Scanning electron microscopy
SEM was used to visualise the structure of LP and EP and their morphological changes induced by the polymer coating. For all samples, the paper fibres were homogeneously distributed (figure S1). LP exhibited an increased level of entanglement (Fig. 3a) due to a higher number of smaller fibrils than EP (Fig. 3b). The polymer coating wrapped the fibres, which was very obvious for the PCLP (Fig. 3c), thereby changing the appearance of the small entanglements from single fibres to agglomerated structures. For EP, the number of entanglements also decreased (Fig. 3d) because fibres with a smaller diameter were adhered onto fibres with a larger diameter. In figure S1, overview SEM images at a 1400 × magnification are given of all paper and paper-polymer systems.
Contact angle goniometry
With the help of static CA measurements, the hydrophobic character of the polymer coating was investigated on a macromolecular scale. The unmodified paper samples adsorbed the water quickly, and videos were recorded to obtain the measurements. These videos were stopped 1 s after the water drop was applied to determine the CA of the different samples. The images of sessile droplets (2 µl of deionised water) on the LP, PCLP, EP, and PCEP sheets for the CA measurements are given in Fig. 4.
LP and EP showed a hydrophilic behaviour. The imbibition was faster in EP (2 s) than in LP (8 s). On both unmodified samples, the CAs were in the same regime, which was also the case for the polymer-coated samples. The CAs of PCLP and PCEP were in the hydrophobic range, with value of 119° ± 10° and 129° ± 9°, respectively, while the CAs of LP and EP showed hydrophilic behaviour with CA values of 79° ± 10° on both materials. On the polymer-coated samples, the water drop was not adsorbed and was much more stable than in the unmodified samples. Additionally, Fig. 3 shows a smoother surface for the polymer-coated samples than for the unmodified samples, which was also observed in the SEM measurements (Fig. 3). These results proved the hydrophobic character of the polymer coating.
The distribution of the polymer on the paper was analysed by fluorescence imaging. The measurements were taken on paper fleeces as well as on small fibre bundles to evaluate the distribution of the polymer. In Fig. 5, overlays of bright field microscopy images (grey) and the corresponding fluorescence microscopy images (yellow) are shown.
A faint and evenly distributed fluorescence could be seen for the unmodified samples, which was caused by the autofluorescence of the lignin (figure S2). A higher autofluorescence was noticed on EP because of the higher lignin content in EP than in LP (Young and Rowell 1986). In contrast, PCLP and PCEP exhibited a strong fluorescent signal that was localised along the fibres. Nevertheless, an analysis of the fluorescent signal showed that the fleece was not homogeneously coated (red areas in Fig. 5 a,d). Additionally, on the level of single fibres, as shown in Fig. 5c,f, the fluorescence signal and thus the amount of coating could vary. The inhomogeneous polymer distribution implies that water could infiltrate not only the unmodified paper but also the coated samples.
Raman measurements were performed to investigate the water uptake in the samples as a function of the moisture and incubation time by observation of the OH band. The reference Raman scans, including the normalised intensity versus Raman shift graphs with the respective vibrational motions, are given in Fig. 6. A table of the peak assignments can be found in Table S1. Briefly, the vibrational motions of carbon and oxygen are located in the lower wavenumber regime (379–1118 cm−1), while peaks at higher wavenumbers (1290–3500 cm−1) are due to vibrations involving carbon, hydrogen, and oxygen.
Raman spectra of polymer-coated samples (Fig. 6, blue spectra) showed the superposed spectra of paper (Fig. 6, red spectra, C# labels) and the polymer (Fig. 6, P# labels). The Raman spectrum of the polymer blend P(S-co-MABP-co-PyMA) (figure S3) shows the superposition of the components PS, benzophenone (BP), and pyrene. For the peak assignments, the peak positions were compared to Osterberg et al. for cellulose (Osterberg et al. 2006), Brun et al. and Hong et al. for PS (Brun et al. 2013; Hong et al. 1991), Babkov et al. for BP (Babkov et al. 2006), and Xie et al. for the pyrene component (Xie et al. 2010). The peak positions for all materials are also listed in Table S1.
To characterise the water uptake, the integrated intensity of the OH band was monitored. Specifically, the OH band was monitored in a region of interest inside the paper fleece but close to the surface. The integrated intensity versus RH graphs in Fig. 7 and figure S4 correlate with the water uptake. In Fig. 7, the integrated intensity is given for room conditions (50% RH), 80–90% RH, conditions in which the sample was fully wetted by a droplet and dry conditions at 2–5% RH.
For the unmodified samples, the integrated intensity of the OH peak was substantially higher after adding the drop of water in comparison to measurements at 80–90% RH for 60 min or 180 min. This led to the result that the time of exposure to 80–90% RH was not long enough to fully wet the paper samples. Comparing the experiments after adding the water drop to unmodified samples, it was observed that the water uptake of EP was higher than that of LP, which agreed with the CA goniometry results. The more pronounced uptake of water for EP was explained by the high content of hemicellulose in comparison to LP (Young and Rowell 1986).
For the modified samples, the change in the integrated intensity was predicted to be less distinct than in the unmodified samples because the water should not enter the paper through the polymer coating. In fact, the hydrophobicity of the coating could be confirmed by observing the sliding motion of drops over the sample surface, therefore not being imbibed. Thus, the addition of the water drop did not lead to as much of an increase in the OH peak intensity as that for the unmodified samples. For an increased RH, the change in integrated intensity was also smaller than for the unmodified samples.
The change in integrated intensity was larger for the unmodified samples than for the polymer-coated samples. The results supported the assumption of a hydrophobic character of the cellulose samples after the addition of the polymer coating.
Atomic force microscopy
To characterise the variation in the mechanical characteristics of the fibres in the different states of hydration, AFM indentation measurements were performed on individual fibres.
The indentation versus time diagrams were divided into three regions, according to three different states of hydration: (I) fibres under room conditions, (II) soaked fibres, and (III) nitrogen-dried fibres. A higher indentation depth at the fixed deflection point was correlated with fibre softening. The depicted changes in percentage were extracted from the average values of the respective region. The results are shown in Fig. 8a,b. EP fibres underwent softening (76 ± 13%) from region I to region II with a consequent continuous increase in the indentation. In contrast, PCEP fibres not only showed an increased stiffness in region I but also a reduced softening in the hydrated state (12 ± 6%) in comparison to EP. The direct comparison of EP and PCEP fibres showed that the polymer coating increased the stiffness of the fibre. The reduced softening was related to reduced water uptake. The indentation values for both samples returned to their respective values (region I) after dehydration (region III). Softening of the LP fibre (38 ± 15%) when hydrated could be measured. The PCLP fibres were in the same range as the PCEP fibres (0.1 – 0.3 nm/nN) and showed the same softening behaviour (12 ± 5%) upon hydration. LP fibres showed the highest stiffness among the samples, which was attributed to the higher content of cellulose contained in LP than in EP. This agrees with Baley et al. and Gassan et al., who stated that the elastic modulus of natural fibres increases with increasing cellulose (Baley 2002; Gassan et al. 2001; John and Thomas 2008). Another possible reason for this result is the high portion of crystalline regions in linters, while eucalyptus mostly has amorphous sites. The crystalline regions are expected to be stiffer than amorphous regions, which could explain the high resistance of the LP fibres against indentation compared to that of the EP fibres.
It can be concluded that the coated samples show less softening when exposed to water than the uncoated samples, which is correlated with less water uptake. Furthermore, the coated samples exhibited a very similar behaviour independent of the underlying fibre material. While the indentation experiments at a single point confirm the functionality of the polymer coating as a water barrier, further AFM experiments are necessary.
In comparison to earlier studies of AFM nanoindentation on fibres (e.g., Ganser et al. 2015), the fibres presented here were freely suspended, and their surface was not supported. This setup introduced further degrees of freedom for the fibres to react towards a load, including bending and twisting. Although deflection versus separation curves that indicated movements of the fibre were not considered for evaluation, the combination of probe indentation with movement of the fibre could not be fully excluded. Thus, nanoindentation experiments were not used to calculate the reduced modulus but to show a force-dependent displacement as a measure of mechanical resistance. The results allowed for a qualitative rather than a quantitative comparison between the different states of hydration and thus supported the presented results of the further applied characterisation techniques.
A more comprehensive understanding of the samples requires two-dimensional maps that show the topography and its correlation with the mechanical properties. An identification of crystalline and amorphous regions and their respective modulus as a function of various states of hydration could support or refute the explanation of a stiffer LP due to crystallinity. Additionally, the acquired distribution of the modulus provided by two-dimensional maps can be used to evaluate the representativeness of the previously presented values. Thus, PeakForce-Tapping mode was applied to freely suspended fibres.
To investigate the mechanical properties of the paper fibres as a function of the RH, AFM measurements were performed on freely suspended single fibres. For the polymer-coated fibres, fluorescence microscopy was applied before the AFM measurements to identify homogenously coated regions of interest on the paper fibre.
The quasi-static PeakForce-Tapping mode allows for the simultaneous mapping of the topography and mechanical properties. Here, cumulative force versus distance curves were measured at every pixel in the recorded map. The mechanical properties were then directly extracted from the force versus distance curves. While the adhesion force represents the minimum in the retraced curve, the DMT modulus was fitted, as shown in Fig. 2b, with the prediction of the contact mechanics by Derjaguin, Muller, and Toporov (DMT) (Derjaguin et al. 1975). In contrast to other contact mechanics models, such as the Johnson, Kendall and Roberts (JKR) model (Johnson et al. 1971), the DMT model includes adhesion outside the contact area and is suitable for high elastic moduli, low adhesion and a small radius of indentation. Therefore, the DMT model was used for the contact mechanics in the measurements. As shown in Fig. 9, the LP fibre exhibited crystalline and amorphous regions in the topography image, but especially in the mechanical property maps, such as the DMT modulus and adhesion, the differences in the crystal structure of cellulose were clearly distinguishable. In the mechanical property maps, the adhesion increased with increasing RH, while the DMT modulus decreased. Overall, swelling of microfibrils was observed in the AFM topography images as the RH was increased. Figure 9a shows the topography image with the microfibrils at 2% RH. Comparing the microfibrils in Fig. 9a to those in Fig. 9d, which shows a fully hydrated fibre, the microfibrils appear to be swollen under wet conditions. The white box (Fig. 9a, d) indicates in detail the swelling of microfibrils due to absorbed water inside the cellulose network. The insolubility of cellulose results in swelling of the microfibrils (Lindman et al. 2010). Water infiltrates the network and diffuses through the pore space system by capillary condensation. Therefore, the volume increases, which leads to an expansion in the crystal structure and eventually breaks the hydrogen bonds in the cellulose network (Cabrera et al. 2011; Gumuskaya et al. 2003; John and Thomas 2008). Thus, variations in the RH can induce changes in the shape of a microfibril due to the not entirely connected cellulose molecules, as shown in Fig. 9d, in the marked areas. Crystalline and amorphous regions showed different values for the DMT modulus, which is in agreement with Cabrera et al. (Cabrera et al. 2011). Cross sections of the DMT modulus versus the position over a crystalline and an amorphous region from Fig. 9 are shown in Fig. 10b,c. The crystalline region showed a DMT modulus of 80–100 GPa, while the amorphous region exhibited a DMT modulus of 20–30 GPa. The transition from crystalline to amorphous structures, and thus the transition in the mechanical properties, seemed to be sharp and well defined
according to Fig. 10a.
A general decrease in the normalised DMT modulus with increasing RH was observed in crystalline and amorphous LP fibres, as well as in EP fibres (Fig. 11a). Up to 40% RH, the crystalline regions of the LP fibres did not show a significant decrease in the normalised DMT modulus, as recorded for the amorphous regions. If the RH was sufficiently high, softening of the crystalline regions was also initiated. More water molecules intrude the fibre network, diffuse to the crystalline areas and break the hydrogen bonds in these ordered areas (Persson et al. 2013). If the fibre reached a fully hydrated state, both the crystalline and amorphous regions lost their stability. A similar increase in the normalised adhesion/radius in crystalline and amorphous regions was observed, as shown in Fig. 11b. When increasing the RH to wet conditions, the crystalline regions had a higher value in the normalised adhesion/radius than the amorphous regions. However, this deviation lies between the error bars and is therefore not interpreted as being distinct. With the increasing amount of water molecules inside the hydrophilic LP fibre, attractive capillary forces accumulate between the AFM tip and the LP fibre surface. Because of the softening the fibre, i.e., the decreasing DMT modulus, the tip was indented deeper into the surface, which led to the requirement of a higher reset force of the tip and therefore to a higher measured adhesion. The EP fibres consist of a more complex system than the LP fibres. While LP fibres possess a cellulose percentage of 95%, EP fibres also have portions of hemicellulose and lignin in their system (Mather and Wardman 2015; Young and Rowell 1986). Cellulose is embedded in an amorphous matrix of hemicellulose and lignin. Hemicellulose is bonded via hydrogen bonds to cellulose and acts as a cementing matrix. Additionally, hemicellulose is not crystalline and has a hydrophilic character. Lignin acts in EP fibres as a hydrophobic network, which increases the stiffness and is totally amorphous. Figure 11a shows a decrease in EP fibres in the normalised DMT modulus when increasing the RH. At 40% RH, the percentage of decrease in the normalised DMT modulus is similar to that of amorphous cellulose. Since the EP fibre mostly consists of noncrystalline parts, the fibre softens in large parts due to infusing water molecules breaking the hydrogen bonds between cellulose molecules. In contrast to simple amorphous cellulose, the EP fibres showed a linear decrease in the normalised DMT modulus when the RH was further increased to a wet condition, as indicated in Fig. 11a by the green line. This linear decrease was assumed to be caused by the lignin network. Since lignin has a hydrophobic network, the swelling of fibres and therefore the related decrease in the normalised DMT modulus was restricted. The green line in Fig. 11b shows an increase in the normalised adhesion/radius with an increasing RH for EP. Comparing these values to crystalline and amorphous cellulose, the increase is minor. It was interpreted that the more amorphous cellulose and its component parts are, the lower the increase in adhesion with increasing RH. Additionally, it was assumed that the hydrophobic lignin network reduced the attractive capillary forces between the tip and the EP fibre surface. Therefore, the adhesion increases in EP fibres were not as distinctive as those in LP fibres with amorphous and crystalline cellulose.
Figure 12 shows the AFM images of a PCEP fibre. As seen in Fig. 12a, for 2% RH, and in Fig. 12b, for the wet condition, there was no significant swelling of the microfibrils of the fibre in the topography images. This is interpreted as the first indicator of a hydrophobic polymer coating. By checking the mechanical properties such as the DMT modulus and adhesion in Fig. 12d,f for 2% RH and Fig. 12e,g for the wet condition, they also revealed no significant changes when increasing the RH. Cross sections of the red framed areas in Fig. 12a-g are shown in Fig. 12h and i. These trends are pointed out in Fig. 11b, where the PCEP fibres are represented by the dashed green line. Even under wet conditions, the normalised adhesion/radius in PCEP fibres is less than that in plain EP fibres. The normalised DMT modulus of the PCEP fibres in 11a with the dashed green line shows almost no change within the error and therefore differs clearly from the EP fibres. This is interpreted as a strong indicator of the hydrophobicity of the polymer coating. The PCLP fibres are represented by dashed red lines in Fig. 11. Similar to the PCEP fibres, the PCLP fibres also exhibited almost no decrease in the normalised DMT modulus and no increase in the normalised adhesion/radius with increasing RH. The changes in the normalised DMT modulus due to the hydration recorded for the polymer-coated fibres were comparable to the values measured during the indentation experiments. Thus, it is concluded that the presence of a homogenous and stable hydrophobic polymer coating can be inferred from AFM measurements. Unmodified fibres exhibit large changes in mechanical properties, such as the DMT modulus and adhesion, as well as the swelling of microfibrils in topographic images. Hydrophobic coated paper fibres show small changes in mechanical properties and almost no swelling in topographical images when the RH was increased.