Professional Documents
Culture Documents
Monitoring Microplastics in Drinking Water An Interlaboratory Study To Inform Effective Methods For Quantifying and Characterizing Microplastics
Monitoring Microplastics in Drinking Water An Interlaboratory Study To Inform Effective Methods For Quantifying and Characterizing Microplastics
Monitoring Microplastics in Drinking Water An Interlaboratory Study To Inform Effective Methods For Quantifying and Characterizing Microplastics
Chemosphere
journal homepage: www.elsevier.com/locate/chemosphere
H I G H L I G H T S G R A P H I C A L A B S T R A C T
A R T I C L E I N F O A B S T R A C T
Handling Editor: Derek Muir California Senate Bill 1422 requires the development of State-approved standardized methods for quantifying
and characterizing microplastics in drinking water. Accordingly, we led an interlaboratory microplastic method
Keywords: evaluation study, with 22 participating laboratories from six countries, to evaluate the performance of widely
Microplastic used methods: sample extraction via filtering/sieving, optical microscopy, FTIR spectroscopy, and Raman
Method
spectroscopy. Three spiked samples of simulated clean water and a laboratory blank were sent to each laboratory
Monitoring
with a prescribed standard operating procedure for particle extraction, quantification, and characterization. The
Spectroscopy
Microscopy samples contained known amounts of microparticles within four size fractions (1–20 μm, 20–212 μm, 212–500
μm, >500 μm), four polymer types (PE, PS, PVC, and PET), and six colors (clear, white, green, blue, red, and
orange). They also included false positives (natural hair, fibers, and shells) that may be mistaken for micro
plastics. Among laboratories, mean particle recovery using stereomicroscopy was 76% ± 10% (SE). For particles
in the three largest size fractions, mean recovery was 92% ± 12% SD. On average, laboratory contamination
from blank samples was 91 particles (± 141 SD). FTIR and Raman spectroscopy accurately identified
* Corresponding author.,
** Corresponding author.
E-mail addresses: hannah.defrond@utoronto.ca (H. De Frond), leahth@sccwrp.org (L. Thornton Hampton), sydk@sccwrp.org (S. Kotar), kristineg@sccwrp.org
(K. Gesulga), cindymatuch@gmail.com (C. Matuch), waynel@sccwrp.org (W. Lao), stevew@sccwrp.org (S.B. Weisberg), charlesw@sccwrp.org (C.S. Wong),
chelsea.rochman@utoronto.ca (C.M. Rochman).
https://doi.org/10.1016/j.chemosphere.2022.134282
Received 25 November 2021; Received in revised form 14 February 2022; Accepted 7 March 2022
Available online 10 March 2022
0045-6535/© 2022 Elsevier Ltd. All rights reserved.
H. De Frond et al. Chemosphere 298 (2022) 134282
microplastics by polymer type for 95% and 91% of particles analyzed, respectively. Per particle, FTIR spec
troscopy required the longest time for analysis (12 min ± 9 SD). Participants demonstrated excellent recovery
and chemical identification for particles greater than 50 μm in size, with opportunity for increased accuracy and
precision through training and further method refinement. This work has informed methods and QA/QC for
microplastics monitoring in drinking water in the State of California.
Microplastics represent a potential threat to human health, with Participants were from 22 laboratories in the United States, Canada,
exposure through the air we breathe (Chen et al., 2020; Gasperi et al., Germany, China, Australia, and Norway. The study benefitted from di
2015), the food we eat (Cox et al., 2019; EFSA Panel on Contaminants in versity in institutions, laboratory practice, and experience by including
the Food Chain (CONTAM), 2016), and the water we drink (Eerkes- laboratories from different countries. The study was conducted ‘blind’,
Medrano et al., 2019). While further research is needed to understand and participating laboratories did not know any details of the spiked
biological fate and toxicity of microplastics in humans (Barboza et al., particle content within each sample.
2018; Carbery et al., 2018; Rist et al., 2018), there is sufficient evidence Laboratories were asked to follow a strict protocol (Appendix A.1), in
to impose a precautionary approach for monitoring and mitigation of which particles were extracted, then counted and identified using ster
microplastics in drinking water (Danopoulos et al., 2020; Leslie and eomicroscopy, followed by chemical identification using either FTIR
Depledge, 2020; Senathirajah et al., 2021). and/or Raman spectroscopy on a subsample of particles (Appendix B.3).
Drinking water exposure has become a focal point for the State of Laboratories self-identified as novice (level 0 = new to microplastics
California, which has a legislative mandate (Senate Bill 1422) to enact analysis, having participated in a three-day training course on these
routine monitoring, including adoption of standard methods for imple methods immediately prior to the study), intermediate (level 1 = < 1
menting that requirement (Wyer et al., 2020). Such methods do not year of experience) or experts (level 2 = > 1 year experience). Each
currently exist, though there is a growing body of research from which to laboratory was also asked to record aspects related to QA/QC proced
draw. Microscopy is a useful pre-screening tool for enumerating ures, time, and cost.
microplastic particles and is a comparatively quick method with low
upfront and equipment costs (Primpke et al., 2020). Fourier-Transform
2.1. Creation of spiked samples
Infrared (FTIR) and Raman spectroscopy are often used to confirm
microplastic particle counts by providing chemical confirmation of
Spiked simulated clean water samples were prepared at Southern
material type. Several studies have previously demonstrated the capa
California Coastal Water Research Project (SCCWRP). Each spiked
bilities of these methods (Cabernard et al., 2018; Käppler et al., 2016; Xu
sample contained a combination of 15 different types of plastic particles
et al., 2019).
(Table S1). Plastic fragments of polystyrene (PS), polyvinyl chloride
Establishment of standard methods for management application re
(PVC), polyethylene terephthalate (PET) and polyethylene (PE) of
quires several additional steps beyond publishing methods in peer-
various sizes and PET microfibers came from the Norwegian Institute for
reviewed journals. Management application requires development of a
Water Research, each in individual gelatin capsules containing sodium
detailed standard operating procedure with sufficient specificity to
bicarbonate and malic acid to facilitate dissolution. Spheres of PE and PS
ensure repeatability across laboratories. It also requires a multi-
(Table S1 - particles 4–7) were sourced from Cospheric LLC© (Santa
laboratory method evaluation study to determine performance charac
Barbara, California, USA). Spiked plastics included some of the most
teristics, including demonstration of whether the precision and accuracy
produced polymer types (PlasticsEurope, 2020) with a range of den
of the chosen methods are suitable for the intended purpose. These data
sities. Spiked plastic particles were white, clear, blue, green, and orange,
serve as a foundation for the development of laboratory accreditation,
and morphologically were fragments, spheres, and fibers. Across all
which define expectations for well-performing laboratories.
samples, the mean plastic content was 609 (±132 SD) particles, with
There have been several interlaboratory studies assessing micro
average particle content of 249 (±60 SD) particles when the 1–20 μm
plastic methods performance across laboratories with blind microplastic
size fraction is excluded. The standard deviation of the spiked particle
samples (Cadiou et al., 2020; Isobe et al., 2019; Michida et al., 2019;
count was calculated by summing all standard deviations across all
Müller et al., 2020; Van Mourik et al., 2021). However, none have been
spiked particle types.
conducted using specific prescribed methodologies, which results in
In addition, four types of natural or ‘false positive’ particles were
high of variability among labs and across methods. In addition, all these
added (mean 80 particles per sample) to represent particles commonly
studies except for one (Müller et al., 2020) have been limited to particles
found within real-world samples which might be mistaken for plastics.
larger than 50 μm, which is problematic because drinking water is
These were fibers (undyed and dyed cellulose, animal fur) and shell
typically filtered to at least this size and monitoring needs to focus on
fragments (Table S1 - Particles 16–19).
smaller particles that can potentially pass through filters (Na et al.,
Simulated clean water samples were created using 1 μm filtered
2021).
(PCTE, Sterlitech) deionized water, referred to as microplastics analysis
Here we present a multi-laboratory validation study using three
grade (MAG) water. Glass containers were previously washed with soap
microplastic measurement methods: stereomicroscopy, Raman spec
and water, pre-ashed at 450 ◦ C to destroy any possible contaminants,
troscopy, and FTIR spectroscopy. For each, we establish performance
and triple-rinsed with 1 μm filtered deionized water. 250 mL of MAG
characteristics, including accuracy and precision. We also establish the
water was placed in a heated water bath. Pills containing microplastics
time requirements to implement such methods, necessary for use in
were added (12 total per sample), and jars were kept at 48 ◦ C for 2 h
required monitoring. The study included laboratories from around the
until all soluble inert components (gelatin, malic acid, sodium bicar
world with a range of experience, and all received samples spiked with
bonate) had dissolved, and the plastic particles dispersed. Each sample
diverse particle shapes, densities, colors and sizes, a strict protocol to
then received an additional 155 mL of MAG water and 45 mL of 1 μm
follow for extraction, particle identification, counting and chemical
filtered (PCTE, Sterlitech) 10% Alcojet® solution to minimize particu
identification. These results collectively inform recommendations for
late agglomeration. The remaining particles were manually added using
microplastic analysis in drinking water, for monitoring programs within
metal forceps. For blank samples, the same process was followed with 12
the State of California and beyond.
empty pills (i.e., without plastic particles) each. Prepared samples were
2
H. De Frond et al. Chemosphere 298 (2022) 134282
covered and stored at room temperature prior to shipping. Throughout methods (e.g., filter material), instrument specifics (e.g., manufacturer,
all stages of sample preparation, measures were in place to minimize software), microscopy settings (e.g., magnification), and spectroscopic
particle contamination (Appendix B.1). To confirm spiked particle method parameters. The full dataset is publicly available to download
counts, three test samples were prepared, extracted, and examined via via the SCCWRP webpage: microplastics.sccwrp.org.
stereomicroscope prior to sample distribution (Table S2). To quantify
contamination from sample preparation, a contamination check was 2.7. Data synthesis and analysis
carried out for every sample batch (Appendix B.2). Each participating
laboratory received three spiked samples and one blank sample. Laboratories were anonymized, and data was analyzed using R
(Version 4.0.3). Total particle counts encompass all suspected micro
2.2. Sample processing instructions plastic particles counted via microscopy, including spiked particles, false
positive natural particles misidentified as plastic, and particles intro
Previous studies noted that recoveries became incomparable when duced from background contamination, i.e., recovery data was not
laboratories were given the freedom to choose their extraction tech corrected based upon blank sample contamination or spectroscopy.
nique (Isobe et al., 2019). Therefore, a standard operating procedure Overall recovery was evaluated by comparing reported total particle
(SOP) (Appendix A.1) for extraction was provided for participants to counts with spiked particle counts for each sample. Precision was
follow precisely. This protocol included a recommended supplies list, evaluated using standard error of recovery. Among samples, there was
and instructions for sample extraction based upon commonly used some variation in the number of manually added particles. This varia
methods (Hidalgo-Ruz, 2012; Masura et al., 2015; Michida et al., 2019) tion was documented and accounted for when calculating recovery for
considering equipment requirements, cost, and time. Laboratories were each sample.
instructed to use sieving and vacuum filtration to extract particles from Results were separated by particles counted above and below the 20
the samples. Participating laboratories were encouraged to practice μm size fraction. For all statistical tests, the significance level was set at
QA/QC measures to minimize laboratory contamination, in line with p < 0.05. Normality of data was tested using the Shapiro-Wilk test and
recommendations from previous studies (Prata et al., 2020a) and other homogeneity of variance was tested using Bartlett’s test. All recovery
practices from personal experience and communication (Appendix A.2). data was non-normal, so Kruskal-Wallis tests were used. The standard
error of recovery within each lab was used to compare the precision
2.3. Sample extraction within individual laboratories. To test if there was a relationship be
tween accuracy (difference between reported recovery and 100%) and
Samples were fractioned via sieving and vacuum filtration into four precision (standard error of recovery) among laboratories, Pearson’s
size fractions: 1–20 μm, 20–212 μm, 212–500 μm, and >500 μm. Par correlation coefficient test was used.
ticipants were given the option to rinse particles from each sieve into To determine the accuracy of microscopy for identifying plastic
glass jars, or to vacuum-filter each size fraction onto filter paper prior to particles, the mean proportion of particles identified as plastic using
stereomicroscopy (hereafter referred to as microscopy). spectroscopy was calculated for each laboratory. Some laboratories re
ported that logistical limitations prevented them from following the
2.4. Microscopy prescribed subsampling procedure. Therefore, these results are treated
as estimates, as the number of particles identified using spectroscopy
Following sample extraction, laboratories used microscopy to count varied widely among laboratories and was not always representative.
all suspected microplastic particles, and to categorize them by color and To determine the accuracy of FTIR and Raman spectroscopy for
morphology based upon keys provided. Particles were also measured chemically identifying material types, the images and raw data sub
(length and width). For chemical characterization, a subsampling mitted by each laboratory were checked. Each particle visually and
strategy was provided using methods outlined in De Frond et al. (2022, descriptively matching a spiked plastic particle by color and
in prep.) where up to 30 particles of each color and morphology com morphology was identified. Results were then filtered to include only
bination (e.g., white fragments, blue fragments) within each size frac confirmed spiked particles chemically identified by spectroscopy. As
tion were selected (Appendix B.3). mentioned, the number of particles chemically identified using spec
troscopy varied among laboratories. Although a comparison of perfor
2.5. Spectroscopy mance among laboratories was intended, this was not possible without
introducing bias from the number of particles identified by each lab.
For laboratories using μ-FTIR (or FTIR spectroscopy, hereafter Therefore, results from all laboratories were pooled into one dataset,
collectively called FTIR) or μ-Raman spectroscopy (hereafter called and the overall proportion of correct chemical identification results was
Raman) to analyze particle composition, all subsampled particles were determined. For example, to determine accuracy of FTIR for identifying
chemically identified. Specific methods for particle preparation and PET particles, all spiked PET particles chemically identified among
instrument settings here were not specified, and the instrument used by laboratories were pooled, and the proportion of correct results was re
each lab differed by manufacturer and model. However, general in ported. Some particle types had the same visible appearance (e.g., Par
structions for conducting manual microplastic analyses using the in ticles 14 and 15), therefore, it was not possible to determine the true
struments available at SCCWRP (Thermo Fisher Nicolet iN10 FTIR polymer type prior to spectroscopic analysis. In these cases, if the
spectrometer and HORIBA XploRA ™ Plus Raman spectrometer) to local chemical identification result was one of the possible polymer types (e.
participants were provided. g., PET or PS) the result was considered correct.
Logistical results for time (per sample and per particle) were
2.6. Data submission compared for all steps in the analyses. The mean result was calculated
among laboratories and compared between methods using Mann-
Participating laboratories submitted data via an online data entry Whitney U tests. Scripts used to generate all statistical analyses, sum
form (Appendix C). Data submission included a description of mary results and figures are available at microplastics.sccwrp.org.
morphology, color, and size fraction for each suspected microplastic
particle counted using microscopy. For subsampled particles, the length, 3. Results
width, particle image, and the chemical identification result from
spectroscopic analyses were required. Additional data was also collected Twenty-two laboratories submitted results for at least one method.
on laboratory information (e.g., QA/QC measures, expertise), extraction All laboratories submitted results for microscopy and counted particles
3
H. De Frond et al. Chemosphere 298 (2022) 134282
in size fractions >20 μm. Of the 22 laboratories, 13 did not count par For particles in size fractions >20 μm (n = 22 laboratories), the
ticles in the 1–20 μm size fraction. Seventeen laboratories submitted standard error of recovery within each laboratory ranged from 1 to 35%
results for their blank sample. For chemical identification, 11 labora (mean SE = 11%, Fig. 1C and Figure S2). For the <20 μm size fraction (n
tories used FTIR spectroscopy and 9 used Raman spectroscopy. = 9 laboratories), standard error of recovery ranged from 1 to 120%
(mean SE = 19%. Fig. 1D and Figure S2). Among all laboratories, there
was not a statistically significant relationship between accuracy (the
3.1. Microscopy: particle recovery
difference between mean recovery per laboratory and 100% recovery,
dependant on the size fractions counted) and precision (SE) of recovery
We measured particle recovery by comparing the number of sus
per laboratory (Figure S3).
pected microplastic particles counted using microscopy to the number of
spiked plastic particles. Across all labs, five (23%) reported mean par
ticle counts that were within the true range (total spike ± SD). Thirteen 3.2. Procedural contamination
laboratories (59%) reported total particle counts lower than the true
value (total spike – SD) and four (18%) reported particle counts above Among laboratories, mean blank sample particle count was 91 ± 141
the true value (total spike +SD). Mean recovery among all laboratories (SD) particles (median = 45) and ranged from 7 to 511 particles (Fig. 2,
and all size fractions was 76 ± 10% (SE), with a significant difference in Table S3). The most common colors reported were black, clear, blue, and
recovery among laboratories (Kruskal-Wallis test). brown (Figure S4). The least common colors were gold, pink, purple and
Given that some laboratories did not submit data for the smallest size silver. The most common particle morphologies reported were fibers,
fraction, results are separated into >20 μm (n = 22) and <20 μm (n = 9) then fragments (Figure S5). All but one laboratory counted fibers within
size fractions. Within the >20 μm size fractions, actual particle sizes as their blank, and all but three laboratories reported fragments. By size
measured by visual microscopy ranged from 50 μm to 2000 μm. The fraction (Figure S6), mean blank particle counts were greatest in the
minimum and maximum particle sizes in the <20 μm size fraction were 20–212 μm size fraction at 34 particles per sample.
3 μm and 20 μm. Mean recovery of particles in size fractions >20 μm was
92 ± 12% among laboratories (Fig. 1A) (mean = 223 particles, median 3.3. Accuracy of microscopy for identifying plastic particles
= 213 particles, Figure S1; Table S3). Recoveries within this group
ranged from 13 to 246%. For the <20 μm size fraction, mean recovery Spectroscopy results were used to assess the accuracy of using visual
was 32 ± 16% (Fig. 1B) (mean = 109 particles, median = 79 particles, microscopy to differentiate between plastic and non-plastic (i.e., natural
Figure S1; Table S3). Recoveries within this group ranged from 1 to or other anthropogenic) particles, determined by the proportion of
158%. particles identified as plastic or non-plastic from the subsample per
4
H. De Frond et al. Chemosphere 298 (2022) 134282
5
H. De Frond et al. Chemosphere 298 (2022) 134282
4. Discussion the spiked plastics and false positive particles analyzed, accuracy of
chemical identification using both Raman and FTIR spectroscopy was
Overall, the methods used in this study are fit for purpose to quantify lowest for fibers compared to other morphologies. The effects of particle
and characterize microplastics in drinking water for most applications. characteristics on method accuracy for both microscopy and spectros
Below we discuss the variables that affected the recovery and precision copy are discussed further in Kotal et al., (2022, this issue), and De Frond
of our results and the applicability of this work to the monitoring of et al., (2022, this issue).
microplastics in drinking water. Combined, these findings help establish whether the methods we
This study included a SOP for all labs to follow, which is unique to examined are fit for purpose. It is important to recognize there are
other method evaluation studies (Cadiou et al., 2020; Isobe et al., 2019; several purposes for monitoring microplastics in drinking water, and
Michida et al., 2019; Müller et al., 2020; Van Mourik et al., 2021). The methods should be evaluated based on their ability to meet each
inclusion of an SOP is likely the reason that variability across labora objective. One of those objectives is assessment of source water
tories was less than in other multi-laboratory studies with blind samples, contamination, for which these methods work well since source water
which reported lower mean recovery among laboratories (Cadiou et al., generally involves description of larger sized particles (e.g., >45 μm in
2020; Michida et al., 2019). Michida et al. (2019) attributed underes size (Zhu et al., 2021)). Another objective may be to determine exposure
timation to particle loss during sample extraction, which we expect was in a risk assessment. For humans, particles <20 μm are considered most
reduced by following the SOP that was tested prior to use to ensure relevant for both exposure and health effects (Prata et al., 2020b).
minimal procedural particle loss (Table S2). Although difficulties in measuring these particles are evident, this study
However, even using an SOP, we found that interlaboratory repeat did provide an estimate for the recovery rate of particles <20 μm (32%)
ability was highly varied, and results were significantly different among making it possible to correct for the bias.
laboratories. As recovery results were not corrected based upon particle While the primary goal of the study was to document performance,
counts within blank samples, it is likely that differences in particle the study also served to identify several areas in which methods could be
contamination may have contributed to the differences in particle re improved. One of those areas is in blank contamination. On average,
coveries among laboratories. Notably, this variation does not appear laboratory contamination from blank samples was 91 particles, although
strongly attributable to experience level, with laboratories of all expe this varied among labs. This variation was most likely due to differences
rience levels having false positive particle counts where they incorrectly in protocols for contamination control and the range in time spent
identified non-plastic materials (Cadiou et al., 2020). It is important to processing samples. Besides the opportunity to reduce blank contami
account for different experience levels in establishing monitoring pro nation, our SOP did not require or prescribe a procedure for blank
grams as novice laboratories will be common in new microplastics correction, which added to variability across laboratories. This proced
monitoring programs, at least initially. This can be done via standard ure was intentionally left open-ended as correction can be complex, as
ized particle identification protocols and training, which are discussed the amount of airborne laboratory contamination can differ among
further by Kotar et al. (this volume). shapes and colors of particles (Prata et al., 2020a). Further, limits of
Although other factors are relevant, particle size was the most quantitation are driven by variability in replicate response and blank
important factor affecting method performance. For particles in size contamination. If adopted, the limit of quantitation will help determine
fractions >20 μm, mean recovery was 92%, whereas performance fell to the volume of water that must be sampled to confidently distinguish a
32% for particles <20 μm. This is in line with findings of previous measured value from zero. Recommendations for minimization and
interlaboratory method comparison studies which reported overlooking reporting of blank particle counts and limits of quantitation are dis
of particles in smaller size fractions when using microscopy (Isobe et al., cussed further in Munno et al., (2022) (this issue).
2019; Michida et al., 2019; Müller et al., 2020). Visual microscopy is Another consideration in determining fitness for purpose is the time
thus appropriate for identifying plastic particles above 50 μm, the expenditures required for implementation, an area identified for
smallest measured particle size within the >20 μm size fractions. Rec improvement. On average, laboratories spent more than 50 h per sample
ommendations for microscopy are discussed in Kotar et al., (2022, this to complete sample processing through to visual quantification followed
issue). Method recommendations for spectroscopy, including priorities by Raman spectroscopy, which was the faster of the two spectroscopic
for future research are discussed in De Frond et al., (2022, this issue). methods (per particle). The times recorded are likely an overestimate as
Particle size s imilarly affected ability to distinguish polymer types our samples were spiked with more plastics than would be found in a
via spectroscopy. FTIR accurately identified microplastics from non- typical drinking water sample. Wong et al., (2022, this issue) have made
plastics and correctly identified polymer type for particles in size frac corrections for sample density and arrived at an estimate that was still
tions above 20 μm but was ineffective below 20 μm. This was expected, more than 40 h per sample for a typical drinking water sample. This
as the reported limit of detection using FTIR ranges from 10 to 20 μm in large time investment leads to consideration of a tiered monitoring
other studies due to wavelength limitations (Primpke et al., 2020; Xu system in which spectroscopic methods are used only when detailed
et al., 2019). Raman was 94% effective in identifying plastics, even particle description is needed, and less expensive methods are used for
including particles in the smallest size fraction in which particles ranged screening level questions (Coffin et al., 2022; this issue). In addition, it
from 3 to 20 μm, consistent with other studies where Raman spectros prioritizes the need for developing method automation, which is in
copy was used to identify small particles (Anger et al., 2018; Araujo progress already for both FTIR and Raman spectroscopy (Primpke et al.,
et al., 2018; Oβmann et al., 2018). However, Raman performed poorly 2017; Xu et al., 2019).
for dyed cellulose fibers. Other works have reported fluorescence
interference from dyes and pigments for Raman spectroscopy, a draw 5. Conclusions
back of this method (Araujo et al., 2018; Lenz et al., 2015; Munno et al.,
2020). In addition, our polymer identification rate probably represents a Method performance was highly dependent on particle size, with
maximum because we did not use environmentally-weathered particles, good recovery for particles >50 μm. Both FTIR and Raman spectroscopy
which can change the visual appearance of microplastics (Song et al., were effective at identifying microplastic particles and differentiating
2017; ter Halle et al., 2017; Veerasingam et al., 2020). from non-plastics but there were performance differences based on
While particle size was the most important factor, color, and shape particle size. FTIR could accurately identify polymer types for particles
also affected performance. White particles, that were most common in in size fractions above 20 μm whereas Raman did so for particles in size
smaller size fractions, were challenging to identify under the light of a fractions above 1 μm. This size issue is of concern for sampling drinking
microscope (Shim et al., 2017) when on a clear glass Petri dish (Michida water systems as they typically filter down to 20 μm, and the primary
et al., 2019) or when mounted on a white or an opaque filter surface. Of target is smaller particles. However, the data provided here quantifies
6
H. De Frond et al. Chemosphere 298 (2022) 134282
7
H. De Frond et al. Chemosphere 298 (2022) 134282
PlasticsEurope, 2020. Plastics The Facts. PlasticsEurope. https://plasticseurope.org Song, Y.K., Hong, S.H., Jang, M., Han, G.M., Jung, S.W., Shim, W.J., 2017. Combined
/knowledge-hub/plastics-the-facts-2020/. effects of UV exposure duration and mechanical abrasion on microplastic
Prata, J.C., Castro, J.L., da Costa, J.P., Duarte, A.C., Rocha-Santos, T., Cerqueira, M., fragmentation by polymer type. Environ. Sci. Technol. 51 (8), 4368–4376.
2020a. The importance of contamination control in airborne fibers and microplastic ter Halle, A., Ladirat, L., Martignac, M., Mingotaud, A.F., Boyron, O., Perez, E., 2017. To
sampling: experiences from indoor and outdoor air sampling in Aveiro, Portugal. what extent are microplastics from the open ocean weathered? Environ. Pollut. 227,
Mar. Pollut. Bull. 159, 111522. https://doi.org/10.1016/j.marpolbul.2020.111522. 167–174. https://doi.org/10.1016/j.envpol.2017.04.051.
Prata, J.C., da Costa, J.P., Lopes, I., Duarte, A.C., Rocha-Santos, T., 2020b. Van Mourik, L., Crum, S., Martinez-Frances, E., van Bavel, B., Leslie, H., de Boer, J.,
Environmental exposure to microplastics: an overview on possible human health Cofino, W., 2021. Results of WEPAL-QUASIMEME/NORMANs first global
effects. Sci. Total 702, 134455. https://doi.org/10.1016/j.scitotenv.2019.134455. interlaboratory study on microplastics reveal urgent need for harmonization. Sci.
Primpke, S., Christiansen, S.H., Cowger, W., Frond, H.D., Deshpande, A., Fischer, M., Total Environ. 772, 145071.
Holland, E.B., Meyns, M., O’Donnell, B.A., Ossmann, B.E., Pittroff, M., Sarau, G., Veerasingam, S., Ranjani, M., Venkatachalapathy, R., Bagaev, A., Mukhanov, V.,
Scholz-Böttcher, B.M., Wiggin, K.J., 2020. Critical assessment of analytical methods Litvinyuk, D., Mugilarasan, M., Gurumoorthi, K., Guganathan, L., Aboobacker, V.M.,
for the harmonized and cost-efficient analysis of microplastics. Appl. Spectrosc. 74 Vethamony, P., 2020. Contributions of Fourier transform infrared spectroscopy in
(9), 1012–1047. microplastic pollution research: a review. Crit. Rev. Environ. 1–63. https://doi.org/
Primpke, S., Lorenz, C., Rascher-Friesenhausen, R., Gerdts, G., 2017. An automated 10.1080/10643389.2020.1807450.
approach for microplastics analysis using focal plane array (FPA) FTIR microscopy Wyer, H., Polhemus, D., Moore, S., Weisberg, S.B., Coffin, S., Rochman, C.M., 2020. Steps
and image analysis. Anal. Methods 9 (9), 1499–1511. https://doi.org/10.1039/ scientists can take to inform aquatic microplastics management: a perspective
C6AY02476A. informed by the California experience. Appl. Spectrosc. 74 (9), 971–975.
Rist, S., Almroth, B.C., Hartmann, N.B., Karlsson, T.M., 2018. A critical perspective on Xu, J.-L., Thomas, K.V., Luo, Z., Gowen, A.A., 2019. FTIR and Raman imaging for
early communications concerning human health aspects of microplastics. Sci. Total microplastics analysis: State of the art, challenges and prospects. Trac. Trends Anal.
626, 720–726. Chem. 119, 115629.
Senathirajah, K., Attwood, S., Bhagwat, G., Carbery, M., Wilson, S., Palanisami, T., 2021. Zhu, X., Munno, K., Grbic, J., Werbowski, L.M., Bikker, J., Ho, A., Guo, E., Sedlak, M.,
Estimation of the mass of microplastics ingested – a pivotal first step towards human Sutton, R., Box, C., Lin, D., Gilbreath, A., Holleman, R.C., Fortin, M.-J., Rochman, C.,
health risk assessment. J. Hazard Mater. 404, 124004. https://doi.org/10.1016/j. 2021. Holistic assessment of microplastics and other anthropogenic microdebris in
jhazmat.2020.124004. an urban bay sheds light on their sources and fate. ACS ES&T Water 1 (6),
Shim, W.J., Hong, S.H., Eo, S.E., 2017. Identification methods in microplastic analysis: a 1401–1410. https://doi.org/10.1021/acsestwater.0c00292.
review. Anal. Methods 9 (9), 1384–1391. https://doi.org/10.1039/C6AY02558G.