Tio₂ and carbon nanotubes composites modify capacitive deionization anodes to improve the dechlorination efficiency in desulfurization wastewater

Desulfurization wastewater mainly contains high levels of supersaturated nitrite, sulfite, suspended solids, heavy metal ions, F⁻ and Cl⁻, etc. (Shuangchen et al. 2016). It is worth noting that the increase in Cl⁻ concentration in desulfurization wastewater will bring many problems (Zheng et al. 2019). On the one hand, it will lead to a decline in the quality of gypsum. On the other hand, the increase in Cl⁻ content will increase the acidity of wastewater and affect desulfurization efficiency. In addition, the corrosion of system materials will be intensified with an increased concentration of Cl⁻. Therefore, the main goal of current wastewater treatment technology is to improve dechlorination efficiency.

The traditional treatment method (Gingerich et al. 2018b; Ma et al. 2019) mainly adopts the chemical dosing method to remove a large number of calcium ions, magnesium ions and some heavy metal ions from wastewater, such as the common triplet-tank treatment technology, double alkali method, etc. There are two main methods of wastewater concentration and reduction: membrane methods and thermal methods. For example, the widely used reverse osmosis (RO) (Conidi et al. 2018), forward osmosis (FO) (Lee et al. 2018b), nanofiltration (NF) (Jia & Wang 2018), electrodialysis (ED) (Luo et al. 2019) and so on have been applied. Thermal concentration technology mainly uses steam for evaporation and crystallization of wastewater, including mechanical vapor recompression (MVR) (Gingerich et al. 2018a), thermal vapor recompression (TVR) (Hassan & Darwish 2014), multi-effect evaporation (MED) (Iaquaniello et al. 2014) and evaporation are in use with waste heat of power plant flue gas. Although much research has been carried out on the treatment of desulfurized wastewater, traditional water treatment technology still faces the problems of large modification of the original system and high investment cost. Based on comprehensive analysis, there is an urgent need to find a new technology for removing Cl⁻ from desulfurization wastewater as key to solving the problem.

As a new wastewater treatment technology in recent years, capacitance deionization technology (CDI) (Porada et al. 2013; Suss et al. 2015; Jia & Zhang 2016) has attracted wide attention from scholars due to its advantages of low energy consumption, convenient operation, simple maintenance and good removal efficiency. CDI technology has been studied in seawater (Jande & Kim 2014; García-Quismondo et al. 2016), brackish water (Tang et al. 2016; Xing et al. 2019), circulating cooling wastewater (Ma et al. 2018), printing and dyeing wastewater (Senoussi & Bouhidel 2018) and other fields. Unlike RO, CDI does not require high pressure or temperature, and it can be operated at room temperature with low voltage. Compared with ED, CDI does not use ion exchange membranes, so it does not cause membrane clogging or require high feed water quality. Indeed, the flue gas concentration tower is more acidic and it very easy to cause corrosion in the equipment. In contrast, CDI is a good way to avoid such occurrences because it does not require dosage and generate secondary contamination. In addition, there is not enough space to build a desulfurization wastewater treatment device at the end of the power plant, and the small footprint of CDI is an effective solution to this problem. Therefore, the use of CDI as a reduction unit for desulfurization wastewater has great advantages and broad application prospects.

The CDI principle is to form an electrical double-layer (EDL) (Jia & Zhang 2016) on the surface of the electrode through the action of electric field force to collect charged ions in water by using the electrochemical characteristics of the surface of the charged electrode. Among them, adsorption materials have always been the research focus of this technology as an important factor of CDI. Carbon materials (Huang et al. 2017) are commonly used for adsorption materials. The general carbon materials include activated carbon (Kim et al. 2016; Hu et al. 2018), graphene (Shi et al. 2016), activated carbon fiber (Tian et al. 2019), carbon aerogel (Quan et al. 2017; Liu et al. 2019) and carbon nanotubes (Lee et al. 2018a). Carbon nanotubes (CNTs) have been widely studied and discussed, because of their large specific surface area and good electrical conductivity. Therefore, how to improve the electrical adsorption efficiency of CNTs is key to the research. Surface treatment or loading of other materials can effectively improve adsorption efficiency. Metal oxides on carbon-based substrates can enhance conductivity, improve specific capacitance (Zhang et al. 2009) and increase charge storage (Boukhalfa et al. 2012). Furthermore, the addition of metal oxides can significantly change the physical and chemical properties of some carbon materials, such as wettability, surface area and zeta potential, which is helpful to improve the deionization performance of capacitors. TiO₂ is a kind of n-type oxide, environment-friendly semiconductor material with high stability, high-quality hydrophilicity, and low cost. The wettability and electrochemical performance of carbon materials can be effectively increased by TiO₂-loaded carbon materials to improve electric adsorption efficiency (Cheng et al. 2019). Therefore, the combination of TiO₂ semiconductor materials and CNTs porous materials as composite adsorption materials can effectively improve the adsorption efficiency of CDI.

Solvothermal methods have been commonly used to coat TiO₂ onto the surface of CNTs (Liu et al. 2013). Muduli et al. (2009) synthesized TiO₂-multiwalled carbon nanotube (MWCNT) nanocomposites using a hydrothermal method, the results showed the TiO₂-MWCNT conjugation through -COOH groups, and that it had efficient charge transfer in the nanocomposites. Yi et al. (2016) reported that TiO₂@CNTs materials had been successfully achieved by a polymer-assisted approach to an aqueous chemical solution method. The polymer, polyethylenimine, aims to combine the Ti⁴⁺ with CNTs for film formation of TiO₂@CNTs. This approach can prevent ion hydrolysis in solution and form a conformal coating for complex structure. Noteworthily, N-methyl pyrrolidone (NMP) is a good polymer solvent (Kim et al. 2006; Sotto et al. 2011). Inspired by these, to explore a simple, low-cost, and energy-efficient synthetic strategy, a novel preparation method of TiO₂/CNTs was developed in this study. This research has prepared Ti⁴⁺ combining NMP along with CNTs and forming a stable and homogeneous precursor solution to fabricate TiO₂/CNTs.

In this study, TiO₂/CNTs were synthesized using a solvothermal method based on Ti⁴⁺ combined with NMP along with CNTs as precursor solution. Materials were used for the first time as adsorption materials for CDI anode to remove Cl⁻ from desulfurization wastewater. Physical characteristics of composites were discussed, and cyclic voltammetry (CV) was used to analyze the electrochemical performance of the composites. Finally, TiO₂/CNTs composites was modified on the surface of the CDI anode to experiment with Cl⁻ removal to find the best composites. The electrosorption method was analyzed and discussed by the adsorption kinetic model and the adsorption mechanism of TiO₂/CNTs was also explored. This study will provide a new direction for the application of TiO₂/CNTs composites as adsorption materials of CDI in the actual desulfurization wastewater.

The process of material preparation is shown in Figure 1. Here, 0.5 g multi-wall carbon nanotubes (MWCNTs, xfm01, Aladdin, Shanghai) were dissolved in 20 mL NMP and used as the polymer solvent (Ma et al. 2019), 40 mL absolute ethanol was added, and then the mixture was fully stirred on a magnetic stirrer (Qiuzo DF-101S, Zhengzhou). In the process of stirring, 2 mL tetrabutyl titanate were added dropwise and used as a precursor of TiO₂ nanoparticles. After stirring for 30 min, the mixed solution was transferred to the polytetrafluoroethylene reactor, and removed after leaving it in the oven at 200 °C for 12 h. After the reaction, the reactor was naturally cooled to room temperature, the precipitates were collected, centrifuged, and washed with water and ethanol three times, dried in a blast drying oven at 60 °C for 12 h, and the resulting solid powder was placed in a muffle furnace and sintered at 500 °C. The final product was the TiO₂/CNTs composite. The composites prepared at this ratio were recorded as 60%T/C (the content of TiO₂ accounts for 60% of the total sample). According to this method, composites with different proportions were prepared, and recorded as 15%T/C and 30%T/C. The chemicals and reagents used were purchased from the Tianjin Beichen Founder Reagent Factory for analytical purity in the experiment.

The anode electrode plates of CDI were prepared by weighing 0.2 g of composites at different ratios, mixing with acetylene black and polyvinylidene fluoride at a ratio of 8:1:1 and dissolving them in an appropriate amount of NMP. The mixed solution was sonicated for 15 minutes with an ultrasonic cleaner (020S, Shenzhen Genen Cleaning Equipment Co. LTD), followed by centrifugation at 5,000 rpm for 10 min and magnetic stirring. The stirred dispersion was added evenly to the 10 cm × 10 cm titanium plate using a rubber-tipped dropper, followed by smoothing the raised portion using a glass rod and drying for 12 hours at room temperature. All experiments were carried out at 25 °C. Similarly, multiwalled CNTs modified CDI cathode electrode plates were prepared.

X-ray powder diffraction (XRD) analysis was performed using an X-ray diffractometer (D8advance, Buker, Germany). CuKα (λ = 0.6491 nm) was used, for which the tube voltage was 40 kV, the tube current was 40 mA, the scan step width was 0.026°, the counting time was 10 s, and the scan angle was 2θ from 5 to 80°. A scanning electron microscope (SEM) (S-4800, Hitachi, Japan) was used to observe the morphology of the adsorbed material. The sample was weighed and dispersed in anhydrous ethanol, and the homogeneous dispersion was prepared by sonication. A small amount of the dispersion was removed and added dropwise on a smooth single crystal wafer substrate and dried naturally. The composites were observed using a transmission electron microscope (TEM) (Jem-2100F, Jeol, Japan). An X-ray photoelectron spectroscopy (XPS) model (250xi, Therom Escalab, USA) with Al target, 100 μm beam spot and 25 W power parameters was used to further analyze the elemental content of the composites surface after the adsorption experiment. The surface groups of the TiO₂/CNTs composites were studied by Fourier transform infrared spectroscopy (FT-IR) using an instrument model (Spectrum 100, Perkin Elmer, Inc., USA). The wettability in terms of static water contact angle (WCA) of materials was tested using an OCA15EC (Dataphysics, Germany) instrument. The specific surface area, pore size distribution, and pore size were measured using the Brunauer–Emmett–Teller (BET) method on the basis of N₂ adsorption–desorption isothermal at 77.35 K using the Auto-sorb-iQA3200-4 (Quantatech Co., USA) instrument.

The capacitance deionization experimental device and system flow chart used in the experiment of CDI device are shown in Figure 2(a) and 2(b) respectively. The device adopted the traditional CDI structure. There were two fixed end plates on the outside, and the negative and positive electrode plates were close to the endplate. A filter was sandwiched in the middle of the two electrode plates to provide a reaction area for the solution to be treated; the outer side was sealed with a rubber pad. The solution entered from the bottom left of the figure and passed through the middle of the two electrode plates from the bottom to the top. The purpose was to allow the solution to be fully adsorbed.

Figure 3(a) shows the XRD spectra of CNTs and TiO₂/CNTs composites. There are obvious diffraction peaks at 26.2°, corresponding to the (002) crystal surface of CNTs (Yi et al. 2016), which indicates that CNTs exist in TiO₂/CNTs composites. In the spectrum of TiO₂/CNTs composites, the diffraction peaks of 25.356°, 37.785°, 48.076°, 53.920°, 55.114°, 62.726°, 70.357° and 75.094° are located at (101), (004), (200), (105), (211), (204), (215) of TiO₂ respectively, corresponding to the characteristic peaks of TiO₂ (JCPDS No. 37-1492) (Wang et al. 2013). This indicated that our samples contained both anatase TiO₂ phase and CNTs as expected. There was no other obvious characteristic peak, which showed that the composites had high purity and good crystal phase.

To further study the surface groups of TiO₂/CNTs composites, Fourier transform infrared spectroscopy (FT-IR) was used to analyze the composites to verify the successful construction of molecular imprinting sites of the composites (Liu et al. 2017). In Figure 3(b), there are vibration bands at 1,632 cm⁻¹ and 1,385 cm⁻¹; the bands at 1,632 cm⁻¹ are related to the C=C bond, while the bands at 1,385 cm⁻¹ are related to the C–C bond. However, at 611 cm⁻¹ the TiO₂/CNTs composites had a large wave band, while CNTs material did not. This is related to the Ti-O structure and O-Ti-O structure in TiO₂ (Nguyen et al. 2016). It can be seen that TiO₂/CNTs composites have the chemical bond characteristics of CNTs and TiO₂ materials.

In order to investigate the pore structure of the TiO₂/CNTs materials, the N₂ adsorption–desorption isotherms are acquired. The tests results are listed in the Table 1 and Figure 3(c). The adsorption–desorption isotherm exhibits the typical feature of a type IV isotherm, indicating the presence of many mesopores inside both samples, this result was also indicated by the pore size distribution curve in Figure 3(d).

Figure 4(a) shows SEM images of CNTs. It can be seen that the CNTs are distributed in disorderly spun filaments. Figure 4(b) shows SEM images of TiO₂/CNTs composites. Compared with a single CNTs, the surface of the TiO₂/CNTs composites is a cluster of coral, the surface is rough and densely distributed. It is easy for ions to adhere to the material and provide more adsorption sites, which indicates that the TiO₂/CNTs composites had high efficiency of electrosorption (Zhu et al. 2015). Due to the high content of powdered TiO₂ which wraps the CNTs material, the shape of spinning is not obvious in the figure.

Figure 4(c) is the TEM diagram of TiO₂/CNTs composites which is the full picture of the composites. It can be seen that there are many TiO₂ nanoparticles attached to CNTs. The staggered lattice stripes indicate the formation of heterojunctions between TiO₂ and CNTs. Figure 4(d) is the high-resolution transmission electron microscope (HR-TEM) image which shows that the nanohybrids TiO₂/CNTs are highly crystalized. The uniform lattice fringes of TiO₂ nanoparticles show that they are highly crystallized, and d is 0.3520 nm corresponding to the (101) crystal surface of anatase TiO₂ (Olowoyo et al. 2019), which is consistent with the XRD results. The above results show that the TiO₂/CNTs composites were successfully prepared.

To determine the chemical composition and valence state of elements in the TiO₂/CNTs composites, the elements Ti, O and C in 60%T/C composites were analyzed by XPS. Figure 5(a) shows the full spectrum of the sample. Figure 5(b) shows the high-resolution XPS spectrum of Ti2p. The binding energies at 458.47 eV and 464.1 eV correspond to Ti2p_3/2 and Ti2p_1/2 of Ti⁴⁺ (Olowoyo et al. 2019). Figure 5(c) shows that the characteristic peaks of the O1s spectrum are at 529.6 eV, which is attributed to Ti-O lattice oxygen (Hafeez et al. 2018). Figure 5(d) shows that the characteristic peak of C1s is located at 284.82 eV, which is attributed to the C–C bond and C=C bond in CNTs (Cong et al. 2011). Table 2 is the composition analysis table of TiO₂/CNTs composites. The peak intensities of Ti2p and O1s were 22031.39 cps and 8066.21 cps, respectively, which were much higher than those of C1s (7282.05 cps). By calculating the atomic ratio, Ti2p:O1s:C1s is 4:2:4. XPS results further prove that the formation of heterojunction between TiO₂ and CNTs is consistent with TEM results.

The wettability of the CDI adsorbent material is critical. The WCA of CNTs and TiO₂/CNTs composites was determined, and the findings are shown in Figure 6(a) and 6(b). It can be seen that the WCAs of CNTs and TiO₂/CNTs composites are 124.6° and 85.9°, respectively. The results demonstrate that the addition of TiO₂ significantly enhanced the wettability of the CDI adsorbent material (Zheng et al. 2018). which could have a positive impact on improving adsorption performance of CDI.

Specific capacitance is an important index to test the electrosorption properties of materials (Nguyen et al. 2016). Figure 7(a) shows the cyclic voltammetric test results of different adsorbed materials in the range of 0–0.8 V at a sweep rate of 100 mV/s. It can be seen from the observation diagram that the images are all in rectangular distribution, which shows that there was no redox reaction in the scanning process. But the EDL capacitance plays a leading role, and the adsorption of ions was due to the interaction between the Coulomb forces rather than the electrochemical reaction (Quan et al. 2017). In addition, the larger is the surrounding area, the larger is the capacitance of the material. Therefore, it is easy to see that the capacitance of the adsorption materials is larger than pure CNTs, and the composites with 30%T/C and 60%T/C ratio had a larger capacitance. Figure 7(b) shows the cyclic voltammogram of the 60%T/C material at 1 mV/s, 5 mV/s, 10 mV/s, 50 mV/s and 100 mV/s scanning speed. According to Equation (1), the specific capacitance of each sweep speed was 87.6 F/g, 42.6 F/g, 32.8 F/g, 2.1 F/g and 1.7 F/g. The observed data show that the specific capacitance was inversely proportional to the scanning rate. This is because, at a higher scanning rate, sodium ions and chloride ions do not have enough time to accumulate and enter the internal holes of the electrode, and the ohmic resistance will increase correspondingly, resulting in a lower specific capacitance (Feng et al. 2019).

Voltage is an important index affecting the performance of CDI. Select the adsorption materials with a 60%T/C ratio to carry out dechlorination experiments under the voltages of 0.4 V, 0.8 V, 1.2 V and 1.5 V under flow rate of 50 mL/min. The data are shown in Figure 8(a). It is expected that the higher the voltage, the better the DE. This is because when the voltage increases, the amount of charge carried on the plate increases (i.e. the electric field force increases), and more chloride ions can be adsorbed in unit time. It can be found in the observation chart that the trend is consistent with the voltages of 0.4 V, 0.8 V and 1.2 V. However, compared with conditions of 1.2 V and 1.5 V, the DE of 1.2 V is better than 1.5 V. This may be due to the hydrolysis reaction (Reddy & Reddy 2006) that occurs when the voltage is about 1.23 V, resulting in a large amount of OH⁻ in the solution; the charge used to adsorb ions will lose part of it to participate in the hydrolysis reaction. Furthermore, OH⁻ and Cl⁻ have a competitive relationship, which will inhibit Cl⁻ adsorption on the electrode surface. To sum up, it is not difficult to see that the composites have the best DE under the voltage of 1.2 V.

To compare the DE of different adsorption materials, under the condition of 1.2 V voltage, 100 mg/L initial concentration and 50 mL/min flow rate, a 200 mL NaCl solution was treated by batch mode. The results are shown in Figure 8(b). It can be seen that with the increase in the TiO₂ ratio, the DE is improved. To sum up, it can be concluded that the optimal composites ratio is 60%T/C. This may be because TiO₂ loading on CNTs can effectively improve the hydrophilicity of the material, and TiO₂ as an excellent semiconductor material can improve the efficiency of catalytic. Furthermore, the increase in titanium atoms can provide more adsorption sites for the adsorption of Cl⁻ (Li et al. 2016).

To compare the adsorption performance of TiO₂/CNTs composites and CNTs material, the desalination capacity of 60%T/C composites and pure CNTs was analyzed. Both experiments were carried out under the conditions of an initial concentration of 100 mg/L NaCl, a flow rate of 50 mL/min and a voltage of 1.2 V. The desalination performance efficiency is shown in Figure 8(c). The conductivity curves of the two materials show a downward trend and tend to be stable after a certain period. However, the decline of 60%T/C composites was much larger than that of pure CNTs. According to Equation (2), the EC of 60%T/C composites and pure CNTs was 6.5 mg/g and 3 mg/g respectively. Furthermore, according to the previous reports of TiO₂/CNTs composites used for CDI (Table 3), it can be seen that the adsorption material shows good electrosorption performance.

The long-term stability of the system is an important indicator to test the performance of the adsorbent material. Figure 8(d) shows the ion adsorption of five cycles of the 60%T/C electrode with an initial concentration of 100 mg/L NaCl under a voltage of 1.2 V and a flow rate of 50 mL/min. Observing the downward linear trend, it can be found that the DE performance of TiO₂/CNTs did not change significantly, which manifests outstanding cycle to cycle durability.

Among them, t is the electrosorption time, k₁ is pseudo-first order model constants and k₂ is pseudo-second order model constants, while C₀, C_t and C_e represent the initial concentration, the concentration at t time and the concentration at equilibrium, respectively.

Table 3 shows the reaction rate constant K-value and correlation coefficient R² of the pseudo first-order and the pseudo second-order kinetic models at different initial concentrations. It can be seen from the comparison of R² that the fitting degree of a pseudo first-order dynamic model to 60%T/C composites was higher. Figure 9(a) is a linear diagram of the adsorption kinetics of 60%T/C composites fitted by a pseudo first-order kinetic model. Combining Table 4 and Figure 9(a), it can be seen that the higher the concentration, the better the adsorption capacity.

The CDI Ragone plot as a functional tool to evaluate desalination performance in CDI, it is more applicable when considering the electrosorption behavior of a material (Kim & Yoon 2015). To this end, the initial salt concentration affecting the DE was examined. As shown in Figure 9(b). Higher salt concentrations shift the curve to the upper right. This means that both the deionizing capacity and the average deionizing rate have increased, which is similar to the conclusion of the kinetic model. The increase in capacity is primarily due to the compaction of the double layer and the subsequent rise in capacitance (Kim & Yoon 2013).

The water sample was selected from the desulfurization wastewater of Shenneng power plant in China after triple tank treatment, and had a high concentration of Cl⁻. The optimal ratio of TiO₂/CNTs composites (60%T/C) and pure CNTs adsorption material were selected for treatment. After diluting the desulfurization wastewater taken out of the power plant, the experiment was conducted under the conditions of a voltage of 1.2 V and a flow rate of 50 mL/min. Figure 10 shows the DE of 60%T/C composites and pure CNTs adsorption material. It is easy to see from the experimental results that the 60%T/C composites still exhibited excellent performance for DE in the actual desulfurization wastewater treatment.

In general, the electrosorption process is divided into the following four parts: (i) diffusion of salt ions in brine; (ii) mass transfer of salt ions between the electrode and brine; (iii) electrosorption and energy storage of ions in the electrode bilayer; and (iv) mass transfer of salt ions in the electrode pores (Yin et al. 2013). In this experiment, the catalytic mechanism of the TiO₂/CNTs material is shown in Figure 11. Overall, CNTs have a highly conductive support matrix that provides a large surface area for the dispersion of TiO₂ (Xu et al. 2020). TiO₂ is a semiconductor material, which makes the material conductive due to the presence of titanium metal and the fact that some electrons can be excited from the valence band to the conduction band driven by thermal energy. The structure of TiO₂ shows that the surface atoms of titanium dioxide nanoparticles are surrounded by a lack of adjacent, which are unsaturated and are easy for other atoms to combine and stabilize. It is worth noting that the metal cation (Ti⁴⁺) on the surface of the material and the positive charge is carried by the electrode that can better attract Cl⁻ in solution. In addition, large numbers of hydroxyl groups are contained in the surface of the TiO₂ material, which can easily adsorb water and presents hydrophilicity and compensates for the lack of hydrophobicity of CNTs (Tan et al. 2020). In summary, it can be seen that the excellent physicochemical and electrochemical properties of TiO₂ accelerate the diffusion of Cl⁻ to the electrode surface, and Cl⁻ is eventually adsorbed in the TiO₂/CNTs material. Thereby, enhanced DE is achieved.

In this study, the modification of CDI anode using TiO₂/CNTs composites was proposed for enhancing DE in desulfurization wastewater. The TiO₂/CNTs nanocomposites were successfully prepared by solvothermal method with NMP. SEM and TEM images showing the good dispersion of TiO₂ on the CNTs. Especially in the case of 60%T/C, it showed marked enhancement in wettability and high surface area. The experimental results showed that the maximum specific capacitance of 60%T/C (87.6 F/g) was obtained when the scan rate was 1 mV/s. Whether in NaCl solution or desulfurization wastewater, the maximum adsorption capacity of 60%T/C composites could reach 6.5 mg/g and demonstrated excellent stability. In addition, the experimental results were consistent with the adsorption kinetics analysis. The catalytic and adsorption mechanisms of TiO₂/CNTs were elucidated. In summary, the DE of CDI was significantly improved due to the synergistic effect of TiO₂ and CNTs. The proposed CDI system with TiO₂/CNTs as the anode has a promising application in the dechlorination of desulfurization wastewater.

This work was supported by the financial support of the National Key R&D Program of China (2018YFB0604305-01).

Shuangcheng Ma, Chang Liu and Lan Ma wrote the main manuscript text, Yongyi Xu and Feng Wang participated in the design of the experiment, and Luyue Hang translated the language of the article.

The manuscript has not been published before or submitted to another journal for the consideration of publication.

The authors declare no competing interests.

All relevant data are included in the paper or its Supplementary Information.