Naturally occurring nanoparticles in two biological wastewater treatment plants in Southern California

Naturally occurring nanoparticles (NONPs) in wastewater are generally considered colloids, but their production and size distribution are not well understood. NONPs are more abundant than engineered nanoparticles (ENPs) in wastewater, where they may cause membrane fouling, harbour pathogens, and transport contaminants to the environment. In this study we aimed to understand the seasonal behaviour as well as the quantity and size of suspended particles (both unltered and ltered through a 450 nm lter) along two water resource recovery facilities (WRRFs, formerly wastewater treatment plants) located in Southern California. We also investigated the factors that correlate with their occurrence. We found that both of the biological secondary treatments investigated, conventional activated sludge process (ASP) and trickling lter (TF), were more ecient in removing suspended particles larger than 450 nm than they were smaller ones. The results show that current treatment processes are not designed to remove nano-sized particles eciently. Additionally, there was a signicant and direct correlation between inuent dissolved chemical oxygen demand (COD) and the abundance of suspended particles both larger and smaller than 450 nm, suggesting that the suspended particles increased with dissolved COD in the WRRFs and thus were biogenically generated during the wastewater treatment.


Introduction
Naturally occurring nanoparticles (NONPs) are found in the environment in concentrations orders of magnitude higher relative to engineered nanoparticles (ENPs). ( Prestel, et al. 2005) this may be due to the fact that the de nitions for colloids and nanomaterials overlap, and there is no established distinction between the two terms. Colloids are de ned by some as small particles that are suspended and do not stabilize well, with sizes ranging from 1 nm to 1 µm. (Lead andWilkinson 2006, Mohlman andPearse 1922) In contrast, nanomaterial is broadly de ned by the US Environmental Protection Agency (US EPA) and the European Commission (EC) as material between 1 and 100 nm in at least one dimension, (EC 2011, USEPA 2021) with recognition that there are naturally occurring nanoparticles. As a result, "colloids" and "nanoparticles" were used interchangeably in past research. In this study, we used 0.45-µm lter and the Zetasizer Nano ZS to aid our understanding of particle distribution in the wastewater, and refer to naturally occurring particles of size range 1 to 100 nm as NONPs.
Nanoparticles and colloids, depending on their physical, chemical, and biological properties, can have adverse effects on wastewater treatment and the quality of treated e uent. (Dean 1969, Dean, et al. 1967 addition, colloids can cause light scattering resulting in inadequate disinfection by UV and also impair the aesthetic quality of the treated e uent. (Christensen and Linden 2003, Gilboa and Friedler 2008, Hassen et al. 2000, Passantino et al. 2004) During chlorination speci cally, colloidal particles may harbour and shield pathogens, (Winward et al. 2008) and if they are organic colloidal particles, they may also increase the chlorine demand of the treated e uent. (De Beer et al. 1994, Falsanisi et al. 2008, Gautheir et al. 1999) Furthermore, smaller particles have high speci c surface area which provides a higher number of potential adsorption sites for many contaminants; (Kowalkowski 2010, Rudolfs andZuber 1953) if the colloids are not removed during the treatment process, they could enhance the transport of contaminants to the environment. Depending on the wastewater composition and the coating on the ENPs, previous studies have shown that a substantial amount of ENPs may remain in the treated wastewater e uent. (Jarvie et al. 2009, Limbach et al. 2008 Previous studies have also shown that organic compounds, such as natural and dissolved organic matter (DOM), are able to stabilize a range of different nanoparticles, both NONPs and ENPs, in laboratory setting or environmental conditions. (Aiken, et al. 2011, Domingos et al. 2009, Mosley and Hunter 2003, Navarro et al. 2009) Therefore, the stabilization of discharged nanoparticles in the environment could aid their transport in the receiving water body upon their release. With rising concerns about contaminant transport on nanoparticles and the impact of nanoparticles on wastewater treatment processes and on the ecosystem, understanding the production and removal of particles of different sizes in water resource recovery facilities (WRRFs, formerly wastewater treatment plants) will thus improve our current knowledge of nanoparticles behaviour and contaminant transport.
Due to the dearth of literature on the production and behaviour of particles of different sizes in WRRFs, this study aims to understand the prevalence and size of particles along the wastewater treatment process and comparatively analyse them with the water quality parameters and activated sludge EPS.

Sampling sites
We investigated the fate and transport of particles at two Southern California WRRFs. Plant 1 had an equalization basin operated with the modi ed Ludzack-Ettinger (MLE) con guration, with an internal recirculation from the end of the activated sludge process (ASP) to the beginning of the ASP (Fig. 1). Full denitri cation was achieved through methanol addition at this plant, which treated approximately 6•10 4 m 3 d − 1 (~ 16 MGD) and had a yearly-average mean cell retention time (MCRT) of 8.5 d as well as an average hydraulic retention time (HRT) of 6 h. Following tertiary treatment using a sand ltration and chlorination, most of the ow was reclaimed. There were several points during the treatment process where chemicals were added in Plant 1: the primary clari er had a ferric dosage of 26 mg l − 1 , and when the ow was more than 5.7•10 4 m 3 d − 1 (~ 12.5 MGD), an additional 0.2 mg l − 1 of polymer was added. Here 8 mg l − 1 of methanol were added at the beginning of ASP into the anoxic denitri cation zone, the return activated sludge (RAS) was fed magnesium chloride (MgCl 2 ) at 20 mg l − 1 to balance the pH, and alum was added to the secondary e uent after the secondary clari er and before tertiary granular ltration. Plant 2 did not have an equalization basin, but it had a Ludzack-Ettinger (LE) layout (identical to Plant 1 but without internal recirculation) and performed partial denitri cation and nitri cation. No supplemental carbon was added in this process. The plant capacity was approximately 2.3•10 4 m 3 d − 1 (~ 6.1 MGD) and had a yearly-average MCRT of 3.3 d as well as an average HRT of 4.4 h (Fig. 1). Approximately 1.9•10 4 m 3 d − 1 (~ 5 MGD) from the primary e uent went to the ASP and the rest went to trickling lters (TF). The stream that went through the TF would go through the TF secondary clari er and would be discharged to the ocean, and the stream from the ASP was reclaimed after being ltered through a mesh cloth. There was alum addition in the secondary e uent after the secondary clari er and before ltration.
Supporting information Table 1 shows the typical primary e uent quality for both sites.

Sampling and laboratory analyses
Samples were taken monthly for a period of one year and were processed the same day. Samples were collected during the same day of the week at the same time of day to minimize inconsistencies. If there was a precipitation event 72 hours prior to sampling, sampling was postponed to at least 72 hours post-precipitation (results from the second February are missing due to continuous rain). Sampling points within both Plants are indicated in Fig. 1. The temperature and oxidation-reduction potential of samples were measured immediately after samples were taken.
For the particle analysis, each sample was allowed to settle for one hour, then the supernatant containing unsettled particles was collected ("larger particles in un ltered sample"). Part of the same supernatant was ltered through a 0.45-µm glass bre lter ("nano-sized particles in ltered sample") to remove larger particles that might interfere with the analysis of smaller particles. The mean particle size (Z avg , nm; intensity weighted mean hydrodynamic size) and count (count rate, CR, kcps) of both ltered and un ltered supernatant were analysed using Zetasizer Nano ZS For the EPS analysis, EPS of the activated sludge were extracted according to Li and Yang (Li and Yang 2007) and separated into loosely bound (LB EPS) and tightly bound (TB EPS) fractions. Brie y, 50 ml of well-mixed sludge samples were transferred to 50-mL polypropylene centrifuge tubes and underwent 5 min of centrifugation at 4•10 3 g. The supernatant was discarded and replaced with the same volume using a 50°C solution of 0.05% sodium chloride (NaCl) which was prepared with 18.2 MΩ cm − 1 water. After resuspending the pellet with a vortex mixer for 1 min, the samples underwent centrifugation for 10 min at 4•10 3 g. The supernatant subsequently obtained here was considered to be loosely bound EPS (LB EPS). The supernatant was replaced with fresh 60°C 0.05% NaCl solution and heated in a water bath at 60°C for 30 min. After centrifugation at 4•10 3 g for 15 min, the supernatant obtained was considered to be the tightly bound EPS (TB EPS). Each EPS fraction was quanti ed by its protein and polysaccharide content using To further investigate the temperature effect on the Z avg and CR, the study period was separated into cooler months (December to April) and warmer months (May to November).

Statistical analysis
Our results were analysed with Pearson's correlation for signi cance (p < 0.05) between two paired parameters and further analysed with principal component analysis (PCA) biplots to understand multi-parameter correlations (MATLAB R2018a). To perform PCA on our complete data set required our data to be free of missing data. However, there were some missing data points in the rst month of our study because some sampling points were added during the second month to have a ner resolution of the suspended particle behaviour, and there were also some reporting irregularities from the instruments (i.e., values shown as out of range; missing data shown as colourless bar spaces in Fig. 2). Therefore, data augmentation was performed to analyse the full data set. The missing data values were augmented as follows: we rst separated cooler and warmer months because the plant operating conditions were different in those months, then we interpolated the missing data found in the warmer months by averaging the data in the previous month and the following month. The same procedure was applied to the cooler months. Since the data missing in the beginning of the sampling campaign could not be augmented with the average from the previous month, we augmented such missing data with the average of the existing data points. The PCA coe cient results from the original and the augmented data were compared using the chi-squared statistical test to check whether they were signi cantly different (p < 0.05).
Results And Discussion

Monthly trend
The particle count and average size of particles in the monthly wastewater samples are plotted in Fig. 2  Particle count rate (CR) In Plant 1, we found that the CR in the un ltered samples was generally much higher in the primary treatment samples relative to samples from the secondary and tertiary treatment (Fig. 2, SI Table 2). The differences observed were possibly due to the ferric and polymer added during the primary treatment. However, after ltering through the 0.45-µm pore size lter, the average CR in the secondary treatment samples was much higher than in the primary treatment samples. This indicated that in Plant 1 most of the nano-sized particles were produced by the ASP and were not already present in the in uent wastewater to the plant. When reviewing monthly variations in the CR for both ltered and un ltered samples, there were slightly more particles observed in the warmer months (CR 1027 ± 1663 and 1163 ± 1113, respectively) than in the cooler months (CR 442 ± 524 and 942 ± 496, respectively), suggesting a possible temperature dependence for particle production.
As a whole, Plant 1 was more e cient at treating larger particles than removing nano-sized particles, removing 93% of larger particles and 59% of nano-sized particles (SI Table 3). The overall secondary treatment of Plant 1 (ASP and secondary clari er) had a higher treatment e ciency for larger particles (95.9% relative removal) than for nano-sized particles (40.1% relative removal), but secondary treatment was the most effective process in removing both the larger and nano-sized particles out of all the treatment processes at Plant 1. The secondary clari er by itself was more e cient in removing nano-sized particles (64.5% relative removal) than larger particles (26.9% relative removal). In addition, while tertiary chlorination process removed some nano-sized particles, the treatment had an increased abundance of larger particles. The results suggest that larger particles may be produced within the tertiary treatment post-secondary sedimentation and post-secondary treatment ltration.
Removal e ciency of both large and nano-sized particles was higher during warmer months in primary treatment and secondary clari er (SI Table 3). The removal of nanosized particles was more e cient in the tertiary treatment during warmer months. However, as mentioned before, the abundance of nano-sized particles increased in ASP in warmer months (SI Table 2). This suggests that temperature might have in uenced the particle production and removal differently during individual unit processes in Plant 1.
In Plant 2, similar to Plant 1, the CR of the primary treatment un ltered samples was generally higher than that of the secondary and tertiary treatment samples (Fig. 2, SI Table 2). Also similar to Plant 1, Plant 2 was more e cient overall in treating larger particles (94.1% removal) than removing nano-sized particles (43.9% removal) (SI Table 3). The overall secondary treatment via ASP and via TF were both more e cient in removing larger particles (95.4% and 88.9% relative removal, respectively) than nano-sized ones (52.7% and 28.3% relative removal, respectively), but the overall secondary treatment via ASP removed more particles than did overall secondary treatment via TF. The results from the secondary treatment are very similar to those found in a report on three treatment plants in the same region of Southern California surveyed for their nanoparticles removal, with their ASP having 63.9% and 85.5% removal and TF 74.5% removal. (Rosso and Rajagopalan 2011) This means the removal e ciencies may vary within a range and small differences between removal e ciencies may not be signi cant. The tertiary chlorination removed additional nano-sized particles (11.1% relative removal) but increased the abundance of larger particles (46.2% relative increase), which also suggests that larger particles may be formed during chlorination as seen at Plant 1.
The similar abundance of nano-sized particles during the primary treatment in Plant 2 and in the beginning of the ASP was not seen in Plant 1 (Fig. 2, SI Table 2), perhaps because there was no addition of chemicals and no equalization basin in Plant 2. Although there was no chemical added, the overall ASP treatment in Plant 2 was more e cient in removing nano-sized particles than in Plant 1 (SI Table 3).
Warmer temperatures increased the abundance of nano-sized particles in ASP secondary clari er in Plant 2 (SI Table 3), suggesting warmer temperatures may have negatively impacted the removal e ciency of nano-sized particles in ASP secondary clari er. However, warmer temperature seemed to increase the removal e ciency of nanosized particles from the TF secondary clari er.
The analysis of potential temperature effect on CR in both Plants will be discussed in the Statistical Analysis section.

Average particle size (Z avg )
The Z avg in Plant 1 was generally smaller in the ltered samples than in the un ltered samples (Fig. 2, SI Table 4).
Within the treatment train, the mean Z avg was the largest in the ASP for both ltered and un ltered samples. In the entire secondary treatment, the un ltered samples had a majority of the particles with a size below 100 nm (particles with size below 100 nm in un ltered samples 15.9% ± 9.0%, 55.1% ± 33.3%, and 38.8% ± 34.9% for primary, secondary, and tertiary treatment, respectively). For the ltered samples, the percentage of particles with size below 100 nm had a different trend compared to un ltered samples (particles with size below 100 nm in ltered samples 40.1% ± 32.5%, 36.1%±28.8, and 58.4% ± 24.9% for primary, secondary, and tertiary treatment, respectively). Although the ltered and un ltered samples showed different trends, the percentage of particles with size below 100 nm increased at the end of the tertiary treatment compared to the beginning of the primary treatment. However, when comparing the Z avg of the particles in the beginning and at the end, the mean Z avg was larger at the end of the treatment than in the beginning.
The increased percentage of particles with size below 100 nm at the end of Ce was not due to larger particles breaking up in the secondary treatment, mass balance-wise, because a larger volume of larger particles was removed than the volume of smaller particles was produced (SI Fig. 1). Instead, it was likely coming from different processes that started in the secondary treatment because after a huge increase of Z avg during the secondary treatment there was a consistent decrease of Z avg along the treatment train, even though the Z avg in the tertiary treatment was larger than in the beginning. The increase of particles less than 100 nm may present a challenge for ltration post-secondary treatment, because sizes less than 450 nm were found to be more important for causing membrane fouling. (Howe and Clark 2002) In terms of the temperature effect, as alluded to by the CR results in Plant 1, the average particle size seemed to be larger during the warmer months than in the cooler months.
In Plant 2, the majority of Z avg was smaller in the ltered samples than in the un ltered samples (Fig. 2, SI Table 4).
For un ltered samples, the largest Z avg of every month was found in the ASP; for ltered samples about half of the largest Z avg of the month occurred in the primary treatment processes, while the other half occurred in the ASP. For ltered samples, the percentage of particles with size below 100 nm increased compared to un ltered samples (volume percentage of particles with size below 100 nm in un ltered samples in primary treatment 10.6%±18.0%, ASP 36.5%±36.7%, TF 22.3%±27.8%, tertiary treatment 53.2%±32.1%; in ltered samples in primary treatment 30.4% ±28.2%, ASP 61.1%±22.6%; TF 47.0%±18.7%, tertiary treatment 61.5%±19.9%). For both ltered and un ltered samples, the percentage of particles with size below 100 nm increased at the end of the tertiary treatment, and this increase is similar the results from Plant 1. Similar to Plant 1, warmer months seemed to produce a slightly larger Z avg than cooler months. This could mean that temperature may have an in uence on the size of the particles. Unlike Plant 1, however, Z avg at the end of the treatment was similar to the beginning of the treatment.
With particles below 100 nm still present in both Plants at the end of the tertiary treatment, and the overall removal of nano-sized particles via ASP being 59.0% at Plant 1 and 43.9% at Plant 2 (1.3% via TF; SI Table 3), our results suggest that current treatment processes are not designed to e ciently remove nano-sized particles, though the overall treatment seems to be quite e cient in removing larger particles (93.0-94.1% removal via ASP, 89.9% via TF; SI Table 3). Interestingly, the removal e ciencies of larger particles seem to also coincide with the removal e ciency of microplastic particles and bers from wastewater (1 nm-5 mm with 100 nm-5 mm being the majority; 84-97% removal). (Magni et al. 2019) Therefore, any improvements to the current treatment processes to remove nano-sized particles should also be cognizant of the removal of other particulate matter of concern.
In a few cases, the Z avg of the ltered samples was larger than un ltered samples in Plant 1 (Fig. 2, SI Table 4), and some Z avg from ltered samples was even larger than the lter pore size (0.45 µm); this contributed to the lower volume percentage of particles with size below 100 nm in the ltered samples from the secondary treatment in Plant 1. This also occurred in some samples in Plant 2. The larger sized particles after ltration could potentially be attributed to two reasons. First, since the larger average particle size after ltration occurred only 19 times in Plant 1 (of those, 16 were ASP samples out of a total 143 samples) and only 5 times in Plant 2 (all were PE samples out of a total 130 samples), some lters might have been faulty and produced larger particles than intended. Another possibility could be that the particles after ltration interacted with each other and agglomerated into larger particles; however, we believe this to be less likely due to their inconsistent occurrence during the study.

Statistical analysis
As seen from the monthly trend, temperature may have some in uence on the size and removal e ciency of the particles during wastewater treatment. In addition, EPS, in uent COD, TSS, and VSS may also have an in uence on the particle size and abundance. Thus, we explored their relationship statistically using Pearson's correlation and PCA.
Since augmentation was used, we performed a chi-squared test on the coe cients both with and without augmentation. The analysis showed that they were not signi cantly different from each other for both Plants 1 and 2, and so augmentation did not alter the PCA results signi cantly.
We used Pearson's correlation to assess the relationship between temperature and particle removal e ciency in Plants 1 and 2, and the only signi cant relationships were from Plant 1 (SI Table 5). The removal e ciency of the particles in the un ltered samples for the entire Plant 1 had a signi cant inverse relationship with temperature. A similar and more signi cant relationship was observed for the removal of the particles in the un ltered samples of the tertiary treatment, although, relative to the entire Plant 1, only a small percentage of un ltered particles were actually removed during the tertiary treatment (SI Table 2). For the un ltered samples, the removal e ciency of the primary treatment, however, had a signi cant direct relationship with temperature. For the ltered samples, the removal e ciency in the secondary clari er also had a signi cant and direct relationship. Thus, lower temperatures may help remove larger particles at Plant 1, but it may impact speci c unit processes differently within the treatment plant.
Increasing temperature seemed to correlate with an increase of CR and Z avg in ltered samples in Plant 1 (Table 1), but it did not correlate with other parameters nor any parameters from Plant 2. In addition, only a very slight correlation of temperature with CR and Z avg was observed in the PCA for Plants 1 and 2 (Fig. 3). This may be due to the relatively small seasonal temperature differences that Southern California experiences, thus the result did not conclusively suggest that temperature had a substantial impact on the particle size and abundance. Table 1 Pearson's correlation coe cients and their signi cance level (p < 0.05 was denoted by *, p < 0.01 was denoted by **, both bolded) among the parameters of un ltered and ltered (through a 450-nm lter) supernatant particles in Plant 1 and Plant 2. The highlighted coe cients were found to be signi cant at both Plants.  Fig. 3), while ltered COD in both Plants 1 and 2 correlated with both un ltered and ltered CR and ltered Z avg . This suggests that the availability of dissolved carbon source may impact the quantity of both the larger particles and nano-sized particles produced during the treatment process, and nano-sized particles tend to be bigger with higher carbon availability. The relationship between the increase of abundance of particles and the increase of dissolved carbon con rms that the most suspended particles, especially nano-sized particles (Fig. 2), were naturally occurring and generated biologically within the wastewater treatment, (Kowalkowski 2010, Lead andWilkinson 2006) most likely by the microbes in the activated sludge. Subsequently, the results again suggest that the current WRRF processes are not e cient in removing nano-sized particles, and the increase of nano-sized particle abundance could also impact downstream water reuse treatment.
We also noted that ltered (i.e., dissolved) COD steadily became the dominating COD instead of particulate COD toward the end of the treatment (SI Figs. 2 and 3). Larger particles contributed substantially to the total COD (un ltered COD) in the primary treatment and the beginning of the secondary treatment (echoing un ltered COD correlating signi cantly with the abundance of un ltered CR, Table 1, Fig. 3), but the total and dissolved COD were quite similar by the end of the treatment (COD f /COD uf ~1) as the larger particles were removed by the treatment processes.
EPS and their components correlated signi cantly with both un ltered and ltered CR and ltered Z avg in Plant 2, but they did not correlate with any CR and Z avg in Plant 1 (Table 1, Fig. 3). This suggests that the production of EPS and their components contributes signi cantly to the quantity of both larger particles and nano-sized particles, and the increase of the size of nano-sized particles in Plant 2.
The differences observed between the two plants may be attributed to the differences of the two plants' treatment processes and operation. Plant 1 was operating with an MLE with recirculation, which would render the ow in the biological treatment similar to a continuous stirred tank reactor (CSTR  Fig. 3). This may also explain the drastic decrease of CR and Z avg during the biological treatment in Plant 2 but not as much change of CR and Z avg in Plant 1 (Fig. 2, SI Tables 1 and 2).

Conclusions
It is important to know the removal e ciency and the production of nano-sized particles in WRRFs in order to accurately model the treatment of nano-sized particles, examine their effect on subsequent treatment processes for water reuse, and investigate the fate and transport of certain contaminants associated with these particles. In this study we found that the two secondary treatment trains, TF and conventional ASP, were more e cient in removing particles larger than 450 nm than smaller particles (on average > 88.9% removal e ciency of larger particles and 40.1-52.7% for nano-sized particles within the ASP treatment). In addition, we found that the WRRFs are not designed to remove nano-sized particles e ciently due to the overall low removal e ciency of nano-sized particles (1.3-59.0%) and the increasing percentage of particles with size below 100 nm along the treatment train. A signi cant and direct correlation was found between the quantity of suspended particles and the quantity of dissolved carbon, suggesting that the majority of suspended particles in the wastewater treatment were generated biologically within the wastewater treatment. The effect of temperature on the size and the quantity of these particles was less conclusive from our study, and this may be due to the small variation of temperature in Southern California.

Declarations
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.
Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.  The principal component analysis (PCA) of selected parameters from Plant 1 and Plant 2. Some parameters were augmented before being analyzed. The inset is the entire PCA of the parameters and a portion of the inset details was enlarged in the graph. Abbreviations: ltered, f; un ltered, unf; volatile suspended solids, VSS; total suspended solids, TSS; chemical oxygen demand, COD; protein, Prot; count rate, CR; extracellular polymeric substances, EPS; temperature, T; polysaccharide, Polys; redox potential, ORP; average particle size, Zavg

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.