Long-Term Stable Lithium Metal Anode in Highly Concentrated

Jun 27, 2019 - Rechargeable batteries play an important role in power leveling, efficient utilization of clean ... Li+ ion flux, resulting in uneven p...
0 downloads 0 Views 2MB Size
Download PDF
Research Article Cite This: ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

www.acsami.org

Long-Term Stable Lithium Metal Anode in Highly Concentrated Sulfolane-Based Electrolytes with Ultrafine Porous Polyimide Separator Yuta Maeyoshi,*,† Dong Ding,† Masaaki Kubota,† Hiroshi Ueda,‡ Koji Abe,§ Kiyoshi Kanamura,†,‡,∥ and Hidetoshi Abe*,† †

ABRI Co., Ltd., Building P-302, Tokyo Metropolitan University, 1-1 Minami-Ohsawa, Hachioji, Tokyo 192-0397, Japan 3DOM Inc., 3-9 Moriya-cho, Kanagawa-ku, Yokohama, Kanagawa 221-0022, Japan § Division of Energy Materials Chemistry, Research Center for Advanced Science and Innovation, Organization for Research Initiatives, Yamaguchi University, 2-16-1 Tokiwadai, Ube, Yamaguchi 755-8611, Japan ∥ Department of Applied Chemistry for Environment, Graduate School of Urban Environmental Sciences, Tokyo Metropolitan University, 1-1 Minami-Ohsawa, Hachioji, Tokyo 192-0397, Japan

Downloaded via BUFFALO STATE on July 17, 2019 at 13:51:24 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

‡

S Supporting Information *

ABSTRACT: Highly concentrated solutions composed of lithium bis(fluorosulfonyl)imide (LiFSI) and sulfolane (SL) are promising liquid electrolytes for lithium metal batteries because of their high anodic stability, low flammability, and high compatibility with lithium metal anodes. However, it is still challenging to obtain the stable lithium metal anodes in the concentrated electrolytes due to their poor wettability to the conventional polyolefin separators. Here, we report that the highly concentrated 1:2.5 LiFSI/SL electrolyte coupled with a three-dimensionally ordered macroporous polyimide (3DOM PI) separator enables the stable lithium plating/stripping cycling with an average Coulombic efficiency of ca. 98% for over 400 cycles at 1.0 mA cm−2. The 3DOM PI separator shows good electrolyte wettability and large electrolyte uptake due to its high porosity and polar constituent of the imide structure, allowing superior cycling performance in the highly concentrated solution, compared with the polyolefin separators. Electrochemical and spectroscopic analyses reveal that the superior cycling stability in the concentrated electrolyte is attributed to the formation of highly stable and Li+ ion conductive solid electrolyte interphase (SEI) layer derived from FSI− anions, which reduces the side reactions of SL with lithium metal, prevents the growth of lithium dendrites, and suppresses the increase in cell impedance over long-term cycling. Our findings demonstrate that polar and porous separators could effectively improve the affinity to the concentrated electrolytes and allow the formation of the anion-derived SEI layer by increasing the salt concentration of the electrolytes, achieving the long-term stable lithium metal anode. KEYWORDS: lithium metal anode, sulfolane, lithium bis(fluorosulfonyl)imide, highly concentrated electrolytes, three-dimensionally ordered macroporous polyimide separator, solid electrolyte interphase

1. INTRODUCTION

Lithium metal is considered as the ultimate anode material for rechargeable batteries because it has the highest theoretical specific capacity (3860 mAh g−1) and the lowest electrochemical potential (−3.04 V versus standard hydrogen electrode) among all possible candidates.5 However, dendritic

Rechargeable batteries play an important role in power leveling, efficient utilization of clean energy, and electrification of mobility, which should suppress the consumption of finite resources and greenhouse gas emissions. The growing demand for energy density has stimulated the development of lithium metal rechargeable batteries coupled with high-voltage/highcapacity cathodes.1−4 © XXXX American Chemical Society

Received: March 28, 2019 Accepted: June 27, 2019 Published: June 27, 2019 A

DOI: 10.1021/acsami.9b05257 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

structure.19 Furthermore, the uniform distribution of pores in a quasi-hexagonal close-packed arrangement of the 3DOM PI separator provides uniform current distribution during lithium plating/stripping processes, resulting in the suppression of the dendritic lithium metal growth.21 Therefore, we have expected that the 3DOM PI separator stabilizes lithium plating/ stripping cycling even in the further concentrated LiFSI/SL electrolytes with higher viscosity. Here, we report long-term stable lithium plating/stripping cycling in the highly concentrated 1:2.5 LiFSI/SL electrolyte using the 3DOM PI separator. The 3DOM PI separator improves the electrolyte wettability and electrolyte uptake, and allows the superior cycling performance in the highly concentrated solution, compared with the conventional polyolefin separators. The 1:2.5 LiFSI/SL solution shows higher average Coulombic efficiency of 98% over long-term cycling, relative to the solutions with lower salt concentrations. The mechanism of the enhanced stability in the highly concentrated electrolyte was investigated by electrochemical and spectroscopic analyses. Finally, we demonstrated the stable charge/discharge cycling of the lithium metal battery with LiFePO4 using the 3DOM PI separator and 1:2.5 LiFSI/SL, in comparison with a conventional 1.0 mol dm−3 LiPF6/ EC:DMC electrolyte.

lithium metal growth and low Coulombic efficiency during lithium plating/stripping hinder the practical applications of the lithium metal anode in rechargeable batteries. Significant side reactions between the electrolyte and the lithium metal anode form an unstable solid electrolyte interphase (SEI) with inhomogeneous morphology and composition, which induces a nonuniform Li+ ion flux, resulting in uneven plating and dendritic lithium metal deposition.6 The dendritic lithium metal can break the unstable SEI and cause the electrolyte decomposition to form a new SEI. Repeated lithium plating/ stripping accelerates the break and repair of SEI, which leads to low Coulombic efficiency, accumulated SEI layers, and increased interfacial resistance, thereby deteriorating the performance of the lithium metal anode. The dendritic lithium metal growth possibly results in the short circuit and thermal runaway of cells. Therefore, the design of electrolyte forming a stable and uniform SEI is crucial to ensure high Coulombic efficiency, long cycle life, and safety of rechargeable lithium metal batteries. In recent years, highly concentrated liquid electrolytes with lithium salt concentrations of over ca. 3 mol dm−3 have attracted great attention due to their unique physicochemical and electrochemical properties.7,8 Increasing salt concentration enhances interactions between Li+ cations and anions/solvents, forming a three-dimensional solution structure. The solution structure triggers the downward shift of the anion’s lowest unoccupied molecular orbital (LUMO). Thus, the anions are reduced more preferentially than the solvents, forming a stable and uniform SEI.7−9 Due to this concept, highly stable lithium plating/stripping cycling has been achieved using the concentrated electrolytes of lithium bis(fluorosulfonyl)imide (LiFSI) in various solvents, e.g., 1,2-dimethoxyethane,10 ethylene carbonate (EC):dimethyl carbonate (DMC),11 fluoroethylene carbonate,12 and sulfolane (SL).13 Among the reported electrolytes, SL is a promising solvent for high energy lithium metal batteries because of its high anodic stability and low flammability.14−17 Furthermore, highly concentrated SLbased electrolytes containing LiFSI enables Li+ ion hopping conduction which suppresses concentration polarization in lithium batteries.18 Ren et al. have demonstrated that a LiFSI/ SL solution with a salt-to-solvent molar ratio of 1:3 shows high Coulombic efficiency of ca. 98% and stable cycling of the lithium metal anode.13 However, the electrolyte solution with molar ratios of SL/LiFSI < 3 has not been investigated yet. In addition, the effect of the salt concentration on long-term cyclability of the lithium metal anode is still unclear. Hence, we have attempted to further increase the salt concentration of LiFSI/SL solution to improve the stability of the lithium metal anode. The concentrated LiFSI/SL electrolytes have much higher viscosity and thus show poor wettability to conventional polyolefin separators,13 which hinders lithium plating/stripping in cells. We have developed an ultrafine porous polyimide (PI) membrane with a similar structure to inverse opal previously.19−21 The inverse-opal structure is the so-called threedimensionally ordered macroporous (3DOM) structure illustrated in Figure S1a. Here, we have defined the membrane as the 3DOM PI separator, though the macroporous structure is slightly random, as can be seen in Figure S1b,c. We have already reported that the 3DOM PI separator has the high electrolyte wettability to conventional liquid electrolytes composed of LiPF6 and carbonates, owing to its high porosity of ca. 70% and the relatively polar constituent of the imide

2. EXPERIMENTAL SECTION 2.1. Materials. LiFSI, SL, and 1.0 mol dm−3 LiPF6/EC:DMC (1:1 by volume) were purchased from Kishida Chemical Co., Ltd. These reagents were lithium battery grade and used as-received. Electrolyte solutions were prepared by dissolving the appropriate amount of LiFSI into SL by stirring at 50 °C in an Ar-filled glove box. The 3DOM PI separator was prepared by a colloidal crystal template method according to our previous report.19−21 An opal structure composed of mono-dispersed silica particles (NIPPON SHOKUBAI Co., Ltd.) was used as the template. The vacancy of the template was filled with N,N-dimethylacetamide solution containing polyamic acid (JFE Chemical Corporation). And then it was heated at 320 °C to convert polyamic acid to PI. The 3DOM PI membrane with the inverse-opal structure was obtained by eliminating the template with 10 wt % HF aqueous solution. LiFePO4 and LiNi0.5Co0.2Mn0.3 (NCM523) electrodes were supplied by The Furukawa Battery Co., Ltd. The electrodes were punched into disks (14 mm in diameter) and dried at 100 °C under vacuum overnight before cell assembly. The mass loading of the active material and the cathode areal capacity were ca. 10.35 mg cm−2 and 1.76 mAh cm−2 for LiFePO4, and ca. 14.72 mg cm−2 and 2.36 mAh cm−2 for NCM523, respectively. Li (300 or 600 μm in thickness) and Li/Cu (20/10 μm in thickness) foils were purchased from Honjo Metal Co. Cu foil (18 μm in thickness) was purchased from Hohsen Corp. 2.2. Electrochemical Measurements. Linear sweep voltammetry was performed by an automatic polarization system (HSV-110, Hokuto Denko Co.) at a scan rate of 1 mV s−1 in a three-electrode cell with a platinum disk (3 mm in diameter) as a working electrode and Li foils as reference and counter electrodes at room temperature. Cu|Li, Li|Li, LiFePO4|Li, and NCM523|Li cells were assembled in a 2032-type coin cell in the glove box. The 3DOM PI membrane was used as a separator. The surfactant-coated PP separator was also employed in Cu|Li cell with the 1:2.5 LiFSI/SL solution to investigate the effects of separators on lithium plating/stripping stability. The thickness of Li foils for Cu|Li and Li|Li cells was 600 and 300 μm, respectively. Li/Cu foils (20/10 μm in thickness) were used as anodes for LiFePO4|Li and NCM523|Li cells. The volume of electrolyte in Cu|Li and Li|Li or LiFePO4|Li and NCM523|Li cells were 80 or 140 μL, respectively, to fully wet the separators and the electrodes. Lithium plating/stripping cycling of Cu|Li and Li|Li cells was conducted with a charge/discharge unit (HJ1001SD8, Hokuto Denko B

DOI: 10.1021/acsami.9b05257 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces Table 1. Ionic Conductivity, Viscosity, and Melting Points of LiFSI/SL Solutions electrolyte

ionic conductivitya/mS cm−1

viscositya/mPa s

melting point/°C

1:10.3 LiFSI/SL 1:3.0 LiFSI/SL 1:2.5 LiFSI/SL

3.04 1.95 1.71

23.1 113 154

−5.3 −8.0 −13.6

Conductivity and viscosity were measured at 30 °C.

a

Co.) at a current density of 1.0 mA cm−2 at 30 °C. The effective area of Cu and Li foils for lithium plating/stripping was 1.77 cm2 (15 mm in diameter). For Cu|Li cell, 0.5 mAh cm−2 of lithium was deposited on Cu foil and then stripped until the potential reached 1.0 V vs Li/ Li+ during each cycle. Coulombic efficiency of lithium plating/ stripping was calculated by dividing the amount of lithium stripped by the amount of lithium deposited on the Cu foil. For Li|Li cell, the lithium plating/stripping capacity was 0.5 mAh cm−2. Galvanostatic charge/discharge tests of LiFePO4|Li and NCM523| Li cells were carried out by the charge/discharge unit at 30 °C. Charge and discharge were performed at the same C rate. The cycling tests were conducted at 0.5C rate after an initial one cycle at 0.1C rate and then four cycles at 0.2C rate in a voltage range of 2.5−3.8 V for LiFePO4 and 2.7−4.3 V for NCM523. The 1C rate corresponds to 170 and 160 mA g−1 based on the weight of LiFePO4 and NCM523, respectively. The theoretical capacity ratios of LiFePO4/Li and NCM523/Li were ca. 1:2.34 and 1:1.75, respectively. The internal resistance of Li|Li and LiFePO4|Li cells during cycling was measured by electrochemical impedance spectroscopy (EIS) with an AC modulation amplitude of ±5 mV in a frequency range from 1 MHz to 0.01 Hz (SI 1287 and 1255B, Solartron Analytical) at 30 °C. 2.3. Characterization. The ionic conductivity of the electrolyte solutions was measured by the EIS with an AC modulation amplitude of ±5 mV in a frequency range from 100 kHz to 1 Hz using a symmetric cell with two platinized platinum electrodes at 30 °C. The viscosity was determined using a viscometer (TVE-35, Toki Sangyo Co., Ltd.) at 30 °C. The liquid phase stability of the electrolyte solutions was evaluated using a differential scanning calorimeter (DSC, DSC-60, Shimadzu Co.). The samples were sealed in aluminum pans in the glove box. The sample pans were cooled to −100 °C and then heated to 100 °C at a rate of 10 °C min−1 under a nitrogen atmosphere. The melting points were determined from the onset temperatures of the heating thermograms. The surface and cross-section of the 3DOM PI separator were observed by a scanning electron microscope (SEM, JSM-7500F, JEOL Ltd.). Electrolyte wettability of polypropylene (PP), surfactant-coated PP and 3DOM PI separators was evaluated by observing how 1:2.5 LiFSI/SL electrolyte droplets spread on the separators at room temperature. The electrolyte uptake of the separators was determined using the following equation

The SEI layers on the cycled Li foils were studied by X-ray photoelectron spectroscopy (XPS) using a PHI 5000 VersaProbe II (ULVAC-PHI, Inc.) with an Al Kα X-ray source. The Cu|Li cells after 50 cycles of lithium plating/stripping were disassembled in the glove box. The obtained Li foils were washed with DMC to remove the residual electrolyte components and then dried under vacuum. The samples were transferred to the XPS chamber using a transfer vessel without exposure to air. The binding energy of each spectrum was calibrated using the standard energy of the C 1s peak of amorphous carbon at 285.0 eV. Curve fitting of spectra was performed with Voigt function (i.e., Gaussian−Lorentzian function) after Shirley-type background subtraction. The morphologies of the cycled stainless steel (SS316L) coin cell cases were observed by the SEM (JSM-6490A, JEOL Ltd.). The NCM523|Li cells after 30 cycles of the charge/discharge tests were disassembled in the glove box. The obtained stainless steel coin cell cases were washed with DMC to remove the residual electrolyte components and then dried under vacuum.

3. RESULTS AND DISCUSSION 3.1. Physicochemical Properties and Anodic Stability of Electrolyte. We prepared three concentrations of LiFSI/SL electrolytes with salt-to-solvent molar ratios of 1:10.3, 1:3.0, and 1:2.5. All LiFSI/SL electrolytes were liquid at room temperature. Their ionic conductivity and viscosity are listed in Table 1. Even at a high salt-to-solvent molar ratio of 1:2.5, the electrolyte exhibited acceptable ionic conductivity of 1.71 mS cm−1 despite a high viscosity (154 mPa s) associated with the solvent and increased salt concentration. Melting points of the mixtures were estimated from DSC thermograms (Figure S2) and presented in Table 1. The 1:10.3 LiFSI/SL electrolyte showed lower melting points of −5.3 °C compared with pure SL (27.5 °C).18 As the salt-to-solvent molar ratio increased to 1:3.0 and 1:2.5, the melting points of the mixtures decreased to −8.0 and −13.6 °C, enhancing liquid phase stability at low temperatures. The anodic stability of LiFSI/SL solutions and the conventional 1.0 mol dm−3 LiPF6/EC:DMC electrolyte was evaluated by linear sweep voltammetry. As shown in Figure S3, the conventional electrolyte showed anodic decomposition above 4.0 V vs Li/Li+, whereas all LiFSI/SL electrolytes remained stable due to intrinsic high anodic stability of SL.14−17 Although low anodic current was observed at ca. 4.2 V vs Li/Li+ for 1:10.3 LiFSI/SL, the concentrated LiFSI/SL solutions with salt-to-solvent molar ratios of 1:3.0 and 1:2.5 exhibited remarkably high anodic stability above 5 V vs Li/Li+, in agreement with the recent report by Alvarado et al.16 In the report, both experiments and quantum chemistry calculations demonstrated that increasing salt concentration promotes complex/aggregate formation of Li+ cations and FSI− anions with Li+-coordinated SL, which slows the decomposition of SL and leads to polymerization under oxidative conditions. This effect should contribute to the significant anodic stability of the concentrated LiFSI/SL solutions. 3.2. Affinity between Electrolyte and Separator. We employed a polar 3DOM PI separator with high porosity to ensure good electrolyte wettability and stable lithium plating/

electrolyte uptake (%) = (Wf − Wi )/ Wi × 100 where Wi and Wf are the weights of the separators before and after soaking in the 1:2.5 LiFSI/SL solution for 2 h at room temperature. The thickness of all of the separators was about 25 μm. The functional groups on the surface of the separators were studied by Fourier transform infrared (FT-IR) spectroscopy (FT/IR-6100, JASCO Co.) with an attenuated total reflectance method at room temperature. The morphologies of the lithium metal deposition were observed by an SEM (JSM-6490A, JEOL Ltd.). A total of 1.0 mAh cm−2 of lithium was deposited on the Cu foil in the Cu|Li cells, and then the cells were disassembled in the glove box. The Cu foils plated with the lithium metal were washed with DMC to remove the residual electrolyte components and then dried under vacuum. The samples were transferred to the SEM chamber using a transfer vessel without exposure to air. The solution structure of the electrolyte solutions was studied by Raman spectroscopy with a 532 nm laser (NRS-1000, JASCO Co.) at room temperature. The solutions were enclosed in a glass tube in the glove box to avoid exposure to air. All spectra were the average of 32 scans collected at a resolution of 2.7 cm−1. C

DOI: 10.1021/acsami.9b05257 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

Figure 1. (a) Photograph of electrolyte wetting behavior and (b) electrolyte uptake of PP, surfactant-coated PP, and 3DOM PI separators at room temperature. A 1:2.5 LiFSI/SL solution was used as an electrolyte solution.

Figure 2. Voltage versus capacity of lithium plating/stripping in Cu|Li cells using LiFSI/SL solutions with salt-to-solvent molar ratios of (a) 1:10.3, (b) 1:3.0, and (c) 1:2.5. (d) Coulombic efficiency versus cycle number of the cells using different concentrations of LiFSI/SL solutions. A current density and a deposition capacity are 1.0 mA cm−2 and 0.5 mAh cm−2, respectively. SEM images of morphologies of lithium metal deposited on Cu foils in LiFSI/SL solutions with salt-to-solvent molar ratios of (e) 1:10.3, (f) 1:3.0, and (g) 1:2.5 at a current density of 1.0 mA cm−2 and a deposition capacity of 1.0 mAh cm−2. The electrochemical measurements were performed at 30 °C.

stripping behavior in the concentrated LiFSI/SL solutions. The concentrated electrolytes have higher viscosity and poor wettability to nonpolar polyolefin separators with low porosity, which may result in high overpotential and unstable lithium plating/stripping behavior. Separators with better electrolyte wettability should resolve this issue. Figure 1a shows the photograph of electrolyte wettability to the three kinds of separators. The 1:2.5 LiFSI/SL solution with high viscosity (154 mPa s) can easily spread over the 3DOM PI separator, whereas the droplets with apparent contact angles remained on

PP and surfactant-coated PP separators. Furthermore, the 3DOM PI separator afforded a much larger electrolyte uptake of 584% relative to that of PP (3%) and surfactant-coated PP (96%) separators, as shown in Figure 1b. The superior electrolyte wettability and uptake of the 3DOM PI separator are due to not only its high porosity of ca. 70% but also the polar constituent of the imide structure.19 Unlike the conventional separators based on nonpolar PP, the FT-IR spectrum of the 3DOM PI separator exhibited strong and sharp absorption peaks at 1775, 1720, 1380, and 725 cm−1 D

DOI: 10.1021/acsami.9b05257 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

Figure 3. (a) Voltage versus cycle number of lithium plating/stripping in Li|Li cells using LiFSI/SL solutions with salt-to-solvent molar ratios of 1:10.3, 1:3.0, and 1:2.5. A current density and a deposition capacity are 1.0 mA cm−2 and 0.5 mAh cm−2, respectively. (b−d) Magnified views of a. (e) Electrolyte resistance (Re) and (f) interfacial resistance of lithium metal (Rint) versus cycle number of lithium plating/stripping in Li|Li cells using different concentrations of LiFSI/SL solutions. The values were obtained by fitting the EIS spectra (Figure S8) to the equivalent circuit model (Figure S9). The electrochemical measurements were performed at 30 °C.

related to CO symmetric stretching, CO asymmetric stretching, C−N stretching, and CO bending, respectively (Figure S4).19 These polar functional groups have a strong affinity with the concentrated electrolyte solution consisting of the polar solvents, Li+ ions, and FSI− anions and thus give the 3DOM PI separator abilities to absorb and retain the electrolyte. The outstanding affinity between the 3DOM PI separator and the concentrated LiFSI/SL solution certainly enables long-term stable lithium plating/stripping cycling. 3.3. Lithium Plating/Stripping Stability and Deposition Morphology. Long-term lithium plating/stripping stability in different salt concentrations of the LiFSI/SL solutions was investigated using Cu|Li cells combined with the 3DOM PI separator, as shown in Figure 2. It can be seen that Coulombic efficiency and cycle life were enhanced with increasing salt concentration. Figure 2a−c displays voltage versus capacity of lithium plating/stripping in Cu|Li cells using different LiFSI/SL solutions. The Coulombic efficiency during cycling is compared in Figure 2d. In dilute electrolyte (1:10.3 LiFSI/SL), Coulombic efficiency was only 68.4% at the 1st cycle, and never reached 80%, indicating significant side reactions between the electrolyte and the lithium metal. The overpotential between lithium plating/stripping greatly increased with the cycle number, resulting in the rapid drop of

Coulombic efficiency within 50 cycles, probably owing to the fast accumulation of the side reaction products on the lithium metal surface. In contrast, as the salt-to-solvent molar ratio of the electrolyte increased to 1:3.0 and 1:2.5, Coulombic efficiency at the 1st cycle markedly improved to 86.1 and 93.2%. The Coulombic efficiency for both electrolytes reached ca. 98% within 100 cycles and remained stable for over 250 cycles. The increase in the overpotential during cycling was also suppressed with increasing salt concentration. These results suggest that a more stable SEI layer is formed in the electrolyte with a higher salt concentration, which limits the side reactions. As a result, 1:2.5 LiFSI/SL showed the higher cycling stability with the average Coulombic efficiency of 98.0% over 440 cycles, compared with those of 1:3.0 LiFSI/SL (97.3% over 370 cycles) and 1:10.3 LiFSI/SL (68.2% over 50 cycles). However, even the concentrated electrolytes resulted in random voltage oscillations and sudden decrease in Coulombic efficiency after long-term cycling, as shown in Figures S5 and 2d, which can be attributed to a short circuit or the high-impedance-related cell failure.10 The morphologies of lithium metal deposited on Cu foils in the different LiFSI/SL solutions were observed by the SEM, as displayed in Figure 2e−g. In the dilute electrolyte, the deposited lithium metal had irregular structures containing E

DOI: 10.1021/acsami.9b05257 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

Figure 4. Raman spectra of LiFSI/SL solutions with various salt-to-solvent molar ratios in (a) 420−480 cm−1 (SO2 twist mode of SL molecules) and (b) 700−780 cm−1 (S−N stretching mode of FSI− anions) at room temperature.

structure of the 3DOM PI separator affords uniform current distribution.21 In addition, the polar functional groups of the 3DOM PI separator can adsorb considerable Li+ ions within the porous structure, reducing the concentration polarization and overcoming the diffusion-limited current, which provides evenly distributed Li+ ion flux as reported previously.23,24 The uniform Li+ ion flux leads to the dendrite-free lithium metal deposition. From the above results, the 3DOM PI separator is essential to attain long-term stable lithium plating/stripping in the concentrated electrolyte. 3.4. Internal Resistance Changes during Lithium Plating/Stripping. To study the detailed internal resistance changes in the LiFSI/SL electrolytes, we performed lithium plating/stripping tests of Li|Li cells coupled with EIS. Voltage profiles of the cells with different concentrations of LiFSI/SL solutions are shown in Figure 3a−d. The increase in the overpotential for lithium plating and stripping processes was suppressed with increasing salt concentration, in accordance with the results in Cu|Li cells (Figure 2a−c). The internal resistance of the Li|Li cells was measured by EIS after 10, 100, 300, and 500 lithium plating/stripping cycles, as exhibited in Figure S8. In the spectra, an intersection of a semi-circle with the real axis at high frequency is assigned to the bulk electrolyte resistance (Re); a depressed semi-circle in middleand high-frequency regions corresponds to the sum of the SEI resistance (RSEI) and the charge-transfer resistance of the lithium metal (Rct).25−27 Here, the sum of RSEI and Rct are interpreted as the interfacial resistance of lithium metal (Rint). Re and Rint values were obtained by fitting the EIS spectra to the equivalent circuit model as shown in Figure S9 and plotted versus cycle number of lithium plating/stripping in Figure 3e,f. In the dilute electrolyte, Re and Rint rapidly increased after 100 cycles, which is attributed to the serious consumption of the electrolyte and the fast growth of the SEI layer. It is considered that the increased Rint is mostly related to the increase in RSEI because the electrolyte is decomposed continuously on the lithium metal to grow the SEI layer during lithium plating/ stripping cycling. Meanwhile, the repetitive cycling certainly increases the surface area of the lithium metal and lowers Rct. In sharp contrast to the dilute electrolyte, both 1:3.0 and 1:2.5 LiFSI/SL solutions showed much lower Rint during cycling, indicating the highly Li+ ion conductive and stable SEI layers formed on the surface of lithium metal. However, Rint in 1:3.0 LiFSI/SL solution increased faster than 1:2.5 LiFSI/SL solution within 300 cycles, finally resulting in the huge increase in Re and Rint by 500 cycles. The 1:2.5 LiFSI/SL solution suppressed the increase in Re and Rint through repetitive lithium plating/stripping, achieving the most stable

needle-like dendrites with a width of a few hundred nanometers (Figure 2e). The lithium dendrites increase the surface area of the lithium metal anodes, which accelerates the side reactions between the electrolyte and lithium metal, resulting in the poor cyclability (Figure 2a,d). Furthermore, the growth of the dendritic lithium metal can penetrate into the porous separators, leading to the short circuit in the batteries.11 By contrast, round-shaped lithium depositions with a grain size of ca. 5 μm were formed in the concentrated LiFSI/SL with salt-to-solvent molar ratios of 1:3.0 and 1:2.5 (Figure 2f,g). Interestingly, the grain sizes of lithium depositions increased with increasing salt concentration. The large lithium depositions reduce the surface area of the lithium metal and minimize the side reactions, leading to long cycle life (Figure 2c,d). Moreover, the round-shaped structures restrict their ability to penetrate into the porous separators and improve the safety of the batteries.10 Such beneficial morphology of lithium metal deposition is attributed to the uniform and high concentration of Li+ ion flux and/or the high Li+ ion conductive and stable SEI layer in the concentrated electrolyte.7 Effects of separators on lithium plating/stripping stability were also investigated using Cu|Li cells with 1:2.5 LiFSI/SL solution. As expected, the 3DOM PI separator showed superior lithium plating/stripping cycling compared with the surfactant-coated PP separator. As shown in Figure S6a, when the surfactant-coated PP separator was used, overpotential between lithium plating and stripping was ca. 120 mV, which was much higher than that of the 3DOM PI separator (ca. 40 mV in Figure 2c), due to its poor electrolyte wettability and small electrolyte uptake, as mentioned above. The high overpotential promotes the undesirable side reactions between the electrolyte and lithium metal, resulting in the lower average Coulombic efficiency of 96.1% over 190 cycles as shown in Figure S6b, in comparison with that of 3DOM PI (98.0% over 440 cycles in Figure 2d). The Coulombic efficiency of the cell using the surfactant-coated PP separator fluctuated after ca. 200th cycle (Figure S6b), which may suggest dead lithium formation.22 Furthermore, the morphologies of the deposited lithium metal in 1:2.5 LiFSI/SL solution differ depending on the separators. In contrast to the large lithium particles for the 3DOM PI separator (Figure 2g), the surfactant-coated PP separator formed smaller lithium deposition with a random structure as shown in Figure S7. The decrease in the particle size of the lithium depositions would be attributed to the nonuniform pore structure and the nonpolar constituent of the surfactant-coated PP separator, which hinders the uniform Li+ ion flux. It has been reported that the ordered macroporous F

DOI: 10.1021/acsami.9b05257 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

Figure 5. (a) Atomic concentration of the SEI layers on Li foils after 50 plating/stripping cycles in LiFSI/SL solutions with salt-to-solvent molar ratios of 1:10.3, 1:3.0, and 1:2.5. (b) Magnified figure for N and F atoms. The values were obtained by the XPS analysis.

Figure 6. High-resolution F 1s, N 1s, and S 2p spectra of the SEI layers on Li foils after 50 plating/stripping cycles in LiFSI/SL solutions with saltto-solvent molar ratios of 1:10.3, 1:3.0, and 1:2.5. Dots and solid lines denote experimental spectra and fitting curves, respectively.

molecules decreased and that of Li+-coordinated SL molecules further increased. Raman spectra corresponding to the S−N stretching mode of FSI− anions are shown in Figure 4b. When the LiFSI concentration increased, the band of FSI− anions gradually shifted to a higher wavenumber, demonstrating the intensified cation−anion association in the concentrated electrolyte via the formation of contact ion pairs (a FSI− anion coordinating to one Li+ ion) and aggregate clusters (a FSI− anion coordinating to two or more Li+ ions), as observed in many studies.9,30−33 The unique networking structure of Li+ cations and FSI− anions with Li+-coordinated SL solvents is crucial for the downward shift of the anion’s LUMO and formation of anion-derived SEI.9,31 3.6. SEI Layer on Lithium Metal. To further elucidate the mechanism of the stable interface of lithium metal in the concentrated electrolyte, SEI layers on the cycled Li foils were studied by XPS. Figure 5 shows the atomic concentration of the SEI layers on Li foils after 50 plating/stripping cycles in the different concentrations of the LiFSI/SL solutions. The C content was much higher than the Li content for the dilute

cycling. These results demonstrate that the relatively stable SEI layer is formed in 1:2.5 LiFSI/SL solution, which prevents the side reactions and suppresses the overgrowth of the SEI layer. However, even the 1:2.5 LiFSI/SL could not completely avoid the side reactions and gradual increase in cell resistance, eventually leading to a short circuit of the cell. 3.5. Solution Structure. To figure out the mechanism of significantly enhanced stability of lithium plating/stripping in the concentrated electrolyte, the solution structures of the electrolytes at various salt concentrations were studied by Raman spectroscopy. Figure 4a shows Raman spectra in the spectral range corresponding to the SO2 twist mode of SL molecules. The Raman spectra of pure SL exhibited a sharp peak of SO2 twist at 443 cm−1 related to free SL molecules.28 When the salt-to-solvent molar ratio increased to 1:3.0, the free SL-derived peak was weakened and a new broad peak appeared at around 460 cm−1. The broad peak is due to different types of SL molecules coordinating to Li+ ions such as monodentate and bridging types, according to the previous reports.13,16,18,29 As the concentration increased, the number of free SL G

DOI: 10.1021/acsami.9b05257 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

Figure 7. Charge/discharge curves for LiFePO4|Li cells using (a) 1.0 mol dm−3 LiPF6/EC:DMC (1:1 by volume) and LiFSI/SL solutions with saltto-solvent molar ratios of (b) 1:3.0 and (c) 1:2.5. The curves of 2nd, 10th, 50th, 100th, and 140th cycles are shown. (d) Discharge capacity and Coulombic efficiency versus cycle number for the cells. (e) Discharge capacity of the cells at various C rates. Charge and discharge were conducted at the same C rate at 30 °C in a voltage range of 2.5−3.8 V. The cycling tests were conducted at 0.5C rate after an initial one cycle at 0.1C rate and then four cycles at 0.2C rate.

decomposition and suppress lithium dendrite growth, contributing to longer cycle life and higher Coulombic efficiency.45,46 These results suggest that the stable lithium plating/stripping processes are achieved through the highly Li+ ion conductive and stable SEI layer containing these effective inorganic components (LiF, L3N, and Li2S2/Li2S). 3.7. Battery Performance of Lithium Metal Battery. To investigate charge/discharge behaviors of lithium metal battery cells using the LiFSI/SL solutions with salt-to-solvent molar ratios of 1:3.0 and 1:2.5, the cells composed of the LiFePO4 electrode with a high areal capacity (1.76 mAh cm−2) and Li/Cu foil (20/10 μm in thickness) were employed (i.e., a theoretical capacity ratio of LiFePO4/Li = ca. 1:2.34). This condition provides severe tests for electrolyte because it requires higher lithium metal utilization of ca. 30% during each cycle, compared with most of the other reports (below 15%).13,47 For comparison, the cells using a conventional 1.0 mol dm−3 LiPF6/EC:DMC electrolyte were also evaluated. As shown in Figure 7a−c, similar charge/discharge performance with a specific capacity of ca. 160 mAh g−1 were obtained from the second cycle of the cells with all electrolytes. However, the conventional electrolyte showed a rapid increase in overpotential with low Coulombic efficiency and the discharge capacity retention of