Make or break: Mg( II )- and Zn( II )-catalen complexes for PLA production and recycling of commodity polyesters †

Recently we reported a series of highly active Al( III )-complexes bearing a catalen ligand support for lactide polymerisation, observing unprecedented activity in the melt. Herein we report diversi ﬁ cation of the metal to furnish a series of well-de ﬁ ned dimeric Zn( II )- and Mg( II )-complexes, which were fully characterised by X-ray crystallography and NMR spectroscopy. The production of biocompatible atactic PLA from rac -LA in solution and under industrially preferred solvent-free conditions was demonstrated, typically observing good activity and M n control with a broad range of dispersities ( Đ = 1.08 – 2.04). Mg( II )-Complexes were shown to facilitate the relatively mild methanolysis of PLA, achieving up to 64% conversion to Me-LA within 8 h at 80 °C in THF. Further kinetic analysis found [Mg( 1,3 )] 2 to have k app values of 0.628 ± 0.0536 {4 wt% cat. loading} and 0.265 ± 0.0193 h − 1 {8 wt% cat. loading} respectively for the rate of consumption of PLA. Preliminary work extended polymer scope to PET from various sources, demonstrating catalyst versatility.

Herein, we report diversification of the metal to Zn(II) and Mg(II) in pursuit of catalysts active for PLA degradation, for which there is existing literature precedent. 109,111,123onsequently, a range of well-defined dimeric Zn(II)-and Mg (II)-catalen complexes were prepared, employing a new and emerging class of ligands in the area.Their application to the ROP of rac-LA in solution, and under industrially preferred melt conditions, is discussed.The relatively mild metalmediated methanolysis of PLA into Me-LA is reported.Preliminary work diversifying polyester scope is also demonstrated.

Polymerisation kinetics
To ascertain a better understanding of metal-ligand cooperative effects on activity, a kinetic study was pursued using [Zn(1)] 2 and [Mg(1)] 2 as a model system.A plot of ln([LA] 0 / [LA] t ) against time exhibited a linear relationship, indicating the reaction to be pseudo-first-order with respect to the consumption of rac-LA (Fig. 3).[Mg(1)] 2 exhibited an apparent rate constant (k app ) of 0.0206 min −1 , over 12 orders of magnitude higher relative to [Zn(1)] 2 (k app = 0.0017 min −1 ), consistent with solution results (Table 2).Indeed, [Zn(1)] 2 also exhibited an induction period of ca.40 minutes, potentially evidencing catalyst aggregation under these conditions.No induction period was observed for [Mg(1)] 2 , highlighting judicial choice of the metal can circumvent such limitations.It is tentatively suggested such aggregation is H-bonding in nature, although metal influence remains poorly understood.GPC analysis of the aliquots retained for [Mg(1)] 2 confirmed the polymerisation to be well controlled and living (Fig. 4).This was demon-strated by a linear increase in M n with conversion whilst maintaining narrow dispersities.A slightly lower M r,monomer relative to ideal PLA was observed (M r,theo = 144.12g mol −1 , M r = 118.27g mol −1 ), which could likely be attributed to minor transesterification, consistent with MALDI-ToF analysis (see ESI †).
Polyester recycling PLA degradation.[Zn(1-3)] 2 and [Mg(1,3)] 2 were investigated in the metal-mediated degradation of PLA into Me-LA in solution at 80 °C (Fig. 5 and Table 3).Whilst Me-LA is a possible green solvent replacement, it is also a potentially valuable chemical to the PLA supply chain since it can be directly converted to lactide. 3,91Commercially available polymer (0.25 g, PLLA cup, M n = 45 510 g mol −1 ) and catalyst were dissolved in either THF or anhydrous toluene under Ar, with heat and stirring assisting dissolution.MeOH was then added and the conversion to Me-LA was determined via 1 H NMR analysis of the methine region (ca.δ = 4.2-5.2ppm).The production of Me-LA has previously been shown to proceed via a two-step process through the intermediate formation of chain-end groups (see ESI †). 109,123onsequently, the methine groups can be categorised as internal (int), chain-end (CE) and those corresponding directly to the alkyl lactate (Me-LA).Conversion of internal methine units (X int ), methyl lactate selectivity (S Me-LA ) and Me-LA yield (Y Me-LA ) are tabulated in Table 3 below.[Zn(1)] 2 exhibited reasonably poor activity at 8 wt%, achieving 19% conversion to Me-LA within 8 h in THF with poor selectivity (Table 3, entry 1).Promisingly, superior activity and selectivity (Y Me-LA = 51%, S Me-LA = 53%,) was observed upon shifting to a non-coordinating solvent, namely anhydrous toluene (Table 3, entry 3).This implies THF competes with the degrading polymeric chain with respect to coordination to the Zn(II)-centre, consistent with the near complete consumption of PLA (X int = 96%).Interestingly, shifting to a more electron withdrawing catalen backbone in [Zn(2-3)] 2 had a detrimental impact on Y Me-LA , achieving 0% conversion to Me-LA under analogous conditions in THF.Whilst contrary to previous work by Payne et al., 110 this was consistent with solution polymerisation results (Table 2).Indeed, previously described solution reactivity trends were retained for PLA degradation.Since this behaviour was retained in anhydrous toluene, it is suggested the afore-mentioned activity loss due to possible catalyst aggregation (Table 2) likely persists under these conditions.It is possible bulky t Bu substituents promote the dissociation of [Zn(1)] 2 in solution, resulting in superior activity relative to [Zn(2-3)] 2 .
[Mg(1)] 2 significantly outperformed its Zn(II)-counterpart, achieving 31% conversion to Me-LA with good selectivity in THF at 8 wt% (Table 3, entry 13).Interestingly, significantly enhanced activity was realised upon decreasing the catalyst loading to 4 wt% (Y Me-LA = 64%, S Me-LA = 66%, X int = 97%), possibly evidencing a reduction in catalyst aggregation due to dilution.To improve industrial feasibility and investigate the limit of this effect, the catalyst loading was further reduced to 2 wt%.Significantly reduced Y Me-LA relative to 4 wt% was observed (Table 3, entries 11 and 12), implying unavailability of the active species predominates, consistent with [Mg(3)] 2 (Table 3, entry 14).In light of this, [Zn(1-3)] 2 were investigated at 4 wt% in both THF and anhydrous toluene at 80 °C, although no activity enhancement was observed.[Mg(3)] 2 exhibited superior activity relative to [Zn(3)] 2 , achieving 42% conversion to Me-LA within 8 h at 4 wt% (    was observed at 8 wt% (Table 3, entry 16).This implies the liberated to be inherently more active relative to [Mg(1)] 2 and that possible catalyst aggregation dominates at 4 wt%.Judicial choice of the metal had previously been shown to circumvent such challenges in the solution polymerisation of rac-LA (Fig. 3).Degradation reactions using [Mg(1,3)] 2 in anhydrous toluene were not pursued in alignment with the 12 principles of green chemistry. 125Overall, mass transfer limitations due to polymer particle size and stirring speeds were considered negligible based on previous work by Román-Ramírez et al., 109 which employed a homoleptic Zn(II)-complex bearing an ethylenediamine Schiff-base ligand.
PLA degradation kinetics.[Mg(1)] 2 and [Mg(3)] 2 were identified as the outstanding candidates and thus pursued for further kinetic analysis.Reaction progress was monitored hourly for the first 4 hours for 1 H NMR (CDCl 3 ) analysis of the methine region.A final aliquot was taken after 8 hours for analysis, totalling 5 data points (Fig. 6).PLA consumption was assumed to adopt pseudo-first-order kinetics in accordance to previous work by Román-Ramírez et al. 109 Consequently, the gradient of the logarithmic plot is equivalent to the apparent rate constant, k app (Table 3 and Fig. 6).[Mg(1)] 2 exhibited a k app value of 0.628 ± 0.0536 and 0.0819 ± 0.0213 h −1 at 4 and 8 wt% respectively in THF, indicating an increase in catalyst loading results in a statistically significant decrease in activity.
[Mg(3)] 2 was found to have a k app value of 0.265 ± 0.0193 h −1 , lower relative to [Mg(1)] 2 , consistent with preliminary methanolysis results (Table 3).Comparable Y Me-LA values (Table 4) were observed relative to Table 3, indicating good reproducibility.Whilst promising, these k app value remain lower compared to previously reported Zn(II)-complexes (k app = 0.44-12.0h −1 ) operating between 50 to 80 °C under analogous reaction conditions. 109,110,123However, to the best of our knowledge, [Mg(1,3)] 2 represent the first example of PLA methanolysis mediated by a well-defined discrete Mg(II)-complex, operating under significantly milder conditions relative to Petrus et al., 111 who relied upon metallic Mg or Mg( n Bu) 2 as pre-catalysts.For [Mg(1,3)] 2 , inspection of the 1 H NMR (Table 3, entries 12 and 16) following solvent removal revealed the formation of a new Mg(II)-species, although its identity remains unclear.It is suggested the dimeric framework dissociates in solution, affording a heteroleptic complex of the general formula Mg(1,3)L, where L could be methoxy, lactyl or higher chain oligomers, as previously described by Jones and coworkers. 123Consequently, [Mg(1,3)] 2 should strictly be regarded as pre-catalysts and this can likely be extended to the remaining Zn(II)-complexes.
PET degradation.Presently, bio-based plastics account for ca.1% of all processed plastics, with PLA accounting for just 13.9% of bioplastic production in 2019. 126Consequently, our attention shifted to PET, a commercial polyester widely exploited in the packaging industry, which consumed 38% of plastics produced globally in 2015, with PET accounting for 22.6% of plastic use in the sector. 13Glycolysis is the most widely used chemical recycling method for PET, characterised by cleavage of the ester bond via insertion of a glycol, commonly ethylene glycol (EG), to produce bis(2-hydroxyethyl) terephthalate (BHET) or higher alcohol derivatives.8][129][130][131] In light of this, preliminary work sought to apply [Zn(1,3)] 2 and [Mg(1)] 2 to the glycolysis of PET.[Mg(3)] 2 was not investigated due to insufficient yield.Typically, high temperatures (180-240 °C) and prolonged reaction times (0.5-8 h) in the presence of a transesterification catalyst, often a metal acetate, are required to achieve appreciable conversion.Whilst numerous metal acetate catalysts have been reported in the literature, zinc acetate is considered the benchmark. 127,132onsequently, Zn(OAc) 2 •2H 2 O (Sigma Aldrich) was chosen as an air-stable, commercially available reference.Additionally, high EG : PET (≥5 : 1) are used to mediate the formation of higher chain oligomers, thus favouring the formation of BHET. 127As such, a reaction temperature of 180 °C in the presence of 8 wt% catalyst and 27.5 equivalents of EG was chosen (Table 5).Two sources of PET were selected: 1.A carbonated drinks bottle (M n ∼40 000 g mol −1 ) and 2. Thin-films, representing waste from the manufacturing industry (Fig. 7).
Typically, a 6-12% reduction in Y BHET (isolated yield) was observed on accounting for residual H 2 O (1.3-3.5 equivalents) ). PVC contamination as low as 100 ppm has previously been reported to adversely impact the quality of the final recycled product, owing to the production of acid catalysts that facilitate chain scission under melt reprocessing conditions (T = 160 °C) routinely employed in industry. 127Thus, this result is particularly promising from an industrial perspective.

Fig. 2
Fig.2Solid-state structure of a tetrameric Mg(II)-complex based on an amine-deprotonated derivative of 2H 2 .Ellipsoids shown at 30% probability.All hydrogen atoms and methyl groups of the t Bu groups have been omitted for clarity.