1. Introduction
Li-ion batteries (LIBs) are widely used in smart watches, electric cars, and energy storage systems [
1,
2]. The energy density of LIBs, comprising a conventional Li transition-metal-oxide cathode and a graphite anode, is higher than that of previous rechargeable battery systems i.e., with nickel-metal hydrides. The current market demands higher energy density for more applications such as wearable electric devices and airplanes; therefore, new electrode materials with high energy density are highly required.
As a next-generation battery material, Si is one of the most attractive anode materials owing to the fact that it is non-toxic, has low cost, and possesses a gravimetric energy density (3579 mAh g
−1, Li
15Si
4) and a volumetric energy density (2190 mAh cm
−3) that are ten times and three times higher, respectively, than those of graphite (LiC
6: 372 mAh g
−1 or 756 mAh cm
−3) [
3]. However, Si also has low electrical conductivity and suffers from large volumetric expansion during lithiation, which has many demerits such as the pulverization of Si and loss of electrical contact among electrode components. To overcome these problems, various approaches, including morphology modification of Si [
4], carbon coating on the Si surface [
5], and exploring new binder systems [
6–
10], have been suggested.
Although changing the morphology and carbon coating resulted in enhanced electrochemical performance, they require additional complicated processes. New binder techniques also contributed to enhanced electrochemical performance; however, they still need further improvement because the conducting agent i.e., carbon, hinders direct contact between the binder and Si particles. Therefore, a conductive binder was suggested as a solution to this issue [
8–
10]. However, the conductive polymer is unstable during the electrochemical reaction owing to its high reactivity with the electrolyte [
11].
On the basis of the above issues, our strategy is to produce a highly conductive and mechanically robust binder. Several binders such as polyacrylic acid (PAA), carboxymethyl cellulose (CMC), styrene butadiene rubber (SBR), polyvinyl acid (PVA), polyvinylidene fluoride (PVDF), polyimide (PI), and conductive polymers are highly utilized for Si-based anodes [
7]. Among the various conventional binders, PVDF, CMC, and SBR binders are not able to accommodate the volume expansion of active materials; as a result, the active materials and conducting agent (carbon) form agglomerates due to weak interfacial interactions [
6,
12]. PAA and conducting polymer binders have been more effective thanks to their multiple hydrogen bonds and the formation of a conducting gel-like framework within the electrodes [
6,
13,
14]. Recently, a Si-based electrode using a new binder i.e., NaPAA grafted to CMC (NaPAA-g-CMC), showed good electrochemical performance with a high initial discharge capacity (2290 mAh g
−1) and stable cycling performance, which originated from its good adhesion force [
15]. However, the Si-based electrodes using NaPAA-g-CMC binder showed relatively low capacity retention for early cycles.
In this study, we propose a design for a highly conductive binder without adding a conducting agent using partially carbonized PAA grafted to CMC (CMC-g-PAA). Utilizing the partially carbonized CMC-g-PAA binder, Si-based anodes with stable cycling performance and robust binding force were fabricated successfully. Additionally, it was possible to increase the initial capacity and efficiency of the Si anodes via optimization of heating conditions and binder ratios.
2. Experimental
2.1 In situ fabrication of Si/carbon/CMC-g-PAA nano-hybrid composite electrode
A slurry comprising nano-sized Si powder (0.34 g, Alfa Aesar), PAA (0.33 g, Aldrich), and CMC (0.33 g, Dai-Ichi Kogyo Seiyaku) in 15 cm
3 distilled water was prepared; this is denoted the P10 electrode. After mixing for 20 min, the slurry was coated onto a Cu foil. Thereafter, the sample was dried at 80°C for 2 h in a convection oven to evaporate distilled water, and then transferred to a tube furnace and heat-treated at 500°C for 5 h under N
2 atmosphere. The Si mass ratio of the P10 electrode was approximately 70 % (
Fig. 1a). A slurry comprising nano-sized Si powder (0.7 g), PAA (0.1 g), CMC (0.1 g), and Super P (0.1 g) in 15 cm
3 distilled water was prepared; this is denoted the R3 electrode. After mixing for 20 min, the slurry was coated onto a Cu foil. Thereafter, the sample was dried at 80°C for 2 h in a convection oven to evaporate distilled water.
2.2 Characterization of the electrode and materials
The microstructure and elemental mapping of the cross-sectional image of the electrodes were analyzed via field-emission scanning electron microscopy (FESEM, JEOL JSM-7000F). The existence of carbon and the remaining polymer structure in the electrodes was examined using Raman spectroscopy (RAMAN, SENTERRA, Bruker) and Fourier-transform infrared spectroscopy (FT-IR, Vertex 70, Bruker), respectively. The adhesive force of the binder was measured by a peel tester (JSV H1000, Japan Instrumentation System Co, LTD), and the thermal decomposition behavior of CMC-g-PAA was analyzed by thermogravimetric analysis (TGA, Diamond TG/DTA, Perkin Elmer). The electrical conductivity of the electrodes was measured by the four-point probe method (Loresta-GP, Mitsubishi Chemical Corporation).The impedance analysis after the C-rate test was performed using Nyquist plots (VSP-3000, BioLogic, Science Instruments). For the ab initio calculation of two models of CMC-g-PAA, the Spartan’16 suite was used with density functional theory (DFT, ωB97X-D) and a 6-31G* basis.
2.3 Electrochemical Testing
The electrodes were cut into disks (12 mm diameter), in which the average loading of the Si active material was 0.8 mg cm−2 and packing density was 0.15 g cc−1. Thereafter, 2032 coin-type half cells were assembled using a polyethylene separator (Toray Tonen) in a dry room with a dew point of less than − 60°C. Subsequently, 1 M LiPF6 in a mixture of ethylene carbonate (EC), ethyl methyl carbonate (EMC), and dimethyl carbonate (DMC) (EC/EMC/ DMC = 2/1/7, volume ratio) was used as the electrolyte solution with 7 wt.% fluoroethylene carbonate (FEC) as the additive and Li foil (16 mm in diameter) as the counter electrode. The assembled cells were aged overnight at room temperature and then electrochemically tested using the TOSCAT-3100U (Toyo System Co.) battery measurement system under the following conditions. The first cycle was operated in a constant current (CC) mode of 100 mA g−1 within a voltage window of 0.005–2.5 V vs. Li/Li+. The test protocol for the next cycle was a constant current followed by a constant voltage (CC/CV) mode for discharging (lithiation). In this mode, 300 mA g−1 was used for the CC step and 0.005 V and 60 mA g−1 for the CV step. For charging (delithiation), the CC mode with 300 mA g−1 was used and the cut off voltage was 1.5 V. For the C-rate test, the cycles were operated in a CC mode with different current densities (100 mA g−1, 200mA g−1, 500 mA g−1, 1000 mA g−1, and 2000 mA g−1) within a voltage window of 0.005–2.5 V vs. Li/Li+.
3. Results and Discussion
PAA and CMC are known as polymer binders, and it is reported that PAA grafted to CMC can be formed at a relatively low temperature of 55°C, as shown in
Fig. 1b [
15]. Therefore, to design highly conductive CMC-g-PAA binders without adding carbon, the CMC-g-PAA binder is partially carbonized by a simple heat-treatment; the schematic diagram is shown in
Fig. 1c.
The CMC-g-PAA binder was analyzed by FT-IR spectroscopy (
Fig. 2a) [
16]. In the range of the studied wave lengths, the peak profiles of P10 and R3 are similar. However, the presence of a P10 transmittance peak at 1637 cm
−1 indicated the double bond between carbon atoms [
17]. The carbon double bond was not in the CMC-g-PAA molecule (
Fig. 1b) and seems to appear after carbonization. The two models of the grafted structure are shown in
Fig. 3a; model 2 showed lower Gibbs free energy than model 1 after energy minimization (
Table 1) and a similar FT-IR spectrum to P10, with regard to peak intensity and order (
Fig. 3b). The DFT calculation results of model 2 show that the peaks were shifted to higher wave numbers (blue-shifted) compared to the P10 profile. Generally, blue-shifting is due to weak intermolecular interactions. Considering that model 2 is very simplified, i.e., it has only one small molecule without intermolecular interactions, the blue-shift is expected. The Gibbs free energies for both models are also compared in
Table 1. Model 2 is more suitable than model 1, with regard to thermodynamics and spectroscopy results, as suggested in a previous study [
15]. The widely branched molecular structure of model 2 enabled us to gain an idea of the strong adhesive force of CMC-g-PAA.
A strong binding force is important to mitigate the volumetric expansion of Si in the electrochemical reaction because a mechanically robust electrode enables good capacity retention during repeated alloying and de-alloying of Si and Li. The use of CMC-g-PAA results in a strong binding force between the electrode components. The carbonization and the electrical conductivity were studied by Raman spectroscopy (
Fig. 2b) and a four-point probe (
Table 2), respectively. As the mass ratio of the binder in the electrodes increased, the electrical conductivity increased and D/G peak intensity ratio decreased. This means that the nano-graphite proportion in the electrode increased [
18,
19] and the conductivity of electrode was increased because of the higher ratio of conductive agent. Interestingly, the P10 electrode showed impressive adhesive force because almost all electrode components remained on the electrode after the adhesion test (
Fig. 2c). On the other hand SBR-CMC electrode showed much worse adhesive force (
Fig. 2d).
To study the carbonization effects on the CMC-g-PAA binder, the results for the P10 electrode were compared with those for the R3 electrode i.e., with 10% of a conducting agent (Super P) without carbonization. The R3 electrode showed higher first cycle capacity and Coulombic efficiency than the P10 electrode (
Fig. 4a). The initial charge capacity at 100 mA g
−1 for the P10 electrode was 1083 mAh g
−1 with 74% Coulombic efficiency, whereas for the R3 electrode, it was 2410 mAh g
−1 with 85% Coulombic efficiency (
Fig. 4b and c). The higher first charge capacity of the R3 electrode originated from lower impedance (
Fig. 4d) and the weaker binding force due to the employment of a conventional binder without carbonization. However, the P10 electrode showed no capacity deterioration during 50 cycles, which is not the case for the R3 electrode (
Fig. 4a). The excellent cycle behavior was attributed to the well-distributed conductive carbon network and the robust mechanical force of the partially carbonized CMC-g-PAA binder.
In a general slurry preparation process for the electrodes, mechanical mixing of the conducting agent (Super P) has a negative effect on the distribution in the electrodes, where the active material and the conducting agent form agglomerates during cycling (
Fig. 5) [
12]. The partial carbonization of the CMC-g-PAA binder affects its chemical cross-linking and improves adhesion. Therefore, the electrode of the partially carbonized CMC-g-PAA binder (P10) still maintained a strong binding force between Si particles and the current collector after cycling (
Fig. 5d–f).
Fig. 5d shows the cross-sectional image of the P10 electrode after 1000 cycles. During cycling, the nanosized Si particles formed agglomerates and expanded. In addition, many cracks were observed on the surface (
Fig. 5e) and the interior (
Fig. 5f) of the electrode, but no traces of loss of active materials were observed. From the elemental mapping data (
Fig. 6b and c), well-distributed P and F from the electrolyte salt (LiPF
6) were also observed in the cycled electrode, which demonstrates that the cracks improved the electrolyte wetting of the electrode.
The loading of the binder severely affected the electrochemical performance of each electrode. As the binder ratio increased, the cycle performance improved, as shown in
Fig. 7a; this was caused by the increased binding force, which limited the expansion of Si particles in the partially carbonized CMC-g-PAA binder. However, the first charge capacity and Coulombic efficiency decreased, as shown in
Fig. 7b. The lower Coulombic efficiency was caused by the insufficient conductivity of the partially carbonized CMC-g-PAA binder. The conductivity of the P10 electrode was 1.780 × 10
−5 S cm
−1; this can be optimized by controlling the heating time and temperature to result in higher carbonization rates in the electrodes.