Exploring optimal formulation for low-temperature aqueous electrolyte
A series of TMP/water hybrid electrolytes with 2 M zinc trifluoromethanesulfonate (Zn(OTf)2) were prepared, where the volume percentage of TMP ranges from 0%, 5%, 10%, 20%, 40%, 60%, 80–100%, and the corresponding electrolyte is marked as TMP–0, TMP–5, TMP–10, TMP–20, TMP–40, TMP–60, TMP–80 and TMP–100, respectively. After 3 h of resting at − 50 oC, the TMP–40, TMP–60 and TMP–80 can maintain the liquid state without deposit or phase separation, whereas crystallization/solidification was observed in the other counterparts (Fig. 2a). Further, the differential scanning calorimeter (DSC) curves show the freezing points of the hybrid electrolytes are lowered to − 56.8°C as the TMP content reaches 40% (Fig. 2b and Supplementary Fig. 1), which can be explained by the breakage of H-bonds network in water molecules by TMP co-solvent. On the other hand, the ionic conductivity negatively correlates to the content of TMP in the temperature ranging from − 30 oC to 60 oC (Fig. 2c, Supplementary Fig. 2 and Supplementary Table 1), possibly due to a slight increase in electrolyte viscosity. However, the TMP–0 exhibits a sudden drop in ionic conductivity below − 30 oC, because of the high freezing point (− 34.8 oC in DSC) that causes the electrolyte solidification. In sharp contrast, the TMP–40 exhibits a slow decline in ionic conductivity as the temperature decreases and achieves a high ionic conductivity of 0.85 mS cm− 1 even at − 50°C (> 0.1 mS cm− 1)26, several orders of magnitude higher than that of TMP–0.
To explore the molecular interaction within the hybrid electrolytes, a series of spectroscopic characterizations were performed. The Raman peaks associated with O–H stretching vibration of water molecules are visible within the range from 3100 to 3800 cm− 1, which can be divided into three peaks including strong, weak and non H-bonds (Fig. 2d, e, and Supplementary Figs. 3–4)7,30. It was found that the probability of non-H bonds increases with TMP content, while the change of strong-H bonds shows the opposite trend (Fig. 2f). Notably, as the TMP content was increased to 40%, the percentage of non H-bond almost reaches the maximum. Moreover, the Fourier transform infrared (FTIR) results in Supplementary Fig. 5 show that the O–H and C–H stretching vibration modes experience significant blueshifts and redshifts with the increase of TMP, respectively, largely ascribed to the breakage of H-bond in water along with the H-bond formation between TMP and water in the hybrid electrolyte8. This can be further confirmed by the 1H nuclear magnetic resonance (NMR) spectra (Fig. 2g and Supplementary Fig. 6), wherein the 1H from H2O and TMP both chemical shift to low field with the increase of TMP. These results reveal that the TMP can interact with water to reform H-bonds, during which the H-bonds in water were largely destroyed, thus affording a low freezing point and high ionic conductivity of the hybrid electrolyte at low temperatures. To maximally inherit the unique merits of aqueous electrolyte, TMP–40 is considered as the optimal electrolyte formulation for low-temperature ZMBs. Also, the TMP–40 endows the separator with high fire retardance (Supplementary Fig. 7), indicating the hybrid electrolyte is safe enough to operate.
Solvation Structure And Sei Characterization
Since the SEI components highly depend on the solvation sheath of Zn2+, we furthered the study of Zn2+ solvation structure in a series of TMP/H2O electrolytes through theoretical calculations and experimental characterizations. The Raman characterizations shows the SO3 stretching band in the OTf−–anions experiences a gradual shift with the increase of TMP concentration (Fig. 3a and Supplementary Fig. 8). The broad peaks can be well fitted into three peaks at ~ 1028 cm− 1, ~ 1033 cm− 1 and ~ 1040 cm− 1, corresponding to the free anion (FA, OTf−), solvent-separated ion pairs (SSIP, Zn2+–(H2O)x(TMP)y–OTf−) and contact ion pairs (CIP, Zn2+–OTf−), respectively31,32, as shown in Supplementary Fig. 9. By calculating the peak area ratio, the CIP percentage increases with the increase of TMP concentration and reaches the maximum value of 51.92% with 40% of TMP (Supplementary Fig. 10), indicating OTf− anion is involved in the Zn2+ solvation sheath. Afterwards, the CIP content decreases with the increase of TMP, possibly due to the strong binding of TMP and Zn2+ that causes more TMP to enter the Zn2+ solvation sheath by substituting partial OTf− anions. Meanwhile, there is a V-shape relationship between FA percentage and TMP concentration, where the lowest FA ratio is 6.59%, suggestive of more OTf− involved in the solvated structure of Zn2+. This also confirms that TMP–40 is the optimized electrolyte formulation for in-situ formation of favorable SEI to suppress side reactions and facilitate Zn2+ transport. For the Raman spectra of the TMP, the P–O–(C) symmetric stretching vibration gradually blueshifts as the increase of TMP (Fig. 3b and Supplementary Fig. 11), indicating more TMP participates in the Zn2+ solvation shell33,34, according well with the above results. This is also supported by the higher binding energy of Zn2+–TMP complex (− 200.36 KJ mol− 1) compared with Zn2+–H2O complex (− 104.54 KJ mol− 1), as shown in Supplementary Fig. 12.
To ascertain the coordination number of anions and solvents in the solvation sheath of TMP–40, molecular dynamic (MD) simulations were carried out. The numbers of Zn2+, OTf−, TMP and H2O in the hybrid electrolytes are summarized in Supplementary Table 2. As shown by the snapshots from the simulated solvation structure (Supplementary Figs. 13–14), some water molecules are squeezed out of the Zn2+ solvation shell in the TMP–40 electrolyte and partially replaced with TMP solvent and OTf− anions. According to the radial distribution functions (RDF) in Fig. 3c, the Zn–O peak in OTf−, TMP and H2O correspond to the distance of 0.19, 0.25 and 0.23 nm, respectively, further validating that the OTf−, TMP and H2O molecules incorporate into the first solvation shell of Zn2+. Accordingly, the respective coordination number was calculated to be 0.85, 0.14 and 5.01, constituting a CIP–type solvation shell of Zn2+[H2O]5.01[TMP]0.14[OTf−]0.85, which favors the in-situ formation of SEI on Zn surface through reductive decomposition. Note that the small amount of TMP and OTf− involved in the solvation shell is beneficial to form a thin SEI layer, which facilitates fast Zn2+ transfer. Moreover, the Zn2+ desolvation energy of TMP–0 and TMP–40 can be obtained by extracting the respective Rct before cycling, where no SEI was formed on Zn surface. As shown in Supplementary Fig. 15, the addition of TMP causes a slight increase in the energy barrier for dissociation of Zn2+, largely ascribed to the strong interaction of Zn2+–TMP and Zn2+–OTf−, which in turn allows for the stepwise formation of ZnF2–Zn3(PO4)2.
With the TMP–40 as the optimal electrolyte formulation, the linear sweep voltammetry (LSV) measurements were first performed to examine the electrochemical stability. It was found that the TMP–40 electrolyte can effectively suppress the water decomposition over a wide electrochemical window and prevent Zn surface corrosion (Supplementary Figs. 16–17). Moreover, the disappearance of the cathodic peak at ~ 0.1 V accompanied with the decrease in current density after five cycles also indicates the SEI was formed at the initial plating and can inhibit the hydrogen evolution reaction (HER) (Supplementary Fig. 18). After 40 cycles of stripping/plating of Zn metal, X-ray diffraction (XRD) peaks corresponding to the zinc triflate hydroxide hydrate (ZnxOTfy(OH)2x−y·nH2O, ZOTH) were detected on the Zn surface in TMP–0 (Supplementary Fig. 19), which can largely restrict the transport of Zn2+ and lead to dendrite growth31,35. In contrast, no byproduct was observed in TMP–40. X-ray photoelectron spectroscopy (XPS) with Ar ion sputtering was further employed to determine the depth distribution of composition in SEI formed on Zn surface. As shown in Fig. 3d, the top SEI layer (before sputtering) is rich in –CF3 species (~ 688.8 eV) and inorganic ZnF2 (~ 684.1 eV) with a tiny amount of Zn3(PO4)2 (~ 134.3 eV). Accordingly, the lattice fringes in the high-resolution transmission electron microscopy (HRTEM) corresponding to the planes of ZnF2 and Zn3(PO4)2 were clearly observed with uniform distribution (Fig. 3e, f and Supplementary Fig. 20), consistent with the XPS results. Based on the previous reports28,31, the –CF3 species arises from either the incomplete reduction of OTf− or the residual salt on Zn surface, while ZnF2 and Zn3(PO4)2 are attributed to the decomposition product of Zn2+–OTf− and Zn2+–TMP complexes. As the sputtering continues, the peak intensity of Zn3(PO4)2 distinctly becomes stronger along with the decrease in ZnF2 peak (Fig. 3d and Supplementary Fig. 21). After 310 s of sputtering, the Zn3(PO4)2 gradually becomes the major composition in the SEI. In sharp contrast, no F or P signals related to ZnF2 or Zn3(PO4)2 was detected in TMP–0 (Supplementary Fig. 22). The XPS analyses provide strong evidences that the TMP–40 electrolyte favors the in-situ formation of gradient interlayer on Zn surface, where ZnF2 and Zn3(PO4)2 dominates the surface layer and inner layer, respectively.
The Kinetic Behavior Of Bivalent Zn On The Electrode/electrolyte Interface
As aforementioned, the SEI with rapid Zn2+ transport kinetics and low Zn2+ desolvation energy is beneficial for ZMBs to stably work at low temperatures. Thus, we performed temperature-dependent electrochemical impedance spectroscopy (EIS) of Zn||Zn cells at temperature ranging from 20 oC to − 30 oC in TMP–0 and TMP–40 after 40 cycles (Supplementary Fig. 23), where a dense SEI should be formed in TMP–40. The charge transfer resistance (Rct) and the resistance associated with Zn2+ crossing SEI (RSEI) can be extracted from the semicircles in mid-frequency region and the high-frequency region, respectively24–26. By Arrhenius-fitting Rct and RSEI over 1000/T, the activation energy of each interface process were obtained, as shown in Fig. 4a, b. Compared with TMP–0, the desolvation energy of Zn2+ in the TMP–40 was greatly reduced (70.2 KJ mol− 1 vs 54.8 KJ mol− 1), indicating the outer ZnF2 facilitate Zn2+ desolvation, agreeing with the reported results36,37. Moreover, the activation energy for Zn2+ transport through the SEI in TMP–40 (Ea,SEI=52.7 KJ mol− 1) is significantly lower than in TMP–0 (Ea,SEI=64.3 KJ mol− 1). This can be well explained by the density functional theory (DFT) calculation results that the Zn3(PO4)2 delivers a much smaller migration energy barrier for Zn2+ (0.38 eV) (Fig. 4d) and higher affinity with Zn2+ (− 1.15 eV) (Fig. 4c) compared with those of ZnF2 (1.12 eV for Zn2+ transport and weak binding energy of − 0.85 eV with Zn2+). That is, the rich Zn3(PO4)2 in the inner SEI serves as the dominant channels for the desolvated Zn2+ across the SEI to deposit on the Zn surface, which can facilitate fast Zn2+ conduction to mitigate voltage polarization under cold environments38,39. Moreover, we prepared SEI containing single ZnF2 or Zn3(PO4)2 to confirm their respective role (Supplementary Fig. 24). Besides, the interface impedance of Zn||Zn cells in the TMP–40 electrolyte can remain stable after 500 cycles, indicating the stable interface due to the formation of ZnF2–Zn3(PO4)2 interlayer (Fig. 4e). In sharp contrast, the TMP–0 electrolyte shows a sharp decrease in the interfacial impedance after 100 cycles, possibly due to the dendrite growth that causes the cell short circuit (Fig. 4f).
In addition, a high mechanical integrity is indispensable for SEI to ensure the long-term cycling and high-capacity plating. To this end, in-situ optical microscopy was performed to observe the morphology evolution at different plating stages. As shown in Fig. 4g, the TMP–40 electrolyte enables the dense deposition without dendrite formation during the whole deposition process and can maintain a smooth surface even at a colossal loading of 50 mAh cm− 2. While for the case of TMP–0, uneven spots appear at 10 mAh cm− 2 and gradually evolve into discontinuous islands as the plating capacity increases, which is consistent with the scanning electron microscopy (SEM) results (Supplementary Fig. 25). The striking contrast in morphology evolution demonstrates that the gradient SEI can effectively suppress the dendrite growth and is highly stable to accommodate the high-loading Zn plating, largely ascribed to the strong bulk modulus of Zn3(PO4)237. Taken together, we reasonably conclude that the rich ZnF2 on the top layer of SEI favors the desolvation of Zn2+ and the robust Zn3(PO4)2 predominating the inner SEI layer facilitates rapid Zn2+ transport. With these admirable characters, it is expected that the as-formed gradient ZnF2–Zn3(PO4)2 SEI can guarantee stable and long-term cycling of Zn metal at low temperatures.
Electrochemical Performance Of Zn Metal Anodes Under Harsh Conditions
It is worth mentioning that these features of the gradient SEI can allow the symmetric Zn cells to stably cycle in the TMP–40 electrolyte at a high current density of 5 mA cm− 2 at 25 oC and 45 oC (Fig. 5a and Supplementary Fig. 26). In sharp contrast, the cells in TMP–0 quickly failed due to severe Zn dendrite formation and aggravated side reactions. This also manifests that the gradient ZnF2–Zn3(PO4)2 SEI is ultra-stable against the high-temperature and high-rate cycling, underscoring the significance of SEI formation on the interface. Then the galvanostatic cycling stability of Zn metal in the TMP–40 electrolyte was studied at low temperatures with different rates. When the operation temperature was fixed at − 30 oC, the cell in TMP–40 exhibits a stable voltage profile at 2 mA cm− 2 with an ultralong cycling life up to 3600 hours, which is around 40–fold improvement in cycle life (Fig. 5b). With the superior kinetics in the electrolyte/electrode interface, the TMP–40 electrolyte enables the symmetric cells to operate over long-term cycles with high rates ranging from 5 mA cm− 2 to 15 mA cm− 2 through a transient activation (Supplementary Fig. 27).
Then the cycling temperature was decreased to − 50 oC. Not surprisingly, the cell in TMP–0 cannot work due to the electrolyte solidification (Fig. 5c). While the TMP–40 electrolyte achieves an ultralong lifespan at 0.4 mA cm− 2 and 0.4 mAh cm− 2 without obvious fluctuation in overpotential over 6000 hours (~ 8 months). As the discharge depth was increased to 1 mAh cm− 2, it is amazing to find that the cell can still maintain an impressive stability without voltage fluctuation over 6000 hours (Supplementary Fig. 28). These observations convectively demonstrate that the as-formed gradient ZnF2–Zn3(PO4)2 SEI with the unique configurations can accelerate the rapid Zn2+ desolvation and conduction at low temperatures, which guarantees the ultra-stable cycling with a negligible polarization at rather extreme conditions (low temperatures with high rates). Notably, the Zn||Zn symmetric cells in TMP–40 achieve a rather competitive cumulative capacity over a wide temperature range (Fig. 5d and Supplementary Table 3), far outperforming those of reported low-temperature aqueous Zn metal anodes8,13,17,19,40,41. Both top and side views reveal that the surface of Zn electrode after 100 plating/stripping cycles in TMP–40 is highly homogeneous and tightly packed, while the one with TMP–0 exhibits severe cracks (Fig. 5e, f), indicating the gradient SEI layer can effectively suppress the Zn dendrite growth.
The reversibility of Zn plating/stripping was further studied by calculating the Coulombic efficiency (CE) of Zn metal onto titanium (Ti) substrate. At − 30°C with a current density of 1 mA cm− 2 and a capacity of 0.5 mAh cm− 2, the CE in TMP–40 electrolyte quickly increases to 99% within 20 cycles and stabilizes at 99.9% over 3800 cycles along with flat voltage profiles (Fig. 5g and Supplementary Fig. 29). Conversely, the cell with TMP–0 electrolyte exhibits a low CE of around 60% and quickly short circuited at the 9th cycle due to dendrite formation. These contrasts can be maximized at different temperatures ranging from 45°C to − 50°C (Supplementary Fig. 30), well illustrating that the gradient SEI layer is stable against the severe side reaction and maintains superior kinetics at low temperatures.
Electrochemical Performance Of Zn–kvoh Full Cells Under Practical Conditions
To evaluate the practical applications of TMP–40 electrolyte, KVOH was employed as cathode to pair with Zn metal for full cells in the TMP–40 electrolyte due to its superior kinetics (Supplementary Fig. 31). The Zn–KVOH full cell with TMP–40 electrolyte delivers an initial capacity of 329.1 mAh g− 1 at 5 A g− 1 and retains a capacity of 323.6 mAh g− 1 after 2300 cycles at room temperature (Fig. 6a), corresponding to a capacity retention of 98.3%. In contrast, the cell with TMP–0 electrolyte exhibits a slightly higher initial capacity of 344.4 mAh g− 1 at 5 A g− 1 possibly due to a higher ionic conductivity, but quickly dropped to 211 mAh g− 1 after 500 cycles, along with a larger polarization in the charge–discharge voltage profiles (Supplementary Fig. 32). The significant performance improvement is also revealed with high loadings of KVOH (6.37 mg cm− 2 and 17.6 mg cm− 2), at various rates (1 A g− 1, 2 A g− 1 and 10 A g− 1) or even at a higher temperature (45 oC, 5 A g− 1), as shown in Supplementary Figs. 33–35. Impressively, with a high areal loading of KVOH up to 17.6 mg cm− 2, the full cell still maintains an areal capacity of 4.37 mAh cm− 2 after 100 cycles, which meets the requirements of a typical commercial Li-ion battery (4.0 mAh cm− 2)42. In view of the inspiring performance, we further evaluated the application of TMP–40 electrolyte in practical situation by controlling lean electrolyte and low Zn excess. As shown in Fig. 6b, when the KVOH loading increases to 33.75 mg cm− 2, the cell still delivers a superhigh initial areal capacity of 9.42 mAh cm− 1 with lean E/C (6.76 µL mAh− 1, the ratio of electrolyte volume to capacity) ratio and low N/P (3.1, the ratio of negative to positive). The corresponding energy density is calculated to be 251.2 Wh kg− 1 (based on the KVOH mass) with a high capacity retention of 93.3% after 50 cycles. The outstanding performance can be ascribed to the superior kinetics and great robustness of the ZnF2–Zn3(PO4)2 SEI that can allow large amounts of Zn2+ to repeatedly strip and plate.
Electrochemical Performance Of Zn–kvoh Full Cells At Low Temperature
By virtue of the favorable SEI formation, the TMP–40 electrolyte can enable the Zn–KVOH full cells to sustain remarkable long lifespan and prominent stability at subzero temperatures. Specifically, when the temperature was decreased to − 30 oC (Supplementary Fig. 36), the discharge capacity of the cell in the TMP–40 electrolyte remains at 120.6 mAh g− 1 after 4000 cycles at 1 A g− 1, far exceeding that without TMP. Even at a higher rate of 2 A g− 1 and 5 A g− 1, the 40% of TMP addition can prompt the cell to charge/discharge reversibly over 10000 cycles and 1500 cycles, respectively. In stark contrast, the cell with TMP–0 electrolyte failed to work, underscoring the critical role of the gradient SEI formed in the TMP–40 electrolyte. These contrasts are more evident at − 50 oC, where the cell with TMP–0 cannot provide any capacities due to the electrolyte solidification. Comparatively, the TMP–40 electrolyte can render the full cell to deliver a stable capacity of 50.8 mAh g− 1 at 0.5 A g− 1 over 12000 cycles (Fig. 6c). The fluctuations at the 800th, 5000th and 9800th cycle are ascribed to the sudden power outages during the long-term test, which also confirm the good temperature adaptivity of the gradient SEI formed in the TMP–40 electrolyte. Compared with the published works on electrolyte modification for aqueous Zn batteries, this work is undoubtedly prominent in achieving long-cycle stability over a wide temperature range (Fig. 6d, Supplementary Tables 4–5)4,8,12,14–17,19,22,40,41,43−46. Post-mortem analyses show that the Zn metal anode after 1000 cycles in the TMP–40 electrolyte can maintain a much flatter and denser surface (Supplementary Fig. 37). The significant performance improvement achieved in the TMP–40 electrolyte at low temperatures strongly validates the gradient SEI with the favorable kinetics can effectively suppress the Zn dendrite growth and ensure the superb cycling behaviors in full cells under extreme conditions.
The exceptional performance of the ZMBs inspired us to further evaluate the low-temperature performance of pouch cell with a size of 4.3 × 5.6 cm in the TMP–40 electrolyte. As shown in Fig. 6e, a reversible capacity of 340.5, 159.4, 122.2 mAh g− 1 was achieved at 25 ℃ (room temperature), − 30 ℃ and − 50 ℃, respectively. Notably, the pouch cell achieves a superior cycling stability at a low temperature of − 30°C with a high capacity retention of 88.6% after 1200 cycles at 0.25 mA cm− 2 (Fig. 6f). Even at − 50°C, it can stably cycle for 180 cycles with nearly 100% of capacity retention (Supplementary Fig. 38). Moreover, five series connected pouch cells with a voltage of ~ 5 V can drive an electro-calculagraph normally working for more than 3 min at an extreme temperature of − 50°C (Fig. 6g and Supplementary Fig. 39), further demonstrating its promising potential in the practical applications under harsh conditions.