Modifications to sorption and sinking capability of microplastics after chlorination

Microplastics (MPs) are generally defined as plastic debris with aerodynamic diameters smaller than 5 mm (Rachman 2018). Natural freshwater bodies, such as rivers (Yonkos et al. 2014; Mai et al. 2020; Stanton et al. 2020), lakes (Bharath et al. 2021; Dusaucy et al. 2021; Saarni et al. 2021), and polar glaciers (Ambrosini et al. 2019; Kelly et al. 2020), have been demonstrated to contain MPs, with the concentrations in the range of 10²–10³ particles L⁻¹ worldwide (Novotna et al. 2019; Triebskorn et al. 2019). These water bodies are important drinking water sources for modern society, and MPs pollution would potentially threaten drinking water safety as MPs entering drinking water treatment plants (DWTPs) cannot be eliminated (Koelmans et al. 2019; Zhang et al. 2020). For instance, finished water from three DWTPs in the Czech Republic contained up to 628 ± 28 particles L⁻¹ MPs, mostly <10 μm (Pivokonsky et al. 2018). A mean concentration of MPs at 400 ± 275 particles L⁻¹ was obtained in 38 tap water samples from different cities in China (Tong et al. 2020). MPs possess not only high-polymer properties but also enrich various trace organic pollutants, heavy metals, and even resistance genes, becoming transport vectors of contaminants (Mato et al. 2001; McCormick et al. 2014). To appropriately deal with the potential threats of MPs to drinking water safety, it is important to examine any modification to the morphology and chemical characteristics of MPs undergoing water treatment processes, from which the environmental behavior and risk of modified MPs can be better understood. However, such information remains scarce.

Conventional treatment processes in DWTPs include coagulation–flocculation, sedimentation, sand filtration, and chemical disinfection. The efficiency for removing MPs by the first three treatments may reach 50–99% (Velzeboer et al. 2014; Pivokonsky et al. 2018; Lapointe et al. 2020), but the contributions to modifying MPs were negligible (Rodriguez-Narvaez et al. 2021). Chemical disinfection is mostly destructive, with chlorination being most commonly employed in DWTPs, which is highly corrosive and oxidative against MPs. Chlorination is typically applied at the end of DWTPs, which lasts several seconds to minutes but kills almost all microorganisms in finished water (US Environmental Protection Agency 1999). Because finished water would often travel a long distance before reaching users, free chlorine (typically at 0.1 − 0.4 mg L⁻¹) (Ministry of Health 2007) is usually used to inhibit the regeneration of microorganisms (Zhang et al. 2019). Thus, residual MPs in the finished water are subject to both short-term and long-term chlorination disinfection (Tinker et al. 2009).

Only a few studies have been conducted on the modifications of MPs during chlorination. Kelkar et al. (2019) found that the surfaces of polypropylene, polystyrene (PS), and polyethylene (PE) were only slightly modified after regular dose (concentration–time (CT) = 75 mg·min L⁻¹) of chlorination in DWTPs, while high-dose (CT > 25 g·min⁻¹ L⁻¹) chlorination substantially degraded MP surfaces. Another study found that ultraviolet (UV) degradation of MPs could generate disinfection by-products during chlorination (Lee et al. 2020). Available data suggest that chlorination can induce modifications in certain MPs, which may alter the characteristics of MPs, such as sorption and sinking capabilities (Kelkar et al. 2019). These unknown modifications could change the risk of MPs to drinking water consumers. For instance, some floating MPs may tend to sink after chlorination and can be preserved in water supply systems. These ‘resident’ MPs may enhance the accumulation of trace contaminants in drinking water. Currently, the potential impact of chlorinated MPs on drinking water safety remains unclear.

To fill the aforementioned knowledge gap, we studied the effects of chlorination on MPs, selecting PE, polyethylene terephthalate (PET), PS, and polyvinyl chloride (PVC), which are abundant in DWTPs, as the model MPs. Before chlorination, MPs were treated by artificial aging. Low and high doses of chlorine were used, simulating short-term chlorination in DWTPs and long-term residual chlorine reaction in drinking water supply systems, respectively. The sorption and sinking capabilities of aged and chlorinated MPs were examined through an analysis of the modifications to the size and morphology, hydrophobicity, and chemical characteristics of MPs. The effects of MP polymer types and particle sizes were also assessed.

Four abundant MP polymers in DWTPs with two shapes made by the same manufacturing technique were acquired from Goodfellow Cambridge Limited (Novotna et al. 2019). Sheet-shaped MPs (thickness = 1.0–1.2 mm) were used for contact angle experiments, and particle-shaped MPs were employed for other experiments. Since small MPs (<10 μm) can penetrate traditional treatment processes within DWTPs (Novotna et al. 2019), the sizes of target MPs were set at 6.5, 200, and 500 μm. Analytical-grade sodium hypochlorite (NaClO) solution (available chlorine ≥ 10.0%) was obtained from Sinopharm (Beijing, China). Methanol and ethanol of high-performance liquid chromatography grade, as well as ultrapure water, were acquired from Fisher (Waltham, MA).

MPs found in the environmental water bodies are reported to be aged by various stresses, such as sunlight (most UV light in the range of 320–400 nm), water flow, wind, and biological effects. Thus, MPs applied in this study were all pretreated by an artificial weathering device. The aging condition was set at 45 °C, 50% relative humidity, and 0.83 W m⁻² UV irradiation at 340 nm for 60 days. These aged MPs were applied as treatment targets in the subsequent chlorination experiments. Modifications reported in this study are based on comparing aged MPs and chlorinated MPs.

The morphology of MPs was determined with a field emission scanning electron microscope (SEM; EVO18, Zeiss, Germany). A micro-Fourier transform infrared (FTIR) spectrometer (Nicolet In10, Thermo Scientific, Waltham, MA) was employed to analyze characteristic carbon–oxygen bonds (Text S1). The element composition and valence bonds were determined by an X-ray photoelectron spectrometer (XPS; K-Alpha, Thermo Scientific). Specific surface area (SSA) was identified by a Brunauer–Emmett–Teller adsorption method.

Microplastic particle size distribution (equivalent spherical diameter (ESD)) was determined using a laser scattering particle size distribution analyzer (LA-960S, HORIBA, Ltd., Japan). All MP particles were approximated as spheres. Subsequently, the ESD was used for the sphere volume formula calculation (⁠⁠; ⁠). Even though some overestimates would occur (Kowalski et al. 2016), they would pose almost no impact on the final conclusions drawn as the aim of the present study is to estimate the relative changes between aged and chlorinated MPs,.

The water contact angle of plastic sheets (100 mm × 50 mm; 5 mm thickness) was measured using an optical contact angle meter (SL200KB, KINO, Boston, MA), which was aimed to analyze the changes in hydrophilicity. Each sample group included 10 pieces of sheets, and each sample was analyzed in triplicates. Before analysis, each plastic sheet was installed horizontally on the platform. Ultrapure water was dropped vertically onto the surface of the plastic sheet by a microliter syringe, with an optical camera taking photos at 1.0 s intervals. After fitting the contact edge into a circle, the contact angle was calculated by the included angle between the tangent line and the horizontal line.

Such high-dose chlorination may not accurately mimic real-world scenarios. However, we believe that it is important to explore the potential effects of long-term exposure to residual chlorine on MPs, particularly in pipeline systems where water can be stagnant for extended periods.

Chlorination of MPs was conducted in 250 mL beakers. One hundred milliliters of the NaClO solution and a given amount of MP particles were added to each beaker. The beakers were oscillated at 120 rpm using vortex mixers. At given time points, the suspension collected was filtered with 0.45 μm hydrophilic polytetrafluoroethylene filter. The aqueous phase and MP particles were separated by filtration, and the particles were washed with ultrapure water and dried overnight at 40 °C before characterization.

The residual chlorine concentrations of the filtrate were determined using Hach powder pillows (Table S1). If necessary, Na₂SO₃ was used for dechlorination of residual chlorine based on residual chlorine concentrations (⁠⁠) (Sathasivan et al. 2017). For example, if 5 mg L⁻¹ (Cl) was added to the finished water matrix, there was still ∼ 4.86 mg L⁻¹ (Cl) of residual chlorine after 15 min (without any MPs addition). Thus, ∼18 mg L⁻¹ of Na₂SO₃ with the same volume of the reaction suspension was added to the water matrix. Water samples were further tested in subsequent sections.

A method in which MPs are filtered by syringe-filter kit was applied for the quantification of MPs based on our previous study (Chen et al. 2021). The total organic carbon (TOC) concentration of the solution was analyzed by a TOC-L analyzer (Shimadzu, Japan).

Sorption capacity was tested using ciprofloxacin (CIP) as the target analyte, following a similar procedure in our previous study (Lin et al. 2020). Both high- and low-dose chlorination tests were conducted. Only 6.5-μm MPs with four different polymer types were used.

Three hundred milliliters of the filtrate obtained in Section 2.4 and the surrogate standard were uniformly mixed in a separatory funnel, and 10 mL of dichloromethane (DCM) was added for extracting target compounds. After shaking for 10 min and standing for 5 min, the DCM was collected. During extraction, the gas in the separatory funnel was released every 2 min. Thirty milliliters of extract DCM was concentrated to 0.1 mL by nitrogen purging. n-Hexane was used to replace the solvent, and the volume was adjusted to 0.5 mL. Anhydrous sodium sulfate was added to remove water. All experiments were conducted in triplicate.

A gas chromatography–mass spectrometry (GC–MS) (Agilent 7890B/5977B) configured with a TG-5Ms capillary column (30 m, diameter 0.25 mm, 0.25 μm) was applied. One microliter extract was injected into GC for detection. A constant flow of helium (purity ≥ 99.999%) at 1.0 mL min⁻¹ was used as the carrier gas. The GC oven temperature was held at 50 °C for 3 min and then increased to 280 °C at 10 °C min⁻¹ (held for 3 min). The inlet, MS transfer line, and ion source temperatures were set at 250, 250, and 280 °C, respectively.

All experiments were conducted with both high- and low-dose chlorination. Because the low-dose treatment induced negligible effects on FTIR, XPS, and SEM, these results are not presented. The concentration of MPs employed in the present study was 1 mg L⁻¹, approximately 80–150 particles L⁻¹ for 6.5 μm MPs, which is close to that (up to 10² particles L⁻¹ (Pivokonsky et al. 2018; Tong et al. 2020; Shen et al. 2021)) in field finished water of DWTPs. The effects of chlorination on MPs are dominated by physical and chemical processes, which can be applied linearly with respect to different MP concentrations. Thus, the modifications to MPs by chlorination presented in the present study can provide insights into information for drinking water treatments.

A preliminary experiment was employed to explore the effectiveness of MPs quantitative method. One hundred milligrams of MPs was weighed accurately and filtered with the syringe-filter kit. After drying, the mass of retained MPs was obtained to calculate the recovery rate (Table S2). In addition, a group of syringe-filter kits was immersed in NaClO of 60 g L⁻¹ for 4 h, and the results confirmed that NaClO had negligible effects on the syringe-filter kit. The results confirmed the stable recovery rate and corrosion resistance of this method (Table S3). The free chlorine and total chlorine concentrations of the experimental water were detected using reagents (2105569 and 2105669; Hach) every day, and the free chlorine and total chlorine concentrations were adjusted according to the measurement results. The chlorinated MPs were washed three times with ultrapure water and dried in an oven at 40 °C to eliminate the effect of residual chlorine on the characterization results. Before the TOC experiment starts, the standard curve was determined to test the precision and accuracy of the instrument, and ensure the reliability of the experiment. In GC–MS analysis experiments, sample blanks and reagent blanks were tested to eliminate background errors. The accuracy and precision of the experiments were ensured by repeating all experiments three times and adding a surrogate standard (deuterated naphthalene).

The FTIR spectra (Figure S1) of four MPs (200 μm) induced by high-dose chlorination (CT = 9,600 mg min L⁻¹) showed that both polymer type and pH impacted the chemical modifications of MPs. Under acidic and neutral conditions, the main component in NaClO solution is HClO, which can be partly decomposed to generate oxygen (⁠⁠), but it was weak as pH increased (Gill et al. 1999). Nascent state oxygen is highly reactive and can oxidize MP surfaces. Mitroka et al. (2013) demonstrated that O₂ generated from chlorination could contribute to carbonyl formation on polymer and boost the degradation of plastics. Thus, both chlorination and oxidation can occur on MPs.

New absorption peaks at 1,560–1,640 and 3,300 cm⁻¹ appeared after chlorination of PE at pH 4 and 7, but no peak was observed at pH = 10 (Figure S1(a)). The surface of PE MPs was oxidized, and new carbonyl groups, such as C–O and C = O bonds, were generated, which are bonded with solvent water. This oxidized surface was enriched with crystal water, generating peaks at 1,560–1,640 cm⁻¹ (Gedde & Ifwarson 1990). It should be noted that FTIR of PET showed little change (Figure S1(b)), probably due to abundant O-containing groups on PET surfaces. Newly formed O-containing bonds were overwhelmed by the background signals.

The FTIR features of PS MPs were only slightly altered after alkaline chlorination, but an absorption peak at ∼1,700 cm⁻¹ was observed at pH 4 and 7 (Figure S1(c)), which is normally attributed to the stretching vibration of aromatic aldehyde or carboxylic acids (O–C = O bonds) (Mao et al. 2020). Herein, oxidation of MPs generated O–C = O, resulting in the peak at ∼1,700 cm⁻¹. Similar peaks were observed for PVC, but with a new absorption peak at 1,620 cm⁻¹ (typically C = O bond) after pH = 10 chlorination (Figure S1(d)). This is due to the higher electronegativity of Cl present on PVC compared to PS or PE, and it is more prone to electrophilic reactions with –OH groups, resulting in a higher number of –OH groups detected on the surface of PVC.

A peak at 3,300 cm⁻¹ was detected in pH 4 and 7 chlorinated PE, and a peak at 3,500 cm⁻¹ appeared on PS and PVC under pH 7. These peaks at 3,300 and 3,500 cm⁻¹ are related to hydroxyl (–OH) with different vibration modes, indicating different association structures. The typical free OH stretching frequency is at 3,648 cm⁻¹, and the intermolecular H-bonded peaks occupy the region of 3,100 − 3,600 cm⁻¹. Intramolecular H-bonded stretching frequency is around 3,612 cm⁻¹ and has a wider band (Brinkley & Gupta 1998). After chlorination, a new peak appeared in the spectrum of PE at 3,300 cm⁻¹, indicating the presence of hydroxyl groups (–OH) that had polymerized intermolecularly. This occurred because the polymerized monomer of PE has a relatively small molecular weight. Once the C–C skeleton was broken, alkyl radicals combined with other PE chains through intermolecular copolymerization to form new polymers, resulting in the appearance of these hydroxyl groups. For PS and PVC, a new peak appeared at 3,500 cm⁻¹, indicating the presence of hydroxyl groups (–OH) that had polymerized intramolecularly. This occurred because PS and PVC contain benzene rings, and the molecular weight of their polymerized monomers is larger. In addition, the C–C bond of PVC and the C–benzene ring bond of PS are difficult to break, so intramolecular polymerization mainly occurred during the free radical reaction process, with OH radicals forming new bonds within the molecule. Since this synthesis occurs only within the PVC or PS molecule, it is called intramolecular polymerization. These results implicate oxidation on the MP surfaces.

By comparison, aged PET already contains abundant C-O and C = O bonds. Chlorination appeared to decompose O-containing bonds on PET and decreased the O1s peak. Only weak Cl2p peaks appeared, suggesting fewer C–Cl bonds were generated, probably due to lower C-bond abundance in the PET skeleton (Reed et al. 2020; Zhang et al. 2021a, 2021b). The surfaces of PS and PVC MPs also contained a large number of O- and Cl-bonds after acidic and neutral chlorination, but the number decreased under alkaline conditions. Although PVC has abundant C–Cl bonds, the intensity of Cl2p peak increased substantially (approximately 10 times) after chlorination under neutral conditions. This indicates that chlorination generated abundant Cl-containing structures on PVC surfaces.

Of note, more significant destructions were confirmed after acidic chlorination. For example, many bubble-like protuberances appeared (Figure S5(a) and S5(c)) and can be attributed to destructed chemical bonds on the surfaces, resulting in the stripping of PE and PET surface layers. Distinct and dramatic corrosion was observed on PS and PVC (Figure S5(e) and S5 ), further corroborating the high destructiveness of chlorination after acidic chlorination arising from the strong oxidizability of HClO. Similar but weaker surface modifications were observed under alkaline conditions. In short, high-dose chlorination would induce degradation, corrosion, and damage on the surfaces of various MPs.

The SSA of four MPs all slightly increased after high-dose chlorination (Table S4). Like mass and volume, acidic chlorination resulted in the most significant increase of SSA compared to alkaline and neutral chlorination. The SSA of PET and PS increased by ∼8 and ∼10% after pH 4 chlorination. By comparison, weaker increases were confirmed for low-dose chlorination. Therefore, chlorination seemed to exert insignificant effects on the SSA of MPs.

It seems that pH is also an important factor for the variation of contact angles, e.g., the maximum contact angle of PE declined by 38% after pH 4 high-dose chlorination. Under alkaline conditions, only a slight decline (<5%) was observed. Similar results were obtained for PET, PS, and PVC. Generally, the variations in contact angle indicated that MP surfaces tended to be more hydrophilic, even after low-dose chlorination (CT at 75 mg min L⁻¹). Although the contact angle results with MP sheets may not be directly applied to MPs particles, they still can assist in better understanding the chlorination mechanisms of MPs. The modifications to hydrophobicity may affect the sorption and sinking capacities of MPs.

After chlorination, particularly with a high dose, the sorption capacity of CIP by all MPs was weakened (Table S6). The effects on PE were more significant (Figure 5(a) − 5(d)). Compared to aged PE, the K_f of chlorinated PE MPs at pH 4 and 7 decreased by ∼50%, but weaker effects were observed at pH 10. For PET and PS, similar patterns were observed, e.g., K_f declined by 38 and 21% at pH 4, respectively, but there were only limited effects for PVC. Low-dose chlorination (CT at 75 mg min L⁻¹) also posed impact on the sorption of CIP (Table S7), which were weaker than those induced by high-dose chlorination. This result suggests that chlorination disinfection can potentially impact the sorption capacities of MPs.

Modifications to surface chemical bonds and surface hydrophobicity appeared to be important for reworking MPs sorption. Chlorination resulted in the deconstruction of the orderly surface skeleton structures of polymers and the generation of disordered oxygen and chlorine bonds. These modifications increased the surface hydrophilicity of MPs and weakened the intermolecular interaction with CIP. The processes were affected by the pH conditions. Chlorination with NaClO is more corrosive and disruptive under acidic and neutral conditions. The FTIR (Figure S1), XPS (Figure 1) and contact angle (Figure 4) data are consistent with each other, i.e., the surface morphology and hydrophobicity of MPs surfaces were strongly modified under these conditions. All these modifications can be the reasons for the weakened sorption capacities of MPs. In general, chlorination weakened the sorption of CIP by MPs, reducing the potential of MPs as a vector for transporting contaminants.

High-dose chlorination (CT = 9,600 mg min L⁻¹) enhanced the sinking capability of PVC, PET, and PS. The sinking ratios of both 200 μm PET and PS increased up to ∼50 and ∼60% under acidic conditions (Figure 6(c) and 6(e)). However, the increase faded as pH increased. These findings were consistent with the variations in surface hydrophilicity (Figure 4). The contact angles of PET and PS decreased by 22 and 25%, respectively, after high-dose chlorination, and moderate oxidation and chlorination occurred on the MP surfaces (Figures 1 and S1). These surface modifications enhanced the affinity of MPs with water and reduced their surface tension, strengthening the sinking capability of MPs. Apparently, high-dose chlorination typically applied in water supply networks can generally accelerate the deposition of most MPs. Such a conclusion, however, was not applicable to PE and small-size PS.

Low-dose chlorination (CT = 75 mg min L⁻¹) increased the sinking ratios of PET, PS, and PVCs with weaker effects than that imposed by high-dose chlorination. Interestingly, the sinking ratios of 200 μm PET and PS still increased by 15 and 30% after pH 7 chlorination. However, no enhancement was observed for PE and 6.5 μm PS. This implies that modifications to surface chemical characteristics by chlorination, simulated with actual doses in DWTPs, would enhance the sinking capabilities of ‘heavy’ MPs, such as PET and PS. In DWTPs, if the MPs tend to sink immediately after the actual low-dose chlorination, they may be trapped in the clean water basin of DWTPs. This would reduce their risk to the subsequent drinking water network. On the contrary, low-density PE MPs remained largely intact. PE MPs have a high abundance in DWTPs, even in the finished water (Pivokonsky et al. 2018; Tong et al. 2020). After chlorination disinfection at the end of DWTPs, the PE MPs continued to float, and still entered the water supply networks. Thus, its potential risk to drinking water safety should be paid special attention.

After chlorination, the TOC concentrations in the suspensions containing four MPs increased. TOC concentration of PE was the lowest, followed by PVC and PS, and PET was the highest (Figure S6). The TOC yield from PE reached 0.77 μg mg⁻¹, while PET was 1.19 μg mg⁻¹. The TOC from the control sample (containing no MPs, only NaClO solution) was only 0.03 μg mg⁻¹. This indicated that MPs released a certain amount of dissolved organic matter into the water during the chlorination reaction.

After a full scan by MS, the products released from MPs during the chlorination were identified. MPs of various polymer types had specific degradation products. There were five, four, three, and four kinds of products detected from PE, PS, PET, and PVC, respectively. The types and structures of products were related to their parent MP precursors. The degradation products from PE contained long-chain alkanes and alkenes, such as 1-decene (872-05-9), decane (124-18-5), and oxidized products, such as decanal (112-31-2) and 1-dodecanol (112-53-8). The skeleton of PE is a long-chain C–C structure. Thus, it is speculated that these products were the fragment products generated after the PE skeleton was corroded and oxidized during the chlorination reaction. PS and PET released a variety of aromatic organic compounds containing benzene structure, such as benzaldehyde (100-52-7) and cumene (98-82-8). PVC released a variety of products containing chlorine or methyl branches, such as 3-(chloromethyl)heptane (123-04-6) and 4-methyl-1-undecene (74630-39-0). These products were related to the backbone structure of their parent MPs, which was consistent with our speculation that the chlorination reaction induced break bonds on MPs, generating various fragmented products.

Chlorination was carried out under neutral conditions. The active species from NaClO in water can be considered as HClO and ClO⁻, both of which are electrophilic reagents. MPs applied in this study contain C–C frameworks and present electrophilic surface properties. Thus, the overall interaction between HClO/ClO⁻ and MPs could be weak. Nevertheless, HClO/ClO⁻ still induced MPs to produce various chlorinated products, such as 1-chlorodecane (1002-69-3) from PE and 4-chlorostyrene (1073-67-2) from PS.

In addition, the ClO⁻ can be transformed into nascent state oxygen. In neutral water, ClO⁻ is relatively unstable. It quickly decomposes and generates nascent state oxygen, which is a strong oxidant. The nascent state oxygen can attack MPs, breaking chemical bonds in its skeleton and fragmentation and generating various organic products. The presence of oxidation products, such as decanal (112-31-2) from PE, benzaldehyde (100-52-7) from PS, and hexanoyl chloride (142-61-0) from PVC, also confirmed the oxidation.

The present study confirmed that chlorination disinfection induced surface corrosion, blistering and cracks of MPs, as well as modifications to size, mass, SSA, and hydrophobicity of MPs. Low-dose chlorination generally contributed limited effects, while high-dose chlorination was more destructive. The sorption capacity of CIP by PET, PS, and PVC MPs was weakened after high-dose chlorination. The treated MP particles also tended to sink after chlorination. Furthermore, chlorination of MPs generated various products, which were the degraded fragments from the MP skeleton. Generally, chlorination posed negligible effects on the sinking behavior of PE, but it still weakened its sorption capacity, consequently reducing its potential as transport vectors of organic contaminants.

This project was supported by the National Natural Science Foundation of China (Grant No. 51778270), National Natural Science Foundation of China (Grant No. 21936004), and Guangzhou Basic and Applied Basic Research Foundation (Grant No. 202102080273).

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.