Radiological investigation of natural carbonated spring waters from Eastern Carpathians, Romania

The current study presents a radiological water-quality assessment on 64 spring water samples from four Romanian counties. The study area is abundant in CO2-rich spring waters consumed by locals and tourists. Gross alpha activities ranged between 21 ± 2 and 7530 ± 658 mBq L−1, with 27% of the samples exceeding the WHO threshold. Gross beta values ranged from 40 ± 2 to 5520 ± 430 mBq L−1, with 29% exceeding the recommended values. Radionuclide activities fluctuated between 0.6 ± 0.08 and 81 ± 6 Bq L−1 for 222Rn, 15 ± 2 to 1154 ± 112 mBq L−1 for 226Ra, and from 18 ± 2 to 64 ± 5 mBq L−1 for 210Po. The annual effective doses attributed to radium varied between 0.002 and 0.23 mSv yr−1.


Introduction
Water is an essential resource and a basic human right. In order to be suitable for human consumption, it has to meet quality standards for microbial, chemical and radiological properties. Radionuclides are naturally present in water, resulting mainly from processes of dissolution, leaching and desorption of the surrounding geological environment (rocks and sediments) [23]. The primary alpha-emitting natural radionuclides present in water are 224 Ra, 226 Ra and 210 Po. Gross beta activities are mainly attributed to 228 Ra, 210 Pb and 40 K [7]. Water is a factor that can potentially increase the internal exposure to natural radiation [37], and for this reason, national and international guidelines were issued to ensure the safety of drinking water [9,13,21,38]. World Health Organization (WHO) [38] recommends a threshold value of 0.5 Bq L −1 for gross alpha activity and 1.0 Bq L −1 for gross beta activity, for water to be considered radiologically safe for consumption. These values were established in regard to the Individual Dose Criterion (IDC) adopted, of 0.1 mSv yr −1 for a 2 L daily water consumption. Romanian law 301/2015 was issued considering the same IDC, and establishes a guidance level of 0.1 Bq L −1 for gross alpha, and 1.0 Bq L −1 for gross beta activity.
In Romania, more than 2000 natural spring waters are documented [3,30], many of which are located in the Eastern Carpathians. This region is characterized by the proximity of the Neogene Călimani-Gurghiu-Harghita and Oaș-Gutâi volcanic ranges. Therefore, manifestations of post-volcanism in the area are leading to the occurrence of sparkling, CO 2 -rich mineral waters [18,34]. Thousands of local inhabitants are consuming raw water directly from the local springs on a daily basis, as their main source of drinking water or for medicinal purposes [6,10]. The water resources in this region are also accounting for 45% of the bottled mineral waters from Romania [10]. Beside drinking, the mineral waters are often associated with balneological practices [17] and is an important factor for tourism in the Eastern Carpathians [8]. Therefore, determining the radioactivity levels of these water resources is important for public health safety, and allows for the assessment of radiation exposure resulted from water ingestion. Few studies on the radiological quality of drinking water were conducted in the Eastern Carpathians, Romania [4,24,25]. Most studies are focusing mainly on 222 Rn and 226 Ra radionuclides, but the present-day state of knowledge in the area remains scarce.
The aim of the present study is to perform a radiological survey on 64 natural carbonated water samples collected from four counties in Romania, namely Covasna, Harghita, Bistrița-Năsăud and Maramureș, located in close proximity to the Eastern Carpathians and the Neogene volcanic ranges that it hosts. Water physico-chemical parameters (temperature, pH, redox potential, electrical conductivity, total dissolved solids (TDS) and salinity) were measured for each sample. Gross alpha and beta activity measurements, along with specific radionuclide determinations ( 210 Po, 222 Rn, 226 Ra, 40 K) were conducted and the results compared with the guidance levels for radioactivity in drinking water, established by WHO [38] and the national law [21]. The annual effective doses attributed to 226 Ra were calculated to assess the exposure to internal radiation.

Study site
Sixty-four water samples were collected from sparkling mineral springs in four counties in Romania (Fig. 1). The first area includes Covasna and Harghita counties (32 samples labeled as CVHR), which are in the proximity of the Călimani-Gurghiu-Harghita volcanic range, and subjected to the post-volcanic manifestations associated with it. The second area, represented by Bistrița-Năsăud and Maramureș counties (32 samples labeled as BNMM), is located nearby the Oaș-Gutâi volcanic range. The samples were collected in 2 L plastic containers and acidified with 65% nitric acid (HNO 3 ), to avoid the adsorption losses of radionuclides on the container walls. At the time of sampling, water physico-chemical parameters (temperature, pH, redox potential, electrical conductivity, TDS and salinity) were measured in 50 ml beakers using an XS-PC5 multiparameter.

Gross alpha/beta specific activity measurements
For gross alpha and beta activity measurements, 1 l of each sample was evaporated to dryness on a hot plate (< 85 °C) without boiling. Subsequently, 0.2 g of the resulted residual material was dissolved with 3 ml 3 M HCl, and mixed with 15 ml GoldStar Quanta scintillator cocktail in a plastic 20 ml vial. The samples were then measured using a TRICARB 2300 TR Liquid Scintillation Counter with a 65% efficiency for alpha and 85% for beta activity (tSIE = 479), a minimum detectable activity (MDA) of 25 mBq L −1 and a measurement time of 21,600 s. The calibration was performed using a 210 Pb standard in equilibrium with its successors, 210 Bi and 210 Po, purchased from the Czech Methodology Institute. The resulting values are representing the total activity of the sample (alpha and beta), as TriCarb 2300 does not allow separation of impulses according to their origin. Gross alpha activities were determined by alpha counting, 0.1 g of the residual was milled and transferred to 45 mm diameter aluminum discs, ensuring the distribution of the mass. The discs were then covered with silver activated zinc sulphide (ZnS(Ag)) sheets for alpha particle detection and were measured using an MEV NP-420 alpha counting system, for 10,000 s. The calibration was performed using a 209 Po for a sample density of 25 mg/cm 2 . The minimum detectable activity for gross alpha activity was 20 mBq L −1 . Gross beta activities were achieved by subtracting gross alpha from the total activity of the sample, previously measured by LSC.

Radon ( 222 Rn) specific activity measurements
The 222 Rn activity measurements were carried out by Liquid Scintillation Counting (LSC) technique [16]. The method involves adding 10 ml of water, at the time of sampling, to 10 ml mineral oil scintillator cocktail (ProScint Rn A/B) in a 20 ml glass vial using a plastic syringe, after which the vial was sealed and shaken vigorously. The samples were then set aside for a minimum of 3 h to allow for the ingrowth of the short-lived progenies of 222 Rn [22,33]. The measurements were carried out using a TRICARB 2300 TR Liquid Scintillation Counter calibrated with a standard NIST RaCl solution with an activity of 6 Bq L −1 , dissolved in water. The MDA was 0.3 Bq L −1 and the measurement time 10,800 s.

Radium ( 226 Ra) specific activity measurements
The 226 Ra activity analysis procedure is based on the liquid scintillation technique, 226 Ra is determined through 222 Rn and its progenies, which are extracted by the scintillation cocktail from the water phase. In this regard, one liter of each sample was acidified with 65% nitric acid (HNO 3 ), and concentrated at least ten folds, in order to reach a lower detection limit, by gentle evaporation on a hot plate, at a temperature of ≤ 85 °C. To avoid the occurrence of precipitation during the pre-concentration phase, all salts were converted in nitrates, which have a high solubility in water. Furthermore, 10 ml of the pre-concentrated sample were transferred in a polyethylene vial, and 10 ml scintillation cocktail (ProScint Rn A/B) were added. The vials were then stored in the dark at a constant temperature (16 °C), for 30 days, to allow for the 222 Rn ingrowth period. After the secular equilibrium between 226 Ra and 222 Rn was established, the samples were measured using a TRICARB 2300 TR Liquid Scintillation Counter, calibrated with a standard NIST RaCl solution, with an MDA of 20 mBq L −1 , and a measurement time of 10,800 s.

Polonium ( 210 Po) specific activity measurements
For 210 Po activity measurements, residue of samples that exceeded a gross alpha activity of 0.5 Bq L −1 was subjected to acidic digestion. To each sample, 0.3 mL (100 Bq·mL −1 ) 209 Po tracer was added. The samples were treated with HNO 3 and HCl. Subsequently, they were brought to 100 ml using distilled water, the pH was adjusted in the 0.1-0.3 range and the samples were left on a hot plate (< 85 °C) for three hours to allow for the spontaneous deposition of 210 Po on high nickel content stainless steel discs [5]. The discs were than analyzed using an ORTEC SOLOIST Alpha Spectrometer System with Ultra ENS-U900 detectors and an active surface of 900 mm 2 with a resolution greater than 29 keV, calibrated using a 209 Po standard solution. The minimum detectable activity for 210 Po was 15 mBq L −1 .

Potassium ( 40 K) activity measurements
The activity of 40 K was measured for 1 g of dry residue from each sample using a Well-type High-Purity Germanium (HPGe) Gamma Spectrometric System (ORTEC GWL-120-15 detector with a resolution of 2.08 keV for 1.33 MeV 60 Co and 1.1 keV for 122 keV 57 Co gamma lines). The detector was calibrated with a Merck KCl solution, and has an MDA of 250 mBq L −1 . The specific activity of 40 K was determined from its 1460.7 keV gamma-ray lines after a measurement time greater than 120,000 s. Spectrum acquisition was performed using the MAESTRO multi-channel analyzer software.

Annual effective dose calculation
The annual effective dose (AED) was calculated for 226 Ra activities from each sample, using the following equation [29]: where: AED Ra-226 stands for 226 Ra Annual Effective Dose (mSv yr −1 ). Λ Ra-226 stands for 226 Ra activity in the sample (mBq L −1 ). IRw stands for the annual ingested volume of drinkingwater (L yr −1 ).
CF stands for the dose coefficient (Sv Bq −1 ). The dose coefficient value of 2.8 × 10 −7 Sv Bq −1 as well as the IRw value of 730 L yr −1 were extracted from Romanian law 301/2015 [21].

Water physico-chemical parameters measured values
The measured values for water physico-chemical parameters at the time of sampling (temperature, pH, conductivity, oxidation-reduction potential (ORP), TDS and salinity) in the studied area are presented in Table 1. Temperature ranged between 11.7 and 25 °C; pH values ranged between 1.71 and 8.18; ORP values were in the −56.7 to 296 mV interval; conductivity values were between 252 and 10,560 µS; TDS ranged from 178.5 to 7500 ppm, and salinity between 56.7 and 5850 mg L −1 .

Specific activity measured values
The specific activity values of gross alpha and beta along with 222 Rn, 226 Ra and 210 Po radionuclides are presented in Table 2. WHO, 2017 and the national law 301/2015 are both stating that initial determinations of gross alpha and beta activities are required as the first step in assessing radiological safety of drinking water. Furthermore, if any guidance level for these activities is exceeded, specific determinations of radionuclides have to be followed. In the present study, gross alpha and beta, as well as 222 Rn and 226 Ra determinations were carried out for the whole set of samples. Subsequently, for the samples exceeding the guidance level of 0.5 Bq L −1 for alpha and respectively 1 Bq L −1 for gross beta activities, stipulated in WHO Guidelines for Drinking-water Quality [38], further investigations were carried out, by measuring 210 Po and 40 K specific activities. In some cases, when the residue mass of the sample permitted further investigations to be performed, 210 Po was also measured for samples with activities below WHO guidance level for gross alpha.
Gross alpha activities (excluding 222 Rn), which are presented in Fig. 2, ranged between 21 ± 2 and 2440 ± 210 mBq L −1 for the samples collected in Covasna and Harghita (CVHR), and between 40 ± 6 and 7530 ± 658 mBq L −1 for those from Bistrița-Năsăud and Maramureș (BNMM). 26.7% of samples exceeded the guidance level of 500 mBq L −1 stated in WHO, 2017 for alpha activity, and 53.5% exceeded the Romanian legislation threshold of 100 mBq L −1 . 12 samples had activities under the detection limit (< 20 mBq L −1 ), and an additional 8 samples could not be measured for gross alpha activity, due to insufficient residue. Therefore, these samples were only measured for gross alpha and beta (total) activity.
Gross beta activities ranged between 50 ± 3 and 4160 ± 398 mBq L −1 for CVHR samples, and between 40 ± 3 and 5520 ± 430 mBq L −1 for BNMM. A total 26% of the samples exceeded the guidance level of 1000 mBq L −1 (stated in both law 301/2015 and WHO, 2017), and 13 samples were below the MDA of 25 mBq L −1 . Subsequently, as guidelines are indicating, 40 K measurements were performed for the samples exceeding the recommended level for gross beta. The potassium activities are radiologically negligible, due to the age-dependent effective dose conversion factor for adults (CF K-40 = 6.2 × 10 -6 mSv Bq −1 ), which is the least of the other radionuclides [12]. Thus, the 40 K contribution should be extracted from the gross beta activity in order to accurately assess safety of drinking water [38]. 40 K activities could be detected in 8 samples from BNMM. The activities ranged from 260 ± 30 to 4368 ± 390 mBq L −1 , with another 7 values under the detection limit. The gross beta activities and 40 K contribution are presented in Fig. 3. It can be observed that after 40 K subtraction, gross beta activities of three samples decreased under the guideline level.
Radon ( 222 Rn) is not regulated by WHO Guidelines for Drinking-water Quality, 2017, as 90% of the dose attributable to radon in drinking-water comes from inhalation rather than ingestion [37]. Thus, controlling the inhalation pathway is the most effective way to control radon internal exposure [38]. Romanian law 301/2015 imposes a recommended value of 100 Bq L −1 for 222 Rn in drinking water. However, radon in not to be included in the gross alpha activity of the samples. The 222 Rn activity values ranged between 0.6 ± 0.08 and 81 ± 6 Bq L −1 for the whole set of samples, and did not exceed the national legislation recommendations.
Radium ( 226 Ra) is one of the most dangerous and widely distributed long-lived α-emitters found in environmental samples [31] due to a combination of its long half-life (T 1/2 = 1602 years) and radiological effects [1]. In the present study, the 226 Ra specific activities ranged between 21 ± 1 and 429 ± 40 mBq L −1 for CVHR and between 24 ± 2 and 1154 ± 112 mBq L −1 for BNMM area. Only one sample has exceeded the WHO guidance level of 1000 mBq L −1 and 5 that of national legislation, which is 500 mBq L −1 . 35 samples had activities under the detection limit (< 20 mBq L −1 ). 226 Ra activity values and their compliance with the guidance level can be observed in Fig. 4.
Furthermore, the annual effective doses (AED) were calculated in order to assess 226 Ra contribution to the received dose, resulting from water ingestion. The recommended maximum value for AED attributed to drinking water, considering a daily consumption of 2L, is 0.1 mSv yr −1 [21,38]. For the calculation, the national legislation dose coefficient (CF) of 2.8·10 -7 Sv·Bq −1 was considered. The values are presented in Fig. 5, and ranged from 0.002 to 0.23 mSv yr −1 , with 5 samples exceeding the guidance value.
Polonium ( 210 Po) activities were generally low, ranging between 18 ± 2 and 64 ± 5 mBq L −1 , with only 4 samples exceeding the minimum detectable activity, and 32 being below it (< 15 mBq L −1 ). The activities did not exceed, in any case, the guidance level for 210 Po, which is 100 mBq L −1 , adopted by both WHO and Romania.

Data analysis
The relationships between gross alpha, gross beta and 226 Ra specific activities with water physico-chemical parameters (T, pH, ORP, conductivity, TDS and salinity) were investigated by performing Pearson's correlations, and the results   are presented in Table 3. Significant positive correlations can be observed between specific activities and salinity, with a correlation coefficient of R = 0.62 for α and R = 0.59 for β and 226 Ra, considering a confidence interval of α = 0.01. Other authors have previously observed such relationships between salinity and dissolved radium [19,20], which could imply ion-exchange dynamics. As salinity and conductivity are related (both measures being increased by dissolved ions), the positive correlations between gross alpha, beta as well as radium activities and conductivity are expected (R = 0.60 for α; 0.57 for β; 0.59 for 226 Ra, α = 0.01). Another positive correlation can be established between specific activities and TDS (R = 0.63 for α and R = 0.59 for β and 226 Ra, α = 0.01). In literature, [32] also noticed positive correlations between TDS and uranium, as well as [27], who found similar relationships for 226 Ra and 228 Ra. TDS, salinity and conductivity have similar R values when correlated to gross alpha, beta and radium activities, which is the result of the strong correlations between these parameters alone. Additionally, no correlations could be settled between the activity of the samples and temperature, pH or oxidation-reduction potential, implying that these parameters do not affect radionuclide concentrations in water. For a better representation and understanding of the activity values, the frequency distributions of gross alpha and beta specific activities were analyzed (Fig. 6). First, an initial distribution revealed that the highest number of cases are contained in the first class interval (between 0 and 1000 mBq L −1 ) for both alpha and beta activities, representing 77% of gross alpha, respectively 71% of gross beta  values from the total set of samples (Fig. 6, panels A and C). The higher class intervals showed a reduced frequency. Considering this observation, further statistical analyses were performed for an in-depth look at this specific class interval. As such, a second frequency distribution performed on the values in the 0-1000 mBq L −1 range highlighted that the highest number of cases are present in the 0-100 mBq L −1 range, with 46% of the total samples for gross alpha activities, respectively 29% of the gross beta activities (Fig. 6, panels C and D) being contained in it. From the frequency distributions presented, it can be concluded that both gross alpha and beta activities are falling mostly in the low activity class intervals, and the cases are exponentially decreasing with activity.
For an enhanced visual representation of the specific activity data, a spatial distribution map was constructed, displaying the location of the samples and their corresponding gross alpha, beta and 226 Ra activity values (Fig. 7). It can be observed that Bistrița-Năsăud (BN) county hosts the highest activity samples for all three parameters, clustered in a"hotspot" area, near Sângeorz-Băi town. The highlighted area includes the samples labeled BNMM 15-20, and 22, which are among the highest activity samples in the dataset, and are also exceeding the guidance levels set by national legislation. An explanation for the cluster formation is that the springs are closely located and similar in water physicochemical parameters values, and thus could be sourced from the same underground aquifer. In contrast, the other studied areas generally showed low specific activities, with sporadically occurrences of higher concentration values and no distinguishable patterns or hotspots.
Furthermore, a comparison between the results obtained in the present study for gross alpha and beta activities and the results of other similar studies in literature has been performed (Table 4). It can be observed that, except for [2], in Jordan, the present study has higher mean values for the gross alpha/beta activities, as well as maximum values, in comparison to the other studies cited. The difference in values between the results of [28] in Galați county, Romania, and the results of the present work could be explained by the geology of the study site. The Neogene volcanic mountain ranges that are located close to the study site could influence the values and lead to higher activities, as radioactivity concentration in soil is relatively higher in areas with volcanic rocks [35]. Beside geology, the high specific activity cluster located in Bistrița-Năsăud county has considerably raised the mean values for both gross alpha and beta, as well as the maximum values.

Conclusions
Sixty-four carbonated water samples collected from natural springs were analysed from a radiological perspective, in order to evaluate their compliance with national [21] and international [38] guidance levels for radioactivity in drinking water. Initial gross alpha and beta specific activity measurements revealed that for alpha activities,   26.7%, of the samples exceeded the WHO, 2017 guidance level of 500 mBq L −1 , and 53.5% exceeded the Romanian law threshold of 100 mBq L −1 . Gross beta common recommended value of 1000 mBq L −1 was exceeded by 26% of the samples. However, frequency distribution analyses indicated that the activities for both gross alpha and beta are generally in the lower distribution intervals (0-1000 mBq L −1 ), and the frequency decreases exponentially with activity. 226 Ra activity values exceeded the national guideline level by five samples (500 mBq L −1 ), of which one was above 1000 mBq L −1 (WHO recommended value). The highest values from the dataset for gross alpha, beta and radium-226 are concentrated in a hotspot area composed of 7 samples, in Bistrița-Năsăud county. Polonium-210 specific activities were low and did not exceed the guidance levels. After calculating the Annual Effective Dose (AED) attributable to radium-226, five samples had a value above the reference dose of 0.1 mSv yr −1 . The mean and maximum values for specific activities for gross alpha/ beta were generally higher than those found in other studies, which could be attributed to the presence of volcanic rocks in the study area, and the post-volcanic emanations associated with the adjacent Neogene volcanic range. The guidance levels, the dose coefficients and the AED calculations performed considered a water consumption rate of 2L/day for the investigated spring waters. Daily consumption from springs exceeding guidance levels should be avoided, and not used as a primary source of drinking water. Being the first investigation performed on the post-volcanic region of Romania, and one of the few conducted in the country, the present work reveals an insight regarding the quality and radiological properties of the natural mineral water sources in this region. By assessing the compliance of the specific activity values of the samples with the recommendations contained in international and national guidelines, the present study provides useful information regarding public health safety. The high percentage of samples exceeding the recommended values for the investigated parameters highlights the importance of studying potable natural water from a radiological perspective.