Introduction

Herein, we investigate solvent selection focusing on three solvent physical properties, namely, dielectric constant, molecule structure, and boiling point. We focus on the effects different solvents have on increasing the ionic conductivity of the inorganic Li7P3S11 electrolyte to contribute to the reactivity of 70Li2S–30P2S5 during liquid-phase synthesis. Even now, the rapid growth of lithium-ion batteries has raised safety concerns regarding commercial lithium-ion batteries with liquid electrolytes. In particular, the use of flammable organic liquid electrolytes can cause leakage from and ignition of the electrolytes1,2. Replacing the organic liquid electrolyte with a solid electrolyte will ensure excellent battery safety. All-solid-state lithium batteries with inorganic solid electrolytes that contain crystalline Li7P3S11 are one of the most promising energy storage technologies due to their high safety and high energy density, especially for electric vehicles3. Among solid electrolytes, sulfide-based lithium-ion solid electrolytes exhibit high ionic conductivity compared with liquid electrolytes. In contrast to oxide solid electrolytes, they also have a mechanically soft nature, which benefits their physical interaction with active materials4. These materials generally contain Li10GeP2S12 (σ25 = 12 mS cm–1)5, Li7P3S11 (σ25 = 17 mS cm–1)6, and Li6PS5Cl argyrodites7,8.

Sulfide materials are prepared under an inert atmosphere because of their moisture-sensitive. Manufacturing of sulfide-based all-solid-state batteries is subjected to the severe limitation that poses major technical challenges9. Commercialization of all-solid-state batteries required a promising process technology toward scalable manufacturing of solid electrolytes and the mixing of electrode composites containing the solid electrolytes, cathode active materials, and conductive additives10,11. For the most part, the synthesis methods for sulfide solid electrolytes can be divided into three categories: (i) solid reaction methods, (ii) mechanochemical reaction methods, and (iii) wet chemical methods. Wet chemical methods are the most effective for obtaining nanosized solid electrolytes, making them advantageous for homogeneous compounding with the cathodic electrode. In addition, wet chemical methods are suitable for large-scale synthesis at low reaction temperatures12. Liu et al. have reported that a wet chemical method based on tetrahydrofuran (THF) solvent is effective for the preparation of β-Li3PS4, a typical sulfide electrolyte13. In addition, Li2S–P2S5 solid electrolytes have been synthesized in organic solvents, such as propionate (EP), butyl acetate (BA)14, and acetonitrile (ACN). Recently, Yamamoto et al. demonstrated the effect of solvent on the ionic conductivity of Li3PS4 synthesized using a liquid-phase process14. The ionic conductivity of Li3PS4 prepared from a liquid phase was found to have a strong correlation with the polarity, \({\delta }_{p}\), of the solvent, reaching 5.09 × 10–4 S cm–1 at room temperature when using BA with a low polarity. This value is consistent with the conductivity of a sample obtained via ball milling. In 2012, Li7P3S11 was synthesized as a high lithium-ion conductor in dimethyl ether (DME) solvent with a maximum ionic conductivity of 2.7 × 10–4 S cm–1 at room temperature15. Following this, Li7P3S11 was prepared using ACN, THF, and anisole solvents; the highest ionic conductivity was achieved using ACN solvent16,17. A wet chemical reaction mechanism for the formation of Li7P3S11 phase in ACN has also been proposed. It was found that two phases, soluble Li2S–P2S5 and solvated Li3PS4, could be converted to Li7P3S11 in the desolvation process during heat treatment18,19. Several research groups have also focused their efforts on the synthesis of Li7P3S11 based on ACN solvent20,21,22,23,24. Xu et al. observed differences between Li7P3S11 synthesized from THF and ACN solvents and suggested that the steric hindrance (ring or short-chain structure) of these solvent molecules results in the presence of residue molecules in the solid electrolyte16. However, due to the absence of systematic studies, no leading theory on the effects of solvent on the synthesis of highly conductive crystalline Li7P3S11 phase has been developed.

In this work, we investigated the effect of solvent on reactivity in the formation for Li7P3S11 solid electrolytes. In particular, we employed 1,4-dioxane (Dox), carbon disulfide (CS), tetrahydropyran (THP), nitromethane (NM), ACN, furfural (FF), succinonitrile (SN), and ethylene carbonate (EC) reaction medium. Using organic solvents with a high dielectric constant led to high reactivity in the Li2S–P2S5 system, forming high-purity crystalline Li7P3S11 from liquid-phase synthesis. However, all solvents with a high dielectric constant except for ACN underwent a side reaction with the Li2S–P2S5 system. Furthermore, they are not suitable for solvent removal via a drying process due to their high boiling point. Here, we demonstrate that ACN solvent is the most suitable solvent for the formation of Li7P3S11 solid electrolyte due to its high dielectric constant, linear structure, and low boiling point.

Results and discussion

Chemical, structural, and microstructural properties

Figure 1 shows the suspensions for Li7P3S11 with each organic solvent after stirring at 50 °C for 3 days. The color of the suspension varied depending on the solvent. The suspensions with ACN and THP were milky white and off-white, respectively. Chemical reactivity in organic solvents generally depends on the bond polarity between solvent molecular and ionic species25. In particular, the dielectric constant and donor number (DN) of the organic solvent play a crucial role in the formation of a solvating complex. For instance, ACN molecules have stronger interactions with Li ions due to their high dielectric constant, as outlined in Table 1. In addition, the coordination of ACN with cations (in this case Li ions) is less affected by steric hindrance because of the linear structure26. The strong interaction between Li ions and ACN indicates their shorter bond length relative to those between Li ions and other solvents, resulting in the white color of the suspension, which is the typical color of the Li3PS4 complex. THP solvent also demonstrated a lower dielectric constant and steric hindrance induced by its cyclic structure, leading to a weaker interaction with Li ions. Using Dox, CS2, and NM as the reaction media resulted in a yellowish-white suspension, indicating poor reactivity to Li3PS4. The mixture of starting materials including FF solvent immediately became discolored (black) after the addition of the solvent, indicating the occurrence of a side reaction between the FF solvent molecules and the the Li2S–P2S5 system. In addition, the suspension with SN solvent gradually became black during stirring, probably resulting in the relatively strong cross-interaction between Li ions and the two CN groups in SN. Figure 2a,b show the X-ray diffraction (XRD) patterns of the precursors and samples, respectively, following heat treatment. The crystal structures of the precursors varied depending on the solvent due to the differences in polarity and structure of the solvent molecules. The XRD pattern of the precursor containing ACN solvent is consistent with that of Li3PS4\(\cdot\)ACN, as reported by Calpa et al27. At the same time, the Li7P3S11 precursor involves two phases of the Li2S\(\cdot\)P2S5 and Li3PS4 complexes18. According to a recent study, following a drying process at 100 °C, Li2S\(\cdot\)P2S5 with ACN exhibits a halo-shaped XRD pattern originating from its amorphous structure28. Therefore, the Li7P3S11 precursor with ACN solvent contains crystalline Li3PS4\(\cdot\)ACN and amorphous Li2S\(\cdot\)P2S5. Residue Li2S was observed in the precursors with ACN, THP, CS, and Dox solvents. The intensity of the peak corresponding to residual Li2S in the 70Li2S–30P2S5 solid electrolyte decreased in relation to increases in the dielectric constant of the solvent. Interestingly, the precursor with EC solvent retained a large amount of unreacted Li2S without being affected by the high polarity of the EC solvent. It has also been reported that Li-sulfur batteries using EC-based liquid electrolytes suppress the dissolution of lithium polysulfides generated during the electrochemical cycling process29. This indicates that EC solvent has low solubility for ionic species containing sulfur, which is consistent with the low reactivity of the EC solvents with Li2S and P2S5 observed in our study.

Figure 1
figure 1

Photographs of 70Li2S–30P2S5 mixtures after stirring for 3 days in ACN, THP, Dox, CS, NM, FF, AN, and EC solvents.

Full size image
Table 1 Physical and chemical properties of ACN, THP, Dox, CS, NM, FF, AN, and EC solvents.
Full size table
Figure 2
figure 2

XRD patterns of (a) precursors after drying under vacuum and (b) 70Li2S–30P2S5 solid electrolytes synthesized using each solvent.

Full size image

Li7P3S11 crystalline phase was successfully synthesized in the samples with ACN, THP, and Dox solvent following heat treatment. The intensity of the first peak originated from Li7P3S11 crystal phase at 29.7 degrees increased in the order of Dox, THF, and then ACN solvents, which showed value of 232, 316, and 386, respectively. This observation suggests that higher dielectric constant of the solvent results in higher purity of crystalline Li7P3S11. In the case of the samples containing THP and Dox, Li3PS4 and Li2P2S6 crystalline phases remained following heat treatment. This is consistent with the trend observed in Li7P3S11 synthesized from the liquid phase in ACN and THF solvents demonstrated in a previous study16. The sample with NM solvent involved an unknown phase, while the samples containing FF and SN exhibited amorphous structure before and after heat treatment at 270 °C. Considering the color of the precursor suspension and the XRD results, it is likely that the FF and SN solvent molecules caused a side reaction with the initial materials. The aldehyde was also activated by the action of Lewis acid as a catalyst. FF solvent is an aromatic compound with an aldehyde group, which indicates that the side reaction was caused by the coordinate bond between Li ions and negatively charged oxygen molecules within the carboxyl group. In addition, Li ions attacked the oxygen molecules of the cyclic compound within the FF solvent molecules, which is associated with ring opening and a chain of side reactions.

Figure 3 shows field emission scanning electron microscope (FE-SEM) images of 70Li2S–30P2S5 solid electrolytes prepared via each solvent. In all samples, the secondary particles ranged in diameter from 50 to 100 μm. The microstructures of the samples containing Li7P3S11 crystal phase created via ACN, THP, and Dox exhibited intriguing textures, including one resembling a broccoli crown. The microstructure of the sample prepared via ACN solvent involved uniform pseudo-spherical primary particles of less than 1 μm, and each particle was aggregated and connected with the others. It is clear that the homogeneous Li7P3S11 was produced through a solid reaction. In the case of THP and Dox, the samples formed primary particles with nonuniform shapes, exhibiting rougher particles than the sample created via ACN solvent. In the case of NM, SN, and EC, the microstructures exhibited a very rough particle surface. Monolithic particles of greater than 50 µm were observed in the samples synthesized via CS and FF, which means that the residue included coordinated complexes and compounds generated from a side reaction, as shown in the thermogravimetry/differential thermal analysis (TG–DTA) results.

Figure 3
figure 3

SEM images of 70Li2S–30P2S5 synthesized via ACN, THP, Dox, NM, SN, EC, CS, and FF solvents.

Full size image

Desolvation process for Li7P3S11 solid electrolyte formation

Figure 4a,b display the TGA and DTA curves, respectively, for the precursors in Dox, THP, and ACN solvents. The precursor powders in ACN and THP both exhibited three steps of weight loss. Several studies have demonstrated that forward DTA scans of Li3PS4 complexes containing organic solvent exhibit large endothermic peaks13,14. Further, the decomposition reaction of Li7P3S11 to Li4P2S6, Li3PS4, and sulfur occurs at a temperature higher than 280°C30,31. Thus, the first step corresponded to the evaporation of solvent molecules coordinated to Li2S–P2S5, the second step corresponded to the evaporation of solvent molecules coordinated to Li3PS4, and the third step corresponded to the evaporation of sulfur with the decomposition of Li7P3S11. The heat treatment temperature for the formation of Li7P3S11 solid electrolytes was determined from the TG–DTA results. The precursors with Dox, THP, and ACN involved the solvent molecule at 3.0 wt%, 32.3 wt%, and 20.7 wt%, respectively. The respective mole ratios of the coordinated solvent molecules of Dox, THP, and ACN in the precursor were 0.17, 2.73, and 3.14 against 70Li2S–30P2S5. The extent of the weight loss is not correlated with the boiling point of the solvent. The precursors with Dox, THP, and ACN solvents exhibited endothermic peaks at 120 °C, 150 °C, and 200 °C, respectively. These observations indicate that the strength of the chemical interaction between the Li2S–P2S5 system and solvent molecules within the precursor structure increases in the order of Dox, THP, and then ACN. This reflects solvent polarity, characterized by the dielectric constant. In the case of the precursors containing the NM, FF, and SN solvents, continuous weight loss during the annealing process was observed up to 450 °C (Figure S1). This experimental result indicates that these solvent molecules remain following heat treatment at 270 °C. It should be noted that these observations may be influenced by the evaporation of sulfur within 70Li2S–30P2S5 during heat treatment.

Figure 4
figure 4

(a) TGA curves and (b) DTA curves of 70Li2S–30P2S5 precursors synthesized using Dox, THP, and ACN as solvents.

Full size image

Ionic conductivity of 70Li2S–30P2S5 system

Figure 5a shows Nyquist plots of the 70Li2S–30P2S5 solid electrolytes synthesized using Dox, THP, and ACN. The total resistance, including the bulk and grain boundary contributions, was determined by the real-axis intercept at high frequency. The resistance of solid electrolytes decreased in the order of Dox, THP, and then ACN. Figure 5b shows the calculated conductivities as a function of the solvent’s dielectric constant. These results revealed the correlation between the dielectric constant of the solvent and the ionic conductivity of the obtained solid electrolytes. The 70Li2S–30P2S5 solid electrolytes synthesized using ACN solvent exhibited the highest ionic conductivity among the prepared samples of 0.8 mS cm–1 at room temperature. The ionic conductivity for 70Li2S–30P2S5 system depends on the crystallinity, and highly crystalized 70Li2S–30P2S5 solid electrolytes exhibits higher ionic conductivity compared to the amorphous 70Li2S–30P2S532,33. The high ionic conductivity can be explained based on the formation of high-purity crystalline Li7P3S11, which was caused by the high chemical reactivity of the 70Li2S–30P2S5 system in ACN solvent. Commonly, the use of high-DN solvents is an effective strategy for activation of the reaction in a solvent34,35,36,37. In addition to the dielectric constant, the donor number may also play a role in reactivity.

Figure 5
figure 5

(a) Nyquist plot and (b) ionic conductivity at room temperature for 70Li2S–30P2S5 system formed via Dox, THP, and ACN solvents as a function of dielectric constant together with those of 70Li2S–30P2S5 system formed via other solvents12,13,14. The white circles indicate reference values.

Full size image

However, the high reactivity of the Li2S–P2S5 system in low-DN ACN solvent cannot be explained by the DN. In the case of Li3PS4 synthesized via a solvent, the use of a solvent with low polarity will lead to higher ionic conductivity since solvents with low polarity are more easily removed and yield a lower crystallinity ratio14,38. In contrast, the intrinsic high ionic conductivity of Li7P3S11 is caused by a high level of crystallinity32, which is explained by the presence of P2S7 polyhedra in the Li7P3S11 structure39. Therefore, a high dielectric constant in the solvent is an essential factor in realizing high conductivity in Li7P3S11. However, the samples using solvents with high dielectric constants other than ACN solvent exhibited ionic insulating properties. The FF, SN, and NM solvents caused a side reaction with the Li2S–P2S5 system, which was responsible for the solvent molecule residue present following heat treatment. Further, in the case of EC, the reactivity of the the Li2S–P2S5 system was low despite it having the highest dielectric constant among the solvents examined. Based on this finding, we believe that differences in the structures of solvent molecules lead to different reactivities with lithium thiophosphates. In addition, high dielectric solvents typically have an extremely high boiling point, which also limits the choice of solvent for the wet chemical synthesis of Li7P3S11 solid electrolytes. The above considerations of the dielectric constant, molecular structure, and boiling point demonstrate that ACN is the most suitable solvent for the liquid-phase synthesis of Li7P3S11 among the investigated solvents.

Conclusion

In summary, we undertook a systematic study of the effect of solvent on the reactivity of an Li2S–P2S5 system during the liquid-phase synthesis of Li7P3S11 solid electrolytes. The XRD results indicated that Li7P3S11 crystal phase formed in the case of ACN, THP, and Dox solvents. In contrast, a side reaction and insufficient reactivity for the synthesis of Li7P3S11 were confirmed for the CS, NM, FF, SN, and EC solvents. The microstructures of the samples prepared via ACN, THP, and Dox had notable textures, such as a broccoli crown, reflecting the formation process of Li7P3S11 solid electrolytes via a solid-state reaction. TG–DTA indicated that the strength of the chemical interaction between the lithium thiophosphates and solvent molecules within the precursor structure increased in the order of Dox, THP, and then ACN. The ionic conductivities of the samples formed via Dox, THP, and ACN solvent increased in relation to increases in the solvent’s dielectric constant, and the highest ionic conductivity of 0.8 mS cm–1 at room temperature was achieved in the sample using ACN solvent. This can be explained based on the formation of high-purity crystalline Li7P3S11, resulting from the high chemical reactivity of the 70Li2S–30P2S5 system in the ACN solvent. These experimental results indicate that the solvent’s polarity, characterized by the dielectric constant, plays an important role in the formation of Li7P3S11 using the liquid-phase method. In addition, a high boiling point and a ring structure that cause steric hindrance were unfavorable solvent attributes for the wet chemical synthesis of Li7P3S11 solid electrolyte. Overall, this study revealed that a solvent’s dielectric constant is a significant factor in obtaining the optimal Li7P3S11 via wet chemical reaction. As it stands, it has been demonstrated that the ACN solvent is most suitable for the liquid-phase synthesis of Li7P3S11 in terms of its dielectric constant, molecular structure, and boiling point.

Methods

Synthesis

Lithium sulfide (99.9%, Mitsuwa) and P2S5 (99%, Merck) were mixed in a mole ratio of 7:3 and then added into Dox (anhydrous, 99.8%, Aldrich), THP (anhydrous, 99%, Aldrich), CS (anhydrous, ≥ 99%, Aldrich), ACN (super dehydrated, 99.8%, Fujifilm), FF (99%, Aldrich), NM (98%, TCI), SN (99.0%, TCI), and EC (anhydrous, 99%, Aldrich) solvents, separately. The mass/volume ratio of the powder/solvent was 1:20 g/ml. The resultant mixtures were stirred at 50 °C for 3 days, and the obtained suspensions were dried under vacuum at 80 °C for 12 h. In the case of FF, SN, and EC, a temperature of 150 °C was required during the drying process. After that, the precursor powders were annealed at 270 °C for 1 h to obtain the Li7P3S11 solid electrolytes. The entire Li7P3S11 synthesis process was handled under an Ar atmosphere.

Material characterization

Powder XRD measurements were carried out under 2θ = 10° − 50° with a step interval of 0.02° and a scan rate of 1° min−1 using a Rigaku Ultima IV diffractometer. The X-ray beam was generated by CuKα radiation (40 kV, 30 mA). We used an XRD holder with a beryllium window (Rigaku). Scanning electron microscopy characterization was carried out using an FE-SEM (S4800, Hitachi), and TG–DTA (EVO II, Rigaku) was performed under Ar flow with a temperature increase of 5 K min−1.

Electrochemical measurements

The total conductivities of the solid electrolytes were measured through samples cold-pressed into pellets with diameters of ~ 10.0 mm under a pressure of 254 MPa. To assemble the cell for electrochemical impedance spectroscopy (EIS) measurements, each sample (~ 80 mg) was filled into a holder made of polyether ether ketone (PEEK) with two stainless steel rods as blocking electrodes. EIS measurements were conducted via alternating-current impedance spectroscopy (SI 1260, Solartron) in a frequency range of 1 MHz to 10 Hz under a dry Ar flow at a temperature of 298 K.