The experimental design to separate MPs from solid sample matrices includes preparation of three different sample matrices and addition of artificial MPs. Separation steps of density separation and digestion of organic matter were performed. Four different solutions were applied according to their density and toxicity. Recovery rates as well as changes in size and surface structure of MPs were analyzed with stereo microscope and ATR-FTIR (Fig. 2). A chemical identification of potential MPs is associated with the cost of an analyzer, but assumed to be mandatory, because of the high error rates of 20–70 % through visual sorting (Bläsing and Amelung 2018). An ATR-FTIR analyzer is relatively cheap e.g. compared to an automated µFPA-FTIR microscope. Therefore, 500 µm was chosen as the lowest size limit as it represents the optical resolution maximum of this technique.
To meet the standards of analytical quality assurance, every combination of plastic mixture, extraction solution and substrate was tested in triplicates with a blank for every series of tests to investigate contamination. Beyond, sources of MPs contamination were excluded (see 2.6).
2.1 Preparation Of Mps
Different types of conventional and biodegradable plastic polymers were produced from mainly post-consumer products (Table 1). Conventional polymers were chosen according to their relevance and the probability of appearance as MPs in the environment. With PE/PP (\(\rho =0.862-1.083\hspace{0.25em}\frac{g}{cm3}\)) as light density polymers and PVC/PET (\(\rho =1.286-1.369\hspace{0.25em}\frac{g}{cm3}\)) as high-density polymers, it can be estimated if the method is able to recover a variety of plastics with different densities. Densities were measured due displacement of volume with a pycnometer (50 ml) and ethanol (99 %; TH. Geyer). Furthermore, three biodegradable polymers consisting of a range of newly produced polymers (PLA, PBS, MB) were applied in density separation validation tests from solid samples for the first time.
Reference MPs were prepared in four steps: first, products were cut. Second, pieces were frozen in liquid nitrogen and ground (Table 1). Third, size classes of 500–800 µm and 800–1000 µm were differentiated by sieving. Fourth, particles were visualized under a stereo microscope (Zeiss Stemi SV8 with Olympus SC30, f = 100, 80x, Oberkochen, Germany). Particles with a maximum length of 1.69 cm \(\pm\) 0.40 cm and a minimum width of 1.08 cm \(\pm\) 0.15 cm in the upper size class and a maximum length of 1.24 cm \(\pm\) 0.33 cm and a minimum width of 0.77 cm \(\pm\) 0.14 cm were gained with this method as the image analysis showed. All MPs were applied as fragments except MB, which was used as a film, because it is disposed as biowaste bags.
Two polymer composite samples were mixed, one with conventional and the other with biodegradable polymers. Concentrations of 400 particles/kg soil and 0.01 % w/w, respectively, were concluded to reflect realistic concentrations of MPs in solid sample matrices and were also applied in other recovery tests (Liu et al. 2019). These assumptions resulted in a MPs count of 25 particles per polymer, consisting of 10 particles in the size range of 800–1000 µm and 15 particles between 500–800 µm. As it is likely that more small particles occur in soil (Braun et al. 2018), more particles of the smaller size class were applied.
Besides the densities of used MPs, the color, chemical spectra as well as the sizes of particles before and after sample preparation were determined (see 2.7).
Table 1
Parameters of preparation of reference MPs
polymer | abbreviation | device | speed [rpm] | origin | color | density |
polyethylene | PE | centrifugal mill (Retsch ZM 200) | 8000 | pellet | bright yellow | 0.862 |
polypropylene | PP | knife mill (Retsch GM 200) | 6000 | pipe | blue | 1.083 |
| | centrifugal mill (Retsch ZM 200) | 6000 | | | |
polyethyleneterephtalate | PET | scissor | | bottle | green | 1.216 |
| | centrifugal mill (Retsch ZM 200) | 6000 | | | |
polyvinyl chloride | PVC | cutting | | pipe | red | 1.369 |
| | centrifugal mill (Retsch ZM 200) | 10000 | | | |
polylactid | PLA | scissor | | coffee-to-go lid | white | 1.300 |
| | centrifugal mill (Retsch ZM 200) | 10000 | | | |
polybutylensuccinate | PBS | centrifugal mill (Retsch ZM 200) | 6000 | pellet (Bionolle) | white | 1.234 |
MaterBi | MB | scissor | | biowaste bag | white- | |
| | centrifugal mill (Retsch ZM 200) | 10000 | (Edeka) | green | 1.241 |
2.2 Preparation Of Solid Sample Matrices
Three solid sample substrates (sand, OECD artificial soil and compost) were examined, representing different environmental compartments (beach/sediment, terrestrial soil and natural sample with high content of organic matter).
An amount of 250 g was applied per sample. This is together with Möller et al. 2021 the largest sample mass that was investigated compared to previously published studies that measure MPs concentrations in soil with density separation.
a) Sand substrates
To determine the influence of different particle sizes in the matrix, sands in different size classes ranging between 60–300 µm, 400–800 µm and 700–1200 µm (Scherf Quarzsand) as well as silicon dioxide in a size range of 53–74 µm (Eijelkamp synthetic sand) were purchased. These were combined to sand sample substrates according to sand fractions in OECD artificial soil (Table 2). In relation to OECD 222, the size fraction of 0.05–0.2 mm has to make up at least 50 % which was evenly divided between the two obtained fractions in this section. The composition of the other 50 % followed the formula of 100 µm \(\equiv\) 5 % of total substrate composition. Therefore, sand was dried at 200°C over night and sieved with a vibratory sieve shaker (Retsch AS 200) for 10 min, amplitude 1.2.
Afterwards, the sand sample substrates were placed in a muffle furnace at 550°C for at least 6 h to remove potential MP contamination. Finally, the composite plastic samples were added to the sand sample substrates. The mixture was stirred manually for 30 s.
Table 2
Composition of OECD artificial soil and sand substrate
| proportion [%] | mass for sand substrate [g] | mass for OECD artificial soil [g] | size fraction [µm] | |
| 25 | 62.5 | 43.75 | 53–74 | sand |
| 25 | 62.5 | 43.75 | 80–200 | sand |
| 5 | 12.5 | 8.75 | 200–300 | sand |
| 10 | 25 | 17.5 | 300–500 | sand |
| 10 | 25 | 17.5 | 500–710 | sand |
| 5 | 12.5 | 8.75 | 710–800 | sand |
| 10 | 25 | 17.5 | 800–1000 | sand |
| 10 | 25 | 17.5 | 1000–1200 | sand |
| | | 35 | | kaolinite |
| | | 15 | | halloysite nanoclay/kaolin clay |
| | | 25 | | sphagnum peat |
total mass [g] | | 250 | 250 | | |
b) OECD artificial soil
The OECD artificial soil is based on elaborations of an artificial soil in OECD 222: Earthworm Reproduction Test (Eisenia fetida/Eisenia andrei). This test was established for the ecotoxicological test of chemicals but shows the advantage of creating a reproducible soil matrix with proportions of soil that can interfere measurements of MPs like clay particles and soil organic matter (SOM) (Thomas et al. 2020).
The OECD guideline presets a content of 20 % kaolin which was split between kaolinite (CAS No. 1318-74-7) and halloysite nanoclay/kaolin clay (CAS No. 1332-58-7) (Sigma Aldrich). Kaolinite (Al2O7Si2 * 2H2O) and halloysite nanoclay/kaolin clay (Al2Si2O5(OH)4 * 2 H2O) are both 1:1-layered aluminosilicate minerals which belong to the kaolin group. The proportion of kaolinite (70 %) was selected to be higher than of halloysite nanoclay/kaolin clay (30 %) as it is the most widespread clay mineral in soils (Dixon 1989).
As there was no further definition of finely ground, the size of sphagnum peat (Floragard Floratorf) was oriented to the descriptions of OECD 232: Collembolan Reproduction Test in Soil which specifies finely ground to a particle size of 2 \(\pm\) 1 mm. Therefore, sphagnum peat was sieved to 1.00-3.15 mm due to available mesh size.
All proportions were dried to constant weight.
A mixture of sand size classes with same proportions as the sand substrate was prepared (Table 2) and placed in a muffle furnace at 550°C for at least 6 h. Afterwards, 35.00 g kaolinite and 15.00 g halloysite nanoclay/kaolin clay were added as well as 25.00 g of sphagnum peat. Additionally, 1.25 g of CaCO3 were added which corresponds to 0.5 % of soil mass for obtaining pH 6.0 \(\pm\) 0.5. The pH was determined according to OECD 222 by mixing 20 g of artificial soil with 100 mL KCl (1M) for 5 min thoroughly. The mixture was left to settle for 2–4 h and then measured with a pH-meter (pH 526, WTW), calibrated at pH 4.0 and 7.0.
The formation of soil aggregates was accelerated by wetting and drying soil. The applied amount of water was oriented to 60 % of water holding capacity as demanded in OECD 222 which was assessed with HYPROP 2 (Meter Group). Two samples of OECD artificial soil (250 g) were put in a sample ring and placed into water for complete saturation for three days. Afterwards, measurements were conducted over 13 days. The water holding capacity at pF 1.8 was calculated to 32.7 %. Therefore, 45 mL of distilled water and the composite plastic sample were stirred manually together with the substrate for two minutes. Afterwards, 5–10 mL of distilled water were used to rinse the spoon. The sample was let stand for four days to create agglomerates with components and dried at 60°C for 24 h.
c) Compost
Compost was derived from a biowaste plant in Lower Saxony, Germany (ALBA, Watenbüttel). The compost consists of biowaste from private households as well as garden waste (66 %) and was fermented at > 55°C for 14 days according to the composting plant operator. Characteristics show a content of organic matter triple the amount of OECD artificial soil. Furthermore, pollution with plastics is documented in the share of foreign matter with 0.08 % dry weight. The moisture content of 38.5 % was obtained by drying compost to a constant weight.
Compost was dried at 60°C and sieved. The composite plastic sample and 100 mL of distilled water were added to 250 g of compost dry weight (size class < 1 mm) and stirred manually for 1 min to create agglomerates containing MPs whereas the amount of added water corresponds to the moisture content before drying. Afterwards, samples were dried again (60°C, 24 h). As compost was not newly mixed from components in contrast to OECD artificial soil, no standing time was applied.
2.3 Preparation Of Density Solutions
Density separation was performed with distilled water (\(\rho =1.00\hspace{0.25em}\frac{g}{cm3}\)) and brine solutions of different densities. Saturated solutions were made by dissolving the respective salt into a volume of 1 l distilled water. In case of NaCl (> 99%, z. A., Ph. Eur.; CHEMSOLUTE®) and NaBr (99%, p.a., ACS; Roth) the amount of salt per water volume (358 g NaCl /L H2O and 905 g NaBr/L H2O) was depending on the specific solubility of each salt. The solutions were stirred in a beaker (2 l) on a magnetic stirring plate overnight resulting in solution densities of (\(\rho =1.19\hspace{0.25em}\frac{g}{cm3}\)) for NaCl and (\(\rho =1.53\hspace{0.25em}\frac{g}{cm3}\)) for NaBr. As there was no information available about solubility of SHMP (65–70% P2O5 basis; SIGMA-ALDRICH), the substance was added to distilled water sequently and stirred until no further salt dissolved and an excess formed at the beaker bottom (\(\rho =1.30\hspace{0.25em}\frac{g}{cm3}\)).
To lower the amount of disposed chemicals as well as costs, separation solutions of SHMP and NaBr were recycled and reused throughout the test series of density separation. Solutions were decanted onto a pleated filter (Macherey-Nagel MN 615 1/4, diameter 385 mm). The salt was retrieved in a rotary evaporator (Rotavapor RE 111; Büchi Labortechnik AG). Nevertheless, solutions that were used for density separation with compost were not recycled as the dark color indicated pollution with humic substances from compost.
2.4 Density Separation, Digestion And Recovery
The recovery rates of MPs in solid sample matrices by density separation with distilled water and different salt solutions were determined. Therefore, spiking experiments were performed, where 250 g of solid sample matrix was spiked with the composite plastic sample. First, 500 mL of the respective separation solution was added into a 2 L glass beaker and stirred with a rectangular stirring bar at 200 rpm. Afterwards, each spiked solid sample was transferred together with another 500 mL separation solution into the glass beaker, resulting in a total volume of 1 L separation solution per sample. Glass beakers were also used in other studies (Chen et al. 2020; Huang et al. 2020; Liu et al. 2018; Zhou et al. 2020) for density separation of MPs in soil as they are cheap, directly to purchase and easy to handle without the need to design complicated individual apparatus assemblies. Thus, they meet the requirement to be available worldwide.
Then, samples were stirred on a magnetic stirring plate at 400 rpm for 30 min as MPs can be strongly sorbed to the matrix (Liu et al. 2018). Even though polymers were artificially added to solid samples in these test series and it is hypothesized that strong sorption effects might not take place, it is likely that this time span is necessary when analyzing real soil samples. Afterwards, the stirring bar was removed and rinsed with the respective density solution as well as the beaker walls. Beakers were let stand to allow density separation for about 24 h, chosen according to (Hurley et al. 2018; Liu et al. 2018). The settling time allows the MP particles to float up to the solution surface and heavier sample matrix particles, like inorganic contents, to sink to the beaker bottom.
The upfloating MPs were taken with a spoon from the top layer of the solution surface. For sand substrates, MPs were then separated with tweezers, washed in distilled water and stored in petri dishes. For OECD artificial soil and compost samples, organic matter was additionally surfacing to MPs. Remaining residues were collected by suction of a peristaltic pump in a second removal step (Masterflex Console Drive easyload 7518-00; Cole-Parmer GmbH; Wertheim; Germany). A glass tube connected to a hose was moved around the solution surface to gain floating MPs. The supernatant was directly filtered on a filter (Whatman 589/1, diameter 125 mm) using a Buchner funnel and a vacuum pump. When filtration was completed, the filter was thoroughly washed with distilled water to remove salty residues and dried at room temperature. Finally, MPs were sorted from the filter visually. Added MPs were identified according to known color and shape. Films were considered according to lighter color and flexibility when touched with tweezers.
H2O2 (min. 30% w/w; CHEMSOLUTE®) was used as a reagent for oxidizing digestion of organic matter. It was successfully applied in many studies recovering MPs from solid sample matrices (e.g. Hurley et al. 2018; Han et al. 2019; Liu et al. 2018; Liu et al. 2019) and showed removal rates of 96–108 % of SOM (Hurley et al. 2018).
Handling and amount of the digestion solution had to be adopted depending on the substrate. Compost samples were placed in a beaker (2 L) and H2O2 (30 %) was added depending on the amount of floating organic matter (Ratio 1:4) and stirred at 50–60 rpm. The samples were heated for 24 h at 60°C and let stand for 3 days at room temperature. OECD artificial soil samples were also heated at 60°C for 24 h, but 100 mL H2O2 were not sufficient to cover the sample, so the digestion solution volume was increased to 200 mL H2O2 (30 %) (Ratio 1:8). Residues were filtered using a vacuum pump onto filters (Whatman 589/1, diameter 125 mm).
Afterwards, MPs were sorted out with tweezers visually with the naked eye and under a stereo microscope. Then, a second density separation was applied. Residues were mixed with 500 mL of the respective density solution, stirred manually and let stand for 1–2 minutes. MPs were then sorted from the surface and washed in distilled water, dried at room temperature and stored in petri dishes for analysis.
Preliminary experiments showed that density separation eliminated most of the sample matrix compared to digestion. There was one exception in compost samples with NaBr as density solution. Thus, density separation was applied before digestion in all experiments except compost with NaBr.
2.5 Tests On Influence Of Sample Preparation
To test the influence of sample preparation on MPs, several potential impact factors were examined individually and also in combination. The individually tested impact factors were dry storage at 60°C, applied separation solutions for density separation, H2O2 (30 %) at room temperature and H2O2 (30 %) at 60°C. Three particles per polymer type in a size > 1 mm were put into a petri dish and tested for eight days for each impact factor except with H2O2 at 60°C which was tested for 3 d. The duration was based on the time periods of the separation process and corresponds to at least twice the time for each condition in the separation process described above. The combination was a sequence of dry storage at 60°C, the separation solution NaBr that extracts the most MP contents and H2O2 at room temperature and at 60°C with same durations as for individual factors, respectively. Afterwards, particles were washed in distilled water, dried and analyzed.
2.6 Contamination Mitigation
To prevent contamination of the samples with MPs during the ongoing experiments, cotton lab coats and butyronitrile gloves were worn in the laboratory and lab surfaces cleaned with distilled water (Torres and De-la-Torre 2021). Equipment was cleaned in a dishwasher and dried in a drying oven. Furthermore, only filtered distilled water (20\(\mu\)m) was applied in test series and for cleaning. All tests of density separation and removal of organic matter were performed under a flow bench and a fume hood. The time a sample was exposed to air was limited as much as possible to prevent atmospheric contamination. Additionally, beakers were covered with aluminum foil during stirring and settling. Furthermore, the air in the lab was filtered (Philips air purifier 2000 AC 2887; Koninklijke Philips N.V.; Amsterdam; Netherlands) as well as the supplied air from outside. Isolated MPs were stored in glass petri dishes until analysis. Besides, a blank sample was performed for each test series.
2.7 Microscopy And Ftir Polymer Identification
MP composite samples and MPs for influence tests were manually sorted onto glass petri dishes and photographed under a stereo microscope (Zeiss Stemi SV8 with Olympus SC30, f = 100, 80x, Oberkochen, Germany) with analysisPRO before and after application of the separation protocol. Special attention was paid to lay all particles the same way to ensure comparability for tests of influence of sample preparation. For other samples of recovery tests, this could not be ensured due to extent of particles.
Then, the photographs were evaluated by automatic computer image analysis to determine the damaging effects on the size of the MPs due to the impact factors. Image analysis was done using the open source computer vision library OpenCV (opencv.org): Contours were localized with the openCV function findContours, and the particle area was calculated from the contour. Maximum particle length was identified by automatic stepwise contour rotation and the openCV function boundingRect. All pictures had identical scale (84 pixels/mm), and contours with an area smaller than 500 pixels (0,07 mm2) were regarded as noise and excluded from the processing. The particles were fragmented and not round which is more likely to reflect their occurrence in the environment. Some particles were also frayed. The fabricated MPs are multidimensional so a two-dimensional picture can only represent a certain aspect of the particles. Thus, two parameters of area and maximum length were considered to reduce effects of observational error.
Besides, MPs were characterized chemically with ATR-FTIR (Bruker Tensor 27 with MVP-Pro, resolution 4 cm− 1, 16 Scans, area 4000 − 380 cm− 1). Several recovered MPs from the MP composite samples were measured with ATR-FTIR to ensure sufficiency of visual classification. In addition, each polymer for the influence tests was determined by measuring three particles on two sides. The following six spectra were accumulated to reference spectra.
2.8 Statistical Analysis
For recovery tests, means, standard deviation and recovery rates of MPs were calculated for each sample and compared for every combination of substrate and solution. Besides, numbers of recovered MPs for the first and second removal from the solution surface were compared. For tests of influence, an analysis of changes in surface structure was performed with ATR-FTIR. Likewise to reference spectra, six spectra per polymer and impact factor were accumulated and afterwards compared with reference spectra. The similarity of spectra was tested with Pearson’s correlation coefficient likewise to Vermeiren et al. 2020.