Unlocking the Polarization and Reversibility Limitations for Stable Low‐Temperature Lithium Metal Anodes

The subzero‐temperature service of lithium (Li) metal anode has been enormously restricted by a large working polarization and a poor reversibility. In this contribution, the overpotential attributions during low‐temperature Li electroplating are deconvolved, and the interplay among the dominating kinetic overpotential, the dynamic solid electrolyte interphase (SEI) chemistry, and the corresponding cycling reversibility of Li metal is established. Specifically, by employing a localized highly concentrated electrolyte as a model system, it is disclosed that ionic concentration gradient plays a predominate role in polarizing the cathodic Li electroplating process at subzero working temperature. Inspired by this, a decoupling electrolyte design strategy is presented to synchronously tame the kinetic polarization and build a dynamically stable anion‐derived SEI, thus boosting a remarkably enhanced Coulombic efficiency of Li with a depressed cell overpotential and a more than three times longer lifespan in practical Li | LiNi0.5Co0.2Mn0.3O2 cells at −20 °C. Herein, the essence to affecting the polarization and reversibility of low‐temperature working Li metal anode is uncovered, affording critical design principles to facilitate a stable dynamic interface for the high‐efficiency cycling of practical Li metal batteries at subzero temperatures.


Introduction
The building of an environmentally friendly and sustainable society highly demands the exploitation and utilization of renewable energy. [1] Electrochemical power sources are critical media for the storage and conversion of these intermittent renewable energy into practical use. [2] More recently, the exponentially expanded market of electric vehicles has aroused ever-increasing interests on the so-called "heart" of the electric vehicle, power batteries. [3] However, the mainstream lithium (Li) ion batteries based on an intercalation chemistry cannot persistently meet customers' demand on energy density, which largely hinders the process toward an electrified modern society. [4] In response, lithium metal batteries based on a high-capacity and lowequilibrium-potential Li metal anode have been substantially revisited to pursue the energy density upper limit of nextgeneration batteries. [5,6] Unfortunately, the practical performance of working Li metal anode is inferior, featured with a low Coulombic efficiency (CE) and a short lifespan, which is to a large extent originated from its unstable electrochemical interface. [7,8] Although graphite anode (in traditional lithium ion batteries) and Li metal anode share the same concept of solid electrolyte interphase (SEI), the SEI on working graphite anode is fairly stable both chemically and mechanically after initial formation cycles. [9] In contrast, suffering from the high reactivity and huge volume fluctuation of Li, the composition and structure of SEI on working Li anode become more complex and dynamic, undergoing persistent crack and repair processes during the whole cell service. [10,11] DOI: 10.1002/sstr.202200400 The subzero-temperature service of lithium (Li) metal anode has been enormously restricted by a large working polarization and a poor reversibility. In this contribution, the overpotential attributions during low-temperature Li electroplating are deconvolved, and the interplay among the dominating kinetic overpotential, the dynamic solid electrolyte interphase (SEI) chemistry, and the corresponding cycling reversibility of Li metal is established. Specifically, by employing a localized highly concentrated electrolyte as a model system, it is disclosed that ionic concentration gradient plays a predominate role in polarizing the cathodic Li electroplating process at subzero working temperature. Inspired by this, a decoupling electrolyte design strategy is presented to synchronously tame the kinetic polarization and build a dynamically stable anion-derived SEI, thus boosting a remarkably enhanced Coulombic efficiency of Li with a depressed cell overpotential and a more than three times longer lifespan in practical Li | LiNi 0.5 Co 0.2 Mn 0.3 O 2 cells at À20°C. Herein, the essence to affecting the polarization and reversibility of low-temperature working Li metal anode is uncovered, affording critical design principles to facilitate a stable dynamic interface for the high-efficiency cycling of practical Li metal batteries at subzero temperatures.
profoundly reduces the amount of electrically isolated dead Li. [20] The joint contributions from these two aspects benefit for a large enhancement on the cycling reversibility of working Li metal anode. [21] Although these emerging advances in electrolyte design explicitly improve the room-temperature performance of Li metal anode to a brand new level, the working batteries are practically subject to varying temperature conditions. [22] Deeply understanding the temperature-dependent Li deposition/dissolution behavior and unrevealing the critical determining factors of the temperature-dependent Li utilization efficiency are of fundamental significance. In this context, some recent investigations already depicted that raising the working temperature within a reasonable range could promote the performance demonstration of Li metal anode in virtue of both interfacial [23] and kinetic [24,25] benefits. In stark contrast, the subzero temperature performance of Li metal anode is far from satisfactory, the major bottleneck of which lies in a large polarization and a poor reversibility upon the reduction of working temperature ( Figure 1). Practically, the complete picture of Li deposition involves bulk ion transport, interfacial Li-ion (Li þ ) desolvation, Li þ diffusion across SEI, and electron transfer. The kinetic rates of all these four processes are inherently temperature dependent as described by Vogel-Tammann-Fulcher (VTF) equation and Arrhenius equation, respectively, while the most resistive step (rate-determining step) still remains debatable. [26,27] Moreover, the reduction of working temperature is commonly accompanied by a remarkably degraded reversibility of Li deposition/dissolution, which is also the case for some electrolytes that demonstrate superior roomtemperature performance. [23,25] Uncovering the kinetic ratedetermining factors during low-temperature Li deposition, and clarifying the interplay between the polarization issue and reversibility issue in working Li metal anode, constitutes the primary step for building stable low-temperature Li metal batteries.
Herein, we deconvolved the overpotential attributions during Li electroplating and built the interplay among the kinetic overpotential domination, the dynamic evolution of SEI chemistry, and the corresponding Li reversibility. By employing a typical LHCE as a model system, it was unveiled that ionic concentration gradient plays an overwhelming role in polarizing the cathodic Li electroplating process at subzero temperature, which not only directly limits the power output of working batteries but also severely degrades the Li deposition uniformity on account of the limited anion decomposition for dynamic SEI formation. Inspired by the results, the design of low-temperature electrolyte was rationally decoupled to simultaneously tame the working overpotential and provide sustainable coordinated anion supply for the dynamic formation of SEI. Consequently, the CE of Li deposition/dissolution at À20°C was evidently enhanced from 95.93% to 98.40% at 1.0 mA cm À2 with a decreased kinetic overpotential, and a more than three times longer lifespan was further demonstrated in practical Li | LiNi 0.5 Co 0.2 Mn 0.3 O 2 (NCM523) cells.

Low-Temperature Li Deposition Behavior in Model Localized Highly Concentrated Electrolyte
A typical LHCE, 1.0 M lithium bis(fluorosulfonyl)imide (LiFSI) dissolved in dimethyl ether (DME)/1,1,2,2-tetrafluoroethyl-2,2,3,3-tetrafluoropropyl ether (TTE) (1/2 by vol.) was developed as a model electrolyte in this work (labeled as baseline LHCE) because of its superior all-round properties as widely documented in literature, wherein the low-temperature performance still remains to be thoroughly examined. [28] First, the Li deposition behavior at 25 and À20°C was systematically compared in Li | Cu cells. As shown in Figure 2a,b, the average CEs obtained at 25 and À20°C were comparable when the current density was as low as 0.2 mA cm À2 , whereas the discrepancy in CEs became intensive after the current density was lifted to 1.0 mA cm À2 , a common rate for practical batteries. [29] Compared to the Figure 1. Schematic illustration about the major bottlenecks for stable low-temperature lithium metal anode, namely, the polarization issue, the reversibility issue, and how these two factors interplay. relatively reversible Li deposition/dissolution at 25°C (CE of 99.12%), the CE at À20°C was pronouncedly deteriorated to 95.93%, which is unacceptable for long-life rechargeable batteries. [30] The Li deposition morphologies at different temperature and current densities were further recorded in Figure 2c,d to figure out the origins of the diverse CEs observed. At a small current density of 0.2 mA cm À2 , the Li deposits presented a favorable large granule-shaped morphology under both 25 and À20°C (Figure 2c), which agreed well with their roughly consistent CEs acquired in Li | Cu cells ( Figure 2a). With increasing the current density to 1.0 mA cm À2 , even though the granule-like Li deposition morphology can be retained at 25°C, it turned into a highly dendritic structure with a high-aspect-ratio needle-like morphology when the working temperature was reduced to À20°C ( Figure 2d). The dramatic transition on Li deposition behavior consequently leads to an obviously degraded reversibility as shown in Figure 2b. As has been widely acknowledged in research community, Li deposition behavior is primarily governed by SEI chemistry, [31] while the formation process of SEI is directly tailored by the solvation structure of Li þ . [32] Based on this perspective, the temperature-dependent Li þ solvation structures were intuitively studied through molecular dynamics (MD) simulation. The simulation results in Figure 3a,b suggested that lowering the temperature reinforces the anion-cation interaction intensity in electrolyte, which can be verified by the stronger peak corresponding to the LiÀO interaction between Li þ and FSI À in the primary solvation shell of Li þ and hence an enhanced average coordination number of FSI À (from 2.61 to 2.76) after reducing the temperature from 25 to À20°C. Furthermore, the microscopic statistics of specific Li þ coordination states revealed that the percentage of Li þ individually solvated by DME molecule (3-0-0) was significantly declined from 5.0% to 0.8% (Figure 3c,d). According to the established theory, a stronger anion-cation coordination will encourage more anion decomposition to facilitate the formation of an anion-derived SEI. [15,32] To confirm this point, electrochemical impedance spectroscopy (EIS) characterization was conducted to probe the interfacial resistance of the symmetric Li cells that had been subject to long-term rest at different temperatures for static SEI formation ( Figure S1, Supporting Information). The cell shelved at À20°C displayed a lower resistance than that shelved at 25°C, while both SEIs had a similar ion transport activation energy (E a ) due to the inorganic-rich nature. The above observations evidently underscored that the change of static Li þ solvation structure should not be the major reason to dictating the transformation of Li deposition manner at subzero temperature, but instead, it was more likely resulted from the dynamic-related interfacial behavior of practical working Li anode.

Identifying the Dynamic Interfacial Limitation of Low-Temperature Li Electroplating
Driven by the above deduction, a time-resolved transient relaxation measurement was employed to analyze the dynamic overpotential attributions during low-termperature Li electroplating (Figure 4), where a short relaxation step (for 10.0 s) was intermittently introduced as a probe in a consecutive galvanostatic Li electroplating process to differentiate the concentration overpotential (η conc ) and iR drop (Figure 4a,b). Critically, the iR drop was defined as the sudden potential drop between the final potential value before the relaxation step and the relaxed potential value www.advancedsciencenews.com www.small-structures.com after a specified rest time. The rest time was rationally selected to be five times of the time constant (τ) regarding interfacial charge transfer process ( Figure S2, Supporting Information) to ensure that the interfacial reaction overpotential (η reac ) has been completely relaxed, so that it can be merged into iR drop. [33] Then, the total iR drop can be accordingly decoupled into ohmic overpotential (η ohm ) and η reac based on an auxiliary EIS test given that η ohm is generally unaffected by the depth of Li electroplating ( Figure S2, Supporting Information). Consequently, the overall Li electroplating overpotentials can be clearly deconvolved in Figure 4c,d. The results seemed contrary to the previous cognition that interfacial process (Li þ transport across SEI or Li þ desolvation) is the kinetic rate-determining step for low-temperature anode; [34,35] it was discovered here that, for the Li deposition at À20°C, η conc contributed the most in total overpotentials under both 0.2 and 1.0 mA cm À2 (Figure 4c,d). Noteworthily, the overpotential domination shifted to η reac when elevating the Li deposition temperature to 25°C ( Figure S3, Supporting Information). This can be partially understood from the remarkable change in ionic conductivity when the temperature was lifted from À20°C (0.21 mS cm À1 ) to 25°C (0.70 mS cm À1 ) ( Figure S4, Supporting Information). Additionally, it was discovered that η rea decreased upon the increase of Li deposition capacity, irrespective of the current density (Figure 4c,d) and temperature ( Figure S3, Supporting Information). The major reason for this is the gradually enlarged active surface area upon the increase of Li deposition depth, which reduces the resistance and overpotential for electrochemical reactions at the interface. [36] COMSOL simulation was further performed to theoretically scrutinize the ionic concentration evolution at low-temperature Li electroplating (Figure 4e,f ). Upon increasing the current density from 0.2 to 1.0 mA cm À2 , the consumption and supplement of Li þ charge carrier became prominently out-of-balance, which resulted in nearly a half reduction in ionic concentration at anode surface after successive Li electroplating (Figure 4f ). On one hand, the exhaustion of Li þ concentration potentially initiates the growth of Li dendrites as stated by Sandy's time model. [37] On the other hand, the evolution of local Li þ concentration should strongly alter the original Li þ Àanion interaction, and the gradually weakened Li þ Àanion interaction under the dynamically decreased Li þ concentration is expected to immensely affect the dynamic SEI formation process. To confirm the latter point, the dynamic consumption of anion and solvent for progressive SEI formation was quantitatively probed via nuclear magnetic resonance (NMR) technique. [38] An SEI-free and structurally stable Li 7 Ti 5 O 12 electrode [39] was intentionally utilized to provide Li source for Li electroplating on Cu in this study to necessarily exclude the electrolyte consumption from counter electrode (Figure 5a). The electrolytes after ten cycles at 0.2 and 1.0 mA cm À2 in Cu | Li 7 Ti 5 O 12 cells ( Figure S5, Supporting Information) were characterized under 19 F and 1 H NMR (Figure 5b), respectively. By tracing the peak intensity evolution of FSI À ( 19 F spectrum) and DME ( 1 H spectrum) relative to those of internal reference reagent, the dynamic consumption of the salt and solvent can be readily quantified. As calculated in Figure 5c, FSI À consumption (9.5%) was more notable than that of DME (7.2%) for the SEI construction at 0.2 mA cm À2 , while DME was selectively consumed at 1.0 mA cm À2 since the loss of DME (11.4%) was prominently higher than that of FSI À (3.8%). The SEI chemistry after Li deposition at 0.2 and 1.0 mA cm À2 was further detected via X-ray photoelectronic spectroscopy (XPS). As displayed in Figure 5d,e, SEI formed at  0.2 mA cm À2 was enriched with the decomposition products of FSI À , including LiF, Li 2 O, Li 3 N, and Li 2 S. However, the amount of these anion-derived SEI species was evidently reduced when the plating current density was increased to 1.0 mA cm À2 . The above results vividly indicated that the dynamic electrolyte consumption for SEI-forming reactions and the resulting SEI chemistry on Li metal can be profoundly modulated by the real-time Li þ Àanion interaction at the interface, and the weakened Li þ Àanion interaction at a larger Li þ concentration gradient is detrimental for an anion-derived SEI formation.
To further confirm either the Sandy's time model or the dynamic interface chemistry model is the determining factor to dictate the Li deposition and cell reversibility, we designed a control experiment by adjusting the volume proportion of DME to TTE ( Figure S6, Supporting Information). When raising the volume ratio of DME/TTE from 1/2 to 2/1, the ionic conductivity at À20°C can be elevated by 38.1% to 0.29 mS cm À1 benefited from the enhanced salt dissociation ( Figure S6a,b, Supporting Information), conducive to a significantly relieved polarization of Li electroplating/stripping ( Figure S6c, Supporting Information). However, such a benefit in accelerating Li þ diffusion did not bring about any expected enhancement on Li reversibility. On the contrary, the average CE was even dropped from approaching 96% to 95.31% ( Figure S6c, Supporting Information), which can be ascribed to the insufficient Li þ Àanion interaction for preferrable anion reduction once reducing the effective Li þ concentration (the concentration of Li þ in DME) in bulk electrolyte. [40] Here, we can safely conclude that the real-time Li þ concentration-dependent Li þ Àanion interaction and the corresponding dynamic SEI formation pathway is the critical factor in determining the Li deposition behavior and hence the cell reversibility at subzero working temperatures (Figure 5e). When the coordinated anions are sufficiently supplied, the cracked SEI can be rapidly repaired by the decomposition of anion. Thus, the granular Li deposition manner can be retained. Nevertheless, when a large ionic concentration gradient is generated, the Li þ Àanion interaction is weakened and the quantity of Figure 5. Determination of the anion and solvent consumption by NMR technique, including a) a schematic illustration of the characterization procedure, b) 19 F and 1 H NMR spectra of the uncycled and cycled electrolytes, and c) the calculated consumption ratio of FSI À and DME. And SEI chemistry analysis of Li deposits at À20°C under XPS, including d) F 1s, O 1s, N 1s, and S 2p spectra, and e) the corresponding atomic concentration. The data were acquired after 120 s sputtering. f ) A schematic illustration of dynamic SEI repair process in the case of sufficient and limited coordinated anion supply, which leads to granular and dendritic Li deposition, respectively.
www.advancedsciencenews.com www.small-structures.com coordinated anions is limited to sustain the persistent repair of SEI. In this case, the anion-derived SEI chemistry gradually turns into a solvent-derived chemistry, which adversely transforms the Li deposition morphology into a needle-like dendritic pattern.

Rational Design of Electrolyte to Regulate the Low-Temperature Working Li Metal Anode
According to the above discussion, the design strategy of low-temperature electrolyte concurrently targeting a low overpotential and a high Li reversibility was further explored (Figure 6a). Intrinsically, the reduction of concentration overpotential requires a lower Li þ to DME ratio to facilitate Li salt dissociation. Oppositely, sustaining a strong Li þ -anion interaction for favorable anion-derived SEI formation demands a sufficiently high Li þ to DME ratio. To surmount this trade-off dilemma, we proposed to separate the functions of rapid Li þ transport in bulk electrolyte and sustained coordinated anion supply for dynamic SEI formation during electrolyte design. In detail, we slightly decreased the concentration of LiFSI salt to 0.8 M in DME/TTE (1/2, by vol.), which promoted a higher ionic conductvity (0.31 mS cm À1 ) and therefore a significantly decreased cell polarization ( Figure S7, Supporting Information). Additionally, a trace amount (0.05 M) of strongly coordinated lithium nitrate (LiNO 3 ) [32] was introduced to compensate the loss of Li þ Àanion aggregation (denoted as advanced electrolyte). It is noteworthy that an enhanced LiNO 3 amount (0.10 M) will enormously compromise the low-temperature ionic conductivity of electrolyte, even lower than that of baseline LHCE ( Figure S7 and S8, Supporting Information). This is in great accordance with the poor low-temperature performance of conventional LiNO 3 -containing electrolytes observed in literature. [23,25] Distinctively, the advanced electrolyte developed in this work largely enhanced the average CE of Li | Cu cells to 98.40% at 1.0 mA cm À2 with a mitigated Li electroplating overpotential (Figure 6b and S9, Supporting Information). Titration gas chromatography (TGC) analysis was further employed to dissect the irreversibility origins of the cycled cells (Figure 6c). It was manifested that the dead Li 0 amount generated in the advanced electrolyte was merely one fifth of that formed in baseline LHCE, which is closely correlated with their distinct Li deposition  morphologies, i.e., the dendritic morphology in baseline LHCE becoming uniform and sphere-shaped when the advanced electrolyte was utilized (Figure 6c). The spherical Li morphology is a featured Li deposition mode under LiNO 3 -derived SEI chemistry ( Figure S10, Supporting Information). [41] Besides, it was shown that the SEI-Li þ capacity loss was comparable in both electrolytes, which can be explained from the fact that although the surface area of Li can be decreased under the spherical morphology, the highly reactive nature of LiNO 3 potentially aggravates the Li consumption in unit surface area. The joint contribution from these two factors finally leads to the negligible variation in SEI-Li þ capacity. Practical Li | NCM523 cells with a limited Li source (50 μm), a high-loading cathode (%3.3 mAh cm À2 ), and a lean electrolyte amount (7.0 μL mAh À1 ) were further assembled for low-temperature cycling (Figure 7). A longer activation period in initial cycling is required for low-temperature cells possibly due to the sluggish kinetics of both electrode wetting and SEI/CEI buildup. At a low rate of 0.1 C, the cell with baseline LHCE can be cycled for 60 cycles with regard to 80% capacity retention (Figure 7a). In sharp comparison, the advanced electrolyte largely reinforced the cycling stability of Li | NCM523 cell, and negligible capacity decay was observed within 200 cycles (Figure 7a). Further increasing the current rate to 0.25 C deteriorated the performance of both cells (Figure 7b). In spite of this, compared to the elusive short service life of the control cell (only 30 cycles), the cell with the advanced electrolyte still delivered a nearly three times longer lifespan (Figure 7b). Such a capacity decay can be straightforwardly reflected in voltage curves (Figure 7c,d), where the polarization augment was identified as the root of the persistent capacity loss of the working cells.  Intriguingly, if the failed cell at À20°C was resubject to roomtemperature cycling, such a large overpotential can be eliminated and the capacity loss was almost totally recovered ( Figure S11, Supporting Information). From this nonpermanent capacity loss phenomenon, it can be inferred that the ever-degrading electrolyte wetting inside the progressively accumulated inactive Li layer should be the origin of the polarization augment of low-temperature Li | NCM523 cells. Such a porous layer consisting of dead Li and cracked SEI can be reinfiltrated by electrolyte once the temperature was elevated, contributing to an alleviation of large polarization and a recovery of cell capacity. This implied that the working polarization of low-temperature batteries is even more sensitive to the inactive Li generation, which is ascribed to the change in physical properties (viscosity, ionic conductivity, etc.) of the electrolyte.

Conclusion
To conclude, this work carefully investigated the overpotential attributions of low-temperature Li electroplating and established the correlation among the overpotential domination, the dynamic evolution of SEI chemistry, and the corresponding Li utilization efficiency by employing LHCE as a model platform. Distinct from the common wisdom, we disclosed that concentration overpotential dominated the overall Li electroplating overpotentials at subzero working temperature, which on one hand limits the kinetic performance, on the other hand degrades the Li deposition uniformity by limiting anion decomposition during dynamic SEI formation. By intricately designing the electrolyte chemistry, a reduction in working overpotential and a sustained supply of coordinated anions for dynamic SEI formation can be simultaneously fulfilled. Consequently, the CE of Li deposition/dissolution at À20°C was evidently enhanced from 95.93% to 98.40%, and a more than three times longer lifespan was further achieved in practical Li | NCM523 cells. This work unveils the actual factor controlling the polarization and reversibility of working Li metal anode in low-temperature operation, affording fresh design principles for the stable and highefficiency cycling of practical Li metal batteries at subzero temperatures.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.