Introduction

Metal sources in the environment can be traced using a variety of methods, including estimating spatiotemporal gradients of metal concentrations and relative enrichment factors, metal association and receptor models such as principal component analysis, and quantitative contribution analysis based on stable isotopic composition, which is a direct indicator of contaminant sources (Chen et al. 2022; Liu et al. 2020; Zhang et al. 2016; Zhong et al. 2021). Although these methods are complementary, a multiple isotope approach can provide integrated information about contaminant sources and transport pathways (Chen et al. 2022; Shiel et al. 2010; Šillerová et al. 2017; Souto-Oliveira et al. 2018; Wang et al. 2021). For example, Pb has four stable isotopes including non-radiogenic 204Pb and three radiogenic isotopes (206Pb, 207Pb, and 208Pb) that are formed through the radioactive decay of uranium and thorium isotopes (Dickin 1995). Therefore, various Pb sources may have unique isotopic compositions that allow the determination of natural or anthropogenic Pb emission source contributions in environmental medias such as sediments (Choi et al. 2007a, b; Townsend and Seen 2012), soils (Klaminder et al. 2011; Morton-Bermea et al. 2011), plants (Bindler et al. 2004; Weiss et al. 1999), and aerosols (Monna et al. 1997). Zn has five isotopes (64Zn, 66Zn, 67Zn, 68Zn, and 70Zn) that are subjected to isotopic fractionation during ore genesis (Sonke et al. 2008; Wilkinson et al. 2005), biogeochemical processes (Arnold et al. 2010; Jouvin et al. 2012; Juillot et al. 2008), and anthropogenic processes such as smelting (Shiel et al. 2010; Weiss et al. 2007), combustion (Mattielli et al. 2009; Ochoa Gonzalez and Weiss 2015), and electroplating (Araújo et al. 2017; Kavner et al. 2008; Tonhá et al. 2020), leading to isotopic fingerprints that allow differentiation from natural isotopic compositions (Yin et al. 2016). High-temperature processes such as ore roasting and smelting tend to fractionate Zn isotopes, with lighter isotopes predominating atmospheric emissions and heavier isotopes accumulating in slags and effluents (Borrok et al. 2010; Ochoa Gonzalez and Weiss 2015; Sonke et al. 2008; Yin et al. 2016). In contrast, high-temperature geological processes such as magmatic differentiation do not significantly fractionate Zn isotopes, resulting in homogeneous Zn isotopic composition in the bulk Earth (Chen et al. 2013). Zn released into the environment is also isotopically fractionated through biogeochemical processes such as adsorption, dissolution, and biological uptake (Arnold et al. 2010; John et al. 2007; Jouvin et al. 2012; Juillot et al. 2008). Thus, Zn isotopes can be used to trace the pathways and sources of contaminants derived from various sources such as smelting or mining, and a combination of isotopic information from different metals (e.g., Pb and Zn) can provide complementary information about contaminant sources and pathways that may be used to quantitatively estimate source contributions.

The upper reaches of the Nakdong River on the southeastern Korean peninsula feature many abandoned non-ferrous metal mines and an active Zn smelter. Because these operations have continued for over 60 years, related heavy metal contamination has been reported in soils (Jeon et al. 2017; Kang et al. 2019), river water (Kim et al. 2016), groundwater (Song et al. 2023), fish (Kwon et al. 2022), and river and lake sediments (Choi et al. 2015). This heavy metal contamination has emerged as a social problem, and several studies have been conducted to identify its sources. Zn concentrations and Pb isotopes have been used as indicators of smelter-generated materials in soils (Jeon et al. 2017; Kang et al. 2019). However, metal source contaminants in sediments have remained controversial. For example, chemical compositions in sediments and river water (Kim et al. 2016) and Pb isotope levels in lake sediments (Park et al. 2012) demonstrated that sediment metal composition was affected by mine tailings, whereas a contamination history of Zn, cadmium (Cd), and Pb in lake sediments showed a shift from mining- to smelting-derived metals as the dominant contaminants over time (Choi et al. 2015). Therefore, new indicators should be developed to discriminate metal contaminant sources and estimate the contribution of smelting-derived materials to river and lake sediments to improve the performance of environmental forensic investigations. In this study, we discriminated smelting and mining activities as sources of Pb and Zn contaminants in a segment of the Nakdong River, determined their pathways to the environment, and estimated the contribution of each source to river sediments according to Pb and Zn isotopic compositions.

Materials and methods

Study area

The study area (36° 43′ 29.3″–37° 5′ 42.3″ N, 128° 50′ 49.2″–129° 5′ 48.3″ E) was an upper segment of the Nakdong River on the southeastern Korean peninsula, approximately 100 km upstream from the entrance to Lake Andong, which is an artificial reservoir created by the construction of the Andong Dam in 1976 (Fig. 1). Approximately 100 mines were in operation within this drainage area during the 1970s and 1980s, but have been abandoned continuously since 1988. To date, 95 mines (55% metal, 39% non-metal, and 6% coal) have been abandoned (Kim et al. 2022). Among these, the Yeonhwa (508,000 tons Pb and 975,000 tons Zn) and Janggun mines (141,000 tons Pb and 264,000 tons Zn) produced the largest volume of Pb–Zn bulk concentrates in Korea, as well as approximately 2,000,000 m3 of tailings and 4,000,000 m3 of rock wastes, which accumulated near the mines and their adjacent rivers (Kim et al. 2022). In 1972, typhoon-related flooding led to a large-scale discharge of contaminants into the Nakdong River, followed by the deposition of 8,840 m3 of mine tailings along the 91 km river floodplain from a Zn smelter to the entrance of Lake Andong (Hong et al. 2016; Kim et al. 2016). Korea’s largest Zn smelter, established in 1970, is still operational, with an annual production capacity of 350,000 tons of Zn ingots through continuous facility expansion. The Zn smelter facilities include ore storage for Zn concentrates imported by truck and rail, as well as a wide exposed area where smelting waste is stored for later generation of by-products. The smelter once produced pure Zn from Zn concentrates derived from local mines; however, > 98% of the total Zn production now depends on imported Zn concentrates.

Fig. 1
figure 1

Map of river water and sediment sampling sites in the study area. A total of 31 water samples (18 main channels and 13 tributaries) and 58 sediment samples (45 main channels and 13 tributaries) were collected in August and May 2020. Smelter and abandoned metal mines are indicated by red rectangle box and open stars, respectively. The outlet of treated wastewater in the smelter, located about 0.50 km downstream from the Zn smelter plant 1 (N12 and W8), is represented by a red star

Full size image

Materials

River waters (18 main channels and 13 tributaries, W1-W31) and sediments (45 main channels and 13 tributaries, N1-N58) were collected from 10.2 km upstream to 90.3 km downstream of the Zn smelter in May and August 2020 (Fig. 1). River waters were collected using a pole sampler or horizontal Van Dorn sampler and then filtered on site into 125 mL acid-washed low-density polyethylene bottles using acid-washed syringe filters (pore size 0.45 μm) and syringes, and the filtrates were acidified with purified HNO3 and transported to the laboratory. To evaluate the possibility of metal-contaminated groundwater from the smelter directly into the river, river waters (n = 5) were also collected from a cross-section of the Nakdong River at station W8 (i.e., at the Zn smelter). Sediments were collected using a plastic scoop and grab sampler, and samples for grain size fraction and total organic carbon (TOC) analysis were wet-sieved through a 2 mm sieve, while samples for chemical analysis were sieved to 150 μm fractions, which is the standard operating procedure for reporting and measuring total heavy metal concentration in soil under the Korean Ministry of Environment. Samples of imported Zn concentrate (n = 7), smelting waste (n = 4), and soil around the Zn smelter (0–2 cm depth, n = 3), which are possible metal contaminants, were collected from the concentrate storage and smelting waste storage located inside plant 1 and from a hillside approximately 0.50 km from the smelter, respectively. For the Korean mine ores (n = 9) and riverside tailings (n = 7), samples were collected from six abandoned metal mines and from tailings deposited at the riverside of a main channel. All sediment and possible metal contaminant samples were transported to the laboratory, freeze-dried and powdered, and stored until chemical analysis.

Methods

Grain size fraction was calculated as the percentage of each fraction to dry weight by treating sediment samples (< 2 mm) with a 30% H2O2 solution and sieving through 150- and 63-μm sieves. TOC was determined using an automated elemental analyzer (Flash 2000 Series, Thermo Scientific, Waltham, MA, USA) after carbonates were removed using high purity water and H2SO3 (Nieuwenhuize et al. 1994; Verardo et al. 1990). Powdered sediment samples (< 150 μm) were determined for lithium (Li), Pb, Zn, and Cd concentrations through inductively coupled plasma mass spectrometry (ICP–MS, ELAN DRC II, PerkinElmer Inc.) after total digestion with mixed acids (HF:HNO3:HClO4, 4:4:1). Filtered and acidified river water samples were directly analyzed for Pb, Zn, and Cd using ICP–MS (ELAN DRC II). Certified reference materials including NRCC SLRS-6 and NIST SRM 1640a for water and USGS BCR-2, NRCC PACS-2, and NIST SRM 2711a for sediment were analyzed with samples together, with metal recoveries ranging from 85 to 110% (Supplementary Table 1 and 2).

Powdered sediment samples (200 mg) for Pb and Zn isotope analysis were weighed in low-pressure teflon digestion vessels (Savillex, Eden Prairie, MN, USA) and digested with 5 mL HF and 5 mL aqua regia at 180 °C for 24 h. After digestion, the strong acids were dried and residues were extracted with 1 M HCl, and then centrifuged at 3,000 rpm for 10 min, and the supernatants were collected. For Pb isotope analysis, the solutions were diluted with 0.5 M HNO3 to achieve a total dilution factor of 2,000. Solutions for Zn isotope analysis were transferred to perfluoroalkoxy alkane vials and evaporated at 120 °C on a hot plate; the residues were redissolved in 1 mL of 2 M HCl for Zn separation. For the Zn column, we used a poly prep column (Bio-Rad, Hercules, CA, USA) packed with 2 mL AGMP-1 resin (Bio-Rad; 100–200 mesh, chloride form). To improve the accuracy and precision of Zn isotope analysis, we minimized potential interference using a second column (Supplementary Table 3). The separated Zn fractions were then evaporated at 120 °C and dissolved in 0.5 M HNO3. Pb and Zn isotopes were analyzed using a multicollector ICP–MS (Neptune Plus, Thermo Scientific) at Korea Marine Environment Management Corp. (Busan, South Korea). To correct mass bias, external normalization method was applied to standards and samples (Chen et al. 2016; Maréchal et al. 1999; Mason et al. 2004). Prior to analysis, all samples were doped with thalium (NIST SRM 997) at a concentration of 10 ng/mL for Pb, and with Cu (ERM-AE 647) at a 1:1 Zn:Cu ratio and concentration of 50 ng/mL for Zn. To minimize concentration effects, the Zn concentrations of standards and samples were matched to within ± 5%.

Zn isotopic compositions are commonly reported using δ notation, which expresses deviation (‰) from the isotopic standard (IRMM 3702) as follows:

$${\delta }^{66}{Zn}_{IRMM3702}\left(\mathrm{\permille }\right)=\left(\frac{{\left(\frac{{}^{66}Zn}{{}^{64}Zn}\right)}_{sample}}{{\left(\frac{{}^{66}Zn}{{}^{64}Zn}\right)}_{IRMM3702}}-1\right)\times 1000$$
(1)

The external reproducibility of these methods was evaluated using NIST SRM 981 for Pb isotopes, and NIST SRM 682, USGS BCR-2 (basalt), NRCC PACS-2 (sediment), and NIST SRM 2711a (soil) for Zn isotopes. The NIST SRM 981 solution was measured once every five samples to ensure that the instrument’s measurement accuracy was maintained; we obtained Pb isotopic values of 207Pb/206Pb = 0.9147 ± 0.0001 and 208Pb/206Pb = 2.1668 ± 0.0004 (n = 75), which align with those reported previously (207Pb/206Pb = 0.9146 ± 0.0001 and 208Pb/206Pb = 2.1671 ± 0.0004; Todt et al. 1996). The repeatability of the Zn isotope analysis was verified by measurements on IRMM 3702 (δ66ZnIRMM3702 = 0.00 ± 0.02‰, n = 102) and NIST SRM 682 (δ66ZnIRMM3702 = –2.66 ± 0.12‰, n = 11); the results were within the analytical error of the certified values (John et al. 2007; Tanimizu et al. 2002). The δ66Zn values of three geological reference materials relative to IRMM 3702 are provided in Supplementary Table 4.

Results

Geochemical properties of river sediments

Grain size fractions and TOC contents in bulk sediments (< 2 mm), and Li concentrations in < 150 μm sediments sampled from 45 main channel sites and 13 tributary sites are presented in Supplementary Fig. 1. In the main channel, the average 2 mm to 150 μm, 150 to 63 μm and < 63 μm grain size fractions were 94.6 ± 7.05%, 5.43 ± 7.05% and 1.34 ± 2.90%, respectively. In tributaries, the average 2 mm to 150 μm grain size fraction was 96.3 ± 2.96%, and that for < 63 μm was 1.34 ± 1.59%. The average TOC content in main channel bulk sediments (< 2 mm) was 0.53 ± 0.47%, with no clear trend; its distribution was independent of the fine-grained (< 63 μm) sediment distribution (r = 0.22). However, the average TOC content in tributaries (0.33 ± 0.31%, n = 13) was significantly correlated with fine-grained sediment fraction (r = 0.83). Li concentrations in sediments (< 150 μm) sampled from the main channel (n = 45) and tributaries (n = 13) ranged from 11.1 to 90.2 mg/kg (average, 50.9 ± 13.8 mg/kg) and 37.2 to 87.0 mg/kg (average, 60.5 ± 14.9 mg/kg), respectively. No clear trend was observed for Li distribution, which was independent of the fine-grained sediment distribution (r = 0.31).

Metal concentrations in river waters and sediments

Dissolved Pb, Zn, and Cd concentrations in river waters collected in the main channel (n = 18) and tributaries (n = 13) are shown in Fig. 2a–c. From -9.73 to -3.11 km upstream of the smelter, dissolved metal concentrations were consistently very low (0.01 ± 0.01 μg/L for Pb, 5.94 ± 0.95 μg/L for Zn, and 0.03 ± 0.01 μg/L for Cd, n = 3). Zn and Cd concentrations increased rapidly from the smelter to downstream reaches, with maximum concentrations at stations W12 (5.18 km downstream of the smelter, Zn concentration, 123 μg/L) and W9 (1.75 km downstream of the smelter, Cd concentration, 3.33 μg/L), and then decreased exponentially further downstream. A slight Pb concentration increase was observed near the Zn smelter; however, Pb concentrations were generally very low (average, 0.04 ± 0.04 μg/L). In tributaries, Pb, Zn, and Cd concentrations were consistently very low, with averages of 0.01 ± 0.01, 4.38 ± 2.36, and 0.01 ± 0.02 μg/L, respectively, which were similar to those in the upper reaches. Dissolved metal concentrations in water sampled near the smelter (station W8, n = 5) were 15-fold higher for Cd and 2-fold higher for Zn on the Zn smelter side of the river cross-section than on the road side, whereas Pb concentrations were consistently very low (Supplementary Fig. 2). These results suggest that groundwater with very high concentrations of Cd and Zn flows directly from the interior of the Zn smelter to the main channel of the Nakdong River, which is consistent with the results of a groundwater leakage investigation that performed bromide and uranine tracer tests and groundwater modeling (Song et al. 2023).

Fig. 2
figure 2

Pb, Zn and Cd concentrations in river waters (a, b and c) and sediments (< 150 μm fraction; d, e and f). Each sampling location is presented along the relative distances from the Zn smelter plant 1 (W8 and N12, respectively), positive and negative for downstream and upstream, respectively

Full size image

Pb, Zn, and Cd concentrations in river sediments (< 150 μm) from the main channel (n = 45) and tributaries (n = 13) are shown in Fig. 2d–f and Supplementary Table 5. Pb, Zn, and Cd concentrations from -10.2 to -2.27 km upstream of the smelter were low and constant, with averages of 78.5 ± 25.0, 699 ± 188, and 3.18 ± 1.16 mg/kg (n = 5), respectively. From station N9 (-1.26 km upstream of the smelter), the concentrations of all three metals increased rapidly as distance to the Zn smelter decreased, reaching maximum concentrations at station N13 for Pb (443 mg/kg), N16 for Zn (7,360 mg/kg), and N19 for Cd (70.5 mg/kg). Downstream of the smelter, Pb concentrations decreased rapidly, and Zn and Cd maintained high concentrations for a distance of ~ 15 km and then decreased linearly to ~ 40 km and maintained low concentrations until the entrance of Lake Andong. Zn and Cd concentrations were much lower in the tributaries than in the main channel, with similar levels to the main channel upstream of the smelter. Pb concentrations were higher in upstream tributaries, but much lower than in the main channel near the smelter. Pb, Zn, and Cd distributions in river sediments (< 150 μm) differed greatly from those of the fine-grained fraction (r = –0.14, –0.20, and –0.22, respectively) and TOC content (r = 0.04, 0.13, and 0.07, respectively), such that high metal concentrations may have been caused by their direct supply from the Zn smelter, rather than compositional relationships such as preferential association of metals with fine-grained sediments and organic matter (Horowitz 1991).

207Pb/206Pb, 208Pb/206Pb, and δ66Zn in river sediments

Pb and Zn isotope ratios in main channel sediments (< 150 μm) varied greatly at the Zn smelter (Fig. 3) in a manner similar to their concentrations. From -10.2 to -2.27 km upstream of the smelter, the average Pb isotope ratios were 0.8322 ± 0.0067 (n = 5) for 207Pb/206Pb and 2.0502 ± 0.0190 (n = 5) for 208Pb/206Pb; the average δ66Zn value was 0.01 ± 0.03‰ (n = 4). From -1.26 km upstream (N9) to 14.8 km downstream (N22) of the smelter, Pb concentrations rapidly increased and then decreased (Fig. 2d), and Pb isotope ratios were consistently high, with averages of 0.8638 ± 0.0052 for 207Pb/206Pb and 2.0960 ± 0.0096 for 208Pb/206Pb (n = 13) in main channel sediments. At Seokporicheon (N10), a tributary near the Zn smelter, Pb isotope levels were also above average. At 47.1–90.3 km downstream of the smelter, Pb isotope ratios were low and highly variable, apparently influenced by nearby tributaries. Zn isotope ratios increased rapidly along with Zn concentrations at -1.26 km upstream of the smelter, and the highest δ66Zn value was observed at 8.59 km downstream of the smelter (N20, δ66Zn = 0.82‰). A slight decrease in Zn isotope ratios occurred from 14.8 to 90.3 km downstream of the smelter (δ66Zn = 0.56 ± 0.12‰), although Zn concentrations decreased exponentially. Interestingly, stations N9, N10, and N11 near the smelter had a δ66Zn value of –0.01 ± 0.00‰, despite their high Zn concentrations (average, 2720 ± 1210 mg/kg, n = 3). Tributary samples had highly variable Pb isotope ratios, with averages of 0.8180 ± 0.0430 for 207Pb/206Pb and 2.0353 ± 0.0953 for 208Pb/206Pb (n = 12). The average δ66Zn was 0.00 ± 0.08‰ (n = 7) from -10.2 km upstream to 53.0 km downstream of the smelter, excluding the Seokporicheon tributary (N10) and 53.0–90.3 km downstream of the smelter, with an average of –0.32 ± 0.06‰ (n = 5).

Fig. 3
figure 3

Pb (a and b) and Zn (c) isotope ratios of the river sediments (< 150 μm fraction). Each sampling location is presented along the relative distances from the Zn smelter plant 1 (N12), positive and negative for downstream and upstream, respectively

Full size image

Pb and Zn isotope fingerprints of sources

Pb and Zn isotopic compositions of potential metal contaminant source materials including imported Zn concentrates, Korean mining ores, smelting wastes, and dust-contaminated soils sampled near the Zn smelter are shown in Fig. 4 and Supplementary Table 6. Imported Zn concentrates (Zn concentration > 40%) used in the Zn smelter had a wide range of Pb isotopic compositions (0.7559–0.9629 for 207Pb/206Pb and 1.8992–2.2367 for 208Pb/206Pb, n = 36; Kang et al. 2019), and many Zn concentrates lie on the worldwide growth curve of Pb ores such as galena in Australia, Peru, Mexico and USA (Cumming and Richards 1975). The Pb isotopic compositions of Korean ores (0.7827–0.8625 for 207Pb/206Pb and 1.9319–2.1464 208Pb/206Pb, n = 98; Jeong et al. 2012; Kang et al. 2019; Park and Chang 2005) and riverside tailings (0.8133–0.8200 for 207Pb/206Pb and 2.0108–2.0233 for 208Pb/206Pb, n = 8; NIER, 2011) are situated along the Korea ore line. The difference between these ore curves was consistent with previous studies that showed higher 208Pb levels in Pb–Zn deposits on the Korean peninsula than indicated by a Pb growth curve (Choi et al. 2007a, b; Mukai et al. 1993). Pb isotopic compositions of smelting wastes (0.8712–0.9053 for 207Pb/206Pb and 2.1120–2.1565 for 208Pb/206Pb, n = 17; MOE, 2016) and residential roof dusts and dust-contaminated soils (0.8590–0.9014 for 207Pb/206Pb and 2.1030–2.1525 for 208Pb/206Pb, n = 20; Kang et al. 2019; MOE, 2016) were similar to the mean values (0.8756 ± 0.0538 for 207Pb/206Pb and 2.1211 ± 0.0739 for 208Pb/206Pb) of imported Zn concentrates (n = 36) used in the Zn smelter.

Fig. 4
figure 4

207Pb/206Pb vs. 208Pb/206Pb ratios of the river sediments. Pb contaminants including global lead ores (Kang et al. 2019), imported Zn concentrates (Kang et al. 2019; this study), Korean ores (Ores from Yeonhwa, Janggun and other mines; Park and Chang 2005; Jeong et al. 2012; Kang et al. 2019; this study), smelting wastes (MOE, 2016; this study), residential roof dusts (Kang et al. 2019), dust-contaminated soils (this study) and riverside tailings (NIER 2011; this study) are indicated

Full size image

The Zn isotopic compositions of imported Zn concentrates were divided into two groups, with δ66Zn of –0.18 ± 0.04‰ (n = 5) and –0.51 ± 0.02‰ (n = 2), respectively (Supplementary Fig. 3). The former value is similar to the global median for Zn sulfide (ZnS) ores, which have an interquartile range (IQR) of –0.35 to –0.08‰ and a median of –0.17‰ (n = 97) (Aranda et al. 2012; Deng et al. 2017; Duan et al. 2016; Gagnevin et al. 2012; John et al. 2008; Kelley et al. 2009; Maréchal et al. 1999; Mason et al. 2005; Mondillo et al. 2018; Pašava et al. 2014; Sivry et al. 2008; Skierszkan et al. 2016; Sonke et al. 2008; Zhu et al. 2018). Although the sample size of the latter group was very small, its value was similar to the mean for Korean mining ores (δ66Zn = –0.55 ± 0.08‰, n = 9). This difference may be attributed to variation in the formation mechanisms of ZnS ores in different Pb–Zn deposits (Zhang et al. 2022). The Zn isotopic compositions of riverside tailings were also divided into two groups, with δ66Zn values of –0.09 ± 0.10‰ (n = 5) and –0.49 ± 0.04‰ (n = 2). The former value is similar to that of bulk silicate earth (BSE) (–0.04 ± 0.05‰ of δ66Zn; Chen et al. 2013) and the latter is similar to the average δ66Zn values of Korean mining ores (Supplementary Fig. 3). Dust-contaminated soils collected near the Zn smelter, had very high Zn concentrations (2900 ± 2260 mg/kg, n = 3) and low δ66Zn (–0.18 ± 0.08‰, n = 3). Smelting wastes had highly variable δ66Zn values (0.73 ± 0.43‰, n = 4), indicating wide variation in Zn isotope fractionation during smelting. However, these values were consistent with the range of δ66Zn values for smelting sludge and effluents reported in the literature, with an IQR of 0.51–1.18‰ and median of 0.69‰ (n = 16; Desaulty and Petelet-Giraud 2020).

Discussion

Identification of Pb and Zn sources and pathways

To confirm Pb and Zn sources, the relationships between metal isotope ratios and the reciprocal of metal concentrations were shown in Fig. 5. For Pb (Fig. 5a and Supplementary Fig. 4), sediments near the Zn smelter (-1.26 km upstream to 26.3 km downstream, n = 17) showed a strong linear relationship with the Li/Pb ratio (Y = –0.09X + 0.88, r2 = 0.88 for 207Pb/206Pb and Y = –0.16X + 2.12, r2 = 0.86 for 208Pb/206Pb), indicating the mixing of two sources. The Y-intercept values indicated Pb isotopic compositions consistent with Zn smelter materials, including imported Zn concentrates, smelting wastes, residential roof dusts, and dust-contaminated soils. However, Pb leaching from smelting wastes to groundwater was ruled out as a source of Pb contamination due to negligible transport of dissolved Pb (Fig. 2a), and discharges of imported Zn concentrates and smelting wastes appear unlikely given the management practices at the Zn smelter (MOE 2016). Thus, atmospheric deposition of smelting dust appears to be a main pathway of Pb supply from the smelter to the river, and since soils within a radius of 2 km from this smelter were severely contaminated by dust emitted from the smelter (Kang et al. 2019), runoff of dust-contaminated soils was considered to be the major source of Pb. According to Fig. 5a, the sediment baseline levels were determined to be 57.6 ± 12.7 mg/kg for Pb concentration, 0.7971 ± 0.0126 for 207Pb/206Pb, and 1.9775 ± 0.0127 for 208Pb/206Pb (n = 10) (Supplementary Fig. 4). Thus, baseline levels were detected in sediments from 26.3 to 44.0 km downstream of the smelter, indicating little influence of Zn smelter materials. However, Pb concentrations were approximately 2-fold higher in the baseline sediments than in the background soils (20.6 ± 4.70 mg/kg for Pb, 0.7794 ± 0.0431 for 207Pb/206Pb, and 1.9385 ± 0.1224 for 208Pb/206Pb; Kang et al. 2019), and their isotope ratios were less radiogenic, so the baseline sediments were defined as contaminated by the riverside tailings, which were frequently observed along the main channel of the river. Metal contamination of river sediments by mine tailings was similar to cases reported at the Sasa mine in North Macedonia, the Mount Polley gold and copper mine in British Columbia-Canada, and an iron mine in southeastern Brazil (Byrne et al. 2018; Duarte et al. 2021; Vrhovnik et al. 2013). Although the Pb isotopic compositions of sediments in tributaries were also plotted along the Korean ore line (Fig. 4), there was a difference in Pb isotopic compositions between tributaries upstream (10.2–45.3 km) and downstream (45.3–90.3 km), which were presumably caused by regional differences in the bedrock and sedimentary rock of the study area (Park and Chang 2005; Supplementary Fig. 5).

Fig. 5
figure 5

207Pb/206Pb vs. Li/Pb ratios (a) and δ66Zn vs. Li/Zn ratios (b) of the river sediments. Error bars represent the 2σ associated with 207Pb/206Pb ratios and δ66Zn values. In figure (a), Pb contaminants including imported Zn concentrates (Kang et al. 2019; this study), smelting wastes (MOE, 2016; this study), residential roof dusts (Kang et al. 2019) and dust-contaminated soils (this study) are indicated, and baseline sediments were also designated. In figure (b), the gray line labeled "BSE" represents the mean of bulk silicate earth (δ66Zn –0.04 ± 0.05‰; Chen et al. 2013). Zn contaminants are shown with a purple box plot for dust-contaminated soils (δ66Zn –0.18 ± 0.08‰, n = 3) and a dark brown box plot for the estimated values (δ66Zn 1.02 ± 0.43‰, n = 4) of Zn adsorbed from groundwater originating from smelting wastes (FeOOH: Δ66Znsolid-solution ~ 0.29‰; Juillot et al. 2008)

Full size image

For Zn, sediments near the Zn smelter (-0.55 km upstream to 26.3 km downstream, n = 13) had very high Zn concentrations (Li/Zn < 0.02), with a wide range of δ66Zn values (–0.01 to 0.82‰) (Fig. 5b). The linear trend in these samples indicates mixtures of two or more contaminant sources: dust-contaminated soils, groundwater contaminated through leaching of stored wastes, and tailings from nearby abandoned mines; these had lighter, heavier, and unfractionated Zn isotopes from BSE (–0.04 ± 0.05‰ of δ66Zn; Chen et al. 2013), respectively. As observed for Pb, dust-contaminated soil (–0.18 ± 0.08‰, n = 3) was a source of contamination, which is similar to a case of metal contamination in river sediments due to contaminated soil wash-out around a smelter in the Baiyin region of China (Zhao et al. 2020). The flow of groundwater contaminated with smelter-derived metals into the main channel of the Nakdong River (MOE 2020a; Song et al. 2023; Supplementary Fig. 2) indicates that Zn derived from groundwater containing metals leached from smelting wastes was adsorbed onto sediments. Baseline sediments (Zn concentration = 476 ± 255 mg/kg, δ66Zn = 0.01 ± 0.07‰, n = 11, blue square and circle symbols in Fig. 5b, excluding Seokporicheon tributary) containing riverside tailings are also an important contamination source. Thus, Zn contaminants are composed of a mixture of three components: dust-contaminated soils, groundwater-derived Zn, and baseline sediments. Sediments from 26.3 to 90.3 km downstream of the smelter contained a minimal contribution from dust-contaminated soils, as shown by the lack of a Pb signal. Thus, groundwater-derived metals were an important Zn source in this zone. This result was similar to those of other studies showing that fluvial dispersion of contaminants from Zn smelter was detected over much greater distances than the effects of atmospheric dispersion (Foucher et al. 2009; Žibret and Šajn 2008).

Estimation of source contributions

If the major metal contaminant sources and isotope ratios are identified, the contributions of two or three components can be quantitatively estimated using mixing equations (Zhang et al. 2016). Using this method, the Pb signals from -1.26 km upstream to 26.3 km downstream of the smelter was interpreted as a mixture of Pb from dust-contaminated soils and baseline sediments. Thus, the quantitative contribution of each Pb source in this zone was estimated as follows:

$${f}_{1} + {f}_{2} = 1$$
(2)
$${f}_{1} \times {R}_{1} + {f}_{2} \times {R}_{2}= {R}_{S}$$
(3)

where f1 and f2 are the contributions of baseline sediments and dust-contaminated soils, respectively; R1 and R2 are the 207Pb/206Pb in baseline sediments (0.7971) and dust-contaminated soils (0.8829), respectively; and Rs is the 207Pb/206Pb in sediment samples. Pb from dust-contaminated soils contributed 80 ± 6% (n = 7) from -1.26 km upstream to 2.90 km downstream of the smelter, and 81% in the Seokporicheon tributary (-0.68 km upstream of the smelter, N10). The contribution of dust-contaminated soils decreased to negligible levels in baseline sediments 26.3–44.0 km downstream of the smelter. In contrast, main channel sediments from 44.0 to 90.3 km downstream of the smelter had low Pb concentrations (49.7 ± 15.9 mg/kg, n = 14; Fig. 2d) that were similar to those of baseline sediments, despite high variation in Pb isotope ratios along the Korean ore line, indicating a negligible contribution of Zn smelter-derived materials.

For Zn, sediments near the smelter (-0.55 km upstream to 26.3 km downstream, n = 13) were interpreted as a mixture of dust-contaminated soils (DCS), groundwater-derived Zn (GDZ), and baseline sediments (BS) (Fig. 5b). The Zn isotopic composition of the baseline sediments was divided into two groups, with δ66Zn of 0.01 ± 0.07‰ (n = 10) and –0.32 ± 0.06‰ (n = 5), respectively. The former, main channel sediments from -10.2 to -2.27 km upstream of the smelter (n = 4) and tributary sediments from -10.2 km upstream to 53.0 km downstream of the smelter (n = 7, excluding Seokporichon tributary sediment) were defined as BS1 due to Zn isotopic compositions similar to bulk silicate earth (–0.04 ± 0.05‰, Chen et al 2013). The latter, tributary sediments from 53.0 to 90.3 km downstream of the smelter (n = 5), were defined as BS2 due to lighter Zn isotopic compositions similar to Korean mine ores (–0.55 ± 0.08‰, n = 9) and riverside tailings (–0.49 ± 0.04‰, n = 2). The influence of DCS was apparent near the Zn smelter (δ66Zn = –0.18 ± 0.08‰, n = 3). GDZ ranges were influenced by isotopic fractionation during Zn leaching from smelting wastes and groundwater Zn adsorption onto Fe-carriers in river sediments. Although Zn released into the environment typically undergoes little isotopic fractionation during dissolution (Δ66Znsolution-solid =  ~ 0‰; Desaulty and Petelet-Giraud 2020), isotopic fractionation during adsorption onto iron oxyhydroxide (FeOOH) can reach ~ 0.29‰ (Juillot et al. 2008). Based on a recent study of heavy mineral assemblages, the δ66Zn endmember of GDZ from smelting wastes was estimated to be 1.02 ± 0.43‰ (n = 4), as goethite (FeOOH) was abundant in the main channel and tributary sediments examined in this study (MOE 2020b).

Thus, the contribution of each Zn source from -2.27 km upstream to 90.3 km downstream of the smelter was estimated as follows (Zhang et al. 2016):

$${f}_{1} + {f}_{2} + {f}_{3} = 1$$
(4)
$${f}_{1} \times {R}_{1} + {f}_{2} \times {R}_{2} + {f}_{3} \times {R}_{3} = {R}_{S}$$
(5)
$$\frac{{f}_{1} \times {C}_{S}}{{C}_{1}}+ \frac{{f}_{2} \times {C}_{S}}{{C}_{2}}+ \frac{{f}_{3} \times {C}_{S}}{{C}_{3}}=1$$
(6)

where f1, f2,and f3 are the contributions of BS1, GDZ, and DCS, respectively, from -2.27 km upstream to 53.0 km downstream of the smelter, and those of BS1, GDZ, and BS2 from 53.0 to 90.3 km, respectively, downstream of the smelter. R1, R2, and R3 are the δ66Zn values of BS1 (average, 0.01‰, n = 10), GDZ (average, 1.02‰, n = 4), DCS (average, –0.18‰, n = 3), respectively, in the former zone, and BS2 (average, –0.32‰, n = 5) instead of DCS in the latter zone. C1, C2, and C3 are the Zn concentrations in BS1 (average, 476 mg/kg, n = 11), GDZ (estimates 10.000 mg/kg), and DCS (average, 2900 mg/kg, n = 3), respectively, in the former zone, and BS2 (average, 206 mg/kg, n = 5) instead of DCS in the latter zone. Cs and Rs are the measured Zn concentration and δ66Zn in main channel sediments. Based on the endmember mixing model, BS1 contributed an average of 25 ± 23% throughout the entire study area, while Zn source materials (DCS + GDZ) from Zn smelter contributed 75 ± 24% (Fig. 6). Of the two source materials from the Zn smelter, the DCS contributed the most (64 ± 16%) within a 2 km radius of the smelter (-2.27 km upstream to 1.96 km downstream), with a 33% contribution into the Seokporicheon tributary sediment (N10) at -0.68 km upstream of the smelter. In contrast, the GDZ contributed 55 ± 15% from -0.55 km upstream to 90.3 km downstream of the smelter and approximately 40% at the entrance of Lake Andong (N58). BS2 from tributaries from 53.0 to 90.3 km downstream of the smelter contributed 11% only at 73.3 km downstream of the smelter (N53). Thus, our heavy metal concentration and isotopic composition analyses estimated the relative contributions of Pb and Zn contaminant source materials, which allowed us to discriminate the effects of Zn smelting from those of mining tailings, groundwater leaching, and dust-contaminated soils.

Fig. 6
figure 6

Contributions (%) of four Zn sources in the main channel sediments. BS 1, BS 2, GDZ, and DCS represent baseline sediment 1 (main channels and tributaries from -10.2 to -0.55 km and -10.2 to 53.0 km, respectively), baseline sediment 2 (tributaries from 53.0 to 90.3 km), groundwater-derived Zn, and dust contaminated soils, respectively

Full size image

Conclusion

This study tried to discriminate the sources and pathways of contaminated Pb and Zn in river sediments influenced by both mining and smelting activities, and to estimate the contribution of each source using stable metal isotopes. Dissolved Zn and Cd concentrations in river waters near the Zn smelter were very high compared to those upstream of the smelter and tributaries, and a cross section near the smelter also showed that the smelter side had much higher concentrations than the road side. This indicates that groundwater with very high Zn and Cd concentrations is discharged into the river. The Pb isotopic compositions of sediments with extremely high Pb concentrations near the smelter were differentiated with those of ores and tailings originating from local mines, and were similar to those of smelter-derived materials such as imported Zn concentrates, residential roof dusts, smelting wastes, and dust-contaminated soils. From the mixing relationships using Pb isotope ratios and Pb concentrations, dust-contaminated soils (DCS) and baseline sediments (BS: tailing contaminated sediments) were regarded as two main contributors of Pb. The δ66Zn values in the sediments near the Zn smelter showed an abrupt increase from –0.01 to 0.82‰, which indicate that there are three Zn sources; lighter, heavier and unfractionated Zn such as dust-contamminated soils (DCS, –0.18‰ of δ66Zn), groundwater derived Zn (GDZ, 1.02‰ of δ66Zn) by leaching of smelting wastes stored inside the smelter, and tailings contaminated sediments (BS, 0.01‰ of δ66Zn) from nearby abandoned mines, respectively. The contribution of each source for Pb and Zn in sediments was estimated using two (Pb) and three (Zn) components mixing models. In sediments within 2 km of the Zn smelter, DCS contributed about 78% and 64% for Pb and Zn, respectively, which decreased to negligible levels after 47.1 km downstream. GDZ, another smelter derived Zn, contributed about 54% of total Zn in main channel sediments downstream of the smelter, and contributed over a longer distances than DCS. This is because parts of GDZ were transported in the dissolved phase while DCS was supplied into the river only in the particulate phase. BS for Pb and Zn contributed lowest within 2 km of the smelter and increased from 47.1 km to the entrance of Lake Andong. Environmental forensic studies using metal stable isotopes could be used for the development of environmental management plans in the catchment, and for the assessment for the impacts of mining and smelting activities on the Lake Andong.