Seasonal distributions of Ti-containing NPs in the main tributaries of the Tamsuei River Basin in northern Taiwan

Currently, widespread applications of engineered nanomaterials have led to a rapid increase in the number of nano-enabled products on the market (Jankovic & Plata 2019). The release of engineered nanomaterials during the manufacturing process, life cycle, and disposal of nanoproducts raises concerns about risks to the environment and human health (Larsson et al. 2019).

Engineered nanomaterials applied to products with high prevalence and frequency of use contain mostly metallic nanoparticles and are more likely to end up into the environment. For instance, titanium dioxide (TiO₂) nanoparticles (NPs), employed in white pigment, UV filters, and photocatalysts (Weir et al. 2012), achieved a global production volume of 100,000 tons/year in 2015, with 75% utilized in the fabrication of personal care products (Jankovic & Plata 2019). Its widespread application dominates the discharge pathway of TiO₂ NPs via domestic wastewater (Keller et al. 2013), leading to the discharge of a significant quantity of TiO₂ NPs into effluent-receiving water bodies (Westerhoff et al. 2011).

Size- and dose-dependent toxic effects of NPs have been reported in both animal and in vitro studies (Sukhanova et al. 2018), underscoring the importance of assessing the mass and particulate number concentrations as well as the size distribution of TiO₂ NPs for risk assessment. Compared to other current analytical techniques for nanomaterial characterization, single-particle inductively coupled plasma mass spectrometry (sp-ICPMS) offers the advantage of simultaneously capturing particle properties, including mass concentration, particulate number concentration, and size distribution (Mourdikoudis et al. 2018). Its high sensitivity renders it suitable for analyzing aquatic environmental samples at low concentrations. Peters et al. (2018) analyzed actual river samples from the Netherlands containing TiO₂ NPs, with measured mass and particulate number concentrations ranging from 0.2 to 8.1 ng/mL and 5 × 10³ to 150 × 10³ part./mL, respectively, validating model-predicted concentrations. In South Africa, particulate number concentrations of Ti-containing NPs behind freshwater dams ranged from 83 × 10³ to 140 × 10³ part./mL (Maiga et al. 2019). Donovan et al. (2016) reported particulate number concentrations of Ti-containing NPs in the Missouri River ranging from 425 × 10³ to 451 × 10³ part./mL. Moreover, a global survey of surface waters and precipitation indicated that, based on measurements from 46 sites across 13 countries, total Ti-containing NP number concentrations (regardless of origin) often fell within the range of 10⁴ to 10⁷ part./mL, corresponding to mass concentrations typically below10 ng/mL, given that measured sizes often exceeded 30 nm (Azimzada et al. 2021). Additionally, several studies have intentionally monitored anthropogenic releases of Ti-containing NPs. For instance, the Old Danube Lake in Austria, a recreational water body likely receiving engineered TiO₂ NPs from sunscreen, exhibited Ti-containing NPs at a mass concentration of 1.38 ng/mL (Gondikas et al. 2014). Elevated concentrations of Ti-containing NPs during the time period of intensive use of TiO₂ NP-containing products were also observed at a bathing area in the Salt River, USA, with overall concentrations ranging from 0.260 to 0.659 ng/mL (Venkatesan et al. 2018). Similarly, the Loire River in France, receiving effluents from anthropogenic activities and sewage discharge, showed Ti-containing NP number concentrations ranging from 134 × 10³ to 803 × 10³ part./mL (Phalyvong et al. 2020).

Quantifying TiO₂-engineered nanoparticles in environmental water presents challenges due to the prevalence of natural background Ti content, with titanium being the ninth most abundant element in the Earth's crust (Hampel 1968). Natural NPs resulting from geochemical or mechanical processes contribute hundreds of times the flux of anthropogenic NPs into the environment and exist in the atmosphere, hydrosphere, and lithosphere (Hochella et al. 2015, 2019). Consequently, Ti-containing NPs in the environment settings are predominantly of natural origin. Establishing background reference levels for Ti-containing NPs in rivers would aid in discerning potential anthropogenic TiO₂ NP releases. Moreover, as the generation, release and transport behavior of Ti-containing NPs are strongly influenced by environmental factors such as pH, chemical concentration, temperature, and hydrology, variations in the quantity and characteristics of Ti-containing NPs in natural water between dry and wet hydrological seasons are expected.

Currently, there is a dearth of studies on the quantification of Ti-containing NPs in rivers in Taiwan. Thus, the objective of this study was to establish baseline data on the mass concentration, particulate number concentration and size distributions of Ti-containing NPs in the Tamsuei River Basin, which encompasses the largest metropolitan in Taiwan. Additionally, given the substantial seasonal variations in river flows in Taiwan, the present study focused on assessing the differences in mass concentration, particulate number concentration and most frequent size of Ti-containing NPs between the dry and wet hydrological seasons.

The Tamsuei River Basin, situated in northern Taiwan, encompasses three main tributaries, spanning a total length of 158.7 km and a drainage area of 2,726 km². Two of the main tributaries, the Dahan River and the Xindian River, converge near Taipei City to form the initial stretch of the Tamsuei River. Subsequently, the Keelung River joins the Tamsuei River from the north in Guandu, marking the lower boundary where the Tamsuei River Basin traverses through the Taipei metropolitan area, before flowing northward into the Taiwan Straits.

For the Xindian River, water sampling sites X1–X7 were situated within a designated water quality protection area, subject to stringent regulations governing anthropogenic activities and sewage discharge into the water body. Specifically, sites X1–X3, located in remote mountain regions predominantly covered by natural forests with minimal residential and agricultural land use, were presumed to experience the least anthropogenic activity impact in this study.

In contrast, the Keelung River is characterized by scattered settlements along its course, primarily serving agriculture and industrial purposes. Water sampling sites K1 and K2 were adjacent to small villages, while site K3 was situated in a more densely populated area. Additionally, an industrial park lies between water sampling sites K3 and K4.

In the case of the Dahan River, all water sampling sites were positioned on manmade structures: site D1 at the Shimen Reservoir, site D2 adjacent to an irrigation ditch, and, as previously mentioned, site D3 at the water intake of a water treatment plant. Coupled with intensive anthropogenic land use within the riverine area, water sampling sites along the Dahan River were deemed most susceptible to anthropogenic influences among all studied river sites.

Water sampling was conducted at all sampling sites along the three tributaries twice: once in early March and again in early June of 2020, representing the dry and wet seasons, respectively. However, for site X8, sampling was performed only once in June.

Samples were collected using a custom-made sampler equipped with a PFA container (SANPLATEC^®, Japan) at a depth of 50–100 cm from the river center. Subsequently, the collected samples were stored in 50 mL centrifuge tubes (SARSTEDT AG & Co. KG). To prevent sample contamination, all containers underwent pretreatment by thorough rinsing and were filled with double-deionized water to remove and prevent particles from adhering to the inner walls.

Replicate samples were obtained from each sampling site and transported back to the laboratory under cooled, dark conditions. For further details regarding the customized water sampler, sampling preparations, and sample pretreatment procedures, please refer to our previous study on river water sampling (Hwang et al. 2021). Concurrently, water quality parameters, including conductivity, total dissolved solids (TDS), salinity, and temperature values, were measured on-site during water sampling using an ExStick EC500 direct-reading instrument (EXTECH, USA).

Total Ti concentrations, as well as the particulate mass concentrations, particulate number concentrations, and particle sizes of Ti-containing NPs, were analyzed using sp-ICPMS (Agilent 8900). The analytical method was adapted from a previous study (Peters et al. 2018). Ti was monitored at m/z 48, as ⁴⁸Ti is the most abundant isotope in nature. The analysis was conducted in mass-shift mode, employing O₂ as a reaction gas to eliminate polyatomic interference (e.g., ³²S¹⁶O⁺) and isobaric interference (e.g., ⁴⁸Ca) on ⁴⁸Ti. The dwell time was set at 0.1 ms, and data analysis was performed in fast time-resolved analysis (TRA) mode. Prior to sp-ICPMS analysis, the samples were diluted to optimal concentration ranges for Ti-containing NP determination using double-deionized water and designated dilution factors. These factors were predetermined based on pretest results. Specifically, samples from the Xindian River, Keelung River, and Dahan River were diluted by factors of 10, 100, and 100, respectively. Subsequently, the diluted samples were filtered through a 0.45 μm pore size PVDF filter (Recenttec^®, Taiwan) prior to the subsequent sp-ICPMS analysis.

The nebulization efficiency of the reference element, Au, in this study was determined from the count per second (cps) values of 0.1 ng/mL ionic and nanosized standards, respectively. This determination served as a basis for calculating particle size in the single-particle mode. Initially, the ionic Au standard of 0.1 ng/mL was diluted from the tuning solution (Agilent, USA) of 100 ng/mL. Subsequently, the nanosized Au standard of 0.1 ng/mL and 30 nm was diluted from a 100 ng/mL stock suspension, which was originally prepared by mixing 0.1 mL 30 nm Au nanospheres (nanoComposix, USA) at 50 ng/mL with 0.8 mL of pure isopropanol, and then diluted to 50 mL with double de-ionized water.

Meanwhile, an ionic Ti standard of 1 ng/mL was established as the reference concentration of the analyte, i.e., TiO₂ in this study utilizing single-particle analysis. A calibration curve was constructed using a 100 nm TiO₂ dispersion standard with designated concentrations of 0.01, 0.02, 0.05, 0.1, 0.2, 0.5, and 1 ng/mL, forming a 7-point calibration curve. The aforementioned ionic Ti standards ranging from 0.01 to 1 ng/mL were diluted with double de-ionized water from 100 ng/mL stock solution, originally diluted from a 10 μg/mL multi-element calibration standard-4 solution (Agilent, USA). As to the 100 nm-sized TiO₂ dispersion standard, it was diluted with double de-ionized water from a stock solution of 1,000 μg/mL, initially diluted from a standard solution with a weight percent concentration of 20% (US Research Nanomaterials, Houston). The detection limit for mass concentration of Ti in the present study was 0.0105 ng/mL.

is the background signals (cps) of double de-ionized water; R is the response factor (cps/ppb) of the analyte; η is the transport efficiency (%); is the flow rate (mL/min); T is the dwell time (s); is the mass fraction of element in the particle; and is the density of the particle (g/cm³). In the preliminary test conducted for the present study, the calculated for Ti-containing particles was found to be 2.32 nm.

Additionally, spiked samples of 0.1 ng/mL were intermittently inserted between every 10 analyses to verify the analysis accuracy and instrument stability. The criteria set for verification between 80 and 120% for the measured number of particle and intensity (cps) compared with the corresponding references, respectively. For further details regarding the operational parameters used for sp-ICPMS determination of Ti-containing NPs, please refer to Table 1.

Moreover, by normalizing the Ti-containing NP mass concentration by the total Ti concentration, it was evident that the proportion of Ti-containing NPs increased significantly during the wet season (Figure 2(d)). Regarding spatial trends, the increases in NP mass concentrations from upstream to downstream were more pronounced in the Keelung and Dahan Rivers during the wet season. However, the spatial distribution of NP mass concentration in the Xindian River exhibited irregularities and even displayed a reverse trend compared with the Keelung River and the Dahan River. This was particularly notable as extremely high values were measured in its upstream tributaries of the Xindian River during both the dry and wet sampling seasons.

Although all the Xindian River sampling sites generally featured the lowest total Ti concentrations among the three study tributaries, the highest Ti-containing NP concentrations, among all Xindian River sampling sites, of 0.31 ± 0.08 ng/mL in the dry season and 1.92 ± 0.00 ng/mL in the wet season were measured at water sampling sites X1 and X2, respectively, situated upstream in the deep mountain area compared with all other Xindian River samples. Conversely, while the lowest mass concentrations of Ti-containing NPs were detected at water sampling site X3 (0.05 ± 0.02 ng/mL and 0.34 ± 0.04 ng/mL in the dry and wet seasons, respectively), the total Ti concentrations at the same sampling site (7.21 ± 0.09 ng/mL and 5.64 ± 0.07 ng/mL for the dry and wet seasons, respectively) were the highest compared with those of all other Xindian River sampling sites.

The distribution pattern of the particulate number concentration of the Ti-containing NPs mirrored that of the Ti-containing NP mass concentration. During the dry season, the Keelung River exhibited the average particulate number concentrations ranging from 170 × 10³ part./mL to 231 × 10³ part./mL, while the lowest concentrations were similarly found in the Xindian River, ranging from 5.32 × 10³ part./mL to 96.4 × 10³ part./mL. Comparable to the distribution of the Ti-containing NP mass concentration, the particulate number concentrations of Ti-containing NPs in the Keelung and Dahan Rivers displayed limited variation from upstream to downstream during the dry season. However, a clear increasing trend from upstream to downstream was evident for wet season samples (Figure 2(b)). The particulate number concentrations of Ti-containing NPs increased from 1,372 × 10³ ± 415 × 10³ part./mL at sampling site K1 to 4,110 × 10³ ± 465 × 10³ part./mL at sampling site K4 and from 666 × 10³ ± 85.0 × 10³ part./mL at sampling site D1 to 2,591 × 10³ ± 655 × 10³ part./mL at sampling site D3. A similar but moderate spatial trend was also observed for water sampling sites X6 to X8 of the Xindian River during the wet season, with concentrations rising from 111 × 10³ ± 9.06 × 10³ part./mL to 268 × 10³ ± 31.8 × 10³ part./mL. However, the highest particulate number concentrations of Ti-containing NPs in the Xindian River for both the dry season (96.4 × 10³ ± 33.9 × 10³ part./mL) and the wet season (417 × 10³ ± 26.0 × 10³ part./mL) were also distributed in the upstream sampling sites X1 and X2, respectively, rather than in the relatively downstream sampling sites.

The most frequent sizes of Ti-containing NPs measured at each sampling site were all smaller than 100 nm in this study. Most of the samples of the Keelung River were characterized with a stable particle size range of approximately 40 nm, while those from the Xindian River mainly fell within the range of 38 − 66 nm. No consistent spatial or seasonal trend was observed in either the Keelung River or the Xindian River. However, as illustrated in Figure 2(c), the most frequent size was larger for dry season samples at all sampling sites in the Dahan River, decreasing from 60 to 72 nm in the dry season to 38 − 52 nm in the wet season.

Additionally, water quality parameters during the water sampling periods are presented in Table 2. Significant seasonal differences were observed in water parameters, with higher temperature and turbidity during the wet season, while conductivity, TDS, and salinity were higher in the dry season. For pH values, no significant seasonal difference or spatial trend was observed among the samples of the Keelung River and the Xindian River, while those in the Dahan River exhibited relatively higher pH values in the downstream samples during both the dry and wet seasons. Furthermore, statistical analysis indicated a significant correlation between turbidity and Ti-containing NP mass concentration, with r² = 0.681 (p < 0.001), equivalently in the dry season (r² = 0.610, p < 0.001), and wet season (r² = 0.628, p < 0.001).

All samples analyzed in the present study exhibited mass concentrations exceeding the detection limit of 0.0105 ng/mL. These measured mass concentrations diverged significantly, showing a tenfold difference from the model-predicted concentrations of TiO₂ NPs in surface water (Gottschalk et al. 2009; Tiede et al. 2016). However, they aligned with findings from various experimental studies, which reported concentrations ranging from low ppb to sub-ppb levels (Gondikas et al. 2014; Peters et al. 2018). This discrepancy might be attributed to regional disparities in key model inputs, such as TiO₂-engineered nanoparticle production, application volumes, and environmental fate. However, a more plausible explanation is that existing modeling studies focus solely on engineered TiO₂ nanoparticles, without accounting for Ti-containing NPs originating from other sources. Many of our observations supported the hypothesis that the contribution of naturally occurring Ti-containing NPs was quite significant, which will be further elaborated on in the following sections.

Spatial trends of Ti-containing NPs implied steady and continuous reception and accumulation downstream in all studied rivers, suggesting that elevated concentrations primarily stem from nonpoint sources. For instance, variations in Ti-containing NP concentrations among rivers could be linked to regional geological characteristics (Wang et al. 2022). Processes such as mineral weathering and biological activities, such as redox reactions mediated by aqueous bacteria, can release ions and often produce nanoscale minerals (Hochella et al. 2008). Natural Fe–Ti oxide minerals are commonly found in igneous rocks, with reports of their presence in the Keelung volcano group and near the Shimen Reservoir on the Dahan River in northern Taiwan (Chen 1990). Catchment areas rich in Ti-bearing minerals may exhibit significantly higher background levels of Ti-containing NPs in rivers like the Keelung and Dahan, compared with the Xindian River. In addition to natural factors, land-use types can also explain variations in Ti-containing NP levels among the three studied rivers. Architectural coatings and white road markings contain TiO₂ pigments release engineered TiO₂ particles ranging from nano to micron sizes during weathering process (Kaegi et al. 2008; Al-Kattan et al. 2013; Wang et al. 2020), making urbanized areas potential diffuse sources of engineered TiO₂ particles. The riparian areas of the Dahan and Keelung Rivers, two of the Tamsuei River's main tributaries, feature higher proportions of built-up land use compared with the Xindian River, which is primarily within a water preservation zone. This suggests that the elevated mass concentrations downstream could also stem from engineered TiO₂ nanoparticles carried by urban runoff entering the receiving river. An anomalous point release event was suspected along the Keelung River during the wet season sampling. A sudden increase in both total Ti concentration and mass concentration of Ti-containing NPs occurred at water sampling site K4 (Figure 2), accompanied by a noticeable rise in turbidity, indicating a release source between sampling sites K3 and K4 (Table 2 and Figure 2(d)). This scenario may be attributed to the presence of an industrial park upstream of the Keelung River, where occasional discharge of industrial wastewater, containing Ti-containing metals and other unknown pollutants, could occur. Furthermore, another exceptional scenario is that the spatial distributions of Ti-containing NP mass and number concentrations in the Xindian River exhibited irregularities, contrary to the profound spatial trends of general increases from upstream to downstream as observed in the Keelung and Dahan Rivers. This phenomenon could be explained from various aspects. First, compared with all other sampling sites of the Xindian River, notably high values of Ti-containing NPs were measured in its upstream tributaries, i.e., sites X1 and X2, which might be attributed to the local geology with relatively richer Ti-bearing minerals, and consequently elevated background levels. Second, as noted in Figure 1, sampling sites X1, X2 and X3, X4 are situated on two different upstream tributaries (creeks), respectively. As expected, there were increasing trends in Ti-containing NP mass and number concentrations from upstream to downstream for both sets of sampling sites, i.e., X1, X2, X5 and X3, X4, X5, respectively, similar to those observed in the Dahan and Keelung Rivers. Third, the concentrations measured at downstream sampling sites X6, X7, and X8 did not consistently exceed those of other samples collected upstream, thus failing to establish a clear upstream–downstream increasing trend. This anomaly is likely due to the locations of these sampling sites, which are within the water preservation zone and close to the water intake for drinking water treatment plants. No residential areas are within the restricted zone and therefore limit the anthropogenic sources contributing to the Ti-containing mass and number concentrations.

Domestic wastewater represents another significant source of engineered nanoparticles entering surface waters. Studies have confirmed the presence of Ti-containing NPs in treated wastewater (Kiser et al. 2009), with increased particulate number concentration observed in receiving rivers adjacent to the wastewater outlets (Phalyvong et al. 2020). In the present study, water sampling site X8 was positioned downstream of the effluent outlets of the Wulai and Zhitan wastewater treatment plants, and one-shot sampling was conducted during the wet season. However, none of the NP analysis items exhibited the expected significantly elevated levels or distribution variations in Ti-containing NPs compared to upstream water sampling sites. This suggests that the contribution of Ti-containing NPs from domestic wastewater may be relatively limited, or that most Ti-containing NPs in wastewater are efficiently removed during treatment processes before effluent discharge. This finding is consistent with previous research indicating that up to 90% of Ti could be effectively removed by wastewater treatment, resulting in treated effluent with substantially lower Ti concentrations than predicted by modeling (Markus et al. 2018). Furthermore, it is possible that a portion of the Ti-containing NPs in the effluent undergo rapid aggregation and settlement prior to reaching the sampling site in the present study (Velzeboer et al. 2014). Similarly, Shi et al. (2016) compared Ti concentrations in both wastewater and in its receiving water, concluding that other sources, such as urban runoff, might be more significant contributors of Ti-containing NPs than wastewater discharge. This conclusion is supported by previous studies attributing TiO₂ NPs sources to precipitation and urban runoff (Nabi et al. 2021; Rand et al. 2021).

In the present study, the particulate mass and number concentrations of Ti-containing NPs exhibited a consistent seasonal trend across all rivers, with significantly elevated levels during the wet season, a pattern consistent with the findings from a previous study conducted in a rural river basin in South Carolina (Nabi et al. 2022). As illustrated in Figure 2(a), there was a noticeable increase in particulate mass concentration from upstream to downstream. For instance, samples collected from the Keelung River during the wet season showed a more than 400% increase in Ti-containing NP mass concentration at site K4 compared with reference from site K1. In contrast, such an increment was only approximately 28% during the dry season. This suggests a potential association between rainfall and the presence of Ti-containing NPs in rivers. As previously noted, TiO₂ NPs might be released from architectural coatings and road markings into facade runoff, with the release amount likely correlating with precipitation intensity, thus leading to an increased presence of TiO₂ NPs in facade runoff during the wet season.

By dividing the mass concentration of Ti-containing NPs by the total Ti mass concentration, the proportion of Ti in particulate form in the total amount of Ti was determined. The present study found that the proportion of Ti in particulate form in all river samples was significantly higher during the wet season (Figure 2(a) and 2(d)), indicating that the increase in Ti during the wet season primarily occurs in particulate form, likely resulting from facade runoff after rainfall. Additionally, the comparatively high partition coefficient between solid and dissolved fractions suggests that Ti in river water is predominantly associated with suspended solids and sediments (Roychoudhury & Starke 2006). The rise in Ti-containing NP concentrations during the wet season is likely the result of higher river flows, which enhance subsequent bed erosion and resuspension of Ti-containing particles in sediments. This phenomenon also explains why, in most cases, the temporal variation and spatial distribution trends of Ti-containing NP concentrations were consistent with the corresponding turbidities in the present study (Figure 2(d)).

In this study, the most frequent sizes of Ti-containing NPs ranged from 36 to 76 nm, which contrasts with previous studies reporting TiO₂ NP sizes in surface water ranging from hundreds of nanometers to micrometers (Gondikas et al. 2014; Peters et al. 2018; Phalyvong et al. 2020). Several factors may contribute to the disparity in particle size ranges across different studies, including diverse origins, transformations influenced by the physical and chemical properties of the surrounding medium upon NP entry into rivers, and variations in sample pretreatment methods for Ti determination. Additionally, high ionic Ti background or isotope (e.g., ⁴⁸Ca) interference can hinder Ti-containing NP detection due to elevated instrumental background levels. Fortunately, the present study encountered relatively low levels of ionic Ti as background in water samples, enabling successful detection of Ti-containing NPs smaller than 100 nm in all studied rivers.

It is noteworthy that, unlike the apparent spatial and seasonal trends observed for the distributions of mass and particulate number concentrations, the most frequent sizes of Ti-containing NPs remained relatively uniform with limited fluctuations. Previous studies have suggested that factors such as the nature of organic matter, salinity, and particle size in water can influence the mode of the most frequent NP size, affecting NP aggregation or disaggregation processes (Domingos et al. 2009; Honda et al. 2014). Natural organic matter present on NP surfaces in aquatic environments can create steric forces that prevent contacts among NPs, thereby reducing NP coagulation (Diegoli et al. 2008; Li et al. 2017). Conversely, French et al. (2009) reported that TiO₂ NPs would increase in size from 4–5 to 50–60 nm upon addition of a NaCl solution of 263 mg/L to TiO₂ NP solutions. Furthermore, high salt concentrations can induce bridging effects among ions, which in turn influence the interparticle distance and result in NP aggregation (Honda et al. 2014; Wang et al. 2019). Additionally, it has been observed that charge interactions play a more significant role in smaller nanoparticles compared with larger ones (Hofman-Caris et al. 2022). Moreover, Zhang et al. (2010) reported that the lower the pH and the smaller the particle size, the higher the solubility of ZnS NPs within the studied pH range.

In the present study, no clear trend relationship was observed between water quality parameters and the most frequent size of Ti-containing NPs. Visual inspection indicated that pH values of samples collected from the Dahan River during the dry season, ranging from 8.13 to 9.52, appeared to be associated with relatively larger most frequent sizes compared with other rivers and seasons (Table 2 and Figure 2). However, no statistically significant association was identified between pH values and most frequent size. The limited fluctuations in the most frequent sizes of Ti-containing NPs observed in the present study may be attributed to various factors. One possible explanation is that the conditions affecting NP aggregation behavior, such as pH, conductivity, total dissolved solids, and salinity, did not exhibit wide variations among sampling sites and seasons, thus not statistically correlating with the most frequent size despite their differential variations among seasons. Furthermore, particle size is associated with the particle distribution in rivers, such as distribution in depth. For instance, a previous study on the Tamsuei River reported that the size of particles resuspended from sediments vertically increases in relation to water depth (Lin et al. 2014). Since the samples in the present study were collected from fixed sampling sites at similar water depths, comparable hydrodynamic conditions might therefore be imposed on the Ti-containing NPs across all water samples, potentially leading to relatively uniform and consistent sizes. However, further study is warranted to thoroughly verify this hypothesis, as well as to explore the potential subtle effects resulting from the water quality parameters, such as pH, salinity, and dissolved organic compounds. These factors could still play a role in influencing the aggregation behavior and size distribution of Ti-containing NPs, albeit perhaps not as prominently as in cases where hydrodynamic conditions vary more widely.

Regarding the determination of the most frequent particulate size, it might be variably reported for the sample of extremely low levels of NP number concentration, given that there are same particulate numbers most frequently observed for certain particulate sizes in the sample. However, only one most frequent particulate size would be reported for such a sample determined with sp-ICPMS. Therefore, caution should be exercised when interpreting the most frequent NP size in samples with extremely low NP number concentration by scrutinizing the original particle size and frequency distributions.

Another limitation of concern in the present study is the difficulty in verifying the main sources of Ti-containing NPs using the method employed. This challenge arises from the inability to differentiate between natural and engineered Ti-containing NPs due to the lack of natural background information on Ti-containing NP concentrations in rivers. Nevertheless, by comparing Ti-containing NP concentrations between dry and wet seasons in different rivers, it is possible to indirectly illustrate the extent of contribution from natural release of Ti-containing NPs into the river, which may be influenced by various factors, such as rainfall, water flow, hydrological, and geological factors. In future studies, to address the challenge encountered in distinguishing between natural and engineered Ti-containing NPs, elemental ratio methods could be employed. These methods involve using rare-earth elements or naturally co-occurring elements as natural trace elements to establish elemental ratios, such as Ti/Nb, Ti/V, and Ti/Y ratios. By utilizing these ratios, it becomes possible to estimate the mass concentrations of engineered Ti-containing NPs by subtracting the contribution of natural Ti-containing NPs from the total Ti-containing NP concentrations (Baalousha et al. 2020; Phalyvong et al. 2020). This approach can help provide more accurate assessments of the presence and contributions of engineered Ti-containing NPs in environmental samples.

In this study, the sp-ICPMS technique was successfully employed to establish seasonal distribution profiles of Ti-containing NPs in aquatic environments, including mass concentration, particulate number concentration, and particle size, across the three tributaries of the Tamsuei River Basin. The sizes of Ti-containing NPs did not exhibit significant variation between dry and wet sampling seasons, likely due to the stable chemical and physical conditions of the surrounding medium. It was inferred that local geology and land use patterns may contribute to diverse Ti-containing NP concentration distributions across different rivers, with such diversity being more pronounced during the wet season.

Future studies should consider integrating complementary analytical techniques such as scanning electron microscopy (SEM) and single-particle inductively coupled plasma time-of-flight mass spectrometry (sp-ICPTOFMS), along with elemental ratios that require knowledge of the composition of other elements in naturally occurring particles (Phalyvong et al. 2020; Wang et al. 2020). These warranted to provide additional insights into the morphology and elemental composition of individual particles, enabling better differentiation of the various sources contributing to Ti-containing NPs in rivers.

This study was supported by a grant from the Ministry of Sciences and Technology, Taiwan, R.O.C. (MOST 109-2314-B-002-162), and authorized by the College of Public Health, National Taiwan University to use the sp-ICPMS for Ti-containing NPs analysis.

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by J.-A.C., Y.-T.C., and C.-H.L. The first draft of the manuscript was written by J.-A.C. and Y.-H.H., and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.