Foam-formed biocomposites based on cellulose products and lignin

We fabricated the foams at our laboratory-scale facilities. First, for all samples, methylcellulose was dissolved in water at 50 \(^{\circ }\)C. When the temperature of the aqueous methylcellulose suspensions reached 30 \(^{\circ }\)C, we added the respective amounts of cellulose fibers, lignin, and water according to Table 1. The suspensions were vigorously stirred and cooled overnight at 3 \(^{\circ }\)C. The foaming of the aforementioned suspensions proceeded through Hele-Shaw flow between two syringes plugged into each other. One syringe was filled with 30 ml of suspension, whereas the other was adjusted to leave a 30 ml free space. Then, the fluid is foamed by exchanging it between the syringes until the volume occupied by the foam reaches 60 ml. The dry foam is produced by printing rod-shaped wet foam structures parallel to each other until a foam sheet of \(27\,\mathrm{cm}\times 10\,\mathrm{cm}\) is formed. The printed wet foam is dried immediately using a heating lamp. A foam block is then produced by layering six foam sheets “glued” on top of each other. The gluing process involves gently spraying water on the foam sheets so that their surfaces bond together (samples are dried again). In a previous publication, we provided more insights into the foam manufacturing process (Reichler et al. 2021). Finally, the density of the foam was measured by cutting cubic specimens of 15 mm per side. We calculated the foam density from the mass and volume of each cube, obtaining a standard deviation no larger than 2 kg/m\(^{3}\).

Before foaming the suspensions, we inspected their rheological properties using SAOS temperature sweep tests. An MCR 302 rheometer (Anton Paar, Austria) equipped with Couette geometry deformed the suspensions at a constant angular frequency (\(\omega\)) of 6.28 rad/s. Within a temperature (T) range from 15 to 70 \(^{\circ }\)C (heat rate of 1 \(^{\circ }\)C/min), a serrated bob tool (CP17/P6) dynamically applied a strain of 1%. Through SAOS tests, we identified the temperature at which the samples transitioned from viscoelastic liquids to viscoelastic solids. The viscoelastic transition temperature (\(T_{v}\)) is calculated from the experimental data of \(\tan {\delta }\) as a function of T, according to the slope intercept method reported by Miranda-Valdez et al. (2022). The loss factor is calculated on the basis of Eq. (1)where \(G^{\prime \prime }\) and \(G^{\prime }\) are the loss shear modulus and the storage shear modulus, respectively. Refer to Ghanbari et al. (2020) for more details about rheometry.

We inspect the foam surface with an optical microscope BX53M (Olympus, Japan) and a camera attachment DP74 (Olympus, Japan). Optical microscopy showed bubbles formed in the surface layers of the foams. We report photographs with a 20\(\times\) objective.

We cut segments of different foam blocks, such as the one in Fig. 1, along the y-direction to observe the bubble structure of their cross-sections. To enhance electrical conductivity, we coated the samples with a thin layer of Au/Pd 80/20. In addition, the foam samples were mounted on carbon tape. The foams were inspected using a field emission scanning electron microscope (FE-SEM) from Zeiss Sigma VP (Germany). The electric potential remained constant (3 keV) and the images were generated from the SE2 signal.

The wettability of the foam surfaces was examined by measuring the contact angle (\({\textit{CA}}\)) on a Theta Flex tensiometer (Biolin Scientific, Sweden). For the \({\textit{CA}}\) test, a 5 \(\upmu\)l drop of water (resistivity of 18 M\(\Omega\) cm) was deposited on the foam top surface parallel to the z-direction. The tensiometer analyzed with a camera the contact angle between the water drop and the surface; a \({\textit{CA}}>90^{\circ }\) is typical of hydrophobic surfaces. The initial \({\textit{CA}}\) was recorded after 1.8 s and the last after 20 s. Each measurement was made in triplicate, and the arithmetic mean is reported.

Infrared spectra of the foams were scanned using a Spectrum TwoTM LiTaO\(_{3}\) spectrometer (Perkin Elmer, United Kingdom) equipped with an Attenuated Total Reflection accessory (Specac Quest, United Kingdom). Spectra were acquired at 25 \(^{\circ }\)C in a wavenumber range of 4000–500 cm\(^{-1}\) with a resolution of 4 cm\(^{-1}\) after 30 scans. We identified the functional groups of cellulose, methylcellulose, and lignin from the spectra. The spectra did not require a baseline correction; however, the spectral intensity was normalized to the absorption intensity at 1056 cm\(^{-1}\). The experiments were replicated three times, and no significant variation in the absorption bands was observed.

We surveyed the thermal decomposition of the foams using a Jupiter STA 449 F3 thermogravimetric analyzer (Netzsch, Germany). The studies were carried out under a protective He gas flow (70 ml/min) while heating from 50 to 700 \(^{\circ }\)C at 10 \(^{\circ }\)C/min. The STA analyzer measured the percentage of mass loss during the heating stage. All experiments were duplicated, and no significant differences were observed in the thermograms. The derivative of the mass loss with respect to the temperature was calculated from the experimental results.

To measure the elastic properties and deformation of the foams, we compressed samples (geometry of \(\approx\) 1.5 cm per side), such as the one shown in Fig. 1. Compression tests were performed on an Electroplus^® E1000 dynamic testing machine (Instron, United Kingdom). The protocol consisted of compressing five specimens of each sample in the y-direction (parallel to the foam rod-like structure) and in the z-direction (perpendicular to the rod-like structure) at a deformation rate of 30 mm/min. We plotted the stress (\(\sigma\)) vs. strain (\(\varepsilon\)) curves using the average and standard deviation obtained from five measurements.

The elastic modulus was calculated by finding the maximum slope of the elastic part of the stress-strain curve and fitting a slope in a range of 20% \(\varepsilon\) before this point. The slope of the fitted line corresponds to the Young’s modulus (E) of the material. The yield point is defined as when the stress diverges 10% from the fitted line. See Fig. S1 to Fig. S3 for visualizing the fitting process and the determination of the yield points. Lastly, the area under the stress-strain curve (integrated from 0 to 70% \(\varepsilon\)) was used to estimate energy absorption (W).

We evaluated the antibacterial and antifungal activity of the foams against Escherichia coli (E. coli) and Saccharomyces cerevisiae (yeast), respectively. All chemical reagents were analytical grade and were used without further purification. E. coli was cultivated in an LB broth (lysogeny) and yeast in a YPD broth (yeast extract peptone dextrose). Ultrapure water with a resistivity of 18 M\(\Omega\) cm was used for all experiments. Briefly, the experiments proceeded as follows. The foams were first weighed and then sterilized with UV light for 20 min. Meanwhile, the overnight cell culture prepared was diluted into a suspension of 106 CFU/ml (CFU stands for colony formation unit). Subsequently, approximately 100 mg of foam materials were added to 20 ml of diluted suspension for overnight incubation. The incubation temperature for E. coli and yeast were 37 \(^{\circ }\)C and 30 \(^{\circ }\)C, separately. The difference in optical density (\({\textit{OD}}\)) of cell suspensions at 600 nm was measured with an eppendorf BioPhotometer plus spectrometer. The results after incubation time were compared with a control cell culture and reported using Eq. (2). All materials were tested in triplicate. The error bars show the standard deviation.

For foam fabrication, it is relevant to evaluate the \(T_{v}\) and elasticity of any precursor suspension, as these characteristics rule out the feasibility of printing and drying. SAOS experiments allowed us to characterize the elasticity of suspensions by measuring \(G^{\prime }\). Furthermore, \(T_{v}\) was estimated from the change in energy dissipation (\(\tan {\delta }\)). Figure 3a shows the former and Fig. 3b the latter. Consequently, the SAOS results for all suspensions demonstrated that, when heated, \(G^{\prime }\) increased sharply (Fig. 3a), while \(\tan {\delta }\) changed oppositely (Fig. 3b). These changes in \(G^{\prime }\) and \(\tan {\delta }\) are more noticeable when the suspensions were heated above \(T_{v}\) (Table 2). The viscoelastic behavior of \(G^{\prime }\) and \(\tan {\delta }\) arises from the aqueous MC matrix of each suspension, which forms a fibril network (Coughlin et al. 2021; Lott et al. 2013). As the MC matrix forms a fibril network, a large number of non-covalent interactions appear between the hydrophobic sites of the polymer (Arvidson et al. 2013).

Regarding the rheological differences in the suspensions studied, Fig. 3a depicts that Foam-CF and Foam-LN showed higher \(G^{\prime }\) than Foam-MC. This effect was expected because the solid content in Foam-CF and Foam-LN is higher than in Foam-MC. Furthermore, \(G^{\prime }\) of Foam-CF and Foam-LN show similarities in shape and magnitude, implying that mechanical reinforcement in suspensions is primarily due to the incorporation of CF. The results reported by multiple authors indicate that the total solid content in a system containing a gel-forming material, such as MC, increases \(G^{\prime }\) and reduces \(T_{v}\) of the systems. For example, Miranda-Valdez et al. (2022) modeled the decrease in \(T_{v}\) as a function of the CF content in an MC suspension. Similarly, Korhonen and Budtova (2019) studied the CF content and its impact on the gelation time of dissolving pulps.

In our case, we observed a decreasing \(T_{v}\) as a function of the solid content of the suspension, Foam-MC being the suspension with the lowest solid content and, therefore, the highest \(T_{v}\) (Table 2). Although the mechanism by which \(T_{v}\) decreases as a function of the content of CF or LN is uncertain, it is believed to occur due to the formation of hydrogen bonds between the fibers and the dissolved matrix (Budtova and Navard 2016; Korhonen and Budtova 2019; Miranda-Valdez et al. 2022). Understanding the gelation mechanism of MC in relation to the CF and LN content is beyond the scope of this article, but we acknowledge its importance. From the plot in Fig. 3b, we can only infer that the activation energy of the gelation process should decrease when CF and LN are added to the MC suspensions. This can be interpreted from the magnitude of energy dissipation (\(\tan {\delta }\)) presented by Foam-CF and Foam-LN, which is less than that of Foam-MC.

The optical microscopy images in Fig. 4 show a bubble-like structure with defined boundaries in Foam-MC (Fig. 4a) and Foam-CF (Fig. 4b). However, this structure is less noticeable for Foam-LN (Fig. 4c). We attribute the disappearance of the bubble boundaries to a densification effect on the foam surface resulting from the addition of lignin to Foam-LN (see the density values in Table 2). Lignin, as a natural binder, can contribute to the formation of hydrogen bonds, improving the adhesion between all the constituents. Examples of these effects can be observed in biomass pellets, where lignin was shown to improve bridging and bonding after drying (Kaliyan and Morey 2010; Nanou et al. 2018). From SEM analysis in Fig. 5 and Fig. S4 to Fig. S6, we point out that Foam-MC, Foam-CF, and Foam-LN possess a closed cell structure with bubbles elongated over the y-direction (quasi-hexagonal prism structure).

In relation to the surfaces presented in Fig. 4, we also measured their wettability by testing the water CA. Foam-LN exhibited an initial water \({\textit{CA}}\) of \(117^{\circ } \pm 4^{\circ }\), which remained above 90\(^{\circ }\) even after 20 s. However, although Foam-MC and Foam-CF had an initial \({\textit{CA}}\) close to 90\(^{\circ }\), after 20 s their \({\textit{CA}}\) decayed to the hydrophilic range. Accordingly, lignin contributed to the production of a hydrophobic surface on Foam-LN. This hydrophobic effect may arise from a combination of surface lignin and surface roughness in Foam-LN (Ferreira et al. 2020; Yang et al. 2006). Looking at Fig. 4a, b, the surfaces of Foam-MC and Foam-CF are smooth, while Foam-LN has many textures. Pictures of the droplet \({\textit{CA}}\) on the foam surface are provided in Fig. S7 to Fig. S9. Lignin in the Foam-LN system may be driven to the surface (liquid-gas interface) as a result of enthalpic changes, which may involve reordering hydrogen bonds during the foam-forming process (Chandler 2005). However, understanding the hydrophobic behavior of Foam-LN requires more studies.

Regarding the chemical structure of the foams, from the FTIR spectra in Fig. 6, we identified the spectral position of the relevant functional groups of cellulose. Table 3 collects the information mentioned above. At 3435–3441 cm\(^{-1}\), the OH stretching vibration appears for all foams. Compared to Foam-MC and Foam-CF, it is possible to visualize the effect of lignin on Foam-LN on the stretching vibration of the CH groups, since the wavenumber is displaced from 2840 to 2850 cm\(^{-1}\). This displacement results from CH located in aromatic methoxyl groups and methyl groups of the lignin side chains (Boeriu et al. 2004). Basically, all foams showed identical FTIR spectra; nevertheless, for Foam-CF and Foam-LN, an absorption band appears at 1740 cm\(^{-1}\). This band is in the carbonyl region and may be related to residual lignin in the BSKP constituting Foam-CF and the organosolv lignin added to Foam-LN. At spectral positions close to 1740 cm\(^{-1}\), lignin typically exhibits the \(\hbox {C}=\hbox {O}\) stretching of unconjugated ketones and carbonyl groups (Boeriu et al. 2004; Derkacheva and Sukhov 2008; Faix 1991).

As the final part of the structural characterization, in Fig. 7, we evaluated the thermal decomposition of the foams. Interestingly, the foam mass was stable at temperatures close to 200 \(^{\circ }\)C. However, the addition of CF and LN accelerated the onset of the decomposition reaction. This can be observed in the zoom-in graph in Fig. 7a for the 225–350 \(^{\circ }\)C range and can be attributed on a part to the hemicellulose content present in CF, which decomposes at lower temperatures than cellulose (Yang et al. 2007; Yu et al. 2017). Regarding Foam-CF, adding cellulose fibers to the foam reduced the carbonization yield, compared to the yield observed for Foam-MC. Further studies are needed to determine the pyrolysis mechanism between MC and CF. In the case of Foam-LN, pure lignin is known to thermally decompose at lower onset temperatures than cellulose and in a wider temperature range (Ma et al. 2015; Trogen et al. 2021; Yang et al. 2007; Yu et al. 2017). In cellulose/lignin blends, Miranda-Valdez (2022) and Trogen et al. (2021) observed that lignin transfers its thermal behavior when mixed with cellulose. Regarding the offset of thermal decomposition (see zoom-in plot in Fig. 7a for 360–430 \(^{\circ }\)C range) of Foam-LN, lignin promoted a higher mass yield. From Fig. 7b, the derivative of the mass shows that the maximum decomposition rate shifted to higher temperatures as CF and LN were added to the foam. Furthermore, the peak minimum shifts to higher temperatures for Foam-CF and Foam-LN. Foam with such thermal properties, such as Foam-LN, would be of particular interest for tailoring biocarbon sponges in the future.

Figure 8a, b depict the foam compression results. The first collects the stress-strain curve obtained from compressing in the y-direction, and the second comes from testing in the z-direction. From Fig. 8, we calculated the mechanical properties, as explained in the methods section, and summarized them in Table 2. In general, the observed mechanical anisotropy in Fig. 8a, b comes from the rod-like structure of the foams assembled parallel to the y-direction and the open mold fabrication process (Gibson and Ashby 1997; Reichler et al. 2021). Furthermore, the closed cell geometry of the foams and the elongated bubbles, as shown in the SEM image of Fig. 5, align parallel to the y-direction, increasing the mechanical anisotropy of the foams (Gibson and Ashby 1997; Reichler et al. 2021).

The compression of the samples in the y-direction (Fig. 8a) showed the three typical deformation stages described by Ashby and Medalist (1983). Initially, (1) as the foams deformed, they showed an elastic region, (2) which is followed by a plateau of deformation (plastic yielding for plastic foams and elastic buckling for elastomeric foams). (3) After the plateau, the closed cell structure collapses, densifying the foams. According to Gibson and Ashby (1997), the stress-strain profile of the foams y-direction (Fig. 8a) would resemble a plastic foam, while the z-direction would for an elastomeric foam (Fig. 8b). For an elastomeric foam, the onset of stress typically begins with a lower strain than for a plastic foam (Gibson and Ashby 1997). The first is the case observed in Fig. S1 to Fig. S3, a fact supporting that the foams exhibit plastic and elastomeric behavior in the y- and z-direction, respectively (Gavin et al. 2018).

Concerning foam composition and its effect on compression behavior, we compare foam samples by referring to their stiffness index (Table 2). Using the stiffness index allows us to include the density effect in the assessment. First, compared to Foam-MC, the fibers in Foam-CF increased stiffness in the y-direction, but not in the z-direction. The stiffness index in the y-direction of Foam-CF follows the fiber-reinforcement principle since both the MC matrix and CF carry a certain proportion of the compressive load. For example, Kerche et al. (2021) reported that microfibrillated cellulose fibers increased the mechanical performance of a polyurethane foam. However, the stiffness observed in Foam-CF and Foam-LN is related to the fiber orientation and length. Due to the fiber length (average 1.47 mm), CF can only align parallel to the y-direction, making it unsuitable to carry a load in the z-direction (Shen and Nutt 2003; Vaikhanski and Nutt 2003). Foam-LN displayed the highest stiffness index. In this case, lignin, as a binding agent, may have improved adhesion between the dispersed fibers and the continuous MC matrix. The effect is similar to the case reported by Xu et al. (2022) for a biocomposite made of CNF containing LN, in which the LN increased mechanical performance.

Compared to other biobased foams, our foams exhibited mechanical performance similar to those prepared with nanocellulose (Cervin et al. 2016; Sehaqui et al. 2010). For example, Cervin et al. (2016) reported a lightweight foam made of CNF with Young’s modulus of 0.2 MPa and \(\rho = 13\) kg/m\(^{3}\). In the y-direction, even our reference foam (Foam-MC) exhibited a superior stiffness index. Our technology demonstrates versatility, as we could add different materials to our reference foam. Both CF and LN increased mechanical performance in the y-direction. However, the z-direction of the foam still requires tuning.

All foams showed antibacterial effects (the \({\textit{OD}}\) of the foams was lower than the control) against E. coli. Foam-MC was the sample that performed best against bacteria. Comparable antibacterial properties were reported by Guibal et al. (2013) for a chitosan/cellulose foam tested against E. coli. However, by observing Foam-CF in Fig. 9, the presence of CF in the foams reduced their inhibition of bacterial growth because \(\varDelta {\textit{OD}}\) of Foam-CF and Foam-LN was lower than for Foam-MC. On the other hand, LN appeared to protect CF from E. coli since Foam-LN \({\textit{OD}}\) is higher than Foam-CF. Ottenhall et al. (2018) reported that CF reduced the inhibition of bacterial growth against E. coli because CF can provide nutrients to E. coli (Ottenhall et al. 2018). On the other hand, against yeast, only Foam-CF and Foam-LN showed antifungal activity. More studies are required to elucidate the inhibition mechanisms.