Binder-free self supporting SiOxCy/C anodes: mechanical reinforcement by biomass fibers and surface fluorination effects on lithium-ion battery performance

1 Introduction

Lithium-ion batteries (LIBs) have garnered substantial success and have undergone a significant transformation over an approximately three-decade period. They have evolved from serving as the preferred power source for relatively compact applications, such as portable electronics, to powering large-scale systems like hybrid electric vehicles (HEVs), and stationary energy storage solutions. This evolution has entailed crucial advancements, notably the adoption of graphite-based anodes (Tarascon and Armand, 2001; Bresser et al., 2014). This transition has played a pivotal role in achieving significantly improved overall energy densities for LIBs, owing to the graphite anodes, low de-/lithiation potential, and its theoretical capacity of 372 mAh.g^-1 (Yazami and Touzain, 1983; Basu et al., 1979).

Nevertheless, despite these remarkable advancements in LIB technology, the current landscape presents a growing demand for enhanced charge power density, primarily driven by the imperative need to reduce recharging times (Palacín, 2009). This heightened requirement frequently surpasses the capabilities of existing LIBs, resulting in challenges linked to safety concerns arising from dendrite formation. In this context, the widespread use of graphite anodes encounters limitations in meeting the demanding charge power-density prerequisites (Deng, 2015; Walter et al., 2020).

In the realm of extensive research on high-capacity candidates, silicon (Si) has garnered considerable attention from researchers as a potential anode material. This interest is primarily due to its high theoretical specific capacity (e.g., 3,580 mAh.g^-1 for Li₁₅Si₄, 4,200 mAh.g^-1 for Li_4.4Si) (Moon et al., 2019; Qi et al., 2020), low discharge potential (0.37 V vs. Li⁺/Li), and abundant reserves (Manj et al., 2020). However, its cycling performance is significantly compromised by the substantial volume expansion (approximately 300%) during lithiation. This phenomenon results in active particle fragmentation, electrode structure disintegration, and an unstable solid-electrolyte interface (SEI) (Li et al., 2023; Adenusi et al., 2023; Peng et al., 2025). All of this contributes to poor cycling performance and presents significant challenges for the widespread adoption of Si in practical applications.

One of the most effective strategies for enhancing the cycling stability of Si is the partial oxidation process, leading to the formation of silicon monoxide SiOx (0 < x < 2), as an anode material. SiOx offers several advantages, including a considerably reduced volume expansion (approximately 150%) and a high theoretical specific capacity up to 1965 mAh.g- (Tarascon and Armand, 2001) for SiO₂ and 2,615 mAh.g^-1 for SiO (Kim et al., 2011; Zhang et al., 2025). Additionally, SiOx exhibits superior rate performance compared to pure silicon, making it better suited for real-world applications requiring fast charging and discharging (Xie et al., 2025; Luo et al., 2023). Due to these notable benefits, SiOx anodes have found practical use in industry as the first commercially available silicon-based anode materials (Wang et al., 2022; Yang et al., 2015). However, it is crucial to underscore a noteworthy challenge faced by SiOx anodes, which pertains to their initial Coulombic efficiency (ICE). During the initial cycle, SiOx undergoes an irreversible reaction with Li⁺ ions, leading to the formation of Li₂O and Li_xSiO_y compounds (Zhu et al., 2023). This reaction not only diminishes the specific capacity but also results in a reduced Coulombic efficiency (ICE) for SiOx. Consequently, this necessitates a higher loading of active materials in the cathode, ultimately leading to a reduction in the overall energy density of the battery. Furthermore, SiOx exhibits intrinsically low electronic conductivity, which further diminishes its electrochemical activity (Zhang et al., 2019; Ji et al., 2024).

Numerous approaches are currently under investigation to address this challenge, including but not limited to the reduction in the size of SiOx particles (Kim et al., 2010), surface coating by Ni (Zhou et al., 2022), Mg, and Polyimide (Wagner et al., 2023; Lee et al., 2016), and the development of composites involving SiOx and other materials. Notably, composites consisting of SiOx with various carbon nanostructures (Zhou et al., 2023; Gaarud et al., 2025; Zhang et al., 2026; Yang Y. et al., 2020), such as nanospheres, nanofibers, and nanosheets, have garnered significant attention. This heightened interest is primarily due to the physical constraints imposed by carbon materials, which effectively mitigate the volume expansion of SiOx by enveloping SiOx particles. The addition of carbon brings the electronic conductivity of the whole material.

In parallel, recent advances in free-standing and binder-free silicon-based anodes have demonstrated improved cycling stability and electrode integrity through tailored interfacial engineering and self-supporting electrode architectures (Han et al., 2023; Gu et al., 2022). Building on previous work on Si, O, and C-based composite materials (Wagner et al., 2026), we designed novel SiOxCy xerogel composites incorporating biomass-derived carbon fibers to simultaneously address electronic conductivity, mechanical self-support, and electrochemical durability. The binder-free electrode design eliminates inactive masses, while a controlled XeF₂ surface fluorination was investigated to modify the electrode/electrolyte interface and improve initial capacity.

For that, SiOxCy/C self-supporting electrodes were successfully obtained by inserting in the resorcinol-formaldehyde-silica sol various organic fiber ratios and followed by gelation, air drying and thermal treatment in inert gas at temperatures up to 1,000 °C. By controlling the global texture and composition of the composite, we demonstrated the mechanical impact of the fibers and the robustness of monolithic materials directly implemented without formulation faced to lithium. Among these, the most promising approach in terms of capacity and durability involved fluorination with XeF₂ to mitigate the formation of the Solid Electrolyte Interphase (SEI) layer, leveraging the exceptional electrochemical and mechanical attributes of LiF, which is recognized as one of the most efficacious constituents of the SEI layer (Charles-B et al., 2019; Jaumann et al., 2015; Hatim et al., 2025).

2 Materials and methods

2.1 Synthesis of the electrode materials

Resorcinol (R), 99% purity, Alfa Aesar, formaldehyde (F), 37% purity, Sigma-Aldrich, tetraethoxysilane (TEOS) (T), 98% purity, Fischer Scientific, 3-aminopropyltriethoxysilane (APTES) (APT), 98% purity, Alfa Aesar and acetone (A), 99% purity, Fischer scientific were used as received for the synthesis of xerogels.

Series of silica-carbon xerogels (SiOxCy/C) were prepared from resorcinol (R), formaldehyde (F), tetraethoxysilane (TEOS) (T), 3-aminopropyltriethoxysilane (APTES) (APT) in the solvent acetone (A). The APTES acts as a basic catalyst for the gelation of RF-gels as well as a silica source for polycondensation with TEOS. The molar concentration of R: F: T: APT: A was maintained at 1: 2.939: 2.949: 0.219: 9.348. The precursors were mixed with the catalyst in acetone. After thorough mixing, the required quantity of the sol was impregnated into the calculated weight of organic kapok fibers (used as received). The fibers concentration was maintained at 2 wt% and 2.5 wt% concerning the mass of the reactant without the solvent. The gelation was performed at 60 °C and the gelation took place in approx. 40–45 min. After the gelation step, the gels were dried at 60 °C for 16 h in air to obtain a xerogel. The xerogels containing organic kapok fibers were later pyrolyzed under N₂ atmosphere at 800 °C or 1,000 °C (heating rate of 5 °C/min and holding for 1 h at 800 °C and 1,000 °C) to obtain carbon-silica monolithic composites xerogels (SiOxCy/C).

2.2 Fluorination

Xenon difluoride (XeF₂) is used as a fluorination agent. It is handled in a glove box under an inert atmosphere (argon) to prevent its decomposition and sublimation. XeF₂ fluorination takes place in a sealed Teflon reactor under an inert atmosphere. In a passivated nickel tray, 100 mg of the SiOxCy/C are deposited. The tray is placed in the center of the reactor and 100 mg of XeF₂ crystals are placed around them without any contact between them. The mass ratio of reaction is 1:1 wt%. The reactor is then hermetically sealed and once out of the glove box, stored in an oven at 120 °C for 15 h. The sample is kept in a glove box under an argon atmosphere with oxygen and humidity rates monitored at around 0 ppm. The material thus obtained is called SiOxCy/C-F.

The nomenclature of the samples is as follows: RFS-Fiber wt%-Pyrolysis temperature °C-F (if fluorinated).

2.3 Material characterizations

2.3.1 Physicochemical characterizations

The morphologies of the catalysts were investigated by scanning electron microscopy (SEM, MAIA3 TESCAN) operated at 3 kV. The surface area and pore characteristics of the catalysts were characterized by using nitrogen sorption isotherms at −196 C with a Micromeritics ASAP 2020 instrument. Before the measurements, the samples were degassed under 10 mmHg at 300 °C for 5 h. The N2-adsorption isotherms were used to determine the specific surface area (S_BET), total pore volume (V_T), and pore size distribution (PSD). S_BET was estimated using the Brunauer–Emmett–Teller (BET) theory. V_T was determined from the adsorption volume at the relative pressure of 0.99. PSD was calculated with the Density functional theory (DFT) method assuming pores with a slit geometry.

The structure of the prepared samples was determined by X-ray diffraction (XRD) using an X’pert pro diffractometer with a Cu Kα radiation source between 20° and 90°. The Raman investigations at room temperature were carried out using an inVia™ InSpect confocal Raman microscope equipped with a charge-coupled device (CCD) multichannel detector. The radiation source was a 532 nm argon laser line. Laser power was carefully adjusted to a low level (1 mW) and an acquisition time of 600 s was set to avoid decomposition of the sample under the beam.

The NMR analyses were performed with a 300 MHz Bruker Advance spectrometer. NMR measurements were carried out on ¹⁹F and ²⁹Si nuclei using Magic Angle Spinning (MAS) using a Bruker probe operating with 4 mm rotors (14 kHz for ²⁹Si and ¹⁹F). Chemical shifts are referenced to CF₃COOH for ¹⁹F. The reference is tetramethylsilane (TMS) for ²⁹Si data. A simple π/2 pulse sequence was used of the following type: pulse from π/2 for a certain time (PL1) then relax (D1). Experimental conditions according to the nuclei are, for ¹⁹F: 282 MHz, PL1 = 3 µs and D1 = 3 s and for ²⁹Si: 59.63 MHz, PL1 = 5.1 µs and D1 = 60 s.

2.3.2 Mechanical characterizations

The mechanical properties of composite with or without fibers were evaluated under the uniaxial unconfined compression mode using an Instron 5,967 testing machine equipped with a 30 kN load cell at room temperature. The load sensor is calibrated in class 0.5 until 1/250 of the maximum load. The samples are cut into the thickness of the monoliths, then the surface is polished with P240 sandpaper to obtain planar and parallel surfaces and the surface to diameter ratio around 1.51.

2.3.3 Electrochemical characterizations

The preparation of SiOxCy/C electrodes involved two distinct approaches. In the first method, slices were directly cut from the materials. Subsequently, these samples underwent a one-hour degassing process in a primary vacuum system to eliminate any potential sources of contamination. These electrodes are referred to as “monolithic electrodes”. The second approach, formulation, involved the creation of electrodes from a powder mixture consisting of our material, acetylene black and polyvinylidene fluoride (PVDF), with a weight ratio of 80:10:10. This mixture was dissolved in N-methyl-2-pyrrolidone (NMP) to form a suspension, which was then deposited onto the current collector. The RF-Silica-Fiber electrodes were subsequently assembled into 2032-type coin cells. Since the monolithic electrode consists solely of the cut composite, without any binder or carbon black additive, the mass loading is defined by the composite density. For the RFS-2.5–1,000 monolith (density of 0.134 g cm⁻³), this corresponds to a mass loading of 13.4 mg cm⁻² of monolithic active material. The electrolyte consisted of 1 mol.L^-1 LiPF₆ in a mixture of ethylene carbonate (EC), propylene carbonate (PC), and dimethyl carbonate (DMC), with a 1:1:3 volume ratio. The counter electrode utilized was composed of lithium metal (reference), while Celgard 2,400 and Whatman glass fiber GF/B materials served as separators.

For galvanostatic charge-discharge assessments, a potentiostat/galvanostat apparatus (MPG2 Biologic) was employed, with the potential window set at 0.005–3.0 V (vs. Li⁺/Li). Capacities were determined relative to the total mass of the active material, which is SiOx and not carbon, with a corresponding current density of 168 mA g^-1 (C/25 with a theoretical capacity considered as 4,200 mAh.g^-1). For these batteries, we calculate the Initial Coulombic Efficiency (ICE), defined as the ratio of the charge extracted during the first discharge to the charge input during the first charge. Additionally, we evaluate capacity retention, which is defined as the ratio of the battery’s capacity at a specific cycle compared to its initial capacity during the first cycle. Expressed as a percentage, it indicates the battery’s ability to maintain its capacity over repeated charging and discharging cycles.

3 Results and discussion

3.1 Physicochemical characterizations of the SiOxCy/C materials

The characteristics of the RF-Silica-Fiber composites and their SiOxCy/C pyrolyzed counterparts are provided in Table 1. RF Silica xerogels are in the form of monoliths having the shape of the synthesis mold. The drying of the gel leads to orange-colored materials whose density is more important as the contraction during gelation was important. This densification is limited by the presence of fibers. The xerogel RFS-0 has a density of 0.627 g cm^-3 and a density of 0.383 g cm^-3 and 0.374 g cm^-3 are obtained for the composite xerogels RFS-2 and RFS-2.5 respectively. A self-supporting black monolithic composite xerogels is obtained after the pyrolysis step in inert gas. The same shape is observed but with shrinkage. Here again, the fibers limit the contraction. RFS-0–800 xerogel has a density of 0.221 g cm^-3 and the densities for the RFS-2–800 and RFS-2.5–800 are 0.141 and 0.131 g cm^-3 respectively. The densities of the 1,000 °C pyrolyzed xerogels viz. RFS-2–1,000 and RFS-2.5–1,000 are 0.150 and 0.134 g cm^-3 respectively. Such densities indicate very high porous materials. For comparison, the density of massive carbons is about 2.2 g cm^-3 and that of SiO₂ is 2.65 g cm^-3. As the pyrolysis temperature is increased from 800 °C to 1,000 °C, there is more shrinkage leading to higher density. For the gels pyrolyzed at 800 °C under N₂ flow, the weight loss after pyrolysis for the RFS-0–800, RFS-2–800, and RFS-2.5–800 is 64, 79, and 81% respectively. The increase in weight loss (from 0% to 2.5% fiber) may be attributed to the higher presence of organic fibers. The composite xerogels pyrolyzed at 1,000 °C have a weight loss of 86% and 90% for RFS-2–1,000 and RFS-2.5–1,000 xerogels respectively. The mass loss during the pyrolysis in inert gas is comprised of between 64 wt% (in the material without fibers) to 90 wt% (2.5 wt% fiber) due to the elimination of the oxygenated groups and carbon atoms reorganization. An example of the non-pyrolyzed and its pyrolyzed counterpart composite xerogels is provided in Figure 1.

TABLE 1. Characteristics of non-pyrolyzed RF Silica fiber composite xerogels and pyrolyzed at 800 °C and 1,000 °C under N₂ flow. – not measured.

The SEM micrographs of the pyrolyzed xerogels with different fiber ratios or different temperatures are provided in Figure 2.

All the materials are macroporous. At 800 °C, the materials are composed of particles with diameters less than 100 nm, poorly differentiated, grouped in aggregates of few particles, bound together by covalent links. The hollow fibers are visible, and their diameters range from 3 to more than 20 µm. The xerogel adheres to the outer surface and appears to be present in the open fiber (Figures 2i,k). Zones of pure silica (EDX analysis) could also be observed in the form of “crumpled sheets”, in materials, with or without fibers. An example is visible in Figure 2b. When the temperature increased from 800 °C to 1,000 °C, the poorly differentiated particles tend to fuse and lead to aggregates larger than 250 nm in size. The texture has also been investigated by nitrogen sorption. The specific surface area (SSA) of the RFS-0–800 without fibers is 400 m²g^-1 (Table 1). This value is in the same order of magnitude with the addition of the fibers. The rise of the temperature from 800 °C to 1,000 °C leads to a decrease of both SSA, around 300 m² g^-1, linked to the increase of the particles size. Analysis of pore size distributions and isotherm shapes indicates that, in addition to macropores (observed by SEM and density value), mesopores (centered around 25 nm) and micropores are present. In addition, part of the porosity is difficult to access as it can be seen on the isotherm’s curves. The adsorption-desorption isotherms and the pore-size distributions of the pyrolyzed composite xerogels are provided in Supplementary Figure S1. The evaluation of the SiO₂ content in the composite xerogels has been done by calcifying the various carbonaceous materials in airflow at 1,000 °C (Table 1). The quantity of silicone is about the same, around 21 wt% for the composites pyrolyzed at 800 °C and around 22.5 wt% at 1,000 °C.

The X-ray diffraction (XRD) patterns of the pyrolyzed composite xerogels, prepared with varying fiber ratios or different temperatures, are presented in Figure 3A. These patterns display a single broad feature in the 2θ = 20°–26° range, attributable to the convolution of two amorphous contributions: the amorphous SiO₂ phase centered around 2θ = 22° and the amorphous carbon contribution typically located around 2θ = 23°–26°. The absence of a sharp, distinct peak at 2θ = 26° confirms that the carbon present in the composite is not graphitic but remains amorphous, consistent with the relatively low pyrolysis temperatures used (800 °C–1,000 °C) and with the disordered carbon nature evidenced by Raman spectroscopy (see below). A faint, broad signal around 2θ = 43° corresponds to the (100) reflection of disordered turbostratic carbon (Cao et al., 2018). As the fiber content increases, the intensity of this broad feature in the 20°–26° range becomes more pronounced. This enhancement is not due to the emergence of a new crystalline phase, but rather reflects the increased contribution of amorphous carbon derived from the pyrolyzed kapok fibers, which overlaps with and reinforces the SiO₂ signal. This behaviour is consistent with the literature on amorphous SiOx/C composites (Zhao et al., 2017; Li et al., 2014; Rahman et al., 2009) and collectively confirms the successful incorporation of fiber-derived amorphous carbon into the SiOxCy/C matrix.

In the Raman spectra (Figure 3B), the two prominent peaks observed at 1,360 cm^-1 and 1,590 cm^-1 correspond to the disordered carbon structures (D band) and the strong tangential stretching mode (G band) of carbon, respectively. The ratio of the intensities of these bands, I_D/I_G, is indicative of the degree of graphitization and the prevalence of defects in carbon-based materials. A ratio of I_D/I_G between 2.4 and 4.4 with an average crystallite size in the plane between 1 and 1.8 nm (calculated by Knight and White formula) suggests low organization of the carbon for all materials (Table 1), representative order-of-magnitude values for RF carbon aerogels found in the literature (Asset et al., 2018). During pyrolysis, the large quantities of oxygen released by the presence of kapok create defects in the carbon (carbon xerogel and carbonaceous fibers), especially at higher pyrolysis temperatures.

These ID/IG ratios confirm the disordered carbon nature of the composite, which is consistent with the moderate electronic conductivity expected for RF-derived carbon materials at these pyrolysis temperatures (Yang H. et al., 2020).

3.2 Mechanical characterizations

Before describing the mechanical response, it is worth clarifying the nature of the interfacial chemistry within the composite, as it directly governs its mechanical integrity. The Si–O–C network in the xerogel is built upon covalent bonds formed during the sol-gel step through Mannich-type reactions between the amine group of APTES, resorcinol and formaldehyde, leading to benzoxazine-type linkages between the organic and silica precursors, as previously evidenced by FTIR on materials synthesized by the same protocol (Wagner et al., 2026). After pyrolysis, the material does not behave as two separate silica and carbon networks but as an intimately bonded Si–O–C matrix, in which Si shares simultaneous bonds with O and C, as confirmed by the ²⁹Si NMR spectra (Figure 8). This covalent inter-network bonding underpins the cohesion of the monolith and is a prerequisite for the mechanical strength observed in the following compression tests. The kapok-derived fibers, located between, on the surface, and partly inside the matrix (Figure 2), then provide additional mechanical reinforcement by accommodating volume variations under compression.

During the lithiation/delithiation, the electrodes are subjected to volume variations and compression within the fixed volume of the cell. Stress-strain dependence is evaluated to observe the impact of the percentage of the fibres on mechanical properties. Uniaxial unconfined compression and loading-unloading compression cycles up to a nominal strain of 0.04% and 0.06% to observe the elastic properties are used to characterize the composites pyrolyzed at 1,000 °C.

The presence of fibers has a visible impact on the behavior of the materials under compression. While materials containing fibers exhibit deformation but retain some cohesion (Figures 4a–e), materials without fibers are brittle and fracture at the first stage of the compression test (Figures 4f–i), consistent with computational predictions of crack initiation and propagation in Si-based anode sheets under mechanical stress (Yuzuki et al., 2025).

This difference is related to large macropores between fibers allowing volume variation without damaging the material (observed by SEM Figure 2h) and to the flexibility of the fiber network under compression. 1,000 °C-pyrolyzed kapok fibers alone have been tested applying force perpendicular or parallel to the main fibers. We showed the total recovery of the initial volume after a compression test whatever the orientation of the fibers (Supplementary Figure S2) and a similar behavior of the evolution of the nominal strain vs. nominal stress of pyrolyzed kapok fibers mat under uniaxial compression for compression applied parallel to the main fibers, or perpendicular to the main fibers (Supplementary Figure S3). This behavior allows the fibers to rearrange themselves under the effect of compression and relaxation, thereby limiting fibers fractures in the composite thanks to their elasticity and the large empty spaces between them. Figure 5 shows evolution of the nominal strain vs. nominal stress of RFS-2.5–1,000, RFS-2–1,000 and RFS-0-1,000 during compression tests. Typical nominal stress-strain curves are obtained for materials with fibers. Stress-strain curves are divided into three parts described by Gibson and Ashby (Gibson, 1997). The first one consists of a linear elastic region at low strains allowing usually determination of compressive modulus E. The second one is a stress plateau corresponding to a progressive buckling of pore walls. Finally, the third one consists in a densification of the material at high strains where the stress increases abruptly. Without fibers, the behavior is different. The material with a higher density (0.345 g cm^-3) is more mechanically resistant (higher maximum stress) but brittle and fails in at very low strains (<0.1%), reaching a maximum stress of approximately 0.40 MPa. The compressive modulus E, calculated from the initial linear portion of the curve, can be significantly affected by a lack of parallelism between the contact surfaces. This defect induces strain localization at the very beginning of the test. To overcome this issue and enable a reliable comparison between materials, a chord modulus was therefore calculated between nominal strains of 0.04% and 0.06%and give the elastic recovery evaluated by the difference between the maximum and the minimum strain values at the second and the third loading-unloading cycle (ΔƐ) (Table 2).

TABLE 2. Mechanical characteristics of RFS-2.5–1,000, RFS-2–1,000 with RFS-0–1,000 composites.

According to the stress-strain curves, the modulus of the cord at 0.4% strain is 10 times higher for the material without fibers, i.e., 30.25 MPa compared to 4.86 and 2.84 for 0, 2, and 2.5% fibers, respectively. Increasing the fiber content from 2 wt% to 2.5wt% does not further improve mechanical efficiency. The same observation applies at 0.6% strain, with no change in classification. This fact is probably due to a texture with a different pore distribution in the two composites and the difference in their density, higher for the RFS 2 1,000 (0.15 g cm^-3) than for the RFS 2.5 1,000 (0.135 g cm^-3) as shown by Yu on carboxymethyl cellulose aerogels (Yu et al., 2025). The elastic properties obtained at 0.4% or 0.6% strain indicate that RFS-2–1,000 (above 2%) shows a higher elastic recovery than the others. The loading/unloading curves do not overlap. The behavior indicates an instantaneous component and a viscous component in all composites with fibers. None of them return to their original shape after the experiment, even those containing fibers, indicating that the matrix, unlike the fibers, is significantly impacted and controls degradation.

From an electrochemical perspective, this mechanical behavior has direct implications for cycling stability. The significantly lower chord modulus of fiber-containing composites (4.86 MPa for RFS-2–1,000 vs. 30.25 MPa for RFS-0–1,000) reflects a more compliant response capable of accommodating the volume changes associated with SiOx lithiation without structural failure. The elastic recovery confirmed by loading-unloading cycles suggests that the conductive network integrity is preserved throughout cycling, which directly contributes to the superior capacity retention observed for monolithic electrodes in the following section.

3.3 Electrochemical characterizations of SiOxCy/C materials

Figure 6A presents the galvanostatic discharge-charge profiles of the RFS-2.5–1,000 formulated powder sample over 100 cycles, with a voltage window set between 0.005 and 3.0 V at a constant current density of 168 mA g^-1. The first cycle reveals an initial discharge capacity of approximately 1,000 mAh.g^-1 and a charge capacity of about 939 mAh.g^-1, indicating an initial Coulombic efficiency of 94%. This efficiency shows the good reversibility of the electrochemical reactions within the material. Subsequent cycles display a notable difference from the first, with an increase in capacity that can be attributed to the stabilization of the SEI (Favors et al., 2014), eventually stabilizing into a consistent discharge capacity near 943 mAh.g^-1 over 100 cycles. This underscores the material’s stability and potential for long-term electrochemical applications.

The composite RFS-2.5–1,000 demonstrates the highest reversible capacity of 1,091 mAh g^-1 upon 5 cycles by comparison with others samples prepared without fiber and/or lower post-treatment temperatures, and exhibits superior capacity retention at 91% over a range of other RF-Silica-Fiber composites with varying fiber contents and pyrolysis temperatures, as summarized in Table 3.

TABLE 3. Electrochemical performance of RF-Silica-Fiber composites (formulated powders).

The table highlights the comparative electrochemical properties of different RF-Silica-Fiber composites, with the RFS-2.5–1,000 composite emerging as a superior material in terms of both energy storage capacity and its ability to maintain that capacity over time.

3.4 Impact of electrode shaping on electrochemical performance

On the sample giving the best electrochemical properties (RFS-2.5–1,000), we compared the behavior of an electrode manufactured from a formulation (Figure 6A) with one where the sample is simply cut and implemented directly into the cell (Figure 6B).

Electrochemical testing of the composite in a monolithic flexible form with a thickness of approximately 1 mm indicated a slight increase in capacities upon cycling. The initial Coulombic efficiency of 90% is marginally lower than the powder form 94%, which can be attributed to the larger exposed surface area of the monolith during the first cycle, promoting additional irreversible reactions between lithium ions and nano-SiO₂ within the composite, including the reduction of SiO₂ to Si and the formation of Li₂O (Xia et al., 2017). Nevertheless, the monolithic form demonstrates excellent long-term electrochemical stability, with a capacity retention exceeding 100% from the 10th cycle onward. This progressive increase beyond 100% reflects a gradual activation of previously inaccessible active mass, consistent with the high tortuosity of the porous network observed in the nitrogen sorption isotherms. At the 10th, 50th, and 100th cycles, the monolithic form delivers discharge capacities of 1,064, 1,068, and 1,070 mAh.g^-1, respectively, indicating minimal degradation over extended cycling. In comparison, the powder form delivers 943 mAh.g^-1 after 100 cycles, confirming the superior performance of the monolithic configuration. A detailed breakdown of reversible capacities at selected cycles for both forms is provided in Table 4.

TABLE 4. Electrochemical capacities of RFS-2.5–1,000 composite in powder and monolith forms at C/25.

This data highlights the comparative electrochemical stability and efficiency of the RFS-2.5–1,000 composite in both powder and monolithic forms, underscoring the monolithic form’s superior capacity retention and performance over extensive cycling. This result is probably linked to the texture of the monolith, where the particles are bound together by covalent links in a flexible monolith capable of accommodating volume variation thanks to its pore size distribution without loss of contact between covalently bonded particles thus maintaining electrode integrity during cycling. Notably, this enhanced performance is achieved without the need for additional binders or conductive carbon additives in the electrode formulation.

The stepped cycle testing profiles for the RFS-2.5–1,000 composite in both monolithic and powder forms (Figure 6C) reveal the behavior of the electrode’s capacity under varying current rates, defined as C/n. At a current of C/25, the monolithic anode of RFS-2.5–1,000 shows an impressive initial specific capacity of 1,151 mAh.g^-1. As the current intensity increases through stages C/12, C/6, C/3, to 2C/3 the specific capacities for the monolithic configuration demonstrate relative stability, with values approximately at 946 mAh.g^-1, 850 mAh.g^-1, 711 mAh.g^-1, and 498 mAh.g^-1, correspondingly. Conversely, the powdered form of RFS-2.5–1,000 starts with a specific capacity of 957 mAh.g^-1 at C/25 and decreases to 541 mAh.g^-1 at 2C/3. Upon reverting to the initial C/25 rate, the monolithic form’s capacity is partially restored to 977 mAh.g^-1, while the powder form sees a slight increase to 960 mAh.g^-1. These results indicate that it is possible to recover the initial capacity, suggesting that the electrode has not experienced degradation due to the use of high current levels.

In the comparative study, we included a Table 5 that juxtaposes the cycling performance of various materials from literature with our current work, both in powder and monolithic forms.

TABLE 5. Comparative analysis of electrochemical performance and carbon composition in electrode formulations.

While direct comparison is limited by differences in applied current densities, our results demonstrate a remarkable initial Coulombic efficiency of 94% (powder) and 90% (monolith) significantly higher than literature values along with competitive capacity retention. Notably, the binder-free monolithic configuration achieves 1,070 mAh.g^-1 after 100 cycles with zero carbon additive in the electrode formulation, highlighting the advantage of this self-supporting design.

3.5 Impact of surface fluorination

3.5.1 Physicochemical characterizations

The study was carried out on monolith materials as it is the most promising shaping for future electrodes avoiding PFAs binders (Savvid et al., 2024). A controlled surface fluorination was carried out by xenon difluoride as a fluorinating agent on the sample pyrolyzed at 1,000 °C with a fiber content of 2.5% which gave the best electrochemical properties (Table 4).

The X-ray diffraction (XRD) pattern of the fluorinated composite RFS-2.5–1,000, depicted in Supplementary Figure S4, confirms that fluorination is solely on the surface. The crystal structure of the material remains unaltered, displaying a significant broad peak at 2θ = 22° associated with silica (SiO₂) and amorphous carbon at 2θ = 24° and around 2θ = 43° indicative of the carbon component.

The Energy-Dispersive X-ray (EDX) spectroscopy mapping, depicted in Figure 7, substantiates the homogeneous presence of fluorine on the RFS-2.5–1,000 composite’s surface. The mappings, executed across various areas, consistently reveal a homogenous spread of fluorine and silicon elements across the composite’s surface. This uniform distribution is indicative of a well-controlled fluorination process. Silicon’s presence is equally pervasive, suggesting that the composite’s matrix remains intact post-fluorination, further corroborating the selective surface modification.

The ¹⁹F and ²⁹Si solid state NMR spectra for the fluorinated RFS-2.5–1,000 sample are shown in Figure 8. NMR provides information on the local environments of silicon and fluorine atoms in the material. The ¹⁹F NMR spectrum reveals a complex fluorine environment with two principal chemical shifts: a major one is located at approximately −58.5 ppm, and a minor one is at −78 ppm. These shifts correspond to the molecular structures O_2/2 SiF₃ and O_4/2 SiF respectively (Lataste et al., 2008; Lataste et al., 2009)indicative of the various fluorine coordination states within the material. Notably, environments indicative of CF bonds between carbon and fluorine were not observed, which aligns with existing literature stating that the fluorination of graphitic carbon typically occurs at elevated temperatures, starting from 250 °C (Chamssedine et al., 2007). However, our fluorination process was conducted at a lower temperature of 120 °C, suggesting a rationale for the absence of CF environments in our samples. This assignment is further supported by thermodynamic and chemical considerations. Si–F bonds are among the strongest single bonds, with bond dissociation energies reported around 565 kJ mol^-1, higher than those of C–F bonds, generally reported around 485 kJ mol^-1 (Luo, 2007). This higher fluorine affinity of silicon-containing environments is consistent with the well-known spontaneous reaction of XeF₂ with silicon surfaces, leading to volatile SiF₄ species (Houle, 1987). Therefore, under the present mild fluorination conditions, XeF₂ is expected to react preferentially with silicon/oxygen-rich domains rather than with the carbon framework. Beyond structural assignment, the formation of Si–F/fluorosilicate surface species may also partially passivate reactive SiOx surface defects, such as under-coordinated silicon atoms, residual silanol groups, or disordered oxygen-rich environments, which are known to promote electrolyte decomposition and irreversible lithium consumption during the first cycles (Cao et al., 2019). The improved initial electrochemical performance observed after XeF₂ treatment may therefore arise from a combination of enhanced electrolyte accessibility within the porous monolith and selective passivation of reactive SiOx surface sites. The structure of the SiO₄ tetrahedra is visible in the ²⁹Si NMR spectrum with a major peak located at about −113 ppm for the composite fluorinated or not and the fluorine peak is not visible on the spectrum of the fluorinated sample.

3.5.2 Electrochemical performance

The data presented in Table 6 and illustrated in Figure 9 underscores the significant impact of fluorination on the electrochemical performance of the RFS-2.5–1000-F monolithic composite. The fluorinated material displays an exceptional initial discharge capacity of approximately 2,381 mAh.g^-1, while the subsequent charge capacity is about 2034 mAh.g^-1, resulting in an initial Coulombic efficiency of 85%. This initial drop in Coulombic efficiency is likely due to the establishment of new chemical bonds between fluorine, silicon, and lithium ions as the fluorination introduces additional active sites for lithium interaction. This new surface chemistry allows better and faster access of the electrolyte into the smallest pores on the surface of the whole composite, leading to higher performance immediately, unlike non-fluorinated composites which require cycling to reach the entire surface (visible with a retention capacity of over 100%). The difference in surface area accessible to the electrolyte during the first cycles, with or without fluorination, leads to greater lithium consumption in the fluorinated material.

TABLE 6. Electrochemical performance of the RFS-2.5–1,000 and RFS-2.5–1000-F monolithic composites.

As cycling progresses, the capacities at the 10th, 50th, and 100th cycles for the RFS-2.5–1000-F demonstrate a notable decline from the initial values, yet they maintain a robust performance compared to the non-fluorinated composite. By the 10th cycle, the charge capacity decreases to 1,662 mAh.g^-1 and the discharge capacity to 2,124 mAh.g^-1. This trend continues at the 50th cycle with a charge capacity of 1,380 mAh.g^-1 and discharge capacity of 2049 mAh.g^-1, followed by the 100th cycle showing a charge capacity of 1,390 mAh.g^-1and discharge capacity of 1786 mAh.g^-1. Despite the reduction, these values indicate that the material retains a substantial capacity, suggesting that the active material undergoes stabilization over time.

The results presented here demonstrate two distinct and complementary effects of XeF₂ surface fluorination. First, fluorination significantly enhances the initial electrochemical accessibility of the porous network, yielding an exceptional initial discharge capacity of ∼2,380 mAh.g^-1 more than twice that of the non-fluorinated monolith. This improvement is attributed to the modified surface chemistry, which promotes faster electrolyte wetting of the smallest pores from the very first cycle, whereas non-fluorinated composites require progressive activation over multiple cycles to reach comparable surface areas. Second, the capacity retention at 100 cycles (75%) remains lower than that of the non-fluorinated monolith (>100%), indicating that the current fluorination conditions introduce additional irreversible reactions that accelerate capacity fading. This trade-off between initial capacity gain and long-term stability is a known challenge in fluorinated electrode systems and is likely sensitive to the XeF₂ dose, reaction temperature, and exposure time. Optimizing these parameters represents a clear and promising avenue to fully leverage the benefits of surface fluorination while preserving cycling durability.

4 Conclusion

We developed a novel multiscale porous composite based on an intimate Si, O, and C matrix reinforced by biomass-derived kapok fibers, enabling self-supporting flexible electrodes that can be directly implemented in lithium-ion cells by simple cutting. This binder-free approach eliminates inactive masses and simplifies cell assembly, while the fiber network provides the mechanical compliance necessary to accommodate SiOx volume changes during cycling. Mechanical characterization confirms that fiber incorporation fundamentally changes the composite response from brittle fracture to elastic deformation, with chord modulus values dropping from 30.25 MPa to below 5 MPa, directly contributing to the superior cycling stability of monolithic electrodes. Electrochemically, the monolithic form consistently outperforms formulated powder electrodes, maintaining a discharge capacity of 1,070 mAh.g^-1over 100 cycles with a retention exceeding 100%, attributed to the progressive activation of the tortuous pore network. Surface fluorination using XeF₂ introduces a modified surface chemistry that dramatically improves initial electrolyte accessibility, yielding an initial discharge capacity of ∼2,380 mAh.g^-1 among the highest reported for SiOx-based binder-free anodes. The fluorination effect is therefore attributed primarily to selective modification of silicon/oxygen-rich surface domains rather than carbon fluorination, with Si–F/fluorosilicate species likely contributing to both improved initial electrolyte accessibility and partial passivation of reactive SiOx surface defects. While capacity retention at 100 cycles currently stands at 75%, this result establishes a clear proof of concept and identifies fluorination conditions as the key parameter to optimize in future work.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fbael.2026.1838088/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Desmond Edem Primus Klenam, University of the Witwatersrand, South Africa

Reviewed by: Xiao-Dong Zhu, Qingdao University of Science and Technology, China

Xiang Han, Nanjing Forestry University, China