Review—Emerging Trends in the Design of Electrolytes for Lithium and Post-Lithium Batteries

More powerful, durable, safer, greener and cheaper batteries are needed to ful ﬁ l the compelling requirements of automotive and grid applications. Addressing most of these requirements passes through the development of new-generation electrolytes able to overcome the issues of the state-of-the-art liquid ones, which are based on highly volatile and ﬂ ammable organic solvents. At the same time, the quest for new electrolytes is strictly related to the attempt of replacing the graphite anode with lithium metal, so opening the way to full exploitation of new post-lithium chemistries, e.g. Li – air and Li – sulfur. Here, we critically review some original concepts which were recently proposed as routes for the development of liquid and quasi-solid electrolytes with improved performances in terms of safety, chemical and electrochemical stability, and capability to sustain high current densities. Attention will be devoted to the problem of lithium dendrites formation, and to the electrolyte families able to eliminate/control their development. Finally, the most promising research directions will be outlined.

Since their industrial launch about three decades ago, lithiumion batteries (LIBs) contributed to change our life paradigms by allowing the development of increasingly powerful and versatile portable electronics. However, present LIBs architecture is not expected to reach the US DOE requirements for automotive, e.g. 235 Wh kg -1 , 500 Wh L -1 and 125 US$ kWh -1 . 1 At the same time, more energy and power density values are also needed to fulfil the requirements of grid applications. 2 The substitution of graphite anode with lithium metal may result in 50% increase of energy density and can allow the exploitation of different cathode-reacting species, e.g. oxygen and sulphur, 3 leading to new and more performing chemistries, e.g. Li-air and Li-S, which could deliver theoretical energy density values of the order of 3.5 kWh kg −1 and 2.5 kWh kg −1 , respectively. 4 Cui and co-workers reviewed the fundamentals of batteries with a lithium metal anode (LMB) and recent key progresses on methodologies, materials and characterization techniques. 5 It must be stressed that the development of LMBs imposes further and more rigid constraints on the electrolyte design. In fact, in addition to the well-known requirements of electrolytes for conventional LIBs, e.g. high conductivity, chemical and electrochemical stability against the electrodes, low flammability, and environmental sustainability, 6 the electrolytes for LMBs must also be able to block the formation of lithium dendrites, e.g. by forming a stable solid electrolyte interface (SEI) towards the metal anode, and/or by constituting a rigid barrier by themselves. 7,8 Furthermore, considering the enormous increase in production estimated for the automotive industry and for the residential storage, the new-generation electrolytes should be designed by devoting attention to the relevant sustainability aspects, such as the development of green industrial processes, the search for materials with reduced or no toxicity, the problems related to battery disposal, recycling and reuse. 9 Most of these concerns-and chiefly those regarding safety, durability, uncontrolled reactivity and environmental impact-are due to the use of liquid electrolytes based on volatile and flammable organic solvents, which can easily undergo drastic degradation processes. [9][10][11] Moreover, in conventional liquid electrolytes the cation transference number, t + , is generally lower than 0.5, which translates into reduced power density, and space charge formation that favours dendrites growth. 8 Many solutions were proposed to overcome these drawbacks, including ceramics, 12 polymer-ceramic hybrids, 13 solvent-in-salts, 14 polymer/gels, 6 as well as electrolytes based on ionic liquids (ILs). 15 Indeed, the industrial exploitation of such systems is challenging, due to hard scalability and high grain boundary/interface resistance in case of ceramics, or to unsatisfactory electrochemical stability, low mechanical strength and poor ability to hinder the dendrite growth in case of polymer/gel electrolytes.
Recently, some new concepts were proposed for the design of novel liquid, quasi-solid and solid electrolytes with improved safety, durability and electrochemical performance. These new concepts include: (i) biomimetic electrolytes 16 ; (ii) ionogels/eutectogels 17,18 ; (iii) patterned membranes obtained by 3D printing and/or photo-or stereo-lithographic processes 19 ; (iv) super-concentrated solutions, 20 which in some cases also show self-extinguishing properties. 21 Whereas several reviews appeared on solid-state electrolytes (e.g. ceramic, polymer-ceramic composites and solvent-free polymer systems), 22,23 to date scarce attention has been devoted to new liquid-like or quasi-solid ion conductors, which can indeed lead to technological breakthroughs in battery development. Here, we define as "quasi-solid" electrolytes those systems were a liquid phase is chemically or physically entrapped into a solid matrix, which is generally nanostructured, e.g. in the case of metal organic frameworks (MOFs) or ionogels.
In this review we aim at discussing the emerging classes of liquid-like and quasi-solid electrolytes. As, in some cases, these systems exploit unusual conductivity mechanisms, we will treat in detail both the fundamental aspects related to ion transport and to the level of performance they currently have achieved. In order to allow a critical comparison and a better comprehension, we will compare in a systematic fashion the relevant physico-chemical and functional parameters of most of the discussed systems (see Table I). Attention will be also devoted to the mechanisms controlling dendrites formation in cells with lithium metal anode, and to the electrolyte classes which appear more promising from this point of view. In conclusion, we will summarize pros and cons of the classes we examined (see Table II) and will critically outline the most important research directions to date active, as far as the perspectives for the medium-term future. However, for the sake of conciseness we will discuss only organic-based electrolytes. Although other very intriguing systems, e.g. water-in-salt 24 and water-in-bisalt 25 electrolytes showing high stability windows towards Li + /Li, have coming recently to the attention of the researchers, they will be not included here but will be the object of a future review.

Liquid Electrolytes
Improving non-flammability and self-extinguishing properties.-Large-scale application of LIBs is severely hindered by safety issues when cells are subjected to mechanical, thermal or electrical abuse. One of the most important concerns is related to the flammability of the organic solvents, e.g. linear and cyclic carbonates, 37,38 which can give origin to the so-called thermal runaway (see Appendix A). In particular, it is well known that linear carbonates, e.g. diethyl carbonate (DEC) and dimethyl carbonate (DMC), have lower flash points than cyclic carbonates, e.g. ethylene carbonate (EC). 10 Here, the flash point is defined as the lowest temperature at which vapours of the material will ignite, when given an ignition source. The attempts to improve the stability of the organic electrolytes by increasing their flash point chiefly included the search for new fluorinated solvents and salts, the employment of phosphate-based flame-retardant (FR) additives, as well as the addition of ILs. 10 The concept of self-extinguishing material is different from that of non-flammable one, in the sense that the first refers to a system that is inherently resistant to flames.
Indeed, a relevant route to reduce the flammability of liquid electrolytes, which at the same time has beneficial effects of their self-extinguishing properties, is based on the incorporation into common organic molecules of more electronegative substituents, such as fluorine, cyano, or sulfone groups. These substituted compounds generally displayed lower HOMO/LUMO levels if compared with pristine ones, which translate into a shift of the electrochemical stability window (ESW) towards higher potentials. Achiha et al. 39 reported the calculated variation of the HOMO/ LUMO energies for several ethers, esters and carbonates upon substitution of hydrogen with fluorine. The decrements in the HOMO and LUMO energies were of the order of few hundred kJ mol −1 (e.g. HOMO from -1099.4 to -1207.0 kJ mol −1 and LUMO from 2.0 to 132.4 kJ mol −1 for diethyl carbonate), and they were found to be roughly proportional to the numbers of the substituted fluorine atoms. These results suggested that the oxidation stability of organic compounds is improved by the fluorine substitution. At the same time, their reduction simultaneously became easy; i.e., their reduction potentials were elevated. This, in principle, is a disadvantage for cell working. However, charge/discharge experiments indicated that fluoro-ethers and fluoro-carbonates gave origin to quick formation of SEI on graphite upon reduction, and resulted good candidates as non-flammable solvents, because they not only improved the thermal and electrochemical oxidation stability of electrolyte solutions, but also increased initial Coulombic efficiencies. Therefore, the advantage of ESW shift induced by chemical substitution of hydrogen with more electronegative elements is twofold: (i) the improved stability towards oxidation renders these solvents advantageous with respect to cell overcharge, (ii) the reduced stability towards reduction may have a beneficial effect on the SEI formation, thus improving safety and cyclability. 39 Although a high number of additives were investigated during the last 15 years, 40 one of the most interesting compounds is still fluorinated ethylene carbonate (FEC), which was proposed as early as in 1999. 26 The irreversible capacity of a cell Li/KS15 graphite cycled at C/20 was reduced from ∼400 to 85 mAh g −1 with an electrolyte 1M LiPF 6 FEC/PC (1:1 v/v).
The design of non-flammable electrolytes must generally come to terms with a downward compromise on battery performance. In fact, the introduction of phosphorus-or fluorine-based FR additives may lead to bad passivation of the LIB carbonaceous anode, causing electrode exfoliation or electrolyte decomposition. [41][42][43] Moreover, to assure a reasonable self-extinguishing time (SET) of the order of 1 s per gram of electrolyte, which is considered to be a good value in determining the safety properties of the electrolyte, it is necessary to employ significant amounts of additive, which may further depress battery performances. 43 As an example, a Li-ion electrolyte with a 70% content of trimethyl phosphate (TMP) was tested on a graphite anode for 10 charge-discharge cycles. 44 The Authors showed that TMP exhibited poor reduction stability towards the graphite anode. They were able to partially overcome this issue with the help of three further additives, and by using graphite partially coated by amorphous carbon, and 2M LiN(SO 2 C 2 F 5 ) 2 salt. However, also electrolytes with a high content of TMP are still subjected to low flash points because of the remaining flammable solvent components. 45 Recently, the group of Winter reported a comparative study of several phosphorous-containing FRs, including (tris(2,2,2-trifluoroethyl)phosphate (TFP), tris(2,2,2-trifluoroethyl) phosphite (TTFPi), bis(2,2,2 trifluoroethyl)methylphosphonate (TFMP), (ethoxy)pentafluorocyclo-triphosphazene (PFPN) and (phenoxy)pentafluorocyclotriphosphazene (FPPN)). 46 The FRs were investigated using a standard electrolyte (1 M LiPF 6 in EC:DMC 1:1 wt%) concerning the structure-property relationships according to their SET, onset temperature of the thermal runaway, chemical and electrochemical stability. Cyclophosphazenes showed superior results concerning their first time of inflammation and thermal electrolyte stability. From the mechanistic point of view, it is accepted that these additives hinder the burning process by chemical reactions. At elevated temperatures, the additives split into radicals, which bind the hydrogen radicals formed during combustion. By binding these hydrogen radicals, reactive educts get consumed and the burning process is stopped.
Interesting opportunities are offered by ILs, which are intrinsically non-flammable, 47 and can improve safety and cyclability either as pure liquids, 27,48 or embedded in polymers to give GPEs. 28,49 The group of Passerini reported a full cell Sn-C/electrolyte/LiFePO 4 where the electrolyte was N-butyl-N-methylpyrrolidinium bis(trifluoromethanesulfonyl)imide (Pyr 14 TFSI) with bis(trifluorome-thanesulfonyl)imide (LiTFSI). 27 The electrolyte displayed an ionic conductivity of 7 mS cm −1 at 60C, and the full-cell delivered a maximum reversible capacity of about 160 mAh g −1 at a working voltage of about 3 V, corresponding to an estimated practical energy of about 160 Wh kg −1 . Beside the intrinsic safety related to the IL based electrolyte, the cell offered 2000 cycles without signs of decay, and satisfactory rate capability. Ferrari et al. 28 early reported a GPE based on poly(vinylidene fluoride)/hexafluoro propylene (PVdF-HFP) copolymer and N-methoxyethyl-N-methylpyrro-lidinium bis(trifluoromethanesulfonyl)-imide (PYRA 12O1 TFSI) ionic liquid, with LiTFSI and mesoporous SiO 2 (SBA-15). Besides the expected non-flammability properties, this electrolyte showed ESW as high as 4.8 V vs Li/Li + (depending on the amount of SBA-15), good reversibility, and 100% coulombic efficiency at C/5 for 200 cycles. In both these cases, the safety of the electrolyte was due to the presence of a relevant quantity of ionic liquid. Despite of their interesting properties, however, ILs are expensive, generally are more viscous than water and organic solvents (ILs viscosity values in the range 30-600 mPa.s were reported 47 ), and must be treated under dry conditions, which limits their industrial use.
Deep eutectic solvents (DESs) share many characteristics and properties with ILs, e.g. they are good media for metal treatments and solvents for green synthesis. As a matter of fact, the terms DES and IL were used interchangeably in the literature, although these are two different types of solvent. DESs are systems formed from a eutectic mixture of Lewis or Brønsted acids and bases which can contain a variety of anionic and/or cationic species. In contrast, ILs are made of systems composed primarily of one type of discrete anion and cation. 50 DESs contain large, non-symmetric ions with low lattice energy and hence low melting points. They are usually obtained by the complexation of a quaternary ammonium salt with a metal salt or hydrogen bond donor (HBD). The charge delocalization occurring through hydrogen bonding between e.g. a halide ion and the hydrogen-donor moiety is responsible for the decrease in the melting point with respect to the melting points of the individual components. 50 Figure 1 (left) shows some of the most common DES components.
DESs based on N-methylacetamide (MAc) with different lithium salts (LiN(CF 3 SO 2 ) 2 , LiPF 6 , LiNO 3 ) were proposed as electrolytes for lithium-ion cells. These systems, however, were liquid at room temperature only for Li molar fraction x < 0.35, which limited their ionic conductivity to about 1 mS cm −1 . 51 Aurbach and co-workers reported the preparation, thermal and transport properties and thermal stability of binary DESs based on mixtures of alkyl sulfonamides (methanesulfonamide, CH 3 SO 2 NH 2 , or MSA, and N, N-dimethylmethanesulfonamide, CH 3 SO 2 N(CH 3 ) 2 , or DMMSA) with lithium perfluoroalkylsulfonimide salts (LiN(FSO 2 ) 2 or LiFSI, and LiTFSI). 29 Ionic conductivity of 3 mS cm −1 at r.t. was reported for DMMSA-LiFSI 4:1 (see Fig. 1 (right)). This DES electrolyte was also cycled at C/10 against LiMn 1/3 Ni 1/3 Co 1/3 O 2 (NMC 111) cathode material. The Authors concluded that only the DMMSAbased glassy eutectics can be further considered for possible use in Li-ion cells, since MSA possesses two active amide hydrogen atoms, which makes this compound unsuitable for this use.
If compared with ILs, DESs may offer clear advantages in terms of cost, as well as of electrochemical properties, e.g. by increasing the Li + transference number because of the possible presence of a single Li + cation. 29 At present, the main limitation of these systems is related to the high reactivity towards the electrodes showed by mixtures, such as ethylene glycol/choline chloride (EG:ChCl), which, otherwise, would be very interesting for their low viscosity, high conductivity, and very low cost. The 0.5 M LiPF 6 /EG:ChCl electrolyte, in particular, displays ionic conductivity of 7.95 mS cm −1 at the room temperature. 52 Improving safety and durability: super-concentrated solutions.-The salt concentration of conventional organic liquid electrolytes generally spans in the range 0.5-1 mol dm −3 . 37 A promising recipe to obtain better performing electrolytes in terms of safety and durability is to use salt-concentrated ("solvent-in-salt") solutions. In fact, the electrochemical nature of organic solutions undergoes drastic changes over ca. 3 mol dm −3 , with some relevant advantages at the anode interface. It was shown that lithium salt/superconcentrated solutions can suppress dangerous dendritic lithium deposition. 53 Another relevant advantage is related to the SEI formation: it is widely recognized that SEI is formed on graphite by decomposition of EC upon reduction, and it effectively passivates the graphite surface and prevents further co-intercalation and decomposition of solvent molecules, allowing only lithium ion to migrate and to be intercalated. Such an effective surface film is not formed in propylene carbonate (PC)-based electrolyte solutions. However, PC is highly attractive as a component for electrolyte solutions for lithium-ion cells because of its superior ionic conductivity. The group of Ogumi early investigated electrochemical lithium intercalation within graphite in PC/LiN(SO 2 C 2 F 5 ) 2 0.82 M and 2.72 M electrolyte solutions. Lithium ion was reversibly intercalated into and deintercalated from graphite in the latter concentrated solution in spite of the use of pure PC as a solvent, whereas ceaseless solvent decomposition and intensive exfoliation of graphene layers occurred in the former solution. 54 The concept of solvent-in-salt was rationalized by Armand, Chen and co-workers, who investigated in a wide composition range (up to 7 mol dm −3 ) an electrolyte system containing Li[CF 3 SO 2 ) 2 N] (LiTFSI), 1,3-dioxolane (DOL): dimethoxyethane (DME) (1:1 by volume) for use in Li/S batteries. These electrolytes inhibited the dissolution of lithium polysulphide, effectively protected metallic lithium anodes against the formation of lithium dendrites and resulted in high lithium cycling efficiency, thus enhancing electrochemical performance (Fig. 2). 14 This approach was largely investigated in the subsequent years, and improved stability and cyclability were reported both for aqueous 55,56 and non-aqueous lithium cells. 14,57,58 These beneficial effects were generally attributed to the formation of a high-quality and stable SEI.
Recently, the group of Yamada achieved both intrinsic nonflammability (practically SET = 0) and excellent carbonaceous anode cyclability in a semi-cell graphite/electrolyte/Li, by increasing the electrolyte concentration up 3.3 M with NaN(SO 2 F) 2 or LiN(SO 2 F) 2 as the salt, and TMP as the solvent. 21 They obtained a threefold advantage with respect to common dilute electrolytes: (i) the highly flammable organic solvents were completely excluded from the cell, so providing a fire-extinguishing capability of the electrolyte up to high temperatures, (ii) the anode SEI was made of inorganic salts, in contrast to the usually solvent-derived organic inorganic hybrid layers, which display poor passivation ability, and (iii) the interactions between the cations and solvent molecules was enhanced, so further reducing the inherent volatility of the solvent. They applied the concept to both Li-ion and Na-ion cells, obtaining good cycling performance and high Coulombic efficiency at C/5 for more than one year (Fig. 3). The same concept was exploited by Shiga et al. 59 who investigated concentrated solutions of lithium bis (fluorosulfonyl) amide in (CF 3 CH 2 O) 2 (NR 1 R 2 )PO (R 1 =CH 3 , R 2 =C 6 H 5 ). Besides flame resistance, also Li stripping/plating performance could be improved by high salt concentration.
Whereas the outstanding functional properties of super-concentrated solutions have been thoroughly demonstrated, more knowledge is needed about ion-ion and ion-solvent interactions. Recently, Deng et al. 60 collected natural abundance 17 O NMR spectra of LiTFSI in EC, PC and ethyl methyl carbonate (EMC), and noticed that ion pairs association occurred in highly concentrated LiTFSI/ EC/EMC/PC solutions. Specifically, Li + ion is coordinated by four oxygen atoms by the organic solvent molecules and the anion, to form a first solvation shell. At high concentrations of LiTFSI, the preferred structures are (LiTFSI) 2 (PC) 4 for LiTFSI in PC, and (LiTFSI) 2 (EMC) 3 for LiTFSI in EMC. Despite of this information, much work must still be done to fully understand the physical chemistry of the concentrated solutions and to unveil the relationships among composition, reactivity and functional properties, chiefly as far as SEI formation is concerned.

Quasi-Solid Electrolytes
Quasi-solid electrolytes are expected to allow efficient strategies towards the improvement of many functional properties of separators in both lithium and post-lithium batteries, including: (i) maximum attainable power density, (ii) chemical and electrochemical stability, and (iii) safety. The power density can be improved by reducing the electrolyte resistance (through the fabrication of thinner films), or its resistivity (by increasing carrier concentration and/or mobility), or even by increasing the cation transference number. Separator stability and safety are largely related to the capability to control formation and growth of lithium dendrites, which makes this topic of substantial interest for battery chemistries based on lithium metal anode, e.g. LMB, Li-S and Li-air.
Improving power density: patternable membranes.-Patterning processes are used in microelectronics to produce well-featured structures at low cost. Basically, this technology allows reaching a fine spatial control even with sub-micron resolution. With a view to the Li battery, the fabrication of electrolytes with non-planar geometries could improve the attainable power density, chiefly by reducing electrical resistance, also enabling novel production processes suitable for integration with microelectronics for the development of on-chip batteries.
Patterned membranes obtained via photo-patterning, lithography and 3D printing were initially proposed as proton exchange systems for polymer fuel cells. Feature sizes of tens of microns on the membrane surface resulted in significant power density enhancement and improved electrode/electrolyte interface. [61][62][63] Hickner et al. investigated the beneficial influence of 3D micro-patterning on the properties, e.g. electric resistance, of photo-polymerized anionexchange membranes. 64,65 Recently, this technology started to catch on also in case of lithium-ion batteries. Dunn and co-workers described a photopatterning process to produce a silica-based membrane. 19 Here, a UV photo-patterning process was adapted to the synthesis of ionogels, which are obtained by confining a liquid electrolyte in a mesoporous inorganic (e.g. silica) matrix (see below, paragraph 2.2.3). After a UV exposition at 365 nm and subsequent aging step, a pattern of ordered circles with diameter of about 250 μm was obtained. The same group developed a photo-patterned, lithium-ion conducting solid electrolyte through the modification of the wellknown negative SU-8 photoresist, with LiClO 4 as the lithium salt. 30 (Figs. 4a-4c).
The resulting m-SU-8 gel showed a micrometer-scale resolution with features of 30 μm in diameter and heights up to 10 μm. The high control achievable on the pattern resolution made it possible to design the electrolyte directly on-chip, and to fabricate modulable arrays. The outstanding mechanical stability of the gel electrolyte also guarantees the retention of the micro-scale patterning. By decreasing the cross-linking degree of the resulting gel electrolyte, and by enhancing the number of sites available for the Li hopping, a conductivity value of about 5 × 10 −5 S cm −1 was obtained at room temperature. Electrochemical stability exceeding 5V and good interfacial stability with Li metal anode were also reported. Microstereolithography was also used to fabricate a separator (UV-cured PEO-based gel polymer electrolyte (GPE)) for 3D lithium ion microbatteries. This method can be directly integrated into cell assembly process, so making easier the micro-battery fabrication. 66 Blake et al. proposed a novel approach for 3D printing Li-ion electrolytes with controlled porosity. 67 A filamentary ink, consisting of poly(vinylidene fluoride) (PVdF) and nanosized Al 2 O 3 , was used to extrude multilayer composite systems which originated gels upon absorption of a liquid electrolyte (LiPF 6 1M in EC-DEC). The result was a separator with electrochemical performances similar to those of Celgard TM even at high C-rates, but with better thermal and mechanical stability. Stable specific capacity of about 160 mAh g −1 was preliminary observed at C/2 for at least 100 cycles in case of a printed electrode membrane assembly (PEMA), prepared by depositing the printed nanocomposite electrolyte directly over an olivine cathode. This technology can be also extended to the fabrication of 3D composite electrodes, enabling complete battery printing (also with interdigitated configuration) and its integration in flexible power sources or onto non-planar surface devices (see Figs. 5a-5e).
Very recently, this method was applied to the fabrication of a hybrid electrolyte made of poly(vinylidene fluoride)-hexafluoropropylene (PVdF-HFP) and an ionic liquid, by using an ink whose rheology was modulated by adding nanosized TiO 2 as the filler. 31 High-temperature printing prevented shrinkage and loss of structural integrity, which in contrast may be encountered in case of similar fabrication methods. The overall interfacial resistance was significantly reduced due to the formation of a dense layer between the porous polymer-based electrolyte and the electrode. The cell performances were significantly better than those of a battery produced by the standard casting method. Coulombic efficiency of about 98% and limited capacity loss of 13% were demonstrated in case of a Li/ink/MnO 2 half-cell.
Although these patternable membranes are still far from full technological exploitation, and at present do not offer relevant increases in current densities, as those obtained by Hickner for fuel cells, 64,65 this technology may offer relevant advantages in terms of system integration (e.g. by improving interfacial contacts) and full 3D battery design.
Ink writing is also useful for the preparation of ceramic electrolytes with 3D architecture. Indeed, such a technique can overcome some critical issues, typical of the die-pressing and tape-casting approaches, as high interfacial resistance, due to poor contact with the electrode, and high bulk resistance, related to excessive electrolyte thickness. McOwen et al. developed multiple ink formulations enabling 3D printing of unique solid electrolyte microstructures with tunable properties. 68 The inks were used to fabricate a variety of patterns (Fig. 6), which were then sintered to give thin, non-planar, complex architectures composed only of Li 7 La 3 Zr 2 O 12 solid electrolyte. The reported ink compositions could be used as a model recipe for preparing other ceramic electrolytes. Optimizing such 3D printing technology could lead to all-solid-state batteries with significantly reduced full cell resistance, higher energy and higher power density. In principle, 3D printing of ceramic electrolytes can indeed simplify the battery preparation, although a sintering step is still needed. We stress that these patterned membranes, as well as several systems below discussed, including most Metal Organic Frameworks, Covalent Organic Frameworks, and soggy sands, are just inert or quasi-inert components that can be used to form a porous membrane/separator where the liquid or liquid-like electrolyte is hosted.
Improving power density: metal organic frameworks (MOFs).-Metal Organic Frameworks are hybrid structures, obtained by reticular synthesis, through which organic and inorganic moieties are strongly linked together. Due to the high component flexibility in terms of chemical nature, size and geometry, more than 20,000 MOFs have been synthesized during the last years. Basically, the organic unit is a di-topic or poly-topic carboxylate which is anchored to a metal-including center, yielding a robust crystalline MOF structure with porosity degree typically higher than 50%. By properly changing the system components, ultra-high porosity may be achieved with pore diameters in the micro-meso domain, namely lower than 10 nm. This results in surface area values ranging from 1000 to 10000 m 2 g −1 , exceeding those typically observed in case of conventional porous materials as zeolites and carbons. In addition, MOFs thermal and chemical stabilities allow post-synthesis functionalization, which enable their use in a wide range of applications, from gas storage, separation and adsorption to catalysis and ion and/ or electron conduction. 69 Recently, MOFs emerged as a promising category of functional materials for electrochemical energy storage and conversion technologies, more specifically in case of fuel cells and batteries. 70 Remarkable proton conduction has been demonstrated in pristine MOFs, functionalized by acidic units, 71 both under water-mediated and anhydrous conditions. Efficient ion-transport is favored by the open-pore channels that act as ion-guest accessible voids. Due to the Li compatibility with these open structures, MOFs are also excellent platforms for designing Li-ion conductors. 72 Basically, the main advantage of MOF electrolytes is that the metallic center is often unsaturated, which can bond the anion of the Li salt, leaving the cation relatively free to move along the pores. Such a feature results in high Li + transference numbers, which can easily be higher than 0.6. 73 In  32 In presence of stoichiometric amounts of Li halide or Li pseudo-halides as the salt, this system undergoes a reversible single crystal transition from the neutral to the anionic form. In consequence of this, the salt anion is bonded to the metal centers and Li ions are free to move through the PC-filled 1D pores. Cationic transference numbers in some case exceeding 0.7 were obtained by means of Bruce-Evans approach. The preferential formation of  anionic MIT-20 demonstrated that Cu 2+ is unique among the firstrow transition metals in thermodynamically favoring the anionic phase over the neutral one, in case of high chlorine concentration. The resulting single-ion conductor exhibited conductivity higher than 4 × 10 −5 s cm −1 at 25°C, which further increased above 10 −4 S cm −1 , with an activation energy of 0.16 eV, in presence of Li + non-stoichiometry coming from the addition of proper amount of LiBF 4 (Figs. 7a-7c). For this reason, the authors classified the MIT-20-LiBF 4 as a novel MOF-based superionic conductor.
Owing to their open microporous structure and high adsorption capability, MOFs were also investigated as a reservoir for liquid electrolytes. 73 In this case, the mechanism for Li + transport is more complex. First, Li ions diffuse through the channel-like pores; then the MOF micro-porosity nano-confines the liquid, which is easily retained in the rigid matrix by capillary effects. Finally, the metal center acts as the anion coordination unit, allowing Li + ion to freely hop from one site to another. The result is a hybrid conductor with high transference number, high conductivity and high structural, mechanical and thermal stability.
In 2011 long's group proposed a Mg-based MOF quasi-solid electrolyte, Mg 2 (dobdc), obtained by soaking it in a solution LiBF 4 -EC-DEC. 74 The ionic conductivity was 3.1 × 10 −4 Scm −1 at 300 K and the activation energy obtained from variabletemperature measurements was 0.15 eV. Such promising results were obtained by allowing uptake of a lithium alkoxide in the MOF, before soaking the liquid component. The authors found that the alignment of the conduction channels in the polycrystalline particles may contribute to enhance the conductivity, and that the ion transport is dominated by intra-particle processes rather than by boundary ones.
More recently, Dunn et al. reported the design and fabrication of MOFs-based pseudo-solid electrolytes with biomimetic ionic channels for fast and efficient Li ion transport (Fig. 8). 75 They started by using, as the scaffold, the well-known MOF HKUST-1, consisting of Cu(II) paddle wheels and benzene-1,3,5-tricarboxylate as linkers. Such a system has 3D pore channels with pore size of about 1 nm. By removing the coordinated water molecules, HKUST-1 with unsaturated (or open) metal centers (OMC) was obtained. In presence of LiClO 4 -PC as the liquid electrolyte, perchlorate anions were spontaneously anchored to the OMC, so forming negativelycharged channels enabling the Li + transport with low activation energy. This approach led to several systems with superionic conduction, in some case exceeding 1 mS cm −1 , activation energy around 0.21 eV and t + > 0.6, which were obtained by changing the OMC and the pores dimension. The best results were obtained by increasing the MOF pore diameter, likely due to a reduction of the tortuosity.
MOFs-based quasi-solid electrolytes were also produced by using ILs as MOF modifiers or even as the liquid component. In the first case, metal-organic frameworks were produced in ILs by electrochemical methods, obtaining an ionogel-like electrolyte doped by a Li salt (LiTFSI), with ionic conductivity ranging from 10 −3 to 10 −2 S cm at room temperature, wide electrochemical stability window and a reversible capacity of about 3000 mAhg −1 for 10 cycles at C/10 in case of Li/MOF electrolyte/Si anodic half-cell. 76 Coulombic efficiency of the cell was observed to be ∼90% which indicated some electrochemical reaction was occurring, resulting in the decomposition of the electrolyte material.
The second route was followed by Fujie et al., who investigated the Li + dynamics of a Li-ion doped IL (EMI-TFSA) incorporated in the microporous structure of ZIF-8 MOF. 77 By means of NMR investigations, they demonstrated that the activation energy for the diffusion of Li + in the micropores of ZIF-8 is comparable to that of the IL bulk. These results could signify that the Li ions diffuse through the micropores via an exchange mechanism with solvating TFSAanions, similarly to what occurs for the protons in case of Grotthuss mechanism. Wang et al. proposed a quasi-solid electrolyte, based on a MOF and a Li-based IL (EMIM-TFSI), where the porous framework ensures a stable 3D open structure where the liquid component retains its high ionic conductivity (∼3 × 10 −4 ohm −1 cm −1 ). Such a 3D framework may provide several direct-contact points between the confined liquid and the cathode, forming abundant nano-wetted interfaces with small resistance to favor Li + transport kinetics. This resulted in good capacity retention over a wide temperature range (from -20C to 150C) in case of LiFePO 4 cathode, uniform Li deposition and excellent mechanical stability (see Fig. 9). 33 Summarizing, MOF-based electrolytes appear to be interesting systems to allow high power density in post-lithium batteries, chiefly thanks to the possibility to obtain high lithium transference numbers. However, the reports of MOF-based electrolytes for LIBs are still rare and more work is required for full systems comprehension, chiefly as far as concerns the electrochemical stability. Very likely, in fact, the Cu(II) ions could be reduced at the potential of the battery negative electrode, and this must be taken in careful consideration, provided that detailed experimental evidences are not yet available in the literature.
Improving Li anode stability.-As stated, the use of lithium metal anode is mandatory to reach the stringent energy density requirements for automotive. Recently, the most relevant strategies to control and engineer the Li/electrolyte interface in order to obtain a robust and stable SEI were reviewed. Attention was devoted to the engineering of both the electrolyte, and the interface at the nanoscale level. 5 A thorough analysis at the cell level has been recently reported by Liu et al., 78 who discussed key factors such as cathode loading, electrolyte amount and Li foil thickness that impact the celllevel cycle life. Furthermore, they identified several important strategies to reduce electrolyte-Li reaction, protect Li surfaces and stabilize anode architectures for long-cycling high-specific-energy cells. On the electrolyte side, concentrated systems or localized highconcentration systems are considered effective in reducing the reactivity towards the electrolyte, however significant research is still needed to promote a dense, not porous Li deposition. Ultrathin, flexible solid electrolytes are desirable to separate the Li metal from the electrolyte. Mechanically and electrochemically stable polymers other than PEO could be studied to this aim. In addition to the experimental work, theoretical studies were also made in terms of continuum linear stability analysis and DFT investigation of the surface energy and interfacial transport of Li + across the interface. 8,79 Following all these analyses, several breakthrough pathways to prevent dendrite growth and to stabilize the Li metal anode can be considered, including: (i) Li surface passivation before cycling to promote fast metal deposition at the interfaces and to protect it from parasitic side reactions with the liquid electrolyte. Such artificial SEI acts as a strong physical barrier (mechanical modulus of the order of gigapascal) against dendrites propagation 80-82 ; (ii) use of film-forming halide salt additives in a conventional liquid electrolyte [83][84][85] ; (iii) use of fluorinated compounds, e.g. fluoroethylene carbonate (FEC), to produce a uniform film which favors Li + transport across the SEI and avoids dendrite propagation 86 ; (iv) design of electrolytes with improved mechanical properties, specifically: (i) liquid/solid hybrid systems with percolating networks; (ii) electrolytes with porosity tailored at the nanoscale; (iii) ionogels.
The strategy (iv) is particularly interesting, as it combines liquidlike fast ion transport resulting from the liquid phase trapped into the solid skeleton/network, and high mechanical modulus due to rigid ceramic or solid polymer frameworks. Whereas the hybrid nature is a common denominator for all these innovative systems, several subclasses can be identified, depending on the design of the inorganic matrix, on the liquid/solid composition ratio and of physical vs chemical nature of the interactions among the different phases.
"Soggy sand" electrolytes.-Soggy-sand electrolytes are a very promising class of Li + conducting liquid/solid composites, with interesting transport and mechanical properties. 87 They consist of a fine dispersion of uniformly nano-sized mono-or bi-functional oxide particles (e.g. Al 2 O 3 , SiO 2 , etc) in a liquid electrolyte, e.g. LiPF 6 -organic carbonate or LiClO 4 -PEG, in properly optimised ratio to give a composite with a soggy sand consistency. 88 Once the oxide component is dispersed into the liquid electrolyte, the salt is dissociated into Li + X − ion pairs, and the anions are absorbed on the oxide surface, so increasing the concentration of free Li + in the space charger layer around the oxide particles. At a specific oxide volume fraction, φ, the colloidal dispersion forms a fractal percolating network through cluster-cluster aggregation, resulting in the overlapping of the space charge and consequent increase in ionic conductivity. The transport mechanism does not depend only on the anion adsorption step on the particle, but is strictly related to a fine equilibrium among several factors, which are crucial for percolation and for an optimal interconnection of the network, including: (i) concentration, (ii) surface energy of the oxide particles, (iii) space charge, (iv) size, shape, volume fraction and nature of the functional group decorating the oxide surface. Since the oxide network is not spatially fixed in the liquid, even a small variation of these conditions may lead to backbone coarsening and, if not kinetically controlled, to eventual sedimentation with a consequent decrease of the overall ionic conductivity. 34 For more details on the ion transport mechanism see Appendix B. The result of a correct design is the optimal matching among electrical properties (conductivity and transference number), and excellent mechanical properties (mechanical stability, shape-ability and electrode wettability). The addition of even small amounts of oxide (e.g. mesoporous or fumed silica) (φ < 0.01) can enhance the ionic conductivity of 1-5 times with respect to the pure liquid electrolytes, depending on parameters related to solvent, salt and filler. Room temperature conductivity exceeding 10 mS cm −1 was observed for solid/liquid composite 1M LiPF 6 -EC/DMC:SiO 2 , resulting from of an effective particle percolation network, for φ = 0.2-0.3 (Fig. 10). 89 The Li + transference number also depends on the parameter φ. Upon oxide particles addition, t + > 0.60 may be observed in case of good percolating networks. High transference numbers may be still observed also at φ values where the conductivity starts to drop because of dilution effects. 34 Further research work is needed to better clarify the relationships among composition, ion-pairs effects and transport properties.
Electrochemical tests on the electrode/electrolyte interface and galvanostatic cycling demonstrated that the system functional properties are affected by the filler surface chemistry, the size of the surface decorating units and, finally, by the spatial coverage of such functional groups. As a rule of thumb, it seems that a homogeneous dispersion of hydrophobic particles improves the stability of the phase boundary more than Janus and hydrophilic particles, so reducing the undesired lithium losses and maximising the cycling stability with respect to heterogeneous distributions. 89 Figure 11 shows the galvanostatic charge-discharge profiles at 25°C and 60 mA g −1 of Li/graphite cells including soggy-sands electrolytes based on hydrophobic silica (SiO 2 ) and silica modified by hydrophobic, hydrophilic and Janus bifunctional moieties. It can be easy noticed that composites containing hydrophobic silica offer minimum irreversible capacity, maximise the cycle stability and the electrode-electrolyte interface. The enhanced stability might derive from the scavenging role of the hydrophobic silica particles in capturing impurities of water and/or liquid solvents. Depending on the SiO 2 volume fraction, the hydrophobic filler could minimise the contact area of both the electrodes towards species as O 2− , OH − , and reduce the passivation processes at the interface.
Despite of such impressive results, further work is still required. The major weakness of soggy-sands electrolytes is the difficult kinetic control of the morphology, which could affect the overall conductivity. A better reproducibility may certainly be attained if a denser percolation network is fabricated, and this could be allowed by several freedom degrees in the composite optimization. The actual challenge is to prepare a composite with high particle fraction with homogenous dispersion, local electrical activity and sufficient space charge percolation, namely a closely packed filler-sphere arrangement still able to offer percolation (Fig. 10c).
Physical nanoconfinement of liquid electrolytes: micro/mesoporous membranes.-Quasi-solid electrolytes may also be designed by exploiting physical nano-confinement of liquids. Here, the immobilization of liquid electrolyte into the matrix takes place through weak interactions, as hydrophobic, π-π stacking or hydrogen bonds, which are formed during the impregnation procedure. The liquid component may a conventional carbonate-based electrolyte or an ionic liquid. 90,91 Several scaffolds, with different microstructure and porosity (meso-, micro-or hierarchical) were tested for the liquid nano-confinement, including inorganic oxides (e.g. SiO 2 , ZrO 2 , CeO 2 ), Covalent-Organic-Frameworks (COFs), and zeolites. 92 However, most of the systems reported in the literature are based on SiO 2 networks, due to its pristine beneficial properties, as high mechanical, chemical and thermal stability, large easy-to-modulate surface area and easiness of preparation. Recently, efficient dendrites suppression was obtained with nano-architectures of silica hollow spheres (HS) able to strongly confine the liquid electrolyte. 93 The HS array was arranged in a close-packing network, which acted as a mechanically-strong separator able to hinder dendrite growth, so preserving the cell from short-circuits. After the scaffold was soaked with the liquid, the resulting system behaved like a quasi-solid electrolyte with conductivity exceeding 1 mS cm −1 at room temperature. Furthermore, stable voltage profiles during charging and discharging at a constant current were collected on a symmetric Li/soaked HS/Li. No voltage drops were observed in consequence of long-term operation at I > 0.16 mA cm −2 . Even at current higher than 1 mA cm −2 there was no evidence of cell failure related to the dendrite proliferation (Fig. 12).
A significant improvement to stabilise LMB operation was recently reported by Archer's group. 94 They prepared hybrid materials by infusing liquid electrolytes into nanoporous ceramics, polymers or their composites (e.g. poly(styrene)-PEO and inorganic porous fillers) and obtained good Li deposition and stable cycling performance at various C-rates by using intercalation cathodes. This approach confirmed a recent theoretical study on Li electrodeposition in elastic media that discussed the relationships between dendrite nucleation and membrane porous structure. 35 Specifically, this analysis suggested that the dendrite growth can be suppressed by constraining the Li deposition on a length-scale below a critical value, λ crit , at which the sign of a growth parameter, ζ, changes from negative to positive (see Appendix C). This may be reached by using porous membranes with shear modulus 1.6-1.8 times higher than that of Li metal (4.2 GPa). Successful results were obtained with γ-Al 2 O 3 scaffolds with channel-like pores, ranging between 20 nm and 200 nm in diameter. With conventional liquid electrolytes and standard cell test protocols, Li electrodeposition became unstable at λ crit ≅ 200 nm, which means that the dendrites start to grow in case of alumina membrane with λ crit > 200 nm. The pore geometry greatly affected ion migration. Ionic conductivity comparable to that of the liquid phase was, in fact, observed for the alumina scaffold with pore diameters of about 160-200 nm. Finally, Archer and coworkers proved that a liquid electrolyte constrained in a porous medium with a small surface charge may act as an "ion rectifier", inhibiting the transport of ions with the same charge of the pore walls. This aspect resulted in an impressive enhancement of t Li + .
Indeed, transference numbers ranging between 0.5 and 0.9 were determined as a function of the electrolyte concentration in case of liquid-like conducting γ-Al 2 O 3 with small pores (Fig. 13). 95 Chemical nanoconfinement of liquid electrolytes: ionogels.-The other side of nanoconfinement is represented by the chemical entrapment of a liquid in a solid matrix. Ionogels are an interesting class of solid-like electrolytes where an ionic liquid is trapped into a solid matrix through a non-aqueous sol-gel process, leading to a  glassy system with high homogeneity, highly cross-linked network and fluid-like dynamics. 90,91 Several inorganic matrices (e.g. SiO 2 , ZrO 2 , CeO 2 ) with different microstructure and porosity (meso-, micro-or hierarchical), were tested for the nanoconfinement of ILs. 92 However, most of the ionogels reported in the literature are based on SiO 2 networks, for the same reasons already discussed in the case of nanoconfined liquid electrolytes (see Section 2.2.2). SiO 2 -based ionogels are generally synthesized by means of the typical one-step reaction among a silicon alkoxide (e.g. tetra methyl ortho silicate, TMOS) as the precursor, an acidic component (e.g. formic acid) as the catalyst, and the ionic liquid-based electrolyte as the solvent. Upon gelation, transparent and homogeneous monoliths with tunable thickness may be obtained. [90][91][92]96 The proper formulation of the three components and the gelation time are crucial factors affecting the pore structure of the scaffold and, consequently, the final mechanical properties of the ionogel, which can be tuned from brittle systems to mechanically-compliant ones. 36 This results in excellent electrochemical properties, in terms of good ion transport and wide voltage window, which, however, are strongly dependent on the correlation among matrix microstructure and physicochemical properties of IL-based electrolyte (Fig. 14).
Several ionic liquids were investigated, including pyrrolidinium, imidazolium, and pyridinium-based systems, giving ionogels with conductivity higher than 1 mS cm −1 at room temperature, even at low IL concentration. The most interesting lithium salt was LiTFSI, due to its high stability against water, acids and oxygen. In some cases, the ionogel conductivity even exceeded the value of the lithium saltdoped IL itself. Such an enhancement was attributed to several reasons: (i) faster Li + migration in the silica network, due to the weakening of the Li + X − interactions induced by the Si-O-Si units, (ii) the presence of well interconnected SiO 2 tetrahedra allowing ease percolation; iii) a large contact area between the silica particles and the IL. 90 Basically, the ion transport has Arrhenian behavior for T > 50°C with activation energies of the order of 0.2 eV, which may be associated to improved fluids dynamics of the confined ionic liquids. In contrast, Vogel-Tammann-Fulcher (VTF) model better fits the conductivity behavior at T < 50°C, where ion migration is likely dominated by the molecular relaxation and by swinging motions of IL anions tethered to the inorganic nanoparticles. 36 Recently, also Deep Eutectic Solvents (DESs) were tested for the preparation of ionogels. Specifically, r.t. conductivity of 1.4 mS cm −1 and a broad electrochemical window were demonstrated on systems made by SiO 2 and LiTFSI-doped n-methylacetamide (NMAC). The Li/ionogel/LiFePO 4 half-cell delivered a stable and reversible specific capacity exceeding 100 mAhg −1 for over 60 cycles, at 17°C and 0.1C (Fig. 14b). 18 These results configure very promising strategies to prepare performing electrolytes for LMBs. However, further experimental and modelling work is needed to better address the correlation between structure and functional properties, which is fundamental for the rational design of ionogels with the required electrochemical properties.
Improving safety: thermo-responsive gel electrolytes.-To assure good electrochemical performances in terms of reversibility, and therefore long cycle life, current LIBs are designed to work on limited range of temperature, current density and voltage. 97 Nevertheless, short circuits due to dendrites, overcharging or other abuse conditions can determine dramatic increases of internal temperature and pressure, which can lead to fire or explosion (see Appendix A). 98,99 To avoid thermal runaway, or at least to limit its effects, the best strategy is indeed the design of inherently safe electrode and/or electrolyte materials. 100 Among the latter ones, gel polymer electrolytes (GPEs) with thermo-responsive properties are of increasing interest. Thermo-responsive GPEs include systems able to undergo reversible phase separation/phase transition (e.g. sol-gel) upon temperature increase. Kelly et al. proposed a thermally-responsive polymer electrolyte based on N-isopropylacrylamide (pNIPAM), which governed the thermal properties, and acrylic acid, which provided the H + ions. 101 As the polymer underwent the phase transition, the local environment around the acid groups was reversibly switched, so decreasing ion concentration and conductivity of the solution. Later, the same group extended this concept to LIBs, by reporting a Li 4 Ti 5 O 12 /GPE/LiFePO 4 cell where the GPE was made of poly(benzyl methacrylate) (PBMA) and 1-ethyl-3methylimidazolium bis(trifluoromethanesulfonyl)imide [EMIM] [TFSI] IL with LiTFSI at various molar concentrations. 102 Upon raising the temperature, the polymer acted as an electronic insulator by coating the electrodes, so determining cell shutdown (Figs. 15a and 15b). However, the GPE performances were strongly influenced by LiTFSI concentration. Sol-gel transition has been recently exploited in model systems based on pNIPAM/acrylamide 103 or PEO-PPO-PEO block copolymer 104 and encompassing H + as the charge carriers, which could find application in supercapacitors or smart windows. Although these concepts are promising, much work must be still done to adapt them to Li + conducting devices, chiefly because of the interactions with lithium salt, which make the shutdown mechanism less efficient. Moreover, it must be pointed out that the reported reduction in conductivity/capacity (about one order of magnitude from 60°C to 150°C, see Ref. 101), is likely not yet enough to envisage use in high power/high energy batteries.
Another route pointed towards the fabrication of shutdown separators. The group of Cui proposed a core-shell microfiber separator made of PVdF-HFP where triphenyl phosphate (TPP), a well-known flame retardant, was encapsulated. 105 In case of thermal runaway, PVdF-HFP undergoes melting, and releases the encapsulated TPP into the electrolyte, so inhibiting the combustion (Figs. 14c and 14d). Encapsulation reduces TPP negative effects on cyclability. Whereas it is well known that the crystalline portion of PVdF-HFP is not soluble in common organic battery solvents like EC and DEC, 106 whereas the amorphous fraction indeed undergoes swelling, the Authors do not specify what is the relative molar fractions of PVdF and HFP, which determine the crystalline/ amorphous ratio.
Jiang et al. proposed a shutdown separator of poly(lactic acid) @poly(butylene succinate) (PLA@PBS) fabricated by a simple coaxial electrospinning process. 107 However, the shutdown separators are "single-time" switches, which somehow limits their practical interest. The thermal stability of GPEs can be improved by adding organic and/or inorganic additives. This route, indeed, is not new, 5 and does not lead, per se, to safer electrolytes. These aspects were discussed in a recent review 108 and will not further treated here.
Thermo-responsive GPEs are indeed a promising route to increase the inherent safety of lithium batteries. Here, the main challenge is to design reversible systems able to switch in very short times. In this sense, although strictly speaking it is not an electrolyte, the groups of Cui and Bao proposed a fast and reversible thermoresponsive switching material consisting of graphene-coated spiky nickel nanoparticles mixed in a polyolefine matrix with high thermal expansion coefficient. 109 The electrical conductivity (∼50 s cm −1 ) of this coated layer decreased within one second by seven to eight orders of magnitude on reaching the transition temperature and spontaneously recovered at room temperature. The sensitivity to temperature was estimated to be 10 3 -10 4 times higher sensitivity than previous switching devices.

Summary and Outlook
It is clear, and widely accepted, that the quest for new-generation electrolytes is closely connected to the effort of replacing the graphite-based anode with a lithium metal one. This will strongly improve the performance of Li-ion batteries, as well as help to make new chemistries, such as Li-air and Li-S, industrially available. To this aim, it is mandatory to master the high reactivity of lithium and its tendency to the formation of dendrites. Cui and coworkers 5 clearly pointed out the need of a careful engineering of the interface and of the electrolyte. In this frame, the use of solid electrolytes with high mechanical modulus and chemical stability could indeed make a good job.
On the other hand, the development of liquid or quasi-solid electrolytes, in case coupled with thin ceramic layers, can ensure high ionic conductivity and better long-term contact, upon cycling, among the cell compartments. Indeed, systems able to control the formation of dendrite must proceed hand in hand with the need to ensure high currents and intrinsic thermal, chemical and electrochemical stability. Likely, it will be necessary to couple different strategies, including: (i) the reduction of the anion transference number, which has beneficial effects on the formation of dendrites, 8 (ii) the design of systems with low reactivity and high conductivity, and (iii) the formation of stable inorganic SEI, without strongly depressing the ionic conductivity. Table II shows the advantages and drawbacks of the electrolyte classes discussed in this review, together with some information on the main issues which are currently limiting their industrial exploitation. Figure 16 classifies the same classes of in terms of their relevant functional property/ properties and physical nature (liquid vs quasi-solid). The systems marked in bold are expected to be the most interesting ones at least in the medium term (<2025).
Super-concentrated liquid electrolytes, possibly with the use of DESs, appear very promising, both to increase cell safety and cycle life by reducing/control dendrite formation and assuring true non-flammability. Moreover, the solvent components could be chosen among green and cheap chemicals. Indeed, the use of liquid (although concentrated) electrolytes should allow exploiting the present separators technology with relevant industrial advantages. Self-extinguishing electrolytes based on ionic liquids have reached such a degree of development as to imagine their use even in the short term. This concept, however, is destined to merge with that of super-concentrated solutions.
MOFs and, in case, COFs (Covalent Organic Frameworks) are also promising in the medium term, chiefly for increasing attainable power density by means of anion blocking. At the same time, their quasi-solid nature can assure good resistance towards dendrite formation. Here, the main issues are related to film processability.
In a future, it is expected that interesting opportunities will be assured by the development and exploitation of quasi-solid singleion conductors with high ionic conductivity. Recently, the group of Di Noto reported the development of lithiated-fluorinated titania nanoparticles entrapping ILs, which showed conductivity in excess of 10 −2 ohm −1 cm −1 . 110 If these single-ion conductors would reveal enough electrochemical stability, chiefly towards reduction, they could envisage a ground-breaking opportunity for the development of post-lithium batteries.
Another strategy to be considered towards more stable batteries is the study of self-healing/self-repairing electrolytes. As stated before, this route has so far been undertaken chiefly for proton conductors and must be extended to lithium and sodium carriers. To date, polymers able of self-repairing through extended hydrogen bonds  networks have been proposed as binders for Si anodes. 111 Supramolecular chemistry concepts 112 could help to extend this approach to separators.
Besides the improvement of the functional performance of electrolytes, it will be also important to concentrate efforts on aspects related to their sustainable production, also in the frame of the modern concepts of circular economy. 113 Therefore, the attention must be devoted to the study of water-based electrolytes, which can be produced with low environmental impact technologies. The problem of the low intrinsic electrochemical stability of water-based systems can be solved with aqueous-organic hybrid systems, as recently proposed by Wang et al. 114 Finally, in order to ensure the sustainability of production processes, it will be necessary to operate in different directions. On one hand, the attention should be directed to systems where lithium can be replaced by cheaper sodium, or with divalent cations (calcium, magnesium), or even halide ions. On the other hand, it will be appropriate to investigate matrices based on biopolymers obtained by bacterial synthesis from solid or quasi-solid waste, from biomass, or even from processing residues (secondary raw materials). Soggy-Sand Electrolytes Soggy-sand electrolytes show synergistic electrical properties, which include the enhanced conductivity of one ion moiety (typically cation) and the decreased conductivity of the counter ion. The ion transport occurs via a complex mechanism involving both a local mechanism and a long-range transport along the inorganic filler network. In a soggy-sand system, the charge carrier concentration in salt-containing solvents is enhanced around the filler particles. This implies: (i) an adsorption force of the filler particles with respect to the counter ions, and (ii) an association/dissociation equilibrium of the ion that locally increases its conductivity in the space-charge zones and decreases that of the counter ion. 87 If the cation transport needs to be improved, the surface of the filler particle must be acidic. Under the assumption that the counterion conductivity may be ignored, the mean space charge conductivity, σ + ex , corresponds to σ total ex , as defined by the following equation: where Θ is the measure of the adsorption strength, σ ∞ is the bulk conductivity and the superscript "ex" indicates the excess value over σ ∞ . The local effect is not sufficient to describe the overall conductivity, because the distribution of the filler particle establishes a long-range transport, especially in case of high surface area and continuously connected surfaces. Therefore, the overall conductivity is obtained by parallel and serial switching of surface and bulk concentration, generally defined as: where β ∞ and β L indicate the proportion of surface and bulk pathways contributing to σ m, respectively, ϕ is the volume fraction of a monolayer of filler particles, and ϕ L is a term including the surface-to-volume ratio.
In soggy-sand electrolytes, the hits and sticks processes involving the liquid and solid phases lead to fractal structures with percolation thresholds which are much lower than those typically observed in case of more conventional composites (around 0.3).
After reaching a maximum, the overall conductivity, σ m , decreases due to the blocking of the space charge layers. Similarly, t + reaches a maximum. After that, the blocking of bulk pathways occurs, both β ∞ and σ m decrease, contrary to the transference number, which benefits of such blocking effects.
Further modelling studies reveal that the percolation threshold tends to 0 for particle diameters very close to 0. This suggests that significant effects are expected in case of nano-sized fillers already at small volume fraction, ϕ.
The ζ sign, resulted by changing the electrolyte parameters (e.g. modulus, salt concentration, porosity), helps to predict the tendency of Li metal to form dendrites. ζ < 0: once the perturbation is formed, it will decay and the electrodeposition will be stable; ζ > 0: the perturbation won't be controlled and the dendrites will propagate.