Enhanced removal of mercury and lead by a novel and efficient surface-functionalized imogolite with nanoscale zero-valent iron material

A novel hybrid nanomaterial, nanoscale zero-valent iron (nZVI)-grafted imogolite nanotubes (Imo), was synthesized via a fast and straightforward chemical procedure. The as-obtained nanomaterial (Imo-nZVI) was characterized using transmission electron microscopy (TEM), electrophoretic mobility (EM), and vibrating sample magnetometry (VSM). The prepared Imo-nZVI was superparamagnetic at room temperature and could be easily separated by an external magnetic field. Sorption batch experiments were performed for single- and multicomponent systems and demonstrated that Hg2+ and Pb2+ could be quantitatively adsorbed at pH 3.0. For multicomponent systems, maximum adsorption capacities of 61.6 mg·g−1 and 76.9 mg·g−1 were obtained for Hg2+ and Pb2+ respectively. It was observed that the functional groups in Imo-nZVI interact preferentially with analytes according to the Misono softness parameter. The higher performance of Imo-nZVI compared with Imo and nZVI is related to the increased number of adsorption sites in the functionalized nanomaterial. The sorption equilibrium data obeyed the Langmuir model, while kinetic studies demonstrated that the sorption processes of Hg2+ and Pb2+ followed the pseudo-second-order model. This study suggests that the Imo-nZVI composite can be used as a promising sorbent to provide a simple and fast separation method to remove Hg and Pb ions from contaminated water.


Introduction
Nanoscale zero-valent iron (nZVI) has recently emerged as a powerful and versatile sorbent for potential use in environmental engineering. Special physicochemical properties, such as a high surface-to-volume ratio, magnetism, and in situ reactivity mean that nZVI-based materials have attracted substantial interest in the scientific community. nZVI has shown application in a wide array of environmental treatments, such as soil, sediment, and groundwater remediation (Crane and Scott 2012;Li et al. 2006;Stefaniuk et al. 2016;Yan et al. 2010). However, nZVI forms aggregates due to van der Waals and magnetic forces, which decrease its efficiency by reducing its surface area and producing a less negative oxidation-reduction potential (Shi et al. 2011). Different approaches based on immobilization techniques have been developed for nZVI stabilization. Diatomite, zeolites, montmorillonite, and cellulose have been used as matrices to increase the stability of nZVI (Arancibia-Miranda et al. 2016;Bossa et al. 2017;Dror et al. 2012;Kim et al. 2013 Suazo-Hernández et al. 2020;Suazo-Hernández et al. 2019;Zou et al. 2016). Furthermore, the sorption capacity of nZVI can also be improved by synergistically combining it with other nanomaterials. In this sense, inorganic nanotubes based on aluminosilicates, such as imogolite (Imo), have attracted much research attention due to their one-dimensional structure, mechanical resistance, and reactive surfaces Arancibia-Miranda et al. 2015). Imo is a hydrated aluminosilicate with the chemical formula (OH) 3 Al 2 O 3 SiOH (Cradwick et al. 1972). Characterized as a nanotubular structure, naturally occurring Imo has a variable length from 100 nm up to several microns, with an internal diameter of 1 nm, and an external diameter of 2 nm. Synthetic Imo, however, has a diameter in the range of 2.3-2.7 nm according to the synthesis method (Arancibia-Miranda et al. 2013b). Imo is amphoteric as the inner surface of nanotubes is lined with isolated silanol (≡SiOH) groups, and the outer surface is covered with both Al-O-Al and Al-OH-Al groups ). Due to its high dispersibility in water and the amphoteric nature of Imo, which is strongly influenced by the pH of the medium (Arancibia-Miranda et al. 2015), cations should be able to interact with the inner surface and anions with the outer surface of the nanotubes (Bonelli 2016). As a result of these properties, Imo has attracted substantial interest in the scientific community for its potential application in the removal of cations and anions from polluted water Bonelli 2016;Levard et al. 2009a).
In this work, we propose the functionalization of Imo with nZVI as a strategy to obtain a powerful and new sorbent nanomaterial with magnetic properties. The combination of both nanomaterials can be used to enhance the extraction efficiency of certain analytes (Bonelli 2016).
In this context, recent research has shown that the high reactivity of nZVI promotes the stabilization of potential contaminants through multiple processes, such as adsorption, reduction, oxidation, precipitation, and coprecipitation (Jiang et al. 2018;Lu et al. 2016). However, it is important to consider that removal mechanisms for nZVI are dependent on conditions of the media (pH and redox potential (E h )), but also on the type, chemical status, and nature of the analyte. In this sense, the sorption preferences and specific interactions between the substrate and the analyte are dependent on (i) the hydrolysis constant of the metal ion (considering the presence of ≡X-OH and ≡X-O-OH groups), (ii) polarizability, (iii) ionic and hydrated radius, and (iv) the tendency to form covalent bonds. Sposito (1989) postulated a relationship for an atom, defined as the Misono softness parameter (Y), to form covalent bonds according to its ionic radius and ionization potential (Sposito 1989). Thus, hard acids have Y values less than 2.8; soft acids are higher than 3.2 and "borderline" metal ions fall in the region of 2.8-3.2 (Misono et al. 1967). Soft acids (Y > 3.2) tend to form strong complexes stabilized by large enthalpy decreases and covalent interaction. Hard acids (Y < 2.8) tend to interact primarily by coulombic interactions. "Borderline" metal ions interactions depend on specific solvent, stereochemical, and electronic configuration factors.
The combination of nZVI and Imo has been studied for the first time in this work to evaluate their affinity for highly toxic metals such as Hg and Pb. It was sought to enhance their individual capabilities in order to obtain a composite material that improves the extraction of pollutants from contaminated waters. Hg and Pb have long been a worrying issue due to their high neurotoxicity and widespread occurrence (Charlet et al. 2012;Jaishankar et al. 2014). The potential health risks from low levels of Hg and Pb in water remain a subject of intense debate. Therefore, removal of these two metals from water supplies is a current challenge. Taking these factors into consideration, along with the potential advantages resulting from the combination of Imo and nZVI, synthesis and characterization of the hybrid nanomaterial and kinetic extraction studies of Hg 2+ and Pb 2+ are presented.

Instrumentation
Elemental detection was performed using a PerkinElmer 5100ZL atomic absorption spectrometer (PerkinElmer, Norwalk, CT, USA) equipped with a pyrolytic graphite tube and a transversely heated graphite atomizer Zeeman-effect background correction system (PerkinElmer, Norwalk, CT, USA). Hg and Pb electrodeless discharge lamps (EDL) (PerkinElmer) operated at currents of 170 mA and 360 mA (modulated operation) and wavelengths of 253.7 nm and 283.3 nm, respectively, with a spectral bandpass of 0.7 nm, were used. All measurements were made based on absorbance signals with an integration time of 5 s. A centrifuge (model 5810, Eppendorf, Germany) was used to accelerate the phase separation process. A reciprocating shaker (Boeco, Hamburg, Germany) was used to mix the reagents. A Horiba F-51 pH meter (Kyoto, Japan) was used for pH determinations.

Reagents
All reagents were of analytical grade, and the presence of Hg was not detected within the working range. A 1000 μg·mL −1 Hg 2+ stock solution was prepared from mercury(II) nitrate (Merck, Darmstadt, Germany)

Synthesis of imogolite
Imo was prepared according to the procedure described in the literature (Arancibia-Miranda et al. 2011).

Synthesis of nZVI and nZVI functionalized imogolite
Pure nZVI was synthesized according to the Arancibia-Miranda et al. procedure (Arancibia-Miranda et al. 2016). The new Imo-nZVI was obtained as follows: the coating process had a theoretical 2:1 mixture of Imo/Fe (w/w), obtained by dissolving 24.41 g of Fe(NO 3 ) 3 ·6H 2 O in 100 mL of EtOH:H 2 O (90:10) in a 500-mL round bottom flask. Then, Imo (2.50 g) suspended in 100 mL of EtOH:H 2 O (9:1) was added. The suspension was stirred for 3 h, leading to a lighter orange-red solution. Then, NaBH 4 (20.0 g) dissolved in 100 mL of H 2 O was immediately added at room temperature, causing the solution to turn black within 60 s, and the reaction mixture was stirred for approximately 1 h under a N 2 atmosphere. The product was collected in a Falcon tube and centrifuged. The supernatant was removed, and the precipitated product was washed with an EtOH:H 2 O (9:1) solution several times. Finally, the product was frozen with liquid N 2 and lyophilized.

Characterization
The products were characterized by powder X-ray diffraction (PXRD), transmission electron microscopy (TEM), electrophoretic mobility (EM), and vibrating sample magnetometry (VSM). The samples were studied on a diffractometer (Shimadzu XRD-6000) with CuKα radiation (l = 1.5418 Å) in the region of 5-80° 2θ. A Zeiss EM 910 transmission electron microscope using an 80 kV acceleration potential on carbon substrates was prepared as follows: A drop of the sample suspended in water was transferred onto the face of a freshly cleaved sheet of mica, and the solvent was allowed to evaporate. Samples were characterized by a field emission scanning electron microscope (FE-SEM) model 200 Nova NanoSEM FEI Company. Analyses were performed under low vacuum conditions with Helix detector and a voltage of 10-18 kV.
The magnetic response of the samples was investigated with VSM operated at room temperature with a maximum magnetic field of 1.2 T and a sensitivity of 10 −4 emu. The isoelectric point (IEP) was determined by measuring the zeta potential (ZP) of particles on a Zeta Meter ZM-4.0 apparatus. Approximately 30 mg of each sample was suspended in 200 mL of a solution with an ionic strength of 0.01 mol·L −1 (KNO 3 ), and the EM was determined as a function of pH. The IEP was obtained from the EM versus pH graph as the pH at which EM = 0. The specific surface areas (SSAs) of Imo, nZVI, and Imo-nZVI were measured by the N 2 method of Brunauer-Emmett-Teller (BET), and pore size was calculated from Barrett, Joyner, and Halenda (BJH) analysis of N 2 adsorption/desorption isotherms at 77 K using an automatic analyzer (Quantachrome Nova Station A, Quantachrome, USA, Florida) (Table 1).

Sorption studies
Batch adsorption of Pb 2+ and Hg 2+ were studied in 40-mL polypropylene centrifuge tubes containing 10 mg of Imo and Imo-nZVI suspended in 10 mL of a Hg 2+ or Pb 2+ solution in 0.01 mol·L −1 KNO 3 . The dependence of metal adsorption as a function of time was studied in a solution of Pb 2+ or Hg 2+ , 60 mg·L −1 , in equilibrium with 10 mg of Imo and Imo-nZVI. The samples were equilibrated using a reciprocating shaker for 5 h. The temperature for the experiments was 298 K, with a pH of 3.0 ± 0.2 for sorption studies for both metals to avoid the precipitation of Hg and Pb (see Table S1) (Burriel et al. 2001). The samples were centrifuged at 10,000 rpm for 10 min, and the supernatant was filtered through 0.22mm Millex-GX membranes and analyzed by electrothermal atomic absorption (ETAAS). The amount of adsorbed metal (q t ) was determined from the difference between the initial (C o ) and final concentrations (C t ) in mg·L −1 , multiplied by the volume of solution in L and divided by the mass of adsorbent in g.

Pseudo-first-order model (PFO)
In this model, integration of kinetic rate equation gives the following: where q e and q t correspond to the amount of the metal adsorbed in equilibrium and at time t, expressed in mg·g −1 , and k 1 is the pseudo-first-order adsorption rate constant (min −1 ), which is a combination of the adsorption (k a ) and desorption (k d ) constants (Boparai et al. 2011;Lagergren 2013;Manquián-Cerda et al. 2017;Rudzinski and Panczyk 2000).

Pseudo-second-order model (PSO)
In this case, the driving force (q e − q t ) is proportional to the portion of activated sites available on the sorbent (Ho and McKay 2000;Rao et al. 2006;Rudzinski and Plazinski 2008). Integrating the kinetic rate equation gives the following expression: The initial adsorption rate of the system can be estimated by this model and is determined by h = k·q e 2 , which can be obtained directly from the intercept of the curve.

Elovich equation
This model is used to describe chemisorption processes on heterogeneous surfaces; it considers that adsorption takes place in two stages: a fast reaction, mainly associated with the movement of the adsorbate over active sites on the outer surface, and the slower diffusion into and out of the adsorbent pores Arancibia-Miranda et al. 2015;Chien and Clayton 1980;Wu et al. 2001Wu et al. , 2009. This model is expressed as follows: The constant α (mg·g −1 ·min −1 ) is regarded as the initial rate, and β (g·mg −1 ) is an indicator of the number of sites available for adsorption, related to the extent of surface coverage and activation energy for chemisorption. Parameters α and β are obtained from the plot of q t against ln(t) as the gradient and intercept, respectively.

Intraparticle diffusion model (Weber-Morris model)
Webber-Morris's pore-diffusion model is a single-resistance model derived from Fick's second law of diffusion (Cáceres-Jensen et al. 2013;Cheung et al. 2007;Weber and Morris 1963). If the sorption process is considered to be influenced by both diffusion in the sorbent, and by convective diffusion in the sorbate solution, then the mathematical dependence of sorbate uptake, q t , at t 1/2 is given by the equation: where k int (mg·g −1 ·min −1/2 ) is the intraparticle diffusion constant and C is a constant related to the thickness of the surface layer. Higher values of C, obtained as the intercept of the plot of q t against t 1/2 , signify greater boundary layer effects, which are related to the intraparticle diffusivity.

Langmuir isotherm
This model assumes that the analyte is adsorbed in a determined number of active sites, forming a monolayer. The Langmuir equation is described by the following: where q m (mg·g −1 ) is the maximum adsorption capacity, q e (mg g −1 ) is the amount of analyte adsorbed, K L is the Langmuir constant, and C e is the equilibrium molar concentration of the analyte (mg·L −1 ).

Characterization of the composite
The samples were analyzed by PXRD to evaluate the changes generated from the coating process of Imo with nZVI, and after Hg 2+ and Pb 2+ removal in both singleand multicomponent systems (Fig. S1). For Imo, four broad peaks were identified at 21.0, 12.0, 3.4, and 2.2 Å, corresponding to the (hkl) (100), (110), (001), and (211) planes, respectively Levard et al. 2009b). According to previous reports, these planes are associated with Imo in monoclinic or hexagonal packing (Arancibia-Miranda et al. 2013a;Govan et al. 2021;Kang et al. 2014). In the case of nZVI, reflections at 2θ = 45.0° (corresponding to the 110 plane) and 2θ = 65.5°  S1) in the nZVI diffraction pattern correspond to magnetite (Fe 3 O 4 ) and lepidocrocite (g-FeOOH), respectively (Manquián-Cerda et al. 2017). The diffractogram obtained for Imo-nZVI showed reflections that can be attributed to the starting materials. The nZVIdoped Imo presents with a lower degree of crystallinity in comparison to pure nZVI. Minor signs of Fe oxides were observed in the composite material, suggesting that a small percentage of nZVI oxidation occurred during the synthesis, mainly due to the traces of oxygen in the solution or by hydrolysis of water. The process of Hg 2+ and Pb 2+ removal in both single-and multicomponent systems caused significant changes in the degree of crystallinity of the composites, altering the structure of the immobilized nZVIs in Imo. After the removal process, the corrosion products magnetite (Fe 3 O 4 ) and lepidocrocite (γ-FeOOH) (Baltazar et al. 2014;Kanel et al. 2005;Liu et al. 2017;Wu et al. 2017) were identified (Fig. S1). Both species could be byproducts of the energetically favorable redox reactions between the nZVI and the analytes, as is described in Eqs. 6-9.
However, according to the significant differences in the standard redox potential for Pb 2+ /Pb 0 (−0.12 V) and Hg 2+ / Hg 0 (0.86 V) couples with respect to the Fe 0 /Fe 2+ (−0.44 V) couple, the removal of Hg 2+ occurs preferentially by a reductive precipitation at the nZVI surface (Lagergren 2013), whereas that removal process of Pb 2+ occurs by sorption with partial chemical reduction (Noubactep 2008).
It is important to highlight that in the multicomponent system, the samples exhibit a degree of amorphicity, suggesting that the presence of both metals significantly alters the surface of Imo-nZVI.
The morphologies of the samples were obtained through a detailed microscopic analysis (Fig. 1, Fig. S2). In the case of the nZVI, diameters fluctuated between 43 and 159 nm with a mean of 76 nm. The nZVI on Imo exhibits a size distribution with diameters between 19 and 70 nm, with a mean of 37 nm.
The dimensions of the pure starting materials were determined by HR-TEM images. For Imo, the external diameter was approximately 2 nm while its length exceeded 500 nm.
For nZVI, two clearly defined areas were observed. The first area presented a higher degree of crystallinity located in the nucleus which corresponds to metallic Fe. The second area, on the external surface of nZVI, was amorphous in nature and is associated with iron oxides, a phenomenon that is widely described in the literature (Kanet et al. 2006;Liu et al. 2017;Manquián-Cerda et al. 2017).
Coating Imo with nZVI resulted in a heterogeneous distribution, in which areas with a high presence of agglomerated nZVI. c Imo-nZVI nZVI are highlighted. This is due to the synthetic conditions of nZVI, where the Lewis acidic character of Fe favors positive charges at superficial sites of Imo, thereby decreasing the adsorption of this cation, a phenomenon similar to that reported by Arancibia-Miranda et al., 2014. The dimensions of nZVI were also sensitive to the process of its immobilization in Imo, and a 20% size reduction compared to pure nZVI was observed.
The surface behavior of the study materials was described through measurements of zeta potential (ZP) (Fig. 2), a technique that is sensitive to the changes in a materials composition. The IEP is considered as an indicative parameter of the changes that occur on the surface of Imo due to being coated with nZVI. Imo has a pH-sensitive surface charge which is positive over a wide pH range with an IEP value of 10.5, favoring anion adsorption (Arancibia-Miranda et al. 2011). Pure nZVI, on the other hand, has an IEP value of 7.7 which is characteristic of this type of material. Coating Imo with nZVI resulted in a reduction the IEP value of pure Imo from 10.5 to 8.3, in addition to a decrease in the magnitude of the generated surface charge. This decrease in the IEP value has a series of effects should be noted. For example, the electrostatic repulsion of the composite is lower compared to that of Imo. Consequently, the energy barrier that Hg 2+ and Pb 2+ must overcome to be absorbed by the composite is lower, since the number of neutral and negatively charged sites is lower than in Imo. Considering the textural properties, Imo-nZVI showed a decrease in surface area of 24.5% compared to Imo. However, the area is 2.7 times larger compared to nZVI, indicating that the coating process dramatically modifies the surface of Imo.
The hysteresis curve of presorption Imo-nZVI shows a typical ferromagnetic curve with a saturation magnetization (M s ) of 16.5 emu·g −1 and a coercive field (H c ) of 100 Oe (Fig. S2). The Pb postsorption magnetization curve presents a decrease in Ms to 7 emu·g −1 . This decrease in magnetization is attributed to the amount of Pb that has been absorbed by the Imo-nZVI nanoparticles, which does not contribute to the magnetization of the sample. A much larger decrease in M s is observed for the postsorption samples with Hg and Pb-Hg, with values of 0.6 emu·g −1 and 0.56 emu·g −1 , respectively. The coercivity of all postsorption samples remains close to 100 Oe. This perturbation on the magnetic properties of the materials in this study may be related to the different precipitation products at the nZVI surface. Since as we mention previously, the removal of Hg 2+ occurs preferentially by a reductive precipitation at the nZVI surface (Lagergren 2013).

Adsorption kinetics
The effect of contact time on the removal of Hg 2+ and Pb 2+ in single-and multicomponent systems was evaluated for the different materials by means of removal kinetics. In the case of Imo, an equilibrium time was achieved after 60 min for both single-and multicomponent systems. Whereas for both nZVI and Imo-nZVI, equilibrium was reached after just 30 min (Fig. 3). Pb 2+ was more effectively removed than Hg 2+ in all cases.
Kinetic models of pseudo-first order (PFO) and pseudosecond order (PSO) (Fig. 4) were used to describe the kinetic behavior of the experimental data. The PFO model demonstrated a poorer fit to the experimental data (Table 2). This behavior could be explained by the theoretical fundamentals that differentiate both models. For example, in the case of the PFO model, a closer fit is observed when the analyte is monovalent and is removed by direct interaction between the analyte and a single surface-active site of the substrate. Better fitting is obtained with the PSO model when an ion is attracted to two active sites in the substrate. The removal of divalent cations, such as Hg 2+ and Pb 2+ , is explained by this model. A chemical-type interaction (chemisorption) is favored, likely forming bidentate complexes ( Table 2).
The sorption capacities (q e ) of Imo obtained from the PSO model (Eq. 5) for Pb 2+ and Hg 2+ were 47.9 and 40.4 mg·g −1 , respectively, in single-component systems. However, due to the competition between cations, the sorption capacity of Imo was reduced by 10% in multicomponent systems. In the case of pure nZVI, the removal of these metals was higher than pure Imo in all cases. This is because this nanoparticle has multiple mechanisms of removal (adsorption, precipitation, coprecipitation, and oxidation-reduction), unlike Imo, for which adsorption is the only mode of removal. The functionalization of Imo with nZVI improved the removal capacity of both Hg 2+ and Pb 2+ compared to the starting materials, where the Fig. 2 ZP vs. pH plot of synthetic Imo, nZVI, and Imo-nZVI q e values in the case of Pb 2+ in single-and multicomponent systems were 81.9 and 77.4 mg·g −1 , respectively. For Hg 2+ , the values for single-and multicomponent systems were 81.0 and 77.7 mg·g −1 , respectively. The removal rates of both metals obtained from the PSO model (Eq. 2) show that the functionalization process provides a factor of 2.6 increase in rate when compared to Imo and nZVI separately. The initial adsorption rate (h), calculated from the PSO model, shows that Pb 2+ is removed by Imo-nZVI in earlier sorption stages than Hg 2+ .
The sorption mechanisms of Hg and Pb analytes in Imo, nZVI, and Imo-nZVI were analyzed using the Elovich equation and Weber-Morris models (Fig. 5).

Elovich equation
The Elovich equation (Eq. 3) is a widely used model in various systems whose reactive sites are energetically heterogeneous. The initial rate (α) obtained from this model was lower in the multi-component systems, indicating that   Fits of the experimental data to the Elovich equation for a Imo, b nZVI, and c Imo-nZVI and Weber-Morris model for d Imo, e nZVI, and f Imo-nZVI the cation competition delays the initial adsorption process. It was observed that α decreased, for both metals, in the order of Imo-nZVI> nZVI> Imo (Fig. 5), which has been reported in similar materials such as zeolite and montmorillonite (Arancibia-Miranda et al. 2016).

Intraparticle diffusion kinetic model
As previously discussed, the immobilization of nZVI in Imo generated changes in textural properties, which could generate changes in the overall removal rate of Hg 2+ and Pb 2+ . The data was fitted to the Weber-Morris model (Eq. 8) to determine whether or not intraparticle diffusion is the limiting step of the removal process (Table 3). Plotting q t against t 1/2 provides insights into the governing sorption processes of the analytes. These processes can be divided into three stages: (1) film or surface diffusion where the sorbate is transported from the bulk solution to the external surface of sorbent; (2) intraparticle or pore diffusion, where sorbate molecules move into the interior of sorbent particles; and (3) adsorption on the interior sites of the sorbent. Multilinearity was observed for all the systems under study. Surface sorption was the principal removal mechanism, with percentages greater than 90% (q e -1 values) of the analytes in all the systems evaluated (Fig. 5). This is a consequence of new adsorption sites (Fe-OH and Fe-O-OH) that are provided by Imo-nZVI. Parameter C of the model (Eq. 4) is related to the thickness of the limiting layer associated with instantaneous adsorption. In multicomponent systems C was reduced for Hg 2+ ions while remaining unchanged, within standard error, for Pb 2+ , which implies that accessibility to the active sites where sorption occurs is dependent on the composition of the solution. This result could be associated with the chemical characteristics of this element (i.e., ionic radius, polarizability, and electronegativity).

Isotherm sorption
The isotherms were studied for all elements in the singleand multicomponent systems (Fig. 6, Table 4) at pH 3.0, for Hg 2+ and Pb 2+ concentrations between 1 and 50 mg·L −1 . In general, the curves of the isotherms showed an "L"-type shape, which indicates good affinity between the substrate and the analytes, as well as a greater removal capacity in the single-component systems. In nZVI and the nanocomposite, the removal of both Hg 2+ and Pb 2+ was greater than in Imo. As discussed, the interactions are favored in nZVI and Imo-nZVI due to the decrease in the IEP and ZP. The experimental data was fitted using the Langmuir model (Eq. 9), where q max corresponds to the maximum removal capacity (mg·g −1 ) and K L is the Langmuir constant associated with the removal capacity (L·mg −1 ). All the materials in the single-and multicomponent systems showed higher values of q max and K L for Pb 2+ than for Hg 2+ . The values of q max for Pb 2+ and Hg 2+ in the singleand multicomponent systems followed the same descending order: Imo-nZVI>nZVI> > Imo. Imo-nZVI showed factors of 1.2 and 1.4 increase for the adsorption intensity (K L ) in comparison to pure nZVI and Imo, respectively. This is most likely associated with the decrease in the size of nZVI that is immobilized on the surface of Imo (Table 4).
According to Misono softness parameters (Y Hg2+ = 4.24 and Y Pb2+ = 3.58) (Misono et al. 1967), the functional groups of nZVI favored a stronger interaction with Hg 2+ than Pb 2+ , due to the fact that Fe has borderline character on the Misono softness scale. This observation is in agreement with the K L value, a measure of the metal ion affinity to adsorption sites on materials, which is enhanced in multicomponent systems (Table 4). The free energy (ΔG°) was determined for each material at a temperature of 25 °C and pH = 3.0 through the following equation: ΔG° = −RTlnK, where R is the gas constant (8.134 J mol −1 ·K −1 ) and T the temperature (K) using the Langmuir isotherms (Table S1). However, Zhou & Zhou (2014) reported that K L cannot be used directly to obtain this parameter because it is expressed as L·mg −1 . The standard equilibrium constant (K°) using the Langmuir equation was calculated to avoid this problem using the eq. K° = K L ·10 3 ·55.5 and replaced in the Gibbs free energy equation. Negative values of ΔG° were obtained for the removal of Pb 2+ and Hg 2+ for Imo, nZVI, and the nanocomposite in both single-and multicomponent systems (Table S1). Hence, the process of removing Pb 2+ and Hg 2+ is spontaneous and thermodynamically favorable.

Role of imogolite as a support for nZVI and its effects on the removal of Pb 2+ and Hg 2+
The structural and surface characteristics of Imo describe how this nanotubular aluminosilicate significantly favors the removal of neurotoxic metals, such as Pb 2+ and Hg 2+ , in single-and multicomponent systems when it is used as a support for nZVI. Several investigations have shown that in situ synthesis of nZVI in different substrates, such as activated carbon, zeolite, kaolinite, and montmorillonite, causes a decrease in size and increases the removal capacity of nZVI;  however, the impact of this process in Imo has important nuances to take into account: Due to the variable charge of Imo, which is positive with ZP values higher than 25 mV between pH 3 and 7, the formation of nZVI occurs mainly at the edges of the nanotube, indicating that the surface sites have higher sensitivity to changes in pH. These sites are neutral or negatively charged according to the synthesis conditions of nZVI, favoring the adsorption of Fe 3+ , thus generating nuclei of attraction for other iron ions.
The high density and magnetic and hydrophilic properties of nZVI prevent it from being suspended, leading to rapid agglomeration and oxidation in aqueous systems, reducing the removal capacity of different analytes (Peng et al. 2018). In contrast, the behavior of Imo is opposite to that described for nZVI since it remains in a stable suspension, even under conditions of unfavorable pH and ionic strength. The porosity and low density of Imo facilitate the suspension of the Imo-nZVI nanocomposite in conditions where nZVI cannot be suspended.
Immobilization of nZVI on Imo increases the reactivity of the material adding new types of surface-active sites with functional groups such as ≡Fe 2+ -OH, ≡Fe 2+ -OOH, ≡Fe 3+ -OH, and ≡Fe 3+ -OOH, generated through the oxidation of the nZVI. These functional groups may preferentially react with Hg 2+ or Pb 2+ according to the softness of the analytes and the polarizability of the Fe 2+ /Fe 3+ group. Thus, hard Pb 2+ would interact preferentially with more polarized ≡Fe 3+ -OOH and ≡Fe 3+ -OH groups whereas the softer Hg 2+ would interact with ≡Fe 2+ -OH and ≡Fe 2+ -OOH groups, contributing to a greater selectivity and specificity (Fig. 7).

Conclusions
The hybrid nanomaterial (Imo-nZVI) synthesized in this work represents a very interesting type of material through its exceptional properties and potential use in a wide range of fields. The synergistic combination of both components was efficiently tested as a sorbent material for neurotoxic metals removal. Based on the speeds and removal capacities obtained, Pb 2+ was preferably removed by all materials, even in systems that coexisted with Hg 2+ . The process of immobilizing nZVI in Imo caused an increase of the speed and removal capacity of both metals compared to the starting materials as the nZVI present in the nanocomposite is smaller in size than in free nZVI. The PSO and Langmuir models demonstrated a better agreement to the experimental data for all the cases under study, which suggests that the removal process occurs via chemisorption. The free energy associated with the removal of Pb 2+ and Hg 2+ indicates that this process is spontaneous in all materials and is more favorable in Imo-nZVI. The excellent percentage of recovery for Pb 2+ and Hg 2+ from water matrices and the magnetic properties of Imo-nZVI make it a promising material for environmental applications. Author contribution All authors contributed to the study conception and design. Material preparation, data collection, and analysis were performed by Estefanía M. Martinis and Nicolás Arancibia-Miranda. Material characterization and sorption kinetic data analysis were performed by Juliano C. Denardin, Raul Calderón, Cristóbal Flores, Karen Manquián-Cerda, and Tamara Maldonado. The first draft of the manuscript was written by Estefania M. Martinis and Nicolás Arancibia-Miranda, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.
Funding This work was supported by Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Agencia Nacional de Promoción Científica y Tecnológica (FONCYT) (PICT-BID), Universidad Nacional de Cuyo (Argentina), Financiamiento Basal para Fig. 7 Schematic representation of the possible interactions that occur on the surface of the Imo-nZVI. The affinity and intensity of removal of Pb 2+ and Hg 2+ with the different groups of the nanocomposite is conditioned by the magnitude of the Misono softness parameters (YHg 2+ = 4.24 and YPb 2+ = 3.58) and the possibility they have of interacting with functional groups with similar characteristics