Occurrence and Sources of Synthetic Musk Fragrances in the Sewage Treatment Plants and the Han River in Seoul, Korea

Levels of synthetic musk fragrances (SMFs) and various personal care products (PCPs) were measured in the Han River and its tributaries in Seoul, Korea. The most abundant SMF in all river and PCP samples was 4,6,6,7,8,8-hexamethyl-1,3,4,7-tetrahydrocyclopenta[g]isochromene (HHCB), followed by 1-(3,5,5,6,8,8-hexamethyl-6,7-dihydronaphthalen-2-yl)ethanone (AHTN), musk ketone (MK), and 1,1,2,3,3-pentamethyl-2,5,6,7-tetrahydroinden-4-one (DPMI). There was a significant correlation between the SMF concentration in the PCPs and the Han River samples. Moving from upstream to downstream in the Han River, the median SMF concentration was 6.756, 2.945, 0.304, and 0.141 μg/L in the sewage treatment plant (STP) influent, effluent, tributaries, and mainstream, respectively, implying that effective SMF removal was achieved during the sewage treatment process, followed by dilution in the receiving water. Four STPs using advanced biological treatment processes had removal efficiencies of 55.8%, 50.6%, 43.3% for HHCB, AHTN, and MK, respectively. The highest SMF concentrations in the tributaries were observed at locations close to the STPs. Our study confirmed that the main source of SMFs in the receiving water were sewage effluent containing untreated SMFs, which are largely originated from household PCPs, especially hair care products (e.g., shampoo) and perfumes.


Introduction
Synthetic musk fragrances (SMFs) are synthetic aromatic compounds that are used in the perfume industry to mimic the scent of deer and other animals [1,2]. Synthetic musks are categorized as nitro, macrocyclic, polycyclic, or alicyclic musk according to their chemical structure and the synthesis methods [3,4]. Nitro musk was synthesized by Baur in 1888, but its use is now restricted due to its proven harmful effects on humans and the environment [5,6]. Macrocyclic musk is difficult to synthesize, and expensive, because it has a complex chemical structure. In contrast, polycyclic musk is relatively inexpensive to synthesize and is known to be safer than nitro musk; it accounts for more than 90% of all synthetic musk used currently [2,7].
Because SMFs are difficult to chemically degrade, they can persist in the environment. Synthetic musk fragrances are also lipophilic, so can accumulate in the adipose tissues of humans and aquatic animals. Several researchers have raised concerns about the safety of HHCB and AHTN following their detection in human fat cells, milk, and even breast milk [7,12]. Other studies have confirmed that nitro musk can act as a weak endocrine disruptor [6]. In a rat study, it was also confirmed that nitro musk compounds can increase the incidence of various cancers, and their toxicity is enhanced when they are present together with other toxic substances, such as benzo-a-pyrene [6]. Therefore, the European Chemical Agency has determined the predicted no effect concentration (PNEC) values for the aqueous phase (PNECwater) for several SMFs (HHCB, 4.4 μg/L; AHTN, 2.8 μg/L;, MX, 1.1 μg/L;, and MK, 1.0 μg/L) [13][14][15][16]. The United States Environmental Protection Agency (US EPA) also lists SMFs on the inventory list of the Toxic Substances Control Act (TSCA) because they have negative effects on human health and the environment, including endocrine-disrupting effects [17]. Musk xylene, HHCB, and AHTN are included in the Substances of Very High Concern (SVHC) list as very persistent and very bioaccumulative (vPvB) substances. The European Chemicals Agency (ECHA) has classified HHCB as an acute and chronic aquatic toxic substance, while MX and MK have been classified as acute and chronic aquatic toxic substances and carcinogens [18]. It is therefore important to determine the levels, and the fate and behavior, of SMFs in the aquatic environment.
Because SMFs are widely used as scented raw materials in household PCPs, more than 90% of the total amount of SMFs used in households enters domestic sewage treatment plants (STPs). Untreated SMFs in STPs can contaminate surface waters through STP effluents. The SMF concentrations in the receiving water can be affected by the sewage treatment process and their removal efficiencies in STPs. Several studies have reported the occurrence of SMFs in sewage effluent in several STPs [7,[19][20][21][22][23]. The SMF removal efficiencies in STPs varied from 50% to 90% [22,[24][25][26][27]. Reiner et al. [28] also reported that HHCB was detected in the concentration range of 1.78-12.7 μg/L in the influent from an urban STP located in New York State, while in the sewage influent in rural Kentucky it was present in the range of 0.043-7.032 μg/L. Surface water monitoring results of SMFs in have been reported in other parts of Korea [8,22,29], but no survey has been conducted in the Han River which passes through Seoul metropolitan area. In this study, we presented annual monitoring results for SMFs in STP influent, effluent, the mainstream of the Han River, and its tributaries. We also compared the SMF concentrations in the Han River samples with the levels found in PCPs. By adopting this approach, we assessed the possible sources and discussed the major removal mechanisms of SMFs in the Han River.
The physicochemical properties of the selected SMFs are presented in Table 1. Certified analytical grade SMF standards were purchased from Dr. Ehrenstorfer GmBH (Augsburg, Germany), and an internal standard, fluoranthene-d10, was purchased from Chiron (Trondheim, Norway). Acetone and water (HPLC grade) were purchased from Thermo Fisher Scientific (Waltham, MA, USA) and NaCl (> 99%) from Tedia (Fairfield, CT, USA).

Sampling
Surface water samples were collected every month from March to November 2018 from five main streams of the Han River (H1-H5) and 17 sites in its tributaries (T1-T17) (Fig. 1). The Han River is the largest river in South Korea. It flows through Seoul and has three large tributaries, with various small streams; the average flow is about 874 m3/s. Surface water samples were collected using pre-washed glass bottles (2 L volume) that were stored in a cool dark location before analysis.
Samples were also collected quarterly at 12 locations (5 influents and 7 effluents) in four STPs (STP1-STP4) located in Seoul. The locations of the collection points are shown in Fig. 1. The total area served by these STPs is approximately 605 km2. The total treatment capacity of the four STPs is 4.98 million m3/d; the largest plant (S4) had a capacity of 1,630,000 m3/d. The total input flow of the four STPs was 4,207,000 m3/d. These STPs mainly treat domestic wastewater (99.4%) with small amounts (< 0.6%) of industrial wastewater [30]. As shown in Table S1 in the Supplementary Materials, after the primary sewage treatment, including the precipitation stage, each STP applies an advanced biological treatment process (a modified Ludzack-Ettinger [MLE] or anaerobicanoxic-aerobic [A2O] process) to remove organics, nitrogen, and phosphorus. Following the advanced biological treatment processes, a chemical coagulant is applied to reduce the total phosphorus (T-P) (< 0.5 mg/L). The treated sewage water is then discharged to the Han River or its tributaries.
To measure SMF levels in household products, various PCPs were purchased from a supermarket in Seoul and the concentrations of nine SMFs in each product were measured. A total of 36 PCPs, including 10 hair care products (e.g., shampoo), 9 dish detergents, 2 body soaps, 3 hand creams, and 12 perfumes, were collected. The perfumes were liquids, while the rest of the samples were creams and gels. The concentrations of the nine SMFs were determined by a combination of headspace-solid phase microextraction (HS-SPME) using a Combi-Pal auto-sampler (CTC Analytics AG, Zwingen, Switzerland) and gas chromatography-mass spectrometry (GC-MS) (QP2010 Plus and QP2020; Shimadzu Corporation, Tokyo, Japan). The SPME fiber was polydimethylsiloxane/divinylbenzen (PDMS/DVB) (65 μm) and was purchased from Supelco (Bellefonte, PA, USA). The fiber was pretreated according to the manufacturer's recommendations before use. A 20 mL screw cap vial sealed with PTFE-silicon was used; for liquid samples, 10 mL of the sample was placed in a vial to which 3.0 g NaCl was added. The vial was sealed and the sample was extracted at 90°C for 30 min at 500 rpm. Liquid perfume in the products was diluted with acetone and analyzed without pretreatment. In the case of cream-and gel-type PCPs, a small amount of sample was aliquoted into a headspace vial and analyzed by SPME-GC/MS. Separation of SMFs was achieved using a DB-1MS column (30 m × 0.25 mm × 0.25 μm: Agilent Technologies). The carrier gas was helium (purity 99.9999%) and the flow rate was 1.0 mL/min. The GC oven temperature was increased from 120°C (2 min) to 160°C (3 min) at a rate of 10°C/min, and then to 185°C 2.5°C/min and finally 250°C (3 min) at 30°C/min. The total analysis time was 24.17 min. The transfer line and ion source temperature of the mass spectrometer was set to 230oC. Thermal desorption on the SPME fiber was performed using an injector at 230°C in split mode, with a split ratio of 5.0. The mass spectrometer was operated in electron ionization (EI) mode at -70 eV. For identification of SMFs, the EI mass spectrum of the compound eluted from the GC column was compared to the EI mass spectra in the National Institute of Standards and Technology (NIST) v 2.2g (2014) database. The quality assurance and quality control (QA/QC) results are shown in Text S1 and Table S2, and the chromatograms and mass spectra of SMFs are shown in Table S3.

Synthetic musk fragrances in personal care products
First, an SMF analysis was performed on the PCPs. Table 2 shows the analysis results for 36 PCPs, including shampoo, detergent, and perfume. The most widely detected SMF was found to be HHCB, which was detected with a frequency of 86.1% among all PCP samples. It was most often detected in hair care products, with a maximum concentration of 4,917 μg/g (median, 395.8 μg/g) ( Table 2). HHCB was especially found in all hair care products and perfumes (100% frequency).
The second most frequently detected SMF after HHCB was AHTN, with a maximum concentration of 442.9 μg/g in 47.2% of all samples (Table 2). Even 80% of hair care products contained AHTN, the AHTN concentration in hair care products was much lower than that of HHCB, with a median value of 1.19 μg/g. The HHCB and AHTN concentrations in PCPs observed in this study were similar to those reported in previous studies in the United States [31], but much higher than those reported previously in China [23].
The most frequently detected SMF after HHCB and AHTN was MK, which was detected at a maximum concentration of 39.3 μg/g only in perfume and hair care products. Not all PCP products were found to contain MA and MX (Table 2). These results were consistent with the trends of domestic SMF usage reported in Korea, with significantly higher HHCB usage (60.6 tons/year) than AHTN (2.0 tons/year) and MK (11.6 tons/year) in 2006 [9]. Among other SMFs, ATII, AHMI, DPMI, and ADBI were observed with a frequency of less than 20%. In particular, DPMI was only detected at concentrations up to 21.9 μg/g in 5 out of 12 perfume products. Hair care products (n = 10) had the highest SMF content per unit weight (median value 455.5 μg/g), followed by perfume (n = 12; median value, 88.273 μg/g). Our result implies that the major sources of SMFs in household PCPs, especially HHCB, are hair care products and perfumes, these SMFs can go into sewage treatment plants through domestic sewages when showering or washing face at home.

Synthetic musk fragrances in STP influent
The SMF concentrations in the influent of the four STPs were between 3.532 and 11.080 μg/L (median, 6.756 μg/L) ( Table 3). Because more than 90% of the inflow water to all STPs was domestic sewage, there was no significant difference in SMF concentration between each STP. Five SMFs (HHCB, AHTN, MK, DPMI, and ATII) were most frequently detected in the sewage influent. The highest concentration (3.200 to 10.110 μg/L) was found for HHCB, with a median value of 6.130 μg/L. On a median basis, HHCB accounted for 90.7% of the total SMFs and was present in significantly higher concentrations than AHTN, MK, and DPMI ( Table 3). The AHTN concentration (median, 0.373 μg/L) was similar to that of MK (0.352 μg/L). The DPMI concentration was half that of AHTN and MK, with a median value of 0.166 μg/L. While ADBI was detected intermittently and at low levels, and, MA, and MX were not detected at all in STP influents.
The HHCB concentration in the STP influent in this study was higher than that previously reported in Italy [32], Spain [33,34], and France [19], and other studies in Korea [22] (Table S4). This might be due to the characteristics of the STPs, which were located in the densely populated city of Seoul, and most of the STP influents were therefore composed of domestic sewage. However, the AHTN concentration in the STP influent was similar to or lower than that reported in previous studies.
The HHCB/AHTN concentration ratio in the sewage influents was 16.4, which was higher than the previously reported values of 4.5 [22], 1.2 [25], and 3.2 [23], and similar to the value obtained by Tasselli and Guzzella [32]. This result implies that significantly higher amounts of HHCB than AHTN are used in PCPs in Korea. Since Table 2 shows that the HHCB concentration was significantly higher in hair care products (e.g., shampoo) and perfumes, these can be the major source of HHCB in sewage influent.  1. *N.D. indicates less than detection limit.

Synthetic musk fragrances in the STP effluents
The total SMF concentrations in the sewage effluent were in the range 0.949-5.880 μg/L, and the median value was 2.945 μg/L ( Table 3). The SMF detection frequencies in the STP effluent were similar to those in the STP influents. Among the SMFs, HHCB, AHTN, MK, DPMI, and ATII were detected in most of the samples, while ADBI, AHMI, MA, and MX were not detected. The SMF with the highest concentration was HHCB, with a median value of 2.546 μg/L, and a range of 0.701-5.390 μg/L. Based on the median value, HHCB accounted for almost 90% of the total SMFs, and its concentration was several tens of times higher than those of AHTN, MK, and DPMI.
The HHCB concentrations measured in this study were slightly higher than previously reported values in Germany [35], Thailand [26], USA [25], and China [23] (Table S4). The distribution of SMFs in the STP effluent was dependent on the SMF concentrations in the STP influent, removal efficiency, and chemical properties of each SMF. The relatively higher level of HHCB in the sewage effluent was a consequence of the high concentration in the sewage influent.
The average removal rate of SMFs in the four STPs using the MLE and A2O processes (Table  S2) was 58.5% for HHCB and 56.8% for AHTN. The HHCB and AHTN removal efficiencies were lower than the values reported in previous studies [20,21,35,36]. Table 3 also shows the concentrations of the nine SMFs (n = 153) in five mainstream sites of the Han River (H1-H5) and 17 tributaries (T1-T17), which were measured every month from March to November 2018. Various pollutants are released into the river as it passes through the densely populated area of Seoul. Following treatment in an STP, pollutants can flow into the mainstream and streams of the Han River in the form of STP effluent. In this study, five SMFs (HHCB, MK, AHTN, DPMI, and ATII) were detected in 17 tributaries; the order of both detection frequency and concentration was HHCB, MK, AHTN, DPMI, and ATII. At all sites, HHCB was detected in samples, but the concentration at each site varied from 0.017 to 2.825 μg/L, with a median concentration of 0.220 μg/L ( Table 3). As with the STP effluent, ADBI, AHMI, MA, and MX were not detected in the Han River samples.

Synthetic musk fragrances in the mainstream of the Han River
The median concentrations observed in the Han River were 0.105, 0.016, and 0.010 μg/L for HHCB, MK, and AHTN, respectively, indicating that HHCB was also the predominant SMF in river samples. The HHCB concentrations were higher, but the levels of AHTN were similar or lower than those reported in China [37] Singapore [38], and rivers in the Hesse region of Germany [39].
The distribution pattern of SMFs in the mainstream of the Han River was similar to that in the STP influent and effluent (Fig. 2). Interestingly, the similar HHCB concentration in the STP effluent compared to those reported in STPs elsewhere indicates that HHCB might be decomposed to a relatively greater extent than in other STPs (Fig. 2). Figure 3 shows the HHCB concentrations in samples taken from the STP influent, effluent, tributaries, and mainstream of the Han River. As shown in Fig. 3, the SMF concentrations, especially that of HHCB, were significantly decreased in the STP effluent compared to the influent, and there was then a gradual decrease from the tributaries to the mainstream of the Han River. This implied that effective SMF removal occurred during the sewage treatment process, which together with dilution in the river represented the two main removal mechanisms of HHCB.  Figure 4 shows the STP discharge volumes, distance to the nearest sampling site, and SMF concentration at each point. As shown in Fig. 4, of the 17 tributaries, a total of 7 sites were affected by the STPs, and had high SMF concentrations (Fig. 4). The sites with extremely high concentrations (T5, T13, T4, and T16) were located in the region affected by the STP effluent. We found that the SMF concentrations at T13 and H2 increased due to the effluent from S1, and the SMF concentration suddenly increased at point T4 due to the effluent from S2. At T5, a high HHCB concentration was observed due to the presence of two STPs upstream, but the concentration gradually decreased toward T7, which was probably a dilution effect, and then rapidly increased due to the discharge from S1. The results shown in Fig. 4 imply that the SMF concentrations in the tributaries were generally higher at sampling sites affected by STP effluent released from upstream locations.

Effect of STP effluent on SMF concentrations in the Han River
Finally, to examine the effect of sewage effluent discharge on the SMF concentration in the receiving water, two factors, i.e., the ratio of sewage discharge to the river flow rate (Qw/Qr) and the distance from the sewer outlet to the sampling point (∆L), were developed. While there was no correlation between SMF concentration and Qw/Qr, we found that there was a statistically strong relationship (r = 0.850, p <0.05) between the 1/∆L value and the SMF concentration in the river. This result implies that the closer the sampling point to the sewage outlet, the higher the SMF concentration in the tributaries or Han River water samples.

Conclusions
Nine SMFs were measured in the influents and effluents of four STPs, 17 tributaries, and five sampling sites in the Han River, and also in 36 different PCPs, including hair care products, perfumes, and detergents. In the PCPs, HHCB was the most abundant SMF, while AHTN, MK, AHMI, DPMI, ATII, and ADBI were detected with a frequency of less than 50%. Among the PCPs, hair care products had the highest SMF content per unit weight, with the highest concentration (median, 395.8 μg/g) being observed for HHCB. The SMF distribution among the four STP influents was similar to that of PCPs, with a median total SMF concentration of 6.756 μg/L and median HHCB concentration of 6.130 μg/L. The median MF concentration in the effluent was 2.945 μg/L, while the median HHCB concentration was 2.546 μg/L, and their distribution pattern was similar to that in the sewage influents, although the MK concentration was slightly higher (0.218 μg/L) than that of AHTN (0.161 μg/L). In the sewage treatment process, the removal efficiencies of HHCB, AHTN, and MK were 58.5%, 56.8%, and 38.1%, respectively. In the 17 tributaries and Han River mainstream, five SMFs were detected at levels similar to that in the STP effluent, but the concentrations decreased due to the dilution effect. The HHCB concentration was 0.220 μg/L in tributaries and 0.105 μg/L in the mainstream. The highest SMF concentrations in the tributaries were observed at locations close to the STPs, implying that sewage effluent was the main source of SMFs in the receiving water. The results of this study confirmed that the main source of SMFs in the receiving water were sewage effluent containing untreated SMFs, which largely originate from household PCPs, especially hair care products (e.g., shampoo) and perfumes.
Conflicts of Interest: The authors declare no conflict of interest.