3D printing of solid polymer electrolytes by fused filament fabrication: challenges towards in-space manufacturing

A new chapter of space exploration is opening with future long-duration space missions toward the Moon and Mars. In this context, the European Space Agency is developing out-of-the-earth manufacturing abilities, to overcome the absence of regular supplies for astronauts’ vital needs (food, health, housing, energy). Additive manufacturing is at the heart of this evolution because it allows the fabrication of tailorable and complex shapes, with a considerable ease of process. Fused filament fabrication (FFF), the most generalized 3D printing technique, has been integrated into the International Space Station to produce polymer parts in microgravity. Filament deposition printing has also a key role to play in Li-ion battery (LIB) manufacturing. Indeed, it could reduce manufacturing cost & time, through one-shot printing of LIB, and improve battery performances with suitable 3D architectures. Thus, additive manufacturing via FFF of LIB in microgravity would open the way to in-space manufacturing of energy storage devices. However, as liquid and volatile species are not compatible with a space station-confined environment, solvent-free 3D printing of polymer electrolytes (PEs) is a necessary step to make battery printing in microgravity feasible. This is a challenging stage because of a strong opposition between the mechanical requirements of the feeding filament and electrochemical properties. Nowadays, PE manufacturing remains a hot topic and lots of strategies are currently being studied to overcome their poor ionic conductivity at room temperature. This work firstly gives a state of the art on the 3D printing of LIBs by FFF. Then, a summary of ionic conduction mechanisms in PEs permits to understand the several strategies studied to enhance PEs performances. Thanks to the confrontation with the specifications of FFF printing and the microgravity environment, polymer blends and composite electrolytes turn out to be the most suitable strategies to 3D print a lithium-ion polymer battery in microgravity.


Introduction
The launch of the Artemis program, in 2019, symbolizes the beginning of a new area in space exploration with targeted humans coming back to the Moon for a durable occupation [1].For the first time, astronauts will spend long-duration missions far from Earth without any regular resupply.Thus, it makes sense to manufacture tailorable devices in space, instead of shipping them through costly and time-consuming spaceship transports [2].That is why researchers, in collaboration with space agencies like the National Aeronautics and Space Administration (NASA) and the European Space Agency (ESA), are working on in-space manufacturing (ISM) methods and in-situ resource utilization [3].The aim is to overcome this dependency, especially for vital needs such as food, energy, and protection against severe constraints of the space environment.Additive manufacturing methods are at the heart of this program as they enable fast fabrication of tailorable architectures with minimal material waste and low cost.Compared to traditional manufacturing methods, this consecutive layer's addition of materials on top of each other allows the precise building of objects with complex shapes.3D printing has been intensively investigated for space applications [3] through the construction of space structures based on lunar or Martian regolith [4], spare parts fabrication, and food preparation [5] onboard space stations or on the Moon/Mars surfaces [2].'In-space' additive manufacturing breakthrough took place in 2014 with the first 3D printed polymer part by fused filament fabrication (FFF), (also called fused deposition modeling) by NASA with the '3D printing in Zero-G' project [6].The first European 3D printer onboard the International Space Station (ISS) has been created, in 2016 by an Italian group, to 3D print polylactic acid (PLA) by FFF [7].The same year, ESA started to develop an FFF printer through the MELT project, to enhance the portfolio of polymers that could be used in such a peculiar environment focusing on polyether ether ketone (PEEK).This demonstration of the feasibility of 3D printing in microgravity has led to permanent installation of the Additive manufacturing facilities (AMFs) onboard the ISS.Production of parts using other high-added value materials, like ceramics and metals, along with new printing techniques have been tested in simulated microgravity environment during parabolic flights demonstrating their compliance toward microgravity: metal 3D printing solutions by Force Metal Deposition [8], Direct Energy Deposition or Electron Beam Fusion and ceramic printing via VAT photopolymerization [8,9].The first automated on-orbit 3D printing of continuous carbon fiber-reinforced thermoplastics have been recently demonstrated by the Chinese Academy of Space Technology [10].Looking at additive manufacturing on the Moon or Mars, processing of regolith as a feedstock material using the power bed fusion technology was investigated [11].
Energy supply is a critical issue in space applications as it is one of the main causes of the space systems lifespan limitation.Energy mainly comes from solar power so rechargeable Li-ion batteries (LIBs) are used for all devices from the ISS primary power system to portable communications devices, going through life support systems [12].That is why batteries are key components that need a fast replacement in case of failure.Thus, the ability to manufacture tailorable energy storage devices is necessary for space exploration.Such capability fits the general strategy to decrease mission dependence from earth supply and will enable new maintenance strategy based on on-demand manufacturing instead of spare parts storage.Indeed, storage of spare parts is volume-consuming and shall be minimized specially in close environment like orbital station, lunar settlement of spacecraft for long travel toward Mars.It has been demonstrated that changing from spare part-based maintenance strategy toward on-demand ISM will decrease by 78% the mass required for manufacturable set [13].
With this in mind, Maurel and coworkers recently introduced the possibility of 3D printing batteries with Moon or Mars resources [12].They underlined the necessity to develop and optimize highly resolution multi-material printers to 3D manufacture batteries out of the earth.The suitability of Na-ion devices instead of Li-ion has been highlighted due to the greater abundance of Na on the Moon and Mars [12].The crossbreed of the desired ability to manufacture tailorable batteries in space, and the knowledge acquired on FFF 3D printing in microgravity opens the way to lithium battery printing by FFF in a space station.LIBs are known as the most efficient type of batteries [14].They are made of five parts: electrodes (cathode and anode), current collectors, electrolyte, separator, and casing.A gradient of potential induces a reversible movement of Li + through the electrolyte between hosting materials in electrodes: the rocking chair movement.Cathodes and anodes are composed of conductive materials to convey electrons, and active materials to allow movement and storage of Li + in their structure.Today's electrolytes are liquids containing organic solvents that enable fast motions of Li + between electrodes, while the separator forms a barrier for electrons to avoid short circuits.However, the microgravity environment impeaches the use of liquid electrolytes and the presence of volatile and flammable organic solvents is hazardous in the confined environment of a space station.Thus, the 3D printing of a polymer electrolyte (PE) is a necessary step to overcome these issues and to achieve one-shot printing of a full polymer battery without any post-treatment (Li-ion polymer battery).This paper will be focused on the 3D printing by FFF of a PE with main challenges and strategies.The first part will give a state of the art on what has been done so far towards the 3D printing of batteries by FFF.Then, ionic conduction mechanisms in polymers will be detailed to understand the landscape of strategies to achieve PEs.The final part will highlight suitable strategies reported to meet the main challenge which is the 3D printing of an efficient PE compatible with a microgravity environment.

3D printing of LIB
Since the 1980s [15], additive manufacturing has been involved in a lot of domains [16] such as medicine [17], aerospace [3], transport, building construction, electronics, food, and energy [18].It consists in successive addition of materials layer to build a 3D dimensional object.This technology could produce complex geometries at a lower manufacturing cost and shorter time than any commercially available process Additive manufacturing processes according to international ASTM standard.Reprinted from [22], Copyright (2020), with permission from Elsevier.[19].Advantages, issues, and feature resolution of each additive manufacturing technology have been more widely detailed in dedicated papers [16,20].Today, they are gathering seven categories, according to the ASTM standard [21] (figure 1).Among them, material extrusion (ME) printing is the most accessible technique to produce polymer parts.

Interests for LIB manufacturing
Since the 2010s, interest in the 3D printing of LIBs is growing because it can help to take up some manufacturing challenges [22]: develop micro-energy storage devices and make energy storage manufacturing processes shorter, easier, and cheaper.The first pioneer industrials in this field, such as SAKUU, started to produce 3D-printed batteries.The breakthrough brought by additive manufacturing of LIB is the freedom of design which happens at three levels in constructing strategies (direct printing or methods with post-treatment), electrode architecture (surface pattern, thin film, 3D network or fiber), and battery configurations (sandwich, in-plane, concentric tubes or fiber) [23] (figure 2(a)).Thus, it enables to: • Manufacture all-solid-state batteries with structural stability.Indeed, each part of the battery is 3D printable in the solid phase to manufacture in one shot an all-solid-state battery.The intrinsic interlayer adhesion, which guarantees the mechanical integrity of printed parts, can provide a high interfacial adhesion between electrolytes and electrodes.That is why FFF has been studied, as a suitable method to create all-solid-state Li-ion polymer batteries in one-shot [24].In the case of lithium-metal battery (LMB), Li metal anode printed by Direct Ink Writing (DIW) can limit dendrite growth thanks to a high specific area, periodic large porous structure, and high Li + /electrons transports [25].• Improve electrochemical performances.3D printing enables to optimize electrodes and electrolyte structures at macroscopic and microscopic scales.Electrodes can be manufactured with specific 3D pattern structures like zigzag lines, periodic micro lattice spirals, or circle grids [18,26].Simulations have shown these structures can give optimized 3D ionic diffusion pathways with shortened ion transport distances for a Li-ion polymer battery [27] (figure 2(b)).It is also possible to play with electrodes and electrolyte thicknesses.Indeed, at same volume, interdigitated or in-plane battery configurations, mean larger surface areas with shorter ion transport distances.In these configurations thicker electrodes can be manufactured while preserving short ion transport distances (figure 2(c)).These tailorable designs contribute to increasing the power density.High macro and micro porosity rates, induced by 3D printing processes and a specific design, facilitate ionic transport in liquid electrolytes through the free space of electrodes.It permits to create thicker electrodes without any limitation on power density [25].Thus, it allows the manufacturing of thicker electrodes with higher mass loadings to obtain higher areal and volumetric energy densities.Optimized structures permit to increase the areal active surface by maximizing the mass loading of active materials and controlling electrode thickness [23,28].Xueliang Sun's group has optimized the 3D electrode pattern to print a thick cathode of 1500 µm in order to achieve energy and power densities comparable to the reported performances of the LiFePO 4 (LFP) cathode [26].In interdigitated battery configuration, higher electrodes (higher number of layers) contribute to the rise in the areal capacity [29].• Manufacture micron-sized batteries with customized shapes.Micro batteries are more and more essential due to technological advances.Resolution of 3D printing in the range of micrometers [16] and tailor-made geometries open new horizons for integrated and flexible energy storage devices in portable technologies [30].For instance, the 3D micro battery of Sun et al [31] has been printed by DIW, with a nozzle of 30 µm.
Chen et al [32] have printed their Li-ion micro battery (7.6 × 3.8 × 3 mm) based on gel PE (GPE) via stereolithography (SLA).The increase in volumetric energy density, brought by 3D printing, helps to create smaller batteries with comparable performances.• Lower manufacturing costs.Thanks to a reduction of material wastage and production time savings compare to a conventional process.Indeed, the conventional battery manufacturing process is made of seven steps whereas battery 3D printing by FFF is composed of two steps: filament preparation and one-shot printing [23] (figure 3).The use of a multi-materials printer can enhance the productivity by avoiding nozzle cleaning steps between each filament.
ME and vat photopolymerization are the two most studied 3D printing categories for energy storage devices manufacturing [25] due to their high versatility, low cost, and ease of process.ME technology consists in dropping off a melted thermoplastic polymer through a nozzle.The feedstock could be filament (Fused Filament Fabrication, FFF), ink (Direct Ink Writing, DIW), or pellets (Fused Granulate Fabrication, FGF).First reported papers on 3D printing for energy storage applications involved DIW because of its great adaptability to multi-material manufacturing and its high resolution (1-250 µm).However, issues with ink rheological behavior have restricted its use [18].Vat photopolymerization also known as SLA provides sub micrometers printing resolution.It consists in curing a photo resin thanks to the energy brought by a laser or UV light.However, it limits the choice of polymers to suitable photopolymers, and operations cost tends to be higher [29].Maurel et al have reviewed main studies about energy devices 3D printing by ME [33].FFF turns out to be interesting as it provides good processability and the possibility of printing an entire battery in one-shot.Moreover, it requires a low-cost set-up, as well as it enables fast fabrication of a wide variety of polymers for relatively small parts.The main drawback is that this procedure suffers from a lower printing resolution (50-200 µm) than other techniques.

State of the art on FFF printing of LIB
In the FFF printing of LIBs, the feeding material is a filament composed of a thermoplastic polymer matrix, loaded with active and conductive materials necessary for the proper working of batteries.Thus, the 3D printing of an entire LIB requires 6 filaments corresponding to each part (figure 4).
Filaments are composed of a thermoplastic polymer matrix blended with conductive materials (carbon black, carbon nanotubes (CNTs) or carbon nanofibers) for current collectors, with conductive and active materials (such as LFP, LiNi x Mn y Co z O 2 (x +y +z = 1) (NMC) for the cathode and Li 4 Ti 5 O 12 (LTO), graphite for the anode) for electrodes.As solvent free method, all of these materials are introduced under the shape of powder or pellets in an extruder.Inside, raw materials are melted and blended, thanks to screw rotation and heating, to obtain a viscous blend.The latter steps out through an extrusion die to create a 1.75 mm diameter homogeneous filament.Conductive materials need to percolate to offer pathway for electrons, meanwhile active materials need to be accessible for Li cations.Thus, the viscosity, wetting coefficient and materials feeding modes (sequencing) are key parameters to reach a convenient morphology.Since 2017, several research groups have tried to achieve this concept [33].First papers were focused on introducing charges in polymer with low loadings to create composite thermoplastic filament for FFF.Commonly used polymers in 3D printing such as PLA or ABS were blended with graphene.Zhang et al [34] as well as Foster et al [35] studied 3D-printed graphene-based PLA negative electrodes [34,35].Their strategy was to limit the content of graphene to obtain a 3D printable filament (<30 wt% of active material) but obviously, filaments suffered from poor electrochemical performances.Between 2018 and 2022, three research groups tried to 3D print each part of the LIB: Reyes et al [30] in the US, Ragones, Vinegrad et al [36] in Israel, and Maurel et al [37] in France (figure 5).All of them have worked with PLA as the host polymer for their composite filaments, and they have tried to progressively increase the content of charges.
For the anode, Reyes, Ragones and coworkers have worked with LTO as active materials.They have succeeded in increasing the percentage of active materials in their filaments: PLA represents respectively 54 wt% and 50 wt% of the filament against 80 wt% for the graphene-based PLA from previous studies.Ragones group has played with carbons additives to enhance electrochemical performances by adding graphite, carbon black, and CNT.In the meantime, Maurel et al succeeded in increasing the loading of active materials up to 49 wt% of graphite in their anode filament with only 33 wt% of PLA.They employed the plasticizer poly(ethylene glycol) (PEG) dimethyl ether (PEGDME)500 to give more flexibility so as to be 3D printable.This strategy permits to obtain a reversible capacity of 200 mAh•g −1 for a current density of 18.6 mA•g −1 .The same strategy was applied to the cathode filament.LFP was chosen as active material by Maurel and Ragones groups and LiMO 2 by Reyes and coworkers.
For electrolytes, two strategies have been considered: 3D printing a separator which must be impregnated with a liquid electrolyte, or 3D printing a PE.Reyes et al have printed and soaked a commercial filament of PLA with a liquid electrolyte whereas Ragones et al tried to soak a polymer blend of PLA and poly(ethylene oxide) (PEO) with ionic liquid (IL) (figure 6).Maurel et al have 3D printed a composite filament of PLA with SiO 2 to improve the separator soaking ability.As for electrodes, they circumvented stiffness issues of the filament thanks to a plasticizer.These two latter groups have also worked on PE printing.Filaments were based on PEO-lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) which is up to now the most studied and efficient single polymer-based electrolyte at temperatures above 60 • C. At a lower range of temperature, it suffers from poor ionic conductivity (10 −5 -10 −6 S•cm −1 ).Moreover, LiTFSI has a plasticizer effect on PEO which turns out to be very difficult to print.The first possibility is to put less LiTFSI and to modify the 3D printer to be able to print a PE at the cost of electrochemical performances [24].Incorporating other   [30].Copyright (2018) American Chemical Society.(c) Reproduced from [38].The Author(s).CC BY 4.0.materials to stiffen the filament is a second strategy which has been tested by Ragones et al.They have blended PEO/LiTFSI with PLA and SiO 2 to obtain a 3D printable filament.In both cases, printed electrolytes suffered from poor performances at lower temperatures and were still very difficult to 3D print [36].
Up to now, studies about 3D printing of liquid electrolyte batteries using FFF have been dedicated to proofs of concept.Replacing the liquid electrolyte with a printed PE is a critical step due to its poor mechanical behavior and its poor ionic conductivity at low temperatures (25 • C).Thus, printing an electrolyte by FFF is the main challenge to successfully manufacture Li-ion polymer batteries.The following section will summarize ionic motion mechanism in PE and its key parameters.

PEs
Organic (polymers) and inorganic (ceramics) materials have been widely investigated to achieve solid electrolytes toward all-solid-state batteries [39].Polymers are easily processable and offer good interfacial properties for a low cost of material.Ceramics are more ionic conductive, confer more rigidity, and provide thermal and electrochemical stability [39].The first use of polymer materials as electrolytes in LMB was reported by Wright and Armand in the 1970s for safety concerns [40,41].More recently, the first commercialization of a PE happened with a system based on PEO and polyvinylidene fluoride (PVDF) for LMB made by the French industrial Bollore [42].Basic PEs are made up of a single polymer matrix and a lithium salt.Polar groups on backbones dissolve lithium salt to allow the movement of Li + .The following electrolyte properties are targeted [43]: • High ionic conductivity: to convey efficiently Li + between electrodes (10 −3 S•cm −1 ).It will impact high C-rate performances, specific capacity, and power density [44].• Electrochemical stability: to remain stable and not be oxidized or reduced upon cycling.
• High transference number (t + ): to avoid polarization during cycling.
• Sufficient mechanical behavior: to be processable, not to crack during cycling, and to maintain good contact with electrodes [45].• Electronic insulator: to avoid short circuits.
• Ease of processing: to ensure fast manufacturing and try to avoid hazardous solvent methods.
Even if polymers have been identified as good candidates to achieve solid flexible electrolytes, their poor ionic conductivity at lower temperatures than their melting point restricts their use.There is a strong contradiction between having solid-like mechanical behavior and reaching a high ionic conductivity.Thus, this section describes ionic motion mechanisms in polymers to understand strategies to improve Li + conductivity in polymers.

Ionic conductivity mechanisms
In organic liquid electrolytes, ionic motion happens via diffusion whereas in ceramics, ions move via crystal defects [46].Unlike these two cases, ionic transport mechanisms in polymers are not commonplace because they are influenced by many factors.Thus, a lot of researchers have dedicated their work to a better understanding of Li + movement through polymers [43,47].In basic systems, polymers can dissolve lithium salt through a polar functional group [48,49] (figure 7).A Lewis complexation of ions (acid-base interactions) is possible thanks to heteroatoms (O, N, S) to form polymer salt complexes.This complexation ability is linked to the donor number and dielectric constant of the polymer [49].The bond's strength is determined for a particular cation by the electron pair donicity of the coordinating groups on the polymer chain [50].General electrostatic interactions (ion-dipole) and non-electrostatic interactions also contribute to ionic solvation [50].Polymer salt complexes are composed of free anions, solvated cations, and aggregated ions.Most polymers have a low dielectric constant which leads to increased ions aggregation, through long-range Coulomb forces, reducing the overall ionic conductivity [47,51].At a defined temperature, we can describe the passage of a current through a PE as a combined contribution of ionic migration over a potential gradient, ionic diffusion due to concentration gradient, and ionic convection due to electrolyte velocity (equation ( 1)).Considering a dilute system, without any concentration gradient and negligible electrolyte velocity, we obtain a form of Ohm's law which corresponds to the following Kohlrausch law with (equation ( 2)) Polymers' semi-crystallinity makes ionic mobility difficult to understand.It could be split in two categories: movements coupled with polymer chain mobility (in amorphous regions, also called free volume theory [52]) and movements decoupled from polymer chain mobility (in crystalline parts also called ion-conduction) [14,47,52].McLin and Angel [53] introduced a decoupling ratio between the structural and the conductivity relaxation time (equation (3)).Below the glass transition temperature, structural relaxation time is long, the decoupling ratio is high: there is no link between ionic conductivity and chain segmental mobility.On the opposite side, above the glass transition, the decoupling ratio is becoming smaller than one: the ionic conductivity occurs thanks to polymer segmental mobility.
Thus, there are two main models for ionic transport in polymers in function of temperature [49]: (1) Arrhenius model (ion conduction): pathways are static so the ionic transport is decoupled from polymers chain mobility.It gives a good description of conduction in polymers below their glass transition temperature as well as in crystalline phases where the polymer chain's mobility is absent [54].
Li + can move through vacancies offered by the crystalline structure.Folded chains form a cylindrical tunnel for Li + to coordinate and be located inside this channel.Li-ion transport mainly depends on the tunnel structure formed by adjacent coordination sites [52].The conductivity increases with temperature by the activation process which enables Li + to move from one site to another.Geji et al [55] have shown that Li + complexation by ether oxygen in crystalline region results in a coordination environment which is quite full.In these regions, there are no obvious static pathways for Li + movement.Thus, ionic movements occur preferentially in amorphous regions [56].However, Gadjourova et al [57] have demonstrated that polymer crystalline structure can have better ionic conductivity than equivalent amorphous phase above the glass transition temperature, with a transference number close to unity.(2) Vogel-Tammann-Fulcher (VTF) model (free volume): ion pathways are dynamic so the ionic transport is coupled with polymer chains mobility.It gives a good description of conduction in polymers above the glass transition temperature and fully amorphous polymers where the segmental mobility is high.It corresponds to the free volume theory in which diffusion can occur when diffusing particles move from one free volume space to another [39].Conductive materials relax more rapidly than non-conductive ones, which means that ionic motion and segmental relaxation of polymers are linked [50].Hallinan and Balsara [43] considered two types of mechanisms: • Segmental motion of the chains surrounding the lithium salt.In the case of a PEO-LiTFSI electrolyte, Li + surrounded by six ether oxygens is the minimum free energy configuration [43].
Segmental motion leads to the modification of configurations and enables Li + to move toward sites of lower free energy [43].It provides more chances for ions to perform faster migration.Ionic jumping can take place on the same polymer's chains (intrachain ionic jumping) or can occur between two different chains (interchain ionic jumping) [58].• Diffusion of the entire polymer chain with coordinated ions, also known as vehicular transfer [43].
It can only occur for low molecular weight polymers (less than 2000 g•mol −1 ).It can explain the dependence of the ionic conductivity of polymers on their low molecular weight seen in the literature [43].
In both cases, Li + motion is due to stronger interactions of cations than anions with ether oxygens [51].However, free volume theory is limited in the case of semi-crystalline systems, incomplete salt dissociation, and glass transition temperature variation due to charge species [51].An enhancement of ionic conductivity at elevated temperatures is visible due to an increase in segmental motion, lower energy barriers for ion transport, and an increase in ion mobility and concentration [52].Maurel et al [24] illustrated such a behavior experimentally for a 3D printed electrolyte PEO:LiTFSI (O:Li = 20:1) system by coupling differential scanning calorimetry and ionic conductivity data.To summarize, 3 main behaviors of ionic conductivity vs. temperature can be identified: 1 • /Arrhenius for crystalline systems, 2 • /VTF for amorphous systems, 3 • /Arrhenius and VTF mix for semi-crystalline systems [39].
Other groups tried to model Li + movement in different ways and at different scales.At a microscopic scale, the Dynamic Bond Percolation (DBP) [59] permits to describe ionic conductivity over a wide range of temperatures [47,60].Ion motion is described as a first-order chemical kinetics.The probability to find a Li + at a site i at the time t is P i (t) and W is the hopping rate from site i to site j.In the Arrhenius case, W = 0 if the jump is forbidden, and W = 1 if the jump is allowed (equation ( 6)).In a VTF case, W depends on the time because segmental motion can make jump forbidden or allowed during a certain amount of time called the renewal time.In this model, the diffusion coefficient is proportional to the average renewal rate which is the rate at which a motion pathway between two sites becomes feasible [60].Even if it is one of the best models for understanding ionic conduction in polymers, it does not consider inertial dynamics and interionic interactions [60].Reducing the renewal time (by increasing segmental mobility) and reducing ion pairing (by increasing the dielectric constant) will increase overall ionic conductivity [47].Quantum mechanics can also explain Li + motions in models in which Li + motion is described as a wavelength λ Li+ (figure 8) [54].According to Zhou et al [54], core factors for improving ionic conductivity are freer Li + and longer hopping distances with lower barriers between 2 coordinating sites.Adam and Gibbs [51] attempted to extend the theory of free volume.In their approach, transport occurs by group cooperative rearrangements rather than by hole-hole jumping motion.Ionic conductivity models for heterogenous systems are more complicated.
According to [47], there are three ways to increase ionic conductivity at room temperature in polymers: • Increase the number of mobile ions (adding Li salt) to increase the prefactor σ 0 in Arrhenius and VTF model (equations ( 4) and ( 5)).• Increase segmental mobility (decrease T g and crystallinity rate) = make VTF mechanism effective at room temperature • Decoupling from the segmental structure motion: create a structure in which Li + can move through static pathways = favor the Arrhenius mechanism * Polymer as an inactive support binder for percolating conductive materials * Polymer-in-salt strategies [61]   * Composite with low crystallinity at the interface which creates pathways * Hard polymers: polymers which are less dense and thus, have a higher free volume [47] The following figure summarizes main concepts of ionic transport in a homogeneous system of polymer and lithium salt (figure 9).

Key parameters to improve the ionic conductivity
Main parameters influencing the above-mentioned ionic mechanisms are: dielectric constant and donor number, polymer structure and molecular weight, temperature, electrolyte architecture, or concentration of Li salt [52].
Dielectric constant and donor number influence the polymer solvation ability.Indeed, isolated electrons of polar groups must coordinate with Li + through Coulomb interactions to dissolve ionic salt [51].Caradant et al [49] provide the donor number and the dielectric constant for seven polymer matrixes depending on their polar functional group.They experimentally showcased that polymer affinity is dominated by the donor number.Ether, ester, and carbonate polymers seem to have good properties at 25 • C. A high dielectric constant permits to reduce ion pairing and increase overall conductivity [50].According to Meng et al [48], four main polar groups display a good ability to dissolve ionic salts: ether group (PEO, PEG, poly(vinyl acetal) (PVA), carbonyl group (poly(vinyl carbonate)) (PVC), poly(propylene carbonate) (PPC), poly(methyl methacrylate) (PMMA)), nitrile group (polyacrylonitrile (PAN)), and fluorine group (PVDF).
The polymer solvation ability is a question of balance: with high dielectric constant and donor number, the ionic conductivity is limited by too strong chemical interactions between Li + and polar groups which reduce the segmental chains motion.In the opposite case, the ionic conductivity is decreased because dissociation of Li salt does not happen.
Polymer structure influences the ionic conductivity through the segmental relaxation of polymers chains.In high molecular weight PEs, ion mobility is inversely proportional to the relaxation time [43].According to Meng et al [48], this mainly relies on polymers' flexibility, inter and intramolecular forces, and crystallinity rate.Polymers' flexibility is linked to backbones and side chains structures.PEO backbones (C-O-C) have a high chain flexibility [58] whereas heterocycles such as poly(vinyl formal) (PVFM) or poly(vinyl butyral) (PVB) provide higher rigidity [48].In amorphous regions, segmental motion is highly increased.Thus, in a semi-crystalline polymer, ionic transport happens in amorphous regions [51,58,62].The low crystallinity rate in polymers is due to the macromolecular structures, simple, ordered, or symmetrical [48].This crystallinity could be decreased by the introduction of fillers, other polymers, or plasticizers.Polymer structures also influence the polymer solvation ability which depends on the steric position of these polymers' groups [48].Devaux et al [63] have shown that end groups own higher mobility than middle segments.It can play a key role in ionic interactions, depending on the end groups chemical composition.
In the case of low molecular weight PEO, the hydroxyl group can solvate Li + which reduces polymer chain mobility.Finally, as described in the previous section, low molecular weight polymer can increase ionic conductivity thanks to the vehicular transfer mechanism.For PEO-LiTFSI at 76 • C, this increase in conductivity is experimentally visible for a molecular weight lower than 2000 g mol −1 [43].
Li salt concentration effect depends on the kind of polymer.In the case of poly(ethylene carbonate) (PEC), the addition of lithium salt improves the ionic conductivity.In the case of PEO, an increase in the lithium salt molar ratio improves the ionic conductivity until a threshold [39].When it is jumped over, undissolved lithium salt acts as a barrier for the motion of Li + which leads to a decrease in ionic conductivity.Increasing ionic salt concentration also reduces the polymer chain mobility because of chemical interactions between the heteroatom of the polymer and lithium salt.These two competing phenomena have been illustrated by Hallinan and co-workers in their review [43].The amount of lithium salt has also an impact on the micromechanical behavior by acting as a plasticizer.It reduces the crystallinity rate of the polymer while degrading the mechanical stability.Thus, there is an optimized lithium salt content, to reach an optimal conductivity without degrading the mechanical stability.For a homogeneous PE (PEO:LiTFSI), the best conductivity is reached for around O:Li = 10:1.However, in this basic homogenous system, a molar ratio of 20:1 is usually used to maintain a sufficient mechanical behavior.In polymers with low dielectric constant, ion-pairing occurs at low salt concentrations, while larger ionic aggregates (most of which are charged) form at higher concentrations [50].Thus, conduction mechanisms should consider the possible exchange of cations and anions between pairs and clusters [50].Ionic motion can happen by jumping from one ionic cluster to another one but also reduce chain segmental mobility.However, the t + number strongly goes down with salt concentration increase for the polyether systems such as PEO [64].
Electrolyte architecture has a fundamental role to play in ionic conductivity.In heterogeneous media, Bouchet et al [65] have shown that the tortuosity of Li + pathways reduces the ionic conductivity.In their polystyrene (PS)-based block copolymer PS-PEO-PS, low molecular weight PEO increases excluded zone proportion (nonconductive zones in PEO near PS sides).Thus, a combination of higher molecular weight PEO and higher volume fraction of the conducting phase (PEO-LiTFSI) contributes to decrease the tortuosity in a block copolymer.In the model proposed by Carman (equation ( 7)), the conductivity of the composite is the bulk conductivity of the conductive phase (σ 0 ) multiplied by its volume fraction (ε) (which depends on excluded zones length λ).The tortuosity is represented by the factor τ (depending on the volume fraction of conducting phase ε and larger than 1).1/ τ is the fraction of conducting volume that has the same transport efficiency as the bulk electrolyte.A percolation threshold of PEO cylinders may exist around 60 wt% of PEO that could explain the conductivity gap at room temperature from 6 × 10 −7 S•cm −1 to 2 × 10 −5 S•cm −1 for an increase from 36 to 74 wt% of PEO [65].
Equation 7. Carman-based model of ionic conductivity in a composite electrolyte [65] In the case of composite electrolytes (polymer matrix + inorganic fillers), high fillers loadings can increase the tortuous lithium-ion path which can decrease the overall conductivity [52].The specific structure of 'hard polymers' , which can contain a greater internal free volume than soft polymers, can offer new static pathways via Arrhenius conducting mechanism [57].This is the case of polydimethylsiloxane which has been used to improve ionic conductivity thanks to its high internal free volume [66].Thus, the aims for future PEs are to decrease the crystallinity rate of the polymer matrix, maximize the Li salt molar ratio, and construct continuous fast ion paths [52].

PEO-LiTFSI PE
Since Armand and Wright's works on PEs, the mix of PEO and LiTFSI has been the most studied single polymer-based electrolyte.A promising polymer for electrolytes should have the following characteristics: (1) functional groups with sufficient electron-donating power to form coordinated bonds with cations (2) suitable distances between such coordinating centers to permit the formation of multiple interpolymer-ion bonds (3) low barriers to the rotation for atoms in the main chain to ensure high flexibility so good segmental relaxation [50].PEO has a strong ability to dissociate and coordinate LiTFSI [48].It has the highest reported lithium salt solvating ability due to its high donor number and dielectric constant [49].Thus, strong chemical bondings are created between ether oxides and Li + [58].PEO also offers a high chain flexibility due to C-O-C backbones which increases the segmental mobility and so the ionic transport according to the VTF model.Finally, as mentioned by Golodnitsky et al [67], PEO has a helical structure with six O-CH 2 -CH 2 groups in two turns of the helix.PEO chains are wrapped around Li + , that is why there could be an increase in ionic conductivity along the helix alignment as suggested by Golodnitsky et al [67].At 25 • C, the reported best ionic conductivities for PEO:LiTFSI is around molar ratio O:Li = 10:1 and O:Li = 8:1 [68].This is mainly due to the reduction of crystallinity rate and the increase in the number of charges carried.Chen et al [69], showcased that from 60 wt% of salt, undissolved residual Li salt acts as a barrier for the Li + transport.
However, polymer-based PEO-LiTFSI electrolytes have residual issues that are limiting their use.Neat PEO exhibits a high crystallinity rate at room temperature (between 75% and 80% [70]) which limits its ionic conductivity to 10 −6 S•cm −1 .Temperatures above the melting point of PEO (T m = 65 • C), which means heating the battery, are mandatory to obtain a working system.Crystallinity rates change by a few percent with the molecular weight [36] and with the addition of Li salt.Thus, reported crystallinity values of PEO-LiTFSI are lower as shown by Zhang et al [68]: 46% for PEO (M n = 600 K):LiTFSI, O:Li = 12:1).Moreover, ether oxide groups create strong chemical bonding with Li + .Trapped EO chains and coordinated cations with dipoles form stable complexes which increase the T g and avoid fast cations migration [64].This phenomenon tends to limit the Li + conductivity at 25 • C [49].In terms of PEO electrochemical stability, the reported voltage stability window is from 0.5 V to 3.8 V vs. Li/Li + [43].Such oxidation potential value makes impossible the use of high potential cathode active materials such as NMC with a working potential higher than 4 V. Zhang et al [71] also warned about the low quality of the solid electrolyte interphase, formed at very low potential with alkali metals.Like other dual-ion conducting polymers, the transference number of PEO/LiTFSI is especially low (around 0.1) [64] which can induce cell polarization.Finally, the plasticizer effect of LiTFSI on PEO can make harder the processability, especially towards FFF 3D printing.In this case, a feeding too sticky and ductile due to the LiTFSI effect can lead to printing failure [24].
Even if PEO/LiTFSI electrolyte remains attractive, its limits lead to a maximal Li + conductivity of 10 −6 S•cm −1 at 25 • C which is far below commercial electrolyte ionic conductivities (10 −3 S•cm −1 ).There is also a strong contradiction between electrochemical and mechanical performances toward 3D printing.

New strategies of PE
Many strategies have been imagined to overcome the low ionic conductivity at room temperature as well as limited electrochemical, thermal, and mechanical stability [72].That is why, the term PE hides several concepts of polymers-based systems from solid PE (SPE) to Quasi SPE (QSPE) such as GPE.Different classifications have been early reported by Fiona Gray [73], Wright [74], Ratner et al [47], and coworkers until the 2000s.They have been mainly created based on different ways to improve ionic conduction through Li + motions mechanisms.More recently, other groups gave actualized classifications based on composition and morphologies [75] (table 1).
The classification of this work gives an overview of all strategies with main groups and subsets (figure 10).Based on previous classifications, three classes can be identified: Solid Polymer Electrolyte (SPE) (line 1, table 1), Composite Polymer Electrolyte (CPE) (line 2, table 1), and Quasi Solid Polymer Electrolyte (QSPE) which gathers additional categories based on a poor mechanical behavior due to high loadings of plasticizer, ionic salt or IL.

QSPE
GPEs are nonaqueous electrolyte solutions contained in a structural polymer matrix such as PVDF, PAN, or PMMA [51].These electrolytes could have a solid-like behavior if the liquid phase is entrapped in the polymer matrix.Li + motions mainly happen by diffusion in these entrapped regions [72].Thus, they reach liquid-like ionic conductivities (10 −3 S•cm −1 ) at room temperature with low volatility and low reactivity.However, GPEs suffer from poor mechanical behavior (poor rigidity and mechanical strength).It can cause damage by leakage of the liquid phase [72].Wookil et al [72] have compared ex-situ and in-situ methods of manufacturing for GPEs (figure 11).According to them, in-situ polymerization by thermal initiator is the most efficient technique.They have reported high ionic conductivities up to 8.82 × 10 −3 S•cm −1 at room temperature for PVFM-based GPE [76].In terms of electrochemical properties, a GPE based on trihydroxymethylpropyl trimethylacrylate reaches an ionic conductivity of 6.15 × 10 −3 S•cm −1 , a 0.1 C discharge capacity of 183.1 mAh•g −1 and 149 mAh•g −1 after 1 and 100 cycles respectively for an NMC811 vs. Li metal cell at 25 • C [77].
Polymeric IL electrolytes (PILEs) are made of polymer, IL, and lithium salt.ILs are organic salts in which ions are poorly coordinated and melt below 100 • C [78].They own high thermal stability, wide electrochemical windows, non-flammability properties, and low vapor pressure.They exhibit high ionic conductivity however, they do not have a solid-like behavior, so they suffer from poor mechanical stability [72,79].Thus, they have been polymerized to form polymers in which either the cation or the anion are incorporated into the polymer backbone (polymerized ILs (PILs)) [78].They have a solid-like behavior but the lower mobility of ions fixed to the polymer backbone results in lower ionic conductivity than ILs [80,81].It could explain why they are often used in electrolytes with IL and Li salt [81].Passerini and his group [82] have studied interactions between PILs, ILs, and Li salt.They have showcased that the addition of IL in PEO or PIL-based electrolytes can improve ionic conductivity up to 5 × 10 −4 S•cm −1 at 20 • C. Higher ratios of IL/polymer are possible with PILs but it does not lead to higher conductivity and it makes electrolytes films sticker.Guo et al [83] have reached an ionic conductivity of 10 −3 S•cm −1 thanks to their flexible ionogel electrolyte made of an immobilized ILs (1,2-dimethyl-3ethoxyethyl imidazolium bis(trifluoromethanesulfonyl)imide (DE-IM/TFSI)) into a hydrogen-bonded network of PIL copolymers bearing ureido-pyrimidinone (UPy) pendant groups (PIL-UPy).The quadruple hydrogen bonds and the electrostatic interactions between DE-IM/TFSI and PIL-UPy permit to obtain satisfactory mechanical strength with a high loading of DE-IM/TFSI (67.3 wt%).
Highly plasticized electrolytes are composed of polymer matrix, lithium salt, and a high content of plasticizers.The latter small molecules can enhance the polymer segmental motion and macro flexibility.Commonly, polar liquids, are added in polymer lithium salt systems.The result is a 100-fold increase in ionic conductivity.Lithium salt such as LiTFSI has also a plasticizer effect on the polymer matrix.PEO-based 2023, [72] 2022, [51] 2021, [75] Solid polymer electrolyte (SPE) Amorphous macromolecular salt complexes  plasticized electrolytes reach ionic conductivity up to 10 −4 S•cm −1 at 25 • C [58].However, most plasticizers are very volatile that is why this solution is temporary, and high plasticizer loadings lead to the degradation of the electrolytes' mechanical behavior.
Polymer-in-salt electrolytes are composed of a small amount of high molecular mass polymers dissolved in low-temperature molten salt in large quantities (>50 wt%) [51].According to Hae-Kwon Yoon and coworkers [84], when salt is added above the critical concentration in these polymers, percolation of ions clusters provides fast cationic transport pathways (10 −3 S•cm −1 at room temperature with 70 wt% of LiCF 3 SO 3 ).Polymers must bring enough mechanical strength while having a high capacity to dissolve Li salt to dissolve Li salt.Polyacrylonitriles (PAN) and polycarbonates are widely used.LiTFSI and LiFSI are often used as Li salt for their ability to be easily dissolved in polymers.However, PAN/LiTFSI-based electrolytes suffer from low backbones segmental mobility and polycarbonate-based systems suffer from deterioration of their mechanical behavior [85].For instance, graphene oxide filler in a PAN-LiTFSI polymer-in-salt system reaches an ionic conductivity of 1.1 × 10 −4 S•cm −1 at 30 • C [86].This kind of electrolyte still suffers from the low chemical and thermal stability of molten salt [85].Thus, hybrid solutions with polymer blends or inorganic fillers addition are developed to improve ionic conductivity and mechanical stability.

SPE
Salt-in-PEs are basic PE composed of a polymer matrix (>50 wt%) and Li salt.This type of electrolyte provides ease of process, flexibility, and good interfacial behavior with electrodes.However, limited conductivity together with a lack of rigidity and a low transference number [58] are the reasons why polymers with Li salt are just the starting point of other strategies to achieve efficient electrolytes.PEO/LiTFSI has been reported as the most efficient and studied couple but the PE exhibits a maximal ionic conductivity at 25 • C around 10 −6 S•cm −1 .Various couples of polymer/Li salt have been tested resulting in different properties (ionic conductivity, electrochemical and mechanical stability) [39,58].Indeed, Li salt does not have the same affinity and stability with polymers.Coordinating Li ions by carbonyl groups of polycarbonates and polyesters has been investigated as attractive alternatives to polyether [39].Aliphatic polycarbonate-based electrolytes can present improved ionic conductivity up to 10 −5 S•cm −1 at 25 • C and higher transference numbers due to weaker bonding between their polar group and cations [39].Polyester electrolytes also provide a higher transference number but showcase limited ionic conductivities on a wide range of temperatures with a maximum value of 10 −6 S•cm −1 at 25 • C and 10 −4 S•cm −1 at 90 • C [39].With an ionic conductivity of 2 × 10 −4 S•cm −1 at 40 • C, Michel Armand's group [87] has shown that a Jeffamine-based poly(ethylene-alt-maleic)/LiFSI electrolyte could outperform the ionic conductivity of PEO/LiTFSI.Besides, the cyclability at 70 • C is improved thanks to the high amorphous rate of Jeffamine polymer.
Polymer blend electrolytes consist of a blend of two different polymers with complementary properties.The aim is to reduce the crystallinity of the ionic conductive polymer and bring enough mechanical strength to improve electrolyte mechanical properties.In the case of miscible polymers, the electrolyte is composed of one phase and both polymers can be involved in ionic transport.Ragones et al [38] have 3D printed an electrolyte based on the miscible blending of PEO and PLA.The overall crystallinity corresponds to the one of PLA which is the lowest and the overall ionic conductivity of such a blend is lower than a classical PEO/LiTFSI.According to the authors, this is due to a non-optimized printing procedure leading to an incomplete mixability of these two polymers.Some Li + could also interact with PLA which has not the same ability to convey Li + , especially at low temperatures due to the high glass transition temperature of PLA (60 • C).Caradant et al obtained the same conclusion with a blend of PVP (T g = 100 • C) and PEO.They have also studied immiscible polymer blend electrolytes processed by extrusion [49].In this case, one phase can be responsible for ionic conductivity and the other can act as a mechanical reinforcement phase.Reported PEO-based polymer blends have an ionic conductivity in a range of 10 −6 -10 −4 S•cm −1 at low temperatures [58].
Single-ion PEs (SIPEs) are able to conduct only one type of ions thanks to covalently bonded anions to their chain or anion acceptor sites to immobilize anions [88] (figure 12(a)).It results in a Li + transference number close to the unity against t + = 0.1 for PEO/LiTFSI electrolyte [64].Thus, it permits to reduce cell polarization upon cycling, to limit the decomposition of the electrolyte and Li dendrite nucleation [75,88] (figure 12(b)).However, SIPEs usually present ionic conductivities lower than 10 −4 S•cm −1 at room temperature because of anions immobilization [88], higher T g [88], and strong ionic interactions with Li + [75].Their poor cycling stability and difficult synthesis process might be also problematic for large-scale applications [51].Recent studies succeeded in reaching higher conductivity up to 10 −3 S•cm −1 by adopting novel strategies.Thanks to a semi-interpenetrating network of PMMA/lithium PS sulfonate (LiPSS) [89], the high transference number of 0.91 demonstrates the ability to inhibit lithium dendrite growth (figure 12(c)).Another strategy is to immobilize anions in the molecular structure by playing with anion size [72].Li et al have synthesized poly(lithium 4-styrene sulfonate)-carbon quantum dots (PLSSCQD) particles and blended them with PEO.The big size of CQD anions and hydrogen bonding immobilized them in the polymer matrix.(figure 13(d)) They depict a room temperature ionic conductivity of 2.20 × 10 −4 S•cm −1 [90].
Copolymer electrolytes contain a homopolymer in which another monomer unit is introduced to offer different properties (stiffness [91], segmental mobility [92], crystallization [93]).This latter can be grafted on a main backbone, randomly or alternatively introduced to the main polymer backbones, or organized in block copolymers [75].Grafted and block copolymers have been the two main studied families so far.Polymers with EO units, such as PEO, are the most attractive conductive parts while PS is widely used for mechanical reinforcement.This strategy can enhance ionic conductivity at 25 • C thanks to crystalline modifications and strengthening of the mechanical behavior.According to Devaux et al [92] linear PS-PEO-PS, is a good compromise between mechanical behavior and ionic conductivity.They showcased that 30 wt% of PS is necessary to obtain percolation of PS and thus self-standing films which exhibit an ionic conductivity of 1 × 10 −4 S•cm −1 at 60 • C.Even if they hinder recrystallization at low temperatures, comb PEO block-copolymers suffer from poor capacity retention [92].Bouchet et al [65] have highlighted a 'dead zone' effect at interfaces between conductive and mechanical blocks.This region is characterized by an absence of conduction and crystallization due to low segmental mobility.A decrease in the proportion represented by these excluded zones can be achieved by increasing the PEO molar weight.It should be noted that, at temperatures above the melting point where crystalline regions disappear, block PEs showcase lower ionic conductivities (lower fraction of PEO domains) and higher mechanical strength due to their mechanical parts.That is why ionic conductivity improvement compared to PEO homopolymer electrolytes happens below the PEO melting point (around 40 • C) [65].Aldalur et al [91] have worked on Jeffamine-PS copolymer.They obtained well-balanced electrolyte properties with 7.9 × 10 −5 S•cm −1 at 40 • C with improved cycling capability due to a stronger mechanical behavior.They have highlighted a drop of conductivity with a PS side moieties content increase which can be related to a lower fraction of conducting volume and a higher tortuosity.Copolymer electrolytes can also be made by in situ synthesis.A comb-like copolymer PLA/PEG with LiTFSI was in-situ photopolymerized by UV light to achieve a PE with an ionic conductivity of 6.9 × 10 −5 S•cm −1 at 25 • C [94].
Crosslinking PEs consist in linking one polymer chain to another one through covalent bonding.Similarly, to polymer blending strategy, it improves the tensile strength and hinders crystalline domains in polymers.However, it permits to obtain more stable morphologies.Most of the crosslinked PEs are made through the thermal decomposition of a crosslinking agent or irradiation by UV light.However, crosslinking of polymers can have a negative effect on the conductivity as it can restrict polymer chains' mobility [56].It could explain why reported ionic conductivities are in the range of 10 −6 -10 −5 S•cm −1 [95].The crosslinking between PEO and a polymer with a lower Li salt dissociation ability results in a loss of O-Li + interactions so a higher transference number.This is the case of the crosslinked electrolytes of PEO and poly(tetrahydrofuran) realized by Mackanic et al [95].Porcarelli et al [96] have reached a higher ionic conductivity of 10 −4 S•cm −1 at room temperature by in-situ UV photopolymerization technique.They have successfully crosslinked PEO and tetraglyme (TEGDME) thanks to a photoinitiator.
Inter-penetrating networks (IPNs) electrolytes are composed of two or more distinct crosslinked polymer networks with no covalent bonds between them.Compared to crosslinked electrolytes, the mechanical stability between the two immiscible phases is only ensured by the interpenetrating character of the structure.They are usually obtained by sequential or simultaneous polymerization.Semi-IPNs are composed of one or several polymer matrixes penetrated by linear or branched polymer at the molecular scale [75].
Multilayered PEs are composed of different layers to mechanically and chemically adapt interfaces with each electrode.This strategy is used to avoid side reactions of the electrolyte with electrodes which can be restrictive in other strategies.Judez et al [97] have developed a bilayer electrolyte composed of PEO/LiFSI + Al 2 O 3 at the Li metal side and PEO/LiFSI + Lithium ion conducting glass-ceramic (LICGC) at the cathode side.This divided electrolyte structure hinders LICGC reactions with Li metal.This strategy is interesting for high-potential battery applications but provides more solid/solid interfaces which increase the overall impedance of the cell.It also makes harder the processability of the cell.
Nanostructured electrolytes are made of specifically structured channels to improve ionic conductivity by providing fast Li + motion pathways.The goal is to control the morphology of the electrolyte to create nanoscale channels to avoid the tortuosity of pathways and to ensure a direct, fast ionic transport.This strategy is used for composite electrolytes where ceramics 3D networks can be obtained by sintering and impregnation or by electrospinning of ceramics nanofibers.Fu et al [98] have made the most ionic conductive garnet additive by creating ceramics fibers thanks to electrospinning.Bae et al [99], have used nanostructured hydrogel to manufacture their 3D garnet framework to reach ionic conductivity of 8.5 × 10 −5 S•cm −1 at ambient temperature.Jeong et al [100] have created a single ion conducting covalent organic framework by a solvent-free method.Their nano conductive channels, with a diameter of 11.8 Å, permit to enhance the ionic conductivity to 2.7 × 10 −5 S•cm −1 and to obtain a t + = 0.9 at room temperature.

Composite PE (CPE)
CPE are made of a polymer/Li salt matrix loaded by inorganic fillers.Added particles can present electrochemical and mechanical performances whereas organic phases can catch their poor processability and interfacial properties up.Since the first reported work in 1982, several studies have been produced toward a better understanding of CPE behaviors [52,101,102].
There are two types of particles: active fillers can conduct Li + in their structure [103] contrary to passive fillers.In both cases, Li + transport can happen in the matrix or at the interface polymer-ceramic.In the polymer matrix, ionic motion mainly occurs in the amorphous phase according to previously described mechanisms.Ionic conductivity in the polymer or at the interface polymer-ceramic is mainly improved by chemical and mechanical interactions between inorganic filler and polymer matrix.In the case of active fillers, there are two more possibilities for Li + conduction: ceramic pathway and hybrid pathway (polymer + ceramic) (figure 13) [102].
Various key parameters have been identified to manufacture CPE: • Type of fillers: passive or active fillers do not enable the same Li + motion pathways [102].
• Surface chemistry: Lewis acid-base interactions with lithium salt facilitate salt dissociation and thus avoid ion clusters [104] and immobilize anions [105].Fillers' surfaces also interact with the polymer matrix (act as a solid plasticizer) [101].Filler surface modification could be an efficient strategy to achieve an ionic conduction pathway at the interface and reduce the fillers/polymer interfacial resistance [80,102].• Size: nanopowder fillers mean a higher specific area hence providing more interactions with the polymer matrix [106] (decrease in crystallinity, rise in free Li + fraction, widen the electrochemical stability window) [106].Fillers granulometry also influences the optimal loading which gives the best conductivity [107].• Shape: 0D (particles) is the most used shape of particles but it is limited by agglomeration issues.1D fillers (nanowire, nanofiber) provide a more continuous path than spherical ones.It is a convenient strategy for active fillers to avoid polymer/ceramic interfaces so as to offer a continuous pathway for Li + in ceramics [104].2D nanosheets/flakes can create larger active interfaces.A 3D framework of active fillers directly conveys Li + .They are usually realized by porous network impregnation or electrospinning techniques of 3D fiber networks [108].
• Concentration: at low loadings (<10 vol%), enhancement of ionic conductivity is linked to a drop of crystallinity, an improved Li + dissociation, and an increase of filler/polymer interfaces [79,109].There is an optimal content of fillers which gives a maximal ionic conductivity.Below this threshold, ionic conductivity is limited because of lower interactions with the matrix and a higher crystallinity rate.Above this loading, percolation can happen meanwhile rising the risk of agglomeration, thus raising the tortuosity of Li + pathways in the polymer.Processability is also becoming difficult with an increase in hardness and brittleness [110].Based on this assumption, the difference can be made between: * Ceramics-in-polymer which are low-loaded electrolytes in which the ionic conductivity is improved, processability is easy but transference number and stiffness remain low.* Polymer-in-ceramic which are high-loaded electrolytes in which stiffness and transference number are high but the ionic motion can happen through ceramic particles only if the interfacial resistance between organic and inorganic phases is decreased [69].They are considered as promising solid electrolytes because they can gather advantages from both inorganic and organic solid electrolytes [111].These electrolytes are more suitable for large batteries because of their better mechanical behavior which makes them safer than ceramic-in-polymer systems [69].• Manufacturing methods: the manufacturing process has also a key role to play to obtain a homogeneous dispersion of fillers and enable the construction of continuous and uniform transport channels for Li-ion migration [79].Mechanical dispersion of fillers inside a matrix is the easiest way but particle agglomeration is hard to hinder.Chemically organized fillers (grafted particles, in-situ synthesis [112]) permit to reduce efficient interfacial resistance between particles and polymers.Finally, mechanically organized fillers (interpenetrating networks, fillers positioning thanks to rheological behavior, 3D ceramics networks backfilled with a PE) create continuous ceramic pathways in the case of active fillers.

Passive fillers
SiO 2 , Al 2 O 3 , TiO 2 , and ZrO 2 have been mainly used as passive fillers to enhance ionic conductivity (figure 15): • In the polymer: The aim is to reduce the crystallinity of the host matrix by mechanical wrapping around fillers and chemical bonding with polymer chains (figure 14).Mechanically, inert nanoparticles contribute to increase segmental motion by increasing the free volume [52].Chemically, Lewis acid groups on the surface of fillers can compete with Li + to complex with oxygens of PEO and hinder its recrystallization [104].Thus it generates a drop in the polymer's crystallization rate and glass transition temperature (T g ) [113].• At the interface: Reported works also highlighted an ionic conductivity enhancement at high temperatures when the polymer is already in an amorphous state.They demonstrated that specific interactions happen between fillers and the matrix.Indeed, Lewis acid groups on the surface of particles can compete with Li + to complex counter anions and help to dissociate Li salt and avoid ion clusters.It promotes ionic transport by creating fast pathways at interfaces [113,114] (figure 15).It also increases the Li + transference number [101] and the number of free Li + [104]., which corresponds to a maximal ionic conductivity.However, the decrease in crystallinity is not significant (18% of crystallinity without fillers) so the main chain dynamic of the host polymer matrix does not change significantly.Dispersed SiO 2 can improve ion-conduction behavior up to a loading of 10 wt% of because of agglomeration issues [52].A chemically organized dispersion of particles could create Li + conduction pathways at the interface ceramic/polymer as well as reduce more drastically the polymer crystallinity.Modified SiO 2 nanoparticles have been widely studied to promote Li salt dissociation and avoid filler agglomeration such as porous vinyl functionalized (PVF)-SiO 2 [116], or walnut-like SiO 2 [117].A deeper investigation of SiO 2 inert fillers/Li salt interactions demonstrated that the functionalized silica surface plays a major role in LiTFSI dissociation.SiO 2 -Li particles induce a drop of one decade in terms of conductivity [105].A threshold of conduction exists at a sufficient amount of fillers inside the matrix.Choudhury et al [118] have used hairy nanoparticles of SiO 2 grafted with hydroxyl end-chain groups of PEO.These nanoparticles were then crosslinked with PPO to bring the mechanical resistance.Such a technique enables to reach high mechanical modulus of the membrane which still needs to be soaked with liquid electrolyte to achieve high ionic conductivity.In-situ synthesis of particles in the polymer matrix is another way to reach this goal.Lin et al [119] have showcased that in-situ synthesis of SiO 2 nanospheres in a PEO (M V =600 K) matrix can reduce polymer crystallinity rate by chemical bonding and mechanical wrapping.Xu et al [110] succeeded in dispersing homogeneously SiO 2 particles by in situ synthesis in a PEO network.The low PEO crystallinity leads to a conductivity of 1.1 × 10 −4 S•cm −1 at 40 • C. In their exhaustive review, Zheng et al [52] have reported ionic conductivities from 10 −6 S•cm −1 to 10 −4 S•cm −1 for passive particles dispersion in a polymer matrix.

Active fillers
Perovskite (LLTO), NASICON (LAGP), sulfide (LGPS) and garnet (LLZO) have been mainly used as active fillers [75].Perovskite-structured fillers provide lower ionic conductivity compared to the other fillers and can be reduced at low potential (vs.Li/Li + ) [80].NASICON-type fillers are good candidates to reach liquid-like ionic conductivity up to 10 −2 S•cm −1 at room temperature but their rigid nature induces interface issues with electrodes and Ti-containing particles are not stable at low potential [120].This is also the case with sulfide additives.Garnet-type particles show ionic conductivity up to 1 × 10 −3 S•cm −1 at room temperature with a wide electrochemical stability domain.Acidic groups on these particles bind to the salt anions which furnishes the shared electron pair as Lewis base [121].It immobilizes anions and therefore increases the transference number.In their extensive review of garnet-type solid-state fast Li-ion conductors, Thangadurai et al [122] reported the highest ionic conductivity at 25 • C, 1 × 10 −3 S•cm −1 , for LLZTO (Li 6.4 La 3 Zr 1.4 Ta 0.6 O 12 ), thanks to its pure cubic garnet structure [123].Zr-and Ta-based materials are stable on a wide potential range (up to 6 V vs. Li/Li + at 25 • C) and appear stable against chemical reactions with Li metal [124].That is why the following section mainly emphasizes PEO/Ta-doped LLZO CPE in which ionic conductivity can be enhanced: • In the polymer: in the same way as passive fillers interact, it hinders the polymer crystallization.
• At the interface: space charge region enhances the ionic conductivity.Free energy difference leads to Li + migration at the surface of fillers, which leaves negatively charged vacancies in the lattice [101].• In the fillers: Li + movement follows the Arrhenius model inside particles.Li cation can also move in both, polymer and fillers, through a hybrid pathway (figure 16).However, the high interfacial resistance between particles and polymers makes this kind of pathway impossible [102].
Dispersion at low loadings: Ionic conductivity increases with the content of fillers up to a maximum of around 10 and 15 wt%, which depends on the powder granulometry [98,111].Hua Zhang and coworkers [68] have shown that crystallinity can be divided by 2 thanks to the addition of LLZTO nanoparticles in a PEO/LiTFSI matrix.This increase in ionic conductivity below the LLZTO percolation threshold shows that dispersed nanofillers can improve the ionic conductivity in the polymer matrix (more free volume for PEO chains motions, and decrease of the crystallinity) [69].However, better ionic conductivity values have been obtained for a PEO:LiTFSI (EO:Li = 8:1-12:1) with 10 wt% of LLZTO nanoparticles, than for a classical PEO-LiTFSI system and PEO-LiTFSI with SiO 2 or ZrO 2 [52].Interactions between PEO, LLZTO particles, and Li salt create Li vacancies at the LLZTO surface, sites for ionic conduction at the interface [120].Around 10 wt% of LLZTO also leads to better cycling stability compared to PEO-LiTFSI due to dendrite suppression and better stability at the anode interface.Indeed, it tends to reduce the PEO area exposed to Li metal and reduce the formation of a passivation layer of Li 2 O. Interfacial effects at the polymer/garnet interface are usually displayed to explain enhanced ionic conductivities.Zhang et al [107] have studied Li salt-free electrolyte PEO/LLZTO to avoid the segmental motion effect and just analyze the effect of the PEO/LLZTO interaction.Their electrolyte provides an ionic conductivity of 2.1 × 10 −4 S•cm −1 at 30 • C for 12.7 vol% of LLZTO nanopowder due to the space charge region at the interface.Zheng et al [125] have found that EO:Li molar ratio has a critical role to play in conductive interface formation.According to them, LGPS particles are softer than LLZO that could explain a less challenging interface formation in their LGPS-PEO (70 wt% EO:Li = :1) electrolyte which exhibits an ionic conductivity of 2.2 × 10 −4 S•cm −1 at 25  hybrid pathways and Li + motion only happens in the organic phase.It explains that higher loadings induce lower ionic conductivity by decreasing the percolating amount of polymer and increasing the tortuosity of conducting pathways.Moreover, higher loading of fillers leads to aggregation due to a surface energy gap between fillers and polymer [126].Zheng and Hu [127] have showcased this phenomenon by studying PEO/LiTFSI electrolytes with LLZO from 5 to 50 wt%.At 20 wt%, no improvement is observed due to the blocking effect of particles.At 50 wt%, the percolation threshold is overtaken and LLZO is the main source of Li + which can be carried.However, there is no significant enhancement of ionic conductivity because of grain boundary interfacial resistance and particle blocking effect [127].It is also the origin of the lower ionic conductivity (1.12 × 10 −5 S•cm −1 at 25 • C) obtained by Zhao et al [121], with their electrolyte composed of 40 wt% of LLZTO particles in PEO/LiTFSI matrix.Despite this, their electrolyte provided a wide electrochemical window (up to 5.5 V vs. Li/Li + ) and a t + of 0.58 at 25 • C. Gupta and Sakamoto investigated interfacial resistance origins [128].A layer of impurities (Li 2 CO 3 ) on LLZTO nanoparticles resulting from proton exchange when the sample is exposed to moisture creates a barrier to Li + and is responsible for electrostatic repulsion with PEO.With an appropriate heat treatment of LLZTO (700 • C for 10 h under argon) and a convenient amount of LiTFSI (EO:Li = 15:1), they succeeded in sharply decreasing the interfacial resistance (figure 17(a)).Kuhnert et al [102] have also tried to lower interfacial resistance in a PEO/LLZTO electrolyte.They have grafted PEO backbones on LLZTO thanks to CTMS (3-chloropropyl) trimethoxysilane.They have been able to reduce the interfacial resistance and enable higher Li + conductivity through the interface (figure 17(b)).CTMS has been also involved in the chemical bonding of LGPS microparticles with PEO backbones which results in an ionic conductivity of 9.83 × 10 −4 S cm −1 at 25 • C with a high Li + transference number of 0.68 (against 2.42 × 10 −4 S cm −1 and 0.58 for a dispersed LGPS in PEO) [129] (figure 17(c)).Li + motion has been greatly enhanced thanks to LGPS bulk properties, weaker Li-sulfide interactions than Li-oxygen, and a reduced grain boundary resistance between particles.Authors have also blended PEO with two widely different molecular weights (2000 and 1 × 10 6 ) to facilitate the bonding with CTMS and provide faster Li + pathways through the vehicular mechanism.Huang et al [130] have also demonstrated that modifying interfaces in the organic-inorganic composite electrolyte is an efficient way to improve ionic conductivity, thermal stability, and electrochemical performances.Their polydopamine coating on LLZTO nanoparticles (80 wt%) allows to reduce strongly the interfacial resistance (4 times lower at 50 • C) with PEO/LiTFSI (20 wt%) which results in a promising ionic conductivity of 1.15 × 10 −4 S•cm −1 at 30 • C.This enhancement effect is mostly due to a more homogenous particles distribution and facile Li + pathways created at the interface.At these loadings, LLZTO-loaded PEO-LiTFSI electrolytes present better performances than ZrO 2 -loaded ones.Mechanically, high loadings of fillers permit to reach sufficient mechanical resistance to prevent lithium dendrite growth [131].For instance, Li et al [111] have extruded a PEO-LiTFSI electrolyte with 75 wt% of LLTO, which exhibits a high tensile strength (3.85 MPa) preventing the formation of dendrites.Long Chen and coworkers [69] have also studied the effect of adding LLZTO microparticles from 0 to 80 wt% in a PEO-LiTFSI (O:Li = 8:1) electrolyte.At 80 wt%, the addition of a plasticizer such as PEG is necessary to have enough flexibility to keep mechanical integrity.Ionic conductivities remain lower for polymer-in-ceramic (80 wt%) than for ceramic-in-polymer (10 wt%) systems.

3D continuous ceramic pathways:
In order to overcome interfacial resistance and take benefit of the ionic transport performances of ceramics, researchers attempted to create 3D continuous Li + pathways.MJ.Palmer et al [109] have created a high-loaded electrolyte made of a 3D framework of conductive ceramics (77 wt%).They sintered a thin layer of LATP microparticles and then backfilled it with a crosslinked PE based on PEO/Jeffamine and LiTFSI to manufacture a 25 µm thick composite electrolyte (figure 18(a)).They obtained an ionic conductivity two orders of magnitude greater than that of a dispersed ceramic system (3.5 × 10 −5 S•cm −1 at 20 • C).Electrospinning is another method widely used to generate high-loaded ceramic-polymer nanofibers.These nanofibers form a 3D ultra-conductive network which can be backfilled by a PE.Mengmeng Zhang et al [132] have reached an ionic conductivity of 1.05 × 10 −4 S•cm −1 at 50 • C thanks to their LLZTO/PVDF/PEO network in a PEO/LiTFSI matrix.A slightly higher conductivity of 2.15 × 10 −4 at 25 • C has been obtained for a similar method with PVP/LLZTO fibers [98] (figure 18(b)).
Thus, main advantages of these different strategies can be classified regarding their: mechanical stability, ionic conductivity, transference number, and electrochemical stability (figure 19).Compatibility of these strategies with FFF printing specifications is discussed below to identify the method which can provide the highest ionic conductivity at room temperature with an electrochemical stability window compatible with an LFP/graphite cell cycling.

FFF printing specifications
Based on this landscape of strategies, the aim is to identify potential approaches for 3D printing a PE by FFF 3D printing.In this process, two counter-rotative rolls carry a polymer-based filament in a heating block through a PTFE duct.Polymer melts in the heating block and filament acts as a piston driving melted materials out of the printing head through the nozzle [133].Obviously, filaments have to contain a minimal amount of thermoplastic polymer which restricts the quantity of fillers in the case of loaded filaments.Despite the great versatility of this process, specific mechanical and rheological behaviors are needed to  qualify a filament as 'FFF printable' .Thermal properties are also important to ensure good adhesion between each printed layer.All of these requirements can strongly go against ionic conductivity properties of electrolytes.
Mechanical properties of the filament will influence the feedability.The filament must have a low ductility not to be crushed by feeding rolls.If it is not the case, it will pass through feeding rolls with a high contact area that induces winding of the filament around rotating rolls [134].Go et al [135] have showcased that it occurs when the extrusion force is higher than a critical force linked to the shear strength of the material and lower than the needed force to push melted material in the heating block (figure 20(a)).However, a too-low ductility filament can be very brittle and break in the print head [130] (figure 20(b)).In the case of composite feedstock, this brittleness increases significantly with the loading of fillers [136].The filament must also have enough stiffness to avoid buckling after feeding rolls [137].It is acting as a piston which has to apply a minimal pressure on the melted material to exit it from the nozzle.According to Venkatarman et al [133], this needed pressure is proportional to the melt viscosity.Thus, they have used the ratio of Young modulus on melt viscosity to predict filament buckling (figure 21(a)).Arrigo and Frache [137] have applied this model to the 3D printing of different commercial PLA filaments.With their parameters (nozzle 0.4 mm, print speed 30 mm•s −1 ), the critical ratio was around 4 × 10 −5 s −1 .A minimal flexibility is also important to obtain a spoolable filament and to avoid filament rupture.Finally, a too-sticky filament can adhere to the rotating roll or the PTFE sheath of the heating block.It results in winding or blockage of melted material in the PTFE sheath.Thus, Maurel et al [24] have tried to adapt the printer by suppressing the PTFE tube to 3D print a PEO: LiTFSI (O:Li = 20:1) system.Studies have been devoted to quantifying the required mechanical properties for the feedability of an FFF 3D printer.Through 3-points bend tests, buckling tests, and hardness tests, mechanical properties have been confronted with the printability.A correlation has been demonstrated between filament hardness and printability [138] (figure 20(c)).These features are coupled with rheological behavior.
Rheological and thermal properties suitable to the process are needed to allow the flow of polymers through the nozzle.FFF3D printing is a non-continuous extrusion process that induces shear rate fluctuations changing the viscosity of the melt [139].Beran et al [136] have modified the printing speed from 20 mm•s −1 to 70 mm•s −1 and they found the equivalent shear rate between 2 × 10 2 and 6 × 10 2 s −1 .According to Arrigo and Frache [137], the printing of a 1.75 mm diameter filament, with a 0.4 mm nozzle at 30 mm•s −1 , corresponds to a shear rate between 9 × 10 2 and 2 × 10 3 s −1 .They have investigated rheological properties required for FFF printing by studying: flowability, filament buckling, shape stability, die swelling, and interlayer adhesion.Non-Newtonian characteristics are beneficial for printing with a low melt viscosity during the flow through the nozzle that increases rapidly when the material exits (zero-shear conditions).It contributes to reduce buckling risk [136] as well as it permits to avoid dripping during the non-printing movement.For Samoro et al [139], high melt viscosity is a drawback for the constant flow during the printing of drug-loaded filament.However, non-Newtonian features provide swelling when materials exit the nozzle so optimized printing parameters are needed [133].As printing involves heating, melting, and solidifying, convenient thermal properties with a sufficient content of thermoplastic are needed [139].For instance, the incomplete relaxation of polymer chains during cooling will affect negatively the welding of each layer [137].Vaes and Van Puyveld [140] have studied FFF printing of semi-crystalline polymers.For these polymers, a drop-in viscosity happens when the melting point temperature is jumped over.A minimal heat transfer is required to lower polymers' viscosity to avoid higher extrusion forces and nozzle clogging [140].It intensifies at higher print speeds which also induces higher extrusion force due to a reduced residence time of the extruded material in the heating block.The maximal extrusion force, which is around 60 N for a classical 3D printer, can be exceeded leading to printing failure [135].
For battery printing applications, researchers have tried to decrease the content of hosting polymer (from 80 wt% [35] to 46 wt% [27]) to maximize electrochemical performances of printed filament.However, fillers have an impact on mechanical and rheological behavior.Beran et al [136] have studied the printability of loaded filaments.A decrease lower than 50 vol% of hosting polymer hampers the filament's printability.Depending on the size and shape of fillers, it could generate clogging during printing.An increase in size results in higher viscosity and therefore the use of a larger nozzle when printing.For a ratio nozzle diameter/filler size lower than 6.2, they have observed systematic clogging of the filament [136] (figure 21(b)).However, a larger nozzle induces higher printing layer height and thus a lower resolution.They conclude that the printability is influenced by the filler volume content, the zero-shear viscosity of the matrix, and printer conditions (printer speed and temperature).In their study on FFF printing of a glass spheres-filled polycarbonate filament through a 0.4 mm nozzle, they have established a viscosity criterion on the zero-shear viscosity of the matrix.Moreover, Zhang et al [134] have worked on drugs-loaded thermoplastic polyurethane filament for FFF printing for the pharmaceutical domain.They investigated the impact of the roughness and variation of filament diameter induce by fillers, which can be the origin of printability issues.

Microgravity environment specifications
For the last 10 years, space agencies have worked to develop on-orbit additive manufacturing facilities.Polymer printing by SLA or FFF, and metal printing are the main studied processes.Major achievements related to additive manufacturing by FFF process suitable for microgravity or being operated in microgravity are reported in table 2. Several suitable 3D printers have been developed to process a range of thermoplastics in microgravity and 3D printed parts have already been manufactured onboard the ISS.Moreover, 3D printing using continuous carbon fiber reinforced PLA composites in space has been achieved by Chinese research institutes on board spacecraft [141].
It is clear from the list of successful projects mentioned (table 2) that this process is fully compatible with a microgravity environment leading to the manufacturing of parts with comparable properties as one manufactured under earth gravity [6,7].So far, the FFF process in orbit has been demonstrated to be stable and efficient.Full process automation is implemented, and the reliability of such 3D printers as well as printing parameters are thoroughly assessed to minimize the need for human intervention besides printed part recovery.The range of polymers used spans from commodity (PLA/ABS) to high-performance thermoplastics (PEEK/polyetherimide (PEI)-based Ultem®).Looking at the range of material processed so far, microgravity-compatible 3D printers are capable of reaching extrusion temperature at least compatible with Ultem® and PEEK (∼380 • C).
From the available literature, all specific printing parameters used that enable the fine-tuning of the printing process as a function of the polymer are not disclosed.The 3DP project used ABS and a 0.4 mm diameter nozzle.In the POP3D project that used PLA, the printing temperature was set between 160 • C and 180 • C with an average flow rate between 1 and 10 mm 3 s −1 [7].For the MELT project focusing on PEEK, a 0.4 mm nozzle was used with a temperature of 380 • C and a printing speed of 20 mm s −1 [142].The IMPERIAL project has assessed the capability to 3D print parts with no size limitation in one direction of the build volume with an extrusion temperature of 400 • C, and heated chamber at 100 • C in order to process PEEK and Ultem®.
To implement such a process under microgravity, there are several key aspects to consider that are linked to the fine-tuning of the printing parameters considering the printer design (nozzle diameter, heating capability, motor accuracy…) and the intrinsic material properties (melting temperature, melt flow index).The first one is the feedstock shape and how the material in its solid state is fed to the extruder to enable the melting process.Most of the experiments performed so far have used filaments that are stored in spools.This implies that the filament of the feedstock needs to exhibit a certain flexibility to be able to be rolled around the spool and later unrolled without any irreversible deformations (bending, kinking, breaking) that might hamper or block the feeding process.Therefore, very stiff filament or composite filament highly loaded with filler might present issues in reaching the desired radius of curvature to be stored in a spool of a convenient diameter.For the feeding process, the potential issues described in table 3, must be avoided to enable a good feeding process with a continuous and controlled flow of material in the nozzle but also to avoid any human intervention or need for extensive maintenance by operators.It is of utmost importance to underline that in orbit, astronaut crew time is very limited and needs to be efficiently used with respect to their mission; any loss of this time to solve such issues will be negatively affecting the efficiency of such process and its overall added value.
On-orbit FFF 3D printing imposes additional constraints compared to on-earth manufacturing.The confined environment of a space station hinders the use of highly volatile species.Safety requirements of the ISS, establish a toxicity hazard level (THL) (from 1 to 4) according to volatility and the hazardous nature of chemical species [147].Even if, all 3D printers are operating in a closed environment including venting and filtering capabilities, it is highly recommended to select materials with low-volatility chemicals.Battery electrolytes are considered to be THL-2 or higher.There are also stringent rules on flammable materials to eliminate fire propagation issues in the ISS environment.The lack of gravity makes the use of fluid critical, especially in the case of wetting a porous media.Particles rearrangements and separations can accentuate wetting instabilities [148].
In summary, the compliance of the FFF process with microgravity is due to a compliant surface tension of melted thermoplastic and its viscosity that enables the melt pool to be extruded and adhere to a primary surface.From these achievements so far one can conclude that any thermoplastics that can find similarities with the ones described in table 2 could be processed efficiently under a microgravity environment.In this respect, the most important parameters to be assessed when it comes to new materials are related to the feedstock mechanical properties for the nozzle feeding process (feedability), the achieved viscosity, shear rate, and pressure during the melting process (rheology) [149] and finally the limits imposed by the 3D printer design in term of printing volume, accuracy, and versatility (multi-material possibility, ease of operation…).

Suitable strategies for PE printing in microgravity
FFF 3D printing specifications are not compatible with all PE strategies (table 3).A minimum amount of thermoplastic polymer is required to meet rheological and mechanical features which is not the case for polymer-in-salt and polymer-in-ceramic strategies.Thus, these strategies are considered unfavorable regarding their rheological behavior.Quasi-solid electrolytes and salt-in-polymer electrolytes display issues of feedability because of their lack of mechanical stability and stiffness.For instance, difficulties in printing a PEO:LiTFSI 20:1 electrolyte have been illustrated by Maurel et al [24].The sticky filament did not provide enough hardness to be printed with classical printer configuration.In highly plasticized PEs, organic solvent volatilization during extrusion, printing, and storage can degrade the mechanical behavior of the filament.For instance, filaments containing PEGDME200 and acetyl tributyl citrate (ATBC), tend to exude plasticizer and become fragile under atmospheric conditions and room temperature, even if reported boiling points are higher (327 • C for ATBC) [150].That is why the use of a volatile plasticizer to have a more flexible filament is a short-term solution [139].Like other strategies involving volatile chemical species, it is strongly going against the above-mentioned safety requirements of the ISS even if mitigation could be implemented due to the availability of a venting line and having a printer hermetically closed.Every strategy with a score lower than 3/5 on figure 19, has been classified as unfavorable for mechanical reasons.Polymer-in-ceramics are also unfavorable due to their high stiffness.In terms of morphology, filament extrusion followed by FFF printing involves the melting of species at different temperatures and shear rates.It results in specific anisotropic morphologies, influenced by many parameters (extrusion and printing parameters, polymers rheological behavior) [140].Gel polymer electrolytes, controlled nanostructured channels, interpenetrating networks, and crosslinking polymer solutions are feasible only if the morphology can be controlled all along the process.Indeed, GPEs require entrapped liquid phases, and crosslinked PEs are mainly manufactured by thermal decomposition which is not compatible with extrusion.Thus, entrapped liquid phases, nanostructured channels, and photo-initiated crosslinking of polymers are considered infeasible by a double extrusion process.That is why they are regarded as unfavorable strategies for morphology issues.With these considerations, four residual strategies appear suitable to print PEs by the FFF process.Multilayered electrolyte strategy requires different filaments to realize the 3D printing which makes the process in case of multi-material printing more complicated.Additional printing heads would be needed to avoid the use of two filaments with the same nozzle which requires cleaning steps.Such multilayered electrolyte exhibits lower ionic conductivity but offers added value in high-potential energy storage applications.It is the same for the single-ion strategies.It reduces the polarity of the cell without displaying the best ionic conductivities at room temperature [88].This strategy is especially interesting for LMB as it permits to prevent dendritic growth.Copolymer solutions are promising strategies requiring important chemistry steps to suit extrusion and 3D printing.
With these assumptions regarding process conditions and microgravity environment, polymer blending and CPE are the two remaining strategies suitable for the 3D printing of Li-ion polymer batteries by FFF in a microgravity environment.
Polymer blend electrolytes by FFF:.Immiscible blends can offer complementary features that can be studied with two polymers: polymer A, containing Li salt, serves as a Li + pathway with a poor mechanical behavior; polymer B, chemically inert, acts as mechanical reinforcement.It could improve ionic conductivity by: 1. Reducing the crystallinity rate thus increasing segmental mobility in phase A 2. Increasing the lithium salt content at an optimized rate in phase A, without degrading the overall mechanical behavior thanks to the mechanical compensation of B 3. Creating preferential pathways of Li + through phase A thanks to the resulting morphology The choice of polymers, salt and proportion (Polymer A:Polymer B:Li salt) are key parameters.An increase in polymer B content would enhance overall filament stiffness, but would decrease the volume proportion of the conducting phase, which could degrade the ionic conductivity.On the opposite, an increase in Li salt content could enhance ionic motion while plasticizing the filament.Process parameters (extrusion and 3D printing processes parameters) have also a major influence on electrolyte performances.In this double extrusion process, temperatures and shear rate affect the morphology, by influencing the viscosity and crystallinity rate, which are two key parameters of ionic motion.Extrusion temperatures must be between the melting and the degradation points of polymers.Indeed, they must be processable in the same range of temperature [151].The speed and rotating mode (co-rotative or counter-rotative) of the screw will modify the applied shear rate.A minimal shear rate is needed to ensure the complete melting of polymers as well as the good homogeneity of the blend.However, a too high shear rate, as well as a too high temperature, lead to polymer degradations.Verdier et al have underlined PEO degradations above 135 • C during extrusion.According to them, polymer degradations should always be considered when the extrusion process is used [151].3D printing, as a discontinuous extrusion step, also impacts crystallinity and morphology as shown by Vaes and Puyvelde [140] in their review.It depends on printing parameters (liquefier temperature, bed temperature, printing speed and strategy) and the type of polymer.Temperature should be high enough to lower the viscosity and avoid filament buckling in the upper part of the printing head.Heated bed and chamber influence the recrystallization and tend to increase the crystallinity rate [152].The ionic conductivity of the 3D printed electrolyte could be improved by a drop of crystallinity provided by a fast cooling (high fan speed and low bed temperature).Polymer degradations are also supposed to happen during printing.Finally, printing strategies should be considered due to the anisotropic morphology of printed parts.In biphasic filament morphology, an elongation of domains along the printing direction can be observed [140].Moreover, macro porosities are visible inside FFF 3D printed parts [153].All of these parameters will modify the viscosity of each phase which has an impact on tortuosity and accessibility of the ionic conductive phase (figure 22).A fine-tuning these parameters is needed to obtain a trade-off between mechanical and electrochemical properties.Petra Potschke and coworkers have investigated co-continuous structure formation in an immiscible blend [154].The co-continuity mainly depends on the viscosity ratio and volumetric phase proportion.Thus, a precise process condition permits to find the inversion phase point at which the co-continuity is reached.Several empirical models have been proposed to describe the phenomenon with a common point of co-continuity at the equivolume and equiviscous point.A co-continuous structure permits a maximal contribution of the mechanical modulus from polymer B while ensuring a continuous pathway for Li + in polymer A.

Composite electrolyte by FFF:
This strategy involves a polymer A, which conveys Li + and fillers, passive or active.In the case of passive fillers, the aim is to avoid filler aggregations, and homogenize filler dispersion in the polymer matrix to enhance the mechanical behavior of the filament and reduce crystallinity.Thus, process parameters such as higher shear rates and temperatures contribute to homogenizing the particle's repartition in the polymer.The sequencing is important in this case.For instance, fillers and polymers premixed without Li salt can favor polymer-filler interactions to avoid aggregation.In the case of active fillers, percolation of particles is required to leverage their high ionic conductivity.If the percolation threshold is overcome, it could improve the ionic conductivity by opening new pathways through inorganic particles.However, more than 40 wt% of fillers would hinder the printability.Thus, it is necessary to lower the percolation threshold to benefit from the filler's ionic conductivity without compromising the printability.An immiscible co-continuous polymer blend can be used to confine particles inside one phase or at the interface.Plattier and coworkers have confined carbon black particles at the interface of a co-continuous blend of PP and poly-ε-caprolactone thanks to a viscosity ratio close to one [155].Shear rate and temperature during extrusion and printing, as well as polymer proportions control fillers localization by modifying viscosities.The 3D printing of CPE, based on inactive fillers, has already been studied. 1 wt% SiO 2 starts to decrease crystallinity and permits improved printability [38].Mejia et al [156] have extruded an electrolyte composed of PEO (Mn = 5 × 10 6 )/LiTFI (EO:Li = 12:1) with additional D-α-tocopherol polyethylene glycol 1000 succinate (TGPS)-coated sepiolite nanofibers (up to 15 nm).It was found that the latter favored mechanical behavior by creating a 3D network of PEO.The ionic conductivity was close to 10 −3 S•cm −1 at 25 • C with a solid-like behavior that makes it extrudable.However, they have used 40 wt% of ethylene carbonate (EC) to highly plasticize their electrolyte which explains the reached ionic conductivity.To avoid the use of EC with its low electrochemical and thermal stability, they have replaced it with a less volatile IL, PYR 14 TFSI.However, the ionic conductivity was reduced by an order of magnitude [157].Two recent papers demonstrate the solvent-free manufacturing and extrusion feasibility of composite electrolytes based on active fillers.Li et al have used extrusion to create an electrolyte membrane of PEO/LiTFSI filled with 75 wt% of LLTO.This extrusion method with a high content of ceramics provides a tensile strength 3 times higher than that of the solution casting method and a transference number 5 times higher than that of PEO/LiTFSI.However, it portrays an ionic conductivity of 2.95 × 10 −5 S•cm −1 at 30 • C and a high interfacial resistance [111].Moreover, such loadings are not suitable for FFF 3D printing.A PEO/LiTFSI electrolyte filled with low loading (10 wt% of LLZTO nanoparticles) has been manufactured by solvent-free hot rolling.It exhibits a similar ionic conductivity of 6.8 × 10 −5 S•cm −1 at 30 • C [68].

Residual challenges
The main challenge towards Li-ion polymer battery printing is the PE manufacturing, which is non-trivial due to a strong opposition between mechanical et electrochemical properties.However, it is not the only challenge to take up so as to 3D print by FFF full polymer batteries.Complementary to this major issue, the following residual challenges stand in the way in terms of all-solid-state cell assembly, process conditions, and performances optimization.
• All-solid-state assembly: Electrodes: In all-solid-state batteries, the ionic transport from the electrolyte interface to the active material located in electrodes is difficult to achieve.The need for clear pathways up to active material makes the presence of solid electrolyte inside electrodes necessary.Thus, improvement of energy density and Coulombic efficiency could be obtained by replacing traditional electrode binders with PE formulation [157].Ragones et al [38] made the hypothesis that their low capacity and sloping charge/discharge profiles were linked to the absence of Li salt and PEO in their printed PLA-based electrodes.Aldalur et al [158] have used a new formulation of amorphous PE based on (propylene oxide (PO)/EO) Jeffamine (4 × 10 −5 S•cm −1 at 25 • C) as a binder in the cathode with LFP (over 150 cycles).
Casing protects active parts of the battery.It must have high mechanical and thermal stability, good adhesion with all the parts, and be chemically and electrically inert.PEEK, nylon, or reinforced filament with fiberglass are therefore good candidates even if they required more stringent 3D printing parameters.Low porosity is also important to ensure the impermeability of the casing.Printing temperatures for the casing have to be chosen not to degrade other filaments.A thermal gradient can also affect the crystallinity and morphology of other filaments which will have an effect on electrochemical properties.
Interfaces: Contact at interfaces are a major concern in the case of all-solid-state batteries.Solid-solid interfaces provide a higher resistance that hinders Li + motions through interfaces.Behavior at interfaces with electrolytes/electrodes must be studied.Interlayer adhesion can provide good contact between each electrode and the electrolyte but might be impacted by the microgravity environment.Process temperatures and cooling behavior must be compatible with each filament.

• Process conditions requirements:
Dry environment: Lithium salt and some polymers are highly hygroscopic species.Extrusion and printing steps have to occur in a low humidity-controlled environment to avoid a drop-in viscosity and degradation of the filament mechanical stability.Multi-material printing: As one-shot printing involves a different feedstock for each part of the battery, it is important to avoid any contamination during the process.Residual conductive or active materials in the printing head can cause short-circuit.Several answers have been imagined to overcome filament pollution.Cleaning sequence with an inert material will increase the manufacturing time.Multi-material printers with one printing head for each filament would permit a one-shot 3D printing while reducing the pollution of each part by another one.
Printing parameters optimization: Temperatures, speed of printing, cooling rate or layer height are parameters which can affect the quality of printing and battery electrochemical performances [33].Greener processes: Extrusion and 3D printing are good candidates to be solvent-free methods of manufacturing.Removing hazardous solvents as well as limiting the use of nanopowders are key factors to ensure safer working conditions.
Microgravity environment: In case microgravity is part of the printing environment, there are a few uncertainties concerning the use of filled thermoplastics and their homogeneity after being melted and extruded.Microgravity may induce local changes in the filler content, dispersion, and homogeneity within the printed material creating a gradient within one deposited layer.If such an effect leads to a decrease of filler content at the interface, printing will have an impact on the overall ionic conductivity by forming, at the microscopic level, a gradient of properties that will differ from the bulk material properties.

• Electrochemical performances optimization:
3D architecture: Optimized architecture could permit to make up for energy density loss induced by the large proportion of thermoplastic.Modeling studies could help to design more efficient batteries with larger specific areas.

High potential materials:
The majority of studies choose LTO and LFP as negative and positive electrode materials, respectively, because of their low volumetric expansion and thermal stability [29].An increase in efficiency could be brought by studying 3D printing of higher potential cathode material such as NMC for the cathode.It will bring new challenges for cathode 3D printing as NMC particles can be bigger than LFP which could create issues of polymer coating and filament homogeneity which can impact cell performances.Concerning electrolytes, higher potential means side reactions with polyether such as PEO.That is why electrochemically stable polymers have to be explored even if they provide less ionic conductivity and low-potential stability.

Conclusion
On the one hand, FFF printing has been successfully implemented on orbit and onboard the ISS, which demonstrates its suitability for a microgravity environment.The target is to be able to answer astronauts' vital needs for future long-mission, by manufacturing on-demand, tailorable, and complex items directly in space decreasing the need for spare parts brought from Earth.On the other hand, FFF is a new promising manufacturing process for energy storage devices.Its freedom of design and ease of process could permit a reduction in cost and time of production.During the last 5 years, FFF 3D printing of electrodes, separators, and current collectors for LIBs have been studied with promising results.However, these cells still need to be impregnated with hazardous and volatile liquid electrolytes.Thus, SPE printing represents the main challenge to take up towards one-shot printing of LIB in microgravity.Up to now, two groups have already 3D printed electrolytes by FFF, however facing printability issues [24] or reaching ionic conductivity 10 −5 S•cm −1 at 60 • C that are low compared to the needs [38].Indeed, Li + motion mechanisms in polymers are non-trivial and their semi-crystallinity makes them inefficient at room temperature.Several strategies have been investigated to overcome these issues.We can gather them into three main categories: QSPE, CPEs, and SPE.QSPEs suffer from poor mechanical behavior, and some strategies from CPE or SPE groups require specific morphologies or a huge amount of fillers.That is why, facing FFF 3D printing mechanical, rheological, and morphological specifications, most of electrolyte strategies are incompatible with the process.Moreover, solutions involving volatile, flammable, or hazardous species are not suitable for the safety requirements of the confined environment of a space station.Among feasible solutions, SPE with polymer blend and CPE seem to represent the best way to successfully 3D prints an electrolyte compatible with a microgravity environment.The main parameters which are materials choice, materials proportion, and extrusion/printing parameters have to be optimized to enhance ionic conductivity without degrading the mechanical behavior.In the case of CPE, active fillers are attractive to improve this ionic conductivity especially if they can be confined in a continuous network.Thus, the polymer blend strategy could be used to confine active fillers in one phase or at the interface between the two domains by playing with the sequencing, viscosities, and wetting coefficients.Future studies could be performed on copolymer strategy which could be an efficient method but it requires chemical synthesis of a particular polymer able to convey Li cations and be processable in a double extrusion process at the same time.On the process side, the utilization of direct extrusion additive manufacturing [139] should be worth considering as it would avoid the use of filament as feeding materials at the origin of strong mechanical constraints.It could bring formulation freedom by directly feeding the printer with pellets or powder and it has not been investigated yet for LIB manufacturing.To conclude solvent-free PE 3D printing by FFF is the crucial step to open the way toward one-shot Li-ion polymer battery printing in a microgravity environment.

Figure 3 .
Figure 3.Comparison of conventional and additive manufacturing steps.

Figure 4 .
Figure 4. Proof of concept of the Li-ion coin cell 3D printing by FFF with liquid electrolyte.Reprinted with permission from [30].Copyright (2018) American Chemical Society.

Figure 5 .
Figure 5. Reported studies on filaments for FFF 3D printing of Li-ion batteries.

Equation 1 .
General equation of current through a polymer electrolyte.

Equation 6 .
Dynamic Percolation Bonding (DPB) with W the hopping rate (a) in Arrhenius case, (b) in VTF case

Figure 9 .
Figure 9. Main mechanisms and models of ionic conduction in a polymer electrolyte.

Figure 16 .
Figure 16.Main mechanisms for Li + motion in CPE with active fillers.

Table 1 .
Latest classifications of polymer electrolyte strategies.

Table 2 .
List of FFF 3D printers operated in and developed for the microgravity environment.

Table 3 .
Compatibility of polymer electrolyte strategies with FFF 3D printing process.