Biocomposites of polyamide 4.10 and surface modified microfibrillated cellulose (MFC): influence of processing parameters on structure and thermomechanical properties

The FTIR-ATR spectra of acetylated MFC show new peaks not observed in the spectrum of raw MFC. Two peaks at 1733 cm⁻¹, that are related to the stretching vibrations of carbonyl group (ν C=O), and a peak at 1249 cm⁻¹ due to stretching vibrations of C–O bond in ester group, clearly demonstrate partial esterification of MFC (Fig. 1). The intensity of peaks is not very strong and indicates low substitution degree of cellulose hydroxyl groups, probably limited to the most reactive hydroxyl groups at C2 and C6 in cellulose mer (Eyley and Thielemans 2014). Strong peaks at 1063 and 1160 cm⁻¹ are related to C–O and C–O–C stretching vibration in cellulose, respectively (Fengel 1993; Gierlinger et al. 2006). Another broad peak at 1644 cm⁻¹ was attributed to the bending vibrations of adsorbed water.

The band appearing between 3700 and 3100 cm⁻¹, which is derived from OH-stretching vibrations, comprises several distinct bands related to cellulose hydroxyl groups forming inter- and intra-molecular hydrogen bonds, free cellulose hydroxyl groups and absorbed water (Hinterstoisser and Salmén 2000; Ali et al. 2001).

The morphology of raw MFC and MFC-Ac was analysed by SEM (Figs. 2, 3). Raw MFC is comprised of anizometric cellulose particles with a wide range of diameters from 1 to 30 µm and low aspect ratio. The chemical modification resulted in major change in material morphology. Microfibers with submicronic diameter and complex morphology exposing individual cellulose crystals became dominant fraction of material. Particles evidenced features of microcrystalline cellulose but nanocrystals were kept in the fibrillar arrangement. Changes in MFC morphology could arise as simultaneous action of chemical process of hydrolysis in acidic environment and physical disintegration, especially of larger particles, during repeated sonication.

Microscopic observations provided only qualitative assessment of the materials’ morphology. SEM microscopy did not allow to determine the reliable values of the fibrils length and diameters distribution by picture analysis due to the entanglement of fibrils. The morphological features of MFC Solex and MFC-Ac samples were also studied by the dynamic light scattering (DLS) method. The results of DLS analysis are presented in Fig. 4.

The dispersions of both unmodified (MFC Solex) and acetylated MFC (MFC-Ac) in water exhibited bimodal size distribution with high contribution of particles with the hydrodynamic diameters around few micrometers. The size distribution profile of acetylated MFC indicated a slight shift towards lower sizes for both fractions. Despite of high demand, the standard methods or validated techniques for the size evaluation of polydisperse nanomaterials with a high aspect ratio, including cellulose nanocrystals and nanofibrils, are still under development (Gamelas et al. 2015; Fraschini et al. 2014). In the very recent work by Gamelas et al. the DLS hydrodynamic diameter, as an equivalent spherical diameter, was used to estimate the nanofibrils length assuming a cylinder with the same volume and with the diameter (width) assessed by AFM (Gamelas et al. 2015). In the work by Fraschini et al. with the anizometric cellulose microfibrils or whiskers, the hydrodynamic radius determined by DLS was considered only as an indicative measure of the ‘apparent’ size of the dynamic hydrated/solvated particle (Fraschini et al. 2014). It was also pointed that in the case of rod-like particles, such as cellulose nanocrystals, a slight change in the length will directly affect the measured DLS size while changes in the cross-section will hardly affect the diffusion velocity. In other words, the DLS method would rather provide information on the length distribution of the nanocellulose fibers and is significantly less sensitive to the diameter variations that were observed on SEM microphotographs. Besides, several other significant limitations of the DLS method must be taken in consideration when interpreting the results. Most important are possible agglomeration of nanosized particles and relatively high content of micrometric agglomerates that may produce multiple scattering. The MFC microparticles might origin from a raw material as well as arise due to agglomeration of nanofibrils during drying or due to hydrogen bonding in the water suspension. Interestingly, Builes and co-workers demonstrated by that by adding an effective surfactant to the water suspension of microfibrillated cellulose the particles diameters, determined by DLS method, were significantly reduced (Builes et al. 2013). The authors explained this phenomenon by surfactant-MFC interactions that prevented formation of larger agglomerates in water environment.

SEM analysis of the MFC indicated significant decrease of particles diameters in the course of chemical treatment while the fiber length was basically preserved, as determined by the DLS method. However, agglomerates displayed by SEM and DLS analysis in the filler were broken into smaller individuals under high shears during melt blending of PA4.10/MFC-Ac composites, as it is discussed in the following part of the study.

The influence of chemical modification of MFC by heterogeneous acetylation on crystal structure of fibrils was estimated by WAXD method. Diffraction patterns of raw MFC, MFC SolEx, and MFC-Ac are collected in Fig. 5.

Crystallinity degree of cellulosic materials was determined according to Segal method (Segal et al. 1959) based on the following equation:where X _CR—crystallinity index (%); I ₀₀₂—intensity of diffraction peak in the range of 22°–24° corresponding to 002 crystal plane, I _AM—intensity of the minimum at 2θ = 18°, which is approximated value of amorphous halo.

The highest crystallinity degree was found in MFC after solvent exchange process (Table 2). This finding indicates extraction of non-crystalline components from the raw MFC material, such as other carbohydrates that naturally occur in the wood pulp, whereby less contaminated crystalline cellulose fibrils remain after repetitive washing. Acetic anhydride action caused only a slight decrease in MFC crystallinity degree as compared to MFC-SolEx. WAXD and FTIR data indicate that the esterification reaction occurred in amorphous regions of microfibrils and on the surface of cellulose crystals, while the core of crystallites remained unchanged. Literature survey on chemical modifications of cellulose nanocrystals shows that relatively small loss of crystallinity can be achieved at moderate concentration of modifier and limited time of reaction (Eyley and Thielemans 2014). Significant damage to the crystal structure for esterified cellulose nanocrystals can be expected at high surface substitution degree (DS_surf) of 1.5 (Braun and Dorgan 2009). The preservation of MFC crystalline structure during modification is crucial for maintaining its reinforcing potential, as well as barrier properties.

The values of X _CR obtained for MFC materials used in the course of this work are similar to that reported in literature, for example for nanocellulose extracted from sisal fibers (Morán et al. 2008). The X-ray diffraction analysis methods, including Segal method, are convenient and widely used for estimation of crystallinity index (CI) of cellulose micro- and nanoparticles (Terinte et al. 2011). However, the simplified character of this method (Park et al. 2010), the influence of disorder in cellulose fibrils (Lindner et al. 2015) and different crystallite dimensions (Leppänen et al. 2009; French and Cintrón 2013) have recently been pointed as sources of significant errors. Nevertheless, Segal’s method is applicable for empirical measurements that allow direct comparison of cellulose-based materials and determination of trends.

The process of thermooxidative degradation of MFC fillers was evaluated by thermogravimetric method. At temperatures below 120 °C the mass loss was due to moisture vaporisation. The initial temperature of thermooxidative degradation for unmodified microfibrillated cellulose was ca. 277 °C, which is within the processing temperature range of polyamides. Thus, successive preparation of composite materials based on PA with raw MFC by melt blending is impossible. Serious degradation of cellulose-based fillers should be expected especially when polyamide 4.10 is applied as a composite matrix due to its relatively high melting temperature. Solvent exchange procedure itself induced significant changes in the thermooxidative stability of MFC, especially in the initial stage of the decomposition process, which is crucial for processability of filler and preservation its beneficial mechanical and barrier properties during melt blending. For MFC SolEx one can observe an increase of thermooxidative stability displayed as an increase in the onset temperature of degradation by over 35 °C and temperature at 10 % mass loss by about 11 °C (Table 3). Surface modification of MFC by acetylation caused further slight increase of the initial degradation temperature. The degradation process of all three MFC occurred in two stages. In the main degradation step (I) the temperatures at maximum rate of mass loss were not shifted by the physical processing and only slightly increased by the chemical modification. However, the overall course of the decomposition in this stage was visibly altered. The degradation of solvent exchanged and acetylated MFC was rapidly accelerated in a narrow temperature range as displayed in the DTG curves (Fig. 6). The second stage of degradation proceeded similarly in both modified samples but was shifted to higher temperatures.

The observed substantial changes in the thermooxidative degradation of MFC after solvent exchange procedure is most probably due to extraction of carbohydrate components (with higher susceptibility to thermooxidative degradation) from the raw product by repeating washing. The raw cellulose pulp used for MFC production is a complex material with cellulose as a main component, but containing also lignin, hemicelluloses and in some cases pectins (Dufresne 2012). Pure crystalline cellulose is expected to present higher thermal stability due to regular crystalline structure and lack of less stable functional groups such as carboxyl groups prone to decarboxylation.

The attempt to obtain a PA composite material with unmodified raw MFC failed due to severe degradation of components but compounding of PA4.10 with MFC after complete solvent exchange procedure and acetylated MFC allowed production of composite material suitable for further tests. Moreover, surface modification clearly enhanced the dispersibility of MFC in PA4.10 matrix and improved the appearance quality of the extruded rods, as evidenced by optical microscope photographs taken in a reflected light mode (Fig. 7).

The microstructure of PA4.10 surface showed typical brittle fracture of polymer material without distinctive structures (Fig. 8). SEM pictures of cryofractured profiles of PA4.10/MFC-Ac-3 % composite produced at different screw speeds are presented in Fig. 9. Both cellulose filler and polyamide matrix are organic polymers and comprise elements having similar atomic mass. Thus, under SEM observation MFC is not very well visible phase. However, differences in carbon and oxygen content between components as well as presence of nitrogen atom in polymer matrix produce sufficient material contrast and allow analysis of composites structure and morphology. Microfibrillated cellulose after modification was mainly composed of rigid linear particles which after processing could be visible in PA matrix as straight whisker-like fibers with regular boundaries. This was a specific morphological feature distinguishing MFC particles from the less regular cracks present in the polyamide matrix.

At screw speed of 50 rpm the microfibrillated cellulose was not sufficiently dispersed in the PA matrix and numerous agglomerates with sizes up to 80 µm were observed (Fig. 9). Also small voids/bubbles were formed as gases could be entrained in the cellulose powder and were not sufficiently evacuated from the extruder under moderate melt pressure generated at low rotational screw speed. Moreover, moisture and volatile products of degradation could be evolved from both polymeric components during melt blending. At higher screw speed one can observe much better dispersion and distribution of micro/nanocellulose particles. Some of them were depicted with arrows in the Fig. 9. Agglomerates and other defects are rather rare in PA/MFC-Ac-3 %_100 material but gas bubbles are more likely to occur in material processed at 150 rpm. Particles of acetylated MFC observed on the fracture surfaces are partially covered with PA which indicates rather good interfacial adhesion.

At higher content of MFC-Ac (5 wt%) the defects, especially gas bubbles, were more frequently observed (Fig. 10). The apparent density of microfibrillated cellulose aerogel is very low and it comprises mainly air which can easily be captured by polymer melt. Processing of such materials by melt blending requires effective degassing system. Larger number of cavities and bubbles might be also indicative of higher extent of thermomechanical degradation corresponding to increased content of thermally sensitive material. The least defects were observed in composite containing 3 wt% and prepared at screws speed of 100 rpm. However, the dispersion and equal distribution of individual MFC particles increased with the screw speed. The final mechanical and barrier performance of biopolyamide/MFC composites is expected to depend on homogenization degree of submicronic filler and the presence of morphological defects of different origin.

SEM microphotographs of a series of PA/MFC-Ac 1 % composites (not shown here) reveal the presence of voids and agglomerates, however, there are less individual MFC particles due to small amount of additive.

Generally, at screw speed of 50 rpm the mixing intensity and backpressure were not sufficient for producing bio-PA/MFC composites with good quality. The higher screw speed was applied the better dispersion degree of MFC in PA4.10 matrix was achieved. The highest number of cavities and gas bubbles was observed at screw speed of 50 rpm. The smaller number of defects was revealed in composites produced at 100 rpm but a further increase of the screws speed up to 150 rpm again resulted in generating of considerable number of bubbles. Defects observed at the highest screws speed resulted probably from local overheating of the melt and thermomechanical degradation of polymers during processing.

Taking into account morphological results one can indicate that processing of temperature sensitive cellulosic filler with bio-PA engineering polymers is a challenging issue, however, it was proved to be possible if proper modification procedure is applied. During the processing of PA4.10/MFC composites the highest possible intensity of mixing that promotes the dispersion of MFC in PA matrix, had to be compromised with the control of excessive heat released due to friction. Further adjustment and control of temperature providing efficient plasticization, homogenization and preventing of material degradation at high throughput can be achieved on the semi-industrial extrusion lines usually equipped with multiple heating and cooling zones and vacuum degassing systems. With growing melt viscosity of the polymer the risk of overheating during processing is higher as a result of more intensive shearing. Thus, melt blending of thermally sensitive MFC or NFC with low viscosity engineering resins would be easier to control. Lower viscosity of polymers and lower molecular weight was reported to favour the dispersion of nanoadditives and this effect was explained by faster melt infiltration by primary agglomerates (Kasaliwal et al. 2011a, b). Therefore different viscosity grades of PA4.10 would be required for industrial applications, such as automotive industry (injection moulded parts, for example engine covers) and packaging sector (extrudable and film forming grades for plastic films with improved barrier properties and balanced CO₂ index).

The FTIR spectra of PA4.10/MFC-Ac nanocomposites prepared under different processing conditions with varying MFC content are presented in Figs. 11 and 12. Due to low content of the filler the infrared spectra of composite materials do not differ much from that taken for the pristine polyamide 4.10. However, several additional bands and changes in intensity of existing bands have been observed in composites spectra. Hence, with the increasing content of MFC-Ac a gradual increase of absorption intensity in carbonyl region at 1742 cm⁻¹ was observed. Therefore, this band was correlated with the carbonyl stretching vibrations in cellulose acetate at 1734 cm⁻¹ (see Fig. 1). Other absorption bands associated with the modified cellulose in biopolyamide matrix were observed at 1034, 1060 and 1113 cm⁻¹, especially in spectra of nanocomposites containing 5 wt% of filler. The bands at 1007 and 1046 cm⁻¹ in the PA4.10 (Figs. 11, 12) are in the spectral range of C–C skeletal stretching vibrations of aliphatic polyamides (Luo et al. 2011). In polyamide 6.10, a homolog to PA4.10, typical absorption bands occur at 1480, 1245 and 940 cm⁻¹ (Ogava and Fujiwara 1985). The strong absorption band at 3299 cm⁻¹, due to stretching vibrations of N–H bound in polyamide, overlaps the stretching vibrations of hydroxyl groups in cellulose. Therefore no significant change can be detected in that region. However, we observed some weak bands which do not overlap strictly with the absorption maxima of pure components. In carbonyl region one can find additional peaks at 1711 and 1729 cm⁻¹. There is probably a shift of absorption frequency for C=O group either in acetylated cellulose or in polyamide due to formation of specific interactions between the components. Polyamide macrochains can form hydrogen bonds with cellulose in the interfacial layer. However, the fraction of filler macromolecules located on the interface is probably not large with respect to total mass of the filler so the polymer matrix-polymer additive interactions are rather weak. Strong changes in FTIR spectra were observed in polyamide-based systems containing low molecular weight additives capable to overcome high cohesion energy E_c of polyamides by destroying its hydrogen bonding and forming interactions with amide groups on a molecular level (Roberts and Jenekhe 1991; Afshari et al. 2008).

On the spectra of PA4.10/MFC-Ac composites some small additional bands contributed to C–C skeletal absorption region at 1020 and 1049 cm⁻¹ can be observed. Changes in the spectra could also reflect formation of additional carbonyl compounds as a result of thermomechanical degradation or oxidation of components during melt blending, as both biopolymers are susceptible to oxidation. Generally, raw materials used in this work are biopolymers with relatively high degree of crystallinity and a major part of biopolyamide as well as cellulose macromolecules form crystals with intra- and inter-molecular H-bonds.

Bio-PA/MFC composites containing different amounts of MFC-Ac and manufactured in different processing conditions displayed characteristic changes in the melting and crystallization behaviour as evidenced by DSC analysis (Table 4).

A slight decrease of melting temperature was observed with an increase of MFC content (Fig. 13). This effect indicates the formation of smaller and less perfect crystallites of PA4.10 in the presence of MFC-Ac. PA4.10/MFC-Ac materials are two phase systems comprising two crystalline polymers and thermal behaviour of polyamide matrix is influenced by the presence of another polymer—a filler. In polymer blends similar effects were interpreted as a result of H-bond formation between polymeric compounds. Higher order interactions increase miscibility of blend components but interfere with the formation of perfect crystals (Zhu et al. 2004; He et al. 2001a, b). A decrease of melting temperature T _m in semicrystalline polymers brings important information about miscibility and its associated with polymer–polymer interaction parameter χ₁₂ (He et al. 2004). It was explained by thermodynamic depression related to a decrease in the chemical potential due to the presence of the polymeric solvent. When two polymers are miscible in the melt, the chemical potential of the crystallizable polymer is decreased due to the addition of the second component. Melting point depression was typically observed in hydrogen-bonded blends if the blend contained a semicrystalline component (Li et al. 2001; He et al. 2001a, b). The degree of mutual interactions of both components in PA4.10/MFC-Ac depends on the interfacial surface arising from the dispersion level. Melting of PA4.10 in composite systems subjected to higher shearing at rotational screw speed of 150 rpm occurred in slightly wider temperature range. This effect may also origin from an increase of molecular weight distribution due to processing operations.

Crystallization temperature of PA4.10 (Fig. 14) was slightly shifted towards higher temperatures by addition of MFC-Ac indicating heterogeneous nucleation of crystallization induced by cellulose filler. The least overcooling to undergo crystallization was necessary for composites with 5 wt% of MFC-AC. The onset temperature of crystallization showed minor changes only for selected composites. The crystallinity degree (Table 4) of polyamide in composites was calculated according to the formula:where X _cDSC —crystallinity degree of polyamide calculated from the DSC measurements, ΔH _m —measured enthalpy of melting, ϕ—mass content of filler (Torbati et al. 2014; Virtanen et al. 2014).

Since the enthalpy of fusion of 100 % crystalline PA4.10 is not reported in literature the enthalpy of fusion of 100 % crystalline PA6.10 (ΔH _m ⁰  = 197 J/g) was taken (Elzein et al. 2002). It is noteworthy that the heat of fusion for the totally crystalline PA6.6 was reported as ΔH _m ⁰  = 191.0 J/g (Haberkorn et al. 1979). We assume thus that ΔH _m ⁰ of PA4.10 is close to the values determined for its homologous aliphatic polyamides. Crystallinity degree of PA4.10 in composite was comparable to that determined for neat sample. Subtle variations were observed that are not expected to influence the final properties of composites.

Dynamic mechanical analysis was used to evaluate viscoelastic properties of polyamide 4.10/MFC-Ac composites as a function of temperature. The mechanical properties of composites showed dependence on the processing conditions that influenced structure and morphology of composites. Storage modulus E′ of all composite materials was higher than that of pure polymer in the whole temperature range (Fig. 15). For a series obtained at 100 rpm, the E′ modulus gradually increased with larger filler content. The highest increase of the storage modulus was observed for composites prepared at the highest applied rotational screw speed, i.e. 150 rpm (Fig. 16). The measured storage modulus of composites reflected the dispersion state of MFC and the presence of structural defects. Blending MFC at only 50 rpm did not provide good composite structure and developed numerous cavities. These structural defects led to deterioration of mechanical properties as compared to reference biopolyamide. Importantly, a higher increase of E′ was observed below glass transition temperature, i.e. in the temperature range of most polyamide applications. For example, the storage modulus at 25 °C increased from 1934 GPa for PA4_150 to 2199 GPa for PA/MFC-Ac 3 %_150 and 2130 GPa for PA/MFC-Ac 5 %_150 sample. On the contrary to glassy state of material, the rubbery plateau of composites showed moderate improvement of storage modulus. The mechanical performance of polymer/MFC nanocomposites can be influenced by the crystalline phase of polymer matrix. For example, Suryanegara et al. obtained clearly higher values of tensile modulus and tensile strength of crystalline PLA/microfibrillated cellulose composites as compared to that of amorphous PLA/MFC system (Suryanegara et al. 2009). Moreover, composites based on crystalline polymer matrix exhibited higher storage modulus in the DMA tests at elevated temperatures over the glass transition of PLA. Elsewhere Braun et al. demonstrated the ability of acetylated cellulose nanowhiskers to act as nucleants for PLA crystallization. The reduction of the crystallization half-time as well as increase of the crystallinity degree of PLA was reported (Braun et al. 2012). In the work by Siqueira et al. the crystallinity of PCL was increased by addition of sisal whiskers modified with n-octadecyl isocyanate while it was almost not influenced by the presence of microfibrillated cellulose bearing the same functional groups (Siqueira et al. 2009). Interestingly, although the crystallinity of MFC-based nanocomposite films was lower as compared to whiskers-based composites, the stiffness of the former was higher. This comparative study indicated the prevalent role of the entanglement in strengthening of semicrystalline polymers by cellulosic nanoparticles. In our work little variations were observed in the crystallinity degree of PA4.10 in all nanocomposites as compared to unmodified polymer (Table 4). Thus, the improvement of mechanical performance of nanocomposites can be ascribed to the reinforcement of PA4.10 by crystalline cellulose nanofibrils with negligible contribution of self-reinforcement by crystalline phase of polyamide.

The dependence of loss tangent versus temperature (Fig. 16) demonstrates the existence of three secondary relaxations, α, β and γ. Glass transition temperature (T _g ) of polyamide can be determined by α relaxation, since it is usually related to the segment movements in the amorphous region (Wang and Zhang 2009). The β relaxation is ascribed to the mobility of the carbonyl group in the amorphous region and the γ relaxation is believed to reflect co-movement of amide and methane groups.

The α transition temperature was slightly affected by the addition of MFC-Ac and by the processing conditions (Fig. 16). The T_g determined for pure PA4.10_100 and PA4.10_150 did not change much and it was 58.5 and 57.7 °C, respectively. For the composites with up to 3 wt% content of MFC-Ac processed at 100 rpm the T_g value was not influenced as compared to the extruded PA4.10 and oscillated around 58 °C. However, for the composites that were processed at higher screw speed (150 rpm), the α transition temperature was slightly shifted towards lower temperatures as compared to unmodified PA4.10. For the composites containing 5 wt% of MFC-Ac the α transition temperatures were lowered by ca. 3 °C regardless of the applied rotational screw speed. In these materials larger interfacial surface might be produced due to higher content of the filler. Complex physical processes with opposite effects could contribute to the final performance of relaxation phenomena in the polymer composites. Firstly, the observed changes could result from the plasticizing effect of acetyl moieties which were located on the surface of modified MFC. For example, the literature reported plasticizing of PLA by surfactant adsorbed on cellulose nanocrystals (Fortunati et al. 2012a), cellulose nanowhiskers grafted by n-octadecyl-isocyanate (Espino-Pérez et al. 2013) and lactide oligomers attached to cellulose nanowhiskers (Goffin et al. 2011). The effect of increased storage modulus and decreased glass transition temperature was found, for example, for PLA/nanocrystalline cellulose nanocomposites by Fortunati et al. (2012b). On the other hand, the polymer-filler interactions are known to have opposite effect of rendering the α relaxation of polymer matrix and shifting it towards higher temperatures (Ahn et al. 2009; Papon et al. 2012; Raftopoulos and Pielichowski 2015). Slight increase of T_g was reported for instance for acetylated cellulose nanoparticles blended with PLA (Abdulkhania et al. 2014; Lin et al. 2011). Effects related to the interfacial area depend on the filler content and degree of dispersion. More distinct changes in T_g, which were observed for 5 wt% content of MFC-Ac and nanocomposites processed at higher screw speed, probably reflect higher surface area of polymer-nanofiller interface. Lowering of T_g might also indicate more pronounced changes in the molecular weight of polymer matrix exposed to high shearing forces as compared to that processed under moderate conditions. The increase of the melt viscosity of composites in relation to pristine polymers is a typical phenomenon that makes the polymeric components more prone to thermomechanical degradation. Therefore stronger changes in molecular weight of PA4.10 might be observed for composites with 5 wt% of MFC-Ac. However, the decrease of T_g with lowering polymer molecular weight, expressed by Fox–Flory equation (Gauthier 1995), is not large in the range of high molecular weight of polymer, which is the case for the commercial engineering polymers. However, a strong decrease of T_g versus mechanical energy that was delivered to the unit mass of extruded material, as well as lowering of the average molecular weight, were reported for the amylose and amylopectin in corn flour. The distinct effect rose from extrusion-induced mechanical degradation of biopolymers (Kaletunc and Breslauer 1993). Prolonged heating applied during extrusion of PA4.10/MFC-Ac nanocomposite with melt recycle is another factor considered as a possible reason of matrix degradation. Significant reduction of molecular weight upon prolonged heating was reported for e.g. PLA copolymers (Garlotta 2001). Moreover, low molecular weight products formed during the thermo-mechanical degradation of polyamide and/or cellulose might dilute in the polymer matrix, leading eventually to the plasticization of polyamide matrix. The values of polyamide’s T_g is also sensitive to water content which in polyamides is known to act as a plasticizer—water shifts the T_g towards lower temperatures (Murthy 2006). On the other hand, molecular orientation during bio-PA processing can also influence the glass transition temperature and, as a result, the oriented segments of polyamide relax at a higher temperature.

tgδ diagrams of all materials under investigation display a small shoulder at ca. −15 °C. Similar small transition peak was observed between α and β peaks of PA11.11 and was defined as β_c relaxation related to the absorbed water (Liu and Zhao 2014).