Designing electrolyte structure to suppress hydrogen evolution reaction in aqueous batteries

: Aqueous batteries (ABs) have attracted increasing attention because of their inherent safety and low cost. Nevertheless, hydrogen evolution reaction (HER) at the anode presents severe challenges for stable and safe operation of ABs. Instead of passivating the anode surface to hinder HER kinetics, a novel design strategy is proposed here to suppress the HER via alternating its thermodynamics pathway. By adding a hydrogen bond acceptor, dimethyl sulfoxide (DMSO), the onset potential of HER can be delayed by as much as 1.0 V (on titanium mesh). Spectral characterization and molecular dynamics simulation confirm that the formation of hydrogen bonds between DMSO and water molecules can reduce the water activity, thereby suppressing the HER. This strategy has proven to be universal in expanding the electrochemical window of aqueous electrolytes.

reduction on the anode, which kinetically prevents sustained reduction of both water and anion, the electrochemical stability of the aqueous electrolyte was improved. However, the onset potential for HER was only reduced by 0.2 V compared to the low-concentration electrolyte (1 m LiTFSI in H 2 O). Although high concentration aqueous electrolytes with different salt species (e.g. nitrate, perchlorate, chloride, acetate and etc.) have been studied [14][15][16][17][18][19] , their inhibitory effects on HER are not satisfactory.
Besides, the suppression of HER in those systems typically relies on the formation of a surface passivation layer, which prevents on the direct contact between the anode and the aqueous electrolyte and avoid excessive water reduction reactions kinetically. Not only extensive optimization of electrolyte formulations is needed to form an effective passivation layer on the anode, but also the passivation layer itself can break to expose new anode-electrolyte interfaces due to anode volume changes in charge/discharge cycles. Besides, using highly concentrated salts (including some highly fluorinated salts, e.g. LiTFSI, lithium bis(pentafluoroethanesulfonyl)imide (LiBETI) 12 , lithiumtri-fluoromethane sulfolate (LiOTF) 13 ), may significantly increase the toxicity and cost of electrolyte, which compromises the economic benefits, and plagues the large-scale application of ABs. There are also other strategies to suppress HER, including the current collector design 20 , anode surface protection 21 , molecular crowding electrolytes 22 , etc. Although some excellent results have been achieved via these strategies, the strict experimental conditions and high-cost hinder the large-scale applications. Taking account of the pitfalls of strategies mentioned, a simple yet universal strategy to decrease the inherent activity of H 2 O in HER is highly desirable.
Here, a novel strategy from the perspective of thermodynamics is firstly introduced to reduce the HER potential of ABs, by manipulating the hydrogen bond (H-bond) structure of water molecules. It was found out that by adding DMSO with a mole fraction of 0.5 into aqueous electrolytes (e.g. 2 m NaClO 4 in H 2 O), the HER potential can be reduced by 1.0 V (from 0.6 V to 1.6 V, versus Ag/AgCl with saturated KCl). Spectral characterization and molecular dynamics simulation (MD) demonstrate that all the hydrogen atoms in water molecules are bound in the DMSO-H 2 O H-bond network, which effectively reduces the activity of water molecular. In addition, DMSO enter the cation solvation sheath directly to replace part of coordinating water molecules, thus inhibiting their involvement in interfacial electrochemical reactions. Benefiting from the greatly suppressed HER in aqueous electrolytes via this strategy, efficient utilization of the V 2+ ↔V 3+ reaction of the Na 3 V 2 (PO 4 ) 3 anode material, which cannot proceed in conventional aqueous electrolytes, can be achieved; in addition, the Coulombic efficiency of Zn metal anode can be largely improved.

Results
Electrolyte structure design. Fig. 1a displays the schematic of HER in aqueous electrolytes. During the charging and discharging process, water molecules migrates to the electrode surface together with cations, and subsequently undergo a deprotonation process to generate hydrogen. This deprotonation step not only changes the chemical bond in water molecules, but also disrupts the intermolecular H-bond. Therefore, HER of water molecules is a competitive process with the H-bond formation process in the solution [22][23][24][25]46 . In principle, modulating the H-bond structure of water can reduce the activity of water, thus suppressing HER in ABs.
DMSO, as a highly polar aprotic solvent, can form strong H-bonds with water molecules and significantly change the original H-bond structure of water molecules 26,27 . As shown in  and DMSO within each solvation sheath around Na + along the simulation trajectories. It is unsurprising to find that, DMSO has penetrated into each sheath as the averaged number of water is reduced from 6 to 2 in the primary sheath and from 18 to 9 in the secondary sheath, respectively. The replacement has been made with 3 DMSO in primary sheath and 6 DMSO in secondary sheath (Fig. 2e, Supplementary Figs. 7, 8 and 9). Carrying lesser water molecules in Na + solvation sheath would also be beneficial for HER suppression. In order to gain further insights into the mechanism of in the H 2 O-DMSO adduct from the water-water adduct (Fig. 2f). This further supports our conclusion that the thermodynamic pathway of HER can be effectively modulated by the H-bond network with DMSO addition. Electrochemical performance test. The overall electrochemical stability window of 2 m-0.5 is expanded to 3.1 V when using titanium meshes as the negative and positive current collectors (Fig. 3a). Particularly, the HER onset potential is pushed from -0.6 V 9 to -1.6 V versus Ag/AgCl. A low hydrogen evolution potential can well satisfy reactions that cannot occur in traditional aqueous electrolytes. For example, although Na 3 V 2 (PO 4 ) 3 (NVP) can undergo multivalent state transitions (V 2+ ↔V 3+ ↔V 4+ ) (Fig. 3b), the V 2+ ↔V 3+ redox process was excluded in traditional aqueous electrolyte systems, because its reaction potential (-1.2 V versus Ag/AgCl) is much lower than the HER potential on NVP (-0.8 V versus Ag/AgCl in 2 m NaClO 4 ). Nevertheless, NVP-based electrode can undergo a highly reversible V 2+ ↔V 3+ reaction in the 2 m-0.5 electrolyte (Fig. 3c). The discharge capacity of NVP reaches as high as 180 mAhg -1 at current density of 0.5 Ag -1 in the 2 m-0.5 electrolyte (Fig. 3d). To the best of our knowledge, this is the first time that NVP achieves its maximum discharge capacity in an aqueous electrolyte.  Fig. 3g. Owing to the greatly improved stability of the electrolyte against HER, the battery achieves an output voltage of 1.7 V and delivers a reversible discharge capacity of ~90 mAhg -1 (at 1 A g -1 , based on the total mass of anode and cathode active materials) and a capacity retention of ~88% after 100 cycles (Fig. 3h). To further verify that this methodology is universal in suppressing HER side reactions, its application in aqueous zinc metal batteries (AZMBs) is studied. AZMBs are very attractive energy storage solutions due to their safety, low-cost, etc. Nevertheless, the competitive HER on Zn anode during the charging cycle raises serious challenge to develop AZMBs with high Zn plating/stripping Coulombic efficiencies (CE) and good cycle stability for practical implementation. In contrast to the clear H 2 evolution and rough Zn plating morphology in the 2 m Zn(CF 3 SO 3 ) 2 electrolyte as shown in Fig. 4a-c, a highly smooth Zn surface without H 2 evolution was observed in the 2 m Zn(CF 3 SO 3 ) 2 -0.5 electrolyte (2 m Zn(CF 3 SO 3 ) 2 added dimethyl sulfoxide (DMSO) with a mole fraction of 0.5) (Fig. 4d-f). More importantly, the average CE of Zn-Ti cell (as calculated from the ratio of Zn removed from the Ti substrate to that of deposited during the same cycle) in the first ten cycles is 99.02% in 2 m Zn(CF 3 SO 3 ) 2 -0.5, which remains stable even after 500 cycles (Fig. 4g, h, black line). In contrast, in 2 m Zn(CF 3 SO 3 ) 2 , the average CE of the Zn-Ti cell in the first ten cycles is only 85.60% (Fig. 4h, red line) and the CE values fluctuates greatly after ≈13 cycles, which is mainly due to the short circuit of the cell (Fig. 4i). The stability of Zn metal anode was also evaluated by long-term galvanostatic cycling of the symmetric Zn-Zn cells (Fig. 4j). Although with a slightly higher polarization, all cells which use 2 m Zn(CF 3 SO 3 ) 2 -0.5 display more stable cycle performance than those using the 2 m Zn(CF 3 SO 3 ) 2 electrolyte.
After cycling for ~70 h at 1 mA cm −2 , the polarization voltage drops suddenly in the DMSO-free Zn-Zn cell, which can be ascribed to short circuit induced by Zn dendrite. On the contrary, the Zn-Zn cell using DMSO-containing electrolyte displays a prolonged cycling stability for more than 1600 h. Even at a high rate of 5 mA cm -2 (0.5 mAh cm -2 per half cycle) Zn-Zn cells still show excellent cycling stability (Supplementary Fig.   10).

Discussion
In this work, we demonstrate that the manipulation of intermolecular H-bonds is an effective and universal strategy to suppress the HER side reactions in aqueous electrolytes. DMSO, as a co-solvent, will participate in the cationic solvation structure, replacing the water molecules in the solvation sheath. At the same time, it can work as a H-bond acceptor to interact with water molecules inside and outside of the solvation sheath.  Table 1). Water molecules were described with TIP3P potential model 37 . A total 100 ns MD production run was performed for each system after minimization run and heating procedure to reach 300 K. The sampling strategies were carried out under NpT condition with 2fs time step. The electrostatic interactions were described with the particle mesh Ewald method (PBE) 38 and cutoff threshold of 10 Å was applied for non-bonded Van der Waals interaction.
The periodic boundary conditions (PBC) and SHAKE algorithm were turned on during the production run 39   Analysis was performed based on 10-100ns simulation trajectories.
A quantum chemistry study of the reduction reaction was performed using Gaussian 16 to distinguish the energy profile of H 2 evolution from water in different H-bond systems captured from MD simulations. Density functional B3LYP 40 with 6-311G* and D3BJ 41 dispersion correction was used for geometrical optimization. The double-hybrid method B2PLYPD3 42 with def2TZVP basis set was applied for thermodynamic correction. The Minnesota functional M052X 43 with equivalent basis set was used to estimate the solvation effects. The SMD solvation model using water solvent parameters was applied in all needed calculation scenarios.
The half cycle of H 2 evolution could be described as: The free energy for the complex of interest denoted as water/DMSO Hbond cluster in solution could be calculated according to the following expression: The reduction potential was calculated as: Where F is Faraday's constant (1.0 eV/V) and SHE is the absolute potential of the standard hydrogen electrode 44 . As we focused on the difference of reduction potential, SHE part was eliminated from the deduction calculation, so there is no need for interpretation of the value of SHE.