3.1 Biochar
Prior to the pyrolysis process, the digestate was characterised by thermogravimetric analysis to investigate the thermal degradation path of the waste. Figure 1 shows that digestate degradation at low heating rates occurs over a wide temperature range, which narrows as the heating rate increases. The derivative thermogravimetric (DTG) curves show four different, well-differentiated peaks, suggesting a complex degradation pathway of the digestate waste, which takes place through four different degradation steps. In fact, referring to the DTG curve with a heating rate of 10°C/min, the first DTG peak at about 104.31°C is related to the removal of moisture content and very high volatile components. Furthermore, the other three peaks can be considered as degradation peaks of a lignocellulosic feedstock. This assumption may be reasonable since digestate is derived from the anaerobic digestion of organic matter. Thus, according to the literature, biomass degradation after thermal stress is essentially in three steps: (i) hemicellulose degradation, (ii) lignin and cellulose degradation, and (iii) lignin decomposition [53]. Thus, at a heating rate of 10°C/min, the digestate shows that hemicellulose degradation occurs at 214.78°C, while lignin and cellulose degradation and lignin decomposition occur at 326.71 and 429.92°C, respectively. In accordance with the DTG curves and a previous study [40], a temperature of 400°C was chosen to pyrolyse the waste as a compromise between the need for weight loss and the need to save energy compared to 429.92°C.
Before using biochar as a filler for PBAT:PLA blends, it was ground with a mechanical pestle and mortar and sieved to less than 45 µm. The final size of the BC particles was measured using an ultrasonic granulometer at two different stirrer speeds, 2000 and 3000 rpm. Thanks to this analysis, it was possible to obtain the size distribution curves of the particles, see Fig. 2. The factors d10, d50 and d90, which represent the maximum diameter values of 10%, 50% and 90% of the particles respectively, were calculated at different stirring speeds.
According also to the SEM morphology (see Fig. 3c), almost all the particles have a diameter less than 100 µm: thanks to the d10 factor, it is possible to state that at least 10 v.% of the particles have a diameter less than 5 µm, with many particles less than 1 µm (Fig. 3b) which, after sonication during the Mastersizer measurement, break up from agglomerates to be led in solution. The sieving process was efficient: the maximum of the particle size distribution curves varies from 52.1 µm to 35.4 µm by varying the stirring speed during the measurements from 2000 to 3000 rpm, and certainly 90% of the particles have a diameter smaller than 87.2 µm.
In order to chemically characterise the biochar, ATR-FTIR analysis of the BC was carried out and compared with the digestate prior to pyrolysis, and EDX measurements were carried out. The ATR-FTIR spectra show a reduction of several characteristic lignocellulosic bands, for example: the broad band between 3300–3600 cm− 1 with a maximum at 3288 cm− 1 (νOH, OH stretching vibrations of hydrogen-bonded hydroxyl groups), the two peaks at 2920 and 2948 cm− 1 (asymmetric and symmetric CH stretching vibrations of aliphatic groups), the peak at 1628 cm− 1 (related to aromatic C = C stretching and C = O stretching modes). Instead, there are bands whose intensity increases after the pyrolysis process: the peak at 1420 cm− 1, related to aromatic structural vibrations, the peak at 1026 cm− 1, normally related to C-O stretching, the two peaks at 876 and 778 cm− 1, related to aromatic C-H deformation modes [54–56].
In addition, EDX analysis was carried out to characterise the elemental composition of the BC surface at two different magnifications: 50x and 1000x to evaluate the elemental concentration of BC qualitatively and quantitatively. The sample results are homogeneous, although some element concentrations differ by a few percentage points. However, all atoms with an atomic weight lower than sodium have an intrinsic error in weight determination. Nevertheless, as shown in Fig. 4 and Table 2, several elements have been detected by EDX spectroscopy: from carbon to iron and titanium. It can be stated that BC is mainly composed of carbon and oxygen. Other elements with significant concentrations are: aluminium, silica, phosphorus, potassium and calcium. In this study, for the characterisation of the composites, it is important to highlight the presence of iron in low concentration [57–59]. The high variety and quantity of inorganic contents are attributed to the inhomogeneity of OFMSW and its inorganic components, such as bones and table salt, present in the initial food waste.
Table 2
|
1000x
|
50x
|
Element
|
wt.%
|
wt.%
|
C
|
62.74
|
49.44
|
N
|
4.09
|
3.68
|
O
|
16.23
|
24.99
|
Na
|
0.72
|
0.85
|
Mg
|
0.78
|
1.11
|
Al
|
1.66
|
2.06
|
Si
|
3.84
|
4.37
|
P
|
1.29
|
1.40
|
S
|
0.34
|
0.37
|
Cl
|
0.69
|
0.61
|
K
|
1.12
|
1.38
|
Ca
|
5.49
|
8.30
|
Ti
|
0.15
|
0.24
|
Fe
|
0.83
|
1.20
|
3.2 Composites
The complex viscosity as a function of angular frequency of bio-blend and bio-composites is shown in Fig. 5. In Fig. 5a, the curves show that the original samples (PBAT:PLA 0:100 and PBAT:PLA 100:0) have the same complex viscosity at low frequency, with a more Newtonian behaviour for pure PLA compared to pure PBAT, which shows a more pronounced decrease in complex viscosity with increasing angular frequency. In fact, unlike PLA, PBAT shows a narrower Newtonian range, exhibiting a shear-thinning behaviour already from 1 rad/s to higher angular frequency tested. Accordingly, in blend formulation, the addition of PBAT limits the Newtonian range, with the shear thinning tendency becoming more pronounced with increasing PBAT content. For blends with a predominance of one of the two matrices (PBAT:PLA 75:25 and PBAT:PLA 25:75), the complex viscosity at low frequency is higher than that of the pure matrix, with the appearance of a stress-strain behaviour at low frequency, suggesting a change in morphology. However, PBAT:PLA 50:50 shows a more balanced behaviour between the two matrices. This phenomenon is classical for immiscible polymer blends and depends on the higher shear that occurs between the two phases. In fact, for immiscible polymers, the formation of droplets of the minor phase in the major phase results in an increase of the complex viscosity. For the blend in which PBAT and PLA are equivalent, as visible later in the morphology section, no droplet formation is visible and the morphology results in an elongated co-continuous phase that result responsible of an easier sliding of the polymer chain under shear [60, 61]. Thus, the blends have a higher complex viscosity at low frequency and a pronounced shear thinning behaviour during the frequency range, suggesting a high interaction between the two matrices [5, 23], with a decrease in complex viscosity at higher frequency due to disentanglement of the molecular chain [18, 24].
As shown in Fig. 6, the addition of 10 wt.% BC to virgin PBAT or virgin PLA determines two different rheological behaviours. In fact, as seen in the previous study [40], 10 wt.% BC on PBAT matrix determines a global increase of the complex viscosity, as shown in Fig. 6a. Differently, as shown in Fig. 6b, for pure PLA the addition of 10 wt.% BC completely modifies the rheological behaviour with a significant reduction of the complex viscosity and a stress-strain behaviour at low frequency. This behaviour is usually seen when a plasticiser is added to a polymer matrix or degradation phenomena occur.
Turning to the study of the composite blends, the rheological behaviour depends on the weight ratio between PBAT:PLA to mean the balance between their two antithetic behaviours. Figure 5b shows the rheological behaviour of composite blends compared to PBAT and PLA based composites. At low frequency, the complex viscosity decreases as the PLA content in the blend increases, similar to that of the PLA composite. The presence of BC particles and their interaction with the polymer blend is more visible in the stress-strain behaviour at low frequency when the amount of PLA exceeds the amount of PBAT in the biopolymer blend. Instead, the PBAT:PLA/BC 75:25/10 complex viscosity has a more pronounced angular frequency dependence. The PBAT:PLA/BC 50:50/10 showed both stress-yield behaviour and a pronounced decrease in complex viscosity with increasing angular frequency.
Figure 7 shows the SEM images of the nitrogen-fractured surface for the PBAT:PLA blends with different composition ratios, showing a typical two-phase structure. In accordance with the rheological behaviour, the morphology changes with increasing PLA content in the PBAT:PLA blend. As shown in the first column of Fig. 7, the morphology changes from spherical droplets (PBAT:PLA 75:25) to a co-continuous elongated structure (PBAT:PLA 50:50) and back to spherical droplets [22, 62]. In composite SEM images, the average size of BC particles in biopolymer blends appears to be smaller than the dimension shown in Fig. 3c. This phenomenon is probably due to the shear stress experienced by BC during melt mixing, which also favours a deagglomeration of smaller particles seen in Fig. 3b, which appear homogeneous and well dispersed in the biopolymer blend.
The reduction in particle size that occurred during melt mixing could cause an increase in the polymer/filler interface area, with a consequent increase in the interaction between BC and the biopolymer blend. Also in this case, and in accordance with the rheological behaviour, this interaction led to a change in morphology depending on the PBAT:PLA weight ratio. Concerning the addition of BC to pristine PBAT matrix, a good adhesion and dispersion between biopolymer matrix and BC particles was shown, as already reported in a previous study [40]. When added to pristine PLA matrix, good dispersion is achieved, but a dissection between polymer matrix and BC occurs, with clear gaps surrounding BC particles. Moreover, in the presence of BC, when PLA is the minor phase (PBAT:PLA/BC 75:25/10), its droplets homogeneously distributed in the PBAT matrix change from spherical shape to "donut" shape, with a global reduction of the mean particle diameter. Conversely, when PBAT is the minor phase (PBAT:PLA/BC 25:75/10), the homogeneously distributed PBAT droplet retains the spherical shape. In this case, the major phase of PLA changes its morphology and shows a "volcano" shape around the PBAT droplet. For PBAT:PLA/BC 50:50/10, a co-continuous elongated structure appears to be preserved, although the PLA appears slightly damaged.
Figure 8 shows the mechanical properties of the blends, namely Young's modulus, tensile strength and elongation at break. As expected, for the blend without BC (black line), a transition from a ductile material to a rigid material is observed with increasing PBAT wt.%, from a higher young modulus and lower elongation at break values for pure PLA to a lower young modulus and higher elongation at break values for pure PBAT. In line with the rheological behaviour and morphology, the presence of BC has (i) a reinforcing role for PBAT that, as it is aspect, moves the mechanical behaviour of PBAT from ductile to rigid, and (ii) a reduction in mechanical properties for PLA. Consequently, for the blend composition, when PBAT is the main phase, BC shows a small reinforcing role, while when PLA is the main phase, biocomposites show a reduction in mechanical properties. Nevertheless, for PBAT:PLA/BC 50:50/10, an increase in Young's modulus, tensile strength occurs without much variation in elongation at break, probably due to the preservation of the co-continuous elongated morphology and a good dispersion of BC in the blend.
The results of the DMA analysis are shown in Fig. 9 as the variation of the logarithm of the storage moduli and the damping factor with temperature. The storage modulus E' and the damping factor (tanδ) for the blend and the blend filled with 10 wt.% BC as a function of temperature are shown in Fig. 9, from room temperature to 140°C, where possible. As expected, the first observation is related to the differences in the storage modulus at room temperature and in the shape of the storage modulus curve between the pure PBAT and the blend with PBAT in the major concentration with respect to the pure PLA and the blend with PLA in the major concentration. Instead, for the PBAT:PLA 50:50 blend, the mechanical behaviour of PLA prevails, with a storage modulus at room temperature comparable to that of pure PLA and a pronounced increase in a E' (with a maximum at ≈130°C) due to the cold crystallisation phenomena of PLA. A second observation is related to the shape of the damping factor curve: the main peak at tanδ is related to the Tg of PLA, since that of PBAT is normally observed around − 30¸ − 20°C. In Fig. 9b it is clear how the Tg of PLA in the blend remains constant and equal to 70.3±0.6°C, and the increase in magnitude results coherent with the increase of PLA in the blend: namely PBAT, the rubbery phase, acts as a stress concentrator. When the blends are filled with 10 wt.% BC, the first observation is the shift towards a lower value of E' of the PBAT:PLA/BC 75:25/10 and PBAT:PLA/BC 50:50/10 blends, coherent with the tensile properties already shown. The storage moduli below the glass transition temperature of PLA are lower for all PBAT:PLA weight ratios, except for pure PBAT. This behaviour is consistent with the tensile properties and rheological behaviour. Moreover, a different shape with a double shoulder in the E' curves, caused by PLA cold crystallisation phenomena. As regards the damping factor, a significant reduction in the size of the main peaks related to Tg is observed, indicating a low damping capacity in the presence of BC. Furthermore, a shift towards a lower value of Tg is observed with the increase of PLA in the blend, from 71.4°C for PBAT:PLA/BC 75:25/10 to a Tg of 66.4°C for PBAT:PLA/BC 25:75/10. The slight decrease in glass transition temperature with increasing PLA in the biocomposite blend and the decrease in peak size is consistent with the behaviour observed with rheological analysis. We hypothesize that the decrease in storage moduli below the glass transition temperature and the decrease in Tg are due to a possible degradation of the PLA polymer chain, caused by the presence of BC in the blend.
In order to evaluate the degradation phenomena likely to be suggested by the previous characterisation, a measurement of the molecular weight was carried out by means of intrinsic viscosity measurements and correlated to the molecular weight Mv by means of Mark-Houwink constants. Table 3 shows the results of the capillary viscosimetry carried out on PBAT and PLA, with and without 10 wt.% BC. To prepare the solution at the concentration of 0.2 wt.%, each material was dissolved in THF with stirring at 50°C for 1 hour. To evaluate the intrinsic viscosity of the composites, each composite was dissolved in THF, the solid fraction was removed with filter paper, the resulting solution was poured and then a 0.2 wt.% solution was prepared. The viscous molar mass of the biocomposites was compared with that of the pure matrix subjected to the same process condition. The table highlights that Mv of PBAT processed in the presence in absence of BC maintains its viscous molar mass constant, with a dimensionless value of molar mass almost equal to one. While for PLA a significant reduction of the viscous molar mass has been highlighted, in fact the dimensionless molar mass results 0.5. For PBAT, the loading of BC as a reinforcing filler leads to an increase in mechanical and rheological properties. Instead, the reduction in viscosity and molar mass of PLA suggests a scission of the molecular chains induced by the presence of BC in the composite. This result is probably due to the fact that BC causes or accelerates the degradation of PLA, thereby reducing the molar mass [39, 63].
Table 3
Viscous molar mass, intrinsic viscosity, and dimensionless values of molar mass of PBAT and PLA with and without 10 wt.% of BC.
|
Mv0
|
ɳ
|
|
Mv
|
ɳ
|
\(\frac{{\varvec{M}}_{\varvec{v}}}{{\varvec{M}}_{\varvec{v}0}}\)
|
|
[g/mol]
|
[dL/g]
|
|
[g/mol]
|
[dL/g]
|
[-]
|
PBAT:PLA 100:0
|
5.32 ×104
|
0.63
|
PBAT:PLA/BC 100:0/10
|
5.79 ×104
|
0.67
|
1.08
|
PBAT:PLA 0:100
|
1.58 ×105
|
1.17
|
PBAT:PLA/BC 0:100/10
|
7.91 ×104
|
0.7
|
0.50
|