1. Introduction
A great emphasis in microplastic (MP) research is put on the spatial distribution, transport and behavior of microplastics in water. The interactions of MP in the aqueous phase are numerous: leaching of potentially toxic additives, adsorption of organic compounds and metals to MP and toxicological effects on organisms are being investigated. With a few exceptions, all these studies use artificial MP dispersions to extrapolate laboratory findings to environmental behavior. For this reason, meaningful results require attention during the preparation of MP dispersions.
MP exposure experiments conducted with only a few details on applied mass, particle size distribution and concentration are hard to compare [
1]. Prerequisites have been formulated for particle characterization used in toxicology studies regarding surface area or size distribution but also homogeneity of dispersion [
2,
3]. Only fully dispersed, stable dispersions with most particles existing separately allow for correct extrapolations of mass/number concentration effects. Dispersion stability is linked to particle size and specific surface area. However, pristine polymer particles as used in many studies are highly hydrophobic and poorly wettable if not chemically stabilized or further processed [
4]. Depending on the material type and applied concentration of these particles, attachment to laboratory glassware, accumulation at the liquid–gas interface, and formation of homo-agglomerates are the consequences [
5]. This reduces the availability of particles to participate in experiments and can strongly bias outcomes. The OECD Guideline concerning stability of nanomaterials highlights the dispersion stability as an important parameter affecting the environmental behavior of nanomaterials (with nanoplastics already being studied by some research groups) [
6]. In many cases, surfactants such as sodium dodecyl sulfate or polysorbate (Tween) 20 or 80 are used and commercial particle dispersions are thereby commonly stabilized [
7,
8]. Surfactants are also added during the synthesis of MP which could alter the surface characteristics and thus the adsorption process [
8,
9]. Nevertheless, surfactants can also bias experiments: Renzi et al. [
10] recorded a higher mortality and immobilization in toxicological experiments with MP when the surfactant Triton X-100 was used. The surfactant itself had negative effects on the organisms but the more stable dispersion of particles could also lead to increased ingestion and toxicity. Xia et al. [
11] reported the effect of non-ionic surfactants on the sorption of organic pollutants onto microplastics. Natural organic matter (NOM) is also used in some studies to stabilize MP and mimic environmental conditions, but the effect of NOM on dispersion stability has not been fully resolved yet. Reported results indicate a stabilizing as well as no effect using NOM standard samples from the International Humic Substances Society (IHSS) in MP dispersions [
12,
13,
14,
15,
16].
Oxidation of MP surfaces as an effect of weathering processes increases polarity, hydrophilicity and induces charge [
17,
18]. Artificial weathering can be achieved by UV photodegradation or chemical oxidation using hydrogen peroxide or ozone. Van der Esch used an alkaline oxidation with potassium hydroxide to simultaneously fragment and oxidize polystyrene (PS), polyethylene terephthalate (PET) and polyactic acid (PLA) polymers to obtain a stable dispersion [
3]. The advantage of surface activation lies in the mode of stabilization which can closely resemble environmental weathering processes and thus create more realistic scenarios. Artificial weathering, however, is a time-consuming process, especially when using UV radiation. Ozone has been applied to stabilize PS particles, but the oxidation caused a size reduction of particles and a fraction of PS was dissolved [
5].
The agglomeration behavior of dispersions has already been broadly studied for nanoparticles [
19,
20], with a focus on ecological investigations [
21]. Pre-treatment for deagglomeration is an important step to ensure size/surface-dependent particle effects for specific applications. In the case of nanoparticle research, pre-treatments for dispersing solids include hot air oxidation, application of ozone, deep-freezing below 195 °C and the use of chemical stabilizers [
22,
23,
24].
Ultrasonic treatment is also commonly used in this field, as it is reported to break down agglomerates in the nanoscale [
25]. Acoustic cavitation as the main process leads to deagglomeration and formation of OH-radicals with temperatures of up to 5000 K and pressures of 1000 atm after bubble collapse. Cavitation and cavitation-induced shockwaves accelerate particles in dispersion which will eventually collide. These interparticle collisions can change the surface morphology, composition and reactivity [
26]. Sonochemistry, therefore, offers interesting possibilities to stabilize MP dispersions through a combination of physical and chemical interactions.
A stable and fully dispersed dispersion is characterized by stronger repulsive forces which kinetically hinder the non-equilibrium state to reach equilibrium in an agglomerated state [
6]. So, either the net forces in the dispersion have to be measured or the state of particles (isolated or agglomerated) has to be monitored. In many studies, scanning electron microscopy (SEM) images are provided to characterize particles used for dispersions [
8]. Static imaging, however, may not be suitable since particles are removed from the test matrix (usually water), dried and further prepared (e.g., sputtered with gold) to meet prerequisites of the imaging process. SEM images still give valuable insights concerning particle size and morphology [
27].
Zeta potential measurements are widely used to study the effects of ion strength, pH and humic substances on the stability of MP and nanoplastic (NP) dispersions [
12,
13,
15,
16]. This parameter is essential to characterize dispersions, but measurements are elaborate and time-consuming and limited to smaller particle sizes depending on the density of the particle. Particle concentration measurements, on the other hand, are fast and can give an insight into the size distribution of a polydisperse system. Through the simple measurement of light extinction, the particle concentration can be determined directly in the liquid as an index of dispersibility. The number of particles per unit volume combined with size measurements are good indicators for the state of a dispersion [
21].
In this work, we evaluated additive-free methods of dispersion for cryomilled PS particles. For this purpose, manual shaking by hand, long-term horizontal shaking and ultrasonic treatment were compared regarding the suitability of dispersing particles, and the effects on particle numbers and size fractions were investigated. Light extinction measurements were used as a quality indicator for the degree of dispersion. By volume calculation of dispersed particles, the dispersion efficiency ED was determined to evaluate the methods of dispersing.
2. Materials and Methods
PS particles were produced by cryogenic ball milling (Cryomill, Retsch, Germany) of pristine PS pellets (BASF, Ludwigshafen, Germany). The procedure was previously documented in detail [
5]. The PS powder was then dry-sieved over a 100 µm stainless steel mesh (Retsch, Haan, Germany) to exclude larger particles. Ultrapure water (ELGA, Celle, Germany) was used for all batches and cleaning steps. In this process, 500 and 1000 mL glass bottles (Schott, Mainz, Germany) were used for the preparation of the batches. Thorough cleaning of all glassware was achieved by multiple washing, including the use of detergent and ultrasonic baths.
PS particle batches were prepared by weighing-in with a lab scale (A200S, Sartorius, Germany) using an anti-static gun and carefully rinsing the particles into the bottles. Ultrapure water was added by weight to obtain a particle concentration of 20 mg/L PS particles. Three different treatment approaches were used to disperse the hydrophobic PS particles: manual shaking by hand, automated shaking on a horizontal shaker and ultrasonic treatment.
Manual shaking was continuously done by hand with a consistent movement so that the water–gas interface was disrupted. Each up and down movement of the bottle was counted as one agitation. Particle measurements were made every 50 agitations up to a total of 250 agitations.
A horizontal shaker was used at 250 rounds per minute for automated shaking. Initially, the batches were measured every 60–80 min, and this interval was increased throughout the experiment for, in total, 754 h to fully evaluate the impact of long-term agitation.
Ultrasonic treatment was done in an ultrasonic bath (Sonorex Super RK 106, Bandelin, Germany) with a rated output of 120 W, and a frequency of 35 kHz was applied in 5–15 min intervals. Bottles were submerged in a deionized/tap water mixture to approximately 2/3 of the height. The uniformity of the ultrasonic field was not tested.
Particle size and number was determined by light extinction measurement using a particle counter (SVSS, PAMAS, Rutesheim, Germany). For each measurement, 3 mL of the sample was analyzed and every sample was measured in triplicate (mean value is shown, error bars are provided). The particle counter groups detected particles into channels, the boundary settings for these channels are provided in the
SI, Section D. Batches were slightly stirred before measurement to resuspend deposited particles and rinse-off particles attached to the upper part and neck of the bottles. Previous experiments showed that magnetic stir bars can increase particle concentration through abrasion of the bars or the bottle bottom, and were therefore not used. For the settling experiment, the batch was not shaken and only moved carefully for measurement purpose. All concentration values shown are averaged from 3 measurements, and no particle settling could be detected between triplicate measurements. Average particle diameter was about 3 µm and differences in the densities of PS and water is low, therefore significant settling in the time frame of a measurement—(approximately 3 min for triplicate analyses)was not expected. As illustrated in
Figure 1, the timespan of measurement is too small for a considerable settling process of particles expected in the size range of 1–20 µm.
The suspended solid volume of PS particles was calculated using particle size data with the lower channel boundary and using spherical shape (see
SI, Section A). In a previous study, the irregular morphology of cryogenic milled PS powder was shown [
5]. The lower channel boundaries were therefore used to partly compensate for the overestimation of volume by assuming a spherical shape of particles. In order to compare the preparation methods, the parameter dispersion efficiency E
D was used (for formulae see
SI, Section C). E
D is the ratio between calculated suspended solid volume V
css and theoretical suspended solid volume V
tss and indicates the state of the dispersion (fully dispersed vs. agglomerated/attached to surfaces etc.). V
tss was calculated using the suspended solid mass and a PS density of 1.05 g/cm
3. V
css was calculated using the particle counter data and the lower channel boundary (see
SI, Section D) and volume formula for spherical shape.
4. Discussion
The aim of this study was to evaluate an additive-free method to disperse model MP particles in water to realistically depict concentration-effect relations of MP particles. Primary MP particles released into the environment most likely start out as a hardly dispersible material, as with the model particles used in this study, and the dispersion of these is still unknown. Therefore, the preparation of these virgin MP particle dispersions for laboratory studies is important to understand the behavior and fate of virgin MP particles in the environment. Using ultrasonic pre-treatment for cryomilled pristine polystyrene particles, reproducible dispersions of 20 mg suspended solid per liter were obtained. The physical particle interactions induced by the ultrasonic bath after 90 min effectively dispersed particles, while no substantial increase in dispersed particle numbers was observed with longer treatment. With further sonication, the number of particles >50 µm fluctuated while particle numbers <50 µm remained constant, showing that ultrasonic treatment did not further disintegrate PS particles. Ultrasonic treatment as a dispersion technique thus increased the numbers of particles in the dispersion and ensured the availability of particles for possible interaction. After 24 h, around 70% of the particles of a dispersion (180 min of ultrasonic treatment) were still in dispersion.
In contrast to ultrasonic treatment, long-term horizontal shaking and manual shaking did not suffice for fully dispersing all particles. Dispersion efficiency was 127%, 104%, and 69% for ultrasonic treatment (180 min), long-term horizontal shaking (754 h) and manual shaking (250 agitations), respectively. This is in accordance with a similar study implementing nanosized silica and alumina particles in which ultrasonic homogenization as well as stirred media milling proved most effective for deagglomeration [
32]. E
D values of above 100% indicated that the presumption of spherical shape for cryogenically produced particles is a simplification, as cryomilled particles are more shard-like fragments [
5]. This overestimated the calculated volume of measured particles when using the Feret diameter given by the particle counter, which was only partly counteracted by using the lower boundaries of the measurement channels.
Using alkaline ultrasonic treatment van der Esch et al. [
3] reported stable dispersions with aged MP but with a high variability (1,5 log-orders) in particle concentrations. For PS, the toxicologically interesting fraction <20 µm ranged from 40 to 80% within three replicates. This emphasizes the need for appropriate production of particles and pre-treatment for dispersions. Chung et al. [
30] reported a maximum dispersing effect after 20–30 min for zinc oxide nanoparticles using ultrasonic baths. Although the materials and therefore binding forces are not directly comparable, a maximum was also found in this study for ultrasonic dispersion after 90 min. Treatment times can possibly be shortened by using a higher input power; however, the potentially achievable deagglomeration is a function of output power of the ultrasonic device and two different ultrasonic devices need to be cautiously compared.
Van der Esch [
3] reported eroded surfaces on 1 cm
2 polystyrene squares cut from larger sheets and additional FTIR-bands mimicking weathering processes in the environment. We did not investigate changes in the morphology through radical attack during sonification. Such changes were not expected since the duration of ultrasonic treatment was much shorter (180 min compared to 15 h) and pH (unbuffered) was around 5.5 (van der Esch: pH = 13). Therefore, surface oxidation during sonication in this study is unlikely. Nevertheless, it needs to be discussed whether polymers pre-treated in an ultrasonic bath even for short periods can still be regarded as pristine if the surface of particles is potentially altered.
Another challenge in dispersion preparation is the weighing-in: a precise execution is difficult due to fine powder in the µm area, low density materials and potential electrostatic forces. Air drafts and charged surfaces of laboratory equipment can easily carry away the finer fraction of polydisperse powders. The particle concentration of six batches of identical weight-in particles still showed a rel. SD of 3.5% after 180 min of ultrasonic treatment. We therefore suggest preparing a more concentrated dispersion which has to be thoroughly treated with ultrasound before further dilution to targeted concentrations.
Concerning the stability of dispersions, no agglomeration behavior could be observed in a dispersion after 168 h. This was expected since larger particles settled quickly while smaller particles apparently remained as single particles in dispersion. Settling could have been disturbed by the high number of particles in dispersion, this would explain particles of the size 3–10 µm still present in the water column after 168 h. The settling of irregularly shaped particles does not necessarily follow Stoke’s law applied on spherically shaped objects, as is the case for natural colloid aggregates [
33]. Ultrasound treatment is therefore suitable for pre-treatment of PS particle dispersions for eco-toxicological studies since the stabilized particles are available for possible ingestion by aquatic species. Nevertheless, the results point out the reluctance of microplastic particles to behave as predicted by models—in laboratory studies as well as in environmental sampling.
Delmas and Barthe [
34] claimed that addition of surfactant for stabilization is still required after the use of ultrasound to stop agglomeration; however, this type of interparticle bonding is weaker after ultrasonic treatment rendering the use of surfactant more effective. Whether the use of surfactant is still required needs to be decided depending on the experiment. The time-dependent stability of dispersions (the tendency to agglomerate) has to be evaluated individually for each experimental setup since particle size and surface chemistry are key factors in agglomeration processes. This is an important issue for toxicological studies as well as the transport behavior of particles during experiments. Regarding a potential stability of over 168 h through ultrasonic treatment, we do not suggest the use of surfactants for microplastic experiments due to possible side-effects, as stated above by some other authors.
In the environment, the dispersion of virgin MP particles through energy input could be achieved by strong currents and turbulence in rivers and other surface waters. Other processes influencing the stabilization of virgin MP particles in natural environment include biofilm formation, adsorption of organic substances and weathering. Once stabilized, these particles are mobile and do not necessarily settle according to model calculations. As a consequence, MP particles are found throughout the water column and not only in the sediment, aggravating possible remediation techniques. The removal of virgin MP in wastewater treatment plants thus seems to depend on inclusion in flocs or foams. Further research needs to focus on the concurrence of environmental processes contributing to the stabilization of virgin MP particles and the resulting behavior in water to evaluate target-oriented measures to remove MP particles from the environment.
This study provides valuable insights for research groups working with model suspensions. Basic understanding of the behavior of model particles in aqueous suspensions is one of the pre-requisites for systematic research and the transfer of laboratory investigations to environmental research outcomes.