Air-exposed lithium metal as a highly stable anode for low-temperature energy storage applications

The demand for cryogenic applications has resulted in higher requirements for the low-temperature performance of energy storage systems. Lithium-metal batteries are the most promising energy storage systems. Lithium-metal anodes have the merits of high capacity and low potential. However, at low temperatures, especially sub-zero, the formation of lithium dendrites seriously hinders their applications. Herein, distinct from the traditional strategies of separating lithium metal from oxygen substances, we propose a new strategy to suppress dendrites by exposing lithium metal to air for short periods to generate a controlled oxidative protective layer in situ that is compact, homogeneous and mainly composed of Li 3 N, Li 2 O, LiOH and Li 2 CO 3 . Symmetrical and full cells are assembled. The air-pretreated Li metal symmetrical cell exhibits an excellent lifespan of up to 4500 h (1 mA cm -2 ) at 30 °C and also shows a smaller voltage polarization of 20 mV at 1.0 mA cm -2 at -20 °C. Importantly, the full cell using the air-pretreated Li metal as an anode and NCM811 as a cathode can charge-discharge normally at -20 and -40 °C. This work provides an efficient and facile approach for developing superior lithium-metal batteries for future utilization at a wide range of temperatures.


INTRODUCTION
Lithium-metal batteries (LMBs) are considered to be the most promising candidates for next-generation electrochemical energy storage technology, attributed to their merits, which include the high capacity (3860 mAh g -1 ) and low potential (-3.04 V vs. standard hydrogen electrode) of the lithium-metal anode. Uncontrollable dendrite growth, significant volume changes during cycling and repeated formation of the solid electrolyte interface (SEI) are the bottlenecks of LMBs. To address these issues, many efforts have been devoted to stabilizing the SEI interface and suppressing dendrite growth, such as surface protection layers [1] , three-dimensional hosts [2] , artificial SEIs [3] and electrolyte engineering [4] . The majority of these efforts have been dedicated to room-temperature conditions [5][6][7][8] . To advance the practical application of LMBs, their performance at low temperatures has attracted more attention as energy storage systems are expected to operate or store under sub-zero environmental conditions. However, the instinct problems of dendrite formation and growth will be aggravated certainly at low temperatures, leading to the large formation of dead Li, safety issues and short lifespans [9] . On this basis, a more efficient strategy should be introduced to suppress dendrite formation and growth at low temperatures. Regarding highly efficient dendritesuppressing strategies at room temperature, the construction of surface protection layers has been studied extensively. Powders, fibers and films have been attached to the surface of Li metal by rolling, sputtering or deposition [10][11][12] . However, the interfaces between Li metal and protection layers or SEI can make the mechanical durability weak [13] . Importantly, the cost will limit the practical applications of the above strategies and extra materials and procedures will be introduced and added to construct a surface protection layer/artificial SEI. Therefore, there is a strong demand for facile and low-cost technology to construct robust layers/SEIs to suppress dendrites at low temperatures.
The main components of a SEI are lithium-based compounds, such as LiF, Li 2 CO 3 , LiOH, Li 2 O and Li 3 N [14] . Typically, the SEI compositions are highly dependent on the additives in the electrolyte, such as LiNO 3 , fluoroethylene carbonate, H 2 O and CO 2 , which are added to regulate metallic Li deposition [15] . Nevertheless, these additives are continuously consumed and exhausted rapidly because their solubility in the widely applied carbonate electrolytes is low [16] . For low-temperature applications, inspired by the merits of the surface protection layer and artificial SEI, we exposed Li metal directly to air and an air-pretreated layer (APL) formed on the surface. Li metal can directly/indirectly react with N 2 , O 2 , H 2 O and CO 2 and form lithium-based compounds (Li 3 N, Li 2 O, LiOH and Li 2 CO 3 ), which are the main components of the SEI. These in situ formed compounds guarantee the strong adhesion between Li metal and the APL, as well as the interactions between the components in the APL. Among the formed compounds, Li 3 N can function as a Li + conductor (10 -3 S cm -1 ) [17,18] , the density functional theory-calculated Young's moduli of Li 2 CO 3 is 36.2 GPa [19] , which can help to avoid the damage to the APL and Li 2 O buffers the direct contact of unstable Li 2 CO 3 from Li metal during cycling [20] . Finally, an in situ stable dense passivation protection layer formed on the Li metal surface and the lithium-rich compounds that exist in the layer have a synergistic effect blocking the direct contact of naked Li and the electrolyte and preventing electron tunneling without sacrificing extra electrolyte. Importantly, the extremely low cost and facile merits of this strategy are particularly attractive for practical applications.
The performance of LMBs must be sensitive to the exposure time in air because the thickness of the protective layer increases with exposure time; thus, the thickness of the APL should be a balance of resistivity and conductivity. Thus, the main problem is controlling the exposure time. Li metal is treated in air with a humidity of ~40% for 10 min, 30 min, 1 h and 2 h, respectively. We then evaluated their electrochemical performance at both room (30 °C) and low temperatures (-20 and -40 °C).

RESULTS AND DISCUSSION
As depicted in Figure 1A, when exposed to air, the highly active Li metal reacts with N 2 , O 2 , H 2 O and CO 2 directly/indirectly and then the APL forms on the surface of the Li metal [21] . Figure 1B shows the X-ray diffraction (XRD) for pure Li, Li-10 min, Li-30 min, Li-1 h and Li-2 h. The XRD peaks 36.19°, 51.97° and 87.89° of pure Li are corresponding to the (110), (200) and (220) facets (JCPDS No. 15-0401), respectively. After exposing Li metal to the air, with increasing exposure time, the peaks of pure Li became weakened gradually, while the peaks of Li compounds became stronger and stronger. The weak peak of the Li 2 O (111) facet at 33.61° was observed in Li-30 min, Li-1 h and Li-2 h, indicating the existence of Li 2 O, which may locate underneath the formed layer and direct contact with Li metal. There are two peaks at ~32.00°, the peak at 32.48° is the (101) facet of LiOH and the peak at 32.53° is the (200) facet of Li 3 N. The Li 3 N (200) peak appears first, then the LiOH (101) peak becomes stronger gradually with increasing exposure time and the same applies to the two peaks at ~36.00°. Finally, the peaks of LiOH and Li 2 CO 3 became the most obvious among the formed compounds and it can be inferred that LiOH and Li 2 CO 3 exist at the outermost region of the APL. Li 3 N should be located in the middle or interior of the APL because Li 3 N is unstable in air. In addition, Li 2 CO 3 can react with Li metal under the internal operating conditions of the battery [20] . The possible distribution of the lithium-based compounds in the protection layer is shown in Figure 1A.
To obtain the morphological variations of the APL on Li metal, the surface was observed through scanning electron microscopy (SEM) analysis [ Figure 1C-G]. The surface of pure Li metal is a relative plane and with increasing exposure time, the surface becomes rougher with raised bulges observed in the zoomed-in area. The enriched cracks indicate the formation of a large amount of lithium-based compounds, because the molecular volumes (cell volume) of Li 3 N, Li 2 O, LiOH and Li 2 CO 3 are 44.88, 25.27, 56.89 and 122.9 Å³, respectively, and that of a Li atom is 42.9 Å³. The molecular volumes of Li 3 N, LiOH and Li 2 CO 3 are larger than that of a single lithium atom. Therefore, with the accumulation of these compounds on the Li metal substrate, the inner stress of the APL increases, cracks appear and the APL of Li-1 h and Li-2 h tends to peel off. As shown in the digital photos in the insets of Figure 1C-G, the pure Li metal has a bright metallic luster, the majority of the Li-10 min surface becomes dark with indistinct metallic Li, while the surface colors of Li-30 min, Li-1 h and Li-2 h are darkened. The uniform distribution of O, C and N within the APL was confirmed by the elemental characterization [ Figure 1H].
To further investigate the components of the APL, we performed high-resolution transmission electron microscopy(HRTEM) tests [ Figure 2A-I] to reveal that the APL mainly contained nanocrystals of lithiumbased compounds. The crystalline grains were identified. The lattice fringe with the interplanar spacing of 0.265 nm was identified to be the (111) crystal plane of Li 2 O and the (110) and (102) Figure 2J, with the diffraction rings ascribed to (-220) Li 2 CO 3 , (211) LiOH, (212) LiOH and (422) Li 2 O. The chemical composition of the APL was also obtained by X-ray photoelectron spectroscopy (XPS). Figure 2K-M show the Li 1s, C 1s and O 1s X-ray photoelectron spectra of APL, respectively. The results also confirm the existence of Li 2 O, Li 3 N, LiOH and Li 2 CO 3 . The peaks at 54.0, 54.9 and 55.2 eV of Li 1s can be ascribed to Li 2 O, LiOH and Li 2 CO 3 /Li 3 N, respectively. The peak of C 1s at 284.8 eV is the hydrocarbon and the peak at 289.7 eV is Li 2 CO 3 . In addition, other species of Li were detected at 286.3 eV, which could be ROCO 2 Li [21] . Based on the above characterization, although the components of the in-situ APL have been revealed, it is also meaningful to confirm its dendrite suppression ability. First, the electrochemical performance was evaluated carefully at room temperature [ Figure 3] using symmetrical cells. The initial electrochemical impedance spectroscopy (EIS) results reveal the low conductivity of APL [ Figure 3A], with charge transfer resistance (Rct) values of 60, 420, 900, 2720 and 6294 Ω for pure Li, Li-10 min, Li-30 min, Li-1 h and Li-2 h, respectively. The Rct increased drastically with the exposure time. The EIS plots after cycling (1st, 2nd, 5th, 10th, 20th, 50th and 100th) were acquired, with the Rct value tending to reduce with cycling, as shown in Figure 3B. The Rct of all samples decreased beneath 200 Ω after the 1st cycle, which could be due to the uniform deposition of Li metal in the APL after cycling. Conductive pathways were then constructed, revealing that the electrochemical performance was only partially determined by the initial resistance. Figure 3C and D show the Tafel plots for evaluating the exchange current density. Pure Li exhibits the largest exchange current density; after cycling, the value of all samples tends to increase, which corresponds to the EIS results. The above data reveal that the conductivity of the APL was enhanced after cycling.  Figure 3E shows the voltage profiles of the symmetrical cells during Li plating and stripping with increasing current densities of 1, 2, 3, 5 and 10 mA cm -2 at a fixed capacity of 1 mAh cm -2 . Li-10 min exhibits almost the same plots as pure Li but with a larger voltage range. With the exposure time increasing to greater than or equal to 30 min, the voltage range was kept in the range of -50-50 mV at 1 mA cm -2 . Among all the samples, Li-30 min exhibited the best voltage control ability and when the current density switched to 10 mA cm -2 , the voltage fluctuation range could still be maintained at -100-100 mV. The ultralong lifespan of up to 4500 h of Li-30 min at the current density of 10 mA cm -2 is shown in Figure 3F. Pure Li has a short circuit at ~240 h (120 cycles) and the voltage of Li-1 h and Li-2 h was unstable. The selected strippingplating curves in Figure 3F further reveal the excellent dendrite suppression ability of the APL. To observe the stripping-plating behavior of the APL more intuitively, liquid cell tests were employed with digital images of the Li/Li-30 min stripping-plating at the current density of 1 mA cm -2 shown in Figure 3G. Bunches of dendrites were formed on the surface of pure Li metal with dots of Li metal observed on the Li-30 min surface. The SEM images of the pure Li surface after short circuit (120 cycles) are shown in Figure 3H, with a typical dendrite morphology observed. As shown in Figure 3I, the surface of Li-30 min after 120 cycles shows that no dendrites or round particles formed; as the cycling time increased to 2000 h (1000 cycles), a dendrite-like structure was observed [ Figure 3J], but the end was round [ Figure 3K], this corresponds to the conclusion that the Li nucleation and early growth is critical to the final deposition morphology [22] . The above analysis reveals the excellent suppression and morphology-controlling ability of the in-situ APL on dendrites.
As four main Li compounds were detected, it is therefore imperative to test the dendrite-suppression performance of each compound. Li 3 N, Li 2 O, LiOH, and Li 2 CO 3 were applied to modify Li metal by cold  Figure 4A]. The Li-30 min symmetrical cell shows stable Li plating-stripping for more than 500 h and exhibits excellent dendrite suppression at low temperatures [ Figure 4B]. Under 30 °C [ Supplementary Figure 2], the capacities of Li-30 min|NCM811 1mg at 0.1 C, 0.2 C, 0.5 C, 1.0 C, 2.0 C, 5.0 C and 10.0 C are 208, 205, 196, 184, 140, 100 and 41 mAh g -1 , respectively, while those of pure Li|NCM811 1mg are 173, 154, 146, 137, 117, 93 and 54 mAh g -1 , respectively, Li-30 min|NCM811 1mg , respectively, exhibiting higher capacity compared to pure Li|NCM811 1mg . This is because the APL of Li-30 min provides the required components for the SEI, while the extra capacity cost to generate the continuous formation SEI films in pure Li|NCM811 1mg , but at 10 C, the capacity of pure Li |NCM811 1mg becomes higher due to the inferior conductivity of the APL on Li-30 min hindering the formation of smooth Li. As a result, the APL performs well at low rates. At 30 °C [ Supplementary Figure 3], a long cycling (300 cycles) test was performed, with the per cycle capacity decay of Li-30 min|NCM811 1mg being 0.05%, while that of pure Li|NCM811 1mg is 0.12%.
The charge-discharge curves of Li-30 min|NCM811 1mg at 30, -20 and -40 °C are shown in Supplementary Figure 4, Figure 4C and Supplementary Figure 5, exhibiting sluggish dynamics and chargedischarge capacity decreases at reduced temperatures. Figure 4D shows the cycling performance of Li-30 min|NCM811 1mg and pure Li|NCM811 1mg at different rates at -20 °C, 200 cycles at 1.0 C, 200 cycles at 2.0 C and 250 cycles at 0.5 C. The corresponding capacities of Li-30 min|NCM811 1mg are 70, 23 and 100 mAh g -1 , respectively, while those of pure Li|NCM811 1mg are 72, 38, and 92 mAh g -1 , respectively. At 2.0 C, Li-30 min|NCM811 1mg exhibits lower capacity, which shows the same trend as that at 30 °C, but due to the sluggish dynamic at -20 °C, the rate performance is inferior. Nevertheless, Li-30 min|NCM811 1mg shows the advantage at small current densities; at 0.5 C, its capacity is higher than that of pure Li|NCM811 1mg . In addition, the per cycle capacity decay at 0.5 C of Li-30 min|NCM811 1mg is nearly zero, while that of pure Li|NCM811 1mg is 0.02%. The rate and cycling performances of Li-30 min|NCM811 1.5mg and pure Li|NCM811 1.5mg were evaluated at -20 °C. The capacities at 0.1 C, 0.2 C, 0.5 C and 1.0 C are 120, 113, 88 and 60 mAh g -1 , respectively, and the capacity recovered to 120 mAh g -1 when current density switched back to 0.1 C [ Figure 4E], reflecting good reversibility. As shown in Figure 4F, Li-30 min|NCM811 1.5mg exhibits excellent cycling stability under -20 °C; after 1000 cycles at 1.0 C, the capacity maintains at ~51 mAh g -1 without decay. The rate and cycling performances were further investigated under -40 °C, as shown in Supplementary Figures 6 and 7. The capacities of Li-30 min|NCM811 1mg at 0.05 C, 0.1 C and 0.2 C are 95, 66 and 30 mAh g -1 , respectively. Li-30 min|NCM811 1mg is unable to charge-discharge when current increases over 0.5 C and the corresponding capacities of pure Li|NCM811 1mg are 53, 47 and 0.3 mAh g -1 , respectively. The cycling performance at 0.1 C (100 cycles) and 0.2 C (50 cycles) indicates the good stability of Li-30 min|NCM811 1mg . In addition, the charge-discharge efficiency is lower at -40 °C.

CONCLUSION
Exposing pure Li to air alters the Li deposition behavior and enables the dendrite-free growth of Li metal at moderate current densities, thereby enhancing the cycling stability and extending the cycle life for highcapacity Li metal batteries at a wide temperature range. In contrast to typical efforts made to protect Li metal from oxidation, we exposed Li to air and made it oxidize. This strategy has the advantages of being extremely low cost, green and has been shown to be applicable to various metal batteries (e.g., Na-, K-and Zn-metal batteries) at wide temperatures.
Preparation of air-pretreated Li metal: Li metal was exposed to air (humidity of 40%). Four samples were then obtained and all samples were named according to their air exposure time, which are Li-10 min, Li-30 min, Li-1 h and Li-2 h.
Structural characterization: X-ray diffraction (XRD) patterns were obtained by using an X'Pert PRO MDP with Cu Kα radiation (λ = 1.5405 Å) with 40 mA and 40 kV. X-ray photoelectron spectroscopy (XPS) was collected at an Escalab 250Xi spectrophotometer. Field emission scanning electron microscopy (FESEM) images were obtained using a JEOL JSM-7001F scanning electron microscope operating at 15 kV. Transmission electron microscopy (TEM) images were obtained using a JEOL JEM-2100F electron microscope operating at 200 kV. Thermogravimetric analysis (TGA) data were collected using a DTA-60 (Shimadzu), annealed under air/N 2 atmosphere at 10 °C min -1 .

Electrochemical measurements:
The CR2032 coin-type cells were assembled in an argon-filled glove box. The electrolyte consisted of a solution of 1.0 M LiPF 6 in EC:DEC (1:1 v/v) and the amount of electrolyte used for the batteries was 40 μL. The cells were cycled galvanostatically in the voltage range of 3.0-4.2 V with a battery testing system (Neware Electronic Co., China). Electrochemical impedance spectroscopy (EIS) was conducted by a bio-logic VMP3 electrochemical workstation.

Authors' contributions
Completing the experiments and preparing the manuscript draft: Zheng S Discussing and revising the manuscript: Geng H, Eliseeva SN Writing-review and editing, funding acquisition, supervision: Wang B