Effect of Short-Chain Polymer Binders on the Mechanical and Electrochemical Performance of Silicon Anodes

Abstract

Polymer binders are crucial components in providing both mechanical support and chemical stability to the structure of porous Li-ion electrodes. Particularly in silicon anodes, the active material undergoes substantial volume expansion of up to 275%. Due to the mechanical constraint of the current collector, these silicon materials tend to expand in the normal direction while exhibiting substantial particle rearrangement and plastic deformation. Conventional rigid binders such as polyacrylic acid (PAA) and polyimide (PI), while providing satisfactory initial capacity, do not eliminate diminished long-term performance. Our research attempts to develop binder formulations that can accommodate sufficient flexibility for the substantial volume changes of silicon particles. Specifically, we explore the use of short-chain polymer binders and a strategic blend of binders with different molecular weights. Experiments have demonstrated that cells combining both long- and short-chain PAA binders delivered an initial capacity of 2200 mAh/g at a 0.1C rate, compared to 1700 mAh/g for pristine PAA cells. Initial work indicated that shorter polymer chains might compromise the adhesion to the current collector, so we developed a multilayer anode (MLA) structure to mitigate this issue. Nevertheless, at this early stage of development, there was no observed increase in cycling performance for the MLA electrodes.

1. Introduction

Silicon has emerged as a highly promising alternative anode material for Li-ion batteries. It has a theoretical capacity of 3578 mAh/g, which is nearly ten times that of graphite at 372 mAh/g [1,2]. Its adoption is hindered by severe volume expansion: fully lithiated silicon (Li₁₅Si₄) swells by ≈275% at room temperature, compared to just 13% for graphite (LiC₆) [3]. Such drastic expansion induces particle fracture, electrode pulverization, and unstable solid electrolyte interphase (SEI) formation, leading to rapid capacity fade and poor cycle life.

Figure 1 illustrates that when active particles increase in volume, particle rearrangement and translation necessarily happen due to the rigidity of the current collector. The figure further presents two types of connections required for proper binder functionality: adhesive and cohesive. Adhesive connections occur when a binder attaches to active materials or the current collector, a process partially determined by distinct functional groups. Cohesive connections occur when binder molecules connect to other binder molecules through polymer chain interactions. During volume-change cycles, the silicon electrode requires the binder to elongate on the order of particle sizes (green arrow in Figure 1), and the bonds must be able to repair or successfully reattach themselves in case of bond release or failure. If connections are too rigid, meaning that the large adhesion and cohesion limits do not allow for strain that will dissipate the accumulated stress in the electrode, this could instead lead to fracture and irreversible structural changes.

Enhancing the binder flexibility and self-healing capabilities of polymer binders is essential for improving the mechanical integrity and long-term cycling performance of silicon electrodes. To enhance flexibility, one approach is to select polymers with intrinsically flexible backbones, such as polyacrylic acid (PAA), polyvinylidene fluoride (PVDF), and styrene–butadiene rubber (SBR), which offer both elasticity and processability [4,5]. However, PVDF is limited by weak van der Waals interactions; therefore, it cannot maintain Si electrode integrity, leading to a rapid capacity fade to 1000 mAh/g after 10 cycles [6,7]. On the other hand, PAA films tend to be brittle when fully dried, limiting their ability to accommodate extreme Si volume fluctuations without cracking [6]. The PAA binder’s performance is also sensitive to slurry pH [6]. Sun and Wheeler [8] found that in the first four cycles at a C/10 rate, the capacity was around 1650 mAh/g, which was improved with a pH of 4.5 (2600 mAh/g).

Another approach is to reduce the polymer chain length or molecular weight. The structure of polymer chains, such as chain length, significantly impacts a binder’s strength, stiffness, and flexibility. Shorter chains exhibit fewer entanglements and greater segmental mobility, enabling facile bond breakage/reformation under stress. For instance, Kasinathan et al. found that the PAA binder with the lowest molecular weight (250 kDa), compared to that with the highest weight (1250 kDa), achieved the best electrochemical and mechanical performance [9]. Yet, excessive chain shortening can weaken adhesion to the current collector, compromising electrode cohesion [10]. To address this adhesion challenge, the design of multilayer silicon anodes (MLA) has emerged as a promising structural strategy. By introducing interfacial/substrate layers, such as carbon nanotube (CNT) scaffolds or graphene interlayers, between the silicon and current collector, researchers have demonstrated the improved mechanical accommodation of volume changes and enhanced charge transfer during cycling [11].

For facilitating self-healing behavior, one approach is to use stronger and reversible interchain interactions, such as hydrogen bonding, electrostatic forces, metal–ligand coordination, disulfide bonds, and boronate ester bonds [12]. These dynamic bonds enable polymer networks to autonomously reconfigure and recover after mechanical disruption, thereby improving structural resilience and electrode lifespan.

In contrast, while native PVDF is a well-known and often-used solvent-soluble binder, it is known not to work well for silicon electrodes. Some research groups have explored chemical modifications to this binder and the silicon surface to control adhesion. Yoo et al. achieved improved electrode performance through the chemical modification of PVDF with hydroxyl groups, leading to a more homogeneous PVDF distribution [1]. Comparatively, Huang et al. introduced a functional buffer layer on silicon particles, enhancing solid–electrolyte interface (SEI) layer formation and maintaining a capacity of around 2000 mAh/g after 200 cycles while using a traditional PVDF binder [2].

Additionally, blending polymers into hybrid or composite networks can induce synergistic effects, enhancing flexibility, interfacial adhesion, and electrochemical performance simultaneously. Such multiphase systems can integrate the mechanical softness of flexible polymers with the robustness or functionality of more rigid components, offering a promising route toward multifunctional binder materials for next-generation batteries [8].

This study aims to achieve three main objectives related to the expansion of silicon particles. First, we shorten commercial PVDF chains and introduce carboxyl end groups via dehydrofluorination and oxidative cleavage. Second, we evaluate the electrochemical and mechanical performance of low-molecular weight polymer binders to see if they adequately balance flexibility and adhesion. Third, we fabricate and test multilayer silicon anodes by incorporating an engineered adhesion layer to maintain interfacial cohesion while leveraging the benefits of low-molecular weight binder systems.

To address the first objective of this work, PVDF is chemically modified. Although controlling the synthesis of PVDF binder from monomeric feedstock is ideal, safely conducting the manufacturing process is challenging due to the high-temperature and high-pressure reaction involving reactive and toxic vinylidene difluoride (VDF) gas precursor. Therefore, we opted to chemically modify off-the-shelf PVDF products. This modification involves two key reactions: dehydrofluorination and oxidative cleavage. These reactions modify PVDF by shortening its polymer chains and incorporating carboxyl functional groups at the chain termini. In the dehydrofluorination step, hydrogen and fluorine atoms are removed from the PVDF polymer backbone, resulting in the formation of carbon–carbon double bonds (C=C). In the subsequent oxidative cleavage reaction achieved through treatment with ozone and hydrogen peroxide, these unsaturated sites undergo bond scission, introducing oxygen-containing functional groups such as carboxyl (-COOH) [3,13,14,15]. The reactions can be summarized as shown in Figure 2 [16,17,18].

For the second objective, we explored the use of low-molecular weight polymer binders, including poly(acrylic acid) (PAA) and Jeffamine^TM D-2000 (Huntsman Corporation, The Woodlands, TX, USA), as well as novel binders such as PVDF-HFP, polyacrylonitrile (PAN), and polyimide (PI). These binders differ in structure, molecular weight, melting point, and glass transition temperature, potentially offering enhanced flexibility to accommodate the significant volume expansion of silicon particles. In

Table 1. Structure, molecular weight, glass transition, and melting point of different binders.

Name	Molecular Structure	Molecular Weight (kDa)	Glass Transition Point (°C)	Melting Point (°C)
PAA(2000)		2	103 [19]	116 [19]
PVDF		534	−35 [20]	177 [20]
JeffamineTM		2	<25 [21,22]	38 [21,22]
PVDF-HFP		400	−35 [23]	143 [23]
PAN		70	95 [24]	300 [24]
PI		500	300 [25]	350 [25]

, the structures and properties of those binders are listed. The glass transition point is the temperature at which a polymer transitions from a hard, glassy state to a soft, rubbery state. For example, at temperatures below the glass transition point, the polymer chains are in a rigid state, exhibiting very limited mobility. In this state, the material is rigid and brittle because the molecular chains are locked in place, preventing them from sliding past one another. This lack of mobility contributes to the stiffness and brittleness of the material. Similarly, the melting point of a binder is essential for determining its thermal processing capabilities, stability, and safety under operational conditions.

Lastly, for the third objective, we designed an adhesion layer which is introduced between the bulk electrode material and the current collector, as shown in Figure 3. This adhesion layer is composed of long-chain polymer binder, such as PVDF and PAA, and carbon black, which enhances adhesion while also maintaining electrical conductivity.

2. Materials and Methods

2.1. Chemical Sources

Carbon black (C45) was obtained from Timcal (Lac-des-Îles, Terrebonne, QC, Canada). Silicon nanoparticles (~100 nm) were purchased from Nanostructured & Amorphous Materials, Inc. (Houston, TX, USA). The following were obtained from Sigma-Aldrich (St. Louis, MO, USA): polyacrylic acid (PAA, MW = 450 kDa and 2 kDa), polyvinylidene fluoride (PVDF, MW = 534 kDa), polyacrylonitrile (PAN, MW = 70 kDa), poly(vinylidene fluoride-co-hexafluoropropylene (PVDF-HFP, MW = 400 kDa), Jeffamine^TM (MW = 2kDa), LiOH, NaOH, NH₄OH, NMP, CH₂Cl₂, and H₂O₂. Polyimide (PI, MW = 500 kDa) (P84 SG Tecapowder) was sourced from Ensinger GmbH (Lenzing, Austria). Annealed copper sheets were purchased from Nimrod Copper Co. (Springfield, VA, USA). Electrolyte (1.0 M LPF₆ in EC and DEC with a volume ratio of 50:50) and fluoroethylene carbonate (FEC) were purchased from Sigma-Aldrich.

2.2. Binder Preparation

The binders were prepared by magnetically stirring the binder powders with deionized water (for PAA) or NMP (for PVDF, PI, PAN, Jeffamine^TM, PVDF-HFP) at a weight ratio of 1:12 for 6 h at room temperature. Each container was covered with a lid to minimize solvent evaporation.

2.3. Si Electrode Preparation

The fabricated silicon electrodes were composed of 60 wt% silicon, 30 wt% binder, and 10 wt% carbon black (C45). Dry silicon and C45 powders were ground with a mortar and pestle for 20 min. The binder solution and ground particles were then mixed in a sealed jar using a high-speed rotating homogenizer (Algimax GX 300, MONITEX Industrial Co., Ltd., New Taipei City, Taiwan) for 24 s. For the coating process, a 7 cm × 9 cm copper sheet was first cleaned using isopropyl alcohol. The electrode slurry was then coated onto the copper sheet using a coating bar to a thickness of 75 μm. The anode was subsequently dried in an oven at 100 °C for 6 h and calendered to achieve a 50% porosity.

The selected anode composition (60% silicon, 30% binder, and 10% carbon black) was based on prior work and in recognition of the challenge of using only silicon as the active material, as compared to silicon–graphite mixtures. We are aware that commercial electrodes would be preferred at a lower optimized amount of inerts, such as around 5–15% binder. However, a larger fraction allows us to evaluate binder performance in a laboratory setting. Nguyen, Yoon, Seo, Guduru, and Lucht [6] noted that when binder content is reduced to more realistic levels (~10%), the relative trends in performance remain similar but overall stability decreases, especially for less effective binders like PVDF. This demonstrates that the high-binder formulation is a deliberately exaggerated but useful test to compare binder chemistries under controlled yet challenging conditions.

2.4. Cell Assembly and Electrochemical Characterization

Half cells were assembled using a silicon composite electrode as the working electrode, a lithium metal counter electrode, a microporous polypropylene separator (Celgard 2325, Celgard, LLC, Charlotte, NC, USA), and an electrolyte composed of 1.0 M LiPF₆ in a solvent mixture: EC:DEC:FEC 45:45:10 (v/v). All cells were allowed to soak in the electrolyte overnight to ensure thorough penetration of the electrodes. Cycling was performed using a Maccor 4300 battery tester (Maccor, Inc., Tulsa, OK, USA) at room temperature; each cell underwent 3 formation cycles at a C/20 rate, followed by 4 lithiation cycles at rates of C/10, C/5, C/3, 1C, and 2C, coupled with a C/3 delithiation rate. The voltage window spanned from 0.05 to 1.5 V. The sequence of cycles is then repeated a second time, followed by a concluding set of C/10 cycles, leading to a total of 44 cycles after formation. All electrochemical measurements were conducted at room temperature.

2.5. Peeling Test

Peeling tests were conducted using an Instron 3343 stress–strain tester to evaluate the adhesion/cohesion of the binders. The laminate side of the electrode, measuring 30 mm × 54 mm, was attached to a 20 mm-wide strip of aluminum-backed adhesive tape. The electrode portion underneath the adhesive tape was removed from its current collector by pulling the tape at a constant displacement rate of 500 mm/min, as described more fully in ref. [26]. The applied force was measured, and load/displacement plots were generated to analyze the results.

2.6. Nano-Indentation Test

Nano-indentation was performed using a Leco Micro Hardness Tester to apply a controlled force of 100 g. The force was applied via a sharp indenter tip, resulting in a small, precise impression on the surface of the electrode. The depth of the indentation was accurately measured using a Keyence VHX-7000 microscope ((Keyence Corporation of America, Itasca, IL, USA)).

2.7. FTIR Test

Fourier-transform infrared (FTIR) spectroscopy analyses were performed to characterize the chemical structure of the modified solid powders. The FTIR spectra were recorded using a Thermo Scientific Nicolett iS50 from Thermo Fisher Scientific. Samples were prepared by grinding the powders and pressing them into pellets. The FTIR spectra were collected over a wavenumber range of 400 to 4000 cm⁻¹, with a resolution of 4 cm⁻¹.

2.8. XPS Test

X-ray photoelectron spectroscopy (XPS) analyses were conducted to investigate the elemental composition and chemical states of the elements present in the synthesized solid powders. The measurements were performed using a Thermo Scientific k-Alpha XPS ((Thermo Fisher Scientific Inc., Waltham, MA, USA). Samples were prepared by dispersing the powders on double-sided tape mounted on XPS sample holders. The binding energy scale was calibrated using the C 1s peak at 284.8 eV as a reference. The data analysis included peak fitting using CasaXPS, where peaks were fitted with a combination of Gaussian–Lorentzian functions, and a Shirley background was subtracted to determine the elemental composition and oxidation states.

2.9. DSC Test

Differential scanning calorimetry (DSC) measurements were carried out to investigate the thermal properties of the modified materials, including their melting and crystallization temperatures. The analyses were performed using a DSC Q2000, TA Instruments (TA Instruments, New Castle, DE, USA). Approximately 5–10 mg of each sample was sealed in an aluminum pan and subjected to a controlled temperature program under a nitrogen atmosphere to prevent oxidative degradation. The temperature program typically involved heating the sample from 100 °C to 200 °C at a heating rate of 2 °C/min, followed by a cooling down process at the same rate.

2.10. Dehydrofluorination and Oxidative Cleavage Steps

The dehydrofluorination steps were conducted as follows: A total of 1.6 g of PVDF powder was dissolved in 18.4 g of NMP and stirred at 70 °C for 30 min in an N₂ atmosphere. Subsequently, 0.64 g of 5 wt% sodium hydroxide solution was added to the PVDF solution. The mixture was stirred at 180 rpm and maintained at 70 °C for durations of 5, 20, and 60 min. This process resulted in a coagulated emulsion with a brown color. The solid products were then rinsed with ethanol and deionized water and subsequently dried in a vacuum at 70 °C overnight.

The oxidative cleavage steps were performed as follows: At the conclusion of the dehydrofluorination steps, the resulting products were dissolved in CH₂Cl₂ at a weight ratio of 1:250. The glass reactor vessel was submerged in dry ice. Ozone was bubbled through the enclosed solution for 10 min at a flow rate of 4 standard cubic feet per h (SCFH), followed by stirring for 60 min. Nitrogen was then purged through the reactor to eliminate excess ozone. The dry ice was then removed, allowing the reactor to return to room temperature. Next, 0.5 g of 50 wt% H₂O₂ solution was added to the reactor, and the reaction was then continued for an additional 60 min. The final solid products were collected by filtering with additional deionized water.

2.11. Post-Mortem Analysis

Post-mortem analysis was performed to evaluate the integrity and degradation of electrode materials, with a specific focus on delamination of the anode surfaces. Cells after cycling were disassembled in an inert atmosphere to prevent additional degradation from exposure to air. The electrodes were then separated from the cell.

2.12. Multilayer Anode Preparation

Two different multilayer anodes were prepared, one made of PVDF and the other made of PAA binders. The binder (dissolved in solvent) was further mixed with carbon black in a mass ratio of 1:0.25, respectively. The mixture was magnetically stirred for 1 h at room temperature. The resulting slurry was then coated to a thickness of 50 μm using a coating bar on the copper sheet, which was previously sanded and treated with 10 wt% oxalic acid in water to remove any potential copper oxide layers. The adhesion layer was then dried at 100 °C for 3 h to eliminate any moisture. Subsequently, a bulk electrode layer comprising a binder, silicon, and carbon black in water was coated on top of the adhesion layer. This electrode was further dried in an oven at 100 °C for 6 h.

3. Results and Discussion

3.1. Modified PVDF

3.1.1. Analysis of the Dehydrofluorination and Oxidative Cleavage Steps

To examine the success in modifying PVDF through dehydrofluorination and oxidative cleavage steps, FTIR, XPS, and DSC techniques were used. FTIR was used to identify functional groups through molecular vibrations. XPS provided insights into the elemental composition and chemical states at the material’s surface. DSC was useful for investigating the thermal stability and composition of materials through their melting and crystallization behaviors.

Names are assigned to the dehydrofluorination products based on their respective reaction times; thus, m5, m20, and m60 represent products obtained after 5, 20, and 60 min of the dehydrofluorination reaction, respectively. Similarly, the final cleavage oxidation products are named f5, f20, and f60, showing which intermediate products were used to obtain the final product.

Figure 4 shows the FTIR spectra of both intermediate samples (m5, m20, and m60) and final products (f5, f20, and f60). Pristine PVDF serves as a baseline for comparison. In the pristine PVDF, there is an absence of C=C and -COOH peaks, which confirms that the observed new peaks in the modified samples are indeed due to the chemical treatments applied. For the intermediate samples, peaks in the region around 1640 cm⁻¹ are typically associated with C=C stretching vibrations, suggesting the formation of double bonds during the dehydrofluorination process in these samples [27]. In addition, the m60 sample exhibits the most intensive peak, indicating a higher concentration of C=C bonds than other samples. In the final products, the peak observed at 1720 cm⁻¹ is attributed to -COOH [26]. The f60 sample also demonstrates the highest intensity, suggesting that it contains the greatest amount of -COOH.

Figure 5 shows the fitted XPS spectra and possible carbon states within each sample. The functional group peak area percentages are given in

Table 2. XPS carbon peak areas (relative percentage) of dehydrofluorination products and final products.

	CH2 (%)	CF2 (%)	C=C (%)	-COOH (%)	Unknown (%)
PVDF	54.1	39.9	0.0	0.0	6.0
m5	39.1	51.3	4.7	0.0	5.0
m20	33.0	52.8	6.9	0.0	7.2
m60	45.8	32.6	14.8	0.0	6.8
f5	23.5	53.6	7.1	5.3	10.5
f20	33.4	56.1	0.0	7.7	2.8
f60	26.4	32.0	4.2	15.0	22.3

. Note that these percentages do not correspond to atomic ratios, although trends from one sample to the next can indicate relative amounts of these carbon groups. Pristine PVDF shows no presence of a C=C bond or a -COOH group. The 6% of unknown groups could indicate the existence of defects in the polymer.

C=C bonds were observed in all m5, m20, and m60 samples, indicating that the dehydrofluorination step was successful. An increase in the C=C ratio is observed as the reaction time increases. For instance, the C=C bond increases from 4.7% to 14.8% as the reaction time increases from 5 min to 60 min. Longer reaction times allow for more extensive HF elimination, thus creating more double bonds.

It was expected that the percentages of these two peaks would decrease the peaks of CH₂ and CF₂ and maintain the same ratio as observed in pristine PVDF for HF formation, according to the reaction described above. However, Table 2 suggests that the ratio of C=C to -COOH varies in each sample. With extended reaction times, other reaction pathways may emerge, potentially causing an imbalance between the formation of new C=C bonds and the reduction of CF₂ and CH₂ groups. Moreover, VDF monomer additions in the polymerization process can occur head-to-head instead of head-to-tail, leading to 5–10% regioirregular defects in the pristine PVDF material [28], which can contribute to these observed variations in our products.

Following the dehydrofluorination step, oxidative cleavage aims to break the carbon-carbon bonds, particularly at unsaturated sites, and introduce oxygen-containing groups like carboxyl (-COOH). These then become the terminal groups on the newly shortened polymers. In all the final products, an increment of -COOH is observed, especially in f60 (15.0%). This indicates the success of cleavage and functionalization of the polymer. The increase in -COOH content from 5.3% in f5 to 15.0% in f60 is consistent with the trend observed in the intermediate products. However, an unidentified peak at a binding energy of 296 eV was observed in sample f60. Additionally, the C=C groups decrease in both m20 and m60 samples compared to their intermediates, confirming that the oxidative process involves the conversion of double bonds into carboxyl functional groups.

The C=C content in f5 is slightly higher than that in m5. It is possible that the oxidative cleavage has low selectivity, especially in the case of m5, where the amount of C=C is low. The oxidative cleavage might have different reaction rates for different double bonds, depending on their accessibility to the oxidizing agent. For instance, internal double bonds might be less accessible and thus less likely to react compared to more exposed double bonds at the ends of polymer chains.

DSC Test

In Figure 6, the DSC results for the final cleavage oxidation products (f5, f20, and f60) and the pristine PVDF are shown. The pristine PVDF exhibits a sharp melting peak, which is indicative of its semi-crystalline nature and uniform crystalline structure. In contrast, the modified samples (f5, f20, and f60) display melting peaks of reduced intensity and exhibit a slightly higher melting temperature compared to pristine PVDF.

The slightly higher melting points could be related to changes in the polymer chain lengths. Shorter chains typically result in a decreased melting temperature due to reduced entanglement and diminished intermolecular forces. However, if the chain scission process is non-uniform, it could lead to a distribution of molecular weights that includes chains close to the original length. These chains could sustain a similar melting temperature to the original polymer state. Additionally, the melting point may not be as sensitive to changes in chain length, contributing to a broader melting range and minimal temperature shifts.

Furthermore, the process of chain scission might introduce other structural modifications to the polymer chains, such as varying degrees of crystallinity or even cross-linking effects, which are unintended but could occur. This variation in molecular structure could lead to heterogeneity in crystal sizes and accounts for the diverse melting behaviors seen across the modified samples.

Moreover, while the incorporation of -COOH functional end groups into the polymer matrix is minimal, it could still contribute to reinforcing the polymer structure through enhanced intermolecular interactions via additional hydrogen bonding, thereby elevating the melting temperature.

Upon cooling, the crystallization peaks for the modified PVDF samples were observed to be less intense than those of the pristine PVDF, with the f60 sample displaying a particularly notable trend. The f60 crystallization peak emerges at a higher temperature compared to the f5 and f20 samples. This suggests that the f60 sample may begin to crystallize more readily. The propensity for earlier crystallization in the f60 sample could be linked to its shorter polymer chains, which are likely to possess greater mobility, thereby enabling a more efficient transition into an ordered, crystalline arrangement. Additionally, the reduction in chain entanglement for the f60 sample may simplify the alignment and packing of the chains, further promoting crystallization at elevated temperatures.

3.1.2. Testing the Products from the Dehydrofluorination and Oxidative Cleavage Steps

The mechanical performance of Si electrodes and the cycling performance of Si cells were carried out to determine the suitability of multiple binders, including modified PVDF binder products obtained through dehydrofluorination and oxidative cleavage steps. Names were assigned to control samples and modified PVDF samples as follows: The Si electrodes are made using PAA, LiOH-PAA(4.5) derived in previous work [29], and pristine PVDF as binders and are respectively labeled as PAA, LiOH-PAA(4.5), and PVDF-PAA for control samples. LiOH-PAA(4.5) is known to work well, and PVDF-PAA will be used as the control and comparison for PVDF-modified samples. The other samples, which are the final cleavage oxidation products, are named f5-PAA, f20-PAA, and f60-PAA. The PAA label indicates that PAA was combined with modified PVDF at a weight ratio of 1:1 to enhance the performance as a binder for Si electrodes. The addition of PAA is intended to leverage PAA’s excellent adhesive properties and its ability to interact favorably with the electrode material, enhancing ion exchange and ionic conductivity.

Mechanical Performance

Figure 7 shows a comparison of the peel strength of the electrodes. We consider this failure mode as strongly indicating not only adhesion to the current collector but also the amount of elongation the composite can endure before failure. The electrochemical failure of anodes is strongly associated with delamination failure, whether for pristine electrodes or for electrodes after cycling [30,31]. In this case, the control electrodes, LiOH-PAA (4.5) and PAA, had the highest average peel strengths (3.9 N/cm and 1.8 N/cm, respectively), while the control electrode, PVDF + PAA, and the electrodes with modified PVDF displayed significant reductions in peel strength, with the trend being f5 > f20 > f60. This decrease in peel strength with reaction time aligns with the chemical modifications observed in FTIR and XPS analyses, suggesting that the shortened polymer chain structure compromises the adhesion of the electrode films to the current collector.

Figure 8 displays the resulting data from the nano-indentation of each electrode. The modified sample, specifically f60 + PAA, with relatively shorter polymer chains, shows an indentation depth of 22 μm, whereas the PAA sample, with longer chains, exhibits a depth of 8 μm.

This observed behavior can be attributed to the reduced entanglement and crystallinity in polymers with shorter chains, which consequently decreases mechanical resistance to deformation, making it less stiff and more prone to changes in shape under stress. In contrast, longer-chain polymers exhibit a more elastic response, resisting deformation and maintaining their shape to a greater extent under the same stress. The higher degree of entanglement in these longer chains forms a more robust network that can effectively distribute and resist the forces of indentation, resulting in shallower penetration by the indentation.

This experiment proved challenging on our samples because the materials’ significantly lower hardness falls outside the instrument’s optimal range. Furthermore, the instrument was unable to accurately measure the indentation depth, so we had to estimate it using follow-up optical imaging. Due to these difficulties associated with testing softer materials, and to keep the scope of this work focused, we have opted not to pursue further analysis.

Electrochemical Performance Cycling

Figure 9 presents the electrochemical performance of Si anode half-cells cycling at different C-rates for lithiation of the Si. This cycling protocol is designed to combine rate-capability testing with cycle-life testing to evaluate how the electrodes work under increasing rate demand. When the sequence of C-rates is repeated, one can discern the amount of degradation that has occurred by comparing to the same C-rate in a previous sequence. In this case, the electrode containing PAA alone maintains relatively consistent performance throughout the cycling tests, suggesting higher stability but lower capacity. On the other hand, the LiOH-PAA(4.5) electrodes achieve higher overall capacities, possibly due to lithium hydroxide enhancing ionic conductivity and benefiting from an optimal pH value, as previously reported [32].

Figure 7 and Figure 9 show that the PVDF-PAA binder exhibits significantly lower peel strength and capacity than LiOH-PAA (4.5) and the solo PAA binder, indicating that the combination of pristine PVDF and PAA does not result in better performance than solo PAA. Moreover, pristine PVDF with PAA exhibits higher capacity than all the modified samples, suggesting that despite the introduction of -COOH groups, which may offer a more flexible binder structure, a shortened polymer chain is more detrimental to overall cell performance. The capacity across the modified samples shows a sequence of f5 > f20 > f60. This pattern aligns with the outcomes of the peeling tests, suggesting that the weak adhesion between the coating and copper current collector could lead to the failure of these modified samples.

Interestingly, sample f60 maintains a more stable capacity than f5 and f20 at high C-rates. This improved stability could be attributed to the physical structure of the matrix, which features shorter chains. Shorter polymer chains and the subsequent introduction of -COOH can increase the porosity and surface area of the electrode, improving electrolyte accessibility and promoting faster ion transport. This structural alteration could facilitate lithium-ion transport to and from the active sites during rapid charge and discharge. As a result, at higher C-rates where diffusion-limited intercalation becomes less efficient, the contribution of the pseudo-capacitive charge storage mechanism becomes more significant. Furthermore, it is possible that there is a favorable interaction between the carboxyl functional groups and lithium ions in the electrolyte, leading to improved lithium solvation in the electrolyte and possibly the SEI.

Figure 10 demonstrates a positive correlation between the average peel strength and the delithiation capacity of the samples tested at cycle 41 (C/10 rate). It is observed that as the adhesion of the electrode in a sample improves, there is also an enhancement in the electrochemical performance of the cell. Linear regression analysis yielded a strong correlation (R² = 0.986, p < 0.001) with the regression equation: y = 533.3x + 321.2, where y represents capacity (mAh/g), and xx is the peel strength (N/cm). This relationship suggests that mechanical adhesion is a key factor in optimizing the overall performance of Si-based electrodes.

Figure 10 demonstrates a positive correlation between average peel strength and the delithiation capacity of the samples tested at cycle 41 (C/10). As electrode adhesion improves, the electrochemical performance of the cell is likewise enhanced, underscoring the critical role of binder–collector interactions. Linear regression analysis yielded a strong correlation R² = 0.96, p < 0.001) with the regression equation, y = 533x + 321, where y represents capacity (mAh/g) and x represents peel strength (N/cm). This strong correlation suggests that mechanical adhesion is a key factor in optimizing the overall performance and durability of Si-based electrodes.

Post-Mortem Analysis

Figure 11 illustrates the condition of the silicon anodes following the cycling tests. It was observed that cells utilizing PVDF as a binder exhibited varying degrees of delamination. Despite the possible enhanced flexibility offered by the shorter polymer chains in modified PVDF, their electrochemical performance was compromised due to diminished contact with the copper substrate. This reduction in adhesion was corroborated by peeling tests and evaluations of cycling performance, which indicated that binders with modified compositions exhibited significantly lower peel strength and reduced electrochemical capacities.

Herein, we propose that the observed failures in modified PVDF binders may primarily result from the weak adhesion at the interface between the bulk coating and the copper current collector. To address these challenges, another strategy was developed to optimize adhesion within the binder system, as described in the following section of this work.

3.2. Different Silicon Binders and Introduction of a Multilayer Anode

In this section, low-molecular weight polymer binders, such as PAA(2000), PI, PAN, PVDF-HFP, and Jeffamine^TM, were tested electrochemically and mechanically for use in Si anodes. Also, we introduced adhesion layers made of PVDF and PAA into the electrodes using binders with a combination of PAA(2000) + PAA. Additionally, PAA(2000) + PAA, PAA, and PAA(2000) binders were tested as control samples. In

Table 3. Name and description of Si electrode samples used for adhesion layer testing.

Sample	Description
PAA(2000) + PAA	Control sample that combines a binder made of PAA with an average molecular weight of 2000 and PAA
PAA	Control sample of a solo binder made of PAA
PAA(2000)	Control sample of a binder made of PAA with an average molecular weight of 2000
JeffamineTM	Solo binder made of JeffamineTM
PVDF-HFP	Sol binder made of PVDF-HFP binder
PI	Solo binder made of PI
PAN	Solo binder made of PAN
PAA(2000) + PAA on PVDF	Combined binder made of PAA with an average molecular weight of 2000 and PAA, with an adhesion layer made of PVDF
PAA(2000) + PAA on PAA	Combined binder made of PAA with an average molecular weight of 2000 and conventional PAA, with an adhesion layer made of PAA

, a summary and description of the samples are presented.

3.2.1. Electrochemical Performance Cycling

In Figure 12, the capacity of silicon anode half-cells is shown using the different samples listed in Table 3. PI demonstrates superior performance, maintaining the highest capacity. Notably, PVDF-HFP and short-chain polymer PAA(2000) and Jeffamine^TM D-2000 show significantly lower capacities compared with PI, which may be related to their poorer adhesion properties, leading to ineffective electronic transport between the current collector and the coating. In addition, PAN displays high initial capacities but undergoes a rapid decline as the C-rate increases. This decline could be attributed to its low molecular weight (70 kDa), which leads to poor adhesion to the current collector. It is also possible that due to the mechanical stiffness, PAN cannot accommodate the volume changes in the silicon during cycling.

To evaluate the impact of adhesion, we tested a mixture of long-chain and short-chain PAA at a 1:1 weight ratio, applied with and without an adhesion layer. PAA(2000) + PAA exhibits excellent initial capacities that are higher than those of pristine PAA. When combining short and long chain PAA together, a synergistic advantage is achieved, offering flexibility from PAA(2000) and adhesion from PAA. However, PAA(2000) + PAA shows a dramatic capacity drop as the C-rate increases. This could be due to the weak adhesion between the bulk film and the copper, as we further confirmed in the peeling test. On the other hand, the PAA(2000) + PAA on PAA cell maintains a more stable capacity performance throughout all the C-rates, although its average capacity is lower than that of PAA(2000) + PAA.

The lower capacity of PAA(2000) + PAA on PAA could be attributed to its manufacturing process because when the water-soluble bulk layer is coated on the similarly soluble adhesion layer, it leads to blending of the two layers. This blending can compromise the homogeneity and integrity of both layers, potentially increasing susceptibility to delamination and adversely affecting the electronic pathways.

In order to overcome the undesired blending between the adhesion and bulk layers, we modified our method by formulating an adhesion layer using PVDF binder dissolved in NMP. PVDF is not soluble in water; therefore, when the subsequent PAA-based layer is added to the adhesion layer, very little blending will occur. The cycling results show that with this new multilayer anode, PAA(2000) + PAA on PVDF, an improved average capacity (around 500 mAh/g higher) can be achieved compared to that of PAA(2000) + PAA on PAA. In fact, at high rates of lithiation, this formulation outperforms all the others observed in this study.

3.2.2. Mechanical Performance

In Figure 13, the average peel strength of the second set of electrodes is presented, indicating the mechanical integrity and adhesive strength of various binder materials to the copper current collector. In addition, we conducted a peeling test on two kinds of adhesion layer alone, namely, PAA and PVDF adhesion layers. The peel strengths correlate well with the electrochemical performance shown in Figure 12, maintaining a similar pattern, where PI exhibits higher peel strength, followed by PAN, and comparatively lower values for the Jeffamine^TM and PVDF-HFP binders. This trend supports the hypothesis that the inferior performance of certain cells could be attributed to the weak adhesion between the electrode coating layer and the copper current collector.

The peel strength of the PAA(2000) + PAA bulk layer with and without the PAA/PVDF adhesion layer was also examined. PAA(2000) shows a minimum peeling strength among PAA-based electrodes, confirming the deleterious effect of short-chain polymers on coating adhesion. Both PAA and PVDF adhesion layers show excellent peeling resistance. The PAA adhesion layer exhibits an average peel strength of 3.45 N/cm. However, when this adhesion layer was coated with a PAA bulk layer, the adhesion dropped dramatically to nearly that of PAA(2000). This confirms our hypothesis that if the bulk and adhesion layers use the same solvent to dissolve the binder, these two layers will dissolve into each other and thus damage the adhesion coating.

This phenomenon is further overcome by PAA(2000) + PAA on PVDF, where dissimilar binders and solvents were used for adhesion and bulk electrode layers. PAA(2000) + PAA on PVDF shows an average peel strength of 1.51 N/cm, which is much higher than that of PAA(2000) + PAA on PAA at 0.14 N/cm. This further indicates success in improving the adhesion.

4. Conclusions

This work presents an exploration of binder engineering strategies aimed at enhancing the mechanical integrity and electrochemical performance of silicon anodes for lithium-ion batteries, with a particular focus on polymer chain length, chemical functionality, and electrode architecture.

The first strategy involved chemically modifying commercial PVDF binders through dehydrofluorination, followed by oxidative cleavage, to introduce carboxyl (-COOH) functional end groups and shorten the polymer chains. FTIR and XPS confirmed the successful incorporation of -COOH groups, while DSC analysis indicated structural and thermal changes consistent with reduced chain length. While we know the modified PVDF samples, particularly those subjected to extended dehydrofluorination reaction times (up to 60 min), produce shorter chain lengths, we did not have a reliable way to quantify the exact chain lengths. These samples exhibited significantly reduced adhesion to the current collector and inferior mechanical strength, as evidenced by peel tests and nano-indentation. This resulted in poor cycling performance in Si anodes, despite minor gains in flexibility that modestly improved high-rate stability, as shown for the shortest-chain variant (f60). These results suggest that excessive chain shortening leads to binder fragmentation into oligomeric structures, compromising interfacial adhesion. What is the ideal chain length? Undoubtedly, this depends on the polymer and the surfaces to which it is bound, but it should be long enough to ensure adequate adhesion/cohesion to the materials potentially at each end of the chain. Continued work is needed to answer this question.

Incorporating chemical modifications into PVDF binders can enhance the performance of silicon anodes. However, these modifications are likely to increase production costs due to higher material expenses. Therefore, while performance improvements are achievable, they must be carefully weighed against the associated cost implications.

However, the modified PVDF samples, particularly those subjected to extended dehydrofluorination reaction times (up to 60 min, resulting in the shortest chain length, showed markedly reduced adhesion to the current collector and diminished mechanical robustness, as confirmed by peel tests and nanoindentation. These structural weaknesses translated into poor cycling stability in Si anodes. Although the most extensively modified variant (f60) exhibited slightly enhanced flexibility, which provided modest improvements in high-rate performance, this benefit was insufficient to compensate for the overall mechanical and electrochemical degradation.

In the second strategy, low-molecular weight polymer binders were evaluated, namely, PAA(2000) and Jeffamine™. These were also compared to alternative binders, PVDF-HFP, PAN, and PI. Of these, PI emerged as the most promising, exhibiting both strong adhesion and superior capacity retention across a range of C-rates. In contrast, Jeffamine™, PAN, and PVDF-HFP-based electrodes demonstrated relatively poor cycling stability and mechanical performance. These deficiencies were attributed to either insufficient adhesion or inadequate mechanical flexibility to accommodate volume expansion. Moreover, blends of short-chain and long-chain PAA binders were tested, revealing an initially promising electrochemical profile; however, they suffered from capacity fade at higher C-rates, likely due to poor adhesion to the current collector.

To address this issue, a third strategy focused on the design and fabrication of multilayer anodes incorporating an adhesion layer between the copper current collector and the Si electrode. When a PAA-based adhesion layer was used beneath a water-soluble bulk electrode layer of similar composition, solvent blending between the two layers resulted in degraded adhesion and electrochemical performance. However, by using a PVDF adhesion layer, which was processed in NMP and insoluble in the subsequent application of an aqueous PAA layer, this interlayer blending was avoided. The resulting multilayer configuration (PAA(2000) + PAA on PVDF) demonstrated a significantly enhanced average capacity (up to 500 mAh/g higher) and improved adhesion strength compared to its single-layer counterpart or the PAA-on-PAA configuration. This finding confirms that solvent incompatibility can be strategically leveraged to construct mechanically robust multilayer electrodes that preserve layer separation and maintain structural integrity under cycling stress.

Taken together, the results underscore the complex interplay among polymer chain architecture, interfacial adhesion, mechanical compliance, and electrode design in Si-based anodes. While flexibility introduced via chain shortening may aid stress dissipation, sufficient chain entanglement and strong adhesion to the current collector remain essential for cycling stability. Moreover, multilayer electrode designs that decouple adhesion from bulk properties offer a powerful route to optimize both mechanical resilience and electrochemical performance, particularly when solvent compatibility is carefully managed. At this early stage of development, we acknowledge that PVDF-PAA has not yet demonstrated a clear performance advantage over previous neat PAA work. We are not presenting it as a fully developed technology ready to replace existing methods, but rather as a step forward in the ongoing learning and development process.