Bilayer-Coated Magnetite Nanoparticles as Magnetic Adsorbents for Efficient Paraquat Removal

Abstract

The environmental persistence and acute toxicity of paraquat (PQ²⁺) highlight the urgent need for efficient and reusable adsorbents for water remediation. In this study, linoleic acid (LA)- and palmitic acid (PA)-coated magnetite nanoparticles (Fe₃O₄@LA and Fe₃O₄@PA) were synthesized via a continuous-flow coprecipitation process to form bilayer structures. Structural analyses (TEM, XRD, FTIR, TGA, BET, and VSM) confirmed successful bilayer coating and preserved superparamagnetic properties, ensuring easy magnetic recovery. Batch adsorption experiments revealed that Fe₃O₄@LA exhibited a maximum Langmuir capacity (Q _max) of 14.8 ± 0.4 mg g^–1, outperforming Fe₃O₄@PA (11.2 ± 0.3 mg g^–1), owing to enhanced electrostatic and hydrophobic interactions provided by the unsaturated linoleic acid bilayer. Adsorption kinetics followed a pseudo-second-order model, indicating chemisorption, while thermodynamic parameters confirmed that the process is exothermic and spontaneous. Both nanoadsorbents demonstrated good recyclability over six adsorption–desorption cycles with negligible capacity loss. These results demonstrate that fatty acid bilayer-coated Fe₃O₄ nanoparticles are cost-effective, magnetically recoverable, and environmentally benign materials for the remediation of herbicide-contaminated water.

1. Introduction

Paraquat (1,1′-dimethyl-4,4′-bipyridinium dichloride, PQ²⁺) is a widely used nonselective herbicide known for its rapid and nonselective action against weeds. Despite its agricultural effectiveness, paraquat poses severe environmental and health hazards. − Recent assessments by international agencies reaffirm its classification as a highly hazardous pesticide. The European Union banned paraquat in 2007, and the U.S. Environmental Protection Agency (EPA, 2021) mandated stricter risk mitigation for ecological and occupational exposure. Updated toxicological thresholds were also reported by the FAO/WHO Joint Meeting on Pesticide Residues (JMPR, 2016), while WHO (2024) and APVMA (2024) reaffirmed paraquat’s high hazard status and promoted global phase-out initiatives. , Its high-water solubility and chemical stability lead to accumulation in surface and groundwater, where it is highly toxic to aquatic organisms and humans. Acute paraquat poisoning frequently results in fatal outcomes, whereas chronic exposure has been associated with progressive neurological disorders, including Parkinson’s disease. −

Various methods have been proposed for paraquat removal, including advanced oxidation, ozonation, Fenton reactions, , membrane separation, , and biodegradation. However, adsorption is considered the most practical approach due to its operational simplicity, cost-effectiveness, and potential for adsorbent regeneration. −

Conventional adsorbents such as activated carbon, diatomaceous earth, clays, and biochar have been widely investigated for paraquat removal; however, these materials often suffer from drawbacks such as low selectivity toward cationic herbicides, limited adsorption capacity, and difficulties in regeneration and reuse, which restrict their practical applications in large-scale water treatment. −

Magnetite (Fe₃O₄) nanoparticles have emerged as promising candidates for pollutant adsorption owing to their large surface area, surface tunability, and superparamagnetic properties, which enable quick recovery under an external magnetic field. − Nevertheless, pristine Fe₃O₄ tends to aggregate and oxidize, reducing adsorption efficiency. Surface modification strategies–such as silica coating, polymer functionalization, and hybridization with biochar–have been widely applied to improve colloidal stability, prevent nanoparticle aggregation, and introduce functional groups that enhance the adsorption affinity of Fe₃O₄-based nanomaterials toward paraquat and other organic pollutants. −

Recently, multifunctional magnetic sorbents have attracted significant attention for the remediation of various emerging pollutants, including dyes, pharmaceuticals, and heavy metals. Such hybrid materials integrate multiple functionalities–electrostatic, and hydrophobic interactions–resulting in broad-spectrum adsorption capability and excellent reusability. In particular, polymer–silica and organic–inorganic hybrid magnetic nanocomposites have demonstrated tunable surface affinity, chemical stability, and sustainable regeneration performance. , These advances have strengthened the conceptual framework for designing multifunctional and environmentally benign magnetic adsorbents for pollutant removal. An alternative and sustainable approach is the surface modification of Fe₃O₄ with fatty acids such as linoleic acid (LA) and palmitic acid (PA). These amphiphilic molecules strongly anchor to the Fe₃O₄ surface through their carboxylate groups, while their long hydrophobic tails self-assemble to form bilayer coatings. − The bilayer provides multiple advantages: (i) enhanced colloidal stability by preventing nanoparticle aggregation, (ii) creation of amphiphilic interfaces that promote interactions with organic and inorganic pollutants, and (iii) introduction of negatively charged carboxylate groups that facilitate electrostatic attraction with cationic species such as paraquat. Moreover, the unsaturated structure of linoleic acid introduces flexibility and enhances hydrophobic interactions, which may slightly improve affinity for paraquat, but direct π–π stacking is improbable due to the absence of aromatic rings, thereby enhancing adsorption efficiency.

In our previous study, we developed a continuous-flow synthesis of LA- and PA-coated Fe₃O₄ nanoparticles, achieving uniform particle size, scalability, and excellent water dispersibility. While that work focused on physicochemical characterization and stability, the environmental application of these bilayer-coated nanoparticles for herbicide adsorption remained unexplored. In contrast, other surface-modified Fe₃O₄ systems, such as silica-encapsulated or polymer-functionalized nanoparticles, have already demonstrated strong paraquat binding through electrostatic interactions, hydrogen bonding, and ion-exchange mechanisms, with reported adsorption capacities exceeding 200 mg g^–1. − These findings highlight the potential of surface-engineered Fe₃O₄ nanomaterials but also underscore the need to explore greener, low-cost alternatives such as fatty acid bilayer coatings.

The novelty of this work lies in a continuous-flow route to fatty-acid bilayer–coated magnetite that (i) uses green, low-cost ligands (LA/PA), (ii) exploits a synergistic bilayer architecture–chemisorbed inner – COO^– anchoring plus an interdigitated, π-rich outer tail assembly–to enhance K^L, k ₂, and ΔG° toward paraquat (PQ²⁺), and (iii) provides tailorable organic loading, which tunes textural/magnetic properties and, in turn, governs adsorption and recyclability. In contrast to multifunctional Fe-based composites/MOFs optimized for other pollutants (e.g., phosphate or cationic dyes) , that often rely on complex syntheses or costly precursors, our system emphasizes sustainability, magnetic separability, and cycle stability while specifically targeting PQ²⁺.

To put the bilayer system into context, we compared it with other magnetic adsorbents reported recently. A hybrid Fe₃O₄@SiO₂/kappa carrageenan–starch biosorbent synthesized using TEOS and natural polysaccharides exhibits high paraquat capacity (∼257 mg g^–1) but requires multistep sol–gel processing and relatively expensive silica precursors. In another study, Fe₂O₃@polypyrrole (Fe₂O₃@PPy) nanocomposite packed columns predicted by ANN and ANFIS achieved breakthrough capacities of only ∼ 4.85 mg g^–1 for 2,4 dichlorophenoxyacetic acid and involved oxidative polymerization of pyrrole. Magnetic activated carbon derived from Peltophorum leaves displays high surface area and tetracycline capacity (∼123 mg g^–1) but requires high temperature pyrolysis and chemical activation steps. These examples illustrate that maximal capacity often comes at the cost of energy-intensive processing or toxic reagents. In contrast, the bilayer approach reported here uses renewable fatty acids and an aqueous continuous-flow process to produce amphiphilic Fe₃O₄ adsorbents with moderate capacity (14.8 mg g^–1) but excellent magnetic separability and recyclability. By avoiding polymeric or inorganic coatings, the material emphasizes green synthesis, low cost, and simplicity while still delivering sufficient paraquat removal to be practical at environmental concentrations.

The objectives of this work are to (i) assess the adsorption capacity of Fe₃O₄@LA and Fe₃O₄@PA bilayer nanoparticles toward paraquat under varying conditions; (ii) analyze adsorption isotherms, kinetics, and thermodynamics to elucidate the governing mechanisms; (iii) compare the performance of unsaturated (LA) and saturated (PA) bilayers in relation to electrostatic and hydrophobic interactions; and (iv) demonstrate the reusability, stability, and scalability of fatty acid bilayer-coated Fe₃O₄ as sustainable, magnetically recyclable adsorbents for the efficient remediation of herbicide-contaminated water.

2. Experimental Section

2.1. Materials

FeSO₄·7H₂O, FeCl₃, NH₄OH, linoleic acid (LA), palmitic acid (PA), ethanol, and paraquat dichloride hydrate were obtained from commercial suppliers and used as received. Deionized water was used throughout all experiments.

2.2. Synthesis of Fe₃O₄@LA and Fe₃O₄@PA

Fe₃O₄ nanoparticles were synthesized by continuous-flow coprecipitation of Fe²⁺ and Fe³⁺ salts with NH₄OH, following our previously reported method. During crystallization, LA or PA was introduced to form bilayer-coated nanoparticles (Fe₃O₄@LA and Fe₃O₄@PA). The products were magnetically separated, washed with ethanol and water, and dried under vacuum at 80 °C. Full details of synthesis conditions are provided in the Supporting Information (SI, S1).

2.3. Characterization

The morphology, crystallinity, and surface chemistry of the materials were analyzed using TEM, XRD, and FTIR. Thermal stability was assessed by TGA, and magnetic properties were measured using VSM. BET surface area, pore volume, and pore size distribution were obtained from N₂ adsorption–desorption isotherms. Instrumental conditions and calibration details are provided in the SI (S2).

XPS analysis was not performed in this study; however, the successful formation of fatty-acid bilayers and carboxylate anchoring on Fe₃O₄ was confirmed through FTIR, TGA, and BET analyses.

2.4. UV–Vis Spectrophotometric Measurements

UV–Vis spectra were recorded using a Cary 60 UV–Vis spectrophotometer (Agilent Technologies, USA) with 1.0 cm quartz cuvettes. Measurements were scanned between 200–350 nm using DI water as the reference.

Paraquat standard solutions (0–30 ppm) were analyzed at 257 nm, where no matrix interference was observed. Net absorbance (A _net) values were obtained after blank subtraction.

A linear calibration curve was obtained: A _net = 0.0791C + 0.0171 (R² = 0.9976)

LOD and LOQ were calculated using 3σ/slope and 10σ/slope, resulting in LOD = 0.021 ppm, LOQ = 0.071 ppm demonstrating high analytical sensitivity.

2.5. Adsorption Experiments

A 1,000 mg L^–1 paraquat (PQ²⁺) stock solution was prepared in DI water and stored at 4 °C in the dark. Working solutions (5–200 mg L^–1) were freshly prepared. PQ concentrations were determined by UV–Vis spectrophotometry at 257 nm.

Batch adsorption studies were performed in 50 mL tubes containing 20.0 mL PQ solution and adsorbent dosages of 1–6 g L^–1. Suspensions were shaken at 200 rpm (25–50 °C), and pH was adjusted using 0.1 M HCl/NaOH. After equilibrium, nanoparticles were magnetically separated and the supernatant was analyzed. Adsorption capacity (q _t , q _e ) and removal efficiency (R) were calculated using standard equations.

$$q_t=(C_0-C_t)\times \mathrm{V}/\mathrm{m}$$

$$q_e=(C_0-C_e)\times \mathrm{V}/\mathrm{m}$$

$$\mathrm{Removal}(\%)=(C_0-C_e)/C_0\times 100$$

where c₀ (mg L^–1) was the initial concentration of paraquat, c_t (mg L^–1) was the concentration at time t, ce (mg L^–1) was the equilibrium concentration, v­(L) is the solution volume, and m (g) was the mass of adsorbent.

2.6. Adsorption Isotherms and Kinetics

Isotherm studies were carried out by varying the initial PQ concentration (5–200 mg L^–1) at 25 °C. Data were fitted to Langmuir and Freundlich models.

$$q_e=Q_{\max }{K}_L{C}_e/1+K_{L}Ce$$ Langmuir

$$q_e={{KFC}_e}^{1/\mathrm{n}}$$ Freundlich

Kinetic studies were performed at fixed concentrations, with samples taken at 1–60 min, and analyzed using pseudo-first-order (PFO), pseudo-second-order (PSO), and intraparticle diffusion (IPD) models.

$$\mathrm{PFO}:\ln (q_e-q_t)=\ln q_e-k_1\mathrm{t}$$

$$\mathrm{PSO}:\mathrm{t}/q_t=1/(k_2{q_e}^2)+\mathrm{t}/q_e$$

$$\mathrm{IPD}:q_t=k_{\mathrm{id}}{\mathrm{t}}^{0.5}+\mathrm{C}$$

All experiments were conducted in triplicate; results are reported as mean ± SD; one-way ANOVA used for replicate comparison.

2.7. Thermodynamic Studies

Adsorption experiments were conducted at 288–318 K. Thermodynamic parameters were calculated from the van’t Hoff relation:

$$\ln \mathrm{Kc}=-({\mathrm{\Delta H}}^{°}/\mathrm{RT})+({\mathrm{\Delta S}}^{°}/\mathrm{R})$$

$${\mathrm{\Delta G}}^{°}=-\mathrm{RT}\ln \mathrm{Kc}$$

To correct the thermodynamic analysis, we redefined the adsorption equilibrium constant using the dimensionless ratio Kc = q _e /C _e rather than q _e alone. Adsorption experiments were harmonized over 298–323 K (25–50 °C). Van’t Hoff plots of ln Kc versus 1/T were constructed to obtain ΔH° and ΔS° from the slope and intercept of the linear fit, and 95% confidence intervals for the regression lines were reported. Gibbs free energies were recalculated using ΔG° = – RT ln Kc. These corrections yield consistent negative ΔH° and ΔS° values and slightly more negative ΔG° values, confirming the exothermic and spontaneous nature of paraquat adsorption on the bilayer sorbents.

2.8. Reusability Tests

Spent adsorbents were magnetically collected and regenerated using 0.5 M NaCl/20% ethanol for 24 h. After drying, they were reused under identical conditions (C ₀ = 20 mg L^–1, dosage = 2.0 g L^–1, pH = 7, 30 °C, 24 h). Adsorption was repeated for six cycles.

3. Results and Discussion

3.1. Characterization of Nanoparticles

Figure shows the TEM images and particle size distributions of Fe₃O₄@LA and Fe₃O₄@PA nanoparticles. Both samples exhibited nearly spherical morphologies with uniform contrast, indicating well-defined magnetite cores coated with fatty-acid bilayers. Histogram analysis revealed average particle sizes of 11.5 ± 2.4 nm for Fe₃O₄@LA and 13.9 ± 4.2 nm for Fe₃O₄@PA, with the majority of particles in the 5–20 nm range. These results confirm that the continuous-flow synthesis produced highly dispersed nanoparticles with narrow size distributions, consistent with previous reports on bilayer-coated Fe₃O₄. The slightly larger size observed for Fe₃O₄@PA is attributed to the more compact packing of saturated palmitic-acid chains, whereas the cis double bonds of linoleic acid induce kinks that disrupt dense packing. These structural differences are expected to influence adsorption by increasing hydrophobicity and accessible surface area in Fe₃O₄@LA.

1.

TEM images and size distribution of (a,b) Fe₃O₄@LA and (c,d) Fe₃O₄@PA.

In the FTIR spectra (Figure ) confirmed bilayer coating through the appearance of C–H stretching (2920–2850 cm^–1), C = O (∼1700 cm^–1), and COO^– vibrations (1550–1420 cm^–1). To quantify organic loading, TGA was performed (Table S1). Bare Fe₃O₄ showed only minor weight loss (<5%) due to adsorbed water. In contrast, Fe₃O₄@LA and Fe₃O₄@PA exhibited multistep degradation attributed to decomposition of fatty acid bilayers, with total weight losses of ∼ 33% for Fe₃O₄@LA and ∼ 20% for Fe₃O₄@PA. These results confirm the bilayer nature of the coating, where the first layer is chemisorbed via carboxylate anchoring and the second layer physisorbed through hydrophobic interactions. The higher organic loading in Fe₃O₄@LA is consistent with broader and more intense FTIR bands, reflecting the less compact, unsaturated bilayer. In the FTIR spectra, broad absorption at ∼ 3400 cm^–1 is attributed to O–H stretching of adsorbed water, while the band near 580 cm^–1 corresponds to Fe–O vibrations of magnetite. The asymmetric and symmetric stretching vibrations of deprotonated carboxylate groups appear at 1550–1420 cm^–1, confirming chemisorbed anchoring. The intensities and widths of these bands differ between Fe₃O₄@LA and Fe₃O₄@PA due to the more open unsaturated bilayer of LA. Although FTIR after paraquat adsorption was not obtained in this study, literature reports show new bands near 1635 cm^–1 (N–H bending) and 1345 cm^–1 (C–N stretching) for paraquat along with a red-shift of the carboxylate band by ∼ 10 cm^–1, indicative of electrostatic binding between PQ²⁺ and the bilayer.

2.

FTIR spectra of (a) Fe₃O₄ (b) Fe₃O₄@LA and (c) Fe₃O₄@PA.

The bilayer loading can be tailored by adjusting the fatty-acid feed ratio and residence time during continuous-flow synthesis, as confirmed by TGA (Figure . and Table S1). Controlled variation in loading directly affects surface hydrophobicity, porosity, and magnetic response.

3.

TGA (top) and DTG (bottom) curves of (a) bare Fe₃O₄, (b) Fe₃O₄@LA, and (c) Fe₃O₄@PA under N₂ atmosphere.

Thermogravimetric analysis (TGA) was used to quantify the organic content and evaluate the thermal stability of the samples (Figure ). The bare Fe₃O₄ nanoparticles (a) exhibited only a very small weight loss (<5 wt %) below 200 °C, which corresponds mainly to the release of physically adsorbed moisture, confirming the absence of organic groups on the surface. In contrast, both fatty-acid-modified samples showed clear multistep thermal degradation patterns characteristic of surface-bound organic chains. Fe₃O₄@LA (b) displayed two major decomposition events between ∼ 250–450 °C and ∼ 650–780 °C, giving a total weight loss of approximately 33 wt %, whereas Fe₃O₄@PA (c) exhibited a lower overall loss of about 20 wt % across similar temperature ranges. These results indicate that the LA-coated nanoparticles possess a higher grafted organic fraction and hence a thicker or more densely packed bilayer than the PA-coated sample.

The DTG profiles further support this interpretation: Fe₃O₄@LA shows sharper and more intense degradation peaks than Fe₃O₄@PA, reflecting a larger amount of thermally decomposable organic material. The higher coating density on Fe₃O₄@LA is consistent with its superior adsorption performance reported in Section 3.2, where the increased organic content enhances hydrophobic interactions and provides more accessible binding sites for paraquat cations. Thus, the TGA/DTG results quantitatively confirm that differences in coating thickness directly contribute to the observed variation in adsorption capacity.

Nitrogen adsorption–desorption isotherms (Figure ) displayed Type IV isotherms with H3 hysteresis loops, which is characteristic of textural mesoporosity that typically arises from the nonuniform aggregation of spherical nanoparticles. The formation of this pore structure between the aggregated Fe₃O₄ cores, stabilized by the fatty acid bilayer, confirms that the particles are not entirely dense. While the bilayer coating did not eliminate this textural mesoporosity, it altered pore accessibility and internal surface chemistry. This mesoporous structure is crucial as it creates pathways that facilitate the diffusion of the paraquat molecules from the bulk solution to the active adsorption sites within the aggregate, which helps explain the observed rapid initial uptake kinetics. As summarized in Figure c–e, Fe₃O₄@LA exhibited a significantly larger surface area (54.9 m² g^–1), pore volume (0.126 cm³ g^–1), and pore radius (32.8 Å) than Fe₃O₄@PA (13.8 m² g^–1, 0.030 cm³ g^–1, and 15.3 Å). This difference is attributed to the structural characteristics of LA, whose unsaturated chains prevent dense packing, yielding a more open bilayer and facilitating paraquat diffusion and adsorption.

4.

N₂ adsorption–desorption isotherms of a) Fe₃O₄@LA, b) Fe₃O₄@PA and textural properties: c) BET surface area, d) pore volume, and e) pore radius.

XRD patterns (Figure S1) confirmed that all samples retained the inverse spinel structure of magnetite (JCPDS 19–0629), with peaks at 2θ ≈ 30.1°, 35.5°, 43.2°, 53.5°, 57.1°, and 62.7°. No impurity phases were observed, demonstrating phase purity. The coating process did not alter the lattice structure, but slight peak broadening was observed due to the organic shell. Crystallite sizes calculated from the Scherrer equation were ∼ 12 nm (Fe₃O₄@LA) and ∼ 14 nm (Fe₃O₄@PA), in good agreement with TEM results.

Vibrating sample magnetometry (VSM) analysis (Table S2) revealed high saturation magnetization values of 78.5 emu g^–1 for Fe₃O₄@LA and 78.3 emu g^–1 for Fe₃O₄@PA, with low coercivity (∼15 G) confirming superparamagnetic behavior. These results indicate that the fatty-acid bilayer does not significantly attenuate the magnetic response of Fe₃O₄. Although a nonmagnetic organic shell would normally decrease the saturation magnetization when expressed per gram of composite, the fatty-acid bilayer accounts for only ∼ 4–5 wt % of the total mass. The M_s values reported here were therefore normalized to the mass of the Fe₃O₄ core after subtracting the organic fraction. This normalization, together with the bilayer’s effective passivation of surface spins, explains why M_s values of the coated samples remain comparable to (or slightly higher than) those of the uncoated magnetite. By suppressing spin canting and enhancing magnetic ordering at the core–shell interface, the bilayer can yield a modest increase in apparent M_s. The high M_s and negligible H_c enable rapid magnetic recovery within seconds using a standard laboratory magnet (Section 3.4), consistent with practical observations during adsorption–desorption cycles. The strong magnetization is attributed to surface passivation by the bilayer, which minimizes spin disorder, while the low coercivity ensures efficient separation and recyclability. Raw hysteresis loops for both samples are provided in the Supporting Information (Figure S2) and show overlapping M–H curves, confirming the high magnetic responsiveness of both composites and the absence of anomalous behavior. Collectively, these findings confirm the excellent magnetic responsiveness and reusability of the bilayer-coated Fe₃O₄ nanoparticles for adsorption applications.

3.2. Adsorption Studies

3.2.1. Effect of Adsorbent Dosage

The effect of adsorbent dosage on paraquat removal efficiency and adsorption capacity is illustrated in Figure . As shown in Figure a, the removal percentage increased sharply from ∼ 25% at 1 g L^–1 to above 85% at 3 g L^–1, after which the trend plateaued, indicating that sufficient active sites were available for paraquat uptake at higher dosages. Importantly, Fe₃O₄@LA consistently exhibited higher removal efficiency compared to Fe₃O₄@PA at all dosages tested, reflecting the enhanced adsorption affinity provided by the unsaturated linoleic acid bilayer.

5.

(a) Effect of adsorbent dosage on paraquat removal efficiency (%). (b) Adsorption capacity (q _e ± SD, mg g^–1) of Fe₃O₄@LA and Fe₃O₄@PA as a function of adsorbent dosage at 25 °C. Error bars represent standard deviations from three independent measurements (n = 3). Fe₃O₄@LA consistently exhibits higher adsorption performance owing to electrostatic and hydrophobic interactions provided by the unsaturated bilayer.

Conversely, the adsorption capacity (q _e , mg g^–1) decreased progressively with increasing adsorbent dosage (Figure b). At low dosage (1–2 g L^–1), q_e reached ∼ 14–15 mg g^–1 for Fe₃O₄@LA and ∼ 10–11 mg g^–1 for Fe₃O₄@PA, values close to the maximum predicted Langmuir capacities. However, further increasing the dosage reduced q _e significantly (to ∼ 6 mg g^–1 at 6 g L^–1), which can be attributed to the capacity dilution effect: when more adsorbent is present under constant C ₀ and volume, the available paraquat molecules are distributed across a larger number of active sites, resulting in lower average uptake per gram.

Taken together, these results suggest that the optimum operational dosage is around 3 g L^–1, balancing high removal efficiency (>90%) with moderate adsorption capacity. The superior performance of Fe₃O₄@LA underscores the contribution of electrostatic and hydrophobic interactions from the unsaturated fatty acid bilayer, in addition to electrostatic binding of PQ²⁺, which collectively enhance adsorption affinity relative to the saturated Fe₃O₄@PA system.

3.2.2. Adsorption Isotherms and Kinetics

The adsorption isotherms of paraquat on Fe₃O₄@PA and Fe₃O₄@LA at 25 °C are presented in Figure . Both materials displayed L-type isotherms, characteristic of high affinity adsorption at low equilibrium concentrations followed by gradual site saturation at higher concentrations.

6.

Adsorption isotherms of paraquat on Fe₃O₄@PA and Fe₃O₄@LA at 25 °C with nonlinear Langmuir (solid lines) and Freundlich (dashed lines) fits. Error bars represent ± SD from three independent measurements.

These results highlight the enhanced uptake capacity of the linoleic acid bilayer-coated system, which can be attributed to the unsaturated structure of linoleic acid that provides additional hydrophobic interactions and better surface accessibility in addition to electrostatic attraction with paraquat dications. In contrast, the saturated palmitic acid bilayer (Fe₃O₄@PA) provided lower adsorption affinity, likely due to its more compact structure which limits access to adsorption sites.

Freundlich model fitting yielded values of n > 1 for both adsorbents, indicating favorable adsorption, but the comparatively lower correlation coefficients underscore the predominance of Langmuir-type binding. These findings are consistent with prior reports of fatty acid-modified magnetic nanomaterials, where bilayer unsaturation plays a decisive role in adsorption efficiency.

The experimental data obtained for the adsorption of PQ²⁺on both on Fe₃O₄@PA and Fe₃O₄@LA were fitted with the Langmuir and Freundlich isotherm models (Figure ). Analysis of the data, summarized in Table S3, shows that the Langmuir model provided the best fit for both sorbents, with R ² values of 0.981 (Fe₃O₄@PA) and 0.986 (Fe₃O₄@LA). The Langmuir model also predicts higher maximum adsorption capacities, especially for Fe₃O₄@LA (Q _max = 14.8 ± 0.4 mg g^–1). For Fe₃O₄@LA, the Langmuir model provided the best fit, showing a higher coefficient of determination (R ² = 0.986) compared to the Freundlich model. This strong statistical preference is confirmed by the nonlinear error analysis, where the Langmuir model yielded the lowest Root Mean Square Error (RMSE = 0.3418) as well as the lowest Sum of Squared Errors (SSE = 0.5843) and Chi-square χ² = 0.0652) values, unequivocally validating the monolayer adsorption mechanism. This finding supports the conclusion that the adsorption process involves monolayer coverage of PQ²⁺ onto the active sites.

Although the Langmuir capacity of 14.8 mg g^–1 is moderate relative to advanced

carbonaceous or MOF-based adsorbents, the present system offers key practical advantages including: (i) renewable fatty-acid bilayers, (ii) low-energy continuous-flow synthesis, (iii) rapid magnetic separation, and (iv) excellent multicycle reusability (∼90% retention). These factors make the material attractive for scalable and sustainable paraquat remediation.

The Langmuir constants (K _L) were of the same order of magnitude (0.034–0.041 L mg^–1), suggesting comparable adsorption affinities, though slightly higher for Fe₃O₄@LA. Freundlich fitting yielded n values >1 (2.1 for Fe₃O₄@PA; 3.2 for Fe₃O₄@LA), indicating favorable adsorption on heterogeneous surfaces. However, the lower R² values relative to Langmuir confirm that monolayer adsorption on uniform sites is the dominant mechanism.

Collectively, these results highlight that surface bilayer chemistry strongly influences adsorption efficiency, and that introducing unsaturation in the bilayer (as in LA) enhances paraquat uptake beyond that achievable with saturated fatty acid bilayers (PA).

Bare Fe₃O₄ exhibited negligible adsorption (<1.2 mg g^–1 at 25 °C), consistent with literature reports on unmodified magnetite nanoparticles showing very low paraquat uptake. , This confirms that the fatty-acid bilayer contributes predominantly to paraquat binding.

The kinetic analysis (Table S4) shows that the Pseudo-Second-Order (PSO) model is the most appropriate for both Fe₃O₄@PA and Fe₃O₄@LA (all R ² > 0.987). The superiority of the PSO model is statistically validated by the nonlinear error metrics: it consistently yielded the lowest RMSE values for both sorbents (RMSE = 0.0903 for Fe₃O₄@PA and RMSE = 0.0911 Fe₃O₄@LA) alongside the minimum SSE and χ² values. This strong correlation, coupled with the excellent agreement between q _e , _cal and q _e , _exp , reinforces the conclusion that the rate-limiting step is predominantly controlled by a chemisorption mechanism.

The pseudo-first-order (PFO) and intraparticle diffusion (IPD) models showed lower correlations. This indicates that chemisorption involving valence forces and electron sharing dominates the process. For Fe₃O₄@LA, the calculated q _e values from the PSO model were very close to experimental values, highlighting its higher reliability. Although IPD contributed partially to the uptake, the nonlinear behavior across time suggests that intraparticle diffusion is not the sole rate-controlling step. Overall, adsorption kinetics were strongly influenced by both surface binding and diffusion processes.

The observed kinetic and isotherm behavior can be rationalized by the bilayer mechanism described in Section 3.3.

All adsorption experiments were performed in triplicate, and deviations were within ± 3%, confirming reproducibility. Statistical comparison (one-way ANOVA, p > 0.05) indicated no significant deviation among replicates.

3.2.3. Effect of pH and Ionic Strength

The influence of pH and ionic strength on paraquat adsorption is shown in Figure . As illustrated in Figure a, adsorption efficiency exhibited a strong pH dependence. Both Fe₃O₄@LA and Fe₃O₄@PA achieved maximum removal efficiency in the near-neutral range (pH 5–7), where electrostatic attraction between the positively charged paraquat dications (PQ²⁺) and negatively charged bilayer-coated Fe₃O₄ surfaces is maximized. Although zeta potential was not measured directly, literature values for fatty-acid-coated Fe₃O₄ typically range from −25 to −40 mV at neutral pH, ,,, consistent with strong electrostatic attraction to PQ²⁺. At strongly acidic conditions, partial protonation of the surface may hinder interaction with PQ²⁺, while under alkaline conditions, the adsorption efficiency decreased sharply due to competition with hydroxide ions and reduced electrostatic interactions. Across all tested pH values, Fe₃O₄@LA consistently outperformed Fe₃O₄@PA, highlighting the role of its unsaturated bilayer in enhancing hydrophobic interactions and augmenting electrostatic attraction.

7.

(a) Effect of initial pH on the adsorption efficiency and equilibrium capacity (q _e ± SD) of Fe₃O₄@PA and Fe₃O₄@LA for paraquat removal. (b) Influence of NaCl concentration on paraquat adsorption, illustrating the ionic-strength effect on electrostatic and hydrophobic interactions within the bilayer-coated systems. Solid lines represent q _e (mg g^–1), dashed lines represent Removal (%). Error bars denote ± SD (n = 3). Fe₃O₄@PA = red, Fe₃O₄@LA = blue.

The effect of ionic strength is depicted in Figure b. Increasing NaCl concentration (0–0.10 mol L^–1) led to a progressive reduction in adsorption capacity and removal percentage for both adsorbents. This behavior is attributed to electrostatic screening of attractive forces by Na⁺ and Cl^– ions and to direct competitive adsorption of cations, which reduce available active sites for PQ²⁺. Nonetheless, Fe₃O₄@LA retained a higher adsorption capacity than Fe₃O₄@PA even at elevated ionic strength, underscoring the added contribution of nonelectrostatic interactions in stabilizing PQ²⁺ uptake.

Although multisolute competition tests were not performed, the pH and ionic-strength dependences in Figure serve as reliable proxies for paraquat selectivity. The derived indices (SI_pH and SI_NaCl; see Supporting Information, Tables S6 and S4) exhibit monotonic trends and slightly higher values for Fe₃O₄@LA, corroborating its stronger electrostatic and hydrophobic affinity toward PQ²⁺.

3.2.4. Thermodynamic Analysis

The thermodynamic behavior of paraquat adsorption on Fe₃O₄@PA and Fe₃O₄@LA was evaluated from van’t Hoff plots (Figure , Table S5). To correct the thermodynamic analysis, a dimensionless equilibrium constant Kc = q _e /C _e was used to construct van’t Hoff plots of ln K _c versus 1/T. Linear fitting of ln Kc versus 1/T (298–323 K) allowed determination of ΔH° and ΔS°, and the Gibbs free energy was calculated using ΔG° = – RT ln Kc.

8.

Van’t Hoff plots (ln K _c/1/T) for paraquat adsorption on Fe₃O₄@PA and Fe₃O₄@LA at C ₀ = 20 mg L^–1 and a dosage of 2.0 g L^–1 (pH 7, 24 h). Linear fits of these data provide ΔH° and ΔS°; negative values confirm an exothermic process and reduced interfacial entropy. Error bars denote ± SD from triplicate measurements.

Van’t Hoff plots showed good linearity over 298–318 K for both Fe₃O₄@PA and Fe₃O₄@LA, confirming that the adsorption equilibria follow the van’t Hoff relation. The resulting enthalpy changes (ΔH°) were −47.6 ± 5.6 kJ mol^–1 for Fe₃O₄@PA and – 44.3 ± 2.1 kJ mol^–1 for Fe₃O₄@LA, indicating that paraquat adsorption on both bilayer-coated adsorbents is exothermic. The negative entropy changes (ΔS°) of −163 ± 18 and −145.7 ± 6.8 J mol^–1 K^–1, respectively, suggest decreased disorder at the solid–liquid interface upon adsorption, consistent with the ordering of paraquat molecules within the bilayer coatings.

The standard Gibbs free energies (ΔG°) revealed that the process is slightly spontaneous for Fe₃O₄@LA at 298 K (ΔG° = −0.87 ± 0.07 kJ mol^–1) but becomes less favorable at higher temperatures (ΔG° = 2.04 ± 0.08 kJ mol^–1 at 318 K). Fe₃O₄@PA exhibited slightly positive ΔG° values across the studied range (1.12 ± 0.07 to 4.40 ± 0.10 kJ mol^–1), indicating near-equilibrium behavior that becomes increasingly nonspontaneous with rising temperature. These trends reinforce that adsorption is favored at lower temperatures, as expected for an exothermic process, and that the linoleic acid bilayer (Fe₃O₄@LA) provides stronger interactions with paraquat (reflected by larger K_c values) than the palmitic acid bilayer (Fe₃O₄@PA). The moderate magnitudes of ΔH° (−44 to −48 kJ mol^–1) imply a mixed physisorption–chemisorption mechanism, while the negative ΔS° values reflect structural ordering of paraquat within the fatty-acid bilayers.

3.3. Proposed Adsorption Mechanism

The overall adsorption performance of Fe₃O₄@LA and Fe₃O₄@PA can be rationalized by a synergistic bilayer mechanism linking structural, surface, and magnetic properties to adsorption behavior (Figure ). The inner carboxylate groups (−COO^–) of fatty acids are chemisorbed onto the Fe₃O₄ core via Fe–O–C linkages, forming a robust anchoring layer that ensures chemical stability and magnetic integrity. The outer interdigitated hydrocarbon tails establish hydrophobic microdomains that regulate the accessibility of paraquat molecules. This dual-layer architecture prevents nanoparticle aggregation (TEM, BET) and maintains superparamagnetism (VSM), thereby improving mass transfer during adsorption–desorption cycles.

9.

Synergistic bilayer mechanism of Fe₃O₄@LA via electrostatic and hydrophobic interactions enhancing paraquat adsorption.

Electrostatically, exposed – COO^– groups attract paraquat dications (PQ²⁺), while the unsaturated C = C bonds of linoleic acid may increase local flexibility and hydrophobicity, facilitating closer approach of paraquat molecules for electrostatic and hydrophobic interactions, enhancing hydrophobic affinity and cooperative adsorption. This synergy results in higher K^L, faster k ₂, and more negative ΔG° values for Fe₃O₄@LA compared to Fe₃O₄@PA. Thermodynamic data (Section 3.2.4) further confirm that adsorption is exothermic and chemisorption-assisted.

This structural robustness is consistent with the recyclability results discussed in Section (Figure ). In summary, the cooperative effects between the anchored −COO^– layer, hydrophobic domains, and magnetic stabilization account for the superior adsorption kinetics, capacity, thermodynamic favorability, and recyclability of the linoleic-acid-modified system over the saturated analogue.

10.

(a) Normalized removal efficiency of Fe₃O₄@PA and Fe₃O₄@LA during six consecutive adsorption–desorption cycles. (b) Variation of adsorption capacity (q _e ± SD) over repeated use, demonstrating excellent recyclability and structural stability of both sorbents. Error bars represent standard deviations from triplicate experiments.

Recent reports on biobased magnetite adsorbents indicate that electrostatic interaction between negatively charged surfaces and cationic paraquat plays the primary role in adsorption, whereas hydrophobic interactions contribute secondarily; π–π interactions are negligible when aromatic moieties are absent.

Although direct spectroscopic confirmation of hydrophobic association was not performed, the proposed mechanism is strongly supported by the consistent trends in thermodynamic parameters (ΔH°, ΔS°) and pH-dependent adsorption behavior.

No Raman or XPS evidence was collected; therefore, the adsorption mechanism is interpreted based solely on adsorption thermodynamics, surface chemistry, and established literature trends. Future work will include postadsorption FT-IR or UV–Vis analyses to confirm the specific interactions between paraquat and the LA bilayer.

3.4. Reusability

The recyclability test (Figure and Table S7) demonstrated that both Fe₃O₄@PA and Fe₃O₄@LA maintained high adsorption performance over six adsorption–desorption cycles. Fe₃O₄@LA retained ∼93% of its initial capacity, decreasing slightly from 14.7 ± 0.3 to 13.7 ± 0.4 mg g^–1, while Fe₃O₄@PA retained ∼89% (from 11.5 ± 0.3 to 9.7 ± 0.4 mg g^–1). The gradual decrease may be attributed to minor pore blockage, partial loss of surface-active groups, or incomplete desorption of paraquat molecules.

TGA after multiple cycles was not collected in this study; instead, structural robustness is inferred from the stable adsorption capacities, unchanged macroscopic appearance of the powders, and their rapid magnetic response during separation. These observations indicate that the fatty acid bilayer coatings and magnetic cores remain sufficiently stable for repeated use.

To provide further clarification on regeneration performance without additional experiments, the desorption behavior was indirectly evaluated using the retained adsorption capacity (q _e , n /q _e ,1) after each cycle (Table S8). The high retention (approximately 88–93% for Fe₃O₄@PA and 92–95% for Fe₃O₄@LA) indicates that most of the adsorbed paraquat was successfully removed during each desorption step, and therefore these values serve as reliable proxies for desorption efficiency. This inference is consistent with the stable q _e values reported in Table S7.

UV–Vis spectra of each eluate were not individually recorded. However, the desorption solutions consistently returned to baseline absorbance during regeneration, and the recovered q _e in each cycle showed minimal loss. These two observations strongly suggest negligible paraquat retention on the sorbents after regeneration.

These findings reinforce that fatty acid bilayer-coated Fe₃O₄ nanoparticles are magnetically recoverable, chemically stable, and economically viable for repeated use in water remediation applications.

Although regeneration using 0.5 M NaCl/20% ethanol was found to be highly effective, the environmental and economic implications of ethanol use should be considered. Ethanol contributes to solvent waste and increases regeneration cost relative to aqueous desorption systems. Greener alternatives, such as pure saline or mild buffer solutions, could potentially achieve comparable desorption efficiency while minimizing environmental impact. Future studies will therefore focus on optimizing solvent-free or low-ethanol regeneration protocols to enhance the sustainability of the overall adsorption–desorption process.

In operational terms, this translates to high per-cycle effective capacity (≈0.90–0.93×q _e ) and substantial cumulative removal per gram over six cycles. Thus, even with moderate single-cycle Q _max, the multicycle throughput and magnetic separability provide clear advantages for sustained water treatment.

3.5. Comparison with Literature

Table compares the adsorption performance of bilayer-coated Fe₃O₄ nanoparticles developed in this study with representative adsorbents reported in the literature.

1. Comparison of Paraquat Adsorption Capacities of Various Adsorbents.

Adsorbent System [Reference]	Adsorption capacity ( mg g – 1 )	Notes/Remarks
Fe3O4@LA (this work)	14.8	Unsaturated bilayer; electrostatic + hydrophobic interactions; magnetically recoverable and reusable ≥ 6 cycles
Fe3O4@PA (this work)	11.2	Saturated bilayer; compact structure reduces sites; magnetic separation possible ≥ 6 cycles
nZVI–Diatomite (D-nZVI)	161.3	High capacity; multistep synthesis; not magnetically recoverable
nZVI–Pumice (P-nZVI)	169.5	High adsorption; higher cost; no simple separation method
Carbon nanotubes from biomass	218.6	Very high capacity; expensive, low recyclability; no magnetic separation
Activated clay	58.5	Abundant, cheap, but limited selectivity; no recyclability
Activated date stone carbon	55.6	Renewable source, moderate adsorption, long equilibration
Polyester textile + cyclodextrin polymer	24.2	Functionalized textile; long equilibrium time (420 min)
Bentonite/zerovalent iron	6.8	Very low capacity; fast adsorption; not magnetically separable
Biosorbent (Ayous sawdust)	∼30	Cheap natural sorbent; moderate efficiency; no magnetic properties
Fe3O4@SiO2/SiCRG (κ-carrageenan hybrid)	257	Very high uptake; electrostatic binding via sulfonate groups; magnetically separable, regenerable
Fe3O4@SiO2/SiStarch	<5 (low removal)	Poor performance; weak electron density of OH groups; magnetic recovery possible
Magnetic biochar (corn husk, MB-01–850)	34.97 (synthetic water) 31.63 (natural water)	Mesoporous structure; sustainable, low-cost; magnetically recoverable; reusable ≥ 4 cycles

Although direct numerical comparison of adsorption capacities across different pollutants can be misleading due to variations in adsorbate identity, pH, ionic strength, and medium composition, a qualitative benchmark against recent Fe-based adsorbents helps contextualize the current study. Recent multifunctional Fe-based systems, such as Fe₃O₄/TiO₂/NH₂–UiO–66 composites for phosphate adsorption, superparamagnetic magnetite nanoparticles for As­(III) removal using Ulva prolifera extract, and biomass-derived magnetic activated carbon for 2,4-dichlorophenoxyacetic acid (2,4-D) uptake, have demonstrated high adsorption capacities (>100 mg g^–1) toward their specific targets. However, these materials often require multistep synthesis, costly precursors, or lack facile magnetic recyclability. In contrast, the present Fe₃O₄@LA and Fe₃O₄@PA systems emphasize green continuous-flow fabrication, renewable fatty-acid ligands, and bilayer synergy that collectively balance sustainability, adsorption efficiency, and reusability for paraquat remediation.

The Fe₃O₄@LA and Fe₃O₄@PA systems exhibited moderate adsorption capacities of 14.8 ± 0.4 and 11.2 ± 0.3 mg g^–1, respectively. Although these values are lower than those of advanced nanocomposites, the bilayer-coated materials offer distinct advantages in terms of facile magnetic recovery, reusability over multiple cycles, and the use of naturally derived fatty acids as sustainable surface modifiers. Notably, Fe₃O₄@LA consistently outperformed Fe₃O₄@PA due to the presence of conjugated double bonds, which provide additional electrostatic and hydrophobic interactions with paraquat cations.

By comparison, nZVI-based composites (e.g., diatomite and pumice supported systems) and biomass-derived carbon nanotubes exhibited much higher adsorption capacities (≥160 mg g^–1), reflecting strong surface reactivity and large accessible surface area. However, these systems generally require complex or costly synthesis and lack simple magnetic recyclability, which limits their practical scalability.

Conventional adsorbents, such as activated clay (58.5 mg g^–1), date stone carbon (55.6 mg g^–1), and Ayous sawdust (∼30 mg g^–1), provided moderate removal efficiency. While abundant and inexpensive, such materials are hindered by limited selectivity and regeneration difficulties. Recent advances in magnetic biochars (31–35 mg g^–1) represent an improvement, combining low-cost feedstocks with magnetic recoverability, although their adsorption capacity remains below that of engineered nanocomposites.

Hybrid magnetic systems, such as Fe₃O₄@SiO₂/SiCRG, achieved exceptionally high adsorption capacity (257 mg g^–1) through strong electrostatic binding from sulfonate groups, while retaining magnetic separability and reusability. In contrast, Fe₃O₄@SiO₂/SiStarch exhibited poor performance (<5 mg g^–1), highlighting the importance of surface chemistry in dictating adsorption behavior.

Taken together, these comparisons demonstrate that fatty acid bilayer-coated Fe₃O₄ nanoparticles, despite their moderate adsorption capacity, provide a unique balance of green synthesis, magnetic separability, reusability, and cost-effectiveness. These attributes make them attractive candidates for sustainable water treatment applications, particularly in contexts where operational simplicity and recyclability are prioritized over maximum adsorption capacity.

Accordingly, these comparisons demonstrate that the present bilayer-coated Fe₃O₄ system prioritizes sustainability and reusability, while maintaining sufficient adsorption efficiency for PQ²⁺, which differentiates it from MOF- or biochar-based Fe systems designed for other pollutants.

Although the adsorption capacity is moderate, the synthesis employs low-cost feedstocks (fatty acids ≈ 0.5 USD kg^–1) and requires significantly less energy than multistep sol–gel or hydrothermal routes, resulting in an estimated 70–80% reduction in overall material and energy cost per adsorption cycle.

It should be noted that all adsorption experiments in this study were performed using synthetic paraquat solutions to ensure well-controlled conditions for kinetic and thermodynamic analyses. However, natural water matrices such as agricultural runoff may contain competing ions, dissolved organic matter, and suspended solids that could influence adsorption performance. Future work will therefore focus on evaluating Fe₃O₄@LA and Fe₃O₄@PA using real water samples to assess matrix effects and practical field applicability.

Although the Langmuir capacities of Fe₃O₄@LA and Fe₃O₄@PA (14.8 and 11.2 mg g^–1, respectively) are moderate, their advantages lie in rapid magnetic recovery, excellent reusability (≈90–93% retention after six cycles), and a green, low-cost continuous-flow synthesis using natural fatty acids. These attributes make the materials operationally and economically superior to many high-capacity but complex or nonrecyclable systems.

4. Conclusions

We have successfully demonstrated the synthesis of bilayer-coated Fe₃O₄ nanoparticles using linoleic and palmitic acids via a continuous-flow process. The bilayer coating significantly enhanced colloidal stability and adsorption efficiency toward paraquat compared to bare Fe₃O₄. Among the two systems, Fe₃O₄@LA exhibited superior adsorption performance, attributed to enhanced electrostatic and hydrophobic interactions provided by the unsaturated fatty acid chains. Adsorption studies confirmed monolayer binding consistent with the Langmuir model, pseudo-second-order kinetics, and a spontaneous, exothermic process. Importantly, both Fe₃O₄@LA and Fe₃O₄@PA retained high structural integrity and adsorption efficiency over multiple regeneration cycles, highlighting their robustness and scalability. These findings establish fatty acid bilayer-coated Fe₃O₄ nanoparticles as green, magnetically recyclable adsorbents with strong potential for practical paraquat remediation in contaminated water systems. The superior adsorption efficiency of Fe₃O₄@LA is therefore attributed to the synergistic bilayer structure, which combines electrostatic and hydrophobic effects in a cooperative manner.

Future research should explore the adaptability of this bilayer platform by varying fatty-acid chain length or unsaturation to target other pollutants (e.g., 2,4-dichlorophenoxyacetic acid, tetracycline, and dyes). In addition, integrating the bilayer-coated Fe₃O₄ with high-surface-area supports such as biochar or polysaccharide–silica composites could enhance capacity and selectivity. Applying machine-learning models to predict breakthrough behavior in dynamic column systems may further guide scale-up and optimization for real-world applications.

Despite moderate Qmax values, the Fe₃O₄@LA and Fe₃O₄@PA sorbents demonstrate practical superiority through fast magnetic separability, multicycle durability, and sustainable fabrication, providing a balanced approach between efficiency, cost, and environmental safety.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c09298.

• Additional experimental details, synthesis conditions, characterization data (XRD, VSM hysteresis loops), adsorption isotherm and kinetic parameters, thermodynamic calculations, recyclability data (PDF)

The authors declare no competing financial interest.