Sampling site and sample collection
In total, there are five conventional DWTPs in Tehran with the same system of water purification, including screening, coagulation and flocculation, sand filtration and disinfection that three of them are fed by different rivers, so these three DWTPs were chosen to be investigated. In this study, the names of DWTPs were not mentioned for confidentiality reasons, so they are named DWTP 1 to 3. The DWTP number 1 is fed from Karaj and Kan River in the northwest of Tehran with the capacity of 7.2 m3 s-1, the source of the second DWTP is the Jajrud river in the northeast of Tehran with a capacity of 4 m3 s-1 and the third DWTP receives water from Lar Dam in the northeast of Tehran with the capacity of 5.7 m3 s-1. For the sampling, dark glass bottles with a capacity of 2.5 L were used to collect water samples from the raw and treated water from the DWTPs. Sampling was conducted in three different time intervals, over five months from April to September, from 12 am to 1 pm. Each time, one sample from raw water and one sample from treated water were taken (18 samples comprising 45 L in total). Subsequently, collected samples were kept in the dark at 4 ºC before sample preparation.
Sample preparation
To prepare the samples, the Wet Peroxide Oxidation method (NOAA, 2015) was applied to digest microorganisms. To explain briefly, a 0.05 M Fe (II) solution was prepared by 15 g FeSO4.H2O and 6 mL concentrated sulfuric acid (Merck Millipore, USA) and one L of deionized water. Every L of samples was added 80 mL of Fe (II) solution and 80 mL 35% hydrogen peroxide (Merck Millipore, USA) (Wang et al. 2020; Pivokonsky et al. 2018; Anderson et al. 2017; Masura et al. 2015). Afterward, One L Erlenmeyer flasks containing samples were placed on a stirring plate at 60 ºC at 300 rpm for 30 minutes to boost the digestion. Subsequently, samples were kept at room temperature for 24 h before filtration. Then they were filtered via a vacuum pump through cellulose nitrate membrane filters with a pore size of 0.2 µm and 47 mm diameter. To remove clay and other inorganic particles from the filters, a density separation was applied. Hence, a zinc chloride solution (Merck Millipore, 5 M, 1.55 g cm-3) (Yang et al. 2019; Mintenig et al. 2019; Quinn et al. 2017) was added to 20 mL centrifuge tubes and the filters were placed into the solution. Subsequently, the tubes were treated with an ultrasonic bath for 10 minutes to detach the particles from the filter. Without removing the filters, the tubes were centrifuged at 4000 rpm for 5 minutes and supernatants were filtered again to have almost pure MPs. Afterward, filters were placed in Petri dishes and dried in an oven at 60 ºC for 1 h. Afterward, the Petri dishes containing the samples were covered by aluminum foil and placed in a desiccator for further quantitative and qualitative analysis.
Quality assurance/Quality control (QA/QC)
Cotton laboratory outfit and nitrile gloves were utilized to minimize the risk of pollution. A negative-pressure ventilation system was functioning during the sample processing to eliminate the risk of depositing airborne MPs on the filters. Working surfaces on which the experiments were conducted were repeatedly cleaned with 1 M NaOH (Merck Millipore, USA). Furthermore, all the glassware for the sample processing were rinsed three times with filtered deionized water to remove potential MPs on their surfaces. For the sampling, dark glass bottles were used to lower the effect of photo-degradation. Plastic bottles were abstained from using in order to minimize the risk of the addition of MPs from the bottles. To minimize sample pollution, a layer of aluminum foil was placed between the bottles and screw caps. Additionally, one blank sample was carried out to ensure if there was sample contamination. The preparation method was applied on one L of previously filtered deionized water.
Quantitative analysis
A scanning electron microscope (Thermo Fisher Scientific, FEI Quanta 200, USA) with an accelerating voltage of 30 kV and detector working distance of 10 mm was used to image the filter surface in order to enumerate and identify the shape and size of MPs. The filters were cut in half and three cut-outs (3 mm ×8 mm) from one of the halves (one in the center, one in the edge and one in the middle) were scanned by SEM (Pivokonsky et al. 2018). Before imaging, a gold layer was sputtered onto the samples to create electrical conductivity. Approximately 60 images from each cut-out were taken (13 mm2 in total) and the number, size and shape of detected MPs were extrapolated to the whole area of the filter. The number, size and shape (fiber, fragment and sphere) of MPs were verified by ImageJ software (Version 1.50e, National Institute of Health, USA) based on one-L samples. Fibers were verified as thin and long MPs and were measured by their thickness, while fragments are particles that have been created by breaking down of larger pieces of plastic via degradation that were measured by their longest end and spherical MPs have an appearance of a sphere. The particles were divided into 5 categories in terms of their size (1-5 μm; 5-10 μm; 10-50 μm; 50-100 μm; >100 μm).
Qualitative analysis
To identify the chemical properties of MPs, a previously cleaned needle was used to transfer MPs on the another half of the filter onto a conductive adhesive copper tape using a light microscope (N-120, Hinotek, China). In total, 107 suspected MP particles with various shapes and sizes were carefully transferred onto the copper tape and sent for analysis. Since Fourier Transform Infrared (FTIR) spectroscopy is used for detecting the spectrum of particles greater than 500 µm and µ-FTIR is utilized for the particles down to the size of 20 µm (Li et al. 2018) and the fact that the majority of the MPs in this study were comprised of particles smaller than 20 µm, µ-Raman spectroscopy (Horiba Scientific, XploRA ONETM, Japan) was used to detect the chemical composition of MPs by Labspecs 6 Software (Horiba Scientific, Japan) and the obtained spectra were compared with Infrared and Raman Users Group (IRUG). The frequency of excitation laser of micro-Raman spectroscopy was 785 nm, 1 accumulation and 100× objective at 10 mW. The grating was 1200 1 mm-1, instrument aperture 100-μm slit, acquisition time 2 s and spectral range was set to 500-3500 cm-1. In qualitative analysis, 62% of suspected particles were MPs, so this amount was subtracted from the MP numbers in quantitative analysis.
Statistical analysis
All statistical analyses were computed using Statistical Package for Social Science (version 16.0, SPSS, Inc.) and the figures were created with Microsoft Excel 2016 for Windows. Prior to statistical analysis, all data were tested for the basic assumptions for normality and homogeneity of variance. The Kolmogorov-Smirnov test was applied to analyze the normality of the data distribution. P < 0.05 was considered statistically significant (Tables S1-S3, supplementary data).