Racemization. In the production of stereochemically pure lactide, formation of the other lactic acid enantiomer and meso‐lactide is unwanted. Higher temperatures, longer reaction times, and increased catalyst levels result in increased rates of racemization [4, 6, 69]. Since temperature and catalyst influence the rate of lactide formation as well, controlling the racemization rate can become quite complex.
Impurities. Data in the literature on the role and fate of impurities from the feed in the synthesis are scarce. Some metal cations such as sodium and potassium in the feed increase racemization risk, while other metals (Al, Fe) are catalytically active in transesterification, resulting in competitive polylactide formation [68, 69]. Through corrosion, metals may be released in the residue and will build up there [6, 75]. Some patents discuss the presence of acid impurities in the process [6, 7, 67, 78]. Mono‐ and dicarboxylic fermentation acids are responsible for stoichiometric imbalance in the lactic acid polycondensation reaction. Consequently, the composition of the obtained lactic acid oligomer chains can differ from pure PLA, resulting in impeded and incomplete catalytic depolymerization of the oligomers into lactide. In PLA manufacture, degradation reactions play a role, mainly via intramolecular chain scission, and this may also affect lactide synthesis.
On the one hand, it can be concluded that the lactide synthesis is straightforward in the sense of making a prepolymer and releasing lactide by thermal catalytic depolymerization at low pressure. On the other hand, it can be concluded that the scale‐up from a lab‐scale process to an economical, large‐scale process with high yield and no compromises on stereochemical purity is a complex multifaceted task.
1.3.3 Purification of Lactide
A lactide synthesis reactor invariably produces a crude lactide stream that contains lactic acid, lactic acid oligomers, water, meso‐lactide, and further impurities. The specifications for lactide are stringent mainly for free acid content, water, and stereochemical purity. Basically, two main separation methods, distillation and crystallization, are currently employed for lactide purification:
Distillation. Splitting the multicomponent mixture consisting of lactide, water, lactic acid, and its oligomers into pure fractions requires considerable know‐how on kinetics and operation of vacuum equipment. Distillates and bottoms may be recycled, but the accumulation of impurities from the feed or the production of meso‐lactide during the process requires careful fine‐tuning of temperatures and residence times. Distillation is well described in the patent by Gruber et al. [68]. The crude lactide from the synthesis is distilled in the first column to remove the acids and water, and then meso‐lactide is separated from lactide in the second column. As the boiling points of all compounds are in the range of 200–300°C, low pressures are used. Since the difference in boiling temperature of lactide and meso‐lactide is quite small, this distillation requires a lot of theoretical stages (>30). The NatureWorks distillation uses a series of distillation columns and is performed continuously [4]. Part of the distillation can also be integrated with the reaction [79].
Solvent Crystallization. A commonly used laboratory method for lactide purification is recrystallization from mixtures of toluene and ethyl acetate [4]. Lactide of extremely high purity can be obtained by repeated crystallization with different toluene/ethyl acetate ratios. Several patents also mention the use of solvents for the crystallization of lactide, but for large scale, melt crystallization without the use of solvents is preferred.
Melt Crystallization. Lactide crystallizes easily, and several patents describe how crystallization can yield lactide with required specifications regarding lactic acid content, oligomers, meso‐lactide, and water. An early patent describes such a crystallization method and includes some information on the thermodynamic equilibria (eutectica) of the lactide/lactic and the lactide–meso‐lactide system, which define the maximum yield as a function of these impurities in the feed [80]. In patents, the use of different types of equipment is mentioned: static equipment, falling film crystallizers, vertical column with scraper to remove crystal mass from the cooled wall, and scraped heat exchanger coupled to a wash column [70, 80, 81]. For large scale, it is a challenge to design and scale‐up the crystallization equipment with respect to the needed heat transfer areas and hydrodynamics, and the possible increase of viscosity of mother liquor by oligomerization of lactide and residual acid.
The choice between distillation, crystallization, or novel separation methods such as absorption or membrane separation is determined by the desired stereochemical purity of the product. Crystallization yields highly pure lactide, suitable, for example, for high‐melting PLLA homopolymer of high molecular weight. Affordable distillation equipment does not fully remove all meso‐lactide, and consequently, a lactide monomer mixture for PLA copolymers with other thermal properties is obtained upon ring‐opening polymerization.
The design of the separation system relies on detailed knowledge of the thermodynamic properties of the compounds and the kinetics of the reactive system. Obtaining such know‐how requires sophisticated analytical methods for lactic acid and its oligomers, lactides, and residues. Impurities can also be formed in lactide synthesis, similar to PLA degradation reactions, and gas chromatography (GC) methods are needed to identify these compounds and determine their fate in the process.
1.3.4 Quality and Specifications of Polymer‐Grade Lactide
The specifications and allowed impurity levels of lactide monomer for PLA are defined by the polymerization mechanism and the applied catalyst. PLA is commercially produced by ROP of lactides in bulk. The tin(II)‐catalyzed process offers good control over molecular weight and reaction rate provided that it is performed in the absence of impurities such as water, metal ions, lactic acid, or other organic acids. Purification of crude lactides is therefore indispensable for the industrial manufacture of high‐molecular‐weight PLA (M w > 100 kg/mol). In fact, lactide is the ultimate form of lactic acid, in its dehydrated and purest form.
1.3.4.1 Role of the Catalyst and Initiator in Lactide Polymerization
The theoretical description of the Sn(Oct)2‐catalyzed ROP of cyclic esters has been studied by many authors, but there does not appear to be a theory that consistently explains all experimental results of the coordination–insertion polymerization [3, 4,82–84]. Different polymerization mechanisms may dominate, depending on polymerization conditions, catalyst and initiator concentration, and the presence of a solvent.
Here, it is assumed that lactide is polymerized in bulk with Sn(Oct)2—a Lewis acid—and that the mechanism follows the model proposed by Kowalski et al. [84]. Since lactide is a cyclic ester, its ring can be opened by nucleophilic attack on the ester bond to start polymerization. Suitable initiators (nucleophiles) are water and alcohols, including the hydroxyl group of lactic acid. One ester linkage of a lactide ring is cleaved by reaction of the OH group of the initiator R—OH, creating a new R—O—C(O)— ester end group and an OH end group (Figure 1.8).
Every initiating molecule is covalently bonded as an end group to each polymer chain [84]. Via transesterification reactions, the 2‐ethylhexanoate ligands of the SnOct2 catalyst will also end up as octanoic ester groups in the polymer. In some papers, the Sn(II) catalyst is indicated as the initiator, presumably because lactide also polymerizes upon addition of that substance, and the effect of impurities is overlooked. An initiator—or coinitiator—is a substance that can start polymerization, in the case of lactide by opening the lactide ring, and thus offers control over molecular weight. This has to be a nucleophile and cannot be the Sn catalyst itself, as supported by the excellent work of Kowalski et al. who proved that SnOct2 needs activation with R—OH (Figure 1.9) [84].