| Home | E-Submission | Sitemap | Contact Us |  
Environ Eng Res > Volume 28(1); 2023 > Article
Zhang and Yang: Single-pass capacitive deionization with a HNO3-modified electrode for fluoride removal


HNO3-modified activated carbon was used to make electrodes for single-pass capacitive deionization for removing F from drinking water. The optimal operating conditions for F removal were studied, and the F removal performance, cycle stability, and charge efficiency of the electrode were investigated. Based on these results, an optimization scheme was proposed for practical applications. After HNO3 modification, the proportion of micropores, specific surface area, and number of oxygen-containing functional groups on the activated-carbon surface increased, resulting in a significant increase in the specific capacitance of the electrode. Under optimal operating conditions, the adsorption capacity of the modified electrode was 13% higher than that of the unmodified electrode, while the charge efficiency increased by 25% and reached a peak value after about 1,100 s. The HNO3-modified electrode had good cycle stability, and maintained 83% of the original adsorption capacity after 5 cycles. Optimizing the adsorption time (1,500 s) and desorption time (900 s), 80% of the specific adsorption capacity was maintained after 5 cycles. In addition, the cycle time was reduced by 32%, and the utilization rate of electric-double-layer adsorption sites was optimized, resulting in a reduction in the energy consumption per unit F removal.

1. Introduction

Fluorine (F) has the smallest ionic radius and strongest nonmetallic properties of all halogen elements. Because of its high electronegativity, it can form compounds with many other elements. Fluoride is mainly introduced into the human body through drinking water, as an appropriate amount of fluoride can form fluorapatite on the surfaces of teeth, inhibit lactobacillus growth in the mouth, prevent dental caries, facilitate the absorption of phosphorus and calcium, and promote bone development [1]. However, excessive intake of fluoride can easily lead to dental and skeletal fluorosis, osteoporosis, joint stiffness, and other diseases, and also harm the nervous system and liver [2]. Studies have shown that a concentration of 0.7 mg/L is optimal for human health [3]. China’s current Sanitary Standard for Drinking Water (GB5749-2006) further increases the appropriate concentration of fluoride in drinking water to 0.5–1.0 mg/L. About 100 million people in China are exposed to excessive fluoride in their drinking water [4]. Presently, common methods for removing fluoride from drinking water include adsorption [5], electrodialysis [6], chemical precipitation [7], coagulation precipitation [8], and ion exchange [9]. However, these technologies have disadvantages, such as secondary pollution caused by additives, high cost, low efficiency, and difficult regeneration.
Capacitive deionization (CDI) is a new desalination technology based on the principle of supercapacitor and electric-double-layer (EDL) theory. When a voltage is applied to an electrode, an electrostatic field is formed, and the ions in the water are adsorbed by the electrode with the opposite charge under the action of the concentration gradient and electric field, thereby reducing the concentration of ions in the water. When the electrode is saturated, the electric field is removed and the adsorbed ions can be desorbed into the solution. CDI has the advantages of low energy consumption, no secondary pollution, simple operation, and low cost. This technology won the “People’s Choice Award” for innovation from the World Association of Industrial and Technical Research Organizations in 2020 [10]. Research on CDI includes two operating modes: batch mode and single-pass mode. In batch mode, the raw water is recycled and processed by the CDI device. This mode is widely used in CDI research as it is easy to operate and convenient for comparing experimental and theoretical analyses. However, in practical industrial engineering, to ensure the continuity of the water supply and a sufficiently high treatment volume, the single-pass mode is mostly adopted, in which the raw water passes through the CDI device only once and is not recycled. To date, there are few theoretical studies on this mode.
The electrode material is the most important factor affecting the CDI performance. Activated carbon (AC) is commonly used as an electrode material because it is easy to prepare, and has low cost, high electrical conductivity, high specific surface area, and a suitable pore-size distribution. However, disadvantages such as a low adsorption capacity, high ash content, and poor hydrophilicity limit the efficiency of AC electrodes [11]. By oxidizing the AC surface, the number of oxygen-containing functional groups on the surface can be increased, the hydrophilicity and wettability can be improved, the diffusion resistance of ion transfer in the pore can be reduced, and the adsorption capacity can be increased [12]. HNO3 is often used to modify AC electrodes as it is highly oxidizing and can add oxygen-containing functional groups to the AC surface and remove some impurities from the AC surface [13]. Huang et al. [14] used HNO3-modified AC electrodes to conduct desalination experiments, and showed that HNO3 modification greatly increased the number of oxygen-containing functional groups on the AC surface, increased the capacitance, reduced the charging resistance, increased the desalination rate by ~15%, and increased the kinetic rate constant of desalination from 0.09208 to 0.09922. Further, Bao et al. [15] modified multi-walled carbon nanotube electrodes with HNO3 and investigated the effect of HNO3 modification on the desalination performance of CDI. The modification had little effect on the morphology, specific surface area, and pore-size distribution of the carbon nanotubes, but significantly increased the number of oxygen-containing functional groups on the electrode surface and the specific capacitance of the electrode, and greatly increased the desalination efficiency. The adsorption process was described well by the Freundlich model. To date, there is no existing study on the removal of fluoride by CDI using HNO3-modified AC electrodes.
In this study, HNO3-modified AC was used to prepare CDI electrodes, and single-pass mode was used to conduct the CDI fluoride removal experiments as it is more relevant for upscaled practical applications. We analyzed the specific surface area, pore size distribution, and the surface functional groups of AC before and after modification. The influence of the operating conditions on the fluoride removal and cycle stability of the modified electrode is discussed. In addition, we compared the adsorption performance and charge efficiency of the electrodes before and after modification, and fit the data with a kinetics model to understand how HNO3 modification affects the adsorption mechanisms of the AC electrode. A feasible optimization scheme is proposed to provide a reference for improving fluoride removal via CDI considering modification of the electrode material and operation mode.

2. Experimental

2.1. Materials

Sodium fluoride (NaF), nitric acid (HNO3), hydrochloric acid (HCl), sodium hydroxide (NaOH), and N,N-dimethylacetamide (DMAC) were purchased from Tianjin Damao Chemical Reagent Factory. Supercapacitor activated carbon was purchased from Fuzhou Yihuan Carbon Ltd. Polyvinylidene fluoride (PVDF) and conductive carbon black were purchased from France Arkema Ltd. Anion/cation exchange membranes (AMI7001/CMI7000) were purchased from Membranes International Ltd., graphite sheet (500 μm thickness) and Cu tape were purchased from Changsha Spring. The reagents used in the experiments were all of analytical grade.

2.2. Modification of Activated Carbon

First, 7.5 g of supercapacitor AC powder was placed in a beaker, and 70 mL of 1 mol/L aqueous HNO3 solution was added, followed by ultrasound treatment for 10 min during which time the solution was stirred using a glass rod stir to ensure that the AC fully dispersed. Then, the solution was heated in a constant-temperature water bath at 50°C with magnetic stirring for 12 h. After cooling to room temperature (~25°C), the solution was vacuum filtered and rinsed with ultrapure water until a neutral pH was achieved. After solid–liquid separation, the AC powder was placed in a thermostatic drying oven at 60°C for 12 h.

2.3. Electrode Preparation

The graphite sheet (as a current collector) was cut into 50 × 50 mm pieces and washed with ultrapure water to remove ash and impurities on the surface, followed by drying in a constant-temperature drying oven, cooling to room temperature, and weighing. Supercapacitor AC powder, PVDF, and conductive carbon black were dissolved in an appropriate amount of DMAC at a mass ratio of 8.5:1:0.5, and magnetic stirring was conducted for 6 h to produce a uniform slurry. The AC slurry was coated with a thickness of 250 μm on the graphite collector with a coating machine (SZQ-500μm, Tianjin Kexin, China). The slurry was dried in a drying oven at a constant temperature of 60°C for 2 h, and then in a vacuum drying oven at 55°C for 2 h to remove the organic solvents. The mass of the electrode was measured after cooling to room temperature (~25°C), and the loading of the electrode was determined as the weight difference between the electrode and collector.

2.4. Experimental Setup

As shown in Fig. 1, the experimental CDI device consists of a beaker containing the raw-water solution, a DC power supply (TPR3005T, Shenzhen, China), a peristaltic pump (BT200D, Shanghai, China), and a CDI reactor. The CDI reactor is made of two 10×10 cm plexiglass plates sandwiched together with silicone gaskets along the four sides between the two plates to form a cavity. The electrodes were placed in the cavity, with a distance of 3 mm between them. The electrodes were attached with Cu-tape connectors to conduct electricity. To increase the hydraulic retention time of the solution, the flow direction of the solution in the reaction tank is downward in and upward out. In the experiment, the same type of HNO3-modified AC electrode was used for both the anode and cathode, and the same unmodified AC was used for both electrodes in the comparative experiment.

2.5. CDI Experiments

NaF solution was used to simulate raw water in each experiment. During the experiments, the raw-water beaker was placed on a magnetic stirrer (HJ-2A, Jiangsu, China) to ensure that the solution was homogeneous. The raw water was transported to the CDI reactor under a constant voltage by a peristaltic pump at a certain flow rate. Samples for further analysis were taken from the outlet pipe (close to the exit of the reactor). Time zero of the measurements was defined as the time when the raw water entered the CDI device. After adsorption for 50 min, the power was turned off, and desorption was analyzed for 30 min. The fluoride ion concentration in the effluent was measured by a fluoride-ion-selective electrode (PXS-F, Zhejiang, China) at certain time intervals.

2.6. Kinetic Fitting

Kinetic fitting is a basic method used to analyze ion transport in CDI electrodes. In this section, the pseudo-first-order kinetic fitting of the CDI fluoride removal process at different raw water concentrations is described, with the aim of determining the transport mechanism.
The pseudo-first-order kinetic adsorption model is as follows:
ln(qe-qt)=ln qe-K1t
Here, qt is the specific adsorption volume at time t (mg/g), qe is the saturation specific adsorption volume (mg/g), and K1 is the adsorption rate constant (s−1). The adsorption (or desorption) capacity Q (mg/g) in single-pass mode is defined by:
Here, C0 (mg/L) and Ci (mg/L) are the influent and effluent fluoride ion concentrations, respectively, V (mL) is the solution volume, and m (g) is the loading of AC on the electrode. The charge efficiency Λ (%) in single-pass mode is defined by:
Here, F (C/mol) is Faraday’s constant and I (A) is the measured current. The desorption rate η (%) of electrode is calculated using:
Here, Qa and Qd are the adsorption and desorption capacities, of the electrode, respectively after electrode adsorption saturation and complete desorption, respectively.
The cyclic voltammetry (CV) curve of the electrode was measured using an electrochemistry workstation (CHI600E, Shanghai). The test conditions were as follows: AC electrode (25 × 50 mm) as the working electrode, Pt electrode as the counter electrode and saturated calomel electrode as the reference electrode. The electrolyte was 0.5 mol/L NaCl solution, and the scanning rate was 10 mV/s. The specific capacitance C (F/g) was calculated using:
Here, v (V/s) is the potential scanning rate and ΔV is the potential difference.
The specific surface area, pore volume, and pore-size distribution of AC powder were determined by low temperature N2 adsorption/desorption experiments (ASAP2460, Micromeritics, USA). The surface morphology of the electrode was observed by scanning electron microscopy (SEM, S4800, Hitachi, Japan). The surface functional groups of the AC powder were analyzed by Fourier-transform infrared spectroscopy (FTIR, Spectrum 10.0, Perkin Elmer).

3. Results and Discussion

3.1. Characterization of Activated Carbon and Electrode

The N2 adsorption/desorption isotherm and pore size distribution of AC before and after modification are shown in Fig. 2, while the corresponding specific surface area, pore size, and pore volume extracted from these data are shown in Table S1. The AC before and after modification both showed type-I adsorption isotherms (Fig. 2(a)). The cumulative adsorption volume increased rapidly in the low-pressure zone (P/P0 < 0.1) with increasing pressure, and changed little in the moderate- and high-pressure zones, with a value close to the adsorption limit. Compared with the original AC, the adsorption volume of HNO3-modified AC was clearly higher. Although the AC before and after modification was dominated by micropores (Fig. 2(b)), HNO3 modification increased the micropore volume and total pore volume, decreased the mesopore volume and average pore size, and increased the specific surface area by 25% (Table S1). The strong etching effect of concentrated HNO3 etched the pores in the AC, which caused partial collapse of the pores and reduced their size. In addition, this process removed ash from the surface of the AC, resulting in more of the AC surface being exposed, which contributed to the increase in the specific surface area, total pore volume, and micropore volume [16].
Fig. 3(a) shows the FTIR spectra of the AC powder before and after HNO3 modification. Compared with the original AC, the infrared spectra of the modified AC had obvious stretching vibration absorption peaks at 1,720 cm−1 (characteristic peak of the carboxyl group, C=O) and 3,430 cm−1 (characteristic peak of the hydroxyl group −OH). After HNO3 modification, the content of carboxyl and hydroxyl groups on the surface of the AC increased, while that of the oxygen-containing functional groups increased significantly. Literature [14] shows that the increase of oxygen-containing functional groups on the electrode surface can improve the hydrophilicity of the electrode, reduce the contact resistance and improve the adsorption performance. In addition, Huang et al. [17] observed an obvious absorption peak at 1,380 cm−1 (characteristic peak of −NO2) on the AC surface after modification with high-concentration HNO3.
Fig. 3(b) shows the CV curves of the AC electrodes before and after HNO3 modification. Both CV curves had a rectangular shape with no obvious redox peaks, consistent with the typical EDL capacitance behavior. Therefore, ion adsorption on these electrodes was dominated by the EDL capacitance, and no pseudo capacitance caused by the chemical reaction was observed. Therefore, HNO3 modification did not change the adsorption mechanism. The specific capacitance of the original AC electrode was 37.25 F/g, and that of the HNO3-modified electrode was 47% higher (55 F/g). This was attributed to the increase in the specific surface area and total pore volume, which provided more adsorption sites on the electrode surface. Meanwhile, the stronger hydrophilic behavior of the electrode allowed better wetting of the surface by the solution, which enhanced contact with the ions and increased the EDL capacitance. This is consistent with the conclusion of Wang et al. [18] that HNO3-modified AC electrodes can obtain excellent specific capacitance.
Fig. S1 shows SEM images of the surface of the AC electrodes. Under the adhesive action of PVDF, the AC and conductive carbon black were uniformly dispersed on the surface of the current collector, and the AC surface had a network structure and porous channels. The strong corrosive effect of HNO3 can remove ash from the AC surface, resulting in a clean and rough surface with a more uniform pore structure. This will facilitate the contact between the solution and the electrode, and make full use of the specific surface area of the AC to improve the adsorption performance [17].

3.2. Effect of HNO3 Modification on Fluoride Removal

The adsorption capacity and charge efficiency are important indexes to evaluate the performance of CDI electrodes. CDI experiments were performed to evaluate the fluoride removal efficiency using the original or modified AC electrodes under the following conditions: raw-water F concentration of 50 mg/L, voltage of 1.5 V, flow rate of 7 mL/min, and adsorption time of 3,000 s. The electrode adsorption capacity was determined from the measured changes in F concentration of the effluent during CDI, as shown in Fig. 4.
As shown in Fig. 4(a), after adsorption began, the effluent F concentration decreased rapidly and reached the lowest value around 400 s, and then began to increase slowly. After 2,000 s, the effluent F concentration reached that of the influent and stayed stable, indicating that the electrode reached saturation. In the initial stage of adsorption (400–900 s), the effluent F concentration of the HNO3-modified electrode was lower, while after 1,000 s, it was similar to that of the original electrode. This could be due to the high adsorption efficiency of the adsorption sites in the initial stage of adsorption, which results in many adsorption sites being occupied in the pore channels, and a high resistance to further ions entering the pores, leading to a slight decline in later adsorption. After adsorption for 3,000 s, the adsorption capacity of HNO3-modified electrode reached 3.58 mg/g, which was 13% higher than that of original AC carbon electrode (3.17 mg/g). The charge efficiency is reduced by the resistance of the electrode itself, the ion transport resistance in the solution and in the electrode channel, the “co-ion effect”, and counterion attraction [19].
As shown in Fig. 4(b), after the voltage is applied to the electrode, a large initial current is generated, which then drops rapidly and gradually stabilizes. When adsorption begins, the charge efficiency increases rapidly, reaches a peak, and then decreases as the electrode reaches saturation. At this point, many adsorption sites in the pores are occupied, resulting in high resistance to ion transport that greatly affects adsorption. This is observed as the current stabilizing and approaching zero, accompanied by a drop in charge efficiency. According to Eq. 3, the charge efficiency of the HNO3-modified electrode after adsorption for 3,000 s was 22.7%, which is about 1/4 higher than that of the original electrode (18.14%). During the adsorption process, the charge efficiency of the original electrode reached a peak of ~30% after ~800 s, while that of the HNO3-modified electrode reached a peak of ~35% after ~1,100 s. The charge efficiency of the electrode in this study was lower than that of the electrode presented by Zhao et al. [20], which may be related to the existence of a leakage current in the circuit. It was observed in the experiment that after the electrode was saturated, the current did not drop to the theoretical value of 0, and a small current was measured, which is proof of a leakage current.
The results shown in Fig. 4 are consistent with the conclusions of other studies [21]. The enhanced performance of the electrode after modification was attributed to the increase in the specific surface area, resulting in an increase in the capacity of the EDL and number of adsorption sites. In addition, the addition of acidic oxygen-containing functional groups on the AC surface increases the hydrophilicity and wettability of the electrode, which facilitates contact between the electrode and the solution and reduces the diffusion resistance of ions entering the EDL. In turn, this increases the specific capacitance, adsorption capacity, and charge efficiency of the electrode.
The existence of a peak charge efficiency during CDI is of great significance for actual CDI processes. If the adsorption is stopped at a high charge efficiency and desorption is initiated, the energy consumption of F per unit mass adsorption can be minimized, providing a high electrode efficiency.
To obtain a deeper understanding of the effect of HNO3 modification on the diffusion of F in the electrode channels during adsorption, raw water solutions with F concentrations of 25, 40, 50 and 60 mg/L were prepared and adsorption was performed for 3,000 s at a flow rate of 7 mL/min and voltage of 1.5 V. The adsorption capacity over time is shown in Fig. 5(a). Kinetic fitting was performed for the CDI fluoride-removal process with the electrode before and after modification at different raw water concentrations; the results are shown in Figs. 5(b)–(d).
During CDI, F mainly undergo physical adsorption on the electrode surface, and chemical reactions are negligible, following pseudo-first-order reaction kinetics. The adsorption rate is proportional to the residual adsorption capacity of the EDL at the interface between the electrode and solution [22]. Fig. 6(b) and (c) show that the results for both the original and HNO3-modified electrode were fit well by the pseudo-first-order kinetic model, with fitting correlation coefficients above 0.97, indicating that the modification did not change the ion adsorption mechanism of the electrode. This is consistent with the discussion of the CV curves.
As shown in Figs. 5(a) and (c), the adsorption capacity and adsorption rate constant (k) increased with increasing F concentration in the influent. When the F concentration increased from 25 to 60 mg/L, the adsorption capacity increased from 2 to 4.12 mg/g. respectively. Raw water with higher ion concentration has higher conductivity, which reduces the resistance of ion migration to the electrode surface, and increases the ion adsorption rate. This is consistent with the observation of a high initial current with high-concentration raw water. In addition, high-concentration solutions have a weaker “overlap effect” of the EDL at the interface between the solution and electrode, resulting in a complete EDL structure and high adsorption capacity [23]. This is in contrast with a previous study [24], which showed that the adsorption capacity of the electrode at different raw-water concentrations was similar. This is attributed to a low concentration of total dissolved solids (TDS) in the raw water used in this study, the short residence time in single-pass mode, the weak concentration gradient between the solution and EDL, and partial utilization of the AC pores. These factors contribute to the adsorption mechanism being highly sensitive to changes in the raw-water concentration. As shown in Fig. 5(a), the adsorption capacity of the modified electrode increased slowly over the initial 250 s, which may be due to incomplete wetting of the electrode surface, similar to the original electrode. When the electrode is completely wetted, the resistance of ions entering the EDL decreases, and the adsorption capacity begins to increase rapidly. In the later stage of adsorption, the residual capacity of the EDL decreases, resulting in the adsorption capacity gradually reducing until saturation is reached.
According to the fitting results shown in Fig. 5(d), when the F concentration in the raw water was below 50 mg/L, the average adsorption rate of the HNO3-modified electrode was higher than that of the original electrode. However, when the F concentration increased to 60 mg/L, the average adsorption rate of the modified electrode was lower than that of the original electrode. This may be due to the enhanced wettability provided by the additional acidic oxygen-containing functional groups after HNO3 modification. However, when the F concentration is high, although ions can quickly enter deep into the AC pores under the high concentration gradient, the modified electrode has a larger proportion of small micropores, which provide higher resistance to ion migration than larger pores, resulting in the average adsorption rate constant of the modified electrode being lower than that of the original electrode. The fitting formula in Fig. 5(d) shows that HNO3 modification increased the surface hydrophilicity of AC compared with the original electrode, where the rate of CDI fluoride removal was more stable with changes in the F concentration, which is conducive to stable operation in practical applications.

3.3. Cycle Stability of HNO3 Modification Electrode

The cycle stability of the electrode is the key factor in the application of CDI for water treatment. Under the conditions of F concentration of 50 mg/L, flow rate of 7 mL/min, and voltage of 1.5 V, F removal experiments were performed with a HNO3-modified electrode. Adsorption was performed for 3,000 s, followed by desorption for 1,800 s. This process was repeated for 5 cycles. The change in the electrode adsorption capacity over the five cycles is shown in Fig. 6. The concentration vs. time curve (Fig. 6(a)) shows that after adsorption saturation and power to the CDI system being turned off, the effluent F concentration increased rapidly as desorption begins, reaching a minimum concentration around 350 s, followed by a gradual increase to the same concentration as the F influent concentration, indicating that electrode desorption was complete. The desorption rate was clearly higher than the adsorption rate. After five adsorption/desorption cycles, the surface structure of the HNO3-modified electrode was intact, and no obvious delamination of the AC occurred. With increasing number of cycles, the adsorption performance of the HNO3-modified AC electrode decreased to some extent. However, the adsorption capacity at the fifth cycle was still 83% of the first cycle, and the single desorption rate always reached at least 94%. In conclusion, the electrode achieved rapid and stable desorption after the power was turned off; therefore, the HNO3-modified AC electrode had good cycling stability.
Cycling tests with the original electrode were performed for comparison. Under the same conditions, the adsorption capacity after 5 cycles was lower, but the cycling stability was higher than that of the HNO3-modified electrode. The adsorption capacity of the fifth cycle was 86% of that of the first cycle. This is similar to the findings of Liu et al. [25]; they observed that the stability of the original electrode was better than that of the modified electrode. This may be because HNO3 treatment increases the number of unstable groups on the AC surface, resulting in a larger decrease in the adsorption capacity of the modified AC electrode after multiple cycles [26]. Furthermore, the reduced pore size of the AC after modification can make ion transport in the channels more difficult. After many cycles, the number of occupied pores increases, making ion transport and desorption of the ions difficult, resulting in a decrease in cycling stability.

3.4. Effect of Operating Conditions on F- Removal by HNO3-Modified Electrodes

The effect of the CDI operating conditions on the F removal performance of the modified electrodes was investigated using a raw-water F concentration of 50 mg/L, voltage of 1.5 V, flow rate of 7 mL/min, and adsorption time of 3,000 s. These conditions were used for all experiments unless stated otherwise.
First the effect of voltage was investigated using voltages of 1.2, 1.5, 1.7, and 2 V. The concentration vs. time curves for the various voltages are shown in Fig. 7(a). As the voltage increased from 1.2 to 1.7 V, the lowest effluent F concentration decreased, the time required for the electrode to reach saturation increased, and the adsorption capacity of the electrode increased from 2.88 to 4.28 mg/g, respectively. A higher voltage results in a higher surface charge density of the electrode, more adsorption sites, and a thicker EDL. Consequently, the attractive force between the ions and the electrode increases, and the ions can be adsorbed deeper into the pores of the AC, which significantly increases the adsorption capacity [27]. However, at an excessively high voltage (e.g., 2 V), the adsorption capacity decreased, perhaps due to modified AC electrodes being prone to stronger Faradaic reactions [28]. In this case, some of the current is not used for adsorption, and instead stimulates chemical reactions, which reduce the charge efficiency, and adsorption capacity, resulting in electrode loss and a drop in the adsorption performance below that at 1.7 V [29]. This is consistent with the conclusion of Sun [30] that the adsorption capacity decreases when the voltage is above 2 V. In addition, the decomposition voltage of water is 1.229 V. Considering the presence of other ions in water, the contact resistance of the power supply, and energy consumption, the voltage for other experiments was set to 1.5 V.
To investigate the effect of the flow rate, the concentration vs. time curves were obtained using flow rates of 5, 7, 10, and 15 mL/min, as shown in Fig. 7(b). Higher flow rates resulted in a shorter the time for the electrode to reach adsorption saturation and the effluent F concentration to reach a minimum. As the flow rate decreased from 15 to 7 mL/min, the adsorption capacity gradually increased from 2.54 to 3.58 mg/g. At lower flow rates, the hydraulic retention time between the electrodes is longer, which ensures that the ions have sufficient time to migrate into the electrode channels. When the flow rate is too high, the structure of the EDL can be disturbed, resulting in a decrease in capacity, and adsorbed ions can be removed with the high flow rates [31]. However, when the flow rate continued to decrease to 5 mL/min, the adsorption capacity of the electrode decreased compared with that at 7 mL/min. When the flow rate is too low, the amount of water treated per unit time is small, and the hydraulic retention time is long. Hence, the adsorbed ions are attracted to ions with an opposite charge in the water, leading to desorption and a reduction in the concentration, and weaker overall adsorption performance [32].
To explore the optimal pH for CDI F removal, HCl and NaOH were used to adjust the raw water to pH 6, 7, 8, or 9. The concentration vs. time curves are shown in Fig. 7(c). The form of fluorine in water is greatly affected by pH. At pH < 2, there is no F in the solution; at pH 2–6, F and HF2 are formed; while at pH > 6, fluorine mainly exists as F. For raw water with neutral pH, the adsorption capacity reached 3.67 mg/g. When the pH was decreased to 6, the adsorption capacity decreased to 2.29 mg/g. It was observed that the pH of the solution decreased continuously during CDI F removal at 1.5 V, probably as a result of Faradaic reactions [33]. Therefore, more HF2 will be produced in acidic raw water, which will result in competitive adsorption of F and Cl introduced by adjusting the pH. Studies have shown that it is more difficult for ions with larger hydration radii to enter the pores, resulting in a delay in adsorption [17]. The hydration radii of Cl and HF2 are smaller than that of F, which will significantly reduce F removal. In addition, under acidic conditions, the electrode is prone to redox reactions, resulting in a lower charge efficiency and adsorption performance [34]. At pH > 7, the adsorption capacity gradually decreases with increasing pH. Under alkaline conditions, OH competes with F for adsorption sites, resulting in a decrease in the adsorption capacity of F. Therefore, neutral conditions are considered the best for F removal in this study. However, as there is little difference between the adsorption performance at pH 7 and that of the original aqueous solution (pH 7.8), it is suggested that pH adjustment should not be performed for practical applications as it adds an extra step and cost.
Finally, the effect of using cation and anion exchange membranes within the corresponding electrodes was investigated. The F removal performance was compared with that without ion exchange membranes, as shown in Fig. 7(d). When the ion-exchange membranes were added to the HNO3-modified electrode, the ion adsorption rate decreased, and the effluent concentration was always higher than that without ion exchange membranes. The addition of the ion-exchange membranes reduced the adsorption capacity to 3.24 mg/g, which is only 90% of that without the membranes. This is different from the conclusion of Hassanvand et al. [35]. They observed that use of ion-exchange membranes can effectively weaken the co-ion effect and counter-ion attraction phenomena, which reduces electrode Faradaic reactions and improves the adsorption performance [20]. The main reason for this is that our study focuses on the removal of F from drinking water, where the TDS of the solution is extremely low, resulting in a low electrical conductivity, while the resistance of the ion exchange membrane itself is high. Hence, the addition of the membrane significantly reduces the electrical conductivity of the solution. In addition, studies have shown that low solution concentrations result in high transfer resistance through the ion-exchange membrane [36]. The ions are subjected to weak attraction and encounter high resistance when passing through the membrane; therefore, the adsorbed ions stay in the EDL on the surface of the channel, and the deeper parts of the channel are not utilized, resulting in a weak adsorption effect.

3.5. Optimized Operating Conditions

Figure 4 shows that the charge efficiency of the HNO3-modified electrode began decreasing gradually after reaching the peak value. Based on this observation, to reduce the energy consumption per unit of F removal, the operating conditions of the CDI system were optimized. The region of high charge efficiency during adsorption was selected, giving an adsorption time of 1500 s, while the desorption time was reduced by half (900 s). Five adsorption/desorption cycles were performed and the results were compared with the results in Fig. 6 that the electrode adsorption saturation and complete desorption. In addition, the cycling stability was explored considering the fraction of retained adsorption/desorption capacity during cycling. The results are shown in Fig. 8.
Consistent with the previous stability results, the adsorption performance of the electrode under optimal conditions gradually decreased with increasing number of cycles. However, the desorption rate of a single cycle under optimal conditions was around 70% (Fig. 8(a)), which is significantly lower than that shown in Fig. 6(a). This may be because in the initial stage of adsorption, ions can be adsorbed deeper within the pores, while in the desorption process, ions in the shallow part of the pore are desorbed first, and those deep within the pores are difficult to desorb. In this experiment, desorption was performed before the electrode was saturated with F ions. Due to the short desorption time, only the shallow parts of the pores were fully occupied by ions, resulting in a low desorption rate. The desorption rate decreased significantly with increasing number of cycles. This may be because the poor desorption performance resulted in stronger adsorption of ions deep in the pores after multiple cycles and greater desorption resistance. However, it was also found that the adsorption performance did not decrease significantly under the conditions of a low desorption rate; when the electrode is not saturated for desorption and regeneration, the adsorption sites in the shallow part of pores are not utilized. When the ions are adsorbed again, they can easily enter the deep parts of the pore. Hence, the utilization rate of the deep adsorption sites increases, and the adsorption rate does not significantly decrease. After the fifth cycle, the adsorption capacity reached 85% of that of the first cycle, and the stability was slightly improved compared with the condition of complete desorption after saturation. This may be due to the fact that the electrode always operates under high charge efficiency and does not reach adsorption saturation, which slows the degradation of the electrode. The adsorption/desorption time was 68% of the saturated adsorption/desorption time, but the cumulative adsorption capacity after 5 cycles reached 80% of the saturated adsorption capacity. This indicates that the energy consumption per unit F removal can be reduced by making full use of the charge efficiency and high-efficiency region of the CDI process, which can slow the degradation of the electrode to some extent. When using this operation mode for F removal, after a certain service cycle, a longer desorption time can be configured to promote desorption of deeply adsorbed ions within the electrode. This will provide a longer service life of the electrode and save the cost of replacing it.

4. Conclusions

The effect of HNO3 modification of AC electrodes on the performance of CDI for fluoride removal from drinking water was investigated. HNO3 modification increased the total specific surface area by reducing the pore size, and added oxygen-containing functional groups that enhanced the surface hydrophilicity of the electrode and promoted ion diffusion. Modification resulted in a 13% increase in the adsorption capacity compared to that of the original electrode under optimal operating conditions, while the charge efficiency was increased by 25%. The cycling stability of the modified electrodes was considered suitable for practical application of single-pass CDI. Kinetic fitting analyses confirmed that the adsorption rate of the modified electrode was less sensitive to changes in the ion concentration of the influent than the original electrode. For practical applications, these electrode improvements correspond to reductions in the operational time and energy requirements of the CDI unit. According to the concentration of fluoride in water and the limit of fluoride concentration to be achieved, the method of setting multiple groups of CDI devices in series or parallel could be used to improve the removal efficiency of fluoride. Hence, the proposed single-pass CDI system with HNO3-modified AC electrodes has considerable potential for the effective removal of fluoride from drinking water.

Supplementary Information


This work was financially supported by the Natural Science Foundation of Shanxi Province (Grant No. 201801D121275) and Social Development Science and Technology Project of Shanxi Province (Grant No. 201803D31050, No.201803D31046). We would like to thank Editage (www.editage.cn) for English language editing.



The authors declare that they have no conflict of interest.

Author Contributions

F.Z. (Associated Professor) analyzed the results and wrote the paper. F.Y. (M.S. Student) conducted the experiments and helped Feng Zhang in drawing the graphics.


1. Mohapatra M, Anand S, Mishra BK, Giles DE, Singh P. Review of fluoride removal from drinking water. J Environ Manage. 2009;91:67–77. https://doi.org/10.1016/j.jenvman.2009.08.015
crossref pmid

2. Liu H, Gao Y, Sun L, Li M, Li B, Sun D. Assessment of relationship on excess fluoride intake from drinking water and carotid atherosclerosis development in adults in fluoride endemic areas, China. Int J Hyg Environ Heal. 2014;217:413–420. https://doi.org/10.1016/j.ijheh.2013.08.001

3. Ghorai S, Pant KK. Investigations on the column performance of fluoride adsorption by activated alumina in a fixed-bed. Chem Eng J. 2004;9:165–173. https://doi.org/10.1016/j.cej.2003.07.003

4. Li Y, Jiang Y, Wang TJ, Zhang C, Wang H. Performance of fluoride electrosorption using micropore-dominant activated carbon as an electrode. Sep Purif Technol. 2017;172:415–421. https://doi.org/10.1016/J.SEPPUR.2016.08.043

5. Maliyekkal SM, Sharma AK, Philip L. Manganese-oxide-coated alumina: a promising sorbent for defluoridation of water. Water Res. 2006;40:3497–3506. https://doi.org/10.1016/j.watres.2006.08.007
crossref pmid

6. Kabay N, Arar Ö, Samatya S, Yüksel Ü, Yüksel M. Separation of fluoride from aqueous solution by electrodialysis: Effect of process parameters and other ionic species. J Hazard Mater. 2008;153:107–113. https://doi.org/10.1016/j.jhazmat.2007.08.024
crossref pmid

7. Li Y, Zhang C, Jiang Y, Wang TJ. Electrically enhanced adsorption and green regeneration for fluoride removal using Ti (OH) 4-loaded activated carbon electrodes. Chemosphere. 2018;200:554–560. https://doi.org/10.1016/j.chemosphere.2018.02.112
crossref pmid

8. Bai Z, Hu C, Liu H, Qu J. Selective adsorption of fluoride from drinking water using NiAl-layered metal oxide film electrode. J Colloid Interface Sci. 2019;539:146–151. https://doi.org/10.1016/j.jcis.2018.12.062
crossref pmid

9. Asquith BM, Meier-Haack J, Ladewig BP. Poly (arylene ether sulfone) copolymers as binders for capacitive deionization activated carbon electrodes. Chem Eng Res Des. 2015;104:81–91. https://doi.org/10.1016/j.cherd.2015.07.020

10. Ji C. Documentary on the innovation of industry-university-research cooperation in the Environmental Technology Transfer Center of the University of New South Wales (Yixing). Sci Technol Ind China. 2021;2:60–61. https://doi.org/10.16277/j.cnki.cn11-2502/n.2021.02.023

11. Ahmed MA, Tewari S. Capacitive deionization: Processes, materials and state of the technology. J Electro Chem. 2018;813:178–192. https://doi.org/10.1016/j.jelechem.2018.02.024

12. Cheng Y, Hao Z, Hao C, et al. A review of modification of carbon electrode material in capacitive deionization. RSC Adv. 2019;9:24401–24419. https://doi.org/10.1039/C9RA04426D

13. Huang Y, Ma Y, Cheng Y, Wang L, Li X. Biosynthesis of ruthenium nanoparticles supported on nitric acid modified activated carbon for liquid-phase hydrogenation of 2, 2, 4, 4-tetramethylcyclobutane-1, 3-dione. Catal Commun. 2015;72:20–23. https://doi.org/10.1016/j.catcom.2015.09.003

14. Huang W, Zhang Y, Bao S, Cruz R, Song S. Desalination by capacitive deionization process using nitric acid-modified activated carbon as the electrodes. Desalination. 2014;340:67–72. https://doi.org/10.1016/j.desal.2014.02.012

15. Bao S, Duan J, Zhang Y. Characteristics of Nitric Acid-Modified Carbon Nanotubes and Desalination Performance in Capacitive Deionization. Chem Eng Technol. 2018;41:1793–1799. https://doi.org/10.1002/ceat.201700448

16. Chen S, Zeng H. Improvement of the reduction capacity of activated carbon fiber. Carbon. 2003;41:1265–1271. https://doi.org/10.1016/S0008-6223(03)00077-0

17. Huang W. Research on Desalination Characteristics of Capacitor Deionization Electrode Materials and Treatment of Vanadium Extraction High-salt Wastewater dissertation]. Wuhan: Wuhan University of Technology; 2014. https://doi.org/10.7666/d.D617936

18. Wang Y, Li H, Yang P. Effects of different ratios on the electrosorption performance of HNO3/TiO2 modified activated carbon electrodes. J Xi’an Institute Technol. 2019;156:30–35. https://doi.org/10.13338/j.issn.1674-649x.2019.02.005

19. Chen Y, Zi F, Hu X, et al. The use of new modified activated carbon in thiosulfate solution: A green gold recovery technology. Sep Purif Technol. 2020;230:115834. https://doi.org/10.1016/j.seppur.2019.115834

20. Porada S, Zhao R, Van Der Wal A, Presser V, Biesheuvel PM. Review on the science and technology of water desalination by capacitive deionization. Prog Mater Sci. 2013;58:1388–442. https://doi.org/10.1016/J.PMATSCI.2013.03.005

21. Zhao R, Biesheuvel PM, Miedema H, Bruning H, Van der Wal A. Charge efficiency: a functional tool to probe the double-layer structure inside of porous electrodes and application in the modeling of capacitive deionization. J Phys Chem Lett. 2010;1:205–210. https://doi.org/10.1021/jz900154h

22. Chen Z, Song C, Sun X, Guo H, Zhu G. Kinetic and isotherm studies on the electrosorption of NaCl from aqueous solutions by activated carbon electrodes. Desalination. 2011;267:239–243. https://doi.org/10.1016/j.desal.2010.09.033

23. Tang W, Kovalsky P, Cao B, He D, Waite TD. Fluoride removal from brackish groundwaters by constant current capacitive deionization (CDI). Environ Sci Technol. 2016;50:10570–10579. https://doi.org/10.1021/acs.est.6b03307
crossref pmid

24. Dykstra JE, Zhao R, Biesheuvel PM, Van der Wal A. Resistance identification and rational process design in Capacitive Deionization. Water Res. 2016;88:358–370. https://doi.org/10.1016/j.watres.2015.10.006
crossref pmid

25. Liu B, Yu L, Yu F, Ma J. In-situ formation of uniform V2O5 nanocuboid from V2C MXene as electrodes for capacitive deionization with higher structural stability and ion diffusion ability. Desalination. 2021;500:114897. https://doi.org/10.1016/j.desal.2020.114897

26. Liang X, Wang X. Activated carbon modification method and its application in water treatment. Water Treat Technol. 2011;37:1–6. https://doi.org/10.16796/j.cnki.1000-3770.2011.08.001

27. Xu P, Drewes JE, Heil D, Wang G. Treatment of brackish produced water using carbon aerogel-based capacitive deionization technology. Water Res. 2008;42:2605–2617. https://doi.org/10.1016/j.watres.2008.01.011
crossref pmid

28. Zhang C, He D, Ma J, Tang W, Waite TD. Faradaic reactions in capacitive deionization (CDI)-problems and possibilities: A review. Water Res. 2018;128:314–330. https://doi.org/10.1016/j.watres.2017.10.024
crossref pmid

29. Li Y, Ding Z, Wang K, et al. Suppressing the oxygen-related parasitic reactions in NaTi2(PO4)3-based hybrid capacitive deionization with cation exchange membrane. J Colloid Inter Sci. 2021;591:139–147. https://doi.org/10.1016/j.jcis.2021.02.013

30. Sun W, Wang Y. Electrosorption of fluoride with flat carbon aerogel electrode. Environ Eng J. 2014;8:2733–2740.

31. Gaikwad MS, Balomajumder C. Removal of Cr (VI) and fluoride by membrane capacitive deionization with nanoporous and microporous Limonia acidissima (wood apple) shell activated carbon electrode. Sep Purif Technol. 2018;195:305–313. https://doi.org/10.1016/j.seppur.2017.12.006

32. Suss ME, Porada S, Sun X, et al. Water desalination via capacitive deionization: what is it and what can we expect from it? Energ Environ Sci. 2015;8:2296–2319. https://doi.org/10.1039/C5EE00519A

33. Zhang C, He D, Ma J, Tang W, Waite TD. Comparison of faradaic reactions in flow-through and flow-by capacitive deionization (CDI) systems. Electrochim Acta. 2019;299:727–735. https://doi.org/10.1016/j.electacta.2019.01.058

34. Ma D, Cai Y, Wang Y, Xu S, Wang J, Khan MU. Grafting the charged functional groups on carbon nanotubes for improving the efficiency and stability of capacitive deionization process. ACS Appl Mater Inter. 2019;11:17617–17628. https://doi.org/10.1021/acsami.8b20588

35. Hassanvand A, Chen GQ, Webley PA, Kentish SE. A comparison of multicomponent electrosorption in capacitive deionization and membrane capacitive deionization. Water Res. 2018;131:100–109. https://doi.org/10.1016/j.watres.2017.12.015
crossref pmid

36. Długołęcki P, Anet B, Metz SJ, Nijmeijer K, Wessling M. Transport limitations in ion exchange membranes at low salt concentrations. J Membr Sci. 2010;346:163–171. https://doi.org/10.1016/j.memsci.2009.09.033

Fig. 1
CDI device and experimental process.
Fig. 2
(a) N2 adsorption/desorption isotherm and (b) pore-size distribution of activated carbon before and after HNO3 modification.
Fig. 3
(a) Infrared spectra and (b) CV curves of the activated carbon electrodes before and after HNO3 modification.
Fig. 4
(a) Effluent concentration as a function of time; (b) current and charge efficiency as a function of time of the electrodes before and after HNO3 modification.
Fig. 5
(a) Changes in adsorption capacity of HNO3 modified electrode with different raw water concentrations. (b) Kinetic fitting model of the original AC electrode. (c) Kinetic fitting model of the HNO3-modified electrode. (d) Relationship between the kinetic rate constant and concentration.
Fig. 6
(a) Adsorption/desorption cycle curve and (b) specific adsorption/desorption capacity of modified activated carbon electrodes. (Adsorption for 3,000 s, desorption for 1,800 s)
Fig. 7
Effect of the operating conditions on the performance of the modified electrode. Effluent F concentration change as a function of time for various (a) voltages, (b) flow rates, (c) pH, and (d) use of ion exchange membranes. Insets: corresponding adsorption capacities.
Fig. 8
Cycling at the optimal operating conditions. (Adsorption for 1,500 s, desorption for 900 s) (a) Effluent F concentration as a function of time. (b) Adsorption/desorption capacity over five cycles.
Editorial Office
464 Cheongpa-ro, #726, Jung-gu, Seoul 04510, Republic of Korea
FAX : +82-2-383-9654   E-mail : eer@kosenv.or.kr

Copyright© Korean Society of Environmental Engineers.        Developed in M2PI
About |  Browse Articles |  Current Issue |  For Authors and Reviewers