Yike Lei,
Yingchuan Zhang,
Yongkang Han,
Jie Ni,
Cunman Zhang and
Qiangfeng Xiao*
School of Automotive Studies, Clean Energy Automotive Engineering Center, Tongji University (Jiading Campus), 4800 Cao'an Road, Shanghai, 201804, P. R. China. E-mail: xiaoqf@tongji.edu.cn
First published on 5th June 2023
The unique anion redox mechanism of Li-rich Mn-based layered oxide (LMLO) cathodes endows them with a higher specific capacity compared with conventional cathodes. However, the irreversible anion redox reactions can cause structural degradation and sluggish electrochemical kinetics in the cathode, resulting in a poor electrochemical performance in the batteries. Thus, to address these issues, a single-sided conductive oxygen-deficient TiO2−x interlayer was applied on a commercial Celgard separator as a coating layer towards the LMLO cathode. After coating TiO2−x, the initial coulombic efficiency (ICE) of the cathode increased from 92.1% to 95.8%, the capacity retention improved from 84.2% to 91.7% after 100 cycles, and the rate performance of the cathode was significantly enhanced from 91.3 mA h g−1 to 203.9 mA h g−1 at 5C. Operando differential electrochemical mass spectroscopy (DEMS) showed that the coating layer could restrain the release of oxygen in the battery, especially from the initial formation process. The X-ray photoelectron spectroscopy (XPS) results demonstrated that the favorable oxygen absorption by the TiO2−x interlayer benefitted the suppression of side reactions and cathode structural evolution and favored the formation of a uniform cathode-electrolyte interphase on the LMLO cathode. This work provides an alternative path to address the issue of oxygen release in LMLO cathodes.
Recently, many different strategies have been proposed to solve the issue of oxygen release in LMLO cathodes. For example, doping was performed to stabilize the crystal structure and reduce the release of oxygen.13 Also, a concentration gradient structure of LMLO was designed to increase the stability of the structure and avoid unexpected surface reactions to reduce oxygen release.14 In addition, various materials, such as oxides,15 phosphates,16 fluorides,17 lithium-ion conductors,18 polymers,19 and functionalized coatings containing oxygen vacancies, were coated on the cathode to inhibit oxygen release by reducing side reactions.20,21 However, doping with foreign elements and/or coating with inert compounds may reduce the capacity, and even reduce the conductivity of the material and diminish the rate capability.21 In addition, both methods involve complex modification processes, which possibly destroy the surface structure of the cathode materials.
Modification of the separator is also commonly employed to improve the electrochemical performance of LIBs. Although the separator interlayer inevitably reduces the energy density of the whole battery, the improvement in electrochemical performance, thermal stability and safety performance by the interlayer of separator have attracted increasing attention from researchers.22,23 Accordingly, researchers have investigated various coatings on the separator to improve the electrolyte wettability and thermal stability of the battery.24,25 Parikh et al. developed a binary ceramic coating consisting of Al2O3 and TiO2 on a thin separator to enhance the thermal stability, thermal conductivity and electrolyte wettability.26 Qi et al. proposed a mesoporous SiO2 (mSiO2)-anchored separator via covalent bonding, where the mSiO2 nanoparticles facilitated the storage of more electrolyte and enhanced the swift lithium-ion diffusion during charge/discharge.27 However, this type of ceramic coating layer on the separator is generally designed to enhance the electrolyte wettability, thermal stability and mechanical properties,28 and there is no report to date on modified separators that can store the oxygen released from LMLO during charge and discharge.
Black titanium dioxide (TiO2−x) has been widely applied in fuel cells, photoelectrochemical sensors and microwave absorbers.29 TiO2−x is a member of the homologous series known as Magneli phases,30 where the substoichiometric titanium oxides consist of two-dimensional chains of octahedral TiO6, with the oxygen atom missing in every nth layer to compensate for the loss of stoichiometry.31 This structure ensures the excellent electrical conductivity and great corrosion resistance of TiO2−x materials.32,33 In addition, TiO2−x has good electrochemical stability in corrosive media (including organic electrolytes).34 Furthermore, TiO2−x has been applied as the separator coating layer of lithium–sulfur batteries due to its high affinity for polysulfides.35,36 However, the use of TiO2−x as a functional interlayer to absorb the oxygen evolved from high-energy cathodes (e.g., LMLO and high-nickel cathodes) has not been reported. The evolution of oxygen is the key issue hindering the commercialization of LMLO cathode materials. As mentioned previously, oxygen evolution can cause structural degeneration, voltage decay and electrolyte decomposition, which can result in inferior electrochemical performances in LMLO.
In this work, we proposed the use of oxygen-deficient TiO2−x as an interlayer to suppress the oxygen release of LMLO for the first time. As shown by the schematic in Fig. 1, Li+ and oxygen are liberated from the cathode during the charge process and migrate to the anode. The oxygen vacancies in the TiO2−x interlayer adsorb the released oxygen, while Li+ migrates to the anode. In addition, the porous structure of the coating layer on the separator facilitates the adsorption of oxygen without the hindrance of lithium-ion transportation. Consequently, this TiO2−x interlayer improved the initial coulombic efficiency and enhanced the cyclability of the LMLO cathode due to its oxygen absorption effect and the mitigated side reactions between oxygen radicals and the electrolyte.
Fig. 1 Schematic illustration of the mechanism of the TiO2−x-coated separator for the inhibition of oxygen release in LMLO cathodes by an TiO2−x interlayer. |
To investigate the effect of the TiO2−x interlayer on the electrochemical performance of the LMLO cathode, the coated Celgard 2400 separators with different TiO2−x loadings were assembled with an LMLO cathode and Li-metal anode in CR2032-type coin cells. The coin cells were charged and discharged at 0.1C for the first two cycles, followed by 0.2C for cyclability testing. Fig. 4a shows the initial charge and discharge curves of the batteries with pristine and TiO2−x-coated Celgard 2400 separators. All the batteries demonstrated the typical features of LMLO materials, namely, a sloping region below 4.4 V, followed by a high voltage plateau at around 4.5 V during the first charge process. The initial coulombic efficiency (ICE) and the discharge capacity of the LMLO cathode with the pristine Celgard 2400 separator was 92.1% and 303.0 mA h g−1, while that for the TiO2−x-1, TiO2−x-2, TiO2−x-3 and TiO2−x-4 samples was 95.5%, 95.5%, 95.8%, and 95.3%, and their initial discharge capacities were 306.8 mA h g−1, 308.9 mA h g−1, 306.2 mA h g−1, and 308.2 mA h g−1, respectively. The ICE of the LMLO cathode with the TiO2−x-coated separators was significantly enhanced compared with the pristine Celgard 2400 separator. Moreover, the discharge capacity of the cathodes with the TiO2−x-coated separators also increased slightly compared with that with that of the pristine Celgard 2400 separator. The higher ICE and discharge capacity demonstrate the excellent reversibility of the LMLO cathode, which is due to the inhibition of oxygen release by the TiO2−x interlayer and favorable Li-ion transport rate. In addition, the charge and discharge curves of the cathode with the pristine Celgard 2400 separator and TiO2−x coated separators almost overlapped, which indicates that the TiO2−x-coated interlayer improved the reversibility of the electrode, while it did not participate in the lithium deintercalation and electrochemical reaction. As shown in Fig. 4b, the batteries with the TiO2−x-coated separator exhibited a similar cycling performance in the first 10 cycles as that with the pristine Celgard 2400 separator. However, with an increase in the cycle number, the discharge capacity of the battery with the pristine Celgard 2400 separator decreased rapidly. In contrast, the discharge capacity of the battery with the TiO2−x-coated separator was stable. After 100 cycles, the cycle retentions of the LMLO cathodes with the Celgard 2400, TiO2−x-1, TiO2−x-2, TiO2−x-3 and TiO2−x-4 separators were 84.2%, 90.6%, 91.7%, 91.5%, and 90.7%, respectively. In addition, the batteries with the TiO2−x-coated separator exhibited lower average voltage attenuation than that with the pristine Celgard 2400 separator. As shown in Fig. 4c, the voltage of the batteries with the coated separator similarly decreased to 3.1271 V after 100 cycles, which is higher than 3.1025 V for the battery with the pristine Celgard 2400 separator. The electrochemical performances of the batteries indicate that the separators with different TiO2−x loadings all effectively enhanced the ICE and cycle stability of the LMLO cathode compared to the Celgard 2400 separator.
The TiO2−x-coated separator also enhanced the battery rate performance when the batteries were charged at 0.1C and discharged under different rates from 0.1C to 5C, and finally back to 0.1C. As shown in Fig. 4d, the battery with the TiO2−x-2 separator exhibited a better performance than that with the TiO2−x-1, TiO2−x-3 and TiO2−x-4 separators and delivered the specific discharge capacities of 305.9, 288.9, 274.4, 262.3, 245.5, 230.9 and 203.9 mA h g−1 at 0.1C, 0.2C, 0.5C, 1C, 2C, 3C and 5C rate, respectively. These results are higher than that of the battery with the pristine Celgard 2400 separator, i.e., 304.8, 286.4, 271.3, 258.5, 238.8, 209.1, and 91.3 mA h g−1 under the same rate testing condition. When the rate returned to 0.1C, the discharge capacities mostly recovered for all the batteries. The corresponding discharge curves of the cathode with the pristine Celgard 2400 and TiO2−x-2 separators at different rates are demonstrated in Fig. 4e and f, respectively. The battery with the pristine Celgard 2400 exhibited a higher polarization and lower discharge capacity than that with the TiO2−x-2 separator. The enhanced rate performance of the battery with the TiO2−x-2 separator can be ascribed to the superior electrical conductivity of the TiO2−x coating layer.40
The gas evolution during the first charge/discharge of the cathode with different separators was analysed by operando differential electrochemical mass spectroscopy (DEMS). The effect of the TiO2−x coating layer on the gas evolution was evaluated by analysing the evolution of O2 and CO2. The voltage curves and the gas partial pressure were examined in commercial ECC-DEMS cells, as shown in Fig. 5, where the baseline of the gas partial pressure was removed to eliminate the influence of the environment.41 As shown in Fig. 5b, the O2 evolution appeared at the end of the charge process due to the irreversible evolution of lattice oxygen, which is consistent with the literature.42 The release of O2 in the battery with the TiO2−x-2-coated separator was significantly weaker than that in the battery with the pristine Celgard 2400 separator, benefitting from the positive effect of the TiO2−x interlayer on the oxygen absorption and reduced side reactions. The CO2 evolution, as shown in Fig. 5c, shows two waves for the battery with the pristine Celgard 2400 separator. The first wave of CO2 evolution at around 4.5 V is due to the decomposition of the alkyl carbonate solvent.6 The second wave of CO2 evolution appeared at the end of the charge, arising from the interaction between the electrolyte solvent and the LMLO lattice.43 Conversely, there was almost no CO2 wave at 4.5 V for the battery with the TiO2−x-2 separator, which may be due to the expanded electrochemical stability window induced by the TiO2−x coating. As demonstrated by the linear sweep voltammetry (LSV) measurement of the cells containing a stainless steel working electrode and lithium foil counter/reference electrode in Fig. S2,† the electrochemical anodic stability of the electrolyte (1 M LiPF6 in FEC/DMC (1/3 vol%) for TiO2−x) was greatly improved by the TiO2−x coating. In addition, the CO2 evolution from the battery with the TiO2−x-2 separator was significantly weaker than that from the battery with the pristine Celgard 2400 separator, which can be attributed to the mitigated side reactions between the electrolyte solvent (FEC) and LMLO lattice due to the TiO2−x coating layer. The behaviour of O2 and CO2 evolution shows that the TiO2−x coating layer could absorb the oxygen released from LMLO and restrain the side reactions between LMLO and the electrolyte.44
The effect of TiO2−x coating layer on the interfacial resistance and charge transfer kinetics during battery cycling was investigated by electrochemical impedance spectroscopy (EIS). The EIS results of the LMLO‖Li batteries after 5 and 100 cycles are exhibited in Fig. 6a and b, respectively. The equivalent circuit of the Nyquist plots for the activated and cycled batteries consist of two semicircles at high and medium frequencies and a short straight line at low frequency, respectively. The high frequency semicircle is associated with the surface film resistance (Rsf) and the medium frequency semicircle is associated with the charge transfer resistance (Rct). The Nyquist plots are fitted with the equivalent circuit model in the inset image in Fig. 6a and the fitting values of EIS are demonstrated in Table S1.† Obviously, the battery with the pristine Celgard 2400 separator demonstrated a higher Rsf compared with the batteries with the TiO2−x-coated separators after 5 cycles. The Rct of the battery with the pristine Celgard 2400 separator is slightly higher than that of the batteries with the TiO2−x-coated separators. The Rsf of the battery with the pristine Celgard 2400 separator improved from 24.2 Ω after 5 cycles to 28.2 Ω after 100 cycles, which is attributed to the side reactions during the cycle process on the electrode surface. The Rsf of the batteries with the TiO2−x-2 and TiO2−x-3 separators after 100 cycles slightly decreased compared with that after five cycles, which can be ascribed to the adsorption of oxygen and inhibition of electrolyte decomposition due to the TiO2−x coating layer during cycling. The evolution of Rct upon cycling can explain the high stability and excellent rate performance of the battery with the TiO2−x-2 separator. The Rct of the battery with the TiO2−x-2 separator increased from 23.8 Ω in the 5th cycle to 65.5 Ω in the 100th cycle, while the battery with the pristine Celgard 2400 separator presented a greater increase from 29.2 Ω to 174.7 Ω under the same testing condition. The value of Rsf slightly increased from 18.0 Ω to 22.3 Ω and the Rct significantly increased from 20.2 Ω to 162.3 Ω after 100 cycles for the batteries with TiO2−x-1, which can be attributed to the insufficient inhibition of side reactions and oxygen absorption effect by the thin TiO2−x interlayer. In the case of the battery with the TiO2−x-4 separator, the Rsf increased from 20.3 Ω to 30.0 Ω and Rct increased from 28.7 Ω to 155.2 Ω after 100 cycles due to the thicker TiO2−x interlayer with a porous structure, which may facilitate the absorption of more electrolyte and higher Li-ion tortuosity than the TiO2−x-2 sample. However, it is worth noting that although the total impedance values of the batteries with TiO2−x-1 and TiO2−x-4 increased significantly after 100 cycles, their impedance values were still smaller than the battery with the Celgard 2400 separator. Besides, the batteries with TiO2−x-1 and TiO2−x-4 demonstrated a higher capacity retention than the battery with the Celgard 2400 separator due to the positive effect of the TiO2−x interlayer on the enhancement of LMLO cathode structure stability during long cycles. Combined with the O2 and CO2 evolution results, the variation in Rsf and Rct also indicates that the TiO2−x coating layer inhibited the release of oxygen and restrained the side reaction on the LMLO electrode.
Fig. 6 Nyquist plots measured after (a) 5 cycles and (b) 100 cycles for the coin cells used Celgard 2400 separator and TiO2−x-coated separators. |
To evaluate the process of oxygen adsorption, the chemical states of titanium and oxygen in the TiO2−x coating layer upon charge/discharge were detected by XPS. The TiO2−x-coated separators were obtained by disassembling coin-cells in the charged and discharged states in an Ar-filled glovebox. The O 1s and Ti 2p XPS spectra of the fresh TiO2−x-coated separator and that in the charged and discharged states from the first cycle are shown in Fig. 7a and b. As shown in Fig. 7a, the fitted peaks located at about 531.6 eV, 530.8 eV and 529.4 eV correspond to absorbed oxygen, oxygen vacancy and lattice oxygen, respectively.45–48 In the case of the fresh TiO2−x-coated separator, oxygen vacancies and lattice oxygen in are present in the O 1s spectra. After charging to 4.8 V, the intensity of the oxygen vacancy peak decreased and adsorbed oxygen appeared, indicating that the TiO2−x coating layer adsorbed the oxygen released from the LMLO lattice.49 The intensity of the adsorbed oxygen peak decreased and the intensity of the oxygen vacancy peak increased after discharge, indicating that some of the absorbed oxygen desorbed during discharge, which may be influenced by the electric field.50,51 Fig. 7b shows the presence of Ti4+ and Ti3+ in the Ti 2p spectra with the binding energy of Ti3+ lower than that of Ti4+, and the existence of Ti3+ confirms the presence of oxygen vacancies in TiO2−x.40 The contents of Ti3+ and Ti4+ were semi-quantitatively analysed by fitting the XPS results. The contents of Ti3+ for the fresh TiO2−x-coated separator and that in the charged and discharged states were determined to be 42.76%, 34.23%, and 41.28%, respectively. The excess electrons produced by the existence of oxygen vacancies could freely hop at room temperature to balance the valence states.52–54 These results demonstrate that the content of Ti3+ decreased during the battery charging process, and subsequently increased during the discharging process, which correspond to the changes in absorbed oxygen and oxygen vacancies. In addition, powder XRD was performed to investigate the possible phase changes in the crystal structure of the TiO2−x coating layer on the separator during the charge and discharge process. As shown in Fig. S3,† the XRD patterns of the fresh TiO2−x-coated separator and that in the charged and discharged states are very similar, suggesting that there are no great or slight changes in the crystal structure of the TiO2−x coating layer during charging and discharging.
Fig. 7 (a) O 1s and (b) Ti 2p XPS spectra of the fresh, first charged and first discharged TiO2−x-coated separators. |
To deeply understand the structural evolution during the first charge/discharge process of the LMLO cathode and the influence of the TiO2−x interlayer on the ICE of the LMLO cathode, ex situ XRD was conducted on the LMLO cathodes disassembled from the cells with a Celgard 2400 separator and with TiO2−x-coated separator at different charge/discharge potentials.55 The ex situ XRD patterns of the two samples between 10° to 45° during the first charge and discharge process are shown in Fig. S4.† 56 As shown in Fig. 8a and b, the (003) and (104) diffraction peaks representing the lattice parameters c and a are magnified to observe the change in lattice parameters during the first charge and discharge. The LMLO crystal is arranged by periodic Li layers or transition metal (TM) layers and oxygen layers. The (003) diffraction peak corresponds to the [001] crystal direction, c represents the lattice distance between the oxygen layer and Li layer or TM layer, and a represents the lattice parameter of the crystal plane composed of lithium, TM and oxygen.57 During the first charge process, the (003) peak shifted to a lower angle before the charge plateau due to the removal of lithium and the increased electrostatic repulsion between the oxygen layers. Then, the (003) peak shifted to a higher angle during the charge plateau due to the activated Li2MnO3 and release of oxygen.58 It was found that the changes in the (003) peak in the LMLO cathode with the Celgard 2400 separator and with TiO2−x-coated separator during the first charge process are similar. During the first discharge process, similar to previous reports in the literature, the lithium ions moved to the opposite direction, the (003) peak moved to a higher angle firstly, and then to a lower angle, which is contrary to the trend during the charge process.57 However, it can be distinguished by comparing the two samples that the (003) peaks of the LMLO cathode with TiO2−x separator exhibited a greater shift to a lower angle than that of the LMLO cathode with the Celgard 2400 separator, especially at a low potential before the end of the discharge. These results are probably caused by the higher reversible anion redox due to the TiO2−x interlayer,59 leading to extra capacity in the low potential region (Fig. 4a).60 In addition, the (104) peak shifted to a higher angle during the first charge process, and then moved to a lower angle during the first discharge process, which is related to the extraction/insertion of lithium ions and the migration of TM ions.57 The LMLO cathode with the TiO2−x-coated separator had a smaller angle shift of the (003) and (104) diffraction peaks at the end of discharge than the LMLO cathode with the Celgard 2400 separator due to the improved reversibility during the first charge and discharge.61
To understand the influence of the TiO2−x interlayer on the structural evolution of the LMLO cathode during the cycling process, the coin cells after 100 cycles were disassembled to compare the XRD patterns of the LMLO cathode with the Celgard 2400 separator and that with the TiO2−x-coated separator. As shown in Fig. S5,† the crystal structure of both LMLO cathodes degraded compared with the fresh LMLO electrode. The split peaks of (006)/(102) demonstrate the ordered layered structure, and the lower intensity ratio of (003) and (104) exhibit the higher cation mixing.21 The careful observation of I(003)/I(104) shows that the cathode with the TiO2−x-coated separator it matched the fresh LMLO cathode, but was higher than the cathode with the Celgard 2400 separator. In addition, the split peaks of (006)/(102) in the LMLO cathode with the TiO2−x-coated separator were noticeable, while that in the LMLO cathode with the Celgard 2400 separator was almost indistinguishable, which may be caused by the devastating release of oxygen.62 These results indicate that the TiO2−x-coated separator mitigated the release of oxygen and improved the stability of the LMLO crystal structure compared to the cathode with the Celgard 2400 separator.
The cycled cathodes with different separators were characterized by XPS to analyse the influence of the TiO2−x interlayer on the CEI of the LMLO cathodes. The fitted peaks in C 1s (Fig. 9a and e) and O 1s (Fig. 9b and f) of CO and C–O are attributed to the decomposition of carbonate and side reactions. By comparing the two samples, there were more CO and C–O species on the cathode with the Celgard 2400 separator than the cathode with the TiO2−x-2 separator. In addition, the cathode with the Celgard 2400 separator demonstrated more Li2CO3 on its surface according to the O 1s spectra, which is harmful to the cathode.63 According to the F 1s spectra in Fig. 9c and g and P 2p spectra in Fig. 9d and h, both samples contained LixPFy and LixPOyFz, which is attributed to the decomposition of LiPF6 during cycling.21 However, the cathode with the Celgard 2400 separator exhibited more TM–F and Li–F species, which may be derived from the strong reactions between the LMLO lattice and the electrolyte. To demonstrate the influence of the TiO2−x interlayer on the CEI morphology and structure decay of the LMLO cathode intuitively, the HRTEM images of the cathode after 100 cycles are shown in Fig. S6† and 10. As shown in Fig. S6,† the LMLO cathode with the Celgard 2400 separator exhibited a discontinuous and nonuniform CEI, while the LMLO cathode with the TiO2−x-2 separator exhibited a continuous CEI with a thickness of about 4 nm, which is thinner and more uniform than that on the LMLO cathode with the Celgard 2400 separator. In addition, Fig. 10 demonstrates the structural decay of the cycled cathodes with the Celgard 2400 and TiO2−x-2 separators. Fig. 10a shows that there are plenty rock-salt structure domains in the surface region, and many voids and spinel-like structures in the inner region together with a few layered structures. In contrast, as shown in Fig. 10b, spinel-like structure domains appeared near the surface and the layered structure was well-maintained in the inner regions due to the absorption of oxygen and inhibition of side reactions by the TiO2−x interlayer.
Fig. 10 TEM images of LMLO cathodes after 100 cycles: (a) cathode with Celgard 2400 separator and (b) cathode with TiO2−x-2 separator. |
Considering the potential effect of the electrochemical shuttle on the cell components,64 the morphology and composition of the Li anode-faced sides of the separator may be valuable to analyze the effect of the TiO2−x interlayer and cathode stability.65 Fig. S7a† shows that a part of the SEI remained on the surface of the Celgard 2400 separator with large broken Li dendrites embedded in the separator, while the surface of the Li anode-faced side of the TiO2−x-coated separator (Fig. S7b†) appeared smoother and had a much smaller quantity of particles. In the XPS spectra shown in Fig. S8,† the products of carbonate decomposition and Li oxide were observed on the Li anode-faced side Celgard 2400 separator, while many Li fluoride components were observed on the Li anode-faced side TiO2−x-2 separator. The aforementioned results further indicate that the TiO2−x coating layer on the separator prevented the occurrence of excessive side reactions and enhanced the cathode stability, which can be attributed to the absorption of oxygen on the TiO2−x coating layer, preventing oxygen and oxygen radicals from crossing the separator and enhancing the electrochemical stability at a high voltage. To determine the influence of the TiO2−x coating layer on the impedance of the cathode and anode electrodes, a three-electrode assembly was used to demonstrate the resistance of the LMLO cathode and Li anode separately.66 As shown in Fig. S9,† the impedance of the three-electrode was roughly consistent with the EIS of the two-electrode coin-cell configuration. In the initial cycles, the impedance values of the Li anode were almost the same, while the impedance values of the LMLO cathode continued to change with an increase in the cycle numbers. This result indicates that the TiO2−x interlayer mainly benefitted the formation of a more conductive CEI at cathode side.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3ra02125d |
This journal is © The Royal Society of Chemistry 2023 |