The Detail Matters: Unveiling Overlooked Parameters in the Mechanochemical Synthesis of Solid Electrolytes

Abstract

The advent of all-solid-state lithium-ion batteries has advanced energy storage technologies with the development of highly conductive solid electrolytes. Numerous researchers have reported the structural and electrochemical performance of solid electrolytes obtained through different production techniques and with different compositions. (1,2) However, even in relatively robust production techniques using ball-milling with the same composition and stoichiometry, only a minute difference in the synthesis process can significantly affect the crystallization mechanisms and resulting ionic conductivity, thereby highlighting the importance of overlooked parameters. This Viewpoint demonstrates the effects of “premixing”─mixing the precursors with a mortar and pestle prior to the mechanochemical synthesis of glassy solid electrolytes, particularly Li₂S–P₂S₅ sulfides and the newly emerging NaTaCl₆ halides─on the structure and transport of the resulting products. Crystal structures and amorphous configurations of sulfide and chloride electrolytes with high ionic conductivities and excellent mechanical properties have been identified. (3−11) These electrolytes are commonly produced through mechanochemical synthesis using ball-milling, which has been widely utilized with various chemical compounds. (12−14) Li₇P₃S₁₁ is recognized as a metastable phase that is nucleated by the subsequent heat treatment of Li₂S–P₂S₅ glasses produced by planetary ball-milling. NaTaCl₆ is recognized as a mixture of crystal and amorphous phases that shows an excellent electrochemical window. In both cases, a wide range of ball-milling experimental parameters have been investigated, including the amount of powder, number of balls, rotation speed, and ball-milling time, leading to the successful synthesis of the target phases. (15−24) This experimental fact makes these synthesis methods for producing Li₇P₃S₁₁ and NaTaCl₆ seemingly very robust. Nonetheless, although the apparent crystal structures evaluated by X-ray diffraction (XRD) are almost identical, there are significant differences in their ion conductivities. (18,25−27) Moreover, only three of 15 studies reported on Li₇P₃S₁₁ syntheses (3,28,29) have described the hand-mixing of starting powders prior to or intermittently during ball-milling, as given in Table S1. Similarly, one out of three studies of NaTaCl₆ have described the hand-mixing of starting powders (Table S2). Although the details of the hand-mixing have been deemed to have negligible effects, this study demonstrates the importance of such a process. This Viewpoint demonstrates the impact of the unspecified details of mechanochemical synthesis on the crystallization mechanisms and ionic conductivities of the products, highlighting the importance of parameters that are often overlooked in such syntheses. Two synthesis routes based on the existing literature (3,6,17,30) were employed to synthesize Li₇P₃S₁₁ using ball-milling, as shown in Scheme 1. The difference between the two routes is the implementation of a hand-mixing step prior to mechanical milling, referred to as premixing. For Sample 1, the hand-ground Li₂S precursor was thoroughly mixed together with P₂S₅ for 20 min using an agate mortar and pestle. In contrast, the precursors of Sample 2 were just stirred with a spatula for 2 min prior to milling. Similarly, the impact of premixing on the synthesis of NaTaCl₆, synthesized by mechanical milling of NaCl and TaCl₅, was assessed through investigating two NaTaCl₆ samples synthesized with or without premixing. The details of the compositions during the synthesis of the powders, including the required reagents and synthesis process, are given in Tables S3 and S4. Figure 1(a, b) displays the ³¹P magic-angle spinning nuclear magnetic resonance (³¹P MAS NMR) spectra of the Li₇P₃S₁₁ glass produced by milling with or without premixing before heating. Despite the broad halo patterns in XRD (Figure S1), which indicate that both the samples, with and without premixing, were completely amorphous, distinctly different local anion coordination is observed in the NMR spectra. The spectra of both samples were deconvoluted into three peaks, where PS₄^3– and P₂S₇^4– are constituent elements of the target Li₇P₃S₁₁ phase and P₂S₆^4– is the local unit of the Li₄P₂S₆ impurity phase. (19) Figure 1(c) shows a comparison of the area-size fractions of the peaks corresponding to the three anion blocks. Although a glassy phase typically contains various local structures as its nature, the amount of P₂S₆^4– in Sample 1 is less than that in Sample 2. Furthermore, the area-size ratios of P₂S₇^4–, and PS₄^3– are 1.70 and 1.24 for Samples 1 and 2, respectively. The area-size ratio of Sample 1 approaches 2.00, which is the theoretical stoichiometric ratio of the two components in crystalline Li₇P₃S₁₁, indicating the anion blocks in Sample 1 are similar to those in Li₇P₃S₁₁ even before heat treatment. Data with a wider range of chemical shifts, confirming no overlap between sidebands and main peaks, are shown in Figure S2. Figure 1. ³¹P MAS NMR spectra and deconvoluted peak profiles of the as-milled powders of (a) Sample 1 and (b) Sample 2. (c) Area-size fractions of Samples 1 and 2. In situ synchrotron X-ray diffraction (SXRD) demonstrates evidently different crystallization processes between the two Li₂S–P₂S₅ glasses (Figure 2). Sample 1 exhibits a single-step crystallization at approximately 220 °C. Conversely, Sample 2 undergoes a two-step crystallization process owing to the sequential formation of Li₃PS₄ at approximately 220 °C, followed by Li₇P₃S₁₁ at approximately 240 °C. Rietveld refinement was performed on the temperature-dependent diffractograms to quantify the Li₇P₃S₁₁, Li₃PS₄, and Li₄P₂S₆ fractions at various temperatures. Sample 1 crystallizes into Li₇P₃S₁₁ without apparent side phases, whereas Sample 2 comprises 11% Li₃PS₄ and 2% Li₄P₂S₆ even at 300 °C, which can degrade its ion-transport properties. (30−33) The diffraction profiles did not show a significant difference in the peaks of the Li₇P₃S₁₁ phase (Figure S3). Differential thermal analysis (DTA) confirmed the difference in the crystallization processes (Figure S4). While a sharp exothermic signal was observed in Sample 1, a broad signal with two distinct peaks was seen in Sample 2. This trend was reproducible even for the same samples in different batches, indicating the local inhomogeneity of the anion blocks in the original glass states, as revealed in the NMR data. Figure 2. SXRD data of the powders heated at 60 °C/min under a N₂ flow and phase ratios from Rietveld refinement of (a) Sample 1 and (b) Sample 2. A distinctly different crystallization process induced by local structural differences leads to variations in the properties of the resulting crystalline Li₇P₃S₁₁ phases. Figure S5 shows the ³¹P NMR spectra of the two samples after crystallization. While the quantification of local units is challenging with the spectra from the heat-treated samples due to the appearance of the cross peaks in Sample 1, the qualitative difference is clearly observed. It should be noted that such cross peaks associated with P₂S₇^4– and PS₄^3– are commonly observed in well-crystalline Li₇P₃S₁₁ phases. (34) To further investigate the differences in the crystalline phases of Samples 1 and 2, pair-distribution function (PDF) analysis was performed on Samples 1 and 2. As the data from both samples showed the appearance of long-range ordering only after heating, the two samples were largely similar. However, a slight change in G(r) in short-range order of ∼3.4 Å appeared, depending on the hand-milling processes (Figure S6). The morphologies of Samples 1 and 2 after heating showed no significant differences (Figure S7). Overall, it is evident that the premixing procedure prior to milling severely impacts the resulting structure. The above-mentioned differences led to significant differences in the ion-transport properties of the resulting materials. The powders crystallized at 280–300 °C were pressed into pellets, and their ionic conductivities were measured by temperature-dependent impedance spectroscopy to analyze the Li-ion transport in the samples. Furthermore, the migration barrier of the Li-ion conduction was determined according to the Arrhenius relation: Figure 3. Arrhenius plots of the temperature and ion conductivity of Samples 1 and 2. As another model system to assess the impact of premixing, we employed a newly found halide-based solid electrolyte, NaTaCl₆, to further investigate the impact of premixing. The ion-transport properties of NaTaCl₆ are reported to significantly vary with its crystallinity. (27) Moreover, the impact of elongated milling time of mechanochemical synthesis on the ion-transport properties has been revealed. (26) In this section, four samples were synthesized, either with or without premixing, and subjected to milling for either 16.5 or 100 h. Figure 4 shows the room-temperature ionic conductivity and activation energies of four samples obtained with different milling times, with or without premixing. The samples without premixing were synthesized via ball-milling; that is, the precursors were directly placed into the ball-mill cup with the milling media. Meanwhile, the samples with premixing were mixed well with a mortar and pestle before ball-milling. Prolonging the milling time from 16.5 to 100 h slightly improved the ionic conductivity. However, more significant improvements in the ionic conductivity were achieved by premixing the precursors well, whereas there is no significant difference in their XRD patterns. Notably, milling for 100 h was insufficient to achieve the same level of ion transport as that of the sample subjected to premixing. These results highlight the importance of the premixing step. Figure 4. Ionic conductivity and activation energies of mechanochemically synthesized NaTaCl₆ samples under different conditions. Each sample was synthesized three times, and the error bars are the standard deviations from the three batches of samples. Li₇P₃S₁₁ and NaTaCl₆ syntheses highlight the crucial effect of the hand-mixing step before mechanical milling on the local structure, crystallization temperature, and ionic conductivity. Although further accumulation of experimental evidence is necessary to precisely elucidate the underlying mechanisms, the current findings are as follows: Significant variations in the fractions of the local units, indicating the local inhomogeneity, were observed without the implementation of the premixing step. These localized compositional variations can lead to diverse anionic blocks during the subsequent crystallization processes, which, in turn, alters the structure and properties of the products. In the case of NaTaCl₆, extended ball-milling durations did not eliminate localized compositional variations, as inferred from the limited conductivity improvement when the ball-milling time was increased from 16.5 to 100 h. Meanwhile, the resulting ion transport was greatly improved by the introduction of a short premixing step before mechanochemical synthesis. While the impact of extended milling may vary with the chemical and mechanical properties of the precursors, the premixing procedure helps to reduce the synthesis time to obtain the target phases. Differential adhesion to the milling media was proposed as a potential cause of compositional inhomogeneity. Ball-milling homogenized mixed precursors when they were trapped between the milling media or between the media and the inner wall of the cup. This process formed powders with layered structures of various compositions. Such a reaction was promoted by mechanical mixing as the interlayer distance decreased, reducing the diffusion length required to form the target phase. However, when a heterogeneous precursor mixture was introduced into the milling cup, softer materials tended to adhere to the milling medium first. Consequently, we hypothesized that this formed thick layers with a constant diffusion length, leading to pronounced local compositional variations. Mechanical properties of starting materials and final products can be critical for proceeding the reactions. (35) As significant variation in the seemingly reproducible properties─ionic conductivity and cycling performance─is one of the big challenges in the field of solid-state batteries and attracts increasing attention, with the recent reports showcasing the critical issue, (36,37) our findings on the drastic impact of the premixing procedure before the seemingly robust mechanochemical synthesis with ball-milling further highlight the importance of the details in the synthesis conditions. The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsenergylett.4c02156. Table of the synthesis process of Li₇P₃S₁₁ and NaTaCl₆, characterization details, measurement results (XRD patterns, NMR spectra, DTA curves, PDF and SEM images) of Li₇P₃S₁₁ (PDF) The Detail Matters: Unveiling Overlooked Parameters in the Mechanochemical Synthesis of Solid Electrolytes 1 views 0 shares 0 downloads Most electronic Supporting Information files are available without a subscription to ACS Web Editions. Such files may be downloaded by article for research use (if there is a public use license linked to the relevant article, that license may permit other uses). Permission may be obtained from ACS for other uses through requests via the RightsLink permission system: http://pubs.acs.org/page/copyright/permissions.html. A. Miura is grateful to Ms. Masae Sawamoto for preparing the powder capillaries for the SXRD measurements. The authors are grateful to Mr. Yuki Chiba at Tohoku University for his support with the solid-state MAS NMR measurements. S.O. gratefully acknowledges Toyota Riken for its financial support through the Rising Fellows Program. This study was partially supported by JST PRESTO JPMJPR21Q8, JST Gtex JPMJGX23S5, and JPMJGX23S2. This article references 37 other publications. This article has not yet been cited by other publications.