Effect of Uniaxial Stack Pressure on the Performance of Nanocrystalline Electrolytes and Electrode Composites for All‐Solid‐State Fluoride‐Ion Batteries

If all‐solid‐state fluoride‐ion batteries want to compete with existing battery technologies, significant improvements in terms of cyclic stability are necessary to fully access the high specific capacities, which this battery concept can provide in theory. Herein, the development of a high‐pressure, high‐temperature battery operation stand for battery cycling under inert conditions inside a glovebox is reported. This stand is then used to investigate the effect of stack pressure on the cell performance of conversion‐based as well as intercalation‐based electrode materials for fluoride‐ion batteries. It is found that cyclic stability as well as energy efficiency is strongly increased compared to nonpressure conditions, which is assigned to sustained interparticle contact. Thus, the cell design must be considered carefully to be able to distinguish intrinsic material properties from percolation‐ and interphase‐related impacts on the cell behavior. Further, the effect of pressure on the ionic conductivity of common solid fluoride‐ion conductors is investigated.


Introduction
All-solid-state batteries (ASSBs) are currently discussed to increase the energy and/or power density and are expected to be one of the next-generation lithium-based battery technologies. [1]However, the cell concept does not come without problems, as volume changes occurring for the active electrode material can lead to significant capacity fading on extended cycling.This mainly originates from a loss of contact of the active materials to the carbon additive and the solid electrolyte.Various research attempts were made to limit capacity fading, among them, the advantageous effect of pressure on the performance of ASSBs has been demonstrated several times [2] and could be assigned to decreased contact resistance and reduction of volume change-induced contact loss between particles during cycling. [3] addition, the application of pressure can have an impact on the ionic conductivity of solid electrolytes since compression results in a lowering of the accessible volume for mobile ions.A representative example for the effect of crystal structure on the pressure dependence of ionic conductivity is (Li, Na, K)-β-Al 2 O 3 .While the conductivity of Li-β-Al 2 O 3 has a positive correlation with pressure, the mobility of larger Na þ ions in β-Al 2 O 3 is mostly indifferent to pressure changes.On the contrary, increasing the ionic radius as in K-β-Al 2 O 3 results in a negative correlation with pressure. [4]s one alternative to lithium-ion batteries, all-solid-state fluoride-ion batteries (ASS-FIBs) have been investigated with increasing number of reports in the past decade.These systems are not only of interest due to their high theoretical capacities (5000 Wh L À1 ), but also due to a wide potential window provided by fluoride conductors. [5]However, the required high operating temperature and capacity fading upon cycling are some of the main challenges in harvesting the full potential of ASS-FIBs. [6]part from few solid fluoride conductors (BaSnF 4 , [7] PbSnF 4 , [8] CsPb 0.9 K 0.1 F 2.9 [9]   ) which show room-temperature fluoride-ion conductivity in a range of 10 À3 to 10 À4 S cm À1 at the cost of reduced electrochemical stability, solid electrolytes with high electrochemical stability such as La 0.9 Ba 0.1 F 2.9 and Ba 0.6 La 0.4 F 2.6 used in rechargeable FIBs still require high operating temperatures of %150-170 °C. [5,10]p to this point, only little is known on the pressure dependence of conductivity for fluoride-ion conductors.One study reports the pressure dependence of the ionic conductivity of cubic and orthorhombic PbF 2 in detail [11] which is frequently used as a reference electrode for FIBs.Remarkably, no study on the effects of stack pressure on the cyclic stability of solid-state FIBs has been reported so far.This is likely related to the fact that cells suitable for ASSBs which can be operated at higher temperatures and higher pressures at the same time are challenging to be built.
In this article, we report the development of a cell setup which can be used to cycle ASS-FIBs at extended pressure up to 2 GPa and temperature up to 200 °C.This cell was then used for a detailed investigation of the conductivity of four common solid fluoride-ion conductors (Ba 0.6 La 0.4 F 2.4 , La 0.9 Ba 0.1 F 2.9 , BaSnF 4 , and PbSnF 4 ) under pressure, with an analysis of the activation volume of the electrolytes at their respective operation temperatures of 170 and 30 °C.In addition, we show that the application of pressure has a strong impact on the cyclic stability of conversion-and intercalation-based FIBs, which allowed us to operate the cells over 15 months.

Construction of a Hot Press for Solid-State Battery Operation
For operation of ASSBs under pressure, a hot press was constructed.A rendering of the design is shown in Figure 1.The design includes space for a hydraulic cylinder to apply uniaxial force on the cell pellet.The cylinder presses onto a holder for disc springs which are used to minimize the force changes due to volume change of the pellet during cycling.This spring holder sits in a steel alignment part with a PTFE liner to electrically insulate the steel pins from the steel frame.Inside the alignment part, the steel pins are split to incorporate a 3 mm-thick alumina disc to allow for insulation in the axial direction as well.The current collector pins are exposed to allow connection of a potentiostat via crocodile clamps.While the entire frame is made of mild steel, the current collector pins are made from 1.2343 hot working steel (equivalent to AISI H11), to avoid plastic deformation up to temperature of 200 °C.Both current collector pins meet inside another steel mantle for alignment, which is lined by a PTFE tube for electrical insulation.Optionally, to increase the range of applicable pressure, the pins diameter can be reduced from 7.5 to 3.5 mm on the electrolyte contacting side.In this case, another alignment ring, made of PEEK with an inner diameter of 3.5 mm, is placed around the smaller-diameter tips of the current collector pins.PEEK was chosen over PTFE despite its lower chemical stability since it is much harder and avoids powder spills out at high pressures.This steel mantle is sized to fit a 25 mm-diameter heating band for high-temperature operation of the cells.Around the steel mantle, there is ample space to fit the heating mantle, a thermocouple, the crocodile clamps for electrical contact, and ceramic fiber insulation.On the bottom of the hot press, below a second alignment part with a PTFE liner and split pins for accommodation of an alumina disc, a force transducer is placed to monitor the applied force on the pellet.The output signal of the force transducer, a voltage which is proportional to the applied force, is amplified using a LM358N operational amplifier, read by an Arduino UNO and sent to a personal computer using serial communication where a python script is used to display and save the data.The amplification circuit can be found in the Supporting Information (Figure S1, Supporting Information), while the python script and the code to be run on the Arduino are also provided in the Supporting Information.
Currently, the cell developed was not airtight and thus required operation under inert condition for its use together with FIB materials because the presence of humidity and/or oxygen can possibly induce material degradations in FIBs.For this purpose, the entire device was placed inside an argon-filled glovebox and the cables for the potentiostat, heating mantle, thermocouple, and force transducer were routed in with a feed-through.

Synthesis
Barium-doped lanthanum fluoride (La 0.9 Ba 0.1 F 2.9 , 10% doping level, 9 g LaF 3 , 0.894 g BaF 2 ) and lanthanum-doped barium fluoride (Ba 0.6 La 0.4 F 2.4 , 40% doping level 5.8 g BaF 2 4.47 g LaF 3 ) were synthesized by high-energy ball milling of stoichiometric ratios of BaF 2 (Alfa Aesar, 99%) and LaF 3 (Thermofischer, 99.99%) in 50 mL zirconia jars with ten zirconia balls of 10 mm diameter for 12 h of effective milling time.The milling was conducted for 10 min intervals at 600 rpm and 20 min breaks to let the ballmilling jars cool down.The precursors were dried prior to milling for 12 h in vacuum at 200 °C.
La 2 Ni 0.75 Co 0.25 O 4.08 was prepared by solid-state synthesis.Stoichiometric amounts of La 2 O 3 (Alfa Aesar, 99.9%, predried at 1200 °C for 12 h), NiO (Sigma Aldrich, 99.99%, predried at 700 °C for 12 h), and Co 3 O 4 (Alfa Aesar, 99.99%) were ball milled for 1 h with the rotational speed of 600 rpm using a small amount of isopropanol as a dispersing agent in ZrO 2 jars.The mixture was placed in a crucible made of corundum and reacted at 1450 °C for 24 h under a flow of argon (99.9%, 0.4 standard liter per minute (SLM) flow), with heating/cooling rates of 2/1 °C min À1 respectively.For the preparation of the La 2 Ni 0.75 Co 0.25 O 4.08 cathode composite, 30 wt% of as-synthesized La 2 Ni 0.75 Co 0.25 O 4.08 was mixed with 60 wt% of La 0.9 Ba 0.1 F 2.9 and 10 wt% of dried carbon black to improve the ionic and electronic conductivity of the cathode material.The mixture was then ball milled for 3 h at a rotational speed of 250 rpm. [12]he Pb-PbF 2 -C anode composite was prepared by ball milling the components in weight ratios of 45 wt% Pb, 45 wt% PbF 2 , and 10 wt% carbon black with the same parameters and ball-milling containers and media as the La 0.9 Ba 0.1 F 2.9 solid electrolyte.
BaSnF 4 and PbSnF 4 were synthesized by mechanochemical milling stoichiometric ratios of BaF 2 (Sigma Aldrich 99.99%)/ PbF 2 (Alfa Aesar, 99% min) and SnF 2 (Sigma Aldrich, 99%) and postannealing.Milling was conducted in 50 mL zirconia vials with zirconia balls under argon atmosphere, and every 10 min milling interval was followed by a 20 min break.For BaSnF 4 , 5 mm zirconia balls and a ball-to-powder ratio of 17:1 were used.After milling at 600 rpm for 12 h effective time, the powder was annealed at 300 °C for 2 h in vacuum in an alumina crucible.The whole procedure was repeated three times to achieve high purity of the product.For PbSnF 4 , ten zirconia balls of 10 mm diameter were used during 12 h milling (370 rpm) for synthesizing 3 g of γ-PbSnF 4 .After milling, the powder was annealed at 200 °C in airtight stainless steel containers under argon for 5 h.
The BiF 3 -BaSnF 4 -CNF cathode composite was prepared by hand grinding the components in a respective weight ratio of 4:5:1, while the Sn-BaSnF 4 -CNF anode composite was prepared by ball milling tin powder and carbon nanofibers in a 5:1 weight ratio and then hand grinding this mixture in a 6:4 weight ratio with BaSnF 4 .

Cell Preparation
For preparation of the pellets for impedance analysis, 50 mg of the desired electrolyte powder were weighed out and placed directly in the 3.5 mm-diameter cavity formed by the PEEK alignment tube and a steel current collector pin.Before pressing with a hydraulic cylinder, a piece of carbon coated aluminum foil was placed on each side of the hand-compacted powder to ensure good contact between steel pins and electrolyte with the possibly uneven surfaces.The pellet was then compacted in the setup for subsequent pressure-dependent impedance analysis and the pellet was not extracted from the PEEK alignment tube before measurement.After the impedance measurement, the total length between both ends of the steel current collector pins was measured to determine the pellet thickness for 50 mg of each powder.Since the length of both pins was 43 AE 0.01 mm, the rounded total distance of 86 mm was subtracted to determine the thickness of the pellets, and the approximate pellet thickness is given in Figure 4.
For galvanostatic cycling, pins without the reduced diameter were used, resulting in a 7.5 mm-circular steel pin as the current collector on both electrodes of the cell.The cell was also assembled directly in the hot press setup.The chosen electrolyte was first filled in and compacted by hand.Afterward, the anode composite was hand compacted on one side of the electrolyte and the cathode composite on the other side.For the La 2 Ni 0.75 Co 0.25 O 4 -La 0.9 Ba 0.1 F 2.9 -C/La 0.9 Ba 0.1 F 2.9 /Pb-PbF 2 -C cell (referred to as La 2 Ni 0.75 Co 0.25 O 4 /La 0.9 Ba 0.1 F 2.9 /Pb-PbF 2 ), 5, 200, and 20 mg were used respectively, while the masses were 5, 200, and 10 mg for the BiF 3 -BaSnF 4 -CNF/BaSnF 4 /Sn-BaSnF 4 -CNF cell (referred to as BiF 3 /BaSnF 4 /Sn).Both cells were compacted at 450 MPa for 90 s before cycling at the chosen pressures.SEM image of the cross-section of one of the cell pellets after cycling under pressure and the thickness of the electrolyte, anode, and cathode layers can be found in Figure S11, Supporting Information, showing that the thickness of the electrolyte layer was sufficient to avoid short circuits even under operation at high pressures.For the current study, an optimization of the overall energy density was not targeted, but could be improved further by reducing the thickness of the electrolyte layer.

X-Ray Diffraction
For X-ray diffraction, all samples were measured on a Rigaku SmartLab Diffractometer.A Hypix-3000 detector and a Cu K α tube without a monochromator were used.The patterns were recorded in 2θ range of 10°-80°at a scan rate of 1.5°min À1 .Measurements were carried out in argon atmosphere using airtight sample holders.All patterns were analyzed via the Rietveld method with Bruker TOPAS software.The data corresponding to these measurements are discussed in more detail in Section 3.1.

Galvanostatic Cycling
Galvanostatic cycling was performed either using a Solartron 1400 CellTest System or a Biologic VSP 150 potentiostat.The BiF 3 /BaSnF 4 /Sn-cell was cycled at 100 °C under %180 MPa stack pressure with a current density of 35 μA cm À2 , and cutoff voltages of 0.05 and 1.2 V were set considering electrochemical stability of electrolyte material.Before cycling, the cell was heated to the desired temperature and held for at least 4 h to reach thermal equilibrium.During heating, a prestack pressure which is a bit lower than desired stack pressure was applied on the cell to minimize internal delamination due to thermal expansion.The La 2 Ni 0.75 Co 0.25 O 4.08 /La 0.9 Ba 0.1 F 2.9 /Pb-PbF 2 cell was cycled at 170 °C under %450 MPa with a current density of 19.2 and 9.6 μA cm À2 for charging and discharging, respectively.The cell was operated at a charging cutoff capacity of 40 mAh g À1 and discharged to 0 V or 40 mAh g À1 against Pb/PbF 2 .

Electrochemical Impedance Spectroscopy
Electrochemical impedance spectroscopy (EIS) was performed with a BioLogic VSP potentiostat between 1 MHz and 500 mHz.In order to investigate fluoride-ion conductivity under respective working temperature of the chosen electrolytes, Ba 0.6 La 0.4 F 2.4 , La 0.9 Ba 0.1 F 2.9 were heated up to 170 °C, while for BaSnF 4 and PbSnF 4 30 °C was used.In case of elevated temperature measurements, the temperature was equilibrated for at least 4 h before the measurements were started.For all the electrolytes investigated in this work, the pellet was compacted at 2 GPa at the desired temperatures which was then held to equilibrate for 4-5 min.By that, it was assumed that almost all the densification (including compaction, plastic deformation of particles, and elimination of macrovoids) but also elastic deformation of the pellet took place, and from the pressure release sweep one can evaluate the activation volume without the effects of further consolidation (refer to Figure S5-S8, Supporting Information).In addition, the densification and the plastic deformation of the surrounding PEEK alignment tube could be nicely observed as the application of a force of 20 kN (corresponding to 2 GPa) was followed by a drop in the order of 500 N.After the force was stabilized, it was assumed that no further densification or deformation took place on the investigation under lower forces subsequently.Nevertheless, a %2 min equilibration period after changing the applied force was applied before every impedance measurement.After pelletization, the first impedance spectrum was recorded and subsequently the force was reduced stepwise.The force-time profile for the measurements on La 0.9 Ba 0.1 F 2.9 is exemplarily shown in Figure 2. The force-time curves for all other measurements can be seen in the Supporting Information (Figure S2-S4, Supporting Information).

X-Ray Diffraction
The refined X-ray diffraction patterns of the synthesized four electrolytes are present in Figure 3.The phase purity of the electrolytes is confirmed by Rietveld analysis.All electrolytes but BaSnF 4 can be refined with a single phase.BaSnF 4 shows a slightly inhomogeneous distribution of composition according to Ba 1AEd Sn 1∓d F 4 for different particles, resulting in asymmetric peaks.This asymmetry cannot be fit with a single phase, but the distribution of phase compositions can be fit with multiple tetragonal phases with slightly different lattice parameters.It is assumed that this inhomogeneity of the barium and tin content of different grains originates from the comparably mild ball-milling conditions and low annealing temperatures.7a,8,10]

Pressure Dependence of Ionic Conductivity
The pressure dependence of the ionic conductivity at 170 °C for La 0.9 Ba 0.1 F 2.9 and Ba 0.6 La 0.4 F 2.4 and 30 °C for BaSnF 4 and PbSnF 4 is displayed in Figure 4. Since the bulk and grain boundary resistances can't be separated in the Nyquist plot, the total  resistance is used to calculate the ionic conductivity of the materials.The force-time curves as well as the Nyquist plots of all measurements can be found in Figure 2, and S2-S8, Supporting Information, respectively.A generally similar behavior is observed for all materials other than La 0.9 Ba 0.1 F 2.9 .Dropping from the highest pressure, the logarithm of ionic conductivity increases linearly until it reaches a maximum.Past the maximum to lower pressures, a nonlinear decrease in the ionic conductivity is observed.A specific optimized stack pressure for ionic conductivity, in the case of Ba 0.6 L 0.4 F 2.4 and BaSnF 4 around 250 MPa and for PbSnF 4 around 500 MPa, can be noticed.For La 0.9 Ba 0.1 F 2.9 , only the linear region is observed without a second regime at low pressures.
The increase in conductivity on releasing the pressure from the highest pressure shows a linear trend on a single-logarithmic plot, which is expected due to the activation volume for ionic conduction in ionic solids.The activation volume can be understood as the volume which a unit cell has to expand to allow an ion to jump between two neighboring sites.For positive activation volumes, applying external pressure gives a hindrance to the unit cell expansion and therefore to the jump, and hence the ionic conductivity drops at high pressures. [4,13]he drop of ionic conductivity at pressures below the maximum conductivity results from loss of contact at the pellet electrode interface and within the pellet itself.13b] It is reasonable to assume that there is no change in density of the pellet when the stack pressure is relieved in the linear regime.At pressures lower than the linear regime, some crack formation happens due to elastic recovery of the pellet (detailed discussion is presented in Section 3.3).The different rates of decrease in ionic conductivity for different electrolytes in the low-pressure regime are Table 1.Lattice parameters of La 0.9 Ba 0.1 F 2.9 , La 0.9 Ba 0.1 F 2.9 , BaSnF 4 , and PbSnF 4 .
La 0.9 Ba 0.1 F 2.9 P-3c1 Values are given as average and standard deviation from three measurements of the same material.(thickness ≈ 0.5 mm), and d) La 0.9 Ba 0.1 F 2.9 (thickness ≈ 0.035 mm).Pellet area for all samples is 9.62 mm 2 .13b] SEM images of the uncompacted powders (Figure S12, Supporting Information) show a particle size of 360 AE 104 nm for La 0.9 Ba 0.1 F 2.9 and 491 AE 153 nm for Ba 0.6 La 0.4 F 2.4 and 888 AE 246 nm for BaSnF 4 and 574 AE 164 nm for PbSnF 4 .Particle sizes have been measured on 40 particles for each image, not including big agglomerates.These measurements show that BaSnF 4 has a significantly bigger particle size than PbSnF 4 which could affect the different behavior in the nonlinear regime.The behavior of La 0.9 Ba 0.1 F 2.9 and Ba 0.6 La 0.4 F 2.4 will be addressed in more detail in Section 3.2.2.

Determination of the Activation Volume
The activation volume ΔV was determined from the slope of the logarithm of the ionic conductivity lnσ against pressure p in the linear high-pressure regime, [13b,14] according to Equation (1) with R = 8.314 J K À1 mol À1 and T being the measurement temperature.13b,14a] ΔV ¼ ÀRT ∂lnσ ∂P T The determined activation volumes are shown in Table S1, Supporting Information and graphically represented in Figure 5.
The difference of activation volume between La 0.9 Ba 0.1 F 2.9 and Ba 0.6 La 0.4 F 2.4 is most likely explained by the different conduction mechanisms and coordination of the mobile ions since both materials have similar cations within their lattices (see Figure S10, Supporting Information).Ba 0.6 La 0.4 F 2.4 crystallizes in a cubic fluorite structure, with partial occupation of the interstitial sites.Mori et al. described that since the tetrahedral site in Ba 0.6 La 0.4 F 2.4 is fully occupied, conduction takes place via jumps between the tetrahedral site and the partly unoccupied octahedral site. [15]Therefore, the bottleneck of the jump is the equilateral triangle formed by three metal cations in the cubic close packing of the fluoride structure. [15]La 0.9 Ba 0.1 F 2.9 has a tysonite structure [16] , in which the fluoride ions occupy three different sites, one of them also showing a tetrahedral coordination (F1).Among the possible jumps between those different sites, the F1-F1 pathway shows the lowest activation energy and therefore likely has the biggest contribution to the ionic conductivity. [17]he comparatively lower activation volume for Ba 0.6 La 0.4 F 2.4 indicates that the jump through a bottleneck in the fluorite structure does not require as large of a deformation of the unit cell as is required for ionic conduction by jumps between F1 sites in the tysonite structure of LaF 3 .This is also reflected by the difference in activation energies of the two materials, where Ba 0.6 La 0.4 F 2.4 shows a lower activation energy (0.62 eV [15] ) than La 0.9 Ba 0.1 F 2.9 (0.73 eV [10a] ).
Remarkably, BaSnF 4 and PbSnF 4 investigated in this work show a big difference in activation volume (refer to Figure 5), even though both materials are structurally similar and crystallize in a tetragonally distorted modification of the fluorite-type structure which originates from cation ordering.In this structure, the cations stack in Sn-Sn, Sn-X, X-X (X: Ba, Pb) layers along the c axis, with F À ions residing in the X-X and the Sn-X layers at F3 sites (immobile F ions) and F2 sites coexisting with vacancies, respectively (see Figure S9, Supporting Information).
In BaSnF 4 only a small fraction of F ions are present within the Sn-Sn layers (F1 sites) and Sn-Ba layers (F4 interstitial site, around 25% occupancy) due to the repulsive interaction between a lone pair of electrons from the Sn 2þ ions and F À ions, and those remaining anions experience exchange with F2 ions leading to a conduction process in the a-b plane. [18]On the contrary, in PbSnF 4 no indication of F À being allocated within the Sn-Sn layers was given by neutron diffraction studies.Instead, a small fraction of F À ions coexists with vacancies at the F4 interstitial site appearing within the Sn-Pb layer. [8,19]As a result, increased mobile anion density in the Sn-Pb layers improves ionic conductivity of PbSnF 4 up to 10 À3 S cm À1 at RT. Also here, the lower activation volume for BaSnF 4 in comparison to PbSnF 4 is in accordance with the reported activation energies of BaSnF 4 (0.34 eV [20] ) and PbSnF 4 .(0.43 eV [8] ).These findings are counterintuitive considering that Pb 2þ is a much more polarizable ion than Ba 2þ , [21] showing the importance of small structural differences on the overall conductivity.Since a smaller activation volume is desirable for a battery since it results in an electrolyte which shows high conductivity with more indifference to pressure, this result reveals that batteries based on BaSnF 4 could be operated at higher stack pressures than PbSnF 4 -based batteries, if other effects can be neglected.Therefore, additional positive impacts of stack pressure on operation of a full battery could be amplified, and the latter will be addressed in detail in Section 3.3 and 3.4.

Behavior in the Low-to-Intermediate Pressure Regime
While the linear drop of the logarithm of the ionic conductivity in the high-pressure regime is interpreted as an atomistic effect, the decrease of conductivity with further pressure decrease from %50-100 MPa to ambient pressure is a microstructural effect.Schneider et al. [13b] attributed this observation to crack opening Figure 5. Activation volume for La 0.9 Ba 0.1 F 2.9 , Ba 0.6 La 0.4 F 2.4 , PbSnF 4 , and BaSnF 4 , data used for calculation is measured using PEEK alignment tubes.
as well as contact loss between the current collector and the electrolyte.
To investigate if this is also of conceptual relevance for the materials investigated in this work, scanning electron microscopy (SEM) images were recorded on the fracture planes of the pellets after they were extracted from the measurement device.Since the PEEK tube showed plastic deformation at the place of the pellet resulting in the pellet indenting the PEEK tube, it was not possible to extract fully intact pellets.However, because the pellets broke in fracture planes at a shallow angle to the pellets surface, it was possible to record images on these fracture planes.
Figure 6a,e shows the photographs of fracture surface of Ba 0.6 La 0.4 F 2.4 and La 0.9 Ba 0.1 F 2.9 respectively, from which SEM images were recorded (Ba 0.6 La 0.4 F 2.4 in Figure 6b-d and La 0.9 Ba 0.1 F 2.9 in Figure 6f-h.For both materials, it was observed that no individual particles are visible on the fracture plane, and the surface looks like a continuous bulk solid.The particle sizes of 360 AE 104 nm for La 0.9 Ba 0.1 F 2.9 and 491 AE 153 nm for Ba 0.6 La 0.4 F 2.4 as indicated in Section 2.2 should be nicely resolved at this magnification.Only some loose powder particles are visible on the surface of both samples, with some indication of few micropores remaining after compaction at 2 GPa.The edge of the pellet, that is, the part of the pellet, which was pushed against the PEEK, is broken down into loose powder (Figure 6d,h).It is also visible, that the center of the pellets experienced significant discoloration from the original white powder, turning dark gray to black.This color change did not take place at the edge of the pellet, indicating that most of the major densification happened in the center of the pellet, while the edge received less pressure due to the deformation of the surrounding PEEK alignment tube.
The most striking difference between La 0.9 Ba 0.1 F 2.9 and Ba 0.6 La 0.4 F 2.4 is the number of cracks and degree of delamination, which can be seen in Figure 6b,c,f,g, as a scale-like feature on the fracture plane.Here, Ba 0.6 La 0.4 F 2.4 shows a higher number of those delaminations on the same area than La 0.9 Ba 0.1 F 2.9 .
Thus, an explanation for the difference in the low-pressure behavior of La 0.9 Ba 0.1 F 2.9 and Ba 0.6 La 0.4 F 2.4 could be as follows.For Ba 0.6 La 0.4 F 2.4 many voids or cracks could be found which are at least partially oriented perpendicular to the axis of force, which would have opened up when the applied stack pressure dropped below a certain threshold value.Since for La 0.9 Ba 0.1 F 2.9 , a much smaller number of those cracks were found, such a loss of conductivity due to crack formation is not to be expected, matching the observation from the impedance measurements.

Galvanostatic Cycling of Conversion-Based Cathode Materials Under Pressure
After a detailed investigation of the behavior of a selection of solid fluoride ion conductors, we can conclude that while stack pressure does lead to a deterioration of the ionic nature of the materials at high pressures up to 2 GPa, the change is not severe, can be compensated by small increases of the cycling temperature, and thus still allows usage of the shown materials in ASS-FIBs.Especially if the chosen stack pressure is limited to the range of nonlinear change of ionic conductivity of the prepared electrolytes, the cell might even benefit from an improved particle contact within the electrolyte, aside from the expected improvement of contact in the cathode composite itself.
To compare the influence of pressure to a nonpressurized condition, a cell which was based on BaSnF 4 as the electrolyte, a BiF 3 -CNF-BaSnF 4 composite as the cathode and a mixture of Sn-CNF-BaSnF 4 as the anode was cycled in a stainless-steel based Swagelok cell, as described elsewhere [22] (see Figure 7c)).Even though cells of this composition can also be operated at room temperature, it has been shown that operation at elevated temperatures helps to reduce overpotentials and therefore allows battery operation at a higher current density than chosen by Mohammad et al. [20] Hereby, it is possible to achieve higher cycle numbers than it would otherwise be possible.This cell shows strong capacity fading from its 1 st discharging capacity of about Figure 6.Fracture planes of Ba 0.6 La 0.4 F 2.4 and La 0.9 Ba 0.1 F 2.9 pellet after 2 GPa pelletization.a) Ba 0.6 La 0.4 F 2.4 on carbon tape of SEM sample holder and e) La 0.9 Ba 0.1 F 2.9 .b,c) SEM images of Ba 0.6 La 0.4 F 2.4 pellet center area and d) Ba 0.6 La 0.4 F 2.4 pellet on edge area.f,g) SEM images of La 0.9 Ba 0.1 F 2.9 pellet center area and h) La 0.9 Ba 0.1 F 2.9 pellet edge area.
203 mAh g À1 to below 40 mAh g À1 on the 10th cycle, and the cell only delivered around 45% energy efficiency during these cycles (see Figure 7d)).
To investigate the impact of pressure, a second cell was cycled at 100 °C and %180 MPa.The voltage-capacity plots as well as the trends of specific capacity and energy efficiency against the cycle number can be found in Figure 7a,b.In the first discharge, a specific capacity of about 222 mAh g À1 , 73% of the theoretical capacity of BiF 3 (302 mAh g À1 ) was delivered, which is remarkably close to the first discharge capacity of the cell with no external pressure applied.This indicates that introducing high stack pressure does not have a large advantageous effect on the first discharging capacity.This would imply that the electrode composites prepared by the ball-milling route are functional in the sense that they form a good percolation network with only a minor fraction of active particles being isolated from others.During the first 30 cycles, the discharge capacity of the cell fades down to around 110 mAh g À1 .From cycle 30 until the cell was stopped at 94 cycles due to a problem with the inert atmosphere in the glovebox, the cell behaved stable, and the approached discharge capacity decreased from 110 to 90 mAh g À1 slowly.The Coulombic efficiency (see Figure 7b) increased continuously and stabilized at around 99% after the 30th cycle.Moreover, it is observed that with high pressure, a higher energy efficiency can be delivered from the first cycle, and it stabilized at around 58% after 30 cycles.The higher energy density in comparison to the nonpressurized cell has two origins: a reduced difference between the charging and discharging plateau of 0.09 V as compared to around 0.2 V and a minor influence from the increased Coulombic efficiency.We also acknowledge that though it cannot be quantified precisely, it must be kept in mind that the pressure distribution in the cell is not homogenous (see discussion in Section 3.2.2),and we estimate that only about 75% of the area of the pellet benefits from the applied pressure.This can partly influence the capacity retention, and one can assume that higherpressure homogenization might lead to an even higher capacity retention.It needs to be mentioned that the second plateau at %1 V of the cell cycled at 180 MPa, while not appearing in the cell at ambient pressure, only affects the first five charge cycles.Therefore, it can be assumed to be irreversible, since it does not show in the discharge curves, and the effect on later cycles is likely to be minimal.We assume the origin of this second plateau to be the fluorination of steel current collector.The reason for the absence of the second plateau in the nonpressure cell is that as compared to the cell applied stack pressure, the surface contact between pellet and current collector is not nearly as perfect as it is with pressure, therefore, stopping major fluorination reactions on this interface.
To show the effect of stack pressure on the microstructure of a BiF 3 /BaSnF 4 /Sn cell upon cycling, we recorded cross-section SEM images at the region of the cathode-electrolyte interfaces of the cell cycled at 180 MPa and the one cycled at ambient pressure, which are shown in Figure 8.
It can clearly be seen from the comparison of Figure 8a-d that the cell cycled at 180 MPa was able to maintain the interparticle contact as well as the contact at the electrode-electrolyte interface much better than the cell cycled at ambient pressure.There are several cracks and delaminations visible in the cathode layer of the cell cycled at ambient pressure, which opened up mainly in the direction perpendicular to the three-layer structure of the pellet.Some of those delaminations also appear at the interface of electrode to electrolyte.Delaminations of this kind are not observed for the cell cycled at 180 MPa.Due to the better contact, the interface between electrolyte and electrode composite is much harder to discern.The absence of carbon fibers in the electrolyte has therefore been used to tell both layers apart.
This degradation of the electrode composite of the cell cycled at ambient pressure also undermines the EIS spectra recorded after cycling (Figure 9).Whereas for both cells, cycled under pressure or at ambient pressure, no semicircle associated with charge transfer resistances could be resolved, both cells show a slight indication of a second semicircle after cycling.Comparing these spectra, the cell cycled at ambient pressure shows a significantly increased impedance, while the cell cycled at 180 MPa only shows a very small change of the overall impedance.The electrolyte resistance shows no change for both cells after cycling, showing that the degradation of the cell performance can be limited to the electrode composites and is not affected by the electrolyte.
From the comparison to the nonpressurized cell, we conclude that the pressure has mainly two impacts on conversion-based cathode materials.1) Pressure helps to stabilize the percolation pathways for ions and electrons, which avoid the contact loss of active material particles and reduce the capacity fading; and 2) The increase of the energy efficiency comes mainly from the plateaus of the charging and discharging process becoming more similar.This indicates that overpotentials which would be induced for the charge transfer between the active particles and the additives are decreased under pressure application.Further, the volume changes for conversion materials in FIBs are very high, emphasizing the increased importance of applying stack pressure.Overall, even though introducing pressure has no significant improvement on the first discharging, the cyclability and reversibility of the cell has been remarkably improved, which outperformed all the reported ASS-FIBs with Bi/BiF 3 as cathode.The data was reproduced with permission, [12] where the detailed crystallographic structural changes of the material has been reported.Energy efficiency was calculated by calculating energies from integration of the charge/discharge curves and subsequent division of the discharge energy by the charge energy of the prior cycle.

Figure 1 .
Figure 1.Rendering of the hot press for solid-state battery operation.

Figure 2 .
Figure2.Force-time profile of the pressure-dependent impedance measurements on La 0.9 Ba 0.1 F 2.9 .

Figure 7 .
Figure 7. Cycling plot of a) cell BiF 3 / BaSnF 4 cycled at 180 MPa and b) corresponding efficiency against cycle number.c) Cell BiF 3 / BaSnF 4 /Sn cycled at ambient pressure and d) corresponding efficiency plot.Energy efficiency was calculated by calculating energies from integration of the charge/discharge curves and subsequent division of the discharge energy by the charge energy of the prior cycle.

Figure 8 .
Figure 8. a,c) Cross-section SEM images of the BiF 3 /BaSnF 4 /Sn cell cycled at 180 MPa and the b,d) cell cycled at ambient pressure.The red dotted line is an indicator to the boundary between the cathode and the electrolyte as judged by the occurrence of carbon fibers which are only present in the electrode composite.

Figure 9 .
Figure 9. Impedance spectra of the BiF 3 /BaSnF 4 /Sn cell before and after cycling at 100 °C and a) ambient pressure and b) 180 MPa.

Figure 10 .
Figure 10.a) All cycles of the La 2 Ni 0.75 Co 0.25 O 4 versus Pb/PbF 2 cell cycled at 450 MPa.b) Charge capacity, discharge capacity, and Coulombic efficiency of the same cell.c) All cycles of the La 2 Ni 0.75 Co 0.25 O 4 versus Pb/PbF 2 cell cycled at ambient pressure.d) Coulombic efficiency of and energy efficiency of the same cell as described.The data was reproduced with permission,[12] where the detailed crystallographic structural changes of the material has been reported.Energy efficiency was calculated by calculating energies from integration of the charge/discharge curves and subsequent division of the discharge energy by the charge energy of the prior cycle.