Water-mediated cation intercalation of open-framework indium hexacyanoferrate with high voltage and fast kinetics

Rechargeable aqueous metal-ion batteries made from non-flammable and low-cost materials offer promising opportunities in large-scale utility grid applications, yet low voltage and energy output, as well as limited cycle life remain critical drawbacks in their electrochemical operation. Here we develop a series of high-voltage aqueous metal-ion batteries based on ‘M+/N+-dual shuttles' to overcome these drawbacks. They utilize open-framework indium hexacyanoferrates as cathode materials, and TiP2O7 and NaTi2(PO4)3 as anode materials, respectively. All of them possess strong rate capability as ultra-capacitors. Through multiple characterization techniques combined with ab initio calculations, water-mediated cation intercalation of indium hexacyanoferrate is unveiled. Water is supposed to be co-inserted with Li+ or Na+, which evidently raises the intercalation voltage and reduces diffusion kinetics. As for K+, water is not involved in the intercalation because of the channel space limitation.

The E ocp of the electrodes in a, b and c follow the same order: black > red > green > blue. E ocp , the open circuit potential of the as-prepared electrode in the electrolytes that is an important indicator of the chemical sates. From the E ocp , we can see that InHCF/Gr is a compound with oxidation form.

Supplementary Figure 3 | Comparison of rate capability between InHCF/Gr and InHCF.
Rate capabilities of InHCF/Gr (a, c, e) and InHCF (b, d, f) in aqueous 0.5 M Li 2 SO 4 (a, b), 0.5 M Na 2 SO 4 (c, d) and 0.5 M K 2 SO 4 (e, f). g is the summary of rate capabilities of InHCF/Gr and InHCF in different electrolytes (C s is the specific discharging capacity, and percentage is the normalized discharging capacity which is compared with the discharging capacity measured at 1C rate). E (V) is normalized to SHE, and 1C equals to 60 mA g -1 .
Supplementary Figure 4 | Plots of E f , E ocp and E 1/2 vs. the composition of mixed-ion electrolytes. a, Li + /Na + mixed-ion electrolytes; b, Li + /K + mixed-ion electrolytes; c, Na + /K + mixed-ion electrolytes. E f , formal potential, the average of the anode and cathode peak potentials (E pa and E pc ); E ocp , the open circuit potential of the as-prepared electrode in the electrolytes that is an important indicator of the chemical sates; E 1/2 , the voltage that is measured when the electrode material has discharged 50% of its total capacity at a rate of 1C. Supplementary Figure 6 | Summary of rate capabilities of InHCF/Gr in various mixed-ion electrolytes. a, Li + /Na + mixed-ion electrolytes; b, Li + /K + mixed-ion electrolytes; c, Na + /K + mixed-ion electrolytes. C s is the specific discharging capacity, and percentage is the normalized discharging capacity which is compared with the discharging capacity measured at 1C rate. 1C = 60 mA g -1 . (InHCF+Gr is the physical mixture of InHCF and graphene with the mass ratio of 10:1, C s is the specific discharging capacity, and percentage is the normalized discharging capacity which is compared with the discharging capacity measured at 1C rate). E (V) is normalized to SHE, and 1C equals 60 mA g -1 .  4 ] + prefer linear, triangular and tetrahedral structure, respectively.

Supplementary Figure 15 | Electronic density of states projected on In atom in InHCF (top), NaInHCF (middle) and NaInHCF-H 2 O (bottom).
Supplementary Table 1. Structural parameter of InHCF in the Fm-3m structure determined from Rietveld method using powder X-ray diffraction data (Rietveld) and ab initio calculations with the GGA+U approximation (GGA+U).

Supplementary Note 1: the equal specific capacity of InHCF/Gr with Li + , Na + and K + .
Our InHCF/Gr is a cation-deficient (oxidation form) compound (see Supplementary Figure 2). An A + -intercalation reaction that occurs at InHCF can be written as follows: InHCF (Ox) + nA + (Aq) + ne   A n InHCF (Red) It can be seen that InHCF (Ox) has the same molecular weight and A n InHCF (Red) has different molecular weights with different A + . Due to its same molecular weight, the specific discharging capacities of InHCF/Gr (oxidation form) with Li + , Na + and K + are the same.

Supplementary Note 2: the coulombic efficiency and voltage charge limits of InHCF/Na + +K + /NaTi 2 (PO 4 ) 3 battery.
From the Supplementary   Figure 3 of the manuscript, the cut-off charging voltage for InHCF/Gr cathode is 1.2 V vs. SHE that is beyond 0.82 V. In terms of TiP 2 O 7 and NaTi 2 (PO 4 ) 3 anodes, their cut-off discharging voltages are -0.56 V and -0.65 V vs. SHE, respectively, which are lower than theoretical reaction potential of H 2 evolution (-0.41 V vs. SHE). Although the cut-off voltage limits for InHCF/Gr, TiP 2 O 7 and NaTi 2 (PO 4 ) 3 are beyond the theoretical reaction potentials of H 2 /O 2 evolution, their coulombic efficiency are around 100% (see Figure 3 of the manuscript), suggesting that water electrolysis reactions (both H 2 and O 2 evolution) are negligible.
The electrolytes solutions before and after cycling have been analyzed by ICP-OES. It is found that after cycling, the total concentration of M + and N + of the electrolytes is 1.04 ± 0.01 M, almost the same as the one before cycling (1.01 ± 0.01 M).

Supplementary Note 3: the fading mechanism of InHCF/Na + +K + /NaTi 2 (PO 4 ) 3 battery.
Our previous studies and other groups have found that oxygen that could react with the discharged-stated anode electrode materials is the primary cause of the capacity fading upon cycling. Although our used electrolytes here are all deaerated solution, we found that TiP 2 O 7 and NaTi 2 (PO 4 ) 3 undergo very slight fading upon cycling in deaerated solution (see Supplementary  Figure 9). It could be attributed to the trace of oxygen in the electrolytes. We also found that InHCF/Gr exhibits excellent cycle life (no obvious decaying is observed). So the fading of InHCF/Na + +K + /NaTi 2 (PO 4 ) 3 battery mainly comes from the NaTi 2 (PO 4 ) 3 anode.

Supplementary Note 4: the role of graphene.
In our InHCF/Gr, the mass ratio between InHCF and graphene is about 10:1. We choose the physical mixture of InHCF and graphene (InHCF+Gr) to study the effect of graphene on the rate performance of InHCF. In InHCF+Gr, the mass ratio of InHCF and graphene is also fixed at 10:1. Supplementary Figure 10 shows that discharging curves of InHCF+Gr at various rates in 0.5 M Li 2 SO 4 . As shown, InHCF+Gr at various rates exhibits higher capacity retention than InHCF, indicating that graphene serves as a 3D electronic network to diminish the resistance between InHCF nanoparticles, which facilitates the electron transfer. And InHCF/Gr shows slightly higher capacity retention than InHCF+Gr. It can be explained by the fact that the average size of InHCF/Gr is smaller than that of InHCF (see the manuscript). The reason is that 2D graphene adsorbs InHCF nanocrystals during the nucleation stage of precipitation reaction, which blocks their further growth up. Similar electrochemical results are also found in Na 2 SO 4 , K 2 SO 4 , and mixed-ion electrolytes. So it is concluded that graphene not only acts as a moderator during nucleation, but also serves as an electronic network.

Supplementary Note 5: the role of mixed-ion electrolytes.
Our previous studies [20,21] have shown that cubic TiP 2 O 7 and rhombohedral NaTi 2 (PO 4 ) 3 can be used as the anodes for AMIB, as a result of their specific ion-selectivity properties (TiP 2 O 7 , ion-selectivity toward Li + against Na + and K + ; NaTi 2 (PO 4 ) 3 , ion-selectivity toward Na + against K + ) and reasonable working voltages. When an intercalation reaction occurs at H with ion-selectivity towards A + as shown in equation (1), the reaction potential (E) can be calculated using the Nernst equation (2): where A + , a H and A H refer to the activities of A + , H and A + -intercalated H (A n H).
As the total concentration of M + and N + in our mixed-electrolytes is fixed at 1 M to ensure the equal conductivity of the electrolytes, the activity of a given cation in mixed-electrolytes (here we briefly consider it as the concentration of the cation) is below 1 M.  [20,21]. Therefore, equations (1) and (2) are valid for the electrode materials with ion-selectivity property. In terms of electrode materials that allow for co-intercalation of two alkali cations such as InHCF, these equations are not valid.
The CV data in our previous paper have also show that Li + (not Na + and K + ) can be intercalated into TiP 2 O 7 , and Na + (not K + ) can be intercalated into NaTi 2 (PO 4 ) 3 [20,21]. The ion-selective properties of above TiP 2 O 7 and NaTi 2 (PO 4 ) 3 can be easily explained by the steric effect: Na + and K + are too large to enter into TiP 2 O 7 three-dimensional framework while K + is also too large for NaTi 2 (PO 4 ) 3 . For AMIB, one important characteristic is that the total concentration of M + and N + in the electrolytes is fixed during charging/discharging, but the M + /N + ratio is changed. We have demonstrated this point in the paper [20].

DFT calculation.
During the charging and discharging of an alkali-ion battery, an alkali A + is inserted or extracted from a host crystal structure A n H. For a battery that operates by shuttling A + ions between the cathode and a pure alkali metal anode, the overall cell reaction can be written as follows: The forward reaction is the cell discharging reaction, while the reverse is the cell charging reaction. The average intercalation potential V vs. A/A + can then be calculated using the following equation [28,29]: Where E b is the binding energy between A + ion and H, E(A n H), E(H) and E(A) are the total energies of A n H, H and metallic A which are calculated using DFT, and e is the absolute value of the electron charge. In aqueous alkali-ion battery, m/n H 2 O accompany the insertion and extraction of one A + during entire charge and discharge process. So the overall cell reaction can be written by equation (3): The average intercalation potential V vs. A/A + can then be calculated using the following equation: