Enabling High-Voltage Polymer-Based Solid-State Batteries Through Reinforcements with LiAlO₂ Fillers

Abstract

Poor ionic conductivity, low Li⁺ transference number, and limited electrochemical stability plague all-solid-state Li-metal batteries based on solid polymer electrolytes (SPEs). One strategy to overcome these hurdles is the insertion of ceramic fillers to generate composite polymer electrolytes (CPEs). These are based either on active (ion-conductive) fillers like Li₇La₃Zr₂O₁₂ or passive (non-conductive) fillers like Al₂O_3. In this work, the effect of passive Li-containing fillers is showcased, exemplified by a CPE platform of poly(trimethylene carbonate) (PTMC:LiTFSI) with LiAlO₂ particles. The inclusion of such fillers shows a strikingly positive effect. The ionic conductivity is greatly improved by one order of magnitude at 20 wt% of LiAlO₂ compared to the pristine PTMC SPE. Moreover, the Li⁺ transference number is significantly boosted and reaches values close to unity (T ₊ = 0.97 at 20 wt% of LiAlO₂), effectively rendering the material a single-ion conductor. The CPEs show outstanding cycling stability vs Li-metal, and electrochemical stability of up to 5 V vs Li⁺/Li. When implemented in a solid-state battery cell with LiNi_0.33Mn_0.33Co_0.33O₂ (NMC111) and Li-metal, a stable cycling performance for over 100 cycles is observed. This demonstrates the potential of using microsized and cost-effective LiAlO₂ fillers in CPEs for applications in all-solid-state Li-metal batteries.

1 Introduction

Compared to conventional liquid organic electrolytes, the use of solid electrolytes constitutes a promising strategy to develop batteries with improved safety characteristics and higher energy density.^[1] These solid electrolytes can be categorized into two main families: solid polymer electrolytes (SPEs) and inorganic solid electrolytes (ISEs).^[2] However, the direct use of either an SPE or an ISE remains challenging in large-scale battery applications,^[3] due to low ionic conductivity, high surface area resistance, moderate chemical and electrochemical stability, difficult processability, poor mechanical strength, etc.^[4] Composite polymer electrolytes (CPEs) are therefore attractive, aiming to find the best combination of properties from both components.^{[4, 5]}

CPEs have a history spanning several decades. Traditionally, passive micro-equivalent or nano-fillers such as Al₂O₃, TiO₂, ZrO_2, or SiO₂ were inserted into ion-conductive polymer matrices, on which mechanical properties were improved and the ionic conductivity was raised. The latter effect was partly due to reduced crystallinity in the polymer matrix, but partly also due to the introduction of novel ionic transport pathways at the surface of the particles. The mechanistic nature of this ionic transport is, however, not thoroughly understood,^[6-9] but proposals were made that the Lewis–Pearson acid-base nature of the particles was decisive for these phenomena.^[10] Within more recent years, the passive fillers have been replaced by ionically conductive ceramic counterparts. The aim has then been to utilize the higher conductivity in the ceramic phase as compared to the SPE phase, and increased ionic conductivity has indeed been seen in some of these systems. In other cases, the reverse effect has also been observed, and it appears that the ceramic phase in such composites is effectively non-conductive, with only the particle surfaces contributing to the increased conductivity.^{[11, 12]}

The most well-studied SPE component in CPE systems has been based on poly(ethylene oxide) (PEO), considering its dominating position among SPE hosts.^{[13, 14]} PEO is a low-T_g polymer, an excellent solvent for Li salts, and compatible with Li metal. PEO systems, however, have low Li⁺ transference numbers (T₊), explained by the strong complexation of Li⁺ by PEO. In contrast, a high T₊ is considered crucial for dendrite-free Li metal batteries, since a nonuniform Li⁺ concentration gradient gives rise to inhomogeneous Li deposition.^[15] In this context, alternative polymer hosts such as poly(trimethylene carbonate) (PTMC) give a significantly higher Li⁺ transference number (T₊ = 0.8 at 60 °C),^{[16, 17]} but unfortunately have lower ionic conductivity. Recently, we made an attempt to improve the ionic conductivity of PTMC by incorporating Li_6.7Al_0.3La₃Zr₂O₁₂ (LLZO) garnet particles.^[18] Despite being a high-cost material, LLZO has been extensively studied as both a solid electrolyte and active ceramic particle in composites, owing to its unique features of high ionic conductivity and wide electrochemical stability.^[19] While some PTMC-LLZO CPEs displayed a significant enhancement in ionic conductivity for certain particle loadings, there was little contribution from the bulk conductivity of LLZO.^{[18, 20]} This raises concern about the strategy of using active ceramic particles to enhance conductivity in polymer-rich CPEs.

Instead, the observed positive effects of incorporating active, Li-containing fillers, into polymer electrolytes seem to largely originate from the creation of specific surface layers that promote Li⁺ conductivity.^[11] Accordingly, one could imagine reaching a similar or greater effect using passive instead of active fillers (such as LLZO), if the passive fillers create equally functional interfacial conduction layers to promote ion transport. Such fillers are also potentially much simpler and less costly than active fillers. The key to achieving this is to explore ceramic particles that show the right surface properties, rather than particles that possess a high intrinsic bulk conductivity–since that bulk conductivity is anyway not utilized when inserted in a polymer matrix.

One such potentially useful yet passive filler is lithium aluminate LiAlO₂ (LAO). LAO exists in three different allotropic forms: α- (hexagonal), β- (monoclinic), and γ- (tetragonal),^[21] where the γ-phase displays better structural, thermal and mechanical stability. Similar to other ceramic fillers, it has been noted that the addition of nano-sized LAO to a PEO-based SPE matrix leads to improvements in both conductivity and cation transference number, as well as enhanced mechanical properties and interfacial stability.^[22-25] While LAO does contain Li cations in its structure, the diffusion of Li⁺ in the lattice is too low to consider LAO as an ionic conductor at moderate temperatures.^{[26, 27]}

The field of polymer electrolytes has progressed significantly since the earlier reports on LAO composites with PEO:LiClO₄ and the introduction of materials such as PTMC has opened up new possibilities of also going beyond the performance of PEO-based materials. In this work, we show how incorporating microsized LiAlO₂ fillers into a PTMC:LiTFSI polymer matrix can improve the ionic conductivity of the material by one order of magnitude, as well as giving the SPE a Li⁺ transference number of 0.97 at 60 °C–effectively rendering it a single-ion conductor. Incorporating LAO also extends the electrochemical stability up to 5 V vs Li⁺/Li, with stable cycling in a high-voltage solid-state Li//NMC battery demonstrated. Most notably, this is all achieved using a widely available and low-cost ceramic material in combination with a simple homopolymer, with clear potential for facile commercial upscaling.

2 Results and Discussion

First, the general electrolyte properties of samples using active, passive, and Li-containing passive fillers were explored. The total ionic conductivity of the PTMC-based CPEs was investigated by incorporating three different ceramic fillers: LLZO (<10 µm) as an active filler, Al₂O₃ (<2 µm), denoted AO, as a passive filler, and LAO (≈5 µm) as the main targeted filler for this study. CPEs were prepared according to the procedure described in Figure 1, and their conductivities were determined by electrochemical impedance spectroscopy (EIS), considering a PTMC-based SPE as a reference. The temperature-dependent ionic conductivity of PTMC:LAO CPEs is displayed in Figure 2a, and shows that on addition of LAO, the conductivity increases and reaches a maximum at 20 wt%, but starts to gradually decrease at higher loadings, likely due to the formation of particle agglomerates. The observed enhancement is of high significance, considering that the conductivity of 20 wt% LAO-based CPE is more than one order of magnitude higher than the filler-free PTMC-based SPE (Figure 2c). Note that PTMC is fully amorphous, so there is no reduction in polymer crystallinity due to the particle addition that can explain the rise in conductivity. As displayed in Figure S2 (Supporting Information), the SAXS data reveals an evident power-law in q⁻⁴ at small q for CPEs with LAO particles at all concentrations. This is another indication of the formation of a marked interface, without porosity, between ceramic particles and the polymer matrix.

Figure 1

Schematic illustration of the preparation of PTMC-LAO CPEs.

Figure 2

a) Ionic conductivity of PTMC:LAO CPEs for temperatures 30–100 °C (measured during increasing temperature) for different particle loadings b) Nyquist plots of CPEs with 20 wt% of different ceramic fillers and a filler-free reference at 60 °C. The inset shows a zoom-in on CPEs. c) Ionic conductivity of CPEs with 20 wt% of LAO, LLZO, and AO in the temperature range 30–100 °C. d) Ionic conductivity as a function of the ceramic filler weight concentration at 60 °C.

To further demonstrate the effect of using a low-cost passive ceramic filler, the conductivity of the resulting CPEs with LAO was compared to similar CPEs incorporating a standard active ceramic filler (LLZO) and a common passive filler (AO) within the same ceramic loading range (10 to 40 wt%). LLZO has already been used as an active filler with the same polymer matrix, but the LLZO phase did not appear to participate actively in the ion conduction.^[18] AO, in turn, is commonly used for fabricating composite electrolytes. In semicrystalline matrices, a large part of the effects of this filler on ion transport arises from a reduction in the degree of crystallinity, thereby causing an enhancement in the ionic conductivity by generating more of the amorphous–and more highly conducting–SPE phase.^[10] It is noteworthy that such an effect would not be valid for fully amorphous polymers. On the other hand, enhancing effects of AO addition on the ionic conductivity have also been noted in the amorphous phase of PEO-based electrolytes,^[28] while no such effect appears to be seen when poly(ε-caprolactone) is used as the host material.^[29]

Figure 2b displays the Nyquist plots of CPEs with 20 wt% of LLZO, AO, and LAO at 60 °C. It is seen that the bulk resistance of PTMC-based SPE is significantly decreased with the addition of all types of ceramic fillers. Nonetheless, it is clear that the effect of different ceramic fillers varies considerably. When zooming in (inset of Figure 2b), a higher bulk resistance for CPE:LLZO is observed as compared to CPEs with LAO or Al₂O₃. Therefore, incorporating active fillers instead of passive ones does not appear to be a more efficient strategy for improving the ionic conductivity of these polymer electrolytes. When comparing the conductivity development with temperature between CPEs with LLZO and Al₂O₃ at the same ceramic loading (Figure 2c), it is evident that AO performs better than LLZO.

Among all investigated CPEs within the entire range of ceramic loadings and types, it is seen that LAO is performing outstandingly better. LAO-based CPEs showed higher conductivity from 10 wt% to 40 wt% compared to LLZO- or AO-based CPEs, as shown in Figure 2d. The observed decrease in conductivity beyond certain loadings is expected, as particles start to agglomerate. This phenomenon should appear regardless of the nature of the incorporated ceramic filler, considering that it is primarily the polymer phase that conducts ions, and it becomes diluted with increasing particle loadings. Nevertheless, the LAO-based CPEs exhibit considerably higher conductivities as compared to the other CPEs also beyond 20 wt%.

To further explore the usefulness of LAO as a ceramic filler for CPEs, the Li⁺ transference number (T₊) was investigated by means of two different methods: electrochemically in Li symmetrical cells using the Bruce–Vincent method,^[30] and spectroscopically through solid-state PFG NMR on the pristine SPE and PTMC with 20 wt% LAO, annotated as CPE20LAO. In the former method, T₊ is calculated using Equation 1 after determining the impedance values (Figure 3a,b) using the equivalent circuit displayed in Figure S3 (Supporting Information). The transference number is notably higher for the PTMC:20 wt% LAO CPE as compared to the PTMC-based SPE, and reaches a value of almost unity (T₊ = 0.97 at 60 °C) as displayed in Figure 3c. By comparison, T₊ = 0.84 for the reference SPE, which is still very high for SPE materials, but consistent with previously published data on PTMC-based SPEs.^[31] A transference number of 0.97 is extremely high, considering that the SPE material contains a binary salt, and puts it on par with most single-ion conductors such as ionomers and ceramic electrolytes. As seen in Figure 3d, the LAO composite exhibits higher T₊ than the other investigated fillers.

Figure 3

Chronoamperometric response of a Li|electrolyte|Li symmetrical cells at 60 °C. Inset: Nyquist plot before and after polarization, where R_b and R_int represent the bulk and interfacial resistances respectively for a) SPE and b) CPE20LAO. c) Cationic transference number evolution vs LAO loading at 60 °C. The difference in current between samples is due to the thickness differences.

Alternatively to the Bruce–Vincent method, the apparent transport number T₊ can also be measured by PFG NMR. Such experiments were conducted at relatively high temperatures (60–80 °C) since the ionic mobility is too low at lower temperatures. The NMR signal decay was fitted to Equation 2 by a single exponential component to yield an average self-diffusion coefficient of all ⁷Li- and ¹⁹F-containing moieties, respectively, and the cationic and anionic transference numbers were determined as shown in Table 1.

Table 1. Self-diffusion coefficients of cations and anions in the investigated samples, and the respective apparent transport numbers, measured by PFG NMR. For the filler-free SPE, it was only possible to measure self-diffusion coefficients at 80 °C, due to the very short T₂ NMR relaxation times at lower temperatures.

T, K	DLi (^7Li), m^2 s^−1	DTFSI(^19F), m^2 s^−1	T+ (Li)	T− (TFSI)
CPE20LAO
353	3.35 × 10^−13	2.57 × 10^−13	0.566	0.434
343	1.48 × 10^−13	1.21 × 10^−13	0.550	0.450
333	5.92 × 10^−14	4.83 × 10^−14	0.551	0.449
Filler-free SPE
353	1.28 × 10^−13	1.06 × 10^−13	0.547	0.453

While the obtained T₊ of 0.55 at 80 °C is considerably lower compared to that obtained by the Bruce–Vincent method at 60 °C, this type of discrepancy is commonly observed due to the electrolyte ideality requirement^[32-34] as, e.g., the mobility of incompletely dissociated ion pairs or ion clusters is included in the NMR experiments.^{[32, 35]} We also note that this is commonly seen for carbonyl-coordinating polymers such as poly(ε-caprolactone), PTMC, and poly(ethylene carbonate),^{[33, 36]} whereas data from the Bruce–Vincent method for such materials appears to agree well with data from electrophoretic NMR.^[37] The trends from PFG NMR data also agree well with the trends obtained by the Bruce–Vincent method. CPE20LAO is again seen to show the highest cationic and the lowest anionic apparent transport number, and the Li diffusion coefficient of CPE20LAO is almost 3 times as high as D_Li of the corresponding SPE, which clearly shows the positive effect of the LAO particles on the Li mobility in the polymer matrix.

The ionic conductivity and the transference number are often considered the most important properties to describe ionic transport in SPEs and CPEs. Such a significant enhancement as observed here for the PTMC/LAO CPEs could potentially be explained by:

1. The creation of additional, efficient, transport pathways along the resulting polymer–ceramic interfaces, which is pivotal for the observed improved conductivity. These pathways are effectively accessible for ionic conduction owing to the absence of any insulating layer that can hinder their access, which for example is the case for LLZO. This would explain the faster ion diffusion observed in the CPEs.

2. Specific interactions between the LAO particles and the anions in the polymer matrix, leading to a reduction in anion mobility and thereby a higher Li⁺ transference number.

Lewis acid–base interactions have been reported as a significant contributor to the properties of CPEs with active or passive ceramic fillers.^[38-40] However, this effect does not always appear and depends on the surface chemistry of the employed ceramic filler. For example, we have previously demonstrated that Lewis acid–base interactions between LiTFSI and La or Zr sites are restricted when employing LLZO as a ceramic filler in PTMC-based CPEs, due to the detrimental effect of the spontaneous formation of Li₂CO₃ coatings on LLZO.^[18]

To uncover the atomic-scale mechanisms behind the improved performance of LAO as compared to AO and LLZO in the PTMC-based CPEs, the ion–ion interactions in the material were investigated. FTIR experiments were carried out on the as-prepared CPEs with 20 wt% filler and the SPE reference analog to study the lithium salt dissociation in the polymer matrix. As can be seen in Figure 4a, the FTIR spectra of all electrolytes display similar characteristic vibrations of PTMC, indicating structural stability of the polymer matrix after particle addition. In particular, the region from 2000 to 1500 cm⁻¹ displays a characteristic doublet assigned to the Li⁺-coordinating and non-coordinating carbonyl groups in the SPE, which is also apparent for the CPEs. However, the addition of ceramic particles to the polymer matrix does not seem to have a significant effect on the coordination chemistry of the polymer matrix, similar to observations previously explored for PTMC-based CPEs with LLZO particles.^[18] On the other hand, when zooming in on the region from 760 to 725 cm⁻¹, the peak assigned to S─N─S vibrations of the anion can be deconvoluted into two features: one corresponding to free cations/anions at lower wavenumbers, and the other attributed to the Li⁺:TFSI ion pairs which appear at higher wavenumbers (Figure 4b,c).^[41-43] The deconvolution enables quantifying the area for each feature, and thereby the ratio of free cations to ion pairs in the system can be determined. As illustrated in Figure 4d, the fraction of ion pairs is decreased after incorporating ceramic particles, but gets gradually higher from LAO to AO to LLZO-based CPEs, indicating a greater effect on the ion–ion interactions by LAO than with any of the other ceramic particles.

Figure 4

a) FTIR spectra of SPE and CPEs at 20 wt% ceramic loading. b, c) Deconvoluted FTIR spectra in the 760–725 cm⁻¹ region. The peaks at ≈740 and ≈745 cm⁻¹ are assigned to the expansion/contraction of “free” TFSI ions and Li–TFSI ion pairs, respectively,^[43-45] and the continuous blue and dashed black lines represent the sum of the fitted peaks and the experimental data, respectively, of (b) SPE and (c) CPE20LAO. d) Proportion of free ions and ion pairs for different ceramic fillers in the CPEs.

On a submicroscopic level, this effect could be explained by the interaction between the polymer matrix and the ceramic particles through Lewis acid–base interactions. Such interactions would enable an accumulation of charges around the surface of particles. In this study, it is most likely the negatively charged species (TFSI anions) that are attracted to the surface of the ceramic particles. Considering the effects of adding ceramic particles into SPEs, it is clear that these surfaces play a key role in determining the overall ionic transport of CPEs, and therefore that the surface chemistry is decisive for the properties of the electrolyte material. This is what appears to differentiate these three different fillers, and explains the impressive performance of LAO. It has previously been seen for LLZO-based CPEs that Li₂CO₃ on the particle surface limits such effects.^[18] With passive Al₂O₃ ceramic fillers, interactions near the AO surface have been reported to depend on the nature of the surface groups, and acidic AO can display relatively better ionic transport properties compared to basic or neutral particles.^[10] If the surface of the LAO particles comprises particularly good centers of interactions with TFSI anions, this would possibly favor increased accumulation of TFSI anions around the surface of the particles as compared to other ceramic particles. To further confirm this, zeta (ζ) potential measurements were carried out on suspensions of the particles in acetonitrile. The ζ-potential reflects the potential at the surface of particles to adsorb positively or negatively charged species.^[46] The results show a ζ-potential of 11.5 and 4.77 for LAO and AO particles, respectively (Figure S6, Supporting Information), indicating that they could both attract oppositely charged species at the surface to accumulate negative charges from the polymer electrolyte matrix (i.e. TFSI anions). Most importantly, that LAO displays the highest ζ-potential value clearly indicates the strongest surface affinity of these particles to the anions, which is consistent with the high cationic transference number observed for CPEs containing this material. When LiTFSI was added to the suspension, the ζ-potential strongly deviated from a positive to a negative value of −23.5 mV for LAO particles, and −0.389 and −0.525 mV for AO and LLZO particles, respectively (Figure S7, Supporting Information), further confirming the accumulation of anions around the surface of the ceramic particles, with a stronger affinity seen for the LAO particles in particular.

Not only the cationic but also the anionic, self-diffusion coefficient increases after incorporating the LAO particles as seen from the data in Table 1. This corroborates the significant increase in ionic conductivity for CPE20LAO compared to the filler-free SPE. In addition, the IR and zeta potential data indicate that there is a strong interaction between the salt species at the surface of LAO particles. However, this does not indicate that the LAO particles solely act as anion immobilizers, but also have the effect of raising the overall mobility in the system.

Apart from the ion transport properties, a wide electrochemical stability window (ESW) of a battery electrolyte is also vital to enable cycling with high-voltage cathodes. Previously, LLZO-based CPEs have shown that the addition of active ceramic particles extends the ESW as compared to the SPE analog, which can be explained by the wide ESW of LLZO itself.^[47-49] Here, when the ESW is investigated using LAO as filler, it is clear that it is extended as well as compared to the SPE counterpart. As displayed in Figure 5a, the apparent ESW obtained from the LSV measurement is extended up to 5 V vs Li⁺/Li, while the SPE starts to degrade at ≈4.5 V vs Li⁺/Li.

Figure 5

a) LSV measurements at 0.1 mV s⁻¹. b) Voltage vs time profile for CPE20LAO in an LFP half-cell, using the CICC method and increasing the upper cutoff voltage every 5 cycles. All experiments were carried out at 55 °C.

LSV does not take the influence of the true working electrode of the battery into account and does therefore not realistically reflect the practical electrolyte stability in a battery cell. For that reason, the ESW of the CPEs was also tested by means of cutoff increase cell cycling (CICC).^[50] This method consists of monitoring the voltage and capacity when cycling galvanostatically and progressively increasing the upper cutoff voltage. Figure 5b illustrates the resulting voltage profile of CPE20LAO in a cell composed of an LFP cathode and a Li metal anode cycled at 0.05C from 2.7 up to a maximum of 5.0 V vs Li⁺/Li. The voltage profile provides relevant information on the realistic onset voltage at which polymer decomposition/degradation can be expected to begin. For CPE20LAO, no sudden cell failure is observed, which confirms the stabilizing effect of LAO addition to the PTMC-based electrolyte. A similar stable voltage profile of CPE20LAO was obtained when also cycling in a cell composed of NMC cathode and Li metal anode to above 4.0 V, indicating a fair electrochemical stability window also with this cathode (Figure S8, Supporting Information). While it is not evidenced how these ceramic particles enhance the electrochemical stability window, it could be speculated that the incorporation of LAO particles has a stabilizing effect on the LiTFSI salt, rendering less free TFSI anions. Their coordination to the LAO particles can render the anions less prone to undergo oxidative decomposition at the electrode. Similar effects have been seen in, for example, highly concentrated electrolytes, where a stronger coordination renders the electrolyte species less prone to undergo decomposition reactions at the electrode surfaces.^[51]

The wider ESW of CPE20LAO compared to the PTMC-based SPE enables exploring the implementation of this composite electrolyte in solid-state batteries with high-voltage cathodes, operating at higher potentials than LFP. To this end, a solid-state battery with CPE20LAO sandwiched between an NMC cathode and a Li metal anode was fabricated and cycled at 0.05C at 55 °C. The cycling performance of the cell is shown in Figures 6a and S9 (Supporting Information), and reveals a steady capacity over 100 cycles, with an initial discharge capacity of 127 mAh g⁻¹ and a capacity retention of 92% after 100 cycles. The rate performance was also evaluated with an LFP cathode and Li metal anode as shown in Figure 6b, and it can be seen that the electrolyte manages to perform also at rates as high as 5C at 55 °C. Compared to previously reported polycarbonate-based SPEs, the composite with LAO particles reported here shows considerably better cycling performance also with a more demanding high-voltage cathode.^{[17, 49, 52]} Compared to standard PEO-based composite electrolytes, where cycling is primarily possible only with LFP and the addition of plasticizers is needed to enable cycling with high-voltage cathodes,^[53-56] it is noteworthy that the LAO-doped PTMC is performing better in full solid-state batteries employing NMC and Li metal.

Figure 6

a) Specific capacity and coulombic efficiency evolution profiles of CPE20LAO with NMC cathode and Li metal anode battery cell, tested at 0.05C rate. b) Rate capability performance of CPE20LAO with LFP cathode and Li metal anode. All experiments were carried out at 55 °C.

The area-specific resistances of the SPE and CPEs were determined in Li symmetrical cells by electrochemical impedance spectroscopy at 55 °C; see Figures 7a and S11 (Supporting Information). It is seen that the interfacial resistance can be decreased by 2 orders of magnitude after the addition of LAO particles as compared to the SPE and is the lowest among all CPEs with different ceramic fillers. This is in good agreement with previous reports that adding ceramic particles in CPEs reduces interfacial resistivity.^{[29, 57, 58]} The reduced interfacial impedance indicates that incorporating LAO particles can contribute to improving the stability against Li metal and inhibit continuous electrolyte decomposition at the Li metal surface.

Figure 7

a) Comparison of area-specific resistance (ASR) between the filler-free SPE, and CPEs with LAO, AO, and LLZO in a Li|Electrolyte|Li at 55 °C. b) Lithium stripping/plating cycling test of Li|CPE20LAO|Li with increasing current density from 0.1 to 1 mA cm⁻² at 55 °C.

The cycling stability of CPE20LAO vs Li was investigated in lithium stripping/plating experiments carried out on Li symmetrical cells at 55 °C from 0.1 up to 1 mA cm⁻²; the results are shown in Figure 7b. It is clear that no evidence of short circuits is seen even at the highest current density and that the voltage is kept stable throughout the entire cycling experiment. Upon increasing the current density, a slight increase in the overpotential is observed, as expected, and can be explained by the formation of an interphase and ohmic losses in the electrolyte.^[59] Nevertheless, the overpotential does not exceed 0.015 V at the highest current density even after over 1000 h of cycling. The overpotential observed in Figure 7b can be considered low, given the relatively high resistance of the CPEs, but is likely due to the low interfacial resistances of these samples. Compared to the filler-free SPE (Figure S12, Supporting Information), only cycling at the lowest current density (0.1 mA cm⁻²) was possible while an immediate cell failure was observed afterward. For the SPE, a higher polarization voltage was also noticed which kept increasing with cycling, indicating an inhomogeneous Li stripping/plating process. This shows that the incorporation of LAO particles into the PTMC matrix inhibits polymer decomposition/degradation at higher current densities while facilitating a useful Li stripping/plating process, likely associated with the improved mechanical stability, higher conductivity, and higher T₊ of the CPE.

While LLZO indeed is a great ceramic electrolyte material owing to its high ionic conductivity and Li transference number, it is less clear if it constitutes a useful component in ceramic-in-polymer CPEs. LAO, on the other hand, is a non-ionically-conducting material, but structurally stable and with few challenges related to interfacial resistivity. Figure 8 qualitatively summarizes some key properties of the CPEs investigated in this study. The incorporation of LAO particles is beneficial for the overall performance of CPEs. LLZO, despite being considered an active ceramic filler, contributes only passively to the overall ion conductivity of CPEs, leaving only conduction along the polymer-ceramic interfaces as a possible effect that alters the conductivity. While AO-based CPEs can surely be further explored, we show that the Li-containing passive ceramic filler renders more promising results than most reported passive fillers (i.e. TiO₂, Al₂O₃, ZrO₂, etc.).^[12]

Figure 8

Radar plot of CPEs with active LLZO and passive LAO and AO ceramic fillers.

It is possible that the Lewis–Pearson acid–base relationship between the ceramic filler and the salt can partly explain the observed phenomena. As seen from the IR data presented, there is an ion separation induced by the introduction of the particles, and the zeta-potential measurements indicate a resulting accumulation of ions on the particle surfaces. Such an accumulation has also been corroborated by recent photoelectron spectroscopy studies of CPEs.^[60] It could be envisioned that anionic accumulation at the ceramic particle surface frees up Li⁺ in the SPE matrix, causing a higher ionic conductivity and transference number while also promoting increased anionic mobility through specific mechanisms in the highly concentrated interphasial region, as seen in the PFG NMR data. As such anion–particle interactions are likely heavily dependent on the surface chemistry, they will be influenced not only by the overall chemical composition of the material but also by the specific surface structure, which should be inherently different for the involved species considering the large differences in crystal structures (see Figure S4, Supporting Information). The differences observed for the different types of fillers would thereby perhaps be best explained by the differences in surface structure between the materials. Nevertheless, the atomic-scale mechanism for this phenomenon needs further studies, and computational efforts can hopefully support such endeavors. It should then, however, be kept in mind that these CPE systems are inherently complex, and great care needs to be taken when applying these methods and constructing models of such systems. Previous computational work has highlighted these challenges.^[61]

3 Conclusion

In this work, we have shown that the incorporation of low-cost, passive, microsized ceramic filler into a polymer electrolyte can generate improved ionic conduction and electrochemical performance, to the point where it approaches a single-ion conductor. By comparison, this is an outstandingly high transference number for an SPE with a binary salt. This constitutes a novel category of CPEs and has here been exemplified by LiAlO₂ in PTMC electrolyte. The comparisons with other fillers, both active and passive, highlight that the surface chemistry and its interaction with the electrolyte components is decisive for the performance. Quantitative determination of the ionic conductivity, the transference number, and the ion–ion interaction reveals that the addition of LiAlO₂ can significantly improve all these three properties through additional transport pathways along the polymer–ceramic interfaces and efficient Lewis acid–base interactions at the particle surface. The fabricated CPE also shows useful stability with Li metal, allowing an extended use without dendrite formation, and an ESW rendering it compatible with NMC111 electrodes. This work thus reveals that considerable improvements can be made for finding novel functional CPEs, without the use of complex and unsustainable ceramic materials, or by resorting to nanoparticles as fillers.

4 Experimental Section

Materials

Poly(trimethylene carbonate) (PTMC) was synthesized through ring-opening polymerization as described previously.^[18] Li₂CO₃ (99.99%, Sigma–Aldrich) and Al₂O₃ (99.99%, Alfa Aesar) were used as received for the synthesis of lithium aluminate (LiAlO₂). Lithium bis(trifluoromethylsulfonyl)imide (LiTFSI, BASF), Al-doped lithium lanthanum zirconate garnet (Li_6.25Al_0.25La₃Zr₂O₁₂; LLZO, MSE supplies), and aluminum oxide (AO, Sigma–Aldrich) powders were dried at 120 °C for 48 h under vacuum before use. Acetonitrile (anhydrous, 99.8%, Sigma–Aldrich) was used as received. LiFePO₄ (LFP, Phostech Lithium), LiNi_1/3Mn_1/3Co_1/3O₂ (NMC-111, CustomCells), carbon black (Imerys, C-ENERGY, super C65), lithium foil (125 µm, Cyprus Foote Mineral Co.) were used as received for electrode fabrication and battery cell assembly.

LiAlO₂ Synthesis and Characterization

γ-LiAlO₂ (abbreviated LAO) was synthesized following a solid-state route.^[62] First, stochiometric amounts of Li₂CO₃ and Al₂O₃ were mixed and ball-milled in a zirconia jar for 6 h at a speed of 450 rpm using a planetary ball-mill instrument (Retsch PM100). The obtained powder mixture was heat-treated at 900 °C for 2 h with a heating and cooling rate of 5 °C min⁻¹. The LAO powder was thereafter transferred to an Ar-filled glovebox and used for composite electrolyte fabrication without further processing. The crystal structure of neat LAO and in the CPEs was investigated by X-ray diffraction (XRD) on a Bruker D8 Advance diffractometer using Cu Kα radiation (λ = 1.54178 Å). XRD patterns were acquired over a 2θ range of 10°–70°. The morphology, particle size distribution, and elemental composition of as-synthesized LAO powder were evaluated by scanning electron microscopy on a Zeiss SEM coupled with an energy-dispersive X-ray spectrometer (EDS).

CPE and SPE Electrolyte Film Preparations

Solid polymer electrolyte (SPE) and composite electrolyte (CPE) films were manufactured through a controlled solvent evaporation solution casting method as previously described.^[18] For the CPEs, a solution of PTMC + 30 wt% LiTFSI was first prepared, to which 10 wt%, 20 wt%, 30 wt%, and 40 wt% of ceramic particles (LLZO, AO, LAO) were added. The polymer-salt-ceramic-solvent mixture was ball-milled at 25 Hz for 15 min and transferred to Teflon molds. After a vacuum drying cycle of 60 h at up to 60 °C,^[18] self-standing films of 16 mm in diameter were obtained with thicknesses between 50 and 150 µm as shown in Figure S1 (Supporting Information).

Cell Assembly

Positive electrodes were prepared by mixing 75 wt % of active material (LiFePO₄ “LFP” or LiNi_0.33Mn_0.33Co_0.33O₂ “NMC”), 10 wt% carbon black, and 15 wt% of PTMC as binder in acetonitrile. PTMC was used as a binder to achieve good compatibility with the electrolyte material.^[63] The mixture was ball-milled at 30 Hz for 30 min and the obtained slurry was coated on a carbon-coated aluminum foil with a 150 µm film applicator (ZFR 2040 4-sided Applicator, Zehntner GmbH Testing Instruments). The slurry coating was dried in a vacuum oven at 80 °C for 12 h before cutting electrodes of 13 mm diameter. The electrodes were thereafter dried in a Buchi oven at 120 °C for 12 h inside an Ar-filled glovebox before use. Electrochemical characterization was performed using a pouch cell configuration with lithium foil as the negative electrode.

Ionic Conductivity

The total ionic conductivity was measured on a Schlumberger SI 1260 impedance/Gain-phase analyzer with the SPE and CPE materials sandwiched between two stainless-steel electrodes in a CR2025 coin cell configuration. The measurements were carried out from 7 MHz to 100 mHz at an AC amplitude of 10 mV while increasing the temperature up to 90 °C with 10 °C intervals. The assembled cells were annealed at 90 °C for 1 h one day before the measurement to improve the interfacial contacts.

Fourier–Transform Infrared (FT–IR) Spectroscopy

FT–IR measurements were carried out using a PerkinElmer Spectrum One FT–IR spectrometer equipped with a ZnSe crystal attenuated total reflectance (ATR) setup at room temperature. FT-IR spectra were recorded from 4000 to 650 cm⁻¹ with a 4 cm⁻¹ resolution measuring 32 scans. Peak fitting and deconvolution were done using a Voigt function after a linear baseline correction in the Origin software.

Zeta Potential

Zeta potential measurements were performed in a Malvern Zetasizer Nano ZS equipment using disposable folded capillary Zeta Cells. Ceramic particles were prepared as a suspension in acetonitrile (4 mg mL⁻¹), and as a suspension together with LiTFSI before being further diluted 1:5 in acetonitrile, and sonicated for better dispersion and elimination of agglomerates.

Transference Number

Cationic transference numbers were investigated electrochemically employing the Bruce–Vincent method on a BioLogic SP-240 potentiostat at 60 °C.^[30] A potentiostatic polarization of 10 mV was applied before and after measuring the cell impedance from 7 MHz to 100 mHz. Symmetrical pouch cells for the measurements were assembled using a SPE/CPE of 15 mm in diameter sandwiched between two 12 mm lithium disks. The cells were kept at 60 °C overnight before measurements. The lithium transference number T₊ was determined from Equation 1:

$$$$\begin{equation}{{T}_ + } = \frac{{{{I}_{{\mathrm{SS}}}}\left( {\Delta V - \ {{I}_0}{{R}_0}} \right)}}{{{{I}_0}\left( {\Delta V - \ {{I}_{{\mathrm{SS}}}}{{R}_{{\mathrm{SS}}}}} \right)}}\end{equation}$$$$(1)

Li⁺ transference numbers were also evaluated through PFG NMR experiments. All samples for NMR measurements were prepared in an argon-filled glovebox, where the films were cut into small pieces and placed into pre-dried NMR tubes. Further, to prevent the permeation of air and moisture into the NMR tubes, they were hermetically flame-sealed. ⁷Li and ¹⁹F PFG NMR experiments were performed on a Bruker Avance III 600 spectrometer (14.1 T, ν(⁷Li) = 233.23 MHz and ν(¹⁹F) = 564.69 MHz) using a DiffBB diffusion probe-head and 40 A gradient amplifiers to investigate the diffusion of cations and anions inside the polymer matrix. A stimulated echo (STE) pulse sequence was used to measure the respective self-diffusion coefficients in a range of temperatures between 333 and 353 K.^[64] The attenuation of the peak intensity is described by Equation 2:^[65]

$$$$\begin{equation}I = {{I}_0}\exp \left( { - D{{\gamma }^2}{{g}^2}{{\delta }^2}\left( {\Delta - \ \frac{\delta }{3}} \right)} \right)\end{equation}$$$$(2)

where I is the observed integral intensity, I₀ is the reference integral intensity (unattenuated signal intensity), D is the self-diffusion coefficient, γ is the gyromagnetic ratio of the observed nucleus, g is the gradient strength, δ is the length of the gradient, and Δ is the diffusion time. Experiments were conducted with a diffusion time Δ of 500–700 ms and a gradient duration δ of 5 ms for the ⁷Li measurements. For the ¹⁹F measurements, a Δ of 500 ms and δ of 1.5–3 ms were used. The gradient strength was varied in 8 steps (with a maximum gradient strength of 11.38 T m⁻¹) throughout the experiments, providing more than 80% signal attenuation. The self-diffusion coefficients were obtained by (lest-squares) fitting of Equation 2 using the TopSpin 3.2 pl5 (Dynamic center) software.

Electrochemical Characterization

Electrochemical stability was investigated employing linear sweep voltammetry (LSV) using CR2025 coin cells, where the CPE/SPE was sandwiched between a stainless steel electrode as the inert working electrode and a lithium disk of 12 mm diameter as the counter/reference electrode. The measurements were performed at 55 °C, and the voltage was swept from 3.0 to 6.0 V vs Li⁺/Li with a rate of 0.1 mV s⁻¹. To investigate the effect of the active material on the estimated electrochemical stability, a cutoff increase cell cycling (CICC) technique was employed, by galvanostatic cycling of cells at a C-rate of C/20 at 55 °C while increasing the upper cutoff voltage of 0.1 V every 5 cycles until 5.0 V vs Li⁺/Li was reached.^[50] Pouch cells for the measurements were assembled with CPE/SPE sandwiched between a cathode of 13 mm diameter and a lithium disk of 15 mm diameter. The cells were rested at 55 °C for 24 h before measurements and galvanostatic cycling tests were carried out with an ARBIN BT-2043 on lab-scale pouch cells. Cycling was performed at C-rates based on a theoretical capacity of 170 and 160 mAh g⁻¹ for LFP and NMC, respectively.

To evaluate the cycling stability of the composite electrolyte against lithium, stripping/plating experiments were carried out at 55 °C with increasing the current density from 0.1 up to 1 mA cm⁻² using a BioLogic MPG instrument. Symmetrical pouch cells were assembled with lithium metal discs of 12 mm diameter and polymer/composite electrolytes of 15 mm diameter. The cells were rested at 55 °C overnight before the measurements.

The practical application of the fabricated CPEs in solid-state batteries was tested through galvanostatic charge–discharge cycling with a high-voltage cathode (NMC111). The battery cell cycling was performed at 55 °C for a C-rate of C/20, and a voltage window of 2.7–4.2 V on an Arbin instrument, employing a pouch cell format with a 16 mm electrolyte film, sandwiched between a cathode of 13 mm diameter and a lithium disk of 15 mm diameter.

The area-specific resistance (ASR) between the metallic lithium electrode and the composite electrolyte was evaluated by electrochemical impedance spectroscopy at 55 °C in the frequency range 7 MHz–100 mHz. SPEs/CPEs of 15 mm diameter were sandwiched between two lithium disks of 12 mm diameter in a pouch cell configuration. The cells were rested for 12 h before the impedance measurements.

SAXS Characterization

Small Angle X-ray Scattering (SAXS) data were collected on the Swing beamline at SOLEIL (Saclay, France). The X-ray wavelength used was 1.033 Å, with a sample-to-detector distance of 6.564 m. Here, the scattering curves were plotted as a function of the scattering vector q. Standard data corrections were applied.

Acknowledgements

This work was supported by the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 860403 (POLYSTORAGE), the European Research Council (ERC) under the European Horizon 2020 research and innovation programme (Grant agreement No. 772777 FUN POLYSTORE), the Knut and Alice Wallenberg Foundation (INTELiSTORE 139501042), the Swedish Research Council (project number 2024-05180) and STandUP for Energy. SOLEIL was acknowledged for the provision of synchrotron radiation facilities and we would like to thank T. Bizien and J Perez for assistance in using beamline SWING.

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

K.E. did conceptualization, major writing, experiments, review, and revision. L.R. did SAXS experiments, review, and revision. A.M. and V.S. performed NMR experiments. K.E. reviewed and revised the manuscript. J.M. and D.B. did conceptualization, review, and major revision.

Open Research

Data Availability Statement

The data that support the findings of this study are available from the corresponding author on reasonable request.