Introduction

Microplastics (MPs) have spread throughout the world for a long time, but only now has their presence come to the public perception. They are defined as plastic debris smaller than 5 mm with different shapes, such as grain, sheet, and filiform (Law and Thompson 2014). MPs have garnered particular attention due to their persistence, mobility, and, most importantly, “cocktail” feature (Guzzetti et al. 2018). That is, they can adsorb various microorganisms, heavy metals, and organic matter, forming contaminant complexes, aggravating the uncertainty of environmental risk (Vethaak and Legler 2021). As an important part of natural environmental systems, the freshwater ecosystem also suffers from MP pollution. Recent evidence indicated the existence of MPs in worldwide lakes (Anderson et al. 2017; Zbyszewski and Corcoran 2011), rivers (Koelmans et al. 2019), ground waters (Mintenig et al. 2019), estuaries (Lima et al. 2014) and sediments (Lima et al. 2014). For example, the concentration of MPs in rivers ranged from 0.021 × 106 to 7.2 × 106 particles m−3 (Di and Wang 2018; Lahens et al. 2018; Liao et al. 2020; Zhang et al. 2017), while the one in lakes ranging from 0.01 × 106–6.8 × 106 particles km−2 (Su et al. 2016; Uurasjarvi et al. 2020). Furthermore, up to 102 particles kg−1 MPs were detected in sediment samples (Ta and Babel 2020; Vianello et al. 2013). MP pollution in freshwaters may pose ecological risks and threats to drinking water safety for modern society.

Along with the source water, MPs are transferred to urban engineered water systems, with the first step at drinking water treatment plants (DWTPs). To improve the removal of MPs and evaluate the potential threats of MPs to drinking water safety, the modifications on characteristics of MPs induced by various treatment processes require further investigation. Pre-treatments are applied at the beginning of DWTPs. The most common technics are pre-oxidation, e.g., chlorination, ozonation, and potassium permanganate (KMnO4) oxidation (Xie et al. 2016), which have been suggested as the assistant methods for DWTPs (20062011). These methods become the first stage to face MP pollutants, and most contain deconstructive chemical reactions, which could change the characteristics of MPs. Lin et al. (2022) reported that polystyrene (PS) tends to float on the water after chlorination, demonstrating a decreased sinking ratio. In our previous study, KMnO4 oxidation was proved to add nano-MnO2 onto MPs, improving the sedimentation of PS, polyethylene terephthalate (PET), and polyvinyl chloride (PVC), but not polyethylene (PE) (Chen et al. 2022). Compared to the pristine MPs, the sinking ratio of treated MPs (6.5 µm) increased by 30% (PET), 20% (PVC), and 30% (PS). Furthermore, the sorption of organic contaminants by MPs was also changed by heat-activated K2S2O8 (Liu et al. 2019) and radical-based UV (Lin et al. 2020). Thus, pre-oxidation could induce evident aging and destruction of MPs, affecting their behavior and removal efficiency in subsequent treatment processes.

In recent years, ferrate (FeVIO22−, Fe(VI)) oxidation, usually with potassium ferrate (K2FeO4), has received increasing attention due to its easily accessible and environmentally friendly nature in water treatment application (Sharma et al. 2015). Ferrate was explored for multipurpose actions, i.e., degradation of organic contaminants (Zhou and Jiang 2015) and inactivation of algae/cyanobacteria (Fan et al. 2018; Ma and Liu 2002). This relied on the highly oxidative activity of Fe(VI) with a redox potential of 2.20 V, which was higher than hypochlorite (1.48 V), ozone (2.08 V), and permanganate (1.68 V) under acidic conditions. Furthermore, the portfolios of ferrates also included coagulants, which relied on the flocs consisting of nascent Fe(III) oxides/hydroxides (Lv et al. 2018), enhancing the removal efficiency of particles in the next coagulation-sedimentation stage (Jun and Wei 2002). Most importantly, ferrate produces no mutagenic by-products (Xie et al. 2016), and even reduces the generation potential of chlorinated disinfection by-products (Jiang et al. 2016). These multipurpose actions endow ferrates with unique advantages in removing organic contaminants and particles from water. However, ferrate-induced modifications on MPs have not yet been realized. Ferrate pre-oxidation may induce multiple unknown interactions with MPs, resulting in unexpected behaviors of MPs in ensuing water treatment processes.

Herein, this study aims to evaluate the modifications to MPs by ferrate(VI) and explore the impacts on sorption and sinking in DWTPs. Investigating its intrinsic mechanisms would be significant for predicting the migration and transport patterns of MPs in the aquatic ecosystems. The lab-scale potassium ferrate oxidation was conducted, while four polymers, including PE, PET, PS, and PVC, were selected as targeted MPs. The surface morphology, chemical, and hydrophobic variations were investigated to determine the MP destruction induced by ferrate oxidation. Furthermore, ciprofloxacin (CIP) was selected as the targeted organic contaminant for the sorption performance since it has been detected in water supplies and found to be environmentally harmful (Hettithanthri et al. 2022). To further understand the behavior of MPs after pre-treatment, the modification of sinking patterns was evaluated.

Experimental

Materials

Four polymers, PE, PET, PS, and PVC, were acquired from Goodfellow Cambridge Ltd. Small MPs (< 1 mm) were detected in DWTPs (Novotna et al. 2019). Thus, MP particles with 6.5, 100, and 500 μm were prepared. Analytical-grade K2FeO4, methanol, and ethanol were obtained from Sinopharm. Ultrapure water was acquired from Fisher (USA). A source water from a DWTP was obtained as the background water matrix (Table S1).

Aging of MPs

To simulate the actual MPs in the natural environment, MPs used for experiments in the current study were all treated by artificial aging. Pristine MPs were sealed in a box with a customized glass cover that could be penetrated by ultraviolet light. The MP-containing box was placed outdoors under natural sunlight irradiation for 60 days.

Ferrate oxidation

Oxidation experiments were conducted in 250-mL glass beakers using vortex mixers (60 rpm, 25.0 ± 0.2 °C), and in phosphate buffer which exhibited little effect on the oxidation ability of Fe(VI) species (Huang et al. 2018). Before reactions, approximately 100 mL of buffered reaction solution was added to each beaker, and then, a number of MPs were added. The MP concentration applied was 1 mg L−1, which contains ~ 2.5 × 102– ~ 5.5 × 102 particles L−1 (6.5 μm). This is close to that detected in DWTP source waters (~ 102 particles L−1) (Novotna et al. 2019). A stock solution of Fe(VI) (typically 100 μM) was prepared immediately by solving K2FeO4 into 5 mM Na2HPO4 and 1 mM borate buffer (pH = ~ 9.1) (Yang et al. 2018). The ferrate oxidation was started by adding a given amount of the stock Fe(VI) solution (pH = ~ 9.1) into the reaction system with rapid mixing. Stock Fe(VI) solution was used within 10 min after being prepared. The initial Fe(VI) concentration was set to 10 mg L−1 (Zhou et al. 2014). After adding the stock Fe(VI) solution and the MPs, the pH of the solution was adjusted as needed. According to the Fe(VI) species distribution and pH function (Sharma et al. 2015), reaction solutions with pH at 3, 6, 8, and 11 were prepared using H2PO4, NaOH, and phosphate buffer. Sampling was conducted at a specific time, followed by filtration with 0.45-μm hydrophilic filters. The obtained MPs were washed with ultrapure water. Finally, the MP samples were separated from filters and dried under 40 °C overnight. All the experiments were performed in triplicate.

Analytical methods for MPs

Morphologies were detected by a scanning electron microscope (SEM). The chemical composition of MP surfaces was analyzed by X-ray photoelectron spectrometer (XPS), and micro-Fourier transform infrared spectrometer (FTIR) was employed to analyze (Text S1). MP density was determined using a PoreMaster33GT mercury porosimeter. To determine the mass of MPs, a method developed in our previous study was applied (Chen et al. 2022).

Two-dimensional correlation analysis

A two-dimensional correlation analysis (2D-COS) was applied for FTIR. The FTIR spectra were recorded in a range of 4000–400 cm−1. Since no apparent response variation was detected in the range of 3000–2000 cm−1, 2D COS was carried out using the FTIR spectra in the range of 2000–600 cm−1 over time. The analysis of 2D-COS followed a method developed by Lasch et al. using 2D Shige software (Lasch and Noda 2019). After calculation, 2D-COS synchronous and asynchronous maps were obtained to decode the evolution of organic functional groups on MP surfaces.

XPS analysis

XPS calculation involved charge correction, smoothing, non-linear Shirley-type background subtraction, and curve fitting. The C1s binding energies at 284.6 eV were used as the standard. A deconvolution was processed for XPS peaks using XPS PEAK with Lorentzian-Gaussian functions after subtraction of a Shirley background (Yamashita and Hayes 2006). Relative intensities (areas) of the sub-peaks were determined using Gaussian–Lorentzian functions embedded in XPS PEAK. As a result, the evolution of different components from an element can be investigated. The relative area ratios (AR) between different forms of Fe (such as Fe2O3 and FeOOH) are calculated by:

$$\mathrm{AR}={\mathrm{area}}_{1}/{\mathrm{area}}_{2}$$
(1)

where area1 and area2 are the peak areas of Fe2O3 and FeOOH, respectively.

Sorption experiment

Ciprofloxacin (C17H18FN3O3, MW = 331.4 g mol−1) was used for the sorption test. The sorption experiment was conducted using a procedure developed in our published study (Lin et al. 2020). The treated MPs were obtained in “Ferrate oxidation” section.

Sinking experiment

The sinking performance was evaluated based on a previously developed method (Lin et al. 2022). Specifically, MPs were treated by ferrate oxidation using the DWTP source water as background matrixes. Subsequently, MPs were separated by syringe-filter kits, washed, and dried at 40 °C. Approximately 100 mg of MPs was soaked in the source water without any chemical addition and stirred (30 min). And then, a 2-h stand was conducted. We separated the floating MPs (Mfloating, on the surface and suspending in the water) and the sinking MPs (Msinking, on the bottom). A customized method was applied to quantify the mass of these two-part MPs (Chen et al. 2022). The sinking ratio of MPs can be obtained:

$$\mathrm{Sinking}\;\mathrm{ratio}=\frac{M_{\mathrm{sinking}}}{M_{\mathrm{sinking}}+M_{\mathrm{floating}}}$$
(2)

Results and discussion

Morphology

All aged MPs presented smooth surfaces with complete and clear boundaries but no cracks and bumps (Fig. 1). After the ferrate oxidation, some surface morphological variations were observed. In the low-magnification SEM images, only a few differences in size and integrity were found. Notably, many nano-scale crystals were attached on all four MP surfaces, forming rough and uneven surfaces. Previous studies reported nano-scale crystal structure after ferrate oxidation (Prucek et al. 2013; Yang et al. 2018). By amplification, these surface crystals presented irregular shapes. As energy-dispersive spectrum (Fig. S1) showed, these attachments mainly contained iron and oxygen, indicating that they could be the nascent state ferric oxides (FexOx) adhering to the MP surface during the ferrate oxidation. Similar ferric oxides were reported in other studies (Kralchevska et al. 2016; Yang et al. 2018; Zheng et al. 2021). Of note, though the MP samples had been washed with ultrapure water after ferrate oxidation, these ferric oxides still existed, indicating their firm attachment on the MP surfaces. Thus, the MPs were transformed into MP-FexOx complexes. In addition, the smooth surfaces disappeared with obvious alternations, including cracks (Fig. 1d and f), wrinkles (Fig. 1h), and protuberances (Fig. 1b). It could be ascribed to surface destruction by ferrate oxidation.

Fig. 1
figure 1

Microscopical image of microplastics after ferrate oxidation. Experimental conditions: [K2FeO4]0 = 10 mg L−1, pH = 6, reaction time = 30 min, MP size = 200 μm. a, b Aged and oxidized polyethylene; c, d aged and oxidized polyethylene terephthalate; e, f aged and oxidized polystyrene, and g, h aged and oxidized polyvinyl chloride. White arrow indicates nascent state ferric oxides, and green arrow indicates destruction on surface

Full size image

Surface chemical variations

The ferrate oxidation facilitated a series of oxidized functional groups on the MP surfaces (Figs. 2 and S2). Generally, a prominent broad peak was observed in the range of 3000–3500 cm−1 for all MPs after ferrate oxidation. This is the characteristic peak of O–H stretch. FTIR data in time gradient were also obtained (Fig. 2). The broad peak (3000–3500 cm−1) intensity gradually increased with time, suggesting continuous oxidation. Some other characteristic peaks were observed, e.g., ~ 1000 cm−1 and 1400 cm−1 for PE, 1560–1640 cm−1 for PS, indicating C–O, O–H, and other oxidized functional groups.

Fig. 2
figure 2

Fourier transform infrared spectroscopy variations of microplastics. Experimental conditions: [K2FeO4]0 = 10 mg L−1, pH = 6, reaction time = 30 min, MP size = 200 μm

Full size image

Two-dimensional COS images (Fig. 3) were applied to further investigate the changes in FTIR characteristic peaks. For PE, two major autopeaks at ~ 1000 cm−1 and ~ 1400 cm−1 were observed in the synchronous map (Fig. 3a), indicating the existence of C–O and –COOH. Based on Noda’s rules (Jin et al. 2018), asynchronous correlation spectroscopy indicates the evaluation order of chemical bonds (Mao et al. 2020). In the asynchronous correlation spectrum, the characteristic peak at Ψ (1000, 1400) was positive (Fig. 3b), implying that the generation of C–O was earlier than –COOH.

Fig. 3
figure 3

Two-dimensional COS images. SYN indicates the synchronous map, and ASY indicates the asynchronous map. Experimental conditions: [K2FeO4]0 = 10 mg L−1, pH = 6, reaction time = 30 min, MP size = 200 μm

Full size image

In the synchronous map of PET (Fig. 3c and d), three major autopeaks at 1070 cm−1 (C–O stretching of primary alcohol), 1100 cm−1 (C–O stretching of secondary alcohol), and 1275 cm−1 (C–O stretching of alkyl aryl ether) were identified. All these peaks suggested that the PET surface was modified after ferrate oxidation. Based on the asynchronous correlation spectrum, the generation sequence was suggested to be 1070, 1100, and 1275 cm−1, indicating that the oxidized PET surface contained various unordered structures.

For PS and PVC, only major autopeaks at 1375 cm−1 and 1400 cm−1 were observed, respectively. Both of them were associated with C–OH functional groups. The FTIR data and 2D COS analysis generally indicated continuous surface destructions on MPs.

Chemical characteristics of MP-FexOx complexes

XPS (Fe spectrum) was obtained to confirm the chemical characteristics of the adhesive FexOx (Figs. 4, S3 and S4). The evolution of four MPs presented similar patterns. Taking PET as an analyzed target, the aged MPs presented no Fe signal. Yet, apparent Fe2p signals were observed after ferrate oxidation (Fig. 4a). Generally, the dominant Fe2p3/2 peak had binding energy in the range of 710.8–711.9 eV, suggesting the existence of iron oxides. To determine the valence state of iron oxides, the shape and binding energy of Fe2p3/2 satellite peak were investigated. Yamashita et al. (2008) found that the Fe2p3/2 satellite peak of Fe(II) presented a shoulder peak, while the one of Fe(III) was usually an independent peak in the range of 715–25 eV. In Fig. 4b, an independent satellite peak was observed for the MP-FexOx complexes. Thus, the dominant iron oxides on MPs were Fe(III) compounds.

Fig. 4
figure 4

X-ray photoelectron spectroscopy variations of polyethylene terephthalate. Experimental conditions: [K2FeO4]0 = 10 mg L−1, pH = 3, 6, 8, 11, reaction time = 30 min, MP size = 200 μm

Full size image

After pH 3 oxidation, the binding energy of Fe2p3/2 was ~ 711.9 eV (Fig. 4), which was associated with FeOOH (Fe(III), 710.9–711.9 eV). As the pH increased, the chemical shift of Fe2p gradually appeared. The binding energy of Fe2p3/2 shifted to ~ 710.8 eV after pH 11 ferrate oxidation, implying the existence of other forms of iron oxides, such as Fe2O3 (710.4–711.6 eV). To further determine the evolution of these two iron oxides, the characteristic peaks of Fe2p were deconvoluted by XPS peak analysis (Fig. 4b), and their peak area ratio (AR, Fe2O3 vs. FeOOH) was calculated (Table S2) (Biesinger et al. 2011). The AR value had a positive correlation with the pH value. For example, the AR of PE increased from 1.39 to 1.85 when pH rose from 3 to 11, suggesting that more Fe2O3 was generated under alkaline conditions than that under acid conditions. This also implied that the content of Fe2O3 in the reaction system gradually increased. In general, the Fe2p spectrum proved that FeOOH and Fe2O3 dominated the attached FexOx on the MP surfaces during ferrate oxidation.

The O1s spectra of aged PET presented slight signals, which may be due to oxidation during artificial aging. Their intensities increased after ferrate oxidation (Fig. 4c). Characteristic O1s peaks at 532.0 eV were observed under pH 3, indicating the existence of carbon–oxygen bonds, i.e., C–O and C = O (Moulder et al. 1992). A chemical shift (1–2 eV) to lower binding energy was observed as pH increased. All MPs presented similar tendencies, indicating that the O–Fe bonds with low-energy O1s spectrum gradually became dominant. This may be due to the chemical characteristic of ferrate oxidation under different pH values. Ferrate is highly oxidative under acidic conditions, resulting in more generation of oxygen containing groups on MPs. It tended to be weak as the pH increased. On the contrary, the generation of FexOx was enhanced as pH raised. Thus, the peak at 532 eV (O–C bond) shifted to the peak at 530 eV (O–Fe bond).

XPS can only detect the chemical variations on nanometer-level surfaces, while FTIR can go deep into several hundred micrometers. This is why their results were different. The surface of MP-FexOx complexes was dominated by nano-FexOx crystals, which XPS determined. Under these crystals, the entire surface of MPs was oxidized. Thus, FTIR reported the existence of C–O and C = O bonds. The transformation from MPs to MP-FexOx complexes could be accompanied by changes in sorption capacity and sinking performance.

Sorption of organic matter

Sorption of CIP by aged MPs followed the Freundlich model (R2 > 0.95, Fig. 5). Aged MPs only presented weak sorption capacities to CIP, with a Kf at 0.226 L g−1 (PET). After ferrate oxidation, the sorption of CIP by these MP-FexOx complexes was dramatically enhanced (Fig. 5, Table S3). The Kf of PE increased from 0.142 to 0.360 L g−1 after pH 6 ferrate oxidation, while Kf of PS rose from 0.206 to 1.062 L g−1. These significant improvements may be attributed to the attachment of FexOx nanoparticles on MP surfaces. These ferric oxides, including Fe2O3 and FeOOH, presented nano-scale and dispersed characteristics with abundant hydroxylated groups on the MP-FexOx surfaces, endowing them the complicated interactions with organic matter and other contaminants in water through hydrogen bonds and other chemical bonds. Yang et al. (2018) used ferrate to degrade p-arsanilic acid, and they found that the released As(V) can be removed through sorption by in itu formed FexOx during the reaction. Kralchevska et al. (2016) tested the removal of phosphate by ferrate. It was confirmed that the phosphates were removed from water by the sorption of FexOx nanoparticles. Herein, the enhanced sorption capacities implied that MP-FexOx had high potential as contaminant-enriched vectors. Furthermore, the ferrate pre-oxidation could also improve the removal of organic contaminants and MPs in water synchronously.

Fig. 5
figure 5

Sorption of ciprofloxacin on microplastics. Experimental conditions: [K2FeO4]0 = 10 mg L−1, pH = 6 and 8, reaction time = 30 min, MP size = 6.5 μm

Full size image

Sinking behavior

Polymer types, associated with different densities, dominated the sinking ratio of aged MPs (Fig. 6, Table S4). Aged PE tended to float since PE has a density of 0.92 g cm−3 close to water. The other three MPs had densities higher than water. Their sinking ratios raised as the densities increased, e.g., nearly all aged PS (1.04 g cm−3) maintained on the water surface, but > 50% of aged PVC (1.42 g cm−3) tended to sink (Fig. 6c). Particle size also affected the sinking. Taking aged PET as an example, its sinking ratio declined as the size decreased. Approximate 55% 500 μm PET sank on the bottom, but nearly all 6.5 μm PET floated on the surface. Because the smaller the MPs, the greater the surface tension in the water, which hinders the sinking of MPs.

Fig. 6
figure 6

Sinking ratio variations of microplastics. Experimental conditions: [K2FeO4]0 = 10 mg L−1, pH = 3, 6, 8, 11, reaction time = 30 min. Standard deviation was obtained from 3 independent samples

Full size image

The sinking behavior of MPs can be explained as a seesaw of buoyance, gravity, and MP-water interface force (surface tension). When the buoyance dominates, MPs float, such as PE and PS. On the contrary, MPs with high density, such as PVC, are dominated by gravity and sink easily. Furthermore, if the magnitude of buoyance and gravity is close here (e.g., PET), the contribution of MP-water interface force becomes important. The decreasing size of PET increases their MP-water interface force, resulting in the declining sinking ratio (Fig. 6a).

The sinking behaviors of MPs were changed by ferrate oxidation. Generally, small MPs were more susceptible to ferrate oxidation. The sinking ratios of aged PS were zero. After pH 11 ferrate oxidation, it increased to 90% for 6.5 μm PS, much higher than 10% for 500 μm PS. This could be due to the generation of MP-FexOx complexes, which increased their densities (Tables S4–S6), resulting in a rising gravity and a higher sinking ratio. For example, the density of 6.5 μm PS increased 40% after ferrate oxidation at pH 11, while it maintained the same for 500 μm one. Similar phenomena were also observed for PET and PVC.

The effect of ferric oxide attachment on the density of large MPs (500 μm and 100 μm) was slight. Thus, variation of MP-water interface force induced by ferrate oxidation became a dominant factor. Ferrate presented a stronger oxidative capacity under acid conditions (Sharma et al. 2015), with a more significant hydrophobicity decrease (Table S7). As a result, the MP-water interface force decreased more dramatically, and a higher sinking ratio was observed (Fig. 6). The sinking ratio of 100 μm PET treated by pH 3 ferrate oxidation increased ~ 40% but weakened as pH increased (Fig. 6a). However, the sinking ratio of 6.5 μm MPs increased as pH raised, suggesting the dominating role of density variation. Higher pH can generate more FexOx attachment (Yang et al. 2018), contributing to a more significant increase in density (Table S6). The gradual improvement of sinking ratios as pH increased was observed for PET, PS, and PVC, which all verified this hypothesis (Fig. 6).

Moreover, surface oxidation on MPs can enhance the MP hydrophilicity (Table S7). The contact angle of MPs declined from ~ 90° to ~ 20° after acid ferrate oxidation, suggesting that the surface of MPs was transformed from relatively hydrophobic to hydrophilic. These variations decreased the MP-water interface force of MPs, contributing to the sinking of MPs.

The sinking performance variation generally resulted from interactions between buoyance/gravity (density) and MP-water interface force (size and hydrophobicity) (Fig. S5). The small MPs showed a greater improvement in sinking after ferrate pre-oxidation, which will facilitate MP sedimentation and removal during subsequent sedimentation processes. It might also be crucial for drinking water treatment since the most persistent MPs in DWTPs were identified as small ones (most < 10 μm). Of note, PE presented high tolerance against ferrate oxidation.

Conclusion

After ferrate oxidation, MPs were transformed into MP-FexOx complexes, mainly Fe2O3 and FeOOH. The attachment of FexOx enhanced the sorption capacities of organic matter on all MPs. Furthermore, it promoted the sinking of PET, PS, and PVC. Small MP sinking performance was enhanced, e.g., the sinking ratio of 6.5 μm PS increased by 70% after pH 6 oxidation. This suggested that some organic contaminants, such as CIP, in water can be enriched onto MP-FexOx complexes and can be easily removed together by sedimentation after ferrate oxidation. However, PE with different sizes was persistently floated, which should be paid specific attention to.