Synthesis of Cyclen- Functionalized Ethenylene- Based Periodic Mesoporous Organosilica Nanoparticles and Metal- Ion Adsorption Studies

The preparation of two cyclens both possessing two triethoxysilyl groups through click chemistry is described. These two cyclens were incorporated into bis(triethoxysilyl) ethenylene-based periodic mesoporous organosilica nanoparticles (PMO NPs) at different proportions of bis(triethoxysilyl)ethenylene/cyclens (90/10, 75/25). The obtained nanorods were analyzed with different techniques and showed high specific surface areas at low proportion of cyclens. The nanorods containing free amino groups of cyclen were then used for Ni(II) and Co(II) removal from model solutions. The kinetics and isotherms of adsorption of Ni(II) and Co(II) were determined, and the materials showed high uptake of metals (up to 3.9 mmol ·g@1). They demonstrated pronounced selectivity in separation of rare earth elements from late transition metals, e. g. Ni(II) and Co(II) by adsorption and even more so by controlled desorption.


Introduction
The development of Mesoporous Organosilica Nanoparticles (MON) has grown alot the last decade for different applications, mainly drug delivery, but also catalysis. The field has been comprehensively reviewed [1] and recent examples have shown elaboration of complex core-shell structures allowing sequential release of hydrophobic and hydrophilic drugs. [2] Regarding the emerging environmental applications field, the extraction of heavy metal ions from water was studied. Hybrid organicinorganic mesoporous silica-based materials in bulk were first investigated. For instance, carbon dots-modified mesoporous organosilica, [3] amino and thiol functionalized phenylenebridged Periodic Mesoporous Organosilicas (PMO), [4] or SBA-15 based mesoporous silica hybrid adsorbents [5] were reported. The topic has been comprehensively reviewed as well. [6] Recently, the use of nanoparticles instead of bulk materials allowed to increase the outer specific surface area of the materials. Yang et al, prepared ferrocene-based MON for the extraction of phosphates and Pb(II). [7] Kaczmarek et al. synthesized amine-modified bis(triethoxysilyl)ethane-based PMO NPs which acted as sensors for Hg(II) and other metal ions after functionalization with Schiff bases. [8] As ophisticated bis(triethoxysilyl)ethane and Fe 3 O 4 -based Janus nanocube was synthesized by Su et al. for the removal of Au(III). [9] As aza macrocycles such as cyclens are very efficient metal ion chelators, [10] we thought of preparing cyclen-based PMO NPs for heavy metal water removal. We present here the synthesis of BOC-protected bis(triethoxysilyl)-functionalized cyclen precursors through click reaction and their incorporation in PMO NPs with bis(triethoxysilyl)ethenylene (E). After characterization of the materials and cleavage of BOC protecting groups, Ni(II) and Co(II) removal from model solutions and their separation from Rare Earth Elements (REE) was studied.

Preparation and characterization of silylated precursors and nanomaterials
Two bis-silylated cyclen derivatives P1 and P2 were envisaged as precursors for the preparation of periodic mesoporous organosilica nanoparticles (PMO NPs).
Alkyne 3 was obtained in two steps from commercial cyclen 1 through the tri-BOC intermediate 2 following procedures described in the literature (Scheme 1). [11] Ac opper-catalyzed alkyne azide cycloaddition reaction (CuAAC) [12] of the alkyne 3 with the corresponding bis(silylated azide), using the system Cu (I)/TBTA in am ixture of Et 3 N/THF at 50°Cu nder anhydrous conditions, provided the disilylated precursors P1 and P2 in 93% and 88% isolated yields, respectively. Note that Cu(II) was not complexed by cyclen as BOC groups were present. Compounds P1 and P2 were fully characterized by 1 Ha nd 13 C NMR spectroscopy ( Figure S1). The complete characterization of the molecules was achieved by 1D and 2D NMR experiments (1D 1 H, 2D 1 H-1 HC OSY, TOCSY and NOESY, 2D 1 H-13 CH SQC and HMBC) and the coordinated analysis of the resulting spectra (see details in the Supporting Information, Figure S2 and Table S1).
With those precursors P1 and P2 in hand, four PMO NPs were prepared. We mixed disilylated precursors E and P1 (leading to E-P1 PMO NPs) at ap roportion of E/P1 of 90/10 or 75/25.M ixing E and P2 lead to E-P2 PMO NPs,a taproportion E/P2 of 90/10 or 75/25 (Scheme 2).T he syntheses were performed in Milli-Q water with CTAB as template under basic catalysis (NaOH). The micellar solution was prepared at 80°Cfor 50 min with as tirring speed adjusted at 1000 rpm. Then, the stirring speed was enhanced to 1400 rpm. Am ixture of (E)a nd different amounts of the corresponding disilylated precursor Pn was quickly added and the mixture was left to react for 120 mins at 80°C. [13] E-P1 90/10 PMO NPs, E-P1 75/25 PMO NPs, E-P2 90/10 PMO NPs and E-P2 75/25 PMO NPs were collected by centrifugation, the template was removed by washing with a solution of NH 4 NO 3 (6 g/L in 96% EtOH).
BOC groups were then cleaved by treatment with TFA to lead to E-P1' 90/10 PMO NPs, E-P1' 75/25 PMO NPs, E-P2' 90/10 PMO NPs and E-P2' 75/25 PMO NPs.The cleavage of the BOC group was assessed by solid-state 13 CM AS NMR with the disappearance of the signals at 156 ppm (C=O), 80 ppm (C q ), 29 ppm (Me groups) (see the Supporting Information, Figure S3 and Figure S4). FTIR also confirmed the cleavage of BOC with the disappearance of the band at 1670 cm À1 (γ C=O) (Figure S5).
The materials were analyzed with different techniques. Transmission Electron Microscopy (TEM) (Figure 1a nd Figure S6) showed that nanorods were obtained (500-700 nm length, 100 nm width) with aw ormlike porosity. Small angle XRD ( Figure S7) confirmed the low organisation of the porosity.

FullPaper
The textural characteristics of E-Pn' PMO NPs were investigated by Atomic Force Microscopy (AFM) (Figure 2). The samples were deposited as dry powder onto ad oubly gluing carbon tape. Aggregated periodic mesoporous organosilica nanoparticles were clearly distinguishable on AFM images. The AFM data of E-Pn' PMO NPs was in agreement with TEM analysis of E-Pn PMO NPs.T he surface of the particles is apparently rough with shapes, indicating regular pore openings.
Data for N 2 sorption experiments, dynamic light scattering (DLS), zeta potential, thermogravimetric analyses (TGA) are reported in Table 1.

FullPaper
DLS data agreed with TEM images showing hydrodynamic diameters ranging from 463 nm to 639 nm ( Figure S8). Positive zeta potentials were observed in agreement with the protonation of the amine groups constituting the materials (Figure S9). TGA analyses showed the loss of water from 0to100°C then decomposition of BOC group occurred from 150 to 250°C, and finally the loss of the remaining organic part was observed from 300 to 800°C( Figure S10).
Brunauer-Emmett-Teller (BET) technique was used to monitor the porosity and specific surface area of the PMO NPs (Table 1). Materials E-P1 90/10 PMO NPs and E-P1' 90/10 PMO NPs ( Figure 3) showed type IV isotherms with an arrow hysteresis at 0.3 P/P 0 characteristic of mesoporous materials with small pore sizes of 2.3 nm (from BJH method). The specific surface area increased by removing the BOC group from 513 to 705 m 2 ·g À1 in agreement with ab etter accessibility of the porosity. E-P2 90/10 PMO NPs and E-P2' 90/10 PMO NPs materials showed between type II and type IV isotherms characteristic of an important microporous contribution to the porosity. The small pore size was 2nm. The same trend was observed as before with an increase of the specific surface area upon removal of the BOC group.
The content of cyclen moiety in the nanomaterials E-Pn' was estimated from the content of nitrogen from EDS analysis (Table 2and Table S2). For example, the cyclen loading of E-P1' 90/10 PMO NPs was 0.56 mmol ·g À1 .I nF igure 4w eg ather the SEM image and the corresponding element mapping of E-P1' 90/10 PMO NPs.E lement distribution is rather even following the apparent morphology of the particles. The peaks corresponding to nitrogen clearly indicate the presence of cyclen moiety in the nanoparticles (see the element mapping for E-P1' 75/25, E-P2' 90/10 and E-P2' 75/25 PMO NPs in Figures S11-S13 in the supporting information).

Adsorption studies of Ni(II) and Co(II)
The adsorption of Ni(II) and Co(II) from water by E-Pn' PMO NPs was assayed. At first, the adsorption kinetics for E-P1' 90/10 PMO NPs was determined (Figure 5and Table S3).
Adsorption kinetics showed that most of the uptake (80%) occurs during the first hour of interaction of metals with E-P1' 90/10 PMO NPs.S lower adsorption continues and equilibrium is reached after 20 hours.

FullPaper
All the uptake isotherms fitted well with Langmuir curves, having high Rs quare values (between 0.98-0.99). The saturation occurred at 5mMo re ven below for E-P1' 75/25 PMO NPs.T able 3s hows the average maximum adsorption capacity for each nanoadsorbent and heavy metal cation (Ni 2 + and Co 2 + ).
E-P1' 90/10 PMO NPs showed the highest metal uptake (Figure S14) in agreement with its highest specific surface area and pore size. It appears that the adsorption capacity results from combined contribution from both the surface area and the ligand content. Indeed if only specific surface area (S BET )o r Cyclen Loading is considered, metal adsorption capacity do not show any clear correlation ( Figure S15).
When set as ac ombined parameter, the content of Cyclen ligand and the surface area reveal linear correlation with the quantity of adsorbed cations (Table S4, Figure S16 and Figure 7).
When the ratio of E and Pn is 90/10, the observed ratio of Cyclen Ligand/Metal is approximately 1:7, while when the ratio of E and Pn is 75/25, the observed Cyclen Ligand/Metal ratio is 1:2-3. This indicates that the uptake is strongly driven by the interactions of metal cations with the external and internal surface of hybrid particles, possibly even in the form of surface silicates. Formation of the latter may be favored by local basic conditions in the pores induced by the presence of Cyclen ligands. It seems that chelation within aC yclen ring is also contributing to the mechanism of uptake. This indicates a considerable difference for the uptake in mesoporous nanoparticles compared to that of dense nanoparticles, where coordination of metal cations most often can be related to complexation in the monolayer on the surface. [14] The mapping of Ni(II) on E-P1' 75/25 PMO NPs and its quantification (Table S5 and Table S7) was performed with EDS (Figures S17-S18) after adsorption of an aqueous solution of Ni(II) at 0.5 and 10 mM initial metal concentration, respectively. We observed a homogeneous distribution of Ni(II) in the nanorods with an adsorption of 0.46 mmol ·g À1 and 0.94 mmol ·g À1 of Ni(II), respectively.
Desorption tests for E-P2' 90/10 PMO NPs and E-P1' 75/25 PMO NPs were performed with 1M nitric acid (Table 4). The desorption rates were analyzed by titration of the metal with EDTA for Co(II) and Ni(II). EDS analysis with E-P1' 75/25 PMO NPs was also performed for the remaining Ni(II) after desorption (Table S6 and Table S8).
Selective adsorption and desorption of organosilica nanoparticles (E-P2' 75/25 PMO NPs)w ere tested with two different metal mixtures, namely cobalt-samarium and nickel-neodymium which are relevant for recycling of components for the REEbased magnets. For this procedure, after the first adsorption/ desorption cycle, 10 mg nanoparticles were mixed with equimolar metal mixtures of NdÀNi and CoÀSm (5 mM initial  FullPaper concentration each) for 24 hours. Subsequently, nanoparticles were separated by centrifugation (7000 rpm), dried under nitrogen atmosphere and the metal content on the surface was analyzed by EDS. Results showed higher selectivity towards Ni ions, with aN i/Nd ratio of approximately 5.5 :1 ( Figure S19 and Table S9). Lower selectivity was observed for cobalt ions and the ratio of Co/Sm was 1.75 :1 ( Figure S20 and Table S10). Selective desorption was carried out with nitric acid at pH = 3.5 in 50 mL falcon tubes. 10 mg nanoparticles were mixed with 10 mL acid solutions and shaken for 24 hours. Subsequently, the nanoparticles were separated by centrifugation (7000 rpm) and dried under nitrogen atmosphere. EDS analysis showed good selectivity towards Co and Ni with high Ni/Nd = 7:1a nd Co/Sm = 8.5 :1 ratios (Tables S11 and Table S12). Therefore desorption was selective for REE. This indicates that the ligand contribution was important and these mesoporous nanoadsorbents could be used in separation of REE from heavy transition metals.

Conclusion
In summary, we have synthesized two disilylated cyclen based derivatives P1 and P2 containing Boc groups through acoppercatalyzed azide-alkyne cycloaddition (CuAAC). Mixed periodic mesoporous organosilica nanoparticles (PMO NPs) were prepared from these precursors Pn and bis(triethoxysilyl)ethenylene, E,a td ifferent proportions of E/Pn (90/10, 75/25). The Boc groups of E-Pn PMO NPs were removed by treatment with trifluoroacetic acid to afford E-Pn' PMO NPs containing cyclen moieties with coordinating amino groups. The obtained nanorods were analyzed with different techniques (TEM, AFM, N 2sorption measurements, DLS, zeta potential, TGA, EDS), and showed high specific surface areas at low proportion of Pn.The E-Pn' PMO NPs nanorods were then used for efficient removal of Ni(II) and Co(II) from aqueous solutions. The kinetics and isotherms of adsorption of Ni(II) and Co(II) were determined, and the nanomaterials showed high uptake of metals of up to 3.9 and 3.8 mmol ·g À1 for nickel and cobalt, respectively. The uptake mechanism is complex, indicating the combined role of the active surface area of mesoporous silica and its pore volume along with the ligand content. The obtained hybrid nanoadsorbents revealed considerable selectivity in separation of REE from late transition metals, especially on controlled desorption. This may be due to the generally accepted difference in adsorption mechanisms based on inner sphere coordination complexation for Ni(II) and Co(II), and outer sphere mostly electrostatic binding for REE. [15] Experimental Section General information. The 13 CC P-MAS solid state NMR spectra, the surface areas, dynamic light scattering (DLS) and Zeta potential were obtained from University of Montpellier. 13 CC P-MAS NMR spectra were recorded with aV arian VNMRS 300 MHz instrument (Les Ullis, France). The surface areas were determined by the

FullPaper
Brunauer-Emmett-Teller (BET) method from N 2 adsorption-desorption isotherms. The samples were measured with aM icromeritics ASAP2020 analyzer (Mérignac, France) after degassing for 12 ha t 80°Cu nder vacuum. The total pore volumes were evaluated by converting the volume adsorbed at p/p o 0.8 to the volume of liquid adsorbed (single point adsorption total pore volume of pores less than 4000 Åatp/p°� 0.98). The pore size distributions for PMO NPs were determined from the desorption branch using the Barrett-Joyner-Halenda (BJH) method, which relies on the Kelvin equation to relate the width of the pores to the condensation pressure. DLS analyses were performed using aC ordouan Technologies DL 135 (Pessac, France) particle size analyzer instrument. Zeta potential measurements were performed on MALVERN Instruments ZETA SIZER Nano series (Orsay, France) with 1-3 mg samples dispersed in 5mLo fd istilled water. The FastScan Bc antilevers with extremely sharp tip size of 5nm were used to achieve high resolution. The estimation of the amounts of the elements was carried out by aS EM-EDS technique using aF lex-SEM 1000 II scanning electron microscope. These two equipments belong to the Swedish University of Agricultural Sciences. If necessary, experiments were carried out in Schlenk techniques with standard high vacuum. CTAB, NH 4 NO 3 ,s odium hydroxide, potassium bromide and some reagents were purchased from Sigma-Aldrich. Ethanol and acetone was purchased from Fisher Chemicals and hydrochloric acid from VWR PROLABO.

FullPaper
Uptake of Ni 2 + and Co 2 + by E-P1' 90/10 PMO NPs: 20 mg of sample and 20 mL of metal solution (10.5 mM and 9.5 mM initial metal concentration Ni and Co respectively) were introduced in a5 0mL plastic tube and put on ashaker for 70 hours. For each heavy metal triplicate experiments were performed and standard errors were measured. After set interval of times (0.5 h, 1h,1 .5 h, 18 h, 22 h, 46 h, 70 h), E-P1' 90/10 PMO NPs were centrifuged (7000 g) for 10 min and 1mLa liquot was separated to determine the metal concentration in the remaining solution. The sample was first diluted 5t imes, and titrated afterwards with EDTA and xylenol orange as an indicator. For each sample the titrations were repeated 3t imes, and the average was calculated. The uptake was calculated according to this equation: Up = (CoÀCe)*V/m. Where: Co is the initial metal concentration which was also measured by titration, Ce is the equilibrium concentration of the metals after set interval of times, Visthe volume of initial metal solution which was kept constant, misthe NPs weight.

Desorption tests for E-Pn' PMO NPs:
The samples E-P1' 75/25 and E-P2' 90/10 PMO NPs were performed with 1M nitric acid in 50 mL tubes. After adsorption experiments the samples were centrifuged and 20 mL nitric acid was then added. The tubes were placed on a shaker for 24 hours. Then nanoparticles were centrifuged and the solution was neutralized with ammonium hydroxide and titrated with EDTA to calculate the desorbed metal amount. The nanoparticles were dried and the remaining metal content was measured with EDS (the metal uptake was measured with EDS also before desorption test).

Selectivity experiment with mixed metal solution:
For this procedure around 10 mg of recycled nanoparticles (E-P2' 75/25 PMO NPs)w ere mixed with equimolar mixture of NdÀNi and CoÀSm (5 mM initial concentration each) for 24 hours. Afterwards, nanoparticles were separated and dried and the metal content was analyzed by EDS spectroscopy.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.