Monomers, 4,4′-(hexafluoroisopropylidene) diphthalic anhydride (6FDA), m-phenylenediamine (mPDA), and 3,5-diaminobenzoic acid (DABA) were purchased from Sigma-Aldrich and they were dried at 200°C in a vacuum oven. N-methyl-2-pyrrolidinone (NMP, anhydrous), acetic anhydride (AcAn), and pyridine were purchased from Sigma-Aldrich and dehydrated with molecular sieves. Silicon alloy (Si₅₀Al₃₀Fe₂₀) was prepared according to our previous study [13].

PI was synthesized by general two-step procedures: polycondensation and chemical imidization. Briefly, a stoichiometric amount of DABA or mPDA was added to a mixture of 6FDA in NMP, and then reacted at 25°C for 48 h under N₂ atmosphere to obtain poly (amic acid) (PAA) solutions. For chemical imidization, a mixture of AcAn and pyridine (1/1, v/v, with a molar ratio of 1:4 to 6FDA) was added into a PAA solution and then stirred at 80°C for 12 h under N₂ atmosphere. The resulting PI was precipitated, washed with methanol several times, and dried at 80°C under vacuum for 24 h.

The chemical structures were analyzed by ¹H NMR (Avance 400, Bruker) and FT-IR (Spectrum Two, PerkinElmer) spectrometers. The thermal properties were investigated by thermogravimetric analysis (TGA, TGA-8000, PerkinElmer, USA) at a temperature range of 50 – 800°C with a heating rate of 5°C min⁻¹ under N₂ atmosphere. The morphologies were obtained by field emission scanning electron microscopy (FE-SEM, S-4300, Hitachi). The adhesion strength was measured using a 180° peel test apparatus (5848 MicroTester, Instron) using a scotch tape (Magic, 3M) at a 20 mm sec⁻¹ (electrode sample of 20 mm width/80 mm length). The gel content of the thermally treated PI was determined by immersing the pre-weighted samples in NMP at room temperature for 24 h, and the insoluble fractions were dried under vacuum for at least 24 h to a constant weight. The gel content was calculated by the following equation: gel content = W₁/W₀ × 100 (%), where W₀ and W₁ are the initial weight of the thermally treated PI and the weight of the insoluble fraction in NMP, respectively.

An anode slurry was prepared by mixing the Sialloy particles (87 wt%), ketjen black (3 wt%), and PI binder (10 wt%) in NMP using a planetary mixer (Corona Mix CRM I.5L TSI Co. Ltd.,). The slurry was coated on the Cu foil current collector using a casting knife, and then dried at 110°C for 2 h in a drying oven. The loading level and electrode density were 1.0 – 1.1 mg cm⁻² and 0.8 – 0.9 g cm⁻³, respectively. The prepared anodes were thermally treated at 400°C for 1 h in a tubular furnace (Ajeon Heating Industrial) under an inert atmosphere to crosslink the PI, and then naturally cooled to room temperature. The 2016 coin cells were assembled in an Ar-filled glove box. The cells were composed of lithium metal as a counter electrode, a polypropylene separator (Celgard 2400), and the prepared anode. An electrolyte solution composed of 1 M LiPF₆ in ethylene carbonate (EC)/ethyl methyl carbonate (EMC) (EC:EMC, 3:7 v/v) with a fluoroethylene carbonate of 5 wt% (Dongwha Electrolyte, Korea) was injected between the two electrodes. As a comparative experiment, an anode was prepared using a commercial PVDF binder in the same manner without any thermal treatment. The constant current cycle experiment was performed in the voltage range of 0.005 – 2.0 V vs. Li/Li⁺ using the battery charge and discharge cycler (WBCS3000Le32, WonATech, Korea). To measure it, we charged it at the constant current of current density 80 mA g⁻¹ (0.1 C) until it reached 40 mA g⁻¹ (0.05 C) with constant voltage CC-CV mode at 0.005 V, and discharged it at the 80 mA g⁻¹ (0.1 C) constant current mode.

The synthetic scheme and chemical structure of the PI are presented in Figs. 1 and 2. The complete imidization of PAA to PI was confirmed by FT-IR and NMR spectroscopy. As shown in Fig. 3(a), all the PAA samples show characteristic peaks for amide groups such as C=O (1,666 cm⁻¹), C-C (1,600 and 1,410 cm⁻¹), and C-N (1,550 cm⁻¹), respectively [14]. After imidization (Fig. 3(b)), these peaks were disappeared, while characteristic peaks for imide groups, such as C=O (1,780 and 1,721 cm⁻¹), C-N (1,377 cm⁻¹), and C-H (1,094 cm⁻¹), appeared newly, indicating the successful imidization [15]. The ¹H-NMR spectroscopy was performed to further confirm the imidization, and the results are shown in Fig. 3(c). All the peaks can be assigned clearly according to the integral values of the intensity and the ¹H-NMR spectra completely agree with the proposed PI structures. These results confirmed the successful imidization of PAA to PI.

The decarboxylation-induced crosslinking of the PI was confirmed by gel content measurement, and FT-IR and TGA analyses, and the results are shown in Figs. 4 to 5. As shown in Fig. 4(a), the gel content increased from 0 to 44% with an increasing DABA content, indicating the crosslinking of the PI with DABA units. As can be seen in Fig. 4(b), the peak intensities for O-H band at 1260 cm⁻¹ and C-O band of COOH groups at 1317 cm^{− 1} significantly decreased after thermal treatment. These results confirm the decarboxylation of the PI after thermal treatment. As seen in Fig. 5(a), there is the initial weight loss probable caused by the removal of residual solvents and moistures. Besides, DABA-containing PI shows significant weight loss in temperatures ranging from 100 to 400°C, caused by the decarboxylation. These results are generally observed for PI containing DABA units [16]. Above that temperature, the thermal decomposition of the PI backbones occurred. To further investigate the decarboxylation, TGA thermograms of the PI after thermal treatment at 400°C were obtained as shown in Fig. 5(b). The weight loss corresponding to the decarboxylation diminished after thermal treatment, indicating the complete decarboxylation and crosslinking [17].

The peel strength of the anode with different binders was measured, and the results are shown in Fig. 6 and Table 1. As can be seen in Fig. 6(a), the anodes with the PI binders exhibit higher strength than the anode with PVDF binder. In addition, the strength increased with increasing the content of DABA possibly due to the strong interactions of the PI binders with the Si-alloy particles as well as the current collectors and the thermal crosslinking of the PI chains. Changes in the surface morphology of the anodes after peel test were observed by optical microscopy, and the results are shown in Fig. 6(b). The surface of the anodes can be divided into two regions. The black and gold-colored regions show the surfaces of the anode materials and Cu current collector, respectively. The anode with PVDF binder was almost detached from the Cu current collector due to the adhesion failure during the peel test. In contrast, the anode with PI (6FDA-DABA) binder was found to be well attached to the current collector due to the enhanced adhesion strength of the binder. These results revealed that the crosslinked PI (6FDA/DABA) binder enhanced the adhesion between active materials, conductive carbon particles, and current collectors, enabling the electrode to maintain good mechanical integrity of the anodes during cycling.

Galvanostatic measurements were conducted to compare the effect of binders on cell performance. The anodes with different binders showed different electrochemical performance in terms of initial discharge capacity and capacity retention. As shown in Fig. 7 and Fig. 8, the initial discharge capacity (636 mAh g⁻¹ with an initial Coulombic efficiency (ICE) of 74.9%) of the anode with PVDF binder significantly decreased to 427 mAh g⁻¹ after the 1st cycle. The initial discharge capacities of anodes with PI (6FDA-DABA/mPDA), PI (6FDA-mPDA), and PI (6FDA-DABA) binders were 816, 791, and 816 mAh g⁻¹, respectively. Consequently, the ICE values were 93.8% (6FDA-DABA/mPDA), 76.9% (6FDA-mPDA), and 80.2% (6FDA-DABA). The ICE values for PI binders were higher than that of the PVDF binder, the binders. In addition, the PI binder containing DABA shows a higher ICE value. After 50 cycles, the capacity retentions of all the anodes are 48.7% PI (6FDA-DABA), 43.59% PI (6FDA-DABA/mPDA), 32.16% PI (6FDA-mPDA), and 2.94% (PVDF), respectively, indicating that the anode with a PI (6FDA-DABA) binder shows the superior electrochemical performance compared to other anodes.

Fig. 9 exhibits the differential capacity vs. potential plots (dQ/dV) after 1 cycle, 2 cycles, and 50 cycles, respectively. The sharp peaks centered at 0.45 V are due to the delithiation process, which is assigned to the presence of the crystalline Li₁₅Si₄ phase and a two-phase reaction with an amorphous Li~₂Si [18]. After 50 cycles, the peaks located at a 0.45 V decreased, indicating the decrease in the discharge capacity due to the degradation of the PVDF anode. However, the peak of the PI (6FDA-DABA) anode at 0.45 V was well-maintained. These results indicate that the PI (6FDA-DABA) is an effective binder and good consistent with the results of adhesion strength (Fig. 6 and Table 1).

Fig. 10 shows rate capabilities of all the anodes in the range of 0.1 – 2 C-rate. Compared with anodes, the anode with a PI (DABA) binder has a highest reversible capacity. The discharge capacities of anode with a PI (DABA) binder were ~900 and ~600 mAh g⁻¹ at C-rates of 0.1 and 2 C, respectively. These outstanding electrochemical behaviors of the anode with a PI (DABA) binder are mainly due to due to the superior mechanical integrity induced by highest adhesive strength.

Fig. 11 indicates the SEM results for anodes before and after 50 cycles. Fig. 11(a) is nothing wrong with the surface of the anode before cycle. In Fig. 11(b), the cracks were generated after 50 cycles due to microstructure change by the weak adhesion of binder. However, the PI (DABA) with an excellent adhesion strength shows the well-maintained microstructure without a crack after 50 cycles.