Hydrogeochemical and Mineralogical Constraints on the Origin of Elevated Uranium and Radon in a Fractured Granitic Aquifer

Abstract

This study investigated the hydrogeochemical and mineralogical factors governing the occurrence of U and Rn in a fractured granitic aquifer. A 120-m-deep borehole was drilled, and depth-discrete groundwater samples were collected during wet and dry seasons for analyses of water quality and naturally occurring radioactive materials; petrographic and mineralogical analyses were conducted on the recovered drill core. U and Rn concentrations reached 488.1 μg L⁻¹ and 1549.8 Bq L⁻¹, respectively, exceeding Korean drinking-water criteria. Both generally increased with depth and were lower in the dry season than in the wet season. Oxic, mildly alkaline groundwater with elevated HCO₃⁻ favored U mobilization through uranyl–carbonate complexation, whereas Rn was primarily controlled by fracture connectivity and mixing between shallow and deep groundwater. Microfractures and pores in metamict U–Th-bearing minerals provided reactive surfaces and release pathways that promoted U leaching and Rn emanation. The occurrence of U and Rn therefore reflects the coupled effects of hydrogeochemical conditions, fracture development, groundwater mixing, and the susceptibility of radioactive minerals to alteration. These findings provide a mechanistic basis for interpreting and predicting naturally occurring radionuclide behavior in fractured granitic aquifers.

1. Introduction

Naturally occurring radioactive materials are hosted by minerals in the Earth’s crust and may pose a potential health concern where groundwater is used as drinking water. Uranium is a radionuclide present throughout natural environments and in numerous industrial activities, and human exposure may occur through dissolved U in groundwater. Its principal health concern is generally associated with chemical toxicity rather than radiotoxicity, particularly nephrotoxicity and accumulation in the skeletal system [1,2]. Radon is a radioactive noble gas whose short-lived progeny emit alpha particles after release to air. Inhalation of residential radon is an established cause of lung cancer, whereas reported associations with childhood leukemia remain less certain because of heterogeneity among epidemiological studies [3]. Ingestion of radon-bearing water contributes to internal radiation dose, although epidemiological evidence for specific gastrointestinal cancers is limited [4]. In Korea, the drinking-water standard for U is 30 μg L⁻¹, and the monitoring level for Rn in drinking water is 148 Bq L⁻¹.

Uranium contamination of groundwater is a globally recognized environmental issue. Elevated concentrations have been reported in India, China, the United States, Pakistan, South Korea, and other countries, where geological conditions, mining activities, and agricultural fertilizer use can produce U concentrations of tens to hundreds of micrograms per liter [5]. Elevated groundwater Rn is also widely reported, including activities of several thousand Bq L⁻¹ in granitic regions of China, Germany, and Spain. Exceptionally high groundwater Rn activities of up to 77,000 Bq L⁻¹ have been documented in parts of Finland, where U-rich crystalline bedrock coincides with fractured granitic geology [6]. In Ad Dali Governorate, Yemen, activities of up to 896 Bq L⁻¹ were attributed to U-rich basement rocks and fault structures that facilitate Rn migration [7].

In Korea, groundwater U and Rn concentrations are reported to be relatively high in Precambrian metamorphic rocks and Jurassic granites [8]. Precambrian metamorphic rocks and Jurassic granites account for approximately 25.0% and 20.7%, respectively, of the geology exposed across South Korea. Their broad distribution provides a regional geological setting in which groundwater containing naturally occurring radioactive materials may be encountered.

Uranium in groundwater may originate from natural water–rock interactions or anthropogenic contamination. Anthropogenic sources include U mining, fertilizer application, and inappropriate waste disposal. Geogenic U is derived from U-bearing minerals in aquifer rocks. In granitic rocks, U occurs in primary U minerals such as uraninite (UO₂) and coffinite (USiO₄), and as a substituent in accessory minerals including zircon (ZrSiO₄), xenotime (YPO₄), allanite ((Ce, REE)(Al, Fe)₃(SiO₄)₃(OH)), and thorite (ThSiO₄). In Korea, high-U groundwater is especially associated with Jurassic granites of the Gyeonggi Massif and Ogcheon Belt, which commonly reflect crustally derived, highly differentiated magmas [9]. The principal factors promoting U release from bedrock are neutral to mildly alkaline pH, oxidation redox conditions, and elevated HCO₃⁻ concentrations [10]. Experimental work further indicates that HCO₃⁻ can be the dominant control on the rate of U leaching [11]. Nitrate may also covary positively with U because it reflects oxidation conditions and, in some settings, recharge influenced by agricultural activity [12]. In reducing environments, U is predominantly retained as U(IV), and its transport is generally restricted to sub-centimeter scales. Under oxidation conditions, however, oxidation to U(VI) produces the uranyl ion (UO₂²⁺), which forms stable uranyl–carbonate complexes such as UO₂(CO₃)₂²⁻ and UO₂(CO₃)₃⁴⁻ and may migrate over much greater distances [13].

Radon is a gaseous radionuclide generated by decay of ²²⁶Ra within the ²³⁸U decay series. Alpha-recoil energy produced during the transformation of ²²⁶Ra to ²²²Rn is the principal mechanism ejecting Rn atoms from mineral lattices. The recoil range within a mineral grain is only approximately 20–70 nm, depending on mineral density and composition. Consequently, Rn emanation increases where mineral grains are fine and have high specific surface area, or where microfractures and pores intersect Ra-bearing domains [14].

Previous studies in fractured granitic aquifers have attributed U and Rn occurrence to three groups of factors. First, hydrogeochemical conditions primarily control U rather than Rn. In Jurassic granite groundwater of the Icheon area, U correlated with Eh, DO, and HCO₃⁻, whereas Rn was largely independent of groundwater chemistry and showed no significant correlation with U [15]. Second, fracture architecture and groundwater mixing influence both nuclides. Fractures provide reactive surfaces and flow paths, and U, Th, and Ra accumulated on fracture surfaces are major sources of dissolved radionuclides and Rn [16,17]. In addition, seasonal recharge of shallow groundwater can dilute bedrock-derived U and Rn, producing temporal variations in their concentrations [18,19]. Third, the type and alteration state of the host minerals determine how readily U and Rn are released. Differences in U-host minerals and their solubility have been proposed to explain contrasting dissolved U and Rn between Jurassic and Cretaceous granite aquifers in Korea; for example, low-solubility monazite can supply Rn but little U [20]. Altered radioactive accessory minerals have also been identified as radionuclide sources in Mesozoic granite aquifers [21].

Several Korean borehole studies combining depth-specific groundwater sampling with drill-core analyses reported marked depth-dependent and temporal variations in U and Rn, with little correlation between the two [22,23,24,25]. In the Icheon area, U and Th occurred mainly as substituents in accessory minerals such as monazite, apatite, and ilmenite. Because the U content of the drill core was not particularly high, physical factors such as fracture intensity and groundwater flow were inferred to be more important than geochemical conditions [24]. However, few studies have examined the alteration state of U–Th-bearing minerals together with depth- and time-resolved groundwater data from the same borehole. It therefore remains unclear how source minerals, fracture-controlled mixing, and seasonal changes in groundwater chemistry together determine U and Rn concentrations.

In this study, we investigated an area where high U and Rn concentrations had been repeatedly detected. A 120-m-deep borehole was established, and groundwater was sampled at 11 depths at 10-m intervals during two wet-season and two dry-season campaigns to construct depth- and time-resolved hydrochemical profiles. These data were integrated with core logging, rock quality designation, and electron-microscopic characterization of U–Th-bearing accessory minerals, including their metamictization, microfracturing, and hydrothermal alteration, from the same borehole. The objectives were to (i) identify the hydrogeochemical conditions controlling U mobility, (ii) evaluate the influence of fracture development and seasonal mixing on U and Rn distributions, and (iii) determine the mineralogical sources and release mechanisms of U and Rn.

2. Materials and Methods

2.1. The Study Area

The study area is located in the central inland Korean Peninsula, close to the boundary between the Gyeonggi Massif (GM) and the Ogcheon Metamorphic Belt (OB) (Figure 1). The Keumwang Fault, which crosses the study area, is a sinistral strike-slip structure that records both a large ductile shear zone and later brittle deformation. The study basin is interpreted as a pull-apart basin formed in association with movement along this fault [26]. The structural setting has produced extensive rock fracturing and locally intense hydrothermal alteration [27]. Basin lithologies include clastic sedimentary rocks such as mudstone, sandstone, and conglomerate; basaltic, andesitic, and rhyolitic volcanic rocks; Precambrian banded gneiss and granitic gneiss; and, most prominently, Jurassic intrusive granites [28,29].

Figure 1. Geological map of the study area, modified from geological maps published by the Korea Institute of Geoscience and Mineral Resources (KIGAM). The lower-left inset shows the location of the study area on the Korean Peninsula. GM = Gyeonggi Massif; GB = Gyeongsang Basin; TB = Taebaeksan Basin; YM = Yeongnam Massif; OB = Ogcheon Metamorphic Belt. The circle indicates the study area.

Previous work in the basin reported groundwater pH values of 5.79–8.97 and electrical conductivity (EC) values of 111–1214 μS cm⁻¹. Ca²⁺, Na⁺, and HCO₃⁻ are the dominant dissolved constituents. This hydrochemical composition was interpreted to reflect weathering of feldspar-rich granite and gneiss, together with progressive water–rock interaction [30].

2.2. Analysis of Groundwater

Groundwater was sampled during four campaigns to evaluate depth- and time-dependent changes in naturally occurring radioactive materials and water-quality parameters: the first wet-season campaign (July 2025; WS1), the second wet-season campaign (September 2025; WS2), the first dry-season campaign (October 2025; DS1), and the second dry-season campaign (November 2025; DS2). Following drilling of the 120-m borehole, samples were collected from 11 discrete depths at 10-m intervals beginning at 15 m.

Before each sampling campaign, the borehole was purged to remove stagnant water in the borehole column, whose chemistry can differ from aquifer groundwater because of contact with the atmosphere and the degassing and decay of Rn. Purging was conducted with a submersible pump installed at a depth of 40 m at approximately 375 L h⁻¹ for more than 5 h, removing approximately 1.9 m³ of water, equivalent to about five well volumes. Drawdown during purging was not measured; however, after a stabilization period of more than 6 h, the water level had fully recovered to its pre-purging level. This stabilization period was intended to allow the water column to re-equilibrate with inflow from the surrounding fractures before depth-discrete sampling. The sampling was conducted by following the sampling procedure outlined in the Korean standard method (ES 04731) [31], and purging when each item stabilized as follows: pH ± 0.2, EC ± 3%, T ± 0.2 °C, Eh ± 10 mV, and DO ± 10%. No pumping was conducted in the borehole between the four sampling campaigns. Depth-discrete samples were then collected sequentially from 15 to 115 m at 10-m intervals using a pressurized discrete interval sampler without packers (Model 425, Solinst, Canada Ltd., Georgetown, ON, Canada). The sampler was pressurized with air at the surface to keep its inlet closed during descent, and the pressure was released at the target depth to allow groundwater to enter. The sampler was then re-pressurized to seal the sample before retrieval, preventing mixing with water from other depths during recovery. Particular care was taken to minimize aeration and Rn loss during collection. Groundwater level was monitored continuously at 5-min intervals throughout the study using a TD-Diver (DI805, Van Essen Instruments, Delft, The Netherlands).

Field parameters, including water temperature, pH, EC, oxidation–reduction potential (Eh), and dissolved oxygen (DO), were measured using a multiparameter water-quality meter (ProQuatro, YSI Inc., Yellow Spring, OH, USA). The instrument was calibrated with standard solutions before each survey following the manufacturer’s instructions. EC was automatically compensated to 25 °C and is reported as specific conductance (SPC). Alkalinity was determined by acid titration using 0.05 N HCl.

All groundwater samples for laboratory analysis were filtered in the field through 0.45-μm membrane filters. Samples for cation and U analyses were acidified to pH < 2 with concentrated HNO₃ to prevent precipitation and stored at ≤4 °C until analysis [32]. Uranium was quantified by inductively coupled plasma mass spectrometry (ICP–MS; NexION 2000, PerkinElmer, Shelton, CT, USA). Cations and dissolved Si (Na⁺, Mg²⁺, K⁺, Ca²⁺, and Si) were measured by inductively coupled plasma optical emission spectrometry (ICP–OES; iCAP PRO, Thermo Fisher Scientific, Waltham, MA, USA), and anions (F⁻, Cl⁻, SO₄²⁻, NO₂⁻, NO₃⁻, and PO₄³⁻) were measured by ion chromatography (IC; DIONEX AQUION, Thermo Fisher Scientific, Waltham, MA, USA). For Rn analysis, 8 mL of groundwater was mixed with 12 mL of scintillation cocktail (MaxiLight+, HIDEX, Turku, Finland) in a 20-mL high-density polyethylene (HDPE) LSC vial and stored in the dark. Because of the short half-life of ²²²Rn (approximately 3.82 d), samples were analyzed within 48 h by liquid scintillation counting (LSC; 300SL, HIDEX, Turku, Finland).

Analytical quality control followed the Korean standard methods for groundwater quality. Calibration curves for ICP–MS, ICP–OES, and IC had coefficients of determination (R²) ≥ 0.9974, and method blanks were below the limits of quantification. Method detection limits were 0.002 μg L⁻¹ for U, 0.001–0.003 mg L⁻¹ for major cations and Si, and 0.005–0.012 mg L⁻¹ for anions. Accuracy, expressed as recovery of standard solutions, was 82.9–107.6%, and precision, expressed as the relative standard deviation of replicate analyses, was <2.0% for all constituents. For Rn, the counting efficiency determined from three standards was 2.62–2.65, and the minimum detectable activity was 0.01–0.06 Bq L⁻¹. Charge-balance errors of all samples were within ±10%.

2.3. Analyses of Rocks and Minerals

A 120-m-deep borehole was drilled into a Jurassic granitic aquifer in the study area, Korea, to obtain representative aquifer-rock samples. The upper 0–15 m consists of alluvial deposits and was not cored; drill core was recovered from the underlying bedrock between 15 and 120 m, corresponding to a cored interval of 105 m. The recovered core was logged by macroscopic and polarized-light microscopic observations and divided into lithologic units. The principal lithology was Jurassic two-mica granite, accompanied by an intermediate-composition dyke (id), a felsic (acidic) dyke (ad), and calcite-bearing breccia (br). The Jurassic two-mica granite was further divided into Jurassic coarse-grained two-mica granite (Jctgr), Jurassic medium-grained two-mica granite (Jmtgr), Jurassic porphyritic two-mica granite (Jptgr), and a Jurassic medium-grained two-mica granite alteration zone (Jmtgraz). Thin sections, polished sections, and powdered samples were prepared from each lithology for petrographic and geochemical analyses.

Thin sections were examined by polarized-light microscopy to characterize grain size and texture of the major minerals. Polished sections were carbon coated for electron-microscope imaging, qualitative elemental mapping, and quantitative analysis of radioactive accessory minerals. Backscattered-electron (BSE) images and semiquantitative analyses were obtained using a high-resolution field-emission scanning electron microscope equipped with an energy-dispersive X-ray spectroscopy detector (HR FE–SEM–EDS; Apreo 2S, Thermo Fisher Scientific, Waltham, MA, USA) at an accelerating voltage of 15 kV. Elemental maps and quantitative compositions were obtained using a field-emission electron probe microanalyzer (FE–EPMA; JXA-8530F, JEOL Ltd., Tokyo, Japan) at 15 kV and 10 nA, with beam diameters of 1–5 μm selected according to grain size.

Powdered samples were prepared using an 8000D Mixer/Mill (SPEX SamplePrep, Metuchen, NJ, USA) and analyzed by X-ray diffraction (XRD) and X-ray fluorescence (XRF). Mineral phases were identified using a D8 Advance A25 diffractometer (Bruker, Billerica, MA, USA) with Cu Kα radiation at 40 kV and 40 mA over 4–79° 2θ, with a step size of 0.02° and a counting time of 0.4 s per step. Whole-rock major-element compositions were determined from pressed powder pellets using an S8 TIGER XRF spectrometer (Bruker, Billerica, MA, USA).

3. Results and Discussion

3.1. Hydrochemistry of Groundwater

3.1.1. Field-Measured Water Quality Parameters

Time-series monitoring during the two wet-season and two dry-season campaigns showed large short-term fluctuations in groundwater level that were not closely related to rainfall. The groundwater level declined after WS1 but rose modestly after WS2 (Figure 2). The decline after WS1 likely reflects increased groundwater abstraction in the surrounding area during summer, including for agricultural irrigation. In contrast, the rise after WS2 likely reflects delayed recharge from rainfall in the latter part of the wet season. The short-term fluctuations are interpreted to reflect pumping from nearby domestic wells, the closest of which was located about 13 m from the borehole, although their pumping rates and schedules were not available. This indicates that the study borehole responds sensitively to local groundwater abstraction.

Figure 2. Time series of daily rainfall and groundwater level during the monitoring period. Blue bars show daily rainfall, and the black line shows groundwater-level variation. Arrows indicate the four depth-discrete sampling campaigns: WS1 and WS2, the first and second wet-season campaigns; DS1 and DS2, the first and second dry-season campaigns.

Water temperature, pH, and Eh varied little with depth and time, and the groundwater was mildly alkaline and oxic throughout the borehole (Figure 3 and Figure S1). DO ranged from about 2.3 to 5.3 mg L⁻¹ with no consistent depth trend, and SPC generally decreased with depth. Mean pH values were 7.94 ± 0.36 during the dry season and 8.17 ± 0.27 during the wet season. Except for a decrease to pH 7.24 at 15 m, depth-related variation was minor, and the groundwater was mildly alkaline relative to the average pH of 6.8 reported for private drinking-water wells in Korea [33]. Mean Eh values were 305.7 ± 47.7 mV in the dry season and 323.1 ± 54.3 mV in the wet season, indicating oxidation conditions. Mean DO concentrations were 3.72 ± 0.64 and 3.96 ± 0.79 mg L⁻¹ in the dry and wet seasons, respectively. Although DO varied, it remained above approximately 2.3 mg L⁻¹ throughout the sampled interval, indicating persistent oxidizing conditions [34]. Mean HCO₃⁻ concentrations were 103.12 ± 22.24 mg L⁻¹ during the dry season and 82.21 ± 7.67 mg L⁻¹ during the wet season, compared with a Korean average of 84.0 mg L⁻¹ [33]. During the wet season, HCO₃⁻ varied little with depth below the 15-m alluvium–bedrock boundary. During the dry season, the higher HCO₃⁻ concentrations may reflect infiltration and subsequent mixing of shallow groundwater affected by soil CO₂ and carbonate dissolution in the alluvial zone [35].

Figure 3. Depth profiles of dissolved U concentration, ²²²Rn activity, and hydrogeochemical parameters in borehole groundwater: pH, Eh, specific conductance at 25 °C (SPC), DO, HCO₃⁻, Na⁺, Mg²⁺, Ca²⁺, F⁻, and NO₃⁻. The borehole lithology is shown at left. Dark blue and cyan lines represent WS1 and WS2; dark red and pink lines represent DS1 and DS2, respectively.

These hydrogeochemical conditions are closely related to U mobility. The observed pH range of 7.24–8.75, Eh range of 146.5–390.8 mV, and elevated HCO₃⁻ concentrations favor formation of uranyl–carbonate complexes such as UO₂(CO₃)₂²⁻ and UO₂(CO₃)₃⁴⁻, which substantially increase U mobility [10]. The oxidation Eh values and consistently measurable DO indicate that dissolved U was more likely to occur as U(VI) than to precipitate as U(IV). On a pH–Eh stability diagram, the groundwater plots within the stability field of uranyl–carbonate complexes (Figure 4), supporting their predominance [36]. During the dry season, U concentrations decreased even though HCO₃⁻ concentrations increased. This pattern indicates that dilution and mixing reduced the relative contribution of U-rich bedrock groundwater more strongly than carbonate complexation increased U mobility.

Figure 4. Eh–pH stability diagram showing the predominant aqueous U species and the stability field of uraninite (UO₂), constructed using thermodynamic data from Krupka and Serne [36]. The upper and lower dashed lines represent the stability limits of water. Blue and red open circles denote groundwater samples from the wet-season (WS1 and WS2) and dry-season (DS1 and DS2) campaigns, respectively.

3.1.2. Major Ion Chemistry

Major-ion compositions were plotted on a Piper diagram using AquaChem (version 4.0; Waterloo Hydrogeologic, Kitchener, ON, Canada) (Figure 5). Groundwater was dominated by Ca–Na–HCO₃ and Na–Ca–HCO₃ facies. These facies indicate that HCO₃⁻ was generated mainly by carbonate dissolution and silicate weathering, whereas Ca²⁺ and Na⁺ were supplied through water–rock interaction and cation-exchange reactions [37].

Figure 5. Piper diagram showing the major-ion composition of groundwater samples from the study area. Symbols distinguish samples collected during the WS1, WS2, DS1, and DS2 campaigns. The diagram was generated using AquaChem 4.0.

Figure 5. Piper diagram showing the major-ion composition of groundwater samples from the study area. Symbols distinguish samples collected during the WS1, WS2, DS1, and DS2 campaigns. The diagram was generated using AquaChem 4.0.

K⁺ and SO₄²⁻ showed no clear depth- or time-dependent variation. During the wet season, the other major ions (Ca²⁺, Mg²⁺, Na⁺, and Cl⁻) also changed only gradually with depth, except at the 15-m boundary between the alluvial and bedrock aquifers (Figure 3 and Figure S1). Compared with the wet-season campaigns, the dry-season samples at 25–95 m showed slightly lower Na⁺ and higher Ca²⁺ and Mg²⁺ concentrations. This pattern may reflect residence and mixing of shallow groundwater enriched in Ca²⁺ and Mg²⁺ by carbonate reactions in the alluvial zone after recharge during the latter part of the wet season, partially modifying the composition of deeper bedrock groundwater.

3.1.3. Minor Ion Chemistry

Among the minor constituents (Si, F⁻, NO₂⁻, NO₃⁻, and PO₄³⁻), NO₂⁻ and PO₄³⁻ were below detection limits at most depths and during most campaigns. In contrast, the other constituents showed both depth-dependent and temporal variations (Figure 3 and Figure S1). With depth, F⁻ during the wet season was lowest at 15 m (approximately 1.0–1.6 mg L⁻¹) and remained relatively uniform below 25 m (approximately 2.3–3.7 mg L⁻¹), whereas the highest concentrations in all campaigns (3.1–4.2 mg L⁻¹) occurred at 105–115 m. Temporally, compared with the wet-season campaigns, the dry-season samples at 25–95 m showed lower F⁻ (mean 1.4 and 2.1 mg L⁻¹ in DS1 and DS2, respectively, versus 2.8–3.0 mg L⁻¹ in the wet season) and higher Ca²⁺, dissolved Si, and NO₃⁻, reflecting dilution by Ca-rich, F⁻-poor shallow groundwater recharged late in the wet season. Accordingly, the inverse relationship between F⁻ and Ca²⁺ was stronger in the dry season (Spearman rank correlation coefficient (ρ) = −0.93 in Table S1) than in the wet season (ρ = −0.44). In DS2, however, F⁻ at 65–95 m returned to wet-season levels (approximately 3 mg L⁻¹), indicating that shallow-water influence was restricted to shallower depths during this campaign. The increase in dissolved Si likely reflects the contribution of shallow groundwater with relatively high Si concentrations, and the increase in NO₃⁻ suggests that nitrate of anthropogenic origin, including agricultural fertilizer, was transported downward and persisted under oxic conditions [38,39]. Fluorite was not detected in the drill core, although F-bearing minerals such as micas, apatite, and bastnäsite–synchysite-group minerals were identified; nevertheless, the presence of fluorite along fractures around the borehole cannot be excluded.

3.1.4. Naturally Occurring Radioactive Materials in Groundwater

Depth-resolved analyses showed that, except at 15–25 m during DS2, U and Rn exceeded the Korean drinking-water standard for U (30 μg L⁻¹) and the monitoring level for Rn (148 Bq L⁻¹) at most depths and sampling times, except for U at 25 m and Rn at 15 m during DS2. Maximum concentrations were 488.1 μg L⁻¹ for U and 1549.8 Bq L⁻¹ for Rn (Figure 3).

The general increase in radionuclide concentrations with depth is attributed to longer groundwater residence times, more extensive water–rock interaction, and sustained supply and release of rock-derived radionuclides along fractures. In granitic aquifers, weathering and fracture development can enhance both U leaching and Rn emanation, while long residence times and oxic conditions promote dissolved U transport [15]. For Rn, deeper groundwater may remain enriched even during longer residence because production and release of ²²²Rn from ²²⁶Ra-bearing minerals can locally balance or exceed radioactive decay losses.

Within each season, the vertical profiles of U and Rn were broadly similar, indicating a first-order control by common fractures and bedrock architecture. Absolute concentrations, however, varied among campaigns, reflecting seasonal changes in the relative contributions of deep bedrock groundwater and shallower recharge (Figure 3). The lower concentrations observed during the later dry-season campaigns are interpreted to reflect delayed mixing with shallow water introduced during the latter part of the wet season, which diluted radionuclide-rich bedrock groundwater. For Rn, reduced groundwater flux may additionally weaken hydraulic connectivity among bedrock fractures and decrease the effective supply of rock-derived Rn [19].

To quantify relationships among U, Rn, and hydrogeochemical parameters, Spearman rank correlation analysis was performed (n = 44; Table S1). U and Rn were positively correlated (ρ = 0.77, p < 0.05), and both were lower in the dry season than in the wet season (mean U 348 vs. 137 μg L⁻¹; mean Rn 982 vs. 617 Bq L⁻¹). Both nuclides were negatively correlated with indicators of shallow groundwater, including NO₃⁻ (ρ = −0.84 and −0.86 for U and Rn, respectively) and Mg²⁺ (ρ = −0.71 and −0.95), and positively correlated with Na⁺ (ρ = 0.66 and 0.92), which is enriched in the deep groundwater. U was negatively correlated with HCO₃⁻ (ρ = −0.64), indicating that HCO₃⁻ variations did not control the spatial and temporal distribution of U within the borehole. Thus, oxic, mildly alkaline, HCO-bearing groundwater keeps U in solution, but variations in U and Rn within the borehole are mainly driven by mixing of shallow and deep groundwater. U and Rn did not, however, respond equally to mixing. At 25–95 m, the Rn/U ratio increased from 2.2–3.7 in the wet season to 3.1–5.7 in the dry season; that is, U decreased more than Rn. This reflects the short half-life of ²²²Rn (3.8 d). After dilution by shallow groundwater, Rn is replenished by continuous emanation from Ra in the aquifer matrix and on fracture surfaces [16]. Rn recovers toward its equilibrium activity as 1 − e^−λt, reaching approximately 94% within four half-lives (~15 d), whereas diluted U is not rapidly resupplied. In every campaign, the highest Rn/U ratio occurred at 102–105 m (4.6–8.9), within the 102–115-m interval where groundwater showed the highest F⁻ concentrations and generally the weakest shallow-water signals (lowest NO₃⁻ and Mg²⁺), suggesting the least-flushed groundwater with the longest water–rock contact. Under such slow-flow conditions, Rn can accumulate to equilibrium with emanation from the surrounding rock, whereas U does not increase correspondingly [16,17]. However, because Ra activities and borehole flow data were unavailable, the causes of the elevated Rn/U ratios at this interval could not be conclusively identified.

3.2. Petrographic and Mineralogical Characteristics

3.2.1. Core Logging and Whole-Rock Geochemistry

Macroscopic logging showed that two-mica granite is the dominant lithology, with subordinate intermediate-composition dykes, felsic dykes, and calcite-bearing breccia. The two-mica granite is medium- to coarse-grained and contains 8–30-mm phenocrysts between 54 and 62 m. It is generally grayish brown to reddish brown, whereas altered intervals at 69–77 m and 100–102 m are greenish brown to dark brown. The intermediate-composition dyke is black and aphanitic, intrudes coarse-grained two-mica granite, and contains calcite veinlets <1 mm thick. Its contact with the granite is sharp and includes a chilled margin characterized by grain-size reduction toward the contact. The felsic dyke has a greenish-brown to grayish-green aphanitic matrix containing whitish phenocrysts approximately 2 mm across and calcite veinlets <1 mm thick. The calcite-bearing breccia is whitish to grayish white and porphyritic, has a matrix similar to that of the felsic dyke, and contains 5–10-mm carbonate-rock fragments. This unit is descriptively termed calcite-bearing breccia. Strongly fragmented core occurs in the coarse-grained two-mica granite at 26–27 m and throughout much of the felsic-dyke interval.

Rock quality designation (RQD) was calculated for 3-m core intervals as the percentage of recovered core pieces longer than 100 mm [40]. RQD provides a first-order measure of discontinuity and fracture development, with lower values indicating more intensely broken core; however, because it is direction-dependent and does not fully represent a three-dimensional fracture network, it was used here as a supporting rather than stand-alone hydrogeological indicator. The mean RQD was 71.2%, corresponding to a “Good” rock-quality class. Intervals near 25 and 79 m were classified as “Very poor,” indicating particularly intense fracturing (Figure 6). Several dry-season vertical profiles of radionuclides and dissolved ions either covaried or varied inversely with RQD, suggesting that groundwater movement and mixing were enhanced in highly fractured intervals. These intervals likely provided preferential pathways for the downward penetration of shallow groundwater.

Figure 6. Depth profile of rock quality designation (RQD) and the corresponding borehole lithology. RQD was calculated for each 3-m core run as the summed length of core pieces longer than 100 mm divided by the core-run length. The RQD classification ranges and calculation used in this study are shown in the figure.

XRF analyses showed that the two-mica granite is dominated by SiO₂ (66.8–72.6 wt%), Al₂O₃ (17.1–20.8 wt%), and K₂O (2.24–3.82 wt%), with low CaO, MgO, TiO₂, and Fe₂O₃ contents (Table S2: XRF analytical results), consistent with typical two-mica granite compositions [41]. The intermediate-composition dyke contains 55.7–57.8 wt% SiO₂, 10.7–12.4 wt% MgO, and 16.0–20.1 wt% Al₂O₃, comparable to intermediate dykes reported elsewhere [42]. The felsic dyke has high SiO₂ (70.9–71.6 wt%), extremely low MgO (0.02–0.05 wt%), and relatively high Al₂O₃ (18.4–18.7 wt%) and K₂O (2.79–2.94 wt%), consistent with a felsic composition [43]. XRD identified quartz, albite, microcline, biotite, muscovite, sericite, and chlorite in the two-mica granite (Figure S2). The intermediate-composition dyke contains quartz, plagioclase, biotite, and chlorite; the felsic dyke contains quartz, albite, orthoclase, and muscovite; and the calcite-bearing breccia contains quartz, albite, microcline, muscovite, chlorite, and calcite.

3.2.2. Major Minerals Characteristics

Polarized-light microscopy showed that quartz and microcline in the representative two-mica granite are mostly anhedral, whereas biotite forms euhedral to subhedral grains 1–2 mm in size (Figure S3a–c). Muscovite, chlorite, and sericite occur as anhedral grains <1 mm. Myrmekitic intergrowths are present in quartz–feldspar domains, and feldspar sericitization and biotite chloritization are widespread. These textures indicate hydrothermal alteration of the two-mica granite.

The intermediate-composition dyke is porphyritic, but the fine-grained groundmass could not be resolved mineralogically by optical microscopy (Figure S3d). Anhedral quartz, feldspar, and calcite phenocrysts are <1 mm, and a calcite vein approximately 0.5 mm thick is present. The felsic dyke is also porphyritic and has a spherulitic groundmass whose constituent minerals are difficult to identify (Figure S3e). It contains euhedral quartz and anhedral calcite phenocrysts <1 mm. The calcite-bearing breccia is likewise porphyritic, with subhedral quartz and anhedral calcite phenocrysts <1 mm in a spherulitic, optically unresolved groundmass (Figure S3f).

3.2.3. Accessory and Radioactive-Bearing Minerals Characteristics

Electron-microprobe analyses identified a diverse assemblage of U- and Th-bearing or U–Th-substituting accessory minerals in the two-mica granite, felsic dyke, and calcite-bearing breccia: uraninite (UO₂), thorite (ThSiO₄), uranothorite (U-rich thorite; (Th,U)SiO₄), xenotime (YPO₄), zircon (ZrSiO₄), pyrochlore-supergroup minerals (PSG; A₂B₂X₆Y), bastnäsite–synchysite-group minerals (BSG; REE(CO₃)F–CaREE(CO₃)₂F), monazite-(Ce) ((Ce,La,Nd,Th)PO₄), allanite ((Ca,REE)(Al,Fe)₃(SiO₄)₃(OH)), and apatite (Ca₅(PO₄)₃(F,Cl,OH)). Only zircon was identified in the intermediate-composition dyke (Table 1).

Table 1. Accessory minerals identified by SEM–EDS in each drill-core lithology.

Lithology	Jctgr 1	Jmtgr 2	Jptgr 3	Jmtgraz 4	Id 5	Ad 6	Br 7
Accessory minerals	Allanite Apatite BSG 8 Monazite Thorite Uranothorite Zircon	Apatite BSG Zircon	Allanite Apatite BSG Monazite Thorite Uraninite Zircon	Apatite BSG Monazite PSG 9 Thorite Uraninite Uranothorite Xenotime Zircon	Zircon	Apatite Monazite Xenotime Zircon	Allanite Apatite Monazite Thorite Xenotime Zircon

Notes: ¹ Jctgr: Jurassic coarse-grained two-mica granite; ² Jmtgr, Jurassic medium-grained two-mica granite; ³ Jptgr, Jurassic porphyritic two-mica granite; ⁴ Jmtgraz, Jurassic medium-grained two-mica granite alteration zone; ⁵ id, Intermediate dyke; ⁶ ad, acidic dyke; ⁷ br, calcite breccia; ⁸ BSG: Bastnäsite (REE(CO₃)F)-Synchysite (CaREE(CO₃)₂F) group; ⁹ PSG: pyrochlore supergroup (A₂B₂X₆Y).

Uraninite was observed in the Jptgr and Jmtgraz samples and contains, on average, 92.04 wt% UO₃ and 4.93 wt% ThO₂ by EDS, confirming its exceptionally high radioelement content (Figure 7a and Table 2). Uraninite crystallized as a primary U(IV) phase under reducing conditions, but interaction with oxidizing low-temperature fluids can convert U(IV) to soluble U(VI) as uranyl (UO₂²⁺). Complexation of uranyl with carbonate then greatly increases its solubility and mobility [44]. Oxidatively altered and leached uraninite may develop connected pore networks that increase the Rn emanation coefficient.

Figure 7. Backscattered-electron (BSE) images of U- and Th-bearing accessory minerals from the study area: (a) uraninite (Urn); (b) thorite (Thr) associated with zircon (Zrn); (c,d) xenotime-(Y)–zircon intergrowths (Xtm–Zrn); (e) altered zircon (Zrn); (f) a pyrochlore-supergroup mineral (PSG) containing uraninite and galena (Urn/Gn) and associated with zircon; (g) a bastnäsite–synchysite-group mineral (BSG); and (h) monazite-(Ce) (Mnz). Additional abbreviations: Ap, apatite; Bt, biotite; Gn, galena; Mc, microcline; Mnz, monazite; Pl, plagioclase; Qz, quartz; Thr, thorite; Urn, uraninite; Xtm, xenotime; Zrn, zircon; PSG, pyrochlore-supergroup mineral; BSG, bastnäsite–synchysite-group mineral. Scale bars differ among panels because of different magnifications.

Table 2. Representative SEM–EDS compositions of selected accessory minerals from the study area (wt%; mean ± SD).

Component	Uraninite (n = 5)	Thorite (n = 35)	Xenotime-(Y) (n = 27)	Zircon (n = 20)	Xenotime-Zircon Intergrowth (n = 30)	PSG 1 (n = 9)	BSG 2 (n = 23)	Monazite (n = 33)
F							5.3 ± 1.3
SiO2		18.5 ± 2.3		31.5 ± 3.4	22.4 ± 9.3
P2O5			33.7 ± 2.9		13.7 ± 8.2			29.3 ± 1.2
CaO					1.5 ± 1.0	11.8 ± 2.1	20.0 ± 2.5
FeO						3.0 ± 1.1
MnO						1.4 ± 0.9
TiO2						25.3 ± 4.3
Nb2O5						33.6 ± 4.8
ZrO2		3.7 ± 1.6		54.9 ± 7.7	29.7 ± 14.5
HfO2				1.9 ± 0.5
Y2O3		2.5 ± 1.5	45.3 ± 5.8		20.8 ± 11.0	4.2 ± 1.4	1.9 ± 0.7
La2O3		2.2 ± 0.9					17.0 ± 3.6	16.6 ± 2.9
Ce2O3							32.5 ± 4.5	31.1 ± 1.9
Pr2O3							3.2 ± 0.6	2.8 ± 0.5
Nd2O3							12.3 ± 2.5	10.9 ± 2.3
Sm2O3							3.5 ± 1.4
Gd2O3			2.8 ± 0.7		2.8 ± 0.5		3.0 ± 0.5
Dy2O3			5.4 ± 1.2		4.2 ± 0.5
Er2O3			3.7 ± 0.6		3.0 ± 0.4
Yb2O3			2.9 ± 1.0		1.3 ± 0.2
ThO2	4.9 ± 0.2	65.9 ± 7.7	4.2 ± 2.2	3.6 ± 2.4	5.6 ± 4.4	5.7 ± 2.9	3.9 ± 2.3	6.4 ± 2.5
UO3	92.0 ± 3.9	9.5 ± 5.6	2.2 ± 1.7	2.1 ± 0.5	2.8 ± 0.9	5.2 ± 2.3		1.5 ± 1.1

Notes: ¹ PSG: Pyrochlore supergroup; ² BSG: Bastnäsite–Synchysite group.

Thorite and uranothorite occur mainly as fillings along grain boundaries and fractures and are associated with monazite, zircon, and apatite (Figure 7b). Their high ThO₂, low SiO₂, REE-bearing compositions, and pervasive internal fractures indicate dissolution–reprecipitation during hydrothermal alteration combined with metamictization. Thorite is particularly susceptible to alpha-recoil damage, and metamict thorite becomes highly reactive to later fluids [45]. Uranothorite in the present samples contains up to 31.0 wt% UO₃ by EDS (Table S3: EDS analytical results). This composition indicates that U was not retained exclusively in primary minerals but was leached, transported, and reincorporated into Th-silicate structures by later hydrothermal fluids [46]. Metamict thorite and uranothorite are therefore plausible direct sources of groundwater U and Rn at the study site.

Xenotime occurs as discrete grains and as complex intergrowths with zircon in the ad and br samples, where both minerals contain abundant micropores and fractures (Figure 7c,d). Their interfaces consist of smoothly connected mosaic domains composed of nanometer-scale crystallites. Partial fluid-mediated dissolution of zircon rims was followed by reprecipitation of P, Y, and REE, nucleation of nanoscale xenotime, and epitaxial growth along the crystallographic orientation of the zircon substrate [47]. Complementary distributions of Zr and Y–P in the EPMA maps support this interpretation (Figure 8). Ca, U, and Th within these domains are interpreted to reflect substitution for REE in the REEPO₄ structure during hydrothermal alteration [48]. Zircon–xenotime intergrowths contain average UO₃ and ThO₂ concentrations of 2.8 and 5.6 wt%, respectively, whereas discrete xenotime contains averages of 2.2 and 4.2 wt% (Table 2).

Figure 8. EPMA elemental-distribution maps of selected U- and Th-bearing accessory minerals. (a) Xenotime-(Y)–zircon intergrowth (Xtm–Zrn; Figure 7d), showing the distributions of U, Th, Ca, P, Zr, and Y. (b) Pyrochlore-supergroup mineral (PSG; Figure 7f), showing the distributions of U, Th, Nb, Ti, and Y. The corresponding BSE image is shown above each set of maps.

Zircon occurs as euhedral to subhedral grains but is strongly porous and fractured because of radiation damage and associated low-temperature hydrothermal alteration (Figure 7e). Low EPMA analytical totals (93.4–97.4%) and average EDS concentrations of 2.1 wt% UO₃ and 3.6 wt% ThO₂ are consistent with advanced metamictization (Table 2 and Table S3) [21]. Metamict zircon is more susceptible to fluid-mediated alteration and U release than crystalline zircon.

Pyrochlore-supergroup minerals (A₂B₂X₆Y) are common accessory phases in granitic rocks. Their A sites can host Na, Ca, Y, REE, U, and Th, whereas Nb, Ta, and Ti commonly occupy the B sites. Depending on U–Th content and accumulated radiation damage, these minerals may be partially metamict; thermal recovery occurs at approximately 500 K for weakly damaged material and approximately 800 K for strongly damaged material [49,50]. The PSG grains observed here contain abundant fractures and pores (Figure 7f). High-BSE domains contain up to 4.1 wt% UO₃, whereas lower-BSE porous domains contain up to 6.8 wt% UO₃; elemental maps likewise show relative U enrichment in pores. Uraninite inclusions occur within PSG grains, and PbS is present along uraninite margins (Figure 7f). This texture may record radiogenic-Pb-related auto-oxidation, in which Pb⁴⁺ oxidized U(IV) to U(VI), was reduced to Pb²⁺, and subsequently precipitated as PbS under locally reducing conditions [13]. The U enrichment in PSG pores is therefore more plausibly attributed to leaching and short-distance redistribution from enclosed uraninite than to direct dissolution of the PSG lattice.

Bastnäsite–synchysite-group minerals are light-REE carbonates in which U and Th may substitute for light REE. They occur as fibrous crystal aggregates with preferred orientations and may show BSE-contrast variations related to changes in REE/Ca ratios within single grains (Figure 7g). Although ThO₂ reaches 8.3 wt%, U is nearly absent (Table S3). Their direct contribution to dissolved ²³⁸U and ²²²Rn at the study site is therefore considered minor.

Monazite-(Ce) contains average concentrations of 1.5 wt% UO₃ and 6.4 wt% ThO₂ by EDS (Figure 7h and Table 2). However, natural monazite commonly undergoes efficient thermal self-annealing and therefore retains a comparatively crystalline and resistant structure. Its direct contribution to dissolved ²³⁸U and ²²²Rn at the study site is consequently considered lower than that of strongly metamict or altered phases [51].

Apatite and allanite commonly occur together and are frequently associated with other U–Th-bearing minerals, including monazite, thorite, xenotime, and zircon. Allanite also occurs as inclusions within apatite. Allanite contains up to 2.1 wt% ThO₂, indicating partial Th substitution during infiltration of late hydrothermal fluids [52].

Taken together, the principal sources of dissolved U and Rn at the study site are likely uraninite, thorite and uranothorite, xenotime–zircon intergrowths, altered zircon, and PSG, in which metamictization and hydrothermal alteration have produced abundant micropores, microfractures, and mosaic boundaries. These features increase the fluid–mineral reaction area and provide pathways for both oxidative U leaching and Rn release. In contrast, monazite, BSG, apatite, and allanite are considered minor contributors because of their comparatively crystalline structures or low U contents.

4. Conclusions

This study integrated depth- and time-resolved groundwater chemistry with petrographic and mineralogical analyses of drill core from a borehole in the study area where high U and Rn concentrations have been repeatedly detected. The groundwater was consistently oxic, mildly alkaline, and relatively rich in HCO₃⁻. These conditions favor uranyl–carbonate complexation and are therefore conducive to enhanced aqueous U mobility. Rn displayed pronounced seasonal variability, and the largest concentration changes occurred in highly fractured intervals, consistent with mixing and dilution along hydraulically connected fractures.

The drill core contains a diverse assemblage of U–Th-bearing minerals, including uraninite, thorite, uranothorite, xenotime, zircon, and pyrochlore-supergroup phases. Metamictization and the development of microfractures and pores increased structural disorder and reactive surface area, facilitating U leaching and enhancing Rn emanation. Textural evidence that U was leached from primary phases such as uraninite and subsequently redistributed into Th-silicate structures further indicates that late hydrothermal alteration exerted a major control on the present radionuclide signature of groundwater.

Collectively, the elevated naturally occurring radionuclide concentrations originate from dissolution and alteration of metamict U–Th-bearing minerals in the aquifer rocks. Their depth-dependent and seasonal distributions reflect the coupled effects of groundwater chemistry, transient mixing between shallow and deep flow components, and fracture architecture in the granitic aquifer. By linking source-mineral reactivity to hydrogeochemical and structural controls, this study provides a geochemical framework for understanding and predicting U and Rn behavior in comparable fractured granitic aquifers.

These findings have implications for the forecasting and quality control of groundwater used for drinking and domestic purposes in fractured granitic terrains. First, because U and Rn generally increased with depth and decreased where shallow groundwater penetrated during the dry season, well depth and the position of open intervals relative to fractured and altered zones should be considered when siting and designing domestic wells. Second, U and Rn concentrations in the same borehole differed between sampling campaigns, so a single sampling event may not adequately represent exposure; monitoring should include repeated sampling across seasons and under different pumping conditions. Finally, areas of Jurassic two-mica granite containing metamict and hydrothermally altered U–Th-bearing accessory minerals, where groundwater is oxic, mildly alkaline, and HCO₃-bearing, may be prioritized for monitoring and, where necessary, treatment before use for drinking.