Introduction

Cellulose nanocrystals (CNCs) are sustainable nanomaterials that have been studied in many applications, for example, in the biomedical area (Wang et al. 2016; Madhusudana Rao et al. 2017), food industry (Agustin et al. 2014), oil/water stabilizers (Visanko et al. 2014) and sensors and transistors (Gaspar et al. 2014). These versatile applications are justified by the interesting properties of these cellulosic nanoparticles such as high modulus, low density, crystallinity, renewability, high aspect ratio and biodegradability. Recently, CNCs has been widely studied as reinforcing agent in polymeric hydrogels. Examples of polymer matrices already studied for this purpose are polyacrylamides (Hebeish et al. 2014), polyethylene glycol (Yang et al. 2013), alginate (Mohammed et al. 2015; Wang et al. 2016) and gelatin (Wang et al. 2016). Besides the CNC properties mentioned above, the growing interest in CNC-based gels is explained by the ability of these nanocrystals to form highly porous and low density aerogels (usually 0.004–0.500 g cm−3) (De France et al. 2017; Yang and Cranston 2014; Xu et al. 2018). In addition, a selective absorption property can be tailored, widening their range of applications (Zhang et al. 2014). These characteristics make the CNC-based gels attractive in different applications such as organic compound removal, drug delivery, insulating materials and templates for other functional materials (De France et al. 2017). CNC-based gels can be prepared by different techniques, such as evaporative concentration, homogenization, hydrothermal heat treatment and sol–gel process (De France et al. 2017). CNCs are nanometric rigid particles that interact mainly through hydrogen bonds. These CNC characteristics result in physically cross-linked gels made up of pure nanoparticles, which show low structural stability.

In order to overcome the limitation of low structural stability of neat CNC gels, some strategies can be used in CNC-based gel preparations. For example, CNC which self-assemble into liquid–crystal structures can be formed by increasing the CNC content. These in turn can act as substrates or templates for physically cross-linked gels (Chu et al. 2017). Alternatively, chemically cross-linked CNC-based gels prepared by CNC surface functionalization with coupling molecules and further formation of chemical bonds between the nanoparticles are also reported (Yang and Cranston 2014). This approach has been extensively studied by using different coupling agents such as anhydrides, diisocyanates, amides, among others (Ly et al. 2008; Eyley and Thielemans 2014), which promote not only covalent bonds between CNCs but also different functionalities for the CNC-based gels (Abraham et al. 2017; Way et al. 2012; Yang and Cranston 2014; Yang et al. 2015; (Abraham et al. 2017; Way et al. 2012; Yang and Cranston 2014; Yang et al. 2015).

Silane agents have also been studied as coupling agent in cellulose materials (de Oliveira Taipina et al. 2013; Thakur et al. 2014; Zhang et al. 2014) as well as for the preparation of silica/cellulose and silica aerogels using cellulose materials as a template (Cai et al. 2012; Xu et al. 2018). Zanini et al. (2016) produced aerogels from silanized cellulose nanofiber (CNF) suspension, using methyltrimethoxysilane (MTMS) in acid media to favor the formation of silanol groups and the reaction with the hydroxyl groups present on the cellulose nanofibers. de Oliveira Taipina et al. (2013) obtained cellulose nanocrystals with a reduced hydrophilic surface character using 3-isocyanatopropyltriethoxysilane, and recently Khanjanzadeh et al. (2017) reported the functionalization of cellulose nanocrystals with 3-aminopropyltriethoxysilane. By using this approach, Wu et al. (2017) developed an aminosilane functionalized cellulose nanocrystal aerogel, applying supercritical CO2 as a drying method.

Organotrialkoxysilanes are molecules that can be readily hydrolyzed and condensed by sol–gel reactions, resulting in the formation of siloxane network, named as polysilsesquioxane (Brinker 1988). Polysilsesquioxanes are glassy materials that present a hydrophobic character. On the other hand, cellulose nanocrystals, due to their hydroxyl superficial groups, present hydrophilic character. The combination of both materials enables the preparation of omniphilic CNC-based gel, i.e., a material capable of absorbing both polar and apolar solvents (Chatterjee et al. 2016). To the best of our knowledge, the use of a silane that can act both as a CNC surface coupling agent and as a chemically cross-linking CNCs through a sol–gel process, has never been reported.

Recently, a patent of macroporous omniphilic sponges was published by Kumaraswamy et al. (2017). The authors pointed out the importance of these materials as an alternative use in pollutant removal due to their versatile absorption characteristics. The opposite of an omniphilic material is an omniphobic material, i.e., it presents lipophobicity and hydrophobicity simultaneously. Cunha et al. (2010) modified cellulose fibers with 3-isocyanatopropyltriethoxysilane, in the presence of 1H,1H,2H,2H-perfluorodecyltriethoxysilane, giving rise to fibers with omniphobic character.

Contrary to the above mentioned uses of silane agents combined with cellulose, the aim of the present study was to prepare chemically cross-linked aerogels based on cellulose nanocrystals. To achieve this purpose, 3-isocyanatopropyltriethoxysilane (IPTS) was used for both nanocrystal surface modification and polysilsesquioxane forming network. The combination of CNCs and polysilsesquioxane network can result in a porous hybrid structure with an omniphilic character. This new cross-linking method for CNC is a promising strategy to produce aerogels with improved performance and novel functionality.

Materials and methods

Materials

Cellulose nanocrystal powder was purchased from The Process Development Center (University of Maine, USA) and 3-isocyanatopropyltriethoxysilane (IPTS, 95% purity) from Sigma Aldrich. Nitric acid (HNO3) was obtained from LabSynth and dimethylformamide (DMF) was purchased from Vetec and dried prior to use.

Methods

Preparation of the chemically cross-linked aerogels

Cellulose nanocrystals were dried under vacuum at 70 °C for 24 h before use. Then, 1 g CNC was added to 9.4 mL and 6.7 mL DMF to prepare suspensions with 10 and 15 wt% CNC, respectively. The suspension was prepared immediately prior to synthesis and maintained under stirring to avoid CNC gelation. After that, the suspensions were kept under argon flux, in order to avoid the presence of water, for 20 min. Next, an IPTS solution in DMF was added, in a 2:1, 1:1 and 1:2 CNCOH:IPTS molar ratios. CNCOH indicates the number of CNC superficial hydroxyl groups. In this case, the number of mols of superficial hydroxyl groups was calculated by the multiplication of 8.13 × 10−3 mmol m−2 value, which was obtained by Lin and Dufresne (2014), and the total CNC superficial area, which depends on the CNC weight used as well as the CNC mean diameter and density. The reaction was maintained under stirring at 60 °C for 72 h. These reaction conditions were well controlled to avoid CNC gelation in this step. Afterwards, a 3 wt% nitric acid aqueous solution, corresponding to a 3:1 HNO3:IPTS molar ratio, was added and the mixture was stirred carefully. The final suspension was cast onto Teflon™ molds, which was heated at 60 °C for 24 h. After gelification, a solvent-exchange procedure was carried out (DMF by deionized water). Finally, the gels were rapidly frozen under liquid N2 and then freeze-dried to obtain aerogels. A proposed reaction between CNC and IPTS are schematically shown in Scheme 1.

Scheme 1
scheme 1

Schematic representation of the chemically cross-linked CNC/PSS aerogel synthesis. Initially, the CNC hydroxyl groups react with the IPTS isocyanate groups, producing modified CNCs (CNC/IPTS) by the formation of carbamate bonds. Then, an aqueous solution of nitric acid was added to the modified CNC to promote triethoxy group hydrolysis and condensation by a sol–gel process, resulting in the polysilsesquioxane network incorporating CNCs (CNC/PSS)

Full size image

Aerogels were labeled as shown in Table 1, where the A or B denotes the CNC suspension weight content (10 or 15 wt%, respectively), and 1, 2 and 3 denote the CNCOH:IPTS molar ratios (2:1, 1:1 and 1:2, respectively).

Table 1 Sample compositions of cellulose nanocrystal/polysilsesquioxane aerogels
Full size table

Characterization

ATR-IR spectra were obtained in a Cary 630 (Agilent) spectrophotometer at a 4000–400 cm−1 wavenumber range and recorded with a 4 cm−1 resolution using 64 scans. This technique was also used in a kinetic study of isocyanate conversion. To do so, aliquots have been withdrawn from the reaction media, monitoring the consumption of the isocyanate band as a function of synthesis time.

X-ray diffraction (XRD) measurements were carried out in an X-ray diffractometer (SHIMADZU XRD 7000). The diffraction patterns with Cu Kα radiation (k = 0.15406 nm) at 40 kV and 30 mA were recorded in the region of 2θ from 5° to 40°. The crystallinity index (CI) of cellulose nanocrystals and the degree of crystallinity of CNC/PSS aerogels were calculated using the Segal method (Segal et al. 1959), that takes into account the ratio between the lowest intensity between the 200 and the 110 peaks (Iam, near to 18°), related to the scattered intensity of the amorphous fraction, and the intensity of the 200 peak (I200, near to 22.5°), related to the crystalline fraction, as shown in Eq 1:

$${\text{CI}} = \frac{{{\text{I}}_{200} - {\text{I}}_{\text{am}} }}{{{\text{I}}_{200} }} \times 100$$
(1)

Cross Polarization-Magic Angle Spinning 29Si Nuclear Magnetic Resonance (29Si NMR) was obtained at room temperature in a Bruker 300 spectrometer operating at 79.5 MHz for the 29Si nucleus. The acquisition time was 40 ms, contact time 3 ms, recycling delay 1 s and spectral width 50 × 103 Hz.

The morphology of the cryogenically fractured aerogels was examined in a field emission scanning electron microscope (SEM) (FEI Quanta FEG-250) operating at a 10 kV accelerating voltage. All samples were iridium sputter coated in a Bal-Tec MD 020 equipment (Balzers).

CNC and aerogel morphologies were observed by transmission electron microscopy (TEM). Drops of a 0.1 mg mL−1 aqueous CNC suspension were deposited on the flat surface of positively charged ultrathin carbon coated copper grids and stained with a 2 wt% uranyl acetate solution. CNC micrographs were obtained in a JEM 1400 plus transmission electron microscope (JEOL) at 120 kV accelerating voltage. Images were recorded using a One View camera (Gatan). CNC dimensions were measured using the ImageJ software. On the other hand, aerogels were embedded in epoxy resin and cut into ultrathin sections (approximately 50 nm thick), at − 100 °C in a Leica EM FC6 cryo-ultramicrotome. The aerogel morphologies were investigated using a Carl Zeiss Libra 120 transmission electron microscope (80 kV) equipped with in-column OMEGA filter. Images were recorded using an Olympus 2 k × 2 k (Cantega) CCD camera.

BET surface area was determined by N2 adsorption and desorption using a Quantachrome NOVA4200 equipment. The samples were first degassed at 100 °C for 24 h prior to the analysis followed by BET analysis at 77 K. The BET surface area and the Barrett–Joyner–Halenda (BJH) pore size and volume were calculated from the obtained isotherms.

Aerogel densities were determined by weighing and measuring the dimensions of each individual aerogel. The aerogel weight was obtained using an analytical weighing scale (0.0001 g accuracy) and their dimensions were measured by a digital caliper (± 0.01 mm) at six different positions. Four samples were used to determine the density of each sample.

The gel absorption capacity and porosity were calculated as the weight ratio of swollen gels (in water and toluene) and dried aerogels. Aerogels were immersed in the solvents at 30 °C for 24 h until the absorption equilibrium was reached and subsequently weighed. Weight measurements were performed in triplicate. The absorption kinetics of the B3 aerogel for water and toluene was determined during different periods of time. The absorption capacity was calculated according to Eq. 2.

$${\text{Absorption}}\;{\text{capacity}}\left( \% \right) = \frac{{\left( {{\text{W}}_{\text{wet}} {-} {\text{W}}_{\text{dry}} } \right) }}{{{\text{W}}_{\text{dry}} }} \times 100$$
(2)

where Wwet is the weight of the swollen aerogel; Wdry is the weight of the dried aerogel.

The aerogel porosity, ε (%), was determined according to Eq. 3 (Liao et al. 2012).

$$\upvarepsilon\left( \% \right) = \frac{{\frac{{\left( {{\text{W}}_{\text{wet}} - {\text{W}}_{\text{dry}} } \right)}}{{\uprho_{\text{solvent}} }}}}{{\frac{{{\text{W}}_{\text{wet}} - {\text{W}}_{\text{dry}} }}{{\uprho_{\text{solvent}} }} + \frac{{{\text{W}}_{\text{dry}} }}{{\uprho_{{{\text{CNC}}/{\text{PSS}}}} }}}} \times 100$$
(3)

where ρsolvent is the pure solvent density (water or toluene) at 30 °C (g cm−3) and ρCNC/PSS is the density of the CNC/PSS matrix, which was considered as being approximately the density of cellulose (ρcellulose = 1.59 g cm−3) (Heath and Thielemans 2010).

Sorption experiments were performed using water and hexane dye solutions. Methylene blue dye was added into the water and Sudan Orange G dye into hexane. The aerogels were submerged in both solutions to verify their hydrophilic and hydrophobic characters.

Results and discussion

In relation to the reaction conditions used in this work, it is important to point out that the absence of humidity in the reaction media prevented the occurrence of hydrolysis and condensation reactions, allowing the isocyanate reaction with the cellulose superficial hydroxyl groups. This was the first step of the gel synthesis, which was carried out to provide the CNC surface modification itself. In the second step, an aqueous acid solution was added to the modified CNC in order to start the hydrolysis and condensation reactions, i.e., the sol–gel process. The addition of water promoted the hydrolysis of the alkoxy groups, forming Si–OH groups, which suffered auto-condensation, leading to the Si–O–Si group formation. The addition of nitric acid (catalyst) was carried out in order to accelerate the condensation step, which is slow. In this context, the gel synthesis was achieved in this last step and resulted in an oligomeric network of polysilsesquioxane cross-linking the cellulose nanocrystals.

After the gel synthesis, a solvent-exchanged method was applied to allow the leaching out of free CNC and PSS, which were not covalently bounded to the gel structure. Samples A1 and B1 presented higher weight loss (~ 50 wt%) after solvent-exchange compared to the other samples that showed weight loss at around 10 wt%, as shown in the Supplementary Material (Table S1). This behavior for the A1 and B1 samples was due to the lowest IPTS fraction. In order to confirm aerogel stability, the hydrogels were tested under sonication and orbital stirring. Figure S1, in the Supplementary Material, shows the photographs of a gel immersed in water, before and after sonication. All gels kept their shapes after these procedures, even the A1 and B1 samples, which also confirmed that the PSS network was covalently bounded to the CNC surface. In addition, preliminary compression tests were carried out at a 60% maximum compression strain and it was observed that fully water saturated CNC/PSS gels recovered their shape once the load was released, as can be seen in the photographs of the A3 gel before, during and after compression (Figure S2 in the Supplementary Material). The compressive stress–strain curves of CNC/PSS aerogels in water are also presented in the Supplementary Material (Figure S3). The compressive stress values obtained for the CNC/PSS gels were in the 160–470 kPa range, which are higher than the previously reported in literature for CNC and CNF aerogels (Yang and Cranston 2014; Jiang and Hsieh 2014).

In order to better present the most important aerogel structural characterization results, samples A2 and B2 were chosen due to being the samples that best represented the characteristics of the sample set. The results obtained for the other samples can be found in the Supplementary Material (Figures S5, S6 and S7).

A kinetic study was carried out using ATR-IR technique, monitoring the band at 2274 cm−1, related to the IPTS isocyanate group stretching. Figure S4, in the Supplementary Material, presents the conversion kinetic of the carbamation reaction on the CNC surface. The results showed that after 72 h, 68% of IPTS isocyanate groups were converted to carbamate bonds.

Figure 1 and S5a (Supplementary Material) show the ATR-IR spectra of the CNC, PSS and aerogels. Pure PSS film was prepared by hydrolysis and condensation of IPTS, resulting in a brittle and transparent material. The PSS infrared spectrum shows a C–H (propyl group) stretching absorption bands in the 2960–2850 cm−1 region. In addition, bands at ca. 1650 and 1550 cm−1 related, respectively, to C=O and C–N stretching from carbamate bonds (de Oliveira Taipina et al. 2013) were also observed as well as a shoulder near to 1700 cm−1, attributed to C=O stretching. These bands are an indication that the CNC modification by the IPTS was successful. Moreover, the condensation of alkoxysilane groups led to a 1110 cm−1 band attributed to the Si–O–Si stretching (de Oliveira Taipina et al. 2013). However, this band was overlapped with cellulose C–O–C stretching bands in the same spectral region (Dai and Fan 2011; Khanjanzadeh et al. 2017). Therefore, the confirmation of the PSS network formation was obtained by 29Si NMR technique, which is explained below. In addition, CNC and PSS characteristic bands were also observed in the extraction solvent, which was used in the solvent-exchange method. This result indicates that fractions of both CNC and PSS, which were not covalently bonded in the aerogel structure, were leached out. These results are presented in the Supplementary Material (Figure S5b).

Fig. 1
figure 1

ATR-IR spectra of the cellulose nanocrystals, PSS and CNC/PSS aerogels

Full size image

CNC and CNC/PSS aerogel crystalline structures were investigated by XRD (Fig. 2). Similar results were observed for the other samples (Supplementary Material, Figure S6). Cellulose nanocrystals presented typical diffraction peaks at 14.7°, 16.4°, 22.5°, and 34.5° (2θ), which are assigned to Cellulose I crystal unit (Segal et al. 1959; Peng et al. 2013; French 2014). Diffraction peaks at 12.1°, 19.8°, 22.5° and 34.5° (2θ) were also observed and are assigned to Cellulose II (Segal et al. 1959; Peng et al. 2013; French 2014). Cellulose I is the most common form of cellulose crystalline structure. However, Cellulose II crystals can also be observed as a characteristic form in natural fibers, or as a result of fiber exposure to high concentrations of alkaline solutions (Oudiani et al. 2011). On the other hand, PSS presented a broad peak centered at around 20°, which is typical of amorphous material. Diffraction patterns for all aerogels revealed that the CNC crystal structures were preserved after the reactions. Moreover, as expected, the degrees of crystallinity, calculated for CNC/PSS aerogels, were lower than CNC crystallinity index, due to the presence of the PSS network.

Fig. 2
figure 2

XRD spectra of CNC, PSS and CNC/PSS aerogels, and corresponding crystallinity index and degrees of crystallinity

Full size image

29Si NMR spectra of the PSS and CNC/PSS aerogels (Fig. 3) were obtained to provide further structural information concerning the polysilsesquioxane network structure. NMR spectra show peaks at − 66, − 57 and − 48 ppm. As described in literature, polysilsesquioxane network can present four different structures, sketched in Figure S7 in the Supplementary Material, which are classified as T0, T1, T2 and T3. The overwritten number is related to the number of siloxane bonds around the central Si atom (Salon et al. 2007). The appearance of T0 (− 44 ppm) indicates that a fraction of the Si–OH groups suffered no condensation. This peak is absent in the NMR spectra of the PSS and aerogel samples. A lower intensity peak associated with T1, at − 48 ppm, is present in the aerogels and corresponds to the Si–O–Si groups in a chain end. T2 and T3 corresponding peaks, at − 57 and − 66 ppm, respectively, are observed in all samples and are related to linear and three-dimensional structures. Therefore, the presence of T1, T2 and T3 peaks, for all CNC/PSS aerogels, confirmed the PSS network formation.

Fig. 3
figure 3

29Si NMR spectra of PSS and CNC/PSS aerogels. Boxes marked in blue, orange and pink indicate chemical shift characteristic ranges for T1, T2 and T3 siloxane structures, respectively

Full size image

A detailed study by NMR characterization of silane network formation, between silane molecules and cellulose fibers, was performed by Salon et al. (2007). The authors showed that the reaction conditions have a significant influence in the efficiency of hydrolysis and condensation steps, which can affect the silane structure. Regarding the pure PSS sample, NMR spectrum showed that the hydrolysis and condensation steps mainly promoted the formation of T2 and T3 siloxane sites, i.e., linear and three-dimensional structures, respectively. In contrast, the presence of CNC promoted the increase of T1 peak. This increase of the T1 peak may be due to the presence of the rigid CNC nanoparticles, which caused steric hindrance in the self-condensation of the alkoxysilane. A similar result was observed by Zhang et al. (2014), who attributed this partial self-condensation to condensation and hydrogen bonding between the polysiloxane and cellulose surface. In order to compare the different siloxane structures present in each aerogel, the 29Si NMR spectra were deconvoluted and the relative area percentages of the different peaks are presented in Table S2 in the Supplementary Material. The fraction of T1 structures was small and ranged from 7 to 17% for all aerogel samples. In addition, a decrease of the T1 structure relative percentage with the CNCOH: IPTS ratio increase is observed for samples containing 15 wt% CNC. In terms of linear siloxane T2 and three-dimensional T3 structures, the ratios between them is almost constant and does not vary with the CNC concentration and CNCOH: IPTS ratio.

SEM micrographs of CNC/PSS aerogels (Fig. 4) allowed the observation of interconnected macropores presenting similar average diameters of approximately 10 μm (Table S3 in the Supplementary Material), which were separated by thin walls.

Fig. 4
figure 4

SEM images of aerogel cross-sections with the respective hydrogel photograph on the top right hand corner and sample codes on the top left hand corner

Full size image

The A1 sample showed a layered-morphology, which is clearly different from the others. This distinct morphology can be due to different factors, such as the combination of low CNC and IPTS concentrations, which could result in a low degree of the PSS network anchoring in the CNC surface, as observed in the soluble fraction results for this sample. On the other hand, aerogels prepared by using a 15 wt% CNC suspension resulted in uniform structured pores, also separated by thin walls. It is well known that the aerogel morphology can be affected by the freeze-drying process of the CNC suspension. As pointed out by Aulin et al. (2010), cooling the CNC suspension with liquid nitrogen contributes to the formation of a sheet-morphology as a consequence of the ice structure.

The morphology of cellulose nanocrystals alone was studied by transmission electron microscopy (Fig. 5a). Average length and cross-section of the CNCs were measured using TEM micrographs. The obtained values are 130 ± 50 nm, 7 ± 2 nm and 19 ± 8, for the average length, cross-section and aspect ratio, respectively. These values were acquired from 730 measurements and are in agreement with those obtained by Reid et al. (2017) for CNCs from the same source.

Fig. 5
figure 5

a TEM micrographs of CNC and b A3 aerogel pore wall. c Schematic representation of the process leading to aerogels comprising polysilsesquioxane network cross-linked with cellulose nanocrystals

Full size image

Figure 5b exhibits TEM image of the A3 aerogel ultrathin section, showing the pore wall in detail. This image shows the porous morphology, in which the walls, of around 300 nm thick, were made up of cross-linked CNC and PSS. The presence of darker domains with nanometric dimensions was observed (white arrows), and is attributed to silsesquioxane network rich domains. Polysilsesquioxane domains at micro and nano scales have already been reported in literature (Junior et al. 2012; Brandão et al. 2013). Although the mesopores were not clearly observed by TEM, bright regions in the pore walls can indicate their presence. A schematic representation of the process leading to aerogels comprising polysilsesquioxane network and cellulose nanocrystals is shown in Fig. 5c. In this model, the modified CNC were distributed inside the PSS network where mesopores could be present. CNC were not observed in the TEM micrograph, probably due to the characteristic low contrast of these nanoparticles.

To confirm the presence of mesopores from the PSS network, pore volume was measured by the BJH analysis of BET data and the results are shown in Table 2. The pore size distribution showed a single peak at around 4 nm (Figure S8 in the Supplementary Material), which were similar notwithstanding the CNC or IPTS content. However, the mesopore volume decreased with the increase of CNC content in the precursor suspension and with the decrease of the IPTS content. Their lower values, compared with physically cross-linked aerogels (Heath and Thielemans 2010), indicate that the CNC modification and cross-linking by the PSS network resulted in a denser aerogel structure due to the hindering of the nanoparticle repulsion forces which are present in the negatively charged CNC surface.

Table 2 CNC/PSS aerogel pore volumes, solvent absorption capacities and porosities
Full size table

The aerogel surface area was also determined by BET analysis. The surface area for the CNC alone was theoretically calculated to be 368 m2 g−1, using the experimentally determined CNC dimensions and the cellulose density (Heath and Thielemans 2010). On the other hand, the experimental aerogel surface areas, determined in this work, were in the 23–90 m2 g−1 range. These smaller surface area values, compared to those for CNC alone, would be expected due to the cross-linking between CNC nanoparticles, promoted by the PSS network.

The total pore volume (Table 2) was calculated directly from the porosity and the aerogel density. Aerogels showed density values in the 0.11–0.17 g cm−3 range (Figure S9, Supplementary Material). These results are similar to the physically cross-linked CNC gels prepared by Heath and Thielemans (2010). Moreover, it is possible to observe that mesopore volume percentage represents a small fraction of the total pore volume (0.8–3.5% range). The adsorption/desorption curves of the CNC/PSS aerogels are presented in Figure S10 (Supplementary Material).

Cellulose has a hydrophilic character and presents high affinity to water and polar solvents. On the other hand, polysilsesquioxane networks have a hydrophobic character. Therefore, with the purpose of evaluating the hydrophilic/hydrophobic character of CNC/PSS aerogels, polar and non-polar solvent absorption experiments were performed and are also presented in Table 2. These measurements were carried out for all samples except for A1 and B1 aerogels, since they showed high weight loss during solvent-exchange procedure. Absorption kinetics of CNC/PSS aerogels in water and toluene are presented in Figure S11 (Supplementary Material), which showed that the absorption is extremely quick at the beginning and is complete in the first minutes of contact with the solvents. Aerogels synthesized from 10 wt% CNC suspensions showed higher absorption capacity when compared to those synthesized from 15 wt% CNC suspension. Absorption capacity reached maximum values for the A2 sample due to the lower CNC and IPTS precursor contents. The A3 sample showed an increase in both water and toluene absorptions when compared to the B3 sample. Regarding the PSS content, 1:1 CNCOH:IPTS molar ratio sample presented higher solvent absorption value, compared to samples with 1:2 CNCOH:IPTS ratio, which can be attributed to the higher PSS network content in the latter sample. Moreover, no significant difference in aerogel absorption capacity was observed regarding solvent polarity. This interesting behavior could be explained by the capillary forces caused by the aerogel interconnected pores. Moreover, the rigid structure of these aerogels limits conformational changes leading to little volume change and consequently limited solvent absorption capacity.

Chemically cross-linked CNC aerogels developed by Yang and Cranston (2014) can absorb 160 g/g water and 72 g/g dodecane. Zhang et al. (2014) prepared aerogels from silylated nanofibrillated cellulose, which present absorption capacity between 42 and 102 g/g, depending on the solvent used. Superabsorbent materials based on cellulose derivatives are also reported in literature (Jiang and Hsieh 2014; Wang et al. 2015b). The aerogels obtained in this work, however, presented lower absorption capacities (between 5 and 8 g/g) than the above mentioned aerogels.

Based on the results herein presented, CNC/PSS aerogels can be considered a tridimensional rigid structure, resulting from the cross-linking between rigid nanoparticles as well as rigid inorganic network. Thus, an approximately 10% increase in the aerogel volume was observed as a consequence of solvent uptake, which is less expressive when compared to polymer aerogels (Figure S12 in the Supplementary Material shows the dried gel and after water sorption). In addition, when the solvent uptake reached equilibrium, it was possible to observe that all aerogels presented opacity in contact with polar solvent. However, when toluene was used as a solvent, the aerogels became transparent. Changes observed in optical properties of the aerogels, after solvent uptake, can be related not only to filled porous structure but also to the higher toluene permeability through the gel walls, compared to water. The high toluene permeability is a consequence of van der Waals interactions between this solvent and the PSS network. In this case, the non-polar solvent permeation can favor the filling of both the macropores and mesopores.

Aerogels based on CNC and PSS, prepared in this work, presented relatively high porosity (Table 2) as a consequence of their characteristic morphology. However, the higher the CNC and IPTS contents in the precursor suspension, the lower the CNC/PSS aerogel porosity. Similar results are also reported in literature (Heath and Thielemans 2010; Yang and Cranston 2014) and are proposed to be a consequence of higher cross-linking density, which gives rise to a more compact structure.

Solvent absorption tests showed that the combination of CNC with PSS network promoted an omniphilic character in the CNC/PSS aerogels. Therefore, to evaluate this characteristic, one aerogel was chosen and one side of it was immersed in an aqueous methylene blue solution and the other side was immersed in a Sudan orange G solution in hexane. Figure 6 shows the CNC/PSS aerogel photograph after immersion in both solutions, where it is possible to observe both stained sides of the aerogel. Interestingly, a green interface is present between them, indicating that both polar and non-polar dye solutions interpenetrated the aerogel pores. The above confirms the presence of the hydrophilic and hydrophobic domains in the aerogel structure. Similar results were obtained for the other samples.

Fig. 6
figure 6

Wettability test of CNC/PSS aerogels. a Freeze-dried aerogel and b aerogel after absorption of aqueous solution of methylene blue (left) and solution of Sudan Orange G in hexane (right)

Full size image

Hydrophilic/hydrophobic character of CNC/PSS aerogels was also evaluated by contact angle measurements. However, it was not possible to carry out these measurements with water or toluene due to the instantaneous solvent absorption. Since the gels are highly porous, the liquid absorption is due to capillary action (Chatterjee et al. 2016).

The capacity of absorbing polar and apolar solvents by cellulose based materials has already been studied in applications such as stabilizations of oil/water emulsions (Visanko et al. 2014; Wang et al. 2015a, b; Yang and Cranston 2014) and water treatment (Jiang and Hsieh 2014). Amphiphilic aerogels have potential for biomaterial application, especially for drug delivery systems, as their capacity to dissolve, incorporate and drive poor water soluble drugs can be exploited (Prasad et al. 2007). Omniphilic aerogels are less common in literature but show potential applications as absorbents for a wide variety of spills (Chatterjee et al. 2016). In the case of CNC/PSS aerogels, these could also be applied to the biomedical area, since CNC and PSS are both biocompatible.

Conclusions

Aerogels based on modified cellulose nanocrystals with a silane agent and subsequent chemical cross-linking are presented. CNC surface modification was confirmed by ATR-IR and the maintenance of the cellulose crystalline integrity by XRD results. The resulting polysilsesquioxane network obtained after hydrolysis and condensation of the silane was mostly made up of T2 and T3 structures, as evaluated by 29Si NMR spectra. CNC/PSS aerogels showed structural shape stability after the freeze-drying process, sonication and stirring, as well as shape recovery after compression. Mesopores and macropores were detected in BET and microscopy analyses, respectively, and aerogel pore walls were observed by TEM, which showed the presence of PSS nanodomains. These aerogels exhibited similar behavior towards water and toluene absorption. These results were attributed to the aerogel omniphilic character and rigid porous structure. In addition to the absorbent characteristics, the biocompatibility of CNCs and PSS encourage the study of these aerogels in the biomedical area. Omniphilic chemically cross-linked aerogels are still little explored and may open new potential applications for CNC-based aerogels.