Modeling and Box–Behnken Design Optimization for the Efficient Removal of Ibuprofen via Heterogeneous Fenton-like Reactions Using a Fe₃O₄/HNTs as a Catalyst

Abstract

A Fe₃O₄/HNTs composite was synthesized, characterized by SEM, TEM, XPS, adsorption–desorption N₂, XRD, FTIR, VSM and Zeta potential, and was used for an ibuprofen adsorption and Fenton oxidation study. The response surface methodology (RSM) and Box–Behnken experimental designs were employed. The effects of pH, contact time, ibuprofen concentration, and Fe₃O₄/HNTs dosage on ibuprofen adsorption were evaluated. Additionally, adsorption isotherms and a kinetic study were performed. The effects of pH, H₂O₂ concentration, and Fe₃O₄/HNTs dosage for IBU removal were also studied. The results of ibuprofen adsorption on Fe₃O₄/HNTs indicate that adsorption was favored at acidic pH. The adsorption followed pseudo-second-order kinetics and a Freundlich isotherm. Under mild conditions (pH 7, 298.15 K) with a Fe₃O₄/HNTs dosage of 1.5 g L⁻¹ and 0.5 M H₂O₂, the heterogeneous Fenton-like reaction achieved 99% ibuprofen removal and 60% mineralization. The Fe₃O₄/HNTs catalyst demonstrated high efficiency for aqueous ibuprofen removal under environmentally mild pH and temperature conditions, and it was easily recoverable and reusable.

1. Introduction

Chemicals of Emerging Concern (CECs) are chemicals recently identified as threats to the environment and human health that require extensive regulation or monitoring [1]. These compounds include pharmaceuticals, industrial chemicals, personal care products, and others that can harm both aquatic and terrestrial ecosystems as well as human health [2]. CECs are often released into the environment through human activities, such as pharmaceutical use, industrial production, and consumer products [3]. Unlike conventional pollutants, which have been studied and regulated for decades, CECs pose additional challenges due to their dynamic nature and the lack of comprehensive data on their long-term effects [4]. For instance, ibuprofen (IBU) is a widely used nonsteroidal anti-inflammatory drug (NSAID) known for its analgesic and antipyretic properties. It is commonly prescribed to relieve fever, pain, rheumatoid arthritis, and osteoarthritis [5]. IBU is consumed extensively in both human and veterinary medicine [6], and it is the third most widely employed drug in the world, with an annual consumption of nearly 200 tons [7]. IBU has been found in water treatment plants in Australia, Canada, Chile, China, Finland, Ghana, Greece, India, Japan, Korea, Mexico, New Zealand, Palestine, Poland, Portugal, South Africa, Singapore, Spain, Switzerland, Thailand, Turkey, United Arab Emirates, United Kingdom, United States, and Vietnam [8,9]. It has also been detected in domestic wastewater in Spain, Denmark, Singapore, the Czech Republic, Germany, the United Kingdom, Sweden, and Mexico [10,11,12,13,14,15,16,17,18,19], and in surface water bodies such as the Madín Dam in Mexico [20,21]. Its presence is primarily due to human and animal excretion and its introduction through wastewater treatment systems. Its removal in wastewater treatment plants (WWTPs) is limited, which favors its release into the environment [7]. This situation is further exacerbated by animal excretion, the disposal of expired medications, and the discharge of untreated water.

IBU has been reported to have several adverse effects on organisms, including cellular and genetic damage to the freshwater bivalve Dreissena polymorpha, delayed hatching of eggs in the freshwater fish Oryzias latipes and freshwater cladocerans such as Daphnia magna and Moina macrocopa, inhibition of photosynthesis in the freshwater diatom Navicula sp., damage to the hemocytes of Ruditapes philippinarum, and an impact on the growth of the freshwater microalga Scenedesmus rubescens [22,23,24,25,26].

IBU has been removed by methods such as adsorption with different materials, including biochar [6], carbon nanotube-based materials [27], nanoparticles with polyamidoamine [28], and magnetic nanoparticles such as Fe₃O₄@hydroxyapatite [29]; biological methods [30,31]; and advanced oxidation processes (AOPs). AOPs refer to a group of chemical treatment processes designed to remove organic and inorganic contaminants from water and wastewater. These processes are typically based on the generation of highly reactive hydroxyl radicals (•OH), which can break down a wide range of pollutants [32].

One of the most well-known AOPs is the Fenton process, which is based on the generation of hydroxyl radicals from the decomposition of hydrogen peroxide (H₂O₂) in the presence of materials containing Fe (II) ions under acidic conditions. This reaction yields hydroxyl radicals (•OH) and Fe (III). Additionally, these ferric ions react with hydrogen peroxide, producing hydroperoxyl radicals (•OOH) and regenerating the catalyst (ferrous ions). If the reaction is initiated by Fe (III) ions, the process is commonly referred to as “Fenton-like.” However, there is a redox cycle that creates a series of redox reactions, causing both species to be present simultaneously, regardless of the starting ion [32,33,34].

IBU has been oxidized through the Fenton process and its various combinations with different sources of iron and other materials, such as sonolysis and sono-Fenton systems with Fe(II) ions, homogeneous modified Fenton-like oxidation with an FeIII-gallic acid complex, heterogeneous Fenton with Fe-zeolite catalysts, plasma-supported Fenton (with Fe-based ordered mesoporous carbon; 88.3% efficiency), electro-Fenton systems, Fenton-like with zero-valent ions, and electro-Fenton with zero-valent ions [35,36,37,38,39,40].

Among ion-based materials, Fe₃O₄ is one of the most reported iron(III) compounds for Fenton-like oxidation studies of IBU [41,42]. Heterogeneous photo-Fenton processes include carbon dots/Fe₃O₄@carbon sphere pomegranate-like composites activated by a persulfate system, heterogeneous photo-Fenton with humic acid-coated Fe₃O₄, sono-electrolytic Fenton, Fe₃O₄ supported on multi-walled carbon nanotubes (MWCNTs), and the Fenton-like graphene oxide-based electro-Fenton process [41,42,43,44,45,46].

Halloysite nanotubes (HNTs) are an economically viable material that can be employed without further purification. HNTs are clay minerals from the kaolin group, typically composed of two layers rolled into a tubular morphology. The outer layer is composed of silicates (tetrahedral crystals), whereas the inner layer of the nanotube consists of alumina (Al₂O₃) octahedra. Due to their large surface area, HNTs are known to be good adsorbents and catalyst supports [47,48,49,50].

This study aims to develop and optimize Fe₃O₄/HNTs composite for the efficient removal of ibuprofen from aqueous systems using a heterogeneous Fenton-type process. It is proposed that integrating magnetite nanoparticles into halloysite nanotubes enhances catalytic performance by optimizing dispersion, increasing the accessibility of active sites, and facilitating electron transfer during H₂O₂ activation, leading to high removal efficiency and medium mineralization under mild conditions. Furthermore, response surface methodology (RSM) is employed to systematically evaluate and optimize the key operating parameters governing the adsorption and oxidation processes.

2. Materials and Methods

2.1. Preparation of Fe₃O₄/HNTs Composite

Halloysite nanotubes Al₂Si₂O₅(OH)₄·2H₂O (HNTs), iron(III) chloride hexahydrate (FeCl₃ 6H₂O) and iron sulfate heptahydrate (FeSO₄·7H₂O) and ammonium hydroxide solution (NH₄OH, 28–30%), hydrogen peroxide solution (H₂O₂, 30% w/w), solutions of hydrochloric acid (HCl, 0.1 M) and sodium hydroxide (NaOH, 0.1 M) were purchased from Merck/Sigma-Aldrich, St. Louis, MO, USA. The catalyst (Fe₃O₄/HNTs) was synthesized using a coprecipitation method using Fe(ll) and Fe(lll) salts in a basic medium in the presence of HNTs as described by Maleki and Sadati [51]. The HNTs and distilled water were added to the 3-neck flask and kept under constant stirring at 60 °C. Then, FeSO₄·7H₂O and FeCl₃·6H₂O were solubilized in distilled water and added to the 3-neck flask. The solution was kept under constant stirring at 60 °C for 1 h. Afterward, NH₄OH was slowly added until the pH reached 11, maintaining stirring, at 70 °C for 2 h. The resulting solution was stored at room temperature for 24 h, forming solid deposition. Neodymium magnets were used to hold solids along with the supernatant liquid. Solids were washed several times with distilled water until the wash water had pH 7. The solid was dried at 70 °C for two days and crushed with a mortar until a fine powder was obtained.

2.2. Characterization of the Catalyst

The HNTs, magnetite, and solid catalyst were characterized using various techniques. The morphology and particle size were determined by scanning electron microscopy (JEOL-6460LV at 20 kV; JEOL, Inc., Peabody, MA, USA) and energy dispersive X-ray spectroscopy (EDX). The high-resolution scanning electron microscopy (HRSEM) technique was used with a JEOL JSM-7600F field emission scanning electron microscope (FESEM) (JEOL, Inc., Peabody, MA, USA). X-ray photoelectron spectroscopy (XPS) spectra were obtained using the VERSAPROBE II (Instrument from Physical Electronics, Chanhassen, MN, USA).

The specific surface area of the samples was determined using the Brunauer–Emmett–Teller (BET) method. The Barrett-Joyner-Halenda (BJH) method at 77 K (Autosorb 1 MP, Quantachrome Instruments; Boynton Beach, FL, USA) was employed to estimate pore volume and pore size distribution. Before measurements, the samples were degassed under vacuum at 573 K for 10 h. The crystalline phases were identified by X-ray diffraction (XRD) using a Bruker AXS X-ray diffractometer (D8 Advanced Plus; Bruker AXS, Madison, WI, USA), with CuKa1 monochromatic radiation, λ = 1.54056 Å). Data collection was performed in the 2θ range from 2° to 70°, with a step width of 0.02° and a count time per step of 1.8 s. Normal operating conditions were 35 kV and 30 mA. Infrared spectra were obtained using a Fourier-transform infrared (FT-IR) spectrometer (Nicolet 6700 with Ge-on KBr beamsplitter, Thermo Scientific, Waltham, MA, USA), and the amount of IBU adsorbed onto the catalyst particles was determined using a Perkin Elmer 283 spectrometer (KBr tablets) (PerkinElmer, Waltham, MA, USA) in the wavenumber range from 400 to 4000 cm⁻¹. Vibrating sample magnetometer (VSM) measurements were obtained using a SQUID magnetometer. Zeta potential was determined using a Zetasizer Nano ZS90 device (Malvern Panalytical, Malvern, UK/US office: Westborough, MA, USA) at pH 7 and 25 °C, with the particles suspended in distilled water.

2.3. Adsorption

Ibuprofen solutions were prepared in deionized water, with concentrations ranging from 1.5 to 15 mg L⁻¹, and the pH was adjusted with 0.1 M HCl or 0.1 M NaOH. Calibration curves for ibuprofen in aqueous solution were established. Measurements were performed at pH values of 2, 7, and 12. Absorbance spectra as a function of wavelength in the 200–300 nm range were recorded for ibuprofen concentrations from 1.5 to 15 mg L⁻¹ using a UV-Vis spectrophotometer (Perkin Elmer Model Lambda 25; PerkinElmer, Waltham, MA, USA). The results showed that the wavelength of maximum absorption for solutions at pH 2 was 222 nm, while for solutions at pH 7 and 12, maximum absorption occurred at 224 nm.

In 10 mL of IBU solutions with concentrations of 9–15 mg L⁻¹ and pH values of 2, 7, and 12, respectively, 0.5–1.5 g L⁻¹ of the Fe₃O₄/HNTs adsorbent was added. The mixtures were continuously stirred at 200 rpm and 293.15 K. At each time interval (1–14 days), the Fe₃O₄/HNTs was removed using an external magnet. The remaining IBU in the solution, or the amount not adsorbed, was determined by UV-Vis spectrophotometry. Each experiment was performed in triplicate. Obtaining a standard deviation of 0.045.

The amount of IBU adsorbed by the adsorbent at any time, $q_t$ (mg g⁻¹), and at equilibrium, $q_e$ (mg g⁻¹), respectively, was calculated based on the concentration in the solution before and after adsorption, according to Equations (1) and (2). The amount of IBU adsorbed by Fe₃O₄/HNTs at any time, $q_t$ (mg g⁻¹), and at equilibrium, $q_e$ (mg g⁻¹), was calculated based on the concentration in the solution before and after adsorption, using Equations (1) and (2).

$$q_t={\displaystyle \frac{\left(C_o-C_t\right)V }{W}}$$

(1)

$$q_e={\displaystyle \frac{\left(C_o-C_e\right)V }{W}}$$

(2)

where $C_0$ (mg L⁻¹) is the initial IBU concentration, $C_t$ (mg L⁻¹) is the concentration after a contact time $t$, $C_e$ (mg L⁻¹) is the remaining concentration at equilibrium, $V$ (L) is the sample volume, and $W$ (mg) is the mass of Fe₃O₄/HNTs used in each test.

The percentage of IBU adsorption onto Fe₃O₄/HNTs was calculated using Equation (3).

$$IBU adsorbed \left(\%\right)={\displaystyle \frac{C_o-C_f}{C_{o}Co}}\ast 100\%$$

(3)

where C_o is the initial concentration of IBU (15 mg L⁻¹) and C_f is the final concentration of IBU after adsorption.

The adsorption results were fitted with pseudo-first order (PFO) kinetic, pseudo-second order (PSO) kinetic, and intra-particle diffusion to determine the adsorption mechanisms.

This PFO kinetic model (Equation (4)) was analyzed because it typically applies to physisorption processes and describes physical interactions between the adsorbate and the adsorbent, such as Van der Waals or dipole–dipole forces. In this model, q_t and q_e (mg g⁻¹) represent the amounts of IBU adsorbed at time t and at equilibrium, respectively; t (min) is the adsorption time, and k₁ (min⁻¹) is the Lagergren pseudo-first-order rate constant [52].

$$Ln \left(q_e-q_t\right)=q_e-k_{1}t$$

(4)

The pseudo-second order (PSO) kinetic model, Equation (5). This model implies that the adsorbent surface exhibits a specific affinity for the adsorbate, and that the active sites progressively become saturated over time, which gradually reduces the adsorption rate until equilibrium is reached. Where, q_t and q_e (mg g⁻¹) represent the amounts of IBU adsorbed at time t and at equilibrium, respectively; t (min) is the adsorption time, and k₂ (mg g⁻¹ min⁻¹) is the pseudo-second-order rate constant [53].

$${\displaystyle \raisebox{1ex}{$t$}\!\left/ \!\raisebox{-1ex}{$q_t$}\right.}={\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$k_2{q}_e^2$}\right.}+{\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$q_e$}\right.}\mathrm{t}$$

(5)

The intraparticle diffusion model (Equation (6)) posits that the adsorption rate is, at least in part, governed by the transport of the adsorbate from the bulk solution into the internal pores of the adsorbent. This internal diffusion takes place through the porous structure of the solid and may become the rate-limiting step when active sites are not readily accessible from the external surface, due to the adsorbent’s pore characteristics. In this model, q_t is the amount of IBU adsorbed at time t (min), k_i is the intraparticle diffusion rate constant (mg g⁻¹ min^1/2), and C is a constant (mg g⁻¹) associated with the boundary layer thickness [54].

$$q_t=k_i{t}^{1/2}+C$$

(6)

The adsorption isotherm data were obtained at 298.15 K. The concentration range is 5–15 mg L⁻¹. Adsorption equilibrium data were analyzed using the Henry, Freundlich and Langmuir isotherms models.

Henry’s isotherm, Equation (7), assumes ideal behavior, where adsorption occurs in a monolayer without interactions between adsorbed molecules and no saturation of adsorption sites. Although simple, the Henry isotherm is fundamental for understanding adsorption at low coverage and is often employed to define the initial slope of more complex isotherms, such as Freundlich or Langmuir [55].

$$q_e=K_H{C}_e$$

(7)

The Freundlich isotherm shown in Equation (8), is an empirical model that can be applied to multilayer adsorption, with a non-uniform distribution of the heat of adsorption and affinities on the heterogeneous surface, is an empirical model that can be applied to multilayer adsorption, with a non-uniform distribution of the heat of adsorption and affinities on the heterogeneous surface [56]. Where qe (mg⋅g⁻¹) is the equilibrium amount of adsorbed IBU, Ce (mg⋅L⁻¹) is the equilibrium concentration of IBU, and K_F (mg⋅g⁻¹) and n are constants for a given adsorbate and adsorbent, per mass unit of adsorbent, at equilibrium.

$$\log q_e=\log K_F+\left({\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$n$}\right.}\right)\log C_e$$

(8)

The Langmuir isotherm, as expressed in Equation (9), assumes that adsorption occurs as a monolayer on a homogeneous surface. In this context, Ce (mg·L⁻¹) represents the equilibrium concentration of the adsorbate in the solution phase, while qe (mg·g⁻¹) denotes the amount of adsorbate retained per unit mass of adsorbent when equilibrium is reached. The parameters q₀ and K_L are the Langmuir constants, which describe the maximum adsorption capacity and the affinity of the binding sites, respectively [57].

$${\displaystyle \raisebox{1ex}{$C_e$}\!\left/ \!\raisebox{-1ex}{$q_e$}\right.}=\left({\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$q_{max}{K}_L$}\right.}\right)+({\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$q_o$}\right.})C_e$$

(9)

2.4. Ibuprofen Removal by Heterogeneous Fenton Reaction

IBU removal was carried out in propylene centrifuge tubes. Solutions of 15 mg L⁻¹ were prepared in deionized water, and the pH was adjusted using either HCl (0.1 M) or NaOH (0.1 M). After pH adjustment, the corresponding dose of Fe₃O₄/HNTs was mixed with 10 mL of IBU solution. The addition of hydrogen peroxide initiated the heterogeneous Fenton-type reaction. After 24 h of reaction, the catalyst was separated from the reaction samples with a magnet. A sample of the supernatant liquid was immediately collected, and the chemical oxygen demand (COD) was determined. Digester equipment for COD vials HI 83980 Hanna was used for the COD. A UV–visible spectrophotometer (UV-Vis Perkin Elmer Model Lambda 25) was employed to determine the Cr³⁺ and Cr⁶⁺ (as dichromate) absorbance in digested samples to match them against the corresponding COD calibration plot.

The total organic carbon (TOC) of the oxidation samples was recorded using a Sievers 900 analyzer (GE Analytical Instruments (Sievers), Boulder, CO, USA) with ultrapure water. This experiment was performed in triplicate. The mineralization percentage was determined using Equation (10). Where initial TOC and final TOC are the TOC values at the start and end of the oxidation reaction, respectively

$$Mineralization \left(\%\right)={\displaystyle \frac{{TOC}_{initial}-{TOC}_{final}}{{TOC}_{initial}}}\times 100$$

(10)

The reusability of the catalysts was evaluated through five cycles of the Fenton reaction. After each cycle, the catalysts were recovered using a magnet, exhaustively washed, dried, and then reused under the same experimental conditions.

2.5. Response Surface Methodology and Experimental Design

For the adsorption and Fenton oxidation studies, Box–Behnken experimental designs were employed. pH, dosage (adsorbent/catalyst), and H₂O₂ concentration were selected as independent variables. These factors were evaluated at two levels and a center point (−1, 0, and +1).

A second-order (quadratic) model was obtained to analyze the interactions between variables (Equation (11)). An analysis of variance (ANOVA) was applied to verify the statistical significance of the model terms.

$$Y={\beta }_o+{\beta }_1{X}_1+{\beta }_2{X}_2+{\beta }_{12}{X}_1{X}_2+{\beta }_{11}{X}_1^2+{\beta }_{22}{X}_2^2$$

(11)

where X_i, X_i², and X_iX_j represent the linear, quadratic, and interaction terms, respectively. β₀, β_ii, and β_ij are the coefficients of the linear model terms, the quadratic model terms, and the interaction terms, respectively, and Y represents the response, which is the percentage of adsorption and removal for Fenton oxidation, respectively. Design Expert software version 11 (Stat-Ease, Inc. Minneapolis, MN, USA) was the software used. Response surface methodology (RSM) was employed as a statistical and mathematical approach to model and analyze adsorption and oxidation processes where multiple variables influence the desired outcome, to optimize the response [58].

3. Results

3.1. Catalyst Analysis

3.1.1. STEM and SEM

Figure 1 shows the STEM and SEM images obtained in the nanometer range. These images reveal clusters of spherical nanoparticles, identified as Fe₃O₄, surrounding the NTHs. The SEM image suggests that these are not crystallites, but rather clusters with an average size of 265 nm. However, the surface spherical particles have an average diameter of 25 to 40 nm.

The morphology and size details of the Fe₃O₄/HNTs catalyst were determined by SEM. Figure 2 indicates that the catalyst particles were nearly spherical and of uniform size; in addition, halloysite nanotubes exhibited irregular needle shape and irregular agglomerations on and around catalyst particles. It might be due to strong inter-particle Van der Waals forces or due to magnetic attraction among the Fe₃O₄ nanoparticles. Also, SEM images confirm the catalyst’s sized structure and morphology. Chemical composition of the catalyst, determined by EDX, confirms the presence of iron, oxygen, silicon and aluminum.

3.1.2. XPS

The XPS spectrum of the Fe₃O₄/NHTs sample is shown in Figure 3. Peaks for iron, oxygen, aluminum, silicon, and carbon were observed. The iron signals (Fe 2p, BE 710.1 eV and BE 711.3 eV) are assigned to octahedral Fe(II) and Fe(III), respectively, which correspond to Fe₃O₄ [59]. The oxygen signal (O 1s, BE 530 eV) is assigned to iron oxides. Peaks for aluminum (Al 2p, BE 74 eV) and silicon (Si 2p, BE 99 eV) were also observed. After integrating and normalizing the spectral areas, the composition of the sample was found to be Fe (51.33%), O (28.13%), Al (7.38%), Si (10.17%) and C (2.99%), respectively, with a Fe₃O₄/HNTs ratio of 60/40. Si/Al ratio of 1.37, like that reported in other works [60].

3.1.3. Adsorption–Desorption N₂

The pore size distribution of the Fe₃O₄, HNTs, and Fe₃O₄/HNTs samples was obtained using the Barret–Joyner–Hallenda (BJH) method [61]. In all cases, mesoporous particles (2–50 nm) and macroporous particles with a pore diameter greater than 50 nm [62]. The pore size distribution for halloysite is trimodal, with maxima at 3.3, 11, and 75 nm, respectively. The pore size distribution for magnetite is monomodal, with a maximum of 7.7 nm. The Fe₃O₄/HNTs catalyst presents a bimodal pore size distribution with maxima at 3.3 and 18 nm. From the nitrogen adsorption–desorption study for halloysite, magnetite, and magnetite/halloysite samples (Figure 4a), it is observed that the three samples presented hysteresis cycles, indicating the presence of mesoporous material. For HNTs, a minimal hysteresis cycle of type H3 is observed, characteristic of mesoporous materials that form agglomerates of particles with pores or a slit-lamellar shape with non-uniform size, characteristic of mesoporous materials that form agglomerates of particles with pores or a slit-lamellar shape with non-uniform size [62], and that has been observed in different HNT samples [63]. Fe₃O₄ and Fe₃O₄/HNTs samples show H1-type hysteresis cycles, which are characteristic of solids consisting of particles crossed by almost cylindrical channels or formed by aggregates (consolidated) or agglomerates (unconsolidated) of spheroidal particles with pores of size and uniform shape [62]. The similarity of the types of hysteresis between Fe₃O₄ and Fe₃O₄/HNTs indicates a higher magnetite content in the catalyst than halloysite. The specific surface area of Fe₃O₄, HNTs, and Fe₃O₄/HNTs was determined using the BET equation, using the results of N₂ adsorption on the surfaces, obtaining values of 90 67.7 m² g⁻¹, 34.4 m² g⁻¹ and 67.7 m² g⁻¹, respectively. The BET area for Fe₃O₄/HNTs was higher than the area of the HNTs because the magnetite is heterogeneously supported on the halloysite and is clustered together.

3.1.4. X-Ray Diffraction

The X-ray diffraction pattern of the Fe₃O₄/HNTs sample is shown in Figure 4b). The crystalline phases identified are magnetite and HNTs according to the Power Diffraction Pattern sheets (PDF) 01-076-0958 and 00-009-0451, respectively. The HNTs were identified with the reflections near at 2θ (11.87°, 20.1°, 24.5°, 35.1°, 38.1°, 54.6° and 62.6°). The most important result marks this identification for d₀₀₁, corresponding to the reflection at 11.87°, which was observed in the pattern. This reflection corresponds to a d space around 7.77 Å, a characteristic result for dehydrated HNTs (7 Å) [64,65]. The magnetite phase was identified with the reflections at 2θ (30.11°, 35.46°, 43.1°, 57.0°, and 62.59°). The most important result marks this identification for d₂₂₀, corresponding to the reflection at 30.11° observed in the pattern. The crystalline domain size (D) was determined using the Scherrer equation. The highest intensity reflections for HNTs and Fe₃O₄ were selected in the X-ray diffraction pattern of the Fe₃O₄/HNTs sample. The crystalline domain size (D) was determined using the Scherrer equation. The highest intensity reflections for HNTs and Fe₃O₄ were selected in the X-ray diffraction pattern of the Fe₃O₄/HNTs sample. The crystallite sizes of the natural HNTs and the magnetite synthesized by the coprecipitation method were 13 and 12 nm, respectively. However, when the synthesis of Fe₃O₄ was carried out in the presence of HNTs, the crystallite size of both HNTs and Fe₃O₄ increased to 25 nm and 40 nm, respectively.

3.1.5. Vibrating Sample Magnetometer

Figure 5a shows the magnetization-magnetic field (M-H) curve of Fe₃O₄ nanoparticles supported on NTH. The sigmoidal shape of the M-H curve indicates the absence of hysteresis. This type of curve is characteristic of small nanoparticles, where coercivity and remanence are practically zero, which is typical of superparamagnetic materials [66]. The magnetization reaches a saturation value (Ms) of around 13 emu/g, suggesting that the magnetite particles are magnetized to their maximum capacity under the applied field. Although the Fe₃O₄ particles exhibit agglomerations with an average size of ~236 nm in SEM/TEM, a nanoparticle array with sufficiently weak interparticle magnetic interactions exhibits superparamagnetic behavior [67].

3.1.6. Zeta Potential

Figure 5a shows a zeta potential distribution plot, where the x-axis represents the zeta potential (mV) in the range of −200 mV to 200 mV, and the y-axis shows the total number of counts. The plot exhibits a narrow Gaussian curve, centered at −34.6 mV. This result indicates moderate electrostatic stability [68].

3.2. Box–Behnken Design for the Adsorption of Ibuprofen on Fe₃O₄/HNTs

A design Box–Behnken was implemented to evaluate the factors used (A: pH, and B: Fe₃O₄/HNTs) and determine their effect on the adsorption of IBU onto Fe₃O₄/HNTs. Three levels were evaluated for each factor (−1, +1, and 0) (

Table 1. Code and levels of Box–Behnken design for the adsorption of IBU on Fe₃O₄/HNTs.

Factor	Code	Low Level (−1)	Central Point	High Level (+1)
pH	X1 = A	2	7	12
Adsorbent dose (g L−1)	X2 = B	0.5	1.5	2.5

). In addition to three center points, a total of 12 experiments were conducted, where the percentage of adsorption was measured in each one (

Table 2. Amount of IBU adsorbed on Fe₃O₄/HNTs.

Run	Fe3O4/HNTs (g L−1)	pH	Adsorbed IBU (%)
1	0.5	2	7.9
2	0.5	7	2.9
3	0.5	12	1.9
4	1.5	2	10.1
5	1.5	7	3.1
6	1.5	12	2.6
7	2.5	2	8.5
8	2.5	7	3.2
9	2.5	12	2.0
10	1.5	7	3.3
11	1.5	7	3.4
12	1.5	7	3.4

). Obtaining a standard deviation of 0.4.

The statistical models, shown in

Table 3. Lack-of-fit test of the optimization models for the adsorption of IBU onto Fe₃O₄/HNTs.

Source	Sequential p-Value	Lack of Fit p-Value	Adjusted R2	Predicted R2
Linear	0.0009	0.0006	0.7401	0.6220
2FI	0.8766	0.0005	0.7085	0.4662
Quadratic	0.0004	0.0122	0.9721	0.8767
Cubic	0.3341	0.0071	0.9758	0.0083

, were obtained using the Design Expert 11 software. The quadratic model best fits the adsorption results, considering different pH values and doses of the adsorbent. p-values less than 0.0500 indicate model terms are significant. In this case, A, A2, and B2 are significant model terms. Values greater than 0.1000 indicate the model terms are not significant.

The ANOVA results are shown in

Table 4. Analyses of variance (ANOVA) for the quadratic model describing IBU adsorption onto Fe₃O₄/HNTs.

Source	Sum of Squares	Df	Mean Square	F-Value	p-Value
Model	83.31	5	16.66	77.53	<0.0001
A-pH	66.40	1	66.40	308.97	<0.0001
B-Fe3O4/HNTs	0.2091	1	0.2091	0.9728	0.3621
AB	0.0576	1	0.0576	0.2680	0.6232
A2	16.62	1	16.62	77.32	0.0001
B2	1.44	1	1.44	6.68	0.0415
Residual	1.29	6	0.2149	--	--
Lack of fit	1.24	3	0.4137	25.71	0.0122
Error	0.0483	3	0.0161	--	--
Corr total	84.60	11	--	--	--

. The model F-value of 77.53 implies the model is significant. There is only a 0.01% chance that an F-value this large could occur due to noise. The lack-of-fit F-value of 25.71 implies the lack-of-fit is significant. There is only a 1.22% chance that a lack-of-fit f-value this large could occur due to noise.

The quadratic model equation that quantitatively describes the response’s relationship with the factors was obtained. It can be employed to determine the optimal experimental conditions for the absorption of IBU on Fe₃O₄/HNTs (Equation (12)).

$$IBU adsorbed \left(\%\right)=3.47-3.33A+0.1867\mathrm{B}-0.12\mathrm{A}\mathrm{B}+2.5A^2-0.7338B^2$$

(12)

According to Equation (12), the main factor in the absorption of IBU on Fe₃O₄/HNTs is A (pH). At more acidic pH values, greater adsorption is obtained: at pH 2, approximately 8%; at neutral pH, around 3%; and at pH 12, it decreases between 2 and 2.5%. Also, a slight increase in adsorption is observed when the amount of Fe₃O₄/HNTs reaches a maximum. These results indicate that at acidic pH, the adsorption of IBU on Fe₃O₄/HNTs is favored because the protons of the carboxylic acid of the IBU form hydrogen bonds with the surface of the magnetite [69]. On the other hand, at basic pH, ibuprofen is in its anionic form due to the deprotonation of its carboxylic group, which presents a negative charge like the surface of Fe₃O₄/HNTs, repelling each other and disfavoring adsorption (Figure 6).

Another significant value was B: Fe₃O₄/HNTs. The increase in adsorption with the rise in the amount of Fe₃O₄/HNTs is due to the increase in the number of active sites for the adsorption of the IBU. Figure 7 shows the response surface graph of IBU adsorption as a function of pH and the adsorbent dose. Where higher adsorption is observed at acidic pH 2.

Finally, the optimal experimental conditions for the absorption of IBU were obtained at a concentration of 15 mg L⁻¹, a temperature of 298.15 K, a pH of 2, and a dose of Fe₃O₄/HNTs of 1.5 g L⁻¹, adsorbing an amount of 1.1 mg g⁻¹ of adsorbent.

3.3. Adsorption Kinetics

Figure 8a shows that the adsorbed amount of IBU on Fe₃O₄/HNTs increased depending on the concentration. This increase, at higher concentrations of IBU, could be due to the shift in the adsorption–desorption equilibrium towards the adsorption process until the highest concentration was reached, which shows a saturation of the active sites of the adsorbent. To know the adsorption rate of IBU on the Fe₃O₄/HNTs adsorbate, adsorption kinetics, and equilibrium adsorption experiments were carried out, where data were obtained on the adsorbed amount of IBU per gram of adsorbent at 297.15 K as a function of time (pH 2 and dose of the adsorbent of 1.5 mg g⁻¹). The kinetic curves show adsorption behavior, where drug adsorption is slow in the initial days of contact with the adsorbent, followed by a low, gradual removal before reaching equilibrium. As can be seen in Figure 8a, approximately 10% of the IBU was absorbed within the first five days, showing slow kinetics. In the following period, adsorption continues at lower rates, reaching equilibrium after 14 days. The adsorption kinetics profile is characteristic of drug adsorption in solution onto different adsorbents [70,71], as well as gases on CuCl/boehmite surfaces [72].

FTIR spectra of the Fe₃O₄/HNTs and adsorbed Fe₃O₄/HNTs-IBU samples are presented in Figure 8b. FTIR spectra show the vibrations for HNTs in the range of 4000–400 cm⁻¹. The bands at 3695 and 3622 cm⁻¹ correspond to the elongation of the HNTs structural O–H groups. The bands 3527, 3456, and 1650 cm⁻¹ are due to the stretching and bending of the water molecules. The absorption bands found at 1091 cm⁻¹ and 1032 cm⁻¹ are due to the presence of Si–O–Si, and the band at 910 cm⁻¹ corresponds to the bending of Al–O–OH, respectively, for HNTs [73]. Also, a vibration is observed at 553 cm⁻¹ that corresponds to magnetite. It is essential to mention that magnetite presents this characteristic vibration in the infrared spectrum in the range of 540 to 570 cm⁻¹ (Fe–O). The absence of the IBU signal in the spectrum Fe₃O₄/HNTs-IBU indicates minimal IBU adsorption on the Fe₃O₄/HNTs surface.

Adsorption experiments were used to investigate the effect of contact time on IBU adsorption. To explain the adsorption of IBU on Fe₃O₄/HNTs, the PFO, PSO, and intraparticle diffusion equation models were employed. The results were fitted to the equations of the PFO, PSO, and intraparticle diffusion kinetic models.

Table 5. Kinetic parameters values for the adsorption of IBU onto Fe₃O₄/HNTs.

Kinetic Model	R2	Associated Constants
PFO	0.973	k1 = 1.9711 × 10−4 min−1
PSO	0.998	k2 = 3.276 × 10−4 g⋅mg−1⋅min−1
Intraparticle diffusion	0.993	ki = 0.00400 min2 y C = 0.56147 mg g−1

presents the fitting parameters of these kinetic models. In the PFO model, the R² value was 0.973. For the PSO and intraparticle diffusion models, the R² values were 0.998 and 0.992, respectively. The coefficient of determination R² closest to unity corresponds to the PSO model, which is the one that best fits the adsorption of IBU onto Fe₃O₄/HNTs. For the sample of 15 mg L⁻¹, k₂ is 3.27 E⁻⁴ (g⋅mg⁻¹⋅min⁻¹). The value of k₂, on the order of thousandths, implies that the adsorption rate of IBU onto Fe₃O₄/HNTs is slow. This model suggests that the adsorbent surface exhibits a specific affinity for the adsorbate, and that progressively becomes saturated over time, which gradually reduces the adsorption rate until equilibrium is reached [53]. PSO models have been regularly obtained for the adsorption of IBU with other adsorbents such as graphene oxide nanoplatelets, doped copper, superparamagnetic silica nanocomposites, magnetic multi-walled carbon, and Fe₃O₄/Rice husk [31,69,71,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90].

In

Table 6. Isotherm parameters for adsorption of IBU on Fe₃O₄/HNTs.

Isotherm	R2	Associated Constants
Henry	0.988	KH (mg L⋅mg−1⋅g−1)	0.068
Freundlich	0.991	KF (mg g−1) (1/n)	0.124 0.837
Langmuir	0.873	KL (L mg−1) qmax (mg g−1)	0.012 5.144

, it can be seen that practically all the IBU adsorption studies on different surfaces fit the PSO model, regardless of the initial IBU concentration or whether the adsorption is carried out at acidic, neutral, or basic pH.

3.4. Adsorption Isotherm

The results obtained for the amount of IBU adsorbed per gram of Fe₃O₄/HNTs, q_e (mg g⁻¹), and the IBU concentration in the adsorption system at equilibrium, C_e (mg L⁻¹), were measured in adsorption experiments conducted at pH 7. These data were fitted to the Henry isotherm (q_e vs. C_e), the Freundlich isotherm (log q_e vs. log C_e), and the Langmuir isotherm (C_e/q_e vs. C_e), as shown in Figure 9d–f. Linear regression was performed for each isotherm (dashed lines), and it was observed that the data fit best to the Henry and Freundlich isotherms, but not to Langmuir.

The parameters obtained for each isotherm are shown in Table 6. The Langmuir model (R² = 0.87356) indicates that the adsorption results do not fit the Langmuir isotherm, while the data fit very closely to the Henry (R² = 0.988) and Freundlich (R² = 0.99126) isotherms.

The Freundlich equation had the best correlation, with a maximum adsorption capacity of 5.1 mg g⁻¹. It assumes heterogeneous adsorption energies on the surface of the adsorbent. The parameter 1/n in the Freundlich equation reflects the degree of surface heterogeneity of the adsorbent. Lower values of 1/n (approaching zero) suggest greater surface heterogeneity, whereas values approaching one indicate a more homogeneous surface. It maintains low heterogeneity since the value of 1/n is close to 1, in accordance with the Freundlich adsorption. It is important to note that with TEM and SEM images, the surface of the Fe₃O₄/HNTs nanocomposite was observed to be heterogeneous, which confirms that IBU adsorption fits the Freundlich isotherm. This is due to the heterogeneous surfaces formed during the synthesis of nanocomposites, as seen in the case of magnetic nanoparticles coated with zeolite [86] and rape straw biomass fiber/β-Cyclodextrin/Fe₃O₄ [89] (

Table 7. Comparison of IBU adsorption in other clays and iron oxides.

Adsorbent	Experimental Condition	Model Kinetic	Adsorption Isotherm	Maximum Adsorption Capacity (mg g−1)	Reference
Cellulosic Sisal nanoparticles	pH 5, [IBU]o = 30 mg L−1, 313 K,	PSO	-	19.45	[31]
Fe3O4/Douglas fir biochar	5 min, pH 8, [IBU]o = 100 mg L−1, 308 K	-	Langmuir	40	[69]
GO nanoplatelets	60 min, pH 2–10, [IBU]o = 100 mg L−1	PSO	Langmuir	2.4–3.7	[71]
Cu-doped MIL-101Fe	360 min, 298 K, pH 2–11, [IBU] = 1–60 mg L−1	PSO	Langmuir	497	[75]
Magnetic multi-walled carbon nanotube	298 K, pH 4, [IBU] = 20 mg L−1	PSO	Langmuir	1.6–11.8	[76]
Fe3O4/Rice husk	60 min, pH 7, [IBU]o = 8 mg L−1	PSO	Langmuir	-	[77]
Mesoporous silica	pH 7, [IBU]o = 4 mg L−1, 298 K	PSO	-	-	[78]
PAMAM SiO2	pH 9, [IBU]o = 1 mg L−1, 298 K	PSO	-	124	[79]
Carbon nanotubes	pH 7, [IBU]o = 42mg L−1, 298 K	PSO	-	231.5 81.6 19.4	[80]
Ordered mesoporous carbons	pH 6, [IBU]o = 100 mg L−1, 298 K	PSO	-	120	[81]
Polyamidoamine SiO2	pH 9, 303 K	PSO	-	9.7	[82]
Zinc oxide nanoparticles	pH 7, [IBU]o = 100 mg L−1, 298 K	PSO	-	1.1	[83]
Nanoclay composite	pH 6, [IBU]o = 10 mg L−1, 298 K	PSO	-	9.7	[84]
Magnetic genipin-crosslinked chitosan/graphene oxide-SO3H composite	pH 6, 120 min, [IBU]o = 10 mg L−1, 308 K	PSO	Langmuir	138	[85]
Magnetic nanoparticles coated with zeolite	pH 8, 20 min, [IBU]o = 20 mg L−1, 303 K	PSO	Freundlich	-	[86]
Fe3O4@graphene nanoplatelets nanocomposite	pH 2, 200 min, [IBU]o = 0.1 mg L−1, 303 K	PSO	-	9.2	[87]
NiFe2O4/activated carbon	pH 2, 240 min, [IBU]o = 100 mg L−1, 328 K	PSO	Sips	261	[88]
Amine-coated magnetic nanocomposite NiFe2O4@SiO2	pH 7, 15 min, [IBU]o = 12 mg L−1, 298 K	PSO	Langmuir	59	[89]
Rape straw biomass fiber/β-Cyclodextrin/Fe3O4	pH 6, 20 min, [IBU]o = 100 mg L−1, 308 K	PSO	Freundlich	48	[89]
Kaolinite	298 K, pH 3, [IBU]o = 60 mg L−1	-	-	3.1	[90]
Goethite	298 K, pH 3, [IBU]o = 60 mg L−1	-	-	6.1	[90]
Fe3O4/HNTs	1440 min, pH 2, [IBU]o = 15 mg L−1	PSO	Freundlich	5.144	This work

).

The adsorption capacity of IBU on Fe₃O₄/HNTs is shown to be 5.1 mg L⁻¹ (Table 6), a very low value attributed to the weak hydrogen bonding interaction between IBU and the magnetite surface [69]. This result indicates that significant adsorption of IBU onto the HNTs does not occur. The minimal adsorption of IBU on Fe₃O₄/HNTs ensures that it remains available in the solution phase, thereby allowing the heterogeneous Fenton reaction to take place on the catalyst surface.

3.5. Ibuprofen Removed by Heterogeneous Fenton-Type Reaction (Monitoring)

A monitoring design was implemented to evaluate the factors used (A: pH, B: H₂O₂, and C: Fe₃O₄/HNTs) and determine their effect on IBU degradation. Two levels were evaluated for each factor (−1 and +1) (

Table 8. Code and levels for monitoring the design of the removal of IBU on the Fe₃O₄/HNTs.

Run	Code	Low Level (−1)	Central Point	High Level (+1)
pH	X1 = A	2.0	7.0	12
H2O2 (M)	X2 = B	0.05	0.275	0.5
Fe3O4/HNTs (g L−1)	X3 = C	0.5	0.75	1.0

). In addition to three center points, resulting in a total of 11 experiments where the percentage of removal of IBU was measured at each one. The results of TOC removal efficiency for Fenton are presented in

Table 9. Results obtained from the IBU removal monitoring design.

Run	H2O2 (M)	Fe3O4/HNTs (g L−1)	pH	Removed IBU (%)
1	0.050	0.5	2	85
2	0.050	0.5	12	54
3	0.050	1.0	2	91
4	0.050	1.0	12	69
5	0.500	0.5	2	87
6	0.500	0.5	12	56
7	0.500	1.0	2	94
8	0.500	1.0	12	72
9	0.275	0.75	7	89
10	0.275	0.75	7	88
11	0.275	0.75	7	88

.

The data of Fenton were fitted to a quadratic model, the significance and adequacy of which were tested by ANOVA (

Table 10. Lack-of-fit test of the monitoring models for the removal of IBU onto Fe₃O₄/HNTs.

Source	Sequential p-Value	Lack of Fit p-Value	Adjusted R2	Predicted R2
Linear	0.7970	0.0141	−0.1621	−0.7955
2FI	0.7379	0.0113	−0.2881	−2.6005
Quadratic	<0.0001	0.5212	0.9251	0.8204
Cubic	0.8415	0.2246	0.8969	−1.0033

). The F-test of the regression models produced very low p-values (<0.0001), indicating that the model was of high significance. The determination coefficients (R²) of the model indicated that 94.51% of the total variability could be explained by the model for the Fenton. The value of the adjusted determination coefficient (adjusted R² = 0.8204) also proved the high significance of the model.

The values of all regression coefficients with the significance levels are given in

Table 11. Analyses of variance (ANOVA) for the quadratic model describing IBU removal.

Source	Sum of Squares	Df	Mean Square	F-Value	p-Value
Model	1707.43	7	243.92	1177.02	0.0008
A-pH	1416.18	1	1416.18	6833.77	0.0001
B-H2O2	12.40	1	12.40	59.84	0.0163
C-Fe3O4/HNTs	237.62	1	237.62	1146.63	0.0009
AB	0.0365	1	0.0365	0.1759	0.7157
AC	40.05	1	40.05	193.27	0.0051
BC	0.6161	1	0.6161	2.97	0.2268
ABC	0.5202	1	0.5202	2.51	0.2540
Curvature	354.52	1	354.52	1565.96	0.0006
Pure Error	0.4145	2	0.2072	--	--
Cor Total	2032.36	10	--	--	--

. The model F-value of 1177.02 implies the model is significant. There is only a 0.08% chance that an F-value this large could occur due to noise. p-values less than 0.0500 indicate model terms are significant. It could be seen that, for the Fenton reaction, the linear coefficients of reaction pH (β2) and the concentration of Fe₃O₄/HNTs (β3), as well as the interaction coefficient of initial pH with concentration of Fe₃O₄/HNTs (β23) are significant factors at a level less than 5% by independent variables in terms of coded factors are show in Equation (13).

$$IBU oxidation \left(\%\right)=76.38-13.31\mathrm{A}+1.21\mathrm{B}+5.45\mathrm{C}+2.24\mathrm{A}\mathrm{C}+0.2775\mathrm{B}\mathrm{C}-0.2550\mathrm{A}\mathrm{B}\mathrm{C}$$

(13)

According to Equation (12) and the Pareto chart (Figure 10), the variable with the main effect on removal is A: pH. With a t-value of 161.8. It has been reported that, at acidic pH and in the presence of H₂O₂, the iron from magnetite dissolves, and that a higher concentration of iron in the solution contributes to the effective production of hydroxyl radicals, increasing the degradation [91].

After pH, the variable with the greatest effect on IBU removed is C, the concentration of Fe₃O₄/HNTs. Hydroxyl radicals are formed on the surface of Fe₃O₄/HNTs when Fe reacts with hydrogen peroxide, thus forming hydroxyl radicals. Therefore, increasing the catalyst concentration increases the surface for hydroxyl radical formation.

It is worth mentioning that the Fe₃O₄ surface exhibits a mixed valence state (Fe²⁺/Fe³⁺), which leads to two possible adsorption sites for H₂O₂. The two modes of H₂O₂ adsorption are shown in Figure 11. The first is molecular adsorption on Fe²⁺/Fe³⁺ sites (tetrahedral/octahedral): The adsorption structure is molecular, and dissociation can occur through the breaking of the O–H or O–O bonds, generating OH or OOH radicals on the Fe₃O₄ surface. This dissociation is controlled both by kinetics (O–H bond cleavage) and thermodynamics (O–O bond cleavage) [92].

The second mechanism of hydroxyl radical formation is dissociative adsorption on Fe³⁺/Fe³⁺ sites (octahedral/octahedral): This adsorption mode favors the breaking of the O–O bond, generating two OH radicals on the Fe₃O₄ surface with a small energy barrier of 0.24 eV. It has also been observed that in these mechanisms, Fe²⁺ is oxidized to Fe³⁺, while the Fe³⁺ site remains in its original oxidation state, forming a mixture of OH anion and OH radical. In contrast, the dissociative adsorption of H₂O₂ on Fe³⁺/Fe³⁺ sites exclusively produces two OH radicals, as no change is observed in the oxidation states of the iron sites [92].

The next significant value was the AC (pH with Fe₃O₄/HNTs) interaction because pH affects the magnetite contained in the catalyst. The point zero charge (PZC) of magnetite is 6.5, so at acidic pH values, the catalyst surface becomes protonated [93,94], favoring the adsorption of IBU onto Fe₃O₄/HNTs and thus degradation, since hydroxyl radicals are formed on the catalyst surface [93].

Finally, the variable with the smallest, but still significant, effect on IBU removal was the H₂O₂ concentration, since the hydroxyl radicals originate from hydrogen peroxide, so increasing the concentration favors degradation. The response surface plots obtained are shown in Figure 12a–c. The highest amount of IBU removal was 94.74% at pH 2, 0.5 M of H₂O₂, and 1 g L⁻¹ of Fe₃O₄/HNTs.

3.6. Ibuprofen Removed by Heterogeneous Fenton-Type Reaction (Optimization)

Once the effects of pH, Fe₃O₄/HNTs dosage, and H₂O₂ concentration on IBU degradation were determined, a Box–Behnken experimental design was implemented to find the optimal experimental conditions for IBU removal through a Fenton-type heterogeneous reaction, maintaining a constant pH 7 and temperature of 298 K. The amounts of Fe₃O₄/HNTs and H₂O₂ were increased because, according to the monitoring design, a higher amount of Fe₃O₄/HNTs and H₂O₂ (Figure 12b) favors IBU removal (

Table 12. Code and levels of Box–Behnken design for degradation of IBU.

Factor	Code	Low Level (−1)	Central Point	High Level (+1)
H2O2	X1 = A	0.25	0.5	0.75
Fe3O4/HNTs (g L−1)	X2 = B	0.5	1.5	2.5

). The results of the removal of IBU are shown in

Table 13. Degradation efficiency of IBU on Fe₃O₄/HNTs.

Run	H2O2 (M)	Fe3O4/HNTs (g L−1)	Removed IBU (%)
1	0.25	0.5	87
2	0.25	1.5	94
3	0.25	2.5	89
4	0.50	0.5	89
5	0.50	1.5	99
6	0.50	1.5	99
7	0.50	1.5	97
8	0.50	1.5	98
9	0.50	2.5	90
10	0.75	0.5	91
11	0.75	1.5	98
12	0.75	2.5	89

.

The removal design results were fitted to the Linear, 2FI, Quadratic, and Cubic equations (

Table 14. Lack-of-fit test of the optimization models for the removal of IBU onto Fe₃O₄/HNTs.

Source	Sequential p-Value	Lack of Fit p-Value	Adjusted R2	Predicted R2
Linear	0.7970	0.0141	−0.1621	−0.7955
2FI	0.7379	0.0113	−0.2881	−2.6005
Quadratic	<0.0001	0.5212	0.9251	0.8204
Cubic	0.8415	0.2246	0.8969	−1.0033

). The data were best described by the quadratic model, with a standard deviation of 1.29 and adjusted and predicted R² values of 0.9251 and 0.8204, respectively. These results indicate reasonable agreement, suggesting a good predictive capability of the model, with a model determination coefficient (R²) of 0.9848. Additionally, the model showed a non-significant lack of fit (p-value = 0.5212), indicating no statistical evidence that the model is poorly fitted; therefore, it is appropriate and reliable.

The Model F-value of 28.16 implies the model is significant. (

Table 15. Analysis of variance (ANOVA) for the Box–Behnken experimental design for the IBU removal with Fe₃O₄/HNTs.

Source	Sum of Squares	Df	Mean Square	F-Value	p-Value
Model	234.83	5	46.97	28.16	0.0004
A-H2O2	11.56	1	11.56	6.93	0.0389
B-Fe3O4/HNTs	0.4760	1	0.4760	0.2854	0.6124
AB	3.44	1	3.44	2.06	0.2009
A2	6.07	1	6.07	3.64	0.1050
B2	167.64	1	167.64	100.51	<0.0001
Residual	10.01	6	1.67	--	--
Lack of fit	4.84	3	1.61	0.9356	0.5212
Pure Error	5.17	3	1.72	--	--
Corr total	244.84	11	--	--	--

). There is only a 0.04% chance that an F-value this large could occur due to noise. p-values less than 0.0500 indicate model terms are significant. In this case, A and B² are significant model terms. Values greater than 0.1000 indicate the model terms are not significant. If there are many insignificant model terms (not counting those required to support hierarchy), model reduction may improve your model. The Lack of Fit F-value of 0.94 implies the Lack of Fit is not significant relative to the pure error. There is a 52.12% chance that a Lack of Fit F-value this large could occur due to noise. Non-significant lack of fit is good—we want the model to fit. The variables in terms of coded factors are shown in Equation (14).

$$\mathrm{I}\mathrm{B}\mathrm{U} \mathrm{r}\mathrm{e}\mathrm{m}\mathrm{o}\mathrm{v}\mathrm{e}\mathrm{d}\left(\%\right)=98.60+1.39\mathrm{A}+0.2817\mathrm{B}-0.9275\mathrm{A}\mathrm{B}-1.51A^2-7.93B^2$$

(14)

At pH 7, both the H₂O₂ concentration and the squared dose of the Fe₃O₄/HNTs catalyst were the most influential factors for IBU removal. Hydroxyl radicals are generated from hydrogen peroxide, so increasing its concentration enhances the removal efficiency. Equation (14) accurately describes the system within the concentration range shown in the response surface (Figure 12d), where 99% degradation is achieved at 1.5 g L⁻¹ of Fe₃O₄/HNTs and 0.5 M H₂O₂.

3.7. Mineralization of IBU

Based on the initial and final TOC values of 4.05 ppm and 1.61 ppm, respectively, the percentage of mineralization was determined using Equation (10), resulting in 60% mineralization when the sample achieved 99% IBU removal.

3.8. Catalyst Reuse

The catalyst was reused for five cycles (Figure 13), with the amount of IBU removed decreasing from the second cycle onward. The removal efficiency dropped to 70% in the fifth cycle. It is noteworthy that the efficiency remained nearly constant during the first three cycles, which can be attributed to the mild conditions of the Fenton reaction under optimal parameters, maintaining the catalyst’s stability [94,95].

Table 16. Performance comparison and reaction conditions of Fe₃O₄/HNTs and other Fe-based catalysts.

Catalyst	Process	IBU Removed	Experimental Condition	Reference
Fe (ll)	Sonolysis and sono-Fenton	50%	pH (2.6–8.0), [Fe2+] = 10 mg L−1, 3 h	[35]
Fe(III)-gallic acid complex	Homogeneous modified Fenton-like oxidation	90.9%	pH 7, Room temperature, [IBU]o = 0.05 M, [FeIII-GA complex] = 0.1 mM	[36]
Activated carbon fibers (ACFs) supported ferric citrate (Cit-Fe/ACFs)	Electro-Fenton	97%	120 min, current density of 7 mA cm−2	[95]
Zero-valent iron	Electro-Fenton	92%	pH 6, current density: 0.5 mA cm−2, time 1 h, [H2O2] in excess	[40]
Zero-valent iron (metallic Fe)	Electro-Fenton	80%	pH 6, [H2O2] = 50 μM, dosage ZVI 0.01 g L−1	[40]
Fe-ordered mesoporous carbon (OMC)	Plasma-supported Fenton reactions	83%	[IBU]o = 50 mg/L, [H2O2]o = 43.3 mg L−1, 2 h, [H2O2] = 21.9 mg L−1	[38]
Fe-zeolite	Plasma-supported Fenton	88%	Fe-zeolite = 1–5 g L−1, “natural” pH	[35]
Fe/ZrO2	Heterogeneous-Fenton	degradation (98%), mineralization (40%)	H 5, 343.15 K, [H2O2] (3%) 25 mL L−1, (0.0880 mol L−1), [IBU]o = 10 mg L−1, 2 h	[96]
Fe/ZrO2	Fenton-like process	80%	pH 3, 343.15 K, [H2O2] = (3%) 30 mL L−1, [IBU]o = 10 mg L−1, 2 h, Fe/ZrO2 = 400 mg L−1	[97]
Iron-based MOF	Photo-Fenton	80%	pH 7.0, 296.15 ± 2 K [IBU]o = 15 mg L−1; [Fe3O4] = 1.84 g L−1; [H2O2]o = 600 mM;	[98]
Fe3O4	Heterogeneouso Fenton-like oxidation	60%	pH 7.0, 293.15 K, [H2O2]o = 10 mM, Fe3O4 = 1 g L−1, 50 h,	[41]
Fe3O4/clay slurry	heterogeneous Fenton-like oxidation	90%	pH 7.0, 296.15 ± 2 K, [IBU]o = 15 mg L−1, Fe3O4 = 2.0 g L−1, [H2O2]o = 600 mM; clay = 4.0 g L−1;	[42]
Carbon dots/Fe3O4@CS	Heterogeneous Fenton-like oxidation	90%	323.15 K, Persulfate = 5 mmol L−1, [IBU]o = 50 μmol, Fe3O4@CS = 0.3 g L−1, 2 h	[43]
Humic acid-coated magnetic particles	Heterogeneous photo-Fenton	80%	pH 3, [H2O2] = 1.0 mmol L−1, [Fe3O4/0.5HA] = 100 mg L−1, [IBU]o = 0.2 mmolL−1	[44]
Pd@Fe3O4	Sono-electrolytical Fenton	3–100%	pH 3, 5.2 y 11, 298.15 K, [H2O2] = 3 mg L−1: Pd@Fe3O4 = 0.1 g L−1; [IBU]o = 0.2 mg L−1, 1 h,	[45]
Fe3O4/MWCNTs	Heterogeneous Fenton-like oxidation	88.7%	pH 3; 20 h	[46]
Fe3O4/HNTs	Heterogeneous Fenton-like	99% Mineralization (60%)	pH 7.0, 293.15 K, [IBU]o = 15 mg L−1, [H2O2] = 0.5 mol L−1, Fe3O4/HNTs = 1.5 g L−1, 24 h	This work

presents advanced processes for the removal and/or mineralization of ibuprofen (IBU). Reported IBU removal efficiencies range from 60% to 98%, with mineralization up to 40%, whereas in our study, a 99% removal and 60% mineralization were achieved. However, some of the reported processes are carried out under extreme pH conditions or require high energy input.

The optimal IBU removal conditions obtained in this study using Fe₃O₄/HNTs were at mild conditions: pH 7, compared to reported processes at pH 3 and 5 [44,45,46,96,97]; temperature of 293.15 K, versus 323.15 K and 343.15 K reported in previous studies [43,97]; lower H₂O₂ concentration (0.5 M) compared to 0.6–0.8 M [96,98]; shorter reaction time (20 h) compared to 50 h for unsupported Fe₃O₄ [41]; and a smaller catalyst dosage (1.5 g L⁻¹ of Fe₃O₄/HNTs) compared to 6 g L⁻¹ of Fe₃O₄/Clay slurry [42].

Furthermore, the combined use of response surface methodology (RSM) and the Box–Behnken design to independently and systematically optimize the adsorption and oxidation stages provides a more comprehensive understanding of the process. Importantly, the system achieves near-complete degradation and partial mineralization under close-to-neutral pH conditions, representing a significant practical advantage over conventional Fenton-based methods, which typically require strongly acidic environments.

Unlike many reported systems (Table 16), where adsorption plays a dominant role, the present material exhibits a relatively low adsorption capacity, which is advantageous for keeping the contaminant in solution and, therefore, available for oxidative degradation. This characteristic allows for more efficient use of the catalytic surface for the generation of hydroxyl radicals.

While the results demonstrate high efficiency in ibuprofen removal, the adsorption process was relatively slow, requiring prolonged contact times to reach equilibrium. Furthermore, although the catalyst showed promising performance, its long-term stability and iron leaching potential were not systematically evaluated in this work. Another aspect that deserves attention is the relatively high concentration of H₂O₂ required to achieve optimal performance, which could have implications for large-scale applications.

Therefore, future research will focus on improving the stability and reusability of the catalyst, evaluating metal leaching under different operating conditions, and optimizing oxidant consumption. As well as identifying the by-products and evaluating their toxicity. Furthermore, it will be crucial to evaluate this system in real wastewater matrices and apply it to other emerging contaminant classes to further validate its practical applicability.

4. Conclusions

In this study, a Fe₃O₄/HNTs composite was synthesized as an alternative adsorbent for the removal of ibuprofen, an emerging environmental contaminant. Elemental analysis using EDS and XPS confirmed the presence of iron, oxygen, silicon, and aluminum in the catalyst, while XRD characterization verified the formation of crystalline Fe₃O₄. The composite exhibited Fe₃O₄ agglomerates supported on the HNTs and demonstrated superparamagnetic behavior. Additionally, the material displayed a meso- to macroporous structure and a negative zeta potential at pH 7, indicating surface stability under neutral conditions.

Adsorption experiments of ibuprofen on Fe₃O₄/HNTs were fitted using a pseudo-second-order kinetic model and a Freundlich adsorption isotherm. These results suggest that the composite possesses a highly active surface with heterogeneous adsorption energies, supporting its effectiveness for pharmaceutical removal in aqueous media.

Additionally, using response surface methodology with a Box–Behnken experimental design and ANOVA statistical analysis, it was found that the adsorption of IBU onto the Fe₃O₄/HNTs adsorbent-catalyst at 298.15 K and an initial IBU concentration of 15 mg L⁻¹ is pH-dependent, being favored under acidic conditions. Optimal adsorption was achieved at pH 2 with a Fe₃O₄/HNTs dosage of 1.5 g L⁻¹, resulting in a maximum adsorption capacity of 1.1 mg g⁻¹.

For IBU removal via the Fenton-type reaction, it was observed that at neutral pH (7) and room temperature (298.15 K), with an initial IBU concentration of 15 mg L⁻¹ and H₂O₂ concentrations ranging from 0.25 to 0.75 M, both the H₂O₂ concentration and the catalyst dosage were important factors. A maximum IBU removal of 99% was achieved under environmentally friendly pH and temperature conditions when the catalyst dosage was 1.5 g L⁻¹ of Fe₃O₄/HNTs, and the H₂O₂ concentration was 0.5 M.

Finally, TOC analysis indicated that the maximum mineralization reached 60%.

Adsorption studies showed partial efficiency, while heterogeneous Fenton-type reactions achieved complete contaminant removal. This methodology offers the advantage of recovering the adsorbent/catalyst through a magnetic field. These results confirm the material’s potential as an efficient alternative for treating water contaminated with IBU, helping to mitigate its environmental impact and health risks.

This study contributes to the development of Fe₃O₄/HNTs composites, which can function as efficient, magnetically recoverable catalysts for the removal of pharmaceutical contaminants such as ibuprofen. The magnetic recoverability of the catalyst, along with the use of low-cost halloysite nanotubes, enhances the potential scalability of the process. Furthermore, the catalytic mechanism, based on the generation of hydroxyl radicals, suggests that this approach could be applied to a broader range of organic contaminants with similar structural characteristics.