Zinc Complex Based Multifunctional Reactive Lithium Polysulde Trapper Approaching Its Theoretical Eciency

The “shuttle effect” of soluble lithium polysuldes (LPS), which causes rapid capacity fading, remains a lingering issue for lithium-sulfur batteries (LSBs). Herein, we report a new type of reactive molecule-based (or molecular) LPS trapper, zinc acetate-diethanolamine (Zn(OAc) 2 ·DEA), which demonstrated a molecular eciency of 1.8 for LPS trapping, approaching its theoretical limit of 2. This is the highest trapping capability among all reported LPS trappers. During discharge the trapped polysuldes are much more thermodynamically favored for reduction compared to the non-trapped ones, while during charge the complex Zn(SLi) 2 ·DEA formed in the previous discharging process can be more easily oxidized due to its lower energy barrier in comparison to Li 2 S, indicating the catalytic effects of Zn 2+ ·DEA on the redox of sulfur species. Zn(OAc) 2 ·DEA is also an excellent binder owing to its multiple intermolecular hydrogen bonds. LSBs using Zn(OAc) 2 ·DEA as a LPS trapper, a binder, and a redox catalyst exhibited excellent long-term cycling stability (with a capacity retention of 85% after 1000 cycles at a rate of 0.5 C) and enhanced rate performance. The work demonstrated the potential of this novel type of multifunctional metal complex-based reactive molecular LPS trappers for high capacity and stable LSBs.


Introduction
The escalating demand for high energy density rechargeable batteries has not only driven the research on the improvement of traditional lithium ion batteries, but also spurred the development of next generation ultra-high energy density battery technologies such as lithium-oxygen and lithium-sulfur batteries (LSBs) 1,2 . In particular, LSBs possessing a high theoretical speci c energy of 2600 Wh kg -1 together with the very low cost of sulfur have attracted extensive attention for energy storage applications in recent years 1,3-6 . However, there are still numerous issues that impede commercializing LSBs. One of them is the so-called "shuttle effect" 7 , whereas soluble long-chain lithium polysul des (LPS, Li 2 S x , 4 ≤ x ≤ 8) intermediates produced during discharge diffuse to and react with the Li anode to form an insoluble and insulating Li 2 S layer on the anode surface [8][9][10] . This causes continuous loss of active sulfur and an increased impedance of the anode surface, leading to fast fading of battery performance. On the other hand, the formation of soluble LPS is also considered necessary for increasing sulfur utilization and thus a higher speci c capacity through a promotion of their subsequent reactions to produce shorter lithium sul des (Li 2 S 2 and Li 2 S) 11 .
To tackle the "shuttle effect", numerous efforts have been devoted to explore various materials that can keep or trap the soluble LPS within the sulfur cathode. Mesoporous carbon materials were rstly employed to trap LPS through the spatial con nement and the physical adsorption of LPS on the large surface area of these carbon materials; however, due to the weak intermolecular interaction, LPS readily diffuse out of the pores and the battery capacity degrades rapidly within tens of cycles 3 . It was found that incorporation of heteroatoms such as N 12 , O 13 and B 14 in carbon materials can signi cantly enhance their LPS trapping capabilities due to the strong polar-polar interactions between these heteroatoms and Page 3/13 LPS. Organic polymers and frameworks containing heteroatoms have also shown good LPS trapping effects 9,15,16 . Many metal compounds such as metal oxides 17 , sul des 18 , nitrides, 19 and carbides 20 exhibit strong polar-polar interactions between the metal cations (or anions) in these compounds and the S x 2-(or Li + ) ions in LPS 21 . Some metal compounds such as metal-organic frameworks (e.g. the Ni-based MOF) 22 and MXenes (e.g. Ti 2 C) 23 have even stronger interactions with LPS to form metal-sulfur chemical bonds.
Nonetheless, since the previously reported LPS trappers are solid aggregates the adsorption of LPS occurs on the surface and thus the quantity of immobilized LPS critically relies on the available surface area of the trappers 24 . Although the surface to volume ratio can be increased laboriously by reducing the particle size or making mesoporous structures 25,26 , there would be always a large portion of inaccessible materials and a large amount of LPS trappers are required to achieve a satisfactory battery lifetime.

Results
Synthesis and characterization of Zn(OAc) 2 ·DEA The Zn(OAc) 2 ·DEA complex was conveniently prepared by mixing zinc acetate and DEA with a molar ratio of 1:1 in ethanol at room temperature without puri cation. Its 1 H NMR spectrum ( Fig. 1A (Fig. S3). Accordingly, we prepared the sulfur cathode slurry using elemental sulfur as active material, Super P as conductor, and Zn(OAc) 2 ·DEA as binder with a mixture solvent of ethanol and water (v:v =1:1), which showed a very good lm forming property, comparable to that prepared using PVDF as binder in N-methyl-2-pyrrolidone (NMP) (Fig. S4). The electrochemical performance of Zn(OAc) 2 ·DEA based electrode was then evaluated in comparison to the PVDF based electrode.
As shown in Fig. 2A, the CV pro le of the Zn(OAc) 2 ·DEA based cathode at a scan rate of 0.1 mV s -1 showed the typical two pairs of cathodic/anodic peaks, corresponding to the two-step reversible redox reaction between elemental sulfur (S 8 ) and Li 2 S 29,30 . The PVDF based cathode also showed two pairs of redox peaks; however, both reduction and oxidation peaks, particularly the second reduction and oxidation peaks, are much delayed. This indicates that the redox kinetics of sulfur species in the Zn(OAc) 2 ·DEA based electrode is enhanced 31,32 . A cell containing Zn(OAc) 2 ·DEA and Super P (1:1) without sulfur was also cycled under the same conditions, which showed that Zn(OAc) 2 ·DEA is stable in the potential range of 1.7-2.8 V that is the typical charge/discharge potential range for LSBs (Fig. S5A).
When the Zn(OAc) 2 ·DEA electrode was cycled on a battery tester at 0.1 C, it exhibited an initial speci c discharge capacity of 1170 mAh g -1 , which increased by 14% compared to the PVDF based electrode (1025 mAh g -1 ) (Fig. 2B). The former also showed a higher capacity in the rst discharge plateau (388 mAh g -1 vs. 340 mAh g -1 ) (Fig. S5B), indicating its higher sulfur utilization 33 . Besides, the dischargecharge plateau gap for the Zn(OAc) 2 ·DEA electrode is only 143 mV, much smaller than that (200 mV) for the PVDF based electrode, implying its fast reaction kinetics 18 , which agrees with the CV results.
Furthermore, no obvious valley between the rst and second plateaus was observed in the discharge curve of the Zn(OAc) 2 ·DEA based electrode, while there is a large overpotential of 39 mV for the PVDF based electrode, which suggests that the interfacial energy barriers for the nucleation and deposition of Li 2 S were reduced in the Zn(OAc) 2 ·DEA based electrode (Fig. S5B) 34 . At 0.2 C, the Zn(OAc) 2 ·DEA based electrode achieved a capacity retention of 95% after 100 th cycles, which is a signi cant improvement compared to that (57%) for the PVDF based electrode (Fig. 2C). Moreover, a much improved rate performance with capacity ~800 mAh g -1 at a rate of 2 C was obtained for the Zn(OAc) 2 ·DEA based electrode due to its enhanced reaction kinetics (Fig. 2D). More importantly, the Zn(OAc) 2 ·DEA based electrode demonstrated an excellent long-term cycling stability with a capacity retention of 85% after 1000 cycles at a rate of 0.5 C, which corresponds to a very low capacity fading rate of 0.015% per cycle ( Fig. 2E). Additionally, the open circuit potential (OCP) of the Zn(OAc) 2 ·DEA based battery was also very stable, achieving 99% retention after ten days, while the PVDF based battery only had 82% OCP retention, indicating the much lower self-discharge of the former (Fig. S5C) 35 .

Page 6/13
The LPS trapping mechanism of Zn(OAc) 2 ·DEA was studied by using 1 H NMR by adding a Li 2 S 6 solution in DOL/DME to a Zn(OAc) 2 ·DEA solution in deuterated methanol (CD 3 OD). When one molar equivalent (eq.) of Li 2 S 6 was added, white precipitates formed immediately (see the movie in S1 and the photo in Fig. S8A). When adding 2 eq. of Li 2 S 6 , the solution rstly became turbid, which is similar to that with addition of 1 eq. of Li 2 S 6 , but soon turned into a clear yellow solution (see the movie in S2 and the photo in Fig. S8B). The 1 H NMR spectrum of the obtained solution of Zn(OAc) 2 ·DEA:2 Li 2 S 6 ( Fig. 1A)  indicating the formation of lithium acetate in the latter (Fig.S2C). Furthermore, the integral ratios of H a' : :H c' are 2:2:3 indicate that the two acetate anions in each Zn(OAc) 2 ·DEA complex were completed removed by 2 eq. of Li 2 S 6 . With further increasing the amount of Li 2 S 6 (up to 7 molar eq.) no noticeable changes in the 1 H NMR spectrum of the resulting mixture of Zn(OAc) 2 ·DEA and Li 2 S 6 were observed (Fig.   S9). Based on the above NMR data, it is most likely that a reaction proposed in Fig. 1B occurred, that is, each Zn(OAc) 2 ·DEA reacted with or trapped two Li 2 S 6 molecules to form a complex Zn(S 6 Li) 2 ·DEA, which agrees well with the result obtained with UV-vis spectroscopy. We also mixed Zn(OAc) 2 ·DEA with 2 eq. Li 2 S in CD 3 OD and observed the formation of white precipitates immediately (see movie S3 and Fig.   S10A), indicating that Li 2 S can readily react with Zn(OAc) 2 ·DEA as well. The 1 H NMR spectrum of the reaction mixture (Fig. S10B) also showed the release of two lithium acetate molecules, indicating that Zn(SLi) 2 ·DEA possibly formed (Fig. S10C). When this reaction mixture was further mixed with 2 eq. of Li 2 S 6 , most of the white precipitates disappeared and a slightly turbid yellow solution resembling that of Zn(S 6 Li) 2 ·DEAwas obtained (movie S4 and Fig. S11A). The NMR spectrum of this solution (Fig. S11B) indeed showed the formation of Zn(S 6 Li) 2 ·DEA, indicating that Li 2 S 6 can readily react with Zn(SLi) 2 ·DEA to form Zn(S 6 Li) 2 ·DEA with the release of Li 2 S. This indicates that Zn(SLi) 2 ·DEA can also trap soluble LPS.
The interactions between Zn(OAc) 2 ·DEA and LPS during discharge and charge were further examined by X-ray photoelectron spectroscopy (XPS). The high resolution XPS spectra of Zn 2p showed a binding energy of 1022.5 eV for the fresh Zn(OAc) 2 ·DEA based electrode (Fig. 3A), which is similar to that of Zn in zinc acetate (Fig. S12). After discharging to 2.2 V and then 1.7 V, the binding energy decreased to 1021.6 eV and 1021.3 eV, respectively, which is most likely due to the formation of Zn-S x bonds with longer (x = 4-8) and shorter sul de chains (x = 1), respectively. When the electrode recharged to 2.8 V, the binding energy returned to 1022.2 eV, which is very close to that of the fresh electrode. This indicates that the sul de anions in Zn(SLi) 2 ·DEA in the discharged cathode were oxidized to form elemental sulfur, while the Zn(OAc) 2 ·DEA complex was regenerated during charge. There is a possibility that instead of acetate anions TFSI or nitrate anions in the electrolyte combine with Zn 2+ , but the Zn binding energy in Zn(TFSI) 2 or Zn(NO 3 ) 2 would be quite different at 1023.1 eV (Fig. S12) and 1021.3 eV 36 , respectively, if these products were formed. After fully discharging to 1.7 V, the S 2p spectra of the cathode showed the formation of Li 2 S (71%) with the rest (29%) being Li 2 S 2 31,32 and Zn(SLi) 2 ·DEA. The formation of elemental sulfur and complete disappearance of Li 2 S were observed after fully recharging to 2.8 V (Fig.   3B) 32 . Surprisingly, a signi cant amount of Li 2 S (13%) was detected in the cathode when the battery was partially discharged to 2.2 V in the rst discharge plateau, at which Li 2 S should be absent 31,37 . This strongly indicates that formation of Li 2 S was promoted in the presence of Zn(OAc) 2 ·DEA.
Based on the above results, we proposed a reaction cycle involving Zn(OAc) 2 ·DEA during the redox (discharge/charge) in the sulfur cathode as shown in Fig. 4 reversible formation and dissociation of intermolecular hydrogen bonds or "a self-healing effect", which may render it an excellent binder that can cushion the large volume oscillation of sulfur species during cycling, which makes a further contribution to the remarkable battery lifetime.
The in uences of Zn(OAc) 2 ·DEA on the redox of sulfur species The electronic states of the sulfur species in complexes (I) and (II) shown in Fig. 4 Fig. 5A, while the geometries of S 8 and corresponding Li 2 S x are shown in Fig. S13 for comparison. The adsorption conformations of these compounds on the graphene substrate 30 , which represents the surface of the conductive carbon, Super P, were also optimized ( Fig. S14 and S15). The Gibbs free energy changes (ΔG) for the reduction reactions of sulfur species are denoted as ΔG 1  respectively (see the detailed reactions in Supplementary Information). Fig. 5B shows the Gibbs free energy changes for the reduction reactions of sulfur species with or without Zn 2+ ·DEA. It can be clearly seen that the reactions of sulfur species with Zn 2+ ·DEA have markedly more negative Gibbs free energy changes compared with all the sulfur species without Zn 2+ ·DEA, revealing that the former reactions are thermodynamically much more favored 31 . It should be mentioned that the magnitude of ΔG 1 for the reduction between S 8 and Zn(OAc) 2 ·DEA to form Zn(S 8 Li) 2 ·DEA is larger than that for the reduction of S 8 to form Li 2 S 8 , indicating that the former may occur in the beginning of discharge (as shown in Fig. 4 and Supplementary Information). Furthermore, while the magnitude of ΔG for the reactions without Zn 2+ ·DEA decreases with decreasing the sul de chain length, that for the ones with Zn 2+ ·DEA remains large. These results indicate that once (I) with long chain sul des formed, it would be more readily reduced to form a complex with shorter sul des in comparison to the free sul des. For the last step reduction from S 2 2to S 2-, in particular, the absolute value of ΔG 5 for the formation of Zn(SLi) 2 ·DEA (II) is more than two-fold that of the reduction from Li 2 S 2 to Li 2 S. This suggests that (II) may form during the rst discharge plateau. As shown in the NMR data, (II) may further react with (or trap) LPS to reproduce (I). Li 2 S is released in this step, which was observed at a discharge potential of 2.2 V by the XPS measurement, while the newly formed (I) starts a new journey towards (II). Therefore, one Zn(OAc) 2 ·DEA may trap numerous LPS during one discharge.
In the charge process, the regeneration of S 8 may start from the dissociation of Li 2 S to yield LiS and Li +18,30 (Fig. S16) or the dissociation of (II) to yield Zn(S 2 Li)·DEA and Li + (Fig. 5C). The energy pro les of the dissociation of Li 2 S and (II) are shown in Fig. 5D. The dissociation energies of Li 2 S and (II) on graphene are 1.77 and 1.81 eV, respectively, indicating that the latter requires a slightly larger net energy to dissociate. However, the energy barrier for the dissociation of (II) is 2.14 eV, which is much lower than that (2.47 eV) for the dissociation of Li 2 S, indicating that (II) can be oxidized (or charged) more easily than Li 2 S 18 . Once Zn(OAc) 2 ·DEA is recovered, it may react with Li 2 S to form (II) again, which can start a new dissociation cycle, or trap longer sul des during charge. This may interpret the promoted charging process (lower charging potentials) as shown in Fig. 2B. The above DFT calculation results support the enhanced reduction and oxidation kinetics manifested in the longer rst discharge plateau, smaller charge/discharge polarization, and excellent high rate performance. Fabrication of sulfur electrodes. Sulfur and Super P with a mass ratio of 6:4 were ground together and the mixture was heated at 155 ºC for 12 h to obtain a sulfur/carbon composite. The Zn(OAc) 2 ·DEA binder based electrodes were prepared by coating a slurry containing sulfur/carbon composite and Zn(OAc) 2 ·DEA with a mass ratio of 9:1 in ethanol and deionized water (v:v=1:1) on a carbon coated aluminum current collector. As comparison, the polyvinylidene di uoride (PVDF) binder based electrodes were prepared in a similar manner except of using PVDF as binder and N-methyl-2-pyrrolidone (NMP) as solvent. The coated electrodes were dried in ambient air overnight and then further dried at 50 ºC in an oven overnight. The electrodes were cut into discs with a diameter of 12 mm and dried again at 50 ºC in a vacuum oven for 12 h before being transferred into an argon-lled glovebox. The average sulfur loading on electrodes is 1.5 mg cm -2 .
Electrochemical measurements. Electrochemical studies were carried out on 2035 coin cells with lithium foil as anode, Celgard 2400 as separator and 1 M lithium bis(tri uoromethanesulfonyl)imide (LiTFSI) with 2 wt% LiNO 3 in a mixture of DME and DOL (v:v=1:1) as electrolyte. The coin cells were assembled in Characterization. XRD measurements were carried out on a Bruker D8 Discover X-Ray Diffractometer using Cu Kα radiation (λ = 1.5418 Å). FT-IR analysis was performed on a Bruker Tensor 27 spectrometer. For XPS measurements, the samples were sealed in vials in the glovebox before being quickly transferred to an ultra-high vacuum chamber and measured with Thermo VG Scienti c ESCALab 250 using a monochromated Al K-alpha X-ray source. UV-vis absorption spectra were measured on a Cary 7000 Universal Measurement Spectrophotometer (UMS). 1 H NMR spectra were collected on a Bruker DPX 300MHz spectrometer.
Theoretical Calculation. The rst-principles density functional calculations (DFT) used PW91 functional within the gradient-corrected (GGA) approximation 40 as implemented in the Cambridge serial total energy package (CASTEP) code 38 . The Vanderbilt ultrasoft pseudopotential was used with a cutoff energy of 350.0 eV 41 . Geometric convergence tolerances were set at a maximum force of 0.05 eV/Å, maximum energy change of 2 × 10 -5 eV/atom, maximum displacement of 0.002 Å and maximum stress of 0.1 GPa. Density mixing electronic minimizer was implemented and the self-consistent eld (SCF) tolerance was set to accuracy of 2 × 10 -6 eV/atom for energy convergence. The Gibbs free energy change of each reaction is obtained by subtracting the Gibbs free energy of the reactants from the Gibbs free energy of the products.