Editors' Choice—Coating-Dependent Electrode-Electrolyte Interface for Ni-Rich Positive Electrodes in Li-Ion Batteries

Pinar Karayaylali; Ryoichi Tatara; Yirui Zhang; Kuei-Lin Chan; Yang Yu; Livia Giordano; Filippo Maglia; Roland Jung; Isaac Lund; Yang Shao-Horn

doi:10.1149/2.0461906jes

Understanding electrode/electrolyte interface (EEI) is crucial to increase cycling performance and safety of Li-ion batteries.^1–6 Lithium nickel manganese cobalt oxides has been promising positive electrode materials due to their increased initial charge capacity as the nickel content increases.^7–9 However, significant capacity loss^7–9 and thermal instability^8,10–13 are observed once the Ni content increases in NMC electrodes. This impedance growth and reduced capacity retention^7,8,14,15 has been attributed to the solvent oxidation^7,9,16,17 and greater salt decomposition.¹⁶

Numerous studies have shown that surface modification of Co-based and Ni-based positive electrode materials can increase capacity retention in Li-ion batteries. Different coating materials have been studied such as metal oxides (Al₂O₃,^18–21 TiO₂,^22–26 Nb₂O₅^27,28), metal phosphates (AlPO₄^21,29–31), and metal fluorides (AlF₃³²). There are several schools of thoughts for the mechanism or physical origin to enhanced capacity retention associated with coated electrode materials. Coatings on the positive electrode would act as a protective layer that reduces parasitic reactions between the positive electrode and electrolyte³³ and a HF scavenger^30,34–36 that suppresses the transition metal dissolution from positive electrolyte surface to the electrolyte. In addition, the coating on some oxides could trigger the formation of metal fluoride on the surface during cycling, which can reduce the electrode reactivity toward the electrolyte and decrease the impedance growth of the positive electrode surfaces.^35–39 Moreover, coating materials can be converted to metal fluorides,^37,40 reducing electrode surface reactivity toward the electrolyte. For example, Myung et al.³⁷ have shown that the Al₂O₃ coating layer might form aluminum fluoride/oxyfluoride using Time-of-Flight secondary ion mass spectrometry (ToF-SIMS) on the Li(Li_0.05Ni_0.4Co_0.15 Mn_0.4) electrode at 60°C after extensively cycled. Similarly, Lu et al.⁴⁰ have studied the AlPO₄- coated LiCoO₂^40–42 electrodes using X-ray photoelectron spectroscopy (XPS) and found Al-containing fluorides and/or oxyfluorides on the cycled coated electrodes.

More recently, density functional theory calculations^43,44 have shown that the driving force for carbonate solvent dehydrogenation on oxides, yielding surface protic species, increased with greater Ni-content in NMC (NMC111, 622 and 811).¹⁶ Ex-situ infrared and Raman spectroscopy have revealed EC dehydrogenation^16,43,44 on charged NMC surfaces, where resulted protic species on charged NMC surfaces can further react with LiPF₆ to generate less-fluorine-coordinated species.^16,45,46 Greater salt decomposition was coupled with increasing EC dehydrogenation on charged NMC with increasing Ni or lithium de-intercalation.¹⁶ It is hypothesized that coating materials might have much lower thermodynamic tendency to dehydrogenate or dissociate carbonate solvents, which can reduce salt decomposition and capacity loss during cycling.^30,33,47

In this study, we investigate the electrochemical performances of uncoated, Al₂O₃-, Nb₂O₅- and TiO₂-coated NMC622 electrodes using LP57 electrolyte (1M LiPF₆ in ethylene carbonate (EC) and ethyl methyl carbonate (EMC) 3:7 wt:wt) by cycling measurements and electrochemical impedance spectroscopy (EIS). Density Functional Theory calculations reveal that the high band-gap compounds such as Al₂O₃ have a lower tendency to bind hydrogen than materials with smaller band gap such as TiO₂ and Nb₂O₅. While on semiconducting materials hydrogen adsorbs as a proton on the oxygen site and one electron is transferred to the metal states, for instance to the metal 3d states in the case of TiO₂, the large band of Al₂O₃ prevents this charge transfer resulting in unfavorable adsorption.⁴⁷ We employ Fourier-transform infrared spectroscopy (FT-IR) and X-ray photoelectron spectroscopy (XPS) to understand the reaction intermediates and salt and solvent decomposition on charged uncoated and coated NMC622 electrodes. Through those efforts, we have linked the electrochemical performances of uncoated and coated NMC622 electrodes to the solvent and salt decomposition on these electrodes, which will give information for rational design of stable positive electrode-electrolyte interfaces.

Experimental

Oxide synthesis and electrode preparation

The pellet electrode was prepared through pelletizing around 48 mg of active materials using 6 mm diameter pressing die set (Across International) for 15 mins. The pellets were then sintered under oxygen flow at 750°C for 6 hours, for NMC622 electrodes. The cooling and heating rates are controlled to be 2°C/min. The pellet electrodes are then broken into pieces of around 3 mg each, then dry in vacuum under 120°C overnight before transferred into Argon-filled glovebox (<0.5 ppm of H₂O and O₂).

The carbon-free binder free electrodes were prepared by mixing active materials with N-methyl-2-pyrrolidone (NMP) (Sigma Aldrich) in a 1:100 mass ratio. After bath-sonication for 30 min, the ink was deposited on ½ inch diameter aluminum discs and dried at 100°C. Composite electrodes were prepared by mixing active material (85%), carbon black (2% Csp, Timcal, 5% KS6) and polyvinylidene fluoride (PVDF) (8%, Kynar) dispersed in NMP with homogenizer (Thinky AR-100). The slurry was then bladed onto aluminum sheet with a gap of 10 μm. Both composite and carbon-free binder-free electrodes were punched and pressed at 6.3 T cm⁻² under a hydraulic press, to favor embedding of the powder in the aluminum disc in the case of the carbon-free, binder-free electrodes. Finally, the electrodes were completely dried at 120°C under vacuum for 24 h. Al₂O₃, TiO₂ and Nb₂O₅ coated NMC622 (LiNi_0.6Mn_0.2Co_0.2O₂) materials prepared by atomic layer deposition (ALD) were obtained from Forge Nano.

Electrochemistry

Electrochemical behavior of the electrodes was confirmed by galvanostatic measurements in two-electrodes cells (Tomcell type TJ-AC) and coin cell (CR2016). For coin-cell cycling, the formation cycle was done at C/10 rate (27.5 mA/g) from upper cut-off voltage (4.2, 4.4 and 4.6 V_Li) to 2 V_Li with constant current constant voltage (CCCV) charging with a cutoff current of C/20 (13.8 mA/g) for two cycles, followed by 1 C rate cycling (275 mA/g) from upper cut-off voltage (4.2, 4.4 and 4.6 V_Li) to 2 V_Li. Cells were assembled in an argon-filled glovebox (<0.5 ppm of H₂O and O₂) and comprised a lithium metal foil as the negative electrode, separated by two pieces of polypropylene separator (2500 Celgard), impregnated with 100 μL of LP57 (1M LiPF₆ in a 3:7 ethylene carbonate (EC): ethylmethyl carbonate (EMC)) electrolyte (BASF). After assembly, the cells rested for 6 h prior to measurement and then were charged with different end-of-charge potentials (4.1, 4.2, 4.4 and 4.6 V_Li) at a C/100 rate, based on the theoretical capacity calculated assuming full delithiation. The cells were maintained at end-of-charge potential for 5 h before disassembly in the glovebox. Carbon-free, binder-free electrodes were gently rinsed with 100 μL of EMC and dried under vacuum at room temperature for 3 h. No rinsing was performed for pellet electrodes to enable probing electrolyte features.

Three-electrode cells were used to investigate Electrochemical Impedance Spectroscopy (EIS) for coated-NMC622 composite electrodes. These cells were assembled in argon-filled glovebox (<0.5 ppm of H₂O and O₂) with a Li metal foil (15 mm diameter), 2 pieces of Celgard 2325 (19 mm diameter, MTI) as the separators, Li₄Ti₅O₁₂ mesh reference electrode (18 mm diameter), 2 pieces of Celgard 2325 (19 mm diameter) again, and composite electrode (1/2 inch diameter) from bottom to top, where a mesh Li₄Ti₅O₁₂ reference electrode was placed between positive and negative electrode with two separators with 200 μL of 1 mol/L LiPF₆ in a 3:7 wt:wt ethylene carbonate (EC): ethyl methyl carbonate (EMC) electrolyte (LP57, BASF) was used as electrolyte. The composite electrodes were charged with different end-of-charge potential at 27.5 mAh/g (C/10 rate based on theoretical capacity 275 mAh/g). After 10 minutes relax, EIS measurements were carried out at open circuit potential with 10 mV amplitude and frequency range from ∼10⁻² to 10⁶ Hz with VMP3 (Potentiostat with frequency response analyzer, Biologic), with temperature fixed at 25°C (Espec, SU-241). Additional details of cell/experimental configuration for three-electrode EIS method can be found in our previous work.⁴⁸ The resistance values from cycling data and the EIS measurements are different, which may arise from responses from negative and positive electrodes in two-electrode cells used for cycling measurements, whereas EIS measurements were made with three-electrode cell setup to measure impedance changes of the positive electrode alone.

FT-IR spectroscopy

The Fourier transform infrared (FT-IR) spectra of the materials were obtained on an FT-IR Tensor II (Bruker) equipped with deuterated triglycine sulfate (DTGS) detector inside an argon-filled glovebox with the H₂O and O₂ levels < 0.5 ppm. The species on active materials formed during (electro)chemical process was analyzed with the diffuse reflectance infrared Fourier transform (DRIFT) accessory (Praying Mantis, Harrick scientific products). Powder samples for DRIFT measurement were prepared by mixing active materials with KBr (>99.9% FT-IR grade, Sigma Aldrich) to give the concentration of active materials of 0.2 wt%. The DRIFT measurements were done with a 4 cm⁻¹ resolution in the 4000–400 cm⁻¹ spectral range at a scan velocity of 1.6 KHz; 256 scans were averaged. All FTIR spectra of the liquid samples were recorded using a single reflection ATR accessory (Pike Vee-Max II, Pike Technologies) with a Ge prism (Pier optics) at an incident angle of 45 degrees. The ATR measurements were performed with a 4 cm⁻¹ resolution in the 4000–400 cm⁻¹ spectral ranges at a scan velocity of 1.6 KHz; 256 scans were averaged. All spectra are shown in the absorbance units defined as log(I₀/I), where I₀ and I represent the background spectra and sample spectra, respectively. The background spectrum I₀ for DRIFT and ATR measurements were measured in the pure KBr powder and blank condition, respectively.

XPS measurements

All carbon-free, binder-free electrodes were transferred from glovebox to XPS chamber using a transfer vessel (ULVAC-PHI, INC.). For each potential, at least two electrodes were charged for reproducibility. All the XPS spectra were collected using a PHI 5000 VersaProbe II (ULVAC-PHI, INC.) using a monochromatized Al K_α source and a charge neutralizer. A pass energy of 23.5 eV was used and adventitious carbon at 285 eV (C1s spectra) was used for calibration of all XPS spectra. After subtraction of a Shirley-type background, photoemission lines were fitted using combined Gaussian-Lorentzian functions, except in the case of the Co, Ni and Mn 2p_3/2 lines where asymmetric line shapes were used. The RSF (relative sensitivity factors) values for C 1s, O 1s, F1s, P 2p, Li 1s, Co 2p_3/2, Ni 2p_3/2 and Mn 2p_3/2, Al 2p, Ti 2p, Nb 3d photoemission lines were given as 0.314, 0.733, 1, 0.525, 0.028, 2.3526, 2.468, 1.792, 0.256, 1.256, and 3.71, respectively. The chemical compositions, binding energies and full width at half maximum of all spectra can be found in Tables S1–S5.

Results and Discussion

Coating-dependent capacity loss and impedance growth of Al₂O₃-, Nb₂O₅- and TiO₂-coated NMC622 electrodes

Figure 1 shows the voltage profiles of uncoated, Al₂O₃-, Nb₂O₅- and TiO₂-coated (4 monolayers) NMC622 electrodes in the first, 50^th and 100^th cycles (after two formation cycles) with 4.2, 4.4 or 4.6 V_Li cut-off potentials. In the first cycle, the charge and discharge capacities of uncoated and coated NMC622 electrodes were comparable (Figure 1a), which increases from ∼160 mAh/g at 4.2 V_Li, ∼180 mAh/g at 4.4 V_Li to ∼200 mAh/g for 4.6 V_Li. Upon cycling to 4.2 V_Li, uncoated and coated NMC622 electrodes were found to have comparable capacities (for 100 cycles examined in this study), where uncoated and coated electrodes were found to have comparable voltage polarization. With increasing upper voltage cutoff, uncoated NMC622 was found to have the highest capacity loss and greatest growth in voltage polarization upon cycling to 4.6 V_Li in Figure 1c in comparison with coated NMC622. Among the coated electrodes, Al₂O₃-coated NMC622 electrodes had the highest capacity retention and minimum voltage polarization after 100 cycles while TiO₂-coated NMC622 electrodes had the highest capacity loss and resistance (voltage polarization).

The observed trend in the capacity loss and cell resistance growth upon cycling to 4.6 V_Li for uncoated and coated electrodes in Figure 1 is further supported by impedance measurements. The Nyquist plots for uncoated, Al₂O₃-, TiO₂- and Nb₂O₅-coated NMC622 composite electrodes cycled between 4.6 V_Li and 2 V_Li and EIS measurements at 4.1, 4.2 and 4.4 V_Li reveal two semicircles (Figure 2), where the high-frequency semicircle was assigned to the impedance associated with ion adsorption and desorption at the electrified interface while the low-frequency semicircle was attributed to the charge transfer impedance for the positive electrode.⁴⁸ This assignment is further supported by the observation that the low-frequency semicircle grew with higher charging potential from 4.1 V_Li to 4.4 V_Li, resulting from EEI layer growth and greater resistance for charge transfer, while the size of the high-frequency semicircle kept nearly constant as reported in our previous work.⁴⁸ Although R_HF value depends on electrolyte ionic conductivity, electronic conductivity of composite electrode and particle size of active materials, the frequency corresponding to the high frequency (left) semicircles are consistent (10³ – 10⁴ Hz) with other previous works.^48–52 In the first cycle, all electrodes had similar sizes for the high-frequency semicircle while uncoated and Al₂O₃-coated NMC622 had smaller low-frequency impedance than those found for TiO₂- and Nb₂O₅-coated NMC622. Upon cycling, the high-frequency semicircle remained unchanged while the low-frequency semicircle was found to grow, as shown in Figure 2. Of significance to note is that Al₂O₃-coated NMC622 electrodes had the lowest impedance growth for the low-frequency semicircle after 50 cycles, followed by Nb₂O₅-coated and TiO₂-coated NMC622 electrodes. This trend is consistent with that in the estimated resistance based on the cycling data collected from two-electrode cells in Figure 1.

**Figure 2.** Nyquist plots for uncoated (a), Al₂O₃- (b), Nb₂O₅- (c) and TiO₂- (d) coated NMC622 (4-monolayer) composite electrodes cycled between 4.6 V_Li and 2 V_Li with EIS collected at 4.1 (light blue), 4.2 (blue) and 4.4 V_Li (darker blue) in the first charge and after 50^th cycles at 25°C. Data were collected from in three-electrode cells with mesh Li₄Ti₅O₁₂ reference electrode and Li metal counter electrode in 1 M LiPF₆ in a 3:7 (w:w) EC: EMC solution. The cells were galvanostatically charged at C/10 at each potential, then relax 10 minutes before EIS measurements. The electrochemical profiles for each EIS cell are shown in Figure S3 and the comparison of electrochemical performance of three-electrode EIS cell and two-electrode coin cell for cycling are shown in Figure S4.

The decreased impedance growth for the low-frequency semicircle and increased capacity retention of Al₂O₃-coated NMC622 relative to uncoated NMC622 upon cycling to 4.6 V_Li can be attributed to reduced (chemical) oxidation of the electrolyte on the electrode, specifically dehydrogenation of carbonate solvents to generate protic species, which can decompose electrolyte salt anion (PF₆⁻).^16,43,53 This hypothesis is supported by much lower driving force for hydrogen adsorption on Al₂O₃ than LiMO₂ (M=Ni,Co) surfaces (Figure 3), which is shown to scale with the dissociation adsorption energetics of carbonate solvents in our recent work.⁵³ While TiO₂ coating has the highest tendency to dehydrogenate carbonate solvent among Al₂O₃, Nb₂O₅ and TiO₂ coating materials (Figure 3),⁴⁷ it has lower driving force for chemical oxidation when it is compared with layered positive electrode materials such as LiCoO₂ and LiNiO₂. This is supported by our cycling measurements of uncoated NMC 622 and TiO₂-coated NMC622 upon cycling to 4.6 V_Li, where uncoated NMC622 was found to have the higher capacity loss and the greatest growth in voltage polarization than all other coated NMC622 electrodes.

**Figure 3.** Hydrogen adsorption energy for positive electrode materials (LiCoO₂ and LiNiO₂) and coating materials (TiO₂, Nb₂O₅ and Al₂O₃ compounds), computed with PBE-DFT with respect to ½ H₂ in the gas phase.⁴⁷

Evidence for coating-dependent dehydrogenation of carbonate solvents and salt decomposition on coated NMC622 electrodes

DRIFT measurements of soaked and charged uncoated, Al₂O₃, Nb₂O₅ and TiO₂-coated NMC622 pellet electrodes (carbon-free and binder-free) revealed evidence of EC dehydrogenation, which became more pronounced when the coating changed from Al₂O₃ to TiO₂. The DRIFT signals, originated from the stretching modes of the EC corresponding to the C=O bonds, collected from charged Al₂O₃, Nb₂O₅ and TiO₂-coated electrodes to different end-of-charge potentials in the first charge are compared with those of soaked electrodes (only exposure to the electrolyte) and LP57 electrolyte in Figure 4, and those of the O-C-O region are shown in Figure S6. Two distinct features at 1807 and 1773 cm⁻¹ of the LP57 electrolyte¹⁶ can be assigned to the C=O stretching of EC and Li⁺-coordinated EC according to our recent work,¹⁶ respectively. Below we discuss the changes from soaked electrodes to the charged uncoated and coated NMC electrodes.

No significant shift for the peaks associated with the C=O stretching (centered around 1807 and 1773 cm⁻¹) was found for charged uncoated (Figure 4a) and Al₂O₃-coated NMC622 electrodes (4-monolayer in Figure 4b and 12-monolayer in Figure 4c) relative to soaked electrodes. In contrast, these peaks were blue shifted and broadened for charged 4-monolayer TiO₂- (Figure 4e) and Nb₂O₅-coated (Figure 4d) NMC622, which remained largely unchanged with increasing charging voltage for 4 monolayers (Figure 4). This blue shift (that induces peak broadening) became more evident with increased coating thickness from 4 to 12 monolayers for charged TiO₂-coated NMC622 but not for charged Al₂O₃-coated NMC622 (Figure 4). The blue shift and peak broadening can be attributed to species derived from EC dehydrogenation formed on NMC622 electrodes as reported in our recent work,¹⁶ which is supported by the calculated spectra for the Li⁺-EC and dehydrogenated Li⁺-EC shown in Figure 4. Similar blue shifts and peak broadening were found for the O-C-O stretching of EC on charged TiO₂-coated NMC622 while no significant shifts were noted for charged Al₂O₃-coated NMC 622 (Figure S6). Therefore, these results revealed evidence for dehydrogenation of EC on charged TiO₂-coated NMC622 but not Al₂O₃-coated NMC622, which is in agreement with greater driving force for hydrogen adsorption and consequently surface dissociative adsorption of carbonate solvents on TiO₂ than Al₂O₃ from DFT results (Figure 3).⁴⁷

Comparing the P-F region of the DRIFT spectra for soaked and charged coated NMC electrodes revealed the formation of less-fluorine coordinated species, which is accompanied with carbonate dehydrogenation. Dehydrogenation of EC or EMC generates protic species such as surface hydroxyl groups,^16,43 which can react with LiPF₆ salt to form less fluorine-coordinated species such as PF₃O or PF₅ species.^16,46 The possible reaction mechanisms include 2^* H + O_lat ⇒ H₂O + O_vac (^*H being H adsorbed at a lattice oxygen site); H₂O_oxide + PF₅ ⇒ PF₃O + 2HF; PF₃O + H₂O_oxide ⇒ HPF₂O₂ + HF; and 2HF +NMC and/or TiO₂/Nb₂O₅ ⇒ NMC-oxide/fluoride and/or titanium oxyfluoride+ H₂O.¹⁶ The sharp peak around 840 cm⁻¹ originated from P-F stretching in LiPF₆ in the LP57 electrolyte was broadened and blue shifted for charged uncoated NMC622 with increasing charging voltage (Figure 4a), which can be attributed to the formation of less-coordinated fluorinated species at higher potentials supported by DFT simulated spectra of PF₃O and PF₆⁻. In contrast, this peak was found largely unchanged for charged Al₂O₃-coated NMC622 with increasing voltage, which is in agreement with reduced reactivity toward dehydrogenation of carbonate solvents as shown by DFT (Figure 3) and FT-IR results in comparison to charged NMC. On the other hand, the P-F stretching peak was found to be broadened considerably for charged 12-monolayer TiO₂-coated NMC622, which was accompanied with increasing blue shift and peak broadening associated with the C=O stretching upon EC de-hydrogenation discussed earlier.

Greater reactivity toward carbonate and salt decomposition for TiO₂–coated NMC622 than Al₂O₃-coated NMC622 was further confirmed by XPS analysis. EEI layers formed on uncoated and Al₂O₃-, TiO₂- and Nb₂O₅-coated NMC622 carbon-free, binder-free electrodes charged to different potentials were studied by XPS analysis of C 1s, O 1s, F 1s, P 2p, Mn 2p, Co 2p, Ni 2p, Li 1s, Nb 3d, Al 2p and Ti2p spectra. The C 1s spectra were fitted to components 285.0, 286.3, 287.6 and 288.8 eV. These peaks can be attributed to adventitious carbon,⁵⁴ C-O bonds like ROLi⁵⁵ or polyethers, C=O/ O-C-O bonds^55,56 and O=C-O bonds⁵⁵. Only small contribution for CO₃ bonds was observed around 290.3 eV.⁴¹ The growth of EEI layer supported by the O 1s spectra that revealed five contributions for charged coated samples. The lowest energy contribution at 529 eV came from O lattice from coated NMC electrodes, which could be combination of O lattice^40,61–63 from coating or the NMC powder. The other four peaks are attributed to ROLi (∼531 eV),^56,58 semicarbonates around 532 eV (ROCO₂Li and CO₃ group)^41,59 and polyethers species around 533.4 eV (O-C=O bonds)^6,56,57 and lithiated fluorophosphates species (Li_xPF_yO_z).⁶⁰ The O lattice peak was shifted to higher binding energy with increasing voltage, which is an indication of the oxyfluoride formation on the surface. These oxyfluoride species can be formed by the protic species attacking NMC622 electrode and/or coating on the surface of the electrode.¹⁶ The indicative O lattice shift to higher binding energy can be correlated with the shifts of Ni 2p, Ti 2p and Nb 3d to higher binding energy which suggests formation of lithium nickel titanium or lithium nickel niobium oxyfluorides on the surfaces of TiO₂ and Nb₂O₅ coated NMC622 electrodes (Figure 7).⁶² We still see formation of lithium nickel oxyfluorides on Al₂O₃-coated NMC622 surfaces similar to uncoated NMC622 electrodes, however since the Al 2p spectra does not shift to higher binding energy, we do not see evidence of formation of aluminum oxyfluoride species which definitely shows less reactivity of carbonate electrolyte on Al₂O₃-coated NMC622 than TiO₂- and Nb₂O₅-coated NMC622 electrodes.

**Figure 7.** XPS spectra of the Ni 2p, Al 2p and Ti 2p photoemission lines for (a) Al₂O₃-coated (4 monolayers) and (b) TiO₂-coated (4 monolayers) for pristine and charged carbon-free, binder-free electrodes to 4.1, 4.2, 4.4, and 4.6 V_Li with 1 M LiPF₆ in EC: EMC (3:7 wt:wt) electrolyte. All spectra were calibrated with the adventitious hydrocarbons at 285.0 eV and background corrected using a Shirley background. The representative electrochemical profile is shown in Figure S18. The Co 2p, Mn 2p, Ni 2p and Li 1s/Co 3p XPS spectra is shown in Figures S19-23, the quantification is shown in Figure S24 and Table S1 and S3.

Increasing tendency toward greater salt decomposition can be further confirmed by XPS analysis of their F 1s spectra in Figures 5 and 6. Figures 5 and 6 show the 4-monolayer and 12-monolayer TiO₂- and Al₂O₃-coated NMC622 electrodes. The intensity of F 1s spectra is from all 4-monolayer Al₂O₃-, Nb₂O₅- and TiO₂-coated NMC622 electrodes quite similar and can be found Figures S20 and S24. As the Al₂O₃ thickness of the coating increases, the F1s spectra intensity does not change from 4-monolayer Al₂O₃ to 12-monolayer Al₂O₃ coating, which is consistent with DRIFT P-F spectra in Figure 4. However, with increasing TiO₂ coating, the F1s spectra intensity increases, especially formation of the lithium nickel titanium oxyfluoride species on the surface. The increasing F 1s intensities can be deconvoluted to lithium or other transition-metal fluorides (LiF and/or LiM_xF_yO_z) at 685.0 eV⁵⁷ and lithiated or transition metal fluorophosphates (Li_xPF_yO_z) at 686.5 eV.⁵⁷ The formation of LiM_xF_yO_z-like species can be accompanied with a higher binding energy shift at Ni 2p spectra from 854.5 eV (pristine NMC622 electrodes) to 856 eV (charged NMC622 electrodes) and Ti 2p spectra from 458.5 to 459.0 eV. These LiM_xF_yO_z-like species can be formed by the HF species attacking the coating material and/or charged NMC electrodes as suggested previously, where HF can be generated by reacting protic species from dehydrogenation of EC with LiPF₆ species.¹⁶ The intensity of the Li_xPF_yO_z-like species with respect to LiM_xF_yO_z-like species was increased considerably for charged 12-monolayer TiO₂-coated NMC622 electrodes with increasing potential, which suggests greater reactivity of LiPF₆ salt with more protic species produced from dehydrogenation of EC. Although the ratio of Li_xPF_yO_z to LiM_xF_yO_z-like species was similar for both 4-monolayer and 12-monolayer Al₂O₃ coated NMC622 electrodes, the ratio of these species has been significantly changed when the thickness of the TiO₂ coating increased from 4-monolayer to 12-monolayer which indicates the increasing salt decomposition triggered by protic species derived from EC dehydrogenation (Figure S25).^16,43

**Figure 5.** XPS spectra of the C 1s, O 1s and F 1s photoemission lines for Al₂O₃-coated 4 (a) and 12 (b) monolayers for the pristine carbon-free, binder-free electrodes and after charging to 4.1, 4.2, 4.4, and 4.6 V_Li with 1 M LiPF₆ in EC: EMC (3:7 wt:wt) electrolyte. All spectra were calibrated with the adventitious hydrocarbons at 285.0 eV and background corrected using a Shirley background. C1s spectra were assigned with the following contributions: C-H/C-C (∼285 eV),⁵⁴ C-O (∼286.3 eV),⁵⁵ C=O/O-C-O (∼287.6 eV),^55,56 O=C-O (∼288.8 eV)⁵⁵ and CO₃ (∼290.3 eV).⁴¹ O1s spectra were assigned with the following contributions: O lattice (∼529.3 eV),⁵⁷ ROLi (∼531 eV),^56,58 surface O/CO₃/O-C=O (∼532 eV),^41,59 C-O/O-C=O/OP(OR)₃ (E_b∼533.4 eV)^6,56,57 and Li_xPF_yO_z (E_b∼534.8 eV).⁶⁰ The F1s spectra were deconvoluted to three different species: lithium or metal fluoride species around 685 eV,⁵⁷ lithium or metal fluorophosphate around 686.5 eV⁵⁷ and lithium hexafluorophosphate around 688 eV.⁴⁰ The representative electrochemical profile is shown in Figure S10. The P 2p, Co 2p, Mn 2p, Ni 2p and Li 1s/Co 3p XPS spectra is shown in Figures S11-12, the quantification is shown in Figure S13 and Table S1-S2.

**Figure 6.** XPS spectra of the C 1s, O 1s and F 1s photoemission lines for TiO₂-coated 4 (a) and 12 (b) monolayers for the pristine carbon-free, binder-free electrodes and after charging to 4.1, 4.2, 4.4, and 4.6 V_Li with 1 M LiPF₆ in EC: EMC (3:7 wt:wt) electrolyte. All spectra were calibrated with the adventitious hydrocarbons at 285.0 eV and background corrected using a Shirley background. C1s spectra were assigned with the following contributions: C-H/C-C (∼285 eV),⁵⁴ C-O (∼286.3 eV),⁵⁵ C=O/O-C-O (∼287.6 eV),^55,56 O=C-O (∼288.8 eV)⁵⁵ and CO₃ (∼290.3 eV).⁴¹ O1s spectra were assigned with the following contributions: O lattice (∼529.3 eV),⁵⁷ ROLi (∼531 eV),^56,58 surface O/CO₃/O-C=O (∼532 eV),^41,59 C-O/O-C=O/OP(OR)₃ (E_b∼533.4 eV)^6,56,57 and Li_xPF_yO_z (E_b∼534.8 eV).⁶⁰ The F1s spectra were deconvoluted to three different species: lithium or metal fluoride species around 685 eV,⁵⁷ lithium or metal fluorophosphate around 686.5 eV⁵⁷ and lithium hexafluorophosphate around 688 eV.⁴⁰ The representative electrochemical profile is shown in Figure S14. The Co 2p, Mn 2p, Ni 2p, Li 1s/ Co 3p, P 2p and Ti 2p XPS spectra is shown in Figures S15-16, the quantification is shown in Figure S17 and Table S3-S4.

There are several reports on aluminum-based coated positive electrodes showing better electrochemical performance than uncoated positive electrodes such as Al₂O₃ coated-LiCoO₂ cycled at 4.3 V_Li⁶⁴ and 4.4 V_Li,⁶⁵ Al₂O₃-coated NMC111 cycled at 4.5 V_Li,⁶⁶ Al(OH)₃ coated – Li[Li_0.2Ni_0.2Mn_0.6]O₂ cycled at 4.6 V_Li.⁶⁷ One important protection mechanism that has been discussed in previous works is called HF-scavenging^30,34–36 where coating material reacts with HF and forms metal fluoride species on the surface (e.g. Al₂O₃ + 6HF = > 2AlF₃ + 3H₂O) and would reduce the electrolyte reactivity on positive electrode surfaces and improve voltage polarization.^27,36 However, in this work, we showed that Al₂O₃-coated NMC622 electrodes had the minimum voltage polarization and lowest impedance growth and DRIFT C=O region and XPS F 1s analysis did not reveal obvious signs of EC dehydrogenation and formation of aluminum fluoride/oxyfluoride on Al₂O₃-coated NMC622 electrodes whereas the worst electrochemical performance was shown by TiO₂-coated NMC622, where we showed clear signs of EC dehydrogenation and greater salt decomposition including formation of lithium-nickel-titanium oxyfluoride species. Dahn et al.²⁶ reported the titanium-based coated on NMC532 electrodes can help preventing transition metal dissolution and improve the cycling performance. This may be the reason why we have also seen better electrochemical performance for TiO₂-coated NMC622 than uncoated NMC622 using two-electrode coin cell setup. However, our three-electrode EIS cell measurements indicate TiO₂-coated NMC622 has higher impedance growth than uncoated NMC622, since we only observe the effect of positive electrode side, mainly the effect of EC dehydrogenation. These results can be further supported by our recent work,⁴⁷ where we show Al₂O₃ has lower tendency to bind hydrogen than TiO₂ and Nb₂O₅ coating materials, so as we lower the Fermi level with respect to O-p band (from Al₂O₃ to TiO₂), the dissociation of EC molecule where protic species such as surface hydroxyl groups are formed becomes more thermodynamically favorable. We propose that the high band-gap insulators such as Al₂O₃ are the best coating materials due to their lower reactivity with electrolyte solvents and salts.

Conclusions

This work combines EIS, DRIFT and XPS spectroscopies together with cycling measurement to understand the reactivity of coating materials on NMC622 electrodes toward carbonate electrolyte with LiPF₆ salt and this reactivity is linked with the impedance and cycling measurements. By combining cycling and EIS measurements, we show coating-dependent electrochemical performance of NMC622 upon cycling to 4.6 V_Li. Al₂O₃-coated NMC622 had reduced capacity loss and impedance growth while TiO₂-coated NMC622 had greater capacity loss and impedance growth than uncoated NMC622. Combining XPS and DRIFT measurements on C=O and P-F regions, we show evidence for dehydrogenation of EC on TiO₂-coated NMC622 positive electrodes, where it forms protic species and these protic species can react with LiPF₆ salt to form less-fluorine coordinated species such as PF₃O-like (DRIFT) and lithium nickel titanium or niobium oxyfluoride species (XPS). Exposing NMC622 surface more to TiO₂ coating makes the surface more prone to dehydrogenation of EC and formation of less-fluorine coordinated species. Through these key findings, we propose key reaction intermediates for TiO₂ and Al₂O₃ coating on NMC622 and we believe Al₂O₃ coating should be used for Ni-rich positive electrodes such as NMC622 due to reduced reactivity toward chemical oxidation of carbonate solvents and the development of stable positive electrode-electrolyte interface during cycling.

Acknowledgment

This work made use of the MRSEC Shared Experimental Facilities at MIT, supported by the National Science Foundation under award number DMR-1419807. Research at MIT related to this work was supported financially by BMW.

ORCID

Ryoichi Tatara 0000-0002-8148-5294

Yang Shao-Horn 0000-0001-8714-2121

Editors' Choice—Coating-Dependent Electrode-Electrolyte Interface for Ni-Rich Positive Electrodes in Li-Ion Batteries

Article metrics

Author affiliations

Author notes

ORCID iDs

Dates

Abstract

Experimental