An Sofie
Narmon
a,
Annelies
Dewaele
a,
Kevin
Bruyninckx
a,
Bert F.
Sels
a,
Peter
Van Puyvelde
*b and
Michiel
Dusselier
*a
aDepartment of Microbial and Molecular Systems, Centre for Sustainable Catalysis and Engineering, KU Leuven, Celestijnenlaan 200F, 3001 Leuven, Belgium. E-mail: michiel.dusselier@kuleuven.be; Web: https://dusselier-lab.org/
bDepartment of Chemical Engineering, Soft Matter, Rheology and Technology, KU Leuven, Celestijnenlaan 200F, 3001 Leuven, Belgium. E-mail: peter.vanpuyvelde@kuleuven.be
First published on 10th March 2021
Bio-based and degradable polymers such as poly(lactic acid) (PLA) have become prominent. In spite of encouraging features, PLA has a low melt strength and melt elasticity, resulting in processing and application limitations that diminish its substitution potential vis-a-vis classic plastics. Here, we demonstrate a large increase in zero shear viscosity, melt elasticity, elongational viscosity and melt strength by random co-polymerization of lactide with small amounts, viz. 0.4–10 mol%, of diethylglycolide of opposite chiral nature. These enantiomerically pure monomers can be synthesized using one-step zeolite catalysis. Screening of the ester linkages in the final PLA chains by the ethyl side groups is suggested to create an expanding effect on the polymer coils in molten state by weakening of chain–chain interactions. This effect is suspected to increase the radius of gyration, enabling more chain entanglements and consequently increasing the melt strength. A stronger melt could enable access to more cost-competitive and sustainable PLA-based biomaterials with a broader application window. Amongst others, blow molding of bottles, film blowing, fiber spinning and foaming could be facilitated by PLA materials exhibiting a higher melt strength.
Certain factors restrain the adoption of PLA. At present, high-molecular weight PLA is synthesized by ring-opening polymerization (ROP) of the dilactone, lactide (LD), which is made by back-biting depolymerization of polycondensed LA.4,9 While dilactones and ROP are indispensable, making LD is a two-step, time and energy consuming process that requires additional purification, driving up the cost of production. Moreover, LD yields are mediocre and the process is confronted with racemization, losing control over the enantiopurity of (L,L)-lactide (L-LD), and thus of PLA.10,11 Recently, some of us developed a zeolite-based catalytic process, enabling the single-step conversion of LA to LD. The highly selective route outperforms the two-step procedure, potentially in cost and certainly in atom efficiency (few side-products) and in preserving stereochemistry (no racemization).12 Other routes to lactide are also being developed and gear toward solving the cost issue.13–15 A more intrinsic drawback of PLA is associated with its processing performance and consequently its applicability. PLA has a poor melt strength and melt elasticity, which are crucial properties for elongational flow dominated processes such as film blowing, blow molding, fiber spinning, foaming, etc.16–18 Different strategies for improving the melt strength properties of PLA have been described. Some efforts focus on the use of chain extenders to create long chain, branched, cross-linked or star-shaped structures of PLA, to enhance chain entanglements.19–25 Although chain extenders can boost numerous properties of PLA, final polymer structures are often difficult to control, and seeking the optimal extender concentration remains challenging. In addition, chain extended polymer structures can exhibit reduced degradability,26–28 whereas extending agents themselves are often not biodegradable or biocompatible and in some cases are even toxic (diisocyanates). A second strategy consists of blending PLA with other polymers, as well as compounding it with different micro- or nanosized fillers (e.g. 5 wt%) to enhance its melt behavior.29 However, homogeneous blends or equal particle dispersion are not readily obtained. Moreover, blending can result in recycling or (bio)degradation limitations.30
Finally, stereocomplex crystallites, induced by mixing PLAs of opposite chiral nature (L-PLA with 5 wt% D-PLA), can act as crosslinking points, resulting in a melt behavior similar to that of chain-extended PLA structures.31–33 However, synthesis of D-PLA requires large scale availability of pure D-LA, which is currently not trivial.34 In contrast to these strategies, studies linking the composition of the backbone (linear lactide-based co-polymers) to rheological properties are scarce, despite the obvious value of microstructure – viscoelasticity relations for these polyesters. One group performed a few rheological studies on syndio- and heterotactically enriched PLAs.35 The effect of incorporating D-LD in L-PLA has been a topic of investigation but without unambiguously clarifying the impact of the D/L ratio or showing a negligible change in polymer melt viscosity.36,37
Here, we describe the unprecedented improvement of melt strength behavior of PLA by ring-opening co-polymerization of LD in the presence of (small amounts of) diethylglycolide (EG) of opposite chiral nature (Fig. 1A). Although diethylglycolide has been polymerized before, its traditional synthesis is homogeneously catalyzed, inefficient and laborious, ending up with a mixture of diastereomers.38–40 Zeolite-based cyclization of α-hydroxy acids here not only generates efficiency and high yields, but also essentially maintains the full stereochemistry of the substrate, which is of utmost importance for this research. Access to enantiomerically pure cyclic ester monomers enables the exploration of a wide range of high-molecular weight (co)-polymers with possibly attractive new properties.
In this research, rheological studies on co-polymers of enantiomerically pure LD and EG revealed a strong increase in melt elasticity and melt strength compared to those of conventional PLA, with clearly observed differences associated with the chiral nature of the monomers. While improvements in melt properties are usually obtained through chain extension, blending or compounding, a simple co-polymerization here suffices, even for small EG incorporation. A promising strategy for large scale implementation thus arises, given the ease of dropping-in 1% of EG into the lactide batch for classic L-PLA production. The gains in melt strength could translate into lower weight of plastic needed per functional item (e.g. blow molding a thinner bottle) and expand the application horizon for PLA.
ROP, based on the work by Fisher et al.,41 was performed with the stannous octanoate (Sn(Oct)2) catalyst and n-dodecanol initiator. This catalytic system ensures preservation of the stereochemistry of the monomers within the polymer chains, high molecular weight and low polydispersity.42,43 Diverse lactide-based polyesters were synthesized with different incorporations of EG, varying between 0.3 and 10 mol%, i.e. n/(m + n) in Fig. 1(A). The polymers vary in stereochemistry of the co-monomers (L-EG or D-EG with L-LD or D-LD, Fig. S4†). 1H-NMR proved successful incorporation of the co-monomers in polylactide (Fig. S5†), and weight-average molecular weights (Mws) varied between 101 and 149 kg mol−1 (Tables 1 and S3† have a full dataset). Various benchmarks were synthesized, i.e. a co-polymer of L-LD with 10 mol% D-LD and pure poly(L-lactide) (P(L-LD)) and poly(D-lactide) (P(D-LD)), or purchased. The latter, a commercial PLA (Ingeo 7001D, NatureWorks) has specifications targeting use in injection stretch blow molded bottle applications.
Build-ina % | GPC | DSC | SAOS | |||||
---|---|---|---|---|---|---|---|---|
Polymer | Schematic chain structure | Co-monomer | M w (kg mol−1) | D | T g (°C) | T m (°C) | η 0 (kPa s) | ω c (rad s−1) |
a Average build-in % of methyl and methine protons (when both visible), determined by 1H-NMR (400 MHz, CDCl3). b Polystyrene standards were used to calibrate weight-average molecular weights (Mw). The experimental molecular weight was corrected considering the Mark–Houwink parameters for PLA, and the same parameters are used as an estimation for the different co-polymers.6 c Commercial grade PLA from NatureWorks. d Large scale polymerization reactions (50–100 g). e Added amounts before reaction; L- and D-LD cannot be distinguished from each other in 1H-NMR. | ||||||||
![]() |
![]() |
— | 135 | 1.7 | 58 | 177 | 9.7 | 62.9 |
![]() |
![]() |
— | 117 | 1.7 | 57 | 179 | 12.1 | 84.5 |
![]() |
— | 83.3 | 1.5 | 58 | 149 | 5.9 | >100 | |
![]() |
![]() |
10e | 128 | 1.7 | 54 | — | 4.8 | >100 |
![]() |
![]() |
9.0 | 146 | 1.6 | 43 | 148 | 8.3 | >100 |
![]() |
0.7 | 133 | 1.4 | 54 | 177 | 23.1 | 46.7 | |
![]() |
![]() |
10.3 | 101 | 1.4 | 46 | 127 | 15.2 | 80.4 |
![]() |
1.2 | 115 | 1.7 | 56 | 169 | 143.3 | 7.83 | |
![]() |
0.7 | 138 | 1.6 | 54 | 170 | 33.9 | 32.1 | |
![]() |
0.4 | 128 | 1.8 | 57 | 165 | 54.8 | 24.3 | |
![]() |
![]() |
0.6 | 142 | 1.6 | 56 | 170 | 27.8 | 32.7 |
![]() |
![]() |
10.5 | 131 | 1.7 | 45 | 118 | 71.8 | 19.1 |
![]() |
4.1 | 129 | 1.7 | 54 | 154 | 108.2 | 8.17 | |
![]() |
2.0 | 149 | 1.3 | 56 | 160 | 17.3 | 51.0 | |
![]() |
1.2 | 118 | 1.7 | 56 | 171 | 33.3 | 24.3 | |
![]() |
0.7 | 128 | 1.5 | 58 | 172 | 191.1 | 6.64 | |
![]() |
0.3 | 129 | 1.6 | 56 | 177 | 22.9 | 76.6 | |
![]() |
||||||||
Large scale reactions | ||||||||
![]() |
— | 106 | 1.6 | 52 | 177 | 7.1 | >100 | |
![]() |
1.2 | 119 | 1.9 | 59 | 170 | 14.8 | 55.9 |
Chain structures of co-polymers are well-known to be dependent on the relative polymerization rates of the different monomers. If propagation rates are comparable, random chain structures are obtained, whereas a significant rate difference leads to block co-polymers. Various researchers have described the decrease in the polymerization rate of alkyl-substituted lactides with increasing steric hindrance of the substituents38,39,44–46 This has been attributed to a more difficult nucleophilic attack of the initiator.38,46 In addition, a decreased polymerisability is observed with increased intramolecular steric repulsion in the polymer chains (less present in the cyclic di-esters) which decreases the polymerization enthalpy.44 Nevertheless, the difference in the polymerization rate between rac-LD and rac-EG has been described by Baker et al. as rather small when comparing the reactivity ratios, indicating a high degree of randomness within their co-polymer chains.38,39 Here, the randomness of incorporation was quasi confirmed by determining the time evolution of the co-monomers during ROP (Fig. 1(D); Table S4 and Fig. S6†). Only insubstantial differences in the relative progression of the conversions are detected. L-EG polymerization progresses (relative to its starting amount of 1%) only slightly slower than D-LD polymerization (99%), most probably giving rise to strongly randomized co-polymers (note that the degree of randomness can vary with the co-monomer ratio).
Differences in the thermal properties of the various co-polymers are observed. In general, a downward trend in glass transition temperature (Tg), melt temperature (Tm) and degree of crystallinity (χc) is obtained with increasing amounts of EG co-monomer (see Tables 1 and S3, and Fig. S7†). Tm and χc show a stronger decrease when co-monomers with opposite stereochemistry are used. Baker et al.47 explained the decrease in Tg of poly(alkyl glycolide) polymers with alkyl side chains of increasing length (2–8 carbons) compared to that of polylactide based on weaker dipole–dipole interactions between polyester chains as due to a stronger screening of the ester groups by the alkyl side chains and thus less chain–chain interactions. The decrease in Tm and χc is explained by the introduction of defects in the regular structure of P(L- or D-LD), preventing chains from folding into close crystalline structures. This effect is expected to be stronger when co-monomers with opposite stereochemistry are used.4
![]() | ||
Fig. 2 (A) Complex shear viscosity (|η*|), (B) elastic modulus (G′) and (C) viscous modulus (G′′) as a function of angular frequency (ω) for co-polymers with 9–10 mol% of co-monomer. (D) Complex shear viscosity (|η*|), (E) elastic modulus (G′) and (F) viscous modulus (G′′) as a function of angular frequency (ω) for co-polymers with 0.6–0.7 mol% of co-monomer. ♦ and error bars at 0.1 rad s−1 in (A) and (D) indicate the average value and standard deviation as calculated from different synthesized homopolymers of pure L or D-LD (see the ESI† for more details). The colors are also traceable in Table 1. |
All polymer melts tend to a Newtonian plateau at low frequencies, while shear-thinning occurs at higher frequencies (Fig. 1(A) and (D)). Several observations emerge from the data by comparing the viscoelastic properties of the co-polymers with those of P(L-LD), a self-made control with a similar Mw, and the commercial grade PLA (Ingeo) designed for injection stretch blow molding. Adding 10 mol% of D-LD to P(L-LD) seems to decrease the viscosity and elasticity of pure P(L-LD) to a small extent, in line with earlier studies noting only subtle differences in the viscoelastic properties of polylactides with various D-LD contents.36,37,49 While the viscoelastic properties do not seem to significantly change with incorporation of 10 mol% L-EG in P(L-LD), co-polymers with 10 mol% EG and LD with opposite stereochemistry exhibit remarkable increases in both complex viscosity and elasticity of the polymer melts (Fig. 2(A–C) and Table 1). While the average η0 of P(L-LD) and P(D-LD) (Table S6†) is calculated to be 12324 Pa s (standard deviation of 7585 Pa s) and the average ωc lies around 72 rad s−1, for an average Mw of 122.5 kg mol−1, the values of P(L-LD-co-D-EG) and P(D-LD-co-L-EG) are estimated to be 15
212 (ωc = 80.4 rad s−1) and 71
783 Pa s (ωc = 19.1 rad s−1), respectively (Carreau-Yasuda model, see S7.1† for more details). Since the melt viscosity and elasticity are strongly dependent on the chain length,19,50 one could argue that the increase in P(L-LD-co-D-EG) is limited due to its rather low Mw (101 kg mol−1). Nevertheless, P(D-LD-co-L-EG) exceeds the zero shear viscosity of P(L-LD) with a factor 6, exhibiting a unique beneficial impact on the viscoelastic properties of polylactide. Interestingly, comparable results were obtained when much lower amounts, viz. 0.6–0.7 mol%, of EG were incorporated in polylactide (Fig. 2(D–F) and Table 1). Notably, co-polymers in which EG and LD have the same stereochemistry, (P(L-LD-co-L-EG) and P(D-LD-co-D-EG)) exhibited a limited increase in viscosity and elasticity, while the increase is most significant for co-polymers of L-LD and D-EG or D-LD and L-EG. To make sure effects are not largely attributable to variations in Mw, a Mark–Houwink correction was applied to the calculated zero shear viscosities shown in Table 1, which can be consulted in Fig. S16.† Note that as an estimate the same parameters were used for all different co-polymers. This correction further strengthens the possible melt-strength improving effects of EG.
To study the effect of the co-monomer ratio (EG:
LD) on the viscoelastic properties of the materials, various co-polymers (P(L-LD-co-D-EG) and P(D-LD-co-L-EG)) with different percentages of EG were analyzed (Tables 1 and S6, and Fig. S12–S15†). It can be noticed that for both P(L-LD-co-D-EG) and P(D-LD-co-L-EG) no clear correlation exists between viscosity or elasticity and the co-monomer ratio. For P(D-LD-co-L-EG) rather low η0s and high ωcs are observed when 0.3 or 2.0 mol% of EG is incorporated, while incorporation of 0.7, 4.1 or 10.5% exhibits a remarkable improvement in viscoelastic properties. For P(L-LD-co-D-EG) very high viscosity and elasticity are seen when 1.2 or 0.4% of D-EG is used. Nevertheless, the co-polymer with 10.3 mol% D-EG shows a much smaller increase, which is likely attributed to its rather low Mw compared to the other polymers. The non-directional trends potentially derive from changes in the mode of incorporation with the percentages (e.g. random vs. block) or other polymer microstructure effects. Different reactivity ratios have been reported when changing co-monomer percentages.54 Although Mws vary to some extent, the variation is limited and unlikely to cause such large effects (see Mw correction Fig. S16†). In addition, these effects seem to be reproducible in different co-polymers of LD and EG. Although the data described here cannot be compared to the literature data directly (due to differences in molecular weight, (D,D)-lactide content, additive mixing techniques, etc.), one can conclude that the zero shear viscosity values described here are in the same range as values obtained in the literature when mixing PLA with 1–3 wt% of Joncryl, a reactive additive.51–53
Extensional viscosity fixture measurements were performed at different Hencky strain rates (0.1, 0.5, 1 and 3.5 s−1) to determine the extensional viscosity (ηe) as a function of extension time (timee) (Fig. 3(A): Hencky strain rate = 3.5 s−1, data at other rates can be found in the ESI (S7.2†)). Extensional viscosity is related to η0 by a factor of three as explained using the Trouton ratio (ηe = 3η0) (see S7.2† for more details).
A comparable extensional viscosity is observed for the commercial PLA grade and the synthesized P(L-LD) sample, while P(L-LD-co-D-EG) confirms its higher viscosity. Although a strong difference in viscosity is noticed, no strain hardening behavior is observed at these Hencky strain rates, in agreement with the literature on extensional viscosity of linear polylactide.55,56
Finally, to directly measure the melt strength, vertical polymer melt strands (Fig. S18†) were spun on a wheel at a linearly accelerating velocity (100–5000 mm s−1) and the extensional force exerted by the materials was determined using a Haul-off extensional rheometer (details in S7.3†). While strong variations were seen in the rotation speed at which the polymer materials broke, forces seemed to be quite constant. P(L-LD-co-D-EG) exhibited an average force of about 24.4 N (Fig. 3(B)), which is about two times higher than the force of P(L-LD) (12.4 N). P(L-LD) and the commercial PLA showed very comparable values. This large scale – from a polymer chemist point of view – measurement consolidates the presumption of a severe rise in the melt strength of polylactide with incorporation of small amounts of EG with opposite stereochemistry.
Yamane et al.32 obtained indicative, but less strong, viscoelastic effects by blending small amounts of low molecular weight P(D-LD) into P(L-LD). This was explained by P(D-LD) acting as branching points on the P(L-LD) chains, forming small stereocomplex crystallites at the contact points of P(L-LD) and P(D-LD) chains. These crystallites stay solid above the Tm of P(L-LD), counteracting disentanglement under shear deformation. Similar interactions may occur between the P(L-LD-co-D-EG) or P(D-LD-co-L-EG) chains in our co-polymers. Stereocomplexation between blends of P(L-α-HBA) and P(D-LA)57 or P(L-α-HBA) and P(D-α-HBA)58 as well as stereocomplexation between stereoblock co-polymers of P(D-α-HBA) and P(L-LA) (note that these are polycondensation-derived) have been described in the literature.59 However, the ring-opening-polymerization-derived co-polymers in our work only contain very small amounts of co-monomers, randomly distributed throughout the PLA chains. Moreover, no proof of stereocomplexation was observed in the DSC measurements, which should be manifested as a melting peak at higher temperatures, i.e. above the melting peaks of both homopolymers of LD and EG (Fig. S7†), rendering the stereocomplex hypothesis unlikely. To further ensure the absence of stereocomplex crystallites, G′ and G′′ values of some of the prepared co-polymers were measured at increasing temperatures up to 230 °C to detect sudden drops in elasticity or viscosity of the materials. Since no remarkable effects were observed (Fig. S17†), presence of such crystallites with higher Tm can be discarded.
Even though the ethyl side groups are unable to foresee strong entanglements among polymer chains, their presence will likely affect the expansion of the polymer coils in molten state due to decreased inter- and intramolecular chain interactions. This effect might increase the radius of gyration (Rg) of the polymer coils (Fig. 4), rendering them more amenable to entanglements, resulting in higher elasticity and melt strength. This hypothesis agrees with the weaker chain–chain interactions between polydiethylglycolide chains compared to those in polylactide chains, as discussed above. Baker et al.47 described the screening effect of alkyl side chains with increasing length (2–8 carbons) on the ester groups in poly(alkyl glycolide) polymers and decreasing Tg as a result of reduced dipole–dipole interactions between polymer chains. One can hypothesize here that such screening effects are stronger when LD and EG are of opposite chirality, as the possibility to stack the chains is reduced. Interchain interactions are expected to dominate in polymer melts as models and theories describing polymer melt associations neglect the effects of intrachain interactions.60–62 However, a recent study by Carrillo et al.63 demonstrates the importance of the often under-investigated intrachain associations or intrachain loops in determining the structure of associating polymers in the melt state and their influence on the viscoelastic properties of these materials. On the other hand, the exact relation between viscoelastic properties and radius of gyration is still under discussion.64 This relation is described as dependent on both Mw and the density of the polymer coils in molten state. While the effect of Mw is already discussed above, the density of polymer coils is expected to be limited when dealing with very comparable polymer structures.65
Footnote |
† Electronic Supplementary Information (ESI) available: Extended information about materials and methods (S1–S2), extended reading (S3–S7), and extended data in Tables S1–S6 and in Fig. S1–S18. See DOI: 10.1039/d1sc00040c |
This journal is © The Royal Society of Chemistry 2021 |