Complex Effects of Functional Groups on the Cotransport Behavior of Functionalized Fe₃O₄ Magnetic Nanospheres and Tetracycline in Porous Media

Abstract

In this study, four types of Fe₃O₄-based magnetic nanospheres were functionalized with distinct surface groups to examine how surface chemistry influences their co-transport with tetracycline (TC) in porous media. The functional groups investigated are carboxyl (−COOH), epoxy (−EPOXY), silanol (−SiOH), and amino (−NH₂). Particles bearing −COOH, −EPOXY, or −SiOH are negatively charged, facilitating their transport through porous media, whereas −NH₂-modified particles acquire a positive charge, leading to strong electrostatic attraction to the negatively charged TC and quartz sand, and consequently substantial retention with reduced mobility. Adsorption of TC onto Fe₃O₄-MNPs is predominantly chemisorptive, driven by ligand exchange and the formation of coordination complexes between the ionizable carboxyl and amino groups of TC and the surface hydroxyls of Fe₃O₄-MNPs. Additional contributions arise from electrostatic interactions, hydrogen bonding, hydrophobic effects, and cation–π interactions. Moreover, the carboxylate moiety of TC can coordinate to surface Fe centers via its oxygen atoms. Molecular dynamics simulations reveal a hierarchy of adsorption energies for TC on the differently modified surfaces: Fe₃O₄-NH₂ > Fe₃O₄-EPOXY > Fe₃O₄-COOH > Fe₃O₄-SiOH, consistent with experimental findings. The results underscore that tailoring the surface properties of engineered nanoparticles substantially modulates their environmental fate and interactions, offering insights into the potential ecological risks associated with these nanomaterials.

1. Introduction

Engineered nanoparticles (ENPs) refer to nanoscale particulates or their aggregates, with the majority of individual particles occupying the 1–100 nm size range [1]. Owing to their high reactivity, ENPs hold substantial promise for in situ groundwater remediation. However, their mobility in natural porous media is often limited, presenting a major constraint for field implementation. Consequently, advancing understanding of the fate and transport of ENPs in subsurface environments and developing strategies to enhance their transport are essential research priorities [2]. Moreover, released ENPs may act as sources of emerging contaminants, raising concerns about potential ecological and human health risks [3]. In the context of groundwater exposure, the risk posed by ENPs is governed by their fate and transport within natural porous media [4]. Magnetite (Fe₃O₄) is among the most prevalent iron oxides found in nature. Fe₃O₄ magnetic nanospheres (Fe₃O₄-MNPs) exhibit notable properties, including strong magnetism, a large specific surface area, and good biocompatibility, enabling diverse applications in sensing, pigment production, biomedical imaging, ferrofluid technology, and environmental remediation [5,6,7]. Nevertheless, widespread use leads to their release into aquatic and terrestrial environments through various anthropogenic pathways [8]. The presence of Fe₃O₄-MNPs in the environment can pose risks to aquatic life and human health, potentially affecting microbial community structure, contaminant chemistry, as well as the bioavailability and transport of both contaminants and nutrients [9]. Interactions among nanoparticles and between nanoparticles and dissolved or colloidal water constituents can drive the formation or dissociation of aggregates, contingent on physicochemical properties such as morphology, surface charge, and distribution of functional groups. The formation and breakdown of aggregates constitute among the most critical surface-driven phenomena encountered in both aquatic and terrestrial environments, and they govern many key environmental processes [10].

Tetracyclines (TC) constitute a class of antibiotics extensively employed in agriculture, animal health therapy, and human medicine [11,12,13]. Consequently, TC are widely present in the environment. More than 70% of TC are excreted and released into the environment in their active form via urine and feces [11,12,13]. Their high hydrophilicity and low volatility confer significant persistence in aquatic environments, leading to environmental and potential human health concerns, including ecological risks [14]. In heterogeneous subsurface settings, processes such as migration and transformation render antibiotic contaminants in groundwater more prone to accumulation than in other aqueous systems, complicating remediation [15]. Additionally, TC residues can promote the evolution of drug-resistant microorganisms; prolonged exposure to low antibiotic levels in the environment can foster antibiotic resistance genes (ARGs) and resistant bacteria in humans, posing health threats [16,17]. TC is an amphiphilic molecule bearing multiple ionizable functional groups (e.g., carboxyl, phenolic, ketonic, and amino groups) that are expected to interact with charged and highly polar substances [18,19]. Significant residues of TC have been detected in surface water, groundwater, sediments, and soils [14]. As a polar, ionizable organic contaminant, the fate and behavior of TC in aquatic environments are largely governed by adsorption to fixed sediments and mobile colloids [20,21,22,23]. Consequently, there exists substantial potential for interactions between dispersed Fe₃O₄-NPs and residual TC in aquatic environments, which may affect their degradation and transport in porous media.

Several studies have separately investigated the transport of iron oxides and antibiotics in porous media. Evidence indicates that iron oxides with varying physicochemical properties (for example, aggregation behavior, distribution of surface functional groups, and surface charge) can modulate interactions with antibiotics. Zhang et al. [24] demonstrated that functional groups on microplastics influence their aggregation, transport, and interactions with contaminants in environmental media; however, there is a lack of systematic work on how different functional groups affect the migration of iron oxides in porous media. In environmental contexts, iron oxides tend to carry a positive charge due to a relatively high point of zero charge [25,26,27,28], making electrostatic attraction a key mechanism for tetracycline (TC) adsorption, particularly between anionic TC and positively charged iron oxides [29]. Additionally, metal cations can influence TC transport and deposition via cation-bridging mechanisms in aqueous environments [30,31]. To date, no studies have examined the co-transport behavior of Fe₃O₄-MNPs modified with distinct functional groups alongside TC in porous media. Carboxyl and amino groups are the two most common functional groups, representing negative and positive charges, respectively [24]. Considering hydrophilicity, epoxy and silanol groups have also been included in some studies [32]. The co-transport of two constituents is substantially more complex than the transport of a single component, and investigations into the co-transport of magnetic nanoparticles are comparatively scarce. Consequently, there is a need to develop more comprehensive and detailed experimental and modeling approaches to elucidate the transport behavior and underlying mechanisms of magnetic nanoparticles coexisting with antibiotics under various surface functionalizations.

Due to the extremely low mobility of positively charged Fe₃O₄-MNPs colloids in porous media, four variants of Fe₃O₄-MNPs functionalized with different groups were employed in this study: carboxyl (−COOH), epoxy (−EPOXY), silanol (−SiOH), and amino (−NH₂). Under typical environmental conditions, all but the -NH2 modified particles carry a negative surface charge, which enhances their transport through porous media. Direct evidence for Fe₃O₄-MNPs–TC interactions was obtained from column experiments, complemented by characterization using scanning electron microscopy (SEM, Ultra-55), ultraviolet-visible spectroscopy (UV), X-ray photoelectron spectroscopy (XPS), and Fourier transform infrared spectroscopy (FTIR). Extended Derjaguin–Landau–Verwey–Overbeek (XDLVO) calculations were performed to assess the roles of magnetic interactions and gravity on both nanosphere–nanosphere and nanosphere–sand interactions. Given the limitations of DLVO theory for non-colloidal substances, molecular dynamics (MD) simulations were conducted to elucidate the contribution of non-DLVO forces to the deposition behavior of Fe₃O₄-MNPs and TC in porous media.

2. Materials and Methods

2.1. Experimental Materials

To investigate the effects of surface functionality, four Fe₃O₄-MNP variants bearing distinct groups were employed: carboxyl (−COOH), epoxy (−EPOXY), silanol (−SiOH), and amino (−NH₂). The substrates comprised Fe₃O₄-MNPs (particle size ~100–200 nm, initial concentration 5 mg mL⁻¹) and tetracycline (TC, purity ≥ 98%). For stock solutions, 100 mg of each chemical were dissolved in 1000 mL of deionized water to yield 100 mg L⁻¹ stock solutions for both Fe₃O₄-MNPs and TC. The stock solutions were subjected to ultrasonic agitation for 2 h to ensure thorough dispersion and stored at 4 °C. Prior to experiments, the stock solutions were diluted with DI water to attain the required working concentrations.

The porous medium used in the column experiments was quartz sand obtained from Longxin Water Purification Materials Co., Ltd. (Gongyi City, China). The sand underwent a preparatory sequence comprising initial rinsing with tap water, immersion in 10% nitric acid to remove surface contaminants and activate the material, rinsing with deionized water until the effluent reached neutral pH, and oven drying at 105 °C prior to use.

To clarify the surface properties of functionalized Fe₃O₄-MNPs, their composition, morphology, and charge were analyzed via multiple techniques: Fourier transform infrared spectroscopy (FTIR, Nicolet iS50, Thermo Fisher Scientific, North Brunswick Township, USA) for functional group identification, X-ray photoelectron spectroscopy (XPS, ESCALAB 250Xi, Thermo Scientific, Manchester, UK) for elemental composition, and scanning electron microscopy (SEM, JSM-7800 F, JEOL, Tokyo, Japan) operated at 2.0 kV accelerating voltage (selected to avoid particle surface damage) for morphological observation. Hydrodynamic diameters of Fe₃O₄-MNPs, tetracycline (TC), and their mixtures at varying mixing ratios were determined by dynamic light scattering (DLS) using a laser-based nanoparticle-size analyzer. In addition, the zeta potentials of Fe₃O₄-MNPs, TC, and their mixtures at the same mixing ratios, as well as the surface of the quartz sand, were measured by electrophoretic methods.

2.2. Cotransport Experiments

To elucidate the co-transport of Fe₃O₄-MNPs and tetracycline (TC) in saturated porous media under varying experimental conditions, column-based transport experiments were conducted. The experiments utilized a glass column with an inner diameter of 2.5 cm and a length of 20 cm (Figure S1). Prior to testing, the inner surface of the column was smoothed by sandpaper polishing to minimize flow heterogeneity and prevent preferential flow. Wet packing was adopted to ensure uniform quartz sand distribution (to avoid preferential flow). First, ultrapure water was added to the column. Afterward, quartz sand was added in columns (20 g per batch) while maintaining the water level above the sand surface. During each addition, a glass rod was used to stir the sand to expel air, and a small hammer was applied for mild compaction (controlling bulk density at 1.65 g·cm⁻³) until the column was fully packed.

To evaluate the uniformity of column packing and to exclude potential contaminants that could affect subsequent migration experiments, the packed column was flushed with 2 pore volumes of ultrapure water to remove residual impurities. A conservative tracer test using nitrate (NO₃⁻) was conducted to confirm complete packing and to ensure that no extraneous contaminants would influence the migration measurements. The column was sequentially injected with 1 pore volume of a 0.1 mmol·L⁻¹ KNO₃ solution, followed by 8 PV of ultrapure water, using a calibrated peristaltic pump. Effluent samples were automatically collected and archived by an autosampler throughout the experiment. KNO₃ concentrations in the effluent were quantified by UV–visible spectrophotometry at 220 nm.

Following the completion of tracer experiments for each column, 2 pore volumes of background solution were pumped to re-establish equilibrium in the column water chemistry. Subsequently, 2 pore volumes of the target solution, comprising a mixture of Fe₃O₄-MNPs and tetracycline (TC), were injected, followed by 5 pore volumes of background solution to flush the system. Seven pore volumes of effluent were collected with an automated sampler. Concentrations of TC and the four different functional-group–modified Fe₃O₄-MNPs were quantified using UV–visible spectrophotometry at 275 nm and 430 nm, respectively. Method details for concentration determination are provided in the Supplementary Materials. The experimental conditions are summarized in Table S1.

2.3. Interaction Energy

The conventional Derjaguin–Landau–Verwey–Overbeek (DLVO) framework is employed to estimate the interaction energy between charged colloids and the surface of the medium, thereby assessing system stability (Figure S2). The DLVO interaction energy comprises van der Waals attraction (Φ_VDW) and electrostatic double-layer repulsion (Φ_EDL). Additionally, the magneto-gravitational energy (Φ_VM) associated with magnetic nanoparticles is incorporated into the classical DLVO description to compute the interaction energy among Fe₃O₄-MNPs, enabling a quantitative analysis of colloidal aggregation and dispersion behavior [12].

Moreover, gravitational forces are recognized as important determinants of colloidal deposition on the surfaces of porous media [4,33]. The extended DLVO (XDLVO) framework is employed to quantify the interaction energies among Fe₃O₄-MNPs and between Fe₃O₄-MNPs and the porous matrix, with the corresponding derivations and numerical implementations detailed in the Supplementary Materials.

2.4. Molecular Dynamics Simulation

In this study, all-atom molecular dynamics (MD) simulations were conducted to investigate the chemical and physical adsorption of tetracycline (TC) on four surfaces of functional-group-modified Fe₃O₄-MNPs. Simulations were carried out using Materials Studio (MS) from Accelrys. The COMPASS force field—condensed-phase optimized molecular potentials for atomistic simulation studies—was employed to describe interatomic interactions. COMPASS is a general-purpose, all-atom force field developed with state-of-the-art ab initio and empirical parameterization techniques and has been validated across a broad range of systems. Owing to its capability to accurately model interactions in organic–inorganic environments, COMPASS was selected to characterize the adsorption reactions of TC on functional-group-modified Fe₃O₄-MNPs. This force field has been widely used in MD calculations for organic small molecules and metal oxides [33,34]. The functional form of the COMPASS force field is provided below [35,36]:

$${\chi }_{total}={\chi }_a+{\chi }_b+{\chi }_c+{\chi }_d+{\chi }_e+{\chi }_f+{\chi }_g+{\chi }_h+{\chi }_i+{\chi }_j+{\chi }_k+{\chi }_l+{\chi }_m$$

(1)

$${\chi }_a={\sum }_b\left[K_2{\left(b-b_0\right)}^2+K_3{\left(b-b_0\right)}^2+K_4{\left(b-b_0\right)}^2\right]$$

(2)

$${\chi }_b={\sum }_{\theta }\left[H_2{\left(\theta -{\theta }_0\right)}^2+H_3{\left(\theta -{\theta }_0\right)}^2+H_4{\left(\theta -{\theta }_0\right)}^2\right]$$

(3)

$${\chi }_c={\sum }_{\mathsf{\varnothing }}\left[V_1\left(1-cos\left(\mathsf{\varnothing }-{\mathsf{\varnothing }}_1^0\right)\right)+V_2\left(1-cos\left(2\mathsf{\varnothing }-{\mathsf{\varnothing }}_2^0\right)\right)+V_3\left(1-cos\left(3\mathsf{\varnothing }-{\mathsf{\varnothing }}_3^0\right)\right)\right]$$

(4)

$${\chi }_d={\sum }_{\chi }{K}_{\chi }{\left(\chi -{\chi }_0\right)}^2$$

(5)

$${\chi }_e={\sum }_b{\sum }_{b^{\prime }}{F}_{b{b}^{\prime }}\left(b-b_0\right)\left(b^{\prime }-{b^{\prime }}_0\right)$$

(6)

$${\chi }_f={\sum }_{\theta }{\sum }_{{\theta }^{\prime }}{F}_{\theta {\theta }^{\prime }}\left(\theta -{\theta }_0\right)\left({\theta }^{\prime }-{{\theta }^{\prime }}_0\right)$$

(7)

$${\chi }_g={\sum }_b{\sum }_{\theta }{F}_{b\theta }\left(b-b_0\right)\left(\theta -{\theta }_0\right)$$

(8)

$${\chi }_h={\sum }_b{\sum }_{\mathsf{\varnothing }}\left(b-b_0\right)\left(V_{1}cos\mathsf{\varnothing }+V_{2}cos2\mathsf{\varnothing }+V_{3}cos3\mathsf{\varnothing }\right)$$

(9)

$${\chi }_i={\sum }_{b^{\prime }}{\sum }_{{\mathsf{\varnothing }}^{\prime }}\left(b^{\prime }-{b^{\prime }}_0\right)\left(V_{1}cos\mathsf{\varnothing }+V_{2}cos2\mathsf{\varnothing }+V_{3}cos3\mathsf{\varnothing }\right)$$

(10)

$${\chi }_j={\sum }_{\theta }{\sum }_{\mathsf{\varnothing }}\left(\theta -{\theta }_0\right)\left(V_{1}cos\mathsf{\varnothing }+V_{2}cos2\mathsf{\varnothing }+V_{3}cos3\mathsf{\varnothing }\right)$$

(11)

$${\chi }_k={\sum }_{\mathsf{\varnothing }}{\sum }_{\theta }{\sum }_{{\theta }^{\prime }}{K}_{{\mathsf{\varnothing }\theta \theta }^{\prime }}cos\mathsf{\varnothing }\left(\theta -{\theta }_0\right)\left({\theta }^{\prime }-{{\theta }^{\prime }}_0\right)$$

(12)

$${\chi }_l={\sum }_{i>j}{\displaystyle \frac{q_i{q}_j}{\epsilon r_{ij}}}$$

(13)

$${\chi }_m={\sum }_{i>j}[{\displaystyle \frac{A_{iJ}}{{r_{ij}}^9}}-{\displaystyle \frac{B_{iJ}}{{r_{ij}}^6}}]$$

(14)

The total potential energy (χ_total) contains the bond stretching (χ_a), angular bending (χ_b), torsion (χ_c), out-of-plane coordinates (χ_d), cross-coupling (χ_e~χ_k), Coulombic interactions (χ_l), and van der Waals interactions (χ_m) terms. K, H, F, and V are the force field parameters; b, θ, ∅, and χ represent the bond length, bending angle, torsion angle, and out-of-plane angle, respectively; the parameters b₀, θ₀, ∅₀ and χ₀ are ideal values at zero energy.

The outermost layer of the oxide substrate was hydroxylated by terminating the surface oxygen atoms in the top layer with hydrogen. Subsequently, the hydroxyl groups were randomly replaced by four functional groups-carboxyl (−COOH), epoxy (−EPOXY), silanol (−SiOH), and amino (−NH₂)-to simulate surfaces modified with these functionalities. The Fe₃O₄ (0 0 1) surface was constructed by cleaving the crystal along the (001) plane and subjected to geometry optimization via a two-stage scheme: an initial optimization using the Steepest Descent method with a convergence criterion of 1000 kcal·mol⁻¹, followed by refinement with the Conjugate Gradient method to a convergence criterion of 10 kcal·mol⁻¹. Convergence was achieved before 10,000 iterations.

In the initial configuration, five energy-minimized tetracycline (TC) molecules were placed on the Fe₃O₄ (0 0 1) surface, with full three-dimensional periodic boundary conditions applied in the simulation box and the z axis aligned with the surface normal. The height of the simulation cell was fixed at 20 Å. Solvation was achieved with approximately 150 water molecules, and the integration time step was set to 0.25 fs [37]. The Fe₃O₄ (0 0 1) surface was restrained prior to the simulation. MD simulations of TC adsorption on Fe₃O₄ (0 0 1) surfaces modified with different functional groups were conducted at 298 K under the NVT ensemble. Temperature was controlled by the Andersen thermostat, with a time step of 1.0 fs. A total MD duration of 2000 ps was performed, capturing the trajectories of all atoms in the TC–Fe₃O₄ system. Equilibration of the interfacial systems was achieved within this 2000 ps window. The interfacial interaction energy between TC and the Fe₃O₄ (0 0 1) surfaces was computed to quantify the adhesion strength.

3. Results and Discussion

3.1. Characterization of Materials

Figure 1 displays SEM images of four types of Fe₃O₄-MNPs modified with distinct functional groups. All particles are approximately spherical. The carboxyl (−COOH) and epoxy (-EPOXY) modified particles show relatively uniform dispersion with slight aggregation, whereas the silanol (-SiOH) modified particles appear more dispersed. In contrast, the amino (-NH2) modified particles exhibit pronounced agglomeration, in agreement with the observed hydrodynamic diameters. At equal concentrations, the hydrodynamic diameters are 1874.93 nm for -COOH, 1755.26 nm for -EPOXY, 783.31 nm for -SiOH, and 2523.18 nmfor -NH₂. Zeta potentials and hydrodynamic diameters under varying conditions are provided in Table S2. Notably, the. significant agglomeration of Fe₃O₄-NH₂ particles (Figure 1d) is a typical manifestation of the synergistic effect between their surface charge properties and the inherent magnetism of Fe₃O₄-NH₂. According to the zeta potential data in Table S2, Fe₃O₄-NH₂ is the only positively charged particle. The positively charged surface significantly weakens the electrostatic repulsion between particles compared to negatively charged particles. Meanwhile, Fe₃O₄-NH₂ itself has strong magnetism. Under the dual effects, particles tend to agglomerate, forming obvious aggregates.

The FTIR spectral images of the four different functional group-modified Fe₃O₄-MNPs before and after TC adsorption are shown in Figure 2. The spectrograms under different aqueous solution mixing ratios primarily exhibit three bands at 1112 cm⁻¹, 1637 cm⁻¹, and 3421 cm⁻¹. During the modification of Fe₃O₄-MNPs, their surfaces easily saturate with hydroxyl groups in aqueous environments [38]; therefore, the spectral bands at 1112 cm⁻¹ and 3421 cm⁻¹ are surmised to correspond to the bending of =Fe-OH and the stretching vibrations of -OH or -NH [34,39]. A weak COO- band intensity can be observed at 1403 cm⁻¹, while the more intense band near 1637 cm⁻¹ may be due to the stretching of carbon double bonds (C=O) [33]. A comparison of the two plots in Figure 2 indicates that the intensities of the spectral bands at 1112 cm⁻¹, 1637 cm⁻¹, and 3421 cm⁻¹ decrease after the addition of TC to the Fe₃O₄-MNPs solution, suggesting that TC adsorbs onto the surface of Fe₃O₄-MNPs and indicates that a certain degree of chemical reaction occurs.

TC is an amphiphilic molecule with multiple ionizable functional groups (Figure S3). TC1, TC2, and TC3⁺ represent the three protonated functional groups in the TC molecule (the tricarbonylmethane group, phenol ketone group, and dimethylamino group), respectively. In an aqueous environment, these groups can undergo protonation-deprotonation reactions to form cationic species (pH<3.3), amphipathic species (3.3<pH<7.68), and anionic species (pH>7.68) [40]. The most characteristic peaks of TC are found between 1200 and 1700 cm⁻¹. The bands at 1637 cm⁻¹, 1614 cm⁻¹, and 1587 cm⁻¹ can be attributed to amides, C=O bond stretching at ring A, and C=O bond stretching at ring C, respectively. The spectrum also displays bands at 1514 cm⁻¹ and 1456 cm⁻¹, attributed to amide and C-C bond stretching vibrations. The bands at 1400 cm⁻¹ and 1230 cm⁻¹ correspond to -CH₃ deformation and C-N bond stretching vibrations, respectively [31].

The elemental composition and functional groups present on the Fe₃O₄-MNPs modified with different functional groups were identified by X-ray photoelectron spectroscopy (XPS), with the XPS scans of the four different functional group-modified Fe₃O₄-MNPs depicted in Figure 3. Three main peaks corresponding to Fe, O, and C are present in the XPS results (Figure 3). Among them, the silanol-based modified Fe₃O₄-MNPs show weak Si atomic intensity at 99.8–103.7 eV, and the amino-modified Fe₃O₄-MNPs exhibit weak N atomic intensity at 397–412 eV. Focusing only on the O1s spectra, the four different functional group-modified Fe₃O₄-MNPs produce peaks of varying intensities at 530 eV, inferred to be caused by lattice oxygen (Fe-O) [41]. The second peak for carboxyl- and silanol-based modified Fe₃O₄-MNPs appears at 532 eV, with the subpeak of carboxyl-modified Fe₃O₄-MNPs inferred to correspond to C=O [42]. Simultaneously, the subpeaks of silanol-based modified Fe₃O₄-MNPs are attributed to Si-O-Si or Si-OH [43]. The second peak of epoxy-modified and amino-modified Fe₃O₄-MNPs appears at 533 eV, with subpeaks of epoxy-modified Fe₃O₄-MNPs deduced as arising from C-O [33]. The subpeaks of amino-modified Fe₃O₄-MNPs are attributed to -OH [44].

3.2. XDLVO Results

To quantify the stability of Fe₃O₄-MNPs in porous media, we applied the Derjaguin–Landau–Verwey–Overbeek (DLVO) theory, which describes interactions between charged colloids via van der Waals attraction (Φ_VDW) and electrostatic repulsion (Φ_EDL), and extended it to XDLVO theory by incorporating magnetic gravitational energy and gravity energy, as Fe₃O₄-MNPs’ strong magnetism significantly affects their aggregation.

To investigate the aggregation and dispersion behavior of strongly magnetic Fe₃O₄-MNPs modified with different functional groups, we calculate the XDLVO interaction energies between Fe₃O₄-MNPs and between Fe₃O₄-MNPs and quartz sand under various experimental conditions (Figures S4 and S5). Figures S4 and S5 illustrate two scenarios: one considering the effect of gravity and the other not. Due to the strong magnetic properties of Fe₃O₄-MNPs, the magnetic attraction is approximately four orders of magnitude larger than the classical van der Waals attraction (Φ_VDW) and double electrostatic repulsion (Φ_EDL). Consequently, Fe₃O₄-MNPs exhibit strong aggregation behavior in aqueous environments.

The results shown in Figures S4 and S5 indicate a general weakening of the magnetic attraction between Fe₃O₄-MNPs as the TC concentration increases, suggesting that TC enhances the dispersion of Fe₃O₄-MNPs in aqueous environments. When gravity is not taken into account, a large positive energy barrier exists between the Fe₃O₄-MNPs and quartz sand, which prevents Fe₃O₄-MNPs from easily depositing on the quartz sand surface. However, due to the strong magnetic mutual attraction among Fe₃O₄-MNPs, the agglomerates possess large diameters and masses, differing from the generally better-dispersed nano-colloids in aqueous environments. As shown in Figure S5, the positive energy barrier between Fe₃O₄-MNPs and quartz sand nearly disappears when considering gravity, making it significantly easier for Fe₃O₄-MNPs to deposit on the surface of quartz sand. This also elucidates the phenomenon of the extremely low mobility of Fe₃O₄-MNPs in porous media.

3.3. The Cotransport of Fe₃O₄ with Different Functional Group Modifications and TC

Previous studies have shown that the adsorption of TC on Fe₃O₄-MNPs is mainly driven by chemisorption rather than nonspecific physisorption [45]. The adsorption of TC on Fe₃O₄-MNPs is attributed to several mechanisms: ligand exchange, chemical complexation between ionizable carboxyl or amino groups of TC and hydroxyl groups of iron oxides, electrostatic interactions, hydrogen bonding interactions, hydrophobic effects, and cation-π interactions [46,47,48,49,50,51]. In addition to these mechanisms, the carboxyl group in TC can coordinate to the magnetite surface in either a monodentate or bidentate manner, with coordination occurring via oxygen atoms [47,52]. The keto groups in TC also interact with the magnetite surface through electrostatic interactions or inner-sphere complexation, further enhancing adsorption efficacy [53].

3.3.1. Monotransport of Fe₃O₄ and TC

Figure S6 shows the breakthrough curves (BTCs) for mono-migration of four different functional group-modified Fe₃O₄-MNPs at a concentration of 50 mg·L⁻¹, as well as those for TC mono-migration at varying concentrations (10, 20, and 50 mg·L⁻¹). The shapes of the BTCs for the four types of Fe₃O₄-MNPs with different functional groups are distinct. For epoxy-modified Fe₃O₄-MNPs, the C/C₀ peak reached 0.89-significantly higher than the other three types, because their epoxy ring hydrolysis exposed -OH groups that reduced electrostatic attraction with quartz sand. Their BTCs showed steady C/C₀ increase after 1 PV, leveling off at 2 PV, which reflects stable transport without obvious retention.

Combined with the XDLVO theoretical analysis (Figures S4 and S5), there is strong mutual attraction between Fe₃O₄-MNPs modified with different functional groups and the quartz sand surface when considering gravitational forces. The order of interaction strength between Fe₃O₄-MNPs and quartz sand is as follows: Fe₃O₄-SIOH < Fe₃O₄-EPOXY < Fe₃O₄-COOH < Fe₃O₄-NH₂. However, as shown in Table S2, due to their strong dispersion in solution, the agglomeration diameter of Fe₃O₄-SIOH is much smaller than that of the other three types of Fe₃O₄-MNPs. This suggests that Fe₃O₄-SIOH, having a larger contact area, is preferentially deposited on the quartz sand surface, leading to a higher amount of deposition and thereby lower mobility.

Since the amino-modified Fe₃O₄-MNPs are positively charged, they exhibit strong electrostatic attraction with the negatively charged TC and quartz sand, resulting in considerable retention within the column and low mobility. The BTCs for TC follow a classical shape, with the peak value of C/C₀ increasing with concentration.

3.3.2. Effect of TC on Fe₃O₄ Transport

Figure 4 shows the BTCs of the co-migration of Fe₃O₄-MNPs modified with different functional groups and TC in porous media, with the associated experimental results presented in Table S3. For the three negatively charged Fe₃O₄-MNPs, the carboxyl and silanol group-modified variants exhibit similar variations with respect to TC concentration. Low TC concentrations promote the mobility of Fe₃O₄-MNPs modified with carboxyl and silanol groups, whereas high TC concentrations inhibit their mobility. In contrast, the epoxy-modified Fe₃O₄-MNPs show a gradual inhibition of changes in TC concentration.

As shown in Table S2, the absolute values of the zeta potential of the mixture of Fe₃O₄-MNPs and TC initially increase, then decrease, and subsequently increase again with rising TC concentration. This suggests that the functional groups on Fe₃O₄-MNPs chemically complex with and electrostatically interact with ionized TC, thereby reducing the electronegativity of TC and the mixture. Once the available reaction sites for TC on the colloidal surface of Fe₃O₄-MNPs become gradually saturated, continued increases in TC concentration lead to a more negative zeta potential for the mixture. As the adsorption of Fe₃O₄-MNPs on TC slows, the relationship between them shifts from mutual agglomeration and co-precipitation to competition for the remaining adsorption sites on the quartz sand.

Fe₃O₄-COOH and Fe₃O₄-SIOH demonstrate a sudden rise with a small peak at 2.8 PV, likely due to saturation of adsorption sites on the surface of quartz sand. The epoxy group-modified Fe₃O₄-MNPs may undergo a nucleophilic reaction in the aqueous environment, leading to the breaking of the oxygen atom, which opens the ring structure and exposes two -OH groups. Therefore, it can be hypothesized that the epoxy group, compared to the carboxyl and silanol groups, is more reactive to TC and thus flows out of the column with TC, exhibiting higher mobility than the other functional group-modified Fe₃O₄-MNPs.

Since the amino-modified Fe₃O₄-MNPs are positively charged, there appears to be strong adsorption between Fe₃O₄-NH₂ and TC, as indicated by the hydrodynamic diameters presented in Table S2 and the molecular dynamics simulation results illustrated in

Table 1. Energies of four different functional group-modified Fe₃O₄-MNPs and TC adsorption systems.

Energy Type	-COOH/TC	-EPOXY/TC	-SIOH/TC	-NH2/TC
	Contributions to Total Energy (kcal/mol):
Total Energy	−1,209,553.541	−1,217,209.116	−1,211,890.528	−1,218,926.599
Valence Energy (diag. terms)	9089.448	15,453.360	10,052.710	3309.592
Bond	2665.652	3497.470	3870.529	2071.052
Angle	6641.437	12,153.575	6560.122	1677.012
Torsion	−239.040	−221.967	−392.768	−461.485
Inversion	21.399	24.282	14.827	23.013
Valence Energy (cross terms)	−152.778	−162.002	−185.945	−170.904
Stretch-Stretch	28.019	−1.092	−5.566	−5.356
Stretch-Bend-Stretch	−127.546	−107.638	−129.826	−114.045
Stretch-Torsion-Stretch	−6.447	−5.760	−5.217	−4.393
Separated-Stretch-Stretch	1.052	0.980	0.874	0.860
Torsion-Stretch	−24.746	−33.142	−24.645	−23.848
Bend-Bend	2.101	2.584	2.033	0.954
Torsion-Bend-Bend	−15.906	−13.288	−13.174	−10.278
Bend-Torsion-Bend	−9.307	−4.646	−10.424	−14.798
Non-bond Energy	−1,218,490.210	−1,232,500.474	−1,221,757.293	−1,222,065.287
Van der waals	−949,359.830	−965,758.372	−953,107.261	−958,078.039
Long range correction	−164.921	−167.228	−166.904	−159.944
Electrostatic	−268,965.460	−266,574.873	−268,483.128	−263,827.304
Adsorption energy	−39,943.840	−40,400.630	−39,180.480	−40,666.840

. Consequently, TC can act as a carrier to enhance the transport capacity of Fe₃O₄-NH₂ in porous media.

3.3.3. Effect of Fe₃O₄ on TC Transport

Due to the stable inner-sphere surface complexation between TC and Fe₃O₄-MNPs, TC exhibits significant hysteresis in TC-iron oxide interactions [51,54]. The deposition of Fe₃O₄-MNPs colloids bound to TC hinders TC migration in porous media. Additionally, these Fe₃O₄-MNPs can inhibit TC transport by providing additional deposition sites.

The carboxyl group-modified Fe₃O₄-MNPs exhibit stronger inhibition of TC mobility as TC concentration increases. In contrast, the other three types of functional group-modified Fe₃O₄-MNPs inhibit TC transport at low concentrations but enhance it at high concentrations. At low TC concentrations, it is hypothesized that TC complexes and aggregates with Fe₃O₄-MNPs co-deposit on the surface of quartz sand, thereby occupying adsorption sites. At high TC concentrations, the adsorption sites available on the surface of Fe₃O₄-MNPs for TC complexation become saturated, leading to competition between Fe₃O₄-MNPs and TC for adsorption sites on quartz sand. Due to their strong magnetic properties, Fe₃O₄-MNPs can deposit more easily onto the surface of quartz sand. The surface of carboxyl-modified Fe₃O₄-MNPs has more available spots for TC reaction than the other three types of Fe₃O₄-MNPs and has not reached full saturation at 50 mg·L⁻¹.

3.4. The Results of Molecular Dynamics Simulation

The potential energy of a system can be expressed as the sum of valence (or bond), cross terms, and non-bonding interactions. Table 1 presents the calculation results of the adsorption energies for four different functionalized Fe₃O₄-MNPs with TC. The order of adsorption energy strength for these Fe₃O₄-MNPs concerning TC is as follows: Fe₃O₄-NH₂ > Fe₃O₄-EPOXY > Fe₃O₄-COOH > Fe₃O₄-SiOH. This suggests that the amino-modified Fe₃O₄-MNPs have the strongest adsorption capacity for TC, primarily due to the strong electrostatic interaction between positively charged Fe₃O₄-NH₂ and negatively charged TC. The epoxy-modified Fe₃O₄-MNPs exhibit the second highest adsorption stability for TC. Our MD simulation results show adsorption energy order Fe₃O₄-NH₂ > Fe₃O₄-EPOXY > Fe₃O₄-COOH > Fe₃O₄-SiOH, which closely match the hydrodynamic diameter data in Table S2 (Fe₃O₄-NH₂ had the largest agglomeration diameter) and the reliability of the adsorption energy sequence. Core drivers of higher adsorption strength: (1) Surface charge-dominated electrostatic interaction-TC was negatively charged due to the dissociation of phenol-ketone groups. Fe₃O₄-NH₂ was the only positively charged particle, and strong electrostatic attraction significantly enhanced the adsorption energy; (2) Differences in functional group reactivity. The epoxy ring of Fe₃O₄-EPOXY exposed two -OH groups after hydrolysis, which could form bidentate coordination and hydrogen bonds with the carboxyl groups of TC. In contrast, Fe₃O₄-COOH had electrostatic repulsion with TC (both negatively charged), relying only on weak hydrogen bonds; Fe₃O₄-SiOH had low hydroxyl density, and its interaction with TC was dominated by van der Waals forces, resulting in the weakest adsorption.

Figure 5 presents a schematic diagram showing the elements, the distribution of electrostatic potentials, and electron densities of different functional groups. The blue region indicates negative potential, while the red region indicates positive potential. Notably, the epoxy group features two -OH group sites, making it more reactive to TC. The molecular dynamics (MD) simulation, which models the surface adsorption of Fe₃O₄-MNPs modified with different functional groups interacting with TC (Figure 6), conforms to the experimental results.

4. Conclusions

This study provides direct experimental evidence for the interactions between functionalized Fe₃O₄ magnetic nanoparticles (Fe₃O₄-MNPs) and tetracycline (TC) through a range of analytical techniques, including column experiments, scanning electron microscopy (SEM), ultraviolet spectrophotometry (UV), X-ray photoelectron spectroscopy (XPS), and Fourier transform infrared spectroscopy (FTIR). The findings highlight that all four types of surface-modified Fe₃O₄-MNPs demonstrate significant aggregation due to their strong magnetic properties, with magnetic attraction surpassing classical van der Waals forces and double electrostatic repulsion by approximately four orders of magnitude.

Notably, under conditions of high TC concentration, the magnetic attraction among Fe₃O₄-MNPs appears to weaken, suggesting that TC facilitates the dispersion of Fe₃O₄-MNPs within aqueous environments. Despite this dispersion, the large diameter and mass of Fe₃O₄-MNPs agglomerates—resulting from their strong magnetic mutual attraction—contrast sharply with the typically well-dispersed nature of nanocolloids in solution. Additionally, the gravitational effects during the deposition of Fe₃O₄-MNPs onto quartz sand surfaces are significant. When these effects are considered, the positive energy barrier between Fe₃O₄-MNPs and quartz sand diminishes nearly to zero, resulting in their facile deposition and markedly reduced mobility within porous media.

The order of interaction strength between the modified Fe₃O₄-MNPs and quartz sand was found to be Fe₃O₄-SiOH < Fe₃O₄-EPOXY < Fe₃O₄-COOH < Fe₃O₄-NH₂. However, due to the high dispersion of Fe₃O₄-SiOH in aqueous environments, its agglomeration diameter remains lower than that of the other functionalized MNPs. The examination of contact area effects reveals that increased deposition of MNPs onto quartz sand surfaces limits their mobility. Furthermore, the strong positive charge of amino-modified Fe₃O₄-MNPs enhances their electrostatic attraction to negatively charged TC and quartz sand, thereby facilitating significant retention within the column.

The adsorption mechanism of TC onto Fe₃O₄-MNPs predominantly involves chemisorption rather than nonspecific physisorption, with the carboxyl group’s coordination to the magnetite surface potentially occurring in monodentate or bidentate forms. Molecular dynamics (MD) simulations corroborate the experimental findings, indicating that the adsorption energies for TC on modified Fe₃O₄-MNPs follow the order: Fe₃O₄-NH₂ > Fe₃O₄-EPOXY > Fe₃O₄-COOH > Fe₃O₄-SiOH. The amino-modified Fe₃O₄-MNPs exhibit the highest adsorption stability, attributed to the robust electrostatic interactions between their positive charges and the negatively charged TC, while the epoxy-modified variants demonstrate the second highest stability. Overall, this research underscores the importance of functional group modification on Fe₃O₄-MNPs in optimizing their interaction with TC, offering insights that could inform future developments in environmental remediation strategies.