Iron (III) chloride hexahydrated (FeCl₃.6H₂O) 98%, Iron (II) sulfate (FeSO₄.7H₂O) 99.5%, tri-sodium citrate (Na₃C₆H₅O₇) 99.0%, tetraethyl orthosilicate (TEOS) 99.99%, ammonia solution 25%, acetone, ethanol 99.7%, fluorescein-5-isothiocyanate (FITC), 3-aminopropyltriethoxysilane (APTS vol) < 98%, dopamine hydrochloride 99%, 2-amino-2-(hydroxymethyl)1,3 propanediol (Tris) 99%, hydrochloric acid (HCl) 38%. All the chemicals were purchased from Sigma Aldrich and Merck and were used without further purification. Malathion 90%, and Chlorpyrifos 97% were purchased from Hebei Hontai Biotech Co., Ltd., China and Shanghai Molotus Chemical Co., Ltd, China, respectively.

A previously published procedure was used to synthesize the Fe₃O₄ nanoparticles by co-precipitating iron (II) and iron (III) salts in a 1:2 molar ratio. Specifically, a two-necked round-bottom flask containing 10 mmol of iron sulphate and 20 mmol of iron chloride in 100 mL of deionized water under N₂ atmosphere was heated at 70°C for 30 min. Then, 33 mL of 25% ammonia solution was poured into the above mixture until a brownish-black precipitate of iron oxide nanoparticles was formed and rinsed with distilled water and ethanol until pH 7 is attained. The resultant product was separated using an external magnetic field and vacuum-dried for 24 h at room temperature. To modify the surface of iron oxide nanoparticles, the synthesized nanoparticles were re-dispersed in 200 mL of deionized water containing 0.3 M sodium citrate and heated at 60°C for 2 h. After washing the nanoparticles with acetone to get rid of any remaining citrate, they were dried for 15 h in a vacuum desiccator.

Silica coating on Fe₃O₄ nanoparticles (NPs) was executed by following a previously published method to prevent aggregation and enhance the stability of NPs (Kralj et al., 2010). In a sonication tub, Fe₃O₄ nanoparticles were dispersed in 50 mL of deionized water. A mixture of ammonia solution 25% (5 mL) and ethanol (140 mL) was poured into an aliquot of 2 mL of water-dispersed iron nanoparticles, followed by dilution with 40 mL of deionized water and stirred for 18 h. A solution of 20 mL of ethanol and 1 mL of TEOS was gradually added into the above-diluted ferrofluid and was kept at room temperature for 16 h. The final product was separated using an external magnetic field, rinsed with ethanol and distilled water several times, and dried in a vacuum desiccator.

The silica-coated fluorescent nanoparticles (Fe₃O₄-SiO₂-FITC nanoparticles) were synthesized using the protocol described in a recent study (Asadi et al., 2016). Briefly, 0.04 g of FITC was mixed in 10 mL of ethanol and 0.22 mL of APTS, and the resultant solution was stirred for 24 h in the dark by wrapping the flask with aluminum foil. Fe₃O₄-SiO₂ nanoparticles were added to the above FITC solution, and the mixture was mechanically agitated for 15 h. Fluorescent silica-coated iron nanoparticles (Fe₃O₄-SiO₂-FITC) were separated from the mother liquor with the help of an applied magnetic field and then dried in a vacuum at room temperature.

Free radical polymerization was used to synthesize the FMMIPs (Yin et al., 2015). Using this approach, 40 mg of tris buffer (2-amino-2-hydroxymethyl 1, 3 propanediol) dissolved in 50 mL deionized water at pH 11 was mixed with 80 mg of fluorescent silica-coated iron nanoparticles, 80 mg of dopamine hydrochloride as a monomer, and 15 mg of each target pesticide (MLT and CPS). The pH of the above solution was reduced to 8.5 by adding concentrated HCl dropwise and leaving this reaction to continue at room temperature (25°C) for 16 h to ensure the polymerization process. Target pesticides (MLT and CPS) attached on FMMIPs surface with hydrogen bonding and π-π interaction due to functionalities developed on polymeric surface. The FMMIPs thus formed was separated using an external magnetic field and rinsed three times with distilled water and ethanol to remove any unbound pesticide. To prepare FMNIPs, the same protocol was adopted as for FMMIPs, but in the absence of pesticide (template).

The maximum absorption wavelengths (λ_max) for MLT and CPS were determined using a UV-Vis spectrophotometer across the range of 200–400 nm, with MLT and CPS concentrations in solutions being measured at their respective λ_max of 230–254 nm. To initiate the experiment, 50 mL of a 50 mg L⁻¹ standard solution of MLT and CPS was added to each of the 32 test tubes. At the time of analysis, the sample was filtered using a membrane filter paper (pore size 0.45 μm, diam. 47 mm) to remove any remining suspended particles. Subsequently, 50 mg of FMMIP was introduced to 16 of these test tubes, while the remaining 16 received 50 mg of FMNIP. Aliquots were sampled at predetermined intervals (5, 10, 15, 20, 25, 30, 35, and 40 min) to determine the binding capacity (q _e ). A gradient of standard solutions of MLT and CPS, with concentrations ranging from 10 to 100 mg L⁻¹, was prepared for subsequent adsorption experiments conducted at room temperature. For each concentration, 50 mL of the solution was combined with 50 mg of FMMIP in a separate test tube. These mixtures were then agitated on a shaker for 30 min at room temperature to reach equilibrium. The solutions were filtered through an HPLC-grade filter (0.45 µm) after agitation. An identical procedure was applied using FMNIP. After separation, the pesticides bound to FMMIP (FMMIP-MLT/CPS complexes) were eluted using an ethanol-water solution (8:2 v/v) in a 100 µL pipette tip at a flow rate of 10 μL min⁻¹. A UV-Vis spectrophotometer quantified the concentrations of the eluted pesticides. The binding capacity of the spiked samples was determined using Equation 1. q e = C i − C f M × V (1) Where q_e is the binding capacity, C_i and C_f are the respective initial and final concentrations of the analyte in solution, V and M represents the solution volume, an FMMIP mass. Further, the effect of pH on the binding capacity of FMMIP was evaluated.

To assess the efficacy of FMMIP 1L of agricultural water sample was collected from the vicinity of agricultural land near Multan district and stored in previously cleaned glass bottles at 5°C. Water samples were firstly filtered through Whatmann filter paper and then a membrane filter paper (pore size 0.45 μm, diam. 47 mm) was employed to remove suspended particulates. The pH was then adjusted to a neutral value of 7. Three separate 40 mL aliquots of water, contained in different beakers, were spiked with MLT and CPS at concentrations of 0.01, 15, and 30 mg L⁻¹. These spiked samples were subsequently subjected to analysis using the previously described method.

The regeneration of an adsorbent is critical for its practical applications. For this purpose, the sample solution was prepared by adding 20 mg of FMMIP in 15 mg L⁻¹ of the target pesticides; MLT and CPS at room temperature. The mixture was stirred for 30 min at 100 rpm and then filtered through HPLC grade filter with a pore size of 0.45 µm and the concentration of the target pesticides in the filtered solution was measured using a UV-vis spectrophotometer at their respective wavelengths (230 nm for MLT and 245 nm for CPS). Afterwards, the absorbent was separated, washed with distilled water and ethanol, and dried at 100°C. The regenerated FMMIP was used in five consecutive adsorption-desorption cycles.

The presence of functional groups in FMMIPs and their unmodified components was detected by Attenuated Total Reflection (ATR) infrared spectra, which were recorded using a Perkin-Elmer FT-IR (Thermo Scientific) within the spectral range of 4,000 to 500 cm⁻¹ at a resolution of 4 cm⁻¹ and a scan rate of 16 scans per second. The UV-Vis absorption spectra were acquired with a Cary 60 UV-Vis spectrophotometer (Agilent Technologies). The surface morphology of the FMMIPs and individual components was visualized using a Scanning Electron Microscope (SEM LEO 1560, Zeiss, Oberkochen, Germany) operated at 10 kV and a Transmission Electron Microscope (TEM, Tecnai Spirit BioTWIN, FEI Company, Oregon, United States) operated at 20–120 kV. Thermal Gravimetric Analysis (TGA) was performed on the iron nanoparticles, core-shell, and FMMIP using a TGA Q5000 V3.17 Build 265 instrument, with the samples being heated for 30 min at 900°C to evaluate their thermal properties.

The FT-IR spectra depicted in Figure 2 demonstrate the characteristic functional groups of various nanomaterials: Fe₃O₄ nanoparticles (NPs), silica-coated Fe₃O₄ (Fe₃O₄-SiO₂) with fluorescein isothiocyanate (FITC), and FMMIPs complexed with MLT and CPS. The spectra for Fe₃O₄ NPs exhibit distinct peaks at 631 and 577 cm⁻¹, attributed to the Fe-O stretching vibrations, confirming the presence of iron oxide (Figure 2A). In the core-shell spectrum (Figure 2B), the peaks at 1,078, 800, and 1727 cm⁻¹ are assigned to Si-O-Si stretching vibrations, Si-O symmetric stretching, and the carbonyl group characteristic of lactones, respectively, while the peak at 2015 cm⁻¹ indicates the isothiocyanate group. These peaks verify the successful coating of silica and FITC on the Fe₃O₄ nanoparticles. The FMMIP spectrum (Figure 2C), the broad bands around 3,600, 3,400, and 2,900 cm⁻¹ correspond to O-H, N-H, and C-H stretching vibrations, suggesting MLT, CPS, and polydopamine incorporation. The pronounced peaks at 1720, 1,690, and 1,631 cm⁻¹ are associated with C=C, C=N, and C=O conjugations, respectively. Additional peaks at 1715, 1,655, 1,211, 980, 778, 730, and 577 cm⁻¹ indicate the ester and amide functionalities, C-Cl stretching, P-O-C, and S-C bending vibrations, alongside the characteristic Fe-O stretch of Fe₃O₄. These spectral features are conclusive evidence of the successful synthesis of FMMIP nanoparticles.

The morphological features of the synthesized FMMIP and their foundational components; including Fe₃O₄ nanoparticles and silica-coated Fe₃O₄ (core-shell), are presented in Figures 3A–C. The particle size distributions of these materials are given in Supplementary Figure S1. The Fe₃O₄ nanoparticles exhibited an average diameter of approximately 60–70 nm with minimal agglomeration (Bojdi et al., 2016; Ali et al., 2022). After the modification with Tetraethyl orthosilicate (TEOS), the Fe₃O₄-SiO₂ nanoparticles displayed an increased average size of 118 nm, indicating successful coating of silica shell. The FMMIPs, when further coated with polydopamine, demonstrated an increase in size to around 250 nm, authenticating the successful fabrication of the FMMIP. The smooth surface of FMMIP confirms the successful polymerization process. Transmission electron microscopy (TEM) analysis provided additional verification of the polydopamine layer on the silica core-shell laden with MLT and CPS, as depicted in Figure 3D. The particle sizes of the various nanomaterials were measured using ImageJ1.52v software with Java 1.8.0_112 (64-bit).

The magnetic behaviors of Fe₃O₄ nanoparticles, silica-coated Fe₃O₄ (Fe₃O₄-SiO₂), and pesticide-loaded FMMIP were investigated using a vibrating sample magnetometer (VSM), as depicted in Figure 4. The Fe₃O₄ nanoparticles exhibited the highest magnetization value, attributed to their pure magnetic composition. The magnetization of the core-shell nanoparticles decreased due to the TEOS and FITC coatings. A further decrease in saturation magnetization was observed for FMMIP, which can be attributed to the incorporation of non-magnetic layers such as the polydopamine coating and the pesticide loading on the core-shell structure (Fe₃O₄-SiO₂-FITC). Nevertheless, the FMMIP nanoparticles retained sufficient magnetic responsiveness to be manipulated by an external magnetic field (Ali et al., 2023).

Thermogravimetric analysis was performed to evaluate the content of magnetite and the extent of cross-linking within the imprinted polymer coatings of the synthesized nanomaterials. Figure 5 illustrates the thermogravimetric weight loss profiles for the iron nanoparticle, silica core-shell, and pesticide-loaded FMMIP.

The TGA curves for Fe₃O₄ nanoparticles, Fe₃O₄-SiO₂-FITC nanocomposites, and FMMIPs are presented in Figures 5A–C. The thermogram for Fe₃O₄ (Figure 5A) exhibited a weight loss of approximately 15% around 280°C, attributed to the desorption of organic materials (such as citrate groups from the synthesis process) and moisture. Additionally, an 8% weight reduction between 300°C and 500°C was observed due to the breakdown of labile oxygen groups, culminating in a total weight loss of 29%. The Fe₃O₄-SiO₂-FITC nanocomposite (Figure 5B) showed a significant weight decrease of around 36.87%, indicative of the decomposition of the silica and FITC layers. In contrast, the FMMIPs (Figure 5C) exhibited three distinct stages of weight loss: about 36% below 400°C and 34% between 400°C and 600°C, which can be ascribed to the evaporation of adsorbed water and the degradation of the organic constituents of the crosslinker and the pesticides. These TGA results confirm the stability and composition of the nanomaterials, with the FMMIPs showing a higher weight loss due to the presence of additional organic layers (Liu et al., 2010).

The solution pH is a pivotal factor affecting target molecules’ loading efficiency of MLT and CPS into the FMMIPs matrix. This phenomenon is primarily due to the pH-dependent electrostatic interactions between the adsorbent and the adsorbate (Oyetade et al., 2018). Adsorption experiments were conducted with 25 mg L⁻¹ of each pesticide across a pH range of 2.0–9.0. Figure 6 graphically illustrates the effect of solution pH on the adsorption of MLT and CPS onto FMMIPs.

As depicted in this Figure, the adsorption rates of MLT and CPS on FMMIP increased with rising pH levels, peaking at pH 7.0. At this optimal pH, maximum adsorption was recorded at 96.5% for MLT and 88.3% for CPS and similar results were observed in previous studies (Dehghani et al., 2017). Since, dopamine is polymerized in slightly basic conditions so, its NH- and OH- groups becomes active to strongly bind with targets (MLT and CPS) in neutral to slightly alkaline conditions, allowing for effective and efficient binding. However, the self-polymerization of dopamine is hindered by acidic or highly alkaline environments, leading to changes in the polydopamine layer. These changes impact the surface interactions with the carrier and result in decreased adsorption capability (Ge et al., 2024). Figure 6 graphically illustrates the effect of solution pH on the adsorption of MLT and CPS onto FMMIPs.

In order, to assess the binding efficiency of FMMIP for MLT and CPS, kinetic studies were conducted. A series of experiments were conducted to obtain kinetic data at different time intervals followed by analysis of the data through pseudo-first-order and pseudo-second-order kinetic models (Figure 7; Table 1). These models are commonly used to study adsorption kinetics. The pseudo-first-order model assumes that the adsorption rate is proportional to the number of unoccupied sites. This model is used when the adsorbate concentration is low. The pseudo-second-order model assumes that the adsorption rate is proportional to the square of the number of unoccupied sites and is used when the adsorption process is more complex.

Pseudo-First-Order Kinetic Model: The pseudo-first-order kinetic model is generally expressed by the Lagergren’s rate equation (Equation 2): q t = q e 1 − e − k 1 t (2) where q_e is the amount of adsorbate adsorbed at equilibrium (mg/g), q_t is the amount of adsorbate adsorbed at time t (mg/g), k₁ is the rate constant of pseudo-first-order adsorption (1/min), t is the time (min).

Pseudo-Second-Order Kinetic Model: The pseudo-second-order kinetic model, on the other hand, is represented by the following Equation 3: q t = k 2 q e 2 t / 1 + k 2 q e t (3) Where k₂ is the rate constant of pseudo-second-order adsorption (g/mg min).

For both models, qe and qt are determined through experiments, and the rate constants k₁ and k₂ are estimate from the nonlinear forms of the respective equations against time t.

The initial adsorption rate (h ₀) is defined as the rate at which the adsorbate is adsorbed at the beginning of the process when time approaches zero. For the pseudo-second-order kinetic model, the initial adsorption rate is given by Equation 4: h 0 = k 2 q e 2 (4)

The half-life t _1/2 of an adsorption process is the time required for half of the adsorbate to be adsorbed onto the adsorbent. For the pseudo-second-order model, the half-life is given by Equation 5: t 1 2 = 1 / k 2 q e (5)

The adsorption kinetics of MLT and (CPS) on FMMIP were elucidated using pseudo-first-order (PFO) and pseudo-second-order (PSO) kinetic models (See Figure 7). The estimated parameters of both kinetics models: the experimental (qexp) and calculated (qecal) adsorption capacities, rate constants (K1 for PFO and K2 for PSO), half-life (t1/2), initial adsorption rate (h0), and the fit quality parameters (R² and χ²) are listed in Table 1 for both MLT and CPS.

The PFO model not only demonstrated high correlation coefficients (R²) of 0.9902 for MLT and 0.9919 for CPS but also provided calculated equilibrium adsorption capacities (q_e,cal) of 62.2 mg/g for MLT and 55.5 mg/g for CPS, which closely approximate the experimentally observed values (q_e,exp) of 54.2 mg/g for MLT and 49.0 mg/g for CPS. These results and lower chi-square (χ²) values of 4.103 and 2.754 exemplify a more precise fit than the PSO model. The PSO model permits the calculation of half-life (t_1/2) and initial adsorption rates (h ₀). The half-life for MLT and CPS was determined to be 14.0 and 13.6 min, respectively, suggesting a rapid adsorption process. The initial adsorption rates (h ₀) were 6.44 mg/(g.min) for MLT and 5.91 mg/(g.min) for CPS. The PSO model is more favorable due to its R² value being closer to 1, low χ² values, and the substantial agreement between the experimental and calculated q_e values, highlighting its relevance in modelling the adsorption behavior on FMMIP and providing a solid foundation for future process optimization and design (Hayat et al., 2024).

Understanding the equilibrium relationship between adsorbate concentrations and the binding capacity of an adsorbent at a set temperature is fundamental to adsorption isotherm studies. In this investigation, the Langmuir and Freundlich isotherm models were applied to interpret the experimental data acquired for the FMMIP. The Langmuir model presupposes the existence of homogenous monolayer binding sites across the adsorbent’s surface with uniform adsorption energies. This model is described by Equation 6: q e = K L q m C e 1 + K L C e (6) where q _e is the amount of adsorbate adsorbed (MLT or CPS) per unit mass of FMMIP, q _m indicates the maximum monolayer adsorption capacity of FMMIP, C _e is the equilibrium concentration of the adsorbate, and K _L is the Langmuir adsorption constant, which relates to the adsorption energy.

The dimensionless separation factor or equilibrium parameter, R _L , is calculated using Equation 7: R L = 1 1 + K L C 0 (7)

Conversely, the Freundlich isotherm model addresses heterogeneous surface binding sites with variable energies of adsorption, expressed as Equation 8: q e = K F C e 1 / n (8)

Here, K _F is the Freundlich adsorption constant related to the adsorption capacity, and n is the heterogeneity factor of the adsorption process.

The fitting of the experimental data to the Langmuir and Freundlich isotherm models was quantitatively assessed using the coefficient of determination (R²) and the reduced Chi-square (χ²). Figure 8 illustrates the FMMIP’s adsorption capacity for MLT and CPS against the equilibrium concentration of the adsorbates. The Freundlich isotherm, marked by the red line, more accurately reflects the data for both substances than the Langmuir isotherm, represented by the blue line.

Corresponding to the higher R² values for the Freundlich isotherm reported in Table 2 (0.9996 for MLT and 0.9965 for CPS), the data points indicate a precise model alignment, confirming an exemplary fit. The Langmuir isotherm demonstrates slightly lower R² values (0.9983 for MLT and 0.9931 for CPS) alongside marginally higher reduced Chi-square (χ²) values, underscoring a less accurate fit than the Freundlich model.

The adherence of the experimental data to the Freundlich model, as shown in Figure 8, suggests that a heterogeneous mechanism better describes the adsorption of MLT and CPS onto FMMIP. The model’s alignment with the observed data supports the presence of varied binding sites on the FMMIP surface, which accommodate a non-uniform and multilayer adsorption, consistent with the nature of the Freundlich model.

The discrepancies between the Langmuir model and the empirical data, particularly noticeable at elevated equilibrium concentrations, highlight the shortcomings of its assumption of homogeneous adsorption sites and uniform binding energies for the adsorbates in question.

Further, discernment is gained from analyzing the Freundlich isotherm’s “n” values—1.329 for MLT and 1.346 for CPS—both of which lie within the optimal range of 1–10, indicating advantageous adsorption. Values of “n” above one implies an increasing efficiency in adsorption as the adsorbate concentration rises due to the likely interactions with the surface.

In parallel, evaluating the Langmuir isotherm’s dimensionless constant, RL, for MLT (0.6297) and CPS (0.6204) reveals values below one, signifying a favorable adsorption process despite the Freundlich model’s superior fit. An RL value within the range of 0–1 is characteristic of good adsorption, suggesting an increased likelihood of adsorption with rising initial concentration until the saturation point is reached at q _m .

In the current study, the Freundlich isotherm demonstrated the best fit, as evidenced by its R² value being close to 1 and the experimental maximum q_e value closely adheres the calculated value. Additionally, the value of n greater than 1 further supports the suitability of the Freundlich model. These findings indicate that FMMIP is an effective adsorbent for MLT and CPS under the tested conditions (Hayat et al., 2023).

The regeneration potential of the FMMIP was assessed for MLT and CPS through multiple adsorption-desorption cycles, gauging the polymer’s applicability for repeated use in analytical applications. Initially, FMMIP demonstrated recoveries of 85.3% for CPS and 80.4% for MLT in the first cycle. There was a subtle decline in recovery efficiency throughout five cycles, with the fifth cycle showing 73.7% for CPS and 71.2% for MLT, as illustrated in Figure 9. These findings underscore the durable nature of FMMIP, confirming its utility and persistent performance for sequential analytical procedures, thereby indicating its considerable value for practical implementation in pollutant extraction.

Quality assurance (QA) parameters are essential in evaluating the selectivity and precision of MIP-based adsorption methods. These parameters encompass precision, linearity, the limit of detection (LOD), the limit of quantification (LOQ), and the relative standard deviation (RSD). Inter- and intra-day RSDs were calculated from measurements taken in the afternoon and then 1 week later to ascertain the precision of the developed method. The intra-day RSD values for MLT were 6.5% (n = 5), with an inter-day RSD of 7.1%. For CPS, the intra- and inter-day RSDs were 7.3% and 7.7%, respectively. Linearity was observed for both MLT and CPS within the 5–25 mg L⁻¹ concentration range. The determined LOQs were 4 mg L⁻¹ for MLT and 3.7 mg L⁻¹ for CPS, while the LODs were calculated to be 1.26 mg L⁻¹ for MLT and 1.22 mg L⁻¹ for CPS. The calibration curves for MLT and CPS, as shown in Figures 10B,D, exhibited excellent regression values of 0.992 and 0.998, respectively, confirming the robustness and reliability of the method.