Acyclic Acetals in Propylene Carbonate-Based Electrolytes for Advanced and Safer Graphite-Based Lithium Ion Batteries (2025)

Table I.Physical properties of selected organic carbonate and acetal-based solvents. Data summarized from the literature.^12,28,30

The used chemicals were purified as follows: TME (WeylChem) and TEE (WeylChem) were distilled and dried over molecular sieve 3 Å with a resulting water content below 20 ppm as determined by Karl-Fischer titration. The conducting salts, LiTFSI (3 M, battery grade), lithium bis(oxalate)borate (LiBOB) (Gelon, battery grade), LiPF₆ (BASF, battery grade), lithium bis(fluorosulfonyl)imide (LiFSI) (Lonza, 99%) were dried under reduced pressure at 1·10⁻³ mbar and elevated temperatures for two days. EMC and PC (both BASF, battery grade) were used as received. The resulting water content of the final electrolyte formulations were lower than 20 ppm. Electrode sheets based on NMC622 or graphite (Custom Cells Itzehoe) were dried at 110 °C for two days under reduced pressure at 1·10⁻³ mbar. All components were stored in a glovebox (O₂ and H₂O ≤ 0.5 ppm).

Galvanostatic, potentiostatic and -dynamic methods were performed on Metrohm Multi Autolab M204 or a BioLogic SAS VMP3 potentiostat, galvanostat. For the three-electrode and two-electrode cell set-ups,³¹ Swagelok^® cells and coin cells (CR2032) were used with Li-metal (Albemarle) as counter- and reference electrode. The separator FS 2226 (Freudenberg Performance Materials Holding) was soaked with 120 μl electrolyte. For galvanostatic cycling measurements, NMC622 electrodes (12 mm, 1.0 mAh cm⁻²) and graphite electrodes (13 mm, 1.1 mAh cm⁻²) were used. Maccor Series 4000 was used as battery cell test system. Linear sweep voltammetry measurements were performed as follow: the current was measured with an applied scan rate of 0.025 mV s⁻¹ with scan range of open current potential (E_OCP) up to 7.0 V (vs Li∣Li⁺) for anodic stability on polished platinum (Ø = 1 mm) electrode and between E_OCP and −0.025 V (vs Li∣Li⁺) for the reductive stability on copper (Ø = 12 mm) electrode in a three cell set-up, at room temperature. All Swagelok^® cells were built in a glovebox. Coin cells were assembled in a dry room (dew point: < −65 °C).

The conductivity measurements were performed on a Microcell HC Basis Package with a 1 ml TSC 1600 closed sample container and a glassy carbon electrode (rhd instruments) with Autolab PGSTAT 30 and a current booster (BSTR 10 A, from Metrohm). The measurement procedure was started at −35 °C and kept for 10 min at the given temperature. Frequencies from 1–100 kHz were applied. The cell constant was determined prior to the measurements by a 0.01 M KCl solution.

Accelerating reaction calorimetry (ARC) measurements were performed on ARC^® 254 (NETZSCH) with heat-steps of 5 °C, rest steps for 30 min and 10 min' search time. Heat-wait search procedure progresses, if after 10 min no detectable temperature changes of 0.02 °C min⁻¹ is observed. When the specific threshold of 0.02 °C min⁻¹ is reached, the mode is switched to exothermic mode where the vessel temperature follows the sample temperature to simulate adiabatic conditions. The overall temperature limit was set to 400 °C. The titanium alloy bomb with wall thickness of 0.6 mm was filled with 2 ml of liquid samples under argon. The sample in the titanium alloy bomb was heated up from bottom, side and top heater. The onset of the thermal runaway was defined as 0.2 °C min⁻¹. After the onset point was reached, the sample was expected to undergo the thermal runaway at a heating rate of 10 °C min⁻¹.^32,33

Thermogravimetric analysis (TGA) was performed on TA Instruments TGA Q5000IR. The samples were hermitically sealed inside an aluminum pan and measured with a nitrogen flow of 10 ml min⁻¹. The sample was heated up to 600 °C with 10 K min⁻¹.

The flash point was determined with Miniflash FLP from Grabner Instruments using 1 ml of the electrolyte. The electrolytes were cooled down to −20 °C. The measurements were repeated three times under dry conditions.

Differential scanning calorimetry (DSC) measurements were performed on a Q2000 by TA Instrument. The samples were hermetically sealed inside an aluminum pan. Helium flow was set to 25 ml min⁻¹ with a heating and cooling rate of 10 K min⁻¹. The temperature window used, was in the range of −150 to 40 °C.

The acyclic acetals show comparable physical properties as linear organic carbonates (see Table I). However, differences can be found in boiling point and melting point as well as in the relative permittivity of the pure solvents, that is affecting the ionic conductivity of the resultant electrolyte.

In an initial step, a process was initiated to determine the optimum salt concentration and PC:TME solvent ratio at 25 °C. Figure 1 (black curve) shows the dependency of conductivity on either the Li-salt concentration or amount of TME (blue curve). Optimal ionic conductivity was achieved for a lithium salt concentration of 0.8 M LiTFSI. The decrease in ionic conductivity for significantly higher concentrations can be explained by a lack of free ionic charge carriers (ion pair formation). At lower salt concentration a lack of ionic charge carrier (solvent excess) leads to the same effect.¹¹ The determination of an optimum TME:PC ratio is more challenging. The ionic conductivity increases with increased PC content. The differences are due to the different relative permittivity of TME and PC, which leads to a higher amount of dissociated lithium salt with increasing the content of PC. As a result, the TME content should be kept as low as possible. However, the TME content cannot be reduced indefinitely. Another reason that must be considered is the ability of the PC:TME ratio to prevent the co-intercalation of PC into graphite layers which leads to PC exfoliation. As already reported by Heß et al., the assumed SEI film forming ability of TME and TEE should be included in the consideration.²⁹ Figure 2 shows the electrochemical response of different solvent ratios of TME:PC (1:x₂ w/w, x₂ = 6, 4, 2, 1.4) in graphite∣∣Li cells. Compositions of 1:6 w/w (TME:PC) and lower lead to PC-exfoliation in the first cycle (see inset Fig. 2). Increasing the ratio to 1:4 w/w (TME:PC), a reproducible constant current cycling (CCC) in graphite-based cells can be achieved. However, the fading from 30th to 50th cycle indicates an insufficient suppression of PC co-intercalation. The composition of 1:1.4 w/w (TME:PC) shows the smallest difference between 30th and 50th cycle and the highest specific discharge capacity. Therefore, 0.8 M LiTFSI, TME:PC (1:1.4 w/w) was chosen as electrolyte for further characterization. The ratio for TEE-based electrolyte was calculated by converting the mass ratio in molar ratio and vice versa for simplification and thus results in 0.8 M LiTFSI, TEE:PC (1:1 w/w).

The temperature dependent ionic conductivities of the acyclic acetal-based electrolytes are shown in Fig. 3. The comparison with established organic carbonate-based electrolytes, such as EC:EMC (1:1 w/w), 1 M LiPF₆ (8.3 mS cm⁻¹), showing a significantly lower ionic conductivity at 20 °C for the pure acyclic acetals TME and TEE with 0.8 M LiTFSI (1.2 mS cm⁻¹). The addition of propylene carbonate as co-solvent increases the ionic conductivity to 3.2 mS cm⁻¹ for 0.8 M LiTFSI, TEE:PC (1:1 w/w). A slightly higher value (4.1 mS cm^-1) can be achieved for TME-based electrolytes due to the higher relative permittivity of TME. By substituting 0.8 M LiTFSI with LiFSI, the ionic conductivity of TEE:PC-based electrolyte increases up to 4.1 mS cm⁻¹. Another benefit of propylene carbonate as co-solvent with acyclic acetal refers to the enabling of low temperature conductivities of 0.4 mS cm⁻¹ (−35 °C) for 0.8 M LiTFSI, TEE:PC (1:1 w/w) which is close to organic carbonate-based electrolyte with 1.1 mS cm⁻¹ (−35 °C). Moreover, the TEE-based electrolyte shows an inhibition of crystallization compared to 1 M LiPF₆, EC:EMC (1:1 w/w) (T_{crystallization} = −54.9 °C) (Fig. 4) which reveals the TEE-based electrolyte as a promising candidate for low temperature applications.

The electrochemical stability of the acyclic acetal-based electrolytes on platinum for anodic stability and on copper for cathodic stability is shown in Fig. 5. For the anodic and cathodic stability limits, a value of 0.01 mA cm⁻² was set. Acyclic acetal-based electrolytes are reductively stable due to the absence of current response until the potential of Li-plating 0 V (vs Li∣Li⁺). For the anodic stability, the optimized electrolyte formulations 0.8 M LiTFSI, TME:PC (1:1.4 w/w) and 0.8 M LiTFSI, TEE:PC (1:1 w/w) are electrochemically stable up to 4.6 V (vs Li∣Li⁺), a value within the typical working potential of NMC-based LIB cells.

As prior mentioned, PC-based electrolytes on graphite electrodes, without the use of SEI-forming co-solvents or additives, lead to exfoliation of the layered structure.^22–25 An important property of a co-solvent or additive is its prior electrochemical reduction compared to PC to ensure an effective SEI formation to inhibit PC co-intercalation. The first reductive peak, which shows a hint on SEI formation, appears at ≈0.8 V (vs Li∣Li⁺) for the 0.8 M LiTFSI, TEE-based electrolyte (see inset in Fig. 6). 0.8 M LiTFSI, TME:PC (1:1.4 w/w) shows a slightly earlier reductive peak, starting at 1.0 V (vs Li∣Li⁺). Both electrolytes enable a reversible insertion-extraction of lithium ions into the graphite structure (see Fig. 6) while the use of EMC with PC has no effect (see Fig. S1 is available online at stacks.iop.org/JES/167/040509/mmedia). However, the TEE-based electrolyte, 0.8 M LiTFSI, TEE:PC (1:1 w/w) shows more clearly the intercalation steps of lithium into graphite than the TME containing electrolyte. The current can be maintained after the three cycles, whereas the TME-based electrolyte gradually decreases which agrees with the observation for single solvent-based electrolyte.²⁹ It is not clear yet, whether acyclic acetals form an effective SEI, they displace PC from the solvate shell of lithium ions thus preventing solvent-co-intercalation or a combination of both effects take place. Nevertheless, the first reduction peak in the cyclic voltammograms (CVs) can be a first hint on involvement of TME/TEE in a SEI formation mechanism. Further investigations must be performed to elucidate and understand the specific mechanism. For the further electrochemical performance in LIB cells, the focus was set on the 0.8 M LiTFSI, TEE:PC (1:1 w/w) due to the better suppression of PC co-intercalation.

The electrochemical performance of the electrolytes were investigated in a NMC622∣∣graphite cell at room temperature as shown in Fig. 7. An investigation of anodic dissolution of Al in acyclic acetals is indispensable due to the elevated voltage range of NMC622∣∣graphite of 4.3 V and the well-known absence of aluminum passivation ability of LiTFSI-based electrolytes.²⁰ The 0.8 M LiTFSI, TEE:PC (1:1 w/w) electrolyte shows anodic Al dissolution starting at potentials >3.6 V (vs Li∣Li⁺) (see Fig. S2).¹⁵ The increasing amount of current per cycle can be explained by the increasing surface area caused by anodic dissolution of Al.^34,35 In order to circumvent this disadvantage, an additive was used to ensure a passivation layer on aluminum. LiBOB was selected as additive of choice in a concentration of 2 wt% of the electrolyte.^36,37 The acyclic acetal-based electrolytes with 2 wt% LiBOB and LiTFSI as conducting salt presents a 1st cycle Coulomb efficiency (CE) of 85.2%, comparable to the benchmark electrolyte with a CE of 84.8% and CEs of ≈99.9% in the following cycles (see Fig. S3). The substitution of LiPF₆ with LiTFSI in the benchmark system and the addition of 2 wt% LiBOB leads to significantly lower 1st cycle CE of 65.7% compared to the TEE-based electrolyte, showing the promising interplay of TEE-based electrolytes with LiTFSI and 2 wt% LiBOB. By substituting the conducting salt LiTFSI with the more conducting Li-salt LiFSI,³⁸ the specific discharge capacity of the TEE-based electrolyte can be maintained, presenting values on the level of the organic carbonate-based electrolyte 1 M LiPF₆, EC:EMC (1:1 w/w) (136 mAh g⁻¹ vs 133 mAh g⁻¹). After 350 cycles, the capacity is still 88.7% of the initial capacity obtained at the 6th cycle. The capacity retention for the normalized specific discharge capacity shows an overlapping curve progression of benchmark and acetal-based electrolyte (Fig. S3), proving the promising performance of TEE-based electrolytes, particularly in combination with LiBOB as additive and their good cycling behavior on NMC622 electrodes that is comparable to the established linear organic carbonate-based electrolytes.

Asymmetric discharge C-rate performance (charge step was kept constant at 0.1 C) was investigated for the most promising electrolyte formulations. The four investigated electrolytes show a quite similar behavior up to 0.33 C (Fig. 8). From 1 C on, the acetal-based electrolyte outperforms the reference electrolyte 1 M LiPF₆, EC:EMC (1:1 w/w) + 2 wt% LiBOB. By increasing the C-rate to 2 C and higher, the trend is more significant. Furthermore, the acetal-based electrolytes compete the level of the 1 M LiPF₆, EC:EMC (1:1 w/w) until 2 C, although the ionic conductivity is ≈4 mS cm⁻¹ lower. 0.8 M LiTFSI, TEE:PC (1:1 w/w) shows a remaining discharge capacity retention of 60% at 4 C which is a promising result for a non-containing linear organic carbonate electrolyte. After 10 C, the acetal-based electrolyte containing cells can be cycled at the initial 0.1 C without significant capacity loss.

Regarding the determination of the safety properties, flash point and ARC measurements were performed. The physical properties of the pure solvents TEE (71 °C) and TME (53 °C) show an increased flash point compared to the linear organic carbonates EMC (27 °C). This trend can be observed for the optimized electrolyte formulations as well, with a flash point of 78 °C and 86 °C (see Fig. 9). In a case of a short circuit, it is also important to detect the onset temperature of the thermal runaway, which was performed by a heat-wait-search test (Fig. 10). The exothermic threshold as well as the onset of thermal runaway (self-heating rate of 0.02 °C min^-1 and 0.1 °C min⁻¹) for the acyclic acetal-based electrolytes is reached at 248 °C, 62 °C higher than 1 M LiPF₆, EC:EMC (1:1 w/w) with 186 °C. The big advancement can also be obtained for the thermal runaway temperature (self-heating rate of 10 K min⁻¹) which is achieved for the linear organic carbonate-based electrolyte at 272 °C and for the acetal-based electrolytes at 385 °C, differing more than 100 °C. Beside the excellent thermal properties, TGA measurements (Fig. S4) show the starting of mass loss at 82 °C for both acetal-based electrolytes. The benchmark starts slightly after room temperature in agreement with the determined flash point values and the volatile character of linear organic carbonates. This enhances the safety and handling properties of considered electrolyte formulations.