Microstructure and density of porous materials
Figure 1 shows the macroscopic optical appearance of all variants produced. While freeze-dried pure MFC, being produced from bleached pulp, is brightly white, its counterpart MFLC is of brown colour due to its significant content of residual lignin. For both variants, increasing content of polymerised furfuryl alcohol manifests itself in progressively darker grades of brown.
Scanning electron microscopy of freeze-dried MFC and MFLC reveals an open-porous structure for both fibril variants with no obvious differences in terms of characteristic fibril size or morphology (Fig. 2). Microporous structures as often described for lightweight materials produced by freeze-drying of microfibrillated cellulose (Josset et al. 2017; Lee and Deng 2011) were present with characteristic pore diameter of approx. 50 µm, but not very well defined in the structures observed. At higher magnifications, residual non-cellulosic cell wall constituents are clearly apparent as amorphous matrix covering the cellulose fibrils present in MFLC (Fig. 2d).
In the FA-modified MFC specimens, cellulose fibrils were abundantly decorated with microspheres with a typical diameter between one and two microns (Fig. 3a, c). It is proposed that these microspheres are PFA, which polymerised during specimen preparation according to reaction 1 (Fig. 4), but apparently lacks wetting and spreading onto MFC. Quite contrarily, MFLC specimens only sporadically show microspheres (Fig. 3b, d). Here, the cellulose fibrillary network shows a denser and much less porous character than the open porous structure seen in MFC. It is suggested that this denser and less porous structure of MFLC specimens is caused by the wetting and infiltration of cellulose fibrils by FA, providing a significant reinforcement effect upon in situ polymerisation. Considering the furanic ring structure present in FA, wetting of lignin-containing microfibrillated cellulose, which was shown to provide considerably improved miscibility with non-polar solvents and polymers compared to MFC produced from bleached pulp (Ballner et al. 2016; Herzele et al. 2016; Winter et al. 2017; Yan et al. 2016), seems highly plausible for furfuryl alcohol. Apart from better wetting and interpenetration, the FA monomer may also directly react with lignin units found in MFLC according to reaction 2 (Fig. 4), ultimately resulting in a covalent bond between the furan polymer and the lignocellulosic fibre.
Since the content of fibrils in the slurry used for production of porous materials was kept constant, addition of furfuryl alcohol resulted in an overall increase in effective solids content of the slurry and, consequently, also in the density of freeze-dried specimens as shown in Fig. 5a and b. The resulting average densities range from minimum values of 40 kg m−3 for variants with no or only small amounts of FA added, up to 85 kg m−3 for the variants with highest FA content. However, the back-calculated density of cellulose fibrils material alone is much less variable (Fig. 5c and d) and rather constant irrespective of the amount of FA added. Thus, the two sets of samples contain roughly the same amount of cellulose and variable amounts of PFA.
IR-spectroscopy and thermogravimetric analysis
ATR FT-IR spectra of porous MFC and MFLC specimens with varying amounts of FA are shown in Fig. 6. Overall, the spectral features of (ligno-)cellulose dominate as opposed to PFA-specific absorption bands. However, at higher FA contents, the absorption spectra of the compounds typically start approaching the spectrum of pure PFA. This is most obvious at 1715 cm−1 where both the PFA and the compounds show a pronounced peak which clearly increases with increasing amount of FA added to the reaction slurry. According to Pranger and Tannenbaum (2008), this peak is assigned to C=O stretching of γ-diketones formed by hydrolytic ring opening of some of the furan rings along the PFA chain. However, these furan ring opening reactions may be considered side reactions leading to intermediate products that do not influence the overall rate of polymerization (Conley and Metil 1963). The peaks at 1615, 1560 and 1507 cm−1 present in the spectrum of PFA are typical to furanics and can be assigned to C=C stretching vibrations in aromatic compounds (Tondi et al. 2015). While compounds with a low FA content do not show significant absorption in this area, clear peaks may be discerned for the compounds containing 33 and 50% FA. The peak at 1507 cm−1 is associated to asymmetric stretching of aromatic C=C–H groups (Tondi et al. 2015) which are, apart from furanics, also found in phenolic lignin compounds. Since this peak is lacking for pure MFC but clearly appears in MFLC, this indicates the presence of residual lignin in the latter case. Between 1450 and 1150 cm−1 the compounds’ absorption gradually increases from MF(L)C to PFA according to the FA content. The pronounced peak at 780 cm−1 in the PFA spectrum is associated to C–H bending of conjugated polyheteroaromatic furan rings (Tondi et al. 2015) and appears also in compounds containing 33 and 50% FA.
The thermal stability of all variants evaluated by means of TGA shows clear changes correlated with increasing content of polymerized FA in the porous structures (Fig. 7). Pure MFC and MFLC both show a typical sigmoidal decomposition curve with an abrupt decomposition starting at an onset temperature of around 320 °C for MFC and 342 °C for MFLC. Thus, pure MFLC showed a slightly higher thermal stability compared to pure MFC which may be explained by differences in cellulose crystallinity (crystallinity index = 0.70 for MFC and 0.74 for MFLC) and/or degree of polymerisation due to different sources and processing conditions. With increasing percentage of FA, the shape of the curves changes to a more gradual decomposition behaviour typical to pure PFA resin. The latter also showed a remarkably high thermal stability with a residual weight of around 45% after heating to 800 °C in air atmosphere. Compounds with a higher FA content initially show a slight mass loss between 100 and 175 °C which is attributed to the evaporation of residual monomeric FA. These specimens also showed a substantial mass loss prior to the degradation of cellulose but retained more weight at temperatures above 400 °C. This can be quantified by comparing the temperatures at which the samples still retain 90% of their initial weight. This temperature equals around 315 °C for pure (ligno)cellulose and declines to about 240 °C for specimens containing 50% FA. Subsequently, pure MF(L)C samples undergo a rapid decomposition and the curves drop to around 18% remaining mass while 50% FA samples retained about 33% of their initial mass right after cellulose degradation occurring at around 400 °C. Differences in the decomposition rate of individual PFA compounds are also clearly obvious from DTG curves (Fig. 7b and d). The decomposition of pure PFA occurs quite gradually resulting in a broad decomposition range. According to Guigo et al. (2009), scission of PFA chains starts at temperatures around 200 °C and reaches local maxima at around 350 °C and 430 °C attributable to scission of methylene and methyne linkages as well as scission of the furan ring together with continuation of methylene scission, respectively. In the present study, the DTG curves of both pure PFA and compounds containing at least 33% FA show a clear shoulder at temperatures above 400 °C which may be attributed to the scission of furan rings as mentioned above. Unlike pure PFA, all compounds containing cellulose fibres show a sharp decomposition peak. It can be noted that the higher the FA content of the compound, the lower its thermal decomposition rate and the higher the release of unreacted FA at temperatures around 150 °C.
From a general view point of mechanics of cellular solids (Ali and Gibson 2013), which is in good agreement with numerous experimental results (Ago et al. 2016; Donius et al. 2014; Jimenez-Saelices et al. 2017; Josset et al. 2017; Sehaqui et al. 2010; 2011; Svagan et al. 2011), the compression strength and stiffness of porous MFC materials is expected to show a positive correlation with increasing density (Fig. 8).
In the present study, no such clear trend was observed neither for strength nor for the modulus of elasticity (Fig. 9). Both the modulus of elasticity and the yield stress derived from compression tests show high variability, spanning a range of values from 1000 to 10,000 kPa and 50–200 kPa, respectively. However, the systematic increase in compound density with increasing furfuryl alcohol content (Fig. 5) does not result in a corresponding linear increase in mechanical performance, as one would expect based on foam mechanics (Ali and Gibson 2013). A more detailed analysis of the results from mechanical testing, considering different amounts of PFA present in the porous material structures, is shown in Fig. 10.
Pure MFC specimens show a modulus of elasticity of 2435 ± 750 kPa and a yield stress of 98 ± 29 kPa. By comparison, pure MFLC shows lower performance with 2005 ± 712 kPa for the modulus and 83 ± 25 kPa for yield stress, respectively, in spite of slightly higher density values for the latter material (Fig. 5). Addition of FA has different effects on the two variants of fibrillated material used. For MFC, addition of FA results in an increase in variability of the modulus of elasticity, and a trend towards higher values. However, no systematic correlation is apparent, or, if present, obscured by high variability. As for yield stress, no significant effect of the presence of PFA in the specimens is observed. By contrast, very clear effects were observed when MFLC instead of MFC was used. Again, a very significant increase in the variability of the modulus of elasticity is observed upon addition of FA. Even so, a positive effect of FA addition on the modulus is obvious. In the variant with highest content of PFA (50% MFLC, 50% PFA), the average modulus of elasticity is 6828 ± 1783 kPa, which corresponds to a roughly threefold increase compared to unreinforced MFLC. With regard to compressive yield stress, effects of FA addition are also evident. In spite of high variability, an initial linear increase of yield stress with increasing amounts of FA added is discernible, reaching maximum values of 177 ± 26 kPa at 33% content of PFA. For the variant with highest PFA content, a decrease in yield stress is observed. To summarise the results shown in Fig. 10 it can be said that the addition of FA has clearly positive effects on the compressive performance of freeze-dried MFLC, whereas this is hardly the case for MFC.
The shape of stress–strain curves recorded during compression testing of MFC and MFLC materials (Fig. 11) provides more insights into the different mechanical behaviour of these materials. Stress strain curves were similar in shape for the pure MFC and MFLC variants, respectively, and also for all PFA-reinforced MFC specimens. Typically, these stress–strain curves were smooth, and an initial quasi-linear section was followed by a transition to a region of further compression at steadily but moderately increasing stress, until stress increased again rapidly in the final densification phase beyond 70% strain. This behaviour is typical of unreinforced foams of fibrillated cellulose (Ali and Gibson 2013; Sehaqui et al. 2010). Quite contrarily, PFA-reinforced MFLC specimens showed a clear first stress maximum after the initial quasi linear elastic region, which was followed by a plateau region of more or less constant stress concluded by the final densification zone at high strain. A similar change in the shape of compression curves from ductile towards brittle fracture is observed when solid wood cell walls are modified with brittle polymer (Gindl et al. 2003).
It is thus proposed that the cellular architecture of freeze-dried MFLC foams is significantly altered by FA addition. SEM images of freeze-dried specimens with FA addition (Fig. 3) confirm this assumption and also indicate a potential mechanism behind the significantly different behaviour of MFC and MFLC observed in this regard.