Electrolyte solution preparation
All chemicals were used as received without purification. Salts and organic solvents were purchased from Sigma-Aldrich. Salts used in this study include Zn(OTf)2 (purity, 98%), ZnSO4·7H2O (purity, 99.5%), Zn(ClO4)2·6H2O (purity not specified) and lithium trifluoromethanesulfonate (purity, 99.995%). Organic solvents include ME (CH3OCH2CH2OH; purity, 99.0%), DME (CH3OCH2CH2OCH3; purity, 99.0%), DEE (CH3CH2OCH2CH2OCH2CH3; purity, 98.0%), diglyme ((CH3OCH2CH2)2O; purity, 99.0%), 1,4-dioxane (C4H8O2; purity, 99.0%), 1,3-dioxolane (C3H6O2; purity, 99.0%), methanol (CH3OH; purity, ≥99.8%), ethanol (CH3CH2OH; purity, ≥99.5%), 1-butanol (CH3(CH2)3OH; purity, 99.8%), dimethyl carbonate ((CH3O)2CO; purity, ≥99.9%), diethyl carbonate ((C2H5O)2CO; purity, ≥99.0%), ethylene carbonate (C3H4O3; purity, ≥99.0%), propylene carbonate (C4H6O3; purity, ≥99.0%), γ-butyrolactone (C4H6O2; purity, ≥99.0%), N,N-dimethylformamide (HCON(CH3)2; purity, ≥99.8%), acetonitrile (CH3CN; purity, ≥99.9%), trimethyl phosphate ((CH3O)3PO; purity, ≥99.0%,), dimethyl sulfoxide ((CH3)2SO; purity, ≥99.9%) and sulfolane (C4H8O2S; purity, ≥99.0%). The zinc salt concentration was controlled at 3 mol kg−1 (3 m) where the mass of the solvent is the mass of water (ultrapure water, approximately 18.2 MΩ cm at 25 °C, purified by a Milli-Q water purification system) and the organic additives in various molar percentage values. Additional 2 m of lithium trifluoromethanesulfonate was added into 3-m Zn(OTf)2 electrolyte solutions with the ether additives to provide Li+ ions for intercalation/deintercalation at the LiMn2O4 and LiVPO4F positive electrode.
Electrode preparation
To synthesize the NaV3O8 positive electrode active material, 3 g of commercial V2O5 powder (Sigma-Aldrich; purity, ≥98.0%) was added into 100 ml of 2 mol l−1 of NaCl aqueous solution (Sigma-Aldrich; purity, ≥99.0%) and stirred magnetically at 400 rpm for 72 h in air under a fume hood in a glass beaker. The resulting orange–red powders were washed with 300 ml of deionized water for three cycles. In each washing step, the powders were dispersed in water and stirred at 400 rpm for 5 min in a glass beaker, followed by collection via centrifugation at 8,000 rpm for 10 min. The obtained powders were subsequently freeze dried at temperatures below −40 °C under a vacuum of 2 (purity, ≥99.0%; water content, ≤50 ppm) and LiMn2O4 (purity, ≥99.0%; water content, ≤800 ppm) powders were purchased from Carnd Technology, whereas LiVPO4F (purity, ≥99.0%; water content, ≤500 ppm) powders were supplied from Advanced Lithium Electrochemistry. All powders were dried at 120 °C under vacuum for 12 h before use. Composite positive electrodes were prepared by mixing the active material powder, electron-conductive carbon additive (Super P, Carnd Technology; purity, ≥99.5%, primary particle size, ≤50 nm; specific surface area, ≥62 m2 g−1) and polytetrafluoroethylene (Carnd Technology, 60 wt% dispersion in H2O) in a mass ratio of 7:2:1 in ethanol, followed by manually grinding using an agate mortar and pestle for 10 min in air at 25 °C. The dough-like electrode slurry was placed onto a titanium (Ti) mesh current collector (Carnd Technology; purity, ≥99.5%; thickness of 0.27 mm, 100 mesh, pore size of 0.15 mm) and roll pressed using a roller press (MSK-2150-H5, MTI). The electrode was pressed repeatedly (typically 5–10 passes) with a controlled roller gap to ensure uniform thickness and good adhesion between the active material and the current collector. The mass loading of the active material was controlled to approximately 12.5 mg cm−2 by adjusting the electrode area after roll pressing. The obtained electrodes were then dried at 80 °C overnight to remove any residual H2O and ethanol. The dry electrodes were cut using a precision disc cutter (MSK-T-10, MTI) before coin cell assembly. The positive electrodes obtained were stored in a vacuum desiccator before cell assembly and tested in a coin cell and pouch cell configuration. Zn foils (purity, ≥99.995%; thickness of 10, 15 or 100 µm) and Cu foil (purity, ≥99.95%; thickness of 9 µm) were purchased from Carnd Technology. The Zn negative electrodes and Cu current collectors were cut into disc-shaped electrodes using a precision disc cutter (MSK-T-10, MTI) before coin cell assembly. Before cutting, the Zn foil was polished with softback sanding sponges (3M) and then wiped with ethanol, whereas the Cu foil was cleaned by wiping with ethanol. For pouch cell assembly, the NaV3O8 positive electrodes, Zn negative electrodes and Cu current collectors were cut using a home-made cutter.
Electrochemical measurements
CR2025 coin-type cells were assembled in air at 25 °C using stainless steel (SS) cases and springs, with glass fibre membranes (Filtech; thickness of 0.66 mm and diameter of 19 mm) as the separator and an electrolyte solution volume of 100 µl. The electrolyte was transferred using a calibrated single-channel air-displacement micropipette (20–200 µl; Thermo Fisher Scientific) equipped with disposable 200-µl polypropylene tips, and applied dropwise onto the separator to ensure uniform wetting. The electrodes used in the coin cells were 12 mm in diameter. The crimping load applied for the coin cell assembly was of 0.5 t. For the Zn||NaV3O8 single-layer pouch cells, the composite positive electrode dimensions were 3 × 4 cm2, with a hydrophilic sulfonated composite separator (Carnd Technology; thickness of 0.152 mm, lateral size of 3.5 × 4.5 cm2, porosity of ≥88% and pore size of ≤20 μm) and a 10-µm-thick zinc foil (lateral size of 3 × 4 cm2). Nickel tabs (Carnd Technology; purity, ≥99.5%) were attached to both positive current collector (Ti mesh) and negative current collector (Cu) by ultrasonic welding (MSK-800-2K, MTI). The single-layer electrode stack was then placed into an aluminium-laminate pouch. The electrolyte was injected into the pouch using a pipette (100–1,000 µl, Thermo Fisher Scientific), with the electrolyte amount controlled at an E/C ratio of 6 g Ah−1. The pouch cell was subsequently vacuum sealed at ~10−2 torr for 60 s and heat sealed at 180 °C for 6 s (MSK-115A-S, MTI). After sealing, the cells were rested for 12 h to ensure complete electrolyte wetting before testing. An external pressure of approximately 0.1 MPa was applied to the pouch cell by sandwiching it between two rigid acrylic plates during testing. Charge–discharge tests of batteries were performed on a Neware battery test system (CT-3008) in an air-conditioned laboratory environment with an average temperature of 25 ± 1 °C. The charge–discharge tests of Zn||NaV3O8 coin cells at −35 °C was conducted in a temperature-controlled climatic chamber (GWS MT3065), with an average temperature deviation of ±0.5 °C. The mass used to calculate the specific current and specific capacity refers to the mass of the positive electrode active material. Three cells were tested for each electrochemical experiment, and consistent performance was observed across all cells. The results presented in the figures correspond to a representative cell; additional data from the other cells are available on request.
The bulk ionic conductivities (σ) and pH values of electrolytes at 25 °C were measured using a pH/conductivity multiparameter benchtop meter (Thermo Orion Versa Star Pro). Electrochemical impedance spectroscopy (EIS) measurements were used to measure bulk ionic conductivities at −35 °C, with the cell constant Kcell determined based on the bulk ionic conductivity at 25 °C. EIS measurements were carried out using SS||SS (Carnd Technology, 304; thickness of 15 μm, diameter of 16 mm, used as received) CR2025 coin cells (assembled as described above) on a VMP-300 potentiostat (BioLogic) using the potentiostatic mode with an amplitude of 5 mV and frequencies ranging from 500 kHz to 10 Hz, collecting 5 points per decade. Before each EIS measurement, the system was stabilized at the open-circuit potential for 300 s to ensure a steady state. The bulk ionic conductivity of an electrolyte solution at −35 °C was calculated using the following equation:
$$\sigma =\frac{L}{{SR}}=\frac{{K}_{\mathrm{cell}}}{R},$$
(1)
where L is the distance between the two SS electrodes, S is the contact area of the SS electrodes and R is the resistance value (in Ω) extrapolated at the intersection between the raw EIS data and the real impedance axis.
The corrosion rate (Corr) was evaluated by monitoring the potential of a Zn@Ti electrode, which was prepared by depositing 0.565 mAh of metallic Zn onto a Ti foil (Carnd Technology; purity, ≥99.89%, thickness of 20 μm and diameter of 12 mm). The measurement was conducted in Zn||Ti coin cells (assembled as described above). The Corr is determined using the following equation:
$$\mathrm{Corr}=\frac{{m}_{\mathrm{Zn}}}{t},$$
(2)
where \({m}_{{\rm{Zn}}}\) represents the total mass of the deposited Zn metal and \(t\) is the corrosion time corresponding to the complete consumption of Zn.
The HER and OER potentials of the electrolyte solutions were investigated using an HPR-40 differential electrochemical mass spectrometer (HIDEN Analytical), coupled with a gold (Au) working electrode, an Ag|AgCl reference electrode and a Pt counter electrode at a scan rate of 5 mV s−1 at 25 °C.
Cyclic voltammetry (CV) tests were conducted on symmetric Zn||Zn coin cells (assembled as described above) over a potential range of −15 mV s−1 to 15 mV s−1 at specified scan rates and 25 °C.
The reduction stability of water and ether additives were investigated by linear sweep voltammetry tests using a Ti working electrode, an Ag|AgCl reference electrode and a Pt counter electrode at a scan rate of 5 mV s−1 at 25 °C.
Tafel tests were conducted on symmetric Zn||Zn coin cells over a potential range of −20 mV s−1 to 20 mV s−1 at a scanning rate of 10 mV s−1 and 25 °C.
Ex situ physicochemical characterizations
The contact angle of the electrolyte solutions on the zinc metal negative electrode and the NaV3O8 positive electrode was measured using an optical tensiometer (Attension Theta, Biolin Scientific). Electrolyte droplets (approximately 3 µl) were automatically dispensed onto the electrode surface using a microsyringe. The contact angle was determined from the droplet profile and recorded after 10 s as a stable value. Each measurement was performed three times at different locations on the electrode surface, and the average value was reported.
FTIR spectroscopic measurements were performed using a Nicolet 6700 Thermo Fisher FTIR spectrometer in the attenuated total reflection mode.
Raman spectra were collected using a LabRAM HR Evolution Raman microscope (Horiba Jobin Yvon) with a 532-nm laser.
SAXS measurements were carried out in the capillary transmission mode at the SAXS/wide-angle X-ray scattering beamline of the ANSTO—Australian Synchrotron.
NAP-XPS was conducted at the TLS 24A1 beamline of the National Synchrotron Radiation Research Center. The chamber vacuum was controlled at 1 mbar.
QCM with dissipation monitoring measurements were performed using 5.0-MHz AT-cut quartz crystals precoated with gold (14.0-mm diameter; Biolin, QX301). Zn powders (40–60 nm and purity of 99%; Sigma-Aldrich) or NaV3O8 powders were mixed with poly(vinylidene fluoride) (Sigma-Aldrich) at a weight ratio of 9:1 in 1-methyl-2-pyrrolidinone (Sigma-Aldrich; purity, ≥99.5%) to form a homogeneous slurry. The slurry was spin-coated onto the quartz crystals at 8,000 rpm and subsequently dried in a vacuum oven at 40 °C for 12 h. The QCM measurements were carried out in a flow-cell configuration. Initially, a baseline was established by flowing ultrapure H2O at a rate of 100 μl min−1 until a stable frequency signal was obtained, where water served as the reference adsorbed species. Subsequently, the electrolyte was switched to the 1.8-mol%-DEE-containing solution, and the resulting frequency changes were recorded. The adsorbed mass of DEE was calculated from the frequency shift using the Sauerbrey equation. The fundamental resonance frequency and its overtones (third, fifth and seventh) were recorded simultaneously, together with the corresponding dissipation factors. Data acquisition and analysis were performed using QSoft401 (v.2.8.4.948) and QSense Dfind (v.1.2.8) software43.
The cycled electrodes for FTIR, X-ray diffraction and scanning electron microscopy (SEM) characterization were harvested by disassembling the cells in ambient air. The electrodes were rinsed three times with ultrapure water to remove residual electrolytes and then dried in ambient conditions for 24 h before analysis.
For the XPS measurements, the cells were disassembled in an Ar-filled glovebox (Ar gas; BOC Australia; 99.999%; H2O and O2 content,
X-ray diffraction measurements of Zn electrodes were conducted using Rigaku Ultima IV with monochromatic Cu Kα radiation, scanning between 5° and 80° at a rate of 10° min−1. Ex situ XRPD measurements of NaV3O8 electrodes was performed at the Powder Diffraction beamline of the ANSTO—Australian Synchrotron, using an X-ray beam with a wavelength of 0.68880 Å. Diffraction patterns were collected using a MYTHEN microstrip detector with an exposure time of 30 s.
Morphological images and surface roughness of electrodes were obtained using SEM (Hitachi SU7000) and a confocal microscope (Olympus LEXT OLS5000 profilometer).
XPS measurements of Zn electrodes were performed on Thermo Scientific Nexsa using monochromic Al Kα radiation.
Flammability tests of electrolyte solutions were conducted by igniting a glass fibre separator soaked with 200 µl of electrolyte using a gas lighter for 3 s. The self-extinguishing time, defined as the time required for flame extinction, was recorded, with each test repeated three times.
Titration tests were performed to evaluate the pH-buffering behaviour of the electrolyte solutions. In each measurement, 3 ml of electrolyte solution was placed in a glass vial and stirred at 100 rpm at 25 °C. A 0.1 mol l−1 of NaOH aqueous solution was then added stepwise in increments of 30 µl using a pipette (10–100 µl; Thermo Fisher Scientific), and the pH value was recorded after each addition using a calibrated pH meter.
The amounts of dissolved vanadium in separators were determined by inductively coupled plasma mass spectrometry (Agilent 8900x QQQ-ICP-MS).
In situ and operando physicochemical characterizations of zinc cells
The in situ observation of the adsorption layer on the Zn surface was conducted using the ATR-SEIRAS technique. Specifically, a thin Zn film with a thickness of 50 nm was deposited on an Au@Si wafer (Shanghai Yuanfang; thickness of 500 μm and lateral dimensions of 1.1 × 0.9 cm2). The obtained Zn@Au@Si wafer was used as the working electrode, with an Ag|AgCl reference electrode and a Pt counter electrode, all assembled in a custom-designed spectro-electrochemical cell. The ATR-SEIRAS measurements were conducted using a Nicolet iS50 FTIR spectrometer, equipped with a narrowband MCT-A detector and an in situ infrared optical accessory (SPEC-I, Shanghai Yuanfang) at an incidence angle of 45°. For static adsorption in the absence of an external electric field (Fig. 3c and Supplementary Figs. 19 and 21), the background spectrum was collected before the injection of the electrolyte solution. Continuous spectral acquisition was then conducted until an equilibrium state was observed. For the Zn plating observation (Fig. 3d and Supplementary Figs. 27 and 28), the background was set as the equilibrium adsorption state. The plating current density was set to be 0.5 mA cm−2.
Operando synchrotron-based XRPD was performed at the Powder Diffraction beamline of the ANSTO—Australian Synchrotron. Zn||NaV3O8 CR2025 coin cells with 4-mm-diameter windows were used to ensure synchrotron beam transmission. The customized Zn||NaV3O8 coin cells (assembled as described above) were tested at 500 mA g−1 between 0.3 V and 1.6 V and 25 °C. Diffraction patterns were collected using a MYTHEN microstrip detector with an exposure time of 30 s, and data were recorded at 3-min intervals.
The wavelength of the synchrotron X-ray beam for operando XRPD experiments was 0.59040 Å, whereas ex situ XRPD experiments used a wavelength of 0.68880 Å. This variation does not affect the reliability of the results, as both experiments provided sufficiently strong X-ray intensity to detect the species present on NaV3O8 electrodes with sufficient sensitivity.
Statistical analysis
The statistical analysis was performed using in-house developed code written in Python language (Python v.3.10.13). Specifically, the Pearson correlation coefficient was calculated using the ‘corrcoef’ function from the Numpy library (v.1.26.2)44. Mutual information was calculated with the ‘mutual_info_regression’ function from the Scikit-learn library (v.1.3.0)45. Feature importance was calculated using the ‘permutation_importance’ function in conjunction with the ‘RandomForestRegressor’ machine learning algorithm, both implemented in the Scikit-learn library (v.1.3.0) with default parameters45. The equation between CE and four descriptors was obtained by fitting a Ridge Regression, as implemented in the Scikit-learn library as well.
Computational methods
All MD simulations were performed using the GAFF force field46. The ACPYPE was used to obtain the GAFF force field topology47. The simulation box size was 5 × 5 × 5 nm3 for all simulation models, which consisted of Zn2+, OTf− and H2O, without and with ME/DME/DEE molecules. The cut-off distance of 1.2 nm was used for the Lennard–Jones potential. The Coulombic potential was measured using particle mesh Ewald with a cut-off distance of 1.2 nm and Fourier grid spacing of 0.12. All bonds were constrained with the LINCS algorithm and periodic boundary conditions were applied in all directions. The MD simulations were started by running initial energy minimization, followed by 1,500 ps of NVT simulation and 1,500 ps of NPT simulation, with an integration time step of 0.001 ps. All simulation systems were finally maintained at 298 K using the Nosé–Hoover thermostat for 30 ns to collect the simulation data. A time constant of 1 ps was applied for the temperature coupling. The calculations of proton diffusion coefficients, hydrogen bonds and partial density were conducted using GROMACS.
The adsorption energy of water, ME, DME and DEE molecules on the Zn surfaces of (101) and (002) facets was investigated using density functional theory. The density functional theory calculations were implemented using the Vienna ab initio simulation package48,49 with the core and valence electronic interactions being modelled using the projector augmented-wave method50,51. The revised Perdew–Burke–Ernzerhof exchange–correlation functional was used52. The wavefunction was expanded with a kinetic energy cut-off of 500 eV and a Γ k-point were used. The dispersion correction was also considered in this study by using DFT-D3 method53. The adsorption energy (Eads) was calculated using the following equation:
$${E}_{\mathrm{ads}}={E}_{\mathrm{Zn}-\mathrm{surface}+\mathrm{adsorbents}}-{E}_{\mathrm{Zn}-\mathrm{surface}}-{E}_{\mathrm{adsorbents}},$$
(3)
where EZn-surface + adsorbents, EZn-surface and Eadsorbents are the total electronic energies for the Zn surface with adsorbed species, clean Zn surface and adsorbed species (including water, ME, DME and DEE molecules), respectively.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
