Bin
Luo
a,
Jintian
Wu
ac,
Ming
Zhang
a,
Zhihao
Zhang
a,
Xingwei
Zhang
a,
Zixuan
Fang
*a,
Ziqiang
Xu
*ab and
Mengqiang
Wu
*ab
aSchool of Materials and Energy, University of Electronic Science and Technology of China, Chengdu 611731, Sichuan, China. E-mail: zixuanfang@uestc.edu.cn; nanterxu@uestc.edu.cn; mwu@uestc.edu.cn
bYangtze Delta Region Institute (HuZhou), University of Electronic Science and Technology of China, Huzhou 313001, Zhejiang, China
cSchool of Chemical Engineering, Sichuan University of Science & Engineering, Zigong 643000, China
First published on 27th October 2023
The poly (vinylidene fluoride) (PVDF)-based composite solid-state electrolyte (CSE) has garnered attention due to its excellent comprehensive performance. However, challenges persist in the structural design and preparation process of the ceramic-filled CSE, as the PVDF-based matrix is susceptible to alkaline conditions and dehydrofluorination, leading to its incompatibility with ceramic fillers and hindering the preparation of solid-state electrolytes. In this study, the mechanism of dehydrofluorination failure of a PVDF-based polymer in the presence of Li2CO3 on the surface of Li6.4La3Zr1.4Ta0.6O12 (LLZTO) is analyzed, and an effective strategy is proposed to inhibit the dehydrofluorination failure on the basis of density functional theory (DFT). We introduce a molecule with a small LUMO–HOMO gap as a sacrificial agent, which is able to remove the Li2CO3 impurities. Therefore, the approach of polyacrylic acid (PAA) as a sacrificial agent reduces the degree of dehydrofluorination in the PVDF-based polymer and ensures slurry fluidity, promoting the homogeneous distribution of ceramic fillers in the electrolyte membrane and enhancing compatibility with the polymer. Consequently, the prepared electrolyte membranes exhibit good electrochemical and mechanical properties. The assembled Li-symmetric cell can cycle at 0.1 mA cm−2 for 3500 h. The LiFePO4‖Li cell maintains 91.45% of its initial capacity after 650 cycles at 1C, and the LiCoO2‖Li cell maintains 84.9% of its initial capacity after 160 cycles, demonstrating promising high-voltage performance. This facile modification strategy can effectively improve compatibility issues between the polymer and fillers, which paves the way for the mass production of solid-state electrolytes.
The SSEs are categorized into solid-state polymeric electrolytes (SPEs), inorganic solid-state electrolytes (ISEs) and composite solid-state electrolytes (CSEs), based on the components.4 CSEs are highly regarded for their combination of the excellent ionic conductivity and mechanical strength of ISEs, along with the low interfacial impedance of SPEs. The CSEs mainly consist of an inorganic filler, polymer matrix and lithium salt. Common polymer matrices include PEO, PVDF, PAN and PPC.5 Among them, PVDF has been widely studied because of its good comprehensive performance, especially for its high dielectric constant (εr).6,7 As for the inorganic fillers, they can be divided into inert fillers (e.g. SiO2, Al2O3, TiO2, ZrO2, etc.) and active fillers (e.g. Li6.4La3Zr1.4Ta0.6O12 (LLZTO), Li1.5Al0.5Ti1.5(PO4)3, Li7Ge2P5S10, Li3PS4, etc.) according to their ability to conduct ions. Among them, garnet-type solid electrolyte LLTZO has excellent conductivity performance, high chemical stability and a good shear modulus.8
However, LLZTO is prone to reacting with water and CO2 in an air atmosphere to form an alkaline layer of Li2CO3 on its surface, which leads to its lithiophobicity and a decrease in conductivity.9,10 Moreover, an alkaline environment resulting from the Li2CO3 layer on the LLZTO surface will cause dehydrofluorination of PVDF, which can be understood by using the following equation,11
2(CH2–CF2)n + Li2CO3 → 2(CHCF)n + 2LiF + H2O + CO2 | (1) |
The generated CC unsaturated bonds may cross-link and reduce the fluidity of the slurry or even gelate, resulting in the components not being homogeneously mixed,12 which blocks the migration of Li+, weakening the ionic transport and mechanical properties of CSE.13,14 In addition, the formation of CC bonds reduces the flexibility of the molecular chains, which increases the interfacial resistance of PVDF and the fillers.14,15 Moreover, Yi et al. found that the defluorination of a PVDF-based polymer impairs the compatibility of the electrolyte membrane with electrodes,15 and gelation is an obstacle to mass production in slurries with PVDF-based polymers.5,15,16 Various strategies such as rapid acid treatment,17 using high-temperature decomposition18 and high-speed mechanical polishing19 have been proposed to solve these problems. More recently, Lee et al. employed dry etching to treat the surface of LLZO to remove Li2CO3.20 This treatment promoted the formation of fast Li+ conductive pathways along the LLZO fillers. The composite electrolyte membrane prepared with a PVDF matrix has good ionic conductivity and electrochemical properties. However, dry etching may lead to a potential risk of partial transition of cubic LLZO into the tetragonal phase. In addition, these methods are too complex and expensive. Therefore, it is critical to develop a facile design strategy to enhance compatibility among components based on CSEs.
In this work, we introduce polyacrylic acid (PAA) with a smaller LUMO–HOMO gap than poly (vinylidene fluoride-co-chlorotrifluoroethylene) (P(VDF-CTFE)) as a sacrificial agent to modify the surface of LLZTO by DFT analysis. The CSEs with ceramic fillers and P(VDF-CTFE) are prepared by the solution casting method. According to the DFT, the oxygen atom in the carboxyl group of PAA is the main electrophilic site, leading to the reaction of PAA with Li2CO3 on the surface of LLZTO. The degree of dehydrofluorination of P(VDF-CTFE) is reduced owing to the removal of alkaline substances, thus avoiding a decrease in the fluidity of the slurry that would prevent the coating of the membrane. The modifications improve the compatibility between the ceramic and matrix and ensure the homogeneous distribution of the filler, facilitating fast and uniform Li+ transport. Therefore, the prepared P(VDF-CTFE)@LLZTO-PAA CSE exhibits an excellent ionic conductivity of 4.53 × 10−4 S cm−1, a Li+ transfer number of 0.336 and an enlarged electrochemical window of 4.5 V (vs. Li+/Li). The as-assembled Li‖P(VDF-CTFE)@LLZTO-PAA CSE‖LiFePO4 can stably cycle 650 times at 1C with an initial specific discharge capacity of 143.8 mA h g−1, and has a capacity retention of 91.45%. This study presents a facile and effective approach for modifying alkaline ceramic fillers to be compatible with polymeric matrices and the large-scale preparation of composite solid-state electrolytes, which have excellent potential for mass production in the industry.
It is generally considered that the vibration modes of the organic functional group are very sensitive to the Raman spectra, and Fig. 1(c) displays the characterization results for LLZTO, PAA and LLZTO-PAA. As depicted in Fig. 1(c), similar phenomena are observed in the Raman spectra. First, the peak belonging to the carboxyl group at 1670 cm−1 can be found in the spectrum of PAA, and the characteristic peaks corresponding to the vibration mode of –CH2 and the stretching mode of –CH2 or –CH can be found at 1100 and 2936 cm−1, respectively.22,25 As for the spectrum of LLZTO, the peaks at 294, 370, 684 and 771 cm−1 are attributed to the garnet phase, which can also be found in the spectrum of LLZTO-PAA, indicating that the modulation of the PAA sacrificial agent does not change the cubic structure of the ceramics.17 In particular, the peak at 1086 cm−1 associated with carbonate vibration which further demonstrates the presence of Li2CO3. In contrast, after PAA modification, the vibrational peak of Li2CO3 disappears from the LLZTO-PAA spectra, indicating that the alkaline contaminated layer is removed from the surface of LLZTO. The peak at 1670 cm−1 associated with the carboxyl group disappears, but a new peak corresponding to the symmetric vibration of –COO− appears at 1423 cm−1 and a peak belonging to –CH2 or –CH is present at 2926 cm−1.26 Both FTIR and Raman spectra indicate that the surface treatment of LLZTO can effectively remove Li2CO3 content. Specifically, the carbonate group on the surface of LLZTO will react with the carboxyl group of PAA, which is represented by the following equation,
CO32− + 2 − COOH → 2COO− + H2O + CO2 | (2) |
The strategy of surficial modification by a sacrificial agent is effective to remove the alkaline and low-conducting Li2CO3 from the ceramic surface, which is beneficial to ensure the ionic conductivity of LLZTO. The results of exploring suitable pH values by using different weight ratios of LLZTO/PAA are shown in Fig. S1.† The initial solution of LLZTO has a pH of 10.10, which gradually decreases with the addition of PAA. When the weight ratio is 3:1, the pH of the mixture decreases to 7.28, and PAA essentially neutralizes the alkalinity of LLZTO.
In order to further investigate the rationality of PAA as a sacrificial agent, the properties of PVDF fragments and PAA are studied, and the energy values of the LUMO and HOMO of the PVDF fragment and PAA are calculated by using DFT, respectively (Fig. 1(d)). PAA exhibits a smaller LUMO–HOMO gap of 0.1841 Ha than the PVDF fragment (0.2717 Ha), indicating that it is more susceptible to redox reactions and electron gain/loss. Meanwhile, combining the results of FTIR and Raman experiments, PAA is considered suitable to be used as a sacrificial agent to remove Li2CO3, and can inhibit the dehydrofluorination of the matrix. To further determine the electrophilic and nucleophilic sites, the electrostatic potentials (EPS) of the PVDF fragment and PAA are calculated and the distribution results of EPS on the molecular surfaces are visualized (Fig. 1(e)). It can be observed that the oxygen atoms in the carboxyl groups of PAA exhibit significantly negative electrostatic potential, primarily due to the strong contribution of its lone pair of electrons to the electrostatic potential. The calculated results of the electrophilic Fukui functions of the PVDF fragment and PAA shown in Fig. 1(f) also indicate that the oxygen atom in the carboxyl group is the main electrophilic site. The results showed that the PAA sacrificial agent is more prone to excitation, and its carboxyl group can effectively react with Li2CO3 and reduce its initiated dehydrofluorination reactions, inhibiting the cross-linking of the matrix, which facilitates uniform dispersion of the components in the CSE membrane and enhances ionic conductivity and mechanical properties.
The microstructure of the LLZTO-PAA particles is analyzed by TEM. The LLZTO-PAA particle has a core–shell structure with a 10–20 nm polymer shell layer attached to the surface of the LLZTO ceramic core (Fig. S2†). As shown in Fig. 1(g and h), a dense amorphous shell layer can be clearly observed, which clings to the ceramic surface (as marked by the green dash lines). The lattice fringe with a spacing of 0.541 nm is measured and displayed in Fig. 1(h), which is well matched to the (112) lattice plane of LLZTO.8,19 However, as for LLZTO without surficial treatment, the floccules are distributed on the surface of LLZTO particles (Fig. 1(j)), which may be correlated with the dislodged Li2CO3 during stirring. The lattice fringes are not uniformly arranged, and the lattice plane of (−202) with a lattice spacing of 0.295 nm belonging to Li2CO3 can be identified (Fig. 1(k)).27 The selected-area electron diffraction (SAED) patterns are also different between LLZTO-PAA and LLZTO. The SAED pattern of LLZTO-PAA exhibits the diffraction spots of the (112), (220), (332), and (004) crystalline planes that are related to the cubic phase of LLZTO (Fig. 1(i)). In contrast, the SAED pattern of LLZTO exhibits a series of diffraction rings, and in particular the (−202) lattice plane attributed to Li2CO3 is again observed (Fig. 1(l)). To investigate the distribution of PAA on the LLZTO surface, EDS mapping of LLZTO-PAA particles is performed (Fig. S3†), where it is observed that PAA can coat LLZTO particles.
To investigate the distribution of LLZTO in the SSE membrane, EDS mappings are performed on the cross section of the membrane, as shown in Fig. S6.† The distribution of the La element from LLZTO in the P(VDF-CTFE)@LLZTO-PAA CSE membrane is more uniform than that of the P(VDF-CTFE)@LLZTO CSE membrane, confirming the improvement of the distribution of ceramic fillers in the polymer matrix after PAA sacrificial agent modification. This result is probably related to the high compatibility of LLZTO-PAA with the polymer matrix and a lower degree of dehydrofluorination of P(VDF-CTFE), which allows for the homogeneous distribution of LLZTO in PVDF and facilitates the fluidity of the slurry to be maintained.
Fig. 3 (a) XRD curves of SSE membranes and ceramic fillers. (b) DSC heating curves, (c) FTIR spectra and (d) Raman spectra of SSE membranes. |
FTIR is used to explore the chemical structure and bonding transitions of the SSEs, as exhibited in Fig. 3(b). As for P(VDF-CTFE) SPE, peaks related to the vibration of the α-phase at 746 and 878 cm−1 are observed, and the peaks at 831, 1071, 1175, 1254 and 1379 cm−1 are correlated with the vibration of the β-phase.32 Among them, the absorption peaks located at 1175 cm−1 originated from the –CF2 vibration, which corresponds to 1166 cm−1 in the pure P(VDF-CTFE) membrane,33,34 due to the interaction of the polymer with the lithium salt.30 Meanwhile, the dissociation of LiFSI in P(VDF-CTFE) leads to new peaks at 1653 and 570 cm−1. Meanwhile, the β-phase content is calculated from the absorption intensities located at 746 and 831 cm−1 by using the equation (ESI†), and the results are shown in Table 1. When LLZTO-PAA is added, the β-phase content increases, which means the interaction between the fillers and P(VDF-CTFE).32 Moreover, the β-phase has high polarity (εr = 10–13), which facilitates the dissociation of the lithium salt.35 In addition, the Lewis acid–base interactions between the LLZTO fillers and the lithium salt also affect the dissociation of LiFSI.36 The changes in the β-phase content are also observed in P(VDF-CTFE)@LLZTO CSE and P(VDF-CTFE)@PAA SPE. The increase of the β-phase in P(VDF-CTFE)@PAA SPE is probably caused by the effect of carboxylic acid groups in PAA on the solidification process of the polymer matrix. However, the β-phase content in P(VDF-CTFE)@LLZTO CSE does not change significantly compared to that of the P(VDF-CTFE) SPE, which is related to the reason that the fillers do not fully come into contact with the polymer due to its inhomogeneous distribution and agglomeration.
SSE membrane | F (β) (%) | T m (°C) | ΔHm (J g−1) | χ c (%) |
---|---|---|---|---|
P(VDF-CTFE) | 45.37 | 161.70 | 20.86 | 21.34 |
P(VDF-CTFE)@LLZTO | 44.93 | 158.49 | 20.79 | 21.30 |
P(VDF-CTFE)@PAA | 46.38 | 154.06 | 19.78 | 20.22 |
P(VDF-CTFE)@LLZTO-PAA | 47.17 | 152.56 | 14.63 | 14.94 |
The DSC measurements are carried out to explore the melting temperature (Tm) and other thermodynamic parameters of the prepared SSE membranes. As shown in Fig. 3(c), the Tm is marked above the DSC curve of the corresponding SSE membranes. In comparison to the P(VDF-CTFE) SPE (161.70 °C), the Tm of the other three SSEs with additives decreases. P(VDF-CTFE)@LLZTO-PAA CSE has the lowest Tm (152.56 °C). According to equations (ESI†), the crystallinity (χc) of the corresponding SSE membranes is calculated (Table 1). The crystallinity of P(VDF-CTFE)@LLZTO CSE (21.30%) is higher than that of P(VDF-CTFE)@LLZTO-PAA CSE (14.94%). This finding is attributable to the non-uniform distribution of LLZTO fillers without surficial treatment in the polymer matrix, and its perturbing effect on the polymer chain segments is not significant.30 The χc of P(VDF-CTFE)@LLZTO CSE is not much different from that of the SSE membrane without a filler (21.34%), which could not contribute significantly to the transport of Li+. Furthermore, this finding suggests that the addition of LLZTO-PAA can effectively decrease the crystallinity of P(VDF-CTFE), which means an increase in the amorphous regions and enhanced chain segment movement within the matrix, leading to improvement of the transport of Li+ and an increase in ionic conductivity.21
Raman spectroscopy is an effective technique for studying structures containing conjugate CC double bonds, which are a sign of dehydrofluorination.37 As shown in Fig. 3(d), there are two main changes in the spectra of P(VDF-CTFE)@LLZTO CSE without PAA modulation compared to those of P(VDF-CTFE) SPE. The first change is the appearance of two new spectral bands at 1507 and 1117 cm−1, which are signatures of the polyene CC stretching vibrational mode. This phenomenon indicates that the polymer P(VDF-CTFE) has undergone a dehydrofluorination process to generate a conjugate CC double bond structure under the influence of alkaline LLZTO.38 In order to investigate the interaction between the PVDF fragment and Li2CO3, theoretical calculations are carried out for the system using DFT (Fig. S7†), and a strong interaction is observed between the PVDF fragment and Li2CO3 with an interaction energy of −1.24 eV. As displayed by the results in Fig. S7(b),† the length of the C–F bond (LC–F) of the PVDF fragment after adsorption is elongated to about 1.53 Å from 1.36 Å, indicating the obvious trend of defluorination, which is consistent with the Raman test results. The second change is a substantial decrease in intensity located at 1432 and 2977 cm−1, corresponding to the –CH tension vibrational and –CH2 bending modes, respectively.39 This finding indicates a decrease in –CH2 groups in an alkaline environment. In contrast, the above two changes are not found in P(VDF-CTFE)@ LLZTO-PAA CSE after the addition of a PAA sacrificial agent, and its Raman spectrum is almost identical to that of the P(VDF-CTFE) SPE and P(VDF-CTFE)@PAA SPE. This result indicates that surficial modification of LLZTO with PAA can inhibit the dehydrofluorination of the P(VDF-CTFE) matrix, which could improve the fabrication process of the electrolyte membrane.11
To study the ion transport in CSE membranes, the σ of the various membranes is shown in Fig. 4(a). The weight ratios of P(VDF-CTFE) to LLZTO and LiFSI are identified to be 10:1 and 3:2 (Fig. S10†), respectively. The σ of P(VDF-CTFE)@PAA SPE is 2.73 × 10−4 S cm−1, slightly surpassing that of P(VDF-CTFE) SPE (2.53 × 10−4 S cm−1), owing to the reduced crystallinity of the polymer matrix and the contribution of the carboxyl group to Li+ transport.39 Moreover, the σ of P(VDF-CTFE)@LLZTO-PAA CSE (4.53 × 10−4 S cm−1) is significantly higher compared to that of P(VDF-CTFE)@LLZTO CSE (1.93 × 10−4 S cm−1). This difference can be attributed to the uniform distribution of LLZTO-PAA, leading to a decrease in the crystallinity of P(VDF-CTFE) and the formation of a fast Li+ transport pathway dominated by the polymer matrix.40,41 Based on the above analysis and characterization results, the increase in conductivity of P(VDF-CTFE)@LLZTO CSE is mainly related to the following factors: (1) the dense structure of the membrane and the homogeneous distribution of the fillers facilitate the continuous transport of Li+; (2) the addition of LLZTO-PAA effectively reduces the crystallinity of P(VDF-CTFE), which promotes the migration of Li+.40 In addition, Fig. 4(b) shows the Arrhenius curves for different electrolyte membranes, where the σ of all the SSE membranes increases with increasing temperature due to the increased ability of the chain segments of the polymer matrix to move. It is observed that P(VDF-CTFE)@LLZTO-PAA CSE exhibits a minimum Ea of 0.206 eV, indicating that the Li+ can migrate through it with a lower migration barrier. In contrast, the Ea of P(VDF-CTFE)@LLZTO CSE is 0.256 eV, which may be related to the agglomeration and inhomogeneous distribution of the ceramic fillers in the CSE.42
The electrochemical window is a crucial parameter for assessing the electrochemical stability of an electrolyte. Fig. 4(c) shows the results of linear scanning voltammetry (LSV) measurements on electrolyte membranes. P(VDF-CTFE)@LLZTO CSE has an electrochemical window of 4.3 V (vs. Li+/Li). The electrochemical window for P(VDF-CTFE) SPE and P(VDF-CTFE)@PAA SPE is 4.1 V, which is less than that reported in other literature studies,43 attributed to the effect of residual solvent (Fig. S11†).44 Significantly, the P(VDF-CTFE)@LLZTO-PAA CSE exhibits a higher oxidation decomposition voltage of 4.5 V, which may be attributable to a more homogeneous distribution of ceramics and restrain activity of DMF.33,40,45 The enhanced electrochemical stability enables the electrolyte to effectively match with a high-voltage cathode to provide high energy density.
Good mechanical properties are essential for the safe use of SSE-based lithium batteries in practice, and are related to the ability of the electrolyte membrane to inhibit lithium dendrites. The mechanical performances of SSE membranes are evaluated by using stress–strain curves (Fig. 4(d)). The tensile strength (TS) of P(VDF-CTFE) SPE is 2.79 MPa and the maximum strain is 27.24%. The TS of P(VDF-CTFE)@LLZTO CSE is only 0.554 MPa after adding pure LLZTO due to the agglomeration of ceramics and poor film formation. However, with the addition of LLZTO-PAA, P(VDF-CTFE)@LLZTO-PAA CSE possesses a TS of 5.22 MPa and a maximum strain of 79.93%. These improved mechanical properties are due to the adhesion between the P(VDF-CTFE) matrix and the ceramic filler. Thus, the addition of LLZTO-PAA endows the electrolyte membrane with better mechanical stability to support the safety of the battery.33
Thermogravimetric analysis (TGA) conducted from 30 °C to 600 °C is performed to assess the thermal stability of the electrolyte membrane (Fig. 4(e)), which is also an essential factor regarding battery safety. The weight loss trend is consistent for all SSEs. The weight losses observed within the temperature ranges of 310 °C to 450 °C and 200 °C to 310 °C are caused by the decomposition of P(VDF-CTFE) and lithium salt, respectively.46 Trace amounts of water retention contribute to weight loss before 55 °C. Moreover, the weight loss between 55 °C and 195 °C can be attributed to the residual solvent DMF.6 Therefore, the amount of DMF can be roughly calculated to be approximately 12.3%, 9.2%, 8.9% and 8.3% in P(VDF-CTFE) SPE, P(VDF-CTFE)@LLZTO CSE, P(VDF-CTFE)@PAA SPE and P(VDF-CTFE)@LLZTO-PAA CSE, respectively. The flammability test is used to assess the safety of electrolyte films further. As depicted in Fig. 4(f), P(VDF-CTFE)@LLZTO-PAA CSE does not continue to burn after being heated by a heat source, which demonstrates its good thermal safety.
The Li‖SSEs‖Li symmetric cells are assembled and tested for constant current cycling to investigate the ability of SSEs to inhibit lithium dendrite growth. Fig. 5(b) shows the lithium plating and stripping behavior of the cells under the conditions of 25 °C and 0.1 mA cm−2. The cycle life of the cells with P(VDF-CTFE) SPE and P(VDF-CTFE)@PAA SPE is 1600 h and 2050 h, respectively. It is reported that the Li polyacrylic acid (LiPAA) formed by the PAA− anion and Li+ could improve the electrolyte's ability to resist volume changes in the lithium anode.31,49 The Li‖P(VDF-CTFE)@LLZTO CSE‖Li cell produces a huge polarization voltage after 850 h of cycling and the cell eventually fails in an open circuit mode. The failure here could be attributable to the lower tLi+, which causes anions to accumulate at the interface and form a space charge layer, resulting in a greater interfacial impedance.50 Interestingly, the Li‖P(VDF-CTFE)@LLZTO-PAA CSE‖Li cell maintains a stable polarization voltage over a 3500 h cycle, indicating its enhanced stability towards lithium metal and effective inhibition of lithium dendrite growth. As depicted in Fig. 5(c), the critical current density of Li‖P(VDF-CTFE)@LLZTO-PAA CSE‖Li is 0.6 mA cm−2, which exceeds that of Li‖P(VDF-CTFE)@LLZTO CSE‖Li (0.4 mA cm−2). The microscopic morphology of the lithium metal surface after cycling Li‖SSEs‖Li for 200h is observed by using SEM (Fig. 5(d and e)). The lithium surface of Li‖P(VDF-CTFE)SPE‖Li exhibits roughness and non-uniformity, while Li‖P(VDF-CTFE)@LLZTO-PAA CSE ‖Li has a flat surface and uniform lithium deposition. Due to the uneven deposition of lithium, the volume of the interface between the SSE membrane and lithium may change, eventually leading to cell failure.47,49 These results suggest that adding LLZTO-PAA fillers effectively promotes the uniform diffusion of Li+ and inhibits lithium dendrite growth between the electrolyte membrane and lithium metal electrode.
Fig. 6(b) demonstrates the cycle performance of the LFP‖SSEs‖Li cells at a rate of 1C. The initial discharge capacity of the LFP‖P(VDF-CTFE)@LLZTO CSE‖Li cell is 134.8 mA h g−1. After 250 cycles, the cell exhibits a specific capacity of only 120.5 mA h g−1, retaining 89.4% of its initial capacity. The fluctuations in charge/discharge efficiency in the middle stage probably originate from the breakage and formation of the unstable solid electrolyte interphase (SEI) layer.49 Notably, the initial specific capacity of the LFP‖P(VDF-CTFE)@LLZTO-PAA CSE‖Li cell at 1C is 143.8 mA h g−1. Subsequently, the capacity increases to 146.5 mA h g−1. This finding can be attributed to a decrease in interface resistance with further activation of the cell and the homogenization process of the Li+ in the CSE membrane.30Fig. 6(c) exhibits the charge/discharge curves for the LFP‖SSEs‖Li cells at the 1st cycle and the 250th cycle. More importantly, the LLZTO-PAA-based cell maintains a specific capacity of 131.5 mA h g−1 and a capacity retention rate of 91.45% after 650 steady cycles. Furthermore, Fig. S13† shows that the LFP‖P(VDF-CTFE)@LLZTO-PAA CSE‖Li battery exhibits good cycling performance at 0.3C. It has an initial capacity of 159.4 mA h g−1, and after 505 cycles, the cell also has a capacity of 148.7 mA h g−1 and a charge/discharge efficiency of 99.5%. The 4.5 V electrochemical window of the P(VDF-CTFE)@LLZTO-PAA CSE also predicts its ability to match with the high voltage cathode; thus we assemble the electrolyte membrane with LCO to form an LCO‖P(VDF-CTFE) SPE‖Li cell. Fig. 6(d) demonstrates the cycle performance of the LCO‖Li cell at 1C, where the cell has an initial specific capacity of 124.1 mA h g−1, and can cycle for 160 cycles with a capacity retention of 84.9%. These findings demonstrate that the P(VDF-CTFE)@LLZTO-PAA CSE has high reversibility and good electrochemical stability due to the excellent ionic conductivity and homogeneous LLZTO-PAA fillers ensuring uniform Li+ flux.
To further investigate the changes that occur in the electrolyte membrane during charge and discharge, the LFP‖SSEs‖Li cells are tested using cyclic voltammetry (CV) from 2.5 to 4 V. As shown in Fig. 6(e), the haphazard and non-overlapping curves of the LFP‖P(VDF-CTFE)@LLZTO CSE‖Li cell show the presence of some side reactions in the cell. In contrast, the LFP‖P(VDF-CTFE)@LLZTO-PAA CSE‖Li cell shows good reproducibility in the CV curves of the first four cycles, indicating the improvement of electrochemical stability of the ceramic filler for the electrolyte membrane.52 The other difference is the voltage difference between the oxidation and reduction peaks in the first cycle, which is 0.33 V for the LFP‖P(VDF-CTFE)@LLZTO CSE‖Li cell and 0.31 V for the LFP‖P(VDF-CTFE)@LLZTO-PAA CSE‖Li cell. This finding suggests that the P(VDF-CTFE)@LLZTO-PAA CSE has better kinetic performance for Li+ transport.
The LFP‖SSEs‖Li cells are cycled 100 times before testing the lithium anode electrode using XPS to explore the chemistry on the interfacial phase layer (Fig. 6(f and g)). As for the C 1S spectrum of LFP‖P(VDF-CTFE)@LLZTO CSE‖Li, the peaks of C–C (284.77 eV), C–H (285.99 eV), N–CO (288.60 eV), and C–F (289.99 eV) are observed, which is mainly attributed to the residue and decomposition of the polymer matrix and DMF.53 N–CO is mainly derived from the amide group in the residual DMF.14 But for LFP‖P(VDF-CTFE)@LLZTO-PAA CSE‖Li (N–CO/–COO: 288.65 eV), the N–CO of residual DMF solvent and the carboxylate groups (–COO) of polyacrylic acid both produce peaks which overlap at ∼288 eV.54,55 In addition, a new peak is observed at 287.46 eV, which is attributed to the C–CO of polyacrylic acid.56 Another difference is found in the O 1S spectra for Li2O, Li2CO3/LiOH and FSI−, which mainly comes from the reaction of the polymer matrix with lithium and the LiFSI residue.31,53 Evidently, the peak area of Li2CO3/LiOH in LFP‖P(VDF-CTFE)@LLZTO-PAA CSE‖Li is reduced (Fig. S14†) due to the removal of trace amounts of water from the membrane by the absorption effect of the homogeneously dispersed LLZTO-PAA fillers, which mitigates the reaction of the impurities with lithium.33 In the O 1S spectrum of LFP‖P(VDF-CTFE)@LLZTO-PAA CSE‖Li, the small bump at 354.16 eV is related to the COO− of polyacrylic acid and its reactants.55 The areas of peaks of C 1S and O 1S spectra using P(VDF-CTFE)@LLZTO CSE are larger than those using P(VDF-CTFE)@LLZTO-PAA CSE (Fig. S14†), indicating that LLZTO-PAA is more beneficial to inhibit electrolyte decomposition.14 These results indicate the disparity in the structure and chemical distribution of the SEI of P(VDF-CTFE)@LLZTO-PAA CSE and P(VDF-CTFE)@LLZTO CSE, leading to different cell performances.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3sc04710e |
This journal is © The Royal Society of Chemistry 2023 |