Characterization of poly(acrylic acid-co-UPy acrylate)-grafted-PEG, PAU-g-PEG 1 and poly(acrylic acid)-grafted-PEG (PAA-g-PEG) 2
1H-NMR and FT-IR spectroscopic analysis was employed for the structural analysis of PAU-g-PEG 1 and the control polymer PAA-g-PEG 2 (Figure S2 and S3).
The peak observed for copolymer 1 near 12.2 ppm was for carboxylic acid (–COOH1), which indicated that the tBA groups had undergone hydrolysis successfully. The three peaks at 7.45, 7.04, and 5.72 ppm were due to the protons of the UPy molecules (namely, H2, H3, H4), and this showed that the monomer including UPy was successfully copolymerized for copolymer 147. The peaks near 3.48 ppm were for the protons in the PEG copolymers (H5), demonstrating that PEG had been grafted onto the polymers. Lastly, the peaks around 1–3 ppm were for the protons of the alkyl chains (H6, H7, H8) in the copolymer backbone. The ratio of UPyA:acrylic acid:PEG-MEMA in the PAU-g-PEG copolymer 1, that is x:y:z, was calculated to be 1:70:1.6 based on the comparative integral ratio of the proton peaks (H1:H2,3,4:H5). This ratio was slightly different from the feed ratio used in the reactions, suggesting different reactivity between the comonomers (Table S1).
The PAA-g-PEG copolymer 2, prepared as a control, was also analyzed by 1H-NMR spectroscopic analysis, and the ratio of acrylic acid to PEG-MEMA was calculated from the integral ratio of proton peaks of carboxylic acid (H1) and PEG (H1) to be 70:1.6 (that is, y:z), the same as the ratio of acrylic acid to PEG-MEMA in PAU-g-PEG 1 (Figure S2 and Table S1).
The FT-IR spectra of PAU-g-PEG 1 and PAA-g-PEG 2 were further analyzed. The peak at 1758 cm−1 (C=O stretching) in the ester group was observed, together with the broad peak at 2,500–3,500 cm−1 (O–H stretching) and peak at 1705 cm−1 (C=O stretching) originated from the carboxylic acid group of PAA, confirming the successful grafting of both UPy and PEG unit onto PAA (Figure S3).
Thermal stability of the polymeric binder
The thermal decomposition behaviors of the synthesized copolymers 1, 2 and PAA were confirmed using the TGA analysis (Figure S4). All three exhibited similar decomposition behavior because they had the same PAA unit as the main chain. The first degradation of PAU-g-PEG, PAA-g-PEG and PAA was at 180–322 °C (weight loss of 22 wt%), 179–322 °C (weight loss of 21 wt%) and 184–280 °C (weight loss of 9 wt%). This weight loss corresponded to dehydration, forming PAA anhydrides. In case of PAU-g-PEG and PAA-g-PEG, decomposition of both UPy and PEG units grafted onto the polymer side chain additionally occurred, causing a higher weight loss54. All three binders (PAU-g-PEG, PAA-g-PEG and PAA) showed high thermal stability up to about 180 °C.
Mechanical properties of the polymeric binders used for silicon electrodes
An 180° peel-off test of the Si electrode made with the newly developed polymer binders was conducted to determine the physical stability, that is the adhesive force, of PAU-g-PEG 1, which was designed to have both self-healing ability and ionic conductivity. The results were compared to that of pristine poly(acrylic acid) (PAA) with a molecular weight (Mw = 100,000 g mol−1) similar to that of copolymers 1, and also with PAA-g-PEG 2, which has Li+-conductivity but no self-healing sites (Fig. 2a). All three electrodes had the same weight ratio (SiNP:Super P:binder = 6:2:2) and controlled thickness.
The results showed that the adhesive force of the polymers was in the order of PAU-g-PEG > PAA > PAA-g-PEG. The PAA itself had relatively high adhesion due to its interactions with the hydroxyl (–OH) functional groups on the silicon surface and the oxidized surface of the current collector. The adhesive force of copolymer PAA-g-PEG 2, prepared by grafting just PEG onto the PPA main chain, was slightly lower than PAA. This can be attributed to the weakened interactions among electrode components in the presence of PEG, which engages fewer hydrogen bonds compared to the PAA copolymer.
The PAU-g-PEG copolymer which was comprised of both PEG grafted onto the PAA main chain and the self-healing UPy molecules showed a high adhesive force of 1.6 N, significantly higher than that of PAA polymers, even with the grafting of PEG (Fig. 2b). The interactions between the electrode components were reinforced by the introduction of UPy, through strong H-bonding, and the UPy molecules further contributed to the enhanced physical properties of the polymers by serving as physical crosslinking sites between the polymer chains and also with Si particles, to form a more stable 3D network, as reported before55.
Ultimately, the PAU-g-PEG copolymer resolved the problem of weakened adhesive force caused by PEG grafting, by introducing UPy molecules, and it achieved higher physical stability than PAA. This enhanced the physical properties of the PAU-g-PEG copolymer binder, and was expected to accommodate the volume expansion of Si more effectively (as will be discussed later).
Self-healing ability of the polymeric binder
Next, we assessed the self-healing ability of copolymer PAU-g-PEG 1, which had been found to significantly enhance the adhesive force of the electrodes through strong hydrogen bonds (Fig. 3). To this end, the synthesized copolymer 1 was firstly hydrogelated using the freeze/thaw method as follows. A 35 wt% PAU-g-PEG solution was prepared and hydrogelated by freezing for 1 h at − 20 °C and thawing for 24 h at room temperature (Figure S5a)56.
In contrast, the PAA 35 wt% solution formed by the same process did not retain its hydrogelated form after the freezing/thawing process, and returned to a liquid state (Figure S5b), suggesting that the UPy molecules of PAU-g-PEG contributed to its ability to remain in gel form.
Next, the self-healing ability of the copolymer PAU-g-PEG 1 was examined as follows.
The prepared PAU-g-PEG hydrogel was first cut into two pieces, one of which was dyed red to help illustrate the self-healing process. The pieces were stored at room temperature in contact with each other and without any special treatment. About 3 h later, it can be seen that the red dye gradually diffused from the surface of the cut hydrogel, and after about 24 h, the dye was completely spread and there was almost no trace of slicing activity. Overall, the self-healed hydrogel maintained the surface contact in a stable manner (Fig. 3).
Despite its low content of UPy molecules at 1.4 mol%, PAU-g-PEG 1 exhibited the characteristics of a self-healable supramolecular polymer, based on strong hydrogen bonds.
Electrochemical cell performance of the silicon electrodes
Silicon electrodes were fabricated using PAU-g-PEG and PAA-g-PEG copolymers as well as PAA as binders, and their electrochemical performance was assessed using both a coin cell and 3-electrode cell (Fig. 4). The coin cell, in the form of a half cell, was comprised of a silicon electrode as the working electrode, and a lithium metal disk as the counter and reference electrode. For the 3-electrode cell, a piece of lithium metal was used as the reference electrode.
To assess their long-term cycling performance during the charge/discharge process, three electrodes with a mass loading level in the range of 0.5–0.6 mg cm−2 were charged/discharged at a low current density of 0.1 C and a voltage range of 0.01–1.5 V for the initial formation cycles (3 cycles), and at 0.5 C in the following cycles (Fig. 4a).
The PAA-based electrode ([email protected]) showed the lowest initial capacity (2,575.7 mAh g−1) and fast capacity fading in the early cycles, but the capacity was stabilized in later cycles.
The PAA-g-PEG-based silicon electrode ([email protected]g-PEG) showed a higher initial capacity (2,716.9 mAh g−1) than that of PAA, but also experienced severe initial capacity decay. The poorest capacity retention, with capacity dropping below that of even the PAA electrode after 200 cycles, was obtained for the [email protected]g-PEG electrode.
Meanwhile, the silicon electrode with the PAU-g-PEG binder ([email protected]g-PEG) displayed the highest initial capacity of 2,946.7 mAh g−1, and also showed the highest capacity retention among the three electrodes.
The capacity and capacity retention of the three silicon electrodes at 350 cycles were 1,450.2 mAh g−1, 53.3%; 356.6 mAh g−1, 18.8%; and 212.8 mAh g−1, 8.8% respectively, or PAU-g-PEG > PAA > PAA-g-PEG when arranged in increasing order. These results are consistent with the physical stability of the electrodes. The hydrogen bonding ability of PAA, together with the self-healing property of UPy, in the PAU-g-PEG electrode is believed to have effectively alleviated the structural deterioration caused by volumetric changes in the silicon during the charge/discharge process.
Moreover, both the PAU-g-PEG and PAA-g-PEG electrodes showed a higher initial capacity than the PAA electrode. This result is ascribed to the increase in ionic conductivity via the Li+ conductive pathway formed by grafting PEG in these two binders (to be discussed later). The PAU-g-PEG binder, in particular, was more physically stable than the other two binders, which was advantageous to the formation of the lithium ion pathway, and helped to achieve the highest initial capacity.
The initial Coulombic efficiency of the three binders was in the order of (PAA-g-PEG: 74.2%), (PAU-g-PEG: 70.4%) and (PAA: 51.6%). The two electrodes fabricated from PEG-grafted copolymers, PAA-g-PEG and PAU-g-PEG, showed good Coulombic efficiency, higher than 70% (Fig. 4b). The PAA-g-PEG electrode had a slightly higher Coulombic efficiency than the PAU-g-PEG electrode due to differences in the activation of the active material, that is Si, arising from differences in electrolyte uptake.
To determine differences in the electrolyte uptake of the silicon electrodes in relation to binder type, the three electrodes were stored in an electrolyte for 48 h under the same conditions, and swelling ratios were measured based on the weight changes of each electrode. First of all, PAA showed the highest electrolyte uptake, due to the strong H-bonding of the carboxylic acid groups in PAA with the carbonate-based electrolyte [ethylene carbonate (EC) and ethyl methyl carbonate (EMC)]. The PAA-g-PEG electrode, on the other hand, showed the lowest electrolyte uptake, due to its weakest H-bonding, caused by the grafting of PEG onto PAA. The PAU-g-PEG electrode, whose reduced hydrogen bonding interactions by PEG-grafting were compensated by UPy molecules, showed a higher electrolyte uptake (Figure S6).
It was, therefore concluded that the [email protected]g-PEG electrode had a relatively low lithiation capacity because of the low electrolyte uptake of this binder (PAA-g-PEG), which limits the activation of the active material in its initial cycle. The [email protected]g-PEG electrode, with its high electrolyte update, had a high lithiation capacity (Fig. 4c).
Afterwards, the [email protected]g-PEG electrode reached a high Coulombic efficiency of 94.0% in just two cycles, higher than that of PAA-g-PEG (92.6%) and PAA (81.1%). After the initial capacity fading, it quickly stabilized, and a Coulombic efficiency higher than 99% was maintained throughout. After 350 cycles, the Coulombic efficiency remained at a high of 99.4% (Fig. 4b). Overall, the [email protected]g-PEG electrode showed outstanding initial activation of silicon, and achieved high reversible capacity by overcoming the physical stress caused by volumetric change, through the stabilized ion conductive pathway.
Next, a C-rate performance test was conducted to determine rate capability under high current densities (Fig. 5a). The current density was fixed at 0.1 C for charging, and varied from 0.1 to 3 C for discharging.
The C-rate performance test revealed that the two electrodes fabricated from PEG polymers (PAU-g-PEG and PAA-g-PEG) had high rate capabilities. In particular, the [email protected]g-PEG electrode with the best cycling performance achieved a high capacity of 2,753, 2,590 and 2,519 mAh g−1 even under a high current density of 1, 2 and 3 C, respectively. When the current density was dropped to 0.1 C again, it demonstrated excellent rate capability, with a recorded capacity of 2,800 mAh g−1, which is close to its initial capacity of 2,900 mAh g−1.
The [email protected]g-PEG electrode also showed good rate capability and recovery (although not better than the PAU-g-PEG electrode) under all C-rate conditions. From these results, we can conclude that grafting PEG onto the PAA played a key role in enhancing ionic conductivity, and in turn, rate capability.
Compared with the recently reported polymer binders for silicon anodes, the electrode made of PAU-g-PEG showed moderate capacity at relatively high C-rate and cycle numbers, and excellent rate performance (Table S2).
The Si electrodes using the PAU-g-PEG binder were also prepared with various mass loading conditions of the active material, and their capacity retention was compared, to further evaluate the electrochemical properties of the [email protected], which exhibited the best cycle and rate characteristics (Fig. 5b). This was performed at high loading levels of 0.8 mg cm−2 and 1.0 mg cm−2, higher than the condition of 0.5 mg cm−2 at which stable cycle performance was obtained. The evaluation was conducted at 0.5 C for all three cells.
Although capacity decreased with increasing Si loading level, the initial capacity fading quickly stabilized for all three electrodes, and they subsequently showed good capacity retention. The electrodes fabricated with mass loadings of Si of 0.8 mg cm−2 and 1.0 mg cm−2 at 200 cycles maintained a capacity higher than 1,400 mAh g−1 and 1,200 mAh g−1 respectively. These results indicate that PAU-g-PEG binders can be utilized in silicon electrodes even at high loading levels, and that such electrodes are suitable for batteries with high energy densities.
The cycling performance of the [email protected] electrode with low binder content (10 wt%) was also obtained at 0.5 C (Figure S7). The capacity and capacity retention of this electrode at 100 cycles were 1,333.3 mAh g−1 and 63.3%, respectively, indicating that the PAU-g-PEG binder showed good performance even with a high active material and a low binder content.
Next, cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) measurements were carried out to examine differences in the electrochemical properties of the silicon electrodes ([email protected]g-PEG, [email protected]g-PEG and [email protected] in relation to their binder type (Fig. 6).
CV measurements were taken for all three electrodes under the same voltage range and scan rate conditions, and EIS measurements were obtained under the same charge/discharge conditions (0.5 C) and frequency range. The CV measurements showed that all three electrodes reached a cathodic peak at 0.21 V, corresponding to the lithiation of Si, and anodic peaks at 0.33 V and 0.50 V, corresponding to the delithiation process. All electrodes also experienced reversible electrochemical reactions (Fig. 6a–c). In addition, irreversible peaks were initially observed for all three electrodes in the range of 1.0–0.5 V due to the formation of an SEI layer in the cathodic scan domain of the first cycle, but such peaks were not seen in the second cycle. This means that side reactions did not occur in the charge/discharge voltage range, which verifies the electrochemical stability of the UPy-functionalized and PEG-grafted PAA, that is PAU-g-PEG, as a polymeric binder for silicon anodes.
Next, EIS measurements were taken for the three electrodes during the 5th and 50th charge/discharge process under a current density of 0.5 C after three formation cycles at 0.1 C. The first semi-circle represents the resistance (RSEI) due to the SEI layer, and the second semi-circle represents the resistance (RCT) due to charge transfer (Fig. 6d–f)57,58.
The results showed that both electrodes fabricated from the PAA-g-PEG and PAA binders showed an increase in RSEI after cycles, but the size of the semi-circle was larger for the [email protected]g-PEG electrode, which also had a higher RSEI after both 5 cycles and 50 cycles than the [email protected] (Fig. 6e,f). This means that the silicon electrode with the PAA-g-PEG binder, which had the weakest physical stability, formed a thick SEI layer, as it was unable to tolerate the stress caused by volumetric expansion during the charge/discharge process.
As for RCT, the [email protected]g-PEG electrode revealed a lower resistance value than [email protected] due to the enhanced lithium ion conductivity provided by the PEG domain. RCT is influenced by both electronic and ionic resistance, since charge transfer involves the transfer of both electrons and lithium ions, and the electrode made of the binder (PAA-g-PEG), which had poor mechanical properties, lost electronic and ionic contact after cycling, leading to an increase in RCT38,59.
The [email protected]g-PEG electrode, which had both ionic conductivity and self-healing ability, was found to have much lower RSEI and RCT values, and maintained an impedance similar to the 5 cycles level, even after 50 cycles (Fig. 6d). This can be attributed to the formation of a strong electrical network between active materials due to the excellent in-electrode mechanical stability of the PAU-g-PEG binder, and low resistance due to the formation of ion transport channels by PEG. In particular, these results are attributed to the effective accommodation of physical stress produced by the volume expansion of Si, by introducing the self-healable Upy unit into the polymer binder.
Overall, the PAU-g-PEG binder alleviated the increase in resistance, by continuously recovering the structural collapse of the Si electrode, and also by maintaining electrical contact between active materials. These results are consistent with the good electrochemical performance data exhibited by the [email protected]g-PEG electrode.
Lithium ion diffusion rate by polymeric binders
In addition, the lithium ion diffusion coefficients (DLi+) of the three electrodes ([email protected]g-PEG, [email protected]g-PEG and [email protected]) were calculated and compared using the Warburg factor values obtained from the EIS measurements and GITT method of their formation cycles (Figure S8).
The calculated DLi+ from the EIS measurements was 8.02 × 10–10 cm−2 s−1, 3.88 × 10–10 cm−2 s−1, and 3.49 × 10–11 cm−2 s−1 for [email protected]g-PEG, [email protected] and [email protected], respectively. That is, the [email protected]g-PEG electrode had a lithium ion diffusion coefficient ten times higher than that of the PAA binder electrode, which is an indicator of the lithium ion conductivity of PEG in the polymeric structure.
Additionally, the average value of lithium ion diffusion coefficients during charging was calculated from the GITT method60. The average DLi+ was 5.81 × 10–12 cm−2 s−1, 3.88 × 10–12 cm−2 s−1 and 3.05 × 10–13 cm−2 s−1 for [email protected]g-PEG, [email protected]g-PEG and [email protected] Although there are some differences in the values from GITT compared to those from EIS, the trend was the same.
Here, it should be noted that the [email protected]g-PEG electrode, comprised of both PEG and self-healable UPy unit, had a lithium ion diffusion coefficient two times higher than that of the [email protected]g-PEG electrode. This can be traced to the excellent mechanical stability of the [email protected]g-PEG electrode, as verified in the earlier peel-off test. That is, the physical crosslinking through the UPy moiety not only formed a more robust 3D network, but also helped to maintain this high structural stability during the charge/discharge cycles. As a result, the lithium ion pathway formed by the grafting of PEG was also well maintained, indicating that the [email protected] electrode had higher ion conductivity than the [email protected] electrode.
Morphological analysis of the silicon electrodes
Morphological changes in the silicon electrodes after cycling were observed by scanning electron microscope (SEM) analyses (Fig. 7). The silicon electrodes fabricated from the three types of binders (PAU-g-PEG, PAA-g-PEG and PAA) all showed uniform morphologies before cycling, demonstrating the binders were well dispersed, due to interactions between the PAA main chain and silicon particles (Figure S10). After 50 cycles, however, very different morphologies were observed in the three electrodes.
At first, while slight cracks were seen in the [email protected]g-PEG electrode after 50 cycles, it was found that the morphology did not change much compared to that before cycling (Figure S10a and Fig. 7a).
Both the [email protected]g-PEG and [email protected] electrodes, on the other hand, not only experienced severe cracking after 50 cycles, but also displayed non-uniform morphologies (Fig. 7b,c). This phenomenon is more apparent in their magnified images. The [email protected]g-PEG electrode maintained a uniform SEI layer, whereas the [email protected]g-PEG and [email protected] electrodes formed thick, non-uniform SEI layers (Fig. 7a′–c′).
Furthermore, thickness changes after formation cycles were analyzed through cross-sectional SEM images (Figure S11). The thickness increase of [email protected], [email protected] and [email protected] electrodes was 8.57%, 19.0% and 31.4% respectively, when compared to thickness of the pristine electrodes. It can be seen that PAU-g-PEG binder showed the lowest thickness increase and maintained the good contact with the current collector, indicating excellent structural stability of electrode.
The above SEM results indicate the advantages of the PAU-g-PEG binder, which was capable of self-healing and had high physical stability due to the formation of a stable 3D network through physical crosslinking. These morphological properties were consistent with the results of the cyclability test and EIS, in which [email protected]g-PEG had the most outstanding capacity retention and lowest resistance.