Tailoring bifunctional hybrid organic–inorganic nanoadsorbents by the choice of functional layer composition probed by adsorption of Cu2+ ions

Spherical silica particles with bifunctional (≡Si(CH2)3NH2/≡SiCH3, ≡Si(CH2)3NH2/≡Si(CH2)2(CF2)5CF3) surface layers were produced by a one-step approach using a modified Stöber method in three-component alkoxysilane systems, resulting in greatly increased contents of functional components. The content of functional groups and thermal stability of the surface layers were analyzed by diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy, and 13C and 29Si solid-state NMR spectroscopy revealing their composition and organization. The fine chemical structure of the surface in the produced hybrid adsorbent particles and the ligand distribution were further investigated by electron paramagnetic resonance (EPR) and electron spectroscopy of diffuse reflectance (ESDR) spectroscopy using Cu2+ ion coordination as a probe. The composition and structure of the emerging surface complexes were determined and used to provide an insight into the molecular structure of the surfaces. It was demonstrated that the introduction of short hydrophobic (methyl) groups improves the kinetic characteristics of the samples during the sorption of copper(II) ions and promotes fixation of aminopropyl groups on the surface of silica microspheres. The introduction of long hydrophobic (perfluoroctyl) groups changes the nature of the surface, where they are arranged in alternately hydrophobic/hydrophilic patches. This makes the aminopropyl groups huddled and less active in the sorption of metal cations. The size and aggregation/morphology of obtained particles was optimized controlling the synthesis conditions, such as concentrations of reactants, basicity of the medium, and the process temperature.


Synthesis of bifunctional nanoparticles with hydrophobic (perfluorooctyl-, methyl-, or n-propyl-) and amine-containing groups in the surface layer
Sample NM (TEOS/APTES/MTES = 3/0.5/0.5). Its synthesis was carried out similar to N4, but APTES and МТЕS volumes were 0.7 ml and 0.6 ml. After the appearance of opalescence in 40 min, ammonia was added. In 3 h after addition the precipitate was centrifuged and washed. The yield was 0.89 g.
Sample NMi (TEOS/APTES/MTES = 3/0.5/0.5). Its synthesis was carried out similar to NM, but in an ice bath at a temperature of 3-4°C. After the appearance of opalescence in 60 min, ammonia was added. In 3 h after addition the precipitate was centrifuged and washed. The yield was 0.52 g.
Sample NMh (TEOS/APTES/MTES = 3/0.5/0.5). Its synthesis was carried out similar to NM, but with heating to 50°С. After the appearance of opalescence in 15 min, ammonia was added. In 1 h after addition the precipitate was centrifuged and washed. The yield was 0.49 g.  (Table TS1). The surface of sample F2 is rough and each particle seems to be composed of smaller particles (10-20 nm in size) (Fig. FS3).
Meanwhile sample F1 with higher relative content of PFES has smooth surface and features particles close to spherical (Fig. FS3). Moreover, relative content of fluorine determined by EDXS analysis for both samples F1 and F2 correspond to the initially desired F/Si relations (Fig. FS4). As well as the content of perfluoroctyl groups, recalculated from elemental analysis on carbon for samples F1 (1.7 mmol/g) and F2 (1.3 mmol/g) also coincide with the theoretically assessed values based on the ratios of reacting alkoxysilanes (1.73 mmol/g for F1 and 1.32 mmol/g for F2). It should be S8 mentioned that specific surface of sample F1 is significantly less than the sample F2 (see Table TS1). Consequently, higher density of surface groups on the surface of sample F1 promotes the formation of smoother spherical particles, apparently via the hydrophobic interactions. Due to the above-mentioned different particle structures of samples F1 and F2, their size comparison would be incorrect.

DRIFT analysis of the surface layers
The assignment of absorption bands in the DRIFT spectra of samples was carried out using references [1][2][3]. In the DRIFT spectrum of the sample N2 ( Fig. FS6, spectrum 1) the absorption band at 1534 cm -1 , resulting from δ(NH 2 ) bending of the amino groups is clearly visible. In addition, the DRIFT spectrum also contains an intense absorption band with a high-frequency shoulder in the region of 1000-1200 cm -1 , which is characteristic of the ν as (SiOSi) stretching vibrations. This indicates the formation of a network of polysiloxane bonds.
The band of medium intensity at 1638 cm -1 refers to δ(H 2 O) bending. The presence of propyl chains (Si-CH 2 CH 2 CH 2 -N) in the DRIFT spectra is indicated by a group of adsorption bands of weak intensity in the region 1390-1440 cm -1 and two adsorption bands of medium intensity in the region 2800-3000 cm -1 . They are typical of CH 2 bending and of stretching vibrations of CH, respectively. Note the presence of the low-intensity absorption band at 1412 cm -1 (see Fig. FS6), which refers to δ(Si-CH 2 ) vibrations of 3-aminopropyl moiety. The DRIFT spectra of the other samples with amino groups are identical to the described above.
Interestingly, heating the N2 sample in vacuum to a temperature of 100°C leads to the disappearance in the DRIFT spectrum of the absorption bands at 1638 cm -1 (Fig. FS6, spectrum 2) and 1534 cm -1 , and the appearance of the absorption band at 1580 cm -1 . The shift of the absorption band from 1534 cm -1 to 1580 cm -1 indicates different surrounding of the amino groups at different temperatures. Thus, the absorption band at 1534 cm -1 is characteristic of amino groups connected with silanol groups via water molecules. During heating, these bonds are destroyed, and amino groups are connected to each S11 other via hydrogen bonds, as evidenced by the appearance of the bands at 1580 cm -1 . 19 In addition, the removal of water, makes possible to identify two low-intensity absorption bands in the 3280-3370 cm -1 range related to ν s,as (NH) stretching of amino groups involved in hydrogen bonds. Finally, it should be mentioned that the presence of silanol groups in the surface layer of the sample N2 is proved by the absorption band at ~3650 cm -1 (Fig. FS6).
The DRIFT spectra of the samples with amino/methyl groups (NMh as example, Fig. FS6, spectrum 3) have sharp absorption band at 1273 cm -1 , which is absent in the DRIFT spectra of other samples and can be attributed to δ s (CH 3 ) of methyl group bound to a silicon atom. At the 1415 cm -1 in the DRIFT spectra of these samples the band of low intensity is observed. This band relates to the asymmetric bending of methyl groups δ as (CH 3 ). S12 Figure S6: DRIFT spectra of samples: S13 The presence of perfluorooctyl groups in monofunctional fluorine-containing samples (sample F2 in Fig. FS6) was confirmed by a band of medium intensity with a frequency of ~1315 cm -1 corresponding to ν as (CF). 4 This absorption band is not observed in samples not containing a fluoroalkyl residue. However, the symmetric stretching band ν s (CF) somewhat overlaps with stretching vibrations of polysiloxane network (broad intense adsorption band of siloxane bonds, SiОSi, in the range 1000-1200 cm -1 ), so it is difficult to identify it clearly. The signal (shoulder) at ~900 сm -1 , overlapping with a broad medium intensity band of (Si-ОН) vibrations at 950 сm, -1 also indicates the presence of CF 3 groups [4,5].Indirectly, the presence of ≡Si(CH 2 ) 2 (CF 2 ) 5 CF 3 groups in samples is testified by absorption bands at ~1364, ~1413 сm -1 (weak), and ~1441 сm -1 in their IR spectra (Fig. FS6), which can be attributed to (CH 2 ), (Si-CH 2 ), and  as (CH 2 ) respectively. Absorption bands characteristic of symmetric and asymmetric stretching vibrations of C-H bonds are also present in the region 2900-2985 cm -1 , but they overlap with broad absorption band of (OH) of adsorbed water at ~ 3000-3400 cm -1 .
The DRIFT spectra of bifunctional samples with fluorine and amine containing groups in the surface layer revealed adsorption bands characteristic of both amino-and perfluorooctyl functional groups mentioned above, thus witnessing their incorporation in the samples (Fig. FS6). For example, the spectra of samples NF3 revealed an absorption band at 1547 cm -1 , which refers to the δ(NH 2 ) bending vibrations of amino groups. Upon heating the sample NF3 to 100°C the δ(H 2 O) bending band of water molecules at 1640 cm -1 vanishes from its IR spectrum, and the δ(NH 2 ) band of amino groups shifts to 1590 cm -1 .
The removal of water at 100°C allows identification of absorption bands in the S14 3260-3370 cm -1 infrared spectrum region, belonging to ν s,as (NH) stretching vibrations of amino groups involved in the hydrogen bonds. Compared with the amine samples (N2 and NMh), the IR spectra of fluorinated samples (F2 and   NF3, Fig. FS6) have easier identifiable absorption band at ~ 3650 cm -1 , which undoubtedly belongs to the silanol groups. Table TS1 presents specific surface areas for some samples calculated from low-temperature nitrogen adsorption isotherms. These data are consistent with the SEM data. Thus, S sp of all amino samples is 10-43 m 2 /g, which is due to rather large size of their particles (about 280-720 nm). In addition, an increase in the synthesis temperature causes an increase in the particles' diameter and, as a consequence, a decrease in the S sp . (samples N4 and N4h). Only sample N4i has diameter of submicroparticles 140 nm, but they agglomerate each other. Bifunctional samples (Table TS1) Table   TS1). This is an indirect confirmation that their SEM images are likely to present the secondary structures, but clearly it can be argued only for sample F2 (Fig. FS3). For this sample it is consistent with structural adsorption analysis. S15

TGA studies
Thermal analysis data indicate the presence of functional groups and water in the synthesized samples. Thus, Fig. FS7 presents thermograms for samples N4h, NMh, F2 and NF4. All thermograms are characterized by weight loss in the temperature range of 90-110°C, which can be associated with the removal of water and residual solvent. The results of the thermal analysis showed that the samples with monofunctional fluorine-containing layer are the most stable.
Their thermal destruction starts above 400°C (sample F2 in Fig. FS7). The processes of destruction of their organic layer are similar to the xerogels synthesized at the same TEOS:PFES ratios [6]. The DTG curves for bifunctional samples contain a peak at lower temperatures about 290°C (FN4 and NMh in Fig. FS7) associated with the removal of surface amino groups.
The decomposition of amino groups in pure amino sample (N4h) starts at slightly lower temperature (270°C, Fig. FS7) and is consistent with the data for xerogels containing 3-aminopropyl groups [7,8].
According to Table TS1, amino groups content in spherical silica particles is in the range 0.5-2.0 mmol/g (at the ratio of TEOS:APTES=3:1), which is about 2 times less than expected from the ratio of reactive alkoxysilanes. The data in this table suggest that several factors determine amino groups content: the components ratio (samples N1 and N2), the order of alkoxysilanes introduction in the reaction solution (samples N2 and N3), the synthesis temperature (samples N4i, N4, and N4h); little effect is produced by the amount of used ammonia (samples N2 and N4) [9]. The types of copper(II) adsorption isotherms for the samples also confirm the above-mentioned observation (Fig. 6). Whereas simple for majority of samples, the copper(II) adsorption isotherms of samples N4i and N2 have clear bends (Fig. 6, FS8). If for the first sample such bend is observed at a low C 0 Cu :C 0 R ratio and in a wide range (see Fig. 7), for the later sample it is abrupt at 1.5. It is worth noting that the synthesis of sample N4i was conducted at low temperature, while monofunctional sample N2 was obtained at room temperature and using different synthesis technique. Obviously, in both cases there is different composition of copper(II) complexes formed in the surface layer of the samples. S19 Figure S9: The EDSR spectra of the copper(II) complexes with monofunctional layers.

Adsorption of organic molecules on the monofunctional perfluoroalkyl functionalized layers
In the case of acetonitrile vapor adsorption isotherm for sample F2, at low fillings it coincides with n-hexane adsorption isotherm. But with increasing P/Ps, acetonitrile adsorption curve is going higher, which may result from different S20 molecular sizes of acetonitrile and n-hexane. Water adsorption isotherm curve is lower, confirming the hydrophobicity of the sample.