Non-hydrolytic sol-gel synthesis of amine-functionalized silica: Template- and catalyst-free preparation of mesoporous catalysts for CO2 valorization

Non-hydrolytic sol-gel synthesis of amine-functionalized silica: Templateand catalyst-free preparation of mesoporous catalysts for CO 2 valorization Thai Q. Bui a , Tomas Pokorny a , Petr Machac a , Zdenek Moravec a , Eva Domincova Bergerova b , Ales Styskalik a,* a Department of Chemistry, Faculty of Science, Masaryk University, Kotlarska 2, CZ-61137 Brno, Czech Republic b Centre of Polymer Systems, Tomas Bata University in Zlin, tr. Tomase Bati 5678, CZ-76001, Zlin, Czech Republic ARTICLE INFO Keywords: Non-aqueous condensation Silsesquioxane Tertiary amine Heterogeneous catalysis Carbon dioxide Epoxide Cyclic carbonate ABSTRACT Carbon dioxide utilization presents an important and topical research topic. However, the performance of catalysts needed for CO 2 transformations does not achieve the necessary levels for their widespread application. To this end, we decided to study non-aqueous condensations providing amine-functionalized silica catalysts, possibly active in CO 2 -epoxide cycloaddition reaction. While non-hydrolytic sol-gel method is well-known for its efficiency in providing highly porous Lewis and Brønsted acid metallosilicates, here we show for the first time its application for the preparation of silica-based catalysts containing basic groups. First, the reaction conditions were screened to reproducibly obtain porous materials with preserved amine moieties. These were identified as follows: silicon tetraacetate and bridging tertiary amine silanes as precursors, toluene as a solvent, and temperature between 160 and 180 ◦C. In such a way, materials with up to 776 m 2 g −1 and 1.58 cm 3 g −1 were obtained in one-step process, without any template, after conventional drying step. Next, the aminefunctionalized materials were tested in CO 2 -epoxide coupling providing cyclic organic carbonates with high selectivity (>99 %) and moderate activity (up to 86 % epichlorohydrin conversion after 1 h at 120 ◦C and 10 bar CO 2 ). The characterization of spent catalysts revealed a presence of cyclic organic carbonates at the catalyst surface as well as conversion of tertiary amine groups to quaternary ammonium moieties. 1. Introduction Organic-inorganic hybrid materials (OIHMs) have been studied extensively in both academia and industry because they can be tailored to possess desirable properties to suit a wide range of applications thanks to possibility to combine the versatility of organic species with the advantages of inorganic components such as excellent thermal stability and robust structure [1]. Generally, OIHMs can be classified into two major classes depending on the nature of interactions between organic and inorganic phases. Class I hybrids contain weak interactions (van der Waals, π – π , hydrogen bonding, electrostatic) while class II hybrids contain strong (covalent) bonds between two constituents. In the field of heterogeneous catalysis, class II hybrids are preferable to class I ones because strong bonds between organic and inorganic building blocks in the former would favor preservation of the hybrid catalysts during catalytic reaction, minimizing leaching of the active organic moiety into reaction media [2]. Mesoporous silica-based hybrids derived from the sol-gel synthesis method are particularly attractive as potential hybrid heterogeneous catalysts due to the wide availability of organosilicon precursors in the market, the ease of preparation and tuning as well as the high accessibility of active sites and the efficient mass transport [2]. So far, these hybrids have been synthesized mainly by using the traditional sol-gel method, aka the hydrolytic sol-gel (HSG), whose mechanism is related to hydrolysis and polycondensation reactions. However, there are two major limitations in this method when water is used as a solvent [3]. First, different metal/silicon precursors (e. g., Si(OR) 4 vs. Ti(OR) 4 or Si(OR) 4 vs. R’ x Si(OR) 4-x ) may have significantly different hydrolysis as well as polycondensation rates under aqueous conditions leading to phase separation or heterogeneity. Second, due to high surface tension of water, the pore structures of wet gel may collapse during the drying process resulting in materials with poor textural properties. To overcome the limitations of the HSG mentioned above, the nonhydrolytic sol-gel (NHSG) can be seen as an alternative method where * Corresponding author. E-mail address: [email protected] (A. Styskalik). Contents lists available at ScienceDirect Microporous and Mesoporous Materials journal homepage: www.elsevier.com/locate/micromeso https://doi.org/10.1016/j.micromeso.2024.113371 Received 2 August 2024; Received in revised form 9 October 2024; Accepted 11 October 2024 Microporous and Mesoporous Materials 381 (2025) 113371 Available online 12 October 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/ ). the sol-gel processes take place in water-free environments. The NHSG method relies on non-hydrolytic polycondensation reactions between metal/silicon precursors and oxygen donors other than water under nonaqueous conditions and leads to the formation of metal/silicon oxides [3,4]. On one hand, such approach brings several difficulties and environmental concerns including application of organic solvents, lengthy procedures, and work under dry N 2 atmosphere (either in the glovebox and/or applying the N 2 /vacuum manifold). However, these downfalls are counterbalanced by decisive advantages. First, The condensation reaction rates are usually lower and levelled off compared to ones under aqueous environments, leading to the formation of materials with well-controlled properties such as homogeneity, composition, morphology, texture, and surface chemistry. Second, a conventional gel drying provides often highly porous materials. Organic solvents exhibit much lower surface tension in comparison to water and therefore the application of templates or supercritical/freeze drying is not necessary. Thanks to these advantages (but not limited to), materials prepared from the NHSG method have found many applications in industry such as heterogeneous catalysts, luminescent materials, catalyst supports, and Li-ion battery electrodes [3,5]. In the field of heterogeneous catalysis, the NHSG method is particularly beneficial for the preparation of mesoporous mixed oxide catalysts compared to the HSG method in terms of simplicity, versatility, and properties control [6,7]. In addition, this method could also be useful for the preparation of organic-inorganic hybrid catalysts [8]. However, to the best of our knowledge, only mixed oxide catalysts functionalized with alkyl and/or aryl groups have been so far reported as class II hybrid catalysts prepared by the NHSG method [8–16]. Therefore, it is still in its infancy and, indeed, needs more efforts to exploit further the potential of the NHSG method in the preparation of class II hybrid materials, particularly in the field of heterogeneous catalysis. One way of broadening their spectra of class II hybrid catalysts derived from the NHSG method is through the introduction of amine groups into inorganic materials structure. Amine is one of the most attractive organic functional groups due to its wide applications. Particularly, tertiary amines and heterocyclic amines are promising metaland halogen-free organocatalysts for the synthesis of cyclic organic carbonates from CO 2 and epoxides [17–21]. This cycloaddition reaction is one of the most promising routes for chemical fixation of CO 2 on industrial scale due to 100 % atom economy and producing valuable cyclic carbonate products. While homogeneous catalysts provide excellent catalytic activity, heterogeneous catalysts are technically preferable in industry due to the ease of catalyst separation and recycling as well as the ease of application in continuous-flow processes. Several types of heterogeneous catalysts containing nitrogen-amine active sites for this reaction have been successfully developed such as N-doped carbons [22], mesoporous melamine-formaldehyde resins [23], covalent organic frameworks (COFs) [24], metal-organic frameworks (MOFs) [25], poly(ionic liquid)s (PILs) [26], and mesoporous amine-silica hybrids [27–29]. Among them, mesoporous amine-silica hybrids are particularly attractive due to metaland halogen-free nature, low-cost, simple preparation, efficient mass transport, and robust structure. So far, mesoporous amine-silica catalysts for the cycloaddition reaction have been prepared mainly by grafting amine precursors on mesoporous silica supports [27–29]. Although this post-modification method exhibited considerable success in the preparation of mesoporous amine-silica catalysts, an effective and simple one-pot synthesis (aka direct synthesis) is highly desired. In this regard, the use of bridged trialkoxysilylated amine precursors for direct sol-gel synthesis of mesoporous amine-silica hybrids is highly promising. Several mesoporous amine-silica hybrids derived from bridged trialkoxysilylated amine precursors showed high porosity, high thermostability, high content of surface amine groups [30–32], and were applied in several applications such as heterogeneous organocatalysts [33,34], CO 2 adsorbents [35–37], perfluorinated compounds adsorbents [38], Hg(II) adsorbents [39], and dyes adsorbents [40]. However, there is little information about applying these materials for the cycloaddition reaction between epoxides and CO 2 . In addition, these materials have been prepared mainly via traditional hydrolytic sol-gel and using structure-directing agents to create mesopores (Table S1). In this report, we applied a template-free and one-pot NHSG method to co-condense mono/bis/tris-trimethoxysilylated amine precursors bearing aliphatic amine moieties with silica precursors (SiCl 4 and Si (OAc) 4 ). Our aim was to prepare mesoporous amine-silica class II hybrids as potential metaland halogen-free catalysts for the CO 2 -epoxide cycloaddition reaction. First, the ideal reaction conditions were unambiguously identified. Second, the structure and the adsorption properties of the resulting materials were described in detail. Finally, the comparison of catalytic performance of amine-silica materials with their homogeneous analogues as well as characterization of spent catalysts revealed an interesting behavior of heterogeneous catalysts prepared by NHSG polycondensation. 2. Experimental General information is written in Supporting Information. 2.1. Synthesis of hybrid amine-silica xerogels Novel hybrid amine-silica materials were prepared in one pot using non-hydrolytic sol-gel (NHSG) method, specifically via alkyl halide and ester elimination routes [3,5]. The structures of amine and silica precursors in this study are presented in Fig. 1. Anhydrous dichloromethane (DCM), tetrahydrofurane (THF), and toluene (TOL) were used as solvents. Typically, we mixed 3 g of a silica precursor SiX 4 (X =CH 3 COO or Cl) with 10 mL of an aprotic solvent (DCM/THF/TOL) in a Teflon-lined stainless-steel autoclave (100 mL) under N 2 atmosphere in a dry box. Subsequently, we added to the mixture a stoichiometric amount of an amine precursor (Equations (1)─4, Fig. 1), which contains either a terminal primary amine group (denoted as N1) or a bridging secondary/ tertiary amine group (denoted as N2/N3-Me/N3). The resulting mixture was magnetically stirred until we got a clear solution. Next, the autoclave was sealed and aged at a desired temperature (140─200 ◦C) in an oven for 4 days under autogenous pressure. After this, the autoclave was cooled to room temperature and then opened inside a dry box. The obtained gel was crushed, transferred to a Schlenk vessel, and dried under vacuum at 120 ◦C overnight to remove volatile products (e.g., CH 3 X) and solvent. The volatile products were identified by using GCMS. 3 SiX 4 +4 (CH 3 O) 3 Si(CH 2 ) 3 NH 2 (N1) →12 CH 3 X+Si 7 O 12 (CH 2 ) 12 (NH 2 ) 4 (eq. 1) 3 SiX 4 +2 [(CH 3 O) 3 Si(CH 2 ) 3 ] 2 NH (N2) →12 CH 3 X+ Si 7 O 12 (CH 2 ) 12 (NH) 2 (eq. 2) 3 SiX 4 +2 [(CH 3 O) 3 Si(CH 2 ) 3 ] 2 NCH 3 (N3-Me) →12 CH 3 X+ Si 7 O 12 (CH 2 ) 12 (NCH 3 ) 2 (eq. 3) 9 SiX 4 +4 [(CH 3 O) 3 Si(CH 2 ) 3 ] 3 N(N3) →36 CH 3 X+Si 21 O 36 (CH 2 ) 36 - N 4 (eq. 4) 3. Results and discussion 3.1. Ester elimination route: The NHSG condensation and porosity of hybrid materials In this route, Si(OAc) 4 was used as a silica precursor together with 4 amine precursors (N1/N2/N3-Me/N3) containing trimethoxysilyl groups to synthesize hybrid amine-silica materials (Equations (1)─4). T.Q. Bui et al. Microporous and Mesoporous Materials 381 (2025) 113371 2 The silica and amine precursors were mixed with an equal number of acetoxy and methoxy functional groups in an aprotic solvent. The presence of methyl acetate as a volatile product in all synthesis experiments was confirmed by GC-MS analysis of residue solvents after the non-hydrolytic sol-gel (NHSG) synthesis. The so-called ester elimination has been reported to provide a broad variety of materials including silicophosphates [41], their hybrid derivatives [42], metallosilicates [43,44], and metal trimethylsiloxides [45]. While Lewis acidity of the metal centers has usually been reported to drive the NHSG condensation [45], such sites are lacking in our case. Based on the fact that there was no gelation between Si(OAc) 4 and trimethoxymethylsilane CH 3 Si (OCH 3 ) 3 under similar synthesis conditions (10 mL of DCM, 180 ◦C, 4 days), we believe that N-sites from amine precursors could play a role as a catalyst for the NHSG synthesis of amine-functionalized silicas. It should be noted that carboxylic acid esters (i.e., methyl acetate, silicon tetraacetate) can react with primary and secondary amines forming secondary and tertiary amides, respectively [27,46]. This reaction should be accompanied by methanol formation. GC-MS analysis confirmed MeOH presence in residue volatiles after NHSG syntheses when applying N1 and N2 precursors. Therefore, the possible amide formation was carefully checked (see section 3.2 Ester elimination route: The structure of hybrid materials). The N 2 adsorption–desorption isotherms and corresponding NLDFT pore size distributions of some selected hybrid xerogels obtained from the ester elimination route under different conditions (amine precursor, aprotic solvent, temperature) are given in Fig. S1. The isotherms adopt type IV typical for mesoporous materials (except for sample N1 which was non-porous, see discussion below). The hysteresis loops are mostly H2-type indicating the presence of irregular mesopores with complex pore structures [47]. Some samples exhibit steep N 2 adsorption at p/p 0 >0.9 indicating the presence of interparticle voids [47]. Indeed, the NLDFT models (Fig. S1, right) show pore sizes ranging in the mesopore region (and to some extent in the macropore region) over tens of nanometers as can be expected for the NHSG synthesis applying no templating agents. The qualitative N 2 adsorption-desorption isotherms evaluation agrees well with the morphology of the materials observed by SEM and TEM (Figs. S2 and S3). The micrographs show irregular sponge-like particles with sizes in the micron range that appear to contain both mesoand macropores. Table 1 summarizes the quantitative textural properties of hybrid amine-silica materials synthesized from the ester elimination route under different synthetic conditions (amine precursor, aprotic solvent, Fig. 1. Chemical structures of silica and amine precursors in this study. Table 1 Textural properties of hybrid amine-silica xerogels obtained from the ester elimination route (Si(OAc) 4 as a silica precursor) under different conditions. Gel Amine precursor Aprotic solvent Temp. (◦C) S BET (m 2 g −1 ) a V total (cm 3 g −1 ) b V micro /V tot. (%) c PS DFT (nm) d PS aver. (nm) 1 N1 TOL 180 <10 n.d. e n.d. n.d. n.d. 2 N2 DCM 180 664 0.43 12.6 2.6 2.6 3THF 180 708 1.18 4.2 10.5 6.7 4TOL 180 624 1.32 3.1 7.0 8.5 5TOL 160 426 0.53 3.6 6.6 5.0 6 N3-Me DCM 180 604 0.58 2.9 2.6 3.8 7THF 180 700 0.63 9.0 2.6 3.6 8TOL 200 303 0.92 1.0 29.4 12.2 9TOL 180 776 1.58 0.7 13.9 8.1 10 TOL 160 666 0.93 1.9 3.2 5.6 11 TOL 140 141 0.47 0.0 7.0 13.2 12 N3 TOL 180 761 1.26 2.7 6.8 6.6 13 TOL 140 396 0.61 0.5 6.8 6.1 a Estimated at p/p o =0.97. b Based on t-plot analyses (Fig. S4). c The maximum value of the pore size distribution curve (NLDFT, ads, cyl. pore model). d 4V total /S BET . e Not determined. T.Q. Bui et al. Microporous and Mesoporous Materials 381 (2025) 113371 3 temperature). It is clearly seen from Table 1 that materials derived from bridging amine precursors (N2/N3-Me/N3) often exhibited high specific surface areas (S BET ; 141─776 m 2 g −1 ), high total pore volumes (V total ; 0.43─1.58 cm 3 g −1 ), and large mesopores. All three bridging amine precursors (N2/N3-Me/N3) reacted with Si(OAc) 4 at different temperatures and in various solvents. The comparison of both S BET and V total clearly highlights 180 ◦C as the ideal synthetic temperature; specific surface areas and total pore volumes are lower at both higher (200 ◦C) and lower (140 ◦C, 160 ◦C) temperatures (Table 1). This observation is probably connected with the condensation degree: the materials synthesized at 180 ◦C exhibit only low signals of unreacted organic groups and, at the same time, no signs of decomposition (see section 3.2 Ester elimination route: Structure of hybrid materials). Regarding the effect of solvents on porosity, materials synthesized in dichloromethane (DCM), tetrahydrofuran (THF), and toluene exhibited similar S BET ranging from 604 to 776 m 2 g −1 . However, the V total was always significantly higher in nonpolar toluene than in polar DCM or THF (Table 1). In contrary to samples prepared with bridging amine precursors (N2/ N3-Me/N3), the material derived from the terminal amine precursor (N1) was non-porous (S BET <10 m 2 g −1 ) under template/additive/ catalyst-free and NHSG synthesis conditions. This observation agrees with reported results from NHSG synthesis of hybrid silicophosphate xerogels [42]. Specifically, the co-condensation reactions between Si (OAc) 4 and terminal precursors R─P(O)(OSiMe 3 ) 2 (R =alkyl or aryl group) produced non-porous materials while the co-condensation reactions between Si(OAc) 4 and bridged ones (Me 3 SiO) 2 (O)P─R─P(O) (OSiMe 3 ) 2 produced highly porous materials (553─617 m 2 g −1 ). Similar results were also observed from the use of terminal and bridged silane precursors (MeSi(OAc) 3 vs. (AcO) 3 Si─R─Si(OAc) 3 , respectively) with tris(trimethylsilyl)phosphate P(O)(OSiMe 3 ) 3 [42]. The improved porosity when applying bridging precursors comes from the additional cross-linking introduced by the organic bridge, while the organosilane precursors with terminal organic groups provide, in fact, a lower connectivity in comparison to both bridged and convenient (i.e., four-connected) silica precursors [42]. To the best of our knowledge, there is no report related to using N1/ N2/N3-Me/N3 amine precursors in the template/additive/catalyst-free and NHSG synthesis of hybrid amine-silica materials. Most importantly, even without using any template/additive/catalyst, the mesoporous materials (with a clear hysteresis loop) derived from bridging amine precursors (N2/N3-Me/N3) using NHSG exhibited comparable or even better textural properties compared to the ones using hydrolytic sol-gel (HSG) approach (Table S1). 3.2. Ester elimination route: The structure of hybrid materials In order to confirm the structural integrity of organic moieties before and after NHSG synthesis under different conditions, we performed solid-state 13 C CP MAS NMR measurements for the hybrid amine-silica xerogels and compared the results with liquid-state 13 C NMR (in CDCl 3 ) spectra of their corresponding amine precursors. The results show that the synthesis performed in toluene at 180 ◦C was the optimum condition in our study to preserve the organic structure of the amine precursors in their corresponding hybrid xerogels (Fig. 2,S5, S6) [30]. The additional signal at ~170 ppm represents carbonyl groups in residual acetoxy or acetamide groups (see explanation below) [41]. The signal of corresponding methyl groups (22.6 ppm in silicon tetraacetate [41]) is overlapping with signals coming from methylene moieties in amine precursors. From Fig. S5 we can clearly see “unusual”peaks in 13 C CP MAS NMR spectra of materials synthesized in DCM, especially in the case of N3-Me precursor, compared to ones synthesized in TOL or THF. The possible reasons for this observation could be due to side reactions between amine sites and DCM under synthetic conditions [48]. Similar patterns were observed in alkyl halide elimination (CH 3 Cl produced as volatile product) and in spent catalysts (reaction with epichlorohydrin and/or [4-(chloromethyl)-1,3-dioxolan-2-one]) and were explained by occurrence of side reactions on amine sites, i.e., quaternization, reverse Menschutkin, and Hofmann reactions (see section 3.6 Recyclability studies and spent catalysts characterization) [48–51]. Regarding the effect of temperature, it can be seen from Fig. S6 that intensities of methoxy and acetoxy groups decrease when the synthetic temperature increases from 140 to 180 ◦C, indicating a higher degree of co-condensation. However, the appearance of unwanted peaks at 65 ppm and 26 ppm together with the significant reducing of intensities of signals at 63 ppm and 45 ppm indicate decomposition of organic moieties in toluene at 200 ◦C (Fig. S6). This observation also agrees with the N 2 physisorption results in which the hybrid xerogels synthesized in toluene at 180 ◦C showed the optimum textural properties (Table 1). Importantly, the quantitative 29 Si MAS NMR spectra (Fig. 3) of 4 hybrid materials synthesized under optimum conditions show that there was a great agreement between the experimental vs. theoretical ratio Ttype silicon atoms over Q-type silicon atoms (theoretical ratio T/Q =4/ 3 or 1.33), indicating the validity of proposed co-condensation reactions (Equations (1)─4). Results from the thermogravimetric analysis (TGA) show that these hybrid amine-silica materials started to decompose at around 200─225 ◦C under air flow (Fig. S7a) while they were stable up to around 350 ◦C under N 2 flow (Fig. S7b). These observations indicate the Fig. 2. Solid-state 13 C CP MAS NMR spectra of 4 representative hybrid materials synthesized under optimum conditions (toluene, 180 ◦C) from the ester elimination route and liquid-state 13 C NMR (in CDCl 3 ) spectra of their corresponding amine precursors. Asterisk mark denotes rotational sidebands. T.Q. Bui et al. Microporous and Mesoporous Materials 381 (2025) 113371 4 presence of organic functional groups in the xerogel structure. We assumed that, first, residue solids at the end of TGA experiments (till 1000 ◦C) under air flow were composed of SiO 2 only (Fig. S7a) and, second, all Si and N atoms from silica and amine precursors were transferred completely to final hybrid xerogels (Equations (1)─4). It should be noted that Equations (1)–(4) are ideal equations where the degrees of condensation (DC) are 100 %. In our second assumption, the real DC (<100 %) does not affect the Si/N molar ratio in the final gel because with different values of DC, the final gel will only have different amount of unreacted methoxy and acetoxy groups (ester elimination route). Based on our 2 assumptions, we calculated N contents of representative hybrid xerogels based on TGA results performed in air and presented them in Table 2. An example of our calculations for N content of SiOAc-N3 is provided in the Supporting Information. Interestingly, these calculated values of N content agreed very well with experimental values obtained from organic elemental analysis (Table 2), indicating the validity of our assumptions as well as agreement with the structural integrity of organic moieties from solid-state NMR results mentioned above (Figs. 2 and 3). Calculated Si contents of hybrid xerogels and mass loss values obtained from TGA-Air are presented in Table S2. XPS and FT-IR measurements were further conducted to determine the surface elemental compositions and structural units of the hybrid amine-silica materials. The survey scan XPS spectra (not shown) revealed the appearance of 4 elements (Si, N, C, and O) on the surface of these materials, indicating the presence of hybrid organic-inorganic structure. The high-resolution XPS profiles of their Si2p, N1s, C1s, and O1s branches are shown in Fig. 4. The Si2p spectra (Fig. 4a) show 2 major peaks at binding energies (BE) ~102.7 and ~103.6 eV corresponding to Q-type silicon (SiO 4 ) and T-type silicon (SiO 3 C) configurations, respectively [52,53]. It should be noted that the T/Q ratio in Si2p spectra was set to 1.33 to be consistent with the theoretical T/Q ratio in bulk materials when performing the curve-fitting by using CasaXPS software. The N1s spectra (Fig. 4b) show that the main N-species in 2 materials SiOAc-N3 and SiOAc-N3-Me was amine-N (BE 399.2 eV) as expected while, in contrast, the main N-species in the other 2 materials SiOAc-N2 and SiOAc-N1 was amide-N (BE 399.9─400.0 eV) [54]. This observation was also consistent with the appearances of amide-C (O=C─N) at 287.8─288.2 eV in the C1s spectra (Fig. 4c) and amide-O (O=C─N) at 531.0─531.4 eV in the O1s spectra (Fig. 4d) [54]. Table S3 summarizes our assignments for high-resolution XPS profiles of 4 representative hybrid materials. The FT-IR spectra agreed well with the conclusions based on XPS spectroscopy. The FT-IR spectrum of SiOAc-N1 (Fig. 5, black) clearly confirmed the presence of secondary amide group with 4 typical bands: N-H stretch (3273 cm −1 ), overtone of N-H bend (3085 cm −1 ), C=O stretch (1633 cm −1 ), and N-H bend (1551 cm −1 ) [55] while the one of SiOAc-N2 (Fig. 5, red) confirmed the presence of tertiary amide group via the only C=O stretch band at 1625 cm −1 . The presence of tertiary amine groups in SiOAc-N3 and SiOAc-N3-Me was also confirmed by their FT-IR spectra (Fig. 5, blue and green, respectively) via the typical band at 2790─2802 cm −1 representing for C-H stretch of methyl/- methylene groups next to N in tertiary amine [30,31,56]. With results obtained from XPS and FT-IR, we conclude that only tertiary amine precursors (N3 or N3-Me) could co-polymerize with Si(OAc) 4 to form hybrid amine-silica materials under our non-hydrolytic sol-gel conditions while primary (N1) and secondary (N2) amine precursors formed hybrid secondary and tertiary amide-silica materials, respectively due to in-situ reactions between primary/secondary amine sites and co-product methyl acetate and/or Si(OAc) 4 precursor [27,46]. In agreement with XPS results, the STEM-EDS analysis of a selected hybrid material (SiOAc-N3-Me) showed the appearance of 4 elements (Si, N, C, and O) in its elemental composition. Most importantly, the STEM-EDS elemental mapping (scale bar 50 nm) also revealed a uniform distribution of these 4 elements with no single clusters (Fig. S8). The FT-IR spectra (Fig. 5) also reveal that SiOAc-N3 and SiOAc-N3Me contained unreacted acetoxy groups in their structure (C=O stretch at 1740─1742 cm −1 ) while the other 2 materials SiOAc-N2 and SiOAc-N1 did not contain them. This observation also agreed with the weak signal of ester-C (O=C─O) at 288.7─289.1 eV in the highresolution C1s spectra (Fig. 4c), the weak signal of ester-O (O=C─O) at 533.8─533.9 eV in the high-resolution O1s spectra (Fig. 4d), and the presence of unreacted methoxy and acetoxy groups in the solid-state 13 C CP MAS NMR spectra (Fig. 2) of SiOAc-N3 and SiOAc-N3-Me [54]. Since the silica and amine precursors were mixed with an equal number of acetoxy and methoxy functional groups at the beginning and the non-hydrolytic sol-gel reactions followed the ester elimination route to form methyl acetate as a volatile co-product together with hybrid gels, we assume that the more unreacted methoxy groups appeared in hybrid materials, the more unreacted acetoxy groups remained as well. Based on results from XPS and FT-IR as well as the absence of unreacted methoxy groups in the 13 C CP MAS NMR spectra (Fig. 2) of SiOAc-N1 and SiOAc-N2 materials, we could also conclude that the peaks at 174.0 &171.5 ppm were not assigned to ester-C (O=C─O) but amide-C (O=C─N) instead in the 13 C CP MAS NMR spectra (Fig. 2) of SiOAc-N1 and SiOAc-N2 materials, respectively. The absence of unreacted methoxy and acetoxy groups of SiOAc-N1 and SiOAc-N2 materials could be due to higher degrees of co-polymerization compared to SiOAc-N3 and SiOAc-N3-Me materials and/or the in-situ formation of amide sites (confirmed by IR and XPS spectroscopy) leading to MeOH (observed by GC-MS) and/or ≡SiOH formation under synthetic conditions. However, it should be noted that the FT-IR spectra (Fig. 5) showed no significant bands in the range 950─850 cm −1 (i.e., absorption band characteristic for Si−OH stretching vibration [57]), indicating insignificant appearance of silanol groups (Si─OH) in our hybrid materials compared to ones prepared via hydrolytic sol-gel method [30,31,56]. Fig. 3. Solid-state 29 Si MAS NMR (one pulse) spectra of 4 representative hybrid materials synthesized under optimum conditions (toluene, 180 ◦C) from the ester elimination route. Table 2 Nitrogen contents of 4 representative hybrid materials synthesized under optimum conditions (toluene, 180 ◦C) from the ester elimination route. Material SiOAc-N1 SiOAc-N2 SiOAc-N3-Me SiOAc-N3 a Calc. N (wt.%) 7.21 4.09 4.37 2.86 b Exp. N (wt.%) 6.99 4.00 4.34 2.80 a Calculated from calc. Si contents (Table S2) and theoretical ratios Si/N of representative hybrid xerogels (Equations (1)─4). b Obtained from Organic Elemental Analysis (CHNS). T.Q. Bui et al. Microporous and Mesoporous Materials 381 (2025) 113371 5 3.3. Ester elimination route: CO 2 adsorption on hybrid materials Fig. S9─12 illustrate CO 2 adsorption isotherms of 4 representative hybrid xerogels and their corresponding Freundlich–Langmuir (aka Sips) model fittings (R 2 =0.99996–1.00000) at 3 different temperatures (0, 15, and 25 ◦C) [58,59]. Fig. S13 illustrates changes of the isosteric enthalpy of CO 2 adsorption (ΔH CO2_ads ) during the adsorption process obtained by Clausius–Clapeyron approach via Sips fit (i.e., Freundlich-Langmuir model) of CO 2 adsorption isotherms [58,59]. Table 3 summarizes main results obtained from these isotherms including the CO 2 uptake at 1 bar of CO 2 and the isosteric heat of CO 2 adsorption (Q st = − ΔH CO2_ads ) at near-zero coverage. As expected, with N sites in the form of amide or tertiary amine and lack of other sites for CO 2 chemisorption, the Q st values of hybrid materials were found in the range of 15–33 kJ mol −1 , representing mainly CO 2 physisorption (Q st < 50 kJ mol −1 ) [60]. The comparable Q st values (~25–27 kJ mol −1 ) of SiOAc-N3 and SiOAc-N3-Me materials could be due to the similarities in specific surface area (761–776 m 2 g −1 ) and surface functional groups (e. g., tertiary amine, acetoxy, methoxy). Meanwhile, with slightly lower specific surface area (624 m 2 g −1 ), the highest Q st value (33.23 kJ mol −1 ) of SiOAc-N2 could be due to the presence of surface functional groups with higher polarities (e.g., tertiary amide, silanol). In contrast, the lowest Q st value (14.54 kJ mol −1 ) of SiOAc-N1 could be due to its non-porosity (<10 m 2 g −1 ) and the dissolution of CO 2 into the material matrix rather than only interactions on the material surface [61]. Basically, the heat of CO 2 absorption into dense rubbery matrices (Q =–ΔH) includes 2 contributions: the heat of binding energy of CO 2 in matrix (Q 1 >0) and the heat of reorganization (Q 2 <0) [62]. Specifically, the Q st value of SiOAc-N1 is quite comparable to the heat of CO 2 dissolution into the polymeric matrix of poly(methyl propyl siloxane) with Q = 14.69 kJ mol −1 [63]. Overall, the CO 2 uptake of hybrid materials decreased when increasing sorption temperature from 0 to 25 ◦C and did not reach a plateau at 1 bar of CO 2 (Fig. S9─12), indicating that the CO 2 uptake could be improved further at higher CO 2 pressure. The CO 2 uptake (mmol/g) at same conditions is in the order of SiOAc-N2 >SiOAc-N3 ~ SiOAc-N3-Me >SiOAc-N1 (Table 3). This order also agrees with the differences in porosity and surface functional groups of materials already mentioned above when comparing Q st values. It should be noted that in carbon dioxide physisorption, CO 2 molecules are attracted by solid adsorbent mainly via van-der-Waals forces as well as electrostatic interactions between CO 2 molecules and polar sites on adsorbent surface thanks to having CO 2 quadrupole moment [64]. Although the similarities in specific surface area (761–776 m 2 g −1 ) and surface functional groups (e.g., tertiary amine, acetoxy, methoxy) of 2 materials SiOAc-N3 and SiOAc-N3-Me, the CO 2 uptake per nitrogen site (mmol CO 2 /mmol N) in SiOAc-N3 is higher than SiOAc-N3-Me (e.g., 0.200 vs 0.125 mmol CO 2 /mmol N at 25 ◦C, respectively, Table 3), indicating minor contribution of tertiary amine for CO 2 uptake at 1 bar of CO 2 . Specifically, comparing to SiOAc-N3 (similar specific surface area), higher content of tertiary amine (higher N content) in SiOAc-N3-Me (4.34 vs 2.80 wt% N, respectively, Table 4) did not lead to higher CO 2 uptake (mmol/g) at 1 bar of CO 2 but lower CO 2 uptake per nitrogen site (mmol CO 2 /mmol N). The hybrid material SiOAc-N2 derived from NHSG also showed a Fig. 4. High-resolution (a) Si2p, (b) N1s, (c) C1s, and (d) O1s XPS spectra of 4 representative hybrid materials synthesized under optimum conditions (toluene, 180 ◦C) from the ester elimination route. A minor component at 105.0 eV in Fig. 4a corresponds to SiO 4 and CSiO 3 moieties in hydrogen bonding (Table S3) [52]. T.Q. Bui et al. Microporous and Mesoporous Materials 381 (2025) 113371 6 comparable or even better CO 2 sorption capacity compared to similar hybrid materials derived from HSG using the same amine precursor (N2) probably thanks to a higher specific surface area of SiOAc-N2 [35,65]. 3.4. Alkyl halide elimination route In this route, SiCl 4 was used as a silica precursor together with 4 amine precursors (N1/N2/N3-Me/N3) containing trimethoxysilyl groups to synthesis of hybrid materials (Equation (1)─4). The silica and amine precursors were mixed with an equal number of chloride and methoxy functional groups in an aprotic solvent. The presence of methyl chloride as a volatile product in all synthesis experiments was confirmed by GC-MS analysis of residue solvents after the non-hydrolytic sol-gel (NHSG) synthesis. Table 4 summarizes the textural properties of hybrid materials synthesized from the alkyl halide elimination route under different synthetic conditions (amine precursor, aprotic solvent). Like the ester elimination route, materials derived from the terminal amine precursor (N1) were also low/non-porous (2─14 m 2 g −1 ) in the alkyl halide elimination route. In contrast, even though toluene (TOL) was the optimal solvent to prepare highly porous materials with preserved organic moieties coming from bridging amine precursors in the ester elimination route (Table 1,Fig. 2 and S5), the material synthesized from SiCl 4 and N2 precursor in TOL was non-porous (<10 m 2 g −1 ). The Fig. 5. FT-IR spectra of 4 representative hybrid materials synthesized under optimum conditions (toluene, 180 ◦C) from the ester elimination route. Table 3 CO 2 uptakes and isosteric heat values of CO 2 adsorption of 4 representative hybrid xerogels synthesized under optimum conditions (toluene, 180 ◦C) from the ester elimination route. Material CO 2 uptake at 1 bar of CO 2a (mmol/g) Q st at near-zero coverage (kJ mol −1 ) 0◦C 15 ◦C 25 ◦C SiOAc-N1 0.174 (0.035) 0.136 (0.027) 0.092 (0.018) 14.54 ±5.41 SiOAc-N2 1.014 (0.355) 0.742 (0.260) 0.584 (0.204) 33.23 ±0.05 SiOAc-N3Me 0.674 (0.218) 0.482 (0.156) 0.386 (0.125) 24.67 ±1.62 SiOAc-N3 0.731 (0.366) 0.543 (0.272) 0.399 (0.200) 26.59 ±3.15 a Numbers in parenthesis are presented in mmol CO 2 /mmol N. The N contents are obtained from Organic Elemental Analysis (CHNS). Table 4 Textural properties of hybrid amine-silica xerogels obtained from the alkyl halide elimination route (SiCl 4 as a silica precursor) under different conditions. Gel Amine precursor Aprotic solvent Temp. (◦C) S BET (m 2 g −1 ) a V total (cm 3 g −1 ) b V micro /V tot. (%) c PS DFT (nm) d PS aver. (nm) 1 N1 TOL 180 <10 n.d. n.d. n.d. n.d. 2DCM 180 14 0.05 0 16.1 13.4 3 N2 TOL 180 <10 n.d. n.d. n.d. n.d. 4DCM 180 351 0.65 5.2 2.6 7.4 5 N3-Me DCM 180 300 0.50 12.0 2.6 6.7 6 N3 DCM 180 575 1.31 4.2 2.6 9.1 n.d. =not determined. a Estimated at p/p o =0.97. b Based on t-plot analysis (Fig. S16). c The maximum value of the pore size distribution curve (NLDFT, ads, cyl. pore model). d 4V total /S BET . T.Q. Bui et al. Microporous and Mesoporous Materials 381 (2025) 113371 7 sample derived from bridging amine precursor N2 and SiCl 4 in dichloromethane (DCM) showed high specific surface area (351 m 2 g −1 ) and high total pore volume (0.65 cm 3 g −1 ). Accordingly, reactions between SiCl 4 and N3 and N3-Me were performed in DCM and exhibited high S BET (300─575 m 2 g −1 ), high V total (0.50─1.31 cm 3 g −1 ), and pore size distribution in mesopore range (Table 4). The solid-state 13 C CP TOSS MAS NMR spectra revealed the complicated organic structures of the hybrid materials obtained from the alkyl halide elimination route either in TOL or DCM (Fig. S14), indicating a negative effect of using SiCl 4 as a silica precursor on the structural integrity of organic moieties after NHSG synthesis. This observation could be mainly due to in-situ reactions between amine sites and co-product CH 3 Cl under synthetic conditions to form a complicated mixture including quaternary ammonium chloride salts as well as their corresponding thermal decomposition products via the reverse Menschutkin and/or Hofmann reactions (described in detail in section 3.6 Recyclability studies and spent catalysts characterization with [4- (chloromethyl)-1,3-dioxolan-2-one] as the model quaternization agent) [49–51]. Noteworthy, the quaternary (alkyl)ammonium halides are often used as co-catalysts for CO 2 cycloaddition reactions with iodides and bromides being highly preferred [66,67]. Therefore, application of SiBr 4 or SiI 4 in alkyl halide elimination might lead to catalytically interesting materials. The survey scan XPS spectra (not shown) revealed the appearance of 5 elements (Si, N, C, O, and Cl) on the surfaces of representative hybrid materials prepared in DCM or TOL by the alkyl halide elimination route. The high-resolution N1s XPS spectra (Fig. 6a) showed 2 major peaks at binding energies (BE) ~399.5 and ~401.9 eV corresponding to amine-N (N A ) and quaternary ammonium N (N QA ), respectively [54]. The N QA /N A ratio decreased from 8.3 to 1.8 when the steric hindrance on N site increased (i.e., from N1 to N3 precursor), indicating the decrease in degree of quaternization of amine sites (Fig. 6a). Meanwhile, the high-resolution Cl2p XPS spectra (Fig. 6b) revealed the presence of Cl − anions from quaternary ammonium salts at BE~197.7 eV as well as C─Cl at BE~200.6 eV in the Cl2p3/2 spectra [54]. The appearance of C─Cl in the structure could be due to side reactions between amine sites and DCM under synthetic conditions as already observed in the ester elimination route (Fig. S15) and/or thermal decomposition products of quaternary ammonium chloride salts (see discussion in section 3.6 Recyclability studies and spent catalysts characterization) [48,49]. 3.5. Investigation of synthesized hybrid materials as heterogeneous catalysts for the CO 2 cycloaddition to epoxides In order to investigate the catalytic potential of synthesized hybrid amine-silica materials as alternative metaland halogen-free heterogeneous catalysts for direct conversion of CO 2 to cyclic carbonates under solvent-free and co-catalyst-free conditions, screening catalytic tests were performed in a custom-made high-pressure batch reactor (Fig. S17) using near-stoichiometric amount of CO 2 (~14 mmol or 10 bar of initial CO 2 pressure) and epichlorohydrin (10 mmol) at 120 ◦C for 1 h. Table 5 summarizes the main catalytic results from selected catalysts synthesized using non-hydrolytic sol-gel (NHSG) method via the ester elimination route in TOL at 180 ◦C for 4 days (SiOAc). Increasing the initial CO 2 pressure from 10 to 15 bar (entries 9 and 10) just improved slightly the epoxide conversion, indicating that 10 bar of initial CO 2 pressure is the optimal pressure for our reaction conditions. For the sake of comparison of catalytic activity, selected catalysts synthesized using non-hydrolytic sol-gel (NHSG) method via the alkyl halide elimination in DCM at 180 ◦C for 4 days (SiCl) were also tested at similar catalytic reaction conditions (50 mg of catalyst, 10 bar of CO 2 , 10 mmol of epichlorohydrin, 120 ◦C, 1 h). Results are summarized in Table S4 even though the chemical structure of these materials (SiCl) was not well-defined due to their complicated organic structures (Fig. S14). Overall, all synthesized hybrid materials were active with high selectivity to cyclic carbonate (>99 %) and low-to-high conversion of epichlorohydrin (11─86 %) under investigated conditions. Commercial porous silica showed no activity (entry 1), indicating the catalytic role of N sites in hybrid materials for the CO 2 cycloaddition to epichlorohydrin. The materials prepared by alkyl halide elimination exhibit higher epichlorohydrin conversions than corresponding samples prepared by ester elimination except for amine-functionalized silica prepared from N3-Me (Table 5 and Table S4). On one hand, the materials prepared from SiCl 4 contain quaternary alkylammonium chloride sites that might be beneficial for CO 2 cycloaddition. On the other hand, poorly defined moieties coming from the decomposition of quaternary ammonium sites might be incorporated in SiCl materials as well. Therefore, it is not possible to explain unambiguously the differences between SiCl and SiOAc samples. Among hybrid materials prepared by ester elimination, SiOAc-N3Me showed the highest catalytic activity (TOF =358 h −1 ). Most importantly, the order of catalytic activity in decreasing order of turnover frequency (TOF) was SiOAc-N3-Me >SiOAc-N3 ≫SiOAc-N2 > SiOAc-N1 (entries 2–4, 8), indicating that tertiary amine-N sites (SiOAcN3-Me and SiOAc-N3) were much more active for the reaction than Fig. 6. High-resolution (a) N1s and (b) Cl2p XPS spectra of 4 representative hybrid materials synthesized in DCM from the alkyl halide elimination route. T.Q. Bui et al. Microporous and Mesoporous Materials 381 (2025) 113371 8 amide-N sites (SiOAc-N2 and SiOAc-N1). This observation also agrees with the use of tertiary amines as highly efficient organocatalysts for CO 2 fixations [17]. While catalytic activity appear to strongly depend on the active sites (tertiary amine-N vs. amide N), the porosity does not seem to play a decisive role: Non-porous SiOAc-N1 exhibited a similar epichlorohydrin conversion as highly porous SiOAc-N2. The catalyst SiOAc-N3-Me was also active for the CO 2 cycloaddition to different epoxides other than epichlorohydrin such as styrene oxide and 1,2-butylene oxide (entries 13–15). Among the 3 epoxides, epichlorohydrin is the most reactive one thanks to the presence of an electron-withdrawing chloromethyl group (─CH 2 Cl) that activated the epoxide ring, rendering it more susceptible for nucleophilic attack by tertiary amine-N sites and/or activated CO 2 [17]. The lowest reactivity of styrene oxide could be due to the steric effect of the phenyl group next to the epoxide ring. This observed trend of epoxide reactivity for mesoporous tertiary amine-silica hybrid SiOAc-N3-Me catalyst is also in agreement with mesoporous melamine-formaldehyde resin MMFR catalyst, where epoxide reactivity decreased in the same order (i.e., epichlorohydrin ≫1,2-butylene oxide >styrene oxide) [23]. Reported mesoporous amine-silica hybrid catalysts in the literature for the synthesis of cyclic carbonates from CO 2 in batch reactor under solvent-free condition are also summarized and compared with this work in Table S5. More specifically, although the N contents of our catalysts are lower than the N contents of Si-Imid and SiO 2 -His materials, the catalytic performances of our catalysts are still comparable or even higher [27,28], indicating that aliphatic tertiary amine sites are more efficient than imidazole or histidine in this case. Significantly higher catalytic activities have been reported only for catalysts containing metals and in the presence of co-catalysts: 16 wt% APTES@ZrO 2 -MCM-41 and SBA-15/N-Au showed more than 90 % of cyclic carbonate yield at 80 ◦C [66,67]. Finally, with epichlorohydrin as an epoxide substrate, our catalysts showed higher catalytic performance than most of the reported mesoporous amine-silica (with/without metal) catalysts except for PT@SBA-16, while it is not the case with styrene oxide or 1,2-butylene oxide as an epoxide substrate [27–29,68]. This observation is also true when comparing our catalysts with the reported catalyst B-SBA-15-NH 2 in the presence of co-catalyst KI [69]. The observation could be explained by in-situ reactions of our catalysts with epichlorohydrin and/or [4-(chloromethyl)-1,3-dioxolan-2-one] to create quaternary ammonium chloride as another active form for the synthesis of cyclic carbonate (see mechanism below in section 3.6 Recyclability studies and spent catalysts characterization) [48–51]. In order to understand better the catalytic activity of tertiary amineN sites in mesoporous tertiary amine-silica hybrids (SiOAc-N3-Me and SiOAc-N3), the organic structure analogues (methyldipropylamine (MDPA) and tripropylamine (TPA)) were also tested as homogeneous catalysts at similar catalytic reaction conditions (entries 16–21) and compared in terms of turnover frequency (TOF), apparent activation energy (E a ) and pre-exponential factor (A) (Table 5,Fig. 7). The mass of homogeneous catalysts MDPA and TPA was taken to have a comparable N content in 50 mg of heterogeneous catalysts SiOAc-N3-Me (0.155 mmol of N) and SiOAc-N3 (0.100 mmol of N), respectively, based on organic elemental analysis. The temperature range for comparison was chosen from 80 to 100 ◦C to avoid very slow or fast initial rate of epichlorohydrin conversion under our reaction conditions. The catalytic reaction was assumed to follow pseudo-first-order kinetics based on Table 5 Summary of batch catalytic experiments conducted on the cycloaddition reaction of epoxides and CO 2 . entry catalyst epoxide temp. ( ◦C) time (h) conv. a (%) sel. a (%) k ini.b (h −1 ) TOF c (h −1 ) 1 Commercial SiO 2 (Aerosil 300) epichlorohydrin 120 1 0 – – – 2 SiOAc-N1 epichlorohydrin 120 1 11 >99 0.12 4.7 3 SiOAc-N2 epichlorohydrin 120 1 14 >99 0.15 10.5 4 SiOAc-N3 epichlorohydrin 120 1 56 >99 1.81 181 5  100 1 34 >99 0.42 41.6 6  90 1 20 >99 0.22 22.3 7  80 1 10 >99 0.11 10.5 8 SiOAc-N3-Me epichlorohydrin 120 1 86 >99 5.54 358 9  100 1 52 >99 0.89 57.6 10 d   100 1 55 >99 n.d. n.d. 11   90 1 38 >99 0.51 33.0 12   80 1 25 >99 0.29 18.6 13 styrene oxide 120 14 51 >99 n.d. n.d. 14   140 14 97 >99 n.d. n.d. 15 1,2-butylene oxide 120 14 55 >99 n.d. n.d. 16 Tripropylamine (14 mg, 0.098 mmol) e epichlorohydrin 100 1 42 >99 g 0.42 43.0 17   90 1 16 >99 g 0.17 17.8 18   80 1 7 >99 g 0.07 7.4 19 Methyldipropylamine (17 mg, 0.148 mmol) f epichlorohydrin 100 1 80 >99 g 1.95 132 20   90 1 62 >99 g 0.65 43.9 21   80 1 32 >99 g 0.39 26.1 n.d. =not determined. a Based on 1 H NMR and GC-MS analysis. b Calculated based on the initial rate of epichlorohydrin conversion. c TOF =kini.*nepoxide nN sites with n N-sites based on Organic Elemental Analysis. d Initial pressure of CO 2 =15 bar. e Organic structure analog of SiOAc-N3, comparable N content based on Organic Elemental Analysis. f Organic structure analog of SiOAc-N3-Me, comparable N content based on Organic Elemental Analysis. g Excluding by-products derived from reactions between amines and epichlorohydrin/cyclic carbonate due to difficulty in distinguishing by NMR. T.Q. Bui et al. Microporous and Mesoporous Materials 381 (2025) 113371 9