Improving the Cycling Performance of Lithium-Ion Battery Si/Graphite Anodes Using a Soluble Polyimide Binder.

Herein, we improved the performance of Si/graphite (Si/C) composite anodes by introducing a highly adhesive co-polyimide (P84) binder and investigated the relationship between their electrochemical and adhesion properties using the 90° peel test and a surface and interfacial cutting analysis system. Compared to those of conventional poly(vinylidene fluoride) (PVdF)-based electrodes, the cycling performance and rate capability of P84-based Si/C anodes were improved by 47.0% (372 vs 547 mAh g-1 after 100 cycles at a 60 mA g-1 discharge condition) and 33.4% (359 vs 479 mAh g-1 after 70 cycles at a 3.0 A g-1 discharge condition), respectively. Importantly, the P84-based electrodes exhibited less pronounced morphological changes and a smaller total cell resistance after cycling than the PVdF-based ones, also showing better interlayer adhesion (F mid) and interfacial adhesion to Cu current collectors (F inter).

Introduction

The progressing fossil fuel depletion and global climate change pose a serious threat to humankind, thus requiring internationally coordinated efforts and attention. Electric vehicles (EVs) and energy storage systems (ESS’s) have gained tremendous attention as one of the promising greenhouse gasses solution.[1−3]

However, the successful implementation of EVs and ESS’s relies on high-energy-density battery systems. Although lithium-ion batteries (LIBs) based on carbon anode materials (graphite, 372 mAh g–1 for LiC6) and transition-metal-oxide cathode materials (e.g., lithium cobalt oxide, LiCoO2) have been commercialized to power mobile electrical devices, their current energy density is still too low to satisfy the high-level requirements of large-scale applications such as EVs and ESS’s.

Although silicon has received much attention as a promising anode material due to its high theoretical energy density (4200 mAh g–1 for Li22Si5 or Li4.4Si), environmental friendliness, and low cost,[4−8] it has not been successfully implemented in commercialized LIBs because of suffering from large volume changes (>300%) during alloying and dealloying. Moreover, the stresses accompanying these changes cause mechanical failure and result in severely degraded electrochemical performances of Si electrodes.

Consequently, Si/graphite (Si/C) composite anodes (Si/C anodes), wherein Si constitutes only a portion of the active material, have been proposed as an alternative to pure Si anodes.[9,10] This approach exhibits certain advantages because even if the theoretical anode capacity increases infinitely, the total theoretical capacity of the battery soon reaches saturation because of the limited theoretical capacity of the cathode.[11]

We have recently succeeded in enhancing the performance of Si-based anodes by using mussel-inspired and co-polyimide-based adhesive polymeric binders, hierarchical Si composites, adhesive Cu current collectors, and adhesive conductive additives.[4,5,12,13] Still, the demand for battery-suitable high-loading Si/C-based anodes requires further progress in this direction. However, as shown in Table S1 (Supporting Information), the importance of active material loading in Si/C anodes has been underestimated by current studies.

P84 is a highly adhesive, soluble, and thermally stable co-polyimide-based binder, exhibiting outstanding intrinsic properties compared to those of conventional polymeric binders and thus improving the performances of battery constituents (anodes, cathodes, and separators), enhancing the cycling performances of Si-based anodes and the high-temperature electrochemical performances of cathode materials and increasing the thermal stability of LIBs that contain layers of ceramic composite separators.[7,14−16]

Herein, we used P84 in high-loading Si/C-based anodes (∼4.0 mg cm–2) and investigated their electrochemical and mechanical properties. Specifically, electrochemical properties such as electrochemical stability, cycling performance, rate capability, and impedance were evaluated using cyclic voltammetry (CV) and alternating current impedance spectroscopy, with the corresponding data recorded using a battery cycler. The mechanical properties of the above anodes were evaluated using the 90° peel test and a surface and interfacial cutting analysis system (SAICAS).

Result and Discussion

The uniformity of Si/C anodes was characterized by scanning electron microscopy (SEM) coupled with energy-dispersive X-ray spectroscopy (EDX). Figure shows that the electrode surface featured uniformly distributed microsized graphite particles, with the gaps between them filled by nanosized Si, confirming that Si/C anodes exhibited good surface uniformity.

Figure 1

(a) SEM image of the P84-based Si/C anode surface and (b) EDX elemental mapping relevant to (a) (blue = Si, red = C).

PVdF- and P84-based Si/C anodes were precycled, and the obtained results were compared. As depicted in Figure , P84-based Si/C showed a higher discharge capacity than that of PVdF-based Si/C (P84-based Si/C: 900.6 mAh g–1 for charging and 602.5 mAh g–1 for discharging; PVdF-based Si/C: 658.7 mAh g–1 for charging and 523.8 mAh g–1 for discharging). In contrast, the former electrode exhibited a lower Coulombic efficiency than that of the latter one (66.9 vs 79.5%, respectively), which seemed to be closely related to the irreversibility of the reduction reaction, as is discussed below.

Figure 2

Potential profiles of P84- and PVdF-based Si/C anodes recorded during precycling.

Electrochemical stabilities were probed by CV, with both electrodes showing almost identical profiles from open circuit voltage (OCV) to 0.9 V vs Li/Li+, whereas P84-based Si/C exhibited a more intense reduction peak below 0.9 V vs Li/Li+ (Figure ). As suggested in previous studies, this behavior is related to the intrinsic properties of imide-based polymeric materials, i.e., the C=O groups of the imide ring in P84 react with Li+ during charging.[7,17] Interestingly, P84-based Si/C showed a stronger oxidation peak than that of PVdF-based Si/C. The above results were in good agreement with those of previous precycling tests relevant to Figure , revealing that P84-based Si/C showed higher discharge and charge capacities than those of PVdF-based Si/C and implying that the irreversible reduction reaction played a positive role in Si/C anodes.

Figure 3

Cyclic voltammetry (CV) profiles of P84- and PVdF-based Si/C anodes recorded at a scan rate of 0.1 mV s–1.

CV profiles of PVdF-based Si/C anodes and P84-based Si/C anodes at 5th, 10th, 15th, and 20th scans were drawn and compared. As shown in Figure , the magnitude of the current peaks increased with cycling due to activation of more material to react with Li in each scan.[18] Furthermore, the first scan shown in Figure differs from those due to a “formation” effect, corresponding to the filming of the electrode during the first Li discharging.

For graphite anodes, the main reactions are believed to be the solvated Li intercalation (at potentials from ∼0.8 to ∼0.2 V vs Li/Li+) and the reversible formation of subsequent stages of LiC6 (at potentials from ∼0.2 to ∼0.005 V vs Li/Li+).[19] The stepped voltage profile of Si can be observed in a high-temperature experiment (415 °C) to form four intermediate equilibrium phases. In contrast, at a mild temperature (25 °C), lithiation shows a single gradually sloping plateau (∼0.1 V vs Li/Li+) and large hysteresis between lithiation and delithiation.[20] This is because the lithiation process results in the formation of a metastable amorphous LiSi phase instead of the equilibrium intermetallic compounds.[20] Considering the overlapping voltage plateau region of graphite (∼0.2 to ∼0.005 V vs Li/Li+) and Si (∼0.1 V vs Li/Li+) during lithiation, delithiation voltage plateau should be carefully considered to distinguish the role of each active material.

As depicted in Figure , the P84-based Si/C anodes showed two overlapping peaks at ∼0.27 and ∼0.32 V and a broad shoulder peak at ∼0.5 V during charging (delithiation) until the 20th scan. In contrast, PVdF-based Si/C showed a peak shape similar to that of the P84-based Si/C anodes up to the 5th scan and showed a single peak at ∼0.27 V without the ∼0.5 V wide shoulder peak from the next scan. Considering the fact that the PVdF-based Si/C showed a rapid capacity decrease due to Si pulverization (Figure ), a long-lasting single peak at ∼0.27 V corresponds to graphite, whereas the other peaks at ∼0.32 and ∼0.5 V correspond to Si. The peak characteristics of Si are in good agreement with those in the previous study.[21]

Figure 4

CV profiles of (a) P84-based Si/C anodes and (b) PVdF-based Si/C anodes during 5th, 10th, 15th, and 20th scans recorded at a scan rate of 0.1 mV s–1.

Figure 5

Electrochemical performances of P84- and PVdF-based Si/C anodes: (a) cycling performance (charge rate = 60 mA g–1, discharge rate = 60 mA g–1). (b) Rate capabilities determined by varying the discharge rate from 60 mA g–1 to 3.0 A g–1 while maintaining the charging rate at 60 mA g–1.

Subsequently, we investigated the effect of polymeric binders on the cycling performance and rate capability of Si/C anodes.

As shown in Figure a, P84-based Si/C showed better cycling performance than that of PVdF-based Si/C, with the respective capacity retentions after 100 cycles equaling 93 and 72% (P84-based Si/C: initial discharge capacity = 586 mAh g–1, discharge capacity at 100th cycle = 547 mAh g–1; PVdF-based Si/C: initial discharge capacity = 511 mAh g–1, discharge capacity at 100th cycle = 372 mAh g–1). Furthermore, P84-based Si/C exhibited a capacity retention of 88% after 200 cycles (Figure S1, Supporting Information; discharge capacity at 200th cycle = 516 mAh g–1).

Furthermore, we investigated the cycling performance of P84-based Si/C anodes at different active material loadings and current densities (Figure S2, Supporting Information), revealing that the electrode with the lowest loading (1.0 mg cm–2) showed the best performance during high-current-density charging (0.3 A g–1). In contrast, Si/C anodes with the highest loading (4.0 mg cm–2) showed poor cycling performance and a low discharge capacity of ∼400 mAh g–1. As depicted in Figure a, a low current density (60 mA g–1) was required for the high-loading Si/C anodes (4.0 mg cm–2) to achieve a discharge capacity (∼600 mAh g–1) exceeding that of graphite (372 mAh g–1). Considering the importance of high-loading anodes (Table S1, Supporting Information) and the fact that long-term ESS’s utilize charging at a rate of C/12, one can infer that the high-loading P84-based Si/C systems with a stable cycling performance and high discharge capacity (Figure a, charging rate = C/10) are suitable for practical applications.[22,23]

As shown in Figure b, P84-based Si/C showed a better rate capability than that of PVdF-based Si/C, with the above electrodes maintaining 81 and 70% of the initial discharge capacity, respectively, after 70 cycles at 5C (P84-based Si/C: initial discharge capacity = 586 mAh g–1, discharge capacity at 70th cycle = 479 mAh g–1; PVdF-based Si/C: initial discharge capacity = 511 mAh g–1, discharge capacity at 70th cycle = 359 mAh g–1).

After precycling, the interfacial resistances of unit cells comprising P84-based and PVdF-based Si/C anodes were measured and compared. In general, the total cell resistance (Rtotal) comprises bulk resistance (Rb), solid–electrolyte interphase (SEI) resistance (RSEI), and charge transfer resistance (Rct), i.e., Rtotal = Rb + RSEI + Rct.[24−26] To explore the origin of electrochemical properties of LIB cells in practice, Rtotal should be closely monitored because, in general, the LIB cells having smaller Rtotal revealed the improved rate capability and cycle performance.[16,26−31] As shown in Figure , we fitted the electrochemical impedance spectroscopy results using an equivalent-circuit model based on the fact that the SEI was composed of multiple layers.[32] The unique feature of this model is the separation of (1) all Ohmic resistant components lumping into R1 and (2) Faradaic nonlinear components into the R2, R3, and R4. To be more specific, R1, (R2 + R3), and R4 represent Rb, RSEI, and Rct, respectively. The calculated resistance is listed in Table S1. The Rtotal of P84-based Si/C cells was much smaller than that of PVdF-based Si/C cells, in agreement with the improved cycling performance and rate capability of the former electrodes. Taking this into account, it can be inferred that P84 effectively prevents Si pulverization and produces thin SEI layers with a smaller resistance.

Figure 6

Impedance spectra of unit cells containing P84- and PVdF-based Si/C anodes after precycling.

As shown in Figure , the initial morphologies of PVdF- and P84-based Si/C anodes were different due to the inherent differences in the physical and chemical properties of constituent polymeric binders. However, the above anodes showed opposite surface morphology changes after cycling, i.e., cycled PVdF-based Si/C anodes featured larger numbers of deep big cracks than those in pristine ones, whereas cycled P84-based Si/C anodes exhibited less surface cracks than those in pristine ones. Because newly formed cracks consume large electrolyte amounts for SEI formation on the fresh active material surface and form dead active Si materials, cracking can decrease electrode capacity and increase interfacial resistance during cycling,[33,34] in agreement with our previous experimental results discussed above.

Figure 7

SEM images of the surfaces of (a–d) P84-based Si/C anodes and (e–h) PVdF-based Si/C anodes before and after the 10th cycle relevant to Figure a.

Adhesion inside electrode composites as well as between the electrode composite and current collectors significantly affects the electrode cycling performance.[35−38] To investigate the effect of polymeric binder type on the adhesion properties of Si/C, they were evaluated using the 90° peel test and SAICAS.

As shown in Figure a, P84- and PVdF-based Si/C anodes showed typical adhesion strength profiles, with the former showing a higher adhesion strength than that of the latter (0.0594 and 0.0346 kN m–1, respectively).[5,8] On the basis of Griffith’s theory of cracking, however, the material fracture is originated due to the presence of microscopic flaws in the bulk material, i.e., if microscopic flaws are present in the material, the fracture stress increases and the materials can be easily fractured and/or torn apart in spite of the inherent mechanical strength.[39] That is, the adhesion strength measured by the peel test cannot represent the adhesion properties of the Si/C anodes properly. In view of the above, the specific adhesion properties of Si/C anodes were evaluated using SAICAS.

Figure 8

(a) Adhesion strength profiles of P84- and PVdF-based Si/C anodes determined by the 90° peel test. (b) Interfacial adhesion strength (Finter) and (c) interlayer adhesion strength (Fmid, measured at 15 μm from the surface) of P84- and PVdF-based Si/C anodes. (d) Finter of P84- and PVdF-based Si/C anodes after 10 cycles relevant to Figure a.

SAICAS allows one to determine the adhesion properties of Si/C anodes at a specific position by adjusting the blade depth relative to the electrode surface. For convenience, the interlayer adhesion (Fmid) in Si/C composites was measured at 15 μm from the surface, with the interfacial adhesion between Si/C composites and Cu current collectors denoted Finter.

As depicted in Figure b,c, P84-based Si/C anodes (Fmid = 0.2422 ± 0.01 kN m–1, Finter = 0.2728 ± 0.004 kN m–1) showed better adhesion properties than those of PVdF-based Si/C anodes (Fmid = 0.2214 ± 0.02 kN m–1, Finter = 0.2214 ± 0.001 kN m–1).

After cycling performance tests relevant to Figure a, both anodes were disassembled after 10 cycles and their Finter values were measured and compared. In both cases, Finter decreased after cycling, with this decrease being less pronounced for P84-based Si/C (92.9% of the initial Finter value, from 0.2728 ± 0.004 to 0.2534 ± 0.01 kN m–1) than for PVdF-based Si/C anodes (83.7% of the initial Finter value, from 0.2214 ± 0.001 to 0.1854 ± 0.01 kN m–1). This result is closely related to the larger mechanical strength of P84, which is less prone to swelling than PVdF.[6,17]

Thus, SAICAS-determined adhesion properties revealed that the firm interlayer adhesion of the Si/C anode composite and its interfacial adhesion to Cu current collectors contributed to its enhanced performance.

Conclusions

Herein, we showed that highly adhesive and soluble co-polyimide polymeric binders (e.g., P84) can improve the electrochemical properties of Si/C anodes such as cycling performance and rate capabilities, additionally facilitating the anode preparation process. Because of the inherent physical properties of P84, P84-based Si/C anodes showed better adhesion properties (Fmid and Finter) and less pronounced adhesion decrease upon cycling than those of PVdF-based Si/C ones, which resulted in the enhanced cycling performances and rate capabilities of the former electrodes. Moreover, P84 diminished the severe morphological changes induced by charging/discharging and achieved a smaller total cell resistance after cycling.

Experimental Section

Electrode Preparation

Si/C anodes were prepared by coating a slurry of 5 wt % Si (30–50 nm, Nanostructured & Amorphous Materials), 75 wt % graphite (CGB-20, Nippon Graphite, Japan), 10 wt % conductive carbon (Super-P Li, IMERYS, Switzerland), and 10 wt % polymeric binder in N-methyl-2-pyrrolidone (NMP, Sigma-Aldrich) onto Cu current collector foil (11 μm, Iljin Materials, Republic of Korea) using a doctor blade technique. Two polymeric binders were used, namely, PVdF (Mw = 350 000, KF-1300, Kureha Battery Materials Co., Japan) and P84 (Mw = 150 000, HP Polymer GmbH, Germany) dissolved in NMP. The cast slurry was dried in air at 80 °C for 2 h and roll-pressed using a gap-control-type roll-pressing machine (CLP-2025, CIS, Republic of Korea). The loading and thickness of Si/C anodes were controlled as ∼4 mg cm–2 and 30 μm in both cases.

Morphological Analysis of Si/C Anodes

The elemental distributions and surface morphologies of the prepared electrodes were examined by field emission scanning electron microscopy (MIRA LMH, TESCAN and FE-SEM, S4800, Hitachi) coupled with energy-dispersive X-ray spectroscopy (EDX; Genesis XM2, EDAX Inc.).

Cell Assembly

Si/C anodes were cut into disks (diameter = 12 mm) and assembled into 2032-type coin half-cells using Li metal (thickness = 200 μm, diameter = 16.2 mm; Honjo Metal Co., Japan) as the counter electrode. Polyethylene separators (thickness = 20 μm, diameter = 18 mm; ND420, Asahi Kasei) were employed, and 1.15 M lithium hexafluorophosphate in ethylene carbonate/diethyl carbonate (3:7, v/v) containing 5 wt % fluoroethylene carbonate (ENCHEM, Republic of Korea) was used as an electrolyte. Cell assembly was conducted in an argon-filled glove box with a dew point below −60 °C.

Electrochemical Measurements

The electrochemical properties of Si/C anodes were probed by cyclic voltammetry (CV) using an impedance analyzer (VSP, Bio-Logic). The prepared 2032-type unit cells (Si/C half-cell) were scanned at a rate of 0.1 mV s–1 in a potential range of 0.005–1.5 V vs Li/Li+, starting from the open circuit voltage (OCV).

The assembled cells were aged for 12 h and then cycled between 0.005 and 1.5 V in constant current mode at a rate of 0.06 A g–1 and 25 °C in both charge and discharge modes using a charge/discharge cycler (PNE Solution, Republic of Korea).

After the above cycling, the total cell impedance was measured using an impedance analyzer (VSP, Bio-Logic) in a frequency range of 1 MHz to 0.01 Hz.

To evaluate the cycling performance, cells were subjected to 100 cycles at a current density of 0.06 A g–1 at 25 °C. Furthermore, to evaluate rate capabilities, the discharge current densities were varied from 0.06 to 3 A g–1, with the charging current density maintained at 0.06 A g–1.

90° Peel Test

A 20 mm × 20 mm anode sample was attached to 3 M adhesive tape, and peel strength was measured using a micro material tester (Instron 5848, Instron Company). The adhesive tape was removed by peeling at an angle of 90° and a constant displacement rate of 10 mm min–1, with the applied load being continuously measured to construct force/displacement plots.

SAICAS Measurements

The interlayer adhesion of the Si/C composite (Fmid) and the interfacial adhesion between Si/C anodes and Cu current collectors (Finter) were characterized by SAICAS (Daipla Wintes Co., Ltd.) utilizing a 1 mm wide boron nitride blade fixed at a shear angle of 45°. For measuring Fmid, the blade was positioned 15 μm from the anode surface. During the test, the blade moved in the horizontal direction at 0.2 μm s–1, maintaining a vertical force of 0.2 N.