3.1 The COD-based optimized Fenton Oxidation
First, the H2O2 dose was tested with theoretically required amount determined by COD and the multiples of that (1, 2, and 5 times for wastewater and 1 and 5 times for sludge samples) were tested by fixing the H2O2 to Fe (II) molar ratio (MR) to 10. Figure 2 shows that, while the COD removal efficiency increases when five times the theoretically required H2O2 dose was used for wastewater samples, it was not performing all that well for the sludge samples. Besides, the theoretical value was found sufficient for both wastewater and sludge. For this reason and to prevent the excessive chemical use, the theoretically required amount of H2O2 dose based on COD analysis was deemed to be satisfactory.
In Figure 3, the removal efficiencies for the Fenton process in WAS are shown along with the change in COD at various MRs of H2O2 to Fe (II). The H2O2 dosage in this experiment was set at the theoretically required level (by COD). At each MR, COD is found to decline over time as would be predicted; with the first 15 minutes showing the greatest decrease (Figure 3(b)). Then, a significantly slower rate of reduction is maintained. Higher molar ratios mean that the amount of Fe(II) used for the reaction gets lower. Fe(II) acts as a catalyst during the peroxide oxidation and speeds up the reaction. Results also indicate that the most efficient molar ratio is MR 10, which removes 90% of COD in 120 minutes. As seen in Figure 3(b), MR 10 also offers the best COD elimination efficiency in the quickest amount of time. Because MPs are subjected to chemicals and oxidative conditions during pretreatment, it is crucial to provide the shortest amount of time that results in high COD elimination. Additionally, Figure 3 demonstrates that for all of the MRs used, the majority of organic substances are degraded at 15 minutes. Despite the fact that the oxidation at MR 30 was generally slower because of the lower amount of Fe(II), performance nearly caught up at the conclusion of the two-hour timeframe. Even if higher removals can be seen at later times (90 or 120 minutes), the additional reaction time is deemed superfluous and prohibitive. So, 30 minutes was determined to be the ideal pretreatment period.
3.2 The optimized extraction and identification procedures
The two methods tested, i.e., single step salt extraction with NaCl only and double step salt extraction with NaCl and ZnCl2 together, yielded different extraction efficiencies. Figure 6 illustrates how the use of a two-step density-based separation exceeded the use of a single salt solution to capture MPs. There were two basic explanations given for this. First, after settling following the Fenton process, some MPs are entrapped in Fe(OH)3 flocs, which results in an underestimating of the MP concentration in the sample. When ZnCl2 is mixed with the settling flocs, the trapped particles become liberated and begin to float upward. As a result, the sample's floating fraction more accurately matched the MPs' content for following analyses. Furthermore, although being among the most commonly reported polymer types in sludge, MPs with high densities including PVC, PET, PMMA, PU, and PES could not float in low density environment achieved by the addition of NaCl. However, all of these polymers are captured in the sludge sample when ZnCl2 is added. Literature also recommends sequential extraction in NaI and ZnCl2 to separate both the low and high-density MPs and accurately determine their number despite the high price and relatively toxic properties of these two salts (Okoffo et al. 2019; Silva et al. 2018).
The optimization studies for the staining protocol resulted that n-hexane is the best one among the three tested carrier solvents due to its good performance in staining MPs without posing any damage to the PCTE filter. The PCTE filter was substantially damaged by acetone, and the dye precipitated as a result (Figure S2). However, due to Nile Red's poor solubility in n-hexane, the stock solution had to be prepared in acetone (Shim et al. 2016). Therefore, a 250 mg/L Nile Red stock solution was made in acetone and subsequently diluted to 10 mg/L in n-hexane. Furthermore, 200 µL of the dye was applied to a black PCTE filter and incubated at room temperature for 15 minutes in the dark to produce the brightest MP fluorescence.
Even though black PCTE provided good contrast facilitating MPs counting, its low-friction flat surface prevents large particles from being captured on the filter surface. This was especially critical for samples to be transferred to another laboratory for further analysis such as FTIR. To ensure capturing all the particles on the surface, filters (i.e., Glass Fiber Filter, GF/A; Cellulose Nitrate, CN; qualitative filter paper, Whatman No 1; and Mixed Cellulose Ester, MCE) with higher surface roughness were tested. Testing for compatibility with Nile Red resulted that all filters except for MCE retained the dye on their surfaces, which caused high background fluorescence hampering visualization of MPs under Fluorescence Microscope (Figure S3). So, MCE filter was integrated into the developed analysis method, which also surpasses the PCTE filter with its affordable price.
All the steps and parameters tested during the method development study are summarized in Figure 4. The optimized and selected parameters of the procedure are shown by green check marks, to show the typical analysis carried out for wastewater and sludge samples (Figure 4). If a wastewater is to be analyzed, then a 400 mL of wastewater sample was taken and used as is without any dilution or adjustment of COD. If a sludge sample is to be analyzed, sample was diluted to a COD value of around 3000 mg/L. For the validation step conducted with the use of organic matrix without suspended particles, KHP, a solution with a COD of about 930 mg/L was prepared. A 0.1 M Fe(II) solution prepared in H2SO4 was added into samples and the pH of the mixtures were adjusted to 3 using 5 M NaOH solution. The reaction was started by adding theoretically required amount of 35% H2O2 solution, yielding [H2O2]/[Fe(II)] = 10. The solution was mixed at 500 rpm at room temperature for 30 min. The reaction was terminated by neutralizing the pH using 5 M NaOH solution. The samples were subjected to two-step density-based separation process, sieved, and collected on filter papers as described in Section 2.2.1. Particles kept on filters were stained with 200 µL of Nile Red solution (10 mg/L) and incubated for 15 min in dark. Following incubation, filters were inspected under Fluorescence Microscope for counting and classification of shiny particles based on their size and shape. Quantification results were then corrected by a blank analysis using distilled water.
3.3 Results of recovery experiments
To check the effectiveness of the optimized method, known plastics at known concentrations were added into KHP (a simple organic matrix), wastewater and then to sludge (i.e., WAS which contains high suspended solids and high organic matter content). The model MPs extracted during recovery experiment are shown in Figure 5. The recovery efficiencies achieved in each scenario are given in Figure 6.
3.3.1 KHP as the sample matrix
The method was intended to be tested in variety of organic matrices in differing complexities. The simplest matrix having organic matter but no suspended solids was KHP. Its initial COD was 937±9 mg/L and 74.5% COD removal efficiency was achieved by applying our optimized Fenton Oxidation process. Table S1 and Figure 6(a) shows that the best MPs recovery results were obtained with the lowest density plastic (i.e., PE). It is seen that the effectiveness of the salts used to extract MPs were highly polymer dependent. For example, NaCl was satisfactory regardless of the size of PE MPs studied. As the density of plastics increased, then the performance of extraction and recovery by NaCl from KHP decreased. Although NaCl alone had been the major contributor at PE; both salts acted as effective extraction solutions. As the size of the PA particles increased, the recovery rate by NaCl increased and that of ZnCl2 decreased. Overall recovery ability of both salts was acceptable, but the higher particle size improved the recovery of PA. At the highest density plastic (i.e., PET), ZnCl2 became the major contributor to overall recovery and NaCl had the minimum contribution. For PET in KHP solution, the individual performance of NaCl was poor and adding ZnCl2 significantly improved the capture rate.
Overall, regardless of the polymer types added into KHP solution, there was a general declining trend in performance with decreasing particle size. Among all the plastics used, PET with the highest density had the lowest recovery in the smallest size. Relatively lower recovery efficiency obtained in KHP can be attributed to its lower initial COD content and hence the less chemical dosage requirement. The chemical dosage (H2O2) corresponding to the theoretically required amount caused a weaker floc formation compared to WAS with three times higher initial COD. Moreover, the absence of particulate matter in KHP, as the center of nuclei for the produced Fe(OH)3 flocs to grow, affected the overall performance, and hence the plastic recovery. It is evident that the presence of particles capturing the plastics before Fenton oxidation stage, might be helping to improve the overall recovery. So, the absence of particulate matter in KHP, made the sinking of higher density plastics like PET possible, and this made the salt solution flotation more difficult.
Still, recovery rates were at similar levels with Cashman et al. (2020) who typically obtained <70% recoveries for most plastics spiked into marine sediments tested with a number of different methods. However, general recoveries obtained in KHP were lower than that is recommended by (ASTM International, 2020b) which is suggested to be >80 %.
3.3.2 Wastewater as the sample matrix
The initial COD of wastewater was 665±10 mg/L; 72% COD removal efficiency was achieved by Fenton Oxidation. Similar to the case of KHP, the performance of NaCl was the best in recovery of PE MPs at the studied particle sizes (Table S2). Also, the integration of ZnCl2 was only needed for the smaller particle size of PE (i.e., 250 – 500 µm); altogether which made the effective and total recovery possible. For PA MPs, slightly different than KHP, significant performance was observed with NaCl alone especially at medium and high particle sizes, showing the importance of the extraction medium. For high particle size of PA, NaCl seemed to be enough; addition of ZnCl2 only mattered and increased recovery for the smallest particle size. The performance of NaCl was again poor for PET MPs, except for the smallest particle size. Oppositely, ZnCl2 added contributed more to the recovery of larger particle sizes.
Overall, in wastewater, the density and size of plastics did not play such a significant role as in the case of KHP; and the general recoveries were in the range of 85-95% for all plastics. This can be due to the effective Fe(OH)3 flocculation and settlement ascribed to the higher initial COD value of wastewater compared to the KHP solution. Moreover, the presence of particulate matter in wastewater possibly formed condensation nuclei and helped with further flocculation and hence the MPs recovery.
3.3.2 WAS as the sample matrix
The initial COD of the WAS sample was 3368±45 mg/L; 97% COD removal efficiency was achieved by Fenton Oxidation. As in line with the previous matrices, the performance of NaCl was satisfactory for the recovery of PE MPs at the studied particle sizes (Table S3). The performance of ZnCl2 was minimal and only helped at the marginal level with the larger particle size of PE this time, different than wastewater. For PA MPs, significant performance with only NaCl addition (close to 80%) was obtained. For the two higher particle sizes, ZnCl2 did not bring any improvement for recovery. Addition of ZnCl2 only mattered and increased the recovery for the smallest particle size of PA, similar to the result obtained for wastewater. In sludge, the performance of NaCl for PA was better than that in wastewater. This may be due to the high concentration of particulate matter – making the medium density inherently suitable for recovery. Generally, ZnCl2 added acted as the smaller contributor to the overall recovery of PA MPs and was more effective in smaller sizes. As in line with the other matrices worked with, ZnCl2 was the major contributor for the recovery of PET, due to its high density. The importance of ZnCl2 even increases as the size of PET increases as the particles become heavier.
In general, recovery efficiencies in sludge were comparable to KHP and wastewater. One can say that as the COD is removed highly, the analysis and visualization of MPs become easier and slightly higher recoveries are obtained. Even though at 80% or slightly lower, the lowest recoveries were experienced with the medium density plastic, PA. PA with density of 1.13-1.15 g/cm3, has the closest density to that of sludge among the three plastics studied. So, the recoveries were lower due to possible separation problems. Similar to wastewater, presence of particulate matter in sludge possibly formed condensation nuclei and helped with further flocculation and recovery. Different than the previous two matrices, the overall recoveries did not seem to be related to particle size, mostly at higher/much higher level than 80%. The theoretical amount of peroxide addition seems to be sufficient for effective recoveries, possibly due to the particulate matter in sludge may form a center of nuclei for the produced flocs to grow, affected the overall capture, and hence the plastic recovery. Furthermore, applying the analysis protocol developed resulted in around 97% of COD removal efficiency. All of these findings were considered appropriate in light of the literature. As a result, using this technique yields accurate results for both a clean sample (KHP) and complex matrices (wastewater and WAS). Two prominent techniques that were incorporated into the strategy are primarily responsible for this accomplishment. First, the improved Fenton Oxidation offers such powerful organic removal efficiency (i.e., COD elimination) that the matrix impact in MPs analysis is eliminated. Second, the two-step density-based separation procedure greatly increases the extraction efficiency of MPs (particularly those with high densities), as the sample is mixed twice on consecutive days, allowing MPs trapped in Fenton Oxidation residuals to float upward.
3.4 Carbonyl index (CI) and crystallinity of recovered microplastics
CI of the selected plastics as their virgin and pretreated states were calculated according to Benítez et al. (2013) for PE, Jakubowicz et al. (2012) for PET and Forsström & Terselius, (2000) for PA and given in Table S4.
CI values for PET in different conditions shows a decrease from 16.2 to 4.1, the lowest value with KHP medium whereas wastewater and sludge mediums show similar results around CI of 6. A similar trend is observed in the case of PA type MPs where the highest CI values is seen in the virgin particles. PE MPs, however, do not show any trend where KHP (lowest complexity medium) and has the lowest CI values while in sludge PE MPs has a minimally decreased CI value according to the virgin state. Highest difference in the CI values when compared with the virgin state is seen in PA MPs in KHP medium which showed around 60% reduction. A comparison in terms of CI shows that only PET type of MPs and PE type MPs in KHP medium show statistically significant (α = 0.05) changes after the recovery procedure as seen in Table S1.
Crystallinity of PET and PE MPs were determined according to methods defined in Hatinoglu &Sanin (2022) and Costa et al. (2001), respectively and results are given in Table S5. PA MPs as mentioned in Lee et al. (2008) cannot be characterized accurately for their crystallinity using only IR spectra so their crystallinity values were not calculated.
From Table S5, it can be seen that PET MPs in KHP medium shows statistically significant increase in crystallinity following pretreatment however crystallinity does not vary significantly in sludge and wastewater mediums. As for PE MPs, it is seen that crystallinity did not change significantly for KHP and sludge mediums, whereas crystallinity decreased considerably in wastewater medium. These changes should further be checked with other methods of crystallinity measurements such as Differential Scanning Calorimetry and confirmed.
Overall, some changes are seen in all mediums in terms of CI and crystallinity especially in CI for PET MPs, however these changes did not cause any problems in identification of said MPs in FTIR equipment. For the most affected MP type, PET, virgin MPs showed a 96.67% match in the library search whereas 69.84% match in KHP that showed up a with the major difference of new peaks arising 2350 cm-1 which is attributed to gaseous carbon dioxide that is not related to the measured sample. Differences in IR spectra of PET MPs in different mediums can be seen in Figure 7 as example spectra. Spectra of all conditions are given in Figure S5.
3.5 Quantification of microplastics in an unknown sludge sample with developed method
The developed method was applied to quantify unknown MPs in a WAS sample and then the counting results were corrected by a blank analysis. The suspected MPs were classified by number as fragment (59.9%), fiber (23.1%), and film (16.9%), which complies with literature findings (Figure 8 (a)). These particles were also grouped into different size classes by number: 38-106 µm (34.6%), 106-250 µm (32.6%), 250-500µm (31.4%), 500-1000 μm (0.8%), and 1000-5000 μm (0.5%) (Figure 8(b)). Smaller MPs in sludge dominated over larger ones (500-1000 μm), which can be attributed to that smaller ones are removed from wastewater at a higher rate by adsorption onto sludge in sedimentation tank (Liu et al. 2019; Magni et al. 2019; Mason et al. 2016; Talvitie et al. 2017).
Fluorescent particles of size >500 µm were subjected to FT-IR analysis and their polymer types are given in Figure 8 (c). The concentration of MPs in WAS determined was 887 MPs/g TS following blank correction (262 MPs/L for laboratory control sample) and 475 MPs/g TS following FT-IR correction (53.6% are plastics). The measured amount is close to the upper limit of the range, 510 - 495,000 MPs/kg TS, determined for sludges in literature (Hatinoglu and Sanin, 2021). The high concentration can be attributed to the source of wastewater, which originated from a metropolitan city with some industrial wastewater contributions to the plant.