**3. Results and Discussions**

*3.1. Melt Fluidity, Morphology and Thermal Properties of PLA/PBSA Blends*

From the MFR and torque results, reported in Figure 1, it can be observed that the increase of PBSA content resulted in a decrease in torque values and an increment in MFR.

**Figure 1.** (**a**) Trend of MFR as a function of PBSA content; (**b**) trend of torque as a function of PBSA content.

In fact, the pure PLA showed a torque of 152.3 ± 3.0 N\*cm and an MFR of 2.3 ± 0.3 g/ 10 min, but the pure PBSA was less viscous, showing a torque of 104.0 ± 7.1 N\*cm and a MFR of 7.1 ± 0.7 g/10 min. Hence the fluidity in the melt and thus the processability of the produced granules can be modulated as a function of composition. This trend is in accordance with the necessity to appropriately modify the extrusion and injection molding for the different compositions, as explained in Section 2.2. Interestingly, the blends containing EO showed an increased value of torque and a significantly lower value of MFR, because of the increase in molecular weight due to the branching reactions of the EO. A simplified scheme of the reactions occurring in the melt between PLA, PBSA and EO is reported in Figure 2.

The lowest MFR value was recorded for the (80/20)J blend, thus the one that required an increase in the mold temperature and injection pressure during the injection molding process (as previously observed and reported in Table 2).

**Figure 2.** Simplified scheme of the reactions occurring in the melt between PLA, PBSA and EO.

The main tensile properties (Young's modulus, yield stress, stress at break, elongation at break and Charpy impact resistance) determined from the tensile stress–strain curves and Charpy impact test, are reported in Table 3.

**Table 3.** Tensile and impact properties for each formulation containing an increasing PBSA content.


The mechanical behavior of the blends changed, passing from neat PLA to binary PLA/PBSA blends. PLA alone had the typical mechanical response of a fragile material with a high stiffness and tensile strength but a low Charpy impact resistance and elongation at break. Neat PLA failed just after the elastic region (no yielding point is observed), as typical in a brittle fracture. The PBSA addition maintained the material stiffness at an acceptable level but at the same time improved its flexibility for contents higher than 5 wt.% of PBSA. In fact, 95–5 blend still showed a fragile behavior without yielding, and a low elongation at break, low Charpy impact resistance and still high stress at break. From 5 up to 40 wt.% of PBSA, the increment in elongation at break and Charpy impact resistance was almost proportional to PBSA content (Figure 3a). A marked decrement of stress at break was registered due to the elastic characteristic conferred by the growing PBSA addition. The variation of stress at break, also considering the values of the deviations reported, can be considered negligible. However, the slight stress at break increase for the 60–40 blend could be attributable to the change in the morphology of the blend which became co-continuous. Hence, also the more ductile PBSA phase results were continuous. Then, in this co-continuous blend both phases fully contributed to the blend mechanical response in all directions and resulted in a more effective stress transfer.

The Young's Modulus decreased by increasing the PBSA content in monotonic way (Figure 3b). Hence, in the investigated range a wide modulation of properties is possible by varying the blend composition.

**Figure 3.** (**a**) Trend of elongation at break (red circles) and Charpy Impact strength (black squares) as a function of PBSA content; (**b**) trend of Young's Modulus as a function of PBSA content.

The results of the mechanical tests are closely related to the materials morphology (Figure 4).

The PBSA, until 20 wt.% in the blends, appeared as a spherical/ellipsoidal dispersed phase in the PLA matrix, indicating the poor miscibility between PLA and PBSA. In some cases, it was possible to observe voids at the interface (debonding). The cause of this debonding could be related to the cryofractured samples, where dilatational stresses were generated to the mismatch of the thermal coefficients between the PBSA particles and the PLA matrix during cooling before fracture, as was also observed for PLA–PBAT system [20]. The 60–40 mixture confirmed what is generally observed in literature [38,41] where a change in morphology occurs (from PBSA dispersed particles to a co-continuous structure); so the remarkable improvement in elongation at break, especially of the impact strength observed, is explained with the achievement of the inversion point region.

A typical two-phase structure, where discrete droplets of the minor phase are dispersed in the matrix, was observed in the samples with PBSA content up to 20%. For these blends it was possible to calculate the *Rn, Rv* and *SD* values, given by the ratio between *Rv* and *Rn*, (reported in Table 4) according to Equations (5)–(7), respectively.

**Figure 4.** SEM micrographs of morphology phase development of PLA/PBSA blends: (**a**) 95–5; (**b**) 90–10; (**c**) 85–15; (**d**) 80–20; (**e**) 60–40; (**f**) (80/20)J; (**g**) (60/40)J.

**Table 4.** *Rn*, *Rv* and *SD* values of the dispersed PBSA phase for PLA/PBSA blends at different compositions up to 20 wt.% of PBSA content.


The average size of the domain was between 1 and 2 μm indicating that PBSA and PLA were thermodynamically incompatible [41]. A slight increment of *Rv* values was observed with the increase in the PBSA content. This behavior is in agreement with the coalescence

theory [73] for which, during the mixing process, the dispersed phase can collide with each other and coalesce, forming bigger droplets. The probability of droplet collision will be more accentuated by increasing the PBSA content, with the number of PBSA droplet collisions that will be proportional to the square of the PBSA concentration [74]. However, this size distribution increment is not as pronounced as could be expected. This behavior was encountered by other authors [70] and can be ascribed to the high shear rate and extensional flow during the extrusion process that limited the particle coalescence favoring the break-up of dispersed droplets. Moreover, the two polymers, being both polyesters, are characterized by a good chemical affinity in full agreement with the evolution of mechanical properties as a function of PBSA content. However, the dispersed phase dimensions are in accordance with the general rule regarding rubber toughening, which states that the dispersed phase must be distributed as small domains in the polymer matrix [75,76]. This aspect, combined with a sufficient interfacial adhesion, would increase the elongation at break and the impact properties of the final material in accordance with the results obtained from the mechanical tests.

The 60–40 blend showed a co-continuous morphology where the dispersed phase coalesced until it formed bigger structures throughout the whole blend [77–80]. Consequently, due to the irregularity of the shape assumed by this structure, the calculation of *Rv, Rn* and *SD* parameters was not feasible. The analysis of the micrographs has anyway shown that the dimension of the two interpenetrating phases is quite low, probably thanks to effective processing and this explain the possibility of a good modulation of properties by acting on composition. Moreover the presence of EO improved the compatibility of the phases: a decrease in phase dimensions can be noticed both in the (80/20)J and in (60/40)J blend (Figure 4f,g) as well as an increased interfacial adhesion [44,81–83] making difficult the observation of these interfaces during the analysis.

The first heating thermograms obtained from DSC analysis and reported in Figure 5a,b show the thermal history of the samples produced by injection molding.

The analysis of these data was preferred to get correlation with mechanical results, measured on injection molded specimens. Moreover, this analysis can provide useful information regarding the peculiar injection molding process adopted for the different blends. In fact, the crystallization occurring in the mold had a significant role in allowing a rapid and efficient ejection of the specimen by the machine without any distortion. Moreover, thermal properties related to the use of EO were yet investigated by Nunes et al. [84] in PLA. EO determined a decrease in the crystallinity in pure PLA and in blends with PBAT because of the introduced disorder due to the branching of chains. Lascano et al. [55] did not determine the crystallinity of their PLA/PBSA but the thermograms they reported were in good agreement with Nunes et al. [84] observations. These considerations can explain the observed necessity to increase the molding temperature in the injection molding of blends containing EO, to counterbalance the reduced tendency to crystallization with a decreased undercooling favoring the extension of the crystalline fraction.

According to literature [85,86], PBSA has a triple melting peak centered at around 87.14 ◦C; the melting behavior can occur with a melting peak numbers that depend on the processing conditions. Consequently, it was not possible to measure with accuracy the PBSA melting enthalpy and the PLA cold crystallization enthalpy due to the overlap between the PBSA melting endotherms, the PLA cold crystallization exotherms (that became more marked with increasing the PBSA content) and the enthalpic relaxation peak occurring in correspondence of the PLA glass transition temperature and ascribed to the specimen's aging.

With the PBSA addition also the PLA double melting behavior started to become observable. These double peaks were correlated to the remelting of newly formed crystallite during heating. Crystallites of disordered α' form (with low melting temperature) recrystallize in a more ordered α form having a higher melting temperature. In any case, the right melting peak of PLA was deemed to be the one which was in a higher temperature range [10]. According to the melting recrystallization model [87] the small and imperfect

crystals would be transformed into more stable and perfect crystals. However, at high cooling rates (like that reached during the injection molding process), the granules passed to the recrystallization region so fast that there was insufficient time for the molten material to reorganize into new ordered crystals generating consequently low melting crystallites [13]. In fact, it is known in literature that the heating rate influences markedly the conversion from α -to α-form for PLA as well as the presence of PBSA double or triple- melting peaks [83,88,89]. The absence of PBSA cold crystallization in the blends and also in the pure material can be ascribed to different factors. The most important is probably correlated to the PBSA macromolecules having a very fast crystallization rate during cooling and this leading to an absence (or very low quantitative) of amorphous domains that could recrystallize again during heating [8].

**Figure 5.** DSC thermograms (**a**) first heating for PLA/PBSA blends; (**b**) first heating for the selected blends with EO; (**c**) second heating for PLA/PBSA blends; (**d**) second heating for the selected blends with EO. Heat flow, reported in J/g, is expressed in arbitrary units to allow the curves shifting.

For all the blends, the cold crystallization temperature of PLA was shifted to a lower temperature if compared to pure PLA (Figure 5a–d); in agreement with Lascano et al. [55], the PBSA melting contributes to the increase in the PLA chain motion allowing PLA chains to arrange into packed structures at lower temperatures. Consequently, the cold crystallization temperature and the melting temperature of PLA decreased with the PBSA addition. However, the most marked decrement of the PLA cold crystallization temperature was observed for the 95–5 blend (especially in the first heating scan, related to the injection molding cycle that could enhance the PBSA nucleating ability because of the high temperature of the mold and high holding time) where the PBSA dispersed phase, present in low content but with a low particle size, seemed to act as nucleating agent [10] thanks to the

13

effect of the extended interface. In fact, in good agreement, by increasing the PBSA content, the PLA cold crystallization temperature rose slightly but still remained lower than that of pure PLA.

An enthalpic relaxation peak above the glass transition temperature due to the aging [90] could be observed in the first heating scan (Figure 5a,b) for all blends, but it was reduced in blends containing EO because of their more disordered structure hindering chain relaxation. However, for both heating scans, it could be observed that the glass transition temperature of PLA remained almost unchanged with PBSA content, confirming the restricted miscibility of the two biopolymers as reported by Lee et al. [13].

In order to better understand the HDT obtained data (Figure 6), showing an almost constant value in all the examined blends, although it was not possible to determine the crystallinity values of the injection molded specimens considering the first heating scan, an estimation of the crystallinity was made on the second run. In this case, having deleted the thermal history, the single PBSA granules were also investigated. In fact, in the second run, thanks to the disappearance of the aging peak and to the lower cooling rate with respect to injection molding, the peaks of PBSA and PLA show an almost null overlap (Figure 5c,d).

**Figure 6.** HDT values for all blend compositions.

In Figure 7 the crystallinity percentage values for PLA, PBSA and the sum of them (that is, the total crystallinity of the sample) are reported. The enhancement of PLA crystallinity with PBSA content, suggests that the PBSA droplets, dispersed into the PLA matrix effectively act as crystallization nuclei for PLA accelerating the crystallization during the heating process [10]. For all the blends the maximum total crystallization, seen as the sum of PLA and PBSA crystallinity percentage, was between 27 and 32%. Hence only small variations of the crystalline fraction were present in the different blends and this was reflected in the HDT values (Figure 6) which were approximately the same for all the examined blends. In fact, it is well known that the HDT of neat PLA is at about its *T*g. The HDT is significantly affected by the degree of crystallization [91]. The low crystallization rate of PLA made it essentially amorphous under the injection molding conditions adopted in this work. Although a quantitative evaluation of the samples crystallinity was not possible for all blend compositions, it could be supposed that the maximum degree of sample crystallinity was lower than that measured during the second heating run. Consequently, the low crystallinity (due to the low crystallization rate) reached by the PLA under practical molding conditions was reflected in a null variation of the HDT that remained almost unchanged for all the blends analyzed (Figure 6).

**Figure 7.** Crystallinity values for PLA/PBSA blends in the second heating scan.
