Influence of Particle Size and Morphology on the Performance of Glass Particle-Based Battery Separators

A common methodology to modify both the thermal characteristics and surface properties of battery separators is the incorporation of particles. At present, the particles employed are predominantly of an arbitrary shape. The enhanced mechanical stability of particle layers comprising platelet-shaped particles is demonstrated and further evaluated on the influence of their aspect ratio (edge length to thickness) on the electrochemical properties of battery separators. Glass is selected as the particle material due to its temperature stability and the freedom of design that it offers when particles are formed directly from the melt. A parametric study of glass particle layers as function of edge length and thickness is conducted. Depending on the electrochemical characterization method, symmetrical cells with stainless steel electrodes and graphite-lithium half cells are used. Particles with an excessively high aspect ratio impede the Li+ diffusion pathway, thereby negatively affecting the performance and stability of the cell. Conversely, if the aspect ratio is insufficient, a deterioration in cell performance can be observed, particularly at elevated C-rates, due to the high specific surface area of the particles. An optimum performance can be achieved by utilizing particles with an aspect ratio of about 10 and a thickness of around 1 µm.

With advantages such as high energy density, low self-discharge, light weight and therefore high specific energy and power, lithium-ion batteries (LIBs) dominate the major energy storage markets, particularly for mobile devices and electric vehicles. One of the key components in liquid electrolyte batteries is the separator. Its function is to physically and electronically separate the two electrodes while allowing free ionic transport. The basic requirements for a separator include chemical and electrochemical stability with respect to the electrolyte and the potential window of the electrodes. Furthermore, high open porosity is required to enable absorption and retention of the electrolyte. [1] In addition, the separator must exhibit the robustness to withstand the mechanical stress of the cell assembly. According to the state-of-the art, a separator should be as thin and lightweight as possible in order not to affect the specific (gravimetric and volumetric) energy density of the battery. [2] On the other hand, the separator represents the safety-relevant component of the cell and must be able to prevent the formation and growth of dendrites during the charging/discharging process. [3] The problem of growing dendrites mainly occurs when metals such as pure lithium are used as anode, which is of particular significance for lithium-metal and lithium-sulfur batteries. [4] Additionally, in fast-charging applications under harsh conditions lithium plating is observed on the graphite anode. [5] Lithium metal batteries (LMBs) are considered as a promising candidate for the next generation of batteries as they offer excellent advantages in terms of high theoretical specific capacity (3860 mAh g −1 ) and the lowest redox potential (−3.04 V versus the standard hydrogen electrode). [6] This is of particular relevance for emerging high-energy applications, such as long-range and heavy-duty electric vehicles, as the highest energy density that conventional LIBs can deliver is far from the demands of these fields. However, the practical application of lithium (Li) metal in secondary batteries is limited by the growth of Li dendrites. Accidental or inhomogeneous deposition of metallic lithium (Li plating) can lead to dendrite growth. [3] If only one dendrite penetrates the separator and forms a conductive path between the two electrodes, a short circuit in the battery leads to its failure. [7] Since battery safety is of paramount importance, the separator is playing a crucial role in the introduction of lithium metal batteries to the market by preventing electronic contact between the electrodes of the battery. [8] During cell operation, the micro-roughness of Li electrode surfaces causes an inhomogeneous distribution of Li ions followed by the formation of Li dendrites. [9] Instead of modifying the lithium surface to allow for more homogeneous plating, [10] a more practical approach would be the modification of the separator with functional and thermally stable particles. [9,11,12] The use of lithium-metal as anode material would offer an enormous increase in energy density compared to current anode materials such as graphite and thus open up new design space for separators that would enable the safe application of these batteries. [8] Even if the additional mass resulting from the separator engineering through particle modification is taken into account, the specific energy density would still exceed that of conventional LIBs to a significant extent. A variety of materials have been put forth for use as particles in separators, the majority of which are alumina or boehmite. [2,3,13] Despite the inherent rigidity of glass materials, previous studies also demonstrated the successful application of glass-based separators in LIBs. [14,15] Glasses offer an optimum compromise between low density and high temperature stability, thereby ensuring safe operation even at higher (dis-)charging rates or in the event of internal damage or external abuse. [16] While the thermal stability of polymers is inferior in comparison to that of crystalline inorganic materials, the density of these temperature-stable materials, such as ceramics, is higher due to their crystalline nature. Amorphous silicate glasses, however, offer a broad stability window with maximum service temperatures of at least 600 °C, which is sufficient for battery applications. Another advantage of glass as separator material is the ability to manufacture particles directly from the melt. [15,17] This allows a high degree of freedom in the choice and control of particle morphology and size. Yuan et al. [18] already proposed that the incorporation of heterogeneous particles within the separator can mitigate and suppress dendrite growth-induced internal short-circuits in solid-state batteries. In their study, the aspect ratio of the embedded particles dominates the dendrite mitigation effect together with the alignment of the particles in layers. Multiple particles with medium aspect ratio, modulated by a specific brick-and-mortar-like arrangement method, can fully mitigate dendrite penetration, while multiple particles with small aspect ratio only elongate the dendrite growth path and delay the short-circuit time. These results reveal that adding heterogeneous blocks with high fracture resistance is a promising approach to mitigate dendrites and reduce short-circuit risk in batteries. Also, Rafiz et al. [11] successfully used boehmite plates to suppress lithium dendrite growth by increasing the tortuosity of their separators. The implementation of these findings requires additional consideration of the actual microstructural effects. In the case of separators for liquid electrolyte systems, these findings have to be adjusted in regard to their influence on the ionic conductivity of the system, especially when using non-conductive inorganic particles. In this case, the porosity and tortuosity of the particle-filled or -coated separator are essential. However, the size and morphology of the utilized particles also have a significant impact on the manufacturing process, like slurry processability and separator formation, and consequently on the electrochemical performance. Although there are many studies that investigated the influence of separator particles of varying sizes, both spherical and arbitrarily shaped, on the electrochemical performance, [19,20] there are only few references for the impact of particle shape, [20,21] none of which include variations of platelet-shaped particles. Inspired by the “brick-and-mortar” toughening mechanism, [18] we propose the strategy of adding platelet-shaped particles to separators for lithium battery systems. In this paper, we evaluate the influence of the morphology of platelet-shaped glass particles in battery separators on the electrochemical properties of the system. A quantitative parametric study of glass particle layers as a function of the edge length and the thickness is carried out in order to improve the electrochemical performance of battery separators comprising these. In the future, our findings may facilitate the development of innovative coatings for polyolefin separators or the introduction of novel particle-based separator concepts.

and discussion In this study, commercially available glass flakes (Glassflake Ltd., Leeds, UK) of various sizes in terms of thickness and edge length, according to the manufacturer’s data sheet (see S1 in supporting information) were analyzed for their suitability as separator materials. The schematic morphology of a glass flake compared to the real morphology obtained by SEM is shown in Figure 1 . In order to further improve the safety of lithium-ion batteries, the advantages of platelet-shaped particles are compared with those of arbitrarily shaped particles. The use of these so-called glass flakes serves the purpose of forming an effective barrier against growing lithium dendrites by establishing a self-stabilizing structure as illustrated in Figure 2 . This barrier prevents or at least effectively delays the penetration of the separator. [11] Morphology of the different sized glass particles The thickness of the glass flakes is determined by the selected production parameters such as the viscosity of the melt and the rotational speed of the crucible, while the edge length is adjusted by milling and classifying. This allowed the characteristics of the separator to be tailored by selecting the optimum flake size. Microstructures of glass flakes with different edge lengths but equal thickness of 1 µm, i.e. different aspect ratios were evaluated by scanning electron microscopy and are shown in Figure 3 . The glass flakes with varying edge length and thickness, used to determine the effects of these parameters by full factorial design of experiments, are illustrated in the supporting information (Figure S4). All glass particles have a sharp-edged morphology devoid of any discernible agglomerations. The glass flakes display a rectangular appearance, with one edge length falling within the range of the nominal edge length, while the other edge exceeds it. The particles range from powder-like structures at low aspect ratio to platelets with a medium aspect ratio, and sheet-like morphologies at high aspect ratio. The preferred fracture mode of the glass platelets is along the edges, with minimal impact on the thickness, resulting in a cube-like appearance after milling. Some residual platelet-shaped particles are also present. No sieve classification could be conducted in this size range due to the unavailability of sieves with a mesh size of less than 10 microns. The determination of a mean particle size of platelets is challenging due to the fact that the measured values represent a mixture of the expansion obtained in each spatial direction, resulting in a wide distribution. [22] However, the characteristic particle sizes determined by laser diffraction are in accordance with the data sheet ( Table 1 ) and the SEM images. The distribution of the milled particles is the broadest observed, which is due to the partially incomplete milling and classifying, respectively. As the particle size is reduced, the specific surface area of the particle collective is observed to increase, while maintaining a relatively moderate level in comparison to that of common separator materials. Evidence of the barrier effect of platelet-shaped particles We previously outlined that the ability of battery separators to resist piercing by lithium dendrites is a crucial attribute for their intended application. As indirect proof of the beneficial effect of particle modification of separators the depth-dependent force for penetration into a compacted particle bed is regarded as a measure of the penetration resistance. To calculate this, the gradient of the force measurement curve is determined in the linear-elastic range of the force measurement graph. In this region, the particle layer is supported by interlocking of the particles before either lateral displacement or excessive compaction occurs. This provides a parameter analogous to the modulus of elasticity, which enables comparison of the particles with varying morphologies, i.e. aspect ratios. Figure 4 illustrates the depth-dependent penetration resistance of glass particles with a thickness of 1 µm, with respect to the aspect ratio of the particles. The penetration resistance of particle layers increases in proportion to the aspect ratio of the particles. A linear correlation can be proven. Consequently, a bulk sample of glass would invariably show the highest penetration resistance. This is a notable discovery, as battery separators need to be flexible for the battery assembly, which often involves processes like winding where a high degree of curvature is applied. The use of particulate systems is therefore essential. Particle modification of separators according to the state-of-the-art employs particles of arbitrary shape. Therefore, the utilization of platelet-shaped particles with an optimized aspect ratio could markedly enhance penetration resistance. Thermal stability In addition to mechanical damage, such as that caused by lithium dendrites, the breakdown of separators is probably a consequence of melting, shrinking and vaporization under conditions of high temperature. This results in the formation of internal short circuits and the onset of thermal runaway within the cell. Separator materials with an intrinsic high melting point represent a promising solution to the safety concerns associated with LIBs. Thermogravimetric analysis (TGA) was used to compare the thermal stability of the glass used with those of other coating material and with a pure polypropylene (PP) separator over a temperature range of 25 to 700°C. The loss of mass is illustrated in Figure 5 . The TGA was performed in a nitrogen atmosphere, which resulted in a delay in the degradation of polypropylene (PP) compared to its behavior under standard ambient conditions. Full breakdown of polypropylene was initiated at approximately 400 °C and continued until the complete sample was decomposed at 480 °C. The mass loss of boehmite at temperatures around 500 °C can be assigned to water release contributing to a remaining mass of only 82%. [23] Water in the battery may react exothermically with lithium metal and the conducting salt of the electrolyte resulting in the evolution of hydrofluoric acid (HF). The glass composition used in this study shows no observable mass loss within the observed temperature range relevant to battery applications. This, combined with the lower density of the glass (2.6 g cm - ³; boehmite: 3.0 g cm - ³), is highly favourable for utilization as a battery separator. Electrochemical stability Linear sweep voltammetry (LSV) was performed to determine the electrochemical stability dependent on the particle material. The LSV curves of glass and boehmite particles for reference exhibit a comparable shape (see supporting information Figure S5). Their currents are below 2 µA for glass, which is even lower than for boehmite (6 µA), indicating that no redox reaction takes place in the system below 5.0 V. Consequently, when these cells are operated within the typical voltage range of 2.75 - 4.5 V, the glass particles are stable and can meet the requirements of a lithium-ion battery. The measured current of the glass particles showed minimal variation across the potential window, even lower than that of the boehmite. These results indicate that the glass particles not only possess electrochemical stability when operated within the relevant voltage range for the intended application but also have a reduced propensity for side reactions compared to boehmite. Ionic conductivity in liquid-electrolyte system A full factorial design of experiments was employed to evaluate four different fractions of particles, characterized by two levels of thickness (1 µm, 5 µm) and edge length (30 µm, 120 µm), with respect to the ionic conductivity of a bulk powder bed containing these particles. The ionic conductivities calculated from impedance measurements for each combination are given in Table 2 . The effect of the “Thickness” parameter is slightly positive, meaning that increasing the flake thickness slightly improves the ionic conductivity of the powder bed. In contrast, the effect of the “Edge length” parameter is strongly negative, i.e. increasing the edge length of the flakes strongly reduces the ionic conductivity of the powder bed. This is consistent with the screening of different flake types in Figure 6a) . The lowest ionic conductivity is obtained with particles of high edge length and low thickness, while a powder bed of particles of low edge length and high thickness provides the highest ionic conductivity. The influence of the edge length is significantly higher than the influence of flake thickness. It can be concluded that a high aspect ratio reduces the ionic conductivity. We consider the influence of thickness and edge length using a thought experiment. A few assumptions are made to illustrate this. The fixed height of the battery separator is 20 µm. In addition, a brick-and-mortar structure is assumed due to the flat orientation of the flakes and the presence of a homogeneous layer structure. The developed influencing factors of the platelet size and aspect ratio on the ionic conductivity in a non-aqueous battery electrolyte are visualized in Figure 6c) . The brick wall modelling clearly shows that for thin particles with short edge lengths there is a high porosity with moderate tortuosity. The porosity of the system is determined by the edge length of the particles, which defines the number of “entrances” for the Li + ions. The tortuosity is represented by the distance, which needs to be covered to pass the glass particle layer for a given total thickness. This distance increases with decreasing particle thickness (see (I) and (II) in Figure 6c)). If the edge length of the particles is too high, the accessibility for Li + ions is blocked, and the ionic conductivity is low as shown in (III) in Figure 6c). Based on these findings, we again attribute the main influence on the electrochemical performance to the edge length of the particles. Accordingly, the following considerations are constrained to particles with a constant thickness of 1 µm and varying edge lengths and aspect ratios, respectively. The ionic conductivities of the powder beds with 1 µm thick flakes and variable aspect ratios (as shown in Figure 3) are compared in Figure 6b) . The arbitrarily shaped particles with an aspect ratio around 1 were obtained by excessive milling of glass flakes to eliminate the platelet shape. Particle layers with low aspect ratios of 1 and 10 exhibit an ionic conductivity of approximately 4.5 mS cm -1 . A slight decline is observed at an aspect ratio of 30, followed by a significant drop at a high aspect ratio of 120. This substantiates the concept of the brick-and-mortar-structure hypothesis introduced in Figure 2. The planar morphology of these platelets impedes lithium-ion transport by blocking access points to the separator layer, increasing the path length and hindering ion transport. On the other hand, the decrease of ionic conductivity between arbitrarily shaped particles and flakes with an aspect ratio of 10 is negligible. Therefore, maintaining the indicated barrier function by using glass flakes with short edge lengths instead of arbitrary shapes is a viable option, which is also consistent with the size ratio of those heterogeneous blocks used to mitigate dendrite propagation in the phase-field based multiphysics model of Yuan et al. [18] . However, a further increase of the glass flake size is not recommended, because the ionic conductivity decreases rapidly when the aspect ratio is greater than 10. Accessible capacity in graphite-lithium half cells To gain insight into the accessible capacities of graphite-lithium half cells containing separators made of glass platelet layers with varying aspect ratios, the cell voltage over the specific capacity in the first lithiation step is compared for given C-rates in Figure 7 . The accessible capacities in lithium half cells vary significantly for the different aspect ratios. The initial lithiation of the graphite anode has been evaluated, demonstrating that at low discharge rates, such as C/10, almost the entire specific capacity of the electrode material graphite (372 mAh g -1 ) can be utilized with separator layers of particles with an aspect ratio of 10 or lower. [24] The following capacities have been achieved: 354 mAh g -1 at aspect ratio 1, 349 mAh g -1 at aspect ratio 10, 321 mAh g -1 at aspect ratio 30 and 175 mAh g -1 at aspect ratio 120. Accordingly, the accessible capacity of the electrode is clearly limited by the morphology of the separator particles when a voltage-related termination criterion (0.01 V) is employed. The utilization of flakes with an aspect ratio of 10 or lower as a separator has resulted in twice the capacity compared to an aspect ratio of 120, at a discharge rate of C/10. At an elevated discharge rate of C/2, the accessible capacities decline significantly for all aspect ratios considered in this study. This is a direct consequence of using bare glass particles in their natural state, without any additional processing to enhance their suitability as separators, such as surface treatment or grafting. Nevertheless, the influence of the aspect ratio is clearly visible. It is evident that at a discharge rate of C/2, the difference in accessible specific capacity among the various aspect ratios is even more pronounced. In this instance, the capacity is three times higher when using flakes with an aspect ratio of 10 compared to those with an aspect ratio of 120, and almost five times higher when using flakes with an aspect ratio of 1 as separator. It is also notable that the ratio between C lith, C/2 and C lith, C/10 gives a first impression of the rate capability of the cells. For example, the capacity at C/2 is only 27.5% of the initial capacity at C/10 for an aspect ratio of 10 and even lower for higher aspect ratios. All the specific capacities related to the mass of active material and the proportion of the remaining capacity at elevated C-rate are summarized in Table 3 . In order to investigate the stability of the electrochemical performance over time, the accessible capacity is once again monitored at the 15th discharge cycle. Now the particle fractions that demonstrated a high initial performance exhibit the highest degree of degradation. After 15 cycles at C/2 the cells with particles of an aspect ratio of 1 reach only 56.3% of their initial capacity, while the quotient C lith, C/2, 1st / C lith, C/2, 15th is 95.2% for an aspect ratio of 120. Flakes with an arbitrary shape (aspect ratio of 1) still remain the optimum choice for achieving high capacity, while particles with an aspect ratio of 10 and 30 demonstrate comparable performance at a level below that of the aforementioned particles. The increased susceptibility to degradation can be explained by the specific surface area of those various particle fractions shown in Table 1. An increase in surface area results in a greater propensity for side reactions during galvanostatic cycling. This encompasses surface related phenomena like the alteration of the surface morphology, increase of contact angle between separator and electrolyte as well as pore blockage caused by electrolyte evaporation and salt precipitation. [25] The volumetric quantity of electrolyte was constant for each cell, resulting in a reduction of the available amount of electrolyte per overall particle surface area with increasing particle related surface area. The drying-out of the electrolyte over time led to a decline in the electroactive surface area and the volume fraction of active material and to the acceleration of capacity fading, as described by Fang et al. [26] . Thus, the effects of capacity loss after cycling are less pronounced with the lower surface area at an aspect ratio of 120. Influence of flake geometry on fast charging behavior High-rate capability of lithium-ion batteries represents a significant challenge in the field of battery research. In particular, separator modification with non-porous and non-ion-conductive particles could potentially compromise battery performance. Therefore, fast charging behavior is analyzed by incrementally increasing the charging currents. Again, layers of glass flakes with different aspect ratios are used as separators in a half cell configuration (graphite/lithium). Figure 8a) illustrates the relative capacity in relation to the first cycle at C/2, with data points representing 1C, 2C and 5C each comprising five cycles. The subsequent recovery of the initial capacity is highlighted by an additional five cycles at C/2 following the C-rate variation. These results are in agreement with the respective ionic conductivities. Particles with a too high aspect ratio (edge length to thickness), such as those with L:D = 120, impede the Li + diffusion pathways leading to lower performances. This is particularly evident at elevated C-rates. While the relative reduction in capacity is consistent across all aspect ratios at 1C, there is no usable capacity at 2C with high aspect ratio particles. The limit of the unprocessed particle layers as a separator was reached at 5C, at which all samples failed to charge. Thus, the enhanced ionic conductivity of the particle layers consisting of flakes with short edge lengths proves to be particularly advantageous for fast-charging applications. The subsequent recovery of capacity to the initial value before the C-rate variation is complete for aspect ratio 10, but incomplete for aspect ratios 1, 30 and 120, respectively. This indicates that the rapid polarization during incomplete fast charge and discharge cycles may have compromised the structural integrity of the cell, potentially leading to phenomena such as lithium plating or electrolyte decomposition, which obstruct the electrode surface. If the aspect ratio is too low, the brick wall-like safety barrier cannot be formed in its entirety. If the aspect ratio is too high, the internal resistance, combined with high currents can compromise the integrity of the electrode material. A comparison of the internal resistance of the half-cells before and after cycling, as illustrated in Figure 8b), serves to confirm the phenomenon of accelerated degradation at low separator particle aspect ratio. The half cells with glass particle-bed separators demonstrate an increase in internal resistance following cycling with C-rate variation for all types of glass flakes. The degree of degradation is defined by the relative increase of internal resistance. It is therefore evident that a low aspect ratio, which may be characterized by a high specific surface area, allows for a greater number of side reactions to occur, as mentioned above, which in turn leads to the observed decline in electrochemical performance. It is worth mentioning that the particles are used as received without any processing steps to enhance their suitability. The observed behavior may also depend on the material and glass composition. However, as the focus of this study was the influence of particle morphology, the long-term stability is considered in further studies.

The hypothesis of using a barrier similar to a “brick-and-mortar-structure” made of glass platelets in order to toughen separators mechanically has been experimentally tested and proven correct. It has been demonstrated that penetration resistance increases in proportion to the aspect ratio of the utilized particles. Therefore, it was essential to determine the optimum size and morphology of the glass particles for use in battery applications. Consequently, we investigated the form factor-dependent properties of glass particles and their effect on the ion transport and fast charge performance of lithium-ion batteries. Glass was selected because of its high degree of freedom in forming particles directly from the melt. In addition, glass compositions adapted to the battery environment provide a reasonable alternative to the particles that are currently used in battery applications. The material density is significantly lower than that of ceramics such as alumina. Glass exhibits superior thermal stability compared to polymers or boehmite. The most suitable morphology of such platelet-shaped particles, exemplified by glass flakes, was investigated by electrochemical impedance spectroscopy (EIS). Thickness and edge length of these flakes were varied according to a full factorial experimental design. Increasing the thickness of the flakes improved the ionic conductivity of the particle layers slightly. However, with increasing edge length of the flakes the ionic conductivity of the electrolyte-soaked powder bed strongly decreased. Consequently, a high aspect ratio (i.e. edge length to thickness ratio) of the separator particles has been observed to diminish the ionic conductivity by blocking the Li + diffusion pathways. The fact that the thickness of the glass flakes has a minor influence is advantageous for processing, as the edge length can be easily adjusted by milling. In addition, thinner particles are beneficial for the volumetric energy density of battery cells because they allow lower separator thickness for a given number of particle layers. Building on this discovery, the aspect ratio of glass flakes with a thickness of 1 µm was varied by changing their edge length in order to identify an optimum compromise in terms of the electrochemical performance. The ionic conductivity of glass flakes with an aspect ratio of 10 is found to be highly comparable to that of arbitrarily shaped particles with an aspect ratio of 1. Also, galvanostatic cycling tests were conducted in a half-cell configuration with a variation of C-rate in order to evaluate the influence of separator particles on the accessible capacity of graphite electrodes. While the aspect ratio of particles has a relatively minor impact at low C-rate (C/10) the loss of accessible capacity is significant for particles of high aspect ratio especially at elevated charge current (C/2). Thus, a trade-off between electrochemical performance and safety barrier function was made by choosing particles with a maximum aspect ratio of 10. This should be considered due to the fact that separator particles with a higher aspect ratio will demonstrate an enhanced degree of piercing resistance by forming a brick-and-mortar structure. These results indicate that optimizing the material and particularly the shape of separator particles has the potential to enhance the physical, thermal, and electrochemical properties of separators in practical applications. Moreover, the structure and properties of separators coated with various particle shapes offers valuable insights into the design of separators with enhanced performance. In sum, the findings of this study can be readily applied to improve the characteristics of separators both physically and electrochemically. This work provides a framework for selecting suitable particle sizes and shapes for incorporation into separator materials. The adaptation of separators to specific applications can thus be facilitated by modifying the aspect ratio of the particles. Experimental Section

The rotary atomization of a glass melt results in a wide range of particle morphologies that can be categorized by shape and structure. This enables targeted functionalization of separators. A detailed description of the processing route can be found in the publications of Kyrgyzbaev et al. [27] and Schadeck et al. [15] . The platelet-shaped particles (glass flakes) produced by this process vary considerably in thickness D and edge length L. Due to the manufacturing process of glass flakes by rotary atomization their thickness is defined by processing parameters such as viscosity, temperature and rotational speed. The edge length is adjusted by subsequent milling and classifying. [15,17,27] Although L1 and L2 are in the same range, they do not necessarily possess the same value. Consequently, the morphology of glass flakes can vary considerably, exhibiting shapes such as square, rectangular, or even needle-like structures, depending on the manner of their fragmentation during milling. The most important attribute for the description of platelet-shaped particles, such as glass flakes, is the aspect ratio, which is defined as the ratio between edge length and thickness (L:D). To investigate the influence of the two factors, particle fractions at high and low levels of thickness and edge length are compared. Furthermore, particles of fixed thickness and variable aspect ratio are selected. In order to provide a reference point, the edge length of a fraction of glass flakes was deliberately reduced as much as possible by excessive ball milling to reduce the aspect ratio and eliminate the platelet shape. The composition of the glass flakes is provided by the manufacturer and is included in the supporting information (see S2). To demonstrate the thermal stability of battery separator materials, commercially available boehmite (γ-AlOOH) particles as well as a polypropylene separator (Celgard®) were selected as references. Cell assembly and cycling All cells for the electrochemical characterization were assembled in an argon-filled glove box (O 2 and H 2 O content < 1 ppm) in a coin-cell-type setup (Swagelok cell, supporting information S3). In order to conduct electrochemical characterization of the glass powders, stainless steel electrodes were employed. Prior to sealing, defined bulk powder volumes were placed within the cell and compressed. The height of the powder beds was adjusted and verified by measuring the height of the cells before and after assembly. Half cells were assembled to test the effective capacities when using a powder bed as separator. To build graphite-lithium cells in half cell configuration a commercially obtained graphite-coated copper foil (Customcells Itzehoe GmbH, Itzehoe, Germany) was used as electrode (spec. capacity: 355 mAh g -1 ). Furthermore, a 12 mm coin of pure lithium metal (750 µm, Alfa Aesar, Haverhill, MA, USA) served as counter electrode. Prior to cell assembly, the graphite electrodes were punched out to 12 mm coins and dried at 110 °C for 16 h together with the glass flakes in Argon atmosphere. To maintain a constant pressure (0.05 MPa), a steel spring (EN material number: 1.4310, Gutekunst springs) and a nickel coin (d = 12 mm, h = 500 µm; HMW Hauner, Röttenbach, Germany) were placed inside the cell stack. For all the cells a 1 m solution of LiPF 6 in a 1:1 (v/v) mixture of ethylene carbonate (EC) and dimethyl carbonate (DMC) (Selectilyte TM LP 30, BASF, Ludwigshafen, Germany) was used as the electrolyte. Physico- and electrochemical characterization techniques SEM The scanning electron microscopy (SEM) characterization shown in this study was performed using a Zeiss scanning electron microscope (LEO 1530, Zeiss, Jena, Germany) at an accelerating voltage of 5 kV. First, the particles were characterized by scanning electron microscopy to reveal their microstructure and illustrate the different aspect ratios. The nominal particle size of the glass flakes was verified by laser diffraction technology using a particle size analyzer (PSA 1190, Anton Paar, Austria). BET The specific surface area of the particle fractions was determined by N 2 physisorption (ASAP 2010, Micromeritics, Unterschleißheim, Germany) according to Brunauer-Emmet-Teller (BET). TGA Thermogravimetric analysis (NETZSCH STA 449 F5 and 449C, Selb, Germany) was conducted up to 700 °C in nitrogen atmosphere (heating rate 5 K min -1 ; gas flow 45 ml min -1 ). Mechanical characterization Mechanical testing was performed using a universal testing machine (inspekt duo 10, Hegewald & Peschke, Nossen, Germany). The aspect ratio-dependent piercing resistance of particle layers was determined in close accordance with the standard ASTM F 1306 using a Swagelok setup. Prior to penetration, 0.5 g of glass particles were compacted at 10 MPa for a period of 1 min. Subsequently, an indenter with a tip radius of 1.6 mm and a conical shaft with an inclination of 6° was driven into the particle bed, and the resistance force was measured with a 500 N load cell. The testing device was set to a traverse speed of 1 mm min -1 . The termination criterion was reached at a measured force of 20 N. Linear sweep voltammetry The electrochemical stability of the particles soaked with liquid electrolyte was investigated by carrying out linear sweep voltammetry (VMP3, Biologic, Seyssinet-Pariset, France) experiments utilizing stainless steel electrodes. These experiments were conducted within a potential range of 0.0 - 5.0 V at a scanning rate of 0.02 mV s -1 . Electrochemical impedance spectroscopy (EIS) Conductivity measurements were performed at 23 °C using galvanostatic impedance spectroscopy (Reference 600, Gamry Instruments, Warminster, PA, USA) from 1.0 MHz to 0.5 Hz with an AC current of 0.1 mA. Each sample was measured three times in a Swagelok cell setup with stainless steel electrodes. The ionic conductivities were then calculated, taking into account the internal resistance, thickness and area of the particle layer. The objective of this analysis was to evaluate the effect of two factors (edge length and thickness) at two levels on the ionic conductivity of electrolyte-soaked powder beds containing glass flakes of different aspect ratios. This was achieved by the implementation of a full factorial design of experiments. Galvanostatic cycling Further electrochemical investigations of the cells were carried out at 23 °C using a BaSyTec CTS Lab (Asselfingen, Germany) battery tester after a 24 h rest period following cell assembly. The galvanostatic cycling protocol for all tested cells started with a formation procedure (two cycles at C/10; two cycles at C/5; one cycle at C/2). After formation and 15 cycles at C/2 the current density was gradually increased to 1C, 2C and 5C (five cycles each), followed by another five cycles at C/2 to evaluate the effects of high current densities on the accessible capacity of graphite-lithium half cells. The discharge and charge conditions were maintained in constant current mode, with a cut-off voltage of 0.01 V and 1.0 V, respectively.

The authors thank the Bavarian Research Foundation (Bayerische Forschungsstiftung, BFS; GlasSeLIB – funding code AZ-1463-20) for the funding. Next, the authors would like to thank the Keylab Electron and Optical Microscopy at the Bavarian Polymer Institute (BPI) for SEM imaging (at Zeiss LEO 1530). Furthermore, P.R. and S.M. thank the graduate school of the Bavarian Center for Battery Technology (BayBatt) for ongoing support. Conflicts of Interest There are no conflicts to declare. Data Availability Statement The data that support the findings of this study are available from the corresponding author upon reasonable request.

Lithium-ion Batteries; Separator; Inorganic; Safety CRediT authorship contribution statement Philipp Rank: Conceptualization, Methodology, Investigation, Validation, Formal Analysis, Data curation, Writing – Original Draft, Visualization. Sebastian Müllner: Methodology, Investigation, Formal Analysis, Writing – Original Draft. Thorsten Gerdes: Resources, Writing – Review & Editing, Supervision, Funding acquisition Christina Roth: Resources, Writing – Review & Editing, Supervision, Funding acquisition ORCID P. Rank https://orcid.org/0009-0008-5334-5578 S. Müllner https://orcid.org/0000-0001-9696-0011 T. Gerdes https://orcid.org/0009-0004-8384-7086 C. Roth https://orcid.org/0000-0003-1159-2956

[1] C. J. Weber, S. Geiger, S. Falusi, M. Roth in AIP Conference Proceedings, 2014, pp. 66–81. [2] K. Liu, Y. Liu, D. Lin, A. Pei, Y. Cui, Sci. Adv. 2018, 4, eaas9820. [3] J. Holtmann, M. Schäfer, A. Niemöller, M. Winter, A. Lex-Balducci, S. Obeidi, J Appl Electrochem 2016, 46, 69. [4] a) Z. Hao, Q. Zhao, J. Tang, Q. Zhang, J. Liu, Y. Jin, H. Wang, Materials horizons 2021, 8, 12; b) S. Kim, W.-G. Lim, A. Cho, J. Jeong, C. Jo, D. Kang, S. M. Han, J. W. Han, J. Lee, ACS Appl. Energy Mater. 2020, 3, 2643. [5] S. Liu, B. Gu, Z. Chen, R. Zhan, X. Wang, R. Feng, Y. Sun, Journal of Energy Chemistry 2024, 91, 484. [6] W. Xu, J. Wang, F. Ding, X. Chen, E. Nasybulin, Y. Zhang, J.-G. Zhang, Energy Environ. Sci. 2014, 7, 513. [7] N. Wassiliadis, J. Schneider, A. Frank, L. Wildfeuer, X. Lin, A. Jossen, M. Lienkamp, Journal of Energy Storage 2021, 44, 103306. [8] J. Jang, J. Oh, H. Jeong, W. Kang, C. Jo, Materials (Basel, Switzerland) 2020, 13 . [9] L. Tan, Y. Sun, C. Wei, Y. Tao, Y. Tian, Y. An, Y. Zhang, S. Xiong, J. Feng, Small (Weinheim an der Bergstrasse, Germany) 2021, 17, e2007717. [10] a) G. Lu, J. Nai, D. Luan, X. Tao, X. W. D. Lou, Science advances 2023, 9, eadf1550; b) Y. Xie, Y. Huang, Y. Zhang, T. Wu, S. Liu, M. Sun, B. Lee, Z. Lin, H. Chen, P. Dai et al., Nat Commun 2023, 14, 2883; c) J. Becking, A. Gröbmeyer, M. Kolek, U. Rodehorst, S. Schulze, M. Winter, P. Bieker, M. C. Stan, Adv Materials Inter 2017, 4 . [11] K. Rafiz, D. R. L. Murali, J. Y. Lin, Electrochimica Acta 2022, 421, 140478. [12] R. Muchakayala, S. Yarramsetti, P. S. Maram, S. Kalluri, F. Ran, S. Sangaraju, Journal of Energy Storage 2023, 68, 107687. [13] a) M. He, X. Zhang, K. Jiang, J. Wang, Y. Wang, ACS applied materials & interfaces 2015, 7, 738; b) W. Mi, G. Sharma, X. Dong, Y. Jin, Y. S. Lin, Journal of Power Sources 2016, 305, 209; c) D. Parikh, C. J. Jafta, B. P. Thapaliya, J. Sharma, H. M. Meyer, C. Silkowski, J. Li, Journal of Power Sources 2021, 507, 230259; d) C. Shi, J. Dai, C. Li, X. Shen, L. Peng, P. Zhang, D. Wu, D. Sun, J. Zhao, Polymers 2017, 9, 159; e) C. Yang, H. Tong, C. Luo, S. Yuan, G. Chen, Y. Yang, Journal of Power Sources 2017, 348, 80; f) Y. Zhang, Z. Wang, H. Xiang, P. Shi, H. Wang, Journal of Membrane Science 2016, 509, 19. [14] U. Schadeck, K. Kyrgyzbaev, H. Zettl, T. Gerdes, R. Moos, Energies 2018, 11, 999. [15] U. Schadeck, K. Kyrgyzbaev, T. Gerdes, M. Willert-Porada, R. Moos, Journal of Membrane Science 2018, 550, 518. [16] W. Lin, F. Wang, H. Wang, H. Li, Y. Fan, D. Chan, S. Chen, Y. Tang, Y. Zhang, ChemSusChem 2022, 15, e202201464. [17] A. R. Frost, Journal of Agricultural Engineering Research 1981, 26, 63. [18] C. Yuan, B. W. Sheldon, J. Xu, Advanced Energy Materials 2022, 12, 2201804. [19] a) D. Takemura, S. Aihara, K. Hamano, M. Kise, T. Nishimura, H. Urushibata, H. Yoshiyasu, Journal of Power Sources 2005, 146, 779; b) N. Kanhere, K. Rafiz, G. Sharma, Z. Sun, Y. Jin, Y. S. Lin, Powder Technology 2019, 353, 230; c) M. R. Asghar, M. T. Anwar, A. Naveed, J. Zhang, Membranes 2019, 9 ; d) H. Liu, Z. Dai, J. Xu, B. Guo, X. He, Journal of Energy Chemistry 2014, 23, 582; e) S. C. Mun, J. H. Won, Front. Energy Res. 2022, 10 ; f) D.-W. Lee, S.-H. Lee, Y.-N. Kim, J.-M. Oh, Powder Technology 2017, 320, 125. [20] D.-E. Hyun, Y.-J. Jung, T.-W. Kim, S.-M. Koo, C. Park, H. S. Kim, S. Kim, W. H. Shin, D.-W. Lee, J.-M. Oh, Ceramics International 2023, 49, 30147. [21] P. Zhang, L. Chen, C. Shi, P. Yang, J. Zhao, Journal of Power Sources 2015, 284, 10. [22] E. J. Garboczi, K. A. Riding, M. Mirzahosseini, Advanced Powder Technology 2017, 28, 648. [23] T. Sun, Q. Zhuo, Y. Chen, Z. Wu, High Performance Polymers 2015, 27, 100. [24] a) Y. Nishi, Journal of Power Sources 2001, 100, 101; b) J. Asenbauer, T. Eisenmann, M. Kuenzel, A. Kazzazi, Z. Chen, D. Bresser, Sustainable Energy Fuels 2020, 4, 5387. [25] a) J. S. Edge, S. O’Kane, R. Prosser, N. D. Kirkaldy, A. N. Patel, A. Hales, A. Ghosh, W. Ai, J. Chen, J. Yang et al., Physical chemistry chemical physics : PCCP 2021, 23, 8200; b) R. Yang, G. Yu, Z. Wu, T. Lu, T. Hu, F. Liu, H. Zhao, Journal of Energy Storage 2023, 63, 107107. [26] R. Fang, P. Dong, H. Ge, J. Fu, Z. Li, J. Zhang, Journal of Energy Storage 2021, 42, 103013. [27] K. Kyrgyzbaev, A. Rosin, M. Willert-Porada, Glass Tech.: Eur. J. Glass Sci. Technol. A 2016, 57, 95. Tables Table 1. Geometric properties of the particles with a thickness of 1 µm and varying edge length | Nominal aspect ratio | 1 | 10 | 30 | 120 | | D10 [µm] | 1.81 | 5.98 | 9.97 | 22.46 | | D50 [µm] | 7.14 | 12.95 | 22.09 | 53.83 | | D90 [µm] | 17.06 | 20.72 | 34.41 | 93.35 | | Specific surface area [m² g -1 ] | 1.01 | 0.81 | 0.71 | 0.29 | Table 2. Evaluation of the effect of thickness and edge length of platelet-shaped particles on the ionic conductivity of an electrolyte-soaked powder bed containing these particles | 1 | 1 | 30 | 3.157 | | 2 | 5 | 30 | 3.350 | | 3 | 1 | 120 | 0.195 | | 4 | 5 | 120 | 1.025 | | Contrast | 1.0230 | -5.2870 | | | Effect | 0.5115 | -2.6435 | Table 3. Comparison of the electrochemical values of the presented particle morphologies | C lith (1st cycle at C/10) | |||| | [mAh g -1 ] | 354 | 349 | 321 | 175 | | C lith (1st cycle at C/2) | |||| | [mAh g -1 ] | 142 | 96 | 70 | 21 | | C lith (15th cycle at C/2) | |||| | [mAh g -1 ] | 80 | 59 | 52 | 20 | | C lith, C/2, 1st / C lith, C/2, 15th [%] | 56.3 | 61.5 | 74.3 | 95.2 | | C lith, C/2, 1st / C lith, C/10, 1st [%] | 40.1 | 27.5 | 21.8 | 12.0 | Figure legends Figure 1. a) Characteristics of platelet-shaped particles; b) SEM microstructures of differently sized glass flakes Figure 2. Schematic representation of the progressive propagation of dendrites in two different types of separators: spherical particle layers compared to “brick-and-mortar-structure” of platelet-shaped particles under piercing stress Figure 3. SEM images of glass flakes with aspect ratio of a) 1; b) 10; c) 30; d) 120 corresponding to a thickness of 1 µm and e) their respective particle size distribution Figure 4. Penetration resistance of particle layers as a function of particle aspect ratio Figure 5. Thermogravimetric analysis (TGA) curves of typical materials utilized in battery separators Figure 6. a) Heatmap of the ionic conductivity of powder beds with platelet-shaped particles of varying edge length and thickness; b) Ionic conductivities of powder bed separators with particles of different aspect ratio; c) Scheme of the effect of particle thickness and edge length on the structural composition and properties of perfectly layered separators with given thickness Figure 7. Cell voltage vs capacity recorded during a constant current discharge at a) C/10 (1st cycle); b) C/2 (1st and 15th cycle) Figure 8. a) Capacity retention of graphite-lithium half cells containing platelet-shaped glass particles as separator during C-rate variation; b) Comparison of the internal resistance of graphite-lithium half cells with particle-bed separators of different aspect ratio Information & Authors Information Version history Peer review timeline Published Batteries Version of Record22 Oct 2025Published Copyright This work is licensed under a Non Exclusive No Reuse License.

Authors Metrics & Citations Metrics Article Usage 379views 172downloads Citations Download citation Philipp Rank, Sebastian Müllner, Thorsten Gerdes, et al. Influence of Particle Size and Morphology on the Performance of Glass Particle-Based Battery Separators. Authorea. 31 March 2025. DOI: https://doi.org/10.22541/au.174343880.08999503/v1 DOI: https://doi.org/10.22541/au.174343880.08999503/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu.