Schi ﬀ base-functionalized mesoporous silicas (MCM-41, HMS) as Pb( II ) adsorbents

Three Schi ﬀ base-functionalized mesoporous silicas: MCM-41@salen, HMS-C12@salen and HMS-C16@salen have been synthesized by a post-synthetic grafting strategy and their sorption capacities toward Pb( II ) from synthetic aqueous solutions have been assessed. FTIR spectra, TG/DSC analyses, X-ray powder di ﬀ raction, X-ray photoelectron spectroscopy (XPS), N 2 adsorption isotherms and HRTEM micrographs were used to characterize the novel functionalized mesoporous silicas. The novel adsorbents have been tested for their capability in the remediation of synthetic aqueous systems containing Pb( II ). The Langmuir maximum values of sorption capacities of these adsorbents toward Pb( II ) are: 138.88 mg Pb( II ) per g MCM-41@salen, 144.92 mg Pb( II ) per g HMS-C12@salen, and 181.81 mg Pb( II ) per g HMS-C16@salen, therefore, the functionalized silicas (MCM-41@salen, HMS-C12@salen, HMS-C16@salen) could be used as e ﬀ ective adsorbents of Pb( II ) ions from wastewater.


Introduction
Modications by post-graing strategy of the silica surface of MCM-41/HMS materials have been widely studied and numerous organofunctionalized materials for selective and advanced separation processes have been obtained. [1][2][3][4][5][6] Conventional routes of silica surface modication by chemical treatment methods are based on the reaction between the surface silanol groups and commercial silane coupling agents as precursors for immobilization of organic molecules. One of the most used silylation agents is 3-aminopropyl trimethoxysilane (APTES) 7,8 and the amount of organic moieties obtained by graing of APTES is usually 1-1.5 mmol g solid À1 . 4,9 Therefore, the post-graing strategy represents a simple route for anchoring the functional chelating ligands into the pores and, consequently, affords the obtaining of a plethora of organo-functionalized mesoporous silicas that could be used as adsorbents of toxic materials.
Among these materials, those having graed salicylaldoxime (salen) ligands represent efficient sorbents and they were extensively used for remediation of wastewater containing heavy metal ions. Soliman et al. 10 obtained mono-and bissalicylaldehyde Schiff base ligands graed on silica surface in a three-step approach: silanization of silica surface with 3chloropropyltrimethoxysilane, replacing the chlorine leaving group with amino-group from diethylenetriamine, followed by the condensation with salicylaldehyde. The as-prepared hybrid materials show high selectivity in extraction of Cu(II) (metal uptake capacity 0.957 mmol g À1 ) and Fe(III) (0.643 mmol g À1 ). In a similar approach, Kim et al. 11 synthesized a chiral salen-mesoporous MCM-41 material that has been further involved in Mn(III) coordination, thus providing a catalyst for styrene epoxidation with high level of enantioselectivity. Sarkar et al. 12 demonstrated that preconcentration of Cu(II), Zn(II), Co(II), Fe(III) and Ni(II) in water could be highly effective by using the silica gel modied with salicylaldoxime. Parvulescu et al. 13,14 reported the preparation of n-propylsalicylaldimine modied SBA-15/SBA-16 mesoporous silicas and the study of their efficiency in sorption of heavy metal ions, they found out that adsorption capacities decrease within the following series: Cu(II) > Co(II) \ Mn(II). Functionalized SBA-15 mesoporous silica particles, bearing ethylenediaminepropyl-salicylaldimine and N-propylsalicylaldimine Schiff base ligands, were examined as adsorbents in solid phase extraction (SPE) of uranyl (UO 2 2+ ) ions from water. 15 The maximum adsorption capacities determined for these adsorbents were 105.3 and 54 mg uranyl per g adsorbent, respectively. Chen et al. 16 prepared a Schiff base immobilized hybrid mesoporous silica membrane for the detection of Cu(II) by immobilizing 4-chloro-2-[(propylimino)methyl]-phenol (4-chloro-salen) onto the pore surface of mesoporous silica embedded in the pores of an anodic alumina membrane. The detection limit for Cu(II) was 0.8 mM. Gao et al. 17 previously used the same ligand to obtain (via a self-assembly process) a mesoporous silica material SBA-15-APTES-4-chloro-2-[(propylimino)methyl]-phenol, as Zn(II)-sensitive uorescent chemosensor. Porous silica functionalized with N-propylsalicylaldimine was used for separation and preconcentration of Cu(II), Cr(III, VI), Cd(II), Pb(II) and Mn(II, VII) from natural waters. 18 In this work, we describe the obtaining of ordered mesoporous silicas (MCM-41 and HMS type) with hemisalen Schiff base graed onto the silica surface (Fig. 1).
FTIR spectra, TG/DSC analyses, X-ray powder diffraction, X-ray photoelectron spectroscopy (XPS), N 2 adsorption isotherms and HRTEM micrographs were used to characterize the novel functionalized materials.
The obtained mesoporous adsorbents have been assessed for the effectiveness in the remediation of synthetic aqueous systems containing Pb(II). The factors that inuence the sorption behavior have been studied: heavy metal ion concentrations, pH of solution, contact time, as well as the adsorption isotherms.

Techniques and materials
All the raw materials were commercially available and used as received.
FTIR vibrational spectra were registered with a Bruker Tensor 27 spectrophotometer equipped with one ATR sampling unit, in the wavenumbers range of 500-4000 cm À1 .
Thermal analysis was carried out with a Netzsch 449C STA Jupiter. Samples were placed in open alumina crucible and heated with 10 degrees per min from room temperature to 900 C, under the ow of 20 mL min À1 dried air. An empty Al 2 O 3 crucible was used as reference.
Powder X-ray diffraction (PXRD) patterns were registered on a Panalytical X'PERT PRO MPD diffractometer with graphite monochromatized CuK a radiation (l ¼ 1.54Å). The samples were scanned in the Bragg angle, 2q range of 2-10 and step size of 0.013 .
For TEM and HRTEM analysis we used TECNAI F30 G 2 highresolution transmission electron microscope operated at an accelerating voltage of 300 kV device. The samples have been prepared by dispersing the particles by ultrasonication in methanol and subsequently collected into a holey carboncoated TEM support grid.
The N 2 adsorption/desorption isotherms for pore size distribution and BET specic surface measurements were recorded on a Micromeritics ASAP 2020 analyzer. Before analysis the samples were outgassed at 120 C for at least 6 hours under vacuum.
X-ray photoelectron spectroscopy (XPS) data were registered on a Thermo Scientic K-Alpha device, fully integrated, with an aluminium anode monochromatic source (1486.6 eV). The survey spectra were registered using a pass energy of 200 eV at bass pressure of 2 Â 10 À9 mbar.
The pH measurements were taken with Agilent 3200P pH Meter. Batch equilibrium experiments on the GFL 3031 Incubating Shaker at 150 rpm were performed. Atomic Absorption Spectrometry on contrAA® 300 Analytik Jena Atomic Absorption Spectrometer was applied to determine Pb(II) ions concentration in aqueous solutions. All the equilibrium experiments were carried out by contacting 0.025 g of adsorbent with 25 mL of synthetic aqueous solution of 100 mg L À1 Pb(II), at room temperature, for 24 hours and 150 rpm. Aer reaching the equilibrium, the Pb(II) concentrations in ltrate samples have been determined by AAS.

Preparation of functionalized mesoporous silicas
2.2.1. MCM-41. MCM-41 mesoporous silica has been prepared by using cetyltrimethylammonium bromide (CTAB) as structure-directing agent and tetraethoxysilane (TEOS) as sol precursor to form silica mesopores, according to an adapted literature procedure. 19 The reaction mixture was composed of 1.0CTAB : 9.21TEOS : 2.55NaOH : 4857H 2 O (molar ratios). CTAB and NaOH formed the initial reaction mixture in water that was further heated at 80 C for 30 min. TEOS was added at pH 12. Aer 2 h at 80 C, white solid MCM-41 precipitated. The product was ltered off, washed with large volumes of water and methanol and dried in oven at 105 C for 6 h. The surfactant (CTAB) has been removed by thermal treatment at 550 C for 6 h.
The syntheses were accomplished at room temperature, the reaction mixtures were aged for 24 h and the obtained HMS solids were recovered by ltration, washed with deionized water and dried at 105 C for 6 h. The primary amines (C 12 H 25 NH 2 , dodecylamine and C 16 H 33 NH 2 , hexadecylamine) have been removed by calcination in air at 630 C for 4 h. 2.2.3. APTES modication of mesoporous silicas: MCM-41@APTES, HMS-C12@APTES, HMS-C16@APTES. Following a general procedure, the mesoporous silicas have been derivatized with 3-aminopropyl triethoxysilane, APTES, in toluene, as follows: 3 g of mesoporous silica (MCM-41, HMS-C12 or HMS-C16, dried at 160 C for 5 h, under vacuum) and APTES (2.5 mL) in toluene (100 mL) were reuxed for 24 h, under inert atmosphere.

FTIR spectra
In the FTIR spectra (Fig. 2, Table 1) of functionalized silica materials, stretching vibrational bands specic to silica network and to organic moieties (i.e. the surfactant or the covalently bonded aminopropylsiloxane linker/salen ligands) are present. The strongest signals in FTIR spectra are assigned to Si-O-Si 1100 cm À1 corresponding with the silica network formed as a result of hydrolysis/condensation reactions between the silica precursors.

XPS data
XPS scans (Fig. 3) carried out on MCM-41@salen, HMS-C12@salen, HMS-C16@salen revealed that they have as constitutive elements: C-C 1s, N-N 1s, O-O 1s, and Si-Si 2p with the corresponding binding energies presented in Table 2. The C-C 1s, N-N 1s are clearly resulted from the organic moieties graed onto silica surface.
The atomic contents (converted in weight percents, Table 2) suggest that about 50% of the APTES moieties (for organic content: APTES ¼ C 3 H 8 N and APTES + salicylaldehyde ¼ C 10 H 12 N)   have been reacted to form a bond between the amino group (APTES) and the carbonyl group (salicylaldehide). Supposing that all the nitrogen and carbon content comes from the APTES and APTES + salicylaldehyde, we found that the theoretical carbon-tonitrogen (C : N) mass ratio is 5.57 : 1 for 1 : 1 APTES : APTES + salicylaldehyde and this ratio agrees fairly well with the experimental ones (6.03 : 1 for MCM-41@salen, 5.77 : 1 for HMS-C12@salen and 5.72 : 1 for HMS-C16@salen, respectively).

Thermal analysis
TG-DSC analyses ( Fig. 4 and 5) also support the conclusions drawn from the FTIR and XPS spectra. All the as-prepared mesoporous silicas have similar thermal behavior with three main mass loss peaks, illustrated by TG curves: volatilization of the adsorbed solvent (below 100 C), decomposition of organic fragments graed in the silica pores   (100-650 C), and densication of the silica matrix (above 650 C up to 900 C). The DSC curves are very similar in shape for all three organofunctionalized silicas: they present one strong exothermic effect centered at 292 C for MCM-41@APTES, at 289 C for HMS-C12@APTES and at 273 C for HMS-C16@APTES, respectively. two strong exothermic effects: rst effect is centered at 321 C for MCM-41@salen, at 340 C for HMS-C12@salen and 284 C for HMS-C16@salen, respectively, and the second effect occured at 578 C (MCM-41@salen), at 580 C (HMS-C12@salen), and at 591 C (HMS-C16@salen), respectively. The former effect could be assigned to the decomposition of organic ligand graed onto silicas and the latter to the removal of the residual combustion compounds originated from the decomposition of organic part.
The TG-DSC data are consistent with the results of asprepared samples XPS measurements. The largest mass loss was found for HMS-C16@salen, similarly to the XPS results where the highest organic content was also found for HMS-C16@salen.
-HMS-C12 (Fig. 6b, Table 3) and HMS-C16 (Fig. 6c, Table 3) exhibit a single low angle reection (100) at 2q ¼ 2.03 (HMS-C12) and 2q ¼ 1.76 (HMS-C16) indicative of the average pore-pore correlation distance in their hexagonal lattices. 20 For these silicas, it is assumed that the framework mesopores are wormholelike with some spatial orientation of the pores within the particles, i.e. local hexagonal symmetry. As the alkylamines used as surfactants are of C12 (HMS-C12) and C16 (HMS-C16)  For HMS-C12@salen (Fig. 6b, Table 3) and HMS-C16@salen (Fig. 6c, Table 3), the post-synthetic graing of organic ligands caused a signicant decrease in the (100) peak intensity and the complete lack of the higher order Bragg reections in their diffraction patterns (Fig. 6b and c), both features indicate the incorporation of organic moieties into the pore channels of the HMS-C12 and HMS-C16 materials, without the collapse of the pore structure. In conclusion, the pore channels in all HMS silicas prepared by us show uniform framework mesopore distribution with short range hexagonal order. The lack of longer range order is explained by the relatively weak hydrogenbonding interactions that operate in the neutral (alkylamine:-TEOS) assembly process.
The values of the corresponding unit cell parameter a 0 were calculated using the formula: a 0 ¼ 2d 100 /3 1/2 , where d 100 represents the d-spacing value of the (100) diffraction peak in XRD patterns of the samples.
HRTEM images for HMS-C12@salen show uniform wormholelike channels (mesopores) distributed homogeneously throughout the bulk phase over a short range hexagonallike order (see arrow, Fig. 7B). HRTEM micrograph of HMS-C16@salen sample (Fig. 7C) revealed that the mesoporous framework is better-dened than that of HMS-C12@salen, the hexagonal order being comparable to the MCM-41@salen structure. It is known that for longer alkyl chains (number of carbon atoms > 12), the structure of the HMS template changes from rod-like micelles to lamellar micelles. 20,22,23
For HMS-C12 and HMS-C16, H2-type hysteresis loops are present and they are associated with slightly larger and less ordered mesoporous structures.
The pore volume and surface area of hemisalen functionalized silicas (MCM-41@salen, HMS-C12@salen, HMS-C16@salen) decrease dramatically in comparison with the parent MCM-41, HMS-C12, HMS-C16 silicas (Table 4). With the exception of HMS-C12@salen, which retains the framework mesoporosity and give a type IV isotherm (with capillary condensation step less visible), the other two silicas, MCM-41@salen and HMS-C16@salen, show type II isotherms and this fact could be attributed to pore blocking by salen ligands.

Pb(II) adsorption studies
Even at low concentration, Pb(II) damages tissues and organs such as kidneys, liver, reproductive and nervous system. Pb(II) has also indirect effects by inducing the production of reactive oxygen species (ROS) with negative consequences to cell macromolecules, proteins and polyunsaturated lipids. 26 The application of conventional removal methods such as chemical precipitation, 27 coagulation-occulation, 28 ion exchange, 29 membrane ltration, 30 reverse osmosis, 31 electrochemical processes, 32 is limited by their high costs and inefficiency, in the case of low levels of Pb(II). Adsorption is one of the most used conventional process to remediate wastewater due to its simplicity, ease and rapid application and manipulation, the potential use of many natural materials (low cost) and synthetic materials as adsorbents. [33][34][35][36][37][38][39] Mesoporous materials (MCM-41/HMS) have high surface area, narrow pore size distribution, tunable texture and surface properties, and a relative chemical inertness. These properties recommend their applications in sorption, catalysis, separation, sensors, hosts and numerous other elds. 40 The MCM-41@salen, HMS-C12@salen, and HMS-C16@salen materials obtained by us have been used to uptake Pb(II) from synthetic aqueous systems and their sorption properties have been compared to those of parent MCM-41, HMS-C12, HMS-C16 mesoporous silicas.
3.7.1. Effect of initial pH on sorption capacity. The remediation of aqueous systems containing Pb(II) by adsorption depends primarily on the pH solution: the adsorbent surface charge, the ionic state of lead and also the ionization of the adsorbent functional groups are strongly affected by the pH solution. 41 The inuence of pH of solution in Pb(II) retention by MCM-41, MCM-41@salen, HMS-C12, HMS-C12@salen, HMS-C16 and HMS-C16@salen has been studied.
All the equilibrium tests were performed over the pH range of 2-6, by shaking 0.025 g of adsorbent with 25 mL of synthetic aqueous system of 100 mg L À1 Pb(II), at room temperature, for 24 hours and 150 rpm. The solutions were ltered and lead concentrations in the ltrate were determined by AAS, aer reaching the equilibrium. Fig. 9 presents the inuence of pH of solution on sorption properties of our silica adsorbents, and their sorption capacity (Q) has been calculated as follows: Q ¼ net amount of adsorbed PbðIIÞ the amount of adsorbent ¼ where Q -Pb(II) uptake (mg g À1 ); C ithe concentration of contaminant in the initial solution (mg L À1 ); C fthe equilibrium concentration of contaminant (mg L À1 ); mthe amount of adsorbent (g); Vthe sample volume (L).  As can be observed (Fig. 9), the amount of Pb(II) retained by silicas increases with the increasing of the pH value. It reaches a maximum value at pH 5 and, then, it decreases slowly when pH solution changed from 5 to 6. At lower pH, a competition between Pb(II) ions and H + from solution for the adsorbents sites may occur. The experiments were conducted at a maximum value of 6 for pH solution. Beyond this pH the formation of lead hydroxo-species (Pb(OH) + , Pb(OH) 2 0 , and Pb(OH) 3 À ) occurs and the precipitation of the Pb(II)-species can compete with sorption.
3.7.2. Effect of contact time. Fig. 10 illustrates the inuence of the contact time on the Pb(II) adsorption from aqueous systems, for silica adsorbents.
The equilibrium was reached aer 300 minutes and the sorption process of Pb(II) onto silica adsorbents can be considered a fast process. The large number of the available active sites of adsorbents determines the adsorption to proceed rapidly in the early stages of the process. With a gradual decrease in the number of active sites, the adsorption process rate decreases.
The experimental values of sorption capacities are: 77.31 mg Pb(II) per g MCM-41, 93.28 mg Pb(II) per g MCM-41@salen, 94.07 mg Pb(II) per g HMS-C12, 99.27 mg Pb(II) per g HMS-C12@salen, 96.34 mg Pb(II) per g HMS-C16, and 99.58 mg Pb(II) per g HMS-C16@salen. As expected, the functionalization of MCM-41, HMS-C12 and HMS-C16 with salen-type ligands has as a result a higher sorption capacity of functionalized silicas. This result can be due to the fact that N and O atoms, from salen ligands graed on the silica surface of MCM-41@salen, HMS-C12@salen, and HMS-C16@salen adsorbents, are Lewis donors complementary to borderline Lewis acceptors Pb(II) ions and, therefore, the organofunctionalized silicas are more efficient in Pb(II) adsorption.
3.7.3. Adsorption isotherms. The adsorption isotherms represent mathematical expressions of the adsorbate-adsorbent interactions. These equations are useful to determine affinity of the sorbent, its surface properties, and also the sorption mechanism. 42 Langmuir, Freundlich, Temkin and Dubinin-Radushkevich isotherms are the most used to describe the adsorption equilibrium. The Langmuir isotherm represents an empirical equation that describes the process during the adsorption of contaminants/metal ions onto a xed number of sorption sites.
Eqn (2) represents the linear expression of the Langmuir equation: 42 where Q e is the amount of contaminant removed at equilibrium (mg g À1 ); C e is the equilibrium concentration of contaminant (mg L À1 ); K L is the Langmuir model parameter regarding energy of adsorption (L mg À1 ) and Q max states the maximum value of sorption capacity (mg g À1 ). 42 The Freundlich isotherm is frequently used to depict the sorption on heterogeneous surfaces, a process in which the concentration of adsorbate on the adsorbent surface increases with the adsorbate concentration and an innite amount of adsorption can occur. The Freundlich model can be expressed as: where K Fconstant for sorption capacity; n -Freundlich constant for sorption intensity.
It can be noticed that the correlation coefficients (R 2 ) are closer to 1 for the Langmuir isotherm. Consequently, the Langmuir isotherm model is more appropriate to describe the sorption process for all the studied adsorbents rather than Freundlich isotherm model.
According to the pseudo-rst-order model, the rate of adsorption on sorbent is proportional to the number of free active sites of sorbent. 48 Lagergren eqn (5) is the mathematical expression of pseudo-rst-order kinetic model: where k 1 represents the constant rate of pseudo-rst order sorption (min À1 ), and Q e , Q t symbolize the sorption capacities at equilibrium and at time t (mg g À1 ). The linear form of this equation is: where Q e and Q t represent the amounts of Pb(II) retained on adsorbent (mg g À1 ) and k 1 is the rst-order sorption rate constant (min À1 ). The slopes and intercepts of plots of log(Q e À Q t ) versus t have been used to calculate k 1 and the correlation coefficient R 2 (Fig. 13a-c).
The pseudo-second-order kinetic model is mathematically summarised as Ho eqn (7): where k 2 depicts the rate constant of second-order adsorption (g mg À1 min À1 ). 49 The line plots t/Q t against t were used for the parameters of the pseudo-second-order kinetic model (Fig. 14a-c). The intraparticle diffusion model is depicted by the mathematical equation:  where k i represents the intraparticle diffusion rate constant (mg g À1 min À0.5 ). The graphic representations of Q t versus t 0.5 (Fig. 15) have been used to evaluate the k i value.
Since the plots for all six silicas are not linear and they do not pass through the origin, the intraparticle diffusion can not be considered the rate-limiting step for the Pb(II) sorption onto MCM-41, MCM-41@salen, HMS-C12, HMS-C12@salen, HMS-C16 and HMS-C16@salen samples. The plots are multilinear with two or three distinct regions. Thus, two or three different kinetic mechanisms are involved. According to literature data, it can be supposed that the initial curved region correlates with the external surface uptake, the second stage represents a gradual uptake reecting intraparticle diffusion as the rate limiting step and the nal plateau region shows the sorption equilibrium. 50 Kinetic parameters show a good correlation with experimental kinetic values for pseudo-second order model (R 2 > 0.991 for all six adsorbents - Table 7). Consequently, the chemisorption is considered to be the rate-limiting stage. 49
Even though the framework mesoporosity partially decreased as a consequence of graing the salen ligands onto the silica surface, the coordinative sites still remained accessible for binding heavy metal ions.
Kinetic studies have been performed to determine the kinetic model that describes the Pb(II) sorption onto MCM-41, MCM-41@salen, HMS-C12, HMS-C12@salen, HMS-C16 and HMS-C16@salen. The lead adsorption was well tted by a pseudo second-order kinetic model (R 2 > 0.991 for all six adsorbents) and a Langmuir isotherm model has been more appropriate to describe the sorption process for all the studied adsorbents rather than Freundlich isotherm model.
The values of Langmuir sorption capacity suggest that the obtained Schiff base-functionalized mesoporous silicas are relevant materials for the removal and recovery of Pb(II) and also could be used as effective adsorbents of Pb(II) from wastewaters.

Conflicts of interest
There are no conicts to declare.