Microporosity and CO2 Capture Properties of Amorphous Silicon Oxynitride Derived from Novel Polyalkoxysilsesquiazanes

Polyalkoxysilsesquiazanes ([ROSi(NH)1.5]n, ROSZ, R = Et, nPr, iPr, nBu, sBu, nHex, sHex, cHex, decahydronaphthyl (DHNp)) were synthesized by ammonolysis at −78 °C of alkoxytrichlorosilane (ROSiCl3), which was isolated by distillation as a reaction product of SiCl4 and ROH. The simultaneous thermogravimetric and mass spectrometry analyses of the ROSZs under helium revealed a common decomposition reaction, the cleavage of the oxygen–carbon bond of the RO group to evolve alkene as a main gaseous species formed in-situ, leading to the formation of microporous amorphous Si–O–N at 550 °C to 800 °C. The microporosity in terms of the peak of the pore size distribution curve located within the micropore size range (<2 nm) and the total micropore volume, as well as the specific surface area (SSA) of the Si–O–N, increased consistently with the molecular size estimated for the alkene formed in-situ during the pyrolysis. The CO2 capture capacity at 0 °C of the Si–O–N material increased consistently with its SSA, and an excellent CO2 capture capacity of 3.9 mmol·g−1 at 0 °C and CO2 1 atm was achieved for the Si–O–N derived from DHNpOSZ having an SSA of 750 m2·g−1. The CO2 capture properties were further discussed based on their temperature dependency, and a surface functional group of the Si–O–N formed in-situ during the polymer/ceramics thermal conversion.

Recently, we reported a novel single source precursor for a silicon oxynitride (Si-O-N) system, ethoxysilsesquiazane ([EtOSi(NH) 1.5 ] n , EtOSZ) [24]. Under inert atmosphere up to 800 • C, this polymer Recently, microporous structure controlling in nitride-based amorphous solid sorbents has been proposed as an alternative approach to harmonize CO 2 capture capacity and regenerability. Zhao et al. reported synthesis of porous carbon nitride (CN) spheres for CO 2 capture [42]. The resulting CN spheres possessed-at 1 bar-a CO 2 capture capacity of 2.90 at 25 • C and 0.97 mmol·g −1 at 75 • C, superior to those of the pure carbon materials with an analogous porous structure. In addition to the abundant nitrogen-containing basic groups and the hierarchical mesostructure with a relatively high Brunauer-Emmett-Teller specific surface area, it was suggested that a large number of micropores and small mesopores could contribute to the enhanced CO 2 capture capacity, owing to the capillary condensation effect. Schitco et al. also reported that polymer-derived ultra-microporous amorphous silicon nitride showed an excellent CO 2 capture capacity of 2.35 mmol·g −1 at 0 • C and 1 bar. They proposed that the observed high CO 2 storage capacities could be achieved in materials with high amount of ultra-micropore volume [15].
In this study, CO 2 capture capacities at 0 to 40 • C of the present ROSZ-derived microporous amorphous Si-O-N materials were evaluated and briefly compared with those in literature, then the capture behaviors under CO 2 atmosphere were in-situ analyzed by using thermogravimetry and infrared spectroscopy.

Precursor Synthesis
The handling of all the reagents and products in this study was performed under an inert atmosphere of pure nitrogen (N 2 ). The starting alcohols, tetrachlorosilane and reaction solvent of dry tetrahydrofuran (reagent grade) were purchased from Wako Pure Chemicals Industry, Osaka, Japan, and used without further purification. The purity of gaseous ammonia used in this study was >99.9% (Sumitomo Seika Chemicals, Osaka, Japan). Eight kinds of [ROSi(NH) 1.5 ] n analogues were synthesized by varying alcohol species through the common synthetic route (Figure 1) in the same manner and under the same conditions as previously reported for the synthesis of EtOSZ [24]. After the first reaction, the ROSiCl 3 was isolated by distillation. The purity of the distillated ROSiCl 3 was monitored by gas chromatography (GC) analysis, and the ROSiCl 3 fraction with the purity higher than 95% was collected. The recovery rate of each ROSiCl 3 was evaluated as the amount of the collected fraction relative to the total fraction. The yield after ammonolysis at second step was determined by measuring the weight of the reaction product, then the total yield of each ROSZ analogue was calculated from those values listed in Table 1.

Pyrolysis and Heat Treatment
The synthesized ROSZ was placed on an alumina tray and pyrolyzed in a quartz tube furnace under flowing N 2 (200 mL/min.) by heating from room temperature to 550-800 • C with a heating rate of 5 • C/min, maintaining the maximum temperature for an additional 1 h and finally furnace cooling down to room temperature to give a product as a solid. The ceramic yield and the appearance of the 800 • C-pyrolyzed samples were listed in Table 2.  29 Si solid state nuclear magnetic resonance (NMR) spectra for the as-synthesized and the pyrolyzed-ROSZ samples were acquired using magic angle spinning (MAS) with a rotation frequency of 15 kHz (Model ECA-400, JEOL, Tokyo, Japan) at room temperature. The resonance frequencies for the 13 C-and 29 Si-NMR spectra recorded in this study were 100 and 79.5 MHz, respectively. The chemical shifts of the peak signals in the 13 C-and 29 Si-NMR spectra were quoted relative to the signals of adamantine (29.5 pm) and 3-(trimethylsilyl) propionic acid sodium salt (2 ppm), respectively. The attenuated total reflection-infra red (ATR-IR) spectra were recorded on the as-synthesized and the pyrolyzed ROSZ with a diamond prism under an incidence angle of 45 • (Model Spectrum 100, Perkin Elmer, Waltham, MA, USA).
The thermal behaviors up to 1000 • C were studied by thermogravimetric (TG) analysis in N 2 with a heating rate of 10 • C/min (Model TG-DTA 6300, Hitachi High Technologies Ltd., Tokyo, Japan), and simultaneous TG-mass spectrometry (MS) analyses (Model STA7200, Hitachi High Technologies Ltd., Tokyo, Japan/Model JMS-Q1500 GC, JEOL, Tokyo, Japan). The measurements were performed under flowing helium (100 mL/min.) with a heating rate of 10 • C/min.
To examine the effect of the molecular size of alkenes formed in-situ on the porosity formation, molecular structure calculations were performed by Spartan'14 using the basis set of ωB97-XD/6-31G*.
Elemental analyses were performed on the pyrolyzed samples for oxygen, nitrogen, and hydrogen (inert-gas fusion method, Model EMGA-930, HORIBA, Ltd., Kyoto, Japan), and carbon (non-dispersive infrared method, Model CS844, LECO Co., St. Joseph, MI, USA). The silicon content in the samples was calculated as the difference of the sum of the measured C, N, O and H content to 100 wt %.
X-ray diffraction (XRD) measurements were performed on the pyrolyzed samples (Model X'pert Pro α1, Philips Ltd., Amsterdam, The Netherlands).
Textural properties of the pyrolyzed samples were measured by gas adsorption/desorption isotherms (Model Belsorp Max, BEL Japan Inc., Osaka, Japan) using argon (Ar) as a probe molecule with relative pressures ranging from 0 to 0.99. Specific surface area (SSA) was calculated from the isotherm data using the Brunauer-Emmett-Teller (BET) method. The micropores (R pore < 2.0 nm) were characterized by the Saito-Foley (SF) method [43].
To study the CO 2 affinity of the ROSZ-derived amorphous Si-O-N, TG analysis under CO 2 atmosphere was conducted (Model TG8120, Rigaku, Tokyo, Japan). Dry CO 2 was used for the runs and high purity Ar was used as the purging gas for sorbent regeneration. CO 2 adsorption-desorption measurements were conducted at 40 • C under continuous run Ar (40 min), CO 2 (40 min) then Ar (40 min), respectively, according to a published procedure [44,45]. CO 2 adsorption-desorption isotherms were also carried out at 0, 20 and 40 • C (Model Belsorp Max, BEL Japan Inc., Osaka, Japan). All samples were dried 1 h at 120 • C under vacuum to remove moisture and CO 2 adsorbed from air prior to the above characterization experiments. For each measurement, the adsorption and desorption were performed at a constant temperature.
Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) analyses were carried out by switching the inlet flow from inert Ar stream to the adsorbing CO 2 stream according to a published procedure [44,45]. Fourier transform infrared (FT-IR) spectra (Model spectrum TM100, Perkin Elmer Japan Co., Ltd., Tokyo, Japan) were recorded at 3 and 10 min and the maximum adsorptions were recorded after 10 min under CO 2 atmosphere. Upon saturation of the sorbent, the inlet stream was switched back to the inert Ar stream and FT-IR desorption spectra were recorded at the same time intervals that the adsorption step. Dry CO 2 was used for the runs and high purity Ar as the purging gas for sorbent regeneration.

ROSZ Preceramic Polymers
Typical ATR-IR spectra of the ROSZ are shown in Figure 2. All the sample polymers exhibited characteristic absorption bands at 3350 (broad), 2800-3000 and 1070 cm −1 attributed to νN-H, νC-H and δN-H that were involved in Si-NH-Si unit [46], respectively. To identify the chemical structure of the ROSZ in more detail, 13 C-and 29 Si-NMR spectroscopic analyses were performed in solid state. Typical results are shown in Figure 3. The 13 C-NMR spectrum of EtOSZ (a1) presented two sharp signals at 58.0 and 18.8 ppm, assigned to methylene (CH 2 ) unit, and terminate methyl (CH 3 ) unit in the ethoxy (OCH 2 CH 3 ) group, respectively. The corresponding 29 Si-NMR spectrum (b1) exhibited a strong single signal at −44.6 ppm that was assigned to SiO(NH) 3 unit. The weak signals at −53.6 and −61.4 ppm were thought to be attributed to the small amount of by-products (below 5%) that could not be removed by the distillation after the alkoxylation of SiCl 4 (Equation (1)). The signals at −53.6 and −61.4 ppm were assigned to (EtO) 2 -Si-(NH) 2 (linear or cyclic) and (EtO) 3 -Si-NH, respectively [24].
The 13 C-NMR spectrum of as-synthesized CyOSZ (a2) showed three sharp signals at 26. This complicated spectrum was assumed to be due to the co-existence of the cis and trans isomers in the starting decahydro-2-naphtol used in this study. On the other hand, the corresponding 29 Si-NMR spectra exhibited almost same feature with a strong single signal at −44.6 ppm assigned to SiO(NH) 3 unit (b2 and b3).

Thermal Conversion to Inorganic Compound
As shown in Figure 4, ROSZ samples showed a similar TG curve with the following two weight loss regions: a slight weight loss of approximately 4.5% up to 200 • C, which could be due to the residual solvent, and a main weight loss at around 300 to 600 • C. The main weight loss increased consistently with the molecular weight of the alkyl group (R) of the OR group in ROSZ, and the ceramic yield evaluated for the 800 • C-pyrolyzed samples followed this relationship (Table 2).  29 Si-NMR spectra of as-synthesized ROSZs. (a1,b1) EtOSZ, (a2,b2) CyOSZ, and (a3,b3) DHNpOSZ.
Chemical compositions of the 800 • C-pyrolyzed ROSZ samples were listed in Table 3. As a reference data, the theoretical compositions of the as-synthesized ROSZ polymers were also listed in this table. In spite of the pyrolysis under inert atmosphere of N 2 , the carbon content remarkably decreased, and the resulting C/Si atomic ratios of the 800 • C-pyrolyzed ROSZ samples were in the range of 0 to 0.08. The nitrogen content also decreased to some extent, and the resulting N/Si atomic ratios were 0.5 to 0.6, while the O/Si atomic ratios were 0.8 to 1.1, close to that of the ideal ROSZ (1.0). Then, TG-MS analysis was performed on the as-synthesized ROSZ samples under He atmosphere. The results were summarized and shown in Figures 5 and 6.   The TG-curves measured in He were quite similar to those in N 2 (Figure 4), and the gaseous species formed in-situ were mainly detected during the dominant weight loss region at 300 to 600 • C, however, compared with the ROSZs having primary alkyl group (Figure 5a), those with secondary alkyl group showed a narrow TICC signal and the gas evolution completed up to 550 • C ( Figure 5b).
As previously reported [24], the gaseous species, m/z ratios at 45 and 16 detected for EtOSZ ( Figure 6a) were assigned to SiNH 3 + and NH 2 + , respectively. These fragment ions could be due to the partial decomposition of the silsesquiazane linkage. This led to the lower nitrogen content observed for the 800 • C-pyrolysed EtOSZ. The m/z ratio at 28 was assigned to ethylene (CH 2 = CH 2 + ) formed in-situ by the C-O bond cleavage of the ethoxy group. As shown in Figure 6, the m/z ratios at 16 (NH 2 + ) and/or 15 (NH + ) were detected for other ROSZ analogues, and the partial decomposition of the silsesquiazane linkage could proceed to some extent in all the ROSZ samples synthesized in this study. Together with these species, gaseous species detected at various m/z ratios are listed in Table 4. As we expected, other m/z ratios detected for the ROSZ analogues with secondary alkyl group (R = iPr, sBu, sHex, Cy and DHNp) were basically assigned as the fragment ions of the corresponding alkenes (C(R 1 ) = CH-R 2+ ) [47] shown in Equation (1).
In the case of ROSZ analogues with primary alkyl group (R = nPr, nBu, nHex), this C-O bond cleavage could proceed in association with some Si-N bond cleavages. The unknown species such as m/z ratios at 67, 79 and 91 were estimated as oligomers originated from the silsesquiazane linkage.

Textural Properties of ROSZ-Derived Amorphous Si-O-N
To study the porous structure formation, textural properties of the ROSZ-derived amorphous Si-O-N samples were characterized by measuring gas adsorption-desorption isotherm. Based on the results of TG and TG-MS analyses, the following three kinds of ROSZ analogues with R = Et, Cy and DHNp were selected. Then, the porosity formation behavior was studied by evaluating the Brunauer-Emmet-Teller (BET) specific surface area (SSA) of the ROSZ-derived Si-O-N samples pyrolyzed at various temperatures from 400 to 800 • C. XRD analysis of the 800 • C-pyrolyzed samples resulted in the detection of a typical amorphous diffraction line, and as shown Figure 7, the SSA of all the X-ray amorphous samples drastically increased above 400 • C to reach its maximum at 550 • C, then decreased consistently with the pyrolysis temperature. The immediate increase in the SSA at 400 to 550 • C was well consistent with the gas evolution behavior of ROSZs. Then, the resulting porous structure was examined for each sample just after the pyrolysis at the gas evolution complete temperature identified and shown in Figure 5, 600 • C for EtOSZ and 550 • C for CyOSZ and DHNpOSZ. The SSAs of the 600 • C-pyrolyzed EtOSZ, 550 • C-pyrolyzed CyOSZ and DHNpOSZ were measured to be 476, 601 and 750 m 2 ·g −1 , respectively (Table 5). These three samples exhibited a typical type I isotherm ( Figure 8) and a unimodal pore size distribution (PSD) curve (Figure 9). The peak top of the PSD curve located in the micropore size range, and the micropore volume evaluated for these samples were listed in Table 5.   To study the effect of the molecular size of gaseous species formed in-situ on the porosity formation, molecular structure calculations were performed on the alkenes detected as a main gaseous component. The sizes obtained for ethylene, cyclohexene and octahydronaphthalene were 0.55, 0.74 and 0.97 nm, respectively. Then, the values listed in Table 5 were plotted as a function of the estimated molecular size of alkene ( Figure 10). It was clarified that the SSA and micropore volume linearly increased with the molecular size of alkene. The peak top of the PSD curve also exhibited a similar correlation with the molecular size.

CO 2 Capture Properties
Generally, CO 2 capture capacity strongly depends on the SSA of solid sorbent [15,44,45] and, in this study, evaluation was performed on the three samples; 550 • C-pyrolyzed DHNpOSZ having the highest SSA together with 600 • C-and 800 • C-pyrolyzed EtOSZ samples.
In order to determine the nature of the interactions occurring at the surface of the present ROSZ-derived amorphous Si-O-N samples, CO 2 adsorption/desorption isotherms were carried out at 0, 20 and 40 • C.
As a result, at 20 and 40 • C, the 600 • C-pyrolyzed EtOSZ (Et(600)) exhibited the highest CO 2 capture capacity among the present three samples.  At 20 • C and 1 bar, compared with conventional Zeolite 13X (3.9 mmol·g −1 ) [30,31], MOF (SIFSIX-2-Cu-i) [35] and another AC (treated under mild condition using KOH) [33] exhibited higher capture capacities of 5.4 and 4.8 mmol·g −1 , respectively. On the other hand, the CO 2 capture capacity at 20 • C of the present ROSZ-derived amorphous Si-O-N was limited to be 2.0 mmol·g −1 at SSA = 476 m 2 ·g −1 (Et(600)); however, this capacity value was found to be compatible with that of some other solid sorbents having much larger SSA, for example, AC (2.1 mmol·g −1 at SSA = 2994 m 2 ·g −1 ) [34] and MOF (SIFSIX-2-Cu) (1.8 mmol·g −1 at SSA = 3140 m 2 ·g −1 ) [35]. Compared with polymer-derived amorphous silica (SiO 2 ) evaluated in our previous study [44,45], present EtOSZ-derived Si-O-N samples also showed higher CO 2 adsorption capacities in spite of their lower SSAs. This seemed to be due to the capillary condensation effect derived from the microporisity of the present Si-O-N samples as previously suggested for the amorphous CN and Si-N by Zhao et al. [42] and Schitco et al. [15], respectively. However, as shown in Figure 11, such an immediate CO 2 uptake was not clearly observed at the very low CO 2 partial pressure region. Then, to investigate the CO 2 capture properties of the present Si-O-N in more detail, CO 2 adsorption/desorption behaviors were in-situ monitored by measuring TG curves under continuous run Ar-CO 2 -Ar at 40 • C. Figure 13 shows the samples weight gain under CO 2 flow, expressed in weight percentage. A sharp weight gain was observed for all the samples. The weight gain due to the CO 2 up-take was in the order of 600 • C-pyrolyzed EtOSZ, 550 • C-pyrolyzed DHNpOSZ and 800 • C-pyrolyzed EtOSZ, which was consistent with the result obtained by the CO 2 sorption isotherm shown in Figure 12. In an attempt to regenerate the samples at the same temperature under Ar flow, 800 • C-pyrolyzed EtOSZ desorbed almost all its captured CO 2 , demonstrated by the total decrease in the weight gain. This desorption behavior was similar to that of SiO 2 which had no chemical interaction towards CO 2 [44,45]. On the other hand, the regeneration under Ar flow of the 550 • C-pyrolyzed DHNpOSZ and 600 • C-pyrolyzed EtOSZ resulted in leaving 0.43 and 0.2 wt % of the adsorbed CO 2, respectively, which suggesting the existence of chemical interaction toward CO 2 .
The nature of the interaction was further investigated for each sample using diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS). Figure 14a-c show the IR spectra of the three samples before and after CO 2 adsorption, then after a regeneration step under Ar flow. In this study, the maximum adsorption rates were recorded after 10 min of CO 2 exposure for the three samples. Flowing CO 2 over 550 • C-pyrolyzed DHNpOSZ sample (Figure 14a) produced an increase in characteristic IR bands associated with the formation of proposed species shown in Figure 14d [48]. The increased intensities in the broad absorption bands at 1633 and 1597 cm −1 were assigned to bridged carbonate and bidentate carbonate (shown in Figure 14d), respectively. Flowing CO 2 on 600 • C-pyrolyzed EtOSZ for 10 min also resulted in the appearance of the increased intensities in the two broad absorption bands at 1633 and 1595 cm −1 assigned to bridged carbonate and bidentate carbonate, respectively (Figure 14b). These samples kept the increased intensities after regeneration under Ar flow for 10 min. This was consistent with the TG analysis which showed a remaining amount of CO 2 that could not be removed at ambient conditions. In the case of 800 • C-pyrolyzed EtOSZ, subsequent regeneration under Ar flow resulted in the further increase in the broad absorption band intensity at 1633 cm −1 assigned to bridged carbonate (Figure 14c). Figure 15 shows the 29 Si-NMR spectrum of 800 • C-pyrolyzed EtOSZ measured in our previous study [24]. One broad peak at around −100 ppm was deconvoluted to three peaks centered at −110, −100, and −90 ppm that were assigned to SiO 4 , HO-SiO 3 (Q3), and SiO 3 N, respectively. Accordingly, the ROSZ-derived Si-O-N samples in this study could have surface OH groups.
At very low CO 2 partial pressures, the surface OH groups provided CO 2 chemisorption sites forming carbonates shown in Figure 14d, which could lead to enhancing CO 2 physisorption at higher partial pressures as previously discussed for the enhanced CO 2 capture capacity of the polycarbosilane-derived amine-functionalized Si-C-H materials [45]. The contribution of the CO 2 chemisorption-i.e., the number of OH groups/surface area of the 550 • C-pyrolyzed DHNpOSZ-was thought to be smaller than those of two other samples. This resulted in the very high CO 2 capture capacity at 0 • C, while the temperature dependence became more pronounced, and the CO 2 capture capacity drastically decreased at 20 to 40 • C.

Summary
In this study, novel polyalkoxysilsesquiazanes were synthesized. The microporosity formation behavior during the polymer/ceramics thermal conversion and the CO 2 capture properties of the resulting X-ray amorphous microporous Si-O-N materials were investigated. The results can be summarized as follows: (1) ATR-IR, 13 C-and 29 Si-NMR spectroscopic analyses revealed that a series of polyalkoxysilsesquiazanes ([ROSi(NH) 1.5 ] n , ROSZ, R = Et, nPr, iPr, nBu, sBu, nHex, sHex, cHex, DHNp) were successfully synthesized in a good yield via two simple steps: reaction of SiCl 4 with ROH to afford ROSiCl 3 , followed by ammonolysis at −78 • C. (2) The simultaneous TG-MS analyses of the ROSZs under a He flow revealed that the cleavage of the oxygen-carbon bond of the RO group was a common and dominant decomposition reaction.
The subsequent evolution of alkene as a main gaseous species formed in-situ lead to the formation at 550 to 800 • C of the X-ray amorphous microporous Si-O-N. (3) The peak of the pore size distribution curve located within the micropore size range, and the total micropore volume as well as the SSA of the resulting X-ray amorphous Si-O-N increased consistently with the molecular size estimated for the alkene formed in-situ during the pyrolysis. (4) The CO 2 capture capacity at 0 • C of the Si-O-N material increased consistently with the SSA, and reached 3.9 mmol·g −1 at the SSA of 750 m 2 ·g −1 , which was achieved for the 550 • C-pyrolyzed DHNpOSZ. (5) DRIFTS monitoring of the CO 2 adsorption under continuous Ar-CO 2 -Ar run at 40 • C revealed carbonate formations at the CO 2 chemisorption site provided by the surface OH groups formed in-situ during the polymer/ceramics thermal conversion at around 550 • C. (6) The CO 2 capture capacity measured at the CO 2 partial pressure, p/p 0 = 1 could be enhanced by the initial CO 2 chemisorption at very low CO 2 partial pressures.