1 Introduction

Polysilsesquioxanes (PSQs) are well-known silicon-based materials characterized by the repeating [R-SiO1.5]n structural unit, where R could be either a polar or nonpolar, reactive or non-reactive organic functional group [1,2,3,4,5]. Moreover, they are characterized by a variety of structural patterns, which make PSQs a very versatile family of organic–inorganic hybrid materials [6,7,8,9]. Three main architectures are generally reported, i.e. random branched, polyhedral oligomeric silsesquioxanes (POSS), and ladder-like polysilsesquioxanes (LPSQs) [10]. A wide variety of POSS cages (RSiO1.5)n, with n ranging from 6 to 18 have been synthesized, with various R groups [3, 11,12,13] and many of them are available for purchase nowadays.

LPSQs are characterized by an inorganic double-stranded backbone. They have rapidly become interesting competitors of POSS and linear polysiloxane, as blend components or fillers in nanocomposites (NCs), due to the tunable aspect ratio of the backbone and the various organic side-chains, which impart interesting physical properties [14], such as good thermal stability [15,16,17], chemical resistance to acids, excellent mechanical properties (e.g. high modulus, hardness) [14, 18, 19], and optical transparency [20]. Due to the large choice of R groups, LPSQs show high solubility in different solvents [21, 22], thus favouring their use in many applications. With the development of less time consuming synthesis methods that allow LPSQs production in good yield, they have been used in blends to increase thermal resistance and mechanical properties of organic polymers, such as polylactic acid [17, 23,24,25] or polyimide [26, 27], and to produce membranes for gas separation, hard coatings, self-healing and photosensitive materials, electrolytes in Li batteries and triboelectric generators [6, 15, 18, 20, 22, 28,29,30].

LPSQs can be synthesized by hydrolysis-condensation of organosilanes under acid or basic conditions, and their chemical-physical properties such as molecular weight, degree of structural order and crystallinity strongly depend on both synthesis conditions and nature of the organic group linked to silicon [10]. Phenyl-silsesquioxanes with different degrees of ordering were produced by varying time and temperature of aging, and using different templating agents, and their thermal stability was correlated with the amount of uncondensed silanols;[31] highly regular ladder polyphenylsilsesquioxanes with high molecular weight were prepared by Zhang et al. [32] using template alkoxysilane monomers in a three-step approach, concluded by an end-capping reaction.

Lee et al. proposed an easy synthetic method using K2CO3 as a mild base catalyst in water/THF mixture to obtain at room temperature either Phenyl-POSS or ladder-like polyphenylsilsesquioxane depending on the initial concentrations of phenyltrimethoxysilane [33,34,35]. Different photo-curable ladder polysilsesquioxanes were obtained [34] varying the ratio of rigid and flexible organic side-groups, namely phenyl and propylmethacrylate groups, obtaining structures with variable molecular weight. Later, the same approach was applied to the synthesis of ladder-like poly(methacryloxypropyl)silsesquioxane (LPMASQ) by hydrolysis-condensation of 3-methacryloxypropyl trimethoxysilane (MPTMS) [7, 36], leading to amorphous fully-condensed, medium-degree order ladder structures with MW ~ 28000 Da.

We have previously reported on the use of LPMASQ as active molecular fillers in the preparation of polybutadiene-based NCs for improving polybutadiene (PB) thermal and mechanical behaviour [37]. LPMASQ was prepared by reacting MPTMS under basic conditions in a one-pot procedure following the approach of Lee et al. [36] and its structural properties were confirmed by infrared and nuclear magnetic resonance spectroscopies. Nanocomposites were prepared by solvent-casting and photo-induced polymerization and we analysed in detail the structural organization of ladders in the elastomeric matrix. Increasing the amount of filler in the PB matrix, polymerization of methacrylate groups leads to gradual assembly of LPMASQ units and produced peculiar dielectric properties along with enhanced thermal and mechanical stability of the final nanocomposites [37].

Exploiting the dispersibility of oxide nanoparticles (NPs) in LPMASQ, it is possible to add new features or further vary the physical properties of the NCs. This approach was recently exploited by our group to modify the surface wettability of PB/LPMASQ NCs obtained by photopolymerization [38] through the addition of MPTMS-functionalized silica nanoparticles. A similar approach was used to produce PB/LPMASQ NCs with active-passive O2-scavenging ability by adding bare and MPTMS-modified TiO2 nanoparticles [39]. Nanocomposites produced with modified TiO2 showed an oxygen uptake rate much higher than that obtained with bare TiO2. This result was discussed based on different mechanistic aspects, namely the production of singlet oxygen favoured by energy transfer mechanisms on the surface of functionalized titania.

In such studies, while a clear assessment of the effective role of LPMASQ for the achievement of peculiar functional properties was provided, the structural modifications of these units as a consequence of the interactions with filler particles was not investigated in detail. As concerns the use of titania NPs, besides imparting peculiar features to the composites, they may in principle influence the LPMASQ photocuring reaction, as recently pointed out for other acrylic and methacrylic systems, [40 and reference therein] where their capability to act as photoinitiators, generating reactive radicals under photoexcitation, has been exploited to produce self-curing composites. However, the possible structural changes derived from photopolymerization of LPMASQ UV-curable ladder-like units in the presence of TiO2 have not been explored so far in the literature.

In this scenario, the present work aims at unveiling the structural effects imparted by titania addition to LPMASQ in NCs obtained by photopolymerization. In detail, LPMASQ and commercial TiO2 NPs, either naked or surface-functionalized with MPTMS were employed as a suitable matrix and filler, respectively, for producing a NC by a simple and rapid solvent-casting technique followed by UV curing. A comprehensive structural investigation of the TiO2 modified LPMASQ composites was performed by X-Ray Diffraction (XRD), attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy and solid state nuclear magnetic resonance (NMR) spectroscopy, focusing on the effects of filler incorporation on the molecular organization. The results, besides evidencing a hindering effect of TiO2 NPs on the curing efficacy, indicate a remarkable impact of the NPs on the structural features of both ladder backbones and organic side chains, shedding light on the properties at the hybrid interface and providing suitable information to better design these composite materials.

2 Experimental

2.1 Materials

3-Methacryloxypropyltrimethoxysilane (MPTMS, 98%) was supplied from ABCR GmbH (Germany). Benzophenone (≥99.0%), potassium carbonate (K2CO3, ≥99.0%), anhydrous toluene (99.8%), tetrahydrofuran (THF, 99.9%) and ethanol (96%) were purchased from Sigma–Aldrich. Irgacure 184 was supplied by Ciba. Titania NPs (Evonik Aeroxide P25, average particle size 30–50 nm, SSABET = 54.7 m2g−1) were obtained from Evonik. All reagents were used without further purification.

2.2 Synthesis procedures

The preparation of LPMASQ (Scheme 1a) was carried out according to our previous reports [37,38,39] by MPTMS hydrolysis-condensation in the presence of K2CO3 catalyst in H2O/THF mixture, adapting the procedure of Lee et al. [36]. In detail, MPTMS (0.08 mol) was added dropwise under nitrogen flow and vigorous stirring to a solution prepared by dissolving K2CO3 (0.04 g) in deionized H2O (4.8 ml) and THF (9 ml). The clear solution was stirred at RT for 240 h in the dark, and after solvent evaporation under reduced pressure at 30 °C yielded a transparent and jelly-like product. 1H, 13C and 29Si NMR (Online Resource Fig. S1 and experimental conditions), ATR-FTIR (Fig. S2) and thermogravimetric (TG) analyses (Fig. S3) were recorded to confirm the ladder-type structure of LPMASQ, in agreement with previous reports [36,37,38].

Scheme 1
scheme 1

a LPMASQ synthesis and (b) preparation of nanocomposites by solvent-casting and UV-induced photo-crosslinking

Full size image

TiO2 functionalization was performed as schematized in Fig. S4, using the procedure reported in [39]. MPTMS (0.063 mol) was slowly added under vigorous stirring to the dispersion of 0.006 mol of TiO2 in 50 ml of toluene. The dispersion was refluxed (120 °C) for 16 h in the dark. Functionalized titania NPs (fTiO2) were recovered by centrifugation at 5000 revolutions per minute (rpm) for 10 min, washed three times with ethanol, and dried overnight at 80 °C. The particles functionalization was evaluated by FTIR spectroscopy (Fig. S5) and the amount of grafted molecules was calculated by TG analysis [39].

The synthesis of LPMASQ/titania NCs was performed according to [39] using a solvent casting technique followed by UV-induced photo-crosslinking (Scheme 1b). Briefly, LPMASQ (1 g) was dissolved in toluene (3 mL) at RT and various amounts of both TiO2 and fTiO2 NPs were added to the solution to finally obtain NCs with 0, 1, 2, or 3 wt% NPs loadings. To improve the curing efficiency, both Irgacure 184 (1 wt%) and benzophenone (1 wt%) were used as photoinitiators [41]. After stirring under nitrogen in the dark for 16 h, the mixtures were cast in Teflon molds (internal diameter 4 cm) and heated at 70 °C for 1 h. The samples were photopolymerized under N2 flow for 50 min by using a mercury vapor lamp (OSRAM, HBO 50 W/AC 39 V, lamp-sample distance 12 cm). The produced samples were stored in the dark under N2 atmosphere. NCs are respectively labeled L_xTiO2 and L_xfTiO2, where L represents the polymerized LPMASQ, and x is the wt% of TiO2 and fTiO2 NPs.

2.3 Characterization techniques

Solid state NMR analyses were carried out with a Bruker 400WB spectrometer operating at a proton frequency of 400.13 MHz. Samples were packed in 4 mm zirconia rotors, which were spun at 8 kHz under air flow. Adamantane and Q8M8 were used as external secondary references. NMR spectra were acquired with cross polarization (CP) and single pulse (SP) sequences under the following conditions: 29Si frequency: 79.48 MHz, contact time 5 ms, decoupling length 6.3 µs, recycle delay 10 s, 2k scans. Single pulse sequence: π/4 pulse 3.9 µs, recycle delay 60 s, 2k scans. 13C frequency: 100.48 MHz, contact time 2 ms, decoupling length 5.9 µs, recycle delay 5 s, 2k scans. For variable contact time (VCT) experiments, the contact time has been varied from 0.05 to 9 ms with a spin-lock field strength equivalent to 63 kHz. To evaluate the cross-relaxation trends, the curves of the magnetization area as a function of CP contact time (t) are fitted with Eq. 1: [42]

$$M\left( t \right) = M0 \cdot e^{ - t/T_{1\rho \left( H \right)}} \cdot \left( {1 - e^{ - t/T_{CH}}} \right)$$
(1)

where M(t) is the peak area as a function of contact time t, M0 is the normalization constant, T1ρ(H) is the proton spin-lattice relaxation time in the rotating frame, and TCH is the cross-polarization time constant. It is worth mentioning that multiexponential decays are also possible and could be explained in terms of the presence of different domains (i.e. homogeneity, segregation) [42, 43].

ATR-FTIR analyses were run in the 4000–550 cm−1 wavenumber range (64 scans, resolution 4 cm−1), with a Varian 4100 Excalibur spectrometer equipped with a diamond ATR crystal.

XRD patterns were collected on polymerized LPMASQ and NCs with a Rigaku D/III max diffractometer (Rigaku), using Cu Kα1 radiation (λ = 1.54056 Å) and a curved graphite monochromator in the diffracted beam, in the 2θ range 3°–80°, with sampling interval of 0.05°, and counting time of 2 s.

Thermogravimetric analysis was carried out on a Mettler Toledo TG50 with heating rate of 10 °C/min in N2 atmosphere.

3 Results and discussion

The structural features of synthesized LPMASQ were studied by multinuclear NMR and FTIR experiments and results were compared with previous reports [10, 34, 36,37,38, 44]. 1H and 13C NMR (Online Resource Fig. S1a, b) spectra of the obtained sample show the typical signals of methacryloxypropyl groups and the absence of unreacted methoxy groups. The 29Si NMR spectrum (Fig. S1c) shows a broad, weak resonance in the region −52 to −58 ppm attributed to residual T2 units (R-Si(OSi)2OH, 9%), and the main complex resonance in the region −60 to −69 ppm due to T3 units [10, 34, 44]. The FTIR spectrum (Online Resource Fig. S2) shows C–H stretching vibrations at 2943 and 2841 cm−1 due to methylene groups of the propyl chain; the signals related to Si-OH groups at 3500 and 960 cm−1 are not observable in the infrared spectrum, in agreement with the low amount of T2 units in the 29Si NMR spectrum. The strong absorption band at 1714 cm−1 and the signal at 1637 cm−1 are due respectively to C = O and C = C stretching vibrations of the methacryl groups. The appearance of two Si-O antisymmetric stretching vibrations at 1100 and 1020 cm−1 indicated the formation of dominant ladder-like structures [10, 34, 37, 44, 45], the high intensity of the absorption band at 1020 cm−1 points out lower structural regularity in comparison with LPMASQ produced by Lee et al. [36]. The thermogravimetric analysis (Online Resource Fig. S3) points out the remarkable thermal stability of synthesized LPMASQ according to the initial degradation temperature found at 375 °C and the shape of the thermal degradation curve is in agreement with those reported for ladder-like structures [31, 34, 36, 44]. FTIR, TG and NMR results overall confirm that the characteristics of the synthesized LPMASQ are similar to those previously reported [34, 36,37,38]. The slight lowering (5 °C) of the initial degradation temperature with respect to the value reported in [36] could be due to silanols, small defective structures, entrapped cycles or a slightly lower molecular weight of synthesized LPMASQ.

The effect of processing, namely solvent-casting and UV curing, on LPMASQ structure was investigated by means of solid state NMR, FTIR and XRD. The structural features of L sample (i.e. polymerized LPMASQ) evaluated by solid state NMR (Fig. 1) appear in agreement with previously published results [37, 38]. In the 13C CPMAS spectrum (Fig. 1b), all the resonances of LPMASQ methacryloxypropyl chains are visible (labelling scheme in Fig. 1a), but both the strong reduction in intensity of the C = C related peaks (e and g) with respect to pristine LPMASQ (Online Resource Fig. S1) and the additional new peaks confirm the occurrence of radical polymerization [37], with partial conversion of the methacrylic double bond. As a matter of fact, signals appearing at 45 and 55 ppm (g’ and e’, respectively) are related to the formation of new aliphatic carbons. Moreover, in the carbonyl region, a second C = O resonance (d’, 177 ppm) is also detected, which is downfield shifted with respect to the typical LPMASQ C = O peak (d, 165 ppm), and refers to the carbonyl close to methine e’ and methylene g’.

Fig. 1
figure 1

LPMASQ structural scheme and carbon labelling (a), 13C (b) and 29Si (c) CPMAS NMR spectra of L sample

Full size image

The features of the inorganic skeleton of L sample were studied by silicon-29 NMR. LPMASQ backbone is characterized by fully condensed T3 units (R-Si(OSi)3) [36, 44], which produce in the 29Si CPMAS spectrum (Fig. 1c) the asymmetric resonance at about −66 ppm. The amount of terminal T2 units (δ -57 ppm, ≈7%) decreases slightly compared to as-synthesized LPMASQ; this result is to be expected considering that the solvent casting step followed by thermal curing at 70 °C allow some condensation of residual silanols. The T3 band is centered at −67 ppm and its asymmetric shape results from the overlapping of the resonances detected in the 29Si NMR spectrum of pristine LPMASQ (Online Resource Fig. S1c) related to T3 units in different conformations [45]. The components linewidth (LW) helps to discriminate among the different conformations assumed by the LPMASQ chains, since it is known that the LW of the T3 peak is a measure of the regularity of the polymer skeleton [44,45,46]. Accordingly, sharp signals with LW below 200 Hz indicate highly ordered and long ladder chains. Thus, following the literature, the observed T3 resonance could be described as a main component (δ -67.1 ppm), representing the completely condensed linear ladders. The downfield and upfield shoulders could then be attributed to ladder-type silsesquioxane chains forming rigid stacks by folding together, short oligomers and defective structures similar to open cages, respectively [8, 44,45,46,47,48,49,50,51].

The analysis of the FTIR spectra recorded on LPMASQ and sample L (Fig. 2a) strengthens the assignment of T3 components in the NMR spectrum, while showing that the structural features of polymerized sample are not substantially modified with respect to LPMASQ. As a matter of fact, the siloxane stretching bands of L and LPMASQ are almost identical, supporting the use of pristine LPMASQ T3 components in the line-shape analysis of polymerized samples. The Si-O antisymmetric stretching vibrations produce a band in the range 1200–900 cm−1, which is the result of overlapping of various components. In detail, the shoulder at 1159 cm−1 is correlated to T7OH3 or T8OH2 open cage structures, the signal at 1100 cm−1 is attributed to cage-like architectures probably due to chain folding of LPSQs, and the components at 1030 and 1009 cm−1 are assigned to linear ladder-type units of different length [44, 45, 47].

Fig. 2
figure 2

a Si-O antisymmetric stretching region in the ATR-FTIR spectra of pristine LPMASQ and L samples (b) XRD pattern of L, with the structural scheme showing X and Y distances

Full size image

The relative intensity of the two bands centered at 1100 and 1020 cm−1 appears almost unchanged, in agreement with the limited secondary condensation shown by 29Si NMR. FTIR shows that minor changes occurred and the ladder-like network overall maintains the structural features of the starting LPMASQ [52].

Finally, the diffraction pattern of L (Fig. 2b) is characterized by two amorphous halos, respectively, centred at 2θ ~ 6°, associated with the intramolecular periodic chain-to-chain distance (denoted as Y in inset of Fig. 2b) and at 2θ ~ 20°, associated with the average thickness X of the Si–O–Si silsesquioxane structure [38, 44]. In addition, a third very broad hump centred at 2θ ~ 11° can be detected. This component has been previously observed in LPMASQ/silica NCs prepared by photopolymerization [38], and is attributed to the distances (Y’) produced by the side chains polymerization, which can lead to a broad distribution of intramolecular chain-to-chain distances as a result of variable side-chain lengths [53].

The NCs, obtained by solution casting and photopolymerization, appear as quasi-transparent homogeneous films that can easily be detached from the substrate. SEM analysis reveals that all samples are characterized by a typical surface waviness, but no defects are detected at low magnification (Online Resource Fig. S6); however, the EDX analysis pointed out that even if TiO2 is distributed throughout the samples, some aggregates are present regardless of NPs surface modification [39]. The functionalization of titania particles with MPTMS was confirmed by FTIR analysis (Fig. S5); the fTiO2 spectrum presents bands at 1716 and 1635 cm−1 due to methacryloxy groups and broad, weak Si-O antisymmetric stretching bands in the range 1200 to 1000 cm−1. The amount of grafted molecules was evaluated by TG analysis through the net weight loss in the range 150 to 1000 °C (2.8%) [39].

Despite the low amount (up to 3%w), the addition of both bare and functionalized TiO2 NPs to LPMASQ clearly affects the polymerization of methacryloxypropyl chains. As a matter of fact, the 13C CPMAS spectra (Fig. 3) show both a slight sharpening (the linewidth difference is about 20%) of the LPMASQ signals and the decrease in area of d’, e’ and g’ resonances compared to L. Surprisingly, g resonance displays both a relevant linewidth decrease and an intensity increase with increasing the TiO2 loading, but at the moment there is not a clear explanation for these phenomena. It should be mentioned that in the case of L_xfTiO2 samples, it is not possible to distinguish the features of LPMASQ side chains and titania surface functions, due to the identical molecular structure and the low amount of MPTMS grafted onto the NPs.

Fig. 3
figure 3

13C CPMAS spectra (contact time: 2 ms) of L, L_xTiO2 and L_xfTiO2 NCs

Full size image

To quantify the methacrylic double bond conversion, the % of d’ is calculated from the relative areas of d and d’ peaks. The values of Table 1 show that the extent of polymerization decreases with increasing titania amount, but a clear dependence on NPs functionalization cannot be argued.

Table 1 Integration of the carbonyl region of the 13C CPMAS NMR spectra shown in Fig. 3. The relative area of d’ with respect to the sum d+d’ gives the extent of LPMASQ polymerization
Full size table

In LPMASQ/silica NCs prepared by photopolymerization [38] we observed a similar reduction of methacrylic double bond conversion, which can be explained by the decrease of energy available for polymerization reactions due to the capability of inorganic particles to absorb or scatter light [54].

The organization of LPMASQ ribbons is also affected by NP addition, according to the changes observed for the inorganic backbone in 29Si spectra (Fig. 4).

Fig. 4
figure 4

29Si CPMAS NMR spectra of the LPMASQ/titania samples; the superposition of two 29Si spectra is shown in the inset to highlight the variation of the lineshape

Full size image

Similarly to what was noticed for the 13C resonances, a narrowing of the T3 peak is observable in LPMASQ/titania NCs, as the full width at half maximum (FWHM) is about 360 Hz compared with the observed value of 449 Hz for L sample. It is worth noting that, while the signal still exhibits evident asymmetry, it appears that the downfield components gradually reduce upon TiO2 addition.

These observations can be deepened by applying a profile fitting of T3 peaks. The line-shape analysis is performed directly on CPMAS spectra, since the comparison between CPMAS and MAS results (Online Resource Fig. S7) indicates that the spectra are completely superimposable and CPMAS spectra present higher S/N ratio. In all samples, the T3 resonance can be successfully fitted with five components, according to the above reported discussion for pure L sample. The results, summarized in Table S1 (Online Resource), point out that the relative amount of the different components is related to TiO2 loading. Only the component at δ -67.1 ppm appears to grow with the TiO2 amount at the expenses of the others (i.e. δ −68.7, −65.3, −64.1 and −62.8 ppm) that slightly decrease or remain constant, causing the observed signal asymmetry. NMR results are in agreement with previously reported FTIR evidence, which pointed out slight differences of the siloxane components in the Si-O antisymmetric stretching vibration band, in particular the decrease in intensity of the peaks at 1105 and 1067 cm−1 respectively assigned to cage-like structures due to chains folding and defective open cage structures [39]. Combining this information and the literature reports [8, 44,45,46,47,48,49,50,51], the components could be assigned as follows: δ -67.1 ppm is due to long linear ladders, the upfield component at −68.7 ppm to defective structures similar to open cages, and the three downfield components at −64.1 ppm to folded ladder chains and −62.8/−65.3 ppm to low molecular weight structures.

According to 13C NMR results, TiO2 negatively affects the extent of LPMASQ polymerization; thus, the increase of the linear component with titania load could be correlated with the lower methacrylic double bond conversion observed in LPMASQ/titania NCs. To validate this hypothesis, in Fig. 5 the intensity of the different T3 silicon resonances obtained from the corresponding 29Si NMR spectra (Online Resource Table S1) are reported as a function of the extent of polymerization calculated from the carbon-13 spectra (Table 1).

Fig. 5
figure 5

Correlation plot between the amount of T3 components (peak area % from Table S1) and the extent of methacrylic double bond polymerization expressed as the d’ relative area (% of the whole C = O signals from Table 1). Symbols and colours identify the specific T3 component; dotted lines are linear tendency curves

Full size image

The component at -67.1 ppm due to long linear ladders rises with the decrease of methacrylic double bond polymerization, while the other T3 components slightly reduce or remain constant, confirming that TiO2 not only negatively affects the polymerization process, but also influences the distribution of ladders’ conformations. Figure 6 summarises the structural effects pointed out by 13C and 29Si NMR studies.

Fig. 6
figure 6

Addition of titania in polymerized LPMASQ leads to the decrease of methacrylic group polymerization and the increase of linear ladders

Full size image

Finally, the comparison between L sample and nanocomposites reveals that the amount of T2 units (Online Resource Table S1) remains constant (≈7%) in the samples produced with fTiO2. The T2/T3 ratio, conversely, decreases slightly (5.5%) with the use of bare TiO2 suggesting that condensation between Ti-OH groups and silanols may have occurred during solvent-casting and thermal curing steps. This reactivity contributes to the complexity of both NCs structure and hybrid interface, since few siloxane ribbons can also be directly bonded to the TiO2 surface in addition to possible copolymerization of methacrylate groups of LPMASQ and grafted MPTMS.

On the other hand, the secondary condensation of LPMASQ silanols, also observed in sample L, could preferentially involve small and defective mobile fragments and together with the unfolding of the folded ladder structures supports the increase in linear ladders.

Since the results show that LPMASQ polymerization is controlled by the amount of TiO2 NPs regardless of functionalization, the study of molecular dynamics of organic side-chains in the slow (kHz) regime through 13C NMR was addressed to shed light on the filler-matrix interactions in NCs produced with bare and functionalized titania [55,56,57]. To fulfil this aspect, the evolution of the magnetization (i.e. the peak area) of L, L_2fTiO2 and L_2TiO2, as representative samples, has been followed as a function of the contact time in 13C CP experiments (Fig. 7a shows the trend for all of the carbon atoms of sample L). VCT curves are governed by kinetics and dynamics of the process of cross-polarization from protons to carbons and may be fitted through different models [43]. The rapid increase in the curves at short time intervals is dominated by 1H-13C polarization transfer (TCH). TCH values depend on the mutual C–H distance and H atoms’ density in the proximity of a given C atom (ms range) but are affected by the group mobility that interferes with 1H-13C polarization transfer [58]. After the maximum, the curve decay is governed mainly by the spin-lattice relaxation in the rotating frame T1ρ(H). The decay constant T1ρ(H) of the protons is a volume property averaged over a distance of ca. 2 nm and provides information on the relative mobility of the H atoms. In homopolymers, high T1ρ(H) values are related to long chains or rigid/crystalline materials and low values are associated with small molecules/domains and the presence of amorphous phases [42, 43]. Moreover, T1ρ(H) provides information on possible regions of morphological heterogeneity. Involving a large number of monomeric units, T1ρ(H) can be significantly affected by the spin diffusion process, which tends to average it to a single mean value in polymeric chains [55, 57].

Fig. 7
figure 7

13C VCT curves (a) of all the resonances in sample L and (b) of signal d’ in the samples L, L_2fTiO2 and L_2TiO2

Full size image

In this work, the magnetization curves of LPMASQ carbons were normalized to have the maximum intensity equal to 100 and the curves were obtained by fitting with Eq. 1, in order to highlight the different behavior of all functional groups. The carbons in sample L clearly present different trends (Fig. 7a). The curve growth, which is described by TCH constant, is characterized by different slopes, depending on the dipolar coupling strength, typical of each functional group. Interestingly, looking at the decay part of the curves expressed by the parameter T1ρ(H), relevant differences among the functional groups can be appreciated that indicate inefficient spin diffusion at RT. Voelkel [43] explained this phenomenon with a high molecular mobility of the system, which averages out the proton-proton dipolar interaction. In our case, this hypothesis can be considered viable, taking into account the narrow signal linewidths (200–300 Hz) observed in the 1H MAS NMR spectrum of sample L (Online Resource Fig. S8) [43, 56].

Similar considerations can be made by comparing the curves of functional groups in LPMASQ/TiO2 NCs. However, the comparison of trends of the same functional group in samples prepared with and without titania highlights different behaviors. Among the functional groups, the most interesting information is obtained from the study of the carbonyl groups, which are the most sensitive to the presence of the filler. Indeed, we have shown above that it is possible to derive information about the polymerization process from the analysis of carbonyl carbons in unreacted and polymerized methacrylate groups, respectively. In addition, the carbonyl carbons are of particular interest also for the study of molecular dynamics, since they are not directly linked to H and the nearest protons are responsible for their relaxation in the rotating frame, providing useful information on the chains’ overall behaviour [55, 56]. Both growth and decay of d’ curves (Fig. 7b) are very sensitive either to the presence of NPs and their surface functionalization. In particular, the d’ curve relaxes faster for L_2TiO2 and even more in the case of L_2fTiO2. Thus, the presence of TiO2 and fTiO2 seems to differently influence the folding/organization of the LPMASQ chains during the photopolymerization process. The NPs probably induce a different arrangement of LPMASQ polymerized side-chains, and the functionalization of the particles’ surfaces enhances this effect. The dynamics of d’ appear to be sensitive to the chemical groups present on the NPs surface. As a consequence, the different surface features, i.e. titanols available for H-bonding or methacrylic groups able to copolymerize, induce conformation rearrangements favouring less ordered chain packing [53, 57], in agreement with the 29Si results.

The curves of carbonyl d (Online Resource Fig. S9) show similar growths, but the different slope of L_2fTiO2 curve suggests higher mobility of the close protons; the decay appears slightly faster for the composites than for L, generally suggesting moderate sensitivity of unreacted chains to the presence of NPs. The different behaviour of d and d’ is highlighted by the comparison of the curves in all the three samples. Interestingly, similar, though less obvious, trends are found for the closest g’ and g carbons (Online Resource Fig. S9). Finally, the propyl carbons a, b and c have fairly similar trends and do not appear sensitive to the presence of filler, due to its vicinity to the LPMASQ inorganic skeleton.

From these results, both organization and dynamics of polymerized side-chains appear strongly affected by NPs and even more by their surface functionalization, while the effect is less pronounced for resonances belonging to unreacted LPMASQ side-chains. Polymer chains suffer structural deformation in close proximity to surfaces, with typically a reduction of the free volume moving away from the NP surface to the bulk, reflected in a chain density variation [59]. Therefore, an alteration in polymer chain organization near the interfaces is possible with respect to the bulk region.

To investigate the effects induced by NP addition at a different length scale and complement the information obtained by NMR at the molecular level, the XRD patterns of L and selected NCs were compared (Fig. 8).

Fig. 8
figure 8

XRD patterns of L and NCs with bare and functionalized TiO2

Full size image

In L_xTiO2 and L_xfTiO2 samples, sharp peaks account for the presence of different amounts of titania (anatase, PDF 21-1272, rutile PDF 21-1276, in agreement with the presence of both anatase and rutile phases in Aeroxide P25 [60]). These reflections are overlapped with the amorphous halos observed in L sample. The peak related to periodic intramolecular chain-to-chain distance Y moves to lower angles with the addition of titania, while the signal related to the siloxane backbone thickness X appears unchanged. Moreover, the third halo (related to distance Y’) that was observed at 2θ ~ 11° in sample L, becomes narrower and shifts towards higher angles (Online Resource Fig. S10), as a consequence of the different polymerized side-chain arrangement highlighted by solid state NMR.

The profile fitting analysis of XRD patterns of selected samples (Online Resource Fig. S11) was run in order to quantify the observed changes. Table 2 reports the calculated d-spacing values and the corresponding FWHM for the three reflections, and the parameter R, defined as the ratio of intensities of Y and X related signals, which gives information on structural regularity and concentration of defects in ladder polysilsesquioxane [61]. The constant X value equal to 0.44 nm is comparable to reported distances in ladder silsesquioxanes and confirms that the siloxane backbone is unaffected by the addition of NPs, as also proved by the unmodified FWHM. The intramolecular chain-to-chain distance Y increases in NCs, but the results do not suggest a clear trend with the amount and functionalization of nanoparticles. TiO2 addition strongly modifies the third halo, which is likely to be related to the polymerization process and consequently to organic side-chains distances.

Table 2 Results of the profile fitting analysis of XRD patterns of L, L_1fTiO2 and L_2fTiO2
Full size table

Accordingly, sample L presents a very broad distribution of such distances as a consequence of random polymerization among LPMASQ propylmethacrylate chains. In NCs, TiO2 addition reduces the distribution of side-chain distances, which become shorter with respect to the case of sample L. This result, together with the observed increase of Y distance and the related signal narrowing, suggests a separation between domains characterized respectively by long ordered and short disordered organic side-chains [53]. This hypothesis is strengthened by the large decrease of R value in NCs, which points out a relevant increase of defects with loss of the structural regularity of ladders.

4 Conclusions

The effect of photopolymerization and addition of a small percentages of TiO2 nanoparticles, both bare and functionalized, on ladder-like poly(methacryloxypropyl) silsesquioxanes was studied by combining solid-state NMR and X-ray diffraction to highlight structural changes at different length scales.

Adding up to 3 wt.% of titania nanoparticles reduces the methacrylate double bond polymerization and changes the distribution of ladders structures, according to 13C and 29Si solid state NMR. In particular, the decrease of the extent of organic side-chain polymerization can be correlated with the increase of linear long ladders, pointing out the effect of TiO2 both on the polymerization ability and the chain conformation of LPMASQ.

The XRD study of the polymerized LPMASQ (L) sample highlights the appearance of a broad distribution of distances Y’ produced by the random polymerization among propylmethacrylate chains, leading to a wide distribution of ladder chain-to-chain distances. This reflection adds to the characteristic amorphous halos due to the typical LPSQ intramolecular chain-to-chain distance Y and the length of the siloxane backbone X. The addition of titania nanoparticles, regardless of functionalization, results in an increase in distance Y and the narrowing of the distribution of Y’ distances, which become shorter. Moreover, the decrease of the parameter R highlights the loss of the ladders’ structural regularity.

Interestingly, the effect induced by the use of functionalized NPs is revealed by approaching the study of the organic chains’ molecular dynamics in the slow regime, here probed through the analysis of cross-polarization kinetics. Above all, this pointed out that the polymerized organic chains display a change of conformations/folding as a result of changing interactions at the filler-polymer interface.

In summary, this study, beyond demonstrating that even the incorporation of very low loadings of TiO2 NPs produce relevant modifications on the architecture of ladder-like silsesquioxanes, constitutes a methodological approach that can be extended also to other composite systems, opening the possibility to better understand and tune their structural and interfacial properties with positive implications on the functional properties.