Low concentration electrolyte with non-solvating cosolvent enabling high-voltage lithium metal batteries.

Developing low cost, yet high-voltage electrolyte is significant to improve the energy density and practicability of lithium metal batteries (LMBs). Low concentration electrolyte has significant merits in terms of cost and viscosity; however, their poor compatibility with high-voltage LMBs hinders its applications. Here, we develop a diluted low concentration electrolyte by replacing solvating cosolvent with a non-solvating cosolvent to facilitate the interaction between BF4 - and Li+, resulting in optimized interfacial chemistry and suppressed side reaction. Thus, the high-loading Li-LiCoO2 full cells (20.4 mg cm-2) deliver outstanding cycling stability and rate performance at a cutoff voltage of 4.6 V. More impressively, a Li-LiCoO2 pouch cell achieves an energy density of more than 400 Wh kg-1 under practical conditions with thin Li (50 μm) and lean electrolyte (2.7 g Ah-1). This work provides a rational approach to design a low concentration electrolyte, which can be extended to other high voltage battery systems.

Introduction

Lithium-ion batteries (LIBs) has gradually approached their ceiling of energy density afforded by intercalation chemistry, falling behind the increasing demands for advanced portable electronics and electric vehicles (Duffner et al., 2021). In this context, lithium metal batteries (LMBs) outperform other candidates in terms of energy density because metallic Li anode possesses the highest theoretical specific capacity (3860 mAh g−1) and lowest redox potential (−3.04 V vs. the standard hydrogen electrode) (Wu et al., 2020b; Xu et al., 2021a; Zhong et al., 2021). Besides, further increasing the working voltage of LMBs is also critical for higher energy density. We note here that layer-structured LiCoO2 (LCO) can achieve a high capacity of 220 mAh g−1 at an escalated upper cutoff voltage of 4.6 V (Wang et al., 2020a; Li et al., 2021). Along with its high Li+/electron conductivity, theoretical density, and compressed electrode density, high-voltage LCO still possesses many competitive advantages in the family of cathode material (Lyu et al., 2021; Chikkannanavar et al., 2014). However, accompanied decrease in cyclability and thermal abuse tolerance renders these approaches challenging, due to the well-known problems of lithium metal anode (LMA), such as inhomogeneous Li+ deposition kinetics and unstable solid-electrolyte interphase (SEI) (Lin et al., 2017; Xu et al., 2021b), as well as the structure collapse and detrimental cathode/electrolyte interface reactions of high-voltage LCO (Lim et al., 2019; Kalluri et al., 2017).

Developing multifunctional electrolytes is an effective way to overcome challenges of high-voltage LMBs (Zhang et al., 2020; Yamada et al., 2019). Typically, various high-efficiency additives ware added to the conventional carbonate electrolytes with a Li salt concentration of 1 M, such as fluoroethylene carbonate (FEC), LiNO3, and nitro-C60 (Zhang et al., 2017; Jiang et al., 2019, 2020a). Besides, high oxidation tolerance solvent, such as sulfone- and nitrile-based electrolytes were used to substitute carbonate solvents (Alvarado et al., 2018;Aspern et al., 2019; Hu et al., 2020). Most strikingly, the strategy of “solvent-in-salt” was employed to broaden the electrochemical window of electrolytes, including various highly concentrated electrolytes (HCE) and diluted high concentration electrolytes (DHCE) with an increased Li salt concentration of 2–10 M (Suo et al., 2018; Chen et al., 2018; Piao et al., 2020).

Nevertheless, recent studies raise valuable reconsideration on the vital role of high salt concentration for energy storage devices. The so called “low concentration electrolyte (LCE)” has received extensive attention because of its huge advantages in cost, compared with the conventional concentration electrolyte (CCE) and HCE (Hu and Lu, 2020). Hu's group put forward the concept of LCE and applied it in sodium-ion batteries (SIBs) (Li et al., 2020). Wu et al. reported that the reduction of salt concentrations to 0.1 M suppressed the polysulfide shuttle effects in lithium-sulfur (Li-S) batteries (Wu et al., 2020a). Xiang and co-workers designed a dual-salt LCE to construct a robust SEI in Li-LiFePO4 full cells (Zheng et al., 2020). However, the use of low concentration electrolytes for high-voltage LMBs still faces great challenges due to the continuous decomposition of free solvents and the instability of the electrolyte/electrode interface. The solvation structure of the electrolyte plays an essential role in determining these properties. Specifically, the amount of free solvents can be effectively reduced by the use of strong coordinated lithium salts (LiFSI, LiDFOB, LiBF4, etc.) or the addition of non-solvating cosolvents (Ding et al., 2021; Weber et al., 2019; Louli et al., 2020), while the introduction of fluorinated reagents can form a stable LiF protective layer (Yan et al., 2020).

It is widely accepted from recent studies that the anion-dominated solvation structure is the key to obtain electrochemical stability of HCEs and DHCEs (Jiang et al., 2021). In light of this, it is possible to develop a low concentrated electrolyte with excellent electrochemically compatibility by replacing solvating cosolvent with an equal volume of non-solvating cosolvent. So, the deficient coordination between Li+ and solvating cosolvent must facilitate the anion-cation interaction and thus promote the generation of aggregates (AGGs, an anion coordinating to two or more Li+). Based on these insights, we design a diluted low concentration electrolyte (DLCE) that enables stable cycling of high-voltage LMBs. In detail, replacing half of the solvents in the typical dult-salt LCE (0.3 M LiDFOB +0.2 M LiBF4 in DEC/FEC (7: 3 by volume)) with non-solvating fluorobenzene (FB) obtains the DLCE (0.3 M LiDFOB +0.2 M LiBF4 in DEC/FEC/FB (3.5: 1.5: 5 by volume)) with significantly improved electrochemical performance. As illustrated in Figure 1A, the addition of FB cosolvent not only improves the physical properties of LCE, but also regulates the solvation structure, which contributes to the formation of LiF-rich interface. Benefiting from these optimized properties, the high-loading Li-LiCoO2 (LCO, 20.4 mg cm−2) cell based on DLCE displays excellent cycle stability and rate performance at the cutoff voltage as high as 4.6 V. More impressively, a Li-LCO pouch cell sustains an estimated energy density over 400 Wh kg−1 under practical test conditions (50 μm Li, 2.7 g Ah−1 electrolyte).

Figure 1

Solvation-structural characterization

(A–C) (A) Solvation and interfacial structures of the LCE and DLCE for Li-LCO batteries during cycling. Raman spectra of different electrolytes in the wavenumber range of (B) 700–750 cm−1 (FEC O-C-O) and (C) 754–774 cm−1 (BF4−).

(D and E) (D) 11B NMR spectra and (E) LSV plots of various electrolytes.

Results and discussion

Solvation structure of electrolyte

FB, as a cosolvent holds the advantages of low density and low cost compared with hydrofluorinated ether (HFE) (Jiang et al., 2021). To ensure that the lithium salt can be fully dissolved, the concentration of electrolyte was selected as 0.3 M LiDFOB and 0.2 M LiBF4 in the DEC/FEC/FB system based on the results of the solubility test (Figure S1). Besides, the CCE (0.6 M LiDFOB +0.4 M LiBF4 in DEC/FEC (7: 3 by volume)) is chosen as a control electrolyte to demonstrate the effect of FB on the physicochemical property and solvation structure. Thanks to the decreased salt concentration and the low viscosity of FB, the DLCE exhibits the lowest density of 1.11 g cm−3, the lowest viscosity of 1.27 mPs·S, and outstanding wettability with contact angle of 21.5°(Figures S2 and S3). As expected, Raman spectrums indicate that FB is barely coordinated with Li+ (Figure S4) and the Li+-solvents interaction is significantly facilitated in the DLCE (Figure 1B), thereby reducing the content of free FEC (Yang et al., 2019). Besides, in all three electrolytes, the peaks of free BF4− are quite weak, and contact ion pairs (CIPs, one BF4− coordinating to one Li+, 763 cm−1) occupy dominant stage owing to the strong association strength of LiBF4 and the poor donor ability of FEC and DEC (Figure 1C). More critically, a certain amount of CIPs convert to AGGs (aggregates) after replacing half of the solvent with FB, suggesting that interaction between Li+ and BF4− becomes even stronger in the DLCE (Figure 1C) (Seo et al., 2012). These positively charged AGGs are supposed to allow more BF4− to enter the electric double layer at the solid-liquid interphase, thus enhancing decomposition kinetics of BF4− to form a LiF-rich SEI. Meanwhile, increasing the salt concentration from 0.5 M (LCE) to 1 M (CCE) has similar effects. The 11B NMR spectra can further verify this conclusion. As shown in Figure 1D, the characteristic peak of BF4− upshifts from −2.36 to −2.41 ppm with increasing salt concentration from 0.5 M to 1 M. More critically, this peak shifts to −2.52 ppm upon more AGGs in the DLCE (Zhou et al., 2011). But, the solvated structure of LiDFOB is almost unchanged in studied electrolytes based on the results of Raman and 11B NMR, which may be owing to a weaker solvation ability for LiDFOB (Figure S5). The change of coordination structures greatly determines the decomposition potential of electrolyte. The electrochemical stabilities of the DLCE at high voltages were evaluated by linear sweep voltammetry (LSV). On the high-voltage side, the LSV scans show that the electrochemical decomposition of electrolyte starts at 4.50 V in the LCE, 5.00 V in the CCE, and 5.25 V in the DLCE (Weber et al., 2019; Louli et al., 2020; Zhang et al., 2002), which should nevertheless be sufficient for most commercial high-voltage cathodes. The improved anodic stability of DLCE can be divided into three aspects; firstly, the content of free solvents is deceased. Secondly, the coordinated solvents may have a lower tendency to be oxidized. Thirdly, the FB may have more excellent anodic stability than FEC and DEC.

The cycling reversibility of LMA in different electrolytes was demonstrated by the Coulombic efficiency (CE) (Xiao et al., 2020). As shown in Figure 2A, under the test conditions of 1 mA cm−2, 1 mAh cm−2, the average CE of Li plating/stripping in the DLCE is as high as 98.3%, considerably higher than 94.2% in the LCE and 98.2% in the CCE, indicating that introducing FB improves the reversibility of Li plating/stripping. Tafel plots further demonstrate that the electrochemical reaction kinetics of Li electrode is quite fast in the DLCE, as evidenced by its highest exchange current density of 1.97 mA cm−2, thus eliminating uneven Li deposition (Figure 2B) (Jiang et al., 2020b). X-ray photoelectron spectroscopy (XPS) was conducted to analyze the composition of the SEI on LMAs. It is seen from Figure 2C that the SEI formed in the DLCE and CCE have roughly similar contents of F, Li, C, and O. The generation of F is mainly attributed to the decomposition of FEC and Li salts in electrolytes. It is worth noting that the SEI formed in the DLCE and CCE all contains less organic compounds (O contained species) and more LiF compared with the LCE, as a result of diminished solvent reduction and significant salt reduction products (Figures 2C, S6, and S7). The corresponding element distribution of Li deposition detected by energy-dispersive X-ray spectroscopy (EDX) also verifies that the SEI formed in the CCE and DLCE contains more F and fewer C elements, implying that the SEI layer derived from anions can effectively inhibit further side reactions of solvent (Figure S8). The morphology of Li deposition was in-depth studied by scanning electron microscopy (SEM) to clarify the mechanism of the performance improvement (Figures 2D–2F and S9). As expected, dense and uniform Li depositions are observed in the DLCE and CCE, and the thickness of Li film is only 15.1 μm and 18.1 μm, respectively, which is very close to the theoretical thickness of Li (1 mAh cm−2 = 4.85 μm). In sharp contrast, large amounts of Li dendrites grow in the LCE, and the thickness of Li film is 30.1 μm. These observations are in good agreement with the variation law of the CE. Besides, the high-resolution transmission electron microscopy (HR-TEM) image of deposited Li in the LCE shows that the LiF is distributed randomly in the SEI together with organic component, while the SEI generated in the CCE and DLCE displays a thin layered-structure of 18 nm and 9 nm, respectively, which is rich in LiF (Figures 2G–2I and S10). The above results suggest that the addition of FB leads to anion recruitment into the Li+ solvation sheath to form more AGGs, thus promoting the formation of a thin and LiF-rich SEI with fast kinetics, thereby significantly improving the performance of LMA.

Figure 2

Li deposition performance and morphology

(A and B) (A) CE tests and (B) Tafel plots of various electrolytes.

(C–I) (C) Quantified atomic ratios of the elements in SEI from different electrolytes by XPS. SEM images of deposited Li on Cu foila in the (D) CCE, (E) LCE, and (F) DLCE (0.5 mA cm−2 for 3 mAh cm−2). The inset is optical photographs of the deposited Li. HR-TEM images of the corresponding deposited Li in the (G) CCE, (H) LCE, and (I) DLCE.

Electrochemical performance of Li-LCO cells

To demonstrate the practicability of the developed electrolyte, the cycling performance of Li-LCO cells were tested at different temperatures. Firstly, as shown in Figures 3A, the Li+ conductivity of DLCE is lower than that of the CCE/LCE in the temperature range from 40°C to −40°C. However, the cyclic voltammetry (CV) curve shows that the LCO cathode with smaller polarization in the DLCE than in the LCE, which suggests that the interfacial property rather than Li+ conductivity determines the electrochemical reaction kinetics in our studied system (Figure 3B). At the conditions of 1 C/25°C, the Li-LCO cell using DLCE can maintain 87.1% of the initial capacity after 1000 cycles, while the cell in the CCE and LCE only has capacity retention of 65.4% and 45.5%, respectively (Figures 3C and S11). The corresponding electrochemical impedance spectroscopy (EIS) plots imply that the interphase resistances (RCEI) and charge-transfer resistances (RCT) in DLCE are significantly lower than that in CCE and LCE during cycling, which further demonstrates that the addition of FB cosolvent improves the interphase stability of LCO (Figure S12). Even at the low temperature of −20°C (Figures 3D and S13), the Li-LCO cell using DLCE still holds outstanding cycling stability with a capacity retention of 71.2% after 500 cycles and an initial specific capacity as high as 158 mAh g−1. In sharp contrast, the cell using LCE only presents the capacity retention of 37.5% and an initial specific capacity of 136 mAh g−1. More importantly, the Li-LCO cell in the CCE exhibits the poorest electrochemical performance with the capacity retention of 34.8% and the initial specific capacity of 118 mAh g−1 at low temperature, indicating that the physical properties of the electrolyte can significantly affect the low-temperature performance of LMBs.

Figure 3

Li-LCO cells cycling at different temperature

(A) Li+ conductivity of various electrolytes under different temperatures.

(B–D) (B) CV curves of the initial cycles for Li-LCO cell under different electrolytes (3–4.3 V, 0.1 mV s−1). Long-term cycling performance of Li-LCO cells in different electrolytes (C) at 1 C/25°C and (D) 0.5 C/−20°C (1 C = 170 mA g−1).

Increasing cathode loading and cutoff voltage are two effective ways to enhance the energy density of LMBs. Therefore, Li-LCO cells using different electrolytes were assembled using commercial LCO cathode (loading: 20.4 mg cm−2) with a cutoff voltage of 4.6 V. Benefiting from the modified solvation structure and interfacial chemistry, the Li-LCO cells show excellent rate performance in both of the DLCE and CCE with a specific capacity of 65 mAh g−1 even at the high current density of 5 mA cm−2 (Figures 4A and S14). In comparison, the Li-LCO using LCE only delivers a specific capacity of 30 mAh g−1. Besides, the addition of non-solvating cosolvent effectively suppresses the occurrence of side reactions at the interface, thereby improving the long-term cycling stability of LMBs. Therefore, the Li-LCO cell using DLCE can operate steady over 120 cycles with the capacity retention of 85.6% at the current density of 1 mA cm−2, while the Li-LCO cells using CCE and LCE display rapid capacity fading after 60 cycles and 30 cycles, respectively (Figures 4B and S15). In addition, fluorine nuclear magnetic resonance (19F NMR) spectra were used to evaluate the consumption of LiDFOB after cycled in high-loading Li-LCO cells with different electrolytes (Figure S16). After 100 cycles, LiDFOB in the LCE is almost exhausted, but it can still be detected in DLCE and CCE, showing the unique solvated structure and interface chemistry of DLCE can effectively reduce the decomposition of LiDFOB and the subsequent gas production issues. Finally, a Li-LCO pouch cell was assembled under practical conditions, including high-loading cathode (20.4 mg cm−2), high cutoff voltage (4.6 V), thin Li foil (50 μm), and lean electrolyte (2.7 g Ah−1). Based on these parameters, the energy density of Li-LCO is calculated to be 400 Wh kg−1 (Figure 4C and Table S1). The Li-LCO cells using DLCE simultaneously satisfy the conditions of high cathode loading and high cutoff voltage, and show superior performance to most previous studies (Figure 4D and Table S2) (Qin et al., 2019; Xu et al., 2019; Chen et al., 2019; Liu et al., 2019, 2020; Ren et al., 2020; Huang et al., 2021; Wang et al., 2019, 2020b; Zhang et al., 2019).

Figure 4

High-voltage Li-LCO full cells at practical conditions

(A and B) (A) Rate performance and (B) long-term cycling performance of high-loading Li-LCO cells under different electrolytes (Cathode loading: 20.4 mg cm−2, 3–4.6 V).

(C) The charge-discharge curves of Li-LCO pouch cell using DLCE under practical conditions (Cathode loading: 20.4 mg cm−2, 3–4.6 V, 50 μm Li, 2.7 g Ah−1 electrolyte).

(D) The performance comparison of Li-LCO cells.

Characterization of LCO after cycled

Finally, we characterized the LCO cathode and their surface layers after cycled to reveal the origin of the excellent electrochemical performance of high-voltage Li-LCO cells in the DLCE. As shown in Figures 5A, 5B, and S17, a thin (6 nm) and homogeneous CEI is observed on the surface of LCO after cycled in the DLCE, while the thickness of CEI formed in the LCE is 48 nm, as a result of the continuous oxidation decomposition of free solvents (Ren et al., 2020). This structural difference in CEI will significantly affect the cycling stability of LCO under high voltage. In detail, a relatively small (003) peak shift is observed in LCO cycled in DLCE at a high voltage of 4.6 V, in contrast with the dramatic (003) peak shift in LCO cycled in LCE, demonstrating suppressed O3 to H1–3 phase transition in the DLCE (Figures 5C and S18) (Zhang et al., 2019; Yoon et al., 2020). Besides, the CEI formed in the DLCE also contains more F and fewer C elements (Figure 5D). The increase in F content stems from the decomposition of FB and salt to produce additional LiF, while the decrease in C content further proves that the decomposition of the solvent in the DLCE is inhibited (Figures 5E, 5F, and S19) (Zhou et al., 2011).

Figure 5

Characterization of cycled LCO cathode

(A–F) TEM images of cycled LCO in (A) LCE and (B) DLCE. (C) XRD patterns of different LCO. (D) Quantified atomic ratios of the elements in CEI from different electrolytes by XPS. F 1s spectra of the cycled LCO in (E) LCE and (F) DLCE (Cycling conditions: 20.4 mg cm−2 cathode, 3–4.6 V, 80 cycles).

Conclusions

In conclusion, we have demonstrated that replacing solvating cosolvent in LCE with an equal volume of non-solvating cosolvent is highly effective to enhance the interaction between BF4− and Li+, thus inducing a BF4−-derived LiF-rich SEI in the diluted low concentration electrolyte. Besides, the side reaction of the electrolyte can be significantly inhibited and LCO is well protected. Thanks to these modified solvation structures and improved interfacial properties, a high loading cell using DLCE presents outstanding cycling stability even under a cutoff voltage of 4.6 V. Besides, the Li-LCO pouch cell exhibits a high specific energy of more than 400 Wh kg−1. This work not only provides a feasible way for the practical application of low concentration electrolytes but also develops a new view for the design of advanced electrolytes.

Limitations of the study

In this work, we developed a diluted low concentration electrolyte for high-voltage lithium metal batteries. Future studies can explore the feasibility of this new insight of electrolytes in other battery systems, such as lithium-ion batteries, sodium-ion batteries, and aqueous batteries.

STAR★Methods

Key resources table

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Jia Xie (xiejia@hust.edu.cn).

Materials availability

This study did not generate new unique reagents.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Other
Li foil	China Energy Lithium Co., Ltd.	CAS#7439-93-2
Fluorobenzene	Aladdin	CAS#462-06-6
LiCoO2 power	Shenzhen Kejing Star Technology Co., Ltd.	CAS# 12190-79-3
Lithium difluoro(oxalato)borate	Dodo Chem Co., Ltd.	CAS#409071-16-5
lithium tetrafluoroborate	Dodo Chem Co., Ltd.	CAS#14283-07-9
fluoroethylene carbonate	Dodo Chem Co., Ltd.	CAS#114435-02-8
diethyl carbonate	Dodo Chem Co., Ltd.	CAS#105-58-8
Commercial LiCoO2 electrode	Zhuhai COSMX power battery Co., Ltd.	http://www.cosmx.com/html/about/about/