Fluorine chemistry in lithium-ion and sodium-ion batteries

The role of fluorine in cathode materials

The voltage trend of transition metal-based cathode materials roughly follows the formal charges of the central atoms in the structure, consisting of the idea of the inductive effect. F has the strongest electronegativity (3.98 on the Pauling scale), thus endowing the metal fluorine (M-F) bond with a high ionic character. The presence of strong M-F bonding stabilizes the antibonding Mⁿ⁺/Mⁿ⁺¹ state through a M-F inductive effect to generate high operating voltage.

Transitional metal fluoride cathodes

Metal fluorides (denoted as MF_n) are a class of cathode materials that have attracted extensive research interests. These materials have a large specific capacity due to the possibility of multiple electron transfer at a single redox center, such as FeF₃, FeF₂ to Fe^0[9], In LIBs and SIBs, an overall chemical reaction equation is given as: nA⁺ +ne^- + MF_n → nAF + M (A denotes Na or Li, M denotes transition metals such as Fe, Cu, etc. if not specified, the same below). Specifically, both insertion and conversion reactions may happen during this progress^[10] [Figure 2A]. Simultaneously, metal fluorides typically provide a higher working potential compared to oxides and sulfides as a result of the high electronegativity of fluorine^[11] [Figure 2A]. To illustrate, in the case of LIBs, FeF₃ boasts a theoretical potential of 2.74 V (vs. Li⁺/Li)^[12], while Fe₂O₃ displays a lower theoretical potential of 1.63 V (vs. Li⁺/Li)^[13]. This highlights the superior performance of metal fluorides in terms of working potential.

Figure 2. (A) Discharge curves of FeF₃ and Fe₂O₃, which show that metal fluorides have higher operation potential than oxides. This figure is quoted with permission from Lemoine et al.^[10]. (B) False-color composite image of the converted electrode showing LiF (red) and Cu (green) phases separate from each other. This figure is quoted with permission from Wang et al.^[14]. (C) Voltage profiles of CuF₂/C (black) and NiO-CuF₂/C (red) electrodes. NiO coating reduces Cu dissolution and improves CuF₂ reversibility. This figure is quoted with permission from Seo et al.^[17]. (D) Schematic illustration of the asymmetric reaction pathway in Cu_yFe_1-yF₂. This figure is quoted with permission from Wang et al.^[18]. (E) Voltage profiles of Cu_0.5Fe_0.5F₂ for the first five cycles at 9.2 mA g^-1. This figure is quoted with permission from Wang et al.^[18].

CuF₂ was initially introduced as a cathode material for primary lithium batteries and possesses a high working potential of 3.55 V, which gives it a high energy density of 1,874 Wh kg^-1[2]. However, the lithium storage in CuF₂ is irreversible: during the first lithiation process, copper will form large isolated particles when the conversion reaction occurs, which separates Cu from LiF and isolates the electron transport path [Figure 2B]^[14,15]. Hua et al. showed that competition between Cu dissolution and Cu+2LiF→CuF₂ reaction occurs during the electrode process due to the presence of a large overpotential, thus consuming the LiF phase formed during discharge^[16]. To solve Cu dissolution, the strategy of metal oxide coating was proposed. To improve the reversibility, Seo et al. used NiO to coat CuF₂ particles (denoted by CuF₂-NiO/C) as compared to original CuF₂ (denoted by CuF₂/C)^[17]. The presence of NiO decreases the interaction between the electrode and the electrolyte, thereby moderating the dissolution of Cu, thus realizing reversible cycling of CuF₂ [Figure 2C]. Another way to improve the reversibility of CuF₂ is to construct a conductive network during its lithiation and enhance the reversibility by forming solid solutions of CuF₂ with other metal fluorides (e.g., FeF₂, NiF₂). Wang et al. constructed a ternary metal fluoride material (Cu_xFe_1-xF₂)^[18]. During the first discharge process, Cu²⁺ is first reduced to metallic Cu⁰ and forms disordered FeF₂, after which Fe²⁺ is reduced to metallic Fe⁰. During the subsequent charging, Fe⁰ is reconversion to FeF₂, then Fe partially oxidized to form a rutile-like iron fluoride framework (with Fe at a valence of Fe^2+/3+), which facilitates the nucleation and growth of Cu fluoride because of their structural similarity [Figure 2D]. The benefit led to an initial discharge capacity of 575 mAh g^-1, with the ability to keep the capacity over 450 mAh g^-1 in subsequent cycles [Figure 2E].

Wang et al. have proposed that iron-based metal fluoride forms a bicontinuous iron network during discharge^[14], which provides a large LiF/Fe interface area and a highly conductive pathway for local electron transport, thus facilitating the conversion in subsequent cycles. In contrast to FeF_x, the converted product from CuF₂ consists of larger Cu nanoparticles isolated from the LiF phase, which is mainly responsible for the lack of reversibility in CuF₂. To avoid the irreversibility issue of CuF₂, various iron-based metal fluorides with much better reversibility have been explored. The ReO₃-type FeF₃ (Figure 3A, the first one) was tested for lithium storage in 1997 with a discharging capacity of 140 mAh g^-1 for the first week in the voltage range of 4.5-2.5 V (vs. Li⁺/Li) and then exhibited a capacity of 80 mAh g^-1 in subsequent cycles^[3,19]. Until 2003, Liu et al. mechanically milled FeF₃ with different types of carbon to obtain carbon-FeF₃ nanocomposites (FeF₃@C) to improve the electrochemical properties^[13]. The material exhibited a reversible capacity of 600 mAh g^-1 at 70 °C. They also found another reversible plateau at 2 V, exhibiting a discharge specific capacity of 367 mAh g^-1. Through in-situ and ex-situ XRD analysis, the following reaction mechanisms were proposed:

Figure 3. (A) Crystal structures of FeF₃, FeF₂, FeF₃·0.33H₂O, and FeF₃·0.5H₂O. This figure is quoted with permission from Liu et al. (ReO₃-type and HTB-type)^[19], Huang et al. (Rutile-type)^[25], and Rao et al. (Pyrochlore-type)^[34]. (B) Reaction pathways for both discharging and charging of the FeF₃ electrode and their crystal structures. These summaries are based on the empirical data and DFT calculation results, where Li, Fe, and F atoms are represented by green, blue, and red spheres. This figure is quoted with permission from Li et al.^[21]. (C) Performance during cycling of FeF₂@CNT under different current densities, which shows cycling stability and rate performance of FeF₂@CNT. This figure is quoted with permission from Zhou et al.^[27]. (D) Voltage vs. capacity profiles of FeF₃·0.33H₂O electrodes for the first ten cycles with a current density of 14 mA g^-1. This figure is quoted with permission from Li et al.^[32]. (E and F) Voltage vs. capacity profiles of FeF₃·0.5H₂O cathodes wired by 5 wt % SWNTs at 0.1 C during the first few cycles (vs. Li/Na Anode, respectively). This figure is quoted with permission from Li et al.^[29].

Li⁺ +1 e^- + Fe^IIIF₃ → LiFe^IIF₃ (4.5-2.5 V vs. Li⁺/Li)

LiFe^IIF₃ + 2Li → 3LiF + Fe⁰ (2.5-1.5 V vs. Li⁺/Li)

Despite the fact that FeF₃ exhibits attractive lithium storage properties, it shows severe polarization and sluggish kinetics during cycling^[20]. Considerable efforts have been undertaken to comprehend the mechanism and, hence, mitigate these issues. Through in-situ X-ray absorption spectroscopy, in-situ transmission electron microscopy, and density functional theory (DFT) calculations, Li et al. proposed a new reaction mechanism: FeF₃ → Li_0.25FeF₃ ↔ LiF + FeF₂ ↔ Fe + 3LiF [Figure 3B]^[21]. In this process, the uneven distribution of the electrochemically active phases brought about a serious ohmic voltage drop. To further, Hua et al. found that FeF₃ undergoes a three-phase reaction upon Li⁺ insertion, generating a small amount of FeF₂ through a replacement-like mechanism^[22]. During the deeper lithiation process, Fe³⁺ is gradually reduced to Fe²⁺ as Li⁺ is inserted, which is accompanied by the migration of Fe to form FeF₂. The reaction mechanism is closely related to the size of FeF₃ in this process: the FeF₃ at the micrometer scale undergoes a process that involves a two-phase reaction, whereas FeF₃ at the nanoscale displays a mechanism based on a solid solution, suggesting the irreversibility of FeF₃ during charging and discharging may be mitigated by rationally designing FeF₃ nanostructures and reducing the Fe migration distance. FeF₃ nanoparticles (40-106 nm) were prepared in carbon nanocages (40-110 nm) by Yang et al. The results showed that the FeF₃ nanoparticles provided an excellent reversible capacity of 410 mAh g^-1 over 120 cycles at 100 mA g^-1[23].

FeF₂ has received attention as an intermediate product of the charge/discharge process of FeF₃. In rutile-type FeF₂, the three-dimensional (3D) network of FeF₆ octahedra is described from trans-chains of edge-sharing octahedra along [001]. The chains are further linked together by opposite vertices (Figure 3A, the second one)^[24,25]. FeF₂ also has an excellent theoretical specific capacity (571 mAh g^-1), but it shows very poor cycling performance, with a severe decrease in capacity after only five cycles^[26]. Its poor cycling performance can be attributed to two main factors. Firstly, the severe volume change that FeF₂ undergoes during the conversion reaction can cause instability. Secondly, FeF₂ dissolves easily in the electrolyte, making it flake off from the current collector. This detachment can disrupt the operation and overall performance. It has been shown that the cycling performance of FeF₂ in microporous/mesoporous carbon improves dramatically. Zhou et al. encapsulated FeF₂ in carbon nanotubes (CNTs)^[27]. The CNTs endure the stress brought on by the volume alteration during Li insertion to preserve the shape of the material and secure the contact between the material, the current collector, and the conductive carbon, which enhances the electrochemical performance of FeF₂ [Figure 3C]. In order to reduce the dissolution of FeF₂, Xiao et al. used lithium bis(fluorosulfonyl)imide (LiFSI)-based electrolytes to form a protective layer in situ to mitigate the dissolution of the cathode, which successfully enabled the Li-FeF₂ battery to exhibit more than 1,000 stable cycles^[28]. Quantum chemical calculations show that the decomposition of LiFSI at the cathode side produces a LiF-rich protective layer.

Another strategy to extend the application of FeF₃ is incorporating structural water. The crystal structure of FeF₃ changes significantly when it contains crystalline water, and several special structures can even accommodate the insertion of Na^+[29]. Hexagonal bronze tungsten (HTB) type FeF₃·0.33H₂O, which has FeF₆ octahedra connected by shared F atoms to form a hexagonal cavity, forms a tunnel structure along the [001] direction (Figure 3A, the third one)^[19,30]. In the tunnel, water molecules occupy 1/3 of the cavity position, and the remaining space can accommodate ion insertion and extraction. This unique structure facilitates rapid ion transport in HTB-type FeF₃^[19,30]. Conte et al. monitored the reaction by Mossbauer spectroscopy to unveil the reaction mechanism^[31]. Their findings implied that during the first discharge, the reduction of Fe³⁺ to Fe²⁺ occurs through two different routes: one portion maintains the original HTB structure, while another portion has a disordered rutile structure. The rutile phase is then restored to the HTB phase during charging. When the voltage is below 2 V, the Fe phase appears, which increases the capacity but decreases the cycling stability. DFT calculations also show that water molecules in FeF₃·0.33H₂O are necessary for structural stability, and only 0.66Li⁺ can be inserted to form Li_0.66FeF₃·0.33H₂O^[30]. In 2010, Li et al. reported a non-aqueous low-temperature synthesis of HTB-FeF₃ based on ionic liquids, which exhibited a capacity of 154 mAh g^-1 in the first ten cycles, with a typical charge/discharge plateau observed at around 2.7 V without significant polarization during the cycle [Figure 3D]^[32]. Zhang et al. packaged FeF₃·0.33H₂O in 3D ordered mesoporous carbon (3D-OMCs) as the cathode material for SIBs, and the first discharge capacity was 386 mAh g^-1 at a current density of 20 mA g^-1, and the capacity was still maintained at 201 mAh g^-1 at a current density of 100 mA g^-1[33]. Such excellent performance was partly attributed to the 3D mesoporous structure of 3D-OMCs, which provides a fast electron diffusion path for Na⁺ insertion and extraction. Additionally, the presence of 3D-OMCs alleviates the drastic volume change of FeF₃·0.33H₂O during charging and discharging.

The one-dimensional channels of HTB-type FeF₃·0.33H₂O indeed facilitate ion transport. However, the single ion channels are easily partially blocked by water molecules, which limits the application of HTB-type FeF₃ in SIBs. When the water of crystallization in FeF₃ increases to 0.5, we get pyrochlore-type FeF₃·0.5H₂O, whose structure becomes similar to that of AlF₃·0.5H₂O. FeF₆, as the basic unit in the crystal, is connected to form hexagonal cavities by shared F atoms, and these cavities are connected to form a 3D porous structure (Figure 3A, the last one)^[29,34]. This structure has a larger cell volume (~1,130 ų) than HTB-FeF₃ (710 ų) and ReO₃-type FeF₃ (310 ų). At the same time, the interconnected 3D channels exhibit a higher pore density, which eliminates the need for stringent considerations in terms of pore size, especially for ions with a large radius, such as Na^+[29]. The water molecules in them are limited by the zigzag channels to have less effect on the ion storage. Li et al. obtained 10 nm FeF₃ nanospheres with graphene coating in the presence of single-walled CNTs (SWNT), Fe(NO₃)₃·9H₂O, and the ionic liquid, BmimBF₄ or C₁₀mimBF₄^[29]. The material exhibited excellent lithium storage performance (143 mAh g^-1 at the first discharge, and the capacity was maintained at 135 mA g^-1 for the next 300 cycles. Figure 3E) and sodium storage performance (250 mAh g^-1 at the first discharge, and the capacity was maintained at 150 mAh g^-1 after 50 cycles. Figure 3F). The charging and discharging curves show a high degree of symmetry. Beneficial from the open 3D channel, pyrochlore-type FeF₃·0.5H₂O exhibits a much smaller polarization than ReO₃-type FeF₃. Jiang et al. proposed a pyrochlore-type cathode material, Fe₂F₅·H₂O, for SIBs. This material, benefiting from its large cell volume and interconnected, open 3D channels, facilitates rapid and reversible insertion/extraction of Na^+[35]. The first discharge capacity reached 189.4 mAh g^-1 at a current density of 20 mA g^-1, and it retained a capacity of 147.3 mAh g^-1 after 30 cycles.

In addition to Fe-based fluorides, Ti, Mn, and Co fluorides were tested for LIBs and SIBs, but none of them can be recharged after the initial discharge^[13]. The electrochemical properties of metal fluoride cathodes are summarized in Table 1. Iron-based fluorides show very high specific capacity during charging and discharging, but the occurrence of the conversion reaction will seriously affect the reversibility of the material, and limiting the discharge cutoff voltage to above 2 V can improve the cycling performance but will reduce the energy density, which requires us to make reasonable trade-offs in practical applications. As it does not contain lithium/sodium ions, it needs to be pre-treated when combined with graphite and hard carbon to form a full battery.

Fluorinated polyanionic cathodes

Metal fluorides have great capacity but are limited by their extremely low conductivity and slow kinetics, bringing difficulties in practical applications. In order to take advantage of the high redox potential brought by the high electronegativity of fluorine while waiving the slow kinetics of conversion reactions, researchers have turned their attention to fluorine-based polyanionic compounds. Polyanionic materials exhibit good thermal stability^[36,37]. In polyanionic cathode materials, a number of anionic groups are reported ((XO₄)_m^n-, X = P, S, Si, As, etc.)^[38,39]. The elevation of the electronegativity of the X atom in the anionic group enhances the ionic character of the redox metal species, thereby increasing the redox potential, which allows us to modulate the working voltage of a material by changing the anionic group type or by combining multiple anionic groups^[40]. lithium-/sodium- fluorophosphates (A_xM_y(PO₄)_mF_n) and fluorosulfates (A_xM_y(SO₄)_mF_n) are the most typical fluorine-based polyanionic compounds in LIBs and SIBs, known for their high operating voltage and excellent thermal stability.

Lithium-/sodium- fluorophosphates (A_xM_y(PO₄)_mF_n)

Exploring fluorophosphate chemistry, LiVPO₄F (LVPF) was first introduced as a cathode material for LIBs by Barker et al. in 2003^[4]. As shown in Figure 4A, LiVPO₄F has a triclinic structure consisting of VO₄F₂ octahedral chains that are interconnected by PO₄ groups to form a 3D framework, of which 18% of the Li occupies the penta-coordination sites (LiO₅) while the remaining 82% occupies the octahedral sites (LiO₆)^[41]. During the two-electron transfer (V⁴⁺ to V²⁺), there is a 15.9% volume change, exhibiting two distinct redox plateaus observed at 4.2 V and 1.8 V (vs. Li⁺/Li, Figure 4B)^[42].

Figure 4. (A) Crystal structures of LiVPO₄F, NaFePO₄F, and Na₂FePO₄F. (B) Electrochemical curves of LiVPO₄F/C composite cycled vs. Li⁺/Li starting in discharge show Li-ion insertion/extraction potential (as shown by the red arrow). This figure is quoted with permission from Ellis et al.^[42]. (C) The characteristic discharge profiles of the 5 wt% ZrO₂ coated Li₂CoPO₄F cathode materials between 2 and 5.2 V at various rates. This figure is quoted with permission from Amaresh et al.^[48]. (D) The charge-discharge curves of NaVPO₄F obtained from different synthesis processes at 15 mA g^-1. This figure is quoted with permission from Xu et al.^[49]. (E) Charge-discharge curves of Na₃V₂PO₄F₃/C. This figure is quoted with permission from Zou et al.^[52]. (F) The initial charge/discharge profile of Li_0.995K_0.005VPO₄F at 0.12 C (1 C = 156 mA g^-1) in the potential range of 2.5-4.6 V (vs. Li⁺/Li). The inset shows the CV curves at a scanning rate of 0.1 mV s^-1. This figure is quoted with permission from Wu et al.^[58].

As analogs to LiVPO₄F, LiFePO₄F and LiTiPO₄F share isostructural structures to LiVPO₄F^[43]. LiFePO₄F can exhibit an insert (extract) Li⁺ potential of 2.71 V (vs. Li⁺/Li) with a first discharge capacity of 129 mAh g^-1 at a current density of 150 mA g^-1[44]. LiTiPO₄F demonstrates two reversible charge/discharge plateaus at 2.8 V and 1.7 V (vs. Li⁺/Li) and can provide a capacity of 157 mAh g^-1[45]. However, their practical application is hindered by the limited energy density caused by a low operating voltage. As fully discharged states (Li-rich states), another type of material known as Li₂MPO₄F (where M can be Mn, Fe, Co, or Ni) holds the potential to reach high energy density by accommodating more than one Li⁺ ion on a single transition metal^[46,47]. In particular, Li₂CoPO₄F exhibits a high discharge plateau above 4.5 V and good performance (109 mAh g^-1) at 2.0-5.5 V (vs. Li⁺/Li, Figure 4C)^[48].

Lithium-/sodium- fluorophosphates are also widely used in SIBs. Similar to LiVPO₄F, NaVPO₄F is used as a cathode material in SIBs. NaVPO₄F has a monoclinic structure with two PO₄ tetrahedra, sharing two O atoms connected to two different VO₄F₂ octahedra^[49]. When used as the cathode for SIBs, NaVPO₄F synthesized by hydrothermal method showed the best performance, delivering a capacity of ~140 mAh g^-1 (at 0.1 C) and a discharge plateau of 3.4 V [Figure 4D]^[49]. In its fully discharged state, Na₂FePO₄F belongs to the orthorhombic crystal system, which has a layered structure with two different Na sites consisting of FeO₄F₂ and Fe₂O₇F₂ octahedra^[43,50]. This material can exhibit efficient Na⁺ insertion and extraction at 3.0 V, with an initial discharge capacity of 120 mAh g^-1. Besides, Na₂MPO₄F (M = Co, Mn) with the isostructural structure was also developed^[51]. Zou et al. synthesized pure Na₂CoPO₄F by spray drying methods, which exhibit a voltage plateau up to 4.3 V and a discharge capacity of 107 mAh g^-1[52]. Wu et al. synthesized carbon-coated hollow spheres Na₂MnPO₄F using the same method, which could observe a plateau of 3.6 V at room temperature and an initial discharge capacity of 102.4 mAh g^-1 at a current density of 15 mA g^-1[53]. As another sodium fluorophosphate with different stoichiometry, Na₃V₂(PO₄)₂F₃ is a 3D network consisting of [V₂O₈F₃] octahedra and [PO₄] tetrahedra. Upon the extraction of two Na⁺ ions from the framework of Na₃V₂(PO₄)₂F₃, the material demonstrates dual voltage plateaus at values of 3.7 V and 4.2 V, indicating a theoretical energy density of up to 507 Wh kg^-1 [Figure 4E]^[54]. Successful extraction of three Na⁺ can be achieved when the voltage is extended to 5 V, but this process causes partial structural degradation^[55].

The inherently low conductivity discussed above greatly limits the practical application of fluorophosphate materials. In order to solve this obstacle and improve the electrochemical performance, three major strategies with respect to nanostructure engineering, surface coating, and metal doping are developed. Law et al. synthesized nano-sized NaVPO₄F with particle sizes ranging from 300-800 nm by the soft template method^[56]. Galvanostatic intermittent titration techniques (GITT) revealed that the range of the solid-phase diffusion coefficient D_Na was between 2.0 × 10^-10-2.5 × 10^-13 cm² s^-1. This diffusion coefficient allows the material to be cycled 10,000 times at 10 C and 20 C with capacity retention up to 81% and 87% (1 C = 142 mA g^-1). To increase the cycling stability, surface coating is a feasible way to limit the reaction between the cathode material surface and electrolyte. Amaresh et al. coated ZrO₂ on the surface of high-voltage Li₂CoPO₄F, and this coating reduced the decomposition of the electrolyte, resulting in an initial discharge capacity of up to 144 mAh g^-1 (10 mA g^-1)^[48]. Metal doping is also an effective method to improve cycling performance, and this method can alter the electronic structure within the material^[57]. Various metal substitutions, such as Mg, Cr, Al, and Co, have been unveiled to improve the cycling stability of fluorophosphates. Wu et al. compared bare LiVPO₄F with K-doped Li_0.995K_0.005VPO₄F and found that the K-doped sample exhibited a higher discharge plateau, lower polarization, and lower interfacial impedance^[58]. Specifically, it obtained a capacity of 130-140 mAh g^-1 at 15 mA g^-1 and an average potential of 3.6 V (vs. Li⁺/Li,Figure 4F). The almost overlapping charge/discharge curves of the first few cycles demonstrate the high reversibility of the material.

Lithium-/sodium- fluorosulfates (A_xM_y(SO₄)_mF_n)

Another promising catalog of the family of fluorosubstituted polyanionic compounds, fluorosulfates, is considered as higher voltage cathode materials for batteries due to the enhanced inductive effect of sulfate than phosphate. Tavorite-type LiFeSO₄F, reported as a cathode material for LIBs, has a triclinic crystal structure consisting of shared-angle FeO₄F₂ octahedral chains oriented alternately along the c-axis. These chains connect isolated SO₄ tetrahedra to form a 3D framework and form diffusion channels along the [110], [010], and [101] directions [Figure 5A]^[59-61]. In LIBs, it exhibits a reversible capacity of 130-140 mAh g^-1 (at 15 mA g^-1) and an average potential of 3.6 V [Figure 5B]. In addition to Tavorite-type LiFeSO₄F, triplite-type LiFeSO₄F is also a reported cathode material for LIBs. Unlike tavorite-type LiFeSO₄F, triplite-type LiFePO₄F exhibits co-edge connections between FeO₄F₂ octahedra and LiO₄F₂ octahedra. This connection changes the conformation of F atoms around Fe, causing electrostatic repulsion, thereby making triplite-type LiFeSO₄F exhibit higher operating voltage (3.9 V vs. Li⁺/Li). A similar structure of triplite-type NaFeSO₄F was also investigated for its sodium storage properties^[5]. In contrast, NaFeSO₄F exhibits a smaller tunnel angle of 75 (85) and a larger tunnel size due to the larger radius of Na^[5,60]. NaFeSO₄F has a high operating voltage (3.7 V vs. Na⁺/Na) but exhibits a coulombic efficiency of about 78% and an energy efficiency of 64% during cycling [Figure 5C], which is mainly due to the severe distortion of the crystal structure occurring upon Na-ion insertion^[62]. Notably, unlike the 3D tunnel of LiFeSO₄F, it has been demonstrated that NaFeSO₄F conducts Na-ions in a one-dimensional manner through the [101] tunnel. Not only Fe-based fluorosulfates but also fluorosulfate materials based on other transition metals have been examined, such as tavorite LiCoSO₄F, triplite LiMnSO₄F, and sillimanite LiZnSO₄F^[63]. Unfortunately, none of them are electrochemically active.

Figure 5. (A) Crystal structures of tavotite and triplite-type LiFeSO₄F and triplite-type type NaFeSO₄F. This figure is quoted with permission from Ati et al.^[59]. and Tripathi et al.^[60]. (B) The tavorite LiFeSO₄F has a stable cycling performance. Charge/discharge profile of LiFeSO₄F cells cycled between 2.5 and 4.2 V at 15 mA g^-1. This figure is quoted with permission from Recham et al.^[61]. (C) Charge and discharge profiles for triplite-type NaFeSO₄F discharge at 0.01 C (1C = 138 mA g^-1). This figure is quoted with permission from Kim et al.^[62]. (D) a-LiFeSO₄F shows a different charge/discharge mechanism compared to tavotite and triplite-type LiFeSO₄F. dQ/dV profile of a-LiFeSO₄F (blue), tavorite (green), and triplite (orange). This figure is quoted with permission from Heo et al.^[64]. (E) Cycling performance of a-LiFeSO₄F at different temperatures measured at a current density of 40 mA g^-1. This figure is quoted with permission from Heo et al.^[64].

Besides the above-mentioned crystalline fluorosulfides, amorphous fluorosulfates have also received attention from researchers. Heo et al. reported a-LiFeSO₄F, a stable and reversible amorphous fluorinated iron sulfate electrode^[64]. During the discharge process, the amorphous LiFeSO₄F consists of two processes that can be represented by the following equation:

a-FeSO₄ + Li⁺ → LiFeSO₄ (Intercalation reaction at ~3.59 V vs. Li⁺/Li)

b-LiFeSO₄F + 2Li⁺ + 2e^- → Fe + LiFeSO₃F + Li₂O (Conversion reaction at ~2.18 V vs. Li⁺/Li)

This mechanism is also evidenced by the dQ/dV profile [Figure 5D], where sharp redox peaks can be observed at 3.6 V and 3.9 V for tavorite- and triplite-type LiFeSO₄F, but a-LiFeSO₄F does not display similar peaks in this interval; instead, it exhibits a clear redox peak at 2.2 V and a capacity of 300 mAh g^-1 at 25 °C (current density: 40 mA g^-1) without significant capacity attenuation in 200 cycles [Figure 5E]. The superior cycling stability can be credited to the distinctive amorphous structure of a-LiFeSO₄F that can remain intact after the conversion reaction and serve as the main host for subsequent Li⁺ insertion.

The electrochemical properties of fluorinated polyanionic compounds are summarized in Table 2. It can be seen that these electrodes generally have high operating voltages. Additionally, they also have good cycling performance and thermal stability due to their stable structural framework.

Table 2

Electrochemical performance of fluorinated polyanionic cathodes

Materials	Structure	Potential (V vs. Li+/Li or Na+/Na)	Theoretical capacity (mAh g-1)	Theoretical energy density (Wh kg-1)	Discharge capacity (mAh g-1)
LiVPO4F[41]	Tavorite	4.2	237	995	106
LiFePO4F[44]	Tavorite	2.71	151	409	129
LiTiPO4F[45]	Tavorite	2.8/1.7	161	326	149
Li2CoPO4F[48]	Orthorhombic	4.5	143	643	91
LiFeSO4F[61]	Tavorite-type	3.6	150	542	130
a-LiFeSO4F[64]	Amorphous	2.5 V	454	1,135	300
NaVPO4F[49]	Tavorite	3.4	142	482	121
Na2FePO4F[43]	Orthorhombic	3.0	124	372	121
Na2MnPO4F[43]	Orthorhombic	3.6	124	449	99
Na3V2(PO4)2F3[55]	Tetragonal	3.7/4.2	128	507	110
NaFeSO4F[62]	Triplite	3.5 V	138	483	83

Functionalized fluorinated electrolytes

Functionalized fluorinated electrolytes could extend the application scenarios of rechargeable batteries, such as high/low temperature operation, high safety, high power, etc., as discussed below:

Previous studies have shown that fluorinated solvents can effectively enhance the low-temperature performance of electrolytes. The substitution of fluorine typically lowers the freezing point of most solvents, enhancing the conductivity of the electrolyte during cold temperatures. This, in turn, improves the battery performance under low-temperature conditions. Fluorine, as a strong electron absorbing group, weakens the electron-giving ability of the solvent/co-solvent when there is a fluorine substitution on them. Fluorine-substituted solvent/co-solvent molecules have lower binding energies (in absolute terms) when coordinated with Li⁺, which means that introducing fluorine into solvent/co-solvent molecules can make it easier to detach Li⁺ from the solvation structure^[65]. Yang et al. proposed an ethyl trifluoroacetate (ETFA)-based electrolyte^[66]. Due to the high electronegativity of the fluorine, the binding energy of this solvent to Li⁺ is much lower than that of ethylene carbonate (EC), which allows the rapid de-solvation of Li⁺ from the solvation structure. Owing to the fast de-solvation process, Li||LTO cells with this electrolyte achieve capacities of 154 mAh g^-1 and 78 mAh g^-1 at -50 °C and -70 °C, respectively [Figure 6A].

Figure 6. (A) Charge/discharge profiles of LTO||Li cells using the ETFA electrolyte under different temperatures. This figure is quoted with permission from Yang et al.^[66]. (B) Schematic diagram of fluorine flame retardant principle; (C) Charge/discharge profiles of the LiNi_0.8Co_0.15Al_0.05O₂/Li half cells at various temperatures, which demonstrates the excellent high temperature stability of fluorinated solvents. This figure is quoted with permission from Shiga et al.^[68]. (D) Nail penetration tests of the 5 Ah graphite||NMC pouch use LiPF₆-EC/EMC (left) and coordination number-rule electrolyte. This figure is quoted with permission from Chen et al.^[69]. (E) The linear sweep voltammetry curves of the electrolyte without and with F-EPE, which shows fluoride electrolytes could operate under high potential. This figure is quoted with permission from Luo et al.^[73].

Introduction of fluorinated molecules into the electrolyte also could extend the high temperature adaptability. Generally, organic molecules become less combustible as the degree of F substitution increases, which makes the electrolyte safer^[67]. In addition, F atoms also have a strong attraction to H atoms, which allows F to absorb H radicals during combustion and quench the chain reaction, increasing the flash point and flame retardant properties of the solvent [Figure 6B]. An ultra-concentrated electrolyte composed of LiFSA and fluorinated alkyl phosphate was prepared by Shiga et al. This electrolyte was self-extinguishing and allowed reversible insertion of Li⁺ into graphite, and the electrolyte also exhibited ultra-high thermal stability. Such features allow graphite/Li half-cells with this electrolyte to operate at temperatures exceeding 100 °C [Figure 6C]^[68]. Our group has proposed an electrolyte using trifluoroethyl phosphate and trifluoroethyl methyl carbonate as solvents. After nail penetration, pouch cells using this electrolyte still keep compact, while pouch cells using commercial 1m LiPF₆-EC/EMC (EMC: Ethyl Methyl Carbonate) electrolytes immediately catch fire [Figure 6D]^[69]. Notably, due to the advantages of the F-substitute solvent described earlier, the pouch battery using the electrolyte can also operate stably within a temperature range of -40-60 °C.

The electron-withdrawing inductive effect results in a decrease in the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) energy levels when fluorine is introduced into the solvent molecule^[70,71]. The decrease of the HOMO energy level enhances the anti-oxidative properties of the molecule, which enables this substance to be compatible with high potential cathodes, thus meeting the high-energy requirements^[72]. Luo et al. expanded the electrochemical window by adding 10% of fluorinated 3-(1,1,2,2-tetrafluoroethoxy)-1,1,2,2-tetrafluoropropane (F-EPE) solvent to 1.2 M LiPF₆ in EC/EMC electrolytes^[73]. With the introduction of F-EPE, the potential window of the electrolyte extends above 5.2 V [Figure 6E]. This high potential window allows Li||LiNi_0.5Mn_1.5O₄ cells to have higher stability and coulombic efficiency.

Fluorinated interface engineering

Briefly, the essence of the formation of the EEI is a redox reaction between the electrode and the electrolyte^[8]. Taking LIBs as an example, SEI has many components, such as Li₂O, LiOH, Li₂CO₃, Li₃N, etc.^[74,75]. Among them, LiF has attracted a lot of attention due to its excellent performance. As a SEI component, LiF has a wide band gap (8.70 eV) to effectively prevent electron transport^[76]. At the same time, the high mechanical strength and low solubility of LiF make the LiF-rich interface extremely stable^[77,78]. The safety of the battery can be enhanced through the efficient inhibiting of dendrite growth on the electrode surface by the high surface energy and low diffusion energy of LiF-rich SEI^[79].

HOMO and LUMO directly reflect the redox stability of the electrolyte. Meanwhile, both the cathode and anode have their own electrochemical potentials^[80]. If the anode potential is higher than LUMO, then the electrons on the anode will tend to transfer to the electrolyte, thus causing the electrolyte to reduce, creating a solid-electrolyte interface and preventing the reaction from proceeding further. A similar situation occurs at the cathode, creating a cathode-electrolyte interface (CEI) [Figure 7A]^[81]. In addition to preventing the electrolyte from further decomposition, the current study also found that a fluorinated interface has a positive role in protecting the electrodes and promoting the uniform deposition of Li [Figure 7B]^[82].

Figure 7. (A) Schematic diagram of the mechanism of SEI and CEI formation. This figure is quoted with permission from Goodenough et al.^[81]. (B) Schematic of how LiF-rich SEI prevents the growth of dendrites. This figure is quoted with permission from Choudhury et al.^[82]. (C) The schematic shows that when heated, the fluoropolymer, CYTOP, gradually breaks down to release pure F₂ gas. This gas then reacts with lithium metal or Li_xSi, resulting in a uniform and dense coating of LiF. This figure is quoted with permission from Zhao et al.^[84]. (D) Long-term cycling stabilities of anode at a current rate of 0.2 A g^-1 in 1m NaClO₄ electrolyte in PC without and with 2 vol.% of FEC additive. This figure is quoted with permission from Wen et al.^[88]. (E) Coulombic efficiency of Li||Cu battery at a current density of 0.10 mA cm^-2. This figure is quoted with permission from Zhang et al.^[86]. (F) Discharge capacities of the Li/LCO cells cycled at 1.0 C (1 C = 274 mA g^-1). This figure is quoted with permission from Lee et al.^[85].

Many efforts have been made by researchers to obtain stable fluorinated interfaces. The straight ward method to get robust EEI is pretreatment by ex-situ coating a layer of LiF on Cu or Li to improve electrochemical performance. Zhang et al. plated LiF on the surface of Cu collectors^[83]. LiF-rich SEIs guided the deposited Li into a stabilized structure, which greatly contributed to the stability of such anodes. Zhao et al. developed a method to form LiF films using the fluoropolymer CYTOP as a precursor to generate F₂ in situ [Figure 7C]^[84]. This compound can release F₂ at less than 250 °C, thereby generating a dense LiF coating on the anode. This film is able to minimize the interaction between lithium metal and carbonate electrolyte, preventing the creation of dendrites and leading to 300 cycles without dendrite formation at a current density of 5 mA cm^-2.

The second way to generate a fluorinated interface is from organic components (either organic additives or solvents). As we mentioned before, fluorine substitution reduces the LUMO energy levels of the components in the electrolyte, which makes the fluorinated electrolyte components easier to reduce and thus generate stable SEIs^[70]. In addition, theoretical calculations also show that fluorinated solvent molecules that are in the primary solvated shell layer of the Li cation are more easily reduced^[8]. EC is the most widely used solvent commercially, and its fluorine-substituted product, fluoroethylene carbonate (FEC), is the most widely used film-forming additive for a variety of batteries, including LIBs, LMBs, and SIBs^[85-87]. Wen et al. prepared hollow carbon nanospheres as anodes for SIBs and achieved a specific capacity of 89.9 mAh g^-1 at 5 A g^-1 by introducing 2 vol.% FEC into the electrolyte [Figure 7D]^[88]. In LMBs, FEC can also form a dense and stable protective film on the lithium metal surface. Zhang et al. obtained a homogeneous SEI by introducing a small amount of FEC (5%) into the EC-based electrolyte^[86]. This homogeneous SEI significantly improved the Coulombic efficiency of the Li||Cu half-cells (Coulombic efficiency of 98% within 100 cycles, Figure 7E). Matching the FEC-protected lithium metal with LiNi_0.5Co_0.2Mn_0.3O₂ resulted in an initial capacity of up to 154 mAh g^-1. Silicon is considered a promising anode for the new generation of high-specific-energy LIBs. However, it undergoes significant volume changes when it charges and discharges. Researchers found that adding FEC to the electrolyte could generate smoother, more stable SEI on the Si anode, and subsequent studies have further demonstrated that FEC in the electrolyte can form a fluorinated interface on the Si anode or Si-graphite, thus preventing further side-reactions during the volume change^[89,90].

To improve the effectiveness of FEC, the researchers increased the amount of FEC in the electrolyte. Lee et al. proposed a series of FEC, DEC-based electrolytes with different FEC contents^[85]. By introducing FEC, the reversibility of the Li anode was significantly improved. This electrolyte effectively improved the coulombic efficiency and cycling performance of Li/LiCoO₂ (LCO) batteries (30% FEC, capacity retention of 97.5% for 900 cycles, Coulombic efficiency: 99.0%, Figure 7F). However, the amount of FEC is not the more, the better; when the FEC content reaches more than 50%, the battery efficiency and capacity will produce a significant decrease, which is due to the formation of a resistive layer on the surface of LCO by FEC and, thus, significantly reduce the discharge capacity.

The third strategy for achieving a fluorinated interface comes from the inorganic component. The salt is another important component in the electrolyte. It has been shown that most of the F-containing lithium salts (e.g., LiPF₆, LiBF₄, LiFSI, etc.) can involve the F in them in the generation of LiF-rich SEI^[91-93]. The decomposition of these salts produces an electrodeless fluorinated interface with excellent physical and chemical properties. Chen et al. designed an electrolyte [2.0 m LiPF₆ in 1:1 v/v mixture of tetrahydrofuran and 2-methyltetrahydrofuran (mixTHF)]^[76]. The use of ether with a low thermodynamic reduction potential (0.0-0.3 V vs. Li⁺/Li) as a solvent allowed PF₆^- to become preferentially decomposed relative to being in a carbonate-based electrolyte [Figure 8A-C]. The SEI produced by the decomposition of inorganic salts greatly improves the cycling stability of the battery.

Figure 8. (A) Schematic of three general categories of solvent-salt interactions. (B) Distribution of the Li⁺ solvates for the LiPF₆-THF (2.0 M) from MD simulations. This figure is quoted with permission from Chen et al.^[76]. (C) Reduction potentials of key electrolyte components with the first Li⁺ solvation shell, which shows that fluorinated salts will decompose preferentially. This figure is quoted with permission from Chen et al.^[76]. (D and E) Calculated electronic structures (projected density of states, pDOS) of dilute D and concentrated E LiTFSA/AN electrolytes, which suggests a gradual transfer of LUMO to anion with increasing concentration. This figure is quoted with permission from Yamada et al.^[94]. (F) Charge/discharge profiles of Na||HC(hard carbon) cells using the concentrated NaFSA/TMP electrolyte. This figure is quoted with permission from Wang et al.^[95]. (G) Cycling performance with different electrolytes at 2 C and 25 °C. This figure is quoted with permission from Piao et al.^[96].

When the salt concentration is increased, the anion decomposition tendency will be improved. DFT- molecular dynamics (MD) simulations illustrate that the LUMO energy level of the electrolyte gradually shifts from the solvent to the anion when the salt concentration is gradually increased [Figure 8D and E]^[94]. Wang et al. developed highly concentrated electrolytes for LIBS and SIBs, using trimethyl phosphate (TMP) as solvents and NaN(SO₂F)₂ (NaFSA) or LiN(SO₂F)₂ (LiFSA) as salts^[95]. Both electrolytes showed excellent performance (almost no degradation for 1,000 cycles, Figure 8F) when the NaFSA and LiFSA concentrations were increased to 3.3 m and 5.3 m, respectively.

The excellent SEI brought by high concentration electrolytes can significantly improve battery performance, but the higher viscosity and cost greatly limit the practical application of high concentration electrolytes. To address these two drawbacks, researchers have proposed an electrolyte known as localized high concentration electrolyte (LHCE). By adding one or more diluents that are immiscible with salt but miscible with solvents to the electrolyte, the electrolyte is saturated at a low concentration and exhibits low viscosity and ionic conductivity. Piao et al. added 1,1,2,2-Tetrafluoroethyl-2,2,3,3-tetrafluoropropylether (TTE) as a diluent to LiFSI/DMC electrolyte to form LHCE^[96]. MD simulations revealed that the introduction of diluent enhances the interaction between Li⁺ and FSI^- so that the FSI^- can be preferentially reduced to form LiF-rich SEI. This electrolyte enables stable cycling of Li||NMC622 cells at both 25 and 60 °C [Figure 8G].