A review on the application of iron(III) fluorides as positive electrodes for secondary cells
- 3.5k Downloads
Among the many industrial and research level applications of fluorides like microelectronics, polymers, agronomy, imaging, dental composites, uranium enrichment processes or catalysis, there is a recent and growing interest for these ionic compounds in the field of energy storage and conversion. Li-ion technology in particular, has been attracted to fluorides due to the possibility of more-than-one electrode reaction. In fact, these reactions can potentially enhance the energy stored in commercial cells (conversion reactions). By reaction with lithium, transition metal fluorides such as MnF3, FeF3, CoF3 or CuF2, can be reduced to the respective zero valent metal/LiF mixture with resulting high theoretical capacities ranging from 3 to 4.2 times the amount currently delivered by commercial LiFePO4 (170 mAh g−1). Since, in practice, the most of the research efforts are being put into iron trifluoride due to chemical (relatively mild synthesis conditions), electrochemical (lower polarization than MnF3 for example) and environmental (Fe is more environmentally friendly than Co) reasons, in this review we will focus on the electrochemistry, synthesis strategies and amelioration techniques for the high oxidation state ferric fluoride and its derivatives.
KeywordsIron fluoride Li-ion Batteries
Metal fluorides are ionic compounds that have been known since the late Renaissance era when they were used to render aluminum slags less viscous during the smelting process (CaF2 fluxes) . More recent applications, both at the industrial and research levels include: etching of silicon wafers for microelectronics, production of fluorinated polymers, agronomic compounds, medical imaging products and dental composites for caries control, uranium enrichment processes, catalysis and energy storage and conversion [2, 3, 4].
In the latter field, in particular, fluorides have attracted attention due to the conversion reaction with lithium ions. In conversion reactions, the transition-metal, M, of an initial MX species (X = H, N, O, F, P or S) is reduced by lithium to give LiX and the pure metal . Since the first report of such kind of reaction in oxides , it has been clear that the involved materials could greatly enhance the energy stored in Li-ion cells. The reason for such high energy densities lies in the possibility for the transition metal to pass through various oxidation states involving more than one electron. For example, put aside the reaction potential, FeF3 would reach 712 mAh g−1 for its complete 3 electron reaction, while in LiFePO4 only the Fe(III)/Fe(II) couple will be active, delivering a theoretical maximum of 170 mAh g−1 (4.2 times less than iron trifluoride).
Recently, more and more research targets the optimization of high energy anode materials (Si, Sn, oxides, etc.) . Some prototypes could be brought to the market in the near future. However, at the current state of research, cathodes still represent the main cell limitation . In fact it is not possible to assemble a performing full cell using a high energy anode if the cathode material cannot match that energy. Considering a cell that would be assembled with a Si anode, a high amount of lithium can theoretically react delivering around 4,000 mAh g−1 (Si + 4.4Li+ + 4.4 e- → Li4.4Si). If, in a purely speculative way, this high capacity can be reached and sustained for several hundreds of cycles, such an achievement would be completely useless when the coupling would be done with LiFePO4-based cathodes for example. In fact since in full Li-ion cells, cathodes constitute the Li+ reservoir (this is the reason why new portable gadgets have to be charged during the very first use), the 170 mAh g−1 of 1 g of LiFePO4 will never be enough for the full reaction of Si. This leaves the manufacturer with two options: assemble the commercial cell with a Si-based anode without using its full energy potential or balance the amount of cathode using 4.4 times more of cathodic powder than anodic one. Such an increase in the amount of active material has both an economic and an engineering impact on the final cells. For these reasons, the development of cathode materials exhibiting reversible, more-than-one electron reactions would constitute one of the major breakthroughs for the realization of the next generation of accumulators, some other requirements being high voltage reactions and flat reaction plateaus.
In this review, we will focus mainly on the electrochemical reaction and synthesis strategies for the high oxidation state ferric fluoride and its derivatives as cathodes for the next generation of electrochemical accumulators.
Early research, characteristics and electrochemical mechanism of FeF3
Fluoride-based cathodes have been investigated very early during the development of Li-based cells. Already in the 1960s, Lockheed investigated CuF2 and CoF3 to develop long-lasting Li-based primary cells for use in space applications [9, 10, 11]. Later, in the 1970s, AgF2, NiF2 and BiF3 were also investigated for primary cells . The first prototypes of the modern Li-ion cells were developed following the proposal (around 1980 ) and commercial application (in 1990 ) of the “rocking chair” systems, effectively clarifying that research in cathode materials would be a prominent field for the following years.
The first investigation of FeF3 was made by Arai et al. in a study published in 1997 . The study was performed on a commercially available product in the region 4.5–2 V versus Li+/Li. The authors identified the topotactic insertion reaction during the first part of the discharge with a maximum reaction of 0.5 Li+. This behavior accounts for half of the currently demonstrated insertion of 1Li+ per formula unit. The most probable reason is the lack of nanostructuration of the commercial FeF3.
After the above-mentioned study, fluoride-based cathodes were somehow left aside for some years to the profit of more immediately interesting compounds like oxides or phosphates. The reason for this slowing down of the research lies in the intrinsic, poor electronic conductivity of the fluoride which is due to the large bandgap induced by the highly ionic character of the metal-halogen bond. Badway et al. [16, 17] extensively studied the FeF3 reaction mechanism in two publications that appeared in 2003 and later in 2007 . Again the study was carried on a commercially available product but this time, high energy ball milling was used both to decrease the fluoride particle sizes down to 25–30 nm and to intimately mix it with carbon-based conducting agents. Moreover, the studied voltage regions were varied from 4.5–2 V versus Li+/Li to 4.5–1.5 V versus Li+/Li which permitted the description of the full electrochemical reaction of FeF3.
Based on XRD studies, it was concluded that since a topotactic insertion should result in no appreciable change in the intensities of XRD reflections, reaction 1 proceeds through two separate mechanisms . The topotactic reaction can be observed up to a composition of roughly Li0.5FeF3 while large variations in unit cell volume can be observed for the remaining part of the curve up to around 1 Li+. The conversion reaction (reaction 2) then follows. On the other hand, the authors could not conclusively account for the charge behavior: unexpectedly identifying a compound related to FeF2 rutile structure at the end of charge, they were only able to exclude the presence of side reactions such as the formation of FeCx or FeOxF2−2x by electrolyte decomposition.
Through density functional theory (DFT), Doe et al.  identified several possible intermediate Li–Fe–F phases to explain both the discharge and charge behavior. Interestingly, they concluded that discharge and charge do not follow the same reaction path with the formation of a higher number of intermediate phases during charge than during discharge. Also, the final charge product resulted to be obtained from the delithiation of a rutile-like Li0.75Fe0.75F3 phase thus forming a rutile-distorted FeF3 with cell parameters very close to those of FeF2.
Experimental proof of such a behavior was given almost concomitantly by Yamakawa et al.  through ex situ 6Li, 7Li and 19F magic-angle-spinning nuclear magnetic resonance (MAS-NMR) among other techniques. The analyzed commercially available iron trifluoride was ball-milled with carbon. In order to clarify the end-of-charge mechanism, the authors used an interesting approach consisting in the comparison of an electrochemically cycled FeF3 sample with a FeF2 sample cycled with the same conditions. The reported results confirm the structural transformation of the initial ReO3–FeF3 phase into a phase structurally resembling a lithiated rutile with a stoichiometry close to Li0.5FeF3. The comparison with pure rutile-FeF2 showed that the Li local environments were similar to those of the FeF3 system lending further support to the experimental data. The transformation evolves initially through a two-phase region (up to x ~ 0.5) and then through a solid-solution reaction that involves insertion of Li into Li0.5FeF3. Furthermore, the material can accommodate more lithium by progressively reducing the Fe and extruding LiF from the matrix reaching a final state where LiF and α-Fe are intimately mixed. Interestingly, NMR spectra showed that the obtained α-Fe is superparamagnetic and not ferromagnetic as one would expect. When the reconversion was studied, it was found that the LiF-Fe mix re-forms a lithiated rutile phase at 4.0 V with a composition close to Li0.5FeF3. The authors conclude then that the conversion reaction is largely, but not completely, reversible.
It is worth noting that the first insertion reaction (reaction 1) was very recently reinvestigated with the proposal of a simpler mechanism other than the separated two-phase region up to Li0.5FeF3 and solid-solution up to LiFeF3. Through in situ XRD, Tan et al.  proposed that lithiation involves a single, continuous, topotactic insertion of Li-ions in the range 0–1 Li+. The insertion would take place along the body diagonal of the trifluoride unit cell only resulting in limited distortion of the lattice parameters as the shifts of the observed diffraction peaks seem to suggest.
While the electrochemical reactions are, at this point, largely understood, the reaction kinetics remained to be addressed, notably the evolution of the polarization during cycling and the interfacial behaviors. Liu et al.  published a complete electrochemical study coupling GITT with electrochemical impedance spectroscopy (EIS) in order to clarify the above-mentioned points. In order to compare with previous studies, a high energy ball-milled FeF3/carbon composite was analyzed in the 4.5–1 V versus Li+/Li potential window. The authors observed that after the initial Li insertion, the formation of the LiF/Fe nanocomposite resulted in a phase separation. This nanocomposite exhibits a large internal interface between LiF and Fe and its EIS signature is characterized by both high resistance (due to LiF insulating character) and high capacitance (due to the high interfacial surface). It is important to stress that this internal interface does not involve the electrolyte in any case and is purely limited to the interface inside the solid state composite. As one can expect, the phase separation results in a large overpotential which is experimentally observed as a strong polarization of FeF3 electrochemical curve. During charge, recombination involves migration of heavy atoms such as Fe and F to recover the rutile LixFeF3 (x ~ 0.5). This step is initially fast at the interface but soon slows down resulting in a limitation to the reaction rate.
As a final observation, the authors also implied that such a behavior arises when the reaction kinetics are faster at higher rates. In other words, rapidly reversing the LiF/Fe nanocomposite is much more effective than let it rest for some time. This observation has, of course, important implications on the cell design should a commercial FeF3-based cathode see the daylight.
Apart the electrochemical mechanisms involved in the conversion reaction, studies devoted to analyzing other problems involving iron trifluoride in secondary cells appeared. For example, Zhou et al. [23, 24, 25] studied the thermal stability of the usually ball-milled commercial iron trifluoride/carbon mixture by thermogravimetry (TGA-DSC). They analyzed the thermal behavior of samples cycled limiting the discharge depths to 1 Li, 2 Li and 3 Li. It was demonstrated that FeF3 is thermally stable and that reaction peaks were observed only due to electrolyte decomposition. No appreciable reaction was observed for 1 and 2 Li reactions except for the formation of a surface film due to the contact electrode–electrolyte (decomposing in the range 100–160 °C). It is interesting to point out, however, that the iron formed at the end of the complete conversion reaction (3 Li) was found to react with PF5 present in the electrolyte (decomposition of LiPF6) at temperatures around 300 °C.
Synthesis techniques, new electroactive phases and approaches for the improvement of FeF3
Besides the well-known mechanotreatment of commercial FeF3 powder, authors have been increasingly interested in the direct synthesis of the electroactive species by soft-chemistry approaches such as sol–gel or hydrothermal/solvothermal methods. Iron trifluoride is generally obtained by treating the anhydrous trichloride in a flow of HF. The reaction at RT yields amorphous FeF3 while crystalline powders can be obtained after heat treatment (at least 600 °C) . This is obviously an expensive and complex procedure if one wants to produce larger quantities at the industrial scale.
The first report of a full synthesis of a FeF3 based material at low T appeared in 2009 with the publication of Wu et al. [27, 28]. By reacting a FeCl3 water solution with NaOH, the hydroxide Fe(OH)3 precipitates. This precipitate easily reacts with an excess of HF water solution to give a number of hydrated iron trifluorides some of which can be partially dehydrated in mild conditions (70–150 °C at atmospheric pressure or under vacuum). The obtained fluoride was then ball-milled together with MoS2 or V2O5 in order to promote the formation of a mixed conducting matrix to provide easier transport of both ions and electrons. Both samples exhibited interesting capacities (around 160 and 200 mAh g−1for the MoS2 and the V2O5 composites, respectively, at 0.1 °C rate) for at least 30 cycles. However, the authors limited the cycles to a relatively narrow potential window of 4.5–2 V vs Li+/Li thus exploiting only the part of the curve relative to the insertion reaction (cf. Eq. 1). As one can imagine, some discrepancies in the electrochemical curves arise since in the reported composites FeF3 is actually hydrated and structurally different from the ReO3-type FeF3 phase. Thus, the intercalation potential is relatively lower decreasing from the 3.4–3 V for the anhydrous ReO3-type FeF3 phase  to the 3–2.6 V for the reported hydrated phase.
Crystallographic parameters of the main iron trifluoride hydrated forms
The picture is complicated also by the fact that both the pyrochlore and the hexagonal tungsten bronze (HTB) structures (see Table 1) can be obtained either in the fully dehydrated FeF3 form  or in the partially hydrolyzed form with substitution of F anions by OH groups. Considering for example the HTB-FeF3·0.33H2O structure, it is possible to obtain it through dehydration of FeF3·3H2O (T > 120 °C ). Further thermal treatment yields anhydrous HTB-FeF3  or an OH substituted structure such as HTB-FeF3 − x(OH)x·yH2O (y < 0.33) [31, 32, 33]. By conventional structural analysis techniques, like XRD, it cannot be clearly distinguished between these three forms so other means have to be deployed to analyze the material in detail. In particular Mößbauer spectroscopy can help due to its sensitivity to the atomic environment as well as by infra-red (IR) spectroscopy after adsorption of small molecules (generally NH3 or pyridine).
The pyrochlore–FeF3·0.5H2O is also built around slightly smaller hexagonal cavities compared to the HTB homologue (around 500 pm across), but the channels run in the  direction instead. However, pyrochlore is also a more unstable structure and it undergoes irreversible amorphisation at potentials below 1.6 V. This phenomenon is due to the loss of structural water dislodged from the channels and has secondary consequences on partial electrolyte degradation . However, the pyrochlore structure seems also to produce much better distributed composites with carbonaceous additives leading to a longer-lived cathode material [30, 36].
The same working group also demonstrated the capability of both structures through sodium intercalation with comparable 140–150 mAh g−1 in the region 4–1.2 V versus Na+/Na independent of the chosen compound [30, 37].
Following the direction traced with the synthetic method developed by the Maier’s group, Li et al.  obtained a HTB-FeF3·0.33H2O/graphene nanosheets (GNS) composite. During synthesis, GNS are stabilized in ethanol by the addition of BmimBF4 ionic liquid which helps to avoid restacking due to the attraction between the [Bmim]+cation and the π-electrons of graphitic layers. The resulting composite material shows much spaced fluoride particles, 10 nm in average, anchored on the graphene sheets even if a small percentage of the carbonaceous source remains after synthesis (less than 10 wt % from elemental analysis). In the insertion reaction electrochemical window (4.5–1.7 vs. Li+/Li) a very good performance at the higher C rates was recorded (90 mAh g−1 at 20 °C rate) confirming the importance of a careful address of the electrical insulating nature of fluoride particles. Prolonged cycling showed a stability lasting for over 200 cycles at 10 °C rate.
Lu et al.  employed the same ionic liquid-based synthesis, but implemented it by microwave irradiation (80 °C for 5 min). The authors obtained the pyrochlore type FeF3·0.5H2O regardless of the amount of ionic liquid added to the reaction medium. However, the sample morphology is clearly influenced by the amount of ionic liquid and in particular with its increase, the resulting particle morphology changes from octahedrons to spheres with porous surface. These latter are most probably secondary particles (around 200 nm in diameter) formed by the agglomeration of smaller primary particles (as already observed by Li et al. [35, 38]). Electrochemical results for these porous spheres in the region 4.5–1.6 versus Li+/Li are comparable to those reported for hydrated fluoride phases with an average of 140 mAh g−1 for 100 cycles at around 0.07 °C rate.
The latter in particular, was easily capable of cycling at its full theoretical capacity after more than 50 cycles but, in a more interesting way, it showed the possibility of delivering a quite stable 100 mAh g−1 at 1 °C rate in the potential window 4.4–1.6 versus Li+/Li.
Very recently, also ball-milling has been revisited for the direct synthesis of HTB-FeF3·0.33H2O-acetylene black composites . Fe(NO3)3·9H2O and acetylene black react with NH4F in an ethanol/polyethylene glycol mixture (1:1 volume) for 10 h in stainless steel milling vessels. A mild heat treatment to remove NH4NO3 follows (120 °C, 24 h). However, unlike in previous examples where a link between the carbon and the fluoride particles can be identified in the form of C–F bonds [30, 41], there is no evidence of chemical reaction for this composite, suggesting a behavior more related to intimate mixing/coating than to real bonding between the fluoride particles and the carbonaceous matrix. Electrochemical results are in line with the pure phase HTB-FeF3·0.33H2O with 150 mAh g−1 delivered at 0.1 °C rate in the potential window 4.4–1.5 versus Li+/Li.
In a recent publication, Liu et al.  compared the electrochemical response of anhydrous FeF3, of HTB-FeF3·0.33H2O and of the highly hydrated FeF3·3H2O. Interestingly, the three phases have similar behaviors in the enlarged voltage region 4.5–1 V versus Li+/Li with an initial plateau appearing in the region 3–2 V related to the insertion reaction and a second plateau below 1.7 V. The authors conclude that the HTB structure shows the best electrochemical performances during the intercalation process, but loss of structural order leads to increased performance degradation when the conversion reaction is included.
Regardless, some studies focused on the fully hydrate phase: FeF3·3H2O. The FeF3·3H2O structure is built around chains of apex-sharing octahedrons running along the  direction. In the apex positions, two fluorine atoms are arranged, while the four basal positions are statistically occupied by either fluorine or water (two of each). Each octahedron chain is then surrounded by four chains of exclusively water molecules, which create large spaces between each octahedron sequence originating roughly rectangular cavities around 46 pm long and 26 pm large. Since, unlike in HTB-FeF3·0.33H2O, water molecules do not occupy the free positions in the channels, the structure is stabilized without further effect on Li+ diffusion. Thus, in principle, during the intercalation process Li+ diffusivity should be higher in FeF3·3H2O than in HTB-FeF3·0.33H2O.
Shi et al.  obtained the tri-hydrate by simple precipitation (FeCl3–cetyl-trimethylammonium bromide surfactant, CTAB, -HF mixed water solution), while composites were obtained after ball milling with carbon black. Analysis based on cyclic voltammetry results (CV) confirmed the assumption of a better Li+ diffusivity demonstrating an increased Li+ diffusion coefficient of 7 × 10−13 cm2 s−1 at RT. For the sake of comparison, Li et al.  reported an average value close to 10−14 cm2 s−1 for the pure, non-milled HTB-FeF3·0.33H2O considering aggregates of around 350 nm. The tri-hydrated fluoride also delivers a robust 210–200 mAh g−1 in the region 4.5–1.5 V versus Li+/Li at 0.1 °C rate for at least 30 cycles. However, unless the structure can be stabilized in some way, the instability due to the loss of structural water results in a tendency to steady performance degradation with cycling.
Synthesis followed by a post-treatment
For the synthesis of water-free ReO3-type FeF3 except for solid state reactions, a post-treatment to eliminate structural water is needed. In most of the cases, a second treatment is also involved, allowing particle size reduction, intimate mix with carbon additives or the formation of specific structures. In this section, we will examine the various reports which have appeared on this specific subject.
Simple heat treatment for water removal
One common synthesis method is the simple precipitation usually involving an ethanol or water solution of FeCl3 or Fe(NO3)3 mixed with an ethanol or water solution of HF or other fluorine source (NH4HF2 for example). Surfactants, like CTAB, or carbon additives (carbon nanotubes, graphene oxide or reduced graphene oxide, etc.) are sometimes added to the reaction mixture in order to limit particle growth or provide the needed electronic conduction enhancement. Products are then submitted to a dehydration step by annealing (generally not above 400 °C).
This publication  reports also an important insight in the electrochemical reaction of these fibers. From TEM observations on fully discharged electrodes, single-crystal α-Fe nanodomains (3–5 nm) are formed consistently with the already described conversion reaction. However, these nanoparticles clearly appear to be interconnected through their common  family of planes to form a continuous network. Such a behavior, although never reported for the trifluoride, was recently demonstrated to occur for a FeF2/C composite and was ascribed to the small diffusivity of Fe ions thus producing a metallic iron backbone encapsulated in LiF . On the other hand, upon charge the usual reconversion to defective trirutile is observed but the reaction is constantly hindered up to complete electrode failure. At the 50th charge, significant amounts of unconverted LiF and α-Fe are present but the exact reason of such degradation is still to be reported.
Other papers involve drying after precipitation either without  or with the presence of carbonaceous additives (for example reduced graphene oxide  or carbon nanotubes ). In the three above cases, however, although the authors provide XRD signatures suggesting full dehydration of FeF3·3H2O to ReO3-type FeF3, the reported electrochemical signatures in the region 4.5–1.5 versus Li+/Li are very close to that of the hydrated phases (HTB or pyrochlore) suggesting either water presence in the compounds or partial dehydration/hydrolysis as discussed for the hydrated phases in the previous section [31, 32, 33]. The group of Maier [37, 38] clearly observed that the final fluoride product is highly influenced both by the initial reaction T (for example, carrying on a reaction at 0 °C yielded a completely amorphous fluoride) and by the reaction medium (for example, the dehydration process carried in a hydrophobic ionic liquid yielded an almost amorphous product). So far, no comprehensive study has appeared on this particular issue which certainly is worth exploring.
Coating: carbonaceous additives and protective oxides
The most common coating process involves the addition carbonaceous additives which are known to greatly enhance electronic conductivity of the electroactive material. For fluorides, graphene oxide/reduced graphene oxide (or graphene nanosheets as they are sometimes called) are the most widely used additives. Although carbonization of a carbon source (such as citric acid, a sugar or other compound) is often used to coat cathode materials such as LiFePO4 at high temperatures (e.g., 650 °C ), these temperatures are not compatible with transition metal fluorides. HF evolution (resulting in structure decomposition) for the HTB structure can be recorded as low as 170 °C under Ar atmosphere, practically as soon as water is removed from the hexagonal channels . For this reason, it is important to study coating procedures that can be carried on at lower temperatures.
Another graphene/reduced graphene oxide composite was reported by Ma et al.  through a rather convoluted synthesis involving various steps. In a first step, Fe3O4/C composites were obtained from graphene oxide, sodium acetate (NaAc) and FeCl3·6H2O dissolved in ethylene glycol (EG) and solvothermally treated at 200 °C for 12 h. An annealing step (500 °C, 2 h, Ar/H2 atmosphere) was carried out in order to reduce (at least partially) graphene oxide. The composite was solvothermally treated a second time with HF (48 wt % H2O solution) to transform Fe3O4/C into FeF3·3H2O/C. Further annealing (200 °C, 4 h, Ar atmosphere) yields FeF3/C. The first annealing step, promotes the formation of few hundreds nanometer iron oxide particles. This greatly hinders the electrochemical reaction with a fast degradation from 600 to 230 mAh g−1 after the first five cycles even at very low current density (4.5–1.5 versus Li+/Li at 0.03 °C). However, the benefic effect of the intimate contact and, possibly, the anchoring on the reduced graphene oxide additive can be observed on the long period when the large crystals have been repeatedly ground by the protracted conversion reaction. In fact, the composites are stable at around 200 mAh g−1 for at least 100 cycles. A similar result can be obtained using commercial Fe2O3  instead of magnetite nanoparticles.
It is worth noting that other examples of addition of carbonaceous additives have been reported in “Hydrated phases”, namely, in the works of Maier [30, 34, 36] and Di Carlo  but we chose to illustrate them in a different section due to the differences in the obtained fluoride product.
Another interesting concept is the coating of cathodic materials with oxides. The reason for coating is the fact that since direct contact with the electrolyte is prevented, dissolution of the cathode material due to the decomposition of LiPF6 is also hindered. Moreover, other benefits have been observed such as: maintenance of ionic conduction pathways, suppression of the release of oxygen and of phase transitions (in case of high-energy materials such as the solid solution Li2MnO3/LiMO2, with M = transition metal), and decrease of cationic disorder .
Following this approach, Zhang et al.  used the oxide coating concept but with the added benefit of increased electronic conductivity. Anhydrous FeF3 was thus coated with a thin layer of semiconductive α-Fe2O3 (hematite) and showed improved electrochemical properties without further addition of carbonaceous additives and/or ball milling. The synthesis of such a composite material involves the synthesis of hematite first (highly agglomerated particles of 100–150 nm) by a sol–gel method (a water solution of Fe(NO3)3·9H2O, citric acid and NH4OH is left to gelify and then heat-treated at 500 °C, 5 h under air). The resulting oxide is then fully fluorinated by further heat treatment (475 °C, 5 h under inert gas containing F2) and rapidly re-oxidized under air (500 °C for a duration varying from 15 to 480 s). The electrochemistry reported in the potential window 4.5–1.2 V versus Li+/Li (0.07C) is the result of the complex interaction between various active species: the primary FeF3, the Fe2O3 coating and possibly interfacial oxyfluoride formed between the two contiguous layers. Even though, the authors state that: “[…] the improvement of electrochemical performance of FeF3 resulting from an in situ Fe2O3 coating is inferior to that by mixing with carbon-based conductive additives”, it is remarkable that such large particles can sustain some cycling without any further treatment to reduce size and/or further enhance electronic conductivity (roughly 260 mAh g−1 are delivered after 10 cycles).
In recent years the benefits of ordered three-dimensional (3D) porous nanostructures over nanopowders have widely taken root in the energy-materials community because they result in higher surface areas and more favorable structural stability over simple non-ordered particles. Moreover, the ameliorated diffusivity of the electrolyte through channels and pores would ensure a high specific surface area in contact with the electrode and hence an increased possibility for Li-ions to react. In the specific case of fluorides, there could also be an additional positive effect due to the minimization of nanoparticle aggregation induced by the conversion reaction .
Among carbon additives, excluding the already mentioned carbon nanotubes, graphene/graphene oxide and ordered mesoporous carbon, activated carbon microbead (ACMB) can be also used as a structuring agent . After the synthesis of ACMB from glucose, the composite material was obtained via a CTAB-assisted precipitation synthesis: a CTAB–FeCl3 mixed water solution was slowly added to an ACMB-HF water solution and stirred for 24 h at RT. The obtained material was then heat-treated at 170 °C for 10 min, Ar atmosphere. The resulting composite is constituted of a mixture of FeF3·3H2O and HTB-FeF3·0.33H2O which seems to be grafted at the surface of ACMB spherules (about 30 μm in diameter) thus keeping the spherical morphology. The authors do not show wide area SEM images, so it is impossible to appreciate how homogeneous the coverage is, also taking into account the small initial amount of ACMB (2 wt % in respect to the overall fluoride quantity). The delivered capacity in the 4.5–2 V versus Li+/Li window was slightly lower than the theoretical 150 mAh g−1 with a stabilization at around 120 mAh g−1 after 50 cycles (at 0.1 °C rate).
As it has been pointed out in “Simple heat treatment for water removal”, surfactants are employed in order to limit particle growth and aggregation and, among them, the most employed in iron fluoride synthesis is CTAB. However, the molecule itself can be also employed as a structuring agent directing the assembly of the particles. In the study of Tan et al. , CTAB was dissolved with Fe(NO3)3·9H2O in an ethanol/HF mixed solution and solvothermally treated for 2 h at 80 °C. The obtained precipitate was then washed and dried at 180 °C for 12 h. The resulting product, a mixture of FeF3·3H2O and HTB-FeF3·0.33H2O, consists of quite uniform prismatic/cylindric hollow rods around 3 μm long and 500 nm thick. Unfortunately, the bare structures were not able to meet the requirement of prolonged cycling with a marked capacity fade after 20 cycles (4.5–2 V vs. Li+/Li at 0.5 °C rate) and so a ball-milling treatment with acetylene black was carried out. The treatment completely destroyed the hollow structures and lead to small particles intimately mixed with the carbonaceous additive with a consequently ameliorated electrochemical response (150 mAh g−1 after 100 cycles in the 4.5–2 V vs. Li+/Li voltage window at 0.5 °C rate) but effectively eliminating every effort initially put into the formation of self-assembled coherent structures.
Another concept has been investigated by Martha et al.  who approached structuration from the point of view of electrode preparation instead of synthesis. The generally accepted method consists of pasting an electrode slurry (active material/carbonaceous additive and plasticizer) onto aluminum foils. Instead of Al, the authors used non-graphitic carbon fibers as current collectors and petroleum pitch as a binding agent between iron fluoride and the fibers, thus eliminating the need of an inactive plasticizer (providing only dead weight to the final electrode). After ball milling of commercial FeF3 and a graphene-based electron conductor, the composite powder was mixed with petroleum pitch and the resulting slurry was coated onto the carbon fibers.The electrodes were then directly used after carbonization (around 450 °C, 5 h under Ar). In the voltage window, 4.5–1 V versus Li+/Li, 600 mAh g−1 have been recorded for the composite at low current density (0.05 °C rate) with good performances also retained at 1 °C rate (a stable 400 mAh g−1 for 10 cycles). Moreover, it is noteworthy that the replacement of aluminum by carbon fibers contributes to further reduce the weight of the electrodes permitting higher overall energy per weight unit and thus highlighting the interest of such a technological approach.
The current, rechargeable battery technology implies the use of a liquid electrolyte, which needs to be imbibed into a separator. This characteristic brings to the commercial models several restrictions in terms of design, size and leak-proof assembly. In recent years, the concept of all-solid-state (micro)batteries has found large support from the scientific community due to the increased flexibility of the design of stand-alone microelectronic devices and the enhanced applicability brought by the complete absence of leakage (for example medical implants would be much safer than what they are today) . In order to advance on such a concept, it is very important to be able to produce thin, homogeneous films of active material/separator so that miniaturization is possible without losing the advantages of energy densities and cycle life, which are now achieved by Li-ion cells.
Unlike the extensive work conducted on oxides and phosphates [67, 68], only few works have been published on iron fluoride thin films applied to Li-ion batteries. In one of them, Makimura et al. [69, 70] used pulsed laser deposition (PLD) to grow iron fluoride thin films from 80 to 150 nm in thickness. The technique consists in pulsing a laser source with a 2 Hz frequency onto FeF3 or FeF2 target. The target is then vaporized creating a thin film on a stainless steel substrate. Typical electrochemical curves for FeF2 samples deposited at the high T (600 °C) and RT are recorded. All the reported films presented a very large reversible over-capacity at low potentials (between 1 and 0.05 V vs. Li+/Li) acting as the main electrochemical phenomenon and granting the very interesting performances reported (around 500 mAh g−1 for 20 cycles). Such a behavior has been explained as an interfacial interaction of lithium within the Fe/LiF matrix leading to local charging (interfacial storage).
Current, commercial Li-ion technology comprises a carbon-based anode and lithium-containing cathode material. Adopting such a configuration was inevitable after the problems deriving from Li dendritic growth during charge . As such, the first generation Li/TiS2 cells were replaced by the so-called “rocking chair” systems as we know them nowadays. At the current state of technology, FeF3 suffers from a more debilitating plague than electronic insulation or reaction kinetics: the lack of lithium in the initial material. One way to solve the problem would be the use of prelithiated anodes which are difficult to manufacture and handle in practice. Research is focusing on how to introduce lithium in the structure so as to actually give a chance to fluorides in the next generation of accumulators.
The direct synthesis of the charge end-product (the Li0.5FeF3 trirutile structure, see “Early research, characteristics and electrochemical mechanism of FeF3”) has been reported in 1969 . Through heat treatment at high temperature (1000 °C) of an equimolar mixture of LiF, FeF2 and FeF3 powders, it was possible to obtain the pure LiFe2F6 trirutile phase. A successive study pointed out that single-crystals can be obtained through hydrothermal synthesis in fluorolytic medium (5 M HF solution at 650 °C) starting from LiF and FeF3 . A more recent study, however, introduces the possibility of mechanical milling of the same equimolar mixture of fluorides although the process requires very high milling times (40 h for the initial product and another hour after addition of carbon black) [76, 77]. The pure product showed a poor intercalation behavior (less than 120 mAh g−1 after 15 cycles in the 4.5–2 V vs. Li+/Li voltage window at 0.08C rate) which could be ameliorated when the synthesis was conducted with a slight excess of lithium (Li1.2Fe2F6.2). Mößbauer spectroscopy pointed out that the mechanical treatment is rough enough to promote loss of fluorine, resulting in an under stoichiometric compound bearing a large number of anionic vacancies. Thus, the authors ascribe the ameliorated electrochemical performance (around 140 mAh g−1 after 30 cycles in the 4.5–2 V vs. Li+/Li voltage window at 0.08 °C rate) to the vacancy effect over structural order.
Since the trirutile Li0.5FeF3 requires both harsh conditions for its synthesis and some defective structure in order to work properly, other approaches are being explored with the aim of producing a prelithiated fluoride cathode. The most direct approach consists in obtaining the end products of the conversion reaction, LiF and Fe (see Eq. 2), as an intimate mix.
Based on the above results, Liao et al. [80, 81] used magnetron co-sputtering to create a 64-sample library of LiF(1−x)/Fex (0 < x < 1) compositions. The principle of the experiment was to rapidly examine a series of possible compositions in order to quantify the optimal LiF/Fe ratio for an acceptable electrochemical result. Without surprise, the best performances were obtained when the ratio of LiF/Fe was 3/1 but the interesting result turned out to be a confirmation of the appearance of interfacial storage (see previous section) for cutoff voltages of 1.2 V versus Li+/Li. In fact, co-deposited LiF/Fe thin layers already possess an interfacial region between LiF domains and Fe domains that enables overcapacity. Moreover, by Mößbauer spectroscopy, the ‘excess’ of Li ions could be univocally proved to be located at the LiF side of the interface with the corresponding electrons enriching the Fe side as a consequence.
Further ball milling techniques were then carried directly on the LiF/Fe 3/1 ratio. Li et al.  used a two-step milling adding to the commercial LiF and iron powders a nanometric TiN. In the second step, further milling was conducted on the LiF/Fe/TiN with addition of graphite. Instead of promoting the formation of LiF with release of a gas, as described in the work of Amatucci (Eq. 4), here the TiN nanoparticles act as both a grinding-enhancer powder and as a structuring agent forming a conductive nanometric core capable of fixing the electroactive components in order to prevent phase segregation and/or aggregation. Electrodes were prepared with the customary addition of conductive agent and plasticizer (10 wt % each) and good performances were recorded in the enlarged potential window 4.5–1 V versus Li+/Li with around 300 mAh g−1 for 20 cycles at 0.7 °C rate.
A third approach, explored by Kim et al. , consists not in the direct milling of the end products of the conversion reaction (such as LiF/Fe in the 3/1 ratio, see Eq. 2) but in mimicking the composition at the end of the insertion reaction (such as LiF/FeF2 in the 1/1 ratio, see Eq. 1). A two-step milling was again used, first adding FeF2 to LiF (both commercial powders) and then further milling after addition of graphite. Electrodes were prepared by slurry (20 wt % of conductive agent and 10 wt % of plasticizer) and around 200 mAh g−1 were delivered in the potential window 4.8–2 V versus Li+/Li. The interesting point here is the peculiar reaction of the FeF2–LiF composite which, although similar in composition to the electrochemically formed LiFeF3, reacts via a different pathway. With the help of X-ray absorption fine structure and near edge structure (EXAFS/XANES), XPS and DRX studies, the authors were able to prove that, during the first charge, the excess of fluorine resulting from Li+ release, would be incorporated into the remaining FeF2 to form a higher oxidation state fluoride analogous to ReO3-type FeF3 although highly defective. The following reaction would follow the scheme reported in Eq. 1 with an insertion inside the newly formed iron trifluoride.
Ma et al.  produced a composite starting from hydrothermally synthesized Fe3O4/reduced graphene oxide ball-milled with LiF. The resulting LiF/Fe3O4/reduced graphene oxide powder was then heat-treated in a reducing environment (550 °C, 4 h, Ar/H2). After the heat treatment, TEM images reveal that magnetite is successfully reduced to α-Fe particles (50–60 nm), grafted onto the reduced graphene oxide surface. At the same time, while no evolution of average particle size is recorded for LiF (100–150 nm), the particles do not seem to be well homogenized with iron, and large void spaces can be observed. However, a good cycling stability with a certain decrease in polarization can be observed in the potential window 4.5–1.5 V versus Li+/Li (0.03–0.14 °C rates) certainly due to the benefic effect of the conductive reduced graphene oxide layers.
The synthesis does not seem to be optimized at the moment as FeF2 and Li5FeO4 impurities have been detected by XRD; morphology is also an issue as the high temperatures favor particle coalescence and growing deeply affecting the electrochemical results.
In addition to the LiF/Fe/C composite, however, iron carbide (Fe3C) was also obtained as a result of the deactivation of the Fe catalyst. Electrochemistry studies showed very encouraging results with a stable capacity of 220 mAh g−1 for 150 cycles in the broad electrochemical window 4.5–1.3 V versus Li+/Li, but at a low current density of 0.03 °C. It is noteworthy that the reported electrodes were used as pure powders without further additives (carbon black or PVdF binder) and that the remarkable properties could be related to the high amount of carbon (around 45wt % from elemental analysis) unfortunately, the inactive iron carbide contributes to some hindering of the electrochemical performance due to the consumption of iron.
Another ferrocene-based synthesis was recently introduced by Zhang et al. . The authors produced fibers from electrospun solutions of LiF, ferrocene and polyacrylonitrile (PAN) which were then heat-treated in two steps: a first annealing at 280 °C in air for 5 h followed by a carbonization at 900 °C (8 h under Ar). The resulting composite yielded freestanding and flexible electrodes with some degrees of 3D structuration. Electrochemical tests were conducted in the window 4.2–0.5 V versus Li+/Li resulting in around 400 mAh·g−1 delivered after 50 cycles (current density of 0.05 °C) even if a marked tendency to performance degradation can be observed after the 30th cycle.
Other iron fluoride compounds could be employed for the development of cathodic materials for the next generation of accumulators. It is the case of lithium fluoroferrate or sodium-iron fluorides which can open new perspectives for the research and development sectors.
Lithium fluoroferrate (Li3FeF6) possesses two polymorphs (monoclinic α-phase and orthorhombic β-phase) with structures closely related to cryolite (Na3AlF6). Both structures consist of isolated and slightly distorted FeF6 octahedra which under tilt rotation produce either one or the other polymorph. These octahedra are linked together by interconnected lithium polyhedra in the tri-dimensional structure. The most active group on the electrochemical study of this compound is that of Garcia-Alvarado [89, 90, 91, 92] who deeply studied the behavior of both phases. They found a better activity of the α-polymorphin in respect to the β-form and a correlation between particle size and delivered capacity. As a general remark, the α-polymorph can be obtained with smaller particle size, and the synthesis was optimized until the full theoretical capacity was obtained (140 mAh g−1). This optimized synthesis involves stoichiometric amounts of reagents (Fe(NO3)3·9H2O, HF water solution and Li2CO3), which lead to the precipitated fluoroferrate product after addition of 2-propanol at 0 °C. A further ball milling treatment (12 h) yielded the final fluoride/carbon composite. The composite material obtained by this synthetic route was able to deliver 140 mAh g−1 in the voltage window 4.5–2.25 V versus Li+/Li at 0.06 °C rate. More recently , simulations pointed out that defects are energetically favoured especially for small crystallites with the most probable defect types being Li-excess (with consequent local reduction of FeIII to FeII) and/or Frenkel defects (lithium insertion into interstitial sites). Moreover, lithium-ion transport in Li3FeF6 seems to be closely related to interstitial migration (low calculated activation energies of 0.4 eV) paving the way for a doping strategy consisting in the introduction of different ions into iron sites, such as divalent Ni, Mg, Co or Mn, to promote Li interstitial compensation.
At the current state, however, the fluoroferrate cannot represent a viable electrode material due the low capacity associated with a 1e- reaction and to problems in lithium extraction. In fact, the charge reaction of Li3FeF6 involves FeIV metal centers and, even if it has been recently calculated that Li2FeF6 is a stable compound, there is currently no experimental observation for the latter also due to the prohibitive reaction potential of 6.1 V versus Li+/Li .
Sodium-iron fluorides (NaFeF3 and Na3FeF6)
Sodium-ion batteries have recently drawn the attention of the scientific community as low cost and low environmental impact devices with widely available starting materials and low toxicity . As it has been highlighted in previous sections, some iron fluorides have found an application also in Na-ion cells; one example is the HTB-FeF3·0.33H2O structure. As it concerns the ReO3-type FeF3, its structure has the potential to accommodate also Na-ions during the intercalation reaction (Eq. 1) although with slower kinetics than Li-ions . In the case of sodium, however, the intermediate NaFeF3 is well known and can be easily synthesized. Although technically not a FeIII compound, NaFeF3 finds easily its place in this review due to the concept it represents. There is not much research concerning this material [96, 97, 98]. An optimized synthesis method, based on the observation that capacity increases with the crystallinity of the final NaFeF3 compound, involves a stoichiometric mixture of FeF2and NaHF2. The reagents are then submitted to a roll-quenching technique where the melt obtained by heating at 1,000 °C (40 s under Ar) is injected into a fast spinning copper roller (2,000 rpm). Ball-milling with acetylene black provides the final composite material which can deliver the full capacity associated with the 1e− reaction (200 mAh g−1 at 0.014 °C rate in the potential window 4.5–1.5 V vs. Na+/Na). A more significative finding was, however, that among the isostructural NaMF3 series (with M = Mn, Fe, Co and Ni), NaFeF3 is the only phase capable of reversible Na+ deinsertion/insertion. This occurrence finds its explanation mainly in the chemical stability of the de-intercalated fluorides, MF3 (with M=Mn, Co and Ni) being known fluorinating agents. Only the affinity of iron toward fluorine ensures the ability of NaFeF3 to sustain cycling.
Yamada et al.  prepared dispersed and uniform particles by liquid-phase synthesis involving the thermal decomposition of iron and sodium trifluoroacetates in a mixture of oleic acid and oleylamine (10/10 molar ratio) at 280 °C for 30 min under Ar. The resulting particles were mixed with acetylene black by ball milling to yield the final composite material. TEM microscopy studies evidence an average particle size 10–20 nm. The composite delivers 115 mAh g−1 at 1° C rate in the potential window 4.5–1.5 V versus Na+/Na.
Just like lithium fluoroferrate, the sodium analogue, Na3FeF6, is related to cryolite structure. Shakoor et al.  prepared sodium fluoroferrate through dry ball milling using stoichiometric amounts of sodium and iron fluorides (3NaF:1FeF3). The composite delivers around 110 mAh g−1 (0.1 °C rate) in the potential window 4.25–0.5 V versus Na+/Na, but capacity faded progressively from the 10th cycle on. The probable reason has been identified by the authors in the difficult-to-reverse conversion reaction since NaF and Fe have been found in the discharged electrode. As a final remark, unless the same concept of interstitial charge compensation, which has been postulated for the lithium fluoroferrate is also valid for the sodium analogue (see Ref ), the charge reaction of Na3FeF6 to Na2FeF6 cannot be the initial step, so a hypothetical device would require a discharge step in order to start its operative life.
Pure FeF3 is highly hygroscopic so that its stability in air at RT is very limited. The compound will accept water molecules producing a series of H2O-stabilized structures until reaching the FeF3·3H2O stoichiometry. This hydration process is fairly common in solution-based syntheses and generally yields HTB-FeF3·0.33H2O or pyrochlore-FeF3·0.5H2O structures;
Water molecules act as strong structural stabilizers during the electrochemical lithiation promoting an intercalation behavior in the same potential window where anhydrous FeF3 would react with the full conversion. This occurrence limits the theoretical capacity (150 mAh g−1 for HTB-FeF3·0.33H2O instead of 712 mAh g−1 for FeF3 at 1.5 V vs. Li+/Li) but permits faster Li+ diffusion with good capacities retained even at 1C rates;
These water molecules can be highly mobile so it is not uncommon to observe loss of structural water resulting in a tendency to steady performance degradation with cycling. One of the possible reasons is that water can hydrolize the LiPF6 salt into LiOH, PF5 and HF. Therefore, careful structural assessment, coating or other structure stabilization has to be envisaged for such materials;
When envisaging the production of a water-free FeF3, the effect of the initial F- concentration has to be taken into account. A large excess of HF compared to Fe, favors the growth of amorphous fluoride wires. The wire morphology favors the formation of contiguous and interconnected 1D Fe domains at the end of the conversion reaction potentially paving the way for the full 3e− reaction;
It is possible to create ordered three-dimensional structures for fluoride-based materials, which would result in ameliorated diffusivity of the electrolyte and possibly in the reduction of the nanoparticles aggregation originating from the conversion reaction ;
The absence of preliminary Li content is a major obstacle for the development of fluoride-based cathode materials. Prelithiation is thus a point to extensively study;
An alternative to prelithiation could be Li+ compensation induced by divalent cation doping. This effect has, so far, been predicted only for the Li3FeF6 compound but it is not impossible that a similar behavior could be induced in structures such as the HTB-FeF3·0.33H2O or pyrochlore-FeF3·0.5H2O;
Open structures and certain derivatives also suggest a viable exploration of the sodiation reaction for the development of cheaper cells.
As far as it is possible to observe in the actual state of research, the most promising iron fluoride-based cathodic materials are obviously represented by the hydrated HTB and pyrochlore phases. In fact, it has been widely proven that these two structures are easy enough to synthesize through different soft-chemistry-based techniques using low-cost or reusable starting materials. The main hindrances to the development of these materials are represented by the relatively low capacity (150 mAh g−1), the lower, sloped reaction potential (3.2–3.3 V vs. Li+/Li) compared to the higher, flatter potential of, for example, LiFePO4 (3.4 V vs. Li+/Li) and the lack of lithium/sodium in the initial compound.
While Fe-based materials offer advantages with respect to environmental and economic considerations, the major difficulty in the electrochemistry of Fe is the lower voltage obtained when compared to other transition metals such as Co or Ni. A way to overcome this issue was demonstrated in 1989 on a comparative study between the three iron analogues Fe2(MoO4)3, Fe2(WO4)3 and Fe2(SO4)3 . Since all three compounds possess very similar metal–oxygen distances and formal charge, the increased redox potential with lithium from around 3.0 V for the molybdate and tungstate to 3.6 V for the sulfate, came as a surprising effect. The “inductive affect”, as it became known, was originating from the significantly larger electronegativity of the sulfate groups resulting in an increase of the ionic character of the Fe–O bonds. Thus, it is thanks to the inductive effect that sulfide-based electrode materials always display a lower potential than the corresponding oxides, as well as why the fluoride analogues show even higher potentials. In this last case in particular, manipulations axed on the increase of Fe–F bond ionic character by replacement of the anion would be complex since fluorine is already the most electronegative element. Instead, one way to act on the iron-fluorine bond could be the application of chemical pressure . When smaller cations are replaced by larger ones in a close-packed structure, the result from the point of view of the smaller ion is comparable to an elongation of the bond thereby increasing its ionic nature. This is why accurate doping with Co, Ni or Cu would probably give some result in this direction.
Both capacity and prelithiation could be ameliorated by the careful application of blending, combining different active materials. Just like spinel/layered mixtures (xLi2MnO3·(1 − x)LiMO2 with M = Mn, Co, Ni or their mixtures)  orphosphate/silicate mixtures (xLiFePO4·(1 − x)Li2FeSiO4)  have demonstrated incremented capacity in regular half-cell tests (usually delivering at least 200 mAh g−1), fluoride mixtures could be easily analyzed in this regard. For example the system xLiFe2F6·(1 − x)Li3FeF6 has not been investigated so far, while the mixture could certainly represent a valid candidate as a cathodic material.
Besides the more advanced concepts expressed hereby, some fundamental questions still remain to be answered for the iron fluoride hydrated phases in particular. It is noteworthy that since the detailed description of the FeF3 behavior by Badway, Doe and Yamakawa, no in-depth structural/mechanistical studies have appeared. Except for the Maier’s group, who carefully assessed not only the synthesis procedure but also the electrochemical characteristics of both the HTB and pyrochlore systems, all other publications have been focused for the best part on the synthesis procedures and/or the effects of addition of various carbon additives. For example, it has been pointed out that in ionic liquid-based syntheses, the final fluoride products are influenced both by the initial reaction temperature and by the reaction medium (nature of the ionic liquid). So far, no study presents a careful and rationalized assessment of the dehydration products for other synthetic mediums or studies their influence on the final fluoride material. Moreover, as the same structure can exist with different hydrated/hydrolyzed states, the relation to water content-electrochemical behavior has still to be elucidated.
Another valid concept is the study of the electrode preparation/optimisation instead of synthesis. Effects such as current collector influence, separator influence, electrolyte decomposition and slurry composition have found very limited and non-organized reports in the current literature.
Finally, we would like to stress again that probably no practical materials would see the light of day if a solution to the prelithiation issue or the exploration of Na-based derivatives is not extensively investigated.
- 1.Agricolae, G.: Bermannus sive de re metallica (on the nature of metals, or minerals). Dover Publications NY (1986)Google Scholar
- 2.Tressaud, A.: Functionalized inorganic fluorides: synthesis, characterization & properties of nanostructured solids. Wiley, New York (2010)Google Scholar
- 3.Louvain, N., Fakhry, A., Bonnet, P., El-Ghozzi, M., Guerin, K., Sougrati, M.-T., Jumas, J.-C., Willmann, P.: One-shot versus stepwise gas-solid synthesis of iron trifluoride: investigation of pure molecular F2 fluorination of chloride precursors. CrystEngComm 15, 3664–3671 (2013)CrossRefGoogle Scholar
- 9.Bauman, H.F.: Limited-cycle secondary battery using lithium anode. In: center, D.D. (ed.). p. 29. Lockheed Missiles & Space Company, Palo Alto (1964)Google Scholar
- 10.Bauman, H.F., Chiton, J.E., Mauri, R.: Lithium anode limited cycle battery investigation. In: center, D.d. (ed.). p. 81. Lockheed Missiles & Space Company, Palo Alto (1966)Google Scholar
- 11.Bauman, H.F., Chilton, J.E., Hultquist, A.E., Adams, G.B.: Lithium anode limited cycle battery investigation. In: center, D.D. (ed.). p. 40. Lockheed Missiles & Space Company, Palo Alto (1966)Google Scholar
- 20.Yamakawa, N., Jiang, M., Key, B., Grey, C.P.: Identifying the local structures formed during lithiation of the conversion material, iron fluoride, in a Li ion battery: a solid-state NMR, X-ray diffraction, and pair distribution function analysis study. J. Am. Chem. Soc. 131, 10525–10536 (2009)CrossRefGoogle Scholar
- 26.Brauer, G.: Handbook of preparative inorganic chemistry, vol. 1, 2 ed. Academic Press, NY (1963)Google Scholar
- 34.Li, C.L., Gu, L., Tsukimoto, S., van Aken, P.A., Maier, J.: Low-temperature ionic-liquid-based synthesis of nanostructured iron-based fluoride cathodes for lithium batteries. Adv. Mater. 22, 3650 (2010)Google Scholar
- 45.Liu, L., Guo, H., Zhou, M., Wei, Q., Yang, Z., Shu, H., Yang, X., Tan, J., Yan, Z., Wang, X.: A comparison among FeF3*3H2O, FeF3*0.33H2O and FeF3 cathode materials for lithium ion batteries: structural, electrochemical, and mechanism studies. J. Power Sources 238, 501–515 (2013)CrossRefGoogle Scholar
- 46.Shi, Y.L., Wu, N., Shen, M.F., Cui, Y.L., Jiang, L., Qiang, Y.H., Zhuang, Q.C.: Electrochemical behavior of iron(III) fluoride trihydrate as a cathode in lithium-ion batteries. ChemElectroChem. doi:10.1002/celc.201300069 (2014, in press)
- 50.Wang, F., Robert, R., Chernova, N.A., Pereira, N., Omenya, F., Badway, F., Hua, X., Ruotolo, M., Zhang, R.G., Wu, L.J., Volkov, V., Su, D., Key, B., Whittingharn, M.S., Grey, C.P., Amatucci, G.G., Zhu, Y.M., Graetz, J.: Conversion reaction mechanisms in lithium ion batteries: study of the binary metal fluoride electrodes. J. Am. Chem. Soc. 133, 18828–18836 (2011)CrossRefGoogle Scholar
- 52.Chu, Q., Xing, Z., Ren, X., Asiri, A.M., Al-Youbi, A.O., Alamry, K.A., Sun, X.: Reduced graphene oxide decorated with FeF3 nanoparticles: Facile synthesis and application as a high capacity cathode material for rechargeable lithium batteries. Electrochim. Acta 111, 80–85 (2013)CrossRefGoogle Scholar
- 68.Knoops, H.C.M., Donders, M.E., van de Sanden, M.C.M., Notten, P.H.L., Kessels, W.M.M.: Atomic layer deposition for nanostructured Li-ion batteries. J. Vac. Sci. Technol. A 30 (2012)Google Scholar
- 74.Portier, J., Tressaud, A., Depape, R., Hagenmuller, P.: Crystallography and magnetic study of an unpublished fluoride of trirutile type. C. R. Acad. Sc. Ser. C 267, 1711–1713 (1968)Google Scholar
- 78.Amatucci, G.G.: Electrode composition useful as positive electrode material of electrochemical cell e.g. lithium battery cell, comprises lithium fluoride compound demonstrating high specific capacity. WO2004051772-A2; US2004121235-A1; AU2003295927-A1; AU2003295927-A8; US7371338-B2; WO2004051772-A3Google Scholar
- 81.Liao, P., Dunlap, R.A., Dahn, J.R.: A Mossbauer effect study of combinatorially prepared Al2O3/Fe and LiF/Fe multilayers. J. Phys.: Condens. Matter 20, 055203 (2008)Google Scholar
- 98.Dimov, N., Nishimura, A., Chihara, K., Kitajou, A., Gocheva, I.D., Okada, S.: Transition metal NaMF3 compounds as model systems for studying the feasibility of ternary Li-M-F and Na-M-F single phases as cathodes for lithium-ion and sodium-ion batteries. Electrochim. Acta 110, 214–220 (2013)CrossRefGoogle Scholar
This article is published under license to BioMed Central Ltd. Open AccessThis article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.