Low-Temperature and Purification-free Stereocontrolled Ring-Opening Polymerisation of Lactide in Supercritical Carbon Dioxide

A stereoselective, solvent-free ring-opening polymerisation (ROP) of lactide (LA) in supercritical carbon dioxide (scCO 2 ) is reported for the first time. The key aim is to exploit scCO 2 to lower the temperature of traditional melt polymerisations, lowering the energy requirement and leading to cleaner polymeric materials. We have utilised a zirconium amine-trisphenolate initiator-stereoselective catalyst [( i PrO)Zr(OPh( t Bu) 2 -CH 2 ) 3 N] to yield highly heterotactic poly(lactide) (PLA) homopolymer (P r = 0.74 – 0.84) from rac -LA, demonstrating control of the PLA microstructure in scCO 2 . In addition, high monomer conversion (86 – 93%) was achieved in short reaction time (1 h), affording poly(lactide) with a very low degree of transesterification and narrow molecular weight distribution. Most importantly, all the reactions were performed at only 80 °C, almost 100 o C lower than the conventional melt process (typically performed at 130-180 °C), representing a very significant potential energy saving.


Introduction
Biodegradable and bio-derived materials are receiving significant research interest, both as potential replacements for petrochemically-derived commodity polymers and for use in biomedical applications; 1-6 polyesters represent one such class of these materials. 7,8 Poly(lactic acid) (PLA), a linear polyester that can be derived from sugars, has shown particular promise through its unique physical properties, biocompatibility and biodegradability. [9][10][11][12] PLA is synthesised through ring-opening polymerisation (ROP) of lactide, the cyclic dimer of lactic acid, which facilitates control of the polymer molecular weight and dispersity, 13 and end-group functionality. 14,15 ROP of LA has been demonstrated utilising a wide variety of metal, [16][17][18][19][20][21] and organo-catalytic species, [22][23][24][25][26][27] and is employed commercially in the production of PLA. 28 The toxicity associated with, and cost of removal of, residual organic solvents mean that PLA is produced industrially in solvent-free, melt-phase conditions at high temperature (up to 180°C). 29 Such high temperatures are required to maintain mobility of the polymer and this leads to high production costs and significant challenges in maintaining control over polymerisation. 30 For example, whilst many catalysts can control the PLA microstructure at lower temperatures, 31 examples of stereoselective ROP of lactide at 180 °C or above are very rare because transesterification reactions are significant at T ≥ 150 °C. 32 To our knowledge, Nomura et al. have reported the only system to date that attains useful levels of stereocontrol at 180 °C. 29 Furthermore, at these elevated temperatures transesterification reactions, in addition to catalyst deactivation, can often occur to produce atactic PLA. Unwanted polymer degradation pathways are also operative, which can impact molecular weight control. 33 Supercritical carbon dioxide (scCO2) has found utility as a "green" solvent. It is abundant, inexpensive, is easily removed at the end of the reaction, and can even be recycled. Furthermore, diffusion of scCO2 into the free volume between polymer chains can weaken intermolecular interactions and increase chain mobility. The resulting reduction in melt and glass transition temperatures and the lowering of polymer viscosity can bring many advantages, including facilitating low-temperature processing. [34][35][36][37] The use of scCO2 as the medium for a variety of polymerisations has been demonstrated, [38][39][40] including ROP of cyclic esters (including lactide) using both metal and organic catalysts. [41][42][43][44][45][46][47][48][49] These synthetic strategies have largely required lengthy reaction times (5 -24 h). 50,51 Significantly, none of these approaches have demonstrated stereocontrolled ROP.
Zirconium amine trisphenolate species [e.g. ( i PrO)Zr(OPh( t Bu)2-CH2)3N] (1)] have been reported as highly heterotactic-selective initiators for the ROP of rac-LA. Notably, they demonstrate enhanced air stability that facilitate manipulation avoiding glove-box technology. 52,53 In this regard, Chmura et al. 53 reported the synthesis of highly controlled and heterotactic PLA. However, to make this happen required either toluene as reaction solvent or high temperatures for the melt polymerization conditions (130 °C); a series of organic solvents (e.g. DCM, methanol) were also required for the purification step of the final product.
In this report, we demonstrate that use of scCO2, as an industrial scalable alternative could facilitate the stereoregular ROP of rac-LA under organic solvent-free conditions at significantly lower temperatures (80 °C) than those used in the conventional melt-phase polymerisation (130 -180 °C). This not only represents a more energy-efficient greener and scalable process for the synthesis of PLA, but also affords products with controlled microstructure.

Monomer and Polymer solubility in scCO2
To investigate the reaction in detail, we observed first the solubility of rac-LA in scCO2 through view cell experiments. Thus, different amounts of monomer were charged into a view cell autoclave, which was then heated and pressurised with CO2. At 80 °C and 240 bar we found that 5 wt/v% rac-LA, was almost completely CO2-soluble. Raising the pressure to ca. 300 bar, and hence raising the density, led to full solubility ( Figure 2). At increased monomer loading (7 wt/v%), solubility was again observed at 80 °C and ca. 300 bar CO2. However, saturation was quickly reached upon further increase in monomer loading (10 wt/v%) and full solubility in scCO2 could no longer be achieved at 300 bar and 80 °C (Figure 2, right). This is consistent with the findings of Stassin and Jerome, who studied the solubility of L-LA at a range of temperatures and pressures showing that scCO2 can lower the melt temperature of L-lactide, that it has partial solubility in scCO2. 54,55 In addition, the Tm of L-LA is lowered from ~100 °C to ~ 50 °C and we recently showed that similarly the Tm of rac-LA can be lowered from ≈ 130 °C to 95 °C in the presence of scCO2. 56 This monomer solubility in scCO2 can be exploited to ensure there are minimal monomer residues in the final polymeric materials. 56 By contrast, the polymers are completely insoluble in scCO2. But, CO2 is soluble in the polymer and this leads to plasticisation 56, 57 and a liquefied polymer system, essentially a polymer melt. These studies and others 54 suggest that ROPs conducted at 80 °C would initially consist of a homogeneous monomer solution in scCO2. As the polymerisation progresses, the polymer precipitates out of scCO2, leading to a biphasic system consisting of a CO2-rich phase and a polymer-rich phase, with lactide monomer likely in both phases. The polymeric products were found to be of moderate molecular weight (entries 2-6: 6700 -9350 Da (triple detection), consistently slightly lower than predicted (Mn(theor)). We expect that the polymer growth may be hindered once precipitated, leading to lower than expected molecular weights and slight broadening of the molecular weight distribution (Ð = 1.05 -1.40), relative to PLA generated in conventional solvent-free ROP. 53

Stereoselective rac-LA Polymerization in scCO2
The polymerisation of rac-LA in scCO2 over a range of temperatures (80 to 130 °C) was conducted (Table 1) utilising initiator 1. Ring opening polymerisations conducted at 80 °C afforded PLA at good conversion after 1 h (Table 1,entry 2: 86%), similar to the room temperature solvent-mediated polymerization with the same catalyst (50% yield in 48 hr) and conventional melt polymerisation at 130 °C (78% yield in 0.1 h). 53 MALDI-TOF mass spectrometry and homonuclear decoupled NMR spectroscopy were used to investigate the degree of control of polymerisations in scCO2. Significantly, the PLA from all polymerisations in scCO2 was found to have high levels of heterotactic enchainment (Pr = 0.73 -0.84), representing the first demonstration of a stereocontrolled ROP of rac-LA in scCO2. There is though a lower temperature limit, and polymerisations conducted at 60 °C did not produce any polymer in one hour (Table 1, entry 1: 0 % conversion). The poor solubility of the monomer at these conditions likely prevents the polymerisation from initiating.        * GPC traces ( Figure S1-S12) and 1 H homonuclear decoupled NMRs (S18- 26) for key entries in Table1 and S1 are reported in Support Information.
Polymerisations at lower temperatures exhibited a greater degree of stereocontrol (Table 1, Figure S5 right) than those conducted at higher temperatures (Table 1,  The data also allow definitive identification of the end group confirming that the i PrO group from 1 is transferred to the growing polymer chain during the initiation of ROP. A low intensity second series of 72 Da mass difference (1 lactyl unit) to the main distribution was identified for some samples and points towards transesterification being just a minor side reaction (Figure 3 RIGHT). The intensity of this second series, and thus the amount of transesterification, was found to increase with reaction temperature, as observed in the spectra obtained for PLA synthesised at 100 °C (entry 3, Figure S1), and 130 °C (entry 4, Figure S2).
In order to probe the polymerisation system further, a series of reactions at different M/I ratios, pressures and monomer concentrations were performed to broaden the initial understanding of the effects of these parameters.
For Pressure: At 60 bar (Table S1 entry 8) no reaction was observed, likely due to the very low solubility of the monomer under these conditions, and also the minimal effect of plasticisation. At 120 bar a conversion of above 80% with a Pr of 0.82 was observed (Table S1 entry 9). This is close to the best yield obtained at our optimal 240 bar, but with only poor control in terms of dispersity. This likely reflects the lower level of plasticisation at the lower pressure (and hence lower density of CO2).
For monomer concentration: when 0.1 g (Table S1 entry 10) of monomer was loaded no reaction was observed, likely due to the high dilution of the specific condition.

For M/I Ratio:
At an M/I ratio of 150:1 (at 80 °C and after 1h, Table S1 entry 11), reasonable agreement was detected between the theoretical and the measured molecular weight (Mn,theo = 13,200 Da; Mn = 14,100 Da) with a Pr of 0.82 and a conversion of monomer into polymer of around 60%. When the M/I was raised to 300:1 (at 80 o C and after 1h, Table S1 entry 12) a drop in monomer conversion to around 45% was observed. Under these conditions, we still see good control in terms of Pr and dispersity but the resulting polymer showed a molecular weight of only 15000 Da. In an attempt to increase the monomer conversion, and the final PLA molecular weight, two further reactions were carried out for longer time periods of 16 and 24h. Despite the higher conversion (circa 85%), we still find that the molecular weights plateau between 10000 and 12000 Da and there is no significant increase. Instead, a direct consequence of the longer reaction time was the disruption of the Pr values that dropped from 0.82 after 1h (Table S1 entry  increased the monomer conversion from 40% to 70% after 1h of reaction (Table S1 entry 15). However, the final molecular weight observed was still limited to around 15000 Da as with all the other attempts with the same M/I ratio of 300; different temperatures and reaction times.
The limitation of molecular weight is puzzling. It could be that residual contaminants e.g. water in the scCO2 are having a deleterious effect. Or, it could be that as the polymer molecular weight increases, we reach a point where the viscosity in the melt is sufficiently high that monomer penetration to the active metal site is limited.
In both cases further studies and optimisation will help us. However, this should not detract from our key message that use of scCO2 effectively lowers the temperature required for ROP of lactide, suppressing transesterification reactions, with a concomitant increase in the stereo-control of the polymerisation. There is also a significant additional benefit with our novel proposed route. On the current industrial scale process, a high temperature devolatilisation step is required to remove residual lactide monomer. In fact, we have already successfully demonstrated that the monomer can be easily extracted from crude reaction mixtures by scCO2assisted extraction (exploiting the monomer solubility and the polymer insolubility in scCO2). 56

Conclusions
Routinely, conventional melt-phase polymerisations require high temperature (over 130°C). On the commercial scale temperatures are much higher and the residence time at high temperature is long and these  shorter than for previous reports of PLA synthesis in scCO2 and affords PLA with little or no detectable transesterification as well as narrower molecular weight distribution. In addition, the MALDI-TOF data allowed definitive identification of the end group confirming that the i PrO group from 1 is transferred to the growing polymer chain during the initiation step remarking the controlled nature of the reaction.
The preliminary results presented in this work suggest that the initiator/catalyst system could be used in organic-solvent-free ROP and will provide access to a range of PLA stereo-architectures at lower temperatures with only minimal transesterification observed at those temperatures. In addition, the current commercial processes require a devolatilization step to remove residual monomer which requires a very significant additional energy input. Since the monomer is soluble in scCO2, we can envisage a completely new approach, lowering the number of steps by exploiting the scCO2 to eliminate the separate energy intensive devolatilisation steps that are currently essential on the commercial scale.