A Review of Modification of Carbon-Based Materials Based on Defect Engineering in Capacitive Deionization

Abstract

Capacitive deionization (CDI) is a novel water treatment technology based on the principle of double-electric-layer adsorption, which stores ions in the solution on the surface of electrodes by applying a low potential difference to achieve desalination. CDI has the advantages of low operating voltage (<1.2 V), small equipment footprint, low energy consumption, low cost and environmental friendliness. The performance of CDI is heavily dependent on the electrode materials. Carbon-based materials are widely used in CDI systems because of the large specific surface areas, lower price, and remarkable stability. To improve the CDI performance, extensive research efforts have been made for the modification of carbon-based materials. Defects in carbon-based materials play an important role in electrochemical processes and the introduction of defects is an important method to modify carbon-based materials. However, there is a lack of systematic summary of modification of carbon-based materials through introducing defects in CDI system. Therefore, this study makes the first attempt to review the modification of carbon-based materials of CDI based on defect engineering. The mechanism of enhancing CDI performance of carbon-based materials with the induction of different defects is analyzed and the future research prospects are proposed.

1. Introduction

Capacitive deionization (CDI) is a novel desalination technology for low-concentration salt water, which has attracted much attention in the past decades due to its advantages such as low cost, low energy, simple structure, east to operate, environmental friendliness [1,2]. CDI is an electrochemical-control method based on double-electric-layer (EDL) capacitance. In CDI system, the anions and cations in the solution can be adsorbed to the oppositely charged electrodes to form EDL capacitance by applying a potential difference between the electrodes (usually not more than 1.2 V), and the desalted solution can be driven out to the pool as the product at a constant feed flow rate. In the next step, by shorting the electrodes or applying an opposite potential difference between the electrodes, the anions and cations can be desorbed from the interfacial EDL region (pores) of the electrodes into the bulk liquid phase, thus obtaining a concentrated solution for waste disposal. The CDI process mainly involves electrostatic interactions, and there is no possibility of irreversible chemical reactions; therefore, the electrodes during CDI operation can go through multiple charge/discharge cycles [3]. Owing to the unique superiority, CDI is considered to play an important role in providing pollution-free drinking water and agricultural water at low cost in many application scope including brackish water desalination [4,5,6], softening [7,8,9], wastewater treatment [10,11,12], and ion-selective separation [13,14,15].

In the CDI system, carbon-based materials are widely used as the electrode materials because of their advantages such as large specific surface area, developed and interconnected pore structure, high chemical stability, safety and environmental friendliness, low cost, and easy availability [16,17]. Despite these advantages, the CDI performance of activated carbon or porous carbon as electrode materials is not satisfactory. For example, Miao et al. used 3D porous nanofiber aerogels to fabricate electrodes, which showed a desalination capacity of 17.21 mg·g⁻¹ in 200 mg·L⁻¹ NaCl solution [18]. Wang et al. prepared porous carbon deriving from pine, with which a desalination capacity of 16.18 mg·g⁻¹ was obtained in 100 mg·L⁻¹ NaCl solution [19]. Except for NaCl, Xu et al. employed an rGO electrode for the removal of ammonium, which exhibited an electrosorption capacity of 5.75 mg·g⁻¹ [20]. Pastushok et al. reported that the recovery efficiency of nitrate ions from low salinity water was 21% in a CDI system and the electrosorption capacity of nitrate ions was 5.5 mg/g utilizing microporous activated carbon [21]. Therefore, the carbon-based materials need to be modified in order to enhance the desalination capacity. Extensive research efforts have been made for the modification of carbon-based materials [22,23,24]. In our previous study, the modification methods of carbon-based materials in CDI were reviewed, which were mainly divided into four categories: element doping, metal oxide modification, chemical treatment, and surface coating [25]. The in-depth analysis of effective modification methods can provide an important direction for the further development of efficient carbon-based materials.

Recently, it has been reported that the use of defect engineering can effectively improve the electrochemical performance of carbon-based materials [26,27,28]. Defects in carbon-based materials play an important role in electrochemical processes such as electrocatalysis [29], electrochemical energy storage [30], and electrochemical desalination [31]. Conventional defect sites in carbon-based materials include the intrinsic defects (processed directly in a conjugated network without any dopants, such as edge, vacancy, and hole or topological defects), and the extrinsic defects (mainly from heteroatomic or mono-metallic atom doping). Previous studies have proved that the performance of CDI can be effectively improved through the introduction of defects [32,33,34]. For example, Gong et al. proposed the defect-rich interconnected hierarchical porous carbons by homogeneous activation strategy, which exhibited a superior desalination capacity of 31.25 mg·g⁻¹ in 100 mg·L⁻¹ NaCl solution and high stability [35]. Peng et al. reported that the defect-containing MoS₂/graphene composites displayed a superb CDI performance as compared to graphene with a desalination capacity of 25.47 mg·g⁻¹ with the voltage of 0.8 V in 200 mg·L⁻¹ NaCl solution and remarkable cycling stability [36]. Zhang et al. prepared boron-doped graphitic carbon nitride which has a large number of defects caused by lattice distortion resulting from boron doping. In CDI system, a maximum electrosorption capacity of 14.70 mg·g⁻¹ was achieved for the removal of sulfate ion from the mine wastewater [37]. However, there is a lack of a systematic summary of modification of carbon-based materials through introducing defects in the CDI system.

To fill this vacancy, this study makes the first attempt to review the modification of carbon-based materials based on defect engineering. We collected papers over 20 years from scientific database for detailed analysis, based on which the introduction of defect can be categorized as intrinsic and extrinsic defects [38,39]. Furthermore, the mechanism of enhancing CDI performance of carbon-based materials with the induction of different defects is analyzed and the future research prospects are proposed.

2. Intrinsic Defect Engineering

For carbon-based materials, intrinsic defects will inevitably exist, although extrinsic defects can be avoided by controlling the synthesis conditions and processes. Exploring the electronic structure and action mechanism of intrinsic defects can better understand the complex carbon-based systems. The excellent properties of undoped carbon-based materials show the important contribution of intrinsic defects. Intrinsic defects can effectively improve the surface chemical state, increase the specific surface area, and regulate the electronic structure of carbon-based materials, therefore, enhancing their CDI performance [40,41]. In CDI desalination process, the common defects of carbon-based materials mainly include topological defects and edge/vacancy defects [42,43], which are analyzed in detail as follows.

2.1. Topological Defects

There are almost topological defects, namely lattice defects, in all carbon-based materials, in order to meet the need of disordering to some extent. The complete carbon network is thought to be derived from the sp2 hybrid structure of pure graphene, such as the hexagonal honeycomb structure. However, as the carbon network is deformed during various synthesis processes, non-hexagonal topologies, such as pentagons, heptagons, octagons, stone defects, and even hanging bonds, can appear in the carbon framework, as shown in Figure 1. The stretching C-C bond at the defect site of the pentagon makes the carbon framework convex, while the shrinking bond of the heptagon or octagon can form the concave framework [44]. Compared with regular basic hexagonal carbon rings, these topological anomaly sites with higher energy distribution can also induce local charge redistribution in the carbon network [45]. Therefore, topological defects can make an important contribution to promoting electrochemical performance of carbon-based materials.

Previous studies [46,47] showed that graphene-like nanoflakes and multi-walled carbon nanotube films have smaller specific surface areas but significantly higher CDI desalination capacity than activated carbon. As shown in Figure 2, Zheng et al. obtained graphene nanosheets by vacuum-promoted low-temperature exfoliation, which exhibited remarkable electrosorption capacity of lead ions. This is due to the generation of topological defects caused by deformation of the carbon network during the preparation of graphene and carbon nanotube. The presence of delocalized π electron cloud in the graphene layer, which can be used as the Lewis base, has a strong affinity for cations, and their strength can be greatly enhanced by heat treatment. Scattered π-electron cloud is thought to be associated with the high CDI performance of graphene and carbon nanotube [48,49]. However, because the formation of topological defects is difficult to control, there are few relevant studies in the field of CDI. The relationship between topological defects and CDI performance should be further studied in the further.

2.2. Edge/Vacancy Defects

Edge defects are the most common intrinsic defects in the carbon framework, because a perfect crystal cannot exist without a boundary. Due to the different bonding states between the edge site and the base surface, the edges usually have various electrochemical and thermodynamic properties [50]. For a typical graphite-hexagonal network, the edges are divided into zigzag and armchair edges. In materials like graphene quantum dots, the edge-effect is remarkable. The unique skeleton arrangement and edge-electron accumulation state lead to a non-uniform distribution of electron density. Furthermore, the presence of oxygen-containing functional groups (such as hydroxyl and carboxyl groups) at the edge greatly increases the electron density, enhancing the field strength at the edge [51]. This enhanced field can promote electron migration, endowing the material with more active sites.

Vacancy/hole defects are another typical class of intrinsic defects in carbon-based materials. Vacancy defects are formed by the absence of atoms in the lattice. Vacancy defects refer to the absence of one or several carbon atoms, while hole defects refer to the absence of a large number of carbon atoms, including micropores, and mesoporous and hierarchical pore structures. For carbon-based materials, the introduction of vacancy defects can significantly increase the adsorption energy. For example, in nitrogen-doped carbon, the adsorption energy of GN-GD with vacancy defects is −3.42 eV, which is much higher than −0.23 eV of GN-G without vacancy defects [52]. Moreover, in materials with vacancy defects, the charge is more likely to accumulate at the defect sites. The effective redistribution of charge density can produce more active sites, which is beneficial to the adsorption of ions. In general, they are always accompanied by marginal sites, so the influence of vacancy/hole defects on CDI performance can also translate into the accompanying effect of marginal defects.

The edge/vacancy defects of carbon-based materials have significant influence on CDI Performance. Firstly, the adsorption sites would increase. Both edge and vacancy defects can increase the number of adsorption sites and thus improve the salt adsorption capacity [32]. Secondly, edge/vacancy defects would also accelerate the transport of the ion and electron. Edge and vacancy defects can optimize the charge density distribution and structural characteristics of materials, thereby accelerating ion and electron transport. For example, the action of carbon-vacancy defects endows the material with high wettability and conductivity, facilitating the diffusion and adsorption of ions. Due to the changes in electronic structure caused by edge and vacancy defects, the ion-adsorption capacity is enhanced [53]. For instance, the spin-polarized density-functional-theory calculations show that the vacancy-defect-containing pyridinic-N can serve as the most effective active site to further promote Na⁺ adsorption, which is of great significance for improving the salt-adsorption capacity in capacitive deionization.

2.2.1. Microporous Activated Carbon

Microporous activated carbon mostly exhibits a large blocky structure, and its pore size is mainly composed of narrow and deep micropores (less than 2 nm). Researchers have studied the CDI performance of different sources of commercially available microporous activated carbons on various salt ions in aqueous solutions. In a symmetric two-electrode cell, the amount of Na₂SO₄ removed from 1.0 mmol·L⁻¹ aqueous solution at a 1.0 mL·min⁻¹ flow rate over 80 min at an applied voltage of 1.0 V was roughly dependent on the specific surface area (SBET) of the microporous activated carbon electrode, as shown in Figure 3. Further, the cation removal and microporous activated carbon electrode regeneration were carried out. The results showed that the recovery rate of alkali metal cation was around 70–80%, and that of alkali earth metal cation was about 50–60% [54]. Hu et al. prepared microporous activated carbon using pistachio shell as the precursor, which was carbonized at 450 °C, activated with KOH at 780 °C, and then calcined at 780 °C under CO₂ atmosphere with different times. The capacitance was measured by cyclic voltammetry (CV) to compare the electrochemical performance of Na⁺ and H⁺. The capacitance results in 1.0 mmol·L⁻¹ NaNO₃ and 0.5 mmol·L⁻¹·H₂SO₄ with different scan rates are shown in Figure 4 [55]. However, the results showed that microporous activated carbon has unsatisfactory CDI performance due to its low electrical conductivity, high electron transfer resistance, and low utilization rate of specific surface area, which cannot meet the needs of practical application [56].

2.2.2. Mesoporous Carbon

In order to study the importance of pore size of carbon-based materials in CDI, mesoporous carbon has been gradually studied [57,58]. Mesoporous carbon tends to have ordered mesopores with pore sizes ranging from 2–50 nm. Ordered mesoporous carbon was prepared from a mixture of tetraethyl orthosilicate (TEOS) and triblock copolymer P123, which was carbonized at 850 °C after the silica being washed off with HF. Electroadsorption tests from a 25 mg·L⁻¹ NaCl solution at an applied voltage of 1.2 V using a two-electrode device revealed that the electroadsorption rate of mesoporous carbon was much faster than that of activated carbon, and the electroadsorption capacity of mesoporous carbon was also higher than that of activated carbon. The material analysis results showed that although the total specific surface area of ordered mesoporous carbon (844 m²·g⁻¹) was lower than that of activated carbon (968 m²·g⁻¹), the mesopore ratio (82%) was much higher than that of activated carbon (41%) [59]. Moreover, the addition of Ni to the mixture of TEOS and P123 can effectively increase the mesopore volume of mesoporous carbon, which resulted in the improvement of electroadsorption capacity for Na⁺ [60].

Mesoporous carbon, which was obtained from carbonization of calcium citrate at 1000 °C and subsequently, the removal of the template Ca with HCl solution was used to prepare the electrodes in CDI system to remove NaCl. The prepared mesoporous carbon showed an electroadsorption capacity of about 23 μmol·g⁻¹ (removal efficiency of about 89.8%) [61]. Liu et al. used a method based on block copolymers combined with electrospinning to prepare mesoporous carbon fibers as CDI electrodes. As shown in Figure 5, this electrode exhibited a high electroadsorption capacity and electroadsorption rate due to its advantages of longitudinal ion conductivity, planar ion mobility, and strong ion adsorption in mesopores [62]. The researchers further assembled graphene and mesoporous carbon into a heterogeneous structure, which improved the accessibility and mutual conductivity of mesoporous carbon to the treatment solution, achieved better ion diffusion behavior, and obtained a high electroadsorption capacity of 24.3 mg·g⁻¹ [63]. Similarly, the researchers prepared high-order mesoporous carbon nanopolyhedra by carbonizing the ordered mesoporous polymers at different temperatures. Due to the unique three-dimensional polyhedral structure and ordered mesopore distribution, this mesoporous carbon exhibited an electroadsorption capacity of 14.58 mg·g⁻¹ with an applied voltage of 1.2 V in 584 mg·L⁻¹ in NaCl solution [64].

2.2.3. Hierarchical Porous Carbon

It has been found from the current literature that the activated carbon with microporous structure has large specific surface area [65,66]. However, these deep and narrow micropores are difficult to effectively contact the solution to be treated, which significantly reduces the utilization efficiency of the micropores [67]. In addition, when ions are adsorbed in deep and narrow micropores for a long time, the regeneration of electrode materials becomes difficult, because the transport, movement, and adsorption of ions in the microporous materials are limited. Compared with microporous carbon, the pore size and pore volume of mesoporous carbon obviously increase, which are more conducive to the diffusion and movement of salt ions [68]. However, mesoporous carbon has low specific surface area and does not have sufficient active sites for ion adsorption, which seriously limit its development in the field of CDI. Therefore, further research on hierarchical porous carbon becomes decreasingly important. Hierarchical porous carbon materials simultaneously contain micropores (less than 2 nm), mesoporous (2–50 nm), and macroporous (greater than 50 nm) in a material system. The combination of these different types of pores allows the material to be fully utilized while having more adsorption sites [69,70].

The preparation methods of hierarchical porous carbon materials mainly include soft/hard template method, erosion activation method, and self-synthesis method. For example, Zang et al. used SiO₂ microspheres as hard template to cover phenolic resin, and prepared hierarchical porous hollow carbon spheres through carbonization and activation of carbon dioxide. The desalination capacity of hierarchical porous hollow carbon spheres reached 17.5 mg·g⁻¹ in 250 mg·L⁻¹ NaCl solution with the voltage of 1.4 V [71]. Qiu et al. used g-C₃N₄ as a two-dimensional template to cover the surface with water through heating together with glucose, and then used layered boric acid for further structural guidance and activation treatment to prepare hierarchical carbon nanosheets. Using this material, a good CDI desalination performance was obtained [72]. The researchers used MnO₂ nanowires as the template to prepare hierarchical porous carbon tubes which were fully mixed with LB AGAR and β-cyclodextrin, and then added KOH for carbonization activation [71]. As shown in Figure 6, the obtained material possessed a unique open hollow tubular structure with micropore–mesopores distributed in the tube wall and excellent N, S doping properties. In 25 mg·L⁻¹ NaCl solution, it had a desalination capacity of 12.05 mg·g⁻¹ [73]. In summary, we can see that the template method is more commonly used in casting hierarchical porous structures for carbon-based materials. However, in order to obtain a larger specific surface area, it often requires additional activation steps, which can also be understood as the strategy of template-activation combination.

Another effective method for preparing hierarchical pore structures is erosion activation. This method is often used to add different activators to the special precursor to fully carbonize, in order to achieve the purpose of the hierarchical pore structure. For example, the researchers proposed a simple microwave-hydrogen peroxide activation method to prepare hierarchical porous three-dimensional structures of graphene for high-performance CDI desalination. In a 500 mg·L⁻¹ NaCl solution, when the applied voltage was 1.4 V, the desalination capacity reached 21.58 mg·g⁻¹ [74]. Kim prepared hierarchical open-pore nitrogen-doped active nanoporous carbon polyhedra using a metal-organic framework ZIF-8 as a precursor by carbonization and KOH activation processes. The open-pore structure was reasonably introduced into the microporous framework, and the pore size distribution was optimized. As a result, the prepared material showed a desalination capacity of 24.4 mg·g⁻¹ in 100 mg·L⁻¹ NaCl solution [75]. Similarly, researchers used ZIF-8 as a precursor to introduce hierarchical pore structure through activation of molten salt KCl/LiCl, which greatly improved the salt removal performance of the material [76]. Further, in order to study the advantage of hierarchical pore structure in the CDI process, hierarchical porous carbon materials, which were mainly composed of large pores and micropores, were prepared by carbonizing polyvinyl alcohol and polyvinylpyrrolidone. As shown in Figure 7, the desalination capacity of hierarchical pore carbon reached 16.3 mg·g⁻¹ in a 500 mg·L⁻¹ NaCl solution with a voltage of 1.2 V, superior to that of activated carbon and ordered mesoporous carbon under the same conditions [77]. Cuong et al. prepared rice husk biochar with hierarchical pore structure by means of KOH activation from rice husk biomass carbon source. Since there is a certain amount of silica in rice husks, it is necessary to first use NaOH to treat dissolved silicon impurities. In 20 mmol·L⁻¹ NaCl solution, the salt removal rate of obtained material can reach 0.92 mg g⁻¹ min⁻¹, and the salt removal amount was 8.11 mg·g⁻¹ [78]. Yan et al. also proposed a method to prepare hierarchical porous carbon using wood as carbon precursor. Using the primary ordered network structure of wood as carbon framework, the three-dimensional ordered hierarchical porous structure was prepared by constructing micropores on its carbon wall through CO₂ activation, which exhibited a high-volume desalination capacity of 2.4 mg·cm⁻³ [79].

In addition to the above two methods for preparing hierarchical porous carbons, the self-synthesis method is the most simple and effective method. It can avoid the need for hard formwork and harsh post-processing, while also eliminating the need for additional activators. The essence of self-synthesis often lies in the precursor of the sample and the carbonization temperature. Using double-stranded metal-organic framework as a precursor, researchers prepared hierarchical porous carbon with a specific surface area up to 1500 m²·g⁻¹ through one-step carbonization, which displayed a desalination capacity of 24.17 mg·g⁻¹ in 500 mg·L⁻¹ NaCl solution [80]. In addition, Xu et al. used polymerized fibers as the template framework, on which ZIF-8 particles were grown in situ. After the internal fibers were removed by DMF solution, a 3D cross-linked multi-stage porous carbon tube network was obtained by carbonization, which has an ultra-high desalination capacity of 56.9 mg·g⁻¹ in NaCl solution [81]. Sun et al. reported that a hierarchical porous carbon material was prepared by one-step carbonization of EDTA, in which sodium ions on the chain played both a template and an activator. The obtained material had a desalination capacity of 30.27 mg·g⁻¹ in a NaCl solution of 40 mg·L⁻¹ [82]. As shown in Figure 8, researchers first used aluminum organic gel as a precursor, and the derived hierarchical porous carbon material was directly used in the CDI desalination after pickling and drying. In a NaCl solution of 500 mg·g⁻¹, the desalination capacity reached 25.16 mg·g⁻¹ with a voltage of 1.4 V [83].

2.2.4. Graphene and Fullerenes

There are also inherent defects in graphene and fullerenes. During the synthesis of reduced graphene oxide (rGO), especially in the process of exfoliation and reduction of graphene oxide (GO), the disruption of the regular hexagonal lattice structure at the edges occurs [84]. These edge-terminated atoms have unsaturated bonds, which endow the edges with high chemical reactivity. In the context of CDI, the high-reactivity edge sites can act as additional ion-adsorption sites [85]. For example, researches [86] demonstrated that rGO with a higher proportion of edge-rich structures showed enhanced ion-adsorption capacity. The edge defects can also increase the surface polarity of rGO, promoting the interaction between the material and polar electrolyte ions, which is beneficial for the initial stage of ion adsorption in the CDI process. Vacancy defects in rGO are formed when carbon atoms are missing from the graphene lattice. They can be generated during the reduction process, for instance, when oxygen-containing functional groups are removed rapidly, leading to the collapse of the local lattice structure [87]. Vacancy defects can significantly change the local electronic structure of rGO. The presence of vacancies creates localized states in the bandgap, which can facilitate the transfer of electrons during the CDI process [88]. Moreover, these defects can also act as nucleation sites for the formation of micropores, increasing the specific surface area available for ion adsorption. Research by Zhang et al. [89] showed that rGO with properly controlled vacancy defects had a higher CDI capacity compared to defect-free rGO, as the vacancies provided additional sites for ion-storage and improved the material’s overall charge-storage ability.

In fullerene-based carbon materials, edge defects can occur during the formation of aggregates or when the spherical fullerene molecules are disrupted to form carbon-based structures [90]. Similar to rGO, the edge-terminated atoms in fullerene-based materials have high chemical activity. These active edge sites can participate in chemical reactions with electrolyte components, creating additional binding sites for ions [91]. For example, in the study of fullerene-derived carbon nanotubes, the edge defects were found to enhance the interaction with metal ions in the CDI system, improving the overall ion-removal efficiency [92]. The edge defects can also change the surface energy of the material, promoting the wetting of the electrode surface by the electrolyte, which is beneficial for ion diffusion and adsorption. Vacancy defects in fullerene-based materials can be generated by mechanical treatment, such as ball-milling, or by thermal treatment [93]. These vacancies disrupt the regular molecular structure of fullerenes. The presence of vacancy defects in fullerene-based carbon materials can lead to the formation of local stress fields, which can affect the electronic distribution within the material. In CDI applications, the vacancy-induced changes in the electronic structure can facilitate the charge-transfer process between the electrode and the electrolyte, enhancing the ion-adsorption performance [94]. Additionally, the vacancies can act as channels for ion diffusion, improving the mass-transfer rate of ions within the material.

Table 1. Summary of the electrodesorption capacity of carbon-based materials with intrinsic defects.

Defect Type	Electrode Material	Experimental Conditions			Electroadsorption Capacity (mg·g−1)	Ref.
		Voltage	Concentration (mg·L−1)	Ion Type
Topological Defects	Graphene nanoribbons	1.2	500	NaCl	10.84	[95]
Edge defects	Plasma-activated biochar	1.2	68	NaCl	6.5	[96]
Vacancy defects	Activated carbon	0.4	580	NaCl	7.2	[97]
Vacancy defects	N-doped carbon nanosheets	1.2	100	NaCl	32	[32]
Vacancy defects	Carbon nanotubes	1.2	500	NH4Cl	460	[98]

summaries the CDI performance of different carbon-based materials by creating different types of intrinsic defects.

3. Extrinsic Defect Engineering

Carbon-based materials contain only a single carbon element and have an ultra-stable structural framework, which makes it difficult to make breakthroughs in electroadsorption capacity, electroadsorption rate, and ion selectivity in the application of CDI. Researchers have experimented with doping other ions or elements into carbon-based materials to improve their electroadsorption performance [98,99]. In addition to the modification of intrinsic defects of carbon-based materials, the introduction of other corresponding elements or functional groups on the surface of carbon-based materials is also a main method to improve the CDI performance of carbon-based materials, which is called the extrinsic defect engineering of carbon-based materials. The enhancement of the electroadsorption performance of CDI is achieved by adjusting the spin distribution or charge transfer on the sp2 carbon plane, which can enhance the infiltration, adsorption, and desorption of the treated salt solution by the carbon-based material, thus facilitating the full adsorption of the salt solution [100,101]. The in-depth study and analysis of the doping effect shows that the location, configuration, and content of dopants all have an important impact on CDI performance, especially in the cutoff distance of the edge effect [102,103]. For metal element doping, it is not to replace the position of the original carbon atom, but to further compound or bond the metal element with the carbon atom. This modified structure can effectively improve the electrical conductivity of carbon-based materials and reduce the resistance of charge transfer, thus improving the stability and charge efficiency of carbon-based materials [104,105].

3.1. Doping of Non-Metallic Elements

3.1.1. Nitrogen Doping

Due to the unique valence electron layer structure, nitrogen can form ionic or covalent bonds with other atoms. This characteristic makes it widely used in surface defect modification of carbon-based materials [106,107]. Doping nitrogen into a graphite network is considered to be one of the best ways to produce N-type conductive materials with improved electrical conductivity. So far, different kinds of nitrogen-doped carbon materials, for example, nitrogen-doped carbon nanofibers, nitrogen-doped porous carbon spheres, nitrogen-doped carbon aerogels, and nitrogen-doped graphene have been developed to demonstrate the enhanced CDI performance. In fact, nitrogen introduced into the carbon framework can improve the conductivity of the carbon-based materials and the wettability of the interface with the electrolyte [108,109]. Therefore, salt ions can be easily captured by the carbon-based materials, which accelerates the electroadsorption rate of salt ions [110].

Nitrogen-doped carbon-based materials are usually synthesized by thermal conversion, hydrothermal treatment, plasma treatment, and flame treatment with various nitrogen precursors. In order to obtain nitrogen-containing porous carbon framework, researchers often used nitrogen-containing MOFs, such as ZIF-8/ZIF-67, as precursors, which were calcined under an inert atmosphere to prepare nitrogen-doped porous carbon polyhedra [111]. Nitrogen-doped porous carbon was prepared by carbonization using soybean skins containing nitrogen as a carbon source and KHCO₃ as a template agent and activator. The obtained material exhibited an electroadsorption capacity of 15.8 mg·g⁻¹ and adsorption rate of 0.37 mg·g⁻¹·min⁻¹ [112]. Another common strategy for preparing nitrogen-doped carbon materials for CDI is to carbonize the various solid nitrogen precursors that are thoroughly mixed with different carbon sources. It has been reported that several types of solid nitrogen precursors, such as melamine, polyacrylonitrile, cyanamide, and urea, can be pyrolyzed to obtain different nitrogen-doped carbon materials with varying CDI results. As shown in Figure 9, in order to improve the CDI desalination ability of ordered mesoporous carbon, nitrogen-doped ordered mesoporous carbon was prepared by calcining the ordered precursor after full mixing with 1, 6-hexamethylenediamine. Effective nitrogen doping can greatly improve the hydrophilicity, electrical conductivity, and pseudo-capacitance behavior of mesoporous carbon, thus further increasing its salt removal capacity from the original 16.35 mg·g⁻¹ to 25.42 mg·g⁻¹ [113]. The researchers prepared nitrogen-doped graphene by carbonizing the mixture of graphite oxide and melamine-formaldehyde, in which melamine-formaldehyde was used as a nitrogen source. The obtained composite showed a layered porous structure with a specific surface area of 352 m²·g⁻¹ and a nitrogen doping content of 10.86%. Remarkable electrochemical capacitance, low internal resistance, and high reversibility enable the nitrogen-doped graphene composite to have an excellent electroadsorption capacity of 21.93 mg·g⁻¹ with a voltage of 2.0 V in a NaCl solution with an initial concentration of 1000 μS·cm⁻¹ [114]. The additional common method is to calcine the precursor in an ammonia atmosphere to obtain nitrogen doping. The researchers directly freeze-dried graphite oxide and then annealed it in an ammonia atmosphere to produce the nitrogen-doped graphene sponge. Its specific surface area was 526.7 m²·g⁻¹, pore volume was 3.13 cm³·g⁻¹, and nitrogen doping content was 8.5%. When the initial NaCl concentration was 500 mg·L⁻¹ and the voltage was 1.2 V, the desalination capacity reached 21 mg·g⁻¹ [115].

Extrinsic doping defects in graphene-based materials are typically introduced by doping heteroatoms such as nitrogen, sulfur, and phosphorus. Nitrogen-doping is one of the most widely studied methods. When nitrogen atoms are incorporated into the lattice of graphene-based materials, they can replace carbon atoms or be located at interstitial positions [116]. This doping can significantly modify the electronic properties of graphene-based materials. Nitrogen-doped graphene-based materials have enhanced electrical conductivity due to the increase in the number of charge carriers [117]. In terms of CDI performance, the nitrogen-containing functional groups on the doped graphene-based materials can form strong electrostatic interactions with electrolyte ions, improving the ion-adsorption capacity [118]. For example, Liu et al. [119] reported that nitrogen-doped rGO electrodes exhibited a much higher CDI capacity than undoped rGO electrodes, especially for the adsorption of cations.

Heteroatom doping is also an effective way to introduce extrinsic defects in fullerene-based carbon materials. Nitrogen-doping in fullerenes can change the molecular orbital energy levels, leading to an increase in the electron-donating ability of the material [120]. This enhanced electron-donating property can strengthen the electrostatic interaction between the fullerene-based material and electrolyte ions, which is essential for efficient charge-storage and ion-removal in CDI [121]. For example, nitrogen-doped fullerene-based electrodes showed a significant improvement of energy storage performance in EC/DEC electrolyte [122].

3.1.2. Multi-Element Doping

In addition to nitrogen, the most commonly used elements to modify carbon-based materials include sulfur, phosphorus, fluorine, and boron. Nitrogen has an atomic radius similar to that of the carbon, but the electronegativity is very different. The sulfur, phosphorus, and boron have different electronegativity and atomic radii. Thus, the introduction of sulfur, phosphorus, and boron can also change the properties of the surface structure of the carbon-based material [123,124]. In the recent study, sulfur was introduced into the graphene framework, which proved to effectively improve the CDI performance of graphene [125].

In recent years, the use of multi-element-doped carbon-based materials for CDI desalination has been widely reported. Sulfur co-doping with nitrogen has been widely studied in CDI electrode materials. Sulfur atoms have a larger atomic radius and lower electronegativity compared to nitrogen atoms. The introduction of sulfur can further increase the interlayer spacing of carbon materials, which is beneficial for ion diffusion [126]. For example, Wang et al. [127] prepared nitrogen, sulfur co-doped porous carbon by pyrolyzing a mixture of thiourea and glucose. The obtained electrode material showed a higher specific capacitance and salt adsorption capacity compared to single sulfur-doped or undoped samples. The synergistic effect of nitrogen and sulfur doping increased the number of active sites on the electrode surface, improved the electrical conductivity, and enhanced the ion storage ability.

Phosphorus co-doping with nitrogen also shows excellent modification effects. Phosphorus doping can introduce more defects into the carbon lattice, and combined with nitrogen doping, it can effectively regulate the electronic structure of the material [128]. Wu et al. [129] synthesized nitrogen, phosphorus co-doped carbon rings (NPC-Rings) through the thermal transformation of MOF/MOF nanorings and followed by the phosphidation activation. The resulting electrodes exhibited a high specific surface area, good electrical conductivity, and enhanced ion adsorption capacity in CDI applications. The co-doping of nitrogen and phosphorus not only improved the surface wettability but also optimized the pore structure, facilitating more efficient ion transport and adsorption.

Graphene aerogels doped with nitrogen and phosphorus were also prepared to extend the capacitive properties and cycle stability of carbon-based materials. Using phytic acid as a phosphorus source and chitosan as a nitrogen source, the graphene oxide was treated by mixing with acid, and the highly hydrophilic nitrogen and phosphorus co-doped three-dimensional layered carbon structure was successfully prepared. Due to its interconnected three-dimensional structure, the material showed a high electroadsorption capacity of 26.8 mg·g⁻¹ when a voltage of 1.2 V was applied [130]. Similarly, the researchers used phosphoric acid to activate graphene oxide to prepare a series of three-dimensional graphene with different pore structures, which showed a high desalination performance of 20.93 mg·g⁻¹ due to its excellent electrical conductivity and specific capacitance [131]. Further, as shown in Figure 10, Pan et al. successfully prepared nitrogen, phosphorus, sulfur co-doped hollow carbon polyhedron by using poly (cyclotriphosphonitron-co-4, 40-sulfonyl diphenol) as carbon, phosphorus and sulfur sources, and ZIF-8 as a hollow structure template and nitrogen source [132]. In a NaCl solution of 500 mg·L⁻¹, the obtained multi-element-doped material showed a desalination capacity of 22.19 mg·g⁻¹.

Boron co-doping with nitrogen is another research hotspot. Boron has a smaller atomic radius and can form strong covalent bonds with carbon atoms. The combination of nitrogen and boron doping can significantly change the electronic properties of carbon materials [133]. For instance, Yang et al. [134] developed N, B-co-doped porous carbon framework materials using a secondary doping strategy that integrates heteroatom doping with hierarchical structural design, which showed remarkable CDI performance. The co-doping increased the number of edge-sites and defect-related active sites, promoted the pseudocapacitive behavior, and thus enhanced the overall desalination performance. The porous carbon sheets co-doped with nitrogen and boron were prepared by hot water treatment and carbonization of glucose, using g-C₃N₄ as a nitrogen source and a template agent and boric acid as a boron source and activator, which displayed remarkable CDI performance [72]. The researchers prepared three-dimensional nitrogen/boron co-doped carbon nanosheets by high-temperature calcination freeze-drying chitosan gel using chitosan as nitrogen and carbon source and boric acid as a boron source [135].

In addition to sulfur, phosphorus, and boron, co-doping of nitrogen with other non-metallic elements, such as oxygen, fluorine, etc., has also been explored. Oxygen co-doping can introduce functional groups on the surface of carbon materials, improving the surface polarity and wettability [136]. Fluorine co-doping can enhance the chemical stability and hydrophobicity of materials under certain conditions [137]. These co-doping strategies provide more possibilities for the design and modification of CDI electrode materials.

3.2. Doping of Metal Elements

In addition to the common non-metallic element doping, metal element doping refers to the doping of metal elements, such as iron, cobalt, and nickel, into carbon-based materials. In general, metal element doping of carbon-based materials can effectively enhance the specific capacitance, reduce the resistance, improve the electrical conductivity, and thus, achieve the improvement of the electrochemical performance of carbon-based materials [138,139].

Gao et al. used ZIF-8 as a precursor to prepare stratified mesoporous carbon doped with nitrogen and titanium through high-temperature calcination. Due to the synergistic effect of nitrogen and titanium, the desalination capacity reached 13.89 mg·g⁻¹ and the desalination rate was 3.53 mg·g⁻¹·min⁻¹ in 500 mg·L⁻¹ NaCl solution with a voltage of 1.2 V [140]. To solve the problem of low conductivity and narrow pore size distribution of ZIF-8-derived carbon-based materials, the researchers wrapped gold nanoparticles in ZIF-8-derived nitrogen-doped porous carbon. The gold nanoparticle-modified porous carbon demonstrated extremely high conductivity and achieved a higher desalination capacity of 14.31 mg·g⁻¹, compared with 8.36 mg·g⁻¹ for the porous carbon derived from ZIF-8 alone. In addition to titanium and gold, iron doping has also been widely studied. Wang et al. put forward their own views on the tolerance of carbon-based materials in CDI applications [141]. In natural brackish water/oxy-saline water, the traditional carbon-based materials would generate hydrogen peroxide due to the oxygen reduction reaction between dissolved oxygen and the electrode surface, which corroded the electrode materials and led to poor tolerance. Based on this, as shown in Figure 11, Xu et al. proposed to use MOF as the precursors and composite with polymer fibers to generate an iron/nitrogen co-doped three-dimensional carbon tube structure [142]. The obtained structure can effectively inhibit the reduction of dissolved oxygen to produce hydrogen peroxide, but directly reduce dissolved oxygen to water, which greatly reduced the erosion of carbon electrodes. In brackish water containing oxygen, the electrode showed excellent desalination performance and cycling stability with no obvious decline in desalination capacity after 200 cycles. In addition, the researchers also demonstrated that the cobalt/nitrogen co-doped carbon-based materials exhibited more structural defects, higher specific capacitance, lower electrochemical resistance, and superior desalination capacity due to the addition of cobalt.

Table 2. Summary of the electrodesorption capacity of carbon-based materials with extrinsic defects.

Doping Element	Electrode Material	Experimental Conditions			Electroadsorption Capacity (mg·g−1)	Ref.
		Voltage	Concentration (mg·L−1)	Ion Type
N	Nitrogen-doped hierarchical porous carbon	1.2	500	NaCl	24.17	[80]
N, P	N/P co-doped eave-like hierarchical porous carbon	1.2	500	NaCl	24.14	[143]
N, Fe	Fe-N-doped hierarchical carbon	1.2	500	NaCl	28.88	[144]
N, Fe	3D interconnected Fe-N-doped carbon tubes	1.2	500	NaCl	40.70	[142]
Zn	ZnO-decorated activated carbon	1.2	23,400	NaCl	123.66	[145]
Zn	Single-atom Zn-doped N-doped carbon	1.2	500	Mn2+	280 mg/g	[146]
S	S-doped activated carbon	1.2	1000	Cr(VI)	90.2	[147]
Co, N	Co-N-doped carbon	1.2	20	Cr3+	15.20	[148]
		1.2	40	Pb2+	20.91

summaries the CDI performance of different carbon-based materials based on extrinsic defect engineering by doping different elements.

Overall, in CDI, carbon-based materials modified by intrinsic defects and doping defects exhibit distinct characteristics and mechanisms.

Table 3. Comparison of modification methods between intrinsic defects and extrinsic defect engineering.

Aspect	Intrinsic Defects	Extrinsic Defects
Defect type	topological defects, vacancy/edge defects	Substitutional or interstitial doping with heteroatoms
Formation method	High-temperature treatment, chemical activation, plasma etching	Pyrolysis doping, hydrothermal synthesis, electrochemical deposition
Effect on structure	Enhances specific surface area and pore volume; introduces lattice irregularities	Alters electronic structure via heteroatom incorporation; may form hybridized networks
Surface chemistry	Increases oxygen-containing functional groups (e.g., -COOH, -OH), conducive to ion diffusion	Introduces polar heteroatom groups, conducive to surface wettability and chemical activity
Conductivity	May reduce conductivity due to disrupted conjugation	Improves conductivity via charge carrier modulation (e.g., N-doping)
Electrosorption performance	High ion adsorption capacity via enhanced double-layer capacitance	Improved adsorption capacity and selectivity through Faradaic reactions
Cycle stability	Prone to structural degradation during cycling due to structural collapse	Exhibits better stability due to robust heteroatom-carbon bonds
Typical application scenarios	Desalination of high-concentration water (such as seawater desalination) Selective ion separation based on pore size sieving (such as Cs+/K+ separation)	Removal of low-concentration heavy metal ions (such As Cu2+, As3+); Selective adsorption of target ions in complex water quality (such as phosphate recovery)
Challenges and limitations	The defect density is difficult to control precisely, which may lead to structural instability; Selectivity depends on physical sieving and has limited ability to distinguish ions of similar sizes	The doping uniformity and stability need to be optimized; excessive doping may reduce the conductivity; Some heteroatom precursors (such as precious metals) are relatively expensive, which limits their large-scale application

makes a comparison between intrinsic and extrinsic defect engineering from different dimensions.

4. Outlook

CDI, as a new water treatment technology, has been widely concerned for its advantages of low energy consumption, low cost, high efficiency, and good renewability. The selection and modification of electrode material is one of the key factors affecting the CDI performance, which has become a hot spot in CDI research in recent years. Defect engineering is an important way to modify carbon-based materials. This paper reviews the modification of carbon-based materials based on defect engineering and analyzes the mechanism of improving CDI performance.

Despite significant advancements in defect engineering for carbon electrodes in capacitive deionization (CDI), several critical research gaps still exist:

(1)

Limited understanding of the correlation between defect types and adsorption mechanisms

Most of the existing studies focus on the influence of a single defect type (such as vacancies, edge defects, or heteroatom doping) on adsorption performance, but there is a lack of systematic comparisons of different defect types and their combined effects. The influence of the spatial distribution of defects (such as surface defects and bulk phase defects) on adsorption performance has not yet formed a unified theoretical framework, and an in-depth analysis needs to be conducted in combination with molecular dynamics (MD) and density functional theory.

(2)

Lack of precise control technology for defect density and distribution

At present, defect introduction methods (such as chemical etching and plasma treatment) are difficult to achieve precise control of defect density, resulting in poor repeatability of material properties. Furthermore, the uniformity problem of defect distribution (such as structural collapse caused by local high defect density) has not been fully solved. It is necessary to develop a defect engineering design strategy based on machine learning to achieve the collaborative optimization of defect density and distribution.

(3)

Limited understanding regarding the dynamic evolution and cyclic stability of defects

The dynamic evolution of defects during the long-term cycling process (such as defect healing, addition, or migration) and their influence on desalination performance have not been systematically studied. Most existing studies rely on electrochemical cycling tests to evaluate stability, while the microscopic mechanisms of defect structure evolution (such as the interaction between defects and electrolytes) still need to be dynamically tracked in combination with in situ transmission electron microscopy (TEM), in situ neutron diffraction, and other techniques.

(4)

Lack of research on defect regulation targeting ion selectivity

The potential of defect engineering in improving ion selectivity has not been fully exploited. Most existing studies achieve selective adsorption of 810 through surface charge regulation. However, the selective mechanisms (such as chelation and intercalation) of different defect types (such as vacancies and doping) for specific ions (such as heavy metals and multivalent ions) still require theoretical analysis. Furthermore, the defect design strategies for multi-ion competitive adsorption in complex water sources (such as high-salt wastewater) still need to be optimized in combination with mathematical simulations (such as the Nernst–Planck equation and the Frumkin–Butler–Volmer model)

(5)

Engineering challenges in Practical Applications to be addressed

The large-scale preparation and long-term operational stability of defect engineering-modified carbon electrodes still face bottlenecks. For example, although methods, such as plasma treatment and extrusion expansion, can introduce defect 1315, the amplification effect of process parameters (such as uneven defect density and decrease in material mechanical strength) has not been fully evaluated. Furthermore, the impact of defect engineering on electrode costs (such as the economy of heteroatom precursors and the energy consumption of plasma equipment) lacks systematic analysis. It is necessary to explore the optimal solution in combination with technical and economic models (such as life cycle assessment).

Based on the research gaps discussed above, suggestions for future research directions are proposed:

(1)

Basic theory of defect engineering: Systematically study the adsorption mechanisms of different defect types and establish a structure–activity relationship model of defect type–density–performance.

(2)

Precise control technology: Develop methods, such as ALD and machine learning-assisted design, to achieve precise control of defect density and distribution.

(3)

Dynamic stability research: Utilizing in situ characterization techniques to reveal the evolution laws of defects during cycling and optimize the defect–structure co-design.

(4)

Selective adsorption optimization: Design defect–functional group collaborative interfaces to enhance the selective adsorption capacity of specific ions.

(5)

Engineering application: Explore low-cost and large-scale defect introduction processes and conduct long-term operational stability tests.

In conclusion, the research on defect engineering in CDI requires breakthroughs in multiple dimensions from basic theory, technological innovation to engineering application, in order to achieve precise design and industrial application of high-performance carbon electrode materials.