The active materials of the NaCoHCF electrodes were synthesized via a citrate-assisted controlled crystallization method [23]. Here, 100 mL of a 0.05 M CoCl₂ and 0.35 M sodium citrate solution was mixed with 100 mL of a 0.05 M Na₄Fe(CN)₆ solution under vigorous stirring for 24 hours. To synthesize the NaNiHCF particle, 100 mL of a 0.05 M NiCl₂ was added into 100 mL of 0.35 M Na-citrate and 0.05 M Na₄Fe(CN)₆ mixed solution under vigorous stirred condition. The obtained solutions were aged for 24 hours, and the precipitates were collected by filtration. During the filtration step, the precipitate was washed twice with 200 mL of distilled water and once with 200 mL of ethanol. The obtained active materials were dried in a 60°C oven overnight.

To fabricate the electrodes used in this study, 80 wt% of the prepared active materials, 10 wt% of a conductive material (Super P, Timcal, Switzerland), and 10 wt% of polytetrafluoroethylene (PTFE, Sigma-Aldrich, USA) were gently mixed with ethanol for 20 minutes. The prepared mixture was placed on a roll presser to create a sheet-type electrode approximately 250 μm thick. The prepared sheet-type electrode was dried in a vacuum oven at 120°C overnight to remove the residual solvent.

The morphologies and elemental distribution of the prepared NaCoHCF were characterized by a field emission scanning electron microscope (FESEM, JEOL JSM-7900F, Japan). X-ray diffraction (XRD, Rigaku, Japan) was used to analyze the crystal structure of the prepared NaCoHCF in the 2θ range of 10° – 70° with a ramping step of 1° min⁻¹.

The electrochemical properties of the NaCoHCF electrode were assessed in a three-electrode configuration. In this cell, a 1.0 cm × 1.0 cm NaCoHCF electrode attached to a titanium current collector served as the working electrode, an electrochemically prepared 2.0 cm × 5.0 cm AgCl electrode was used as the counter electrode, and a commercial Ag/AgCl electrode (in Saturated KCl) was used as the reference electrode, with a 1.0 M NaCl solution used as the electrolyte. Cyclic voltammetry (CV) scans were obtained using a potentiostat (PARSTAT 2273, Princeton Applied Research, USA) operated at a scan rate of 0.1 mV s⁻¹. A galvanostatic charge/discharge test was conducted from 0.05 V to 1.10 V under a constant current condition (± 0.1 A g⁻¹) using a battery cycler (WBCS3000, WonATech, Republic of Korea).

The two 2.0 cm × 2.0 cm area of NaCoHCF or NaNiHCF electrodes were attached onto each titanium current collector (thickness: 200 μm, Sigma-Aldrich, USA) with carbon paint (DAG-T 502, Ted Pella, USA). To ensure that one of the prepared electrodes was in a fully charged while the other was in a discharged state, one was discharged to −0.2 V (Na⁺ intercalated) and the other was charged to 1.1 V (Na⁺ de-intercalated) for 30 min in a 1.0 M NaCl solution using a three-electrode configuration with an Ag/AgCl (in saturated KCl) reference electrode.

A rocking-chair capacitive deionization (RCDI) cell was employed in this experiment. This cell consists of the prepared fully discharged and charged electrodes, two polyamide woven spacers, and an anion exchange membrane (AMV, Selemion, Japan). It is divided into two chambers, denoted here as chamber A with the fully discharged electrode and chamber B with the fully charged electrode, according to the positioning of the anion exchange membrane. The cell was covered by two PTFE and silicone gaskets to prevent the solution from leaking.

A volume of 0.4 mL of a 500 mM NaCl solution was placed in each chamber (total volume of solution in the cell: 0.8 mL). This cell was operated under constant current operation (± 1.0 mA cm⁻²) over a voltage range of −1.1 V to 1.1 V using a battery cycler (WBCS3000, WonATech, Republic of Korea). After the cell voltage reached the limit voltage, the solutions in both chambers were refreshed with new NaCl solution in amounts of 500 mM. The cation concentration of the extracted solution after each step was measured by means of ion chromatography (ICS-1100, Thermo Fisher Scientific Inc., USA).

The electrochemical desalination performance parameters were calculated based on the results of a full cycle, defined as charging and discharging steps. The ion removal efficiency was calculated as follows:

The salt adsorption capacity (SAC) was calculated using the Eq. (2).

where M.W_NaCl^is the molecular weight of NaCl (58.4 g mole⁻¹), m_e is the total weight of both electrodes including the Super P and PTFE (g), ΔC_i is the concentration change of step i (mM), and V_c is the volume of the chamber used (mL).

The charge efficiency (Λ) was calculated as Eq. (3).

where F is the Faraday constant, z_i is the valance state of i, Δn_i is the molar change of i, and ∑ is the transferred total charge.

The specific energy consumption (SEC) was calculated via Eq. (4).

where I is the applied current, m is the mole of the removed NaCl, t_cycle is the total operation time, and V_cell is the cell voltage during operation.

Fig. 2 shows the crystalline structure and morphology of the prepared NaCoHCF electrode.

As shown in Fig. 2(a), the XRD diffraction line of the synthesized NaCoHCF is clear and in good agreement with those in previous studies, indicating that NaCoHCF with highly crystalline face-centered cubic (FCC) structure was synthesized. Moreover, the uniformly distributed C, N, O, Na, Fe and Co elements in the energy dispersive spectroscopy (EDS) mapping images support the synthesis of high-purity NaCoHCF (Fig. 2(b)). In a detailed comparison with the standard diffraction pattern of Prussian blue (JCPDS 52-1907), it can be seen that the lines are split by doublets at 220, 420, 440, and 620, which means that PBA is in a rhombohedral state. The result of the rhombohedral state reveals that the synthesized NaCoHCF material is in a state where two sodium ions are intercalated within the PBA structure. Typical PBA has a cubic structure, but it is known that when secondary sodium is inserted into the PBA frame, the cubic structure is distorted and changes to a rhombohedral state [23].

As can be seen in the FESEM image in Fig. 2(b), a highly crystalline NaCoHCF material of approx. 5 μm was synthesized. These synthesized particles were larger than regular PBAs, with particles as small as 50 nm. While the PBA particles prepared by the conventional method are small in size due to their fast reaction rate and low solubility, we used a kinetically controlled crystallization method to synthesize NaCoHCF with a higher order and a larger particle size. This method was reported in a previous study using Na-citrate as a chelating agent to slow the reaction rate through the slow release of Co²⁺. Finally, a sheet-type electrode with a thickness of approx. 200 μm was well fabricated. Through a FESEM analysis, for the fabricated sheet-type electrode, it was additionally confirmed that the active material (NaCoHCF), the conductive material (Super P), and the binder (PTFE) were well mixed, as shown in Fig. 2(c).

To clarify the electrochemical properties of the NaCoHCF electrode further, cyclic voltammetry (CV) and galvanostatic charge/discharge testing were conducted with a 1.0 M NaCl solution and a three-electrode configuration (Fig. 3). As shown in Fig. 3(a), the CV curve presents two distinguishable oxidation/reduction peaks at 0.4 V and 0.9 V. The shape of this curve is the typical CV curve of intercalation materials, which demonstrates that Na⁺ can be reversibly de-/intercalated. More importantly, the presence of two redox peaks indicates that the electrode follows the 2-Na reaction mechanism. The 2-Na reaction mechanism of NaCoHCF originates from two-redox active site reactions of the Co²⁺/Co³⁺ and Fe²⁺/Fe³⁺ couples at 0.4 V and 0.9 V, respectively. The charge-spin lattice strength capabilities of the two transition metals differ due to the fact that Co is nitrogen-coordinated and Fe is carbon-coordinated, which causes the two redox active sites to have different operating potentials [24]. This redox potential differs significantly from the typical electrode potential of Fe(CN)₆⁴⁻/Fe(CN)₆³⁻, and is even higher than that of the Co²⁺/Co³⁺ couple, due to the unstable condition of the Fe²⁺/Fe³⁺ couple in the Co–Fe framework [25]. In addition, the Na⁺ insertion/extraction reaction during the Co²⁺/Co³⁺ couple can lead to a significant volume fluctuation to accommodate the structural change and this broad peak between 0.4 to 0.65 V expected to show by the volume changes [26]. Consequently, in the Co-Fe framework Prussian blue, the redox potential of Fe²⁺/Fe³⁺ is higher than that of Co²⁺/Co³⁺ pair.

Fig. 3(b) shows the galvanostatic charge/discharge profile at a current density of 0.1 A g⁻¹ in the 1.0 M NaCl solution. It was confirmed that the potential profile had two plateaus at 0.4 V and 0.9 V. This result is consistent with the CV results, implying that the NaCoHCF electrode follows the 2-Na reaction mechanism at a current density of 0.1 A g⁻¹. This 2-Na reaction mechanism leads to the very high specific capacity of 88 mAh g⁻¹ in this case (active material: 110 mAh g⁻¹). This specific capacity of the NaCoHCF electrode is the highest among faradaic electrodes applied to electrochemical desalination thus far, as shown in Table 1. In addition, the Coulombic efficiency, representing the ratio of the discharge capacity and the charge capacity, is 99.7%, indicating that the NaCoHCF electrode operates reversibly without accompanying side reactions such as water decomposition. Consequently, the properties of the NaCoHCF electrode with superior specific capacity and reversibility suggest that it is a promising cation-reactive electrode material for CDI.

Fig. 4 shows the desalination performance of RCDI using the NaCoHCF electrodes under constant current operation at a current density of 1.0 mA cm⁻² in a 500 mM NaCl solution. As shown in Fig. 4(a), the system operates reversibly in the cell voltage range of −1.1 V to 1.1 V in the 500 mM NaCl solution. During the charging and discharging steps, the cell voltage profile of this system gradually changed in the range of −0.9 V to 0.9 V, whereas it changed rapidly at both ends. This result implies that the full capacity of the NaCoHCF electrodes is utilized at the set cell voltage limit. The potential profile of each electrode demonstrates that the electrode uses its full capacity at the set cell voltage limit. The measured potential profiles of the both electrodes have two plateau regions at 0.4 V and 0.9 V, meaning that the aim of this study was achieved by using the two redox active sites of the NaCoHCF electrodes within the set cell voltage range.

Fig. 4(b) shows the resulting ion concentration changes after each step. As described above, an important characteristic of RCDI is that desalination and concentration occur simultaneously in one step. In step 1, concentration occurs in chamber A and desalination occurs in chamber B. Step 2 works exactly the opposite of step 1. After the completion of steps 1 and 2, the Na⁺ concentrations in chambers B and A were 55 mM and 68 mM, resulting in 88% salt removal on average in 500 mM of saline during one cycle. This demonstrates close to a 10% increase in salt removal compared to the rates in previous studies that utilized Ag/AgCl electrodes, which to the best of our knowledge represents the highest salt removal rate of any CDI system applied to saline concentrations above 500 mM [15]. The high salt removal rate of 88% is attributed to the high specific capacity of the NaCoHCF electrodes (88 mAh g⁻¹) stemming from the use of two redox active sites.

Fig. 5 shows a comparison of the desalination performance outcomes of the NaCoHCF cell and the NaNiHCF cell at 1 mA cm⁻² in 500 mM of NaCl. To ensure a fair comparison of the system performance changes due to the introduction of the NaCoHCF electrode, NaNiHCF, one of PBAs applied to a typical CDI system, was operated under identical conditions apart from the cell voltage range. The cell voltage ranges were set differently because the working potential ranges of the NaCoHCF (0.4 – 0.9 V vs. Ag/AgCl) and NaNiHCF (0.5 – 0.7 V vs. Ag/AgCl) electrodes are different. Fig. 5(a) confirms that the cell capacity per unit electrode weight of the NaCoHCF cell (272 C g⁻¹) was approximately 1.5 times higher than that of the NaNiHCF cell (180 C g⁻¹). The higher specific capacity of the NaCoHCF electrode is attributed to the use of two redox potential regions of the NaCoHCF electrode. The four plateau regions of the NaCoHCF cell in the cell voltage profile in Fig. 5(a) indicate that two redox potential active sites were used in both charge/discharge steps. This result is well supported by the voltage potential profile of each electrode in the NaCoHCF cell. On the other hand, the voltage profile of a NaNiHCF cell with two plateau regions in the charge/discharge steps implies that one redox active site was employed.

Fig. 5(b) compares the changes in the quantitative electrochemical desalination performance capabilities of the cells with the NaCoHCF electrodes and the NaNiHCF electrodes. The NaCoHCF cell achieved approx. 88% ion removal and a SAC value of 123 mg g⁻¹, representing a 48% improvement in the ion removal efficiency and a 43.7% higher SAC value compared to the NaNiHCF cell. This result indicates that the NaCoHCF cell not only allows a longer operating time compared to the NaNiHCF cell but that the longer operating time was in fact used to increase the desalination capacity. In addition, it was confirmed that the charging efficiency, which is the ratio of the amount of charge actually used for desalination to the amount of charge applied to the system, was higher in the NaCoHCF cell (88.8%) than in the NaNiHCF cell (82.5%). This result means that the NaCoHCF electrode applied to the RCDI cell is a very efficient for desalting highly concentrated saline water. In particular, the SAC value of 123 mg g⁻¹ (active material: 154 mg g⁻¹) of the NaCoHCF cells is the remarkable value in relation to CDI, as shown in Table 1. The high SAC value of the NaCoHCF system is due to the high capacity of the NaCoHCF electrode and the continuity at which desalination can occur in both steps 1 and 2, a unique characteristic of the RCDI system [20].

Although the NaCoHCF cell showed the most improved desalination performance compared to the NaNiHCF cell, the specific energy consumption increased by 2.5 times. The NaCoHCF cell has a higher SEC value than the NaNiHCF cell because the resistance of the electrolyte increases as the ion removal rate increases. The final concentration of the NaCoHCF cell is approx. 60 mM, whereas the final concentration of the NaNiHCF cell is approx. 200 mM. Accordingly, it can be seen that the solution resistance of the NaCoHCF cell is much larger. This result is also well supported by the voltage drop difference results between the NaCoHCF cell and NaNiHCF cell. A voltage drop occurs when going from step 1 to step 2 in the cell voltage profile. The voltage drop of NaCoHCF cell was 0.63 V, which is 2.5 times higher than the voltage drop of the NaNiHCF cell (0.25 V). The main causes of the cell voltage drop are known to be polarization through the solid-liquid interface, the resistance of the electrode and ion exchange membrane, and the electrolyte resistance. In this case, the difference in the voltage drop between the two cells is expected mainly to stem from the electrolyte resistance according to the difference in the ion removal rate. It is a fact that the operation condition of current density (1.0 mA cm⁻²) is not enough for sufficient rate performance compared with conventional CDI systems, and that is one of the main limitations of the RCDI using battery electrode materials [18]. Given that an increase in SEC according to an increase in the ion removal rate is inevitable, it is considered that a follow-up study to find the optimal condition between the ion removal rate and SEC value is required.