Selective adsorption of high ionization potential value organic pollutants in wastewater

Significance

The metal-pyridine-N material facilitates selective adsorption of high ionization potential (IP) organic pollutants by establishing an efficient electron transfer pathway. Through metal bridging, electrons are transferred from carbon to Pyridine-N, then to the metal, attracting electron-deficient pollutants. This design enables prolonged and efficient adsorption and degradation of targeted pollutants, overcoming catalytic challenges in heterogeneous Fenton-like reactions.

Abstract

It is imperative to devise effective removal strategies for high ionization potential (IP) organic pollutants in wastewater as their reduced electron-donating capacity challenges the efficiency of advanced oxidation systems in degradation. Against this backdrop, leveraging the metal-based carbon material structure meticulously, we employed metal-pyridine-N (M-N-C, M=Fe, Co, and Ni) as the electron transfer bridge. This distinctive design facilitated the ordered transfer of electrons from the adsorbent surface to the surface of high IP value pollutants, acting as a “supplement” to compensate for their deficient electron-donating capability, thereby culminating in the selective adsorption of these pollutants. Furthermore, this adsorbent also demonstrated effective removal of trace emerging contaminants (2 mg/L), displayed robust resistance to various salts, exhibited reusability, and maintained stability. These findings carry substantial implications for future carbon-based material design, offering a pathway toward exceptional adsorption performance in treating water pollution.

The treatment of organic wastewater constitutes a pivotal concern within the realms of environmental conservation and sustainable development. Notably, the wastewater contains a substantial concentration of high IP (IP ≥ 9.0) organic pollutants (1–3). These pollutants require a considerable amount of energy for ionization, thus possessing chemical stability and antioxidative properties, posing a challenge to traditional degradation methods (4, 5). Presently, the primary treatment methods for organic wastewater involve chemical processes such as photocatalysis (6, 7), Fenton reaction, and ozone oxidation (8, 9). However, the short lifetime and limited migration range of the generated free radicals, compounded by the surface characteristics inherent to high IP value pollutants, impede their adsorption onto the catalyst surface, leading to unsatisfactory degradation efficiency of such pollutants. The substituent groups in organic compounds typically exert a significant influence on their ionization potential: Electron-acceptor groups such as hydroxyl and amide groups diminish the IP of organic pollutants, while electron-withdrawing groups like carboxyl and nitro have the opposite effect (10–13). Hence, in theory, delivering external electrons to high IP value pollutants holds promise in destabilizing their stability. The adsorption method could serve as a direct and effective strategy for eliminating high IP value contaminants from water, employing adsorbents endowed with expansive surface areas and hydrophilic functionalities, notably carboxyl groups [such as biochar (14–16), activated carbon (17), clay (18, 19), and microplastics (20)]. They operate through physical forces such as electrostatic interactions, π–π conjugation, and hydrogen bonding to adsorb pollutants. However, these adsorbents have often been inadequately designed concerning their electron-donating mechanisms, leading to suboptimal performance in treating wastewater containing high IP pollutants. Therefore, in the design of adsorbents, it is imperative to thoroughly consider the electron-donating capacity of their surfaces and explore methodologies for electron transfer to pollutants, thereby achieving highly selective and efficient chemical adsorption (13, 21–24).

In this study, utilizing a systematic stepwise synthesis approach, we have achieved the successful fabrication of a carbon material adsorbent denoted as M-N-C (M=Fe, Co, and Ni). Remarkably, the adsorbent demonstrates exceptional selectivity and high-efficiency adsorption toward pollutants with high IP. Taking the Fe-N-C adsorbent as a representative, our investigation unveils that through precise modulation of the coordination mode between the adsorbent carrier and the metal, coupled with the control of the metal’s specific form, iron (Fe) assumes a pivotal role as a “bridge.” This orchestration facilitates the efficient transfer of electrons from the adsorbent to the surface of pollutants, ensuring a stable chemical adsorption process (13, 25–28). Experimental analysis and density functional theory (DFT) calculations elucidated the significant influence of the carbon support structure on the electron transfer mechanism of the Fe bridge. By regulating the active sites of the adsorbent, we can adjust the distribution of surface charges, thereby enhancing the chemical interaction between the adsorbent and pollutants. This strategic approach aims to streamline the efficient pretreatment of specific resilient waterborne contaminants through adsorption, establishing a robust foundation for the implementation of advanced pretreatment strategies.

Results

Synthesis of Fe-N-C Adsorption Materials.

An iron-based carbon material adsorbent (Fe-N-C) was prepared using a grind-assisted pyrolysis method (Fig. 1A). The mixture gradually transformed from a yellow solid to a viscous black paste (SI Appendix, Fig. S1). FeCl₃·6H₂O, when combined with dopamine hydrochloride (DA), acted as the precursor with intrinsic moisture adsorption and oxidizing capabilities, inducing rapid polymerization of DA to obtain polydopamine (PDA). The abundant catechol moieties in PDA were well coordinated with the Fe ions. Subsequently, calcination pyrolysis was performed under an inert atmosphere to obtain Fe-N-C adsorption materials.

Fig. 1.

(A) Schematic diagram of the Fe-N-C preparation process. (B) XRD characterization of Fe-N-C obtained at different pyrolysis temperatures. (C) SEM (i), TEM (ii), and HR-TEM (iii) image of 800-Fe-N-C₂, (iv), Elemental mapping for 800-Fe-N-C₂. (D) Raman characterization of Fe-N-C at different pyrolysis temperatures. (E) S_BET of Fe-N-C at different pyrolysis temperatures.

The crystal structures of Fe-N-C at different pyrolysis temperatures were characterized by XRD (Fig. 1B). At relatively low temperatures (≤700 °C), the diffraction peaks observed at 30°, 35°, 43°, 56°, and 62° correspond to Fe₃O₄ (JCPDS PDF#85-1436), indicating that the main form of iron present is magnetite. When the pyrolysis temperature was further increased, the diffraction peaks were observed at 44° and 65°, corresponding to Fe⁰ (JCPDS PDF# 87-0721). This suggests that at a relatively high pyrolysis temperature, the amorphous carbon facilitates the reduction of FeCl₃, leading to the conversion of Fe³⁺ to Fe²⁺ and ultimately to Fe⁰ (Eqs. 1–3) (29, 30). The SEM images of Fe-N-C obtained at different pyrolysis temperatures are shown in Fig. 1C and SI Appendix, Fig. S2. Fe-N-C obtained by 600 °C exhibited a blocky structure, while higher pyrolysis temperatures resulted in a fragmented folded lamellar structure with rough surfaces. These stacked lamellar structures facilitate the retention of contaminants, enhancing adsorption performance (31). Additionally, TEM and HRTEM were employed to characterize 800-Fe-N-C₂, unveiling finely crystallized Fe nanoparticles embedded within carbon nanosheets, exhibiting a lattice fringe of 0.227 nm, corresponding to XRD findings. Mapping analysis further validated the uniform dispersion of Fe species across the adsorbent surface (Fig. 1C).

$$3{\mathrm{Fe}}_2{\mathrm{O}}_3+\mathrm{C}\triangleq 2{\mathrm{Fe}}_3{\mathrm{O}}_4+\mathrm{CO}\uparrow ,$$ [1]

$$2{\mathrm{Fe}}_2{\mathrm{O}}_3+3\mathrm{C}\triangleq 4\mathrm{Fe}+3{\mathrm{CO}}_2\uparrow ,$$ [2]

$${\mathrm{Fe}}_3{\mathrm{O}}_4+2\mathrm{C}\triangleq 3\mathrm{Fe}+2{\mathrm{CO}}_2\uparrow .$$ [3]

The functionality of adsorbent carriers and the diversity of functional groups play a crucial role in the adsorption performance of adsorbents, particularly for carbon-based materials. The degree of graphitization in these materials and the introduction of functional groups in the adsorbent can significantly impact electron transfer capability, thereby influencing the adsorption performance of the adsorbent. Hence, we employed Raman spectroscopy to investigate the graphitization degree of adsorbents annealed at different temperatures (13, 32). As observed in Fig. 1D, with the increase in pyrolysis temperature from 400 °C to 800 °C, the I_D/I_G ratio escalated from 0.75 to 0.91, indicating an elevated degree of carbonization within the Fe-N-C material, consistently favoring graphitic structures. Concurrently, an augmentation in iron content led to an increased I_D/I_G value in the Fe-N-C adsorbent, implying that the inclusion of iron also facilitates carbon material carbonization (SI Appendix, Fig. S3), thereby promoting the formation of graphitic structures, enhancing electron transfer capability. Simultaneously, from FT-IR analysis, we observed abundant functional groups across all adsorbents: hydroxyl or carboxyl (–OH) (3,500 cm⁻¹), carbonyl C═O (1,650 cm⁻¹), and aromatic C═C (1,600 cm⁻¹). The stretching vibration of the Fe═O bond at 550 cm⁻¹ confirms successful iron integration, with an increase in ─OH and C═O peak intensities with rising Fe content (SI Appendix, Fig. S4). This illustrates that the formation of magnetic iron nanoparticles not only alters the surface properties and structure of carbon materials but also fosters the formation of organic functional groups, thereby providing additional chemisorption sites (25, 33).

Furthermore, with the increase of pyrolysis temperature, the reaction between carbon and iron oxides, along with the evolution of CO and CO₂, resulted in a remarkable increase in the specific surface area (S_BET) of Fe-N-C. As depicted in Fig. 1E, when the pyrolysis temperature was increased from 400 to 700 °C, the S_BET of Fe-N-C increased significantly from 15.50 m²/g to 434.77 m²/g. This enhancement occurred because the organic components in the polydopamine (PDA) underwent vigorous evaporation with the rising temperature, leading to the formation of rich lamellar structures. However, with further increase in the pyrolysis temperature, there was a slight decrease in the specific surface area of the Fe-N-C material. This indicates that at higher temperatures, there may be a risk of the adsorbent framework collapsing, resulting in a reduction in the specific surface area of the adsorbent, which is detrimental to its adsorption capability. Hence, in synthesizing the adsorbent, maintaining an optimal temperature is crucial—it enhances graphitization without risking framework collapse.

Selective Removal of High IP Value Pollutants Based on M-N-C.

Under optimal reaction conditions (SI Appendix, Figs. S5 and S6), the adsorption performance of M-N-C adsorbents was investigated for a range of 10 organic pollutants with IP values ranging from 6.8 to 10. These pollutants included Rhodamine B (RhB), Bisphenol (BPA), phenol (PN), Aniline (AN), Quinoline, sulfamethoxazole (SMX), p-Chlorophenol (4-CP), nitrobenzene (NB), benzoic acid (BA), and p-nitrophenol (4-NP), each with their respective molecular formulas and IP values as shown in the Fig. 2A. The adsorption effects and adsorption rates (k_obs values) of these 10 pollutants in different M-N-C (Fe, Co, Ni) systems were classified, revealing distinct adsorption capacities for the M-N-C system when exposed to various organic pollutants with different IP values (Fig. 2B and SI Appendix, S7–S9). Taking the Fe-N-C system as an example, pollutants with electron-donating groups such as hydroxyl and amine (e.g., BPA, PN, AN) showed low adsorption efficiency in the Fe-N-C system (removal rate < 50%, k_obs < 0.02 min⁻¹). In contrast, pollutants with electron-withdrawing groups such as nitro groups (e.g., NB, BA) demonstrated high adsorption efficiency in the Fe-N-C system (removal rate > 70%, k_obs > 0.4 min⁻¹). Pollutants with both electron-donating and electron-withdrawing groups (SMX, 4-CP, 4-NP) exhibited intermediate adsorption efficiency. The Co-N-C and Ni-N-C systems demonstrated similar adsorption characteristics to the Fe-N-C system but with lower efficiency. According to the Hund's rule of electron distribution, the electron-donating abilities of Fe²⁺, Co²⁺, and Ni²⁺ follow the sequence: Fe²⁺ > Co²⁺ > Ni²⁺ (SI Appendix, Fig. S10). The differences in electron-donating abilities result in variances in the adsorption performance of Fe-N-C, Co-N-C, and Ni-N-C adsorbents toward different IP organic compounds. Overall, as the ionization potential of organic pollutants increased, the adsorption capacity of the M-N-C system also increased. The adsorption correlation between M-N-C and pollutants with varying IP values was extensively examined. Our findings revealed a pronounced correlation between the two factors (IP value and lnK_obs), as illustrated in Fig. 2C (R_Fe² = 0.8822, R_Co² = 0.8940, and R_Ni² = 0.8542). This once again substantiates the selective adsorption of M-N-C for pollutants characterized by high ionization potential values. The adsorption performance of NC material obtained by pure DA grinding and calcination on various pollutants is depicted in SI Appendix, Fig. S11. Metal-free NC exhibited removal rates of less than 10% for all pollutants tested, and the adsorption efficiency showed no direct correlation with the IP values of the pollutants. To further explore the role of the loaded metal in the adsorbent, we employed strong acid to remove Fe from the Fe-N-C adsorbent, reducing the electron-donating capacity of the adsorbent. Subsequently, we assessed its adsorption performance on pollutants with different IP. As shown in SI Appendix, Fig. S12, the experimental results indicate that after the adsorbent lost its iron content, the electron bridge between the adsorbent and pollutants disappeared. This led to a varying degree of reduction in the adsorption efficiency for pollutants with different ionization potentials, and the selective adsorption capability for high IP pollutants was also lost. This experimental outcome emphasizes the indispensable role played by iron in the adsorption process. Interestingly, Fe-N-C exhibited intriguingly high specificity in adsorbing BPA, with its adsorption performance significantly improved after acid treatment. Given the molecular structure of BPA, which includes two hydroxyl groups, the role of hydrogen bonding necessitates exploration (34). Upon introducing NH₄F into the Fe-N-C/BPA adsorption system to disrupt the hydrogen bonds between the adsorbent and BPA, the adsorption efficiency of Fe-N-C for BPA decreased from 66% to 30.21%, emphasizing the importance of hydrogen bonding in BPA adsorption (SI Appendix, Fig. S13). Following acid treatment, there is a notable increase in the surface hydroxyl and carbonyl groups of the adsorbent, leading to a discernible enhancement in its adsorption performance for BPA. Under identical conditions, the adsorption capabilities of the traditional carbon material, graphene oxide (GO), were further assessed for different IP pollutants (SI Appendix, Fig. S14). Despite GO having a large specific surface area and containing numerous hydrophilic groups (such as carboxyl and amino groups) capable of physical or hydrogen bond interactions with pollutants, its binding forces are ultimately limited. This outcome emphasizes the challenges faced by conventional carbon adsorbents in efficiently capturing and selectively targeting pollutants with elevated ionization potential values. The divergent adsorption performance observed between the adsorbent and GO underscores the superior efficiency and selectivity of the M-N-C adsorbent, laying the foundation for a more effective and precise removal of pollutants.

Fig. 2.

(A) Pollutants with different IP values and their molecular formulas. (B) The adsorption of organic pollutants with different IP by M-N-C ([adsorbents] = 0.15 g/L, [organic pollutants]₀ = 10 mg/L, reaction time = 1 h). (C) The correlation between LnK_obs (Fe-N-C, Co-N-C, Ni-N-C) and IP values. (D) Langmuir and Freundlich isotherm models fitting for SMX adsorption of Fe-N-C. (E) The adsorption properties of the adsorbent and the variation of zeta potential with pH ([SMX]₀ = 10 mg/L, pH_initial (pH_i) = 5.0, T = 35 °C, 800-Fe-N-C₂ dose = 0.15 g/L). (F) Adsorption of SMX with NH₄F, KSCN, and 1,10-phenanthroline. ([SMX]₀ = 10 mg/L, [Fe-N-C]₀ = 0.15 g/L, [NH₄F, KSCN, 1,10-phenanthroline]₀ = 10 mM, pH_i = 5.0, T=35 °C).

Investigation of the Adsorption Process.

Numerous studies indicate that an IP of 9.0 eV serves as a delineation for assessing the degradability of pollutants (35). Hence, we utilized SMX as a model pollutant closely associated with this criterion, with Fe-N-C serving as the adsorbent, to investigate the adsorption process occurring within the system. As shown in SI Appendix, Fig. S5C, there are two stages in the adsorption process: rapid adsorption and adsorption equilibrium. Rapid adsorption occurred within the initial 30 min, with removal efficiencies of SMX ranging from 12.18% to 71.77% for adsorbents produced at different temperatures, respectively. Subsequently, due to adsorption saturation effects, an equilibrium state was gradually reached within 60 min (36). Notably, 800-Fe-N-C₂ exhibited superior adsorption performance, being 4.85, 2.93, 1.75, and 2.18 times more effective than 400-Fe-N-C₂, 600-Fe-N-C₂, 700-Fe-N-C₂, and 900-Fe-N-C₂, respectively. Intriguingly, 700-Fe-N-C₂ with a larger S_BET displayed inferior SMX adsorption compared to 800-Fe-N-C₂, potentially due to the higher degree of graphitization in 800-Fe-N-C₂, indicating that chemisorption plays a more significant role than physisorption.

The kinetic model fitting can provide a better description of the migration process of SMX from the bulk solution to the adsorbent surface and offer insights into the adsorption mechanism. Therefore, three kinetic models, pseudo-first-order (PFO), pseudo-second-order (PSO), and Elovich, were used to fit the kinetic data. The obtained kinetic models and all parameters calculated from the three kinetic models are given in SI Appendix, Figs. S15–S17 and Table S1. Both the PFO (R²=0.9728 to 0.9965) and PSO models (R² = 0.9796 to 0.9981) effectively fit the experimental kinetic data, with the PFO rate constant (k₁) and the PSO rate constant (k₂) both significantly below 1, indicating a rapid adsorption process. The q_e,_cal. calculated by the PFO is closer to the predetermined values of q_e,exp. obtained from our experiments, while the PSO model has goodness of fit (R²), which points to the fact that chemisorption and physical adsorption complement each other and work together in the adsorption of SMX to Fe-N-C₂. Moreover, chemisorption is the key to the adsorption of SMX by Fe-N-C₂, while physisorption determines the adsorption capacity of Fe-N-C₂, aligning with the earlier discussion (37). The Elovich kinetic model also shows good fitting results (R² = 0.9541 to 0.9992), indicating that the adsorption process is a nonuniform diffusion process with chemisorption as the dominant process (38).

The equilibrium adsorption of the adsorbent increased with the rise in the initial SMX concentration (SI Appendix, Fig. S18). To gain a comprehensive understanding of the adsorption mechanism between Fe-N-C and SMX, different isotherm models were utilized to fit the experimental results, with 800-Fe-N-C₂ as a representative. The fitting results and parameters are shown in Fig. 2D and SI Appendix, Table S2. Evidently, the experimental results were better fitted by the Freundlich model (R² = 0.9608) than the Langmuir model (R² = 0.8441), suggesting that the adsorption between SMX and 800-Fe-N-C₂ occurs as a multilayer adsorption on a nonhomogeneous surface. Thus, the physical adsorption is indispensable in this system. Furthermore, the maximum adsorption capacity calculated by the Langmuir model served to compare the adsorption capacity among different adsorbents. The q_m showed the maximum adsorption capacity was 275.48 mg/g. As shown in SI Appendix, Fig. S19 and Table S3, we compared our results with previously reported works and observed that although the adsorbent may not have the highest maximum adsorption capacity, it exhibits optimal performance when considering the maximum adsorption capacity per unit specific surface area (q_m/S_BET). This suggests that chemisorption plays a vital role in this adsorbent.

Adsorption Mechanism Investigation.

The electrostatic interaction between adsorbents and organic pollutants significantly influences adsorption performance, with pH playing a crucial role in determining the surface charge of adsorbents and the form of the existing organic compounds. For the amphoteric molecule SMX, its primary forms in solution are SMX⁺, SMX⁰, and SMX⁻, corresponding to pH < 1.8, 1.8 < pH < 5.6, and pH > 5.6, respectively, as shown in SI Appendix, Fig. S20 and Table S4. To confirm the impact of electrostatic interactions on adsorption performance, the adsorption properties of the adsorbent and the variation of zeta potential with pH are illustrated in Fig. 2E. The point of zero charge (pH_zpc) of Fe-N-C falls between 6 and 7. When pH < pH_zpc, the surface of Fe-N-C carries a positive charge, and conversely, it becomes negatively charged. The optimal adsorption performance was reached at pH = 5. At this point, SMX primarily exists as SMX⁰, and the adsorbent surface is predominantly positively charged, so the electrostatic interaction between SMX and Fe-N-C is not the primary driving force for adsorption. Furthermore, despite RhB and Tartrazine possessing opposite charges, the adsorption efficiency of the adsorbent exhibited no significant disparities, standing at 9.50% and 19.02%, respectively (SI Appendix, Fig. S21). This further indicates that electrostatic effects are not the primary driving force in this adsorption system. Instead, the excellent adsorption performance of SMX is mainly attributed to chemical forces.

Analysis of the above results shows that physisorption and chemisorption collaborate with each other and occur simultaneously in the adsorption of SMX by Fe-N-C. Increasing the pyrolysis temperature and FeCl₃ content improves the S_BET of the carbon substrate and promotes physical adsorption. In SMX, the sulfonamide group (-SO₂NH-) with its strong electron-absorbing effect and the strong electronegativity of O make the aromatic ring electron deficient, turning it into a π-electron acceptor. The Fe-N-C has the most graphitized surface and acts as a π-electron donor, enabling the formation of π–π interactions with SMX (39). Given its capacity to disrupt π–π bonds, exploring the impact of ultrasonic treatment on both adsorption and desorption processes is essential. In comparison to the adsorption process under stirring conditions, ultrasonic treatment reduces the efficiency of SMX adsorption by Fe-N-C from 87.64% to 81.88%. Furthermore, desorption experiments conducted on adsorbents saturated with SMX using ultrasonic treatment resulted in a desorption rate of only 6.41% (SI Appendix, Fig. S22). These experimental findings suggest that π–π conjugation does not exert a predominant influence in this adsorption process. In addition, FT-IR analysis revealed the presence of a large number of oxygen functional groups on Fe-N-C, which can form hydrogen bonding with the -NH₂ and -NH- groups on SMX. To further verify the generation of hydrogen bonding between the adsorbent and SMX, we added a certain amount of NH₄F to the adsorption system and found that the adsorption effect was slightly weakened (Fig. 2F), Indicating minimal presence of hydrogen bonding within the system. Furthermore, upon comparing the XPS of the adsorbent before and after the reaction (SI Appendix, Figs. S23 and S24 and Table S5), it can be observed that the content of oxygen-rich functional groups C–O and O–C=O decreased while the content of C–O increased after the reaction. This suggests that the oxygen-rich functional groups participated in the adsorption through hydrogen bonding. Additionally, the oxygen content of the reacted Fe-N-C was elevated from 7.85% to 19.8%, which may be caused by the oxidation of Fe or adsorption of SMX (SI Appendix, Fig. S24).

To further investigate the role of iron species on the SMX adsorption performance, we conducted experiments to study the effect of adding KSCN as well as 1,10-phenanthroline to the adsorption system (Fig. 2F). SCN⁻ interacted with the Fe³⁺ site, leading to a decrease in adsorption efficiency to 70.29%. On the other hand, 1,10-phenanthroline could interact with the Fe²⁺ site, which completely inhibited the adsorption of SMX. This highlights the crucial involvement of the iron state, particularly Fe²⁺. At the same time, the XPS spectra of Fe 2p in Fe-N-C before and after the reaction were compared (Fig. 3A and SI Appendix, Table S5). The peaks of Fe 2p could be divided into Fe⁰ (708.37 eV), Fe²⁺ (711.22 eV), and Fe³⁺ (713.06 eV). Following adsorption, the distinctive peak corresponding to Fe⁰ vanished, accompanied by a reduction in Fe²⁺ content from 56.75% to 51.44%. Concurrently, the Fe³⁺ content increased from 36.12% to 48.56%, implying electron transfer occurred during the adsorption process. In our previous research, we found that pyridine N is more prone to donating electrons compared to pyrrole N, making it an important active site (40). To better elucidate the N sites, the N 1s XPS results were compared before and after adsorption. As shown in the Fig. 3B, the N 1s signal of Fe-N-C can be deconvoluted into three peaks at 398.67 eV, 400.1 eV, and 401.2 eV, assigned to pyridine-N, pyrrole-N, and graphite-N, respectively (Fig. 3C). Following the reaction, the content of pyridinic-N significantly decreased (from 40.89% to 25.77%), while the content of pyrrolic-N increased substantially (from 35.09% to 50.67%). This suggests that during the adsorption reaction, pyridine nitrogen participates in the electron-donation process as an adsorption site. For the adsorption systems of Co-N-C and Ni-N-C, a similar trend in changes was observed. After the reaction, there was an increase in the content of the high-valent state of the metal and a decrease in the pyridine-N content (SI Appendix, Figs. S25 and S26). This suggests a commonality in the M-N-C adsorption systems. Based on these findings, we can surmise that the pyridinic-N, in cooperation with the Fe²⁺ site on Fe-N-C, is essential for high IP pollutants adsorption. The Fe²⁺ loaded on pyridine N acts as an electron donor and interacts with the strong electron-withdrawing groups in high IP pollutants to promote the adsorption of high IP pollutants by the adsorbent. This electron transfer mechanism enhances the adsorption performance of M-N-C for high IP pollutants.

Fig. 3.

Comparison of the XPS spectra of (A) Fe 2p; (B) N 1s for the Fe-N-C before and after the reaction. (C) Different coordination modes of Fe cluster and N. The adsorption energies of (D) aniline and (E) nitrobenzene on different Fe/N sites in Fe-N-C. (F) The differential charge of the adsorption of aniline and nitrobenzene on Fe²⁺-pyridine N sites (The yellow regions indicate electron accumulation, while the blue regions indicate electron depletion. Δρ is the difference between bader charge of pollutants and valence electron count).

To gain further insights into the role of different N coordination methods and iron sites, we conducted density functional theory (DFT) calculations to elucidate the adsorption mechanism of organic compounds by the Fe-N-C adsorbent. To model the adsorbent, we anchored pristine iron clusters to a monolayer of carbon material. Interestingly, when the pristine iron oxide clusters were anchored to graphite-N, the adsorption energy was positive, indicating that composites from this configuration would be difficult to maintain stability. In contrast, the adsorption energies for the pyridine-N and pyrrole-N configurations were negative, suggesting the possibility of both anchoring sites. Thus, in the subsequent discussion, we focused on the adsorption of contaminants using the pyridine and pyrrole configurations (Fig. 3C and SI Appendix, Fig. S27 and Table S6). To investigate the adsorption sites in the Fe-N-C adsorbent that favor the adsorption of pollutants, we selected two model organic pollutants (aniline and nitrobenzene) with different functional groups (electron-donating group -NH₂ vs. electron-withdrawing group -NO₂). Their respective adsorption energies on different Fe/N sites in the Fe-N-C were calculated (Fig. 3 D and E and SI Appendix, Table S7). Notably, compared to Fe³⁺, Fe²⁺ exhibited more negative adsorption energies (ΔE_ad is more negative) for both aniline and nitrobenzene, regardless of whether it was anchored on the pyridine-N site or pyrrole-N site, indicating that Fe²⁺ loading is more favorable for the adsorption of organic pollutants. Furthermore, a comparison of the adsorption energies of Fe²⁺ loaded on pyridine-N and pyrrole-N sites for the two pollutants revealed that the adsorption energy was lower on the pyridine-N site. Specifically, nitrobenzene on Fe²⁺-pyridine N (ΔE_ad = −1.3 eV) exhibited a more negative value than that of aniline (ΔE_ad = −0.65 eV). This difference may be attributed to variations in electron transfer between Fe²⁺-pyridine N and different functional groups. To further support this hypothesis, we simulated the change in charge distribution of the system upon the adsorption of aniline and nitrobenzene on Fe²⁺-pyridine N sites, as depicted in Fig. 3F and SI Appendix, Fig. S28 and Table S8. The yellow regions indicate electron accumulation, while the blue regions indicate electron depletion. From this observation, it became evident that electron transfer relies on iron, thereby reconfirming the crucial function of the iron bridge. This indirectly underscored the reason why the adsorbent lost its ability to selectively adsorb pollutants with high IP values after the removal of iron. When the nitrogen atom (N) on the benzene ring binds with Fe, the decrease in bader charge (Δρ_Fe = −0.12 e) suggested electron transfer from a single aniline molecule to Fe²⁺-pyridine N. Conversely, it was identified that when the oxygen atom in the nitrobenzene molecule bonded with the iron atom (Δρ_Fe = 0.35 e), the nitrobenzene molecule exhibited electron acceptance. Since Fe²⁺-pyridine N more readily donates electrons, this observation underscored the propensity of electron-rich Fe²⁺-pyridine N to facilitate electron transfer to molecules bearing electron-deficient functional groups, such as nitrobenzene. Similar to the charge transfer observed during the adsorption of AN or NB on Fe-N-C, AN demonstrated a negative electron transfer density upon interaction with Co-N-C or Ni-N-C (Δρ_Co = −0.1 e, Δρ_Ni = −0.09 e), whereas NB exhibited a positive electron transfer density with Co-N-C or Ni-N-C (Δρ_Co = 0.34 e, Δρ_Ni = 0.29 e) (SI Appendix, Fig. S29 and Table S9). This demonstrated that pollutants with low IP value readily undergo electron loss at the M²⁺-pyridine-N sites, whereas those with high IP value tend to gain electrons more readily. Moreover, the electron transfer density sequence among the three metal center adsorbents and NB was Δρ_Fe > Δρ_Co > Δρ_Ni. This indicated that Fe exhibits a greater propensity to donate electrons compared to Co and Ni, aligning with the experimental findings of Fe-N-C, Co-N-C, and Ni-N-C adsorbents for organic compounds with different IP. Through dual validation via experiments and DFT calculations, it was confirmed that the selective adsorption of high IP value pollutants by the M-N-C system was attributed to the electron transfer between M²⁺-pyridine N sites and functional groups of pollutants (Fig. 4A).

Fig. 4.

(A) The role of electronic bridges in iron species. (B) Desorption with different solvents. (C) Adsorption efficiency of SMX with different ions and (D) different water quality ([SMX]₀ = 20 mg/L, [HA] = 100mg/L, [NH₄⁺, Cl⁻, SO₄²⁻, CO₃²⁻, PO₄³⁻] = 10 mM). (E) Adsorption efficiency of BTR and BPSIP ([BPSIP, BTR] = 2 mg/L, [adsorbents]₀ = 0.3 g/L, 50 mL 1% ethanol–water solution). (F) Removal of COD from actual wastewater by different adsorbents ([adsorbents]₀ = 1.2 g/L, 50 mL actual wastewater, T = 1 h). (G) Schematic illustration of the cross-flow filtration system and images of Fe-N-C/PVDF membrane. (H) NB removal performance of the Fe-N-C/PVDF system.

Stability and Practicability Assessment.

To assess the stability of the adsorbent, extensive testing was conducted on Fe-N-C, including desorption experiments, cyclic experiments, and long-term regeneration experiments. Desorption experiments were performed on saturated Fe-N-C using five different solvents (Fig. 4B). Ultrapure water, as a polar solvent, exhibited a low desorption efficiency of only 19.09%, indicating that saturated Fe-N-C can more stably persist in natural aquatic environments. To investigate the impact of solvent polarity, less polar organic solvents, namely methanol, ethanol, and acetonitrile, were selected for desorption. The desorption efficiency with these solvents ranged from 75.24% to 82.35%, surpassing that of ultrapure water. This improvement can be attributed to similar solubility effects between the solvents and pollutants. After five adsorption-desorption cycles, the adsorption efficiency of Fe-N-C for pollutants showed a slight decrease but remained above 70% (SI Appendix, Fig. S30). Moreover, the recycled Fe-N-C samples were characterized by SEM, XRD, and XPS. As illustrated in SI Appendix, Figs. S23, S31, and S32, SEM characterization revealed that the recycled sample retained its sheet-like structure. XRD analysis indicated a slight decrease in peak intensity, which contributed to the observed reduction in activity during the recycling experiment.

To simulate real operational conditions, long-term degradation of NB in the Fe-N-C system was performed by periodically adding NB (SI Appendix, Fig. S33). In the initial five cycles, the added NB was effectively removed. However, as the adsorption sites were continuously covered by pollutants, the continuously added NB became challenging to remove effectively, resulting in a halving of the removal rate by the eighth cycle. To re-expose the adsorption sites, peroxymonosulfate (PMS) was introduced into the reaction system, aiming to degrade NB on the surface of the adsorbent through the activation of Fe by PMS. Interestingly, upon the addition of PMS, the concentration of NB in the reaction solution increased momentarily. This was due to the preference for electron transfer of Fe to activate PMS, weakening the electron interaction between NB and the adsorbent, allowing the adsorbed nitrobenzene to return to the solution. Furthermore, while Fenton treatment could partially restore adsorption activity, it was challenging to maintain, as PMS tends to oxidize Fe²⁺ to Fe³⁺, making it difficult to provide electrons again (SI Appendix, Fig. S33A). Given that NaBH₄ has a strong reducing effect, it was reasonable to hypothesize that the addition of NaBH₄ could reduce Fe³⁺ back to Fe²⁺ to reactivate the adsorbent. Upon comparison of the XRD and XPS spectra of the adsorbent before and after the reduction of NaBH₄, it was revealed that the reduction process had no discernible impact on the original crystal structure of the adsorbent and could increase the electron-donating ability of the adsorbent. Specifically, the Fe²⁺ content increased from 51.06% to 56.22%, and the pyridine N content increased from 22% to 31% (SI Appendix, Figs. S34 and S35). Subsequent adsorption experiments demonstrated that following reduction, the adsorbent regained its activity, thereby providing further validation for the proposed adsorption mechanism (SI Appendix, Fig. S33B).

Considering the potential presence of various substances in real water environments that may influence the removal of pollutants, additional research was conducted to investigate the impact of ionic species (NH₄⁺, Cl⁻, SO₄²⁻, CO₃²⁻, and PO₄³⁻) and humic acid (HA) on adsorption. As shown in Fig. 4C, HA, NH₄⁺, Cl⁻, and SO₄²⁻ had a negligible impact on the adsorption performance of Fe-N-C. In contrast, high concentrations of CO₃²⁻ and PO₄³⁻ showed a distinct inhibition of pollutant adsorption. The addition of CO₃²⁻ led to an increase in pH to 11, causing strong electrostatic repulsion between the pollutant and the adsorbent. On the other hand, the hindrance of pollutant adsorption by PO₄³⁻ was attributed to the chemical interaction between PO₄³⁻ and the iron species present, which interfered with the adsorption process. This further underscores the importance of iron sites in facilitating the adsorption of pollutants. In the complex matrix of real water samples, Fe-N-C consistently exhibits excellent pollutant adsorption performance, and extending the reaction time further enhances the adsorption efficiency (Fig. 4D). These results underscore the resilience of Fe-N-C to interactions with various components present in real wastewater, indicating its broad potential for practical applications in wastewater treatment.

Applications in Wastewater Purification.

In aquatic ecosystems, organic pollutants are typically present in trace concentrations. As technology continues to advance, an increasing number of persistent neopollutants [e.g., diphenylamine (DPA), benzotriazole (BTR), 4-hydroxyphenyl-4-isopropoxydiphenylsulfone (BPSIP)] continue to emerge. These pollutants exhibit low solubility in water but high solubility in organic solvents like ethanol. However, the presence of ethanol in the reaction system significantly inhibits the generation of free radicals, such as hydroxyl radicals (•OH). This inhibition makes it challenging to achieve the effective degradation of these emerging pollutants in radical-dominated reaction systems. Consequently, it places even greater demands on wastewater treatment technologies. We selected two emerging contaminants, BTR and BPSIP, to assess the adsorption capacity of the adsorbent 800-Fe-N-C₂ for these trace emerging pollutants, as shown in Fig. 4E. It can be observed that the adsorbent can completely adsorb and remove BPSIP within 1 h, while the adsorption effect on BTR is almost negligible (only 13.93%). This difference is attributed to the molecular structure of BPSIP, which includes an electron-absorbing group (-SO₂R), facilitating electron transfer with the adsorbent. In contrast, in the case of BTR, the nitrogen atoms on the five-membered ring of the molecule have lone pairs of electrons, making it more likely to release electrons during the adsorption process. Consequently, this results in less significant electron transfer and reduced adsorption efficiency for BTR. Considering the concentrations of emerging pollutants in surface or drinking water, we adjusted the initial concentrations of BPSIP and BTR to the microgram level for the adsorption experiments. Due to the low pollutant concentrations, which did not reach the adsorption limits of the adsorbent, the removal rates of both pollutants exceeded 90% (SI Appendix, Fig. S36A). It is noteworthy that the presence of electron transfer interaction between Fe²⁺-pyridine N and the functional groups of BPSIP resulted in a higher adsorption rate constant for BPSIP compared to BTR (SI Appendix, Fig. S36B).This finding further underscores the selective adsorption capacity of the adsorbent.

Considering micro- and nanoplastics are widely recognized as emerging pollutants of significant concern in daily life, we specifically targeted polystyrene (PS) spheres of two sizes, 500 nm and 100 nm, and utilized fluorescence spectroscopy to investigate the adsorption process of nano-plastics by the adsorbent. As depicted in SI Appendix, Fig. S37, the adsorbent effectively removed PS spheres within 2 h. Subsequently, SEM characterization of the adsorbent postreaction revealed a notable accumulation of spherical structures on the surface, consistent with the morphology of PS spheres (SI Appendix, Fig. S38). These experimental results further underscore the extensive potential application of the adsorbent in water treatment.

As is commonly understood, the challenge of achieving in-depth treatment in the actual wastewater treatment process intensifies as the chemical oxygen demand (COD) continues to decrease. To investigate the in-depth treatment capabilities of various carbon material adsorbents for actual wastewater, we conducted a comparison, and the specific results are shown in Fig. 4F. After 4 h of adsorption treatment, the Fe-N-C system reduced the COD of actual wastewater from 185 mg/L to 20 mg/L, a level that traditional activated carbon materials could not achieve. Nevertheless, it is crucial to emphasize that the adsorbed pollutants necessitate further treatment. Previous studies have indicated that introducing oxidants, such as H₂O₂, PMS, etc., can initiate Fenton-like reactions, facilitating the degradation of pollutants on the adsorbent surface (27, 41). Consequently, we compared the experimental results of different homogeneous and heterogeneous Fenton systems for the advanced treatment of simulated wastewater (phenol). As shown in SI Appendix, Fig. S39, the Fe-N-C/PMS system efficiently removed 90% of phenol within 15 min, outperforming the conventional iron-containing system with the same molar amount. These experimental results demonstrate the promising potential of this adsorbent for the deep treatment of actual wastewater and hold significant implications for the future design of carbon-based materials with exceptional adsorption performance and in situ regeneration capability for water pollution treatment. More profoundly, we can deposit the Fe-N-C adsorbent onto commercially available PVDF membranes and evaluate its continuous pollutant removal performance using a circulating system (Fig. 4G and SI Appendix, Fig. S40). As the reaction time progresses, the circulating adsorption system can achieve over 90% removal efficiency of NB (Fig. 4H). Overall, the catalyst synthesized in this study demonstrates outstanding adaptability, durability, and oxidation capability in complex wastewater environments, solidifying its excellent potential for practical applications.

Conclusions

The metal-carbon materials synthesized in this study successfully established an electron transfer bridge within the M-N-C (M=Fe, Co, and Ni) framework. This accomplishment enables the preferential adsorption of pollutants characterized by high ionization potential values in polluted water, significantly enhancing the adsorption efficiency of carbon-based materials for pollutants. Experimental and theoretical simulations further validate the synergistic coordination effect of M²⁺-pyridine N, enhancing electron transfer from the adsorbent to pollutants and facilitating the efficient removal of trace emerging pollutants. This investigation presents a unique approach for the preliminary treatment of high IP pollutants in real wastewater, providing innovative methods for designing the interfacial processes in adsorption reactions and offering fresh strategies for the subsequent application of Fenton-like treatment technologies in environmental settings.

Materials and Methods

Chemicals.

All the chemical reagents which were used for catalyst syntheses and experiments evaluation in this work are listed in SI Appendix.

Preparation of Adsorption Materials.

Iron-based carbon material adsorbent (Fe-N-C) was prepared using a grind-assisted pyrolysis method. Dopamine hydrochloride (DA) and FeCl₃·6H₂O were mixed in a certain molar ratio, and the mixture was ground for 10 min under ambient air conditions until the solid powder transformed into a viscous liquid. Subsequently, the resulting mixture was calcined in a tube furnace under a nitrogen atmosphere with a heating rate of 2 °C min⁻¹ for 120 min to obtain the T-Fe-N-C_x adsorbent (where T represents the pyrolysis temperature, T = 400, 600, 700, 800, and 900 °C; x is the molar ratio of FeCl₃·6H₂O to DA, x = 1, 1.5, 2, and 2.5). The other M-N-C samples (including cobalt and nickel) were prepared by using the similar procedures except for different metal precursors.

Adsorption experiment.

Disperse 0.015 g of the adsorbent into 100 mL of a 20 mg/L SMX solution with an initial pH of 5.0 for the adsorption experiments. Samples were withdrawn at specific intervals, filtered through a 0.22 μm filter, and measured by high-performance liquid chromatography (HPLC) at a wavelength of 262 nm with no delay. Other organic pollutants [RhB, Bisphenol, phenol, Aniline, Quinoline, p-Chlorophenol, nitrobenzene, benzoic acid, p-nitrophenol, benzotriazole (BTR), and 4-hydroxyphenyl-4-isopropoxydiphenylsulfone (BPSIP)] were treated in a similar way, and their varying concentrations were identified by HPLC or UV-Vis spectrophotometer (SI Appendix, Table S10). All the batch experiments were repeated three times, and the graphs with error bars were presented.

DFT Calculation Details.

The specific calculation items were presented in SI Appendix.

Statistical Analysis.

All data were gained directly from the source experiment and processed using Origin 2021 and Excel 2021. All experiments were carried out in duplicate. Full details for DFT computations, control experiments and materials characterizations were introduced in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.