Synergistic Carbon Nitride‐Coated Bismuth Composites for Sustainable Water Purification

ABSTRACT

The development of high‐performance electrodes is crucial for advancing capacitive deionization (CDI) technology. This work presents a rationally designed bismuth trioxide/graphitic carbon nitride (BO‐CN‐20) composite anode that achieves a high salt adsorption capacity of 32.63 mg g⁻¹ and good initial cycling performance (91% retention after 10 cycles). Combined experimental and density functional theory (DFT) studies reveal that the enhanced performance originates from a synergistic effect between the two components. This synergy involves efficient Cl⁻ storage via reversible redox reactions on bismuth trioxide (Bi₂O₃), coupled with the structural integrity ensured by the graphitic carbon nitride (g‐C₃N₄) framework. The composite exhibits an optimal Cl⁻ adsorption energy, which facilitates both high capture and facile release of ions. This study not only demonstrates a high‐performance CDI electrode but also offers valuable insights into the design of synergistic composite materials for advanced desalination.

The BO‐CN‐20 composite anode addresses the fundamental trade‐off between high salt adsorption capacity and long cycling stability in capacitive deionization (CDI) through a synergistic effect, where the structural confinement by g‐C₃N₄ and the reversible chloride ion storage via Bi₂O₃ together achieve simultaneous enhancement in both aspects.

1. Introduction

With the global population expanding swiftly and industrial development pressing forward, water shortage has emerged as a critical issue confronting the entire world. The salinity of seawater not only threatens ecosystems but also presents potential risks to human health, necessitating the development of efficient and sustainable seawater desalination technologies [1, 2, 3]. Conventional methods for extracting fresh water from seawater, including processes like reverse osmosis (RO) and multi‐effect distillation (MED), are often hindered by issues such as significant energy requirements, membrane contamination, elevated operational expenses, and intricate system upkeep, all of which constrain their large‐scale deployment [4, 5, 6]. CDI, a desalination technique characterized by low energy consumption and high efficiency, is increasingly acknowledged as a potentially viable solution to the worldwide freshwater scarcity problem. Its working principle depends on the generation of electrical double layers at the electrode‐electrolyte boundary, where an imposed electric field promotes ion adsorption onto the electrode surface, thereby achieving rapid and reversible desalination [7, 8]. Electrode materials serve a vital function in CDI systems, directly influencing desalination efficiency and cycle stability. Conventional electrode materials, including activated carbon, graphene, and carbon nanotubes, are widely employed in CDI systems. However, they present significant limitations in ion‐selective adsorption and desalination capacity (typically less than 20 mg_NaCl g⁻¹), which hampers their ability to meet the performance requirements of efficient desalination [9, 10, 11, 12, 13]. Thus, recent research efforts have concentrated on developing electrode materials that effectively facilitate Faradaic reactions. Considerable advances have been achieved in sodium storage materials, which demonstrate outstanding specific capacity and fast charge‐discharge kinetics [14, 15, 16]. In contrast, highly efficient anode materials for chloride ion storage remain limited and underexplored.

In recent years, silver (Ag) has become a popular choice as the main electrode material for chloride ion (Cl⁻) storage in seawater desalination processes [17, 18, 19, 20]. Nonetheless, the elevated cost of silver and the poor electrical conductance of its transformation product, AgCl, impose notable restrictions on its potential for practical application. Compared to silver‐based materials, bismuth‐based materials are increasingly favored as effective and practical electrodes for Cl⁻ capture, primarily because of their cost‐effectiveness and reduced energy consumption. Bi/BiOCl has shown great promise as a chloride ion storage electrode material [21, 22, 23]. For example, Choi et al. [24] utilized bismuth foam as a promising Faraday electrode for chloride ion removal in electrolytic deionization (EDI), demonstrating a chloride ion removal capacity of 82.9 mg g⁻¹. Additionally, Shi et al. [25] introduced a nitrogen‐doped porous carbon framework embedded with bismuth nanoparticles (Bi/N‐PC) for CDI systems. Under a current density of 250 mA g⁻¹, the Bi/N‐PC material exhibited a specific capacity of 410.4 mAh g⁻¹, a desalination capacity of 113.4 mg g⁻¹ at 100 mA g⁻¹, and a desalination rate of 0.392 mg g⁻¹ s⁻¹. However, existing bismuth‐based electrodes suffer from issues such as approximately 158% phase transition volume expansion, poor conductivity of BiOCl, and inadequate hydrophilicity. These defects lead to the adsorption and release of chloride ions, causing bismuth electrode powdering during multiple cycles, severely damaging the active components of the electrode and resulting in rapid performance degradation [26, 27, 28].

To overcome the intrinsic limitations of bismuth‐based electrodes, recent studies have explored the incorporation of carbon materials into bismuth‐based systems for CDI applications. The resulting composites exploit synergistic interactions, yielding increased specific surface area, improved conductivity, and enhanced chloride ion adsorption. Graphitic carbon nitride (g‐C₃N₄) constitutes a nitrogen‐abundant porous 2D substance featuring an sp²‐hybridized carbon skeleton, exhibiting excellent electronic properties and maintaining superior stability under extremely high thermal, photochemical, and chemical erosion conditions [29, 30, 31]. In recent years, it has garnered significant attention in fields such as optoelectronic devices, catalysis, and energy storage [32, 33]. Bismuth oxide (Bi₂O₃), as a semiconductor material, can effectively enhance charge separation efficiency and reduce charge recombination through morphology control, metal doping, and heterostructure construction, demonstrating excellent electrochemical performance during capacitive deionization processes [34, 35]. Currently, the construction of heterojunctions such as Bi₂O₃/ZIF‐67 and δ‐Bi₂O₃@PCNF can effectively suppress charge recombination, allowing for the efficient elimination of tetracycline and phosphate from aqueous solutions via capacitive deionization [36, 37]. Additionally, bismuth‐carbon composite materials prepared using a bismuth‐carbon interface coupling strategy can effectively enhance the performance of individual components and mitigate the loss of Bi particles during electrosorption [38].

Herein, we demonstrate an interlayer insertion confinement strategy using g‐C₃N₄ as a host to uniformly load Bi₂O₃ nanoparticles. This approach effectively inhibits Bi₂O₃ agglomeration and exposes abundant active sites. By systematically analyzing the surface morphology, hierarchical structure, and valence state distribution, the optimal composite BO‐CN‐20 was identified. When tested as a capacitive deionization anode at 1.2 V in 500 mg L⁻¹ NaCl, the BO‐CN‐20 electrode delivers a salt adsorption capacity of 32.63 mg g⁻¹ and a salt adsorption rate of 1.05 mg g⁻¹ min⁻¹. The electrode retains 91% of its capacity after 10 cycles, which is attributed to an integrated mechanism that couples electrical double‐layer adsorption with pseudocapacitive Bi⁰/Bi³⁺ redox reactions. Density functional theory calculations further confirm an optimal chloride adsorption energy in the composite, enabling a good balance between efficient capture and facile desorption of chloride ions. Moreover, the layered confinement effect of g‐C₃N₄ effectively prevents bismuth detachment during repeated Cl⁻ insertion/extraction. This work not only develops a high‐performance BO‐CN composite electrode but also establishes a synergistic design strategy combining optimal adsorption energy and structural confinement for durable desalination materials.

2. Materials and Methods

The first step involves the preparation of g‐C₃N₄ material. Five grams of melamine are directly pyrolyzed at 550°C in a muffle furnace for 4 h, with a heating rate of 5°C per minute, to form block‐shaped CN. The obtained bulk CN powder was then manually pulverized using a mortar and pestle. The second step of the synthesis involves embedding Bi₂O₃ into g‐C₃N₄ to prepare Bi₂O₃/g‐C₃N₄ (BO‐CN). This process includes adjusting the amount of Bi₂O₃ used for synthesizing the composite material. Specifically, a total of 10 milligrams of g‐C₃N₄ and x milligrams of Bi₂O₃ (x = 5, 10, 20, 30, and 40) were combined and finely ground in a mortar, followed by calcination at 550°C in a muffle furnace for 3 h. Ultimately, the composite material labeled BO‐CN‐x was successfully obtained through the synthesis process. Figure 1a illustrates the preparation process. Comprehensive details on the chemicals used for material synthesis, the primary instruments employed for material characterization, electrochemical analysis of BO‐CN, along with the experimental parameters and analytical approaches for assessing the deionization performance of BO‐CN hybrid capacitors, are provided in the electronic Supplementary Materials section.

FIGURE 1.

(a) Preparation process of BO‐CN. SEM images of (b) Bi₂O₃, (c) g‐C₃N₄, and (d) BO‐CN‐20. (e) TEM images of BO‐CN‐20. (f) XRD patterns of g‐C₃N₄, Bi₂O₃, and BO‐CN‐20. (g) FTIR patterns of g‐C₃N₄, Bi₂O₃, and BO‐CN‐20. (h) EDX elemental mapping.

3. Results and Discussion

3.1. Structure and Composition Analysis

Scanning electron microscopy (SEM) images of Bi₂O₃, g‐C₃N₄, and BO‐CN‐20 are presented in Figure 1b–d. The surface of monocomponent Bi₂O₃ particles is evidently rough, with an irregular, agglomerated morphology. Most particles have a small surface area, but some larger particles are still present, indicating that the sample may exhibit a certain degree of agglomeration. g‐C₃N₄ consists of an irregular block‐like structure constructed through the stacking of numerous layered nanosheets, which are densely distributed and have a smooth surface. The majority of the BO‐CN‐20 sample exhibits a plate‐like nanoplate morphology, which features a substantial specific surface area, offering additional adsorption sites. In some regions, Bi₂O₃ particles are observed to be intercalated between g‐C₃N₄ layers, which facilitates closer interface contact, thereby boosting electron transport and ion migration during capacitive deionization. Higher‐magnification SEM images of the three samples are provided in Figures S5‐S7, where the detailed microstructure and the intercalation morphology can be clearly seen.

To examine the microstructural features of the synthesized materials, the BO‐CN‐20 sample with the highest adsorption performance was selected for transmission electron microscopy (TEM) analysis, as presented in Figure 1e. The grey regions stem from layered g‐C₃N₄, whereas the dark particles are attributed to Bi₂O₃ nanoparticles dispersed on the surface of g‐C₃N₄, with diameters ranging from approximately 50 to 100 nanometers. This arises from the greater atomic mass of bismuth compared with that of carbon and nitrogen, thereby preventing the electron beam from penetrating the Bi₂O₃ layer [39]. Furthermore, the HRTEM micrograph shown in Figure 1e demonstrates the existence of BO‐CN‐20 composites, where the observed lattice fringes exhibit interplanar spacings of 0.328 nm, 0.317 nm, and 0.269 nm. These measurements match the (120), (012), and (202) crystalline planes of Bi₂O₃, respectively. These results align with the XRD patterns presented in Figure 1f. Notably, while the (002) peak of g‐C₃N₄ is also capable of generating lattice fringes with an interplanar spacing of 0.323 nm, such fringes are absent in Figure 1e. This is because the (002) peak originates from the stacked layers of g‐C₃N₄ that lie parallel to the screen. In summary, from TEM and HRTEM analyses, it can be deduced that a distinct interface has formed between g‐C₃N₄ and Bi₂O₃, which is beneficial for salt ion adsorption.

To further verify the elemental makeup of the synthesized BO‐CN‐20 composite, EDS (Figure S8) and EDX (Figure 1h) measurements were performed. Element mapping characterization revealed that Bi, C, N, and O elements are evenly distributed across the composite material. These findings further validate the effective formation of BO‐CN‐20 composite through a simple co‐calcination approach.

To explore the crystalline structure of composite materials thoroughly, X‐ray diffraction (XRD) analyses were conducted. Figure 1f and S9 present the XRD spectra of g‐C₃N₄, Bi₂O₃, and BO‐CN composites with varying proportions. Distinct diffraction peaks of g‐C₃N₄ were observed at 2θ = 12.89° and 27.63°, corresponding to the (100) and (002) lattice planes, respectively [40, 41, 42]. The first peak originates from the ordered arrangement of condensed triazine units, while the second is attributed to the stacking of conjugated aromatic hydrocarbon layers in the interlayer regions of g‐C₃N₄. From the XRD pattern of Bi₂O₃ (ICDD Card No. 01‐071‐0465), diffraction peaks at 25.749°, 26.906°, 27.392°, 28.005°, 32.245°, and 46.342°, corresponding to the (002), (111), (120), (012), (202), and (041) crystal planes of the monoclinic phase of Bi₂O₃, demonstrating that the synthesized Bi₂O₃ possesses excellent crystallinity, in agreement with previous reports in the literature. In the composite material BO‐CN‐x samples, due to the overlap of the (120) lattice plane of Bi₂O₃ at 27.392° and the (002) lattice plane of g‐C₃N₄ at 27.63°, only the diffraction peaks attributed to Bi₂O₃ are detectable. The incorporation of nitrogen and carbon elements typically enhances conductivity, thereby significantly improving desalination capacity. Notably, compared to the pure Bi₂O₃ sample, the BO‐CN‐x composite exhibits two new peaks at 37.95° and 39.62° (marked with black triangles), indicating the formation of a new metallic Bi⁰ phase upon introducing g‐C₃N₄ into Bi₂O₃. Although the characteristic diffraction peak intensity of elemental bismuth is not prominent, it verifies that elemental bismuth exists in the BO‐CN composite structure. The generation of elemental bismuth stems from the reduction of Bi₂O₃ by carbon generated during the carbonization process. Subsequent EDS and XPS analyses verify the existence of bismuth and g‐C₃N₄ within the composite material.

Fourier transform infrared (FTIR) spectroscopy was utilized to analyze the structural features of g‐C₃N₄, Bi₂O₃, and BO‐CN‐20. As depicted in Figure 1g, the peak at 810 cm⁻¹ in g‐C₃N₄ corresponds to the bending vibration of the s‐triazine unit [43, 44]. The pronounced bands observed between 1650 and 1230 cm⁻¹ are linked to C‐N bond stretching in nitrogen‐containing heterocycles, while the broad absorptions appearing in the 3400‐3000 cm⁻¹ region are associated with O‐H and N‐H stretching vibrations [45, 46]. These spectral features collectively verify the presence of g‐C₃N₄ in the sample. For pure Bi₂O₃, the distinct absorption peaks detected at 430 and 506 cm⁻¹ are associated with Bi‐O bond vibrations within the Bi₂O₃ lattice framework [47]. In comparison with the spectrum of the pristine sample, the BO‐CN‐20 composite material exhibits stretching vibrations corresponding to both g‐C₃N₄ and Bi₂O₃ functional groups. However, the intensity and sharpness of the characteristic bands in the BO‐CN‐20 composite material's spectrum are slightly reduced, which can be attributed to the reduction in particle size and the uniform dispersion of Bi₂O₃ across the g‐C₃N₄ surface. This conclusion is further corroborated by porosity analysis derived from N₂ adsorption‐desorption isotherms and pore size distribution curves (Figure S10). While Bi₂O₃ exhibits negligible N₂ uptake, indicating a nonporous structure, both g‐C₃N₄ and the BO‐CN‐20 composite show type IV isotherms with H3‐type hysteresis, characteristic of mesoporous materials. The specific surface area and pore volume data (Table S1) confirm that the integration of Bi₂O₃ with the g‐C₃N₄ framework effectively tailors the textural properties of the composite, consistent with the XRD and FTIR results.

To further determine the surface chemical composition of the BO‐CN‐20 composite, X‐ray photoelectron spectroscopy (XPS) was employed to examine the chemical states of the constituents in Bi₂O₃, g‐C₃N₄, and BO‐CN‐20. Figure 2a displays the full XPS spectrum of the composite material. Characteristic peaks corresponding to C, N, O, and Bi were observed in the BO‐CN‐20 structure, further confirming the existence of Bi₂O₃ and g‐C₃N₄ components in the hybrid structure. Figure 2b compares the C 1s binding energies of the pure g‐C₃N₄ and BO‐CN‐20 composite materials, with no significant differences observed between the two. A binding energy at 284.8 eV corresponds to carbon atoms involved in C‐C bonding within sp²‐hybridized regions of g‐C₃N₄. Conversely, the signal at 288.3 eV is linked to sp²‐hybridized carbon in the triazine ring structures (N‐C = N) characteristic of the g‐C₃N₄ framework [48, 49]. As depicted in the N 1s spectrum shown in Figure 2c, distinct signals are identified at binding energies of 398.5 eV, 400.4 eV, and 404.3 eV. The N 1s binding energy of 398.5 eV in the composite matches that of pure g‐C₃N₄, corresponding to sp²‐hybridised nitrogen (C = N‐C), which confirms the presence of graphite‐like sp²‐bonded g‐C₃N₄. The peak at 400.4 eV corresponds to the tertiary nitrogen N‐(C)₃ bonded to aromatic carbon atoms [50]. Additionally, a weak π‐excitation is detected at 404.3 eV [51]. Figure 2d presents the high‐resolution Bi 4f spectrum of BO‐CN‐20 and Bi₂O₃. Four peaks were observed in BO‐CN‐20, with the peaks at 157.6 and 162.9 eV corresponding to Bi‐Bi bonds in metallic bismuth, indicating that part of the Bi₂O₃ in the composite material has been reduced to metallic bismuth. In pure Bi₂O₃, the peaks at 164.1 eV and 158.8 eV correspond to the Bi 4f_5/2 and Bi 4f_7/2 orbitals, respectively, signifying the existence of Bi³⁺ in bismuth oxide [52, 53]. In the BO‐CN‐20 composite material, these two peaks shift to 164.3 eV and 158.9 eV, respectively, with electrons transferring from Bi₂O₃ to g‐C₃N₄. This indicates that electronic interactions occur between Bi₂O₃ and g‐C₃N₄ in the course of composite synthesis, resulting in a reduction in the electron density of bismuth atoms. It further suggests that a composite structure has formed between the two, rather than a simple physical mixture.

FIGURE 2.

(a) Full XPS spectra of g‐C₃N₄, Bi₂O₃, and BO‐CN‐20. (b‐d) High‐resolution XPS spectra of C 1s, N 1s, and Bi 4f for the relevant samples.

3.2. Electrochemical Performance

The electrochemical properties of the samples were assessed via a three‐electrode system, with the specific experimental protocol outlined in the Methods section. Figure 3a presents the cyclic voltammetry (CV) profiles of BO‐CN‐x measured at a scanning rate of 50 mV s⁻¹. According to the data presented in the figure, all BO‐CN composites exhibit an oxidation peak close to 0.2 V and a reduction peak at approximately ‐1.2 V. Figure 3b illustrates the CV profiles of BO‐CN‐20 recorded at a series of scan rates between 1 and 50 mV s⁻¹. At a scan rate of 1 mV s⁻¹, a pronounced redox pair is observed between ‐0.79 V and ‐0.021 V. This signal is characteristic of the reversible Bi⁰/Bi³⁺ redox couple, confirming the highly reversible nature of the electrochemical process. With increasing scanning speed, the redox peaks gradually shift toward more positive and negative potentials, suggesting accelerated ion transport and facilitated charge transport across the electrode‐electrolyte boundary. At elevated scan rates, Cl⁻ fails to diffuse from the electrolyte into the inner part of the material promptly and is thus confined to interact with the outer active surface.

FIGURE 3.

(a) CV curves of BO‐CN‐x at 50 mV s⁻¹. (b) CV curves of BO‐CN‐20 at different sweep rates. (c) Diffusion‐ and capacitive‐controlled contributions of BO‐CN‐20. (d) The capacitive kinetics analysis of BO‐CN‐20. (e) GCD curves of BO‐CN‐20 at different current densities. (f) Nyquist plots of BO‐CN‐x.

The CV curve can be deconvoluted into two kinetic contributions: a capacitive (surface‐controlled) process and a diffusion‐controlled mechanism. The former arises from rapid, non‐Faradaic interfacial charge storage, while the latter is governed by ion diffusion through the bulk of the electrode material, typically associated with slower Faradaic processes. As illustrated in Figure 3d, the b‐values corresponding to the oxidation and reduction peak currents of the BO‐CN‐20 electrode are 0.57 and 0.66, respectively. These figures indicate that the charge storage behavior is dominated by a combined contribution of diffusion‐controlled and capacitive‐controlled processes. This dual‐mode ion storage characteristic is beneficial for realizing efficient desalination. Figure 3c illustrates the proportion of capacitive and diffusion‐controlled contributions to the overall charge storage of BO‐CN‐20 at various scan rates. A reduction in scan rate leads to a progressively stronger influence from diffusion‐controlled processes, implying an enhanced involvement of the Bi/BiOCl conversion mechanism (Bi + Bi₂O₃ + 3Cl⁻ ↔ 3BiOCl + 3e⁻). In contrast, the capacitive contribution associated with g‐C₃N₄ shows a declining trend. This phenomenon arises because at elevated scan rates, the redox‐induced conversion reaction cannot proceed fully. Under such conditions, the CV profile delivers a more accurate depiction of the intrinsic ion‐storage characteristics of the electrode material.

BO‐CN composites with different proportions were subjected to galvanostatic charge‐discharge (GCD) testing at a current density of 1 A g⁻¹. Charge‐discharge plateaus were observed for all samples, indicating the occurrence of a distinct Faradaic reaction, as shown in Figure S11. This observation verifies that Cl⁻ storage in the Bi‐based electrode occurs via redox processes, which aligns well with the findings derived from CV curve analyses. It is noteworthy that the BO‐CN‐20 composite exhibited a markedly larger charge‐discharge plateau region compared to other compositions, suggesting superior specific capacity. Under a current density of 1 A g⁻¹, the BO‐CN‐20 electrode delivered a specific capacitance of 92.5 F g⁻¹, reflecting its efficient charge‐storage capability. The GCD curves recorded at 1, 2, 5, and 10 A g⁻¹ are presented in Figure 3e, all of which display good reversibility even at elevated current densities. Furthermore, long‐term GCD cycling tests were performed on BO‐CN‐20 for 100 consecutive cycles at 5 A g⁻¹ to further evaluate its intrinsic electrochemical stability. As shown in Figure S12, the GCD profiles remained generally stable during cycling. The slight extension of the charge‐discharge duration in later cycles may be attributed to gradual electrode activation and improved accessibility of electroactive sites. After 100 cycles, the BO‐CN‐20 electrode exhibited negligible attenuation, indicating good electrochemical reversibility and cycling stability.

To gain an in‐depth understanding of ion transport behavior and assess the inherent resistance properties within the electrode configuration, a systematic electrochemical impedance spectroscopy (EIS) investigation was conducted. Figure 3f illustrates that these curves lack distinct semicircles, signifying that the BO‐CN electrode possesses an extremely low charge transfer resistance. The slope of BO‐CN‐20 is greater than that of other material compositions, indicating faster redox reaction rates at the electrode‐electrolyte boundary, which efficiently facilitates ion diffusion and migration capabilities, thereby enhancing the overall desalination efficiency of CDI. In summary, BO‐CN‐20 demonstrates excellent conductivity and electrochemical properties.

3.3. Desalination Application

To systematically evaluate the performance of BO‐CN‐x electrodes in capacitive deionization technology, we set up a CDI testing system and conducted batch experiments in constant current mode. In this system, BO‐CN‐x functions as the anode, whereas AC serves as the cathode. Cl⁻ is intercalated into the BO‐CN‐x anode electrode through a pseudocapacitive energy storage mechanism, while Na⁺ is adsorbed and stored on the AC cathode mainly through electrical double‐layer adsorption (principle depicted in Figure 4a). Since the same AC cathode and identical operating conditions were employed throughout all BO‐CN‐x‐based CDI tests, the observed differences in desalination performance can be mainly attributed to the chloride‐storage capability of the BO‐CN‐x anode. Accordingly, Figure 4b presents the desalination performance of the BO‐CN‐x electrode operated at 1.2 V in a NaCl solution with a concentration of 500 mg L⁻¹. Notably, the BO‐CN‐20 electrode exhibits an exceptional SAC of 32.63 mg g⁻¹, significantly outperforming other electrode configurations. This performance enhancement is primarily attributed to the composite structure formed by g‐C₃N₄ and Bi₂O₃, which facilitates ultrafast electron transport within the conductive pathways. Figure S13 further illustrates the solution concentration‐time variation curves of the BO‐CN‐20 electrode at an operating voltage of 1.2 V with NaCl electrolyte concentrations spanning 300 to 1000 mg L⁻¹. At initial salt levels of 300, 500, and 1000 mg L⁻¹, adsorption equilibrium was reached within 28, 31, and 42 min, respectively. As illustrated in Figure 4c, the highest SAC is observed at a NaCl concentration of 500 mg L⁻¹, suggesting that this concentration represents the most favorable condition for ion uptake.

FIGURE 4.

(a) Schematic diagram of the CDI working principle. (b) SAC of BO‐CN‐x. (c‐d) SAC and Kim‐Yoon curves of BO‐CN‐20 at different NaCl concentrations. (e‐f) SAC and Kim‐Yoon curves of BO‐CN‐20 at different voltages. (g) Cycling performance of BO‐CN‐20. (h) Comparison of the salt removal performance between the BO‐CN‐20‐based CDI cell and other systems utilizing previously reported electrode materials.

Although higher concentrations of NaCl solution theoretically provide stronger electric field driving forces, thereby enhancing adsorption capacity, once the level reaches 1000 mg L⁻¹, ion crowding at the electrode surface may occur due to excessive ion accumulation, limiting effective adsorption. Additionally, the charge efficiency decreases at high salt concentrations, resulting in fewer salt ions being adsorbed per unit charge. In contrast, ion migration is more thorough and charge utilization is higher at an appropriate NaCl concentration. To further optimize desalination performance, we examined the functioning of the CDI system at different operating voltages. Figure S14 presents the system performance evaluated under imposed potentials ranging from 1.0 to 1.4 V in a 500 mg L⁻¹ NaCl solution.

The results indicate that as the imposed voltage rises, the salt removal capability of the NaCl solution gradually enhances, reaching a peak at 1.2 V. At applied potentials of 1.0, 1.2, and 1.4 V, the system achieved salt adsorption capacities of 14.65, 32.63, and 22.31 mg g⁻¹, respectively (Figure 4e). This enhancement could stem from the activation of extra redox‐active sites at higher voltages, enhancing the Cl⁻ insertion process driven by Faradaic reactions, thereby improving the diffusion‐controlled ion removal efficiency. This mechanism differs from traditional double‐layer adsorption and helps overcome the co‐ion repulsion effect. Prior research has demonstrated that once the CDI system voltage surpasses 1.2 V, it can lead to water oxidation reactions, thereby reducing the overall system efficiency.

Therefore, the maximum operating voltage in the desalination experiments was set to 1.2 V to balance desalination efficiency and stability. To further evaluate the comprehensive desalination efficiency of BO‐CN‐20, Figure 4d,f show the Kim‐Yoon plots, considering the balance between SAR and SAC [54, 55]. The curves corresponding to 1.2 V and a NaCl concentration of 500 mg L⁻¹ are situated in the upper right quadrant, suggesting that the BO‐CN‐20 electrode delivers both the greatest salt removal capacity and the most rapid ion adsorption rate under these operating conditions.

To evaluate the stability of BO‐CN‐20 in the CDI system, the leaching of bismuth ions into the solution was examined. Following the tenth desalination/regeneration cycle, the bismuth concentration in the solution was determined via ICP‐OES. The findings showed that the bismuth content in the wastewater after cycling was below the detection threshold, indicating that no ions dissolved from the electrode into deionized water, even following prolonged charge‐discharge cycles. Additionally, cyclic stability is an essential parameter reflecting the viability of capacitive deionization systems for practical applications, and it plays a pivotal role in ensuring long‐term operational reliability. As shown in Figure 4g, the BO‐CN‐20 electrode retained more than 90% of its adsorption capacity after ten consecutive cycles in NaCl solution, indicating good initial cycling stability. However, extended cycling tests revealed gradual capacity decay during prolonged CDI operation, with the retention decreasing to approximately 50% after 30 cycles (Figure S15), indicating that the long‐term stability of the present system remains limited. The observed decay is likely associated with co‐ion interference and the incomplete reversibility of the Bi/BiOCl conversion during repeated CDI operation, indicating that the long‐term operational stability of the present system still requires further improvement. Figure 4h further compares the adsorption performance of the BO‐CN‐20 electrode developed in this work to that of previously documented electrodes, with detailed data summarized in Table S2 [56, 57, 58, 59, 60, 61, 62]. These findings reveal that the BO‐CN‐20 electrode significantly outperforms most reported materials in terms of salt adsorption capacity, making it a highly promising capacitive desalination material.

3.4. Desalination Mechanism of BO‐CN‐20 Composites

The interaction mechanism between the BO‐CN‐20 composite and Cl⁻ was systematically investigated using XRD, SEM, XPS, and density functional theory (DFT) calculations. The exceptional cycling stability of the BO‐CN‐20 electrode can be attributed not only to the efficient chloride ion storage capability of bismuth but also to the confinement effect provided by the carbon nitride layers. To validate this finding, SEM morphological analysis was conducted on the BO‐CN‐20 electrode after 10 cycles. The results indicate that the electrode's surface structure remains largely preserved (Figure 5a, inset), with no significant morphological changes, further corroborating its excellent structural stability. The layered structure of g‐C₃N₄ remained intact, without noticeable cracking or delamination. Meanwhile, Bi₂O₃ particles were uniformly distributed, showing no evident detachment or agglomeration. These observations suggest that the g‐C₃N₄ layered framework effectively provides structural support for Bi₂O₃ particles, thereby mitigating structural degradation during cycling. Based on the above analysis, the capacity decay of the BO‐CN‐20 electrode during long‐term CDI operation is primarily attributed to the co‐ionic effect associated with the absence of an ion‐selective membrane. This effect compromises charge efficiency and hinders the complete electrochemical reduction of BiOCl in subsequent cycles, thereby resulting in a progressive accumulation of inert BiOCl and an irreversible decay in desalination capacity [63].

FIGURE 5.

(a) Comparison of the XRD patterns and corresponding SEM images for the BO‐CN‐20 electrode before and after 10 cycles. (b‐c) High‐resolution Bi 4f and Cl 2p XPS spectra of the BO‐CN‐20 composite after Cl⁻ adsorption. (d‐f) Adsorption configurations of Cl⁻ on (d) g‐C₃N₄, (e) Bi₂O₃, and (f) BO‐CN‐20. (g) The corresponding adsorption energies of Cl⁻ for the different structures.

To further elucidate the desalination mechanism, the BO‐CN‐20 electrode was subjected to XRD analysis in both the adsorbed and desorbed states. As shown in Figure 5a, the XRD pattern of the electrode in the adsorbed state displays distinct diffraction peaks corresponding to BiOCl, indicating a phase transformation from Bi/Bi₂O₃ to BiOCl during chloride ion capture. Notably, these BiOCl diffraction peaks nearly disappear upon desorption, suggesting the effective reduction of BiOCl and demonstrating good reversibility of the reaction. In addition, time‐dependent XRD patterns collected at different adsorption stages further reveal that the overall diffraction profiles remain largely similar during the intermediate adsorption stages, which may be attributed to the relatively low amount of generated BiOCl and the overlap of some BiOCl‐related peaks with those of the parent phases. More discernible changes appear at the saturated adsorption state, suggesting that the accumulation of the BiOCl phase becomes more evident as chloride adsorption approaches saturation, as shown in Figure S16. Complementary XPS analysis was employed to probe the surface chemical states (Figure 5b). Consistent with previous findings, the XPS spectrum of the pristine BO‐CN‐20 electrode confirms the coexistence of metallic Bi and Bi₂O₃. After Cl⁻ adsorption, however, the Bi 4f spectrum undergoes a significant change, dominated by new peaks at 159.8 eV (Bi 4f_7/2) and 165.2 eV (Bi 4f_5/2), which are characteristic of the BiOCl phase [64, 65]. Corroborating this, the Cl 2p spectrum (Figure 5c) displayed a pair of spin‐orbit doublets at 198.6 eV (Cl 2p_3/2) and 200.2 eV (Cl 2p_1/2), which are attributable to Cl⁻ in the BiOCl lattice. These results demonstrate that chloride ions are stabilized on the electrode surface through the formation of BiOCl. Collectively, the data indicate the following reversible conversion reaction during adsorption/desorption:

$$\mathrm{Bi}+\mathrm{B}{\mathrm{i}}_2{\mathrm{O}}_3+3\mathrm{C}{\mathrm{l}}^{-}\leftrightarrow 3\mathrm{BiOCl}+3e^{-}$$

This reaction facilitates a reversible phase transition between Bi/Bi₂O₃ and BiOCl, enabling the cyclic storage and release of Cl⁻.

To elucidate the enhanced capacitive deionization mechanism at the atomic level, DFT calculations were employed. The interaction energy between Cl⁻ and the g‐C₃N₄, Bi₂O₃, and BO‐CN‐20 structures was assessed through structural optimization of the corresponding models. The corresponding atomic models are presented in Figure S17. As revealed by the computed adsorption energies in Figure 5d–g, the three materials exhibit distinct characteristics. g‐C₃N₄ shows weak physisorption (‐0.255 eV), corresponding to electric double‐layer capacitive behavior. In contrast, Bi₂O₃ exhibits strong chemisorption (‐2.669 eV), consistent with its pseudocapacitive nature that originates from reversible bismuth redox reactions. Notably, the BO‐CN‐20 composite exhibits an intermediate adsorption energy of ‐1.538 eV. This intermediate value strikes an optimal balance between its constituent materials. The optimized binding energy is pivotal, as it circumvents the kinetic limitations of ion release inherent to the excessively strong binding in pure Bi₂O₃, while simultaneously compensating for the intrinsically low charge storage capability of g‐C₃N₄. Thereby, the composite facilitates both efficient chloride capture and highly facile release, providing a mechanistic foundation for its superior salt adsorption capacity and exceptional cycling stability observed experimentally.

4. Conclusion

In summary, Bi₂O₃‐doped g‐C₃N₄ composite materials were successfully synthesized and applied in a CDI system for efficient chloride‐ion removal. By optimizing the composition and operating conditions, the BO‐CN‐20 electrode achieved a SAC of 32.63 mg g⁻¹ and a SAR of 1.05 mg g⁻¹ min⁻¹ in a 500 mg L⁻¹ NaCl solution at 1.2 V, together with favorable electrochemical reversibility. Mechanistic analyses based on XRD, XPS, and DFT calculations reveal that the enhanced desalination performance arises from the synergistic coupling between Bi₂O₃ and g‐C₃N₄. In this composite, Bi₂O₃ serves as the redox‐active component for reversible chloride storage through the conversion between Bi/Bi₂O₃ and BiOCl, whereas the g‐C₃N₄ framework contributes to structural stabilization, suppresses particle aggregation, and modulates interfacial chloride adsorption behavior. More broadly, this work demonstrates that combining redox‐active chloride‐storage phases with stabilizing host frameworks is an effective strategy for designing advanced CDI electrodes. This synergistic composite design may also provide useful guidance for the development of other Faradaic desalination or ion‐storage systems that require a balance among capacity, reversibility, and structural robustness.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: asia70819‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.