Milling and grinding

It is almost impossible in the laboratory to mimic or reproduce all of the degradation processes simultaneously acting up an item of plastic debris in the natural environment. However, some individual degradation mechanisms can be reproduced in the laboratory quite accurately and effectively, with the potential for more than one to occur at the same time. A good starting point is to grind or mill pieces of plastic as finely as possible, and this process has generally been shown to produce polydisperse particles with a range of sizes, highly irregular shapes and surface morphologies representative of plastic particles found in the environment (Takacs, Reference Takacs2002; Seghers et al., Reference Seghers, Stefaniak, La Spina, Cella, Mehn, Gilliland, Held, Jacobsson and Emteborg2022). The two primary forms of milling available are rotary milling and ball milling, with the former acting more as a ‘cutting’ technique and the latter acting as a ‘smashing’ technique. Crucially, milling approaches allow the plastic-associated chemical composition of the start material to be preserved, as long as any subsequent fractionation or particle size range isolation step avoids the use of water or other solvents. For the production of MP >30 μm in size, these methods have proven to be relatively cheap and fast (Seghers et al., Reference Seghers, Stefaniak, La Spina, Cella, Mehn, Gilliland, Held, Jacobsson and Emteborg2022), although the production of large quantities of material can involve some investment of time. It is also important to note that different polymer types respond differently to the process, which impacts the final yields and particle size/morphology.

A critical issue with common grinding and milling techniques, however, is that they are not able to grind the plastics finely enough to produce reasonable yields of sMP and NP (Eitzen et al., Reference Eitzen, Paul, Braun, Altmann, Jekel and Ruhl2019; Ciobotaru et al., Reference Ciobotaru, Marcu, Maria, Ivanov, Savin, Moncea, Tociu and Deák2020). While particles in this size range are often reported as being produced, they typically represent yields of <<1%, if they are calculated and reported at all. It was found that approximately half of the reviewed papers applied some form of grinding or milling process to produce sMP and NPs, while most others utilised previously ground or milled particles as a starting material (Table 1). For thermoplastic materials, which may soften in ambient temperatures, a common approach has been to cool the material (cryomilling), making it more brittle and facilitating the fragmentation process into smaller particles sizes. Cryomilling is conducted at low temperatures (e.g., using liquid nitrogen), which increases the yield of fine structured particles and allows for rapid grain refinement (Suryanarayana, Reference Suryanarayana2001). Cryogenic conditions can be applied to both ball and rotary milling, either by pre-cooling the sample to be milled or by pre-cooling the sample holder and maintaining a low temperature throughout the milling process (Eitzen et al., Reference Eitzen, Paul, Braun, Altmann, Jekel and Ruhl2019). While cryomilling has been shown to be effective for producing larger MP reference materials (>30 μm) in reasonable yields (Kühn et al., Reference Kühn, van Oyen, Booth, Meijboom and van Franeker2018), the technique is still unable to produce larger quantities of sMP and NP particles (Eitzen et al., Reference Eitzen, Paul, Braun, Altmann, Jekel and Ruhl2019).

It has recently been demonstrated that conducting cryomilling and ball milling at ambient temperature in series can help to increase the production of smaller particles, especially NPs (Caldwell et al., Reference Caldwell, Lehner, Balog, Rhême, Gao, Septiadi, Weder, Petri-Fink and Rothen-Rutishauser2021). Although yields were not explicitly reported, the authors describe stock solutions of the final NP materials at concentrations ranging from 4.8 to 40.8 μg mL⁻¹, which was sufficient for conducting toxicity tests in the study. A combination of dry and wet milling in water, solvent or surfactant has also been shown to improve the yield of lower size ranges (Bhagia et al., Reference Bhagia, Gallego, Hiremath, Harper, Lowden, Lowden, Pu, Vaidya, Ozcan and Ragauskas2021), and has been applied in most of the studies reviewed herein (Table 1). The approach has achieved the production of non-homogenous (in size and shape) particles for a range of polymers, including PS, PP, PE, PET, polybutylene adipate terephthalate and polyether ether ketone. Wet high-speed friction grinding has recently been investigated as an option for the production of MP and sMP, and a significant improvement in the yield of PE, PP and PLA particles in the lower size range was obtained by modifying the aqueous grinding media with 0.5% guar gum (Bhagia et al., Reference Bhagia, Gallego, Hiremath, Harper, Lowden, Lowden, Pu, Vaidya, Ozcan and Ragauskas2021). Minimal impact on the physicochemical properties was observed by this process.

Ultrasonication

Exposing plastic materials and particles to ultrasonication, where sound waves create mechanical forces that can help to break down plastic particles, has the potential to produce MP and NP test/reference materials. To date, only a few studies have been conducted using this approach, but they have demonstrated promising results, including the production of particles in the nanoscale (von der Esch et al., Reference von der Esch, Lanzinger, Kohles, Schwaferts, Weisser, Hofmann, Glas, Elsner and Ivleva2020). Ultrasonication is typically conducted by suspending the material to be fragmented in a solvent, where different chemical characteristics can further influence the fragmentation process. For example, using alkaline conditions has been proven to be effective (von der Esch et al., Reference von der Esch, Lanzinger, Kohles, Schwaferts, Weisser, Hofmann, Glas, Elsner and Ivleva2020). A significant advantage of ultrasonic-based methods is the wide availability of ultrasonication equipment in laboratories globally, increasing the possibility for many researchers to produce their own test materials. These methods are also cost and time efficient. However, there are still some limitations with the technique, including sonication being a time-demanding method and that it requires further development and harmonisation in order to achieve reproducible results when using different sonicators (von der Esch et al., Reference von der Esch, Lanzinger, Kohles, Schwaferts, Weisser, Hofmann, Glas, Elsner and Ivleva2020).

Oxidative degradation

An environmentally relevant method for degrading polymers and plastic materials is to expose them to UV radiation. Many studies have investigated the environmental UV degradation of plastics in the laboratory under simulated sunlight conditions, typically reporting significant fragmentation and chemical (polymer and additive) composition changes to the start material (Gigault et al., Reference Gigault, Pedrono, Maxit and Halle2016; Lambert and Wagner, Reference Lambert and Wagner2016; Sait et al., Reference Sait, Sørensen, Kubowicz, Vike-Jonas, Gonzalez, Asimakopoulos and Booth2021; Sørensen et al., Reference Sørensen, Groven, Hovsbakken, Del Puerto, Krause, Sarno and Booth2021). The irradiation causes photocatalysed oxidative degradation, resulting in broken polymer chains and a reduction in the molecular weight of the particles (Yousif and Haddad, Reference Yousif and Haddad2013). In such studies, the UV wavelengths are usually limited to those present in natural sunlight (e.g., UVA and UVB; >280 nm), but the time scales required to achieve significant degradation typically range from months to years (Sait et al., Reference Sait, Sørensen, Kubowicz, Vike-Jonas, Gonzalez, Asimakopoulos and Booth2021). The process can be sped up by conducting the UV exposure continuously and using high UV intensities but kept within natural ranges. However, shorter wavelengths are more effective at degrading plastic materials, and exposing them to light containing wavelengths that include the UVC region (200–280 nm) can significantly speed up the degradation process (Lee et al., Reference Lee, Busquets, Choi, Lee, Kim and Campos2020; Doğan, Reference Doğan2021), as can using intensities above natural levels. It should be noted that UVC exposure may lead to the formation of different degradation products and oxygenated radicals compared to UVB exposure (Doğan, Reference Doğan2021).

One way of achieving this enhanced degradation rate in the laboratory is to expose the starting polymer material in a UVC-Ozone (UV-O) chamber (Sarkar et al., Reference Sarkar, Rubin and Zucker2021), such as those commonly used for sterilisation purposes. The UV-O exposure facilitates the UV irradiation of the particles while simultaneously oxidising them via ozone, leading to radical reactions that break down the particles into smaller fragments (Sarkar et al., Reference Sarkar, Rubin and Zucker2021). The application of UV-induced fragmentation for the production of test materials has been investigated for polyhydroxybutyrate and PS, in two different studies, producing sMP and NP, respectively (González-Pleiter et al., Reference González-Pleiter, Tamayo-Belda, Pulido-Reyes, Amariei, Leganés, Rosal and Fernández-Piñas2019; Sarkar et al., Reference Sarkar, Rubin and Zucker2021). While the approach has been demonstrated to produce sMP and NP particles, yields and production volumes were not reported in either study. Advanced oxidation techniques such as UV-O typically require costly specialist equipment, and while degradation rates may be fast for UV-labile polymers, the method is not applicable to UV-stabile polymers. The applicability of such methods for the harmonised production of test and reference materials is therefore likely low.

Laser ablation

Magrì et al. (Reference Magrì, Sánchez-Moreno, Caputo, Gatto, Veronesi, Bardi, Catelani, Guarnieri, Athanassiou, Pompa and Fragouli2018) applied laser ablation (irradiation wavelength 248 nm) to create top-down NP fragments of PET at concentrations of up to 300 μg mL⁻¹. Small NP size ranges were isolated by filtration (0.2 μm), and the mean diameter was found to be ~30 nm, one of the smallest sizes among the studies addressed herein. The benefits of this protocol would include not needing to add any particular solvent or stabiliser, whereas the downsides include the low yield and application of (costly) specialist instrumentation and time consumed. Upon characterisation, the produced particles were found to be nearly spherical or spherical, and oxidised at the surface compared to the original material. It is hypothesised by the authors that the oxidised surface is representative of (environmentally) degraded PET, but this is not proven. This technique potentially allows the production of NPs without impurities, chemical precursors and the byproducts associated with bottom-up processes.

Combined degradation protocols

As highlighted above, combining multiple degradation mechanisms has the potential to be more effective at degrading plastic materials and generating sMP and NP test materials in usable quantities (e.g., environmental fate and effects assessment). In an attempt to produce sMP particles in the range 1–10 μm, Sarkar et al. (Reference Sarkar, Rubin and Zucker2021) first used cryomilling to produce PS particles of <100 μm, followed by a sequential combination of thermal treatment (70°C) followed by UV(C)-ozone exposure (both dry) and probe sonication in ethanol–water suspension. The application of the post-cryomilling steps increased the number of particles in the 1–10 μm size range, although the yields were not determined. While not measured and reported directly by the authors, the size distributions presented indicated that an increase in the NP size fraction was likely also obtained using this method. The produced sMP particles were found to be comparable in size, shape, surface morphology and functionality to particles subjected to environmental weathering. However, the measured surface charge (zeta potential, ZP) was significantly different between the sMPs produced by this accelerated protocol and those produced from environmental weathering. Compared to the spherical model PS and environmentally weathered particles, the PS sMPs from the accelerated protocol represented an intermediary ZP. A major downside to the protocol may be the difficulty in scaling up the process to facilitate the bulk production of sMP (and potentially NP). Furthermore, the approach was optimised particularly for PS, known to be ‘easily’ UV-degradable due to its carbon–carbon backbone (Yousif and Haddad, Reference Yousif and Haddad2013; Gewert et al., Reference Gewert, Plassmann and MacLeod2015), and so its applicability for other polymer types is unknown.

Most recently, Schmitt et al. (Reference Schmitt, Altmann, Fengler and Gehde2023) used a combination of UV exposure and mechanical fragmentation using a custom made air-based fragmentation device to produce high yields of PE MP (~80%) using a thin film/foil as a starting material. The approach was able to produce PE flakes with irregular shapes, where a decrease in size occurred with longer UV exposure times (up to 2.5 months). Although a minimum average particle size of 110 μm was reported in the study, the relationship between particle size and UV exposure times, combined with the high yields, suggests the approach may have potential for further development and assessment for its applicability in producing sMP and NP materials. A common downside to both protocols presented above is the need for specialist instrumentation that is not readily available in most laboratories (e.g., UV-O chamber and custom-built fragmentors), making them difficult to reproduce in other laboratories without significant cost and commitment. On the positive side, both methods appear to be able to produce polymer fragments of environmental relevance, due to their partially degraded nature.

Solubilisation and precipitation methods

Several studies have reported the production of nanosized PET through the dissolution of larger PET particles in a strong solvent, followed by precipitation in (mainly) aqueous media (Rodríguez-Hernández et al., Reference Rodríguez-Hernández, Muñoz-Tabares, Aguilar-Guzmán and Vazquez-Duhalt2019; Johnson et al., Reference Johnson, Mecham, Krovi, Caffaro, Aravamudhan, Kovach, Fennell and Mortensen2021; Elhady et al., Reference Elhady, Elbarbary, Gad and Fathy2022). Another study applied a similar approach for PE, replacing direct precipitation by emulsification with a biosurfactant in seawater (Balakrishnan et al., Reference Balakrishnan, Déniel, Nicolai, Chassenieux and Lagarde2019). Although the methodologies are reasonably simple and transferable, they are highly similar to conventional bottom-up synthesis approaches that produce mainly spherical particles with a narrow size range, and which have limited environmental relevance. Additional downsides include the use of toxic solvents and the need to remove them before any particles can be utilised in toxicity studies. It also remains unclear and undocumented whether such approaches to test/reference material production sufficiently retain the additive chemical composition of the starting material. This is increasingly considered critical for achieving a more accurate and environmentally relevant assessment of sMP and NP.

Most recently, Peller et al. (Reference Peller, Mezyk, Shidler, Castleman, Kaiser, Faulkner, Pilgrim, Wilson, Martens and Horne2022) applied partial solubilisation with small volumes of long-chained solvents to successfully transform larger particles of several polymers (PE, PET and PS) into sMP and NP directly in aqueous media. The relatively simple approach involved the addition of plastic granules, particles or pieces to water, followed by the addition of a small volume (10–30 μL) of long-chain alkanes. Vigorous shaking created a cloudy suspension with average particle sizes of 1.3–4.4 μm for the different materials. Subsequent ultrasound treatment resulted in an average particle size of 338–724 nm. Differences were observed in the particle morphology, with some materials producing smoother and more spherical particles, and other materials producing more irregularly shaped and rougher-surfaced particles. Again, there was no attempt to assess whether the (additive) chemical composition of the resulting particles accurately reflected that of the start material. Given the challenges in producing bulk quantities of sMP and NP particles using mechanical (e.g., cryomilling) and UV-based degradation techniques, dissolution and partial dissolution approaches appear an attractive alternative. The protocols can be easily scaled and facilitate reasonably fast production of particles without the need for specialist instrumentation. However, more work is required to characterise the physicochemical properties of the resulting particles and compare them with those found in the natural environment. Furthermore, there is a need for assessing such techniques for their applicability to a much wider range of polymer types, again reflecting those found in the natural environment.

Yields of sMP and NP

One of the key requirements of test and reference materials is being sufficiently available at a low or reasonable cost. Unfortunately, only a small number of papers report yields or produced mass in a way that allows calculation of yield relative to the starting amount of plastic. Where reported, yields are (on a mass basis) very low (<<1%). Dissolution–precipitation methods appear to be much quicker and easier to perform and may have the potential to be scaled up to provide higher sMP and NP yields. In contrast, entirely top-down methods may appear more desirable in terms of their potential for producing sMP and NP with a higher degree of environmental relevance, but are typically costly, time consuming and only able to produce small amounts of material. For a variety of reasons, very few studies reporting methods for the production of sMP and NP directly assess (or attempt to assess) the environmental relevance of the particles from a physical and chemical perspective (especially the latter). In line with previous assessments, we also see that most reported methods (10 papers) have focussed on production of PS (Figure 1), although this is not the most abundant polymer in the environment (Pradel et al., Reference Pradel, Catrouillet and Gigault2023). A similar number of studies have produced PET particles, whereas only a handful have produced PP and/or PE particles. This probably reflects not only the applicability of the methods to the various polymers, but also the interest in their production. We believe that it is important for methods to be able to produce comparable particles of several/most polymer types found in environmental plastic.

Figure 1. Overview of the number of peer-reviewed manuscripts reporting methods for the production of sMP and NP comprised of various polymer types (left) and the resulting particle sizes (right). Polymer types studied were polystyrene (PS), polyethylene terephthalate (PET), polyethylene (PE), polypropylene (PP), polylactic acid (PLA), polyhydroxybutyrate (PHB), polybutylene adipate terephthalate (PBAT), polyether ether ketone (PEEK) and polycarbonate (PC).

Size and shape of sMPs and NPs

Top-down methods generally produce irregularly shaped particles, whereas particles formed through dissolution and re-precipitation tend to be more spherical in morphology (Johnson et al., Reference Johnson, Mecham, Krovi, Caffaro, Aravamudhan, Kovach, Fennell and Mortensen2021). The studies reviewed here show that the weathering process that plastics undergo in the natural environmental can be reproduced relatively well in the laboratory and that the secondary MP and NP formed share many similarities with particles isolated from environmental matrices. The applied techniques and combinations of techniques typically resulted in a mixture of highly irregular particle morphologies, which is a critical property of naturally weathered plastics. The majority of the top-down studies, (mechanical-, UV- and dissolution-based approaches) were able to produce a continuum of particle sizes, including a proportion of particles in the ranges required for a hypothetical sMP and NP test/reference material (i.e., just over 1 μm and <1 μm, respectively; Figure 1). Particles <100 nm are only reported in a few studies, but this can be caused both by the lack of formation and the lack of appropriate characterisation methods for such small sizes. As such, there is potential for further processing of the products to isolate and characterise specific particle size ranges for use as test materials. Across the reported studies, such top-down approaches have been applied to a variety of polymer types, with varying degrees of success being reported. However, it is important to highlight that such approaches are typically limited by very low yields, low reproducibility, the need for costly instrumentation and a significant time investment.

All studies employed some form of particle size and/or size distribution characterisation, for which a range of techniques was applied. In the reviewed studies, nine used DLS and five used SEM, with other reported techniques including nano tracking analysis (NTA), asymmetric flow field flow fractionation (AF4) fitted with either (i) a multi-angle light scattering detector, (ii) static light scattering (SLS) or (iii) inductively coupled plasma-mass spectrometry, laser diffraction or granulometry, transmission electron microscopy and Raman microscopy. Several studies employed a combination of particle size characterisation methods. It should be noted that the different techniques are based on different principles for assuming size and so the final determined values are not always directly comparable. For example, DLS generally provides larger sizes than optical techniques such as SEM, likely due to a combination of (i) presence of surfactants on the particle surface, (ii) calculation of hydrodynamic radii assuming spheres when actual sMP and NP are not spherical and (iii) particle aggregation (Ji et al., Reference Ji, Wang, Wang, Fu, Man and Chen2020; Caldwell et al., Reference Caldwell, Lehner, Balog, Rhême, Gao, Septiadi, Weder, Petri-Fink and Rothen-Rutishauser2021; Johnson et al., Reference Johnson, Mecham, Krovi, Caffaro, Aravamudhan, Kovach, Fennell and Mortensen2021). Conversely, NTA provides lower, possibly more accurate, average particle sizes than DLS (Ekvall et al., Reference Ekvall, Lundqvist, Kelpsiene, Šileikis, Gunnarsson and Cedervall2019).

Influence of fragmentation methods on particle and polymer physicochemical properties

Both naturally occurring environmental degradation mechanisms and those degradation processes simulated in the laboratory have the potential to impact or modify multiple physicochemical properties of plastic materials, including crystallinity, glass transition and phase transition temperatures (Ainali et al., Reference Ainali, Bikiaris and Lambropoulou2021). From the perspective of environmental relevance of test and reference materials, it is important to have a robust characterisation relative to the start materials in order to ensure they reflect the properties of sMP and NP found in the environment. It is only recently, however, that a few studies have started to document the effects of degradation methods on particle properties beyond their size and shape. For example, Lionetto et al. (Reference Lionetto, Esposito Corcione, Rizzo and Maffezzoli2021) investigated if there were changes in the polymer crystallinity when producing PET NPs through a combination of ultra-centrifugal milling and ball milling. The milling process was found to lead to a progressive reduction in the degree of crystallinity and an increase in the proportion of amorphous material, resulting in an overall change in material properties. Considering that the amorphous regions of a polymeric material are often more susceptible to different degradation mechanisms (e.g., UV degradation and microbial degradation) than the crystalline regions, it is likely that some differences are observed between the parent material and the resulting sMP and NP test materials. For the production of environmentally relevant materials in terms of behaviour, this needs to be considered further and compared to eventual changes in the same properties under environmental or simulated environmental degradation (Ainali et al., Reference Ainali, Bikiaris and Lambropoulou2021; Conradie et al., Reference Conradie, Dorfling, Chimphango, Booth, Sørensen and Akdogan2022). It is also worth highlighting that top-down test/reference material production methods have an increased likelihood of preserving the additive chemical profile of the start material, thus increasing its environmental relevance.

Stability of particles in aqueous suspension

Beyond the initial preparation of the sMP and NP test and reference materials, their dispersion stabilisation and longer-term stability in aqueous media is another key consideration that must be addressed. It is well established that common surfactants can be used to maintain small particles in aqueous dispersion, avoiding aggregation at concentrations that are higher than the environmentally realistic concentrations typically used for test and reference material stock solutions (Ji et al., Reference Ji, Wang, Wang, Fu, Man and Chen2020; Johnson et al., Reference Johnson, Mecham, Krovi, Caffaro, Aravamudhan, Kovach, Fennell and Mortensen2021). However, there are some critical drawbacks to using surfactants when the particles are to be utilised in environmental fate and effects assessment. Any particles previously exposed to surfactants are unlikely to behave in a ‘natural’ way in environmental fate studies, whereas the surfactants may influence the outcomes of toxicity studies. Despite these drawbacks, there has been an unfortunate overuse of stabilised PS spheres for environmental fate and effects studies conducted to date (Pradel et al., Reference Pradel, Catrouillet and Gigault2023) and it is critical that research moves towards more realistic assessments.

It is possible to look towards the natural world for inspiration when it comes to the dispersion stabilisation of sMP and NP test/reference materials. Most small particles present in natural aquatic environments, including sMP and NP, can be naturally ‘stabilised’ by the sorption of natural organic matter or through biofilm formation (Junaid and Wang, Reference Junaid and Wang2021; Pradel et al., Reference Pradel, Ferreres, Veclin, El Hadri, Gautier, Grassl and Gigault2021). While several of the studies reviewed herein have used synthetic surfactants to stabilise sMP and NP suspensions (Rodríguez-Hernández et al., Reference Rodríguez-Hernández, Muñoz-Tabares, Aguilar-Guzmán and Vazquez-Duhalt2019; Caldwell et al., Reference Caldwell, Lehner, Balog, Rhême, Gao, Septiadi, Weder, Petri-Fink and Rothen-Rutishauser2021), success in using biosurfactant has also been reported (Balakrishnan et al., Reference Balakrishnan, Déniel, Nicolai, Chassenieux and Lagarde2019). Interestingly, Ji et al. (Reference Ji, Wang, Wang, Fu, Man and Chen2020) found that bovine serum albumin provided better suspension and less agglomeration than SDS, suggesting that natural surfactants are perfectly suitable for this purpose. It is therefore possible to consider producing test and reference sMP and NP materials that are stabilised using different natural surfactants depending on the intended use of the particles in experimental studies. We therefore propose that future work on the development of sMP and NP reference materials should include a focus on using natural surfactants to ensure dispersion and dispersion stability.