Products used in this study are listed in Table 1. PBS and PLA pellets were supplied by Natureplast (PBE 003 and PLA 005 grades, respectively). l-lactic acid/d-lactic acid content for PLA is 94:6 mol% (Dorigato et al., 2012). Trihexyl(tetradecyl)phosphonium chloride (denoted in the text as P-Cl) was purchased from Cytec. Tris(2-hydroxyethyl)methylammonium methylsulfate (denoted in the text as A-M) was purchased from Sigma Aldrich.

Polymer pellets were dried in an oven at 70°C for 12 h. Pellets and additives were extruded using a 15g-capacity DSM microextruder (Midi 2000 Heerlen, Netherlands) with co-rotating screws (L/D ratio of 18) at 190°C, with a 100 rpm speed, for 3 min. PBS/PLA (60:40 wt. ratio), PBS/PLA/P-Cl (60:40:1 wt. ratio) and PBS/PLA/A-M (60:40:1 wt. ratio) were extruded. Extruded samples were injected in a 10 cm³ mould at 30°C to obtain 2 mm thick and 4 mm wide dumbbell-shaped specimens. 0.2 mm thin films were processed under compression at 210°C and cut for ageing experiments.

Degradation experiments for 3 months were performed considering films exposed to various environments. Every month, two samples of each formulation were recovered, washed with deionized water, and gently dried on absorbent paper. Then, they were dried in a vacuum oven at 30°C for 12 h and kept in a closed desiccator. Weight losses were calculated as follows ( W 0 being the weight before degradation experiment and W t the weight after degradation experiment): w e i g h t   l o s s   ( % ) = W 0   − W t W 0 ∗ 100 (5)

TEM micrographs (Figure 1) show examples of the dispersion of PLA phases (white domains) in the PBS matrix, confirming that PBS and PLA are immiscible (Bhatia et al., 2007). PLA phases in PBS/PLA blend without IL are irregularly dispersed (Figure 1A), and their maximum surface area is about 0.8 µm². The non-uniform dispersion of PLA domains has been observed by Deng et al., and it was assumed that, since PBS exhibits a lower melt viscosity than PLA, it is not able to break down PLA droplets (Deng and Thomas, 2015). Similar compatible blends have been prepared in previous studies, including PBAT/PLA/LI and PBAT/PLA/lignin/LI blends (Lins et al., 2015; Livi et al., 2015) where it was clearly demonstrated that ionic liquids act as compatibilizing agents. Indeed, ionic liquids were located at the interface of the PLA domains and when increasing the amount of ionic liquids, a migration of ionic liquids in the minority phase was observed (Megevand et al., 2016). In the present study, the addition of ionic liquids induces the presence of larger PLA domains (up to 27 μm² with P-Cl). Furthermore, PLA domains have ovoid shapes. Then, inclusions of PBS are observed in PLA domains. Since the PBS/PLA weight proportion is 60/40, phase inversion is likely to occur. Indeed, Wu et al. reported that, if the weight proportion of PBS is 60% or more, it becomes the continuous phase, and the inversion point occurs at around 50/50 (Wu et al., 2012), which can explain the increase of PLA phase. At molecular scale, ionic liquids are observed at the interfaces.

The fact that adding ionic liquids induces larger PLA domains in the PBS matrix was not expected regarding our earlier studies. Lins et al. added 1% of phosphonium ILs in PBAT/PLA blends and showed that ILs generated interfacial zones due to a strong attraction to ester groups, leading to a decrease of the interfacial tension (Lins et al., 2015).

Tensile properties of PBS/PLA blends, reported in Table 2, show that neat PBS/PLA exhibits a rather high Young’s modulus due to brittle PLA phases, acting as reinforcing agent, and a high strain at break due to PBS being ductile (Qiu et al., 2016). In presence of IL, complex morphologies, at ratio PBS/PLA 60/40 close to the inversion phase make the discussion rather difficult to establish relationship between tensile properties and morphology of the blends. Young’s modulus and strain at break are slightly improved with ammonium IL, which could suggest the IL acting as an interfacial agent. The increase of modulus can also be explained by the fact that PLA domains are larger, and hence constitute more effective reinforcing agents. However, regarding PBS/PLA blend with phosphonium IL, the stiffness is reduced while elongation at break is significantly higher. PLA domains do not behave as a reinforcing agent in this blend. Elongation at break is probably enhanced due to a plasticization process, which could be induced by small PLA chains and phosphonium IL. It has been seen in the literature that phosphonium IL can plasticize PLA, as well as catalyze its degradation during processing (Park and Xanthos, 2009). By using Phosphonium IL when processing the PBS/PLA blend, it is not necessary to add a preliminary step aiming at cutting commercially available PLA chains in order to facilitate biodegradation. Hence, the increase of strain at break could be a consequence of low molecular weight PLA chains ( M n ¯ of 1.7 kDa, see SEC measurements on Figure 2) and/or of phosphonium IL plasticizing PLA domains. The hypothesis of short PLA chains could explain the loss of stiffness as well, since PLA domains provide stiffness in the PBS/PLA blends. It was seen by Deng et al. that PBS/PLA blends exhibit a low ductility from 40 wt% to 80 wt% of PBS, attributed to a poor adhesion between the two polyesters at such proportions (Deng and Thomas, 2015). Hence, a better adhesion between PBS and PLA, provided by the ILs is another way to explain the gain in elongation at break with ILs. Nevertheless, Young’s modulus decreases in presence of IL P-Cl, due to the depolymerization of PLA, while PBS is much less affected (Figure 2D). In addition, molecular weight being a parameter affecting the degradability of materials, faster degradation is expected for PBS/PLA blends containing IL P-Cl and a “predegraded” PLA phase, with a compromise between mechanical properties and degradability.

Considering environmental and safety issues, it is well known that ILs, which catalyze the glycolysis of PET, can be used to depolymerize many polymers such as polyethylene terephthalate (Kamber et al., 2010). In our study, only a small amount of ILs is used (1 wt%), acting as interfacial agents and migrating into PLA matrix, which considerably limits environmental and safety issues. In our work, in presence of ionic liquid, the degradation products of PLA with the lowest molecular weights, might be lactic acid, that is neither considered dangerous nor very volatile (Song et al., 2014; Elsawy et al., 2017).

Data acquired with dynamic mechanical analyses are displayed in Table 3. Two distinct alpha-relaxation temperatures confirm that PBS and PLA are not fully miscible at the studied proportions (Deng and Thomas, 2015). A slight merging of the alpha-relaxation temperatures of PLA and PBS with phosphonium ionic liquid is observed. This small shift may be due to partial miscibility of low molecular weight chains (Lins et al., 2015). The decrease of the Tα of PLA is also consistent with a migration of P-Cl ionic liquid in the PLA phase, leading to reduced molecular weights (Figure 2) and lower Tα. Ammonium IL does not have a significant impact on the alpha-relaxation temperatures. These results are in agreement with partial miscibility and/or perhaps transesterification in presence of phosphonium IL.

Number average molecular weight ( M n ¯ ), weight average molecular weight ( M w ¯ ), and dispersity ( M w ¯ / M n ¯ ) are summarized in Figures 2A–C. PLA and PBS are not distinguishable on SEC chromatograms. Hence, to assess the average molecular weight M n ¯ of specifically PBS or PLA phases, ¹H NMR spectroscopy (end group analysis) is used as well (Figure 2D). Molecular weights measured with SEC (Figure 2A) are higher than the ones calculated with end group analysis (Figure 2D), where it is assumed that the studied polyesters are not branched. Hence, values obtained with end group analysis may be underestimated. For this reason, trends, rather than absolute molecular weights, are discussed in this work.

Regarding SEC measurements on injected samples, PBS/PLA blends with phosphonium IL exhibit significantly lower molecular weights than PBS/PLA without additive. Using NMR titration, it is observed that the M n ¯ of PLA drastically decreases from 14.8 kg.mol⁻¹ to 1.7 kg.mol⁻¹ with phosphonium IL, while it slightly increases to 18.0 kg mol⁻¹ with ammonium IL. M n ¯ of PBS decreases with both ILs (from 26.3 kg mol⁻¹ to 15.8 with P-Cl, 19.0 with A-M). This suggests the phosphonium IL degrades the polymers during processing, as observed by Park et al. with neat PLA (Park and Xanthos, 2009). This could be due to transesterification reactions, causing random chain scissions at first (Lins et al., 2015). It was observed by Freyermouth that no or few transesterification reactions take place between PBS and PLA in PBS/PLA blends, at 180°C for 105 min (Freyermouth, 2014), however, these conclusions could differ with the addition of ionic liquids. Nonetheless, no additional peak, corresponding to a possible transesterification phenomenon, was detected using ¹H NMR spectroscopy. It is yet clear that P-Cl induces thermal degradation during processing, leading to lower molecular weight chains (especially PLA), which can explain partial miscibility, in agreement with a slight merging of alpha relaxation temperatures in DMA (Table 3). These results are consistent with thermogravimetric analyses (TGA) (Figure 3; Table 4), that show that PLA phases in PBS/PLA/P-Cl blends degrade at lower temperature than in PBS/PLA and PBS/PLA/A-M blends. Regarding PBS phases, degradation temperatures do not change considerably.

As reported in our previous article (Delamarche et al., 2020a), temperature at 1% degradation (T1% deg.) and at the maxima of the derivative curves of the weight loss as a function of temperature, were 304 and 409°C for PBS; 319 and 379°C for PLA. This is consistent with the fact that PLA phases degraded to a greater extent during processing with P-Cl. Results are also in agreement with differential scanning calorimetry (DSC) results (Table 5), where it can be seen that melting temperature of PLA, during the second heating ramp, decreases with phosphonium IL (from 174.4°C on the first heating ramp to 169.5°C), while for neat PBS/PLA and PBS/PLA with ammonium IL, the melting temperature remains similar between the first and second heating ramp. This can be explained by a degradation of PLA phases taking place during the measurement, at the end of the first heating ramp, leading to smaller molecular weight PLA chains. Melting temperature indeed depends on the molecular weights: under a critical value of molecular weight, the melting temperature drops, whereas above this critical value, the melting temperature is constant (Fatou and Mandelkern, 1965; Sánchez-Soto et al., 2002). Xiang et al. observed that melting temperature of PLA diminishes as the molecular weight diminishes (Xiang et al., 2016).

For thermogravimetric analyses, as the IL contents are only 1%, we have assumed their own thermal stability is not influencing significantly the blends behavior. PBS/PLA blend with ammonium IL exhibits slightly higher thermal degradation temperatures (+11.9°C for PLA, +5°C for PBS), which is consistent with a higher molecular weight of PLA measured earlier, and a possible transesterification phenomenon. This could also be due to good interfacial interactions between the ammonium IL and the polyesters, which increase the activation energy necessary to cause degradation (Lins et al., 2015). Therefore, thermogravimetric analysis, like molecular weight measurements, show that the phosphonium IL degrades more specifically PLA phases during processing, which can no more play the role of a reinforcing agent: it can be assumed that PLA chains were degraded to an extent where PLA domains could not provide stiffness to the blend. Ammonium ionic liquid, however, leads to a better stability of the blend.

In PBS/PLA blends, glass transition temperatures appear to be difficult to determine using DSC, because, in the case of PBS, they are wide transitions, and in the case of PLA, they are close to a low intensity melting endotherm of PBS, known in the literature as an annealing peak (Wang et al., 2007). However, we can still observe similar tendencies as the ones observed for alpha relaxation temperatures measured using dynamic mechanical analysis: the Tg of PBS increases by 4°C with the addition of ILs P-Cl, by 1.8°C with IL M-A, whereas, the Tg of PLA phases do not show significant changes (Table 5). As discussed earlier, the shift of glass transition temperature of PBS can be due to a partial miscibility of PLA oligomers in the PBS matrix. Indeed, it has been seen in the literature, that PBS and PLA can exhibit partial miscibility in the amorphous phases to a small extent (Deng and Thomas, 2015; Ostrowska et al., 2019).

Melting and crystallization curves (second ramp) are plotted in Figure 4. Both polymers lead to two melting peaks looking like those of each neat polymer (Bhatia et al., 2007). Melting peak appearing at Tm(PBS) exhibits a more pronounced separation with ILs at second heating ramp. This peak division is typical of PBS melting behavior, and usually appears after crystallization at temperatures above 80°C, with the generation of crystals that have different thermal stability (Wang et al., 2007). Here, adding IL in PBS/PLA blends seems to have the same effect on the crystallinity of PBS as an annealing treatment. This is in agreement with a certain affinity of ILs with PBS. Regarding PLA, melting peak of PLA in PBS/PLA/P-Cl blend appears at lower temperature than for neat PBS/PLA blend on the second heating ramp, suggesting the presence of small crystallites, probably due to low molecular weight PLA chains that were formed during processing and first heating ramp. The melting curve of PBS/PLA blend with ammonium IL is very similar to PBS/PLA without additive, except for a slight separation of melting peak of PBS.

It is well-known that the use of IL-modified nanoparticles may induce a heteregenous nucleation effect as well as a lamellar ordering effect in the polymer matrix (Du et al., 2007; Xie et al., 2016). But in this case, we have used only 1 wt% of ILs (without nanoparticles) and we have demonstrated in a previous paper that a small amount has no significant effect on the crystallization temperature of PBAT matrix (Livi et al., 2014).

Neat PBS/PLA blend shows two crystallization peaks, a broad one looking similar to the melting peak of PBS alone, and a small peak that merges and appears at higher temperature. PBS/PLA blend with ammonium IL shows a similar curve, although the small peak appearing at higher temperature is more pronounced, yet not completely separated. Phosphonium IL induces a crystallization peak division. The larger peak, appearing at lower temperature might be PBS crystallization, while the small peak could be associated with PLA crystallization. Another explanation can be that PLA crystallites act as nucleating agents for PBS once they are formed.