Introduction

With the expanding global population and diminishing freshwater resources, developing advanced wastewater treatment and reuse technologies has become critically important for sustaining water availability1,2,3. Among the water contaminants, HA—formed from the decay of plant and animal matter—significantly contributes to natural organic matter (NOM) in freshwater4,5,6,7,8. HA is a high molecular weight organic substance with a complex and stable structure that is refractory and resists biodegradation under natural conditions9,10. If left untreated, HA in water supplies can form stable complexes with heavy metal ions, enhancing their persistence and mobility, which leads to the accumulation of pollutants in sediments and organisms11,12,13,14,15,16. Furthermore, during chlorination, HA can form carcinogenic, teratogenic, and mutagenic disinfection byproducts such as trihalomethanes and haloacetic acids, severely compromising water quality and human health17,18,19,20,21,22. Hence, the effective removal of HA during wastewater treatment processes is crucial for ensuring safe and clean water. Currently, three main approaches exist for HA removal: biological methods (e.g., biological treatment23,24,25), chemical methods (e.g., electrolysis26, flocculation/coagulation27, and catalytic oxidation28,29,30), and physical methods (e.g., membrane filtration31,32 and adsorption33,34). Among these, adsorption is widely employed due to its cost-effective, simple design, operation ease, environmental friendliness, and high efficiency35. However, traditional adsorbents such as resins and activated carbon are limited by their low adsorption capacity, low specific surface area, and poor regeneration properties36,37.

Perovskite oxides, represented by the formula ABO3, are a diverse family of mixed metal oxides with highly tunable compositions and physicochemical properties, enabling applications in solar cells38,39, fuel cells40,41, electromagnetic materials42,43,44, chemical sensors45, and catalysts46,47. In the perovskite lattice, A-site elements, typically rare earth metals or alkaline earth metals (e.g., La, Pr, Gd) with larger ionic radii, occupy the central position of the cubic lattice and stabilize the framework, while B-site elements, usually transition metals (e.g., Ti, Mn, Fe, Co, Cr) with smaller ionic radii, occupy corner-sharing octahedra and predominantly govern the electronic structure, redox behavior, and surface chemistry48. This structural flexibility offers a powerful platform for tailoring properties relevant to water treatment, including surface charge, hydrophilicity, oxygen vacancy concentration, and magnetic separability49. For instance, Ti doping in LaFeO3 perovskites prepared by a sol-gel method enhances catalytic activity, improving the removal and mineralization of chlorophenols50, and can also modulate magnetic properties by influencing Fe spins through the unpaired electrons of Ti3+ ions51. Similarly, Co doping introduces abundant active sites that accelerate methane activation, whereas Sr doping increases oxygen vacancies, facilitating oxygen-ion migration and improving reactivity and resistance to coking52. However, despite extensive work on such doping strategies, the use of perovskite oxides as integrated adsorbent–regenerable materials for pollutant removal, including HA, remains underdeveloped. Conventional approaches to perovskite-based adsorbent enhancement still focus primarily on maximizing adsorption capacity, with comparatively less consideration to regeneration efficiency, energy input, or compatibility with continuous operation. This leads to a persistent gap between material-level optimization and process-level requirements. Moreover, the rational design and mechanistic understanding of dual B-site doping remain limited, even though exploiting complementary effects between metal ions of differing ionic radii and electronic structures could, in principle, simultaneously enhance multiple functionalities and improve material stability during extended use53,54.

Here, we implement a dual B-site doping strategy in LaFeO3, simultaneously introducing Ti and Co to obtain a perovskite-type adsorbent (LFCTO) for efficient remediation of HA-contaminated water. LFCTO exhibits rapid HA uptake and a high maximum adsorption capacity of 381 mg g−1 in batch adsorption tests, which is attributed to its increased specific surface area and improved surface hydrophilicity. Concurrently, the Co/Ti centers act as Fenton-active sites for H2O2 activation, allowing in situ oxidative degradation of adsorbed HA and efficient regeneration of the adsorbent under mild conditions. The dual-doping strategy also enhances the ferromagnetism of LFCTO, enabling efficient magnetic separation and reuse. By coupling systematic experimental characterization with Density Functional Theory (DFT)and related quantum-chemical calculations, we elucidate the atomic-level synergistic effects of dual B-site doping that underpin both adsorption and regeneration. We demonstrate that LFCTO adsorbent can be regenerated in situ within a continuous adsorption-regeneration cycle, or magnetically recovered for ex situ chemical regeneration. LFCTO maintains its crystalline structure and stable HA removal performance for over 280 h of continuous operation, enabling an energy-efficient, one-step regeneration route that avoids the high-temperature calcination or solvent-intensive elution required by conventional adsorbents.

Results

Fabrication of LFCTO absorbents

As a typical ABO3-type perovskite oxide, LaFeO3 has an atomic ratio of approximately 1:1:3 for La, Fe, and O55,56. In the case of dual-doped LFCTO, La occupies the A-site, while Fe, Co, and Ti occupy the B-site (Fig. 1a). The LF0.55C0.45-xTxO (x = 0.1, 0.2, 0.3, and 0.4) oxides were successfully synthesized using a facile sol-gel method. The resulting perovskite oxide-type adsorbents exhibited a uniform block-like porous structure (Fig. 1b and Supplementary Fig. 1a–d) and the high-resolution TEM (HRTEM) image (Fig. 1c) depicted lattice fringes of 0.28 nm, corresponding to the (110) plane of LaFeO3. Elemental composition analysis using energy dispersive X-ray spectroscopy (EDX) (Supplementary Fig. 1e) confirmed an atomic ratio of La: (Fe + Ti + Co): O was approximately 1:1:3, verifying the successful synthesis of LFCTO via the sol-gel method. X-ray diffraction spectroscopy (XRD) (Fig. 1e) revealed that the diffraction peaks of LFCTO closely matched the characteristic peaks of LaFeO3 (JCPDS no. 75-0541), consistent with its crystal structure, which was also confirmed by the selected area electron diffraction (SAED) pattern (Fig. 1d)57,58. Analysis of the (110) crystal plane (Fig. 1f) showed that as the Ti doping ratio increased, there was an overall left shift in the diffraction peaks, indicating an increase in crystal plane spacing59. Interestingly, at a doping ratio of x = 0.3, the diffraction peak shifted to its original position, maintaining the original configuration of LaFeO3. This shifting phenomenon, attributed to changes in the metal-oxygen bond lengths induced by doping, suggests that Ti/Co double doping can stabilize the perovskite crystal configuration. The Fourier transform infrared spectroscopy (FTIR) (Fig. 1g) revealed peaks at approximately 576 cm−1, indicating the presence of metal oxide bonds, specifically attributed to the Fe-O stretching vibration characteristic of the octahedral FeO6 group in LaFeO360.

Fig. 1: Physicochemical and structural characterization of adsorbents.
Fig. 1: Physicochemical and structural characterization of adsorbents.The alternative text for this image may have been generated using AI.
Full size image

a LFCTO crystal configuration. b TEM. c HRTEM. d SAED pattern of LFCTO-0.3. e XRD patterns. f Variation in the most intense peak (110). g FTIR spectra. h Ferromagnetism property (M-H curves) of LFCTO. The spin density viewed from the side and top of (i, l), the pristine LaFeO3 crystals, (j, m), Co-doped LaFeO3 crystals and (k, n), Ti-doped LaFeO3 crystals. Wavefunction iteration calculations and visualization analyses were performed on the stationary‑point structures within the DFT framework at PBEsol+U/DZVP-MOLOPT-SR-GTH level.

Most adsorbents are difficult to separate from wastewater streams, which poses challenges for their reuse. Consequently, the magnetic separation of suspended adsorbents has been extensively researched61. The M- H loop behavior of LFCTO is studied from a vibrating sample magnetometer at room temperature varying the applied magnetic field (H) during the range of 0 to ± 30 kOe exhibit, as shown in Fig. 1h and Supplementary Table 1 with the saturation magnetization measured as the primary indicator of their recovery capability62. As the Ti doping ratio increased, the magnetization initially increased and then decreased. In LFCTO, Ti can exchange-couple with Fe, shifting the Fe spin moment, which facilitates small-angle rotation tilt, thereby enhancing overall ferromagnetism due to uncompensated Fe spins. As the Fe spin moment increases, the presence of uncompensated spins allows free motion of the domain walls, further enhancing magnetization. However, high concentrations of nonmagnetic Ti can increase the antiferromagnetic spin alignments of Fe3+-O-Fe3+ and Fe4+-O2--Fe4+ through super exchange interactions, inhibit domain wall motion, resulting in decreased magnetization63. The results indicated that LFCTO-0.3 and LFCTO-0.2 exhibited clear ferromagnetic behavior (MS = 7.36 emu g−1 and 5.24 emu g−1, respectively), which is advantageous for recycling and utilization in water treatment processes64.

To further elucidate the effect of Ti/Co doping on the magnetic properties of the materials, we performed theoretical validation using Density Functional Theory (DFT) calculations. LaFeO3 is a G-type antiferromagnetic crystal, and its weak magnetism observed experimentally has been attributed to Oxygen vacancies introduced during synthesis (Fig. 1i, l). Pristine LaFeO3 crystals lack unpaired electrons, resulting in zero spin density throughout the crystal lattice, and are thus non-ferromagnetic. Introducing Co dopants significantly alters the electronic structure of the closed-shell LaFeO3 crystal. This is due to the electronic configuration of Co (III), which is [Ar]4s03 d6. According to Hund’s rule, the 3 d orbitals of Co(III) have four unpaired electrons, leading to a net magnetic moment. The difference in the number of unpaired electrons compared to Fe(III) disrupts the originally stable G-type antiferromagnetic structure when Co is incorporated into the lattice. Figure 1j, m show that Co-doped LaFeO3 crystals exhibit prevalent spin density, indicating the potential to become paramagnetic or ferromagnetic materials. Conversely, higher doping levels of Ti reduce the magnetism of the system because the electronic configuration of Ti (III) is [Ar]4s03d1, with only one unpaired electron in the 3d orbital. This results in a system with fewer unpaired electrons, leading to non-magnetism or weak ferromagnetism. Figure 1k, n further confirm this, showing only one unpaired electron present in the system.

Microstructure and metal ion state analysis of LFCTO absorbents

The specific surface areas and pore size distributions of LFCTO, which influence its adsorption capacity and rate, were determined by N2 sorption porosimetry. As shown in Fig. 2a, the N2 adsorption/desorption isotherms of LFCTO exhibited typical type IV curves with hysteresis between P/P0 = 0.5 and 1.0, indicating a mesoporous structure65. The pore size distribution (Fig. 2b) revealed that LFCTO had a pore size ranging from 2 to 30 nm, with LFCTO-0.3 possessing the smallest pore size among them. These results indicate that LFCTO-0.3 had the largest specific surface area of 28.01 m2 g−1 compared with LFCTO-0.1, LFCTO-0.2 and LFCTO-0.4. Further analysis of pore volume showed that LFCTO-0.3 had a high total pore volume of 0.121 cm3·g−1 (Supplementary Table 1). This suggests that LFCTO-0.3 could exhibit enhanced adsorption compared to other LFCTO samples.

Fig. 2: Microstructure and metal ion state analysis of adsorbents.
Fig. 2: Microstructure and metal ion state analysis of adsorbents.The alternative text for this image may have been generated using AI.
Full size image

a BET analysis and (b) pore size distribution of LFCTO. XPS analysis of (c), survey, (d), Fe 2p of LFCTO-0.3 and LFO. e Co 2p of LFCTO-0.3. f O 1 s of LFCTO-0.3 and LFO.

We performed X-ray photoelectron spectrometry (XPS) to determine the surface chemical composition and chemical states of LFO and LFCTO-0.3 (Fig. 2c–f). The XPS survey spectra (Fig. 2c) confirm that Ti and Co are successfully doped into the LFCTO-0.3 lattice. The A-site La element, which stabilizes the crystal structure, is present as La3+ (Supplementary Fig. 2a). The characteristic peaks of Ti 2p at 458.1 eV and 463.8 eV correspond to Ti 2p3/2 and Ti 2p1/2 of Ti4+, respectively (Supplementary Fig. 2b). Compared to LFO, the Fe2+/Fe3+ ratio increases (Fig. 2d), and the redox dynamics between Fe2+/Fe3+ and Co2+/Co3+ promote the generation of surface oxygen vacancies to maintain charge neutrality. This is further confirmed by the O 1 s XPS spectrum (Fig. 2f), where peaks below 530 eV correspond to lattice oxygen and peaks above 530 eV correspond to chemisorbed oxygen66. Compared to LFO, the increased ratio of lattice oxygen to chemisorbed oxygen in LFCTO-0.3 indicates that Ti/Co doping promotes the creation of more oxygen vacancies in the LFO lattice, thus improving the adsorption performance of the material67.

Electron microscopy revealed a uniform, block-like porous structure in the adsorbents. The crystalline structure was confirmed by XRD and electron diffraction, showing a well-defined perovskite structure and successful incorporation of Ti and Co through doping. The N2 sorption measurements quantitatively demonstrated a mesoporous structure, where LFCTO-0.3 exhibited the highest specific surface area (28.01 m2 g-1) and pore volume (0.121 cm3 g−1). XPS analysis provided crucial insights into the surface chemistry, revealing that Ti/Co doping not only modified the surface composition but also promoted oxygen vacancy formation through Fe2+/Fe3+ and Co2+/Co3+ redox couples. FTIR spectroscopy confirmed the presence of metal-oxygen bonds, which are essential for electron transfer during adsorption. In addition, magnetic measurements showed optimal magnetization (7.36 emu g−1) at a Ti doping ratio of 0.3, facilitating practical separation and reuse. These structural features, including high surface area, abundant oxygen vacancies as active sites, and efficient mass transfer channels through mesopores, work synergistically to enhance the material’s adsorption capacity.

Static HA adsorption on LFCTO adsorbent

Using HA as the target contaminant, Supplementary Fig. 3 illustrates the differences in adsorption performance among various adsorbents: unmodified (LFO), single-doped (LFTO and LFCO), and dual-doped (LFCTO). The removal rates of HA in water revealed that adsorption efficiency improved with the doping of Ti and Co metals, with the dual-doped LFCTO demonstrating the most significant effect. Among these, LFCTO-0.3 exhibited the highest adsorption performance. This improvement is attributed to dual doping at the B sites, which effectively enhances the specific surface area, increases the number of adsorption sites for pollutant molecules, and further boosts adsorption. The simultaneous doping Ti and Co at the B-site promotes the formation of oxygen vacancies on the surface of lanthanum ferrite (Fig. 2f and Supplementary Fig. 4). These oxygen vacancies facilitate the redox cycle at the B-site, generate electron holes, promote charge transfer, and enable subsequent regeneration through hydrogen peroxide. Nevertheless, it should be noted that all HA adsorption experiments were conducted in the absence of H₂O₂, so that HA removal in this stage proceeds exclusively via adsorption onto LFCTO; H₂O₂ is introduced only in the subsequent regeneration step, where it participates in a Fenton-like process to oxidatively degrade the pre-adsorbed HA and restore the adsorption sites.

Adsorption experiments were carried out with different concentrations of HA (CHA = 10, 30, and 50 mg L−1) at a fixed adsorbent dosage (Cads = 0.2 g L−1) and pH = 7 to compare the removal efficiency and adsorption capacity, aiming to identify the dual-doped perovskite oxide adsorbent with optimal performance (Supplementary Text 2). The results showed that all dual-doped LFCTO samples exhibited prominent adsorption for HA. Notably, LFCTO-0.3 demonstrated higher removal efficiency across different concentrations compared to other doping ratios (Fig. 3a), likely due to its largest specific surface area and smallest pore size among all dual-doped modified adsorbents (Fig. 2b and Supplementary Table 1). As the HA concentration increased, the adsorption sites were more fully utilized, further enhancing the adsorption effect. When the concentration of HA increased from 10 to 30 mg L−1, the adsorption capacity of LFCTO-0.3 increased significantly. However, as the HA concentration was increased to 50 mg L−1, the increase in adsorption capacity became less pronounced (Supplementary Fig. 5). Therefore, for subsequent experiments exploring varying conditions, an HA concentration of 30 mg L−1 was selected.

Fig. 3: HA adsorption performance and mechanistic analysis.
Fig. 3: HA adsorption performance and mechanistic analysis.The alternative text for this image may have been generated using AI.
Full size image

Adsorption performance of LFCTO-0.3: the effect of (a) HA concentration, (b) the amount of adsorbent and (c) pH. d The zeta potential of LFCTO-0.3 and HA. e PSO model fitting to static adsorption capacity (qt). The solid lines are linear regressions of the kinetic model. R² > 0.99 for all linear fittings. f Langmuir isotherm and Freundlich isotherm fitting. g The electrostatic potential of HA. h Thermodynamic for HA adsorption on LFCTO-0.3. Error bars represent the standard deviation of three replicate measurements. i LF-NMR spectra of LFO and LFCTO-0.3 using water as a probe molecule. The weak interactions within HA-LaFeO3 and its doped systems. Inset: T2 relaxation decay curves. Schematic: three distinct temporal domains of water in the perovskite oxide. j HA-LaFeO3 system; (k) HA-Co doped LaFeO3 system; (l) HA-Ti doped LaFeO3 system. Wavefunction iteration calculations and visualization analyses were performed on the stationary‑point structures within the DFT framework at the B3LYP‑D3BJ/6‑311 + G*‑CPCM(water) level (g), and at the PBEsol + U/DZVP‑MOLOPT‑SR‑GTH level (j–l).

The amount of adsorbent is a critical factor affecting adsorption performance. As shown in Fig. 3b, the HA removal efficiency by LFCTO-0.3 increased with an increase in adsorbent dosage. When the adsorbent dosage was raised from 0.1 g L−1 to 0.2 g L−1, the removal efficiency of HA significantly increased from 82% to 97%. However, further increasing the dosage to 0.4 g L−1 only marginally increased the removal efficiency to 99%. This plateau is attributed to adsorbent aggregation at higher dosages, which reduces the accessible surface area and decreases the total number of active adsorption sites. In addition, the limited number of HA molecules relative to the increased adsorbent restricts further improvements in removal efficiency. Meanwhile, the adsorption capacity gradually decreased as the adsorbent dosage increased from 0.1 to 0.4 g L−1. This reduction is due to the presence of unsaturated adsorption sites at higher dosages, leading to a decrease in per-unit adsorption capacity. Considering both the removal efficiency and adsorption capacity, 0.2 g L−1 was selected as the optimal adsorbent dosage. In addition, TOC analysis was performed on the aqueous solutions before and after the adsorption process to quantify the organic matter removal efficiency (Supplementary Table 2). TOC decreased from 7.85 mg L−1 (pre-adsorption) to 1.83 mg L−1 (post-adsorption) while IC only slightly changed (0.91 → 0.77 mg L−1), indicating that HA is removed from the aqueous phase mainly via adsorption onto LFCTO-0.3.

To optimize the adsorption process for a wide range of real-world wastewater conditions, the effect of pH on adsorption was investigated (Fig. 3c). The adsorption performance of LFCTO-0.3 under various acidic and alkaline conditions is influenced by three primary factors: the surface charge of the adsorbent, the solubility of HA, and changes in HA molecular configuration68,69. The results indicated that the removal performance of HA by LFCTO-0.3 gradually increased as the solution pH ranged from 3 to 7. Under acidic conditions, HA molecules tend to polymerize and often exist as undissociated molecules or weakly ionic species, resulting in lower solubility. This enhances the interaction between HA and the adsorption sites on LFCTO-0.3, thereby improving the adsorption performance. The zero point charge (pHpzc) of LFCTO-0.3 was determined to be 7.5 through Zeta potential measurements (Fig. 3d). This critical parameter governs the surface charge characteristics of the adsorbent: when solution pH is below 7.5, the LFCTO-0.3 surface becomes positively charged, enabling favorable electrostatic attraction with negatively charged species (formed through the dissociation of HA: HA + H2O → H3O+ + A-). This electrostatic interaction mechanism explains the enhanced adsorption performance observed under acidic conditions70,71. As the pH increased from 7 to 11, the removal efficiency of HA by LFCTO-0.3 gradually decreased. This reduction is primarily due to electrostatic repulsion between the negatively charged adsorbent surface and HA molecules under alkaline conditions. In addition, the increased solubility of HA in alkaline solutions further diminishes the adsorption capacity of LFCTO-0.372. Hence, LFCTO-0.3 demonstrated notably effective adsorption performance for HA under both acidic and neutral conditions. Furthermore, under high pH conditions, LFCTO samples show robust structural stability of LFCTO under elevated pH, which is attributed to its robust crystal structure and strong metal-oxygen bonds in the framework (Supplementary Fig. 6). Considering that the pH of HA solutions is generally around 7, the highest adsorption performance of LFCTO-0.3 at pH 7 indicates that highly efficient removal of HA can be achieved without the need for pH adjustment.

Electrostatic and thermodynamic mechanisms of HA adsorption

To investigate the enhanced adsorption and reaction mechanisms between HA and LFCTO-0.3, quantum chemical calculations were performed on HA molecules in different protonated states under a water environment (Fig. 3g). The results revealed significant differences in the electrostatic potential properties of HA molecules under various protonation states, corresponding to different pH conditions. At lower pH levels, both carboxyl groups of the HA molecule are protonated and electrically neutral. The electrostatic potential projection on its van der Waals surface shows a maximum surface electrostatic potential is 61.9 kcal mol−1 and a minimum of − 39.6 kcal mol−1, indicating significant polarization. The negatively charged regions are located near the carboxyl and nitro groups, while the positively charged regions are near the saturated cycloalkane structures. As the ambient pH increases, the HA molecule undergoes further ionization. At least one of the carboxyl groups becomes deprotonated, imparting a negative charge to the molecule. The van der Waals surface potentials becomes entirely negative, with the highest value decreasing to − 5.3 kcal mol−1 and the lowest to − 148.1 kcal mol−1. With a further decrease in ambient proton concentration, the maximum electrostatic potential shifts to − 39.6 kcal mol−1 If the ambient proton concentration is reduced even further, all active hydrogens on the HA molecules dissociate, significantly enhancing the molecule’s negative charge. The van der Waals surface then exhibits strong negative potential, with the highest being − 80.3 kcal mol−1 and the lowest − 242.5 kcal mol−1. These calculations indicate that HA molecules tend to adsorb onto positively charged sites of the adsorbents due to electrostatic interactions. Consistent with the zeta potential test results, the increase in pH leads to HA molecules becoming more electronegative. Thus, as pH increases from 3 to 7, the enhanced negative charge of HA strengthens the electrostatic attraction with the positively charged surface of LFCTO-0.3, enhancing their removal from water. However, as the pH continues to increase beyond 7, the removal efficiency significantly decreases. This reduction is primary due to the adsorbent surface becoming less positively charged or even negatively charged at higher pH levels, leading to electrostatic repulsion between the adsorbent and the negatively charged HA molecules. Furthermore, it was demonstrated that LFCTO-0.3 not only possesses a wide range of adsorption adaptability but also exhibits appreciable removal performance for various types of contaminants in water, especially for anionic contaminants (Supplementary Fig. 7).

The adsorption kinetics depend on the physical and chemical properties of the adsorbent, which influence the adsorption mechanisms. To model the experimental data of HA adsorption onto LFCTO-0.3, pseudo-first order kinetic (PFO), pseudo-second order kinetic (PSO), and intra-particle diffusion models were employed (Supplementary Text 2)73. The fitting curves are shown in Fig. 3e and Supplementary Fig. 8, and the corresponding kinetic parameters are presented in Supplementary Table 3. The ID model fitting yield correlation coefficients (R2) ranging from 0.341 ~ 0.945, indicating that intra-particle diffusion was not the rate-determining step in the adsorption process74. In comparison, the PFO model showed lower correlation with the experimental data (R2 = 0.834 ~ 0.990) than the PSO model, which exhibited a prominent fit with high correlation coefficients (R2 > 0.996). In addition, the equilibrium adsorption capacities calculated using the PSO kinetic equation (Supplementary Equation 4) were consistent with the experimental data. Therefore, the adsorption of HA onto LFCTO-0.3 is better described by the PSO model, suggesting that the adsorption process is predominantly governed by chemisorption, with physisorption playing a secondary role. To investigate the relationship between adsorption capacity and equilibrium concentration, both Langmuir and the Freundlich isotherm models were applied75. The correlation coefficients (R2) and corresponding parameters for both models are listed in Fig. 3f and Supplementary Table 4. The Langmuir model, with a correlation coefficient of R2 = 0.983, provided a better fit to the experimental data better than the Freundlich model (R2 = 0.979). This indicates that the adsorption process follows monolayer adsorption, primarily characterized by chemisorption76.

Thermodynamic analysis further elucidated the significant role of temperature in the adsorption process from an energy perspective. For practical applications, thermodynamic changes were studied at 298, 308, and 318 K. As the adsorption temperature increased, the adsorption performance of LFCTO-0.3 for HA improved, and the thermodynamic equilibrium constant (Kd) also increased. The results (Fig. 3h and Supplementary Table 5) showed that ∆G < 0, indicating that the adsorption process is spontaneous and fall within the range of physical and chemical adsorption (− 20 ~ − 40 kJ·mol−1)77. Furthermore, the absolute value of ∆G increased with increasing temperature, confirming the spontaneity of the adsorption process at higher temperatures. The positive value of ∆S suggests an increase in the enthalpy at the solid-liquid interface during adsorption, reflecting greater dispersal and interaction of HA molecules on the LFCTO-0.3 interface. This enhanced the solid-liquid interfacial area, promoting further adsorption. In addition, ∆H > 0 indicates that the adsorption process is endothermic, implying that higher temperatures favor adsorption. These results demonstrated that the adsorption of HA onto LFCTO-0.3 is a spontaneous, endothermic chemical process with improved performance at elevated temperatures.

Hydrophilicity and structural defects in LFCTO adsorbent

Low-Field Nuclear Magnetic Resonance (LF-NMR) was used to quantify the surface hydrophilicity of water trapped in the absorbents. Compared to pure water, the presence of the adsorbent causes the water to enter nanoconfined regions. The schematic in Fig. 3i illustrates the water-binding capacity and surface hydrophilicity of the perovskite oxide, representing three distinct temporal domains: sub-nanometer confinement space I, nanoconfinement II, and free space III. The LF-NMR characterization revealed that the T2 relaxation times of LFO and LFCTO were 932.4 and 142.8 ms, respectively (Fig. 3i and inset). Given equal amounts of adsorbent and solvent in both samples, the transverse relaxation time (T2) exhibited a negative correlation with the bound water content78. From LFO to LFCTO-0.3, protons adsorbed on the particle surfaces become more tightly bound, significantly shortening the NMR relaxation time and increasing the bound water content. This increase is primarily due to the improved hydrophilicity of the LFCTO-0.3 surface achieved through doping modification. The presence of more bound water on and within the adsorbent indirectly enhances its adsorption properties. This finding aligns with the type and content of surface oxygen in perovskite oxides determined by XPS O 1 s spectroscopy (Fig. 2f). The ratio of oxygen vacancies to lattice oxygen (OV/OL), which indicates the abundance of oxygen vacancies in the material, increased from 0.61 in LFO to 1.05 in LFCTO-0.3 (Supplementary Table 6). The dual doping of Ti and Co in LFCTO-0.3 promotes the formation of structural defects, generating more oxygen vacancies and active sites, thereby improving adsorption performance. Compared with LFO, LFCTO-0.3 possesses a larger number of oxygen vacancies and active sites, significantly enhancing its adsorption capacity.

The Independent Gradient Model based on Hirshfeld partition (IGMH) is a computational interaction method that utilizes the electron density gradient and has gained widespread popularity due to its universality and robustness79. We applied IGMH to calculate the adsorption behavior of HA molecules on the surface of LaFeO3 crystals. In the diagrams, the red isosurfaces represent regions of attraction, while blue denotes repulsion areas, with their size and color intensity being directly proportional. The results indicate that pristine LaFeO3 crystals possess notable adsorption capacity for HA molecules (Fig. 3j), mainly driven by interactions between Fe and O atoms, with HA molecules being adsorbed onto the crystal surface through two or more Fe-O interactions. As discussed previously, Co doping leads to a transition of the crystal material from an antiferromagnetic to a ferromagnetic state. Given that closed-shell organic molecules like HA are diamagnetic, the enhanced adsorption performance of HA on the Co-doped LaFeO3 surface is relatively modest (Fig. 3k). In contrast, Ti doping reverts the material back to an antiferromagnetic state, restoring these interactions. Due to the B-site dual-doping of Co and Ti during synthesis, the number of adsorption sites and anisotropy of the crystal increase, leading to a stronger adsorption capacity compared to the pristine LaFeO3 crystals (Fig. 3l). The dual-doping effectively enhances the crystal’s ability to adsorb HA molecules, showcasing a notable improvement in adsorption performance.

Building on the findings from the aforementioned studies, we describe the adsorption mechanism of HA onto LFCTO as follows. The adsorption capacity of LFCTO is attributed to the well-structured surface, rich oxygen-containing functional groups, and enhanced specific surface area and pore volume through dual metal doping with Ti and Co. The surface of the adsorbent undergoes protonation or deprotonation processes, generating pH-dependent surface charge distributions. These charge variations create electrostatic interactions between the adsorbent and negatively charged HA molecules, which facilitate the efficient capture of HA through selective surface adsorption mechanisms. Furthermore, doping modification induces charge compensation mechanisms and crystal structure strain, both of which contribute to the formation of oxygen vacancies. The presence of these oxygen vacancies greatly enhances electron migration within LFCTO, significantly increasing its chemical reactivity. This not only accelerates the adsorption kinetics but also provides favorable conditions for the regeneration of the adsorbent, enhancing its overall performance and economic viability.

Cyclic performance and regeneration of LFCTO adsorbent

We evaluated the reusability of LFCTO-0.3 for the adsorption of HA to assess its potential for industrial applications. Figure 4a illustrates a schematic diagram of the adsorption-desorption cycle. Under static conditions at pH 7.0, 0.2 g L-1 of absorbent was added to 200 ml of HA solution (CHA = 30 mg L−1). After the adsorption process was complete, LFCTO-0.3 was recovered using a strong magnet. The adsorbent was then immersed in a diluted hydrogen peroxide (H2O2, 0.15 M, 200 mL) solution to facilitate chemical desorption for 60 min, leveraging the Fenton properties inherent in ferrite to generate hydroxyl radicals that break down the adsorbed HA (Supplementary Fig. 9). Following three washes with deionized water and drying at 60 °C, the adsorbent was reused in the next adsorption experiment. After five adsorption-desorption cycles, the removal efficiency of LFCTO-0.3 to HA remained at 89.87% (Fig. 4b), demonstrating that the magnetic adsorbent retained its adsorption performance after chemical desorption. This also suggests that LFCTO-0.3 maintained its Fenton activity and magnetism throughout the cycles. Free radical trapping experiments reveal that significant ∙OH was observed, confirming effective H₂O₂ activation by the Fe²⁺/Fe³⁺ and Co²⁺/Co³⁺ redox pairs in LFCTO. No hydroxyl radical (∙OH) signals were detected when either LFCTO adsorbent or H2O2 existed independently (Supplementary Fig. 10). During the adsorption stage, the active sites on the LFCTO surface are partially occupied by target pollutants, reducing the number of sites available for ∙OH generation. In the subsequent degradation stage, the generated ∙OH is consumed in the H₂O₂-driven oxidative degradation of the adsorbed HA (Supplementary Fig. 11). To clarify the dominant HA removal mechanism, additional O2∙⁻ trapping and oxygen vacancies experiments were conducted (Supplementary Figs. 12, 13). The O2∙⁻ signal did not disappear during either adsorption and degradation. In fresh LFCTO, the oxygen vacancy (OV) fraction is 24.90%, increasing to 30.02% after HA adsorption and further to 32.69% following H₂O₂-mediated HA degradation, consistent with the Electron Paramagnetic Resonance (EPR) characterization trend. The slight increase in O2∙⁻ is attributed to the exposure of previously hidden vacancies via H₂O₂-induced mild oxidative reconstruction (removing surface-adsorbed HA residues or hydroxyl groups) and to the structural flexibility imparted by Ti/Co dual doping, which prevents excessive oxidation-induced lattice collapse and enables stable accumulation of oxygen vacancies. To further understand the molecular interactions, DFT calculations were performed to determine the molecular orbital structure (Supplementary Fig. 14), which confirmed the feasibility of using H2O2 for the oxidative removal of HA molecules. The calculations indicated favorable interactions between H₂O₂, LFCTO-0.3, and the adsorbed HA, facilitating the oxidative degradation of HA during the desorption process.

Fig. 4: Cyclic performance and stability of perovskite oxide-type absorbent.
Fig. 4: Cyclic performance and stability of perovskite oxide-type absorbent.The alternative text for this image may have been generated using AI.
Full size image

a Static cycle step diagram. b Static cyclic adsorption performance of LFCTO-0.3. c Dynamic cycle step diagram. d Dynamic cyclic adsorption-desorption experiments. e Comparison of HA maximum adsorption capacity (qmax) of LFCTO and the other adsorbents. The shapes and colors of the data points denote different regeneration methods: purple diamond for calcination, blue star for the Fenton reaction, orange pentagon for solvent elution, and gray square for no regeneration. The shaded regions show the typical performance ranges associated with each regeneration method. The source data, including the methodological information can be found in Supplementary Data 1.

Moreover, the concentrations of dissolved metal ions gradually decreased with each cycle. After the fifth cycle, the concentrations of La3+, Co3+, Fe3+, and Ti4+ were 0.017 mg L−1, 0.048 mg L−1, 0.026 mg L−1, and 0.012 mg L−1, respectively (Supplementary Fig. 15). These levels are below the national wastewater discharge standards, indicating minimal leaching of metal ions during regeneration. The crystal structure and surface elemental composition of LFCTO in the fresh, post-adsorption, and post-degradation states were characterized by XRD and XPS (Supplementary Figs. 16, 17). Comparison of the patterns and spectra across these states shows only minor variations in peak intensity and surface element ratios, with no evidence of phase transformation or lattice collapse after degradation, demonstrating the structural stability of the perovskite under H₂O₂ regeneration. Furthermore, the crystal structure of the perovskite adsorbent remained unchanged after five cycles, as confirmed by XRD analysis, indicating good structural stability (Supplementary Fig. 18).

To investigate the regeneration and stability performance of LFCTO-0.3 during the dynamic adsorption process, we conducted five cycles of adsorption-desorption experiments using a fixed bed column packed with LFCTO-0.3, as illustrated in Fig. 4c. The breakthrough curves (Fig. 4d) indicate that LFCTO-0.3 maintained high adsorption performance after five cycles. As the cyclic experiment progressed, both the time to reach the breakthrough point and the saturation time decreased, suggesting a reduced adsorption capacity of LFCTO-0.3 for HA. This reduction is likely due to incomplete desorption within a fixed recovery period of 20 min. Despite this, LFCTO-0.3 exhibits good stability and regenerative capability, as confirmed by XRD analysis.

Notably, LFCTO-0.3 ranks among the highest in adsorption capacities for HA when compared to other adsorbents reported in the literature (Fig. 4e and Supplementary Data 1). It rapidly achieves HA removal with a maximum adsorption capacity of 381 mg g−1 under 2 hr (Supplementary Data 1; Detailed comparison and methodology are provided in the Supporting Information). In addition, traditional regeneration methods for reported adsorbents typically involve high-temperature calcination or solvent elution using agents like NaOH, resulting in elevated energy consumption and potential secondary pollution80,81,82. In contrast, our adsorbent can be regenerated in under 20 min (Fig. 4d) through the Fenton reaction, facilitated by LFCTO’s ability to generate radicals. The impressive adsorption and regeneration properties of LFCTO-0.3 make it particularly suitable for use in adsorption columns. Its easy dispersion and straightforward magnetic separation also allow its application in existing Advanced Oxidation Processes (AOP) tanks without requiring structural modifications to sewage treatment facilities. The versatility of LFCTO-0.3 supports flexible application scenarios. These findings suggest that LFCTO-0.3 could serve as an environmentally friendly and magnetic adsorbent with marked HA removal performance, offering significant advantages in ease of regeneration, stability, and reduced environmental impact compared to conventional adsorbents.

Discussion

We developed LFCTO, a dual-doped perovskite oxide adsorbent synthesized via the sol-gel method, featuring Co and Ti doping at the B-site. LFCTO exhibits superior adsorption performance for NOM, particularly HA. Kinetic and thermodynamic analyses confirm that the adsorption mechanism is predominantly chemical, augmented by spontaneous endothermic physical interactions. LFCTO’s appreciable Fenton catalytic activity and magnetic properties enable efficient regeneration and reuse through magnetic separation, fixed-bed adsorption, and the Fenton reaction. This process circumvents the high energy consumption and potential secondary pollution associated with traditional regeneration methods. The versatility of LFCTO allows seamless integrated into adsorption columns and AOP tanks without requiring structural modifications to existing sewage treatment plants. Its easy dispersion and straightforward magnetic separation further broaden its application potential in diverse wastewater treatment scenarios. Overall, LFCTO emerges as a highly promising adsorbent for water treatment, offering significant advantages in adsorption efficiency, ease of regeneration, stability, and reduced environmental impacts. Its adoption across diverse wastewater treatment settings could substantially enhance the removal of NOM, enhance water quality management, and promote more sustainable treatment processes.

Method

Chemical reagents

HA was purchased from Alfa Aesar (China) Chemical Co., Ltd. Methylene blue, Congo red, Rhodamin B, sodium hydroxide (NaOH), and anhydrous ethanol were purchased from Sinopharm Chemical Reagent Co., Ltd. Oxytetracycline hydrochloride, sodium citrate, ammonia, hydrochloric acid, and hydrogen peroxide were purchased from Aladdin Reagent (Shanghai) Co., Ltd. Tetrabutyl titanate and polyethylene glycol 600 were purchased from Shanghai Lingfeng Reagent Co., Ltd. La(NO3)3·6H2O, Fe(NO3)3·9H2O, and Co(NO3)3·6H2O were purchased from Shanghai Macklin Biochemical Co., Ltd. The chemicals used in this investigation were of analytical grade, and no additional purification steps were conducted on any of the reagents. All solutions were prepared using deionized water (DI).

Synthesis of LFCTO adsorbents

LaFeO3 (LFO) and LaFe0.55Co0.45O3 (LFCO) perovskite oxides were synthesized using a sol-gel method followed by calcination. First, the metal compounds of La(NO3)3·6H2O, Fe(NO3)3·9H2O, and Co(NO3)2·6H2O were mixed, stirred, and dissolved in a beaker. The molar ratio was La: Fe = 1:1 for preparing LFO and La: Fe: Co = 1:0.55:0.45 for preparing LFCO. Sodium citrate was added to the mixed metal solution to achieve a molar ratio of 1:5 between metal cations and citrate anions. The pH of the metal solution was adjusted to 4 using 20% ammonia solution. The mixed metal solution was stirred continuously at 80  °C until it transformed into a sol-gel state. The resulting wet gel was placed in a preheated oven and dried at 120  °C overnight. The dried gel was then placed in a muffle furnace and heated from 20  °C to 300  °C at a rate of 5  °C min−1. It was then held at 300  °C for 3 h before being heated further to the final temperature of 700  °C, where it was held for 5 h. After natural cooling of the muffle furnace, the desired product (LFO and LFCO) was obtained.

LaFe0.55Ti0.45O3 (LFTO) and LaFe0.55Co0.45-xTixO3 (LFCTO) were synthesized by adding titanium dioxide sol to the above mixed metal solution. Titanium dioxide sol was prepared by adding tetrabutyl titanate, anhydrous ethanol, PEG600, and deionized water in a molar ratio of 3:12:3:1 to form a transparent sol. The prepared Ti dioxide sol was slowly added into the mixed metal solution and stirred sufficiently for 1 hour at 50  °C. The molar ratio was La:Fe:Ti = 1:0.55:0.45 for preparing LFTO. For the preparation of dual-doped LaFe0.55Co0.45-xTixO3 (LFCTO, x = 0.1, 0.2, 0.3, and 0.4), the molar ratio is La:Fe:Co:Ti = 1:0.55:0.45-x:x. For example, the preparation of dual-doped LFCTO-0.3, the corresponding molar ratio is La:Fe:Co:Ti = 1:0.55:0.15:0.3. Sodium citrate and ammonia solution were then added sequentially to the mixed metal solution in the same steps as above. The mixed metal solution was stirred continuously at 80  °C until it transformed into a sol-gel state. The perovskite oxide-type adsorbents, LFTO and LFCTO were subsequently obtained using the same dry and annealing conditions.

Material characterizations

The surface morphology of the metal oxide was characterized by scanning electron microscopy (SEM, HITACHI S-4800, Japan). The elemental composition of the metal oxide was obtained by energy-dispersive X-ray spectroscopy (EDX, Horiba, Japan). The morphological structure was further analyzed by transmission electron microscopy (TEM, FEI Tecnai F20, USA). The crystal morphology of the perovskite oxide-type adsorbents was analyzed by X-ray diffraction spectroscopy (XRD, RIGAKU MiniFlex600, Japan). The functional groups of perovskite oxide-type adsorbents were identified by Fourier transform infrared spectroscopy (FTIR, Thermo Nicolet8700, USA). The textural properties of the perovskite oxide-type adsorbents were assessed by N2 adsorption/desorption analysis (BET, Micromeritics ASAP 2460, USA). Metal cation states and oxygen forms were investigated by X-ray photoelectron spectrometry (XPS, ESCALAB 250XI, USA). Ferromagnetism was measured using a vibrating sample magnetometer (VSM, Quantum Design Dyna Cool, USA) at room temperature. The zeta potential of perovskite oxide-type adsorbents in aqueous solution was determined by the laser scattering method (Malvern Nano ZS90, UK). In addition, the absorbance of HA at 254 nm was determined by UV-visible spectrophotometry (Perkin Elmer Lambda950, USA). The dissolution concentrations of leached metal ions were measured by inductively coupled plasma spectrometry (ICP, Optima 7000DV, USA). The hydrophilicity and water distribution characteristics of the adsorbents were characterized by low-field nuclear magnetic resonance spectroscopy (LF-NMR, NM42-40H-I, Suzhou Niumag). The oxygen vacancies and free radicals were characterized by Electron Paramagnetic Resonance (EPR, Bruker EMXplus-6/1, Germany).

Computational simulation

DFT calculations were performed on the small molecules, lanthanum ferrate crystals, and their doped variants involved in the experiments. The wavefunction analysis, molecular visualization, and structural modeling were carried out with Multiwfn 3.8, VMD 1.9.3, and ChimeraX 1.90 software, respectively. The details are described in the Supplementary Text.