert conditions by adding 1 mol.L^-1 of LiNTf₂. [EC:DEC][LiPF₆] (BASF) was stored in a glove box and used as received. The water content was lower than 60 ppm for all electrolytes.

Li₄Ti₅O₁₂ composite electrodes (LTO) were prepared by coating an aqueous-based slurry on aluminum current collector foil (20 μm thickness). The composition of the slurry was 90 % LTO, 5 % SuperC65, 1 % carboxymethylcellulose and 4 % latex, affording an active mass loading of 1.3 mAh.cm⁻². LiNi_1/3Mn_1/3Co_1/3O₂ composite electrodes (NMC) were prepared by coating a N-methylpyrrolidone-based slurry on aluminum current collector foil. The composition of the slurry was 92 % NMC, 2 % SuperC65, 4 % polyvinylidene fluoride and 2 % vapor grown carbon fibers, affording an active mass loading of 1.1 mAh.cm⁻². After roll pressing both positive and negative electrode thicknesses were 60 μm, and their porosity was 35 %. The electrodes were stamped out in 16 mm diameter discs, dried at 80°C under vacuum for 48 hours and stored in an argon-filled glove box.

LTO // NMC lithium-ion coin cells (∼ 2 mAh) were assembled by encapsulating negative electrode, glass fiber separator (Whatman GF/A), 150 μL of electrolyte, positive electrode and inox plate into 2032-type coin cells in an argon-filled glove box. LTO // NMC lithium-ion pouch cells of 9 cm² (∼ 10 mAh) were assembled as follow. The negative electrode (35 mm side), glass fiber separator (Whatman GF/A, 4 cm side) and positive electrode (32 mm side) were encapsulated into three layer PE/Al/PE film in a dry room. The system was dried overnight under vacuum at 55°C, and 0.5 mL of electrolyte were added in an argon-filled glove box. To ensure the complete wetting of the electrodes and separator by the electrolyte, each cell was kept at 60°C for 12 h before the electrochemical experiments.

The cells were galvanostatically cycled at 60°C with an Arbin multi-channel potentiostat / galvanostat, between the cut off voltages of [1–3.5 V]. The capacities were reported in mAh.g⁻¹, based on the limiting active material mass of NMC. Overcharge of LTO // NMC battery consisted in a charge up to 4.5 V (6 V vs Li) at C/10-rate followed by a float voltage for 20 hours at 60°C.

Volumes were determined by weighing the pouch cells in air and in ethanol, and the difference between these two masses was divided by ethanol density, 0.789 g.cm⁻³. The volume increase was reported after subtraction of initial volume (before any cycling). Gases were taken up from the pouch cell with a gas syringe, through a septum and in an argon-filled glove box. They were transferred into the injector of GC-IR apparatus (Trace GC Ultra and Nicolet 6700 from Thermo Scientific), maintained at 150°C. Gases were brought by helium carrier gas into a QBond column (l = 30 m, d = 0.32 mm) where organic gases were retained and sent first to an infrared spectrometer and then to a FID detector. Permanent gases were separated by a carboxene column (l = 15 m, d = 0.32 mm) and then sent to a TCD detector.

Solid state analyses of the electrodes were performed at RT after opening the cells and washing the electrodes in dimethylcarbonate in an argon-filled glove box. The crystalline phase of the material was investigated by XRD analyses, performed on Bruker D8 Advance diffractometer at 33 kV and 45 mA with Cu-kα radiation (λ = 0.15418 nm). A Bragg-Brentano geometry was used with PSD Vantec-1 detector where Kβ radiation was filtered by nickel foil. The diffraction data were recorded at room temperature for 2θ angles between 5° and 70°. The XPS experiments were carried out in a KRATOS AXIS Ultra DLD spectrometer equipped with a hemispherical analyzer and a delay line detector. The base pressure in the analysis chamber was 5.10⁻⁹ mbar. All the data were acquired using monochromated Al Kα X-rays 1486.6 eV, 150 W, at a normal angle with respect to the plane of the surface. The survey scans were obtained at 160 eV pass energy, while high-resolution core-level spectra were measured at 20 eV pass energy. The fit of N-bonded C 1s peak is wide as it includes both C-C-N and N-C-N from the imidazolium cation.