Understanding the role of transition metal single-atom electronic structure in oxysulfur radical-mediated oxidative degradation

Abstract

The ubiquity of refractory organic matter in aquatic environments necessitates innovative removal strategies. Sulfate radical-based advanced oxidation has emerged as an attractive solution, offering high selectivity, enduring efficacy, and anti-interference ability. Among many technologies, sulfite activation, leveraging its cost-effectiveness and lower toxicity compared to conventional persulfates, stands out. Yet, the activation process often relies on transition metals, suffering from low atom utilization. Here we introduce a series of single-atom catalysts (SACs) employing transition metals on g-C₃N₄ substrates, effectively activating sulfite for acetaminophen degradation. We highlight the superior performance of Fe/CN, which demonstrates a degradation rate constant significantly surpassing those of Ni/CN and Cu/CN. Our investigation into the electronic and spin polarization characteristics of these catalysts reveals their critical role in catalytic efficiency, with oxysulfur radical-mediated reactions predominating. Notably, under visible light, the catalytic activity is enhanced, attributed to an increased generation of oxysulfur radicals and a strengthened electron donation-back donation dynamic. The proximity of Fe/CN's d-band center to the Fermi level, alongside its high spin polarization, is shown to improve sulfite adsorption and reduce the HOMO-LUMO gap, thereby accelerating photo-assisted sulfite activation. This work advances the understanding of SACs in environmental applications and lays the groundwork for future water treatment technologies.

Graphical abstract

Highlights

• •The electronic structures of transition metals are key on sulfite activation.
• •The intense spin polarization of Fe leads to its strong activation of sulfite.
• •The rate constant for the Fe single-atom exceeds that of Ni and Cu.
• •Oxysulfur radical-triggered oxidative degradation mechanism was revealed.
• •Fe single-atom catalyst presents promising prospect for sulfite activation.

1. Introduction

Advanced oxidation processes (AOPs) based on SO₄^•− for water treatment have recently grown in popularity. Compared with ^•OH, SO₄^•− boasts a wider effective pH range, longer lifespan, better contaminant selectivity, and is less susceptible to matrix interference [1,2]. The most common method used to generate SO₄^•− for water treatment is via persulfate activation. However, disadvantages such as high energy input and leftover environmentally hazardous persulfate residuals constrain its further application [3]. Some studies have attempted to employ more cost-effective sulfite ($0.35 kg⁻¹ vs. $1.83 kg⁻¹), a common industrial by-product of desulfurization, fossil fuel extraction, and tannery lime process, to generate SO₄^•− [[4], [5], [6], [7]]. Making full use of sulfite aligns with the principle of green chemistry, which is significant for water treatment [8,9].

Transition metals (TM), such as Fe, Ni, and Cu, have exhibited favorable performance for sulfite activation, proceeding via the formation of a TM-sulfite complex and the release of SO₃^•− [10,11]. However, difficulty in recovery, risk of potential secondary contamination, and a narrow operating pH range limit the large-scale application [12]. To solve these issues, TM-based heterogeneous catalysts, mainly including zero-valent iron [13,14], TM oxide [12,15], TM sulfide [16], and spinel ferrite [17,18], have been developed for sulfite activation. Nevertheless, it remains a challenge to enhance the catalytic activity of the reaction, improve atomic economy, and reduce metal leaching.

Single-atom catalysts (SACs) with high atomic dispersion are considered an essential bridge between homogeneous and heterogeneous catalysis, which combine the advantages of an isolated active site of the former and the easy separation of the latter [19,20]. SACs have garnered attention for their high activity, maximum atomic utilization efficiencies, and modulable reaction selectivity [[21], [22], [23]]. Compared to gas-phase catalysis, less attention has been given to SACs for water treatment applications, especially in the activation of sulfite [24]. The materials boast high active site dispersity, which may significantly improve sulfite activation performance. However, high TM loadings are still required to obtain favorable catalytic performance, introducing greater catalyst cost and risk of secondary contamination. Therefore, environmentally friendly solar energy can be introduced to reduce the required metal loading while keeping high catalytic efficiency.

Photocatalytic processes generate electron-hole pairs and are considered environmentally friendly [25,26]. These photogenerated holes (h⁺) can oxidize sulfite to SO₃^•− and initiate a series of reactions to generate reactive oxygen species with strong oxidizing capacity in the presence of dissolved oxygen (Table S1, equations S1–8) [27]. Among various supports with visible light response, g-C₃N₄ is considered a favorable carrier for SACand can interact strongly with TMs due to the abundance of anchoring sites on its surface [20,28,29]. In addition, g-C₃N₄ is considered to have favorable application prospects due to its structural stability and low economic cost. Based on these advantages, g–C₃N₄–based SACs have been widely studied and remained a research hotspot over the past two years [[30], [31], [32], [33]]. Moreover, it is also considered a favorable photocatalyst with good visible light response and suitable band gap, making it effective in photocatalysis [34,35]. Hence, g-C₃N₄ was selected as SACsupport in this work.

The SACs could simplify the catalyst model greatly and improve understanding of catalyst mechanisms [36]. Although some studies have reported the application of g–C₃N₄–based SACs in the field of environment and energy [37,38], some critical issues remain to be explored in depth. In terms of water treatment application, high metal loading for g–C₃N₄–based SACs is typically required to achieve favorable catalytic performance, yet brings the inevitable problems such as increased catalyst preparation costs and potential risk of high metal ion leaching [39,40]. In terms of catalytic mechanisms, an in-depth investigation of the electronic structure properties of TMs in SACs is useful for elucidating catalytic mechanisms [41]. In addition, SACs of various TMs are compared and widely used in persulfate activation for water treatment [28,42,43]. However, the in-depth role of various TMs for oxysulfur radical formation remains unclear, and research on the performance of various TMs for sulfite activation is limited only to homogenous catalysis. It was concluded that the difference in sulfite activation performance by different TMs could not be explained simply by different redox potentials, and a more in-depth mechanism should be investigated further [44]. Therefore, it is significant to investigate the relationship between TM type and capacity for oxysulfur radical generation and the corresponding mechanism.

In this work, different SACs with various TMs anchored on g-C₃N₄ were developed to activate sulfite (denoted as TM/CN, where the TM is Fe, Ni, and Cu). To the authors’ knowledge, this work is the first to study the role of electronic structures and spin polarization of TM SACs and elaborate in-depth interaction mechanisms between different TM SACs and g-C₃N₄ for sulfite activation through both experiment and theoretical calculation, which could be significant and serve as a reference for future SACs design in the field of sulfite activation. The main objectives of this work are: (1) to differentiate the role of various SACs (Fe/CN, Ni/CN, and Cu/CN) on sulfite activation and contaminant degradation, (2) to clarify the key parameters of different TMs, including Bader charge, d-band centers, spin polarization to further elaborate the electron donation from TM to g-C₃N₄ and the close connection between the propensity for sulfite activation with electronic structure of TM based on DFT calculation and experimental results, (3) to reveal the enhanced oxysulfur radical-triggered mechanism and promoted electron donation-back donation process of sulfite activation by TM/CN under the visible light, and (4) to investigate the performance and application potential of Fe/CN on the sulfite activation under the influence of water matrix.

2. Method and materials

2.1. Preparation of TM single atom anchored on g-C₃N₄ catalysts

Chemicals and materials employed in this work are listed in Text S1. TM single atoms anchored on g-C₃N₄ catalysts were prepared as follows: 0.2 mmol of precursor (Fe(NO₃)₃·9H₂O, Ni(NO₃)₂·6H₂O, or Cu(NO₃)₂·3H₂O) was mixed with 5 g of melamine in 80 mL of deionized water under constant stirring at 80 °C. After drying, the sample was ground and calcined in a tube furnace at 550 °C for 3 h at a heating rate of 5 °C min⁻¹ under continuous N₂ flow. After cooling to room temperature, the prepared sample was ground to 150 mesh in a ball mill (Fig. 1a). The prepared sample was noted as TM/CN (where TM is Fe, Cu, and Ni). The pristine g-C₃N₄, denoted as CN, was prepared by the same procedure except for adding a metal salt precursor.

Fig. 1.

a–g, Preparation procedure of various TM/CN (a); XRD spectra (b); UV–vis spectra (c); [αhv]² vs. hv plots (d); Fe 2p XPS spectra (e); Ni 2p XPS spectra (f); and Cu 2p XPS spectra (g). h, XANES spectra of Fe/CN and reference samples. i, XANES spectra of Ni/CN and reference samples. j, XANES spectra of Cu/CN and reference samples.

2.2. Degradation experiments

The as-prepared catalyst of 0.02 g was added into a solution containing 5 mM NaHSO₃ and 10 mg L⁻¹ acetaminophen (ACT). Afterwards, the system was irradiated with a 300 W Xenon lamp (CEL-PE300L-3A, Beijing Education Au-light Co., Ltd.) equipped with a 400 nm cutoff filter. Identical degradation experiments were also performed under dark conditions for comparison. Samples were constantly stirred using a magnetic stir rod. The 1.5 mL sample was taken at certain time points and filtered through a 0.22 μm membrane. Error bars denote the standard errors obtained by conducting trials in triplicate.

2.3. Catalyst characterization and analysis

Material morphology was observed through scanning electron microscopy (SEM, Zeiss Ultra60, Germany), transmission electron microscopy (TEM, FEI Tecnai G20, USA), and high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM, EM-ARM300F). X-ray absorption spectroscopy (XAS) tests were carried out at beamline XAFCA of the Singapore Synchrotron Light Source through fluorescence mode. Zero field cooled (ZFC) and field cooled (FC) measurements were conducted to investigate the magnetic moment of various TM on a magnetic property measurement system (MPMS-3, Quantum Design) at an external field of 1 kOe from 2 to 400 K. Further experimental details of other characterization and analysis method can be found in Text S2.

2.4. Density functional theory calculations

Density functional theory (DFT) calculations employing Vienna Ab-initio simulation package (VASP) 5.4.4 were performed to investigate the mechanism of sulfite activation in different systems. Generalized gradient approximation Perdew-Burke-Ernzerhof (GGA-PBE) was employed as an exchange-correlation functional. Supercell models (2 × 2 × 1) of monolayers of Fe/CN, Ni/CN, and Cu/CN were built. The cut-off energy was set as 450 eV. The 3 × 3 × 1 k-point grid was applied in all calculations except for the use of 15 × 15 × 1 k-point grid in density of state (DOS) calculation. The effect of interlayer interactions between layers was eliminated by a 30 Å vacuum layer. The convergence criteria for energy and force during the structural optimization process was 1.0 × 10⁻⁵ eV atom⁻¹ and 0.01 eV Å⁻¹, respectively. The adsorption energy of sulfite on the catalyst was calculated by equation (1):

$$E_{\text{ads}}=E_{\text{substrate}+\text{molecule}}-E_{\text{substrate}}+E_{\text{molecule}}$$ (1)

where $E_{\text{substrate}+\text{molecule}}$, $E_{\text{substrate}}$, and $E_{\text{molecule}}$ represent the total energy of the adsorption system, substrate, and molecule, respectively.

The charge transfer between the catalyst and sulfite ($\Delta {\rho }_{\text{ads}}$) was explored through the deformation charge density based on equation (2):

$$\Delta {\rho }_{\text{ads}}={\rho }_{\text{substrate}+\text{molecule}}-{\rho }_{\text{substrate}}+{\rho }_{\text{molecule}}$$ (2)

where ${\rho }_{\text{substrate}+\text{molecule}}$, ${\rho }_{\text{substrate}}$, and ${\rho }_{\text{molecule}}$ represent the total charge density of the adsorption system, substrate, and molecule, respectively.

The d-band center was calculated based on equation (3):

$${\epsilon }_{\mathrm{d}}=\frac{{\int }_{-\infty }^{\infty }{n}_{\mathrm{d}}(\epsilon )\epsilon d\epsilon }{{\int }_{-\infty }^{\infty }{n}_{\mathrm{d}}(\epsilon )d\epsilon }$$ (3)

where ε and $n_{\mathrm{d}}$ (ε) represent the band position in the DOS diagram and the density of state, respectively.

3. Result and discussion

3.1. Catalyst characterizations of TM/CN

X-ray diffraction (XRD) test was conducted to further explore the crystal structure of various TM/CN (Fig. 1b). The diffraction peak at 27.2° (002) corresponded to interlayer stacking of conjugated aromatic structure, while the peak at 13.0° (100) was assigned to heptazine ring repeating structural units. The weakening and broadening of the (002) peak was observed in the TM/CN XRD pattern compared to that in CN, attributed to inhibited polymeric condensation caused by incorporating various metal ions [45,46]. Furthermore, no peak corresponding to TM oxide was observed, demonstrating that TM existed in an ionic form rather than as a compound. These results support the existence of interactions between TMs and the CN skeleton.

A lack of adsorption bands belonging to TM oxide in the Fourier transform infrared spectroscopy (FTIR) spectra further illustrates the existence of TMs in the ionic state (Fig. S1). The similar absorption spectra of CN and TM/CN could be ascribed to low metal doping content [47]. The sharp peak at 804 cm⁻¹ was ascribed to the characteristic absorption peak of the triazine ring [46]. The broad absorption band containing several peaks from 1240 to 1650 cm⁻¹ was assigned to the vibrational stretching of the aromatic structure, while the broad absorption peak around 3170 cm⁻¹ was attributed to the stretching vibration peaks of –NH₂/NH groups and their intermolecular hydrogen bonding interactions [48]. Overall, XRD and FTIR could fully characterize the structure of CN and point towards its successful synthesis.

Brunauer–Emmett–Teller (BET) testing was performed to investigate the effect of TM doping on the specific surface area of various TM/CN (Fig. S2). Compared to pure CN, the introduction of Ni, Fe, and Cu increased the BET-specific surface area from 9.429 to 14.333, 12.168, and 11.149 m² g⁻¹, respectively. These results were attributed to the role of TM in CN interlayer exfoliation [49]. Overall, the enhanced specific surface area of TM/CN may increase the number of active sites and better facilitate the activation of sulfite.

The introduction of Fe led to a redshift in the absorption spectra compared to pure CN, broadening the useable wavelength range (Fig. 1c). However, the introduction of Ni and Cu did not significantly change the absorption edge despite changing the light absorption properties. The band gaps of various TM/CN were calculated via the Kubelka-Munk equations [50]. The introduction of Ni and Cu had a negligible effect on the band gap, while Fe reduced the band gap from 2.71 to 2.61 eV (Fig. 1d). Overall, the introduction of Fe enhanced the visible light response of CN and thus promotes better photocatalytic sulfite activation performance.

X-ray photoelectron spectroscopy (XPS) results showed elemental Fe, Ni, and Cu had similar loadings on TM/CN at a molar fraction of 0.21%, 0.20%, and 0.22%, respectively. XPS spectra of various TMs are employed to explore their chemical states (Fig. 1e–g). The 2p XPS spectra of all TMs could be divided into 2p_3/2, 2p_1/2, and satellite peaks, indicating the presence of Fe(II/III), Ni(II/III), and Cu(I/II) in their respective TM/CN [[51], [52], [53]], with faint signals being attributed to low metal content [54]. In addition, CN and various TM/CN had similar N and C 1s XPS spectra, attributed to the similar functional groups found on CN (Fig. S3). The N 1s spectra could be deconvoluted into three peaks, including C–N═C (398.66 eV), N–(C)₃ (400.13 eV), and NH_x amino groups (401.23 eV). The peaks belonging to C═C (284.80 eV), C–NH_x (286.14 eV), and N–C═N (288.07 eV) were observed in C 1s spectra. K-edge X-ray absorption near-edge structure (XANES) spectra were employed to investigate the chemical state of various TM further (Fig. 1h–j). The adsorption edge of Fe/CN was between that of iron (II) phthalocyanine (FePc) and Fe₂O₃, demonstrating the chemical valence state of Fe in Fe/CN to be between +2 and + 3 [23,55]. Similarly, the adsorption edge of Ni/CN was slightly higher than that of nickel (II) phthalocyanine (NiPc), indicating the valence state of Ni in Ni/CN was slightly higher than +2 [56], and the adsorption edge of Cu/CN slightly higher than that of Cu₂O, corresponding to a valence state of slightly higher than +1. Overall, XANES results were consistent with XPS.

TM/CN sheet layer structures of μm size were observed by SEM images (Fig. 2a–c). No significant difference in morphology between the different TM/CNs was observed. EDS mapping showed low loadings of elemental TM to be uniformly distributed across the CN sheet (Fig. S4). TM/CN exhibited a more porous structure than CN, and the absence of particles in the TEM image further proved TM non-agglomeration (Fig. S5). HAADF-STEM was performed to clarify the dispersion of atomic metal further. Several bright spots could be identified on the catalyst surface in Fig. 2d, indicating Fe was atomically well-dispersed other than in the existence form of nanoparticles on the surface of CN [57]. The Fourier-transformed extended X-ray adsorption fine structure (FT-EXAFS) spectrum of Fe/CN displayed an intense first shell peak at around 1.7 Å, which could be attributed to the Fe–N bond [28]. The absence of Fe–Fe or Fe–O peaks illustrates that Fe did not exist as Fe clusters or oxides in Fe/CN. EXAFS-fitting further indicates that Fe bonding in Fe/CN was largely composed of Fe–N coordination, where the coordination number of Fe was around 4. Details of the EXAFS-fitting can be found in Fig. S6 and Tables S2–S4. All the above results demonstrate that single atom Fe was well-dispersed in the Fe/CN. Similarly, the HAADF-STEM and EXAFS results for Ni and Cu (Fig. 2e and f, Fig. 2k and j, and Fig. S6) showed similar results for their respective TM/CN. In general, TM in the as-prepared TM/CN exhibited high atomic-level dispersion on CN support, demonstrating the successful preparation of TM/CN SACs.

Fig. 2.

a–c, SEM image of Fe/CN (a); Ni/CN (b); and Cu/CN (c). d–f, HAADF-STEM image of Fe/CN (d); Ni/CN (e); and Cu/CN (f). g–i, FT-EXAFS spectra of Fe/CN (g), Ni/CN (h), Cu/CN, and relevant reference sample (i). j–l, Corresponding EXAFS R-space fitting curve of the Fe/CN (j); Ni/CN (k); Cu/CN and relevant reference sample (l).

3.2. Role of TM/CN electronic structure on sulfite activation

CN and various TM/CN performance for the degradation of various systems was investigated via DFT calculation (Fig. 3a–c, and e) and experiments (Fig. 3b–d). It was noted that the ACT experienced negligible adsorption to all CN and TM/CN under dark conditions (Fig. S7). Under dark conditions, sulfites without a catalyst aid showed slight activity towards ACT degradation due to their mild oxidation capacity [5]. The lack of enhancement in degradation efficiency following the introduction of CN proved CN's inability to active sulfite under dark conditions. Compared with CN, the ACT degradation performance of Cu/CN was significantly enhanced, showing a degradation efficiency of 20.2% within 60 min in the presence of sulfite in the dark, indicating the importance of the Cu²⁺/Cu⁺ redox cycle in sulfite activation. Furthermore, the ACT degradation efficiency of Ni/CN and Fe/CN reached 35.3% and 50.2% within 60 min in the presence of sulfite under dark conditions, respectively. The corresponding reaction rate constant (k) of various catalytic systems was calculated using a pseudo-first-order decay kinetic model [50]. The k values of Fe/CN, Ni/CN, and Cu/CN in the presence of sulfite were 0.013, 0.008, and 0.004 min⁻¹, respectively (Fig. 3b). It was noted various TM/CN exhibited significant differences in the performance of sulfite activation.

Fig. 3.

a, Sulfite adsorption on various TM/CN and corresponding charge density deformation. Brown, green, pink, blue, grey, yellow, red, and white spheres represent Fe, Ni, Cu, N, C, S, O, and H atoms, respectively. Isosurface level is set as 3.0 × 10⁻³ e Bohr⁻³. Yellow represents the accumulation of charge, and blue represents the depletion of charge. b, k_obs for ACT degradation by various TM/CN in the presence of sulfite. c, PDOS of the various TM 3d orbitals. d, Temperature dependence inverse susceptibilities. e, HOMO-LUMO gap of sulfite before and after adsorption on TM. f, Relationship between k_obs for ACT degradation and E_ads, Q, and HUMO-LUMO gap. g, Mechanism of various TM/CN on sulfite activation.

Spin-polarized DFT calculations were conducted to probe the mechanism of sulfite activation by various single-atom TM/CN (TM = Fe, Cu, Ni). DFT models were constructed by placing a Fe, Cu, or Ni atom in the six-fold cavity structure of the single-layer CN, which was reported to be the most stable structure (Fig. S8) [48]. The first task in understanding the TM element-dependent sulfite activation is to clarify the electronic structure of TM/CN. Interestingly, charge-density-difference calculations evidenced the electron donation from all three TM elements to the CN support (Fig. S9). Specifically, the net charge from Fe to CN was the greatest, with a value of 1.10 e, while those for Ni and Cu were 0.76 and 0.74 e, respectively, indicating the strongest electron transfer from Fe to CN. The net charge transferred from TM to CN substrate positively correlated with the sulfite activation performance of the TM/CN as the ease of metal-CN electron transfer greatly affects the redox cycle rate of various metals, which is crucial for the activation of sulfite. Since sulfite activation is an oxidation process with electrons donating from sulfite to TM/CN, electron-deficient active sites on catalysts are more favorable. In this regard, the single-atom TM sites are supposed to be responsible for the sulfite activation. This point was further supported by the activity test that pure CN without TM atoms showed negligible promotion effect on the sulfite activation.

To check the proposed sulfite activation on single-atom TM sites, adsorption configurations, and energies were first computed. All TM (Fe, Ni, Cu) sites could stabilize the sulfite absorbate to form sulfite-TM-CN adducts, with corresponding adsorption binding energies (E_ads) of −3.51 eV for Fe–CN, −3.36 eV for Ni/CN and −2.37 eV for Cu/CN, respectively (Fig. 3a). Subsequently, Bader charge of sulfite-TM/CN adducts were investigated, since the redistribution of charge density is an indicator of sulfite activation upon the sulfite adsorption. Fe/CN had the greatest charge to sulfite with a Q value of 0.2793 e. In comparison, sulfite received less charge from Ni/CN (0.2123 e) and Cu/CN (0.1994 e). Deformation charge density calculations further indicated the greatest charge transfer from HSO₃⁻ to Fe/CN. The results fully indicate that the essential sulfite activation typically involves electron donation from sulfite and electron acceptation by TM/CN catalysts. It is also concluded that various TMs serve as active sites, following the electron transfer path from HSO₃⁻ to TM to CN substrate.

To reveal the source of the difference in the strength of the interaction between various TM and sulfite, the electronic structures of TM/CN and sulfite were further explored through DOS. Generally, the electrons in the d-orbitals of TMs play a determining role in their catalytic properties. According to previous literature, d-band center (ε_d) is a good indicator of the adsorption force between the metal and the adsorbate as it is indicative of their bonding interaction, where an ε_d close to the Fermi level results in strong adsorption and a low ε_d corresponds to weak adsorption [58]. The ε_d of Cu, Ni, and Fe were calculated to be −1.53, −1.29, and −1.12 eV (Fig. 3c), respectively, consistent with the result of adsorption energy calculations. In addition, the difference between spin-up and -down projected density of states (PDOS) reflects the spin polarization intensity of the TM, and an intense asymmetry corresponds to a strong spin polarization [59]. The asymmetry of PDOS follows the order Fe > Ni > Cu, indicating the highest spin polarization of Fe while the lowest spin polarization of Cu. The PDOS of various TM 3d orbitals divided into d_xy, d_yz, d_xz, d_z2, and d_x2-y2 orbitals follow the same pattern of asymmetry (Fig. S10). The spin densities of various TMs further demonstrate the above conclusion (Fig. S11). Also, ferromagnetism highly depends on the degree of spin polarization [60]. According to DFT calculation results, the magnetic moments (μ_eff) of Cu, Ni, and Fe were 0.99, 2.00, and 4.00 μ_B. ZFC-FC test (Fig. S12) was performed to further investigate the μ_eff of various TM/CN based on Curie-Weiss law, where the μ_eff can be obtained by a linear fit of the χ⁻¹-T curve (Fig. 3d). The relevant details can be in Text S3. The μ_eff value of Fe based on ZFC-FC results was 4.17, followed by 1.76 and 1.24 μ_B for Ni and Cu, respectively. Noteworthily, the μ_eff value obtained from ZFC-FC was consistent with that obtained from DFT calculation. The results fully indicate the highest spin polarization of Fe and the lowest spin polarization of Cu from both experimental and computational results.

The PDOS of sulfite before and after TM activation was employed to further explore the effects of TM spin polarization (Fig. 3e). The 3d orbital of the TM overlaps with the molecular orbital of sulfite, demonstrating an orbital hybridization within sulfite-TM-CN adducts and thus redistributions of electron densities. The consequent highest occupied molecular orbital-lowest unoccupied molecular orbital (HOMO-LUMO) gap of activated sulfite changed significantly depending on the TM-dependent orbital hybridization, which can serve as an indicator to quantify the degree of sulfite activation. The HOMO-LUMO gap of sulfite decreased from the original value of 0.98 to 0, 0.07, and 0.16 eV following activation by Fe/CN, Ni/CN, and Cu/CN, respectively, indicating the strongest activation of Fe for sulfite and the weakest activation of Cu for sulfite. This result indicated that stronger spin polarization of TM promoted the activation of sulfite, which could be attributed to the stronger bonding strength between oxygen-containing adsorbates and TM and enhanced charge transfer [43,61,62]. In general, the k values of Fe/CN, Ni/CN, and Cu/CN for sulfite activation from experimental results make a close connection with E_ads of sulfite on various TM/CN, Q between sulfite and TM/CN, and HOMO-LUMO gap of sulfite before and after TM/CN activation obtained from DFT calculation (Fig. 3f). The higher k for TM/CN corresponds to a lower E_ads, higher Q, and lower HOMO-LUMO gap.

In this work, the activation of sulfite with the order Fe/CN > Ni/CN > Cu/CN could be attributed to the strong interaction of TM with CN, the closeness of TM ε_d to the Fermi level, and the strong spin polarization of TM. The electron donation from the HOMO orbitals of sulfite to the lower energy level of TM and the electron back donation of electrons from the higher energy level of TM to the LUMO of sulfite was also involved in our sulfite activation system (Fig. 3g). As reported [63,64], the strong binding of TMs with molecules can be attributed to an electron donation-back donation mechanism. The strongest adsorption force on sulfite with Fe could also exhibit the greatest electron donation-back donation with sulfite compared to other TMs, fully indicating the importance of spin polarization of TM on the sulfite activation force and catalytic ability. These comprehensive investigations on DFT calculation and experimental results could provide insight into the interaction between TM single-atoms and CN and the role of the electronic structure and spin polarization properties of TM single-atoms for the activation of sulfite.

3.3. The enhanced mechanism of photo-assisted sulfite activation

Visible light was introduced to enhance the performance of various TM/CN to further activate sulfite (Fig. S13). In the absence of sulfite, about 17.6% of ACT was degraded in the presence of CN under visible light irradiation within 60 min due to photocatalytic effects. In comparison, the photocatalytic efficiencies of Cu/CN, Ni/CN, and Fe/CN for ACT degradation reached 10.3%, 14.9%, and 21.8% over 60 min without sulfite. It was worth noting that the introduction of Cu and Ni did not improve the photocatalytic performance due to the nearly unchanged light absorption edge between CN and the TM/CNs. The slight decrease in photocatalytic activity could be attributed to the fact that TM might act as the composite center of photogenerated carriers, leading to a decrease in quantum efficiency [65]. Compared with CN, the slight enhancement of Fe/CN in photocatalytic activity could be mainly attributed to the expanded absorption range of light.

After the introduction of sulfite, the ACT degradation efficiency of CN increased to 42.2% over 60 min, which is attributed to the effect of photogenerated h⁺ on sulfite activation. The degradation efficiency of Cu/CN, Ni/CN, and Fe/CN increased to 47.2%, 60.8%, and 86.2% over 60 min, respectively. The k values of Fe/CN, Ni/CN, and Cu/CN increased to 0.039, 0.017, and 0.011 min⁻¹ after the introduction of visible light, respectively (Fig. 4a). Overall, photocatalytic activity was not the dominant factor in sulfite activation performance in light. TM/CN ability to activate sulfite under both dark and light conditions followed the order CN < Cu/CN < Ni/CN < Fe/CN, demonstrating the importance of TMs and their interactions with CN in the sulfite activation.

Fig. 4.

a, k_obs for ACT degradation of various systems. b–c, Scavenging test of Fe/CN on sulfite activation under dark conditions (b) under visible light (c). d–f, ESR spectra of DMPO-SO₃^•− (d); DMPO-SO₄^•− and DMPO-^•OH (e); and DMPO-O₂^•−(f). g, Proposed ACT degradation pathway.

The roles of various active species generated from the sulfite activation system were explored (Fig. 4b and c). Under the dark condition, the addition of tert-butanol (TBA) and ethanol (EtOH) reduced degradation efficiency by 2.3% and 4.8% within 60 min, respectively. Both continuous N₂ and the addition of aniline (AN) exhibited significant inhibitory effects on ACT degradation, indicating the essential role of SO₅^•−. Under light conditions, the addition of TBA showed a negligible inhibitory effect on ACT degradation, demonstrating the insignificance of ^•OH to the reaction. However, the addition of chloroform (CF) reduced degradation by 28.2%, indicating the crucial role of ^•O₂⁻. The addition of ammonium oxalate (AO) decreased degradation efficiency from 86.2% to 46.9%, demonstrating the importance of h⁺ in the degradation process due to its direct attack on ACT and its activation role to produce SO₃^•− (equations S1 and S2). The addition of EtOH exhibited a more significant inhibitory effect with a degradation efficiency of 44.3% compared with TBA, indicating the key role of SO₄^•−. Notably, among all the quenching agents, AN had the greatest inhibitory effect, reducing degradation efficiency to a mere 11.2% and demonstrating the importance of SO₅^•−, which serves as a key intermediate in the free radical generation chain. Significant inhibition under continuous N₂ conditions was also observed, further indicating the DO was integral to the sulfite activation system (equations S3 and S11) [66]. In addition, the N₂ results also prove the insignificance of SO₃^•− in ACT degradation, as SO₃^•− generation still occurs without oxygen.

The electron spin resonance (ESR) spectra confirmed the active species generated in the Fe/CN-activated sulfite system (Fig. 4d–f). Faint signals of several active species, including ^•OH, SO₃^•−, and SO₄^•−, were detected in the dark. However, no ^•O₂⁻ ESR signal was detected as light is an indispensable condition for its production. Notably, the signals of all free radicals were significantly enhanced by light and gradually increased with illumination time. The signal of 5,5-Dimethyl-1-pyrroline N-oxide (DMPO)-SO₃^•− adduct was detected [67], as shown in Fig. 4d, and the characteristic peaks exhibited in Fig. 4e were assigned to DMPO-SO₄^•− and DMPO-^•OH complexes, respectively. The four characteristic peaks with similar intensity were attributed to DMPO-^•O₂⁻ adduct (Fig. 4f) [68]. Furthermore, the higher valence band position of the Fe/CN valence band compared to the redox potential of ^•OH/H₂O (+2.27 eV) results in the inability of h⁺ to oxidize water to ^•OH. Instead, based on equations S9–14, ^•OH could be generated from ^•O₂⁻ and SO₄^•−.

The total organic carbon (TOC) removal efficiency of ACT in the Fe/CN-activated sulfite system was tested and found to be much lower than that of ACT degradation efficiency (Fig. S14), with 38.3% over 180 min, which was considered to be a common issue in heterogeneous sulfite activation system [12,69,70]. The reason could be attributed to the insufficient oxidizing capacity and self-quenching effect of oxysulfur radicals. Increasing the role of ^•OH in the system might be a solution to the mineralization issue [10]. Furthermore, the transformation products (TPs) of the ACT degradation process were identified through LC-MS following the attack of various active species (Table S5). The ACT molecule was transformed into smaller molecular weight TPs after going through direct hydroxylation, demethylation, and ring-opening reaction and eventually oxidized to CO₂, H₂O, NO₃⁻, and NH₄⁺. Based on the above, the degradation pathway in Fig. 4g was put forward.

The sulfite activation process of various TM activation mechanisms was proposed in Fig. 5 and can be summarized as follows. Regarding the types of active species produced, TM and sulfite first form a complex and then release SO₃^•−, which triggers the oxysulfur radical cycle under dark conditions. The cycle of transition metals in different valence states is accelerated by the generation of photogenerated carriers under light conditions. Moreover, light could increase both the amount and type of oxysulfur radicals. Photogenerated h⁺ on the CN surface enhances the generation of SO₃^•− while photogenerated e⁻ promotes the generation of O₂^•−, fully indicating the enhanced performance of sulfite activation by light.

Fig. 5.

Enhanced mechanism of photo-assisted sulfite activation.

The light could also enhance the electron donation-back donation process between sulfite and TM/CN. Based on the mechanism mentioned above, photogenerated e^-jumping from lower to higher energy levels in TM further strengthens the electron donation-back donation process after light irradiation, leading to enhanced sulfite activation.

3.4. Factors influencing Fe/CN for sulfite activation and stability test

The effect of different water matrix parameters on sulfite activation by Fe/CN under light conditions was investigated, and the corresponding k values are displayed in Fig. 6 and Fig. S15. The solution pH dictates the degree of protonation on sulfite and contaminants and the charges and types of active species on the catalyst surface. The k values for ACT degradation under pH 3, 5, 7, 9, and 11 were 0.031, 0.042, 0.039, 0.010, and 0.006 min⁻¹, respectively (Fig. 6a). Overall, ACT degradation efficiency was higher in acidic and neutral conditions than in alkaline conditions. Though ACT (pKa = 9.38) is more susceptible to degradation in its deprotonated form [71], the observed lower ACT degradation performance at pH 11 illustrated that contaminant protonation does not heavily impact the reaction rate.

Fig. 6.

k value of various factors on ACT degradation in the Fe/CN-activated sulfite system. a, pH. b, DO. c, Anion. d, Water matrix.

Sulfite is mainly present in the form of HSO₃⁻ (pH 1.86–7.2) and SO₃²⁻ (pH > 7.2), respectively (equations S15 and S16). Hence, SO₃²⁻ serves as the main species at pH 9 and 11 and carries a greater negative charge than HSO₃⁻. Meanwhile, the catalyst's surface carries a negative charge between pH 7–11, a positive charge between pH 3–5, and the point of zero charge (PZC) of g-C₃N₄ lies in the range of pH 5–6 [27]. Thus, under basic conditions, the repulsive force between the catalyst and sulfite is significantly stronger and impedes catalyst sulfite activation. As a result, the generation of oxysulfur species was greatly inhibited, which accounted for the poor performance of ACT degradation at pH 9 and 11. In contrast, the catalyst and sulfite are oppositely charged at pH 3 and 5. Under these conditions, the coulomb attraction promotes sulfite activation on the catalyst and causes greater ACT degradation.

Moreover, solution-phase pH has an essential effect on the bonding of metal-sulfito complexes. Metal-sulfito interactions occur via S-bonding at pH 9–11 and O-bonding at pH 3–7, respectively [72], and, compared to S-bonded metal-sulfito complexes, O-bonded metal-sulfito complexes are more inclined to decompose to oxysulfur radicals and cause greater ACT degradation [73]. In general, the Fe/CN sulfite activation system exhibits favorable prospects for use under both acidic and neutral conditions. However, more research will be needed to solve the problem of poor TM sulfite activation under alkaline conditions.

Dissolved oxygen (DO) also plays a key role in the TM redox cycle by influencing the TM's different valence states and via the oxysulfur radical chain. ACT degradation was compared under air, N₂, and O₂ conditions, with the respective k values of 0.039, 0.006, and 0.041 min⁻¹ (Fig. 6b). Almost complete inhibition of ACT degradation under N₂ indicated the importance of DO, which is indispensable for the generation of ^•O₂⁻ and SO₅^•− (equations S3 and S12). In addition, the interruption of the Fe³⁺/Fe²⁺ redox cycle by the lack of DO was also responsible for the reaction inhibition. These results also demonstrate the negligible contribution of SO₃^•− to ACT degradation due to its mild oxidation capacity. No significant difference in degradation rate in air versus pure O₂ was observed after 60 min of reaction time, though the latter had significantly faster degradation in the first 30 min. This could be explained by the rapid consumption of DO at the beginning and the subsequent recovery of DO [5].

The k values of ACT degradation in the presence of 10 mM NO₃⁻, Cl⁻, HCO₃⁻, and HPO₄²⁻ were 0.035, 0.031, 0.029, 0.025 min⁻¹, respectively, lower than that of the control experiment (0.039 min⁻¹) as displayed in Fig. 6c. The inhibition of ACT degradation by NO₃⁻ and HCO₃⁻ could be attributed to their scavenging effects on both h⁺ and SO₄^•− (equations S18–21) [74,75]. The reaction between Cl⁻ and SO₄^•− led to slightly inhibiting degradation efficiency (equation S22) [76]. HPO₄²⁻ was the strongest inhibitor of ACT degradation across all anions due to its strong ability to capture h⁺ and SO₄^•− (equations S23 and S24). In addition, the strong affinity between Fe and P was also responsible for significant inhibition [77]. Overall, Fe/CN exhibited favorable sulfite activation in the presence of other anions.

The k values of Fe/CN-activated sulfite system in deionized, tap, and river water were 0.039, 0.032, and 0.024 min⁻¹, respectively (Fig. 6d), and the corresponding water parameters were displayed in Table S6. Compared with deionized water, the efficiency of ACT degradation in river and tap water decreased by 9.4% and 6.3%, respectively. The inhibition was caused by a combination of factors, such as quenching effects on free radicals, competition between contaminant and coexisting substances, and light impediment on the catalyst surface [50]. Overall, it was concluded that the Fe/CN-activated sulfite system demonstrated favorable prospects for contaminant treatment in real water bodies.

Catalyst recyclability was investigated (Fig. S16). The degradation efficiency of ACT at 60 min decreased from 86.2% to 71.3% after five cycles, which was attributed to the exudation of small amounts of Fe ions. Nevertheless, at the 5th cycle, the ACT removal efficiency was able to reach 84.1% when the system was allowed to react for 90 min, demonstrating the longevity of Fe/CN's useful lifetime. XRD and FTIR spectra of the catalyst remained unchanged after use, demonstrating high catalyst stability (Fig. S17). Moreover, the concentrations of Fe, Ni, and Cu ions in the treated water within 60 min were measured to be 0.158, 0.107, and 0.132 mg L⁻¹, respectively, indicating negligible leaching. In addition to catalytic performance, the cost of the catalyst is also an important factor for practical usage. Fe has the most abundant content on the earth (58000 g ton⁻¹), much higher than those of Ni and Cu (both <100 g ton⁻¹) [78], which enables Fe/CN to have favorable application prospects in water treatment.

4. Conclusion

This work employed a simple preparation method to develop various single-atom catalysts, namely Fe/CN, Ni/CN, and Cu/CN, to activate sulfite for ACT degradation. Among them, Fe/CN exhibited the best performance under dark conditions with a rate constant of 0.013 min⁻¹, 1.64 and 3.13 times higher than Ni/CN and Cu/CN, respectively. Compared to Ni/CN and Cu/CN, Fe single atoms in Fe/CN had the strongest interaction and charge transfer between TM and CN and the strongest adsorption with sulfite. Furthermore, Fe/CN had an ε_d closest to the Fermi level, and Fe single-atoms on the material's surface were found to have the highest spin polarization compared to the other tested TM/CN systems. The relationship between the propensity for sulfite activation and the electronic structure of TM was established. After introducing visible light, the k value of Fe/CN remained the highest out of the tested catalysts, increasing to 0.039 min⁻¹, 2.31 and 3.50 times higher than those of Ni/CN and Cu/CN, respectively. The photo-induced h⁺, ^•O₂⁻, SO₃^•−, SO₄^•−, and SO₅^•− species were determined to be responsible for ACT degradation in the catalytic system. The introduction of light promoted the electron donation-back donation process between sulfite and TM/CN, further enhancing sulfite activation and oxysulfur radical-triggered oxidative degradation.

Introducing a small amount of Fe at 0.21% molar fraction significantly increases sulfite activation performance, leading to a lower catalyst preparation cost and metal ion leaching. This fully demonstrates the advantages of SACs in sulfite activation for water treatment. The agreement between experimental and DFT results in this work provided novel insight into the interaction between TM single-atoms and CN supports. Also, a greater understanding of the influence of electronic structure and spin polarization of TM single-atoms on the activation of sulfite was achieved. As such, this work can potentially guide the intelligent design of SACs based on knowledge of their mechanism for sulfite activation.

CRediT authorship contribution statement

Guanshu Zhao: Investigation, Methodology, Writing - Original Draft. Jing Ding: Supervision, Writing - Review & Editing. Jiayi Ren: Data Curation, Formal Analysis. Qingliang Zhao: Supervision, Funding Acquisition. Chengliang Mao: Visualization. Kun Wang: Conceptualization. Jessica Ye: Writing - Review & Editing. Xueqi Chen: Investigation. Xianjie Wang: Data Curation. Mingce Long: Visualization, Writing - Review & Editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article.