Synthesis of glycol-modified poly(ethylene glycol) (PETG) particles
The initial focus for polyester particles was commercially available glycol-modified poly(ethylene glycol), PETG (Table 1). This copolymer can be considered a non-crystalline PET analogue,(23, 24) and is frequently used as a durable, transparent and inexpensive material in applications where the crystallinity of PET is not desirable. The approach pursued here was that a relatively small fraction of block copolymers of a main-chain polyester or polycarbonate and a hydroxy-terminated water-soluble polymer can be prepared through transesterification (see Scheme 1A). This small amount of amphiphilic block copolymer can then act as a stabiliser of solution droplets or particles upon dispersion into water, depending on the water-miscibility of the solvent used (see Scheme 1C).
Transesterification has been reported using a variety of catalysts.(25, 26) Here, zinc acetate was used due to its low toxicity. Anisole was chosen as a solvent for PETG (as well as for polycaprolactone (PCL), polylactic acid (PLA) and polycarbonate (PC), see below) because it was found to dissolve PETG well at high temperatures, while at the same time having low reported toxicity. In addition, it has a significant vapour pressure despite its relatively high boiling point of 154 ºC. The high boiling point allows conducting the reaction at high temperature, while the high vapour pressure allows relatively efficient removal of solvent residues by reduced pressure at lower temperatures, thus avoiding particle degradation. The direct transesterification with poly(ethylene glycol) monomethyl ether (MPEG) is shown in Scheme 1A.
Initially the rate of transesterification was investigated by size exclusion chromatography using a 1:500 molar ratio between hydroxy groups from MPEG and ester groups (corresponding to 4.4% w/w MPEG). It should be emphasised that all reagents were deliberately used as received, hence some additional hydrolysis may be expected due to residual water content in polymers, catalyst and solvent. In the SEC traces of the sample taken at t = 0 (prior to addition of catalyst, Fig. 1A, black trace), the sharp signal from free MPEG is clearly visible along with the trace of the broad PETG signal. After 1 h most of the MPEG signal has disappeared (Fig. 1A, dark grey trace), although the number average molecular weight of the mixture has not changed significantly (Fig. 1B). These results are consistent with transesterification, since this reaction will not affect the total number of molecules.(27) At longer reaction times, Mn is seen to decrease along with an increase in the MPEG signal, which indicates that concurrent hydrolysis occurs during this period. The exact reason for the delay in hydrolysis was not investigated, but it may be related to the high temperature and low solubility of water in anisole, which reduces the amount of water available for hydrolysis.
After establishing that transesterification was essentially complete in 1–2 hours, the dispersion was effectuated using the setup shown in Scheme 1C; After cooling the polymer solution to 80 ºC, it was transferred into 5 volumes of room temperature water while continuously agitating with a high-speed disperser.
The resulting white dispersion consisted of droplets that were mainly smaller than 10 µm in diameter (see Fig. 2B, top left image). Removal of anisole and water gave PETG particles of comparable size that could be directly redispersed in water (see Fig. 2B, bottom left image).
Particle analysis using both light microscopy and a light scattering method(28) gave comparable results (see Table 2) with no macroscopic precipitation observed, indicating efficient stabilisation and that the image is representative for the entire sample.
Optimisation of stabiliser content
Figure 2A shows the size of dispersed droplets and dried particles as a function of stabiliser content. Below 4.4% w/w MPEG, the particles increase in size and size distribution, which indicates poor stabilisation. On the other hand, further increasing the stabiliser content to 19% w/w (corresponding to an alcohol:ester ratio of 1:100) only leads to a small decrease in the as-dispersed droplet size, whereas the re-dispersed particles appear to be slightly larger. Thus, for preparing small particles while maintaining a significant molecular weight and employing as little stabiliser as possible, the ratio of 4.4% w/w appears to be a good compromise.
Synthesis of particles of poly(ethylene terephthalate), polycaprolactone, polylactic acid and polycarbonate
Following the successful preparation of PETG microparticles, the same procedure was adapted to prepare microparticles of other commonly used polymers. Specifically, polycaprolactone (PCl), polylactic acid (PLA) and polycarbonate (PC) were used as substrates (Scheme 1A), since these polymers were all soluble in refluxing anisole and should in principle be able to undergo transesterification/transcarbonation reaction with hydroxy-groups. In particular, the molar ratio between alcohol and ester was kept constant in order to maintain the effect on molecular weight during transesterification. Since the MPEG groups act as stabilisers and thus are predominantly placed on the interface between polymeric and aqueous phase the maximum required stabiliser amount is related to the volume fraction rather than the molar fraction. Therefore, the amount of stabiliser can probably be reduced for most of these polymers, since the molecular weight of the repeating unit is smaller than for PETG while the densities are comparable.
As seen in Fig. 3, PCL, PLA and PC formed spherical, micron-sized droplets with a size comparable to those formed by PETG, illustrating the versatility of the method.
Due to PET not being soluble in anisole, the formation of PET particles was carried out in dimethylsulfoxide (DMSO) instead. Amorphous PET dissolves easily in DMSO near the boiling point, but eventually precipitate at lower temperatures. However, it is desirable to cool the reaction mixture below the boiling point of water to avoid hazardous over-heating during the subsequent mixing step. In addition, DMSO is miscible with water, which means that particles, rather than dispersed droplets, are formed immediately. At 150 ºC the dissolved PET did not precipitate from solution. Relatively slow transfer of this heated solution into excess water under the action of a high speed disperser led to particle formation. As seen in Fig. 3, Fig. 4 and Table 1, the resulting particles are on average less than 2 µm and easily redispersed.
Table 1
Details on synthesized particles.
Name | Added MPEG % w/w | Measured MPEG % w/wa | Mn g/mol b | Ð b | Tg ºC | TC ºC | wC % c | DMicro µm d | DNC µm e |
---|
MPEG-PETG | 4.4 | 3.4 | 17,000 | 1.62 | 87 | - | 0 | 2.7 ± 0.6 | 2.5 |
MPEG-PET | 5.0 | 2.6 | N/M | N/M | 59 | 250 | 20f | 1.4 ± 0.08 | 1.5 |
MPEG-PCL | 8.7 | 6.5 | 13,200 | 1.44 | -43 | 55 | 69 | 2.7 ± 0.8 | 1.9 |
MPEG-PC | 4.2 | 2.3 | 11,100 | 1.82 | 118 | - | 0 | 5.9 ± 3.4 | 2.0 |
MPEG-PLA | 13 | 11 | 13,300 | 1.53 | 40 | 145 | 25 | 3.0 ± 1.5 | 1.5 |
MPEG-PE | 18 | 1.3 | N/M | N/M | N/M | 98 | 37 | 22 ± 15 | 3.9 |
PET, No Stabiliser | 0 | 9.0g | N/M | N/M | 98 | 256 | 40 f | 2.0 ± 1.1 | 3.2 |
a Measured by 1H NMR. b Measured by SEC relative to polystyrene standards. c Determined as measured peak enthalpy divided by standard polymer heat of fusion according to literature.(29) d Diameter of particles dried and redispersed in water, analysed using optical microscope. The standard error is given as the uncertainty. e Diameter of particles dried and redispersed in water, analysed using a Nanocuvette™ S. f Crystallinity determined using modulated DSC. g This value corresponds to residual DMSO as measured by 1H NMR |
Synthesis of polyethylene (PE) particles
Polyethylene is one of the most common commodity plastics, and hence PE-based microparticles are of significant interest as a model microplastic.(30) Polyethylene is of course a polyolefin and not a polyester, why the method presented here is not directly applicable. However, the use of commercially available polyethylene grafted with maleic anhydride (MA-PE) should allow attachment of stabilising MPEG functionalities through reaction between anhydride and alcohol (see Scheme 2B).
The MA-PE used has an acid number of around 6 mg KOH /g according to the manufacturer, which corresponds to around 10− 4 mol acid/g. Since two acids make up one anhydride, MPEG was added stoichiometrically to the anhydride as 5•10− 5 mol hydroxy groups, corresponding to 18% w/w (see Table 2). In this case no catalyst was used and the reaction was carried out in refluxing toluene rather than anisole. The reaction mixture was readily dispersed into ethanol to give discrete particles. The ethanol was then exchanged with water, which led to rapid creaming of the MPEG-PE particles with lower density than water. As seen in Fig. 3, Fig. 4 and Table 1, the resulting particles are significantly larger than observed for particles based on the polyesters and also appear to aggregate more. The light scattering results shown in Table 1 appears to contradict this, but this is probably a consequence of the size of the particles; the larger particles rapidly rises, effectively removing them from the light path. As a consequence, only a small amount of well-dispersed particles dispersed are analysed. It should be noted that a control experiment omitting the MPEG led to macro-phase separation and essentially no discrete particles, indicating the importance of the stabilising groups being present.
Analysis of the resulting particles by 1H NMR (see supporting information) reveals that the actual MPEG content is significantly lower compared to the targeted amount at only 1.3% w/w (see Table 2). This low MPEG content indicates either poor reaction efficiency or low stability of the resulting ester bond and probably explains the larger particles observed. The particles were nevertheless included in the present study because of the wide-spread use of PE.(30) If hydrolysed, the anhydride contributes with carboxylic groups in addition to the MPEG stabilising groups, and this distinguishes these particles from ‘pristine’ polyethylene particles. These may, if neutralised, contribute to charge-stabilisation of the particles. In addition, such carboxylic groups are found in polyethylene exposed to UV irradiation(31) and are therefore likely to be found in environmentally generated microplastics. The acid groups may act as a substrate for CoEnzyme A,(31) why these particles are likely to be more prone to biotic degradation than untreated polyethylene.
Synthesis of PET particles without stabiliser
In order to compare the influence of stabiliser on degradation, PET particles without a stabiliser were synthesised. As the role of the stabiliser generally is to prevent particle fusion, preparing comparable particles without stabiliser is not straightforward, although there are literature examples such as using trifluoroacetic acid as a volatile stabiliser(32) or through extensive milling.(33) However, both these methods have drawbacks: Trifluoroacetic acid is a strong acid, which sets demands on equipment, especially if large amounts of particles must be prepared and in addition it may react with the polyester leading to end-group functionalisation,(34) which alters the polymer composition somewhat. On the other hand, milling leads to formation of micron-sized particles but the preparation of particles with a comparable size to those reported here requires extended milling times of around 8 hours,(33) which will probably lead to some amount of polymer degradation.
Here we found that cooling a 10% w/v solution of an amorphous PET film dissolved in DMSO to room temperature led to the formation of a precipitate of crystalline PET particles with sizes comparable to those obtained from the use of MPEG as a stabiliser when redispersed in water (see Table 1 and Fig. 4). This procedure was found to work well for PET, but not for any of the other polyesters. The particle formation is presumably driven by the poor solubility of crystalline PET in DMSO at room temperature, which is emphasised by the relatively high degree of crystallinity of these particles compared to those prepared using the stabiliser (40% vs 20%. See Table 1). In the former case, the amorphous part of the PET is plasticised by DMSO and may reorganise to form crystals, whereas the stabilised particles are quickly dispersed into water, which is a non-solvent to both amorphous and crystalline PET. As a consequence, crystals have a lot less time to form, leading to a lower degree of crystallinity.
The resulting particles were found to have less than 10% residual DMSO (see Table 1) as determined by 1H NMR.
Aerobic biodegradation
Selected polymer particles were chosen for the study of aerobic biodegradation in seawater. In particular, PET particles prepared with and without stabiliser were chosen to assess possible effects of the stabilising groups. In addition, PCL was chosen as a degradable polyester known to degrade under these conditions,(35) whereas PLA was chosen as an aliphatic polyester, known to be biodegradable under industrial composting conditions but not to degrade under aerobic aqueous conditions.(36) Finally the biodegradation of stabilised PE particles was investigated due to the extended use of this polymer and consequently high occurrence in the environment.(37)
Figure 5A shows the degradation results after 2 and 4 weeks. The positive controls, sodium acetate and microcrystalline cellulose degrades as expected, which confirms that the microorganisms are active. For the PET particles with and without stabiliser, as well as for the stabilised PE particles and the stabilised PLA, some degradation is apparent, which at a first glance is surprising as neither of these are expected to degrade under the conditions employed. On the other hand, the stabilised PCL particles degrade as expected (although the value after 15 days has been omitted because the maximum capacity of 80 mg/L BOD of the system is exceeded).(36)
Since the MPEG stabiliser is situated mainly on the particle surface, it is the degradation of this entity that is observed on the otherwise non-degradable particles, since microbial degradation of MPEG is well known.(38) In order to examine this hypothesis, the data in Fig. 5B shows the biodegradation assuming that only the MPEG (where present) contribute to the theoretical oxygen demand. For the MPEG-stabilised PET and PLA, the observed degradation correlates well with the maximum calculated ThOD for the MPEG only (Red line in Fig. 5B), which supports that at least for these two sets of measurements, MPEG degradation is observed, whereas the bulk of the polymer mostly remains unaffected.
Similarly the apparent degradation of the non-stabilised PET particles corresponds well with the residual DMSO content.
As shown in Fig. 5A, MPEG-PE particles show a small, but significant, biodegradation. This degradation is not entirely due to the presence of MPEG groups on the surface, since the actual degradation is significantly larger than what can be explained by MPEG alone (Fig. 5B). However, as mentioned above, biotic degradation of polyethylene has been reported to be promoted by the presence of carbonyl groups,(31) indicating that it is the presence of functional groups on the polymer backbone that makes these polyethylene particles more prone to degradation.
Further insight into the degradation behaviour can be obtained by considering the entire time series in Fig. 6. In the first two days, low activity is observed for all polymers, where the biodegradation for all polymers except for MPEG-PE have very similar values. Such lag phases have been explained by adaptation of the microorganisms to produce enzymes that can depolymerise the particles.(36) After two days, degradation of MPEG-PCL commences, eventually reaching around 40% degradation at 15 days. It is worth noting that this is significantly faster than comparable data for particles with sizes above 100 µm previously reported.(36) As such it is consistent with smaller particles degrading faster due to their larger surface area, although the MPEG surface groups and differences in microorganisms are also likely to have an influence.