Reversible Electrochemical Energy Storage Based on Zinc-Halide Chemistry.

The development of rechargeable Zinc-ion batteries (ZIBs) has been hindered by the lack of efficient cathode materials due to the strong binding of divalent zinc ions with the host lattice. Herein, we report a strategy that eliminates the participation of Zn2+ within the cathode chemistry. The approach involves the use of composite cathode materials that contain Zn halides (ZnCl2, ZnBr2, and ZnI2) and carbon (graphite or activated carbon), where the halide ions act both as charge carriers and redox centers while using a Zn2+-conducting water-in-salt gel electrolyte. The use of graphite in the composite electrode produced batterylike behavior, where the voltage plateau was related to the standard potential of the halogen species. When activated carbon was used in the composite, however, the cell acted as a hybrid Zn-ion capacitor due to the fast, reversible halide ion electrosorption/desorption in the carbon pores. The ZnX2-activated carbon composite delivers a capacity of over 400 mAh g-1 and cell energy density of 140 Wh kg-1 while retaining over 95% of its capacity after 500 cycles. The halogen reaction mechanism has been elucidated using combinations of electrochemical and in situ spectroscopic techniques.

Introduction

Aqueous rechargeable batteries are a promising class of batteries for grid-scale electrochemical energy storage owing to their low cost, ease of fabrication, high ionic conductivity, and high operational safety.[1−3] Research on aqueous batteries in recent years has been gaining momentum from application in low-voltage divalent zinc–ion batteries (ZIB) to high-voltage monovalent lithium-ion batteries (LIBs).[4−6] In particular, ZIBs have attracted substantial interest as one of the most promising next-generation technologies because: (i) they depend on an Earth-abundant metal, which is air-stable unlike Li; (ii) their low cost, safety, and environmental benignancy is attractive for grid-scale energy storage, and (iii) the volumetric energy density is approximately 3 times higher than that of Li.[2,7,8] Due to these favorable properties, zinc has been used as an anode material in a series of battery technologies both in conventional static cells (zinc–manganese dioxide batteries, zinc–air batteries, or Zn-graphite dual ion batteries) and in redox flow configurations (Zn–bromine or Zn–iron cells).[9,10]

The development of ZIBs is, however, hindered by a number of factors relating to aqueous electrolytes, the formation of Zn dendrites at the anode, and lack of efficient cathode materials.[7] Furthermore, the codecomposition of water molecules during the deposition of Zn2+ is known to affect the reversibility of the Zn stripping/deposition and depletes the electrolyte due to the sustained water consumption. Wang et al. used water-in-salt electrolytes (WiSEs) to enhance the electrochemical window of water and obtained dendrite-free Zn plating/stripping with near 100% Coulombic efficiency.[7] WiSEs contain a high concentration of the desired salt so that the hydrated ions outnumber free water: as there is no free water to react at the electrode surface, the overall cell voltage can be increased. The combination of small cations and large fluorinated anions in water alters the hydration behavior of the ions where the cation is strongly solvated, but the anion is not. The less solvated fluorinated anions can be reduced to form a passivating solid electrolyte interface (SEI) on the electrode surface.[5] This SEI formation significantly suppresses the hydrogen evolution reaction and is largely responsible for the overall electrochemical stability window of WiSEs.[5] The highest voltage window (4.9 V) recorded at the hydrophobic graphite of WiSEs contains small metal cations (Li+) and large fluorinated anions such as bis(trifluoromethanesulfonyl)imide ([TFSI]−) and trifluoromethanesulfonate ([TFO]−).[6]

Although the electrochemical reversibility of Zn stripping/deposition was enhanced and Zn dendrite formation was suppressed using WiSEs, the lack of an efficient cathode material for ZIBs remains a severe challenge.[11,12] Some of the cathode materials developed to date are Prussian blue analogs,[13,14] manganese oxides,[15−18] and vanadium oxides.[11,12] Each of these materials suffers from limited specific capacity, far below the theoretical capacity of the Zn anode (820 mAh g–1), a low-voltage plateau (<1.4 V), and a low rate capability due to their poor electronic conductivity. The low performance of these cathodes is mainly attributed to the high polarization of the divalent Zn2+ ions leading to strong binding with the host lattice and sluggish solid-state migration dynamics.[1,12] Therefore, the development of efficient cathodes should use different combinations of battery chemistry, for example, reversible deposition and stripping of Zn as the anode half-reaction and reversible atomic intercalation or plating of non-Zn-ion species as the cathode half-reaction. This strategy not only avoids the intercalation of Zn2+ to the host cathode material but also allows the flexibility to fine tune and increase the output voltage, which is the main barrier limiting the overall performance.

Herein, we describe a new class of composite cathode materials that contain Zn halides (ZnCl2, ZnBr2, and ZnI2) and carbons (graphite or activated carbon), where the halide ions act both as charge carriers and as redox centers. It should be noted that our approach differs from the conventional Zn–bromine flow battery, where bromide is oxidized to bromine and stored in an external tank.[19] The approach we pursue in the current study uses Zn halide immobilized on carbon hosts as the cathode, where the redox activity of the halides is exploited to store more charge. In other words, the halides undergo a “conversion–intercalation/adsorption” reaction inside the carbon structure when the cell is charged using a Zn2+ conducting water-in-trisalt (WiTS) gel electrolyte. Our approach also removes the halogen cross-over seen in Zn–bromine cells as the halogen is confined (intercalated or adsorbed) within the carbon structure upon oxidation. The variety of the halogen species means that the cell voltage can be fine-tuned and increased. The standard reduction potentials of I–/I2, Br–/Br2, Cl–/Cl2, and F–/F2 redox couples are 0.54, 1.09, 1.36, and 2.8 V vs standard hydrogen electrode (SHE), respectively.[20] The combination of the zinc-halide–carbon cathode with a Zn anode can therefore generate an open-circuit voltage that ranges between 1.3 and 3.5 V depending on the halide type. The flexible and non-flammable semisolid WiTS gel electrolyte, which is compatible with the halogen cathode, exhibited fast electrode kinetics for Zn oxidation and reduction without the formation of Zn dendrites. We will show that both the identity of the Zn halide and carbon structure in the cathode produces electrochemical energy storage devices that fundamentally differ from one another. The resultant composite electrodes can deliver a capacity of 480 mAh g–1 at 0.05 A g–1 with a corresponding energy density of 140 Wh kg–1. The conversion–intercalation/adsorption mechanism of the individual ZnX2–carbon composite in the WiTS gel electrolyte is characterized by electrochemical, in situ Raman spectroscopy and ex situ X-ray photoelectron spectroscopy (XPS) techniques and analyzed in detail.

Results and Discussion

WiTS Gel Electrolyte Formulation and Characterization

The robust Zn2+ conducting gel electrolyte was formulated from water-in-trisalt (WiTS) electrolyte by mixing ZnSO4, Zn(TFO)2, and LiTFSI (in a 1:1:2 mass ratio, respectively) in 20% water using 10% poly(tetrafluoroethylene) (PTFE) polymer binder. The resulting white gel is a highly flexible, yet semisolid, electrolyte that can be molded to any shape (inset of Figure A for optical image).

Figure 1

(A) Thermogravimetric analysis (TGA) traces recorded for a WiTS gel electrolyte obtained by ramping the temperature from 30 to 800 °C at a rate of 10 °C min–1 under N2. The inset shows the photographic images of a WiTS gel electrolyte, the scale bar is 5 cm. (B) Cyclic voltammetry (CV) recorded at a 3 mm diameter glassy carbon (GC) electrode at 10 mV s–1 using a WiTS gel electrolyte between −0.2 and 4.2 V. (C) CVs recorded with a symmetrical Zn|Zn cell at 1 mV s–1 in a WiTS gel electrolyte before and after cycling for 40 h using galvanostatic charge–discharge. (D) Galvanostatic charge–discharge curve obtained using a symmetrical Zn|Zn cell in a WiTS gel electrolyte and 2.0 M ZnSO4 (aq) at 0.2 mA cm–2. (E) Scanning electron microscopy (SEM) image of a fully discharged Zn substrate, the scale bar = 20 μm. (F) SEM of a fully charged Zn substrate, scale bar = 20 μm.

Each of the components contributes to the unique properties of the gel. LiTFSI was used as it is soluble in water to a high concentration (when compared to Zn(TFO)2 or ZnSO4) to reach the water-in-salt regime. It is also used as a source of the [TFSI] anion since it is believed that the reduction of [TFSI]− is responsible for the formation of the passivating SEI, which extends the overall electrochemical window.[5,7] We also note that in the absence of ZnSO4 the gel is very sticky to handle, while in the absence of Zn(TFO)2, a rigid dry material is formed, as excess ZnSO4 pulls water from the mixture. The thermal decomposition of the gel electrolyte was studied by thermogravimetric analysis (TGA), which shows four mass losses due to the loss of H2O (<200 °C), decomposition of [SO4]2– (200–284 °C, ∼10%), [TFO]− (323–450 °C, ∼34%), and [TFSI]− (>450 °C).[21−23] The TGA also shows the gel contained about 15% water. The ionic conductivity of the gel electrolyte was determined using alternating current (AC) impedance (Figure S1) and was found to be 6 mS cm–1, which is comparable to that of nonaqueous electrolytes (9.0 mS cm–1) used in commercial LIBs.[5] The electrochemical window of the gel, as well as the reversibility of Zn plating and stripping, was investigated using cyclic voltammetry (CV) at a glassy carbon (GC) disk electrode. The gel electrolyte exhibited fast kinetics for Zn oxidation/reduction and achieved a potential window of 3.0 V (Figure B), which is comparable to WiSE-based on 21 m LiTFSI.[5]

The long-term electrochemical reversibility of the Zn plating and stripping processes in the WiTS gel electrolyte was investigated using a Zn|Zn symmetric cell under galvanostatic and CV methods. Figure C shows the CV obtained before and after several charge–discharge cycles; and in each case, the gel electrolyte exhibited reversible Zn redox chemistry with the ratio between the anodic (ip,a) and cathodic peak currents (ip,c) being one which indicates that the Coulombic efficiency of the cell is near 100%. The kinetics of Zn oxidation/reduction, however, significantly improved after cycling, as exemplified by the peak-to-peak (ΔEP) separation that decreased from ∼0.4 to 0.2 V. The decrease in ΔEP with cycling is most likely due to the removal of surface oxides from Zn, which impede electron transfer. In addition, the current due to Zn oxidation/reduction is increased by a factor of 5 after cycling, which could be due to an increase in the active surface area. The charge–discharge curve (Figure D) also showed similar behavior where the overpotential (η) for Zn stripping/plating decreased from about 0.6 to 0.2 V after 10 cycles and stabilized at 0.2 V even after the cell was cycled for 40 h at 0.2 mA cm–2. The cell can operate continuously without short circuiting (no formation of dendrites) for over 400 h at 0.1 mA cm–2 with a much lower η for Zn|Zn2+ redox reactions (see Figure S2). In sharp contrast, in a dilute electrolyte (2 M ZnSO4), a rapid polarization started to occur after nine cycles most likely due to the formation of Zn dendrites and other associated problems including oxide formation (Figure D).[24] These observations demonstrate that the gel electrolyte is an excellent Zn2+ conductor with very facile Zn stripping/plating kinetics. Both the stability and the kinetics of the Zn couple in the WiTS gel electrolyte are superior to those in solution-based aqueous and nonaqueous ionic liquid-based electrolytes.[25−28]

The presence or absence of Zn dendrite formation in this gel electrolyte is also examined using SEM after galvanostatic cycling of a Zn|Zn symmetric cell for 40 h. The Zn-plated substrate exhibited a dense and uniform layeredlike structure with the absence of any substantial dendrites (Figure E). After Zn stripping, this dense structure is completely removed and the original surface is retained without the formation of ZnO according to X-ray diffraction (XRD) (Figure S3), indicating the reversibility of Zn chemistry in the gel electrolyte.

In summary, this Zn2+-conducting WiTS gel electrolyte provides the advantage of safety (as it is nonflammable) and enables certain battery components to be removed (such as the polymer separator). The elimination of the separator from the cell will significantly reduce the contact resistance of the interface. Then, we consider the cathode chemistry involving the carbon host.

Aqueous Zn|(Graphite–ZnX2) Cell

The electrochemistry of confined Zn halides within a graphite electrode was examined using the WiTS gel electrolyte in full-cell Zn batteries. The free-standing graphite–ZnX2 cathodes were prepared by mixing the desired halide and natural graphite at a mass ratio of 1:3 with 5% PTFE binder. Figure A shows the CVs recorded at the cathodes of graphite–ZnI2 (G–ZnI2), graphite–ZnBr2 (G–ZnBr2), and graphite–ZnCl2 (G–ZnCl2) combined with Zn anodes in cells using the WiTS gel electrolyte. Significantly, both G–ZnI2 and G–ZnBr2 showed reversible redox reactions at the characteristic formal potentials of I–/I2 (1.17 V vs Zn/Zn2+) and Br–/Br2 (1.67 V vs Zn/Zn2+). The oxidative redox reactions are therefore attributed to the conversion of the halide ion (I–, Br–) to elemental halogen (I0 or Br0), which is stabilized by sequential intercalation/adsorption into graphite galleries to form a solid graphite intercalation compound (GIC) (eq ).[6,29−31] This oxidation process releases Zn2+, which is transported through the gel electrolyte to replace the ions reversibly plated on the Zn anode (eq ). The reduction process at the cathode is due to the deintercalation and reduction of I0/Br0 to recombine with Zn2+ (n is the molar ratio of carbon atoms to the intercalated/adsorbed halogens in the GIC). The ratio between the ip,a and the ip,c being one in Figure A demonstrates the high reversibility of eq

Figure 2

(A) Cyclic voltammograms recorded at 1 mV s–1 in the WiTS gel electrolyte using coin cells constructed from G–ZnX2 positive electrodes and a Zn foil negative electrode. Ex situ XPS of G–ZnCl2–ZnBr2 obtained for (B) fully charged and (C) fully discharged cells. (D) Galvanostatic discharge curves vs capacity obtained at a current density of 50 mA g–1 using the positive electrodes (G–ZnX2) shown and the Zn negative electrode in the WiTS gel electrolyte.

In contrast, the CV obtained using the G–ZnCl2 cathode showed a lower current and a lower degree of reversibility with a sharp undefined oxidation peak and a small reduction peak. The poor reversibility of G–ZnCl2 suggests the formation of an irreversible product during battery charging. This is in contrast to G–LiCl which has been reported to display a reversible Cl– conversion–intercalation process in WiSEs.[6] The fact that the conversion–intercalation of ZnCl2 is irreversible suggests that the metallic counterions have a strong impact on the conversion–intercalation process. It has been shown that the chemical intercalation of ZnCl2 into graphite forms a strong complex with the carbon species to form ZnCl2–carbon.[32] Indeed, characterization of a fully charged G–ZnCl2-containing electrode using XPS showed the formation of C–Zn–C bonds at a low binding energy (283.0 eV) when analyzing the high-resolution C1s spectrum (Figure B).[33] Furthermore, the signal due to Zn carbide is still present and increased for the fully discharged cathode, which confirms the irreversibility of the process, manifested in the CV response (Figure C). The slight shift to high binding energy (283.5 eV) during discharge indicates a change in the environment of C1s presumably due to the increase in the concentration of chloride along with Zn inside the carbon. This data demonstrates that ZnCl2-based cathode materials cannot be combined with sp2-carbon to form a secondary ZIB.

Figure D presents the discharge curves at 50 mA g–1of the ZIB full cells with the WiTS gel electrolyte for different halide-based cathodes. As expected, G–ZnCl2 showed a very low capacity with the absence of any useable voltage plateau due to the irreversible reaction. The G–ZnI2 and G–ZnBr2 cathodes showed voltage plateaus that correspond to their respective redox reactions, G–ZnI2 at 1.17 V and G–ZnBr2 at 1.67 V vs Zn/Zn2+, in agreement with the CV data. The cathode made from the equimolar mixture of ZnCl2 and ZnBr2 showed two discharges voltage plateaus: a small one at 1.90 V, due to the Cl–/Cl2 redox reaction, and the other at 1.67 V due to the Br–/Br2 redox reaction. Nonetheless, the specific capacity of each cathode is much lower than the theoretical capacity of a halogen GIC (309 mAh g–1 for MBrn and 632 mAh g–1 for MCln).[6] Among the cathodes tested, the best specific capacity (55 mAh g–1) was obtained using G–ZnCl2–ZnBr2 with the others being lower than 30 mAh g–1. However, the capacity decayed by more than 50% after 200 cycles due to the continual formation of zinc carbide species (Figure S4). The specific capacity quoted is based on the total mass of the cathode (mass of graphite plus mass of halide).

In situ Raman spectroscopy was used to understand the halide intercalation mechanism and to rationalize the poor performance of the G–ZnX2 using the WiTS gel electrolyte. Figure shows the fully charged–discharged Raman spectra for each cathode. The free-standing sample for each cathode showed a similar response at open-circuit potential (OCP), with the characteristic graphite bands, G-band at 1580 cm–1 and a small D-band at ∼1350 cm–1, being shown and no other bands associated with Zn halides. The fully charged G–ZnI2 electrode displayed an intense Raman signal at 172 cm–1 due to surface-bound iodine species.[34,35] However, the absence of a G-band splitting suggests that there is no intercalation of the iodide species into the graphite galleries (see the inset of Figure A). While other halogens including chlorine and bromine intercalate into graphite, iodine has a strong affinity for adsorption rather than intercalation.[36,37] G–ZnBr2 also exhibited similar behavior when fully charged where it showed a signal at 240 cm–1 due to the stretching mode of Br2.[6,38] The G-band of the material, however, was split into two Raman modes: the E2g2i mode at ∼1580 cm–1 due to the interior unintercalated original layers and the E2g2b at 1604 cm–1 due to the bounding layers next to the intercalants. The fact that the intensity of the E2g2i mode is twice that of the E2g2b together with the higher wavenumber for the E2g2b reflects the dilute staging of bromine species into graphite galleries.[39] The opposite trends were observed during discharge that involves the desorption/deintercalation of the halogen at each cathode (Figure B) characterized by the absence of halogen-related bands, further confirming the reversibility of the process. This in situ Raman spectroscopy data shows that reversible surface adsorption/desorption is the dominant reaction mechanism in G–ZnI2 and G–ZnBr2 electrodes. This observation is similar to Na chemistry at a graphite electrode where Na plates are on the graphite surface rather than intercalating due to its size and weaker chemical interaction with the graphite planes.[40] The use of hard carbon in the composite may improve the cell performance.

Figure 3

In situ Raman spectral series of a Zn|G–ZnX2 cell in the WiTS gel electrolyte during full (A) charge and (B) discharge. The insets in (A) and (B) show the graphite G-band region. (C) Comparison of Zn|G–ZnX2 at different voltages.

The electrode that contained ZnCl2 (neat or mixed with ZnBr2) on the other hand showed a more intense E2g2b at 1609 cm–1, approximately twice that of the E2g2i, a characteristic of the formation of a stage-3 GIC.[39] However, the G-band splitting remained after the battery was fully discharged, which demonstrates that the intercalation process is nonreversible, in agreement with the XPS analysis (inset of Figure B). Furthermore, the fully charged G–ZnCl2 or G–ZnBr2–ZnCl2 did not show the band associated with Cl2 (expected in the region of 530–570 cm–1) intercalant due to the reaction of ZnCl2 with the graphite host as previously discussed. Figure C shows the Raman spectrum response for the G–ZnI2–ZnBr2 composite electrode. It is interesting to note that significant Raman frequency shifts for both iodide and Br-species were observed when analyzing the G–ZnI2–ZnBr2 sample. When the cell was charged to 1.4 V, a broad band at 172 cm–1 due to surface-bound iodine species (also the case for G–ZnI2 charged to 1.9 V) was seen. When the voltage was increased to 1.9 V, sharp bands at 180 and 189 cm–1 were observed along with the G-band splitting. The formation of these new bands is most likely due to the formation of interatomic IBr intercalants.[41] A frequency downshift is often observed for surface-bound halogen when compared to free halogen due to the interaction of halogen with host materials, which weakens the interatomic bonds of the intercalants.[6,41] In contrast, the G–ZnCl2–ZnBr2 sample did not show the BrCl formation due to the reaction of ZnCl2 with the graphite (Figure A).

Overall, the analysis of Raman spectra indicates that the size of halogens significantly impacts the reaction mechanism at the graphite cathode. The conversion–adsorption process occurs when the halogen is larger, for example in G–ZnI2, and the conversion–intercalation process occurs for smaller halogens such as G–ZnCl2. Even though ZnCl2 can reach a reasonably high intercalation staging, the irreversible reaction is the hindering factor for use in practical ZIBs. The low capacity of the full-cell battery at each electrode is explained by the conversion–adsorption reaction mechanism, which needs a high surface area carbon rather than the low surface area graphite.

Aqueous Zn|(Activated Carbon–ZnX2) Cell

Given that the conversion–adsorption process is the dominant reversible process in Zn halide–carbon composites, the working hypothesis was that variation of the surface area of carbon would determine the extent of cell performance. To this end, various Zn halides were mixed with high surface area-activated carbon (AC) and their performance was tested in full-cell coin cells. Figures A and S5 show the representative charge–discharge curves of Zn|AC–ZnX2 batteries using the WiTS gel electrolyte at various current densities between 0.1 and 2.0 V. Significantly, the charge–discharge curves possess a near-triangular shape with little deviation from an ideal capacitor response. This implies that the kinetics of the halogen conversion–adsorption reaction at the activated carbon electrode is extremely facile in the WiTS gel electrolyte. However, the CV response at each electrode was not strictly pseudocapacitive due to the presence of redox peaks (see Figures D and S6A). The specific capacities for Zn|AC–ZnCl2, Zn|AC–ZnBr2, and Zn|AC–ZnI2 cells are approximately 281, 232, and 196 mAh g–1, respectively, at a current density of 0.05 A g–1 (compared to <100 mAh g–1 for bare AC in the same electrolyte). These capacities decreased to 102, 90, and 65 mAh g–1 when the current density was increased to 1.0 A g–1, and the corresponding capacity fade is over 60% for each cell (Figure B). The capacity loss is most likely due to the low conductivity of the activated carbon and future work will focus on the use of conducting additives, such as carbon black, to further optimize the performance. The capacities obtained at these electrodes are nonetheless higher than Zn-ion capacitors using an activated carbon cathode and Zn anode.[42,43]

Figure 4

(A) Charge–discharge curve obtained using coin cells constructed from the AC–ZnCl2Br2I2 positive electrode and the Zn foil negative electrode at gravimetric currents indicated. In each case, the voltage range was between 0.1 and 2.15 V vs Zn/Zn2+and the gel electrolyte was WiTS. (B) Specific capacity vs gravimetric current using positive electrodes indicated. (C) Gravimetric capacitance vs gravimetric current using AC–ZnCl2Br2I2 and bare AC positive electrodes. (D) CVs recorded at bare AC and AC–ZnCl2Br2I2 cathodes using the Zn anode at 3.0 mV s–1 in the WiTS gel electrolyte.

The AC composite cathode made from the combination between ZnCl2, ZnBr2, and ZnI2 with the equimolar ratio of the halides achieved specific capacities twice that of individual ZnX2 at all current densities studied (Figure B). The improved capacity in the Zn|AC–ZnCl2Br2I2 cell relative to pure Zn|ZnX2 could be due to a denser adsorption of halide ions on the carbon microstructure. The enhancement in performance of the mixed halide cathode is possibly due to the reduction of Coulombic repulsion of the adsorbed halides and an electrocatalytic effect. It has been shown that the intercalation/adsorption density of halogen on a carbon surface is approximately twice as large as Li–GIC. This is because the oxidation state of the halogen is close to zero, which reduces the average effective charge per halogen atom. This minimizes the Coulombic repulsion and, in turn, increases the adsorption density of halogen on the carbon pore.[6] Furthermore, the Coulombic repulsion is expected to be further reduced within the carbon microstructure host when the three-halide species (Cl–, Br–, and I–) are adsorbed next to one another. The low Coulombic repulsion in the AC–ZnCl2Br2I2 electrode over individual AC–ZnX2 can enhance the packing density of halogens on the carbon surface. Indeed, Yang et al. also observed that the performance of individual LiBr–carbon or LiCl–carbon cathodes is inferior to the binary combination of the LiClBr–carbon cathode.[6]

The halide ion conversion–adsorption reaction is an electrocatalytic process where the reaction is greatly affected by the surface composition of the electrode.[44,45] CVs were used to see if this electrocatalytic effect is in play, and Figures D and S6A show the CVs recorded for each electrode. At the AC–ZnCl2Br2I2, a sharp oxidation peak at 1.18 V due to the conversion and adsorption of iodide species was seen along with a corresponding reduction peak at 1.06 V due to the reduction and recombination of iodide with Zn. Electrochemical data also suggest that the I–/I2 redox reaction is a surface-controlled process at the AC electrode as both ip,a and ip,c are proportional to the scan rate (Figure S6C). The ΔEP for the I–/I2 redox couple at AC–ZnCl2Br2I2 was 0.12 V in contrast to over 1.0 V at AC–ZnI2. This demonstrates that the I–/I2 redox reaction is kinetically much faster for the AC–ZnCl2Br2I2 case than the AC–ZnI2 system. Similarly, the ΔEP for the Br–/Br2 redox couple decreased from 0.65 V at AC–ZnBr2 to 0.1 V at AC–ZnCl2Br2I2, which confirms the electrocatalytic effect of the ternary halide mixture over the individual halides. In other words, the preadsorption of iodine, due to its low redox potential, on the AC surface improved the kinetics of bromine adsorption, which in turn improves the kinetics of chlorine adsorption similar to an electrodeposition process. Ex situ XPS was used to gauge the surface composition of fully charged Zn|AC–ZnCl2Br2I2 cell as shown in Figure S7. XPS shows the presence of Cl–, Br–, and I– for the sample that was analyzed at OCP. However, the XPS signal due to all of the three-halogen species disappeared for the fully charged sample. This indicates that the halide ions are fully oxidized to molecular halogen and coadsorbed to the AC pore, which can sublime under the ultrahigh vacuum of the XPS. Nevertheless, the presence of preadsorbed halogen on the AC surface was confirmed using in situ Raman spectroscopy using AC–ZnI2 as an example electrode. As shown in Figure S8, the fully charged electrode displayed bands associated with surface-bound I2 species as well as the typical defective (D) and G-band of AC. The SEM images of the AC–ZnCl2B2rI2 electrode at OCP and after full charging are also presented in Figure S9. The images after charging showed a pore swelling with significant size expansion due to the adsorption of halogens. Each of the expanded pores contained several smaller holes. The small pores could be created by the evaporation of the adsorbed halogen under the SEM-operating vacuum that leaves holes behind.

The specific capacity of Zn|AC–ZnCl2Br2I2 is 479 mAh g–1 at 0.05 A g–1, which decreased to 230 mAh g–1 at 1.0 A g–1. The gravimetric capacitances of Zn|AC–ZnCl2Br2I2 ranged from 400 to 930 F g–1 depending on the applied current density (Figure C). This figure also demonstrates the importance of the Zn halides to the cathode chemistry, proving that the cathodic process involved the immobilized Zn halides. Based on the mass of the cathode (mass of activated carbon plus mass of ZnX2), the energy densities of 422 and 160 Wh kg–1 at power densities of 122.8 and 1071.7 W kg–1 were obtained. These values are higher than the energy density of all other cathode materials reported to date for ZIBs including MnO2,[46,47] V2O5,[12,48] Zn3V2O7·2H2O,[49] Zn0.25V2O5·nH2O,[11] CuHCF,[50] and VS2 nanosheets.[51] The cell energy density ranged between 55 Wh kg–1 (at 1.0 A g–1) and 140 Wh kg–1 (at 0.05 A g–1), assuming that the weight of the cathode material within a pouch cell configuration is a third of the total mass of the cell.[46] These values are much higher than those of typical commercial supercapacitors (5–10 Wh kg–1), lead-acid batteries (30–40 Wh kg–1), and Zn-ion capacitors (17–30 Wh kg–1).[52−56] The energy density of the Zn|AC–ZnCl2Br2I2 cell is even higher than a Li-ion capacitor, where the energy density varies between 30 and 90 Wh kg–1.[57] Li-ion capacitors often use intercalation-type anodes and adsorption-type cathodes (adsorption of large complex anions ([PF6]−, [TFSI]−, [BF4]−, etc.)), where their overall energy density is limited by the capacitor-type electrode.[57] The advantages of using halide ion conversion–adsorption within carbon cathodes are (i) they are smaller than most organic anions so that the migration/diffusion of ions is faster, (ii) they are inside the carbon structure so that they do not have to diffuse to the surface from bulk electrolyte, and (iii) they undergo reversible fast redox reactions, which substantially provide an extra charge, unlike inert anions.[6] The combination of these factors is responsible for the high performance of the Zn|AC–ZnCl2Br2I2 cell. The cell also truly combines the characteristic high energy density of a battery with the high power density of a supercapacitor device. For example, this cell can be fully charged within a few minutes (6 min) at high power and can be discharged for over 5 h at lower rates (see Figure A).

Figure 5

(A) Zn|AC–ZnCl2Br2I2 coin cells charged at 1.0 A g–1 (left-hand vertical axis) and discharged at 0.05 A g–1 (right-hand vertical axis) using the WiTS gel electrolyte and (B) capacity retention and Coulombic efficiency of the Zn|AC–ZnCl2Br2I2 cell cycled at 0.75 A g–1 using the WiTS gel electrolyte.

The Zn|AC–ZnCl2Br2I2 cell also exhibited excellent cyclic stability when the cell was cycled at 0.75 A g–1. The cell capacity retention is 95% after 500 cycles with 99% Coulombic efficiency throughout the cycles (Figure B). Although the capacity initially decreased by 10%, a subsequent increase in capacity was observed after 250 cycles and the ohmic drop of the cell decreased with increasing cycling (Figure S10). The increase in capacity could be due to the gradual activation of the electrode, which increases the number of active electrochemical sites for ion adsorption. Finally, the WiTS gel electrolyte performance was tested using a traditional α-MnO2 cathode and Figure S11 shows the charge–discharge curve obtained for a Zn|α-MnO2 cell. The gel electrolyte exhibited a Coulombic efficiency of over 99% with specific capacities that were increased from 162 to 210 mAh g–1 after 50 cycles. This indicates that the WiTS gel electrolyte is as efficient as traditional aqueous electrolytes (such as 1 M ZnSO4), but with additional advantages such as the absence of parasitic water reduction reactions, elimination of the need for a separator, and other advantages that semisolid state devices provide, e.g., flexibility. The combination between the WiTS gel electrolyte and ZnX2–carbon composite cathode in conjunction with aqueous gel electrolytes offers the tantalizing possibility of solving the issues of poor ZIB performance. This study also provided the very first fundamental understanding of halogen conversion chemistry inside crystalline and amorphous carbon.

Conclusions

A Zn-conducting water-in-trisalt gel electrolyte and halogen-incorporating cathode have been successfully developed and used in Zn-based electrochemical energy storage for the first time. The benefits of using confined halogen within the carbon structure as a cathode are (i) elimination of the irreversible binding of Zn2+ to the host structure within the cathode chemistry, (ii) the provision of substantial extra charge through their conversion–intercalation/adsorption process, and (iii) obviating the need for ions to diffuse to the surface from bulk electrolyte as they are already inside the carbon structure. The most significant findings emerging from this study are that the identity of the Zn halide and carbon structure in the cathode composite produces electrochemical energy storage devices that are fundamentally different from each other (battery vs supercapacitor). The use of graphite in the composite electrode produced batterylike behavior, where the voltage plateau was related to the standard potential of the halogen species. In situ Raman spectroelectrochemistry revealed that the identity of halides determines the mechanism of charge storage, intercalation vs electrosorption. In contrast, when activated carbon was used in the composite, the cell acted as a hybrid Zn-ion capacitor due to the fast reversible halogen species electrosorption/desorption in the carbon pores. In this case, the overall capacity is related to the number (binary or ternary) of Zn halides present; and the Zn|(activated carbon–ZnCl2Br2I2) cell exhibited a high specific capacity and energy density as well as good cycling stability. The combination of the Zn|(activated carbon–ZnCl2Br2I2) cell with the WiTS gel electrolyte is promising for the development of high-performing, low-cost, and environmentally friendly energy storage device based on Zn. Future work should focus on the microstructural design of carbon pores to match the size of halogens as well as investigating other carbon structures such as heteroatom-doped carbon or high surface area hard carbon materials.

Experimental Methods

Materials and Apparatus

All chemicals are of analytical grade and obtained from Sigma-Aldrich, Fluorochem or Alfa Aesar and used as received. X-ray photoelectron spectroscopy (XPS) was performed using a Kratos Axis Ultra DLD spectrometer with a monochromated Al Kα X-ray source (E = 1486.6 eV, 10 mA emission). Scanning electron microscope (SEM) analysis was carried out using a FEI Quanta 650 FEG environmental SEM. Powder X-ray diffraction analysis was performed using a Philips X’pert PRO diffractometer with Cu Kα radiation (λ = 0.154 nm) and operating at 40 kV and 30 mA.

WiTS Gel Electrolyte Preparation

Two grams of ZnSO4·7H2O (99.99%, Sigma-Aldrich), 2 g of zinc trifluoromethanesulfonate (98%, Fluorochem), and 4 g of lithium bis(trifluoromethanesulfonyl)imide (99%, Fluorochem) were mixed together with 1.5 g of ultrapure water (Milli-Q, 18 MΩ cm resistivity) in a mortar and pestle until a uniform white paste formed. Subsequently, 1.3 g (10% of total salt) of a 60% PTFE suspension was added and mixed with a mortar and pestle. The mixture was then heated at 80 °C on a hot plate for 30 min to remove any excess water. A semisolid elastic gel was formed, which can be changed into any desired shape (Figure A).

Electrode Preparation

For the preparation of the carbon–ZnX2 cathode, first the desired amount of ZnX2 (anhydrous ZnCl2, ZnBr2, ZnI2, or mixture of two or more halides) was dissolved in 0.5 g of water followed by slow addition of natural graphite (−325 mesh, 99.8%, Sigma-Aldrich) or activated carbon (YEC-8B, Fuzhou Yihuan Carbon Co., Ltd) while mixing homogeneously with a mortar and pestle. The mass ratio between ZnX2 and carbon was 1:3. PTFE suspension (5% with respect to the mass of carbon) was added to the thick slurry and mixed to uniformly coat the mixture with the polymer binder. The excess water was removed by heating the mixture on a hot plate. The resulting carbon–ZnX2 clay was quite flexible and could be made as a free-standing film or rolled onto a prepunched (15 mm diameter) titanium (99.99%, Alfa Aeser) foil current collector. The composite electrode was dried in a vacuum oven at 80 °C overnight. The typical mass loading of the carbon–ZnX2 composite electrode ranged from 2 to 5 mg cm–2.

Battery Assembly and Electrochemical Measurements

The full cells were assembled in CR2032-type coin cells using carbon–ZnX2 as the cathode and Zn foil as the anode. The flexible gel electrolyte was spread onto the Zn foil with an approximate thickness of 0.5–1.0 mm and acted as both the electrolyte and separator. The coin cell was sealed using a hydraulic crimping machine (MSK-160D) in an ambient atmosphere. Three-electrode cell electrochemical measurements were conducted using a WiTS gel electrolyte that was rolled onto a microscopy glass slide for electrode connection. A glassy carbon working electrode, Pt wire counter electrode, and Zn metal reference electrode were used. Electrochemical measurements were performed using an Autolab potentiostat (model PGSTAT302N, Metrohm Autolab, The Netherlands). The charge–discharge battery tests were carried out using a Basytec Cell Test System (BaSyTecGmbH, Asselfingen, Germany) with 32 independent test channels. The average capacity of three different coin cells was used to report capacity/capacitance.

In Situ Raman Spectroscopy Measurement

Raman spectra were obtained using a Renishaw inVia microscope with a 532 nm excitation laser operated at a power of 0.274 mW with a grating of 1800 lines/mm and 50× objective. The in situ Raman cell was obtained from ECC-Opto-Std (EL-Cell GmbH, Hamburg, Germany) and the cell was comprised of a free-standing carbon–ZnX2 positive electrode and a Zn foil negative electrode with a WiTS gel electrolyte. A titanium foil that contained a small hole in its middle (diameter ca. 1 mm) was used as a current collector for the positive electrode. The exciting laser beam was shone through a thin glass window onto the rear of the free-standing carbon–ZnX2 film through the small hole in the center of the Ti foil. Spectral scans were collected in a backscattering configuration. The Raman measurements were collected at various voltages as the cell charged and discharged at 1 mV s–1.