Layered Iron Vanadate as a High-Capacity Cathode Material for Nonaqueous Calcium-Ion Batteries

Calcium-ion batteries represent a promising alternative to the current lithium-ion batteries. Nevertheless, calcium-ion intercalating materials in nonaqueous electrolytes are scarce, probably due to the difficulties in finding suitable host materials. Considering that research into calcium-ion batteries is in its infancy, discovering and characterizing new host materials would be critical to further development. Here, we demonstrate FeV3O9∙1.2H2O as a high-performance calcium-ion battery cathode material that delivers a reversible discharge capacity of 303 mAh g−1 with a good cycling stability and an average discharge voltage of ~2.6 V (vs. Ca/Ca2+). The material was synthesized via a facile co-precipitation method. Its reversible capacity is the highest among calcium-ion battery materials, and it is the first example of a material with a capacity much larger than that of conventional lithium-ion battery cathode materials. Bulk intercalation of calcium into the host lattice contributed predominantly to the total capacity at a lower rate, but became comparable to that due to surface adsorption at a higher rate. This stimulating discovery will lead to the development of new strategies for obtaining high energy density calcium-ion batteries.


Introduction
Lithium-ion batteries (LIBs) have been an essential energy storage device for mobile applications in the last three decades and have received particular attention for their use in recently commercialized electric vehicles, primarily because of their outstanding energy densities and power [1]. Nevertheless, safety issues, high cost, and the maldistribution of lithium resources are driving researchers toward developing new materials, such as magnesium, calcium, zinc, or aluminum, that can be used as the carrier ion [2]. Among these, calcium-ion batteries (CIBs) have emerged as a highly promising post-LIB technology [3][4][5], particularly due to the recent innovative developments in their anodes [6][7][8].
Calcium ions can transfer two electrons per ion, and thus, in principle, the capacity of a host material can double due to the divalency of calcium, provided the host material can release and accept the transferred electrons. The cost of the final battery product is expected to be competitive, owing to the abundant calcium reserves. The redox potential of calcium is slightly higher (by 0.15 V) than that of lithium, enabling a high cell voltage. The larger ionic radius of calcium (1.0 Å) compared to those of other divalent ions (0.60-0.74 Å) could be a disadvantage because a narrow diffusion pathway sufficient for smaller ions will not allow the passage of larger Ca ions. However, the larger ionic radius of calcium ions (Ca 2+ ) ensures a lower effective intercalant-ion charge density (0.49 e/Å 3 ) than Zn 2+ (1.18 e/Å 3 ), Mg 2+ (1.28 e/Å 3 ), and Al 3+ (4.55 e/Å 3 ), which can be rather advantageous for the diffusion in the host materials and in the electrolytes [9].
Herein, we report that layered iron vanadate FeV 3 O 9 ·1.2H 2 O is a potential cathode material for nonaqueous CIBs. Calcium ions could be reversibly intercalated into the material with a high Coulombic efficiency at room temperature. The material has the highest capacity (303 mAh g −1 ) reported thus far, and has an excellent cyclability at room temperature. It is also the first example where the capacity exceeds those (150-250 mAh g −1 ) of conventional LIB cathode materials.
Iron vanadates are naturally abundant compounds and include fervanite (Fe 4 V 4 O 16 · 5H 2 O) [28], navajoite (FeV 9 O 24 ·12H 2 O) [29], kazakhstanite (FeV 3 O 9 ·2.6H 2 O or Fe 5 V 15 O 39 (OH) 9 ·9H 2 O) [30]. Among these, kazakhstanite has a layered structure with a large interlayer spacing (d 002 = 10.6 Å). It was first reported in 1989 as a new mineral, and its synthesis was reported later in 2002 [30,31]. Although it has been studied for lithium and zinc ion storage, a detailed material characterization is lacking and the intercalation mechanism is still elusive [32,33]. In this study, we attempted to synthesize kazakhstanite, but obtained a compound with the formula FeV 3 O 9 ·1.2H 2 O (FVO) under our synthetic conditions, which is similar to kazakhstanite in structure except for the crystal water content.

Synthesis and Materials Characterization
Layered iron vanadate FeV 3 O 9 ·1.2H 2 O was prepared via a facile co-precipitation method [33]. Three millimoles of NH 4 VO 3 (99%, Alfa Aesar, Haverhill, MA, USA) was dissolved into 100 mL of distilled water with stirring for one hour at 90 • C. Then, 10 mL of 0.1 M Fe(NO 3 ) 3 ·9H 2 O (98%, Sigma-Aldrich, St. Louis, MO, USA) solution was slowly added to the NH 4 VO 3 solution and kept for one day at 90 • C. The initial solution was transparent yellow, and the insoluble colloidal products were opaque brown. The brown precipitates were collected by filtration with aspiration and carefully washed with distilled water three times before being dried at 80 • C in air overnight. The analyses of morphology and elemental composition were performed using field-emission scanning electron microscopy (FE-SEM, Hitachi SU-8020), field-emission transmission electron microscopy (FE-TEM, Hitachi HF-3300) with an energy dispersive X-ray spectroscopy (EDX) attachment, and inductively coupled plasma optical emission spectroscopy (ICP-OES, Varian 700-ES). The amount of crystal water in the samples was determined by thermogravimetric (TG) analysis (Rigaku TG 8120). The trace amount of water in the electrolyte was analyzed by the Karl Fischer titrator (831 KF coulometer, Metrohm, Herisau, Switzerland).

Electrochemical Characterization
Customized homemade cells were used for the electrochemical measurements ( Figure S1). The cathode (working electrode) was composed of FeV 3 O 9 ·1.2H 2 O, conducting carbon (Super P, Timcal Graphite & Carbon), and poly(vinylidene fluoride) binder (W#1300, Kureha Co.) in an 8:1:1 weight ratio. A slurry was prepared by dispersing the mixture in N-methyl-2-pyrrolidone (NMP) using a planetary centrifugal mixer, and coated on 20 µm stainless steel foil (SUS-316L, Wellcos Co., Korea). The NMP solvent was removed by drying the coated electrode at 60 • C in a vacuum oven overnight before the electrode was pressed by an electrode rolling press (Wellcos Co., Seoul, Korea). The loading of FeV 3 O 9 ·1.2(H 2 O) was~2.3 mg on each electrode with an area of 1.53 cm 2 . As the counter and reference electrodes, the activated carbon pellet (~0.8 g) was used. As the organic electrolyte, 0.5 M calcium perchlorate (Reagent grade, Alfa Aesar) in acetonitrile (99.8%, Samchun Chemicals, Pyeongtaek, Korea) was used. The calcium perchlorate was dried at 220 • C for 3 days [34]. Molecular sieves were used to remove water in the electrolyte before use. The water content in the electrolytes was 27 ppm according to water analysis, while it was 38 ppm after the first cycle. As a separator, glass fiber (GF/A, Whatman) was used.
AC is stable in nonaqueous electrolytes, enabling us to use it as a reference electrode [35,36]. The relatively high loading of AC (~0.8 g) compared to the cathode loading (~2.3 mg) led the AC potential to be almost constant during the cycles. On discharge, calcium ions are intercalated into the cathode, while the anions in the electrolyte are adsorbed on the surface of the AC counter electrode. On charge, the reverse reaction occurs. The AC-pellet voltage was estimated as 3.08 V vs. Ca/Ca 2+ by comparing two CV curves: one using the Ag/Ag + reference electrode and the other using the AC reference electrode. Finally, the ferrocene/ferrocenium redox couple voltage was used to calibrate the Ag/Ag + electrode voltage.

Characterization of the Synthesized Materials
The phase of the FVO product was confirmed from the Le Bail fitting of the data (Figure 1a), which exhibited no visible impurities. Though the detailed crystal structure has not been determined yet, FVO is considered to have a layered structure including crystal water between layers [32,33] (a proposed structural model in Figure S2). The refined lattice parameters shown in the inset were very close to those for kazakhstanite (JCPDS No. 00-046-1334) [30]. In the Fourier transform infrared (FTIR) transmittance spectrum (Figure 1b), the bond at 1004 cm −1 corresponded to V=O stretching, while the absorption bands at 1621, 3191, and 3417 cm −1 indicated the existence of adsorbed water and crystal water in the host structure. The SEM and high-resolution TEM images further confirmed a nanosheet morphology (Figure 1c,d and Figure S3). The TEM image of the nano-flake edge showed layered fringes, indicating a lattice spacing of about 1.0 nm (Figure 1d, inset). The thickness of the nano-flake is less than 10 nm. The EDX elemental mappings obtained showed a uniform distribution of Fe, V, and O (Figure 1e). The ICP elemental analysis confirmed that the Fe:V atomic ratio was 1:3 (Table S1). Combining all the above analyses together with the crystal water content determined from thermogravimetric (TG) analysis ( Figure S4), the synthesized material (FVO) was formulated as FeV 3 [30] has more crystal water than we obtained in the lab, the amount of crystal water can be varied depending on the synthetic conditions. The influence of the water content on the properties should be investigated as a future work.
crystal water than we obtained in the lab, the amount of crystal water can be varied d pending on the synthetic conditions. The influence of the water content on the properti should be investigated as a future work.
The discharge-charge profiles of the second cycle at various current densities a shown in Figure 2b. The discharge capacity decreased significantly with increasing cu rent density. For instance, the discharge capacity at 200 mA g −1 was 32% of that at 20 m g −1 (96 mAh g −1 at 200 mA g −1 and 303 mAh g −1 at 20 mA g −1 ; see Figure S7 for rate perfo mance with cycles). It was also noted that the charge capacity contribution of the constan voltage step also decreased with increasing high current density. These phenomena ind cated the low rate-performance of the material due to the slow calcium intercalation k netics. However, an excellent cycling stability (Figure 2c) was observed-the first di charge capacity (105 mAh g −1 at 200 mA g −1 ) decreased notably till around the first te cycles but soon stabilized and remained so up to around the 300th cycle. It then gradual decreased up to the 400th cycle (83 mAh g −1 which is 79% of the first cycle).
The absence of any apparent plateau in the galvanostatic profile suggests that th total capacity might have a significant capacitive contribution. To quantify the contrib tions from the bulk-intercalation and surface-capacitive reactions [37,38], cyclic voltam mograms (CVs) were recorded in the voltage range between −1.5 and 1 V (vs. AC) at sca rates ranging from 0.05 to 0.5 mV s −1 (Figure 2d). From the current-rate relationship f

Electrochemical Performance of FeV
The galvanostatic discharge-charge profiles for FVO are presented in Figure 2a. The initial discharge and charge capacities were 334 and 297 mAh g −1 , respectively, at 20 mA g −1 .
The discharge-charge profiles of the second cycle at various current densities are shown in Figure 2b. The discharge capacity decreased significantly with increasing current density. For instance, the discharge capacity at 200 mA g −1 was 32% of that at 20 mA g −1 (96 mAh g −1 at 200 mA g −1 and 303 mAh g −1 at 20 mA g −1 ; see Figure S7 for rate performance with cycles). It was also noted that the charge capacity contribution of the constant-voltage step also decreased with increasing high current density. These phenomena indicated the low rate-performance of the material due to the slow calcium intercalation kinetics. However, an excellent cycling stability (Figure 2c) was observed-the first discharge capacity (105 mAh g −1 at 200 mA g −1 ) decreased notably till around the first ten cycles but soon stabilized and remained so up to around the 300th cycle. It then gradually decreased up to the 400th cycle (83 mAh g −1 which is 79% of the first cycle).
The absence of any apparent plateau in the galvanostatic profile suggests that the total capacity might have a significant capacitive contribution. To quantify the contributions from the bulk-intercalation and surface-capacitive reactions [37,38], cyclic voltammograms (CVs) were recorded in the voltage range between −1.5 and 1 V (vs. AC) at scan rates ranging from 0.05 to 0.5 mV s −1 (Figure 2d). From the current-rate relationship for each reduction and oxidation peak ( Figure S8) [37,38], it is evident that the contribution from intercalation was about 81% at 0.05 mV s −1 (Figure 2e), and the contribution decreased with increasing scan rates (58% at 0.5 mV s −1 ). Thus, calcium intercalation was dominant at a lower rate. each reduction and oxidation peak ( Figure S8) [37,38], it is evident that the contribution from intercalation was about 81% at 0.05 mV s −1 (Figure 2e), and the contribution decreased with increasing scan rates (58% at 0.5 mV s −1 ). Thus, calcium intercalation was dominant at a lower rate. According to the ICP elemental analysis (Table S2), the amount of calcium in the discharged FVO electrode was 1.9 Ca per formula unit, which was about 81% of the value (2.33 Ca) calculated from the discharge capacity (334 mAh g −1 ). Interestingly, the numerical value of the contribution (81% of the capacity) was coincident with the amount of bulkintercalated calcium (81% at 0.05 mV s −1 ) described above. Most probably, the surfaceadsorbed calcium atoms were washed out during the ICP sample preparation, thereby explaining the reduced calcium content. It was also noted that a small portion (0.2Ca) of the intercalated calcium atoms was trapped in the structure even after being charged (Table S2).

Elemental Analysis of FeV3O9•1.2H2O
It is noteworthy that the water content remained almost constant during a cycle, as revealed by the TG analysis ( Figure 3a); only the discharged sample showed an increased dehydration temperature, which resulted from the higher structural stability owing to the bonding between the intercalated calcium and the host layers. The retention of water could be an advantageous characteristic of this material when we consider a contrasting case of the potassium intercalation into VOPO4•2H2O, where the crystal water is removed from the structure upon discharge [39]. FTIR spectra also confirmed the existence of water at all the states of the samples during a cycle (Figure 3b). The SEM images showed that the morphology of FVO was well maintained, without any notable changes (Figure 3c). According to the ICP elemental analysis (Table S2), the amount of calcium in the discharged FVO electrode was 1.9 Ca per formula unit, which was about 81% of the value (2.33 Ca) calculated from the discharge capacity (334 mAh g −1 ). Interestingly, the numerical value of the contribution (81% of the capacity) was coincident with the amount of bulk-intercalated calcium (81% at 0.05 mV s −1 ) described above. Most probably, the surface-adsorbed calcium atoms were washed out during the ICP sample preparation, thereby explaining the reduced calcium content. It was also noted that a small portion (0.2Ca) of the intercalated calcium atoms was trapped in the structure even after being charged (Table S2).

Elemental Analysis of FeV 3 O 9 ·1.2H 2 O
It is noteworthy that the water content remained almost constant during a cycle, as revealed by the TG analysis ( Figure 3a); only the discharged sample showed an increased dehydration temperature, which resulted from the higher structural stability owing to the bonding between the intercalated calcium and the host layers. The retention of water could be an advantageous characteristic of this material when we consider a contrasting case of the potassium intercalation into VOPO 4 ·2H 2 O, where the crystal water is removed from the structure upon discharge [39]. FTIR spectra also confirmed the existence of water at all the states of the samples during a cycle (Figure 3b). The SEM images showed that the morphology of FVO was well maintained, without any notable changes (Figure 3c). Figure 4a shows the powder XRD patterns of the FVO electrodes at different states (pristine and charged/discharged states) during cycles, clearly evidencing the reversibility of the electrochemical calciation during the cycles. The shift of the (00l) peak to a higher 2θ angle on discharge indicated a decreased interlayer space, owing to the attractive interaction between the inserted calcium and the host layer. Upon charging, the peaks re-shifted to a position the same as that of the pristine state, revealing the structural reversibility. The high-resolution TEM image clearly showed a disordered lattice fringe upon discharge, which disappeared on charging (Figure 4b). Batteries 2021, 7, x FOR PEER REVIEW 6 of 9  Figure 4a shows the powder XRD patterns of the FVO electrodes at different states (pristine and charged/discharged states) during cycles, clearly evidencing the reversibility of the electrochemical calciation during the cycles. The shift of the (00l) peak to a higher 2θ angle on discharge indicated a decreased interlayer space, owing to the attractive interaction between the inserted calcium and the host layer. Upon charging, the peaks reshifted to a position the same as that of the pristine state, revealing the structural reversibility. The high-resolution TEM image clearly showed a disordered lattice fringe upon discharge, which disappeared on charging (Figure 4b).
This is in good agreement with the broadened peaks and recurring sharp peaks in the XRD patterns. The TEM EDX elemental mappings of pristine, discharged, and charged electrodes are shown in Figure 4c. A uniform distribution of calcium was clearly visible in the discharged FVO particles; this is hardly seen on the charged state, suggesting that calcium ions were the main transfer ions for the redox reaction. Figure 4d-f display the Xray photoelectron spectroscopy (XPS) survey spectra of Ca 2p, V 2p, and Fe 2p spin-orbitals. Calcium ions appeared during discharging and mostly disappeared upon charging, indicating a partial reduction of the vanadium ion, which disappeared on charging. The peak maxima of iron shifted slightly to lower binding energies during discharging and reshifted upon charging. The XPS spectra indicated that both vanadium and iron atoms were the redox centers for the reversible calciation-decalciation reaction.

Conclusions
In summary, FeV3O9•1.2H2O has been demonstrated as a new potential high-performance cathode host material with high capacity and excellent cyclability for nonaqueous calcium-ion batteries at room temperature. The material was synthesized via a facile coprecipitation method and showed high electrochemical performance as an calcium-ion battery cathode material. It exhibited a reversible discharge capacity of 303 mAh g −1 at 20 mA g −1 with a good cyclability and an average discharge voltage of ~2.6 V (vs. Ca/Ca 2+ ). It is the highest reversible capacity among those reported for calcium-ion battery cathode materials. Bulk intercalation contributed dominantly to the total capacity at lower scan rates; however, it became comparable to surface adsorption at higher rates. X-ray diffraction, elemental, electrochemical, and XPS analyses unambiguously evidenced the reversible calcium insertion into the host material. As a further study, the atomic-scale crystal structures of the synthesized and cycled phases should be determined to deeply understand the calcium storage mechanism of this material. This is in good agreement with the broadened peaks and recurring sharp peaks in the XRD patterns. The TEM EDX elemental mappings of pristine, discharged, and charged electrodes are shown in Figure 4c. A uniform distribution of calcium was clearly visible in the discharged FVO particles; this is hardly seen on the charged state, suggesting that calcium ions were the main transfer ions for the redox reaction. Figure 4d-f display the X-ray photoelectron spectroscopy (XPS) survey spectra of Ca 2p, V 2p, and Fe 2p spinorbitals. Calcium ions appeared during discharging and mostly disappeared upon charging, indicating a partial reduction of the vanadium ion, which disappeared on charging. The peak maxima of iron shifted slightly to lower binding energies during discharging and re-shifted upon charging. The XPS spectra indicated that both vanadium and iron atoms were the redox centers for the reversible calciation-decalciation reaction.

Conclusions
In summary, FeV 3 O 9 ·1.2H 2 O has been demonstrated as a new potential high-performance cathode host material with high capacity and excellent cyclability for nonaqueous calciumion batteries at room temperature. The material was synthesized via a facile co-precipitation method and showed high electrochemical performance as an calcium-ion battery cathode material. It exhibited a reversible discharge capacity of 303 mAh g −1 at 20 mA g −1 with a good cyclability and an average discharge voltage of~2.6 V (vs. Ca/Ca 2+ ). It is the highest reversible capacity among those reported for calcium-ion battery cathode materials. Bulk intercalation contributed dominantly to the total capacity at lower scan rates; however, it became comparable to surface adsorption at higher rates. X-ray diffraction, elemental, electrochemical, and XPS analyses unambiguously evidenced the reversible calcium insertion into the host material. As a further study, the atomic-scale crystal structures of the synthesized and cycled phases should be determined to deeply understand the calcium storage mechanism of this material.
In spite of the several potential advantages of using calcium, the development of a high-performance calcium-ion battery cathode material was known to be practically challenging to date. Therefore, our findings will provide a stimulus to the on-going research and will open up unexplored opportunities for discovering superior cathode materials for calcium-ion batteries.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/batteries7030054/s1, Figure S1: Scheme of the homemade cell used for electrochemical characterizations, Figure S2: A proposed crystal structural model for FeV 3 O 9 ·1.2H 2 O, Figure S3: TEM images of the synthesized FeV 3 O 9 ·1.2H 2 O particles, Figure S4: Thermogravimetric analysis for pristine FeV 3 O 9 ·1.2H 2 O sample under nitrogen flowing atmosphere, Figure S5: CV of FeV 3 O 9 ·1.2H 2 O electrode at 0.2 mV s −1 in 0.5 M Ca(ClO 4 ) 2 in AN with Ag/Ag + reference electrode in a threeelectrode system, Figure S6: Ferrocene/ferrocenium redox couple CV test to estimate the AC voltage vs. Ca/Ca 2+ , Figure S7: Cycle performance with various C-rates, Figure S8: (a) Log−log plot of cathodic/anodic peak current dependence on the scan rate. (b) Plot of peak current (i/v 1/2 ) dependence on the square root of scan rate (mV 1/2 s −1/2 ). Table S1: Elemental ratios estimated from ICP analysis for FeV 3 O 9 ·1.2H 2 O sample, Table S2: Elemental ratios estimated from ICP analysis for the pristine FeV 3 O 9 ·1.2H 2 O, first discharged, and first charged electrodes.