Introduction

Organo-bridged alkoxysilane, [(RO)3Si]2R’, is an important silsesquioxane precursor because the functionality of the resulting polymers can be easily controlled by changing the physical and chemical properties of the bridging groups. These precursors are used as mesoporous materials [1], separation membranes [2, 3], flexible aerogels [4], adsorbents [5], photo-induced materials [6], and so on. These materials are synthesized using colloids and particles, which are prepared by the sol–gel reaction of organo-bridged alkoxysilanes. Their gelation time and structure can be changed by changing the structure of bridging groups [7, 8]. Low-molecular-weight compounds (dimers and trimers) have also been synthesized and isolated from hydrolysis–condensation reaction of organo-bridged alkoxysilane [9, 10]. The compounds, sols (colloids and hydrolyzed alkoxysilanes), and gels (particles and insoluble polymers) obtained using organo-bridged alkoxysilane have been well investigated. However, their soluble polymers are not understood completely because of their difficult preparation and isolation.

Yamamoto et al. synthesized soluble ethane-bridged silsesquioxane polymer (sEBSP) by the hydrolysis–condensation reaction of bis(triethoxysilyl)ethane (BTESE) under nitrogen flow [11]. Herein, we focus on bis(trimethoxysilyl)ethane (BTMSE) because the molecular weight of its polymers and the colloidal size can be controlled by controlling the hydrolysis rate of alkoxy groups [12,13,14]. However, the structural units of sEBSP have not been identified yet. Although the use of organic–inorganic composites containing silsesquioxanes as sols (colloids, particles) [15,16,17], polyhedral oligomeric silsesquioxanes (POSS) [18,19,20], and cyclic silsesquioxanes [21] has been investigated, not much information is available on organic–inorganic composites fabricated using soluble silsesquioxane polymers.

In this work, we investigate the preparation and characterization of sEBSP. We also prepared an organic–inorganic composite comprising sEBSP and poly(methyl methacrylate) (PMMA) and poly(bisphenol A-co-epichlorohydrin) (PBE). The sEBSP was characterized by nuclear magnetic resonance (NMR), Fourier-transform infrared (FTIR) spectroscopy, and gel permeation chromatography (GPC), while the organic polymer–sEBSP composite was characterized by FTIR spectroscopy and thermal investigation.

Experimental

Measurements

NMR spectra were recorded using a JEOL Resonance JNM-ECP 300 spectrometer (JEOL, Tokyo, Japan; 1H at 300.53 MHz and 29Si{1H} at 59.70 MHz). The chemical shifts were reported in ppm using chloroform-d (CDCl3; for 1H: 7.26 ppm in residual chloroform) and tetramethylsilane as internal standards (for 29Si{1H}: 0.00 ppm). To obtain the 29Si{1H} NMR spectra, Cr(acac)3 was added to the samples as a paramagnetic relaxation agent. The broadening factor (BF) was 20 Hz. GPC was performed using an HPLC system (LC-20AD, Shimadzu, Kyoto, Japan) attached to a PLgel 5 µm Mixed-D column. Tetrahydrofuran (THF) was used as the eluent (1 mL min−1), and RID-20A was used as the detector at 40 °C. The molecular weight was calculated using polystyrene standards. The FTIR spectra of sEBSP were recorded on an FTIR spectrophotometer (FT/IR-6100, JASCO, Tokyo, Japan) with a neat sample sandwiched between KBr aperture plates. The FTIR spectra of the composites were recorded using the same spectrophotometer with attenuated total reflectance (JASCO ATR PRO 0450-S, ZnSe prism). Thermogravimetric analysis (TGA) was performed using a TG–DTA analyzer (2000SE, Netzsch Japan, Kanagawa, Japan). The samples were heated to 1000 °C at a rate of 10 °C min−1 under air flow. Differential scanning calorimetry (DSC) was performed using DSC 3500 Sirius (Netzsch Japan). Both heating and cooling were performed at a rate of 5 °C min−1 under nitrogen flow. This process was performed three times.

Materials

Methanol and THF were purified by standard methods and stored over activated molecular sieves. BTMSE was purchased from Tokyo Chemical Industry Ltd. (Tokyo, Japan). A 6 mol L−1 HCl aq solution was purchased from FUJIFILM Wako Pure Chemical Corp. (Osaka, Japan). PMMA (Mw = 997,000 g mol−1) and PBE (Mw = 40,000 g mol−1) were purchased from Sigma-Aldrich (Tokyo, Japan). Water was purified using the PURELAB Flex 5 water purification system (ELGA, Marlow, UK).

Hydrolysis–condensation of BTMSE

BTMSE (1.35 g, 5.0 mmol) and methanol (1.61 g) were placed in a 100-mL four-necked flask equipped with nitrogen inlet and outlet tubes and a mechanical stirrer. The mixture was stirred at 150 rpm and cooled in an ice bath for 10 min. Then, 6 mol L−1 HCl aq solution (0.96 g, HCl/BTMSE = 0.05) and H2O was added. The resulting mixture was stirred in the ice bath for 10 min and then at 22 ± 3 °C for 10 min, followed by heating at 70 °C for 3 h. The hydrolysis–condensation reaction was carried out under a nitrogen flow of 360 mL min−1. The sEBSP was obtained as a colorless and viscous liquid.

Preparation of free-standing films

First, 0.12 g of organic polymer (PMMA or PBE) in THF (5 mL) was added to the sEBSP (13 mg) and stirred for 24 h at 22 ± 3 °C. The precursor was poured into a 50-mmϕ Teflon Petri dish followed by drying at 22 ± 3 °C for 1 day and curing at 120 °C for 1 day.

Results and discussion

Preparation and characterization of sEBSP

The sEBSP was prepared by the hydrolysis–condensation of BTMSE catalyzed by hydrochloric acid. The hydrolysis–condensation reaction was carried out under a nitrogen gas flow to remove the catalyst [13, 22, 23] and to accelerate the condensation rate by evapolating the solvent [22, 24]. The results of the hydrolysis–condensation of BTMSE are shown in Table 1. The yield of sEBSP and the content of methoxy group in sEBSP (determined by 1H NMR spectra) decreased when the amount of water is increased because the hydrolysis ratio depends on the amount of water used. Compared with BTESE, BTMSE was easily hydrolyzed and formed a gel (gelation point: H2O/BTMSE > 3.8, H2O/BTESE > 5.0 [11]). The sEBSP showed high solubility in acetone, THF, chloroform, and toluene and only a slight solubility in methanol, ethanol, and 2-propanol, similar to the BTESE polymer [11]. GPC analyses of sEBSP are summarized in Table 1 and Fig. 1. The number average molecular weight (Mn), weight average molecular weight (Mw), and polydispersity index (PDI) of sEBSP increased with the increasing ratio of H2O/BTMSE because the polycondensation rate was dependent on the hydrolysis ratio in our process [13, 24]. The PDI value was very large.

Table 1 Results of the hydrolysis–condensation of BTMSE
Full size table
Fig. 1
figure 1

GPC charts of sEBSP

Full size image

The 29Si{1H} NMR spectra of BTMSE and sEBSP are shown in Fig. 2. The 29Si{1H} spectrum of BTMSE showed a signal assigned to the T0 unit at − 42.2 ppm in CDCl3. The sEBSP contains no unreacted BTMSE, hydrolyzed monomers, [CH2Si(OMe)3−x(OH)x]2 (x = 0–3), or an intramolecularly cyclized monomer ([CH2Si(OMe)2]2O) because no signals were observed in the region from − 37 to − 42.2 ppm. Moreover, no signals of the T3 unit ([CH2Si(OSi)3]) were observed from − 64.5 to − 66.6 ppm [25,26,27]. For bis(triethoxysilyl)ethane, bicyclic T1–OEt and T2–OEt appeared at − 51 and − 53.3 ppm, respectively [28]. Because the chemical shift of Si–OMe is 2–4 ppm smaller than that of Si–OEt [26, 29], we assigned the signals at − 47.6 and − 51.4 ppm to bicyclic T1–OMe and T2–OMe, respectively. Loy et al. reported that a xerogel with a bicyclic structure showed a signal at − 60.5 ppm [8, 10]. The sEBSP exhibited a signal at − 60.6 ppm, which corresponds to the bicyclic T3 unit. The cyclic T2 unit appeared downfield compared with the branched T2 unit [30, 31]. Therefore, the signals at − 56.7 and − 54.4 ppm can be assigned to branched T2–OMe and cyclic T2–OMe, respectively. The signal at − 51.8 ppm is due to linear T1–OMe, and that at − 50.4 ppm might be due to cyclic T1–OMe. These assignments are summarized in Table 2. The end-capping-sEBSP (cap-sEBSP), prepared by the reaction of sEBSP (H2O/BTMSE = 3.4) with Me3SiCl, was measured to identify the silanol group (Tn–OH). The 29Si{1H} NMR spectrum of cap-sEBSP showed a new signal at 10.1 ppm. However, this signal intensity was very weak, implying that the amount of silanol is low in sEBSP. The structural units of sEBSP are illustrated in Scheme 1. The sEBSP has two types of cyclic units: I (number of Si ≥ 4) and II (number of Si = 3). However, the exact number of the units could not be determined.

Fig. 2
figure 2

29Si{1H} NMR spectra of BTMSE and sEBSP in CDCl3

Full size image
Table 2 Assignments for 29Si{1H} NMR spectra of BTMSE and sEBSP
Full size table
Scheme 1
scheme 1

Structural units in sEBSP through hydrolysis–condensation of BTMSE

Full size image

The FTIR spectra of BTMSE and sEBSP were measured for the sample sandwiched between KBr aperture plates (Fig. 3). Their vibration wavenumbers are assigned based on the vibrational assignments of silsesquioxanes [10, 32,33,34,35,36,37,38,39,40,41] and are summarized in Table 3. The sEBSP showed absorption bands formed by stretching vibration of siloxane bonding (Si–O–Si) at 1000–1100 cm−1. These bands overlapped with three absorption bands. In general, branched, cyclic, and linear Si–O–Si appear at 1080, 1050, and 1020 cm−1, respectively [36,37,38]. Moreover, bicyclic Si–O–Si exhibits a band at 1007 cm−1 [10]. The FTIR results show that the sEBSP was composed of a branched siloxane unit (at 1087 cm−1), a cyclic siloxane unit (at 1049 cm−1), and a linear/bicyclic siloxane unit (at 1011 cm−1). No difference was observed in the molar ratio of H2O/BTMSE for FTIR, which implies that the skeleton structure will be similar. The sEBSP has more methoxy groups than the ethane-bridged silsesquioxane colloid and gel [32, 34, 38]; hence, sEBSP can be isolated and obtained as a soluble polymer. The sEBSP showed a peak at 3410 cm−1 for silanol (Si–OH) and a shoulder peak at 913 cm−1. In contrast, the cap-sEBSP showed a decreasing intensity for the silanol band, which implies that it has no silanols.

Fig. 3
figure 3

FTIR spectra of BTMSE, sEBSP, and cap-sEBSP

Full size image
Table 3 Assignments for FTIR spectra of BTMSE and sEBSP
Full size table

Hence, the results suggest that (i) the sEBSP contained methoxy and silanol groups and (ii) its structure contained hybridized linear, branched, cyclic, and bicyclic siloxane units. Scheme 2 illustrates the hydrolytic condensation reaction of BTMSE in the initial stage. At least six types of structural units (linear, branched, cyclic I (SiOC and SiO rings), cyclic II, and bicyclic-type) were formed by cyclization, bicyclization, and crosslinking condensation from the first-stage unit. Therefore, the PDI of the sEBSP increase because of the complicated condensation reaction.

Scheme 2
scheme 2

Possible formation of ethane-bridged silsesquioxane by hydrolysis–condensation in the first stage

Full size image

FTIR studies of organic polymer–sEBSP

Organic polymer–sEBSP was prepared by a solution–melt blending method. The free-standing films were characterized by FTIR spectra, as shown in Fig. 4. The spectrum of pure PMMA exhibited absorption bands due to νC = O at 1722 cm−1, δCH at 1386 cm−1, νC–O–C at 1142 cm−1, and skeletal mode of C–C at 749 cm−1 [42, 43]. The absorption bands assigned to the PMMA matrix of PMMA–cap-sEBSP correspond to those of pristine PMMA. In comparison, the FTIR of PMMA–sEBSP showed that the vibration wavenumbers of C = O (at 1728 cm−1) and C–O–C (at 1148 cm−1) shifted 6 cm−1 higher than that of pristine PMMA. The vibration wavenumbers of C–H and skeletal mode had the same values. These results show that the sEBSP will affect the carbonyl group of PMMA. In general, the vibration wavenumber of the hydrogen bond between the carbonyl group and silanol (νC = O∙∙∙HO–Si) is lower than that of the carbonyl group that forms a non-hydrogen bond [21, 44]. However, the carbonyl group of PMMA–sEBSP showed a higher wavenumber, which implies the absence of C = O∙∙∙HO–Si. To investigate the hydrogen bonding mode of carboxylic acid, the vibration wavenumber of the carboxyl group that forms hydrogen bonds with the hydroxy group (CH3O–H∙∙∙O(H)–C = O) was calculated and was observed to upshift (to 11 cm−1), compared with the carboxyl group that forms nonhydrogen bonds [45]. Based on these results, we suggest that hydrogen bonds (SiO–H∙∙∙O(Me)–C = O) are formed between PMMA and sEBSP, as shown in Fig. 5.

Fig. 4
figure 4

FTIR spectra of PMMA–sEBSP and PBE–sEBSP

Full size image
Fig. 5
figure 5

Proposed hydrogen bonding interaction in PMMA–sEBSP and PBE–sEBSP

Full size image

The FTIR spectrum of pure PBE exhibited absorption bands due to νC = C–H at 1506 cm−1, νC–O–H at 1413 cm−1, δCH2 at 1361 cm−1, and νC–C–O–C at 1228 cm−1 [46, 47]. The absorption bands due to the PBE matrix for PBE–cap-sEBSP corresponded to those of pristine PBE. In contrast, νC–C–O–C of PBE–sEBSP shifted to a higher wavenumber (1241 cm−1) than that of pristine PBE. Moreover, νC–O–H of PBE–sEBSP was observed at the same wavenumber as that of pristine PBE, which indicates the absence of any hydrogen bond with the hydroxy group. These results show that the PBE–sEBSP would form a hydrogen bond between silanol and the oxygen atom of the phenoxy ether moiety in PBE, as illustrated in Fig. 5. Transalkoxylation between PBE and sEBSP can be confirmed by the vibration stretching of Si–O–alkyl: νSi–O(–Me) at 699 cm−1 and νSi–O(–iPr) at 686 cm−1 [41]. Although νSi–O(–Me) was observed at 697 cm−1, νSi–O(–iPr) was not observed; therefore, no transalkoxylation between PBE and sEBSP occurred.

Thermal analyses of organic polymer–sEBSP

The TGA of organic polymer–sEBSP under air is shown in Fig. 6. The temperatures for 5% weight loss (Td5) are summarized in Table 4. Degradation of pure PMMA occurred through two steps: (i) scission of head-to-head linkages [48] or degradation by radical transfer to the vinyl chain end [49] at 150–200 °C, and (ii) cleavage of the main chain and oxidative degradation at 270 °C [50]. The PMMA–sEBSP was more thermally stable than pristine PMMA, probably because of the radical scavenging effect caused by groups such as metal oxides [51] or the restriction of the main chain [52]. This behavior was similar to that of PMMA–POSS [18], PMMA–organosilsesquioxane oligomer [21], and PMMA–titanium phosphonate cluster composites [53]. The Td5 value of the PBE–sEBSP was lower than that of pristine PBE. This tendency was similar to that of PBE–titanium phosphonate cluster hybrids [46, 54]. It is possible to eliminate methanol while forming Si–O–iCH(CH2)2 by transalkoxylation of sEBSP and PBE with heating.

Fig. 6
figure 6

TGA traces of PMMA–sEBSP and PBE–sEBSP under air

Full size image
Table 4 Td5 and Tg of PMMA–sEBSP and PBE–sEBSP
Full size table

The miscibility of organic polymers and sEBSP was confirmed by DSC analyses. If these composites are immiscible, the glass transition temperature (Tg) values will be similar to those of pristine organic polymers [55]. Moreover, the Tg values of organic–inorganic hybrids/composites depend on the type of the formation system. The following types of formation systems are known: (i) composites, in which organic and inorganic domains are combined over long lengths (at least at micrometer level); (ii) nanocomposites, in which composites are combined over the nanometer order; and (iii) hybrids, characterized by weak interactions and strong bond formation (from atomic level to 50 nm) [56]. For PMMA–inorganic domains, the Tg values decrease in the order of nanocomposites < composites ≈ pristine PMMA < hybrids [57,58,59]. PBE–inorganic domains showed a similar tendency [60, 61]. The DSC analyses of PMMA–sEBSP and PBE–sEBSP are shown in Fig. 7, and the Tg values are summarized in Table 4. The Tg values of pure PMMA and PBE were 115 and 98 °C, respectively. The Tg values of PMMA–cap-sEBSP and PMMA–sEBSP were slightly higher than those of pristine PMMA. In contrast, the Tg values of PBE–cap-sEBSP and PBE–sEBSP were lower than those of pristine PBE. Therefore, sEBSP was miscible in organic polymers, and PMMA–sEBSP and PBE–sEBSP were classified as a hybrid and nanocomposite, respectively. Hence, the sEBSP and cap-sEBSP exerted similar thermally stable effects on organic polymer matrices. No difference was observed in the thermal effect between sEBSP and cap-sEBSP because of the low content of silanol group in sEBSP.

Fig. 7
figure 7

DSC traces of PMMA–sEBSP and PBE–sEBSP

Full size image

Conclusions

The sEBSP was prepared by a hydrolysis–condensation reaction of BTMSE and showed solubility in organic solvents (acetone, THF, chloroform, and toluene). Its methoxy group content decreased, while its Mw and PDI decreased when the amount of water was increased. When H2O/BTMSE was 3.8, the ethane-bridged silsesquioxane was obtained as a gel. The structure of sEBSP was determined by NMR and FTIR. The sEBSP contained hydroxy and methoxy groups on the silicon atoms and was composed of linear, branched, cyclic, and bicyclic units.

The organic polymer–sEBSP composites were prepared by solution–melt blending of the organic polymer with sEBSP. The FTIR results showed that the PMMA–sEBSP can be formed by hydrogen bonding with the methoxy group in the PMMA matrix (SiO–H∙∙∙O(Me)–C = O). The PBE–sEBSP was formed by hydrogen bonding with the phenoxy moiety of PBE (SiO–H∙∙∙O(R)–C6H4). In contrast, the organic polymer–cap-sEBSP (with no silanol) was observed without any hydrogen bonding interaction. The Td5 values of PMMA–sEBSP and PMMA–cap-sEBSP increased in contrast to that of pristine PMMA. The PBE–sEBSP and PBE–cap-sEBSP had lower Td5 values than those of pure PBE. We proposed that transalkoxylation between the hydroxy group of PBE and the methoxy group of sEBSP involved the elimination of methanol. The sEBSP and cap-sEBSP exerted a similar thermally stable effect on organic polymers. The Tg value of PMMA–sEBSP was slightly higher than that of pristine PMMA, while that of PBE–sEBSP was lower than that of pristine PBE. The sEBSP showed miscibility in organic polymers. PMMA–sEBSP and PBE–sEBSP were classified as a hybrid and nanocomposite, respectively.