Experimental and theoretical study of a new CDI device for the treatment of desulfurization wastewater

According to the characteristics of desulfurization wastewater, A new capacitive deionization (CDI) device was designed to study the desalination characteristics of desulfurization wastewater in this paper. The experiments investigated the desalination efficiency under different conditions which find that the best desalination efficiency is achieved at a voltage of 1.2V, pH=11 and 50°C. Besides, ion adsorption is more favorable under acidic and alkaline conditions. The anion and cation removal performance experiments showed that the order of cation removal is Mg2+>Na+>Ca2+>K+ and the order of anion removal is Cl->CO32->NO3->SO42->HCO3-. The mechanism of CDI was studied and analyzed by the isothermal adsorption model and COMSOL simulation software. It was found that the Freundlich model and Redlich-Peterson model have a good fit with the experimental results. The experiments show that the CDI device has excellent stability. CDI device was used to treat actual desulfurization wastewater. Furthermore, the study provides theoretical support for the industrial application of CDI for desulfurization wastewater treatment in the future. Graphical abstract Graphical abstract


Introduction
At present, wet limestone-gypsum flue gas desulfurization (FGD) is widely used in most coal-fired power plants and is a mature technology with high efficiency of desulfurization (Shuangchen et al. 2016). The proper treatment of FGD wastewater has been the key to achieve zero emissions from power plants (Zheng et al. 2019). Traditional treatment methods for desulfurization wastewater include chemical precipitation (Fu and Wang 2011), biological treatment technology (Huang et al. 2017), electrodialysis (Cui et al. 2017), and thermal (Luo et al. 2019;Conidi et al. 2018;Jia and Wang 2018;Lee et al. 2018) and membrane (Iaquaniello et al. 2014;Gingerich et al. 2018;Hassan and Darwish 2014) methods. However, these methods have many problems, such as high cost, high pre-treatment requirements, easy to scale, and clogging. The quality of the prepared industrial salt is difficult to guarantee and affect its application. Therefore, there is an urgent need to find a new water treatment technology to meet the needs of deep treatment of water pollution in power plants.
Capacitive deionization (CDI) is a newly emerged wastewater treatment technology in recent years (Gao et al. 2015;Anderson et al. 2010), in which ions can be captured from the water phase by simple electrostatic interactions between the ions and the electrode surface forming an electric double layer (EDL) (Hong et al. 2020). Currently, CDI has been studied in many wastewater treatments, for example, seawater , brackish water (Chong et al. 2018;Huo et al. 2020), nuclear power wastewater (Zhou et al. 2020), dye wastewater (Liu et al. 2020a), wastewater containing heavy metals (Bharath et al. 2019;Jin and Hu 2020;Thamilselvan et al. 2018;You et al. 2020), etc. But CDI has rarely been studied for the treatment of thermal power plant desulfurization wastewater. CDI treatment of desulfurization wastewater has significant advantages over other processes: unlike reverse osmosis and distillation, CDI does not require high pressure or high temperature so that it can flow continuously at room temperature. CDI operates under a small voltage. Compared to electrodialysis, CDI does not use ion exchange membranes, so there is no membrane fouling problem and the requirement for feed water quality is not high. Compared with membrane concentration and reduction, CDI does not require membrane replacement, has lower operating costs, and is less prone to fouling and clogging. Compared with flue gas evaporation concentration technology, CDI is less prone to fouling and the flue gas concentration tower is more acidic, which can easily cause corrosion to the equipment. In addition, there is not enough space to set up desulfurization wastewater treatment device in the tail section of many power plants, and the small space occupied by CDI can effectively solve such problems. Therefore, the use of CDI as a reduction unit of desulfurization wastewater to achieve zero discharge of terminal wastewater has great advantages and broad application prospects.
The system structure and adsorption materials are important factors affecting the desalination performance of CDI. As an important component of CDI, a large number of adsorbent materials have been widely studied. The current research is mainly focused on porous carbon electrode materials with specific surface area greater than 1000 m 2 /g. Carbon materials such as activated carbon, graphene, carbon spheres, carbon nanotubes, carbon fibers, porous carbon, and carbon-based aerogel are widely used in CDI electrodes because of their low cost, large specific surface area, good stability, and nontoxicity (Zhu et al. 2018). Noteworthily, as research continues, many innovative structures of CDI have emerged, for example, inverted CDI (i-CDI), flow-through CDI, hybrid CDI (HCDI), and flow-electrode (FCDI). The fast proliferation of architectures available for CDI begs the question of which are the most promising. There is no clear answer as each of the architecture has unique advantages, and none should be neglected in future research work. However, the inevitable problems of any CDI structure are that the device is prone to leakage, difficult to clean, and not easy to replace and add binder resulting in lower mass transfer efficiency. The above problems have received less attention in previous studies. Therefore, the new CDI is designed with a "pool-type" structure of outside which is prepared with plexiglass; it can effectively avoid the problem of water leakage. The electrode plates are inserted on the plexiglass instead of the traditional extrusion method. This makes it easy to replace the plates. It is noteworthy that this CDI does not add binder, and the spacing between activated carbon sponge, electrode plate, and grid is reasonably controlled so that they are closely integrated. This study designed a new type of CDI device for the water quality characteristics of desulfurization wastewater which investigated the desalination of CDI device under different conditions (voltage, pH, temperature) in order to determine the best operating conditions. Afterward, an analysis was conducted on the removal rule of the major ions present in the desulfurization wastewater. In addition, the stability of desalination of the device was also verified experimentally which provides basic data for the application of CDI in the treatment of desulfurization wastewater.

Materials
The chemicals and reagents used were purchased from Tianjin Beichen Founder Reagent Factory for analytical purity in the experiment. The raw material of the electrode plate was 316 L stainless steel. The honeycomb activated carbon sponge (from Jiangsu Maohang Carbon Technology Co., Ltd.) was selected as the adsorption material. The pH and conductivity of the solution were tested by Hach HQ40d water quality analyzer. The solution was kept under constant temperature control by a collector thermostatic magnetic stirrer (Qiuzo DF-101S). The peristaltic pump was used by Lead Fluid (BT101L-YZ25). The power supply is an adjustable DC voltage stabilizer (MT-152D). Furthermore, the CDI device used in this experiment is self-made and the external material is plexiglass.

The design of CDI device
The CDI device designed in this experiment is designed for the characteristics of desulfurization wastewater which has large water volume and high salt content. So, it needs high flux and large adsorption capacity of CDI device. The top view and section of the device are shown in Figure 1(a) and (b), respectively. The CDI unit is an open opening above the system which is rectangular solid, similar to a "pond." It is easy to operate and observe the water flow state. The inlet is located at the bottom of the device. The sample liquid is discharged from the top outlet on the other side of the device after it flows through the entire device so that the sample liquid can be in full contact with the adsorbent material. After the water flows into the CDI unit, there is designed a vacant zone in order to the sample can be thoroughly mixed. The plexiglass column has a recess into which the electrode plate can be mounted. This is designed for easy disassembly and replacement. The adsorption material is an activated carbon sponge which is placed close to the electrode plate. The adsorption material is separated by net. In summary, the CDI device has the features of large processing capacity, high concentration of sample liquid, easy disassembly, and no water leakage.

Analysis
Scanning electron microscope (SEM) (S4800 Hitachi) was used to observe the morphology of the adsorbed material. The specific surface area of nitrogen adsorption/desorption at 77.350 K at the AC electrode was determined using a surface analyzer (Auto-sorb-Iqa3200-4, Quantatech Co, USA). The cations ion content of the solution was determined using inductively coupled plasma (ICP, Optima 5300DV, Perkin Elmer) and anions ion content of the solution was determined using ion chromatograph (ICS-500, Thermo Fisher Scientific). The surface groups of the adsorbed material were studied by FT-IR of spectrum100, PerkinElmer, Inc., USA. Electrochemical performance was tested using a CHI760E electrochemical workstation. The specific capacitance of the adsorbed material was estimated by using cyclic voltammetry (CV) which can be determined for voltages ranging from -0.6 to 1.2 V by the following equation : Desalination experiments The cyclic sample injection was used for the test of desalination performance. Besides, the volume of sample liquid treated was 2L, as shown in Figure 2. The desorption was performed by inverted polar operation and the device was backwashed with deionized water. The concentration of the solution was calculated based on the equation between conductivity and concentration (Li et al. 2009), and the electrical adsorption capacity was calculated according to the following equation : Result and discussion

SEM and BET
Scanning electron microscope was performed in order to survey the surface morphology of the activated carbon sponge (Divyapriya et al. 2020). Figure 3(a) and (b) shows the SEM images of the activated carbon sponge without adsorbed desulfurization wastewater at low and high magnification, respectively. The activated carbon sponge has an overall coral shape with large voids and rough surface, which makes it easy , the surface of the carbon material after the procedure is enriched with many particles, which is due to the adsorption of a large number of ions from the desulfurization wastewater. Pore volume and specific surface area of the material were determined by Brunauer-Emmett-Teller (BET). BET-specific surface area (SBET) is given by using Brunauer-Emmett-Teller theory. In addition, the micropore volume, external surface area, and micropore area are determined by the t-Plot method (Rezma et al. 2019), and the pore size distribution is used by the BJH (Barrett-Joiner-Halenda) model. Table 1 shows the test results of the activated carbon sponge. After calculation, the porosity of the adsorbent material is 22.86%. The analysis of the obtained data shows that the pore size of the material is dominated by ordered mesopores and a few micropores.

FT-IR
The FT-IR characterization was used to qualitatively detect the activated carbon sponge, and it can roughly know the surface functional groups in the adsorption material by analyzing the characteristic peaks. The results are shown in Figure 4. The characteristic peak at 2958.63cm -1 is caused by C-H stretching vibration of methyl (-CH 3 ), methylene (-CH 2 ), and methylene (-CH) in activated carbon sponge (Chen et al. 2005); the characteristic peak at 1728.62 cm -1 is related to the stretching vibration of C=O (Koch et al. 1998). Then, the characteristic peak located around the wave number 1444.19 cm -1 is related to the symmetric bending vibration of C-H on methyl or the deformation vibration of C-H connected to the benzene ring (Wang et al. 2014). Besides, the characteristic peak at around 1231.96 cm -1 is attributed to C-O bending vibration (Jaramillo et al. 2010). And the weak characteristic peaks between 700 cm -1 and 900 cm -1 are associated with C-H bending vibrations outside the plane of the aromatic ring. In conclusion, it can be seen that activated carbon sponge contains abundant surface functional groups, which can provide a large number of adsorption sites for the adsorption of pollutants. After the completion of CDI treatment, it can be observed that the wave number of the material of activated carbon sponge moves from 2958.63 cm -1 , 1728.62 cm -1 , 1444.19 cm -1 , 1231.96 cm -1 , and 1016.26 cm -1 to 2957.81 cm -1 , 1728.86 cm -1 , 1437.61 cm -1 , 1236.36 cm -1 , and 1019.74 cm -1 , respectively. It shows that the ions in solution interact with the functional groups on the surface of the material after being adsorbed.

Electrochemical characterization
The desalination properties of CDI are related to the generation of a double layer by the accumulation of ions in the   (Senoussi and Bouhidel 2018). The ion adsorption ability of the carbon electrode can be characterized by the CV test (Cai et al. 2017). The CV graph of the activated carbon electrode is shown in Figure 5. A 1-M sodium chloride solution was used as the electrolyte.
Besides, the scanning rates were 50 mV/s and 100 mV/s, respectively. The shape of the CV curve is almost rectangular which represents the energy storage capacity of the electrode material (Gaikwad and Balomajumder 2017). According to the calculation of Eq. (1), the specific capacitance is 67.06 F/g and 42.53 F/g at the sweep rates of 50 mV/s and 100 mV/s, respectively. In addition, no significant oxidation/ reduction peaks are observed in the CV curve. This phenomenon suggests that the capacitance mainly originates from the electric double layer efficiency formed by Coulomb interactions rather than the Faraday reaction .

Desalination performance under different conditions
There are three experimental conditions that were selected: voltage, pH, and temperature in order to investigate the optimal operating conditions of the CDI device. The experiments were carried out by cyclic injection. Furthermore, NaCl solution as sample liquid, the treatment volume was 2L, and the initial concentration was 500mg/L. The treatment was carried out until the concentration no longer changed significantly. After calculation and analysis, the concentration gradient change and the electrical adsorption capacity at equilibrium are obtained. Figure 6 shows the desalination performance of CDI at different voltages which can be found that higher voltage is better for the desalination of CDI. It is due to the electrostatic attraction will increase as the voltage increases, thereby the     (Yang et al. 2001). The best desalination efficiency was obtained when the voltage was 1.2 V, with an electro-adsorption capacity of 2.59 mg/g. It is worth noting that the desalination efficiency decreased at 1.5 V due to the hydrolysis reaction of water occurs when the applied voltage is higher than 1.23 V (Quan et al. 2017). Therefore, the optimal applied voltage for the CDI device is 1.2 V.
The pH of the solution to be treated has an important influence on the desalination performance of CDI Hassanvand et al. 2018;Hemmatifar et al. 2017). Thus, this section investigates the desalination performance of CDI at different pH. Moreover, pH was adjusted by using HCl solution and NaOH solution. The results are shown in Figure 7. CDI exerts good desalination properties under both acidic and alkaline conditions. This is a result of the surface chemistry of the adsorbent material is altered under acidic or basic conditions, and the surface chemistry affects the interaction between the polar or non-polar adsorbents of the activated carbon sponge. From the previous FT-IR analysis, it was shown that Na + and Clwere adsorbed and bound to functional groups on the electrode material during the adsorption process, indicating that the affinity between Na + , Cl -, and functional groups is a key factor affecting ion desorption. As the pH increases, the carboxyl concentration decreased significantly due to the deprotonation effect. Thus, the affinity determined by the hydrogen bond strength significantly diminishes due to the decrease in the carboxyl concentration on the electrode (Wang et al. 2018a). In particular, the maximum electrical adsorption capacity (3.67 mg/g) was obtained at pH = 11. The pH of the desulfurization wastewater treated by the triplex tank was alkaline. Thus, it indicates that desulfurization wastewater can be treated better by using this CDI unit.
Temperature is an important factor that cannot be neglected in the process of industrial wastewater treatment. Noteworthily, the temperature of desulfurization wastewater is generally about 45°C. For this reason, the CDI device was used to treat simulated sample liquid under isothermal conditions of 30°C~50°C in this experiment. The results are shown in Figure 8. It can be found that the desalination capacity gradually increases with the increase of temperature, and the best desalination efficiency is obtained at 50°C. The desalination capacity of the adsorbent increases gradually with the increase of temperature. Temperature is an important factor affecting the effect of ion adsorption, which has a significant effect on the removal rate of ions and the morphological distribution on the adsorption medium. From the experimental results, it is easy to see that the electric adsorption process is endothermic, and the increase in temperature facilitates the reaction. This may be due to the fact that as the temperature increases, the internal diffusion transport rate of Na + and Clto the adsorbent surface is accelerated. Thus, the increase of the active site on the adsorbent surface leads to an increase in the   Adsorption capacity curves for different cations treated with CDI electrical adsorption capacity due to the increase in temperature. In addition, it also provides higher kinetic energy to increase in the ion migration rate with the increase of temperature (Huang and Tang 2020).

Removal law of anions and cations
According to the main pollutants present in the desulfurized wastewater, K + , Na + , Ca 2+ , and Mg 2+ were selected as cations representatives. Besides, Cl -, NO 3-, SO4 2-, HCO 3-, and CO 3 2are anion representatives. In this section, the electrical adsorption properties of CDI for cations and anions were investigated at 1.2V, 100mL/min, and the initial ion concentration of 0.01mol/L. The results obtained are shown in Figures 9 and  10, respectively. From Figure 9, it can be seen that the electrical adsorption capacity of each cation is as follows (from the largest to the smallest): Mg 2+ (2.86mg/g) > Na + (2.36mg/g) > Ca 2+ (2.34mg/g) > K + (2.14mg/g) when the adsorption equilibrium is reached. The hydration radius is shown in Table 2.
Noticeably, the maximum amount of electrical adsorption should be K + , but this is contrary to experimental results which is consistent with the previously reported findings (Mossad and Zou 2012;Xu et al. 2008). This phenomenon can be roughly explained from the electronegativity point of view, where (from largest to smallest): Mg 2+ (1.31) > Ca 2+ (1.00) > Na + (0.93) > K + (0.82). The electrical adsorption capacity of Na + was slightly higher than Ca 2+ , but the electronegativity was lower than that of Ca 2+ , probably due to the fact that the ionic radius of Na + was smaller than that of Ca 2+ and the smaller ionic radius was easier to enter the pore structure. What is more, the electrical adsorption capacity of each anion is from largest to smallest: Cl -(2.36mg/g) > CO 3 2-(1.74mg/g) > NO 3 -(1.50mg/g) > SO 4 2-(1.47mg/g) > HCO 3 -(0.98mg/g). From the experimental results, it can be seen that the best adsorption effect was obtained for Cl -, which is probably due to the fact that Clhas the smallest ionic radius. For anions with the same charge, the adsorption efficiency is mainly affected by the difference of the radius. In terms of electronegativity, Cl is 3.16, which is higher than N (3.04),   It is obvious that Clhas a higher ability to attract electrons than other ions. Moreover, the increase of the active site on the adsorbent surface leads to an increase in the electrical adsorption capacity due to the increase in temperature. In combination with the hydration radius in Table 2, it was found that most of the ions conformed to this law. In particular, the higher adsorption of CO 3 2may be due to the ease of binding to hydroxyl groups in solution to generate CO 2 and H 2 O (Bai et al. 2019).

Adsorption kinetics
Using the pseudo-first-order model and pseudo-second-order model, the electrical adsorption process of CDI can be described. The pseudo-first-order model and pseudo-secondorder model are calculated according to Eq. (4) and Eq. (5) (El-Khaiary and Malash 2011), respectively.
Validation of the kinetic model using the average relative error equation (ARE) (Kumar et al. 2011): The data obtained from the desalination of samples at different concentrations (100mg/L, 200mg/L, 300mg/L, 400mg/ L, and 500mg/L) were brought into the pseudo-first-order model and pseudo-second-order model, and the results were shown in Figure 11. The parameters of the pseudo-first-order model and pseudo-second-order model are shown in Table 3.Compared to the pseudo-second-order model, the pseudo-first-order model exhibits a lower ARE and a higher R 2 value. It was shown that the experimental results of the CDI device were more consistent with the pseudo-first-order model. It indicates that the electrical adsorption process is electrostatic interactions between ions on the electrode and ions in solution, and similar conclusions were reached by prior study (Gaikwad and Balomajumder 2017).

Isothermal adsorption model
Isothermal adsorption model studies focus on optimizing adsorption and understanding the driving forces that control the interaction between adsorbent and adsorbent. The Langmiur model and Freundlich model being the most commonly used, Langmiur model is based on the ideal assumption that the adsorbed surface is perfectly homogeneous. Whereas Freundlich model is being used for highly heterogeneous surfaces. The Peterson equation model is a combination of the Langmiur model and Freundlich model (OZKAYA 2006). Langmiur model: Freundlich model: Peterson equation model: MPSD is the parameter used to value the adsorption equilibrium and can be used to evaluate the isothermal model, which can be obtained by the following equation: The Langmuir, Freundlich, and Redlich Peterson isothermal models are used to fit the obtained experimental results. Figure 12 shows the fitting results of the three isothermal adsorption models, and the obtained parameter estimates are shown in Table 4. It can be seen that the Freundlich and Redlich-Peterson models fit better with the experimental results by comparing the R 2 and MPSD. Because they have higher R 2 and lower MPSD values. This may be owing to the fact that the adsorption of ions by activated carbon sponge is a multimolecular layer adsorption. Indeed, the adsorption material is a mesoporous dominated carbon material which contributes to the pore adsorption.

COMSOL analysis of rectangular CDI unit
Computer simulation has proven to be one of the most technically advantageous methods for solving a variety of engineering problems. CFD provides a great deal of information about the flow of fluids as well as the simulation details of the geometry. It also allows for easy parameter changes which make them very powerful design systems (Sousa et al. 2014). In this paper, COMSOL Multiphysics 5.5 software is used to analyze the fluid flow behavior during the desalination process of CDI. Set the CDI as a rectangular structure with a size of 50cm×10cm×2cm, the inlet and outlet are located at the center of the narrow side, and the aperture is 1cm. Figure 13 shows the velocity profiles of the CDI rectangular flow channel at different flow rates. It can be seen that the velocities on the outside are higher than the velocities on the inside at high flow rates unlike some previous reports (Gaikwad et al. 2020;Jeon et al. 2017). That means that the distribution of the solution is not uniform so that the desalination efficiency of the outside of the CDI is higher than that of the inside. This may cause the outside to reach adsorption saturation and the inside is still unsaturated. But a large amount of solution passes through the outside for which there is a decrease in the overall desalination efficiency. Furthermore, the flow field is more evenly distributed in the case of low flow rate which enables efficient desalination of each part. It also reveals that large dead zones appear at the corners and eddy current is produced at high flow rates from the flow distribution graph (Figure 14).

Stability of CDI desalination performance
The long-term stability of the system is an important indicator to test the performance of the CDI device Moustafa et al. 2020;Wang et al. 2018b). In this experiment, a NaCl solution with an initial concentration of 500 mg/L was selected and continuously adsorbed and desorbed 10 times under a voltage of 1.2V. The adsorption and desorption were performed by inverted polarity operation and rinsed with deionized water. The time ratio of adsorption and desorption is 1:1. Figure 15 shows the conductivity change of the CDI device with 10 cycles of desalination. It can be seen that the downward trend of the CDI device during the 10 runs is approximately the same with the conductivity dropping from about 1000 μs/cm to about 850 μs/cm. Therefore, the CDI device has a good stability of desalination performance.

Desalination efficiency of actual desulfurization wastewater
Desulfurization wastewater generally has high salt, hardness content and is prone to scaling which is due to the high content of Ca 2+ , Mg 2+ , Na + , and suspended particulate matter. And the desulfurization wastewater weakly acidic and corrosive is caused by higher content of Cl -. For this reason, the CDI device was used for treating actual desulfurization wastewater focusing on the removal effect of the main ions contained. The water sample is selected from the desulfurization wastewater of a power plant after the triple tank treatment. The sample was diluted five times and then used as the solution to be treated. 2 L of the solution to be treated was injected in a cyclic manner, and the real-time desalination effect was observed and recorded. The change of and electrical adsorption capacity are shown in Figure 16. It can be seen that the conductivity is decreased after treatment, and the removal effect on the high concentration of desulfurization wastewater is not very satisfactory which is owing to the small size of the device. Afterward, the main anions and cations present in the treated sample liquid were detected. Table 5 records the values of each index in the desulfurization wastewater (before and after treatment). By comparing the changes of ions before and after treatment, it is found that the removal of Ca 2+ , Na + , and Clis relatively good which is consistent with the conclusions of previous experiments. Rather surprisingly, the water samples selected for this study contained less Mg 2+ which is not a reference value. In addition, there is a corresponding decrease in TDS as well as pH value.

Conclusions
In this paper, the following conclusions are drawn from the study of a new CDI unit for the treatment of desulfurization wastewater.
(1) Activated carbon sponge was selected as the adsorption material and characterized by SEM, BET, FT-IR, and CV tests. The results show that it has good pore structure, large specific surface area (27.4536 m 2 /g), and good specific capacitance (67.06F/g) obtained at a sweep speed of 50mV/s (2) The desalination was investigated under different voltage, temperature, and pH conditions. Consequently, the best desalination efficiency is obtained at 1.2V, 50°C, and pH=11. In addition, this study also found that the desalination and adsorption were more effective under acidity or alkalinity conditions. Furthermore, the removal efficiency of different anions and cations was investigated. The results indicate (3) The adsorption kinetics and isothermal adsorption models were studied and analyzed. It was concluded that the experimental results of the CDI device were more consistent with the pseudo-first-order model. Furthermore, the Freundlich and Redlich-Peterson models are better fitted with the experimental results. The CDI device designed has dead zones at high flow rates in this paper. In other words, low flow rates are more favorable for desalination through COMSOL software simulation (4) According to the requirements of practical applications, the stability of the CDI device was tested in several sets of desalination experiments. Consequently, the experimental results showed that the stability of the device is excellent. In addition, the CDI device was also used to treat actual desulfurization wastewater so that the device exhibited a certain desalination capacity Nomenclatures Q 0 , Constant in Langmuir model (mg/g); b, Constant in Langmuir model (L/mg); C e , Concentration of adsorbate in solution at equilibrium (mg/L); q e , Electrical adsorption capacity at equilibrium (mg/ g); q t , Electrical adsorption capacity at t (mg/g); K F , Constant in Freundlich model (mg/g)/(mg/L) 1/n ; n, Constant in Freundlich model;