Lithium-Rich Cobalt-Free Manganese-Based Layered Cathode Materials for Li-Ion Batteries: Suppressing the Voltage Fading

: Lithium-rich layered oxides are recognized as promising materials for Li-ion batteries, owing to higher capacity than the currently available commercialized cathode, for their lower cost. However, their voltage decay and cycling instability during the charge / discharge process are problems that need to be solved before their practical application can be envisioned. These problems are mainly associated with a phase transition of the surface layer from the layered structure to the spinel structure. In this paper, we report the AlF 3 -coating of the Li-rich Co-free layered Li 1.2 Ni 0.2 Mn 0.6 O 2 (LLNMO) oxide as an e ﬀ ective strategy to solve these problems. The samples were synthesized via the hydrothermal route that insures a very good crystallization in the layered structure, probed by XRD, energy-dispersive X-ray (EDX) spectroscopy, and Raman spectroscopy. The hydrothermally synthesized samples before and after AlF 3 coating are well crystallized in the layered structure with particle sizes of about 180 nm (crystallites of ~65 nm), with high porosity (pore size 5 nm) determined by Brunauer–Emmett–Teller (BET) speciﬁc surface area method. Subsequent improvements in discharge capacity are obtained with a ~5-nm thick coating layer. AlF 3 -coated Li 1.2 Ni 0.2 Mn 0.6 O 2 delivers a capacity of 248 mAh g − 1 stable over the 100 cycles, and it exhibits a voltage fading rate of 1.40 mV per cycle. According to the analysis from galvanostatic charge-discharge and electrochemical impedance spectroscopy, the electrochemical performance enhancement is discussed and compared with literature data. Post-mortem analysis conﬁrms that the AlF 3 coating is a very e ﬃ cient surface modiﬁcation to improve the stability of the layered phase of the Li-rich material, at the origin of the signiﬁcant improvement of the electrochemical properties.


Introduction
The energy density (E in Wh g −1 ) of a battery is determined by the amount of charge it can hold (Ah kg −1 ) and by the voltage across its terminal (V); optimization of its electrochemical performance is achieved by an appropriate selection of anode and cathode materials [1]. In this context, layered Li-rich materials are attracting a lot of attention, because their capacity is the best among the lithium intercalation materials. In addition, they are not expensive and non-toxic when they are cobalt-free. Layered lithium-rich cobalt-free manganese-based oxides (LLNMOs) have been widely considered as new cathode materials for Li-ion batteries . They are constituted by the architectural solid solution yLi 2 MnO 3 (1−y)LiMO 2 (M = Mn, Ni), which includes the monoclinic Li 2 MnO 3 (C2/m space group) and the rhombohedral LiMO 2 (R3m space group) phase. Both R3m and C2/m phases have a Several strategies have been employed to enhance the electrochemical properties of any cathode materials and reduce the voltage decay rate upon long-term cycling. The first approach is surface modification, since the coat can protect the active particles against corrosion and dissolution, and improve the electrical contact between powders (for a review, see [1]). Various substances have been deposited on the surface of active particles such as Li + -conductor (Li2ZrO3, LiAlO2) [13,14], inert compound (AlF3, Al2O3) [15], wide-gap semiconductor (ZnO) [16], and carbon [17,18]. Li1.2Ni0.2Mn0.6O2 nanoparticles (100-200 nm in size) coated with a thin layer of Li2ZrO3 obtained through a synchronous lithiation strategy delivered a capacity of circa 169 mAh g −1 over 100 cycles at current density 200 mA g −1 [13]. ZnO coating also improved the electrochemical properties of 0.5Li2MnO3·0.5LiNi0.5Mn0.5O2; a specific discharge capacity of 215 mAh g -1 was reported after 10 cycles for ZnO treated LLNMO against 181 mAh g −1 for pristine electrode [19]. Enhanced electrochemical properties were observed for Li1.2Ni0.2Mn0.6O2 synthesized nickel enrichment is a second way to mitigate the voltage fading. For example, Shi et al. [20] demonstrated that Ni ions act as stabilizing ions to inhibit the Jahn-Teller effect of active Mn 3+ ions in 0.5Li2MnO3 0.5LiNi0.8Co0.1Mn0.1O2. Doping or cationic substitution is another successful method to reduce the voltage drop [21,22]. Substitution of a small amount of M 3+ (M = Al, Cr, Fe, Co) for Mn 4+ and Ni 2+ in Several strategies have been employed to enhance the electrochemical properties of any cathode materials and reduce the voltage decay rate upon long-term cycling. The first approach is surface modification, since the coat can protect the active particles against corrosion and dissolution, and improve the electrical contact between powders (for a review, see [1]). Various substances have been deposited on the surface of active particles such as Li + -conductor (Li 2 ZrO 3 , LiAlO 2 ) [13,14], inert compound (AlF 3 , Al 2 O 3 ) [15], wide-gap semiconductor (ZnO) [16], and carbon [17,18]. Li 1.2 Ni 0.2 Mn 0.6 O 2 nanoparticles (100-200 nm in size) coated with a thin layer of Li 2 ZrO 3 obtained through a synchronous lithiation strategy delivered a capacity of circa 169 mAh g −1 over 100 cycles at current density 200 mA g −1 [13]. ZnO coating also improved the electrochemical properties of 0.5Li 2 MnO 3· 0.5LiNi 0.5 Mn 0.5 O 2 ; a specific discharge capacity of 215 mAh g -1 was reported after 10 cycles for ZnO treated LLNMO against 181 mAh g −1 for pristine electrode [19]. Enhanced electrochemical properties were observed for Li 1.2 Ni 0.2 Mn 0.6 O 2 synthesized nickel enrichment is a second way to mitigate the voltage fading. For example, Shi et al. [20] demonstrated that Ni ions act as stabilizing ions to inhibit the Jahn-Teller effect of active Mn 3+ ions in 0.5Li 2 MnO 3 0.5LiNi 0.8 Co 0.1 Mn 0.1 O 2 . Doping or cationic substitution is another successful method to reduce the voltage drop [21,22]. Substitution of a small amount of M 3+ (M = Al, Cr, Fe, Co) for Mn 4+ and Ni 2+ in Li(Li 0.2 Mn 0.6 Ni 0.2 )O 2 sensitively influences the length of the plateau region due to the changes in the metal-oxygen bond covalence [23]. Ba doping improves the first-cycle Coulombic efficiency of Li 1.2 Mn 0.6 Ni 0.2 O 2 by mitigating the oxidation of O 2− ions [24]. The synergetic effects of coating and doping were used by Liu et al. who synthesized Li 1.2 Ni 0.2 Mn 0.6 O 2 by co-precipitation technique with 3 wt.% LiAlO 2 coat and 0.8% Cr doping [14]: (i) the spinel phase was suppressed, (ii) the midpoint voltage increased from 2.746 to 3.128 V, and (iii) higher D Li contributed to good rate capability. Nevertheless, even though doping was able to significantly improve the electrochemical performance of the material, it did not solve the problem of side reactions with the electrolyte, dissolution, and phase transition of the surface layer to the spinel phase. Surface modification remains mandatory. However, the efficiency of the coating is also debated, because it depends on several parameters, including full surface coverage, thickness of the coating layer, the composition of the coat, and the morphology of the particles. This last parameter also plays an important role. In particular, use of nano-sized particles is needed for any cathode element to shorten the diffusion path. However, in the particular case of the α-NaFeO 2 structure of Li 1.2 Ni 0.2 Mn 0.6 O 2 considered here, it is also necessary to increase the size of the active (010) lattice planes at the surface.
In the present work, the voltage decay in long-term cycled Li 1.2 Ni 0.2 Mn 0.6 O 2 is investigated for pristine and AlF 3 -coated samples. AlF 3 has been chosen as the inert material to coat the particles. The hydrothermal method has been chosen not only to minimize the fading rate, but also because it is well suited for the preparation of nano-sized powders with a spherical-like shape that is considered as the best morphology to optimize the electrochemical properties. The structure and morphology of the as-prepared sample are analyzed by X-ray diffraction (XRD), Raman spectroscopy, Brunauer-Emmett-Teller (BET) specific surface area method, scanning electron microscopy (SEM), and high-resolution transmission electron microscopy (HRTEM). Electrochemical properties are studied using galvanostatic charge-discharge (GCD) and electrochemical impedance spectroscopy (EIS) experiments. Chemical diffusion coefficients of Li + ions in LLNMO lattices are evaluated. Finally, the rate capability and cycle ability of LLNMO electrodes are further investigated and discussed. As a result, importantly, both the rate capability and cycle ability have been improved, with a capacity of 248 mAh g −1 stable over the 100 cycles. Moreover, the voltage fading rate has been reduced to 1.40 mV per cycle. These results prove the efficiency of the AlF 3 coating to protect the surface layer of the Li-rich particles. Li-rich layered Li 1.2 Ni 0.2 Mn 0.6 O 2 powders were prepared by a two-step process, i.e., a hydrothermal synthesis followed by a solid-state calcination. Lithium and transition-metal (TM) acetates (analytical grade from Merck KGaA, Darmstadt, Germany) were mixed in distilled water at the stoichiometric composition and stirred for 15 min. A homogeneous slurry was obtained after adding citric acid drop-wise to the solution under continuous stirring for 6 h with molar ratio 1:1 of (Li+Ni+Mn):C 6 H 8 O 7 . The pH value was adjusted to to 7-8 with ammonium hydroxide. The slurry was transferred into a 100 mL Teflon-lined stainless-steel autoclave maintained at a constant temperature of 180 • C for 12 h. Then the product was dried and calcined at 500 • C for 3 h, and thereafter post-treated at 900 • C for 12 h in air. The AlF 3 -coating layer was fabricated by the chemical deposition method. The as-prepared Li 1.2 Ni 0.2 Mn 0.6 O 2 powders were immersed in the Al(NO 3 ) 3 dilute aqueous solution heated to 50 • C and stirred vigorously. NH 4 F dilute solution was then added into the solution with the molar ratio of Al to F of 1:3 corresponding to the 5-wt.% AlF 3 coat. After drying at 60 • C in a vacuum oven, the obtained products were heated at 400 • C in ambient argon for 4 h to prevent the formation of Al 2 O 3 and secure the formation of AlF 3 -coated Li 1

Material Characterization
The crystalline phase was identified by XRD using a Philips X'Pert diffractometer (PANalytical, Orsay, France) equipped with CuKα radiation source (λ = 1.5406 Å). The diffractograms were recorded at room temperature in the 2θ-range 10-80°. The experimental patterns were refined by the Rietveld method with FULLPROF software program. The variance of the relative amounts of atomic components in the samples were analyzed by inductively coupled plasma-atomic emission spectrometry (ICP-AES, OPTIMA 7000DV, Perkin Elmer Co., Ltd., UK). The morphology and composition were studied by SEM (ZEISS model ULTRA 55) coupled with an energy-dispersive X-ray attached analyzer (EDX). HRTEM images were obtained using an electronic microscope JEOL model JEM-2010. Raman spectra were taken using a micro-Raman spectrophotometer (Horiba, Longjumeau, France) equipped with an optical microscope (×100 objective) and a 633 nm He-Ne laser excitation line. Data were recorded in steps of 1.6 cm −1 with an acquisition time of 30 s. The wavenumber calibration was regularly verified by using the Raman peak at 520 cm −1 of a silicon crystal as a reference. BET specific surface area was measured by nitrogen adsorption/desorption at 77 K in a relative pressure P/P0 = 0.0-1.0, with P and P0 being the equilibrium and saturation pressure, respectively, using a Quantachrome NOVA automated gas sorption analyzer (Anton Paar GmbH, Blankenfelde-Mahlow, Germany). The pore size distribution and pore volume were estimated using the Barrett-Joyner-Halenda (BJH) method [46].
Electrochemical tests were conducted on CR2025-type coin cells assembled in a glove box (moisture and oxygen content ≤5 ppm) under argon atmosphere. The procedure to fabricate the cathode was as follows: 85 wt.% active material, 10 wt.% carbon black as conductive medium, and 5 wt.% polyvinylidene fluoride (PVDF) in N-methyl pyrrolidinone (NMP) solvent were mixed and ground to generate a homogeneous slurry. The slurry was then spread onto an aluminum foil current collector and dried at 80 °C for 2 h to remove the solvent before being pressed. Using such a

Material Characterization
The crystalline phase was identified by XRD using a Philips X'Pert diffractometer (PANalytical, Orsay, France) equipped with CuKα radiation source (λ = 1.5406 Å). The diffractograms were recorded at room temperature in the 2θ-range 10-80 • . The experimental patterns were refined by the Rietveld method with FULLPROF software program. The variance of the relative amounts of atomic components in the samples were analyzed by inductively coupled plasma-atomic emission spectrometry (ICP-AES, OPTIMA 7000DV, Perkin Elmer Co., Ltd., UK). The morphology and composition were studied by SEM (ZEISS model ULTRA 55) coupled with an energy-dispersive X-ray attached analyzer (EDX). HRTEM images were obtained using an electronic microscope JEOL model JEM-2010. Raman spectra were taken using a micro-Raman spectrophotometer (Horiba, Longjumeau, France) equipped with an optical microscope (×100 objective) and a 633 nm He-Ne laser excitation line. Data were recorded in steps of 1.6 cm −1 with an acquisition time of 30 s. The wavenumber calibration was regularly verified by using the Raman peak at 520 cm −1 of a silicon crystal as a reference. BET specific surface area was measured by nitrogen adsorption/desorption at 77 K in a relative pressure P/P 0 = 0.0-1.0, with P and P 0 being the equilibrium and saturation pressure, respectively, using a Quantachrome NOVA automated gas sorption analyzer (Anton Paar GmbH, Blankenfelde-Mahlow, Germany). The pore size distribution and pore volume were estimated using the Barrett-Joyner-Halenda (BJH) method [46].
Electrochemical tests were conducted on CR2025-type coin cells assembled in a glove box (moisture and oxygen content ≤5 ppm) under argon atmosphere. The procedure to fabricate the Energies 2020, 13, 3487 5 of 22 cathode was as follows: 85 wt.% active material, 10 wt.% carbon black as conductive medium, and 5 wt.% polyvinylidene fluoride (PVDF) in N-methyl pyrrolidinone (NMP) solvent were mixed and ground to generate a homogeneous slurry. The slurry was then spread onto an aluminum foil current collector and dried at 80 • C for 2 h to remove the solvent before being pressed. Using such a cathode (loading of~2 mg cm −2 ), the electrochemical cells fabricated with lithium sheet as the counter electrode, Celgard 2500 film as the separator filled with 1 mol L -1 LiPF 6 in a mixture of ethylene carbonate (EC)/dimethyl carbonate (DMC) (1:1) solution (LP30, Merk) as electrolyte. The galvanostatic charge-discharge GCD) cycles were performed using a potentiostat/galvanostat (workstation VMP3 Bio-Logic, Seyssinet-Pariset, France) in the potential range 2.0-4.8 V. EIS data were collected using a phase sensitive multimeter (model PSM 1700, UK) in the frequency range 0.01 Hz-100 kHz using a voltage bias of 5 mV.

Structure and Composition
The relative concentration of atoms was measured by ICP-AES. Results listed in Table 1 show that the chemical composition of the LLNMO samples is close to the theoretical stoichiometry. Moreover, the amount of Al is consistent with the value expected by the surface treatment. The~2% loss of Li in pristine LLNMO should be attributed to its departure from the surface during the calcination process. Figure 3 displays the Rietveld refinement of the XRD patterns of pristine and AlF 3 -coated Li 1.2 Ni 0.2 Mn 0.6 O 2 oxides. For both samples, these results suggest the structural co-existence of the Li 2 MnO 3 (C2/m S.G.) and LiNi 0.5 Mn 0.5 O 2 (R3m S.G.) phases. XRD diagrams exhibit the typical diffraction patterns of the α-NaFeO 2 -type layered structure with the R3m space group (hexagonal setting) [47]. Additional reflections appeared in the 20-21 • 2 θ-region that are the fingerprints of the Li 2 MnO 3 -type C2/m phase. Such a superstructure suggests the presence of Li + ions in the transition-metal layer of Li 1.2 Ni 0.2 Mn 0.6 O 2 . The existence of this secondary phase is also detected by Raman spectroscopy, as it will be evidenced in the following. For the coated sample, the XRD features are unchanged in agreement with other works [48,49]. were collected using a phase sensitive multimeter (model PSM 1700, UK) in the frequency range 0.01 Hz-100 kHz using a voltage bias of 5 mV.

Structure and Composition
The relative concentration of atoms was measured by ICP-AES. Results listed in Table 1 show that the chemical composition of the LLNMO samples is close to the theoretical stoichiometry. Moreover, the amount of Al is consistent with the value expected by the surface treatment. The ~2% loss of Li in pristine LLNMO should be attributed to its departure from the surface during the calcination process. Figure 3 displays the Rietveld refinement of the XRD patterns of pristine and AlF3-coated Li1.2Ni0.2Mn0.6O2 oxides. For both samples, these results suggest the structural co-existence of the Li2MnO3 (C2/m S.G.) and LiNi0.5Mn0.5O2 (R 3 m S.G.) phases. XRD diagrams exhibit the typical diffraction patterns of the α-NaFeO2-type layered structure with the R 3 m space group (hexagonal setting) [47]. Additional reflections appeared in the 20-21° 2 θ-region that are the fingerprints of the Li2MnO3-type C2/m phase.
Such a superstructure suggests the presence of Li + ions in the transition-metal layer of Li1.2Ni0.2Mn0.6O2. The existence of this secondary phase is also detected by Raman spectroscopy, as it will be evidenced in the following. For the coated sample, the XRD features are unchanged in agreement with other works [48,49].  From the Rietveld refinement results (Table 2), the lattice expansion occurred in both the a and c directions when LLNMO particles are coated with AlF3. Thus, the volume of the elementary unit cell increases slightly with the coating. In stoichiometric Li1.2Ni0.2Mn0.6O2, the valence state of Mn and Ni ions are +4 and +2, respectively [50,51]. The primary effect of the AlF3 coating is a reduction of the loss of Li ions at the surface of particles that provokes an increase of the Li2MnO3 component content and, consequently, an increase of the amount of Mn 3+ ions. The slight change of the lattice parameters is explained by consideration of the ionic radii of Mn ions. The increase in a and c is attributable to the bigger ionic radius of Mn 3+ ions in octahedral coordination (58 pm against 53 pm From the Rietveld refinement results (Table 2), the lattice expansion occurred in both the a and c directions when LLNMO particles are coated with AlF 3 . Thus, the volume of the elementary unit  [50,51]. The primary effect of the AlF 3 coating is a reduction of the loss of Li ions at the surface of particles that provokes an increase of the Li 2 MnO 3 component content and, consequently, an increase of the amount of Mn 3+ ions. The slight change of the lattice parameters is explained by consideration of the ionic radii of Mn ions. The increase in a and c is attributable to the bigger ionic radius of Mn 3+ ions in octahedral coordination (58 pm against 53 pm for Mn 4+ ions) [52]. It also gives evidence that an internal composite structure exists in the Li-rich oxide framework. This increasing concentration of Mn 3+ has been reported to be beneficial to the electrochemical performance. The existence of Mn 3+ causes a gradual activation of Li 2 MnO 3 , which results in a capacity increase upon cycling [53]. For the R3m phase of Li 1.2 Ni 0.2 Mn 0.6 O 2 , the refinement involved fixing the Mn occupancy and varying the ratios of the Ni and Li occupancies on the TM and Li sites. The degree of Ni occupancy in the interlayer Li sites (cationic mixing) is rather low at 1.33% and 1.25% for pristine and coated LLNMO samples, respectively. There are several indicators to evaluate the presence of Li-Ni anti-sites (cation mixing). First, the c/a lattice parameter ratio evidences the deviation of the rock-salt structure (i.e., c/a > 4.90). The second indicator is the peak intensity ratio R 1 = I 003 /I 104 ; the higher R 1 , the lower amount of undesirable Li-Ni anti-sites and better hexagonal structure [54]. The third fingerprint of the hexagonal ordering is defined by the factor R 2 = (I 006 + I 102 )/I 101 ; the lower the R 2 , the better the hexagonal ordering [55,56]. The values of R 1 and R 2 reveal that the hydrothermally synthesized LLNMO samples have a perfect layered structure. As seen in Table 2, the Rietveld refinement results are consistent with the direct observation of the XRD peak intensity ratios showing the lower cation mixing with well-ordered rhombohedral structures for the AlF 3 coated LLNMO sample. The local deformation of the crystal structure is investigated by the analysis of the broadening of the XRD reflections according to the Williamson-Hall relationship [57,58]: where B is the full-width at half-maximum (FWHM), K is the crystallite shape factor (assumed to be 0.9), <e> is the micro-strain field, L c is the size of the crystallites (coherence length), θ is the diffraction angle and λ is the X-ray wavelength of the source. The microstrain is extracted from the slope of the plot of B cosθ vs. sinθ in Figure 4, and L c is determined by extrapolation of the straight line to sinθ = 0.
Both samples have identical L c of 65 ± 5 nm. The results show that the microstrain of 2.45 × 10 −2 for pristine LLNMO slightly reduces to 2.12 × 10 −2 for the coated sample. The smaller value of <e> in the AlF 3 -coated sample proves that the thin layer stabilizes the lattice structure, a consistent result with the fact that the coating process has reduced the number Li-Ni anti-sites by~7%. The decrease of the strain field gives evidence of a better structural stability of the coated sample than the pristine one, as expected from the above discussion devoted to the lattice parameters and the consideration of ionic radii.
Energies 2020, 13, x FOR PEER REVIEW 7 of 22 The porosity of hydrothermally synthesized LLNMO samples was studied by N2 adsorption/desorption tests. Figure 5a,b presents the N2 adsorption/desorption isotherm and the pore-size distribution (inset) for pristine and AlF3-coated LLNMO samples. An increase in the amount of adsorbed N2 with increase of the P/P0 value and the appearance of a hysteresis loop are observed in the isotherm curves. Results obtained from BJH analysis are listed in Table 3. The average BJH pore width is in the range 4.5-5.5 nm, which indicates that these holes are mainly mesopores with total pore volume of 0.45 cm 3 g −1 . These features are fingerprints of a hierarchical mesoporous structure of LLNMO powders that promotes the Li + ion kinetics [59]. Suib et al. [60] described the formation of mesoporosity by aggregation of nanoparticles along their lateral faces. In the same way, the mesopores (~5 nm) correspond here to the voids existing between randomly-packed nanoparticles of ~150 nm in diameter inside secondary particles (agglomerates) of ~600 nm in size, as we will see on TEM images reported later in this work.
BET analysis shows that pristine and AlF3-coated LLMNO samples have a specific surface area of 11.5 and 9.2 m 2 g −1 , respectively, which is advantageous for enhancing the transport of Li + ions in the secondary particles. These values are comparable with those reported by Chang et al. [61] for Li-rich Li1.27Cr0.2Mn0.53O2 oxides prepared by milling-assisted solid-state process. Xiang et al. [62] reported the influence of polyvinyl alcohol (PVA) as chelating agent on the BET surface area for Li1.2Ni0.2Mn0.6O2 prepared by sol-gel technique (varying from ~10 m 2 g −1 for PVA-0% to ~33 m 2 g −1 for PVA-15%). Zhang et al. [63] showed that 3-µm sized secondary particles, which consisted of self-assembled crystallites ~150 nm in size, demonstrated optimized mesostructuration partially mitigating the voltage fading. The equivalent particle size of the samples can be calculated from the BET data and compare with SEM images. The average particle diameter (in nm) is expressed by the relation [64]: where SBET is the specific surface area (in m 2 g −1 ) measured by BET experiments and d the gravimetric density (4.53 g cm −3 for NMO). The BET surface area and the average particle size results are found in Table 3. We observe a relatively good agreement between particle size values LBET and LTEM evaluated from BET and measured from TEM images, respectively. Table 3. Brunauer-Emmett-Teller (BET) specific surface area, calculated average particle size from BET data (LBET) using Equation (2), mean particle size measured using TEM imaging (LTEM) and results obtained from Barrett-Joyner-Halenda (BJH) analysis for hydrothermally prepared LLNMO samples.

LBET LTEM
Pore Size Pore Volume The porosity of hydrothermally synthesized LLNMO samples was studied by N 2 adsorption/desorption tests. Figure 5a,b presents the N 2 adsorption/desorption isotherm and the pore-size distribution (inset) for pristine and AlF 3 -coated LLNMO samples. An increase in the amount of adsorbed N 2 with increase of the P/P 0 value and the appearance of a hysteresis loop are observed in the isotherm curves. Results obtained from BJH analysis are listed in Table 3. The average BJH pore width is in the range 4.5-5.5 nm, which indicates that these holes are mainly mesopores with total pore volume of 0.45 cm 3 g −1 . These features are fingerprints of a hierarchical mesoporous structure of LLNMO powders that promotes the Li + ion kinetics [59]. Suib et al. [60] described the formation of mesoporosity by aggregation of nanoparticles along their lateral faces. In the same way, the mesopores (~5 nm) correspond here to the voids existing between randomly-packed nanoparticles of~150 nm in diameter inside secondary particles (agglomerates) of~600 nm in size, as we will see on TEM images reported later in this work.
BET analysis shows that pristine and AlF 3 -coated LLMNO samples have a specific surface area of 11.5 and 9.2 m 2 g −1 , respectively, which is advantageous for enhancing the transport of Li + ions in the secondary particles. These values are comparable with those reported by Chang et al. [61] for Li-rich Li 1.27 Cr 0.2 Mn 0.53 O 2 oxides prepared by milling-assisted solid-state process. Xiang et al. [62] reported the influence of polyvinyl alcohol (PVA) as chelating agent on the BET surface area for Li 1.2 Ni 0.2 Mn 0.6 O 2 prepared by sol-gel technique (varying from~10 m 2 g −1 for PVA-0% to~33 m 2 g −1 for PVA-15%). Zhang et al. [63] showed that 3-µm sized secondary particles, which consisted of self-assembled crystallites~150 nm in size, demonstrated optimized mesostructuration partially mitigating the voltage fading. The equivalent particle size of the samples can be calculated from the BET data and compare with SEM images. The average particle diameter (in nm) is expressed by the relation [64]: where S BET is the specific surface area (in m 2 g −1 ) measured by BET experiments and d the gravimetric density (4.53 g cm −3 for NMO). The BET surface area and the average particle size results are found in Table 3. We observe a relatively good agreement between particle size values L BET and L TEM evaluated from BET and measured from TEM images, respectively. The crystal structure of Li-rich layered oxides is currently discussed assuming either the existence of solid solution according to the substitution of M in LiMO2 [65,66] or the formation of a composite of LiMO2 and Li2MnO3 [67,68]. The X-ray diffraction analysis clearly pleads for the second model. This hypothesis is also confirmed by Raman spectroscopy, which is efficient in probing the short-range order (local structure) of materials. Regarding the vibrational modes of Li1.2Ni0.2Mn0.6O2 (0.5Li2MnO3·0.5LiNi0.5Mn0.5O2 in layered notation), two Raman-active modes (A1g + Eg) are expected for LiMO2 (M = Ni or Mn, D3d 5 symmetry) [69,70], while six Raman active modes (4Ag + 2Bg) are predicted for Li2MnO3 (C2h 3 symmetry) [71]. In Figure 6, we observe two peaks at 603 and 482 cm −1 , which are assigned to the stretching ν(MO6) and bending δ(O-M-O) vibration with M = Mn 4+ , Ni 2+ belonging to the LiNi0.5Mn0.5O2 phase and three peaks at 587, 437, and 416 cm −1 attributed to the Li2MnO3 phase. The extra-peak at ca. 650 cm −1 is attributed to the stretching vibration ν(Mn 3+ O6). The calculated spectrum is shown in Figure 6 (black line). The Raman spectrum of the surface-modified LLNMO sample does not exhibit any frequency shift with respect to the pristine one. This indicates that the AlF3 coating does not influence the Li-rich local structure, and that neither Al nor F penetrated into the samples in agreement with the XRD analysis.

Morphology
The morphology of LLNMO powders has been investigated by SEM and HRTEM imaging. Figure 7 presents the SEM images combined with the EDX spectrum of the surface of AlF3-coated sample. The SEM images show regular particles with similar morphology (almost monodisperse). The homogeneous size distribution of agglomerates ~200 nm is noticeable without difference between pristine and AlF3-coated particles. The EDX spectrum (Figure 7c) of the AlF3-coated sample exhibits only the additional peak of Al, because the peak of fluorine merges with that of Mn. No  Table 3. Brunauer-Emmett-Teller (BET) specific surface area, calculated average particle size from BET data (L BET ) using Equation (2), mean particle size measured using TEM imaging (L TEM ) and results obtained from Barrett-Joyner-Halenda (BJH) analysis for hydrothermally prepared LLNMO samples. The crystal structure of Li-rich layered oxides is currently discussed assuming either the existence of solid solution according to the substitution of M in LiMO 2 [65,66] or the formation of a composite of LiMO 2 and Li 2 MnO 3 [67,68]. The X-ray diffraction analysis clearly pleads for the second model. This hypothesis is also confirmed by Raman spectroscopy, which is efficient in probing the short-range order (local structure) of materials. Regarding the vibrational modes of Li 1.2 Ni 0.2 Mn 0.6 O 2 (0.5Li 2 MnO 3· 0.5LiNi 0.5 Mn 0.5 O 2 in layered notation), two Raman-active modes (A 1g + E g ) are expected for LiMO 2 (M = Ni or Mn, D 3d 5 symmetry) [69,70], while six Raman active modes (4A g + 2B g ) are predicted for Li 2 MnO 3 (C 2h 3 symmetry) [71]. In Figure 6, The crystal structure of Li-rich layered oxides is currently discussed assuming either the existence of solid solution according to the substitution of M in LiMO2 [65,66] or the formation of a composite of LiMO2 and Li2MnO3 [67,68]. The X-ray diffraction analysis clearly pleads for the second model. This hypothesis is also confirmed by Raman spectroscopy, which is efficient in probing the short-range order (local structure) of materials. Regarding the vibrational modes of Li1.2Ni0.2Mn0.6O2 (0.5Li2MnO3·0.5LiNi0.5Mn0.5O2 in layered notation), two Raman-active modes (A1g + Eg) are expected for LiMO2 (M = Ni or Mn, D3d 5 symmetry) [69,70], while six Raman active modes (4Ag + 2Bg) are predicted for Li2MnO3 (C2h 3 symmetry) [71]. In Figure 6, we observe two peaks at 603 and 482 cm −1 , which are assigned to the stretching ν(MO6) and bending δ(O-M-O) vibration with M = Mn 4+ , Ni 2+ belonging to the LiNi0.5Mn0.5O2 phase and three peaks at 587, 437, and 416 cm −1 attributed to the Li2MnO3 phase. The extra-peak at ca. 650 cm −1 is attributed to the stretching vibration ν(Mn 3+ O6). The calculated spectrum is shown in Figure 6 (black line). The Raman spectrum of the surface-modified LLNMO sample does not exhibit any frequency shift with respect to the pristine one. This indicates that the AlF3 coating does not influence the Li-rich local structure, and that neither Al nor F penetrated into the samples in agreement with the XRD analysis.

Morphology
The morphology of LLNMO powders has been investigated by SEM and HRTEM imaging. Figure 7 presents the SEM images combined with the EDX spectrum of the surface of AlF3-coated sample. The SEM images show regular particles with similar morphology (almost monodisperse). The homogeneous size distribution of agglomerates ~200 nm is noticeable without difference between pristine and AlF3-coated particles. The EDX spectrum (Figure 7c) of the AlF3-coated sample exhibits only the additional peak of Al, because the peak of fluorine merges with that of Mn. No

Morphology
The morphology of LLNMO powders has been investigated by SEM and HRTEM imaging. Figure 7 presents the SEM images combined with the EDX spectrum of the surface of AlF 3 -coated sample. The SEM images show regular particles with similar morphology (almost monodisperse). The homogeneous size distribution of agglomerates~200 nm is noticeable without difference between pristine and AlF 3 -coated particles. The EDX spectrum (Figure 7c) of the AlF 3 -coated sample exhibits only the additional peak of Al, because the peak of fluorine merges with that of Mn. No other element was detected, giving evidence of the absence of any impurity in all the samples. From this experiment, the concentration of Al is found to be 5.15 wt.%.
Energies 2020, 13, x FOR PEER REVIEW 9 of 22 other element was detected, giving evidence of the absence of any impurity in all the samples. From this experiment, the concentration of Al is found to be 5.15 wt.%. The HRTEM images for all the samples are reported in Figure 8. The 200 nm-thick agglomerates of the Li1.2Ni0.2Mn0.6O2 samples are made of well-crystallized particles 95 nm in size. The comparison between HRTEM images shows that the coating had no effect on the size of particles. There is no effect on the shape, confirming that the coating of only 5-7 nm thickness does not change the structure nor the morphology of the powders [72,73]. This is due to the relatively low temperature of 400 °C used during the coating process, which is much smaller than the temperature at which the particles could start to coalesce. The distance between neighboring fringes shown in insets is 0.47 nm, which is assigned to (003), (001), or both planes for rhombohedral, monoclinic, or both phases, respectively. Such features are commonly observed in Li-rich samples of different compositions [74].

Electrochemical Properties
Electrochemical properties of LLNMO samples have been investigated by galvanostatic charge-discharge and electrochemical impedance spectroscopy. In this study, we consider the total charge delivered by each component of the LLNMO cathodes (xLi2MnO3·(1−x)LiNi0.5Mn0.5O2). As x = 0.5 for Li1.2Ni0.2Mn0.6O2, the theoretical specific capacity is 369 mAh g −1 (229 mAh g −1 for 0.5Li2MnO3 + 140 mAh g −1 for 0.5LiNi0.5Mn0.5O2). This value is exactly the theoretical capacity, based on complete extraction of lithium (1.2 Li per Li1.2Ni0.2Mn0.6O2 formula unit) [75][76][77]. The theoretical discharge capacity, based on the mass of the discharged rock-salt product after the first charge/discharge cycle, is 280 mAh g −1 . The comparison between HRTEM images shows that the coating had no effect on the size of particles. There is no effect on the shape, confirming that the coating of only 5-7 nm thickness does not change the structure nor the morphology of the powders [72,73]. This is due to the relatively low temperature of 400 • C used during the coating process, which is much smaller than the temperature at which the particles could start to coalesce. The distance between neighboring fringes shown in insets is 0.47 nm, which is assigned to (003), (001), or both planes for rhombohedral, monoclinic, or both phases, respectively. Such features are commonly observed in Li-rich samples of different compositions [74]. The comparison between HRTEM images shows that the coating had no effect on the size of particles. There is no effect on the shape, confirming that the coating of only 5-7 nm thickness does not change the structure nor the morphology of the powders [72,73]. This is due to the relatively low temperature of 400 °C used during the coating process, which is much smaller than the temperature at which the particles could start to coalesce. The distance between neighboring fringes shown in insets is 0.47 nm, which is assigned to (003), (001), or both planes for rhombohedral, monoclinic, or both phases, respectively. Such features are commonly observed in Li-rich samples of different compositions [74].

Electrochemical Properties
Electrochemical properties of LLNMO samples have been investigated by galvanostatic charge-discharge and electrochemical impedance spectroscopy. In this study, we consider the total

Electrochemical Properties
Electrochemical properties of LLNMO samples have been investigated by galvanostatic chargedischarge and electrochemical impedance spectroscopy. In this study, we consider the total charge delivered by each component of the LLNMO cathodes (xLi 2 MnO 3 ·(1−x)LiNi 0.5 Mn 0.5 O 2 ). As x = 0.5 for Li 1.2 Ni 0.2 Mn 0.6 O 2 , the theoretical specific capacity is 369 mAh g −1 (229 mAh g −1 for 0.5Li 2 MnO 3 + 140 mAh g −1 for 0.5LiNi 0.5 Mn 0.5 O 2 ). This value is exactly the theoretical capacity, based on complete extraction of lithium (1.2 Li per Li 1.2 Ni 0.2 Mn 0.6 O 2 formula unit) [75][76][77]. The theoretical discharge capacity, based on the mass of the discharged rock-salt product after the first charge/discharge cycle, is 280 mAh g −1 . Note that for unknown reasons, some authors used the value of 200 mAh g −1 .
For clarity, we use the current density expressed in mA g −1 with the correspondence 1C = 369 mA g −1 for all the electrodes. The capacity corresponding to the oxidation of Ni from Ni 2+ to Ni 4+ (below 4.5 V) is ca. 100 mAh g −1 for both electrodes. During the initial discharge, the pristine and coated LLNMO electrodes delivered a specific capacity of 242 and 260 mAh g −1 , respectively, which was maintained at 219 and 231 mAh g −1 after 50 cycles. After 100 cycles of charge-discharge at 600 mA g −1 current density, the capacity retention of~91% for the coated electrode is higher than that of pristine LLNMO (~59%).
Energies 2020, 13, x FOR PEER REVIEW 10 of 22 Note that for unknown reasons, some authors used the value of 200 mAh g −1 . For clarity, we use the current density expressed in mA g −1 with the correspondence 1C = 369 mA g −1 for all the electrodes. Figures 9a,b display the GCD curves of pristine and AlF3-coated Li1.2Ni0.2Mn0.6O2 electrodes, respectively, performed at 0.15C rate in the potential window 2.0-4.8 V. The shape and change after the first activation cycle are consistent with literature [9,41,44,50,[78][79][80][81][82]. The initial charge occurred with a well-defined voltage plateau at 4.5 V, which correspond to the irreversible extraction of Li2O from the Li2MnO3 component [8]. Thus, the strong oxidation of O 2-is considered to occur at ca. 4.5 V.
The capacity corresponding to the oxidation of Ni from Ni 2+ to Ni 4+ (below 4.5 V) is ca. 100 mAh g −1 for both electrodes. During the initial discharge, the pristine and coated LLNMO electrodes delivered a specific capacity of 242 and 260 mAh g −1 , respectively, which was maintained at 219 and 231 mAh g −1 after 50 cycles. After 100 cycles of charge-discharge at 600 mA g −1 current density, the capacity retention of ~91% for the coated electrode is higher than that of pristine LLNMO (~59%).  Figures 10a-c present the differential capacity or incremental capacity (IC) plots (−dQ/dV vs. V) for discharge curves of LLNMO electrodes at the 3rd and 50th cycles, which not only specify the reduction potentials but also clearly display the voltage fade issue. In IC plots, the peaks correspond to the pseudo plateaus in the GCD curves. The IC curve of the parent LiNi0.5Mn0.5O2 (Figure 10a) confirms that the peak at ca. 3.74 V is due to the reduction of Ni 4+ to Ni 3+ during the insertion of Li + ions in the layered framework. This reduction peak (A) occurred at 3.75 and 3.77 V in the discharge profile of pristine and coated electrodes, respectively. It is shifted down to ~3.65 V after 50 cycles for pristine Li1.2Ni0.2Mn0.6O2, whereas a shift by only 70 mV is observed for the AlF3-coated electrode. Note that for unknown reasons, some authors used the value of 200 mAh g −1 . For clarity, we use the current density expressed in mA g −1 with the correspondence 1C = 369 mA g −1 for all the electrodes. Figures 9a,b display the GCD curves of pristine and AlF3-coated Li1.2Ni0.2Mn0.6O2 electrodes, respectively, performed at 0.15C rate in the potential window 2.0-4.8 V. The shape and change after the first activation cycle are consistent with literature [9,41,44,50,[78][79][80][81][82]. The initial charge occurred with a well-defined voltage plateau at 4.5 V, which correspond to the irreversible extraction of Li2O from the Li2MnO3 component [8]. Thus, the strong oxidation of O 2-is considered to occur at ca. 4.5 V.
The capacity corresponding to the oxidation of Ni from Ni 2+ to Ni 4+ (below 4.5 V) is ca. 100 mAh g −1 for both electrodes. During the initial discharge, the pristine and coated LLNMO electrodes delivered a specific capacity of 242 and 260 mAh g −1 , respectively, which was maintained at 219 and 231 mAh g −1 after 50 cycles. After 100 cycles of charge-discharge at 600 mA g −1 current density, the capacity retention of ~91% for the coated electrode is higher than that of pristine LLNMO (~59%).  Figures 10a-c present the differential capacity or incremental capacity (IC) plots (−dQ/dV vs. V) for discharge curves of LLNMO electrodes at the 3rd and 50th cycles, which not only specify the reduction potentials but also clearly display the voltage fade issue. In IC plots, the peaks correspond to the pseudo plateaus in the GCD curves. The IC curve of the parent LiNi0.5Mn0.5O2 (Figure 10a) confirms that the peak at ca. 3.74 V is due to the reduction of Ni 4+ to Ni 3+ during the insertion of Li + ions in the layered framework. This reduction peak (A) occurred at 3.75 and 3.77 V in the discharge profile of pristine and coated electrodes, respectively. It is shifted down to ~3.65 V after 50 cycles for pristine Li1.2Ni0.2Mn0.6O2, whereas a shift by only 70 mV is observed for the AlF3-coated electrode. Figure 10a-c present the differential capacity or incremental capacity (IC) plots (−dQ/dV vs. V) for discharge curves of LLNMO electrodes at the 3rd and 50th cycles, which not only specify the reduction potentials but also clearly display the voltage fade issue. In IC plots, the peaks correspond to the  (Figure 10a) confirms that the peak at ca. 3.74 V is due to the reduction of Ni 4+ to Ni 3+ during the insertion of Li + ions in the layered framework. This reduction peak (A) occurred at 3.75 and 3.77 V in the discharge profile of pristine and coated electrodes, respectively. It is shifted down to~3.65 V after 50 cycles for pristine Li 1.2 Ni 0.2 Mn 0.6 O 2 , whereas a shift by only 70 mV is observed for the AlF 3 -coated electrode. The main difference between LiNi 0.5 Mn 0.5 O 2 and Li 1.2 Ni 0.2 Mn 0.6 O 2 is the occurrence of the low-voltage cathodic peak (B) at ca. 3.25-3.35 V, which is assigned to the reduction of Mn 4+ to Mn 3+ in MnO 2 formed during the first charging reaction [38,59,83]. Note that the reduction peak of Mn 4+ is weaker for AlF 3 -coated Li 1.2 Ni 0.2 Mn 0.6 O 2 than Li 1.2 Ni 0.2 Mn 0.6 O 2 ; however, the voltage decay of the B peak is higher than that of the A peak. Figure 11 presents the rate capability and the cycling stability of LLNMO electrodes synthesized by the hydrothermal method assisted with citric acid. Experiments were carried out in the voltage range 2.0-4.8 V at various current densities in the range 30-900 mA g −1 . As shown in Figure 11a, specific capacities of 116 and 153 mAh g −1 are obtained at the 900 mA g −1 rate during the discharge process of pristine and AlF 3 -coated electrodes, respectively. The decay in the specific capacity was observed without significant change in the S-shape of the charge and discharge curves. From the cycling performance shown in Figure 11b, the loss of capacity calculated upon 100 cycles is found to be 0.75 and 0.20 mAh g −1 per cycle for the pristine and AlF 3 -coated electrodes, respectively. The good reversibility of the AlF 3 -coated Li 1.2 Ni 0.2 Mn 0.6 O 2 electrode is confirmed by the Coulombic efficiency remaining around 99.4% over 100 cycles (Figure 11b). The rates of capacity fade for the AlF 3 -coated electrode are 0.29 and 0.41 mAh g −1 per cycle, when tested over 80 cycles at current densities of 300 and 900 mA g −1 , respectively (Figure 11c). The evolution of the Coulombic efficiency for pristine and AlF 3 -coated LLNMO electrodes is presented in Figure 11d. In comparison, the Coulombic efficiency of 76% for pristine LLNMO in the first cycle has increased to 87% after AlF 3 coating. Similar behavior was reported by Liu et al. [43] Figure 11 presents the rate capability and the cycling stability of LLNMO electrodes synthesized by the hydrothermal method assisted with citric acid. Experiments were carried out in the voltage range 2.0-4.8 V at various current densities in the range 30-900 mA g −1 . As shown in Figure 11a, specific capacities of 116 and 153 mAh g −1 are obtained at the 900 mA g −1 rate during the discharge process of pristine and AlF3-coated electrodes, respectively. The decay in the specific capacity was observed without significant change in the S-shape of the charge and discharge curves. From the cycling performance shown in Figure 11b, the loss of capacity calculated upon 100 cycles is found to be 0.75 and 0.20 mAh g −1 per cycle for the pristine and AlF3-coated electrodes, respectively. The good reversibility of the AlF3-coated Li1.2Ni0.2Mn0.6O2 electrode is confirmed by the Coulombic efficiency remaining around 99.4% over 100 cycles (Figure 11b). The rates of capacity fade for the AlF3-coated electrode are 0.29 and 0.41 mAh g −1 per cycle, when tested over 80 cycles at current densities of 300 and 900 mA g −1 , respectively (Figure 11c). The evolution of the Coulombic efficiency for pristine and AlF3-coated LLNMO electrodes is presented in Figure 11d. In comparison, the Coulombic efficiency of 76% for pristine LLNMO in the first cycle has increased to 87% after AlF3 coating. Similar behavior was reported by Liu et al. [43] for ZnAl2O4-coated Li1.2Ni0.2Mn0.6O2 material. The present electrochemical performances of LLNMO electrodes are comparable with previous data. However, there are some differences depending on the morphology of the active particles. Hydrothermally synthesized rod-like hierarchical nano/micro Li1.2Ni0.2Mn0.6O2 delivered a capacity of 212 mAh g −1 after 30 cycles at 1C rate (250 mA g −1 ) [36]. Yang et al. [32] reported a specific capacity of ~190 mAh g −1 at 1C rate (250 mA g −1 ) for agglomerated Li1.2Ni0.2Mn0.6O2 particles with a large pore volume of 9.2 × 10 −3 cm 3 g −1 prepared by sol-freeze-drying method. Li1.2Ni0.2Mn0.6O2 nanoparticles (100-200 nm in size) coated with a thin layer of Li + -conductive Li2ZrO3 coating obtained through a The present electrochemical performances of LLNMO electrodes are comparable with previous data. However, there are some differences depending on the morphology of the active particles. Hydrothermally synthesized rod-like hierarchical nano/micro Li 1.2 Ni 0.2 Mn 0.6 O 2 delivered a capacity of 212 mAh g −1 after 30 cycles at 1C rate (250 mA g −1 ) [36]. Yang et al. [32] reported a specific capacity of 190 mAh g −1 at 1C rate (250 mA g −1 ) for agglomerated Li 1.2 Ni 0.2 Mn 0.6 O 2 particles with a large pore volume of 9.2 × 10 −3 cm 3 g −1 prepared by sol-freeze-drying method. Li 1.2 Ni 0.2 Mn 0.6 O 2 nanoparticles (100-200 nm in size) coated with a thin layer of Li + -conductive Li 2 ZrO 3 coating obtained through a synchronous lithiation strategy delivered a capacity of circa 169 mAh g −1 over 100 cycles at current density 200 mA g −1 [13].
To further investigate the effect of the AlF 3 coating on resistances and lithium kinetics in LLNMO, EIS measurements were carried out. Figure 12 presents the EIS spectra investigated after the first cycle performed at 0.15C rate. As shown in Figure 12a, all of the Nyquist plots of the cells are composed of three contributions that can be modeled using the equivalent circuit shown as inset: (i) the intercept of x-axis at high-frequency corresponds to the ohmic resistance (R s ), i.e., electrodes and electrolyte contributions; (ii) a depressed semicircle in the high-frequency region is ascribed to the charge transfer impedance (R ct , CPE ct ) at the electrode/electrolyte interface, in which the constant phase element is expressed by CPE = 1/T((jω) p ) with ω the angular frequency, T a CPE constant and p an exponent (0 ≤ p ≤ 1, p = 1 for pure capacitance) and j is the imaginary number (j = √ -1); and finally (iii) an inclined line (with a slope of~45 • ) at low-frequency reflects the diffusion-controlled process characterized by a Warburg impedance taken in the infinite limit and expressed by the relation Z W (ω) = σ w (1 − j) ω −1/2 , where σ w is the Warburg factor [84]. The real part of the impedance Z'(ω) is the sum of the components: From the slope of the Z' vs. ω −1/2 curves (Figure 12b), we can calculate the apparent chemical diffusion coefficient of Li + ions (D Li +) in the electrode using the relation [85]: where R is the gas constant, F is Faraday's constant, T the absolute temperature, A the surface area of the electrode-electrolyte interface, and C Li the molar concentration of Li + ions in the electrode (0.8 mmol cm −3 ). The fitting parameters are reported in Table 4. Results can be summarized as follows. The value of R ct for the cell with AlF 3 -coated Li 1.2 Ni 0.2 Mn 0.6 O 2 (65.5 Ω) is smaller than that of pristine Li 1.2 Ni 0.2 Mn 0.6 O 2 (100.3 Ω), demonstrating that the AlF 3 coating layer effectively decreases the interface contact resistance. In addition, the exponent p of the CPE component associated with the charge transfer is close to unity for both cells, which means that the CPE contribution is essentially capacitive. This gives evidence of a good homogeneity of the electrode. Note that the contribution of the SEI layer can be neglected. As a result, the apparent diffusion coefficients of Li + ions are found to be 2. synchronous lithiation strategy delivered a capacity of circa 169 mAh g −1 over 100 cycles at current density 200 mA g −1 [13].
To further investigate the effect of the AlF3 coating on resistances and lithium kinetics in LLNMO, EIS measurements were carried out. Figure 12 presents the EIS spectra investigated after the first cycle performed at 0.15C rate. As shown in Figure 12a, all of the Nyquist plots of the cells are composed of three contributions that can be modeled using the equivalent circuit shown as inset: (i) the intercept of x-axis at high-frequency corresponds to the ohmic resistance (Rs), i.e., electrodes and electrolyte contributions; (ii) a depressed semicircle in the high-frequency region is ascribed to the charge transfer impedance (Rct, CPEct) at the electrode/electrolyte interface, in which the constant phase element is expressed by CPE = 1/T((jω) p ) with ω the angular frequency, T a CPE constant and p an exponent (0 ≤ p ≤ 1, p = 1 for pure capacitance) and j is the imaginary number (j = √-1); and finally (iii) an inclined line (with a slope of ~45°) at low-frequency reflects the diffusion-controlled process characterized by a Warburg impedance taken in the infinite limit and expressed by the relation ZW(ω) = σw (1 − j) ω −1/2 , where σw is the Warburg factor [84]. The real part of the impedance Z'(ω) is the sum of the components: From the slope of the Z' vs. ω −1/2 curves (Figure 12b), we can calculate the apparent chemical diffusion coefficient of Li + ions (DLi+) in the electrode using the relation [85]: where R is the gas constant, F is Faraday's constant, T the absolute temperature, A the surface area of the electrode-electrolyte interface, and CLi the molar concentration of Li + ions in the electrode (0.8 mmol cm −3 ). The fitting parameters are reported in Table 4. Results can be summarized as follows. The value of Rct for the cell with AlF3-coated Li1.2Ni0.2Mn0.6O2 (65.5 Ω) is smaller than that of pristine Li1.2Ni0.2Mn0.6O2 (100.3 Ω), demonstrating that the AlF3 coating layer effectively decreases the interface contact resistance. In addition, the exponent p of the CPE component associated with the charge transfer is close to unity for both cells, which means that the CPE contribution is essentially capacitive. This gives evidence of a good homogeneity of the electrode. Note that the contribution of the SEI layer can be neglected. As a result, the apparent diffusion coefficients of Li + ions are found to be 2.8 × 10 −15 and 1.1 × 10 −14 cm 2 s −1 in pristine and AlF3-coated Li1.2Ni0.2Mn0.6O2 electrodes, respectively. The DLi + calculated values of our hydrothermally synthesized LLNMO samples are slightly higher than the values reported from Yang et al. [39] and Lin et al. [44]. DLi + of 2.34 × 10 −17 and 1.03 × 10 −14 cm 2 s −1 was reported for pristine LLNMO and LLNMO/graphene structures synthesized by The D Li + calculated values of our hydrothermally synthesized LLNMO samples are slightly higher than the values reported from Yang et al. [39] and Lin et al. [44]. D Li + of 2.34 × 10 −17 and 1.03 × 10 −14 cm 2 s −1 was reported for pristine LLNMO and LLNMO/graphene structures synthesized by sonication in ethylene glycol solution [39]. D Li + measured for LLNMO prepared by a sol-gel method assisted with carbon felt as reactive carrier was 2.49 × 10 −16 cm 2 s −1 before charging, which is slightly smaller than 4.71 × 10 −16 cm 2 s −1 obtained under charge at 4.8 V [44]. In contrast, Sasakawa et al. [86] reported a higher D Li + value of 1.6 × 10 −13 cm 2 s −1 in the discharge state of LLNMO particles prepared by combustion method. The same group of researchers showed that D Li increased with increasing the charge rate in the potential range 2.5-4.4 V with an average value of 4.5 × 10 −13 cm 2 s −1 at mid-voltage. Generally, a high D Li + value implies a well-ordered layer structure of the cathode material and a minimum cationic disorder in the Li slabs [87]. Such an Li + diffusion enhancement is attributed to the presence of the thin AlF 3 layer, which promotes the ionic transport at the electrode-electrolyte interface and partly suppresses the spinel layer formation. Liu et al. [29] reported D Li values of 8.  [88] proposed a new Li diffusion model based on the correlation between asymmetric TM migration and Li mobility that explained the non-equivalent intercalation and de-intercalation kinetics. Chang et al. [61] suggested that the high Li-ion diffusion coefficient of 3.9 × 10 −10 ·cm 2 ·s −1 in Li-rich layered cathode material is caused by the mesoporous network composed of tiny primary particles.
To further verify the improvement of the electrochemical properties of the electrodes, EIS measurements were conducted at different discharge cycles. Figure 13a,b presents the variation of the transport parameters R s , R ct , and σ w for both LLNMO samples. Among those parameters listed, R ct is the most important since it represents the charge transfer within the cathode material. After 50 charge-discharge cycles, R ct increased by a factor of 2 for both electrodes but seems to saturate at ca. 145 Ω for the AlF 3 -coated LLNMO material. These results show that σ w for pristine LLNMO increases almost linearly upon cycling, while it reaches a value of 50 Ω s −1/2 for the AlF 3 coated electrode after 30 cycles.
Energies 2020, 13, x FOR PEER REVIEW 13 of 22 sonication in ethylene glycol solution [39]. DLi + measured for LLNMO prepared by a sol-gel method assisted with carbon felt as reactive carrier was 2.49 × 10 −16 cm 2 s −1 before charging, which is slightly smaller than 4.71 × 10 −16 cm 2 s −1 obtained under charge at 4.8 V [44]. In contrast, Sasakawa et al. [86] reported a higher DLi + value of 1.6 × 10 −13 cm 2 s −1 in the discharge state of LLNMO particles prepared by combustion method. The same group of researchers showed that DLi increased with increasing the charge rate in the potential range 2.5-4.4 V with an average value of 4.5 × 10 −13 cm 2 s −1 at mid-voltage. Generally, a high DLi + value implies a well-ordered layer structure of the cathode material and a minimum cationic disorder in the Li slabs [87]. Such an Li + diffusion enhancement is attributed to the presence of the thin AlF3 layer, which promotes the ionic transport at the electrode-electrolyte interface and partly suppresses the spinel layer formation. Liu et al. [29] reported DLi values of 8.2 × 10 −13 and 3.8 × 10 −12 cm 2 s −1 for pristine and Li1.3Al0.3Ti1.7(PO4)3 (LATP) coated Li1.2Ni0.2Mn0.6O2, respectively. The samples were synthesized by co-precipitation with the LATP layer ~4 nm thick and exhibited a cationic mixing of 2%. Liu et al. [43] estimated DLi values of 8.2 × 10 −14 and 7.6 × 10 −12 cm 2 s −1 at mid-voltage (3.5 V) for pristine and Li1.2Ni0.2Mn0.6O2 coated with 1 wt.% ZnAl2O4. Recently, Ku et al. [88] proposed a new Li diffusion model based on the correlation between asymmetric TM migration and Li mobility that explained the non-equivalent intercalation and de-intercalation kinetics. Chang et al. [61] suggested that the high Li-ion diffusion coefficient of 3.9 × 10 −10 ·cm 2 ·s −1 in Li-rich layered cathode material is caused by the mesoporous network composed of tiny primary particles.
To further verify the improvement of the electrochemical properties of the electrodes, EIS measurements were conducted at different discharge cycles. Figure 13a-b presents the variation of the transport parameters Rs, Rct, and σw for both LLNMO samples. Among those parameters listed, Rct is the most important since it represents the charge transfer within the cathode material. After 50 charge-discharge cycles, Rct increased by a factor of 2 for both electrodes but seems to saturate at ca.
145 Ω for the AlF3-coated LLNMO material. These results show that σw for pristine LLNMO increases almost linearly upon cycling, while it reaches a value of 50 Ω s −1/2 for the AlF3 coated electrode after 30 cycles. Raman spectroscopy is a non-destructive and sensitive method to perform post-mortem studies of cycled cathode materials [89]. This analytical technique allows the investigations of structural changes at the particle surface, the electrode degradation, and the formation of spinel-like layer Raman spectroscopy is a non-destructive and sensitive method to perform post-mortem studies of cycled cathode materials [89]. This analytical technique allows the investigations of structural changes at the particle surface, the electrode degradation, and the formation of spinel-like layer [28,80,83,90]. In which case, such surface modifications are not detected by diffraction techniques. Figure 14 shows the ex situ Raman spectra of cycled pristine and AlF 3 -coated LLNMO electrodes after 50 and 100 cycles. The formation of the spinel-like layer in all cycled LLNMO electrodes is indicated by the shoulder at 650 cm −1 . After long-term cycling (100 cycles) of the pristine LLNMO sample, the intensity of this spectral feature becomes relatively strong, revealing the formation of a disorder on the particle surface. Simultaneously, a broadening and a frequency shift of the peak at 416 cm −1 , related to the monoclinic Li 2 MnO 3 phase, are observed. In contrast, the AlF 3 -coating retards the transformation to the spinel structure of the Li 1.2 Ni 0.2 Mn 0.6 O 2 cathode surface. These studies confirm that the AlF 3 coating is a very efficient surface modification to improve the stability of the layered phase of the Li-rich material, at the origin of the significant improvement of the electrochemical properties.
Energies 2020, 13, x FOR PEER REVIEW 14 of 22 [28,80,83,90]. In which case, such surface modifications are not detected by diffraction techniques. Figure 14 shows the ex situ Raman spectra of cycled pristine and AlF3-coated LLNMO electrodes after 50 and 100 cycles. The formation of the spinel-like layer in all cycled LLNMO electrodes is indicated by the shoulder at ~650 cm −1 . After long-term cycling (100 cycles) of the pristine LLNMO sample, the intensity of this spectral feature becomes relatively strong, revealing the formation of a disorder on the particle surface. Simultaneously, a broadening and a frequency shift of the peak at 416 cm −1 , related to the monoclinic Li2MnO3 phase, are observed. In contrast, the AlF3-coating retards the transformation to the spinel structure of the Li1.2Ni0.2Mn0.6O2 cathode surface. These studies confirm that the AlF3 coating is a very efficient surface modification to improve the stability of the layered phase of the Li-rich material, at the origin of the significant improvement of the electrochemical properties.

Structural Stability
The structural and electrochemical stability of LLNMO strongly depend on synthesis conditions and operating parameters. The stable performance in relation to initial charge condition shows that a fast rate of 3C induced a decrease of the Rct resistance and slight increase of the Li + diffusion coefficient in the bulk. Our results confirm the efficiency of the AlF3 coating with the optimized concentration of 5 wt.% [86]. This layer is thick enough to protect the electrode material against the side reactions with the electrolyte, which reduces the resistance of the SEI layer, and increases the rate capability. The coating layer also stabilizes the layered network, with two beneficial effects: (i) it increases the cycle ability, which improves the cycle ability and (ii) it increases the diffusivity of Li + , which contributes to enhancing the rate capability of the coated particles. In addition, the mesopores increase the effective surface area in contact with the electrolyte, which explains the high capacities reported in this work. Improvement of calendar and cycling life together with safety issues of lithium-ion batteries require sophisticated technology to prevent degradation mechanisms. The surface modification of the electrode materials is part of the solution [91]. Among the unwanted degradation mechanisms occurring in LLNMOs, the creation of the solid electrolyte interphase (SEI), growth of a layered-to-spinel surface layer, unreacted residual lithium ingredients on the particle surface, formation of corrosion pits, and occurrence of inter-granular cracks inside primary particles have been widely documented [92][93][94][95][96][97][98].
To date, few works report the use of AlF3 as a deposit on the surface of Li-rich Mn-rich particles [48,71,[98][99][100]. Sun et al. [71] consider the AlF3 coating layer as a "buffer" through which the activity

Structural Stability
The structural and electrochemical stability of LLNMO strongly depend on synthesis conditions and operating parameters. The stable performance in relation to initial charge condition shows that a fast rate of 3C induced a decrease of the R ct resistance and slight increase of the Li + diffusion coefficient in the bulk. Our results confirm the efficiency of the AlF 3 coating with the optimized concentration of 5 wt.% [86]. This layer is thick enough to protect the electrode material against the side reactions with the electrolyte, which reduces the resistance of the SEI layer, and increases the rate capability. The coating layer also stabilizes the layered network, with two beneficial effects: (i) it increases the cycle ability, which improves the cycle ability and (ii) it increases the diffusivity of Li + , which contributes to enhancing the rate capability of the coated particles. In addition, the mesopores increase the effective surface area in contact with the electrolyte, which explains the high capacities reported in this work. Improvement of calendar and cycling life together with safety issues of lithium-ion batteries require sophisticated technology to prevent degradation mechanisms. The surface modification of the electrode materials is part of the solution [91]. Among the unwanted degradation mechanisms occurring in LLNMOs, the creation of the solid electrolyte interphase (SEI), growth of a layered-to-spinel surface layer, unreacted residual lithium ingredients on the particle surface, formation of corrosion pits, and occurrence of inter-granular cracks inside primary particles have been widely documented [92][93][94][95][96][97][98].
To date, few works report the use of AlF 3 as a deposit on the surface of Li-rich Mn-rich particles [48,71,[98][99][100]. Sun et al. [71] consider the AlF 3 coating layer as a "buffer" through which the activity of extracted oxygen is reduced; thus, the electrolyte decomposition is avoided at voltages above 4.5 V. AlF 3 is preferable to Al 2 O 3 , which is unstable in Li cells and can be peeled off from the cathode surface by conversion into AlF 3 in the presence of traces of HF in the electrolyte. AlF 3 is also preferable to Li 3 PO 4 . Kang and Thackeray reported a surface coating of Li-Ni-PO 4 , leading to reversible capacity of 200 mAh g −1 at 1C. They speculated that one of the components Li 3 PO 4 acts as an excellent lithium ion conductor as well as an efficient protective layer stabilizing the electrode surface, though it is not clear how Li 3 PO 4 stabilizes the electrode surface [101]. Then, Liu et al. used this coating to stabilize the interface and reduce the side reaction between the material and electrolyte and thus improve the cyclic stability of a Li-rich, Mn-containing layered material [101]. Lee et al. [102] synthesized a Li 3 PO 4 -coated Li 1.2 Mn 0.6 Ni 0.2 O 2 , which delivered 240 mAh g −1 at 0.1C (25 mA g −1 ), and 205 mAh g −1 after 50 cycles at room temperature, which corresponds to 85% capacity retention over 50 cycles, compared to~91% capacity retention after 100 cycles at 2C in our AlF 3 -coated Li 1.2 Ni 0.2 Mn 0.6 O 2 electrode. More recently, Li-rich@Li 3 PO 4 was synthesized via a template of polydopamine [103]. The nanoparticles were 100-200 nm thick, comparable to the particles in the present work (secondary particles 200 nm, primary particles 95 nm). The beneficial protection effects against side-reactions with the electrolyte were confirmed. Actually, almost all coatings that are chemically and electrochemically stable with the carbonate electrolytes (including Li 3 PO 4 and AlF 3 ) have this function (see [104] for a review). However, the results were not competitive with the results obtained with other coatings such as AlF 3 . For instance, at 0.2C (1C = 200 mA g −1 ) in a potential window from 2.0 to 4.8 V (the same conditions as in the present work), Li-rich@Li 3 PO 4 delivered a capacity of 230 mAh g −1 , which decreased to 130 mAh g −1 after 100 cycles. Therefore, the Li 3 PO 4 coat fails to reduce the kinetics of the phase transition to the spinel phase or Mn dissolution in the surface layer of Li 1 [98] analyzed the fundamental functions of the AlF 3 coating and concluded that the coating mitigates the acidic attack by the electrolyte of spinel-like networks formed on the particle surface. Figure 15 illustrates the transformation mechanism of an active particle of Li-rich layered oxide. Upon repeated cycling, the morphology of the uncoated particle is modified with the formation a spinel phase and corrosion pits, whereas a thin layer of coating protects the particles against corrosion and postpones the spinel growth. The efficiency of the AlF 3 coating is also evidenced by the fact that our results were obtained with only 5-wt.% AlF 3 , while effective protection of cathode elements often need much thicker coats; for instance, 12 wt.% in the case of LiAlO 2 /Al 2 O 3 -coated nano-LiCoO 2 [107].  [98] analyzed the fundamental functions of the AlF3 coating and concluded that the coating mitigates the acidic attack by the electrolyte of spinel-like networks formed on the particle surface. Figure 15 illustrates the transformation mechanism of an active particle of Li-rich layered oxide. Upon repeated cycling, the morphology of the uncoated particle is modified with the formation a spinel phase and corrosion pits, whereas a thin layer of coating protects the particles against corrosion and postpones the spinel growth. The efficiency of the AlF3 coating is also evidenced by the fact that our results were obtained with only 5-wt.% AlF3, while effective protection of cathode elements often need much thicker coats; for instance, 12 wt.% in the case of LiAlO2/Al2O3-coated nano-LiCoO2 [107]. Figure 15. Scheme of transformation of active particle of Li-rich layered oxide upon repeated cycling. Reproduced with permission from [97]. Copyright 2014, American Chemical Society. Figure 15. Scheme of transformation of active particle of Li-rich layered oxide upon repeated cycling. Reproduced with permission from [97]. Copyright 2014, American Chemical Society.

Voltage Decay
Capacity fading and voltage decay are the important drawbacks of Li-rich cathode materials; they are the main challenges for the commercialization of LLNMOs. The voltage decay originates from a gradual phase transformation of the layered phase to a 3 V spinel-like phase. [13,76,[108][109][110][111][112][113]. From GCD curves and dQ/dV plots, Lee and Manthiram showed that the length of the plateau region during first charge is the primary governing factor in the voltage decay process [111]. From first principles calculations, Xu et al. [112] have suggested that both the formation of Li tet -V(TM)Li-Li tet dumbbell structures (Li tet refers to the lithium ion in a tetrahedral site and V(TM)Li refers to the lithium-ion vacancy in the TM layer) and TM ion migration into the lithium layer can facilitate the formation of a 3 V spinel-like phase during charging in the high voltage region. The formation of such Li tet -V(TM)Li-Li tet dumbbell structures can be expected to increase with the length of the plateau region during the first charge process since lithium ions from the TM layer will be extracted once all of the lithium ions are extracted from the lithium layer.
From IC plots ( Figure 10) the voltage decay is found to be 2.80 and 1.40 mV per cycle for pristine and AlF 3 -coated electrode, respectively, after 50 cycles. The sophisticated modification made by a three-in-one surface treatment consisted of synchronous design of oxygen vacancies, spinel-phase integration, and N-doped carbon monolayers. The remarkable results by Zheng et al. deserve to be cited [11]. The voltage fade was mitigated by improving the atomic level uniformity of elemental distribution at the Li 1.2 Ni 0.2 Mn 0.6 O 2 surface. Studies of the influence of the synthesis methods (i.e., sol-gel (SG), co-precipitation (CP), and hydrothermal (HT)) on the voltage-decay rate were carried out using secondary ion mass spectroscopy (SIMS), and it was suggested that one of the main reasons for the high voltage-decay rate in HT samples is the non-uniform distribution of Ni ions with a Ni-rich particle surface leading to an easy reduction of the Mn 4+ ions. Ni-segregated materials also exhibit poor cycling stability.

Conclusions
In summary, this work has shown that significant improvements of the electrochemical performance of Li 1.2 Ni 0.2 Mn 0.6 O 2 electrode materials synthesized by hydrothermal technique have been obtained by an optimized AlF 3 coating layer (5 wt.%) deposited on the surface of the 200-nm thick particles. This coating improved all aspects of the electrochemical properties: discharge capacity, capacity retention (~91% over 100 cycles for the coated electrode in contrast with~59% for the pristine one), and rate capability. The mesoporous morphology of hydrothermally synthesized Li 1.2 Ni 0.2 Mn 0.6 O 2 electrode materials contributed to enhancing the capacity. Another advantage of the AlF 3 coating is the smaller loss of the operating voltage, half that of the pristine sample. Finally, despite certain claims in the literature, these studies have demonstrated the effectiveness of this coating in terms of structural stability, migration of Mn 3+ cations, attenuation of cross growth in the spinel phase, and protection against side reactions with the electrolyte. Note the results have been obtained by coating pristine Li 1.2 Ni 0.2 Mn 0.6 O 2 . The next step should then be to coat doped-Li 1.2 Ni 0.2 Mn 0.6 O 2 using the doping processes mentioned in the introduction, which improved the rate capability by increasing the diffusion coefficient of Li + ions in the structure and mitigated the oxidation of O 2ions, in order to benefit from the synergetic effects.

Conflicts of Interest:
The authors declare no conflict of interest.