Roll-to-roll double side electrode processing for the development of pre-lithiated 80 F lithium-ion capacitor prototypes

Lithium-ion capacitors (LICs) show promise to help lithium-ion batteries (LIBs) and electrical double layer capacitors (EDLCs) in giving response to those applications that require an energy storage solution. However, pre-lithiation is a major challenge that needs to be tackled in order to develop efficient and long-lasting LIBs and LICs. In this work, we report for the first time the scale-up and utilization of sacrificial salts (dilithium squarate, Li2C4O4) as a pre-lithiation strategy in a LIC prototype fabricated in a pilot line. The synthesis of the sacrificial salt is scaled-up to produce 1 kg and is later incorporated in the positive electrode during the slurry formulation. After in-depth process optimization, 12 meter of a double side electrode are fabricated, achieving a high mass loading of 5.5 mg cm−2 for the HC negative electrode, and 14 mg cm−2 for the positive electrode accounting both the activated carbon and the dilithium squarate. On account of the satisfactory mechanical and electrochemical behaviour of the electrodes, multilayer pouch cell LIC prototypes are fabricated reaching 80 F each. Pre-lithiation is completed during the first ten cycles and after the required gas exhaustion, electrochemical performance of prototypes is also satisfactory. Moreover, fabricated pouch cells overcome a float test of 1600 h at 50 °C showing a capacitance retention of 84.3%. These results give clear evidence for the potential use of this strategy in real products and can foster research in the field to promote pre-lithiation by means of sacrificial salts as the final solution to the pre-lithiation step, both for LIBs and LICs.


Introduction
The large amount of greenhouse gas emissions, originated by a modern society relying on an intensive fossil fuel based industrial activity, has accelerated the climate change in the last decades and aggravated the negative impact derived from it.A shift in the energy paradigm is necessary at personal, local, national, and global level to meet the Net Zero Emissions by 2050 Scenario in order to stem the tide and mitigate the impact of global warming.Among many, two major challenges are envisaged in such a normative: (i) a transition from non-renewable energy sources to renewables, and (ii) the electrification of transport.The former is hampered by the discontinuity of the energy sources (e.g.sunlight and air, two of the main expanding renewable sources), and grid stability issues, while the latter is even a more complex case, combining low battery ranges, lack of infrastructure network, administrative obstacles and/or the origin of the energy that powers the electric vehicles.Two major challenges for which, different suitable high-end electrochemical energy storage systems will become key enabling technologies since both applications require high-energy, high-power, and long cycle life.In the case of electromobility, these features need to additionally be combined with excellent gravimetric and volumetric performance to minimize weight and size.Most of all, they need to be competitive in costs.
Beyond market stablished lithium-ion batteries (LIBs) and electrical double layer capacitors (EDLCs), lithium-ion capacitors (LICs) show promise to contribute on the transition from fossil fuels to renewables and transport electrification.LICs have already been proven in several applications in photovoltaics, windmills, or electric transport, where high-energy delivered simultaneously at high-power can offer the best cost of ownership solution for grid stability or heavy-duty machinery among many other applications.Unfortunately, despite being a commercially available solution, LICs still have some barriers to overcome towards market uptake.Similar to batteries, LICs utilize a carbonaceous negative electrode such as graphite or hard carbon (HC), but contrary to batteries, the positive electrode is based on high specific surface area carbons, where charge storage is based on a capacitive mechanism instead of a faradaic process, showing excellent sustainability but leaving the electrolyte as the only source of lithium ions in the system.Hence, an extra lithium source is necessary to compensate the loss of ions during the solid electrolyte interphase (SEI) formation and the irreversible lithium-ion insertion without starving the electrolyte.Thus, pre-lithiation of LICs, is nowadays considered one of the most important challenges of the technology in order to develop a cost competitive solution.Along the last decades, different pre-lithiation strategies have been studied [1][2][3][4], starting by utilizing an auxiliary lithium metal electrode [5], increasing the concentration of electrolytes [6], incorporating stabilized lithium metal powder (SLMP) in the anode [7,8], lithium metal stripes [9], Li 3 N [10,11], or a large variety of oxides and sacrificial salts [12][13][14].However, most of these studies lack of a scalability assessment, and hence, even that they might have shown outstanding solutions, technology transfer is not often possible.Among them, the use of lithium metal stripes and SLMP were an exception, and efforts were done in order to implement those solution in prototypes.
Zheng et al reported the incorporation of lithium metal stripes into the surface of 46 × 46 mm HC negative electrodes.Pre-lithiation of HC was achieved in that case by soaking it in a certain electrolyte for 18 h.Shellikeri et al also analysed the difference on utilizing lithium metal stripes (45 µm thickness) or SLMP as pre-lithiation agent in both 49 × 49 mm in size HC and graphite electrodes.Similar to the previous work, the insertion of lithium into the structure of the carbonaceous anode was induced by immersing the electrodes in an electrolyte for a certain time.In this case, owing to the different kinetics of both materials, graphite required larger time to reach the same lithiation stage.Results showed that SLMP enabled to reach higher lithiation degree in the first 10 min, whereas on the following hours, the lithiation kinetic was slowed down.Thus, the potential use of both strategies in both materials was shown, concluding that depending on the desired lithiation degree, lithium loading and soaking time need to be appropriately determined.However, long pre-lithiation times are needed, and the use of metallic lithium requires of strict safety measures [15].Yan et al investigated the use of ultra-thin (i.e.15-20 µm) lithium metal films incorporated in the negative electrode of LIC pouch cells.Authors claimed that using those thin lithium films, LICs' capacitance could be increased by a 3.8% and the equivalent-series-resistance be reduced in a 16% [16].
As previously described, main works followed at pouch cell level were focused on utilizing lithium metal as lithium source.However, the use of an air sensitive material requires manipulation in a dry room or a glove box, which significantly increases the cost of the technology.Even though, the frontrunner LIC fabricated by Musashi Energy Solutions Co., as far as it is known, also relays in metallic lithium as a pre-lithiation agent [17].Nonetheless, an air-stable solution is desirable, as could be the use of some oxides or sacrificial salts in the positive electrode, since it will make the final product more competitive in terms of cost.Previously reported by our group [18][19][20][21], the use of dilithium squarate (Li 2 C 4 O 4 ) as sacrificial salt has shown potential to be a scalable and industrializable solution for high performing LICs.In this alternative strategy, the incorporation of the sacrificial salt in the positive electrode is followed during the slurry preparation, which does not necessarily alter the manufacturing process.The potential of such approach surpassed technology borders and attracted the attention of the LIB community as well.In the recent years, the use of silicon-based anodes for the development of higher energy density LIBs has raised the interest of a pre-lithiation solution.As for LIC technology, several different pre-lithiation strategies for LIBs were investigated [22], however, the use of sacrificial salts did not kept much attention, despite some works were published [23].Yet, recently, Gomez-Martin et al reported about the challenges and opportunities of the same Li 2 C 4 O 4 sacrificial salt as pre-lithiation additive for the positive electrode in NMC622||silicon/graphite LIB, showing the high potential of dilithium squarate for next generation LIBs [24].These demonstrates that pre-lithiation strategies followed in LIC technology can be transversal to other technologies, if properly tailored from one technology to the other [25].
Thus, this work focuses on the technological scalability assessment and the potential real use of dilithium squarate based pre-lithiation strategy.To this aim, the investigation is focused on developing multilayer pouch cell prototypes fabricated by using tools that are compatible with industrial processes.Mixing, coating, and drying parameters of two different materials such as a HC, and an activated carbon (AC) together with Li 2 C 4 O 4 have been investigated and processed using apparatus and quantities adapted for pilot line fabrication.Double side roll-to-roll (R2R) electrodes with high mass loadings (i.e.5-15 mg cm −2 ) were fabricated.Best manufacturing parameters were determined by the physicochemical and mechanical analysis of the prepared electrodes.Those electrodes were used to assemble 20.25 cm 2 electrode size 80 F multilayer LICs.Electrochemical evaluation includes standard charge/discharge profiles but parameters such as self-discharge, floating at elevated temperature (50 • C), or internal resistance together with energy density and power densities were also evaluated, obtaining a satisfactory performance.
All the electrode slurries were prepared using a disperser-homogenizer (Dissolver DISPERMAT, VMA Getzman) in N-Methyl-2-pyrrolidone (NMP) solvent.The viscosity of all slurries was measured by a rheometer (Haake Series 1 Rheometer) employing a 35 mm diameter plate and 20 mm cone measuring geometry.The coating process was performed in a R2R coating line from B&W Megtec.The adhesion and cohesion of the obtained electrodes was studied by 90 • peel test utilizing a dynamometer from Instron.Their microstructure was analysed by scanning electron microscopy (SEM) and energy dispersive spectroscopy (EDX) utilizing a FEI Quanta 200 field-emission gun SEM.
Pouch cells of were assembled using 2 single side and 3 double side negative electrodes with a mass loading of 5.5 mg cm −2 for each HC layer, and 4 double side positive electrodes with a mass loading of 9.8 mg cm −2 AC and 4.9 mg cm −2 Li 2 C 4 O 4 in each layer.The mass ratio between positive and negative electrode is of 1.8 (AC:HC) in order to appropriately compensate the higher capacity of the negative electrode respect the positive one.Moreover, the ratio between sacrificial salt and HC has being set in 0.9 (Li 2 C 4 O 4 :HC) to ensure the appropriate pre-lithiation degree.All electrodes were 20.25 cm 2 size and cells were assembled by the Z-holding technique with cellulose separator, accounting for a total area of 162 cm 2 .They were immersed in a certain amount of 1 M LiPF 6 (ethylene carbonate:dimethyl carbonate, 50:50 vol.%).
Electrochemical characterization was followed using Swagelok half-cell configuration (i.e.1.13 cm 2 electrodes) for the performance of negative and positive electrodes individually, while LICs were investigated as full cells in pouch format.

Slurry processing: HC and AC-Li 2 C 4 O 4
A R2R electrode manufacturing process starts by selecting the best mixing technique and tool for the slurry processing [26].At lab-scale, when few ml (i.e.2-5 ml) of slurry are processed, magnetic stirrer or rotor-stator mixing tools are most utilized.However, when scaling to higher volumes (>30 ml), a tank and an impeller that conformed a disperser are preferred.Dispersers are used when fine solids need to be integrated in a fluid.The dispersing process between materials and solvent involves several steps.Ideally, first, particles are wet by the solvent, second, the mechanical mixing breaks them (agglomerates or aggregates), leading to smaller particles, and finally, these particles are stabilized avoiding renewed association (flocculation).Thus, in order to achieve a good dispersion, the selection of a proper vessel and impeller, as well as the density of the material, the solid content, and the viscosity of the slurry should be taken into deep consideration.For safety reasons, the tank should be 30%-60% bigger than the mixing volume (figure 1(a)), while a different shape of impellers could be selected depending on the shear rate of the fluid.When a low shear-rate mixture is processed, lightweight disks could be used (figure 1(b)), while heavy duty dissolver disks (figure 1(c)) should be selected for high shear-rate mixtures.The height where the impeller is positioned is also of high importance to ensure a good dispersion.Figure 1(d) describes the parameters that should be considered.Moreover, a cooling jacket is also needed to maintain the temperature constant along the mixing procedure (see figure 1(a)).
During the slurry preparation of the HC, a 1 L jar with a lightweight disk was selected.Parameters identified in figure 1(d) were followed and its correct position was determined.Meanwhile, for the AC-Li 2 C 4 O 4 , 1 L jar was also utilized with a heavy-duty disk.In this case, a different disk was selected as the slurry processed was more viscous.This higher viscosity is related to the lower apparent density of the AC (i.e.0.6 g cm −3 AC vs. 1.2 g cm −3 HC), which is a microporous material with high specific surface area (see table S1).Thus, in order to achieve high density electrodes, viscosity of the AC slurries should be kept higher than that of the HC.
The order of mixing the materials is also of utmost importance.Some authors have studied the influence of varying the steps along the mixing procedure.Lee et al proposed two different procedures for the preparation of a LiCoO 2 slurry.The main difference was based on the incorporation of all the solvent (i.e.NMP) from the beginning or introducing 2/5 of NMP together with the binder (step 1), and then the rest together with the conductive carbon and LiCoO 2 (step 2).The multi-step process shows lower viscosity than the one-step process.This is the result of the break-down of the network structure taking place in the initial period, what also conducts to a better distribution of the LiCoO 2 and conductive carbon along the slurry, leading to lower electrode polarization, better rate capability and cyclability [27].Grießl et al also investigated the influence of different mixing protocols on a graphite anode.In this case, it is an aqueous-based processing where the influence of kneading concentration, pre-dispersion of conductive carbon, mixing time and mixer type are investigated.The main result analysed was the pore size distribution of the fabricated electrodes.Smaller pore size in the electrode structure was the responsible of the higher ionic diffusion resistance, while a broadening of the distribution lead to a higher packing density and higher contact area between anode coating and current collector, resulting in lower interface resistance [28].However, generally, the most common and systematic approach is followed by: (i) preparation of a binder solution with a pre-defined binder concentration, (ii) incorporation of the conductive carbon, and (iii) addition of the active material.Each step is proceeded by a determined mixing velocity and time.This protocol ensures an appropriate dispersion of all, inactive and active materials along the slurry.
Considering all the above-mentioned, both HC and AC-Li 2 C 4 O 4 slurries were processed following the general protocol (see figure 1(e)).For the HC, first a 3.5 wt.% PVdF-NMP solution was prepared, while for the AC-Li 2 C 4 O 4 , a 2.45 wt.% concentration was determined to be the most appropriate one in order to avoid introducing more solvent in the coming steps.As second step, the conductive additive was introduced in the PVdF-NMP dissolution.In the case of the positive electrode, the conductive carbon introduced in the slurry also acts as the carbon coating of the sacrificial salt.Hence, a dry pre-mixing step was followed for the C65 and Li 2 C 4 O 4 in a vibratory ball-milling.This way, not only the carbon coating but also the obtention of fine particles was ensured.After this second step, each respective active material was slowly introduced in each solution.
Another important point during the slurry preparation is focused on prevention of the formation of agglomerates.This is directly related to the nature (particle size, wettability, surface tension) of the materials.It is important to ensure that big agglomerates are not introduced from the beginning, as high energy would be required to disintegrate them.At this point, mixing speed and time play a crucial role.Both parameters, but mainly time, will be directly dependant on the solid-liquid content utilized.In the case of the HC negative electrode, a total mass of 250 g solid content was employed with a 42% of solid and a 58% of liquid content in the slurry fabrication.Meanwhile, in the positive electrode, a total mass of 180 g solid content was employed with a 34% of solid and a 66% of liquid content.The formulations utilized are 90:5:5 (HC:C65:PVdF) for the negative electrode and 85-x:x:4.5:5(AC:x = Li 2 C 4 O 4 :C65:PVdF) for the positive electrode, all in weight percentages.The amount of utilized sacrificial salt (i.e.x wt.%) should be defined according to the pre-lithiation degree required by the anode.The binder solution is preferably prepared at low mixing rates in order to ensure that the polymer chains are not modified during the mixing process, while the time will be determined by the total dispersion of the polymeric material.In the case of PVdF, being a highly non-reactive thermoplastic fluoropolymer, is possible to apply a speed of 700 rpm.However, with the solid amounts utilized, long times of 90-120 min are required to ensure its correct dispersion along the NMP solvent.Next, the conductive carbon is introduced, and the mixing velocity is increased, as higher shear-rate will be necessary for a homogeneous dispersion [29].However, it is of high importance to control and analyse the velocity-time parameters.Conductive carbons are normally on the size of nanometres and agglomerates of the same size could be generated [30].However, if high-speed and during more than the necessary time is applied to that dispersion, these small agglomerates can be disintegrated into small aggregates or primary particles, and later, owing to their cohesive forces, re-agglomerate in even higher clusters [31].Hence, the optimum velocity-time needs to be find in this step.In the case of the bare conductive carbon introduced in the negative electrode slurry, 900 rpm and 90 min were determined to ensure an appropriate dispersion along the PVdF-NMP solution, while for the C65-Li 2 C 4 O 4 , the velocity was slightly reduced in order to avoid overpassing on the reduction to small aggregates, that later might re-agglomerate.In the final step, the active material, either the HC or the AC, is introduced and the applied speed is increased, as higher shear-rate would be necessary.The viscosity at this point needs to be slightly adapted by adding some more solvent if required.Once the mixing process is concluded, a degassing step in a roller is followed overnight (i.e. 15 h) in order to remove the gas bubbles created during the strong mixing.This degassing step might be detrimental at the point that if those above-mentioned nanoparticles are present in the slurry, the cohesive forces between them will tend to re-agglomerate the particles.In figure S1 a clear example of a non-optimized slurry process (figure S1(a)) and the already optimized one (figure S1(b)) can be observed.The evaluation of these agglomerates can be followed in the laboratory by using a grind meter.Figure 2

Coating and drying of HC-based electrodes (single side and double side)
Once the slurry preparation procedure is optimized, next step is to optimize the coating and drying parameters.When working at lab-scale with few ml of slurry, the most common approach is to laminate the slurry into the current collector-that is fixed by vacuum-by using a blade, and the wet thickness and the coating speed are determined.Afterwards, this laminate is dried during some hours at a certain temperature (sometimes even under vacuum conditions).It is worth to highlight that in this work, in addition to the slurry up-scaling process, the coating of large electrodes has also been transferred to a pilot-line (pre-commercial production line) for R2R fabrication of the electrodes.Herein, coating, and drying parameters must be defined precisely to obtain high-quality electrodes.The major difference is related to the coating method.As shown in figure 3, and as described by its name, a blade is utilized at lab-scale when coating the slurry by using the doctor blade (figure 3(a)).Meanwhile, in our R2R equipment, the slurry is dumped in a reservoir and slip by the comma bar in a first roll (i.e.backing roll) and then, transferred by the so called 'roll-to-roll' process to the current collector foil that will be already in the second roll (i.e.bump roll) (see figure 3(b)).At this point, the wet thickness and the web speed are the critical parameters that need to be fixed.The distance between the backing and the bump roll must also be well-adjusted to ensure that the slurry is well deposited in the current collector foil.Next important parameters are the temperature, the airflow, and the pressure of the dryers, which will be selected depending on the nature of the slurry (e.g.viscosity, solvent boiling point, …), but will be conditioned by the number and the length of the dryers in the equipment.As shown in figure 3(c), ovens are positioned one after the other, and each of them is allowed to have a certain temperature, airflow, and pressure.These three parameters are dependent between them.The airflow and pressure will determine how is the hot air moving along the oven.Fast airflows are not desired as they might cause a turbulent flow into the coating and disorder the solid content by its density.Meanwhile, it is also important to keep a moderate temperature gradient among ovens, as a large temperature gradient, mainly in the first oven, can cause a detrimental binder migration which later results in bad adhesion and cohesion among particles [32].Figure 3(d) explains the solvent evaporation process that takes place inside the ovens.There have been several studies on LIB manufacturing investigating different approaches to find the most appropriate drying technique.Von Horstig et al evaluated different drying technologies for simultaneous double-side coating [33].Classified by the higher to the lower technological readiness level, convection, infrared, laser, microwave, and joule heating technologies were proposed.Apart from the convection technique, where the heated air is an indirect drying medium, other technologies directly introduce the energy into the electrode, increasing the efficiency of used energy, and hence, reducing the energy demand.However, still the convection method is the most common drying method utilized during the electrode fabrication process.Several experiments were performed to determine the best coating and drying conditions for the HC-based negative electrode.The selection of the parameters was ad-hoc adjusted to the specifications of our pilot line, such as the number and the length of the ovens, that determine the drying time of the coating.Hence, different web speeds, temperatures, and airflows have been studied mainly for the first oven (see table S2), while keeping those parameters constant for the second one.The quality and mechanical stability of the electrodes was evaluated by 90 • adhesion tests and cross-section SEM images.First experiments were followed to determine the best web speed.As shown in figure S2(  adhesion.Thus, these first experiments allowed to set the web speed in 0.5 m min −1 , the first oven temperature in 80 • C, and the airflow at 8 m s −1 .Though, with the aim of trying to reduce the energy consumption, a lower temperature (i.e.70 • C) was also investigated (see figure S3).However, as observed in the results from the adhesion test in figure S3(a    and 60 mAh g −1 at 10 C are obtained.These gravimetric capacity values are slightly lower than some of the values reported in literature.However, to be comparable with few milligrams' electrodes, areal capacity provides a more significant metric.Figure S5(b) shows the comparison of the areal capacity values of the electrodes with the mass loading in the range of 3-5 mg cm −2 .
Once both coating and drying parameters of the HC-based single side electrodes were defined, double side electrodes were fabricated (see figure 5). Figure 5

Coating and drying AC-Li 2 C 4 O 4 electrodes (single side and double side)
The manufacturing of the positive electrode represents a major challenge when referred to LIC technology.As previously mentioned, the high specific surface area and the microporosity of the AC make it more difficult to process compared to the HC.In addition, an extra ingredient is added (Li 2 C 4 O 4 ) to meet the pre-lithiation challenge of the system, which implies to develop very high loading and thick positive electrodes containing a combination of organic and inorganic materials for the first time.The viscosity had to be adapted to allow its R2R coating, and since a larger amount of solvent was found to be needed as well, new drying parameters had to be fixed.Figure S6 describes the results of the first attempt, showing the consequences of the unappropriated selection of the fabrication parameters, which were similar to those ones used for the HC-based electrode.The nature of the cracks observed in figure S6 could be ascribed to several things, namely rapid evaporation of the solvent, the lack of binder, or the bad tension of the current collectors and consequently the bending of it, among others.However, as in most of the cases, the manufacturing of a new material requires several loops in the search of the proper initial parameters.In the case of the AC-Li 2 C 4 O 4 -based electrodes, a higher amount of solvent needs to be evaporated, hence, larger drying time is required.To this aim, the web speed was reduced, thus, electrodes will spend longer time through the ovens.In this case, 0.2 m min −1 was set as the most appropriate web speed.As summarized in table S3, different experiments were followed varying the temperature of both ovens with the aim of analysing the effect of the temperature gradient among them.Figure 6 describes the appearance of the dried coating showing the importance of these parameters.In figures 6(a) and (b), T 2 was kept at 100 • C while T 1 was increased from 80 • C to 90 • C. In both cases, some cracks appear on the lateral of the coating.Those cracks are more significant when selecting T 1 = 90 • C than T 1 = 80 • C. The nature of these cracks is described by the rapid drying rate taking place in the first oven, which is confirmed by the experiments in figures 6(c) and (d).In both cases, where T 1 = 90 • C is selected as the first drying temperature, large cracks starting from the lateral and going to the centre of the coating are visible, what can be directly ascribed to the rapid evaporation of the solvent.This first drying step is of key importance to ensure the correct distribution of the particles along the coating.A rapid drying rate induces a rapid evaporation of the solvent, which at the same time, causes the migration of the binder to the surface of the electrode, affecting the adhesion and cohesion properties of the coating [34].Meanwhile, the temperature of the second oven is not that critical, but its optimization is also crucial to ensure the total solvent evaporation after the drying step.Hence, in this case, with a slurry with previously mentioned properties, a T 2 = 120 • C is necessary.SEM images in figure 6 show the microstructure of each coating, while figure S7 reveals the binder distribution along the different electrodes by the EDX mapping of fluor .In concordance with the visual evaluation, on the one hand, those electrodes with T 1 = 90 • C show major F intensity, describing that the PVdF content in the surface is higher and that a rapid migration of the polymer has taken place.On the other hand, at T 1 = 80 • C the binder content in the surface is much softer, what means that the binder migration has not been that pronounced.Moreover, setting T 2 = 120 • C shows to migrate the polymer along all the surface, in comparison with T 2 = 100 • C. Thus, T 1 = 80 • C with 8 m s −1 airflow combined with T 2 = 100 • C with 12 m s −1 seems to be the most appropriate temperature and airflows for the developed positive electrode.Adhesion tests of this last positive electrode was performed (see figure S8).Contrary to the HC-based electrodes, due to the nature of the AC-Li 2 C 4 O 4 electrodes and the required high loading, obtaining an excellent adhesion and cohesion is much more challenging.However, it was possible to fabricate robust R2R AC-Li 2 C 4 O 4 electrodes.
Similar to the HC-based electrode, in the AC-Li 2 C 4 O 4 -based electrode, once the coating and drying parameters of single side electrodes were selected, those parameters were also utilized for the double side coatings.Figure 7  Finally, the final step during the electrode manufacturing process is the calendering step.The calendering step is not trivial since it is in charge of ensuring a good contact among particles which enhances the electronic conductivity, adapt an adequate tortuosity improving lithium-ion transport in pores, and increase the density of the electrodes reducing the charge transfer resistance [35].Those are some of the parameters that determine the final electrochemical response of the electrodes as well as their mechanical strength.Thus, calendering is a step that can spoil all the good job done previously.Electrodes are usually calendered at a certain temperature and velocity in order to achieve the targeted properties [36,37].In this work, the HC-based electrodes were calendered at 80 • C and fixed to reduce their thickness in a 30%, while AC-Li 2 C 4 O 4 -based electrodes were calendered only to reduce the 5% of their thickness.In capacitive-type electrodes the calendaring should only ensure the correct cohesion among particles and uniformity.Special care should be taken to prevent pore blocking, structure collapse and resistance increase [38].

Multilayer LIC pouch cell assembly
The R2R processing of double side HC negative electrodes and AC-Li 2 C 4 O 4 positive electrodes, make possible the fabrication of multilayer LICs utilizing a sacrificial salt as extra lithium agent.Two single side HC electrodes in the terminals were combined with four double side AC-Li 2 C 4 O 4 and three double side HC electrodes to build a 80 F multilayer LIC prototype.Figure 8 describes the electrochemical response of the cell voltage in a standard voltage window of 2.2-3.8V applied to commercial cells, together with the potential swing of the electrodes at the applied voltage vs. Li + /Li. Figure 8(a) shows the impedance measurement after the resting/wetting hours at open circuit voltage, together with those registered after some cycles at different current densities, all of them measured at the discharge voltage (i.e.2.2 V).The semicircle presented after the pre-lithiation step describes the formation of the SEI, which represents an additional charge transfer resistance that it is kept along cycling.Figure 8(b) describes the first charge step of the LIC up to the upper voltage of 4.1 V to complete the pre-lithiation step, where in the positive electrode, the sacrificial salt is reversibly oxidized (see inset reaction), and those lithium ions are inserted in the negative electrode, being part of the SEI formation agents.The potential of 60 mV vs. Li + /Li achieved by the negative electrode shows an appropriate pre-lithiation degree of the device.However, in figure 8(c), the positive electrode shows a non-linear profile in the next pre-conditioning cycles, what reveals that still some salt is being oxidized in the second and third cycles.Nonetheless, in the subsequent cycles, the triangular shape of a capacitive electrode is recovered, showing that all the salt has been successfully oxidized.During this oxidation process, as it has been reported before [19,20], some voids are left in the positive electrode, which enhance ions diffusion through the electrode and improve its electrochemical response.At the same time, the negative HC electrode shows to work in a very stable potential swing between 500 and 100 mV vs. Li + /Li, meaning that a good and stable SEI has been formed.Before continuing the characterization of the cell, a degassing step must be followed in order to eliminate the gas generated along the oxidation of the sacrificial salt (see figure S10).Afterwards, the multilayer LIC has been characterized between 2.2 and 3.8 V at different current densities.Figures 8(d)-(f) show the LIC cell profile at different current densities, going from low (i.e.20 min) to high (i.e.<1 min) discharge times.As observed, with the increase of the applied current density, the internal resistance, described by the ohmnic drop, also increases, therefore, there is still room for improvement both in the electrode and the cell configuration.One the one hand, part of the resistance might be the consequence of non-optimum contact among components.On the other hand, the electrode configuration might need some more fine-tuning (e.g.formulation, processing parameters, porosity, tortuosity…).However, the scope of the work is to demonstrate the scalability of the pre-lithiation strategy, while the system undergoes a policy of continuous improvement.
Figure 9 shows capacitance, energy, and power values obtained from these galvanostatic charge/discharge measurements, as well as the self-discharge of the system, together with a floating test carried out at 50 • C. Figure 9(a) compares the capacitance of a monolayer with a multilayer pouch cell both cycled in the cell voltage of 2.2-3.8V.As expected, the capacitance among the different discharge time is increased in proportion to the increased number of electrodes, allowing to develop a prototype that reaches 80 F at low discharge times.However, while discharge time decreases, capacitance also decreases, hence, also the energy (see figure 9(b)).Thus, 26 Wh kg −1 electrode energy density was obtained at 69 W kg −1 electrode power density, and 14 Wh kg −1 electrode at 1000 W kg −1 electrode .Moreover, owing to the better compacting of the electrodes in the multilayer configuration when applying vacuum for sealing, a lower internal resistance is achieved and hence, higher energy output is obtained at higher power in comparison with the monolayer LIC pouch cell (see figure 9(b)).Figure 9(c) describes the self-discharge of a multilayer LIC pouch cell prototype at room temperature compared to a commercial cylindrical LIC cell from LICAP (200 F), measured following the IEC 62576:2018 standards.For the commercial cell only the voltage of the cell is monitored.However, since our LIC prototype contains a lithium reference to monitor the potential of both positive and negative electrodes operando, the swing of those electrodes is also shown.As it can be observed, both cells were charged up to 3.8 V.While the commercial cell keeps its voltage almost constant in 3.75 V, the prototype developed in this work decays to 3.5 V after 150 h.Analysing the graph, it can be observed that the self-discharge of the full cell mimics the self-discharge profile of the positive electrode, as it has been also confirmed by other authors [39].The later can be the result of different processes, charge leakage, charge loss by electrolyte-electrode reactions, or redistribution of charge, however, recently using in-situ nuclear magnetic resonance (NMR) spectroscopy, Hess et al. ascribed it mainly to the diffusion of ions and impurities at the surface of the electrodes [40].Instead, the self-discharge of faradaic materials is normally a slower process, loosing <5% of the stored capacity over 1 month.However, it has been also demonstrated that depending to the characterization tests it has been performed, the self-discharge rate can be increased [41].Thus, special attention should still be paid during positive electrode processing, especially in those steps related to the purity of materials (i.e.avoid large amount of functional groups that might generate parasitic reactions) or drying (i.e.avoiding humidity) that might affect the overall purity of the electrodes.To conclude the electrochemical evaluation of the system, figure 9(d) shows a floating test carried out at 50 • C. The results show that when this prototype was maintained at constant 3.8 V for 1600 h, the capacitance decay was only 16%.Overall, the sum of the results shown by the electrochemical evaluation reveals the system might still need some fine-tuning adjustment for best performance, but more importantly, it demonstrates that the sacrificial salt strategy is a valid technological approach to overcome the pre-lithiation challenge.

Conclusions
This work shows the challenges behind the development of an 80 F multilayer LIC pouch cell and the way to solve them.On the one hand, the mixing procedure, slurry properties and coating/drying parameters have been investigated.It has been demonstrated that coating and drying parameters are of key importance to obtain high quality electrodes.Moreover, it has been shown, by utilizing two different materials such as a HC and AC combined with a sacrificial salt, that the nature of the material forces to tune all the process for each material.To the best of our knowledge, it is the first time where the integration of a sacrificial salt has been done at pilot line level for the development of a positive electrode with industrial standards.Thus, this work demonstrates the compatibility of this pre-lithiation strategy with an industrial fabrication process, not only for the manufacture of LICs, but also for LIBs.Furthermore, the R2R developed double side electrodes have been validated by fabricating 80 F multilayer LIC pouch cells.The results, in terms of energy-to-power values, together with the self-discharge and float measurement at 50 • C during 1600 h, prove the competitiveness of the developed technology.

Figure 1 .
Figure 1.(a) Different size tanks, (b) lightweight disk, (c) heavy duty disk, (d) equations to determine the best position of the disk in the tank (D = container diameter, d = dissolver disk diameter, a = distance of the dissolver disk to the container bottom, f = tank volume), (e) Slurry process of the negative and positive electrodes (x = Li2C4O4 wt.%).
(a) describing the HC-based slurry, shows no presence of agglomerates while figure 2(b), describing the AC-Li 2 C 4 O 4 , shows that still some agglomerates of 30 µm are present in the optimized slurry.
a), the adhesion test reveals that the highest speed applied (i.e.0.75 m min −1 ) shows the most irregular and lower force required along all the peel distance, what is confirmed by eye in figure S2(b).Meanwhile, a web speed of 0.5 m min −1 with an airflow of 8 m s −1 , shows the best adhesion properties.Both results are described by cross-sectional images.Figure S2(c) shows poor contact of HC particles with the copper current collector, while figure S2(d) shows better

Figure 3 .
Figure 3. (a) Lab-scale coating method by doctor blade, (b) comma bar coating method at pilot line, (c) drying line, and (d) solvent evaporation procedure in the drying step.
) and in images in figure S3(b), setting the temperature of the first oven at 70 • C is not high enough to completely evaporate all the solvent, and adjusting the second one to 100 • C shows a high migration of the binder to the surface of the HC-based electrode, what is the origin of the bad adhesion observed.SEM images in figure S3(c) confirm those results.The selection of the airflow is also of high importance, as if a high value is preset, the turbulence created inside the microstructure of the electrode might re-agglomerate those smallest particles (i.e.C65) in big aggregates, as shown in figures S3(d) and (e).

Figure 4
describes the adhesion properties of electrodes dried with a different first temperature.As observed, those dried at T 1 = 80 • C or T 1 = 90 • C show better adhesion, measured by the peel force required (see figure 4(a)), while by eye (figure S4), those dried at 80 • C clearly show the best adhesion properties, since most of the material remains in the electrode after the peel test.Moreover, SEM images (see figures 4(b) and (c)) confirm the high quality of the electrodes, showing good dispersion among all the components, good adhesion with the current collector and good cohesion among particles.Electrochemical characterization, in terms of rate-capability is described in figure S5(a).At low rates of C/10 (being C = 372 mAh g −1 ), a specific capacity of 180 mAh g −1 is delivered, while 100 mAh g −1 at 2 C
(a) shows an image of the wet slurry during the coating process before the electrode enters the oven, while figure 5(b) shows the electrode dried after it gets out the ovens.SEM cross-section image depicted in figure 5(c) show that both sides have similar thicknesses, particles are homogeneously distributed with very little agglomerates and good adhesion to the current collector, confirming the high quality of the fabricated double side HC-based electrode.The fabricated HC-based coating is 180 mm wide, 70 µm thick, and with a weight of 5.5 mg cm −2 .That coating was obtained by R2R processing, utilizing a 0.5 m min −1 web speed, wet thickness of 95 µm, T 1 = 80 • C with an airflow of 8 m s −1 , and a second oven with T 2 = 100 • C and 10 m s −1 .
(a) shows the deposition of the slurry while figure 7(b) shows the electrode after the slurry is dried.SEM cross-section image in figure 7(c) confirms the good quality of the fabricated double side electrodes by R2R processing, showing homogeneous distribution of the sacrificial salt along the electrode, no segregation of binder and conductive additive particles, and a good adhesion to the current collector.Moreover, the second side is the mirror image of the first.The obtained AC-Li 2 C 4 O 4 -based electrodes are of 180 mm wide, 150 µm thick, and 14 mg cm −2 .Specific capacity output at different current densities is summarized in figure S9.At low current densities the specific capacity output shows values close to 60 mAh g −1 , while at higher rates of 5 A g −1 where the discharge time is <10 s, it shows around 45 mAh g −1 .This capacity retention along different current densities might still be improved by further electrode engineering optimization.

Figure 8 .
Figure 8. Electrochemical characterization of multilayer LIC pouch cells: (a) EIS measurements, (b) first charge step in terms of specific capacity, (c) pre-conditioning cycles, and LIC prototype cell profile at different current densities: (d) 0.02 A g −1 , (e) 0.05 A g −1 , and (f) 0.10 A g −1 .

Figure 9 .
Figure 9. Electrochemical characterization of multilayer LIC pouch cells: (a) capacitance at different discharge times, comparison of monolayer and multilayer cells, (b) ragone plot, (c) self-discharge in comparison with a commercial cell, and (d) float test at 50 • C.

Table 1 .
Viscosity values of HC and AC-Li2C4O4 slurries at different shear-rates.However, after changing mixing disk, time, and speed, this is the best result obtained so far.The rheological behaviour of both slurries is described in figure2(c) and viscosity values at different shear-rates are summarized in table 1.As described, at low shear-rates the viscosity of the AC-Li 2 C 4 O 4 is higher, while at the shear-rate of interest regarding the R2R deposition (i.e. 10 s −1 ) both slurries show similar values.However, due to the lower density of AC-Li 2 C 4 O 4 composite, it is necessary to keep it slightly more viscous to ensure its appropriate coating and drying.