The material of the blankets was identified by FTIR as polyethylene terephthalate (PET) polyester. The fabric of the deep-red fleece blanket consisted of a ground textile weft-knitted fabric made from texturized, delustered polyester multifilament yarn with a fine filament titer of approximately 2.5 dtex (diameter approx. 15 μm). The loop piles for the double-sided plush consisted of texturized PET microfiber (fiber diameter 10 μm—approx. 1 dtex) multifilament yarn with at least 200 filaments, which was cut or raised to a loop height of approximately 10 mm. The structure is shown in Supp. Inf. Fig. S4. A SEM micrograph of the microfiber is shown in Fig. 1.
Three series of washing experiments were carried out: (1) without additions, (2) with detergent, and (3) with detergent and fabric softener. In each series, two blankets were separately washed and dried 10 times (Supp. Inf. Fig. S5).
Plots in Fig. 2 (Table 1) show the average relative release of fibers. Results of the first washing varied the most, 0.008–0.021 wt% fibers released, which is attributed to differences between the as-purchased blankets. The differences quickly decreased during subsequent washing cycles. The two parallel experiments done in each series were generally in good agreement (detailed results Supp. Inf. Table S1, Fig. S6). Fiber release stabilized during the last few washing experiments. The average values from cycles 8, 9, and 10 were taken as an estimate for a stable release expected on a long-term basis: 0.00108 wt% (no additives), 0.00140 wt% (detergent), and 0.00124 wt% (detergent + softener). These certain, however low, differences indicate that additives are not a main factor in fiber release but rather a mechanical stress. The average of all three series is 0.00127 wt%. By using a rougher filter than the paper filter used by Browne et al. (2011), we prevented clogging and were able to evaluate the effect of washing additives.
The wastewater after filtering contained no visible fibers. To verify the efficiency of a filter with relatively large openings (200 μm) relative to fiber thickness (10 μm), we further filtered two samples of effluent water using a paper filter with 2–3-μm pores. The quantity of collected fiber fragments was very low (we estimate a maximum of several % of the released fibers), but we were not able to quantify it due to large volumes and clogging problems. The fragments were mainly in the 20–200 μm range (Supp. Inf. Fig. S7) with very few long fibers (max. approx. 700 μm). These results show that installation (and maintenance) of a relatively simple and robust filter could prevent most of the emissions.
To verify our results, we washed a five-year-old PET fleece jacket. Microfiber release was 0.00111 wt% (no additive), 0.00123 wt% (detergent), and 0.00136 wt% (detergent + softener) giving an average of 0.00123 wt%. This result is in good agreement with our experimental estimate for long-term release obtained with the blankets, confirming it as an acceptable long-term release value.
We attempted to correlate the mass of the fibers released to their number, however found it impossible to separate the intertwined fibers (Supp. Inf. Fig. S8). We were however able to disentangle some fibers; the range of lengths was 0.3–25.0 mm, with an average length of 5.3 mm; however, the average length may be underestimated due to the particular difficulty of disentangling long fibers. A number of released fibers were very lengthy (up to 25 mm) even after several washings, which is most likely a function of the plush fabric. These lengths indicate disentanglement of full-length fibers covering piles on both sides of the fabric (approx. 10 mm each) and the part fixed in the ground textile. Among all inspected fibers, only one filament from the ground textile (2.5 dtex) was observed while all others originated from pile fibers (1 dtex). Using the average fiber length (5.3 mm), a 1-dtex diameter (1 g per 10.000 m), and the 0.0012 wt% release, we can calculate that a 500-g piece of fabric will release 6 mg of fibers or 11.3 × 103 fibers (although this number may be overestimated due to a likely underestimation of the average fiber length).
Although fibers released during spin-drying are not released into the wastewater, we monitored the quantities (Table 1, Supp. Inf. Table S1). Release of fibers during drying was in all cases higher by an approximate factor of 3.5 than the release during washing. The plot of average releases in Fig. 2 indicates that the long-term release value was not yet reached since values continue trending lower. This assumption was confirmed by the results obtained with the old fleece garment, which gave releases of 0.00111, 0.00103, and 0.00123 wt% during the three drying cycles. The average of 0.00112 wt% is significantly lower than the average 0.00394 wt% we obtained with the blanket.
Our results confirm findings of previous studies indicating fiber release during washing and support the numerous reports of synthetic fibers found in natural marine and freshwater habitats (Thompson et al. 2004; Klein et al. 2015; Gallagher et al. 2015) as well as in organisms (Rochman et al. 2015). The estimated number of fibers released in our experiments (even when taking into account a possible overestimation) is significantly higher than the 1900 fibers per garment/washing reported by Browne et al. (2011) but is much lower than predicted by Bruce et al. (2016)—up to 250 × 103 fibers per garment/washing. It is however significant that we used the 1-dtex mass/fiber-length value which we consider appropriate as opposed to Sundt et al. (2014) and Bruce and al. (2016) using the 300-dtex value. Our weight percent release ratios are significantly lower than the 0.039 wt% reported by Dubaish and Liebezeit (2013) and the very broad range (0.007–0.216 wt%) reported by Bruce et al. (2016) for new garments and front-loading washing machines. Bruce et al. (2016) used two filters (333 and 20 μm) which can attribute for only part of the difference as the smaller mesh collected a minor part of the total fiber release. Partially, the very mild washing conditions used in our experiments probably lead to a conservative fiber release estimate. However, the much longer fibers released in our experiments, 5.3 mm compared to 0.7 mm by Bruce et al. (2016) who used fleece jackets, strongly support the conclusions of Petersson and Roslund (2015) who concluded that fabric structure is the most important factor influencing fiber release. As we are still collecting the first sets of fiber release data, we will need to establish in more detail the influence of washing conditions (e.g., temperature, duration, load size) and fabric properties (fiber type and material, fabric structure, etc.) in order to come to more reliable estimates of the quantitative extent of this type of pollution.
A key result of this study is the indication that fibers are emitted throughout the lifetime of the garment. The importance of our estimated 0.0012 wt% of loose fibers released into the wastewater during each washing lies in the cumulative effects. We performed a rough assessment of emissions for a northern climate with the following assumptions: each resident has one polyester blanket (small size 350 g) washed four times a year and one fleece jacket (500 g), washed eight times a year. Based on our results (0.0012 wt% loss per washing), the mass of released fibers corresponds to 4.5 mg for the blanket and 6.5 mg for the jacket resulting in 70 mg of microfibers released annually per person. For Slovenia with just over 2 million inhabitants, this leads to emissions of approximately 144 kg a year. We believe these are conservative estimates since no new items (with an initially higher release) were considered, and the average person is most likely to own more items made of synthetic fibers (sports clothing, gloves, caps, pet items, etc.). Considering a material density of 1.38 g/cm3 and a fiber diameter of 10 μm, we can calculate that this quantity has a surface of 41,700 m2. Although it was already shown that the majority of fibers released during washing is removed in wastewater treatment plants where these are used (Talvitie et al. 2015; Mintenig et al. 2014) and despite PET absorbing lower quantities of POPs than polyolefins (Wright et al. 2013), we nevertheless believe that the large specific surface of microfibers qualifies this form of microplastics as an important class with a notable contribution to the overall problem of pollutants carried by microplastics (Rios et al. 2007).