Increasing the Lithium Ion Mobility in Poly(Phosphazene)-Based Solid Polymer Electrolytes through Tailored Cation Doping (2024)

For achieving higher practical relevance, it is important to further increase the ionic conductivity of solvent-free MEEP-based electrolytes. At room temperature the total ionic conductivity σ_total of solid MEEP with different amounts of dissolved lithium salts can reach around 10^–4 S cm^–1, but as the lithium ion transference number t_Li+ in such systems is typically lower than 0.1, the partial lithium ion conductivity σ_Li+ is often below 10^–5 S cm^–1. ^52,53 The main reason for this unsatisfactory performance is that the lithium cations are coordinated not only by the oxygen atoms in the side chains but also by the nitrogen atoms in the backbone, which leads to pocket-like coordination structures. The strong coordination to the nitrogen atoms along the poly(phosphazene) backbone traps lithium ions and therefore greatly hinders their propagation through the polymer electrolyte. ^52–54

Thus, the trapping of lithium ions in these coordination structures needs to be prevented to increase the lithium ion mobility in MEEP-based SPEs. To gain a better understanding of the lithium ion coordination situation at the MEEP backbone, density functional theory (DFT ⁵⁵ ) calculations have been performed here. First, a lithium coordination by only ether oxygen atoms in the side chains of MEEP was compared to a coordination directly at the backbone, in which case the lithium is coordinated by a nitrogen atom from the backbone plus additional ether oxygen atoms from the surrounding side chains. This comparison revealed a distinct energetic preference of the backbone-coordination towards lithium. Starting from such a lithium coordination at the MEEP backbone, the lithium cation was replaced by various other alkali and alkaline earth ions and the energy changes for these theoretical "cation exchange reactions" were computed. It was found that all heavier alkali metal ions showed an energetically more favorable coordination to the nitrogen atoms at the MEEP backbone than the lithium cation, with the strongest interaction observed for the potassium ion. This leads to a cation displacement hypothesis, stating that it should be possible to free the lithium ions from the polymer backbone by blocking the nitrogen coordination sites with other alkali metal ions. This would release the lithium ions and allow them a higher mobility along the weakly coordinating polyether side chains, similar to the literature-known lithium ion conduction mechanism in PEO, ⁴⁰ enabling a higher partial lithium ion conductivity through the MEEP-based SPE.

To investigate the proposed cation displacement hypothesis, quantum chemical calculations have been performed with the software Turbomole 7.3 ⁵⁶ using the PBE functional and a def2-TZVP basis set. ^57–59 Borodin et al. and Bieker et al. have both compared calculations with the computationally expedient PBE functional with more accurate but computationally expensive G4MP2 calculations, and they reported good agreements for both lithium and magnesium complexes with various popular electrolyte solvents, including ethers. ^60,61 Following these findings, the PBE functional has also been used within this work. First, different coordination scenarios regarding lithium in MEEP were investigated using DFT computations. A cluster continuum approach was used, in which the lithium ion and the solvent within the first solvation shell were treated explicitly while the solvent in further distances to the central ion was included via an implicit solvent model. ^62,63 The Conductor-like Screening Model (COSMO) implemented in Turbomole was used within this work for the implicit solvent calculations with a dielectric constant ε = 20. ^60,61,64 This value was chosen following previous studies by Hall et al., who calculated dielectric constants for a range of different solvent blends typically used in LIBs, and Borodin et al., who showed that the stabilization effect of cationic clusters resulting from the implicit solvent largely saturates for ε > 20. ^60,65

Within the performed calculations, the MEEP polymer was approximated by a MEEP oligomer consisting of five monomers in order to make the calculations computationally feasible. No vibrational frequencies were calculated; therefore, all energetic terms correspond to internal potential energies. Figure 2 shows two different coordination situations for lithium in MEEP: a backbone-coordination, where the lithium is coordinated by a backbone nitrogen atom plus three oxygen atoms from surrounding side chains (pocket-like coordination), and a PEO-like side chain coordination, where the lithium is coordinated by only oxygen atoms from two side chains. Comparing the potential energies of these geometry-optimized structures yields a very distinct energetic preference of 10 kcal mol^–1 towards the backbone-coordination. This finding strongly supports the earlier-mentioned hypothesis of the lithium ions being captured and locked within the pocket-like coordination sites at the backbone, hampering ion mobility. ⁵⁴

Our approach to increase lithium ion mobility within MEEP is based upon releasing the lithium ions from those backbone coordination sites and transferring them towards the more mobile and less strongly coordinating side chains. To this end, we decided to introduce further cations other than lithium into the system. Those other ions should feature an even higher energetic preference towards the backbone coordination sites to occupy the backbone slots and force the lithium ions towards the remaining side chain coordination sites. To find suitable cations (X), structures of MEEP with backbone coordination to lithium, sodium, potassium, and cesium, as well as magnesium cations were computed. In addition, the corresponding ethylene carbonate (EC) complexes X(EC)₄ were also simulated and were used to compute virtual reaction energies ΔE of the (purely theoretical) cation exchange reactions

The choice of this reference system was motivated by the fact that MEEP can also be used as a gel polymer electrolyte matrix, i.e. including additional liquid electrolyte solvents like EC or DMC. Here, the lithium mobility is similarly hindered by the strong coordination to the MEEP backbone and a release of lithium ions into the liquid electrolyte would be desirable. With the help of this virtual reaction energy, the energetic preferences of the coordination of the different cations to MEEP can be compared. A negative value of indicates an energetic preference of the corresponding cation-MEEP coordination over the lithium-MEEP coordination. Figure 3 displays this comparison and indicates a higher energetic preference of the X-MEEP coordination for all alkali metal ions heavier than lithium (X = Na, K, Rb, Cs). While the value for the sodium ion is at a comparatively small −8.8 kcal mol^–1, all other investigated alkali metal ions feature values at ≈−15 kcal mol^–1. The highest energetic preference towards the coordination at the MEEP backbone is found for the potassium ion (−16.4 kcal mol^–1). In contrast, the magnesium-MEEP coordination displays a positive of 10.8 kcal mol^–1 indicating a lower preference than for the lithium-MEEP coordination. These results suggest the possibility of a lithium ion displacement at the nitrogen coordination sites of the polymer backbone by doping the system with additional, heavier alkali metal ions, which might then increase the lithium ion mobility by forcing the lithium ions to occupy the less strongly coordinating sidechain coordination sites in a fashion similar to the one observed in PEO. ⁴⁰

It should be noted that these calculations ignore possible entropic contributions to the coordination behavior in MEEP. Furthermore, only an approximation of the MEEP polymer in the form of a short oligomer was studied and possible anion effects were ignored by omitting the anions from the calculations. Nonetheless, the energetic preferences shown in Fig. 3a are significant, such that the results presented here can be seen as an applicable indication, keeping in mind that further calculations might be helpful in order to achieve a profound understanding of the ion coordination and mobility in MEEP. To verify the qualitative validity of the presented theoretical findings, experimental studies with MEEP-based SPE systems were conducted. Based on the promising results from the DFT calculations, KBOB was chosen as the representative salt to confirm the assumed cation doping induced lithium ion displacement mechanism.

Suppliers

Lithium bis(oxalato)borate (LiBOB, 97.3%) and 1-butyl-1-methylpyrrolidinium bis(oxalato)borate (Pyr₁₄BOB, 99.9%) were obtained from Solvionic. Poly(ethylene oxide) (PEO, M_v = 4 000 000 g mol⁻¹), sodium hydroxide (NaOH, 99.99%), potassium hydroxide (KOH, 99.99%), cesium hydroxide (CsOH, 90%), oxalic acid (C₂H₂O₄, 99%), boric acid (H₃BO₃, 99%), dimethyl sulfoxide (DMSO, 99.9%, anhydrous), and tetrahydrofuran (THF, 99.9%, anhydrous) were all purchased from Sigma-Aldrich. Benzophenone (C₁₃H₁₀O, 99%) was obtained from Acros. LiBOB, Pyr₁₄BOB, PEO, and benzophenone were dried using a HiCube 80 Eco vacuum pump (Pfeiffer Vacuum, Aßlar, Germany) connected to a Büchi Glass Oven B-585 drying oven (BÜCHI Labortechnik AG, Flawil, Switzerland). All other components were used as received.

Synthesis of poly[bis(methoxy-ethoxy-ethoxy-)phosphazene] (MEEP)

The polymer MEEP was prepared as has been described by Jankowsky et al. ^53,66

Synthesis of the alkali metal bis(oxalato)borate salts

The synthesis scheme used to prepare potassium bis(oxalato)borate (KBOB) was adapted from the patent by Wietelmann et al. concerning the synthesis of LiBOB. ⁶⁷ Oxalic acid (14.41 g, 0.16 mol) and potassium hydroxide (KOH, 4.50 g, 0.08 mol) were suspended in 120 ml of deionized water and heated to 50 °C. Separately, boric acid (H₃BO₃, 4.95 g, 0.08 mol) was dissolved in 100 ml of deionized water at 50 °C. Over the course of 15 min the H₃BO₃ solution was added dropwise into the constantly stirred KOH-containing mixture. The solution was heated to boiling and vigorously stirred for ≈ 4 h until the water was totally evaporated. This yielded a white precipitate that was pre-dried at 110 °C for 12 h. Further drying at 80 °C was successively executed at pressures of 510⁻² mbar and 10⁻⁶ mbar for 48 h each, which yielded the final product. The above procedure was also tested for the synthesis of NaBOB and CsBOB using NaOH (3.20 g, 0.08 mol) or CsOH (13.44 g, 0.08 mol) instead of the KOH.

Preparation of MEEP membranes

In order to fabricate a MEEP SPE membrane, first benzophenone (30 mg, 0.165 mmol, 23.4 mol%), pristine MEEP (200 mg, 0.706 mmol), and variable amounts of KBOB (between 2.5 mol% and 10 mol%, ratio of potassium ions to oxygen atoms in MEEP is between 1:240 and 1:60) were dissolved in THF (200 μl) and DMSO (100 μl) under stirring. LiBOB (20 mg, 0.103 mmol, 14.6 mol%, ratio of lithium ions to oxygen atoms in MEEP is 1:41) was added to the mix and the resulting viscous liquid was dried over night at 10^–1 mbar at room temperature before heating it up to 80 °C for 30 min at 10⁻² mbar. The now highly viscous mixture was placed between two siliconized poly(ethylene terephthalate) (PET) foils (bo-PET, PPI Adhesive Products Ltd, 100 μm thickness) and manually pressed between two flat metal plates to form a polymer layer with a thickness of 150 μm. For cross-linking, each side of the sample was exposed to UV-A irradiation in a UVACube 100 (Dr. Hönle AG) for 10 min. This resulted in a flexible membrane that was carefully separated from the foil layers using a 100 μm thin razor blade.

Preparation of PEO membranes

Nuclear magnetic resonance (NMR) spectroscopy

NMR measurements were performed employing an AVANCE III HD spectrometer (BRUKER, USA) and a broadband probe (PA BBO 400 MHz, BRUKER) at a frequency of 100.61 MHz (¹³C) or 128.38 MHz (¹¹B). All samples were solvated in DMSO-d₆, and all measurements were executed at a temperature of 293.3 K.

X-ray Diffraction (XRD)

A Bruker D8 ADVANCE device (BRUKER, USA) was used to execute powder X-ray diffraction (XRD) measurements. The experiments were performed in step scan mode at a step size of 2θ = 0.02° and a step time of 1 s per step with a range from 2θ = 10° to 80°. A copper target was employed as the source of the Kα radiation with a wavelength of 1.54 Å used for the measurement.

Scanning electron microscopy (SEM)

SEM images were acquired using a Carl Zeiss AURIGA Scanning Electron Microscope (Carl Zeiss Microscopy GmbH, Jena, Germany). Images were recorded using magnification factors between 100 × and 2500 × using an accelerating voltage of 3 kV and a working distance of 3.8 mm. An In-Lens detector was used to acquire the images.

Energy-dispersive X-ray spectroscopy (EDS)

EDS was executed using an Oxford Instruments XMax silicon drift detector (Oxford Instruments plc, Abingdon, UK) that is attached to the above-mentioned Carl Zeiss AURIGA SEM device. The elemental mapping was done at a fixed magnification factor of 250 × using a working distance of 5.0 mm. The applied accelerating voltage was 20 kV.

Thermogravimetric analysis (TGA)

A TA Instruments TGA Q5000 device (TA Instruments, New Castle, USA) was used for thermogravimetric analysis (TGA) of the synthesized BOB salts. The samples were heated from 50 °C to 400 °C at a heating rate of 10 °C per minute. During operation, the samples were flushed with nitrogen gas at a constant gas flow rate of 10 ml min^–1.

Transference number measurements

Ionic conductivity measurements

To investigate the ionic conductivity of the polymer samples, 12 mm wide membrane disks were placed between two stainless steel disks with a diameter of 15 mm in coin cells (CR2032). A separator ring was placed around the polymer disks to eliminate unwanted contact between the stainless steel electrodes. After resting for 24 h, the cells were connected to a Novocontrol impedance spectrometer equipped with an AN-alpha analyzer, a POT/GAL 20/11 electrochemical testing station, and a Novocontrol Quatro cryosystem temperature control unit. The samples were first heated from 20 °C to 80 °C and then cooled down again to 20 °C in steps of 10 °C. Three impedance measurements were executed at each temperature step, with resting intervals of 15 min between the individual measurements. From these measurements, the total ionic conductivity (σ_total) values of the polymer electrolyte membranes were calculated. Furthermore, the σ_Li+ of the samples was calculated by directly multiplying t_Li+ with the σ_total obtained for a temperature of 60 °C. As with the transference number measurements, four identical cells were used for each batch and the final results represent averages of the data acquired from each of the four systems.

All experiments were carried out using alkali metal BOB salts as both conducting agent and additive instead of other typical alkali metal salts due to the reportedly high interfacial compatibility of MEEP polymers with LiBOB content towards lithium metal as well as LiBOB's positive influence on the formation of an effective SEI. ^72–76 A known synthesis route for LiBOB was modified to prepare KBOB in this work, which has not been described in literature thus far. ⁶⁷ The synthesized KBOB was analyzed with ¹³C and ¹¹B NMR spectroscopy to confirm the successful synthesis; the spectra are depicted in Fig. 4. The ¹¹B NMR spectrum shows only one sharp boron signal which proves that all the boron supplied to the synthesis reaction through the H₃BO₃ was successfully integrated into the bis(oxalato)borate complexes and no further boron compounds were formed. This is further backed up by the absence of a signal at a chemical shift of δ = 19.4 ppm that would indicate remaining H₃BO₃ according to literature. ⁷⁷ Similarly, the depicted ¹³C NMR spectrum shows only one product signal at δ = 158.4 ppm, with the only other visible feature of the spectrum being the solvent signal of the used DMSO-d₆ at δ = 39.5 ppm. The observed NMR shifts are in good agreement with literature data recorded for BOB salts. ^78,79 The XRD spectrum in Fig. 4c that was obtained from the synthesized KBOB is typical for this compound. The recorded diffraction pattern matches well with literature reported data, as denoted in the graph. ⁸⁰ The absence of surplus reflexes in relation to the literature pattern indicates a high purity of the analyzed sample. Furthermore, the TGA analysis of the synthesized KBOB (Fig. S3) shows a decomposition onset at ≈330 °C which is in good agreement with values reported earlier. ⁷⁹ Judging from the aforementioned analytical results, it can be concluded that the presented synthesis route represents a reliable yet simple way to obtain KBOB. In a similar fashion, it was confirmed that NaBOB and CsBOB are also accessible via the described experimental route; TGA results for these compounds can be found in the Supporting Information. Since potassium ions were found to be most favorable for the desired displacement of lithium ions from the trapping coordination sites at the MEEP backbone via cation doping (see Fig. 3), all further experiments were focused on the usage of KBOB only.

The chemical stability of the prepared KBOB containing MEEP-based SPE systems towards lithium metal is investigated by monitoring the evolution of the impedance values of symmetric Li│MEEP-SPE│Li cells over an extended period of time. The impedances acquired for cells containing polymer membranes with 0 mol% and 2.5 mol% of KBOB are compared in Fig. 5. As can be seen, the bulk resistance of the MEEP-based SPEs is the same for both compositions, remaining unchanged over 90 h of storing the cell at 60 °C. The KBOB containing cell obviously shows the same impedance evolution pattern as the reference cell, achieving continuously lowered impedance values for 70 h of resting. After 70 h, all equilibration processes inside the membrane systems are finished and the impedance values for both cells remain stable. This confirms that the addition of KBOB to MEEP-based SPE membranes does not impair their stability against lithium metal.

The mobility of lithium ions in the polymer membranes containing various amounts of KBOB alongside LiBOB was investigated by determining the t_Li+ of the samples. According to the cation displacement hypothesis proposed herein, one can expect an increase of the lithium ion mobility when KBOB is added to the MEEP-based membranes, which should lead to a distinct increase of the t_Li+. This would verify the proposed change in the lithium coordination situation upon addition of further alkali cations to the MEEP polymer. For comparison, the same measurements were also executed with KBOB-doped PEO membrane systems. Since PEO is exclusively composed of an ether structure with carbon and oxygen atoms, the comparison of systems based on MEEP and PEO can help to verify that the observed effects are indeed caused by the typical MEEP backbone structure.

Figure 6 shows the t_Li+ values of SPE membranes based on MEEP and PEO that contained between 0 mol% and 10 mol% of KBOB. MEEP SPE samples without KBOB showed a low transference number of ≈0.08, which is in line with literature data. ⁷⁴ When 2.5 mol% of KBOB were included in the MEEP mixture the t_Li+ value drastically increased to 0.18. No other studies have reported comparably high t_Li for MEEP-based SPEs to this date, and these values are rather impressive for an all-solid-state MEEP electrolyte system without the addition of plasticizers. However, further enhancements of the KBOB content lead to a steady decline in t_Li+ down to about 0.125. This leads to the conclusion that when comparing the composition ratios investigated in this work the best prerequisites for the movement of lithium ions through the MEEP-based SPE material are present at a KBOB content of 2.5 mol%, which equals to a molar potassium-to-lithium ratio of ≈0.17. According to the cation displacement mechanism introduced earlier, it is expected that the potassium ions used for the cation doping are located in the backbone coordination pockets of MEEP. This turns many formerly trapped lithium ions into mobile ions, which can explain the rise in the t_Li+ value. Furthermore, the lithium ion mobility is likely also boosted by the effect of KBOB on the inner structure of the MEEP. Both the potassium cation and especially the BOB anion are sterically demanding, which could lead to a widening of the polymer network. This would increase the free volume and allow for an easier movement of the small lithium ions through the polymer matrix, similar to what has been reported for other polymer electrolytes. ^81,82 On the other hand, upon increasing the KBOB content in the polymer mixture, the overall share of free lithium ions among all the mobile ion species gets lower until the point where the lithium ion movement through the material could be outweighed by the mobility of the other ions. This way, at higher KBOB concentrations the contribution of the lithium ions to the overall ion mobility would be lowered, which would then be reflected in declining t_Li+ values. Also, at high salt concentrations, the ion pairing tendency in the SPE is likely increased, which should lead to a higher rate of neutral lithium salt molecules that do not contribute to t_Li+ anymore.

In comparison to the drastic reaction presented for the MEEP membranes, the t_Li+ of the PEO-based membranes shows no significant changes upon the addition of KBOB, as displayed in Fig. 6. All investigated PEO-based membrane compositions, including the one without any KBOB, feature t_Li+ values of ≈0.03 to 0.04. In relation to MEEP, the different effect of KBOB on t_Li+ of the PEO-based membranes is because the structure of PEO does not feature MEEP-like nitrogen-based backbone coordination slots with strong lithium ion trapping properties. Therefore, no lithium cations can be set free upon the addition of KBOB, as has been reasoned for MEEP. The difference between the t_Li+ curves of MEEP and PEO caused by the structural difference between the two polymers confirms the validity of the cation displacement model that was the starting point of this work.

The overall ionic conductivities σ_total of SPE membranes based on MEEP or PEO with KBOB contents between 0 mol% and 10 mol% are displayed in Fig. 7. Aside from an expected change with increasing temperature, an influence of the KBOB content on the σ_total values is discernible. Among the MEEP-based samples investigated herein, the highest σ_total values are recorded for a KBOB content of 2.5 mol%. Further additions of KBOB lead to a continuous decrease in σ_total values with the result that samples containing 10 mol% of the salt additive show a lower σ_total than the membranes with no KBOB present. Similar to what has been discussed for the transference numbers, the initial boost of the overall ionic conductivity is due to the release of free lithium ions from the trapping coordination sites at the MEEP backbone upon the addition of KBOB. The overall increase in mobile charge carriers when KBOB is added also contributes to the σ_total value. Adding to this, the aforementioned widening of the polymer structure in reaction to KBOB content allows for a generally easier ion transport, especially with regard to the low size and weight of the released lithium ions that makes for their particularly swift movement through the widened polymer network. On the other hand, the reduced KBOB dissociation rate at higher salt contents leads to surplus salt mostly residing in the polymer matrix in a neutral and undissociated state. This KBOB is not contributing to the ionic conductivity and even hampers the movement of the other ions by blocking the diffusion pathways in the polymer, which explains the reduced σ_total values at higher KBOB contents. Moreover, with high amounts of KBOB, undissociated salt molecules tend to aggregate and form needle-like crystals with a length of up to 100 μm, as can be seen in the SEM and EDS images in Figs. 8a–8d. The EDS-based visualization of potassium in the sample with 10 mol% of KBOB shows that these crystals are rich in potassium, which confirms that they represent local agglomerations of KBOB. Furthermore, the phosphorus signals in Fig. 8b are diminished at the mentioned locations. Since phosphorus is present only as part of the MEEP backbone in the examined system, its absence at the places of high potassium content implies that the KBOB crystals interrupt the local polymeric structure. Consequently, the ion movement through these areas is blocked, which further contributes to the decrease in the σ_total values. In contrast, Figs. 8e–8h show that a MEEP-based membrane with 2.5 mol% of KBOB is smooth and does not feature any KBOB agglomerations or other irregularities.

While the PEO-based SPE membranes show a stronger temperature influence on their σ_total, they seem to perform best when used without KBOB. Any addition of the salt to the polymer formulation leads to reduced σ_total values of the produced membranes. As discussed before, this is largely due to the absence of trapped lithium ions that could be set free via the stated ion displacement mechanism to facilitate an increase in ion mobility. Instead, the increased salt loading with more and more KBOB being added leads to a higher ion pairing tendency which effectively reduces the general ion mobility through the samples, as has been reasoned earlier. Also, the share of potassium ions among the mobile charge carriers gradually increases with higher KBOB contents, which leads to a decline in overall ion mobility since potassium ions are significantly heavier and thus less mobile compared to lithium ions.

By multiplying the recorded t_Li+ and σ_total values, σ_Li+ of the analyzed SPE samples is obtained. The σ_Li+ curves displayed in Fig. 9 illustrate the effective transportability of lithium ions through the polymer membranes. Without added KBOB, the samples based on MEEP show a σ_Li+ of ≈0.02 mS cm^–1, which is just above the values exhibited by the PEO-based membranes. However, while the membranes based on PEO feature an unchanged or even slightly diminished performance with different amounts of KBOB added, the σ_Li+ value of the MEEP system is increased by a factor of four upon the incorporation of 2.5 mol% of KBOB. The measured σ_Li+ of ≈0.08 mS cm^–1 is almost three times as high as literature values of 0.028 mS cm^–1 measured at the same temperature, which represent the state-of-the-art for MEEP SPEs with LiBOB as the conducting salt. ⁷⁴ On the other hand, increasing the KBOB content of the MEEP-based membranes to 5 mol%, 7.5 mol%, and 10 mol% leads to a gradual performance decline down to 0.024 mS cm^–1 at 10 mol% of KBOB. Since σ_Li+ is calculated directly from t_Li+ and σ_total, the obtained changes of the values are due to the same phenomena as already discussed before. It is concluded that the σ_Li+ of MEEP-based SPE membranes can indeed be effectively boosted by adding small amounts of KBOB, which might lead to an increased performance in a battery application scenario, which will be investigated in future research. By comparing the observed performance increase to the fundamentally different response of PEO-based membranes upon the addition of KBOB, the existence of the proposed cation displacement mechanism induced by alkali ion doping is verified.