Analytical
Methods
View Article Online

Published on 29 September 2021. Downloaded by Auburn University on 8/5/2025 6:19:58 AM.

PAPER

View Journal | View Issue

Cite this: Anal. Methods, 2021, 13, 5299

Microplastic extraction from sediments
established? – A critical evaluation from a trace
recovery experiment with a custom-made density
separator†
Maurits Halbach,
Böttcher *

Christin Baensch, Sonka Dirksen and Barbara M. Scholz-

By now, microplastics are present in every environmental compartment of which sediments are considered
one major sink. As a result, several approaches for their enrichment from sediments have been established
in microplastic analysis. At the same time, the smaller microplastics gained increasing attention regarding
their ecotoxicological relevance. A customized sediment separator was evaluated with trace amounts of
small microplastic particles (150–300 mm) of the nine most common polymers. Separation was
performed with sodium bromide (r ¼ 1.5 g cm3). The experimental recovery comprises pristine as well
as incubated polymers to include early biofouling effects. Polymer quantification was achieved
exclusively using pyrolysis-gas chromatography-mass spectrometry. The results reflected an overall
mean recovery of 65%. Interestingly, the observed behaviour seems to be density related. While
polymers of higher densities revealed higher average extraction efficiencies (74–97%), those of less
dense polymers are reduced and span between 34 and 65%. These observations hypothesize possible
polarity related surface interactions as a relevant factor for microplastic particle extraction. In contrast,
the density of the separation fluid seemed to be of subordinate relevance, if small microplastic particles
were extracted in trace amounts. Early biofouling enhanced recoveries of some polar polymers, whereas
the effect on apolar polymers was even negative in some cases. In a comparative synopsis with other
published density separation approaches, a limited number of comparable experimental setups
concerning particle size, polymer density range and polymer concentration were revealed. Nonetheless,
some related experiments point to similar density/polarity driven extraction behaviour. In conclusion, the
Received 8th June 2021
Accepted 27th September 2021

presented study suggests a re-evaluation of current separation approaches for extraction of low
number/mass concentrations of small microplastics from sediments to enable a more comprehensive

DOI: 10.1039/d1ay00983d

insight into factors that influence surface properties for microplastics extraction. Concurrently, it raises
the question of how an ideal environment relevant recovery experiment can be designed.

rsc.li/methods

1. Introduction
Plastics have become indispensable in modern society. Their
low price, durability, and versatility lead to an ever-growing
demand of plastic products since the 1950s.1 Since then, plastics enter the environment through various pathways and parts
of them end up in the ocean.2–4 In particular the smaller 5 mm
so called microplastic (MP) fraction is proven to be present in
every environmental compartment by now.5
Sediments are considered a major sink for MPs in the environment.6,7 Nonetheless, a dened picture of MP transport
Institute for Chemistry and Biology of the Marine Environment (ICBM), Carl von
Ossietzky University of Oldenburg, P. O. Box 2503, D-26111 Oldenburg, Germany.
E-mail: bsb@icbm.de
† Electronic supplementary
10.1039/d1ay00983d

information

(ESI)

available.

This journal is © The Royal Society of Chemistry 2021

See

DOI:

processes from and into sediments and their resulting distribution is still lacking. A high pollution of sediments implies
a potential threat in particular to fertile riverine and coastal
habitats.
MPs smaller than 1 mm are thereby of special concern since
their particle counts increase with declining particle size8 that is
related to an elevated toxic potential.9,10 Thus, small MP particles are an essential target for microplastic analysis in sediments.
State-of-the-art
established
polymer-specic
quantication techniques (e.g., FTIR-, Raman-spectroscopy,
and pyrolysis-gas chromatography-mass spectrometry (Py-GC/
MS)) allow a comprehensive analysis even for very small MP
particles. However, regarding environmental samples
a successful application of these techniques is inevitable
adherent to a preceding sample purication if complex sediment samples are in the analytical focus.11

Anal. Methods, 2021, 13, 5299–5308 | 5299

View Article Online

Published on 29 September 2021. Downloaded by Auburn University on 8/5/2025 6:19:58 AM.

Analytical Methods
Sediments are one of the most challenging compartments
for MP analysis.12–14 The challenge to maintain polymer integrity
while reducing complex organic and inorganic matrices resulted in diverse purication approaches.12–14 Inorganic components are commonly removed by elutriation or density
separation.12–14 These techniques take advantage of the density
differences between microplastics and inorganic components
like quartz and other minerals.
In particular, the focused small MP particles (1 mm) show
less clear analytical behavior in density-dependent extraction
procedures. Here, the overall specic polymer density becomes
less important.15–17 Density separation of small particles needs
longer equilibration times to reach oatage18 and is more
affected by adhesion and buoyancy changes. Observed buoyancy changes are related to aggregation19,20 and biofouling.21–25
Only few evaluation studies on density separation techniques accounted for small particle sizes,26,27 while none
addresses aggregation or biofouling in their recovery experiment. Furthermore, recoveries were oen performed with visual
controls26–33 but were rarely associated with reliable polymerspecic quantication techniques (e.g., FTIR, Raman, and PyGC/MS).34
In this study we put forward a density separator that has
been developed within the framework of a former project
simultaneously with several other separators following the
increasing demand of MP data from environmental sediments.6,35 Ever since these times the small MP particles have
become more and more relevant in the ecotoxicological context
and gained increased attention regarding their analytical
behaviour such as potential inuence of aggregation and
biofouling.
The aim of this study is to re-evaluate our separator by performing recovery tests for a broad range of relevant MP types,
with low, environmentally relevant concentrations and small
particle sizes. We thereby want to complement our two previous
successful tests on the reproducibility of environmental sample
analysis36 and on the extraction of 1 mm particles from different
sediment types (Kögel et al., in prep.).
Our recovery experiments compare the extraction of pristine
and incubated polymers from an articial sediment. The incubation experiment should thereby reect the effect of biofouling
on the recovery efficiency. All recoveries are quantied exclusively by Py-GC/MS and accordingly reect real analytical
conditions.
The results are critically discussed in contrast to the current
literature, and with a comparative look on currently used
separation procedures. Here, we intend to review all separation
procedures and related recovery experiments with special focus
on the shied and future needs in small MP analysis of
sediments.

2.
2.1

Materials and methods
Technical setup

The separator assembly consisted of three units. (1) A modied
3 l glass beaker (Fig. 1) tted with a discharge funnel and
a skimmer. (2) A displacement cone (Fig. S1†) that is placed in

5300 | Anal. Methods, 2021, 13, 5299–5308

Paper

Separation beaker with a stirring device and displacement
cylinder.

Fig. 1

the centre of the beaker and connected to an electric stirring
device (Hei-TORQUE Value 200, Heidolph Instruments). (3) A
second 800 ml glass beaker to collect the skimmed solution.
The 3 l beaker was modied by a professional glassblower
(specications in Fig. S1†). All parts are easy to handle, easy to
cover with alumina foil during the separation process and easy
to clean aerwards. During the density separation procedure
buoyant particles rise to the solution surface where they are
displaced by the cylinder to the outer ring of the solution
surface. A constant rotation of the cylinder pushes the particles
together into the skimmer which channels them down into the
discharge funnel. Finally, the particles are retrieved in the
collection beaker.
2.2

Experimental setup

Sample preparation. To test the efficiency of the density
separation nine relevant microplastic standards (Tables 1 and
S4†) were added to ne quartz sand (grain size distribution cf.
Table S1†). These widely demanded and utilized polymers represented a wide density range (Table 1). Spiking sets were
prepared by manually preselecting particles of the respective
pristine polymers with similar sizes (150–300 mm). They were
weighed into aluminium weigh boats with a Cubis Ultramicro
balance MSE2.7S-000-DM (Sartorius, Germany). The respective
polymer weights (3–45 mg, equivalent to particle numbers
between 5 and 20 cf. Table 1) were adjusted to enable an optimal
quantication within the individual Py-GC/MS calibration
curves and outside any detection limits (Table S2†). The particle
size range was chosen to ensure the particle transfer into the
sediment by optical control via a binocular.
For every replicate, one separation beaker was lled with
800 g of quartz sand (pre-treated in a muffle furnace at 550  C
for 12 h) and spiked with the prepared sets. The polymer
particles were rinsed from the aluminium boats into the separation beaker with 200 ml ltered water, representing a 1 kg wet
weight sediment.
To induce biolm production on the particles in the incubation experiment the spiked sediment in the beaker was
additionally spiked with a 50 ml homogenized mixture of articial seawater and marine benthos microalgae (Thalassiosira
eccentrica, Amphora coffeaeformis and Diploneis sp.) representative of a typical marine benthic community (North Sea, Germany). The water level was then further adjusted to create

This journal is © The Royal Society of Chemistry 2021

View Article Online

Paper
Table 1

Analytical Methods
Overview on properties of polymers and their characteristics in the experimental setup

Published on 29 September 2021. Downloaded by Auburn University on 8/5/2025 6:19:58 AM.

Polymer properties

Experimental setup

Abbreviation

Polymer

Density (g cm3)

Particles (n)

Weighed portion (mg)

Size (mm)

PP
PE
PS
PA6
PMMA
PC
MDI-PUR
PET
PVC

Polypropylene
Polyethylene
Polystyrene
Polyamide-6
Polymethyl-methacrylate
Polycarbonate
MDI-polyurethane
Polyethylene-terephthalate
Polyvinyl chloride

0.9
0.96
1.04
1.13
1.19
1.2
1.23
1.38
1.43

7–9
7–15
5–10
6–10
5–7
5–7
7–10
5–9
15–20

43.1
42.9
20.3
45.2
20.6
5.0
44.8
15.0
42.8

300
250
210
300
250
150
280
190
200

a water layer on top of the sediment to allow microalgae growth.
The microalgae have been pre-cultivated for ten days and
autoclaved with nutrient and vitamin enriched articial
seawater. The replicates were further incubated for another
three days at room temperature to allow biofouling.
Density separation procedure. To the spiked sediment
replicates 1375 ml of a sodium bromide (NaBr, Grüssing GmbH)
solution (>1.5 g cm3) was added. With respect to the given
water content of the sample 340 g of washed NaBr was added to
adjust the nal density to >1.5 g cm3. The suspension was
manually stirred for a few seconds with a stainless-steel spoon.
This was repeated 8 times over a period of two hours. Subsequently, the solution was le to settle overnight. Now, the stirring unit was positioned into the beaker in a way that its
displacement cylinder was at the level of the skimmer and
overow. The level of the NaBr solution was elevated right up to
the overow. The collection beaker (600 ml) was placed beneath
the overow and the stirring device was adjusted to a speed up
to 30 rotations per minute. Over a two-hour time span the sides
of the beaker, the displacement cylinder and the skimmer were
rinsed successively with saturated NaBr solution from a PFTEbottle every 15 minutes (Fig. S2†). Subsequently the skimmed
solution was transferred thoroughly from the collection beaker
into a separation funnel (500 ml). This was shaken vigorously to
disintegrate potential aggregates and adhered particles then
rinsed from the glass walls with NaBr solution from a PFTEbottle. The solution was le to settle overnight. Then, the
heavy fraction at the bottom (approximately 2/3 of the solution)
was discarded. Meanwhile, the sediment extraction procedure
was repeated, and the resulting solutions were combined in the
respective funnel. Density separation in the funnel was
continued until no particles were le to settle. The residual
solution was vigorously rinsed onto a glass bre lter (1 mm
pore size, Ø 15 mm) with ltered water and 10% ethanol and
stored in glass Petri dishes (5 cm).
The modular setup between the separation beaker and stirring unit allowed the parallel extraction of four samples.
Contamination measures. To keep any secondary contamination as small as possible glassware and laboratory equipment
were rinsed with ltered water and 10% EtOH from Teon
bottles. All solutions were ltered through glass bre lters (1

This journal is © The Royal Society of Chemistry 2021

mm pore size). Glass bre lters were muffled at 400  C (4 h)
before use. All experiments were performed in a frequently
cleaned, but not activated fume hood. Plastic lab ware was
avoided to the most possible extent. All washing bottles used
were made of PTFE, not included in the polymer types analysed.
In particular plastic lids were substituted via alumina foil. All
processing steps were performed under alumina foil protection
and uncovered for short intervals only if necessary. Only cotton
clothes were allowed to be worn in the laboratory along with
a cotton lab coat.
To exclude secondary contamination from the commercial
NaBr it was prewashed and recrystallized. Solid NaBr was lled
into the separation beaker of the separator, covered with preltered, saturated NaBr solution and stirred with a constant
addition of solution generating an overow of the upper solution layer, which removes potential MP particles. Subsequently
the saturated NaBr solution was decanted and the wet but
crystalline NaBr was transferred into cup-sized aluminium
bowls that allows their adjacent portioning.
Although the sand matrix used for the experiment was
treated at 550  C to remove any organics, the recovery experiments were complemented by a procedural blank. This procedural blank was processed in parallel and analogous to the
incubation recovery experiments to assess any possible
secondary contaminations throughout the whole density separation procedure. It is supposed to indicate any potential
contamination issue, and carried out here, it takes the extended
handling process adhered to the incubation step into account
as well. Even though the detected polymer contents are very low
or even absent (cf. Table S9†), they were subtracted from the
respective incubation recovery raw data on an area basis. The
subtraction from the pristine experiment was discarded due to
high frequent intensity variances between the two Py-GC/MS
measuring sequences. Procedural blank data from the same
sequence for a meaningful subtraction would be needed.
Nonetheless, the quantication of the procedural blank with
respect to the polymers (cf. Table S9†) underlines its minor and
almost negligible impact on the overall results. The quantied
blank functions as a contamination estimate for the pristine
recovery (Table S9†).

Anal. Methods, 2021, 13, 5299–5308 | 5301

View Article Online

Published on 29 September 2021. Downloaded by Auburn University on 8/5/2025 6:19:58 AM.

Analytical Methods
Quantication – Py GC/MS. The extracted samples were
prepared for Py GC/MS measurement according to Fischer
et al.36 The glass bre lters were folded into pyrolysis cups and
spiked with 4 different pyrolysis injection standards (Table S5†)
and tetramethylammonium hydroxide (TMAH, 25% in methanol (MeOH), Fluka, Germany) for on-line derivatisation.
The prepared sample cups were analyzed using a microfurnace pyrolyzer (EGA/Py-3030D, Frontier Lab) connected to
a gas chromatograph (Agilent 7890B) and a mass spectrometer
(Agilent MSD 5977A). All measurement conditions rely on
Fischer et al.36,37 and given in detail in the ESI (Table S6†).
Polymers were identied via polymer specic indicator ions
(Table S8†) and then quantied by external calibrations from
peak areas and peak area ratios. Basic calibration curve
parameters are given in the ESI (Table S7†).
Potential contaminations identied in the procedural blanks
(Table S9†) were compensated by peak area subtraction from
the sample peak area before any quantication. Potential
outliers were identied by the Grubbs test implemented in
Origin (OriginLab Corporation, Version 2020).

3.
3.1

Results & discussion
Recovery experiments

Polymer-specic extraction. All 9 polymers were successfully
extracted by the sediment separator regarding all recovery
experiments (4 pristine and 4 incubation experiments; see the
following subsection). The overall quantied MP mass recovery
regarding the sum of spiked polymers was 68  35%. The
observed polymer specic recovery efficiency varied highly
between the different types. The highest average recoveries were
achieved with PVC (97%) whilst the lowest average recovery was
achieved with PE (34%) (Fig. 2).
Interestingly, arranging the polymers in accordance with
their respective density, a so far inconspicuous trend becomes
obvious. The differences amongst polymer recoveries seemed to
be density driven and polymer recovery consistently increased
with polymer density (Fig. 2). Opposite to any expected scenario

Combined polymer-specific recoveries from the pristine and
incubation experiment, including 8 replicates. cum. ¼ all recoveries
from every experiment and polymer. Outliers are marked as diamonds;
* one outlier out of bounds at 280%.

Fig. 2

5302 | Anal. Methods, 2021, 13, 5299–5308

Paper
that would favour a better polymer recovery with decreasing
density and increasing separation uid density, the reverse
effect is observed. High-density polymers represented by PET
and PVC but also by PMMA, MDI-PUR and PC as well (Table 1)
are extracted and quantied with mean efficiencies above 70%.
This emphasizes that a density of 1.5 g cm3 represented by the
utilized NaBr solution is sufficient to extract the high-density
polymers included in the experiment.
In contrast, the lower density polymers like PE, PP and PS
show surprisingly low recovery rates (around 50%). A possible
reason for this behaviour might be a combined effect of low
density on one hand and low polarity on the other. Low density
causes buoyancy of particles and results in their accumulation
at the solution surface as well as its air–glass boundary. Additionally, the apolar character of the polymers might induce
a repulsion from the polar density agent to the glass surface and
result in adhesion.38,39 These interaction processes could occur
at different steps of the separation procedure, in the separator
itself, the collection beaker, and nally the separation funnel. In
the last step, the reduction of density solution level in the tilted
glass walls of the separation funnel further promotes adhesive
surface interactions for polymer particles which might result in
MP removal, despite intensive rinsing.
This suboptimal density and potentially polarity dependent
extraction behaviour has not been described in the microplastic
literature to our knowledge before. The inverted and almost
expected trend would be easily traced back to an insufficient
density of the selected density uid. The usage of very low
concentrated, rather small particles of the respective polymer
might have revealed so far not recognized but crucial gaps of
knowledge in the eld of density driven microplastic extraction.
In contrast to larger particles small particles are more affected
by surface properties or e.g., drag within the solution, while
large particles are mainly affected by density differences.18,40 It
therefore might appear logical that small particles are less
effectively separated from sediments than large particles.
Further, high amounts and percentages of MP can directly
inuence the extraction behaviour.41 The oen-used high
amounts might result in a kind of saturation effect, which
shields MP particles from adhesion or aggregation with other
material and therefore allows better extraction.
We take the high reproducibility of the density/polarity
dependent results as an incentive to pay closer attention to
MP extraction validation studies with low concentrations of
small microplastics in sediments (Section 3.3).
The same separator and separation uid had been used in an
independent approach of a joint project. In this experiment the
recovery was performed with 1 mm spherical, pristine particles.
The respective results are part of a much broader study and will
be published in detail in another context (Kögel et al. in prep).
With respect to the applicated particles, three polymer types
(PE, PS and PET) were overlapping with the presented study.
The determined recoveries of these 1 mm spheres from three
different sediments and nine extractions in total were 92.2% 
13.9 for PE, 93.3%  10.0 for PE and 94.4%  7.2 for PET. These
feasible satisfactory recoveries did not show any indication of
a potential polymer discrimination. Accordingly, the supposed

This journal is © The Royal Society of Chemistry 2021

View Article Online

Published on 29 September 2021. Downloaded by Auburn University on 8/5/2025 6:19:58 AM.

Paper
impact of surface properties on small particle behaviour implies
a potentially different extraction behaviour compared to big
particles.
Pristine versus incubation experiment. For most polymers,
the achieved extraction efficiencies and observed variances were
similar, disregarding whether they were pristine or incubated
before. While PE, PP, PS, PA6, PUR and PVC were extracted in
the same order of magnitude, the extraction efficiency for
PMMA and PET clearly increased in the incubation experiment
(Fig. 3).
Within polymer types, minor differences between pristine
and incubated were observed. The variance between experiments decreased for PE and PS if particles were incubated, but
was almost unchanged for PVC, PS, PA6, PUR, PC and PP. An
increased variance between recoveries appeared in the case of
PET and PMMA aer incubation.
Early alteration processes like biofouling are known to
change particle behaviour.24,42 A biofouling layer on the particles might reduce adhesion at glass surfaces as well as inducing
or reducing aggregation-affinity with other particles (organic
and inorganic) and directly inuence its density.20,25,43 Accordingly, surface adhesion and polymer particle buoyancy are
supposed to have a direct effect on polymer recoveries and their
respective variation coefficients.
The incubation experiment of PET and PMMA indicates that
early biofouling effects resulting from very basic conditions
represented by the performed experiment already have a visible
positive effect on extraction recovery. Since both polymers are
polyesters this polymer-specic effect might be even accompanied by polymer-specic biofouling.44,45
A deduction of any general extraction behaviour of polymers
from such incubation experiments needs large reservation
regarding environmental samples. Variations in natural organic
and inorganic matter might affect microplastic particles
differently concerning aggregation.21,44 Further, the specic
polymer composition, shape and alteration degree affect

Fig. 3 Polymer-specific recoveries with pristine polymers (a), incubated polymers (b) and in comparison (c); * 4th replicate did not
include PP; ** one outlier out of bounds at 280%.

This journal is © The Royal Society of Chemistry 2021

Analytical Methods
biofouling, surface properties, aggregation behaviour, and
sinking speeds.18,40,43
Critical evaluation of the experimental setup. Disregarding
the observed and discussed effects, some aspects regarding the
experimental setup are supposed to be of general relevance for
the reported surprising recoveries and partial variances of
studied polymer types. Here, in particular the comparable low
number of particles (Table 1) applied in the experiments is
a weak point and must be stressed. In the experiment, they
represent a compromise to get a low absolute polymer mass
(<50 mg) on one hand, representative of real environmental
conditions, but have still a size that allows an optical control
(binocular) for an accurate spiking process on the other.
Subsequently no further optical counting was possible since the
MP particles were not distinguishable from any sand particles.
Accordingly, the loss of just one single particle would already
result in a 5% loss e.g. for PVC (20 particle approach) and up to
20% loss relevant for most of the polymers (5–7 particle
approach) in overall recovery, determined by Py-GC/MS exclusively (Table S3†). Thus, the applied low number of spiked
particles, and the loss of one or a few particles have an immediate effect on both recovery and variance as well.
Further on, for the Py-GC/MS quantication of the given
process, the standard deviation of the respective polymer calibration varied between 3% and 13% if projected on the
respective polymer recovery (Fig. S3;† calculation example).
Consequently, the quantication process comprised an
inherent methodical variation. Both aspects needed critical
consideration regarding the observed variances in recovery data
of the polymer types as well.

3.2

Environmental samples

Embedded in the successful improvement and application of
the Py-GC/MS method for the MP quantication in different
complex environmental samples the same separator and
procedure were already applied for MP pre-concentration from
muddy sediments. This application on real sediments underlines the feasibility of the separation procedure in general.36
However, it reveals potential difficulties with respect to environmental samples as well. Such are size, shape and density
dependent inhomogeneity of polymer particles in the natural
environment and accordingly a challenge of environmental MP
analysis.36,46,47
Overall, the study reected a successful quantication of 7
high- and low-density polymer types in a muddy, intertidal
sediment (Fig. 4). It gave a meaningful order of magnitude of
the expected resulting polymer types, highly relevant for realistic recovery experiments. The summarized MP concentrations
of the four replicates varied between 48.4 mg (C) and 166 mg (D)
per kg wet weight (Fig. 4). The MP concentrations of individual
polymers were almost constantly less than 50 mg per kg sediment which was implemented in the presented spiking experiments. Sizes of suspected MP particles in these environmental
sample lters hardly exceeded 200 mm and are reected by
particle sizes in the presented recovery experiment.

Anal. Methods, 2021, 13, 5299–5308 | 5303

View Article Online

Published on 29 September 2021. Downloaded by Auburn University on 8/5/2025 6:19:58 AM.

Analytical Methods

Fig. 4 Microplastic content of 4 subsamples from a muddy intertidal
sediment in mg g1 wet weight (adapted from Fischer et al., 2019).

The overall complexity of natural samples and the associated
alterations of particle properties in natural environments
implicate problems if an objective transfer of extraction efficiencies is deviated from rather articial, much less complex
extraction experiments on natural sediment samples. Besides
the already addressed aspects, different degrees of physicochemical as well as biological alterations have to be considered
as well.
However, the results of a previous study36 were considered as
guidance for the presented recovery experiment and its capabilities to extract high- and low-density polymers at lower ppb
levels.

3.3

Critical reection with respect to literature data

As suspected sinks for MPs their investigation in sediments,
sludge and soil is of increasing importance. Several isolation
approaches have been developed almost simultaneously over
the last years. Highly surprised about our own unexpected but
reproducible results, we reviewed the existing literature for any
indicators of similar density or polarity dependent extraction
behaviours in former studies, in evidence with minor recoveries
particularly for small MP particles at low concentrations.
This review focused on extraction methods that included
recovery experiments. To fulll comparability of the broad
spectrum, it was restricted to elutriation and density separation
approaches. The results are provided in Table 2. It is subdivided
into three major sections: (1) the technical setup; (2) the
experimental setup, and (3) parameters regarding the overall
process evaluation. Here, the most meaningful and characteristic sub-criteria were emphasized. These assigned criteria
describe either determined aspects mostly associated with the
technical setup, or exible/adaptable aspects, mainly related to
the experimental setup or the evaluation process. All of them
can easily be changed and adapted in a future re-evaluation
process.
The criteria close to or matching our presented study were
marked blue, additionally. In the following discussion, we
focused almost only on the comparison with our results but use
this line-up to emphasize crucial aspects with respect to MP
analysis.

5304 | Anal. Methods, 2021, 13, 5299–5308

Paper
General experimental criteria. The technical setup reects
variable materials and applicable sample volumes. Oen the
tailored separator volume is focussed on a certain sample type
and its expected MP load. Here, the sample size (Table 2, A2) has
a direct impact on data quality and the achievable limit of
detection, although the latter is nally dened by the subsequent analytical detection principle for the MP (Table 2, C1, C2).
With respect to an inhomogeneous distribution of particulate
MP already discussed,46,47 the analysed volume should be as
representative as possible for the respective sample type.
Reecting on realistic environmental sediment sample
concentrations (300 particles per kg (ref. 48–50)) a volume above
500 ml should be used for feasible extractions. Only 5 of 13
studies presented separators with a respective volume.27,28,32–34
In principle, setups including relevant plastic parts bear an
inherent risk of contamination with the corresponding polymer.26,30 Accordingly, stainless steel or glass setups are
preferred. Here, glass setups have the advantage of enabling an
overall visual control.
All extraction procedures reported in Table 2 employed
sufficiently dense separation uids (Table 2, A4). However, nontoxic and less expensive density agents with sufficient density
capabilities (e.g. $1.5 g cm3) are favourable, e.g. NaBr over
ZnCl2.51 NaBr is commonly used with a density between 1.37
and 1.4 g cm3.52 Its extension to densities $1.5 g cm3, as
shown in the presented approach, assures the recovery of even
denser polymers (e.g. PVC/PET).
The experimental setup needs thorough examination to
identify comparable conditions and thus potentially similar
extraction behaviours to our study (Table 2; blue cells). Here, the
most crucial factors effecting extraction behaviour are particles
size (Table 2, B1), polymer density (Table 2, B2) and polymer
concentration (mass; particle count) (Table 2, B3–B4).18,40,41 An
ideal experimental setup needs to include polymers of different
densities/polarity to attribute respective effects.
Our study applied particle sizes between 150 and 300 mm.
This makes it one of the studies that consider particle sizes at
the lower end of the range studied in the contemplated
approaches (Table 2, B1). Although, several studies include this
size range (Table 2, B1), only ve studies reported recoveries for
high- and low-density polymers within this size class as well
(Table 2, B2).
If very high particle counts (103–106 per kg) and high or very
high masses (0.5–900 mg per kg) are applied, they are suspected
to mask any of the described effects. However, since almost all
of the referred studies used visual counting or balancing to
determine their recoveries such high concentrations are
warrantable. Focussing on realistic environmental conditions,
studies reported that environmental sediment samples rarely
exceeded 1000 particles per kg in the MP size range between 100
and 300 mm. More oen particle counts were even below 300
particles per kg.48–50 Environmental MP mass concentrations
commonly range below <200 mg kg1.34,36
With respect to particle sizes and concentrations only three
studies are comparable to our study.28,32,34 Although, no study
perfectly matches the conditions of small MP particles and low

This journal is © The Royal Society of Chemistry 2021

This journal is © The Royal Society of Chemistry 2021

Condensed literature overview on microplastic density separation approaches

a
Merged range of particle sizes (more details in respective study); (L) & (kg) ¼ number of particles per litre or kilogram; L/AP ¼ low-density/apolar; H/P ¼ high-density/polar; blue colouring ¼
similar criteria characteristic compared to our study; grey colouring ¼ insufficient information.

Table 2

Published on 29 September 2021. Downloaded by Auburn University on 8/5/2025 6:19:58 AM.

View Article Online

Paper
Analytical Methods

Anal. Methods, 2021, 13, 5299–5308 | 5305

View Article Online

Published on 29 September 2021. Downloaded by Auburn University on 8/5/2025 6:19:58 AM.

Analytical Methods
concentration in both particle count and mass for high- and
low-density polymers, three studies can be classied as
comparable.
In most approaches, process evaluation is performed by
visual counting or weighing. Here, in particular, the optical
preselection can be humanly biased.30 However, some of the
reviewed studies ensured their evaluation by additional or solely
instrumental identication techniques (Table 2, C1), but almost
none conducted the quantication by a common polymerspecic quantication technique (Table 2, C2). The resulting
polymer recoveries averaged around 90%, but reported efficiencies span broad ranges (54–100%).
Implications on extraction behaviour. The potential impact
of the reviewed studies on MP extraction behaviour is given in
the column “extraction behaviour” with respect to the available
information (Table 2). Those studies not using polymers of
different densities29,53 or did not specify polymer specic
recoveries33 were excluded from nal discussion (Table 2, D1).
The derived extraction behaviours either showed no density/
polarity dependent recoveries,26–28,31,33,41,52 lower recoveries for
high-density/polar polymers52,54 or the most interesting indication of lower recoveries for low-density/apolar polymers.30–32,34
Lower recoveries for high-density polymers were observed for
extraction procedures with a weak otation mechanism e.g.,
light density solution (NaCl) or weak elutriation.54–56 Here,
apparently the buoyancy induced by the weak otation mechanism is insufficient for high-density particles and causes
minor recoveries. This is most likely causing respective effects
in the reviewed studies by Quinn et al. and Mani et al.52,54
The majority of the studies indicate no density/polarity
dependent recoveries (8 studies) (Table 2, D1). It is expected
and found in our previous experiments, that large particles are
more affected by density differences then small particles. Small
particles are more affected by surface properties and thus less
effectively separated from sediments than large particles.18,40
This could explain no observed effects of four
approaches.26,27,31,33 Two studies used high polymer particle/
mass concentrations spiked into only small matrix
aliquots.41,52 This divergent proportion of high polymer loads to
low matrix volume might result in a more sufficient extraction
due to the saturation effect which shields MP particles from
adhesion or aggregation to other materials. Additionally, any
loss of a few particles is neglectable.41 Both arguments might
have prevented any observations with respect to density/polarity
driven behaviour. The remaining one study by Claessens et al.28
used a combination of otation and elutriation mechanisms for
MP separation and not gravity only. Here, a combined effect of
dense NaI solution and the elutriation might favour higher
recovery of small light (here PE) particles (250 mm) even at low
particle numbers (50 per polymer) and resulted in no density/
polarity dependent recoveries.
At least an initial suspicion that less dense, apolar polymers
might be more affected by density separation is indicated in the
data presented by Nakajima et al.31 They reported a higher
variance in PE regarding the small particle fraction. This might
be implied by two other studies as well, since they observed
slightly less effective recovery rates of PE.30,34 Unfortunately,

5306 | Anal. Methods, 2021, 13, 5299–5308

Paper
both studies give insufficient details on either particle size or
process evaluation, which hinders further discussion.
The most distinct evidence for a different behaviour of
apolar and low density polymers is indicated in the work of
Enders et al.32 Their experimental setup resulted in a 17% less
effective recovery of PE compared to PA6. The study stands out
by using a closely related setup to our study with almost
comparable polymer concentrations (particle; mass) while
utilizing small polymer particle sizes. Unfortunately, it was only
restricted to these two polymer types.

4. Conclusion
Recovery experiments from articial sediments spiked with
pristine and biofouled MPs were performed with a custommade density separator, leading to unexpected results. While
the precision was absolutely convincing with respect to the
complex procedure a variable, polymer specic accuracy was
obvious. Low density, non or less polar polymers revealed lower
recoveries compared to those of higher density and polarity. The
experimental setup focussed on low concentrations (5–50 mg),
low particle numbers (N ¼ 5–20), and small particle sizes (150–
300 mm) and a broad polarity polymer range was adapted to
mimic realistic environmental concentrations. This so far
undescribed, counterintuitive and unanticipated extraction
behaviour was suspected to indicate a density/polarity related
and concentration dependent particle size/surface phenomenon inside the separation system. The fact that high-density
polymers were more effectively recovered changed the determining focus from density effects of the separation uid,
towards other relevant factors in particular regarding small
microplastic particles.
Published studies of MP density extraction validation studies
covered almost no comparable recovery experiments concerning particle size, polymer density range and polymer concentration. However, a few studies focused on smaller particle sizes
(<300 mm, low concentration ppb range/low counts), and
different density polymer concentrations at least indicate
evidence of this density/polarity driven extraction behaviour as
well.
The presented results emphasize the complexity of microplastic analysis in particular, when complex pre-concentration
steps are involved. Since authentic environmental sediments
will introduce additional factors that further affect particle
extraction behaviour the perfect microplastic extraction technique at trace levels and respective recovery experiment are still
under construction.

Conflicts of interest
There are no conicts to declare.

Acknowledgements
We would like to thank Joanne Yong (ICBM Planktology
working group) for providing the benthic algae community for
our incubation experiment as well as Anke Müllenmeister and

This journal is © The Royal Society of Chemistry 2021

View Article Online

Paper
Sebastian Neun for their support in the laboratory. The study
was kindly supported by the German Federal Ministry of
Education and Research (Bundesministerium für Bildung und
Forschung, BMBF) in the joint research projects BASEMAN
(grant ID 03F0734D; JPI-Oceans) and PLAWES (grant ID
03F0789E; FONA III, MH).

Published on 29 September 2021. Downloaded by Auburn University on 8/5/2025 6:19:58 AM.

References
1 R. Geyer, J. R. Jambeck and K. L. Law, Sci. Adv., 2017, 3,
e1700782.
2 R. C. Thompson, Y. Olsen, R. P. Mitchell, A. Davis,
S. J. Rowland, A. W. G. John, D. McGonigle and
A. E. Russell, Science, 2004, 304, 838.
3 K. L. Law and R. C. Thompson, Science, 2014, 345, 144–145.
4 J. R. Jambeck, R. Geyer, C. Wilcox, T. R. Siegler, M. Perryman,
A. Andrady, R. Narayan and K. L. Law, Science, 2015, 347,
768–771.
5 Z. Akdogan and B. Guven, Environ. Pollut., 2019, 254, 113011.
6 L. C. Woodall, A. Sanchez-Vidal, M. Canals, G. L. J. Paterson,
R. Coppock, V. Sleight, A. Calafat, A. D. Rogers,
B. E. Narayanaswamy and R. C. Thompson, R. Soc. Open
Sci., 2014, 1, 140317.
7 H. Zhang, Estuarine, Coastal Shelf Sci., 2017, 199, 74–86.
8 C. Lorenz, L. Roscher, M. S. Meyer, L. Hildebrandt, J. Prume,
M. G. J. Löder, S. Primpke and G. Gerdts, Environ. Pollut.,
2019, 252, 1719–1729.
9 A. Haegerbaeumer, M.-T. Mueller, H. Fueser and
W. Traunspurger, Front. Environ. Sci., 2019, 7, 17.
10 T. Kögel, Ø. Bjorøy, B. Toto, A. M. Bienfait and M. Sanden,
Sci. Total Environ., 2020, 709, 136050.
11 S. Primpke, M. Fischer, C. Lorenz, G. Gerdts and
B. M. Scholz-Böttcher, Anal. Bioanal. Chem., 2020, 8283–
8298.
12 M. E. Miller, F. J. Kroon and C. A. Motti, Mar. Pollut. Bull.,
2017, 123, 6–18.
13 F. Stock, C. Kochleus, B. Bänsch-Baltruschat, N. Brennholt
and G. Reifferscheid, TrAC, Trends Anal. Chem., 2019, 113,
84–92.
14 J. C. Prata, J. P. da Costa, A. C. Duarte and T. Rocha-Santos,
TrAC, Trends Anal. Chem., 2019, 110, 150–159.
15 A. Stolte, S. Forster, G. Gerdts and H. Schubert, Mar. Pollut.
Bull., 2015, 99, 216–229.
16 F. M. C. Fazey and P. G. Ryan, Environ. Pollut., 2016, 210,
354–360.
17 T. J. Hoellein, A. J. Shogren, J. L. Tank, P. Risteca and
J. J. Kelly, Sci. Rep., 2019, 9, 3740.
18 K. Waldschläger and H. Schüttrumpf, Environ. Sci. Technol.,
2019, 53, 1958–1966.
19 M. Long, B. Moriceau, M. Gallinari, C. Lambert, A. Huvet,
J. Raffray and P. Soudant, Mar. Chem., 2015, 175, 39–46.
20 A. Porter, B. P. Lyons, T. S. Galloway and C. Lewis, Environ.
Sci. Technol., 2018, 52, 7111–7119.
21 R. Leiser, G.-M. Wu, T. R. Neu and K. Wendt-Potthoff, Water
Res., 2020, 176, 115748.
22 N. Wu, Y. Zhang, W. Li, J. Wang, X. Zhang, J. He, J. Li, Y. Ma
and Z. Niu, Sci. Total Environ., 2020, 717, 134431.
This journal is © The Royal Society of Chemistry 2021

Analytical Methods
23 D. Lobelle and M. Cunliffe, Mar. Pollut. Bull., 2011, 62, 197–
200.
24 D. Kaiser, N. Kowalski and J. J. Waniek, Environ. Res. Lett.,
2017, 12, 124003.
25 T. H. Nguyen, F. H. M. Tang and F. Maggi, PLoS One, 2020,
15, e0228209.
26 R. L. Coppock, M. Cole, P. K. Lindeque, A. M. Queirós and
T. S. Galloway, Environ. Pollut., 2017, 230, 829–837.
27 M.-T. Nuelle, J. H. Dekiff, D. Remy and E. Fries, Environ.
Pollut., 2014, 184, 161–169.
28 M. Claessens, L. van Cauwenberghe, M. B. Vandegehuchte
and C. R. Janssen, Mar. Pollut. Bull., 2013, 70, 227–233.
29 E. Hengstmann, M. Tamminga, C. Vom Bruch and
E. K. Fischer, Mar. Pollut. Bull., 2018, 126, 263–274.
30 H. Nel, S. Krause, G. H. Sambrook Smith and I. Lynch,
MethodsX, 2019, 6, 2656–2661.
31 R. Nakajima, M. Tsuchiya, D. J. Lindsay, T. Kitahashi,
K. Fujikura and T. Fukushima, PeerJ, 2019, 7, e7915.
32 K. Enders, R. Lenz, J. A. Ivar do Sul, A. S. Tagg and
M. Labrenz, MethodsX, 2020, 7, 100784.
33 H. K. Imhof, J. Schmid, R. Niessner, N. P. Ivleva and
C. Laforsch, Limnol. Oceanogr.: Methods, 2012, 10, 524–537.
34 A. Gomiero, K. B. Øysæd, T. Agustsson, N. van Hoytema,
T. van Thiel and F. Grati, Chemosphere, 2019, 227, 705–714.
35 M. A. Browne, P. Crump, S. J. Niven, E. Teuten, A. Tonkin,
T. Galloway and R. Thompson, Environ. Sci. Technol., 2011,
45, 9175–9179.
36 M. Fischer and B. M. Scholz-Böttcher, Anal. Methods, 2019,
11, 2489–2497.
37 M. Fischer and B. M. Scholz-Böttcher, Environ. Sci. Technol.,
2017, 51, 5052–5060.
38 R. H. Perry and D. W. Green, Perry's chemical engineers'
handbook, McGraw-Hill, New York, 8th edn, 2008.
39 K. L. Mittal, Advances in Contact Angle, Wettability and
Adhesion, John Wiley & Sons, Inc, Hoboken, NJ, USA, 2015.
40 K. Waldschläger, M. Born, W. Cowger, A. Gray and
H. Schüttrumpf, Environ. Res., 2020, 191, 110192.
41 J. Lin, X.-P. Xu, B.-Y. Yue, Y. Li, Q.-Z. Zhou, X.-M. Xu, J.-Z. Liu,
Q.-Q. Wang and J.-H. Wang, Sci. Total Environ., 2021, 760,
144316.
42 M. Long, B. Moriceau, M. Gallinari, C. Lambert, A. Huvet,
J. Raffray and P. Soudant, Mar. Chem., 2015, 175, 39–46.
43 M. van Melkebeke, C. Janssen and S. de Meester, Environ.
Sci. Technol., 2020, 54, 8668–8680.
44 T. Artham, M. Sudhakar, R. Venkatesan, C. Madhavan Nair,
K. V. G. K. Murty and M. Doble, Int. Biodeterior. Biodegrad.,
2009, 63, 884–890.
45 F. Lagarde, O. Olivier, M. Zanella, P. Daniel, S. Hiard and
A. Caruso, Environ. Pollut., 2016, 215, 331–339.
46 M. Tamminga, S.-C. Stoewer and E. K. Fischer, Environ.
Pollut., 2019, 254, 112970.
47 S. M. Abel, S. Primpke, I. Int-Veen, A. Brandt and G. Gerdts,
Environ. Pollut., 2021, 116095.
48 P. Doyen, L. Hermabessiere, A. Dehaut, C. Himber,
M. Decodts, T. Degraeve, L. Delord, M. Gaboriaud,
P. Moné, J. Sacco, E. Tavernier, T. Grard and G. Duos,
Environ. Sci. Pollut. Res. Int., 2019, 26, 28010–28021.
Anal. Methods, 2021, 13, 5299–5308 | 5307

View Article Online

Published on 29 September 2021. Downloaded by Auburn University on 8/5/2025 6:19:58 AM.

Analytical Methods
49 J. Gago, O. Carretero, A. V. Filgueiras and L. Viñas, Mar.
Pollut. Bull., 2018, 127, 365–376.
50 K. Zhang, H. Shi, J. Peng, Y. Wang, X. Xiong, C. Wu and
P. K. S. Lam, Sci. Total Environ., 2018, 630, 1641–1653.
51 J. Gohla, S. Bračun, G. Gretschel, S. Koblmüller,
W. Maximilian and C. Pacher, Potassium carbonate (K2CO3)
– a cheap, non-toxic and high-density oating solution for
microplastic isolation from beach sediments, 2020.
52 B. Quinn, F. Murphy and C. Ewins, Anal. Methods, 2017, 9,
1491–1498.

5308 | Anal. Methods, 2021, 13, 5299–5308

Paper
53 M. Kedzierski, V. Le Tilly, P. Bourseau, H. Bellegou, G. César,
O. Sire and S. Bruzaud, Mar. Pollut. Bull., 2016, 107, 315–323.
54 T. Mani, S. Frehland, A. Kalberer and P. Burkhardt-Holm,
Anal. Methods, 2019, 11, 1788–1794.
55 M. Kedzierski, V. Le Tilly, G. César, O. Sire and S. Bruzaud,
Mar. Pollut. Bull., 2017, 115, 120–129.
56 A. L. Lusher, K. Munno, L. Hermabessiere and S. Carr, Appl.
Spectrosc., 2020, 74, 1049–1065.

This journal is © The Royal Society of Chemistry 2021