Reaction Behavior of a Silicide Electrode with Lithium in an Ionic-Liquid Electrolyte.

Silicides are attractive novel active materials for use in the negative-electrodes of next-generation lithium-ion batteries that use certain ionic-liquid electrolytes; however, the reaction mechanism of the above combination is yet to be clarified. Possible reactions at the silicide electrode are as follows: deposition and dissolution of Li metal on the electrode, lithiation and delithiation of Si, which would result from the phase separation of the silicide, and alloying and dealloying of the silicide with Li. Herein, we examined these possibilities using various analysis methods. The results revealed that the lithiation and delithiation of silicide occurred.

Introduction

The development of high-performance lithium-ion batteries (LIBs) with a high energy density, a long cycle life, and an adequate level of safety is critical to establish a sustainable society because LIBs are used as a power source in electric vehicles and as stationary batteries for utilizing renewable energy.[1,2] Silicon (Si) has great potential as an active material for negative-electrodes in next-generation LIBs due to its higher theoretical capacity (3580 mA h g–1 for Li15Si4) than that of the currently used graphite (372 mA h g–1 for LiC6). However, Si electrodes show poor cycling performance, which is largely caused by a massive volume change in the Si electrode during lithiation (charge) and delithiation (discharge).[3,4] Si has additional drawbacks, namely, a low Li+ diffusion coefficient and high electrical resistivity, thereby preventing the practical application of Si electrodes.[5,6] To solve these issues, many researchers have put in a lot of work toward the construction of new active materials including the following: synthesizing nanostructured Si materials that accommodate volume expansion,[7,8] coating of Si with conductive materials to reduce the electrical resistivity,[9,10] doping of Si with impurities including phosphorous, boron, and so on, to improve its electronic conductivity and/or change its properties, such as its phase transition, crystallinity morphology, and Li+ diffusion,[6,11−13] preparing composites consisting of elemental Si and binary or ternary transition metal silicide to cover the shortcomings of Si,[14−17] and pre-lithiation of Si to decrease the relative volume change in Si during charge/discharge reactions.[18]

An electrolyte contributes toward improving the performance and safety of rechargeable batteries.[19−24] In particular, ionic liquids have superior physicochemical properties as electrolyte solvents: negligible vapor pressure, nonflammability, and a wide electrochemical window.[25−28] Silicide electrodes have shown poor cycling performance in conventional organic-liquid electrolytes.[15,16,29] In contrast, we have reported that some binary silicides can be used as novel negative-electrodes for LIBs in certain ionic-liquid electrolytes; for example, NiSi2, FeSi2, and Cu3Si electrodes maintain high reversible capacity in 1 mol dm–3 (M) lithium bis(fluorosulfonyl)amide (LiFSA) dissolved in N-methyl-N-propylpyrrolidinium bis(fluorosulfonyl)amide (Py13-FSA).[29] Additionally, the volumetric energy density of LIBs using silicide electrodes will inevitably be high because the density of silicides (approximately 5 g cm–3) is more than double that of silicon (2.33 g cm–3). However, the reaction mechanism of silicide electrodes has not yet been clarified. In contrast, the lithium storage mechanism of other Si-based electrodes has been reported.[30,31]

Possible reactions at the silicide electrode are as follows: (1) deposition and dissolution of Li metal on the silicide (Li+ + e– → Li), (2) lithiation and delithiation of Si, which would result from the phase separation of the silicide (FeSi2 + xLi+ + xe– → LiSi2 + Fe and/or FeSi2 + xLi+ + xe– → LiSi + FeSi), and (3) alloying and dealloying of the silicide with Li (FeSi2 + xLi+ + xe– → LiFeSi2). In this study, we examined these possibilities using solid-state 7Li magic-angle spinning (MAS) nuclear magnetic resonance (NMR), X-ray diffraction (XRD), Raman spectroscopy, transmission electron microscopy (TEM), and energy-dispersive X-ray spectroscopy (EDS).

Experimental Section

Synthesis of Silicide Powder and Electrode Preparation

Binary silicide of FeSi2 was synthesized by a mechanical alloying (MA) method. Commercially available silicon (99.9%) and iron (99.9%) powders were purchased from FUJIFILM Wako Pure Chemical Corporation, Ltd. A mixture of these powders was put in a zirconia vessel together with balls. The Si/Fe molar ratio was 3.0 and the weight ratio of the balls to the mixture was approximately 15:1. The MA was done using a high-energy planetary ball mill (P-6, Fritsch) at 380 rpm for 70 h. Although the mixture ratio was higher than the stoichiometric ratio, no elemental Si was detected in the XRD pattern and Raman spectrum. Hence, a residue on the ball and/or inside wall of the vessel should include excess Si.[29]

An FeSi2 electrode was prepared by slurry coating and gas deposition (GD) methods for NMR and TEM measurements, respectively. We used acetylene black (AB), styrene–butadiene rubber (SBR), and carboxymethyl cellulose (CMC) as the conductive agent, binder, and thickener, respectively. The ratio of FeSi2/AB/SBR/CMC was 70/15/5/10 wt %. Deionized water was used as a dispersing agent. The prepared slurry was coated on a Cu current collector and was dried at 120 °C to form an active material layer. The GD method requires no conductive agent or binder, and its detailed conditions have been reported in our papers.[6]

Cell Assembly and Charge and Discharge Testing

A 2032-type coin cell was assembled in an Ar-filled glovebox (Miwa MFG, DBO-2.5LNKP-TS) with a dew point below −90 °C and an oxygen content less than 1 ppm. An FeSi2 electrode, a glass fiber filter (Whatman GF/A), and a Li metal sheet (thickness: 1 mm, 99.90%, Rare Metallic Co., Ltd.) were used as the working electrode, the separator, and the counter electrode, respectively. The ionic-liquid electrolyte used was 1 M LiFSA/Py13-FSA. Electrolyte preparation was also performed in the glovebox.

Galvanostatic charge–discharge testing was performed using an electrochemical measurement system (HJ-1001SM8 or HJ-1001SD8, Hokuto Denko, Co., Ltd.) in the potential range between 0.005 and 2.000 V vs Li+/Li at 30 °C with no capacity limit. The current density was set at 50 mA g–1. Then, the cell was discharged and charged to adjust the state of charge (SOC) for each point of the NMR measurement. The pre-cycling was performed by the following procedures to form a stable surface film on the FeSi2 electrode: the electrode was charged from an open-circuit potential to 0.500 V vs Li+/Li at 50 mA g–1, and maintained at 0.500 V vs Li+/Li for 12 h. Then, it was discharged from 0.500 to 2.000 V vs Li+/Li at 50 mA g–1.

Solid-State 7Li MAS NMR

The coin cell was disassembled in the Ar-filled glovebox after predefined charge–discharge testing. The FeSi2 slurry electrode was washed thoroughly with diethyl carbonate (DEC) before drying. The active material layer was exfoliated from a Cu current collector using a spatula to seal into a 3.2 mmϕ sample rotor for the NMR measurement. 7Li MAS NMR spectra of FeSi2 at each SOC were obtained using an NMR system (11.7 T magnet, DD2 Agilent Technologies Inc.) at a MAS frequency of 14 kHz. The chemical shifts were referenced to a 1 M aqueous LiCl solution. The peak fitting of NMR spectra was carried out by Origin Pro 8.5.0J (LightStone) software.

TEM Observations

After charge–discharge testing, the cell was disassembled in the glovebox, and the FeSi2 GD electrode was washed with propylene carbonate and DEC before drying. We sliced the electrode into thin sections using focused ion beam scanning electron microscopy (FIB-SEM, SMF2000, Hitachi High-Tech Science Corp.) or FIB (JIB-4501, JEOL, Co., Ltd.). For the FIB process, the electrode surface was protected against damage by the Ga ion beam by carbon coating. The sliced electrode was not exposed to the atmosphere until it was introduced into the chamber of the TEM instrument (JEM-ARM200F, JEOL, Co., Ltd.) using a transfer vessel. TEM observation was performed at 200 kV and energy-dispersive X-ray spectroscopy (EDS) was performed with 10 scans.

Results and Discussion

Possibility of the Deposition and Dissolution of Li Metal on the Silicide

Figure a shows the second charge–discharge curves of the FeSi2 slurry and GD electrodes in 1 M LiFSA/Py13-FSA. Herein, we displayed the second curves because the first charge profile included the reductive decomposition reaction of the electrolyte. The charge–discharge profile of the slurry electrode is similar to that of the GD electrode and hence, other contents, such as AB, SBR, and CMC in the slurry electrode, did not contribute to the electrochemical activity. Additionally, it was confirmed that the charge–discharge reaction of the FeSi2 slurry electrode was stable because the 5th and 10th curves were almost the same as the second curve.

Figure 1

(a) Second charge–discharge curves of FeSi2 slurry and GD electrodes in 1 M LiFSA/Py13-FSA at 50 mA g–1. For comparison, the 5th and 10th curves of the slurry electrode are also shown. (b) 7Li MAS NMR spectra of fully lithiated FeSi2 slurry electrodes during each cycle in 1 M LiFSA/Py13-FSA at 50 mA g–1 and (c) their magnified view.

Figure b shows the 7Li MAS NMR spectra of the FeSi2 electrodes in a fully lithiated state during each cycle in the ionic-liquid electrolyte (1 M LiFSA/Py13-FSA). No peak arising from Li metal appeared at approximately 260 ppm (Figure c),[32] whereas we could confirm a main peak at approximately 0 ppm (discussed below) and some spinning sidebands. Consequently, we concluded that no deposition or dissolution of Li metal occurred on the silicide electrode.

Possibility of Phase Separation of FeSi2

To clarify whether the phase separation of FeSi2 into Si and Fe and/or Si-poor Fe–Si phases occurs (FeSi2 + xLi+ + xe– → LiSi2 + Fe and/or FeSi2 + xLi+ + xe– → LiSi + FeSi), we investigated the phase transition of the FeSi2 electrode during charge–discharge cycling. Figure provides the XRD pattern of the fully lithiated FeSi2 electrode during the 10th cycle. The pattern showed no new peaks that would represent Si, Fe, FeSi, and Li15Si4 phases.[33] In fact, all peaks looked the same as those assigned to the FeSi2 phase before cycling; additionally, these peaks did not shift, as shown in the inset. The results revealed that FeSi2 maintains its crystal structure before and after charge–discharge cycling. Additionally, we have previously confirmed that no peaks assigned to crystalline and/or amorphous Si appear in the Raman spectra of FeSi2 after cycling.[29] However, if the phase separation occurs at an extreme surface of the FeSi2 electrode and ultrafine particles (100 nm or less) of Si form, the particles will not be detected due to the finite resolution of the Raman spectrometer. Thus, we observed a cross section of the FeSi2 electrode after charge–discharge testing by TEM.

Figure 2

XRD pattern of the fully lithiated FeSi2 electrode during the 10th cycle in 1 M LiFSA/Py13-FSA at 50 mA g–1 and its magnified view. The XRD pattern before cycling is also shown.

Figure displays the bright-field (BF) TEM images of the FeSi2 electrode after the fifth cycle and the corresponding EDS mappings of Fe and Si. The active material layer consisted of Fe and Si, as shown in Figure a–c, and these elements are also evenly distributed in the enlarged views (Figure d–i). To confirm whether the observed active material layer is FeSi2 and includes Si that results from phase separation, we measured a selected area electron diffraction (SAED) pattern at area III in Figure a (Figure S1), and the obtained d-spacings are summarized in Table S1. Because all d-spacings were assigned to FeSi2, the active material layer was still FeSi2 after cycling. The XRD, Raman, and TEM results demonstrated that no phase separation of FeSi2 into Si and Fe and/or Si-poor Fe–Si phases occurred and hence, the reversible capacity of the FeSi2 electrode did not result from the lithiation and delithiation reactions of Si.

Figure 3

(a) BF-TEM image of the FeSi2 electrode after the fifth cycle in 1 M LiFSA/Py13-FSA at 50 mA g–1 and the corresponding EDS mappings of (b) Fe and (c) Si. The (d–f) and (g–i) images correspond to areas I and II in part (a), respectively, and area III denotes the SAED measurement point.

Verification of the Alloying and Dealloying of the Silicide with Li

Finally, we expected that FeSi2 itself reacts with Li. Figure a shows the 7Li MAS NMR spectra of the FeSi2 electrodes at various state of charge (SOC) values during the fifth cycle in the ionic-liquid electrolyte. At a high SOC, a shoulder at approximately 11 ppm appeared in addition to a main peak at approximately 0 ppm. Based on the deconvolution of the peak, we defined the peaks at 0 and 11 ppm as Peak A and Peak B, respectively. The appearance of two peaks indicated two possibilities: (a) Li existed in two different chemical environments in the same matter, that is, FeSi2, or (b) Li reacted with two different substances (e.g., Si and FeSi). Although Peak A only appeared at an SOC of 0%, Li should not be stored in FeSi2 because the electrode potential would be too high (Figure S2); thus, Peak A at an SOC of 0% should be assigned to Li in the surface film formed on the FeSi2 electrode through the reductive decomposition of the electrolyte or the Li salt in the electrolyte (LiFSA). Extremely sharp NMR spectra of LiFSA appeared at approximately −1 ppm and we washed the electrode thoroughly with DEC. Additionally, it has been reported that Li-containing compounds, such as Li2O, Li2CO3, in the surface film on the electrode exhibited an NMR peak at approximately 0 ppm.[34−36] Therefore, we concluded that Peak A at an SOC of 0% could be attributed to Li in the surface film.

Figure 4

(a) 7Li MAS NMR spectra of FeSi2 from an SOC of 0–100% during the fifth cycle in 1 M LiFSA/Py13-FSA at 50 mA g–1 and the correlation between the SOC and (b) the peak area or (c) chemical shift.

The intensity of Peaks A and B increased with an increase in the SOC (20–100%), whereas their peak position was approximately constant, as shown in Figure b,c. Because the fifth Coulombic efficiency was 98.2% (Figure S3), it was unlikely that the growth of the surface film caused an increase in Peak A. Key et al.[37,38] reported ex situ 7Li NMR spectra of Si, namely, Li12Si7 and Li15Si4 phases exhibit peaks at 18.0 and 6.0 ppm, which are assigned to Li surrounding small Si clusters and Li near isolated Si, respectively. The peak centered at 18.0 ppm shifts toward a higher magnetic field (lower chemical shift) with a decrease in cluster size (planar Si5 rings, trigonal–planar Si4 clusters, and Si–Si dumbbells), thereby forming Li7Si3 and/or Li13Si4 phases. Additionally, a peak arising from overlithiated Si (Li15+δSi4) phases appeared with a negative chemical shift.[37] The chemical shift of Peak A did not correspond to the above Li–Si alloy phases; therefore, Peak A between an SOC of 20 and 100% is attributed to Li stored in FeSi2.

There remains a possibility that Peak B could be assigned to the Li13Si4 phase because the Li–Si alloy showed an NMR peak at approximately 11.5 ppm. However, the results of Figures and 3 showed no presence of Si during charge–discharge testing; thus, Peak B would arise from Li stored in the different chemical environments of FeSi2. Therefore, the NMR results revealed the existence of two different chemical environments for Li storage in FeSi2, thereby showing that the lithiation and delithiation of the FeSi2 electrode occurred (FeSi2 + xLi+ + xe– → LiFeSi2). We consider that the reaction occurs not only in the ionic-liquid electrolyte but also in the organic-liquid electrolyte.[29]

Because the lithiation of a crystalline or amorphous Si phase exhibited different NMR spectra,[32] we confirmed the 7Li MAS NMR spectra of the FeSi2 electrodes at various SOC values during the first cycle (Figure ). To form the surface film on the electrode at an SOC of 0%, we performed pre-cycling (Figure S4). The appearance and position of Peaks A and B were almost the same as those during the fifth cycle (Figure ), which indicated that FeSi2 maintained its crystallinity, as confirmed in Figure ; thus, no amorphous FeSi2 phase was formed.

Figure 5

(a) 7Li MAS NMR spectra of FeSi2 from an SOC of 0–100% during the first cycle after pre-cycling in 1 M LiFSA/Py13-FSA at 50 mA g–1 and the correlation between the SOC and (b) the peak area or (c) chemical shift.

Although the lithiation and delithiation reactions of the silicide occurred among the possible reactions at the electrode, it is interesting that no peak shifts resulted in the XRD pattern (Figure ). For instance, if LiFeSi2 formed, it would be expected that the lattice constant of LiFeSi2 becomes larger than that of FeSi2. These results indicate that the bulk of FeSi2 did not participate in the lithiation, and rather its surface was involved in alloying with Li. Therefore, the reaction sites (Peaks A and B in the NMR spectra) and reaction areas (bulk or surface) of the silicide are under further study.

Conclusions

We investigated the reaction behavior of a silicide with Li in an ionic-liquid electrolyte. The XRD, Raman, and TEM results revealed that no phase separation of FeSi2 into Si and Fe and/or Si-poor Fe–Si phases occurred during charge–discharge cycling; thus, the reversible capacity of the FeSi2 electrode did not result from the lithiation and delithiation reactions of Si. The NMR results demonstrated that no Li metal was deposited or dissolved on the FeSi2 electrode but that the lithiation and delithiation of FeSi2 did occur. Additionally, there were two different chemical environments for Li storage in FeSi2. Although the reaction sites (Peaks A and B in the NMR spectra) and reaction areas (bulk or surface) are still unclear, our results provide critical insights into the reaction mechanism of novel alloy-based negative electrodes for LIBs.