Coupling Water‐Proof Li Anodes with LiOH‐Based Cathodes Enables Highly Rechargeable Lithium–Air Batteries Operating in Ambient Air

Abstract Realizing an energy‐dense, highly rechargeable nonaqueous lithium–oxygen battery in ambient air remains a big challenge because the active materials of the typical high‐capacity cathode (Li2O2) and anode (Li metal) are unstable in air. Herein, a novel lithium–oxygen full cell coupling a lithium anode protected by a composite layer of polyethylene oxide (PEO)/lithium aluminum titanium phosphate (LATP)/wax to a LiOH‐based cathode is constructed. The protected lithium is stable in air and water, and permits reversible, dendrite‐free lithium stripping/plating in a wet nonaqueous electrolyte under ambient air. The LiOH‐based full cell reaction is immune to moisture (up to 99% humidity) in air and exhibits a much better resistance to CO2 contamination than Li2O2, resulting in a more consistent electrochemistry in the long term. The current approach of coupling a protected lithium anode with a LiOH‐based cathode holds promise for developing a long‐life, high‐energy lithium–air battery capable of operating in the ambient atmosphere.


Chemical and reagents
Dimethyl sulfoxide (DMSO, >99.9%, Macklin) and tetraethylene glycol dimethylether (TEGDME, ≥99.9%, Dow Chemical) were dried over freshly activated 4 Å molecular sieves for weeks. Lithium bis(trifluoromethanesulfonyl)imide (LiTFSI, purity of >99.99%, Solvay) was used as the lithium salt after drying in vacuo at 160°C for 24 h. All chemicals were stored in an Ar-filled glove box (Mikrouna) with both H2O and O2 concentrations lower than 0.01 ppm before use. Electrolytes of 1 M LiTFSI in DMSO or TEGDME were prepared and stored in a glove box under an Ar atmosphere. Various added water contents were based on volume percentage. For electrolyte with 18 O-enriched water, 18 O-enriched water (with enrichment levels of 97% H2 18 O and 3% H2 16 O, Shanghai Maotu Gases Co., Ltd.) was used as purchased.

Preparation of Ru/SP
Procedures were based on our previous work. [1] Briefly, 300 mg of Super P (SP, Timcal) was dispersed in 100 ml of ethanol (EtOH, ≥99.7%, Macklin) containing 46.6 mg of RuCl3 (purity of 99.99% metals basis, Macklin), and stirred continuously until the solution evaporated completely. The resulting mixture was subsequently reduced in a 5% H2/Ar mixture atmosphere at 230℃ for 2 hours. Then, the final product Ru/SP was collected.

Preparation of electrodes
The cathode film composed of Super P and polytetrafluoroethylene (PTFE, a dispersion of 60wt%, Macklin) with a ratio of 80:20wt% was kneaded and rolled into a thin film. The film was subsequently punched out to form free-standing electrodes of various sizes (0.5-2 cm 2 ) and further dried in vacuo at 160°C for 12 hours before the battery assembly. Typically, the areal loading of the electrode is around 1 mg/cm 2 . For preparation of Ru/SP cathodes (SP/Ru/PTFE=75:5:20wt%) and lithium iron phosphate/ferric phosphate electrodes (LFP (Canrd)/FP (Aladdin)/SP/PTFE = 40:20:20:20wt%), a similar procedure was conducted.

Preparation of PEO-LATP-Wax protective layers
25 mg of PEO (Polyethylene oxide) (average Mv ~600,000, Aladdin) were added to 1 ml of TEGDME (tetraethyleneglycol dimethyl ether) and stirred for 6 h at 80℃ until a clear solution was obtained. Lithium sheets (diameter, 15.6 mm) were held by a flat tweezer and immersed in the solution for seconds. Then, lithium sheets were placed on a hot plate for 24 hours and then transferred into vacuum overnight to remove the excessive solvent. After drying, PEO-coated lithium metal anodes were obtained. LATP (lithium aluminum titanium phosphate) powder (~300 nm in diameter, Shenzhen, Kejing) was pressed onto mirror-polished stainless-steel sheets under a pressure of around 12 MPa and sintered at 850 °C for 10 h at a heating rate of 3 °C min −1 in air. Subsequently the LATP films were transferred into Ar filled glove box and pressed onto the Li-PEO at 60℃.When pressed, the soft PEO layer conforms to provide a good interface with LATP and it also avoids LATP degradation if in direct contact with Li metal.
Liquified wax was then infiltrated into the porous LATP film at 65℃, so that the microcracks of LATP were filled up. After cooling, the multifunctional protective layer has been fabricated.
The areal loading of the composite protective layer is estimated to be ~5.6 mg/cm 2 .
Preparation of a PEO-wax layer on Li metal has been reported previously. [2] Briefly, in our case, 25 mg of PEO (average Mv ~600,000, Aladdin) and 100 mg of wax (paraffin wax, Damas-beta) were added to 1 ml of toluene and stirred for 6 h at 80℃ until a clear solution was obtained.
Lithium sheets (15.6 mm in diameter, 0.6 mm or 0.15 mm in thickness) were immersed in the solution for seconds. Then, lithium sheets were placed on a hot plate for hours and then transferred into vacuum overnight to remove the excessive solvent. After drying, PEO-waxcoated lithium metal anodes were obtained.

Cell assembly and electrochemical measurements
All batteries were assembled using a Swagelok design in a glove box. A typical half-cell was assembled by successively stacking the LFP/FP counter electrode (diameter, 22 mm), a piece of glassy microfiber filter paper (GF/A, Whatman) with around 100 µL of the electrolyte, a SP or Ru/SP cathode and finally a stainless-steel mesh as the current collector (open area of 33%).
To fabricate a full cell, a Li anode with the protective layer, a piece of glassy microfiber filter paper (GF/A, Whatman) with around 100 µL of the electrolyte, a Ru/SP cathode and finally a stainless-steel mesh as the current collector were stacked in sequence. In the two-compartment full cells, the following items were stacked in sequence: a lithium sheet as the anode, a piece of glassy microfiber filter paper wetted with 1 M LiTFSI/TEGDME (anolyte), a disk of lithium ion conducting glass ceramic (LICGC, Ohara) membrane preventing the anode from the influence of water and/or carbon dioxide contamination, a piece of glassy microfiber filter paper wetted with 1 M LiTFSI/DMSO with 5 vol% added water (catholyte), and finally the Ru/SP cathode.
Before each galvanostatic test, the cells were rested for 30 min. For all batteries, galvanostatic electrochemical measurements were carried out using the battery testing system CT-4000 (Neware) at room temperature. All potentials in this study were referenced to Li/Li + . After the test, the electrode was extracted from the cell and rinsed by dry acetonitrile (< 1 ppm H2O) several times (each time with 2 ml for 10 minutes), and dried under vacuum for further characterization. Electrochemical impedance spectroscopy (EIS) has been performed using an electrochemical workstation (Ivium, Vertex.C.EIS). Electrochemical impedance spectra were acquired at open circuit voltage by applying a sinusoidal wave with an amplitude of 10 mV in a frequency range from 0.01 to 1,000,000 Hz.

Scanning electron microscopy (SEM) and transmission electron microscopy (TEM)
Morphologies of various cathodes were acquired by a Hitachi-S4800 scanning electronic microscopy (SEM) and a JEOL-2010 high-resolution transmission electron microscopy (TEM), both equipped with energy-dispersive spectroscopy (EDS). The dried cathodes were taken to the SEM sample loading chamber in a sealed sample holder to minimize air exposure. The time from opening the sealed sample holder to finishing sample loading into electron microscopes was < 10 seconds.
The Ru/SP powder was dispersed in ethanol, and it was sonicated for 10 minutes to obtain the well dispersed solution. Then, the solution was dropped onto a Cu grid pre-coated with the Lacey carbon and excessive solvent was removed in vacuum overnight.

X-ray diffraction (XRD)
XRD measurements were performed using a DX-2700B powder diffractometer (Dandong Haoyuan Instrument Co., Ltd.) operated at 40 kV and 30 mA with Cu Kα as the irradiation source (λ = 1.5405 Å). The dried electrode was put into a home-made X-ray sample holder that was sealed with Kapton polyimide films. XRD is employed to investigate the samples in the 2θ range of 19-39°, using a step size of 0.02° and a scan rate of 0.4° per minute.

X-ray photoelectron spectroscopy (XPS)
XPS measurements were conducted on a Kratos AXIS-HS spectrometer equipped with a monochromatic Al Kα X-ray source (150 W, 15 kV, 10 mA). The pass energy for the fixed analyzer transmission mode is 40 eV, and the scanning spot size is 20 μm × 20 μm, with a step size of 5 μm. C 1s, Ru 3p and Ti 2p spectra were acquired at a step size of 0.1 eV/sec over 270-300 eV, 446-498 eV and 448-479 eV, respectively.

Nuclear magnetic resonance spectroscopy (NMR)
Solution 1 H-NMR and 13 C-NMR spectra were acquired using an ADVANCE III HD spectrometer (400 MHz), chemical shifts are quoted in ppm referenced to an appropriate reference solvent peak. Typically, the glass fiber separator after battery cycling was immersed into DMSO-d6 (0.7 ml, 99.8%, Acros) solvent for 15 minutes. The resulting solution was transferred into an NMR tube and then the tube was sealed with a cap and parafilm.
A rotor (3.2 mm) synchronized Hahn-echo pulse sequence was used to acquire 7 Li magic angle spinning (MAS) spectra with a spinning speed of 55-60 kHz, with a recycle delay of 20 s, and an RF field strength of 125-170 kHz. 7 Li shifts was externally referenced to Li2CO3 at 0 ppm.

Operando Electrochemical Mass spectrometry (OEMS)
A newly designed OEMS system was described in Figure S1. The OEMS system consists a 1/16" polypropylene tube carrying a working gas from an electrochemical cell to a mass spectrometer (QMG250M1, Pfeiffer, Linglu Instrument Co., Ltd). The cell design consists in a 1" Swagelok-type cell with inlet and outlet tubes (1/16") welded to the top plunger. The entire system was hermetically sealed. The mass spectrometer was calibrated to determine the partial pressure of a standard mixture of 500 ppm O2 and 5000 ppm CO2 in Ar. The flow velocity of the working gas (21.2% O2, 0.038% CO2 and 78.76% N2) and carrier gas (Ar, ≥99.999%) was controlled at 0.09 ml/min and 2 ml/min via two mass flow controllers (Bronkhorst), respectively.
Before in situ DEMS characterization, the residual gas in the cell was purged by Ar until stable O2 and CO2 partial pressures were obtained, and the OEMS cells were cycled using a battery tester (CT-4000, Neware). The dashed line between valves 1 and 2 represents a tube, which will be connected when the battery is not in operation, so as to prevent air pollution to the OEMS system.

Fourier transform infrared spectroscopy (FTIR)
Fourier-transform infrared spectra were recorded on a Bruker ALPHA spectrophotometer in the region of 4,000~400 cm −1 with a resolution of 2 cm −1 .

Preparation of discharged cathodes
The titration quantification conducted here is modified based on previous work. [3] A cathode was removed from the discharged cell and placed in a glass vessel in an Ar-filled glove box without exposure to air. Typically, the time between the end of cell discharge and cathode extraction was less than 0.5 h. The vial was then sealed with a silicone septa lid and transferred out of the glove box. 20 mL of ultrapure water (18.2 MΩ cm, Millipore) was injected into the sealed vial using a syringe. The vial contents were then vigorously shaken for 30 seconds. The resulting solution was used for the quantification of Li2CO3, Li2O2 and LiOH.

Li2CO3 calibration
To calibrate the amount of Li2CO3, lithium carbonate (Li2CO3) solutions with known quantities were prepared by dissolving Li2CO3 (>99.99%, Macklin) of different mass in ultrapure water.
The diluted Li2CO3 solution (0.5 ml) was injected into the cell using a syringe under an Ar flow (2 ml/min). When a stable CO2 partial pressure was obtained after injecting Li2CO3 solution, and then HCl (1 M, 1 mL) was injected into the cell to obtain CO2 signal. The amount of Li2CO3 was correlated to the amount of evolved CO2 (Equation S1) which was detected by the mass spectrometer ( Figure S2a). By integrating the amount of evolved CO2, the total number of moles of Li2CO3 was calculated using the standard curve ( Figure S2b).

Li2O2 calibration
The TiOSO4-based ultraviolet-visible titration was used for the quantitative determination of Li2O2. [4] Li2O2 was firstly hydrolyzed to form LiOH and H2O2 (Equation was obtained, as shown in Figure S3a. By measuring the absorbance of peak at 405 nm, the number of moles of Li2O2 was calculated using the standard curve ( Figure S3b).

Evaluation of measurement errors in titration
Li2CO3 with different known masses had been weighed, and subjected to titration to evaluate the accuracy. The results are given in Table S2: the errors for Li2O2, Li2CO3, LiOH are <10%, <15%, and ~15%, respectively. Our measurement error is similar to those in others' report. [5] 3 showed that as the electron/O2 molar ratio increased from 2 to 4, O2 and CO2 evolution were both reduced, consistent with LiOH formation and decomposition becoming the dominant reactions during cycling. Because this cell was in a strictly sealed system, any rise in water in the cycled electrolyte should be originated from the battery chemistry itself. Previous studies [6][7] have suggested that the formation of LiOH in anhydrous DMSO electrolyte was due to the nucleophilic attack of O2·to the electrolyte, and subsequent LiOH decomposition on charging would generate water due to Ru catalysts. [1] 1 H NMR analysis ( Figure S6d) of the cycled electrolytes further confirmed that indeed the water signal has increased compared to that of a pristine electrolyte. It therefore suggests that even in an anhydrous cell using Ru/SP catalysts, the cell reaction initially being Li2O2-dominated will be converted to LiOH in nature. In order to further characterize the discharged products, 7 Li solid state nuclear magnetic resonance (NMR) and Fourier transformed infrared spectroscopy (FTIR) measurements have been performed to characterize the formation and decomposition of discharge products. In Figure S7a, 6 cells terminated at discharged/charged states of different cycle numbers were prepared and subjected them to 7 Li solid state NMR tests. Consistent experimental parameters (number of scans, discharge/charge capacities) and sufficient recycle delay were applied to allow quantification of the cell reaction. A single resonance at 1.1 ppm ( Figure S7b) for 7 Li, characteristic of LiOH, was observed in discharged samples and completely disappeared after recharge. This additional result further supports that the cell reaction is predominantly based on LiOH formation and decomposition, consistent with the rest of our report. The 1 st discharged cathode was prepared to characterize the discharged products via FTIR. Although the peaks of Li2CO3 at 857 and 1400 cm -1 can be observed ( Figure S7b), FTIR characterization of discharged cathode suffers from difficulties for quantification (only sampling a small fraction of the cathode). Furthermore, the most intense peaks situate at the far infrared region (200-600 cm -1 ) for all potential products (LiOH, Li2CO3, Li2O2) and are difficult to be discerned and differentiated in common mid-IR spectra. Therefore, in this work we primarily rely on quantitively analysis such as chemical titration, OEMS and the newly added solid state NMR to verify the nature of the cell reaction. To reveal the impact of CO2 accumulation on Li2O2-based electrochemistry over extended cycles, the corresponding O2 (m/z=32), CO2 (m/z=44) signals during cycling were obtained via OEMS. As shown in Figure S8a, there was no resting periods between charge and the following discharge, which was attempted to be kept consistent with Figure 2c. In order to focus on the details, regions representing the transition from 1 st charge to 2 nd discharge ( Figure S8b (i)), and from 7 th charge to 8 th discharge ( Figure S8b (ii)), have been enlarged.

CO2 accumulation impact on
As shown in Figure S8b(i), O2 evolution occurred at the beginning of the 1 st charge, but started to decrease beyond 2/3 of the charge capacity (indicated by the dashed arrow), at which point CO2 evolution began to rise. During the following discharge, oxygen consumption and a concomitant CO2 signal drop were observed. Of note, a small first step in the O2 consumption profile was recorded, which corresponded to a level of four electrons per reduced O2; moreover, a rapid drop in the CO2 signal (bending of the CO2 profile, as more clearly revealed in Figure   S8b(ii)) was observed in accordance to a small discharge step in the electrochemical profile.
These observations suggest that in the presence of higher CO2 concentrations, cell discharge was initiated by electrochemical CO2 reduction, probably via CO2 + O2 + 4Li + + 4e -→2Li2CO3. [8][9][10][11][12] During the decomposition of Li2CO3 on charging, the O2 released was far below the value expected by the reaction stoichiometry above, the O2 signal falling to zero when CO2 evolution peaked, suggesting that L2CO3 formation is not reversible [9,11,[13][14] . At the 7 th charge, the amount of O2 evolution decreased, whereas Li2CO3 decomposition and CO2 evolution increased considerably, implying accumulation of Li2CO3 upon cycling. It is worth noting that the length of the discharge plateau related to electrochemical CO2 reduction in OEMS experiments was much shorter than that in Figure 2c, which is likely due to the constant working gas flow through the cell rapidly dissipating the released CO2. In the absence of a dynamic gas flow (Figure 2c), the CO2 concentration localized at the electrode-electrolyte interface would be much higher, promoting direct CO2 reduction for a larger portion of the discharge process. Therefore, the main route for Li2CO3 in a cell with SP cathodes and anhydrous electrolytes (Li2O2-based electrochemistry) in air is primarily via direct electrochemical reduction of CO2. In order to compare the chemical stability of the two discharge products (Li2O2 and LiOH) under dry air, the resting period between battery discharges and charges has been varied from 0 to 6 hours. The corresponding O2 (m/z=32), CO2 (m/z=44) signals during cycling was obtained via OEMS. As shown in Figure S8i, for cells using SP cathodes and anhydrous DMSO electrolytes, the discharge process showed around two electrons per reduced O2, consistent with dominant Li2O2 formation. With the increase of rest time (from 0 to 6 h), O2 release was decreased (from 3.43 to 2.58 μmol), and CO2 release was increased (from 0.26 to 1.48 μmol).
For cells using Ru/SP cathodes and DMSO electrolytes with 5 vol% H2O ( Figure S9ii), however, the discharge process showed around four electrons per reduced O2, consistent with dominant LiOH formation. No O2 release was observed on subsequent charge processes due to DMSO2 formation, consistent with previous reports. [1] Meanwhile, with the increase of resting time, CO2 release was also increased (from 0.27 to 0.81 μmol) for Ru/SP case. Nonetheless, for the same resting period, the cell with SP cathode revealed more CO2 evolution than that with Ru/SP cathode, which is consistent with the higher thermodynamic driving for CO2 reaction with Li2O2 than LiOH to form Li2CO3. [15] 8. Factors affecting the fraction of Li2CO3 formation Figure S10.  Figure S11. Evolution of XRD patterns acquired from a lithium disc exposed in air for different durations in ambient air. The XRD patterns of the sample holder and relevant reference compounds are also presented for comparison.

XRD characterization for Li metal exposed in ambient air
For a fresh lithium metal, the diffraction pattern featured two dominant Bragg reflections at 36.8 and 52.0 degrees. After 6 hours, new peaks associated with LiOH could be seen, which continued to intensify and had become the main feature in the diffraction pattern after a day of exposure. At this point, weak reflections linked to Li2CO3 formation could also be identified.
These observations imply that moisture in air is the most aggressive component that react with lithium metal, forming LiOH as the prevailing surface product. We initially employed a dual compartmental design by an Ohara ceramic glass separating the anolyte and catholyte parts ( Figure S12a). In dry air, this setup permitted stable electrochemical cycles ( Figure S12b), the discharge-charge profile resembling that of half cells (Figure 2d).

Compartmental Li-air full cell using
OEMS measurement further supported that the cell reaction was based on an ORR, consistent with LiOH formation. Nevertheless, this dual compartmental setup is not adequate to enable a stable cycling in real air. The battery failed in just two days, because of lithium corrosion by air. To investigate whether wax increases the resistance to Li + diffusion, we compared the EIS results obtained on two setups, one with PEO-LATP and the other with PEO-LATP-wax electrolyte (note that the PEO-LATP-wax was prepared from the same PEO-LATP disc) in an all-solid-state Li symmetric battery (see scheme in Figure S13). The electrolyte resistance in both cases were found to be very close (slight variation may be due to different pressures applied to the battery during measurements), indicating that Li + diffuse across the protective layer via LATP-PEO and wax does not induce additional impedance ( Figure S13). For ease of practice and improved robustness of the setup, a thicker solid-state protective layer was applied here, and hence the unusually higher resistances observed.

Protection strategy of Li anodes
7a37d268446ccb8808891a5f5de1472a.mp4 Movie S1. Videos comparing the air-stability and reactivities to wet electrolytes/water of Li anodes. Vigorous gas bubbling has been observed in all cases except for the lithium anodes protected by the PEO-LATP-wax composite layer. In order to study the potential of high humidity operation, we constructed a lithium-air full cell using a thin lithium foil (0.15 mm), a wet (5 vol%) LiTFSI/DMSO electrolyte and Ru/SP cathode, cycling in 99% maximum humidified air ( Figure S14a). When the humidity was increased to 99%, the voltage profiles showed no difference ( Figure S14b) from the same cell operating in ambient air ( Figure S14c). A fully discharged/charged cell to a larger areal capacity is also presented in Figure S14d These additional experimental results confirmed the validity of the lithium protection strategy by PEO-LATP-wax, and further verified the robustness of the LiOH formation/decomposition mechanism against air and water.   H2O gradually reverted to that with an anhydrous electrolyte ( Figure S17). The corresponding OEMS results further suggested that the reaction had gradually transformed from LiOH formation to Li2O2 formation, as evidenced by the increasingly larger O2 consumption during discharge and more CO2 and O2 evolution on recharging. Previous work [1] by some of the authors showed by using 1 H, 13  and δ( 13 C), respectively), [1] which was absent in the pristine electrolyte. This indicates that DMSO solvent in the electrolyte appeared to be oxidized during repeated charging (as there was no oxygen evolution). On the contrary, the Ru/SP cathode after cycling revealed little chemical change, as evidenced by XPS results; this reiterates a better interfacial stability between LiOH and carbon cathodes. The surface of the mesoporous cathode was not covered by a large amount of solid side reaction products (Figure 7f(iii-vi)), unlike the typical case cycling via Li2O2 formation. In addition, 1 H measurement shows that the water content in the electrolyte increased after cycling (resonance at 3.38 ppm for δ( 1 H)), which may result from the charging reaction continually regenerating water [1] and some ingression of moisture from the ambient air over time. Therefore, the failure of the full cell was likely related to electrolytes being depleted rather than cathode clogging or H2O being depleted. Table S3 shows that hardly any prior studies have demonstrated a highly rechargeable lithiumair full cell could operate under the harsh conditions investigated in this work. In terms of electrochemical performance (cycle life, areal/specific capacity), the results shown in this study are competitive to some of the best work reported (Table S4). hydrophobic graphene LFP and NCM-811 [16] 2 2017 PDMS film LFP [17] Adv. Mater.