Understanding adsorption mechanisms and metal ion selectivity of superparamagnetic beads with mesoporous CMK-3 carbon and commercial activated carbon

Microporous and Mesoporous Materials 374 (2024) 113159 Available online 3 May 2024 1387-1811/© 2024 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Understanding adsorption mechanisms and metal ion selectivity of superparamagnetic beads with mesoporous CMK-3 carbon and commercial activated carbon Lisandra de Castro-Alves a , * , Susana Y´ a˜ nez-Vilar a , Manuel A. Gonz´ alez-Gom´ ez a , b , Pelayo Garcia-Acevedo a , ´ Angela Arnosa-Prieto a , Yolanda Pi˜ neiro-Redondo a , ** , Jos´ e Rivas a a Nanotechnology and Magnetism Lab — NANOMAG, Materials Institute – iMATUS, Health Research Institute - IDIS, Department of Applied Physics, Universidad de Santiago de Compostela, E-15782, Santiago de Compostela, Spain b Nanostructured Materials Group, International Iberian Nanotechnology Laboratory-INL, Av. Mestre Jos´ e Veiga s/n, 4715-330, Braga, Portugal ARTICLE INFO Keywords: Heavy metals Adsorption Sodium alginate Multimetal Water CMK-3 mesoporous carbon ABSTRACT This study investigates, the adsorption capacities of synthesized magnetic hybrid beads containing ordered mesoporous carbon (CMK-3) and commercial activated carbon (commercial AC) for Cd (II), Ni (II), and Hg (II) ions adsorption in single and ternary systems. The order of maximum adsorption capacity was Cd (II) >Ni (II) > Hg (II) in both beads and systems. L13 beads containing CMK-3 exhibit superior adsorption efficiency compared to L3 beads with commercial AC, attributed to the large surface area, enhanced accessibility to adsorption sites, and the presence of O=C bonds, as well as other elements such as F − . Adsorption of all metal ions was described by Freundlich isotherm in L3 beads in both systems, meanwhile in L13 beads, Langmuir and Freundlich models described adsorption differently in both systems and among metal ions, indicating a more complex process. According to the reported maximum adsorption capacity values, L3 beads exhibit a greater affinity for Hg (II) ions, while L13 beads show a higher affinity for Cd (II) and Ni (II) ions. Adsorption in both beads occurred through several mechanisms involving surface complexation, ion-exchange, precipitation, physical and chemical processes. These findings highlight the importance of material design in optimizing adsorption performance for environmental applications, with potential for further enhancement of adsorption capacities and selectivities through future research. 1. Introduction The contamination of environmental systems by heavy metals such as mercury, cadmium, and other metal ions poses a significant environmental challenge. Elevated concentrations of these heavy metals induce toxic effects and lead to their accumulation within the food chain, thereby impacting aquatic ecosystems and posing risks to human health. Consequently, effective remediation strategies are essential for the removal of these pollutants from water. Recent efforts have focused on the development of cost-effective and highly efficient adsorbents, including activated carbon [1], polymers [2–5], and aluminosilicates [6], due to their cost-efficiency, versatility, and efficacy [7]. Activated carbon, characterized by its prevalent microporous structure with pore sizes typically below 10 nm, is a widely recognized and extensively utilized adsorbent. However, alternative carbon-based materials such as carbon nanotubes, carbon aerogels, and ordered mesoporous carbon have garnered significant interest over time [8]. Ordered mesoporous carbons (OMCs) can be synthesized through two main routes: “hard-templating,” which involves the carbonization of a template, and “soft-templating,” a simpler one-step method utilizing triblock copolymers as templates [9]. OMCs exhibit chemical inertness, robust mechanical strength, thermal stability, an ordered pore structure, and a hydrophobic surface, facilitating the effective adsorption of long-chain organic molecules [10]. In recent years, CMK-3, a specific type of OMC, has widespread application in adsorption [11–13] and catalysis [12], including CO 2 adsorption from industrial coal ash [14] and the removal of NOx at low temperatures [15]. Nevertheless, CMK-3 and powdered materials used as adsorbents suffer from drawbacks such * Corresponding author. ** Corresponding author. E-mail addresses: [email protected] (L. de Castro-Alves), [email protected] (Y. Pi˜ neiro-Redondo). Contents lists available at ScienceDirect Microporous and Mesoporous Materials journal homepage: www.elsevier.com/locate/micromeso https://doi.org/10.1016/j.micromeso.2024.113159 Received 13 March 2024; Received in revised form 16 April 2024; Accepted 1 May 2024 Microporous and Mesoporous Materials 374 (2024) 113159 2 as particle size, low mechanical strength, and difficulties in separation from aqueous matrices [16]. These challenges can be mitigated through the immobilization of adsorbents onto porous natural polymers, enhancing characteristics such as porosity, rigidity, ease of separation, and mechanical strength. Alginate, a naturally occurring polysaccharide derived from brown algae, is a versatile biopolymer utilized as an immobilization matrix due to its inexpensive production, non-toxic nature, and biodegradability [17]. Numerous alginate-based adsorbents exhibit high affinity and binding capacities for metal ions in aqueous solutions, primarily attributed to the abundance of carboxylic and hydroxyl functional groups [18]. However, the separation of alginate-loaded materials presents a challenge. Thus, a novel strategy involves incorporating magnetite nanoparticles (Fe 3 O 4 NPs) into a polymer matrix such as sodium alginate, resulting in hybrid magnetic beads. This approach offers an effective and rapid means of magnetic separation following the removal of metal ions from water [19–21]. The removal of heavy metals from mixed systems is crucial as wastewater often contains multiple metals, leading to competition for adsorption sites. Investigating adsorption interactions among different metals within mixed systems is vital for predicting metal adsorption in realistic scenarios and optimizing experimental studies. Various isotherm models are employed to elucidate and analyze the adsorption of contaminant systems at equilibrium [22,23]. The Langmuir and Freundlich isotherm models are commonly used to predict the adsorbed material’s quantity as a function of concentration at constant temperature [24]. Usually, the mentioned adsorption models are solved by linearization, and the linear regression coefficient (R2)close to unit is used as indicative of best fittingness of the model. Nevertheless, transformation of the nonlinear equations into their linear form can lead to errors, and potentially violate the assumptions of standard least squares [25]. Recently, more investigations are relaying on the nonlinear method since it proved to give better fitting results than the linear method for isotherm models [26,27]. This research aimed to study the adsorption capacity of magnetic alginate beads incorporating CMK-3 mesoporous carbon and commercial activated carbon (commercial AC) for the removal of metal ions from mixed solutions. The physicochemical properties of both beads and carbons were fully studied to understand the metal ions selectivity and adsorption mechanisms in the ternary system. Besides this, it was determined the best fitting relationship between adsorbate and adsorbent based on error calculation between the experimental and predicted equilibrium adsorption data of the linear and nonlinear isotherm models of Langmuir, Freundlich, Dubinin-Kaganer-Radushkevich (DKR), and Temkin. The developed magnetic alginate beads offer a practical solution for the removal of heavy metal ions from mixed solutions, with potential implications for enhancing water purification and environmental remediation efforts. 2. Materials and methods 2.1. Materials All chemicals used were of analytical grade and without purification. Ferric chloride (FeCl 3 ⋅6H 2 O) was obtained from Alfa Aesar (Madrid, Spain), cadmium chloride hemi(pentahydrate) (CdCl 2 ⋅2.5H 2 O, 99 %), and mercury (II) chloride (HgCl 2 , 99.5 %), from Acros Organics (Geel, Belgium), nickel (II) chloride hexahydrate (NiCl 2 ⋅6H 2 O, 99.9 5 %) from Alfa Aesar, (Kandel, Germany), surfactant Tween 20 from Fluka (Steinheim, Germany), commercial activated carbon powder (MW = 12.01 g) was obtained from PANREAC (Madrid, Spain), ferrous sulfate (FeSO 4 ⋅7H 2 O), calcium chloride (CaCl 2 ), ammonium hydroxide (NH 4 OH, 28 %), triblock copolymer Pluronic P123 (PEO20-PPO70PEO20), tetraethyl orthosilicate (TEOS, 98 %), sucrose (99 %), sulphuric acid (H 2 SO 4 , 95–97 %), Hydrofluoric acid (HF, 48 %), hydrochloric acid (HCl, 37 %), and sodium alginate were purchased from Sigma Aldrich (Saint Louis, MO, USA). 2.2. Synthesis of mesoporous silica (SBA-15) and mesoporous carbon (CMK-3) SBA-15, was prepared according to the procedure described by Zhao et al. [28] with some modifications, using triblock copolymer, P123 as the surfactant, and TEOS as the silica source. Typically, 4.0 g of P123 were dissolved in 150 mL of HCl (1.6 M) at 35 ◦C for 3 h. Posteriorly, 8.5 g of TEOS was added dropwise under vigorous stirring (500 rpm) for 20 h at 35 ◦C. The resultant white precipitated was aged at 80 ◦C for 20 h without agitation. The sample was filtered, washed abundantly with water, and calcined at heating rate of 1 ◦C/min to a temperature of 500 ◦C for 6 h, and cool down at a rate of 3 ◦C/min. CMK-3, was synthesized using SBA-15 as hard template impregnated with a sucrose solution, according to previous work [29]. Briefly, 1.0 g of SBA-15 was added to a previously prepared solution of 1.25 g of sucrose, and 0.14 g of H 2 SO 4 in 5.0 g of H 2 O. The mixture was place in a drying oven for 6 h at 100 ◦C (1 ◦C/min), and subsequently to 160 ◦C for 6 h (1 ◦C/min), after that it was cold down at a rate of 25 ◦C/min. The sample was treated again at 100 ◦C and 160 ◦C with the temperature rates, after the addition of 0.8 g of sucrose, 0.09 g of H 2 SO 4 in 5.0 g of H 2 O. The sucrose-SBA-15 sample was then carbonized under nitrogen flow with a heating rate of 2 ◦C/min to 500 ◦C and then 5 ◦C/min to 900 ◦C for 6 h, cooling down at a rate of 3 ◦C/min). Afterwards, present silica in the sample was removed using a solution of 10 wt% HF under magnetic agitation for 24h at ambient temperature. The sample was then filtered, washed with water and dried at 120 ◦C for 12 h. 2.3. Synthesis of magnetic alginate beads Magnetic alginate beads were prepared by extrusion method of Fe 3 O 4 NPs with sodium alginate using calcium chloride as cross-linking agent. Firstly, Fe 3 O 4 NPs were synthesized by reverse coprecipitation method [21]. Briefly, 15 mL of 1.0 M FeCl 3 ⋅6H 2 O, and 0.5 M FeSO 4 ⋅7H 2 O were mixed. Afterwards, the iron salt solution was added dropwise into a 20 mL of 3.5 M NH 4 OH at 60 ◦C, and mechanically stirred for 30 min. The magnetic nanoparticles were then washed, and re-dispersed in distilled water. For the beads, 2.0 g of sodium alginate was subsequently added to the previously prepared Fe 3 O 4 NPs solution (35 mL). Two sets of magnetic beads were prepared, one with commercial AC, and another with previously prepared, CMK-3 mesoporous carbon. For this, 3.0 g of commercial AC (34.3 wt%), and 0.2 g of CMK-3 (7.69 wt%), were added to sodium alginate/Fe 3 O 4 NPs solutions, and mechanically stirred for 4 h. The obtained magnetic solutions were added dropwise into 100 mL coagulation bath of 0.13 M CaCl 2 with 450 μ l Tween 20 under continuous magnetic agitation (450 rpm) using a New Era NE-300 syringe pump. Beads were instantly formed and left in the bath for 20 min. Afterwards, the beads were cleaned with distilled water, and dried overnight at 60 ◦C. The beads with commercial AC, and CMK-3 mesoporous carbon, were named as L3 and L13, respectively. 2.4. Adsorption studies Batch adsorption studies of Cd (II), Ni (II), and Hg (II) were performed to determine the adsorption capacity of the beads in a single and ternary system. Solutions of 15 mL, with initial metal concentration between 10 and 250 mg/L and pH value of 4.5. The effect of the metal ions initial concentration was studied with the following conditions (magnetic agitation of 300 rpm, 14 mg of adsorbent dosage, and a contact time of 6 h). Metal concentration in solution was determined by ICP-OES. Metal ions removal efficiency was calculated using the following Eq. (1): L. de Castro-Alves et al. Microporous and Mesoporous Materials 374 (2024) 113159 3 %R=(C0−Ce) C0 x100 (1) The equilibrium adsorption capacity, qe, was determined using Eq. (2): qe=(C0−Ceq)∗V m(2) Where, R (%) is the removal efficiency, C0 (mg/L) is the initial metal concentration, Ce (mg/L) is the equilibrium metal concentration present in solution, V (L) is the volume, m (g) is the adsorbent dry weight [30]. 2.5. Adsorption isotherm models A set of isotherm models have been developed to describe the adsorption phenomena. Langmuir, Freundlich, Temkin, or DKR isotherm models are some of the few that consider distinct characteristics of both the adsorbent material and the adsorbed species. 2.5.1. Langmuir isotherm model The Langmuir isotherm model assumes a monolayer, and homogeneous adsorption onto a surface. The adsorbent surface contains a finite number of adsorption sites that possess an equal affinity for the adsorbate. Once a site is filled, no further adsorption can occur in that site, indicating surface saturation. The isotherm is expressed by the following non-linear Eq. (3) [23]: qe=qmKLCe 1+KLCe (3) The linear form of the Langmuir isotherm is given by Eq. (4): Ce qe=1 KLqm+Ce qm (4) where, qe (mg/g), is the amount adsorbed, Ce (mg/L), is the adsorbate concentration in solution, both at equilibrium, KL (L/mg), is the Langmuir adsorption constant, and qm (mg/g), is the maximum adsorption capacity for monolayer formation on the adsorbent. The adsorption equilibrium data was plotted as Ce/qe against Ce. 2.5.2. Freundlich isotherm model The Freundlich isotherm assumes that the adsorption takes place on a heterogeneous multilayer surface with no limited degree of adsorption sites, and the energy is non-uniformly distributed. The isotherm is represented by the non-linear Eq. (5) [31]: qe=KFC1/n e(5) The linear form of the Freundlich isotherm is given by Eq. (6): log qe=log KF+1 nlog Ce(6) where, KF and n, indicate the adsorption capacity and the adsorption linearity, respectively. The equilibrium data for the adsorption is plotted as log qe against log Ce. The two constants, KF and n are calculated from the curve. Deviation of n constant from unity indicates a nonlinear adsorption that takes place on heterogeneous surfaces [32]. If n value is equal to unity (n =1), adsorption is linear. If (n >1) adsorption is favourable, and is a physical process. Adsorption is considered to fit the Freundlich approach when exponent n takes values within the range 1–10 [33]. 2.5.3. Temkin isotherm model The Temkin isotherm model assumes that the heat of adsorption (ΔH ads ) of adsorbate present in the surface layer decreases linearly as a result of increased surface coverage [34]. Also, the equation reveals a uniform distribution of binding energies. Adsorbed quantity, qe, was plotted against ln Ce, and the constants were determined from the slope and intercept. The model is represented by the following non-linear Eq. (7) [35]. qe=btln(KTCe)(7) where, bT (J/mol), is the constant related to the adsorption heat and KT (L/g), is the Temkin isotherm equilibrium binding constant. However, is commonly used the rearranged linear form of the Temkin isotherm equation, represented by Eq. (8): qe=RT bT ln KT+RT bT ln Ce(8) where, R,(8.314 J/molK), is the universal gas constant and T (K), is the absolute temperature. 2.5.4. DKR isotherm model The DKR isotherm model assumes that the adsorption curve depends on the adsorbent’s porous structure, and that adsorption adopts a multilayer pore filling character, where electrostatic interactions (van der Waals forces) are the main physical forces during the process [36]. The non-linear equation form is expressed as: qe=qmexp (−β ε 2)(9) where, qm (mg/g), is the adsorption capacity (maximum theoretical adsorption capacity for the formation of a monolayer). β, is the DKR constant representing mean adsorption energy (mol 2 /kJ 2 ) and ε , is the Polanyi potential, which can be calculated through Eq. (10): ε =RTln(1+1 Ce)(10) The linear form of this model is described as: ln qe=ln qm−β ε 2(11) Furthermore, the E value describes the type of adsorption, if the value is under 1–8 kJ/mol or within 8–16 kJ/mol, the adsorption process is physical (electrostatic interactions) or chemical (covalent interactions), respectively. The value of E can be calculated using Eq. (12) [37]: E=1  2β √(12) 2.6. Linear and nonlinear isotherm models analysis Linear regression has been used to verify the theoretical assumptions of various models by the best fitting relationship between adsorbate and adsorbent. The accuracy of the fitting is only assessed by the coefficient of determination (R 2 ), where the values of R 2 close to unity are considered a best fit. The adsorption isotherm was solved in Origin 2018, and the model’s constants were calculated from the slope and intercept of the linear regression curves. However, the transformation of the nonlinear regression to the linear regression of the isotherm models produces bias errors (poor linearity). For this reason, to avoid these errors, the nonlinear regression method was used to determine the adsorption constants. This employed process of using the nonlinear regression instead of the linear regression has been recommended by several researchers [38,39]. The software MATLAB (Math Works Inc. Release r2020b, Perpetual License number 40903113, Universidad de Santiago de Compostela, Spain) was used to fit the experimental data to nonlinear equation models using the function (lsqcurvefit), which solves the curve-fitting data in the least-square method. Several error functions were used for isotherm equations. In each case, parameters were determined by minimizing well-known error functions to calculate the error deviation between the experimental and predicted equilibrium adsorption data of L. de Castro-Alves et al. Microporous and Mesoporous Materials 374 (2024) 113159 4 the linear and nonlinear models mentioned above. The following five statistical error functions were used: Sum of the Square of the Errors (ERRSQ), Composite Fractional Error Function (HYBRID), Derivative of Marquardt’s Percent Standard Deviation (MPSD), Sum of the Absolute Errors (EABS), and the Average Relative Error (ARE). In addition, the chi-square test ( χ 2 ) was calculated to measure the fittingness between experimental, and calculated equilibrium adsorption data, using the following Eq. (13): χ 2=∑ n i=1(qe.exp −qe.calc)2 qe.calc (13) where, n=5 corresponding to the initial metal concentrations (10, 70, 150, 200, and 250 mg/L). A significance level of 0.05 was selected since it assumed a 5 % risk of concluding that a difference exists within our experimental data. The tabular chi-square value at confidence level at 0.05 level with n−1 degrees of freedom was 9.488. Only the MSPD error and chi-square were presented on the result Table SI 1–4 for brevity. 2.7. Nanoparticles and beads characterizations Structural studies were performed by X-ray diffraction (XRD) using a Philips diffractometer (Panalytical, Callo End, UK) with Cu K α radiation (λ =1.5406 Å), with a step size of 0.02◦and a counting time of 2 s per step from 10◦to 80◦(2θ). Small-angle X-ray scattering (SAXS) patterns were recorded on a PANalytical X’Pert Powder Empyrean, with a step size of 0.01◦and a counting time of 5 s per step from 0.25◦to 6◦(2θ). The Fourier transform infrared spectra (FT-IR) was recorded on a Varian FTIR 670 (Varian, Palo Alto, CA, USA) spectrophotometer in the range 400–4000 cm −1 . Raman spectrums were obtained from a WITec Raman confocal microscope (model =ALPHA300R+) using 532 nm diode laser excitation with 0.3 s integration and 450 scans. The BET specific surface area and the pore size distribution of the samples were characterized under N 2 adsorption-desorption isotherms at 77 K using BET Micromeritics Gemini 2360 instrument (Micromeritics, GA, USA). The morphology of Fe 3 O 4 NPs, carbons, and magnetic beads were characterized by transmission electron microscopy (TEM) using a JEOL JEM1011 microscope (JEOL, Tokyo, Japan) operating at 100 kV, and by scanning electron microscopy (SEM) analysis using a ZEISS FE-SEM ULTRA Plus (30 kV) microscope (Zeiss, Oberkochen, Germany).Thermogravimetric analyses (TGA) were performed from 50 to 850 ◦C at 10 ◦C/min under nitrogen flow (20 mL/min) using a TGA PerkinElmer Pyris 7 (Perkin, Waltham, MA, USA). The composition and chemical bonding present on the samples was analyzed by X-ray photoelectron spectroscopy (XPS) using a Thermo Scientific K-Alpha ESCA instrument equipped with aluminum K α monochromatized radiation at 1486.6 eV X-ray source. Magnetic properties were evaluated through hysteresis loop measurements using a vibrating sample magnetometer (VSM) (DMS, Massachusetts, USA) at room temperature, employing an applied field ranging from −10,000 to 10,000 Oe. The concentrations of Cd (II), Hg (II), and Ni (II) ions were quantified using inductively coupled plasma optical emission spectrophotometry (ICP-OES) with an emission spectrometer, specifically the PerkinElmer Model Optima 3300 DV (Perkin, Waltham, Massachusetts, USA). Fig. 1. SEM images of commercial AC (a), SBA-15 (c), and CMK-3 (e) with TEM images of commercial AC (b), SBA-15 (d), and CMK-3 (f), respectively. L. de Castro-Alves et al. Microporous and Mesoporous Materials 374 (2024) 113159 5 3. Results and discussion 3.1. Morphological characterization The microscopy surface morphology of commercial AC, SBA-15, and CMK-3 mesoporous carbon were studied by SEM and TEM. Fig. 1a and b, show the structure of commercial AC, revealing compact layers with heterogeneous sizes ranging from approximately 40 μ m–320 μ m. Mesoporous silica SBA-15, in Fig. 1c, displays a morphological structure of interconnected short bars, which is characteristic of this material. Similarly, in Fig. 1e, the CMK-3 carbon maintains the morphology, and porous structure of its silica precursor. The presence and arrangement of the porous systems in both SBA-15 and CMK-3 was clearly observed, revealing a well-ordered assembly of pores in Fig. 1d and f, respectively. Fig. 2, shows the SEM micrographs of the external surface and crosssection of the magnetic beads, L3 and L13. Both beads exhibit a spherical morphology, differing significantly in surface, and internal structure. In Fig. 2a, L3 magnetic beads have a porous layer-like surface structure, while, Fig. 2c reveals that the L13 magnetic beads surface features a steep porous valleys indicative of CMK-3 mesoporous carbon. Crosssection SEM images in Fig. 2b and d show the internal architecture of L3 and L13 beads, respectively. Both demonstrate highly porous internal structure, however L3 (Fig. 2b) shows a heterogeneous layer-by-layer arrangement, and manner, L13 (Fig. 2d) consisting of interconnected pores. 3.2. Structural, textural, and microstructural characterization The structural properties of silica SBA-15, commercial AC, and CMK3 mesoporous carbon were characterized by SAXS to assess the porous structure and to get more information about nanometer-scale heterogeneities. Fig. 3a, show diffraction peaks of the precursors used, SBA-15, CMK3, and commercial AC. The SBA-15 spectra reveal three diffraction peaks at 2θ =0.95, 1.65 and 1.91◦that can be indexed to (100), (110) and (200) reflections. This typical pattern of SBA-15 indicates a well-ordered hexagonal mesoporous structure with high degree of symmetry. However, in CMK-3 spectra it is only observed a peak at 2θ =0.95◦indexed to (100) suggesting a less ordered pore structure compared to the SBA15. Meanwhile, absent diffraction peak in the commercial AC diffractogram indicates that no crystalline phase is present in the sample. Raman spectra in Fig. 3b, shows two distinctive peaks at 1346 cm −1 and 1594 cm −1 , which are characteristic of polycrystalline graphite, referred as D peak (disordered induced phonon mode, sp 3 ), and of a single graphite crystal, which is referred as G peak (graphite lattice mode, E 2g sp 2 ), respectively. The D peak corresponds to the structural defects of carbon during carbonization, while G peak indicates stretching vibration of C–C bonds. The ratio between D and G peaks (I D /I G ) used to measure the crystalline dimension and in-plane defects in carbon materials was equal to 0.366 in commercial graphite [40]. In this study, the (I D /I G ) ratio for the CMK-3 and commercial AC was 0.84 and 0.77, respectively, revealing low graphitization degree compared to the commercial graphite value. However, our synthesized CMK-3 carbon (I D /I G ) value of 0.84 is reportedly lower compared to several studies which reported values higher than 1.0 [41,42]. Commercial AC and CMK-3 carbon were used as precursors for the synthesis of magnetic beads, L3 and L13, respectively. The X-ray diffraction pattern of the beads is shown in Fig. 4a, where five characteristic peaks appearing at 2θ =30.1, 35.8, 43.1, 57.2, and 63.1◦can be indexed to (220), (311), (400), (511), and (440) planes corresponding to the magnetite crystalline phase (ICDD: 98-015-8743). The beads chemical topography was analyzed by FTIR spectroscopy as shown in Fig. 4b. Diffraction peaks at 1585 and 1288 cm −1 correspond to symmetric and asymmetric C–H stretching vibrations of the carboxyl groups of sodium alginate, respectively. In addition to the broad peak at 3228 cm −1 and at 1076 cm −1 , which are attributed to the –OH and C–O–C stretching vibrations, respectively [43]. One important band at 558 cm −1 was detected, which corresponds to the stretching vibration of Fe–O bonds in Fe 3 O 4 NPs [44]. The surface properties of both carbon materials and beads were further investigated by full scan-XPS. Fig. 5a presents the survey scan spectra obtained from the surfaces of commercial AC and CMK-3. In both samples, the carbon (C) signal dominates due to their predominantly carbonaceous nature. Quantitative analysis of the elemental composition revealed atomic percentages of 89.9 % and 91.5 % for carbon in commercial AC and CMK-3, respectively. Additionally, it was also detected as major elements, oxygen (6.8 % and 5.6 %), nitrogen (1.8 % and 0.7 %), and fluor (0.78 % and 1.48 %). Minor elements such as sulfur, phosphorous, and chlorine were also detected, each with atomic percentages lower than 0.5 % for both commercial AC and CMK-3, Fig. 2. SEM images of the surface and cross-section structure of L3 (a, b), and L13 (c, d) magnetic beads, respectively. L. de Castro-Alves et al. Microporous and Mesoporous Materials 374 (2024) 113159 6 respectively. Fig. 5b presents the survey scan of beads L3 and L13, revealing their elemental composition. Oxygen, carbon, and calcium are identified as major elements, with respective atomic percentages of 35.3 %, 56.2 %, and 2.93 % for L3 beads, and 36.5 %, 57.5 %, and 2.93 % for L13 beads. The presence of calcium, as well as minor elements such as iron and sodium, is attributed to sodium alginate and Fe 3 O 4 NPs. In Fig. 5d, the C 1s peak is deconvoluted into four main peaks for both carbon materials. The strongest peak at 248 eV corresponds to graphitic carbon (C-sp 2 ), followed by peaks representing the (C–C) bond at 284.8 eV, the (C–O) bond associated with alcohols or ether groups at 286.4 or 286.6 eV, and the (C=O) bond of carbonyl or carboxyl groups at approximately 289 eV. Commercial AC exhibits a higher proportion of (C–C) bonds (18.6 %) compared to CMK-3 (13.1 %), whereas CMK-3 display relatively higher percentages of (C–O) and (C=O) bonds (12.0 % and 5.8 %) compared to commercial AC (10.4 % and 4.9 %), respectively. Furthermore, the high-resolution C 1s spectra of beads L3 and L13 in Fig. 5d also indicate the presence of (C–C), (C–O), and (C=O) bonds. L3 beads exhibit the highest percentage of (C–H, 9.5 %) and (C–OH, 18.8 %) bonds, compared to L13 beads, with (C–H, 7.6 %) and (C–OH, 17.1 %). Conversely, the carbonyl bond (C=O) percentage in L13 beads (67.3 %) notably exceeds that of L3 beads (56.8 %). 3.3. Analysis of hysteresis loops Hysteresis loops of magnetic beads were measured using a vibrating sample magnetometer, which operated within the temperature range of room temperature (−10000 Oe to 10000 Oe). Fig. 6 displays the magnetization curves for Fe 3 O 4 NPs, L3, and L13 beads. The magnetization data was subjected to a normalization process relative to the magnetic mass content, determined through TGA. Specifically, the magnetic content percentages were quantified as 86 % for Fe 3 O 4 NPs, 23 % for L3 beads, and 43 % for L13 beads. This methodological approach ensures accurate comparison of the magnetic properties of all samples, contributing to a comprehensive understanding of their characteristics. The saturation magnetization (Ms) values obtained were 59.29 emu/ g Fe 3 O 4 for Fe 3 O 4 NPs, 48.62 and 49.43 emu/g Fe 3 O 4 for L3 and L13 beads, respectively. Comparatively, the magnetization of the beads was lower than that of Fe 3 O 4 NPs. Notably, negligible coercivity (Hc <30 Fig. 3. Small angle X-ray diffraction patterns (a) and Raman spectra (b) of SBA-15, CMK-3 carbon and commercial AC. Fig. 4. (a) X-ray diffraction (a) and Infrared spectra (b) of sodium alginate and of the magnetic beads L3 and L13. L. de Castro-Alves et al. Microporous and Mesoporous Materials 374 (2024) 113159 7 Oe) and absent remanence were observed, indicating minimal resistance to changes in the magnetic field strength. This reduction in magnetization aligns with earlier studies, that have observed analogous variations attributing them to the effect of NPs encapsulation by a polymer [21, 45]. However, it is essential to highlight the preservation of the superparamagnetic behavior of Fe 3 O 4 NPs within the material. This unique property is an advantage preventing magnetic aggregation and subsequent flocculation in the absence of an external magnetic field. The magnetic material demonstrates remarkable suitability for rapid separation from solutions, providing significant practical utility across diverse industrial applications. 3.4. Surface area and pore volume Fig. 7 presents a comparative analysis of the N 2 adsorptiondesorption isotherm curves obtained for the precursors SBA-15, CMK3, and commercial AC, along with the magnetic beads L3 and L13. The N 2 adsorption-desorption curve of commercial AC and L3 beads exhibits characteristics of a type II isotherm with a H3 hysteresis loop, as shown in Fig. 7a. This type of isotherm and hysteresis loop is frequently associated with nonporous powders or microporous structures with nonrigid aggregates of plate like-particles. The isotherm curve demonstrates additional adsorption above the relative pressure (p/p 0 ) of 0.8, which is indicative of interparticle void filling. Fig. 7b shows the N 2 adsorption isotherm of SBA-15, which has a type IV isotherm with H1 hysteresis, suggesting pore uniformity [18]. Conversely, the CMK-3 carbon and respective L13 bead exhibit a pore Fig. 5. XPS survey scans (a, b) and deconvolution spectra of the C 1s (c, d) of commercial AC/CMK-3, and of L3/L13 beads, respectively. Fig. 6. Magnetic hysteresis loops of Fe 3 O 4 NPs, L3, and L13 beads at 25 ◦C. L. de Castro-Alves et al. Microporous and Mesoporous Materials 374 (2024) 113159 8 filling step with the relative pressure range (p/p 0 ) of 0.4–0.8, attributed to mesopores filling. The shape and presence of the hysteresis loop indicate that both CMK-3 and L13 bead have characteristics of a type IV isotherm with an H3 hysteresis loop. The surface areas of SBA-15, CMK-3, and commercial AC were equal to 716, 940, and 849 m 2 /g, respectively. CMK-3 carbon had a remarkably higher surface area than SBA-15 and commercial AC, suggesting differences in pore structure, and pore size distribution. Concerning the magnetic beads, L3 beads showed a surface area of 107 m 2 /g and L13 of 40.48 m 2 /g. 3.5. Adsorption studies 3.5.1. Effect of initial metal concentration To gain insights into the adsorption capacity of both beads, the influence of initial metal concentration was examined in both single and ternary systems. Aqueous solutions were prepared with initial metal concentrations ranging from 10 to 250 mg/mL, maintaining a pH value of 4.5, which had been previously in another study [20]. The experiments were carried out using 14 mg of adsorbent (L3 and L13 beads) Fig. 7. Nitrogen Adsorption-desorption isotherms of commercial AC/L3 beads (a), and CMK-3/SBA-15/L13 beads (b). Fig. 8. Adsorption capacity expressed in (mg/g) in the single (a) and ternary system (b) and expressed in (mg/g C) in the single (c) and ternary system (d) of Cd (II), Ni (II) and Hg (II) by L3 and L13 beads. L. de Castro-Alves et al. Microporous and Mesoporous Materials 374 (2024) 113159 9 under magnetic agitation with a stirring velocity of 300 rpm at room temperature. Adsorption capacity was calculated based on the carbon content present in each bead (34.3 % AC commercial in L3 beads and 7.69 % CMK-3 in L13 beads). Fig. 8 shows the adsorption capacity in (mg/g) and in (mg/g C) of both beads in the single and ternary systems. Fig. 8a and b show the adsorption capacities (expressed in mg/g) of both beads in the single and ternary systems, respectively. The adsorption capacity values exhibit clear differences in both systems. Specifically, L13 beads demonstrated superior adsorption capacity values for all metal ions compared to the L3 beads in the single system, whereas this trend was reversed in the ternary system. In the single system, L3 beads achieved the highest adsorption capacities, of approximately 55 mg/g for Cd at 250 mg/L, 35 mg/g for Ni at 250 mg/L, and 20 mg/g for Hg at 150 mg/L. Conversely, L13 beads exhibited slightly lower but still notable adsorption capacities, reaching approximately 50 mg/g for Cd at 150 mg/L, 45 mg/g for Ni at 200 mg/L, and 40 mg/g for Hg at 150 mg/L. However, in the ternary system, a decrease in adsorption capacities was observed, as shown in Fig. 8b. The maximum adsorption capacity values for L3 beads were approximately 50 mg/g at 150 mg/L for Cd, 35 mg/g for Ni at 150 mg/L, and 30 mg/g for Hg at 150 mg/L for L3 beads. In contrast, for L13 beads, the corresponding values were approximately 33 mg/g for Cd at 250 mg/L, 25 mg/g for Ni at 150 mg/L, and 3 mg/g for Hg at 200 mg/L. In order to more accurately evaluate the adsorption capacity of the beads relative to the amount of carbon in their composition, the adsorption capacity was calculated by taking into account the percentage of carbon used in both bead types. Fig. 8c and d show the adsorption capacities (expressed in mg/g C) of both beads in the single and ternary systems, respectively. The adsorption capacity of L13 beads is significantly higher compared to L3 beads, showing a hundredfold increase for all tested metal ions in both single and ternary systems. Consistently, Cd (II) showed the highest adsorption, followed by Ni (II) and Hg (II) in both systems and particle types. Adsorption of all metal ions was more favored in the single system compared to the ternary system with removal efficiency above 20 % (Supporting information, Fig. S1). In the ternary system, a slight decrease in adsorption capacity suggests a higher competition between metal ions for available binding sites. In the single system, a plateau in Cd (II) and Ni (II) adsorption was observed on the surface of L13 beads due to saturation of adsorption sites, while an increasing trend in Cd (II) and Ni (II) adsorption was observed on L3 beads, indicating ongoing adsorption. Furthermore, the amount of Hg (II) ions adsorbed decreased after reaching maximum adsorption at the initial metal concentration of 150 mg/L in the single system, a phenomenon observed in both bead types. In the ternary system, a maximum adsorption capacity of Cd (II), Ni (II), and Hg (II) at the initial concentration of 150 mg/L was observed only in L3 beads. Equilibrium was achieved faster for Ni (II) and Cd (II) ions compared to Hg (II) ions in L13 beads, whereas no equilibrium was reached in the L3 beads. The significant difference in adsorption capacity between the two bead types can be attributed to the carbon content in their composition. The mesoporous structure of CMK-3 carbon within L13 beads may have facilitated increased adsorption of metal ions due to the interconnected channels characteristic of the carbon structure. Mesoporous carbons are known for their high specific surface area, large pore volume, and superior adsorption capacity for various contaminants [46,47]. Despite containing only 7.69 wt% CMK-3 compared to 34.3 wt% commercial AC in L3 beads, L13 beads revealed superior adsorption capacity. One reason for this superior adsorption capacity can be attributed to the presence of a higher proportion of mesoporous within L13 beads structure, in contrast, L3 beads were primarily composed of microporous, as observed by the isotherm type determined by BET. 3.5.2. Metal ions competition In order to assess the competition between metal ions, the chemical states of all metal ions on the beads surface after adsorption in both single and ternary system were evaluated using XPS. Table 1 shows the binding energies and atomic percentages of Cd, Ni, and Hg in both single and ternary systems for L3 and L13 beads, providing valuable information regarding the chemical states of the metal ions on the bead surfaces under different experimental conditions. The corresponding binding energies of Cd and Ni elements shifted to lower values from the single to the ternary system at the L3 beads, suggesting competition between the metal ions. In contrast, the mercury peak (Hg 7f7/2) shifted from 101.4 eV in the single system to 101.6 eV in the ternary system, indicating that the adsorption of Hg (II) ions was favored in the presence of Cd (II) and Ni (II) ions. On the other hand, the observed shifts in binding energies at the L13 bead, from the single to the ternary system, were relatively minor for Cd (406.7 eV–406.8 eV), Hg (101.5 eV–101.6 eV), and Ni (857.4 eV–857.2 eV). This observation led us to believe that L13 beads offered a higher abundance of binding sites available for adsorption to take place in a competitive system. 3.5.3. Linear or nonlinear isotherm models: selection of best model fitting The experimental data of Cd (II), Hg (II), and Ni (II) ions adsorption by L3 and L13 beads in the single and ternary systems were fitted to the nonlinear and linear Langmuir, Freundlich, Temkin, and DKR models. The calculated parameters are listed in Table SI 1–4 and plotted in Figs. 9 and 10. In order to determine the most suitable model for describing the adsorption experimental data, error functions were computed for each model, but only χ 2 errors are displayed in the graphics. Fig. 9 compares the results obtained from both linear and nonlinear fitting of the experimental data to the Langmuir and Freundlich models. The linear Freundlich model better describes the adsorption of all metal ions by L3 beads in both adsorption systems, evidenced by the low χ 2 values. Notably, lower χ 2 values were observed in the single system compared to the ternary system, which is potentially attributed to the competitive interactions between all metals in the ternary system, as shown in Fig. 9c and d. Contrarily, the adsorption behavior of metal ions by L13 beads exhibited notable differences between the single and ternary systems. In both systems, the adsorption of some metal ions was described by both Table 1 Binding energies and atomic percents of elements Cd, Ni, and Hg in the single and ternary system of adsorption by both beads, L3 and L13. Sample Element Peak Binding energy (eV) Atomic percent (%) Single syst. Ternary syst. Single syst. Ternary syst. L3 Cd Cd 3d3/2 412.7 412.3 0 0 Cd 3d5/2 405.9 405.6 100 100 Ni Ni 2p 861.7 861.9 50.5 48.5 Ni 2p2/3 856.2 855.9 49.5 51.5 Hg Hg 4f5/2 105.3 105.5 0 0 Hg 4f7/2 101.4 101.6 100 100 L13 Cd Cd 3d3/2 413.5 413.6 0 0 Cd 3d5/2 406.7 406.8 100 100 Ni Ni 2p 862.5 862.1 41.9 57.8 Ni 2p2/3 857.4 857.2 58.1 48.3 Hg Hg 4f5/2 105.4 105.5 0 0 Hg 4f7/2 101.5 101.6 100 100 L. de Castro-Alves et al.