Functional behavior of AlF₃ coatings for high-performance cathode materials for lithium-ion batteries

1.   Introduction

The research on optimized electrode materials for Li-ion batteries aim at enhancing the energy density and the power density, improve the calendar and the cycling life, without compromising the safety of the cells ^[1]. In spite of these efforts, practical applications are still hindered by severe capacity fading during cycling and poor rate capability due partly to side reaction at the electrolyte–electrode interfaces (EEI). During long-term cycling, several unwanted degradation mechanisms occur (Figure 1a) such as oxidation of the electrolyte at high potentials, formation and growth of the solid electrolyte interphase (SEI), transition-metal dissolution, growth of a layer-to-spinel surface layer, unreacted residual lithium ingredients on the particle surface (i.e., LiOH, Li₂CO₃, which are dominant with Ni-rich oxides), formation of corrosion pits and occurrence of inter-granular cracks inside primary particles and severe etching/corrosion by the acidic species in the electrolyte (HF attack) ^{[2,3,4,5,6,7,8]}. Note that the formation of HF originates from the reaction of the lithium salt LiPF₆ used in common electrolyte of Li-ion cells and trace amounts of water (or alcohols) described by ^[9]:

The existence of water, even in very small concentration, generates HF and causes breakdown of the electrolyte. If decomposition of the electrode material occurs by HF attack, byproducts are inevitably gathered. During extensive cycling, the HF attack on LiCoO₂ electrode, for instance, can generate byproducts from Co on the surface of active particles. Then, these byproducts having Co–F bonding can be detected by spectroscopy assuming the relations ^[10]:

These byproducts would adhere at the surface of the separator and enhance the internal resistance of the Li//LiCoO₂ cell.

Several strategies have been directed to ameliorate the cycling performance of cathodes. Four representative tactics are currently used: (ⅰ) surface coating/modification ^[11,12], partial substitution of cations ^[13], partial substitution of oxygen anion ^[14] and utilization of electrolyte additives ^[15]. Moreover, some works have successfully associated doping and coating to enhance the electrode performance ^[16,17]. In most cases, surface engineering was realized by the deposition of a thin protective layer that covers uniformly the active particles (Figure 1b). Optimizing the surface region of a layered oxide is of prime importance because the outer area of a particle experiences the Li-ion path for insertion/deinsertion reaction. Hashem et al. ^[18] demonstrated that one function of coating is the crystallization of the thin disordered layer on the surface of LiNi_1/3Co_1/3Mn_1/3O₂, which improves the electrochemical properties. Other functions of the coating layer can be described considering two mechanisms: (ⅰ) the coat acts as a barrier to the electrolyte infiltration effect that reduced the electrolyte-electrode reactions and (ⅱ) the coating works as an HF scavenger. Numerous oxide materials used as protective coatings have proved to maintain the stability of the electrode–electrolyte interface in Li-ion batteries ^[11,12].

Electrode materials have been successfully coated with metal fluorides, i.e., LaF₃, MgF₂, ZrF₄, CaF₂, SrF₂, YF₃, AlF₃ ^{[19,20,21,22,23,24,25,26,27,28]}, and metal oxyfluorides, i.e., ZrO_xF_y, BiOF and YOF ^[29,30,31]. Generally, fluoride coatings demonstrate higher stability than most metal-oxide coats and potent erosion resistance against the HF in the electrolyte. Lee and Park ^[22] showed that a MgF₂ coating layer deposited by co-precipitation technique onto LiCoO₂ thin-film electrode reduces the dissolution of Co and the surface damage during cycling at 45 ℃. The 1 wt% MgF₂ coating of LiCoO₂ by chemical deposition improves the electrochemical performance with was optimized of MgF₂ but fails to increase significantly the thermal stability ^[32]. The crystalline 2 mol% SrF₂ coating on LiMn_1/3Ni_1/3Co_1/3O₂ (NMC333) powders suppresses the increase of the charge-transfer resistance during cycling over the voltage window 2.5–4.6 V vs. Li⁺/Li ^[26]. Coating NMC333 with LiF eliminates the increase in the charge transfer resistance that occurs for long cycling and during storage in the charged state ^[33]. Coating LiCoO₂ particles with ZrO_xF_y by a chemical deposition method leads to similar results, in which case the ZrO_xF_y deposit was obtained via adding NH₄F in Zr⁴⁺ aqueous solution with a controlled Zr:F ratio of 1:4 ^[29].

Note, however, that the beneficial effect of the layer coating is strongly dependent on the nature and morphology of the coating material. As the MF₃ fluorides are more stable than most of the other compounds under the corrosion of HF, the main effect of the coating with fluorides is an improvement of the electrochemical performance owing to the decrease of the dissolution of the active electrode material and protection against HF attack. Indeed, LaF₃ coating has significantly improved both the rate capability and the cycling performance of LiCoO₂ ^[19] and LiMn₂O₄ ^[34] as well. Another good reason to use AlF₃ as a buffer layer is its excellent protection for the aluminum collector from corrosion by the conventional electrolyte including LiPF₆ as the lithium salt. Moreover, Al is a light and cheap element. These features explain the superiority of AlF₃ and its success over the other metal fluorides as a protective coat of the electrode materials.

Metal fluorides can be used in lithium batteries in many ways. On a historical point of view, they have first been studied already in the 1960s for application as electrodes in primary lithium batteries, because they act as conversion materials, forming lithium fluoride during discharge, leading to large capacities and energy densities reported in different reviews ^[35,36]. However, the huge change of volume associated to the conversion reaction and the fact that the reaction is only partially reversible prevent their practical use as electrodes for secondary (i.e., rechargeable) batteries. Some recent progress has been made by mixing nanoparticles of metal fluorides such as BiF₃, FeF₂, FeF₃ ^[37,38] and AlF₃ ^[39] with carbon to buffer the product of the reaction. However, LiF is a product of the conversion reaction, which is one of the most insulating material on earth, which results in large overpotentials and high resistance, so that the electrode works only at very low C-rates, i.e., at very small currents. For example, the AlF₃–carbon composite only works at 0.1C, and still only 4% of the initial capacity is maintained after 10 cycles ^[39], precluding the practical use of this material as an electrode in the rechargeable lithium-batteries considered here. In practice AlF₃ is suitable only for electrode protection because it is rather inert and the Al³⁺-ion cannot be reduced or oxidized in battery conditions ^[11].

Hereunder, we review the different effects of AlF₃ coating, namely (1) electrode protection agent, (2) safety protection medium, (3) Al corrosion inhibitor, and (4) HF scavenger, and the resulting improvement of the electrochemical performance depending on the cathode material. After a brief description of the AlF₃ phase including synthesis methods and transport properties, we consider the different AlF₃-coated materials used in Li-ion batteries.

2.   Aluminum fluoride coating material

2.1.   Structural properties of AlF₃

Aluminum fluoride possesses five known varieties, i.e., α-, β-, η-, κ-, τ-AlF₃ ^[40], out of which α-AlF₃ is the most thermodynamically stable phase (perovskite-like form). It crystallizes with a rhombohedral structure (space group R-3c) and is thermodynamically stable at room temperature. The lattice has two equivalent stable interstitial sites per primitive cell, which can be occupied by lithium ions; they are located at the center of each distorted cube of α-AlF₃ at (${}^{1}\!\!\diagup\!\!{}_{4}\;,{}^{1}\!\!\diagup\!\!{}_{4}\;,{}^{1}\!\!\diagup\!\!{}_{4}\;$) and (${}^{3}\!\!\diagup\!\!{}_{4}\;,{}^{3}\!\!\diagup\!\!{}_{4}\;,{}^{3}\!\!\diagup\!\!{}_{4}\;$) coordinates. The β-AlF₃ metastable phase crystallizes with an orthorhombic structure (Cmcm space group). α-AlF₃ is a dielectric compound with a low refractive index, wide bandgap ~10.8 eV and strong Lewis acidity ^[41]. Due to its interesting optical, electrical and magnetic properties, a metal fluoride such as AlF₃ is an interesting compound for a wide variety of applications including optical designs, ionic conductors, protective coatings for electrodes in Li-ion batteries, heterogeneous catalyst and ferroelectric components ^[42,43,44]. With a low Gibbs free energy of formation, the advantage of AlF₃ against the amphoteric Al₂O₃ coating has been demonstrated from the exposure to trace of HF ^[45]. Therefore, a non-negligible advantage is that the AlF₃ coating layer appears to be a good Li⁺-ion conductor after lithiation. Interestingly, LiPF₆ is the lithium salt actually used in the commercial Li-ion batteries. LiPF₆ is chosen not only because the very good ionic conductivity, but also due to its ability to form an AlF₃ layer protecting the aluminum current collector against electrolyte corrosion. The efficient effect of passivation of the current collector surface by AlF₃ coat is known since a long time ^[46,47].

The corrosion of Al current collector (CC) foil has been recently revisited by Ma et al. ^[48]. The natural Al₂O₃ passivation layer promotes the deprotonation of the cationic radical of ethylene carbonate (EC⁺), which provokes the decomposition of PF₆^- anion and forms a proton-resistive AlF₃ layer on the Al surface. XPS measurements of the binding energy show that a CC foil anodized at 3.9 V for 20 h displays the conversion of the Al₂O₃ layer (75.6 eV) to stable AlF₃ passivation film (77.2 eV). Methyl difluoroacetate is used as fluorinated solvent that facilitates the formation of AlF₃ to passivate the surface of Al foil ^[49].

2.2.   AlF₃ deposition methods

Various methods have been utilized to prepare AlF₃-coated lithiated materials: (a) chemical deposition or coprecipitation process ^[45], (b) atomic layer deposition ^{[50,51,52,53]}, and (c) solid-state reaction ^[54]. The most popular technique used for AlF₃-coating is the chemical precipitation route. In this process, aluminum nitrate dehydrate, Al(NO₃)₃ and ammonium fluoride, NH₄F, are separately dissolved in distilled water as precursor solutions. The pristine powders are immersed in Al(NO₃)₃ dilute solution, whose pH is adjusted at value of 7.0 by adding ammonia solution, heated to 50 ℃ and stirred vigorously. NH₄F dilute solution is then added into the solution drop by drop. The molar ratio of Al to F is controlled to be 1:3, and the various amounts of AlF₃ are 2–15 wt% for coated Li-rich layered oxides. The mixed solution containing the cathode powders is stirred at 80 ℃ for 5 h to ensure the slow evaporation of solvent, and then dried at 60 ℃ in a vacuum oven. The final coated powders are obtained by heat treatment at 400 ℃ under flow of argon (or nitrogen) for 4 h to prevent the formation of Al₂O₃ and secure the formation of AlF₃-coated Li-rich oxide. The overall reaction for the AlF₃ deposit can be expressed by:

which occurs in the solution reaction at room temperature, and the departure of NH₄NO₃ during the heat treatment in nitrogen atmosphere at 400 ℃:

In the atomic layer deposition (ALD) process, trimethylaluminum (TMA) and anhydrous hydrogen fluoride are developed for the deposition of AlF₃ at substrate temperatures between 100 and 200 ℃. Gaseous HF source (pressure of 10 kPa at room temperature) is HF-pyridine, which enables the safe handling of anhydrous HF ^[50]. A typical deposit process is designed by 1 s dose of TMA injection, 30 s of N₂ purge, 1 s dose of TMA injection and 30 s of N₂ purge that produces a linear growth of AlF₃ with a mass gain of ~30 ng·cm^-2 per cycle. Jackson optimized the ALD AlF₃ coating onto LiNi_0.5Mn_0.3Co_0.2O₂ cathode powders using a Al(CH₃)₃/TaF₅ precursor combination that obviates the utilization of highly toxic HF(g) ^[54]. The freestanding LiCOO₂/multiwall carbon nanotube (LCO-MWCNT) electrodes were coated with 2 ALD cycles of AlF₃ conducted at 150 ℃ ^[55]. TMA and HF derived from HF–pyridine solution were used as precursors in the ALD process according the reactions ^[51]:

showing that AlF(CH₃)₂ fluorinated by the HF exposure is the key intermediate reaction to produce AlF₃ growth. The AlF₃ coating realized by solid-state reaction consists in the sintering at 970 ℃ for 15 h of the mixture of AlF₃ powder (particle size ~400 nm) with the precursor of the electrochemically active material and Li₂CO₃ (5 wt% excess) ^[54]. AlF₃- and LaF₃-coating LiCoO₂ electrodes were also prepared by spin-coating method and thermally treated at 400 ℃ for 30 min ^[56].

2.3.   Transport in AlF₃ coating

Lithium transport through surface coatings has been theoretically examined by several groups using first-principles calculations based on density functional theory ^[57,58,59]. A host supercell containing diluted concentration of Li atoms was considered for amorphous AlF₃ (a-AlF₃) with N_Li/N_host ratio of 100:80, 000 ^[57]. The self-diffusion coefficient (D_s) vs. temperature was represented by an Arrhenius law:

where D₀ is the diffusion prefactor and E_eff the effective activation barrier, with the fitting parameters D₀ = 7.7 × 10^-9 m²·s^-1 and E_eff = 0.65 eV for a-AlF₃ coating (Figure 2). Xu et al. ^[59] studied the diffusivity of interstitial lithium in crystalline α-AlF₃ and amorphous a-AlF₃. For an electronically insulating conformal coating, the transport of Li⁺ ions, which are the unique mobile charge species, is driven by an electric field. To balance the positive charge of Li⁺, there is a compensating negative charge in the coating, which is assumed to be immobile. This model referred to as "electrolyte model" ^[59] has been shown to be consistent with experimental works ^[60,61,62]. Li ions can diffuse by hopping from one interstitial site surrounded immediately by 8 oxygens to nearest site at a distance of 3.59 Å throughout a saddle point and must cross a barrier height E_m = 0.929 eV in the process. Predicted transport properties for crystalline and amorphous AlF₃ coatings are summarized in Table 1 ^[59], which illustrates that the lithiated AlF₃ contains Li–Al–F bonds and acts as a stable lithium conducting solid electrolyte. A mechanism has been proposed for the accommodation of the Al-F coating at the electrode interface ^[63]. As a consequence of the small charge transfer resistance, the Al-F coating layer facilitates the Li⁺ insertion/extraction at the interface. After heat treatment, the coating film is formed with an inner amorphous Li-Al-F layer such as LiAlF₄, which has been reported as a good lithium fast-ion conductor with a conductivity of 1 × 10^-4 S·m^-1 at 25 ℃ ^[63]. The LiAlF₄ structure is constituted by a network of corner sharing AlF₆ octahedra and Li ions are primarily associated with non-bridging fluorine atoms ^[64,65]. Among the thermodynamically possible reactions, AlF₃ converts to the high intrinsic Li-ion conducting Li₃AlF₆ according to:

at potential E = 1.28 V Li⁺/Li ^[66]. Li₃AlF₆ is a stoichiometric ternary mixture of LiF and AlF₃ exhibiting a high ionic conductivity of 10^-6 S·cm^{-1 [67]}. Therefore, the morphology and microstructure of the final phase of AlF₃ are affected by the deposition methods. As an amorphous-like phase is requested for preventing low migration barrier values, using a moderate post heat-treatment of the AlF₃ deposit, the co-precipitation synthesis is the best method that generates a non-crystallized material. Similar precautions are currently taken into account for coating made by ALD technique, for which a low substrate temperature is maintained below 200 ℃. The amorphous state of AlF₃ coating has been evidenced by several authors using HRTEM measurements (see Ref. ^[8] for example).

2.4.   Relative energies in Li-ion battery

Figure 3 is a schematic representation of the relative energies in a Li battery. The equilibrium working voltage V⁰ is given by the difference in chemical potential between anode (μ_A) and cathode (μ_C) for a given redox couple, i.e., 3.6 V vs. Li⁺/Li for the Ni^2+/3+ couple in the LiNi_1/3Mn_1/3Co_1/3O₂ host ^[1]. The electrochemical stability of a Li battery is determined by the electrochemical stability window (ESW) of the electrolyte or bandgap energy E_g:

where E_HOMO and E_LUMO are the energy of the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (HOMO), respectively. The condition for stability is given by:

where n is the number of electrons involved and F is the Faraday constant. Since μ_A of the lithium anode lies above the LUMO, the electrolyte is reduced to form a passivating SEI film that allows the diffusion of Li ions through the layer under a uniform electric field and prevents the decomposition of the electrolyte. On the cathode side, μ_C acts as a redox couple for the active transition-metal cation containing localized d-electrons in the d-manifold, e.g., Ni^2+/4+ in the NMC host material. During the charge/discharge process, the Fermi level E_FC is moved from one formal electronic state to another upon oxidation/reduction of the host. However, for the positive electrode in the charged state (E > 4.5 V), E_FC approaches the bottom of the Co^3+/4+ redox couple (μ_Cc potential), which interferes with the top of the anion p band (O^2-: 2p⁶) of the host and destabilizes the host by release of oxygen ^[68]. At this point, the cell voltage is also limited by the ESW of the electrolyte and a situation becomes similar to the anode side: μ_Cc lies below the HOMO level (located at 4.3 eV below μ_A(Li)) implying decomposition of the electrolyte unless a passivating SEI film blocks the reaction ^[69]. A thin insulating layer applied on the surface of the cathode acting as buffer isolates the host from the electrolyte. As shown in Figure 3, the AlF₃ coating possessing a bandgap of 10.8 eV has an E_HOMO energy level much lower than that of the electrolyte, even in the upper state of charge (SOC) of the cathode. Note that cyclic voltammetry shows the ESW of LiAlF₄ is 2.75–4.5 V vs. Li⁺/Li. However, the presence of such a film causes an increase in the internal resistance of the cell and consumes a portion of the lithium amount of the cathode ^[70].

3.   AlF₃-coated materials for Li-ion batteries

AlF₃ coating was used to enhance the performance of various cathode materials for Li-ion batteries including layered structures such as LiCoO₂ (LCO) ^{[28,53,71,72,73]}, LiNi_1-yMn_yO₂ (NMO) ^[74,75], LiNi_xMn_yCo_zO₂ (NMC) ^{[76,77,78,79,80,81,82,83,84,85,86]}, Li-rich NMC (LNMC) ^{[54,66,75,87,88,89,90,91,92,93,94,95,96,97,98,99]}, LiNi_1-x-yCo_xAl_yO₂ (NCA) ^{[100,101,102,103,104]}, spinel structures such as LiMn₂O₄ (LMO) ^{[105,106,107,108,109,110]}, LiNi_0.5Mn_1.5O₄ (LNM) ^{[111,112,113]}, olivine structures LiFePO₄ (LFP) ^{[114,115,116]} and LiCoPO₄ (LCP) ^[117]. AlF₃ coating has been also experimented on anode materials such as graphite and Li₄Ti₅O₁₂, in which case the deposit suppresses the gas generation ^{[118,119,120,121]}.

3.1.   AlF₃-coated LCO

LCO cathodes show significant degradation of electrochemical behaviors above 4.5 V. At such high voltage, AlF₃ coating is a solution, while Al₂O₃ is not, because of the partial conversion of the layer to metal fluoride by HF attack producing H₂O ^[122]. The beneficial effect of the surface modification of LCO particles by AlF₃ coating was first reported by evaluating laminated-type full Li-ion cells ^[123]. While the capacity of the uncoated LCO/C cell vanished rapidly after 500 cycles, the AlF₃-coated LCO cell showed excellent capacity retention of 91% at 1C rate in the range 3.0–4.5 V. Following this report, early analyses of AlF₃ coating on LiCoO₂ cathode particles was reported in Refs. ^[28,71]. The 0.5 mol% AlF₃ deposit shows a thin layer (5–10 nm thick) on the surface of LiCoO₂, which becomes thicker around 20–30 nm when the coating amount is 1 mol% ^[28]. This small deposit amount is sufficient to realize good performance with capacity retention of 97.7% after 50 charge–discharge cycles and capability of 93.1% at 5C rate. An LCO electrode coated with 2 mol% AlF₃ was able to deliver 208 mAh·g^-1 at 4.54 V upper cutoff voltage. The amount of Co dissolution of this electrode charged at 4.5 V as a function of the storage time in electrolyte at elevated temperature of 55 ℃ was compared with that of pristine material. Results show a concentration of 47 ppm for pristine LCO stored for 6 weeks, against only 19 ppm for the coated electrode ^[72]. DSC measurements of LiCoO₂ electrodes charged at 4.4 V and uniformly encapsulate by a 10–15 nm thick AlF₃ layer show that the exothermic peak associated with the thermal decomposition is at 273.5 ℃ and the reaction enthalpy is 392 J·g^-1, against 208.5 ℃ and 544 J·g^-1, respectively, for the bare materials ^[73]. This result demonstrates that the AlF₃ layer improves the thermal stability. This AlF₃-coated electrode allows a charge regime at high voltage of 4.6 V vs. Li⁺/Li delivering a specific capacity of 160 mAh·g^-1. Recently, Zhou et al. ^[53] fabricated freestanding LiCoO₂/multiwall carbon nanotube nanocellulose fibril (LCO-MWCNT-NCF) composite electrodes coated with AlF₃ layer. The AlF₃ deposit was realized with 2 ALD cycles, leading to a high specific capacity of 216 mAh·g^-1 at 4.7 V for LCO-MWCNT-NCF electrodes. In contrast, when coated with 2 ALD cycles of Al₂O₃, LCO-MWCNT-NCF electrodes cannot be cycled above 4.5 V due to the higher LUMO energy level.

3.2.   AlF₃-coated NMO

The single-phase LiNi_1-xMn_xO₂ (NMO) exhibits a specific capacity of 150 mAh·g^-1 with the formal charges of Ni and Mn of 2+ and 4+, respectively, only the Ni^2+/4+ redox couple being available. However, the electrochemically inactive Mn⁴⁺ provides significant structural stability ^[124]. LiNi_0.5Mn_0.5O₂ cathode materials prepared by quenching treatment and coated with AlF₃ by conventional coprecipitation method show initial discharge capacity of 175 mAh·g^-1, decreasing to 153 mAh·g^-1 after 50 cycles at 0.1C rate (28 mA·g^-1). At high current density of 5C rate (1.4 A·g^-1), this coated electrode delivered 120 mAh·g^-1 at in the voltage range 2.5–4.5 V ^[125]. Time-of-flight secondary ion mass spectroscopy was used to investigate the effect of a ~10 nm AlF₃ deposit on LiNi_0.5Mn_0.5O₂ electrode ^[74]. The insulating LiF outer layer originating from the LiPF₆ decomposition was suppressed by the AlF₃ coating with the benefit of the reduction in the charge transfer resistance. Figure 4 shows the electrochemical features of pristine and AlF₃-coated Li[Ni_0.5Mn_0.5]O₂. The surface modification using 2 wt% AlF₃ of layered oxide improved significantly the cycling stability of the electrode ^[75], which retained a specific capacity of 175 mAh·g^-1 after 50 cycles at C/2 rate in the voltage range 2.0–4.4 V (Figure 4a). This corresponds to a capacity retention of 93% with an excellent coulombic efficiency of 99.5% much higher than that of the pristine material (Figure 4b).

3.3.   AlF₃-coated NMC

Cathodes based on Li[Ni_xMn_yCo_z]O₂ (NMC) compounds with x + y + z = 1 are actively studied as the new generation of stable high-specific-energy cathode materials for Li-ion batteries ^[126]. In an early work, Woo et al. ^[76] investigated the electrochemical performance of the AlF₃-coating of Ni-rich LiNi_0.8Co_0.1Mn_0.1O₂ (NMC811) cathode materials and examined the stabilized electrolyte/electrode interface. A 10-nm thick coating on NMC333 powders was obtained using 0.25 mol% of AlF₃ in the sol-gel solution ^[77,78]. Regardless the cycle life of AlF₃-coated LiNi_1/3Co_1/3Mn_1/3O₂ electrodes, Sun et al. ^[77] determined that the charge transfer resistance (R_ct) was stable under the high voltage cutoff of 4.6 V. The surface of the layered LiNi_1/3Co_1/3Mn_1/3O₂ cathode materials was modified by covering the particles with a 10 nm thick AlF₃ layer deposited by the conventional coprecipitation technique. Electrochemical tests showed an improvement of the performance at high cutoff voltage of 4.5 V, with a capacity retention of 81% after 800 cycles and a decrease of the area specific impedance even at the discharge rate of 10C ^[79]. About 3 nm thick uniform AlF₃ coating (1.5 wt%) on the surface of LiNi_1/3Co_1/3 Mn_1/3O₂ limited the capacity fading at 3% when cycled 80 times in the voltage range 2.8–4.3 V at 2C rate ^[80], which is better than the 93.5% capacity retention obtained at 1C at 60 ℃ after 50 cycles for the LiF-modified LiNi_1/3Co_1/3 Mn_1/3O₂ electrode ^[81]. AlF₃-coated LiNi_0.45Mn_0.45Co_0.10O₂ as cathode material ^[82] delivered an almost stable specific capacity of ~145 mAh·g^-1 when discharge at the rate of 0.1C (20 mA·g^-1) in the voltage range 2.5–4.5 V. The thermal behavior of AlF₃-coated Li_xNi_1/3Co_1/3Mn_1/3O₂ was studied in the charged state, i.e., in the delithiated Li concentration x = 0.35. TGA and high-temperature XRD data showed that the phase transition from rhombohedral Li_0.35MO₂ to spinel Li_0.7M₂O_1.933 structure appearing at 450 ℃ is induced by a slight amount (~0.07 mol) of released oxygen. The decrease of oxygen loss (~0.04 mol) was detected for the AlF₃-coated Li_0.35Ni_1/3Co_1/3Mn_1/3O₂ electrode, which retards the formation of the spinel phase. It was concluded that the AlF₃ coating layer played two roles at the particle surface: (i) a protection against oxygen loss and (ii) a sacrificing for the formation of Li-Al-O and Li-Al-F complexes ^[83]. Using dual deposition process of metal fluorides, the co-effect of 1 wt% AlF₃ and 1 wt% MgF₂ coated layers on the electrochemical properties of NMC333 was studied under high voltage in the range 2.8–4.7 V. In that case, the coated electrode delivered an initial capacity of 207 mAh·g^-1 at C/5 discharge rate ^[127]. Commercial LiNi_0.5Mn_0.3Co_0.2O₂ powders (NMC 532, from ECOPRO Co., Korea) were coated with either AlF₃ or LiLaPO₄ to fabricate a blended cathode ^[84]. The synthesis was realized with 0.25 wt% AlF₃ of the NMC532 powder in the solution, which increases the thermal stability of the electrode charged at 4.6 V. Owing to the coating, the exothermic DSC peak shifts from 230 to ~280 ℃ and the heat generation decreases from 186 to 100 J·g^-1 compared to uncoated particles. Spherical NMC532 particles were coated with different AlF₃ concentrations in the molar ratio range 0.25–2.0 ^[85]. A combination of XRD data, galvanostatic charge/discharge tests and EIS measurements showed that 0.5 mol% AlF₃ coating is the best content providing a high specific capacity of ~160 mAh·g^-1 at 6C rate in the voltage range 2.8–4.5 V. Yang et al. ^[85] coated the surface of LiNi_0.5Mn_0.3Co_0.2O₂ (NMC532) by coprecipitation method, the amount of AlF₃ being set at molar ratios n_AlF3/n_NMC532 in the range from 0.25 to 2%. It was demonstrated that oxygen vacancies favor the Li⁺-ion transport and increase the capacity in forthcoming cycles. An important advantage of the AlF₃ coating layer is the termination of the increase of cell impedance upon charge–discharge process by preventing directly the SEI formation. The AlF₃-coated NMC532 operating at 4C over 2.8–4.5 V had capacity retention of 98% (initial specific capacity of mAh·g^-1) after 100 cycles. Ni-rich LiNi_0.8Mn_0.1Co_0.1O₂ electrodes (NMC811) with AlF₃ interfacial layer deposited by ALD technique from AlCl₃ and TiF₄ precursors kept at 120–130 ℃ were investigated in the voltage range 2.75–4.5 V vs. Li⁺/Li. The average specific capacity of AlF₃-coated NMC811 cathode was 184 mAh·g^-1 at 20 mA·g^-1 current density and 106 mAh·g^-1 at 200 mA·g^{-1 [66]}.

3.4.   AlF₃-coated Li-rich layered NMC cathodes

The lithium-rich Li[Li_δNi_xMn_yCo_z]O₂ (NMC) compounds with δ + x + y + z = 1 layered oxides (LNMC) are inter-grown composite structures xLi₂MnO₃·(1 - x)LiMO₂ (M = Mn, Co, Ni), which have attracted a lot of interest as positive electrode (cathode) materials for Li-ion batteries because since the their high energy density of about 250 mAh·g^-1 in the voltage range 2.0–4.8 V vs. Li⁺/Li ^[128]. Nowadays, they are considered as promising cathode element and used for powering electric vehicles (EVs). Surface analysis showed that surface‐modified Li[Li_0.2Ni_0.17Co_0.07Mn_0.56]O₂ nanoparticles with 3 wt% of AlF₃ deposited by chemical process (2 nm thick) exhibits a largely suppression of the undesirable growth of SEI layer. This 3 wt% of AlF₃ coating enhanced the capacity retention to 83% and the cathode delivers a specific discharge capacity of 182 mAh·g^-1 after 50 cycles when charged to 4.8 V vs. Li⁺/Li. In relation to the effect of the coating on the SEI formation, a detailed description of the complex surface chemistry of Li-rich layered cathodes can be seen in Ref. ^[86]. This decrease of impedance is responsible in turn for the improvement of the rate capability. For instance, the reversible capacity of the AlF₃-coated Li(Li_0.17Ni_0.25Mn_0.58)O₂ at 5C rate was raised to 104 mAh·g^-1 after 200 cycles, much larger than before coating. Electrochemical tests exhibited improved rate capability and life cycle behavior at elevated temperature of 55 ℃ with the best results for 0.5 mol% of aluminum fluoride deposit. Furthermore, the thermal stability studied by DSC showed a shift of the exothermic peak from 247 to 308 ℃ with lower exothermic heat release ^[87]. Using a combination of scanning/transmission electron microscopy (SEM/TEM) and electron energy loss spectroscopy (EELS) characterizations, Sun et al. ^[45] evidenced that the coating of Li[Li_0.19Ni_0.16Co_0.08Mn_0.57]O₂ cathode enhanced the overall electrochemical features in contrast with the typical shortcomings of Li‐rich positive electrodes due to the Li chemical leaching effect of the AlF₃ deposit transforming the initial Li₂MnO₃ layer to a spinel phase. Optimized electrode with 2 wt% AlF₃ coating concentration delivered a specific capacity of about 206 mAh·g^-1 when cycled between 2.0 and 4.6 V at a current density of 0.5C rate (125 mA·g^-1) over 100 cycles. From the thermal stability view point, the 2 wt% coating reduces the exothermic reaction with a generated heat of 274 J·g^-1 at 262.5 ℃.

A core shell material Li_1.2Mn_0.54Co_0.13Ni_0.13O₂@AlF₃/C (LMSAC) was synthesized by coating AlF₃ and carbon hybrid deposit via sol-gel process. The LMSAC cathode delivered the initial specific capacity of 230 mAh·g^-1 at 1C rate in the potential range of 2.0–4.8 V vs. Li⁺/Li and exhibited a capacity retention of 96% after 50 cycles ^[88]. The AlF₃ surface-treated Li_1.1(Ni_0.15Co_0.1Mn_0.55)O_1.95 layered oxide was cycled at 25 and 55 ℃. At elevated temperature, a capacity gain of 10% was attributed to the improved thermodynamic activation ^[89]. Li-rich Mn-based cathode Li[Li_0.2Fe_0.1Ni_0.15Mn_0.55]O₂ particles doped with iron were artificially protected by 7% AlF₃ nano-coating obtained by coprecipitation method ^[90]. It was stated that the fluorine source, NH₄F, acts as a chelating agent that promotes a slow reaction with aluminum and more complete coating film. Upon 3-nm AlF₃ coating, great structural changes were observed by TEM and electron diffraction measurements. An integrated layered-spinel structure was formed. An excellent performance with a specific capacity of ~120 mAh·g^-1 was observed after 100 cycles at 20C rate. As AlF₃ coat appeared to be uniform and thin (compared with the 15-nm Al₂O₃ layer), this protecting deposit is preferred to Al₂O₃ for the electrochemical behaviors that can be triggered by tuning the amount of coating. Zheng et al. ^[8] demonstrated the efficiency of AlF₃ coating on Li-rich Mn-rich Li_1.2Ni_0.15Co_0.10Mn_0.55O₂ electrode cycled at C/3 rate in the voltage range 2.0–4.75 V vs. Li⁺/Li upon cycle number. Quantitative analysis of the pre-peak of the O K-edge and the linear relation of L₃/L₂ EELS intensity with the Mn valence state at the particle surface and in the bulk showed the reduced formation of oxygen-deficient surfaces and the stabilization of the spinel-like phase (Figure 5).

Li_1.17Ni_0.25Mn_0.58O₂ with 5–7 nm thick AlF₃-coating provided a high specific capacity > 250 mAh·g^-1 at 0.2C rate, maintained at 104 mAh·g^-1 when discharged high current density of 5C after 200 cycles. Such a material displayed improved coulombic efficiency from 76.4 to 89.5%. EIS analyses revealed that the AlF₃ layer induces a pre-activation of the Li-rich layered oxide with a maintenance of more active Li sites during the insertion/extraction reaction ^[91]. The effects of AlF₃ modifying cyclic performance of Li_1.24Ni_0.12Co_0.12Mn_0.56O₂ cathode materials (LNCM@AlF₃) were recently reported by Ding et al. ^[92]. A combined set of electrochemical tests showed that the best results are obtained with a proper coating of 2 mol% AlF₃ (~7 nm in thickness). This homogeneous layer favored the diffusion ability of Li⁺ ions at the electrode/electrolyte interface and largely decreased R_ct from 350 Ω in pristine to 59 Ω in LNCM@2%AlF₃ electrode. LNCM@2%AlF₃ operating at 55 ℃ delivered a specific discharge capacity of 219 mAh·g^-1 at current density of 20 mA·g^-1 after 50 cycles. Kim et al. ^[63] reported the positive effect of a controlled AlF₃ coating on the Li-rich NMC electrode, namely 0.5Li₂MnO₃-0.5LiNi_0.5Mn_0.3Co_0.2O₂ (or Li_1.2Mn_0.52Ni_0.2Co_0.08O₂). The authors attributed this improvement to the fact that the coating (nominal amount of AlF₃ about 0.25 mol% of the LNMC material) leads to a Li–Al–F interface that acts as a stable lithium conducting solid electrolyte. At discharge rate of 0.5C in a voltage range of 2.0–4.6 V vs. Li⁺/Li, the Al-F coated electrode had an improved capacity retention (98% of the 216 mAh·g^-1 initial capacity) compared to the pristine one (93% of the 206 mAh·g^-1 initial capacity). The surface modification of Li[Li_0.11Ni_0.33Mn_0.56]O₂ with various amounts of AlF₃ (0 ≤ AlF₃ mol% ≤ 10) was realized by the precipitation routine setting the Al:F ratio to 1:3. After coating with 5% Al/F and heat treatment at 400 ℃, the particle had a core-shell structure: a disordered layer separated the crystalline LNMC bulk from the amorphous AlF₃ outer coating. The electrode with low coating content (~1 mol%) had a specific capacity of 215 mAh·g^-1 on charging to 4.6 V ^[93]. Wang et al. ^[94] claimed excellent electrochemical behaviors of the AlF₃-coated Li_1.2Mn_0.534Ni_0.133Co_0.133O₂ cathode that exhibited a discharge capacity of 222 mAh·g^-1 (capacity retention of 96.5%) after 50 cycles at 1C rate. Li et al. ^[95] prepared an AlF₃-coated Li_1.2(Mn_0.54Ni_0.16Co_0.08)O₂ cathode materials via wet process. Optimized powders with 5–7-nm thick AlF₃ layer exhibited a capacity retention of 72.4% after 50 cycles at 1C rate. Amalraj et al. ^[86] considered the detailed chemistry of the coating/active material and the electrochemical performance of Li and Mn rich xLi₂MnO₃-(1 - x)LiMn_yNi_zCo_wO₂ powders (0.4 ≤ x ≤ 0.5) with an emphasis on the effect of 2–3 wt% AlF₃ coating. TEM studies showed that the coating layer is formed of nano-crystals with tetragonal structure t-AlF₃ (P4nmm space group) regularly distributed. Additionally, amorphous clusters Al-containing species, like AlF₃, AlF_xO_y and Al[FOH] were detected by solid-state NMR measurements. Such AlF₃-coated material exhibited specific capacity of ~250 mAh·g^-1 at a C/5 rate, and high lithium storage capability at 60 ℃. Li- and Mn-rich layered oxide Li_1.4Ni_0.25Co_0.10Mn_0.65O₂ was coated with a small amount of AlF₃ (~11 vol%) by the conventional sol-gel route ^[96]. It was shown that during preparation the environmental Li excess caused the formation of Al₂O₃ and LiF that led the surface conversion from Li₂MnO₃ to self-generated LiMn₂O₄ spinel layer, which upgrades the electronic conductivity of particles.

AlF₃ has been chosen as the inert coating material to protect the Li- and Mn-rich xLi₂MnO₃-(1 - x)LiMn_0.5Ni_0.5O₂ particles with x = 0.3 (Li_1.134Ni_0.3Mn_0.566O₂) and x = 0.2 (Li_1.2Ni_0.2Mn_0.6O₂) ^[74]. Thickness of the amorphous AlF₃ coating layers (5 wt% of the Li-rich layered oxide) uniformly deposited by the conventional coprecipitation method is 5 nm for Li_1.2Ni_0.2Mn_0.6O₂ particles (200-nm size) against 2 nm for Li_1.134Ni_0.3Mn_0.566O₂ ones (Figure 6). Figure 7 presents the galvanostatic charge/discharge profiles carried out at C/10 rate (30 mA·g^-1) in the potential range 4.5–2.0 V vs. Li⁺/Li for (a) pristine and (b) AlF₃-coated Li_1.134Ni_0.3Mn_0.556O₂. The cell voltage vs. specific capacity of Li-rich layered oxides have been analyzed in prior works (see for instance Ref. ^[128]). The first discharge process occurs as activation of the rhombohedral LiMn_0.5Ni_0.5O₂ (R-3m) and monoclinic Li₂MnO₃ (C/2m) phases by several steps: (ⅰ) the delithiation of the R-3m phase with a voltage slope corresponding to the oxidation of Mn³⁺ and Ni²⁺ cations; (ⅱ) activation of the C/2m phase at ca. 4.5 V associated with the oxygen loss; (ⅲ) structural reorganization due to Li⁺ ions removal evolving O₂. Following the first charge at ca. 5 V, the new active material contains presumably lithiated MnO₂ and LiMn_0.5Ni_0.5O₂ phase that provides a specific capacity of 230 mAh·g^-1 decreasing down 200 mAh·g^-1 after 55 cycles. The voltage fade issue was analyzed in terms of differential (or incremental) capacity as shown in Figure 8, which compares the -dQ/dV vs. cell voltage plots of the 2^nd and 50^th cycle for bare and AlF₃-coated Li_1.2Ni_0.2Mn_0.6O₂. These graphs display a shift down to ~3.67 V after 50 cycles for the bare material, while stable voltage is shown for the coated electrode. All these results indicate that a 5-nm thick coating is enough to protect the electrode particle against the attack of the electrolyte.

Zheng et al. ^[8] have investigated in more details the functional mechanism of the Li-rich and Mn-rich cathode materials, by a comparison between Li_1.2Ni_0.15Co_0.10Mn_0.55O₂ uncoated and coated with a 10 nm-thick AlF₃ layer. First, the AlF₃-coated material exhibited significantly reduced capacity decay above 3.5 V, like in Figure 9, indicating a reduction of layered to spinel-like phase transformation. Second, the AlF₃-coated material showed only 10.1% (0.40 V, inset of Figure 9) decrease in average discharge voltage after 100 cycles, which is smaller than the 12.3% (0.47 V) observed for uncoated material. This result, in agreement with other data ^[74,75], indicates that AlF₃ coating not only alleviates the undesirable phase transformation but also maintains the structural stability of the transformed spinel-like phase. The structural evolution of the coated and uncoated samples was investigated by S/TEM and EELS characterization in ^[8] after 100 cycles. The results are the following. In contrast with that of the uncoated sample, there is no serious corrosion observed for the AlF₃-coated material so that the cycling does not destroy the coat. There is no thick SEI layer on the coated sample, the inert AlF₃ coat prevents direct contact between the electrode material and the electrolyte, which thus largely suppresses the side reactions between them observed in the uncoated sample. This result also proves that in spite of the oxygen extraction during activation of the Li₂MnO₃ component, the coating layer functions as a buffer layer to reduce the reactivity of evolved oxygen species and thus decrease the oxidation of electrolyte. Less oxidation of the carbonate electrolyte translates to less formation of acidic species, the reason why no serious corrosion phenomena, such as etched surfaces and corrosion pits, are detected for AlF₃-coated material. The reduced acidic corrosion and the limited accumulation of SEI layer results in an enhancement of the structural stability of the spinel-like phase and the electrode/electrolyte interface, which further enables the reversible lithium ion intercalation/deintercalation processes in the spinel-like phase. Indeed, although the phase transformation from layered to spinel-like was still observed near the surface of the coated sample, the inert AlF₃ coating layer could alleviate or delay this unfavorable phase transformation. The valence states of the Mn/Co/Ni ions in the AlF₃-coated material after cycling are much more stable as compared to those in uncoated material. This result substantiates that the coating layer depresses the attack by acidic species and significantly reduces the formation of oxygen-deficient surfaces. In particular, the suppression of the formation of lower valence (Mn³⁺) could largely decrease the disproportionation of Mn³⁺ in the spinel-like phase and thus promise the improved capacity retention.

Pang et al. ^[54] modified the Li-rich Li_1.2Mn_0.54Ni_0.13Co_0.13O₂ powders using 5 wt% AlF₃ synthesized via a scalable solid-state reaction. The final product exhibited an Al and F co-doping core and a uniform LiF nano-coating that delivered a specific discharge capacity of 211 mAh·g^-1 after 99 cycles at 0.2C rate compared with 160 mAh·g^-1 for pristine powders. EIS results showed a 4-time increase of R_s due to the formation of resistive LiF coat, while R_ct is 3 times lower than that of the non-modified LNMC.

Recently, Zhao et al. ^[97] have prepared LiAlF₄- and AlF₃-coated Li_1.2Ni_0.2Mn_0.6O₂ (LNMO) cathodes and compared them. 1 wt% coating was realized by the coprecipitation method, in which LiNO₃ was added as lithium source for the LiAlF₄ coating. The final product heated at 400 ℃ for 6 h under Ar atmosphere showed the formation of regular and stable protective layers (~5.2 nm thick). XPS measurements carried out in the binding energy range 528–536 eV showed the evolution of the O 1s peaks corresponding to the lattice O^2- (O 1a_a) and non-bonding O^n- surface species (O 1a_b, O 1a_c). Upon AlF₃ coating, the O 1a_a peak at 529.3 eV slightly declined, while the other peaks at 531.6 and 533.3 eV, respectively, increased importantly. Rate capability performed at 5C rate demonstrated respective capacity retention of 23.9%, 40.8% and 52.4% (133 mAh·g^-1) for LNMO, LNMO@AlF₃ and LNMO@LiAlF₄. The ultralong stability of LNMO@LiAlF₄ upon 3000 cycles is attributed to the heterostructured LiAlF₄ coating, in which nonbonding O^n- species were generated. Chen et al. ^[99] prepared Li-rich Li_1.2Ni_0.13Co_0.13Mn_0.53O₂ coated by AlF₃ (2-nm thick) by centrifugation of the precursor with the NH₄F solution and washed with ethanol. To avoid the formation of Al₂O₃, an excess of NH₄F compared with Al(NO₃)₃ was used, i.e., with NH₄F:Al(NO₃)₃ ratio of 3:1 in molar. A composite cathode was prepared by mixing Li-Rich@AlF₃ with graphene in the weight ratio of 95:5. This electrode delivered a specific capacity of ~110 mAh·g^-1 at 2C rate after 250 cycles.

3.5.   AlF₃-coated NCA

As a derivative of LiNiO₂, LiNi_1-x-yCo_xAl_yO₂ (NCA) exhibits enhanced properties due to the presence of Co and Al ^[129]. Surface of NCA particles was fluorinated with NH₄F (with F:NCA ratio of 1 at%) by solid-state reaction at 300–700 ℃ for 2 h. As evidenced by XPS measurements, the surface metal–oxygen (M–O) bonds were partially replaced of by metal–fluorine (M–F) bonds. The XPS O 1s spectrum displayed a binding energy shift towards lower energies for the fluorinated NCA sample of 0.21 and 0.27 eV for the lattice oxygen (M³⁺–O at the surface) peak at 528–530 eV and the peak due to chemically absorbed species (LiOH or Li₂CO₃) at 531–533 eV, respectively. This fluorination protects the M–O bonds from side reaction, reduces the release of oxygen and lowers the charge transfer resistance ^[99]. In the coprecipitation method, the molar ratio of F to Al was kept at 1:7 and the 0.25 mol% to the NCA powder in the solution produced a 10-nm thick coating layer ^[100]. Park et al. ^[101] deposited a uniform AlF₃ thin film of ~10 nm (0.25 mol%) on the surface of NCA powders and tested the cycle performance in the temperature range from -10 ℃ to 55 ℃. It is believed that the coating reduced the catalytic effect of Ni⁴⁺ and the formation of Li_xNi_1-xO-type oxide on the NCA surface operating at temperatures higher than 55 ℃. Zhang et al. ^[102] modified the surface of NCA particles using AlF₃ and AlPO₄ deposits. The cycleability of the AlF₃-coated NCA was quite stable, with a capacity retention of 96.3% of its initial after 30 cycles. A 50-nm thick AlF₃ layer was deposited on NCA particles by simple dry coating process using AlF₃ synthesized from an aqueous solution of Al(NO₃)₃-9H₂O and NH₄F with a molar ratio of 1:7 ^[103]. Electrochemical tests carried out at a current density of 20 mA·g^-1 (0.1C rate) in the voltage range 2.7–4.3 V at 55 ℃ showed improved cycling behavior with ~85% capacity retention (initial capacity of 200 mAh·g^-1). The long-term cycling performance of a full Li-ion cell with mesocarbon microbeads as anode displayed a capacity retention of 86% after 1000 cycles performed at 1C rate over the voltage range 3.0–4.2 V at room temperature. TEM bright field images of the coated electrode after 500 cycles evidenced the absence of cracks and particle pulverization. Furthermore, analysis of metal-ion dissolution by ICP for electrodes charged at 4.2 V and stored at 55 ℃ for 4 weeks showed a drastic decrease of dissolved contents of Ni and Co, i.e., 10 and 1.2 ppm, respectively, against 24 and 4.1 ppm for the pristine NCA.

3.6.   AlF₃-coated spinel frameworks

It is well known that the discharge capacity of spinel LiMn₂O₄ suffers from irreversible capacity loss due to Mn dissolution during cycling at elevated temperatures exceeding 55 ℃ ^[130]. The AlF₃-coated Li_1.1Mn_1.85Al_0.05O₄ spinel-like electrode was synthesized by coprecipitation method via calcination at 400 ℃ for 5 h ^[104]. The 10-nm thick coated appeared to be beneficial for cycling at 55 ℃ in the range 3.0–4.3 V. A specific discharge capacity of 95 mAh·g^-1 was delivered at 5C rate compared with 90 mAh·g^-1 for the pristine material. This coated electrode sustained an excellent capacity retention of 96.2% after 100 cycles and was protected against Mn dissolution in the fully charge state (4.3 V), since after 4 weeks at 55 ℃, the dissolution was reduced to 83 ppm against 280 ppm for pristine Li_1.1Mn_1.85Al_0.05O₄. Complete blocking of Mn³⁺ dissolution from the LiMn₂O₄ surface can be insured by AlF₃ coating ^[105]. Tron et al. ^[106] investigated LiMn₂O₄ coated with 2 wt% AlF₃ as cathode material in 1 mol·L^-1 Li₂SO₄ aqueous electrolyte. A 4-Ah Li-ion battery with graphite anode and AlF₃-coated LiMn₂O₄ spinel cathode was investigated in the voltage range 3.0–4.2 V vs. Li⁺/Li at 0.2C rate ^[107]. With 0.5 mol% (2.2 wt%) AlF₃, the coated LiMn₂O₄ powders displayed an enhanced capacity retention of 5.4% after 100 cycles. The improved performance of the AlF₃-coated LiMn₂O₄/graphite battery was evidenced by the storage test at 25 ℃, leading to a slight increase of the charge transfer resistance R_ct from 121 to 145 Ω after 28 days, while R_ct increased up to 386 Ω for the uncoated spinel. Wang et al. ^[108] fabricated sophistical Al-F codoped spherical spinel particles (15 μm size), LiMn_1.96Al_0.04O_3.94F_0.06, via solid-state reaction followed by 1 mol% AlF₃ coating via coprecipitation method. Due to the strong and stable Al–F bonds, the 20-nm thin layer of AlF₃ exhibited several qualities such as high ionic conductivity and chemical and electrochemical stability, which can negate the oxygen activity on the electrode surface during cycling. Therefore, AlF₃ appears to be an efficient coating material for LiMn₂O₄ in aqueous rechargeable lithium batteries. The best cycling behavior od the Li-rich layered-spinel structure Li_1.3Mn_4/6Ni_1/6Co_1/6O_2.4 was obtained with a AlF₃ coating of 2 wt%. The good capacity retention of 90% after 50 cycles at 0.6C rate over the voltage range 2.0–4.8 V was mainly attributed to the stable charge transfer resistance ^[109].

Spinel LiNi_0.5Mn_1.5O₄ (LNM) high-voltage cathode was modified by different amounts of AlF₃ using coprecipitation coating process ^{[110,111,112]}. EIS measurements on the 1 wt% AlF₃-coated LNM demonstrated the inhibition of the growth of the SEI and the prevention of Mn³⁺ dissolution in the electrolyte, resulting in improved reversibility ^[110]. The best cycling behavior of the Li-rich layered-spinel structure Li_1.3Mn_4/6Ni_1/6Co_1/6O_2.4 was obtained with a AlF₃ coating of 2 wt%. The good capacity retention of 90% after 50 cycles at 0.6C rate over the voltage range 2.0–4.8 V was mainly attributed to the stable charge transfer resistance ^[113]. Li et al. ^[111] reported the improved performance of LiNi_0.5Mn_1.5O₄ cathode material coated with various amounts of AlF₃ (1, 2 and 4 mol% of the LMN powders). 4 mol% AlF₃-coated sample exhibited a high specific capacity of 115 mAh·g^-1 at 2C rate (300 mA·g^-1) for 40 cycles in the voltage range 3.0–4.9 V compared to 84 mAh·g^-1 for the uncoated sample. Ke et al. ^[112] successfully coated well-crystallized LNM powders with amount of AlF₃ up to 3 mol%. Results show that, delivering a specific capacity of 109 mAh·g^-1 at 10C rate, this cathode is suitable for high-power batteries for electric vehicles.

3.7.   Other cathodes coated with AlF₃

The surface of the olivine-type LiFePO₄ (LFP) electrode was partly modified by a coating of nano-sized AlF₃ and Al₂O₃ deposits, which appeared to be interfacial stabilizer by improvement of the overall electrochemical behaviors ^[113]. Song et al. ^[114] examined the influence of AlF₃ nano-coating (nominal 2 wt%) on the electrochemical behaviors of LFP cathode at 60 ℃ in Li-ion battery including graphite as anode. The 5-nm thick coating maintains a specific capacity of 130 mAh·g^-1 at 1C rate after 100 cycles over 2.2–3.8 V vs. Li⁺/Li (capacity retention of 92%). Tron et al. ^[115] prepared an LFP electrode coated with 1 wt% AlF₃ deposit via coprecipitation method. This material used as a cathode in an aqueous rechargeable lithium battery with 1 mol·L^-1 Li₂SO₄ dissolved in water as electrolyte delivered a specific capacity of 132 mAh·g^-1 at 1C rate and a capacity retention of 93% after 100 cycles. Electrochemical properties of optimized 4 mol% AlF₃-coated LiCoPO₄ olivine-like cathode indicated improved cycleability and rate capability with initial discharge capacity of 159 mAh·g^-1 at C/10 rate and 91% capacity retention over 50 cycles ^[116]. AlF₃ coating of Li₃V₂(PO₄)₃/C composite material was prepared via a conventional precipitation method. The capacity retention rate of LVP/C-AlF₃ was 85% after 50 charge/discharge cycles at 0.2C rate in the voltage range 3.0–4.8 V that compared to 76% of the uncoated material ^[131].

LiV₃O₈ (LVO) with monoclinic structure was considered for use in lithium polymer batteries (LPBs) ^[132]. A 2.9 wt% AlF₃-coated LVO prepared by conventional coprecipitation method was formed as core-shell particle with an intermediate Li_xV₂O₅ thin layer. It was suggested that NH₄F in the coating precursor reacts with LiV₃O₈ during calcination ^[133]. Cycling performance was investigated at a current density of 150 mA·g^-1 in the voltage range 1.8–3.8 V. The AlF₃-coated LVO demonstrated an initial capacity of 231 mAh·g^-1 and a capacity retention of 91.8% over 50 cycles. It was suggested that the AlF₃ layer prevents the capacity loss due to the incomplete reversibility of the transformation between Li₄V₃O₈ and LiV₃O₈ phases and better retains the discharge voltage plateaus.

4.   Characterization of the coating/electrode interface

4.1.   Raman spectroscopy

Generally, the XRD pattern of AlF₃-coated electrode does not display any trace of Al-F compound because the deposited coat is a thin amorphous phase. The non-destructive Raman spectroscopy is a powerful analytical method to probe the surface of coating layers compared to XRD technique especially in the case of highly disordered or amorphous materials ^[134]. It is also used to study the environment of lithium ions in lithiated metal oxides because the internal modes of LiO₆ occur in the spectral range 200–300 cm^-1, while those of LiO₄ appear in the range 350–500 cm^{-1 [135,136]}. The Raman spectrum of α-AlF₃ displays four fundamental active vibrations including external (in the low-frequency range) and internal modes. In the Mulliken notation of the D_3d symmetry, they decompose in 1A_1g + 3E_g among which E_g modes are doubly degenerate. The rotation mode of AlF₆ octahedra (E_g) is located at 96 cm^-1, the A_1g rotation mode of AlF₆ at 157 cm^-1, the E_g bending mode of Al–F–Al bonds at 382 cm^-1 and the E_g stretching vibration of Al–F bonds at 478 cm^{-1 [137]}. Note that the external modes have a strong intensity, while internal modes are very weak in the ratio 1:25. For this reason, the external Raman bands of the AlF₃ coating are solely observed as shown in Figure 10. Due to the lower symmetry and inequivalent atomic positions in the elementary unit cell of β-AlF₃, the larger number of Raman active modes is classified as 8A_g + 7B_1g + 7B_2g + 5B_3g. Experimental Raman data display only two spectral features located at 75 (strong) and 411 cm^-1 (moderate) assigned to rotational mode of AlF₆ (A_g), and the bending mode of Al–F–Al bonds (A_g), respectively ^[138].

4.2.   X-ray photoelectron spectroscopy

X-ray photoelectron spectroscopy (XPS) is currently used as analytical tool to determine the composition and the sputter depth profile of AlF₃ films ^[50]. Makarowicz et al. ^[139] distinguished the local geometric structure of α- and β-AlF₃ surfaces using XPS with surface Al³⁺ sites contributing to the Al 2p binding energy peak at 77.1 and 76.1 eV, respectively. The binding energy of F 1s was reported to be 686.7 eV in α-AlF₃, which deviates to 686.7 eV in β-AlF₃. Table 2 lists the Li 1s, Al 2p and F 1s XPS data for LiF, LiAlF₄ and AlF₃ ALD films. Note that the F 1s peak in AlF₃ and LiAlF₄ are shifted to higher binding energy compared to that in LiF due to the higher electronegativity of Al compared to Li. In the F 1s XPS spectrum, the band appearing at 687.3 eV is attributed to the C–F bonds of the PVdF binder ^[130].

4.3.   Electrochemical impedance spectroscopy

One of the most powerful tools to characterize the evolution of the electrical and electrochemical properties of a battery electrode is the electrochemical impedance spectroscopy (EIS), which can reveal the improved performance of an electrochemical system. EIS is widely used to analyze the kinetics of electrode reaction and the effect of EEI on kinetics ^[140]. A Nyquist plot, i.e., Z'(ω) vs. Z"(ω), is the response of a sample or a cell excited by an a.c. signal analyzed in terms of complex impedance Z(ω) = Z'(ω) + jZ"(ω) (or admittance Y(ω) = Y'(ω) + jY"(ω)) in a wide frequency range from 10 MHz to 0.01 Hz ^[141]. EIS experiments are a good tool to detect the improvement of protective coatings using adequate equivalent circuits. An equivalent circuit that describes the spectrum is not unique but a good fit to the experimental data must represent an accurate physical model of the studied system ^[142]. Experimentally, four regions can be distinguished in an EIS spectrum: (ⅰ) the intercepts with the Z' axis at high frequency is related to the ohmic resistance of the cell (R_Ω), (ⅱ) the semicircle at high frequency corresponds to the passivating SEI surface film and coating layer (R_S), (ⅲ) the intermediate-frequency semicircle is related to the charge transfer resistance R_ct at the electrode/electrolyte interface and (ⅳ) the low-frequency range is ascribed to the ion diffusion process in the bulk electrode, i.e., designed as Warburg component. Thus, the real part of the total impedance is described by:

where ω is the frequency and σ_w is the Warburg factor. The linear fit of Z' vs. ω^-1/2 plot provides the values of the Warburg impedance factor.

Generally, EIS results are attributed to the fact that, during the charging and discharging process, the coating layer can suppress the increase of impedance by impeding a direct contact between the highly delithiated electrode and the electrolyte, thus reducing the charge transfer resistance, R_ct. Several reports have shown the efficiency of in situ EIS experiments. Zheng et al. ^[143] have drawn considerable attention on the cycling performance at relatively high rate for Li- and Mn-rich layered electrodes. They reported the improved electrochemical performance of the AlF₃-coated Li[Li_0.12Mn_0.54Ni_0.13Co_0.13]O₂ electrode, i.e., capacity retention of 87.9% against 67.8% for pristine material after 80 cycles at C/2 rate, and the better thermal stability as well. The "buffer" layer effect provided by the AlF₃ deposit analyzed by electrochemical impedance spectroscopy (EIS) and in situ differential electrochemical mass spectrometry (DEMS) was tentatively attributed to the stabilization of the electrode/electrolyte interface by reduction of the active extracted oxygen and suppression of the electrolyte decomposition at voltages < 4.5 V vs. Li⁺/Li. Figure 11 presents the variation of the charge-transfer resistance, R_ct, as a function of number of cycles for pristine and AlF₃-coated Li[Li_0.2Mn_0.54Ni_0.13Co_0.13]O₂. R_ct of pristine material increases faster than that of the AlF₃-coated electrode upon cycling. DEMS experiments demonstrated that AlF₃ coating prevents O₂ evolution during potential scan: leakage of O₂ occurs at a voltage of 4.30 V in the pristine material, while it starts at 4.75 V for the coated electrode, beyond the oxidation of Co³⁺–Co⁴⁺. Sun et al. ^[70] reported a lower R_ct and reduced cobalt dissolution in AlF₃-coated LiCoO₂ charged at the cutoff voltage of 4.5 V. Using a 10 nm-thick AlF₃ layer, R_ct was maintained at 20–27 Ω during cycling. On the other hand, R_ct rapidly increased from 74 to 2044 Ω at the 50th cycle for pristine LiCoO₂, due to the formation of a SEI film composed of alkyl carbonate, LiF, Li_xPF_y and Li_xPF_yO_z compounds. R_ct is a function of several parameters: the amount of AlF₃ coating, i.e., the film thickness, but also the uniformity and the mesoporosity of the deposit. The same group of workers showed enhanced electrochemical performance associated with stable charge transfer resistance, which in turn limited the increase of the total cell resistance in AlF₃-coated LiNi_1/3Co_1/3Mn_1/3O₂ ^[28].

During long life cycling, an abrupt increase in charge-transfer resistance of the cathode is currently observed and it is considered as the main reason for capacity fading ^[144]. Enhancement of the stability of the interfacial resistance between the cathode and the electrolyte has been widely demonstrated by EIS. One reason for the stable cathode/electrolyte interface after AlF₃ coating is its strong resistance to HF, so that AlF₃ act as a stabilizer protecting the oxide structure from damage. Using a time-of-flight-secondary ion mass spectroscopy (ToF-SIMS), Myung et al. ^[10] showed that the coating layer works also as HF scavenger.

4.4.   Area-specific impedance

Because of the different combined factors that change the overall cell potential, the basic concept of the area-specific impedance (ASI, expressed in Ω·cm²) is an efficient test for the performance of a lithium-insertion electrode. ASI is determined from the change in the equilibrium potential as a function of the depth of discharge (DOD) after pulse of current for a given passed charge by the relation ^{[145,146,147,148,149,150]}:

where A is the cross-sectional area of the electrode and I is the current passed throughout the cell. ΔV = OCV - V_cell is the potential change during current interruption for 60 s at each DOD step. Belharouak et al. ^[147] reported an ASI of 50 Ω·cm² for a NMC333 electrode subjected to a pulse of 18 s. The LiCoO₂ coating (~20 nm thick) of Li_1.05Ni_0.35Co_0.25Mn_0.4O₂ highly improved the ASI value (38 Ω·cm² against 55 Ω·cm² measured for a cathode discharged at 90% DOD when cycled between 2.8–4.5 V) ^[148]. These results show that, during battery charging, the charge-transfer resistance is dependent of DOD. Note that ASI can be described in terms of cell polarization and power dissipation generating heat ^[151]. Figure 12 presents the variation of ASI vs. DOD of AlF₃-coated LiNi_0.5Mn_0.5O₂ as a function of the coating concentration. These results show that the lower ASI value is obtained for AlF₃-coating of 4%, which is beneficial for the long-life cycling behavior. Similar results have been reported for AlF₃-coated Li_1.2Ni_0.2Mn_0.6O₂ electrode. For 90% DOD, the ASI decreased from 37 to 30 Ω·cm² for the coated material ^[74].

4.5.   Thermal stability

Safety of lithium batteries is one of the main concerns associated to the risk of thermal runaway and battery fire ^[152]. The thermal stability is currently investigated by differential scanning calorimetry (DSC). As an example, Figure 13 shows that the Li[Li_0.2Mn_0.54Ni_0.13Co_0.13]O₂ electrode is unsafe when at charged state to 4.8 V (delithiated state). The DSC profile of the pristine material exhibits the onset temperature of thermal decomposition from at 206.6 ℃ with an associated exothermic heat of 924 J·g^-1, which is greatly reduced to 538 J·g^-1 for the AlF₃-coated electrode. Also, the thermal decomposition increased to 223.8 ℃ in the surface-modified layered material ^[143].

To avoid their individual electrochemical drawbacks LiCoO₂ and LiNi_1/3Mn_1/3Co_1/3O₂ were associated to form an AlF₃-coated (LCO/NMC) blended cathode material ^[78]. The specific discharge capacity for the blended electrodes was found to be in the range 180–188 mAh·g^-1 at a current density of 90 mA·g^-1 (0.5C rate) in the electrochemical window 3.0–4.5 V vs. Li⁺/Li, depending on the blend ratio, the best composition being LCO:NMC of 7:3.

Figure 14 shows the effect of AlF₃ coating on the thermal stability of blended LCO/NMC electrodes in the charge state, i.e., at potential of ~4.5 V vs. Li⁺/Li. For pristine LiCoO₂ and LiNi_1/3Co_1/3Mn_1/3O₂ electrodes, the DSC curves exhibit an exothermic peak at 212 and 261 ℃, which generated an exothermic heat of 2785 and 1998 J·g^-1, respectively. For both AlF₃-coated materials, the exothermic peak shifts toward high temperatures and the resulting heat is reduced. Meanwhile, the better thermal stability of the NMC333 powders is attributed to the electrochemically inactive tetravalent Mn in the layered lattice. The DSC curve of the LCO/NMC blend electrode demonstrates the smaller exothermic peaks at 227 and 277 ℃ with a total lower generated heat of 1665 J·g^-1. It is assumed that the insulating AlF₃ layer suppresses the oxygen release from the blended particles.

The cycle performance of typical AlF₃-coated cathode materials is summarized in Table 3. Although for many electrodes the intrinsic properties of the AlF₃ coating are not provided, several remarks can be made in relation with the electrochemical performances of positive electrodes for LIBs. Until then, the electrodes with layered structure (LCO, NCA, NMC and Li-rich NMC) have been the most studied to increase the capacity retention. Some spectacular results deserve to be mentioned.

The beneficial effect of AlF₃ coating on LiCoO₂ particles is twofold. First, an extended charge cut-off voltage from 4.2 to above 4.5 V vs. Li⁺/Li allows a large enhancement of the practical specific capacity due to the surface preservation of the deposit against nucleation of irreversible cubic spinel phase. Using a 10-nm thick AlF₃ deposit (2 mol%), the best performance of LCO powders is an initial specific capacity of 210 mAh·g^-1, which is retained at 190 mAh·g^-1 after 50 cycles at 0.5C rate ^[72]. Second, the amount of Co dissolution of the electrode charged at 4.5 V is limited to 18 ppm over 6 weeks. An interesting analysis of the microstructural properties versus AlF₃ content for coated LiNi_0.5Co_0.2Mn_0.3O₂ shows that the XRD peak ratio I₀₀₃/I₁₀₄ is optimum for a coating of 0.5 mol% corresponding to the minimum Li/Ni cation mixing ^[85], which improves significantly the electrochemical performance due to the better Li⁺ ion diffusion. So, this material is a good candidate for electric vehicle power batteries. Similar situation was reported for AlF₃-coated LiNi_1/3Mn_1/3Co_1/3O₂, for which a 10-nm thick AlF₃ layer deposited by the aqueous coprecipitation process remarkably enhances the capacity retention to 96% after 50 charge–discharge cycles at 0.5C rate ^[79]. As discussed before, the nanoscaled AlF₃ layer must be amorphous and homogeneous that are features obtained at a low-temperature annealing of about 400 ℃ ^[77].

One of the best results obtained so far, is from the 50-nm AlF₃ coated LiNi_0.8Co_0.15Al_0.05O₂ ^[104]. At 25 ℃, this electrode delivered 86.2% of its initial capacity over 1000 cycles at 1C rate (current density of 190 mA·g^-1) against 66.5% for the pristine materials. When tested under the high-temperature regime at 55 ℃, AlF₃-coated NCA exhibited a capacity retention of 55.9%, which is almost 5 times that of the pristine NCA electrode. Concerning the new generation of cathode materials, i.e., Li-rich NMCs, the recent work by Amalraj et al. ^[86] reports that xLi₂MnO₃-(1 - x)Li[Mn_yNi_zCo_w]O₂ electrodes coated with extremely small AlF₃ nanocrystals (5–8 nm in size, tetragonal P4nmm symmetry) exhibit a capacity fade of 25% over 200 cycles at C/5 rate and demonstrate a high Li storage capability at 60 ℃. The authors suggest that conductive species such as Li_1-xAlF_3+x and Al[FOH] are formed during the first charge process that promote the low-resistive electrode–electrolyte interface. Finally, note that several authors have investigated the electrochemical performance of AlF₃-coated electrode at temperature of 55 ℃, which is the critical test for checking the material stability ^[85,92,99].

5.   Conclusions

In this paper we have examined the beneficial effect of aluminum fluoride nano-coatings on electrochemical properties of positive electrode materials for Li-ion batteries. The role of nanoscale surface modification of cathode materials for Li-ion batteries has shown that the coating increases the energy density, improves the calendar and cycling life and enhances the thermal stability. AlF₃ is thought to be a standout amongst the most promising coating materials for the improvement in thermal stability as well as electrochemical performances, because AlF₃ can form more stable coating layer than oxides and supports faster Li⁺ deintercalation/intercalation. Deposition of AlF₃ thin layer is very easy using the popular coprecipitation method. Recently, a more sophisticated synthesis such as ALD has been used. It is believed that fast-ionic conducting solid layer LiAlF₄ can be created that favors the Li⁺-ion transport throughout the electrode/electrolyte interface. The excellent efficiency of the AlF₃ coating is highlighted for the Li-rich layered NMC oxides, which can deliver a high specific capacity superior to 250 mAh·g^-1 at moderate C-rate. Therefore, a comparison of the electrochemical performance is difficult because the particle size distribution is rarely documented. Among the specific characterization methods Raman spectroscopy, electrochemical impedance spectroscopy and differential capacity have probed the stabilization of the host structure and the electrode/electrolyte interface.

Conflict of interest

There is no conflict of interest related to this document.