Chemical modeling of groundwater in the Banat Plain, southwestern Romania, with elevated As content and co-occurring species by combining diagrams and unsupervised multivariate statistical approaches

Chemosphere 172 (2017) 127e137

Contents lists available at ScienceDirect

Chemosphere journal homepage: www.elsevier.com/locate/chemosphere

Chemical modeling of groundwater in the Banat Plain, southwestern Romania, with elevated As content and co-occurring species by combining diagrams and unsupervised multivariate statistical approaches Sinziana Butaciu a, Marin Senila b, Costel Sarbu a, Michaela Ponta a, Claudiu Tanaselia b, Oana Cadar b, Marius Roman b, Emil Radu c, Mihaela Sima d, Tiberiu Frentiu a, * a

Babes-Bolyai University, Faculty of Chemistry and Chemical Engineering, Arany Janos 11, 400028, Cluj-Napoca, Romania National Institute for Research and Development of Optoelectronics Bucharest, Research Institute for Analytical Instrumentation, Donath 67, 400293, ClujNapoca, Romania c National Institute of Hydrology and Water Management, Bucuresti-Ploiesti 97, 013686, Bucharest, Romania d Romanian Academy, Institute of Geography, Dimitrie Racovita 12, 023993, Bucharest, Romania b

h i g h l i g h t s
Groundwater naturally enriched with inorganic arsenic species.
Classification of groundwater sources using Fuzzy Hierarchical Cross-Clustering. 3
PO3 4 eAsO4 ion exchange and water-rocks interactions as sources of As.
NaþeFepH cluster as marker for As naturally enriched groundwater.
Co-occurrence of As species and F under the influence of HCO 3 and alkaline pH.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 15 July 2016 Received in revised form 21 December 2016 Accepted 26 December 2016 Available online 29 December 2016

The study proposes a combined model based on diagrams (Gibbs, Piper, Stuyfzand Hydrogeochemical Classification System) and unsupervised statistical approaches (Cluster Analysis, Principal Component Analysis, Fuzzy Principal Component Analysis, Fuzzy Hierarchical Cross-Clustering) to describe natural enrichment of inorganic arsenic and co-occurring species in groundwater in the Banat Plain, southwestern Romania. Speciation of inorganic As (arsenite, arsenate), ion concentrations (Naþ, Kþ, Ca2þ,   2 3  Mg2þ, HCO 3 , Cl , F , SO4 , PO4 , NO3 ), pH, redox potential, conductivity and total dissolved substances were performed. Classical diagrams provided the hydrochemical characterization, while statistical approaches were helpful to establish (i) the mechanism of naturally occurring of As and F species and the þ 2 3 anthropogenic one for NO 3 , SO4 , PO4 and K and (ii) classification of groundwater based on content of  arsenic species. The HCO3 type of local groundwater and alkaline pH (8.31e8.49) were found to be responsible for the enrichment of arsenic species and occurrence of F but by different paths. The PO34 eAsO34 ion exchange, water-rock interaction (silicates hydrolysis and desorption from clay) were associated to arsenate enrichment in the oxidizing aquifer. Fuzzy Hierarchical Cross-Clustering was the strongest tool for the rapid simultaneous classification of groundwaters as a function of arsenic content and hydrogeochemical characteristics. The approach indicated the NaþeF-epH cluster as marker for groundwater with naturally elevated As and highlighted which parameters need to be monitored. A chemical conceptual model illustrating the natural and anthropogenic paths and enrichment of As and co-occurring species in the local groundwater supported by mineralogical analysis of rocks was established. © 2016 Elsevier Ltd. All rights reserved.

Handling Editor: X. Cao Keywords: Groundwater Arsenic speciation Cluster analysis Principal component analysis Fuzzy hierarchical cross-clustering

* Corresponding author. E-mail address: [email protected] (T. Frentiu). http://dx.doi.org/10.1016/j.chemosphere.2016.12.130 0045-6535/© 2016 Elsevier Ltd. All rights reserved.

128

S. Butaciu et al. / Chemosphere 172 (2017) 127e137

1. Introduction Groundwater is one of the most significant sources for drinking water as well as for human activities in industrial, agricultural, household, recreational and other fields. In arid and semi-arid regions groundwater is often the sole source of drinking water for local human communities. Because of the groundwater-mineral matrix interaction, some chemical parameters can occur in high concentrations even in the absence of anthropogenic sources. In these circumstances the assessment of groundwater quality and health risk was the subject of several studies (Aksever et al., 2015; Cuoco et al., 2015; Esmaeili-Vardanjani et al., 2015; Al-Harbi et al., 2014; Kelepertzis, 2014; Shankar et al., 2011; Carrillo-Rivera et al., 2002). One of the most common chemical species naturally occurring in groundwater is arsenic in inorganic form following water-rock interaction. Thus more than 500 million people living in the so-called „arsenic endemic” regions such as West Bengal India, Bangladesh, China, Vietnam, Taiwan, Northern Chile, Mexico, USA, and Central and Eastern Europe are chronically exposed to arsenic from drinking water (1e10000 mg L1) (Sarkar and Paul, 2016; Panagiotaras et al., 2012; Amini et al., 2008; Smedley and Kinniburgh, 2002). In Europe, the Pannonian Basin including eastern Hungary, western Romania, northern Serbia, northeastern Croatia and southern Slovakia, is characterized by an elevated As content, which poses health risks to more than 1000000 inhabitants via drinking water (Romanian National Institute of Statistics, 2012; Rowland et al., 2011; Dimkic et al., 2010; Gurzau and Gurzau, 2001). Examination of groundwater with elevated As concentration and the mechanisms related to its occurrence are public health and environmental concerns in the world (Chakraborti et al., 2016; Frederick et al., 2016; Sarkar and Paul, 2016; Rahman et al., 2015a; Biswas et al., 2014a; Sorg et al., 2014; Bonte et al., 2013; Ghosh and Sar, 2013; Rango et al., 2013). Besides the well-known carcinogenicity of inorganic arsenic species, studies indicated the risk associated to other diseases such as anemia, arsenicosis, vascular and heart diseases, neurological, or neuro-physiological diseases or risk for spontaneous pregnancy loss via drinking water consumption from wells with arsenic below or above of maximum contaminant level (MCL) (Chakraborti et al., 2016; Sarkar and Paul, 2016; Neamtiu et al., 2015; Rahman et al., 2015a, 2015b; Surdu et al., 2015; Bloom et al., 2010, 2014). World Health Organization (WHO) and U.S. Environmental Protection Agency (EPA) adopted the value of 10 mg L1 arsenic as MCL in drinking water (U.S. EPA, 2001 and WHO, 2004). Based on Water Framework Directive (WFD) (2000/60/EC), EU countries adopted the Groundwater Directive (GWD) with the aim to protect all bodies of surface water and groundwater from pollution and established measures and Environmental Quality Standards (EQS) so that inland and coastal water resources to achieve “good status” (Crane and Babut, 2007; Crane et al., 2007; Directive 2006/118/EC). Member states have established threshold values (TVs) for groundwater bodies depending on hydrogeological conditions and natural background levels (NBLs) at least for contaminants referred to as risk factors from natural or anthropogenic sources (As, Cd, Pb, 2   Hg, NHþ 4 , Cl , SO4 , NO3 ) and synthetic substances (trichloroethylene, tetrachloroetylene) (Annex 3 to Directive 2006/118/EC). In Romania, there were established TVs for all groundwater bodies according to NBLs (Order 621/2014; Radu et al., 2010). The simple examination of the results for hazardous chemicals in groundwater in relation with guideline values is not satisfactory for explaining and understanding their behavior in groundwater. Multivariate supervised/unsupervised statistical approaches are needed to reveal hidden relationships between chemical parameters and to establish the relevant characteristics for classification, grouping and delineation of groundwater sources (Cuoco et al.,

2015; Esmaeili-Vardanjani et al., 2015; Spanos et al., 2015; Sener and Sener, 2015; Al-Harbi et al., 2014; Kelepertzis, 2014; Kim et al., 2014; Sappa et al., 2014). The aim of this study was to provide a chemical modeling of groundwater in the Banat Plain, south-western Romania, for a deeper understanding of occurrence of high level of As and co3  occurring species F, SO2 4 , PO4 and NO3 . The model is based on a combination of classical diagrams (Gibbs, Piper and Stuyfzand Hydrogeochemical Classification System (SHCS)) and unsupervised chemometric methods such as Cluster Analysis (CA), Principal Component Analysis (PCA), Fuzzy Principal Component Analysis (FPCA) and Fuzzy Hierarchical Cross-Clustering (FHCC). Usefulness of the model was demonstrated on water samples collected from shallow groundwater body (GW-ROBA03) and depth groundwater body (GW-ROBA18) labeled according to water body delineation in Romania. Besides speciation of inorganic As (arsenite, arsenate) other hydrogeochemical characteristics were investigated such as 2 3   ion concentrations (Naþ, Kþ, Ca2þ, Mg2þ, HCO 3 , Cl , F , SO4 , PO4 ,  NO3 ), pH, redox potential (Eh), electrical conductivity (EC) and total dissolved substances (TDS). The chemical conceptual model describing the pathway and enrichment of As and co-occurring species in the local groundwater was coupled with hydrogeochemical and mineralogical analysis of the aquifer rocks and land use activities. In this model both natural via water-rock interaction and ion-exchange, and anthropogenic sources associated to land use activities were identified. The present study was an attempt to examine the occurrence of the elevated As species in an oxidizing aquifer rather than to assess the quality of the groundwater. 2. Materials and methods 2.1. Site description, water sample collection and preservation The study-area is situated in the Banat Plain, Bega and Timis Rivers Basin, south-western Romania and belongs to the eastern margin of the Panonian Basin that accumulated very thick deposits during the Neogene and Quaternary. The Banat Plain was formed as a result of the activity of rivers during the Quaternary that transported and deposited detrital sediments as a succession of cross bedded layers and alluvial fans. The thickness of the Quaternary deposits may reach tens to hundreds m and they are generally coarse-grained with discontinuous finer intercalations (Mutihac, 1990; Enciu et al., 2014). Beneath the Quaternary deposits, and completely covered by them, the Pliocene fluvio-lacustrine sediments are also detrital, generally finer gained. Their thickness increases from east to west from about 100 m to more than 1000 m. The two superposed groundwater bodies, GW-ROBA03 and GWROBA18, were identified in the Banat Plain by the Romanian Water Authority and classified as shallow/phreatic aquifer and deeper confined aquifer (ANAR, 2015). GW-ROBA03 is developed until 15 m in flood plains and terraces, and until 30 m in interfluves and the overlaying strata are represented by clays, sandy clays, silty clays, silts, argillaceous silts and sandy silts. The main lithological units of the aquifer GWROBA03 are presented in Supplementary material 1 (Radu and Radu, 2011). The shallow aquifer is exploited mainly through low yield domestic wells. The main flow direction is from north-east to south-west, with a hydraulic gradient of 0.1e2.0‰. The effective porosity is between 5 and 25% and the hydraulic conductivity ranges between 6 and 68 m d1. The depth of the water table is less than 5 m in most cases. The aquifer is recharged through precipitations and hydraulic connection with the rivers. Due to the permeability of the covering sediments, GW-ROBA03 is relatively vulnerable to contamination through the human settlements in the

S. Butaciu et al. / Chemosphere 172 (2017) 127e137

area, municipal waste land filling, chemicals of agricultural use and livestock farming. The deep aquifer GW-ROBA18 (Upper Pleistocene-Lower Quaternary) is porous-permeable and from lithological point of view the deposits consists of sands, rarely gravels, sometimes with lens aspect, alternating with clay, sandy clays and marls (ANAR, 2015; Pascu, 1983). The lithological succession determines the presence of a variable number, up to 8e12 in some cases of water-bearing layers to depth of 350 m. GWROBA18 is a confined aquifer, the potentiometric surface is close to the ground level, or even artesian in some cases. The aquifer is intensively exploited for drinking, industrial and agricultural use. Although better protected against pollution than the shallow aquifer, there is a risk of contamination due to the discontinuity of the confining layers. The climate is temperate, with Mediterranean and oceanic influences with the average temperature of 10e11  C and precipitation average around 600 mm. The main land use in the region is linked to maize and wheat production, which could represent an exposure risk of groundwater by fertilizer application. Although previously equipped for irrigation, currently almost no irrigation is performed in the region and thus there is no associated return flow risk. The sampling sites for the groundwater are presented in Fig. 1. Water samples were collected by the Research Institute for Analytical Instrumentation Cluj-Napoca from 18 locations (1 L each) of which 4 GW-ROBA03 (samples 1; 2; 5; 17) and 14 GWROBA18 after 10 min pumping. Water samples were filtered (<0.45 mm) on-site in polyethylene bottles previously cleaned, then pH, Eh and alkalinity were determined. After that, the samples were flash-frozen on field, transported to laboratory and stored at 20  C until analysis. Arsenic speciation, anions, conductivity and TDS measurements were conducted in samples without any treatment, whereas the metals were determined after acidification to 2% (v/v)

129

HNO3. A number of 10 clay and 20 sand samples collected in different locations of the studied area were investigated for mineralogical composition using X-Ray Diffraction (XRD) and elemental analysis by X-Ray Fluorescence Spectrometry (XRF). The XRD analysis revealed the presence of quartz, aluminosilicates of Na, K, Ca, Mg, etc., carbonates, phosphates, sulphates and fluoride (Supplementary material 2). Elemental composition of the mineral matrix and As are presented in Supplementary material 3. 2.2. Reagents and certified reference materials (CRMs) Hydrochloric acid, 30% (m/m), ultrapure, 0.1 M HCl, NaBH4 pro analysis, NaOH suprapur, L-cysteine for biochemistry, stock solution of 1000 mg mL1 As(V), ICP multielement standard solution IV 1000 mg mL1 and stock solution of 1000 mg mL1 F, Cl, NO 3, 3 SO2 4 and PO4 respectively, were purchased from Merck (Darmstadt, Germany). The solutions of 3% (m/v) L-cysteine in 0.01 M HCl (pH ¼ 2.00 ± 0.01), 0.01 M HCl (pH ¼ 2.00 ± 0.01) used as carrier and 0.5% (m/v) NaBH4 stabilized in 0.5% (m/v) NaOH were daily prepared. Standard solutions in the range 0e100 ng mL1 As in 0.01 M HCl and 0.3% (m/v) L-cysteine were used during calibration and the solution of 0.3% (m/v) L-cysteine in HCl 0.01 M was used as blank. Standard solutions in the range 0e5 mg mL1 F and PO3 4 2 and 0e20 mg mL1 Cl, NO 3 and SO4 were prepared to calibrate the ion chromatograph. ZoBell's solution prepared by dissolving 1.4080 g K4Fe(CN)4$3H2O, 1.0975 g K3Fe(CN)6 and 7.4555 g KCl to 1000 mL solution was used as reference standard for the measurement of the redox potential of water (Eaton et al., 2005). Milli-Q water prepared in laboratory (Millipore Corp, Bedford, USA) was used throughout the study. Five certified reference materials of water (ERM CA011b Hard

Fig. 1. Location of groundwater sampling points in the study area.

130

S. Butaciu et al. / Chemosphere 172 (2017) 127e137

Drinking Water UK - Metals, LGC 6010 Hard Drinking Water, ERMCA615 Groundwater, SRM 1643e Trace Elements in Water and LGC 6019 River Water-River Thames), METRANAL-32 Light Sandy Soil and METRANAL-34 Loam reference materials all purchased from LGC Promochem (Wesel, Germany) were used to check As, Naþ, Kþ, Ca2þ and Mg2þ quantification in water and total As and elements in the mineral matrix by XRF, respectively. Certified SPS-NUTRWW1 Waste water (Oslo, Norway) was used to validate the chromatographic method for anions determination. 2.3. Preparation of water samples for As determination and speciation Determination and speciation of arsenic as arsenite and arsenate (As(III) and As(V)) were performed using a nonchromatographic approach based on selective hydride generation capacitively coupled plasma microtorch optical emission spectrometry (HG-mCCP-OES) (Frentiu et al., 2014; Mihaltan et al., 2013). The speciation scheme of inorganic arsenic using the HG-mCCP-OES approach involves: (i) selective determination of As(III) immediately after sample mix with L-cysteine; (ii) determination of total As after the prereduction of As(V) and (iii) determination of As(V) as difference. The working conditions for arsenic speciation are presented in Supplementary material 4. 2.4. Preparation of water samples for the measurement of metals, anions and electrochemical parameters The concentrations of Ca2þ, Mg2þ, Naþ and Kþ were determined by capacitively coupled plasma microtorch optical emission spectrometry (mCCP-OES) using sample introduction by pneumatically nebulization (Frentiu et al., 2011). The standard addition method (3 additions) was used to compensate for the matrix effects. The an 3 ions (Cl, F, SO2 4 , NO3 and PO4 ) were quantified with 761 Compact IC Methrom ion chromatograph (Herisau, Switzerland) following a standardized protocol (ISO 10304-1, 2007). The pH was measured using the 540 GLP pH-meter (WTW GmbH Weilheim, Germany), while EC and TDS were determined using a Sunleaves TDS Expert Plus (Bloomington, USA). To convert from EC to TDS it was used the 442™ conversion factor recommended in the case of natural water. The Eh was measured according to the protocol described in reference (Eaton et al., 2005). Total water alkalinity expressed in mg L1 HCO 3 was determined by titration with 0.1 M HCl in the presence of methyl orange, in compliance with a standard protocol (SR EN ISO 9963e1, 2001). 2.5. Preparation of sand and clay samples for elemental and mineralogical analysis Sand and clay samples were dried at 105  C and sieved (<100 mm). Elemental and mineralogical analysis were carried out on powder pressed pellets by XRF using the Tracer 5i handheld Xray spectrometer and XRD using the Bruker D8 Advance Diffractometer Bruker (Massachusetts, USA). To highlight whether the water-rock interaction could be the source for the elevated As content and anions in groundwater, a leaching study in high purity water was conducted at a solid-to-liquid ratio of 1:2 in a Heidolph shaker at 16 rpm for 24 h at room temperature according to SR EN ISO 12457-1:2003. 2.6. Quality assurance/Quality control laboratory and on-site procedures The results obtained for the determination of As and metals in certified reference materials of water are presented in

Supplementary material 5. The accuracy of As determination by HG-mCCP-OES was of 100 ± 9%, while precision in the range 1.2e7.1% for As(III), 3.1e9.5% for As(V) and 2.9e8.5% for total As. Recovery of As species using the selective derivatization method was within 98e105% for As(III) and As(V), and 98e104% for total As (Cordos et al., 2006). The mCCP-OES method was validated for the determination of Ca2þ, Mg2þ, Naþ and Kþ in water with recovery in the range 93e107% and precision of 2.0e3.9%. The analysis of a Certified SPS-NUTRWW1 Waste water by ionic chromatography (Supplementary material 6) provided recovery in the range 95e107% for all anions. Standard deviations of repeatability in the analysis of water samples (n ¼ 3 replicates) (%) were: 3.3e11.1 F; 3 2 4.0e8.5 Cl-; 3.1e5.7 NO 3 ; 4.2e7.7 PO4 and 3.7e7.2% SO4 . Precision measurements of alkalinity was within 2.9e6.4% for n ¼ 3, while of the redox potential in the range 0.1e1.1%. The XRF method was validated with recovery in the range 80e140% and precision of 2.8e18.2% (Supplementary material 7). 2.7. Multivariate statistical methods CA and PCA are excellent exploratory tools for complex data interpretation and evaluation of water quality (Esmaeili-Vardanjani et al., 2015; Spanos et al., 2015; Al-Harbi et al., 2014; Sappa et al., 2014; Miller and Miller, 2000). Beside these approaches, we used in the present study FPCA and FHCC. The robust FPCA algorithm improves the PCA approach by fuzzification of the matrix data thus diminishing the influence of the outliers and poor linear correlation between variables. These facts result in greater accounting for total variance, low-dimensional matrix data and sharper delineation of principal components (Cundari et al., 2000). The efficiency of the FPCA algorithm was illustrated on a data set concerning the water quality of the Danube River by Sarbu and Pop (2005). The Ward's linkage method and Euclidian distance as a measure of similarity was chosen to group sites and physico-chemical parameters respectively, in CA. The PCA was based on the eigenvalues of the correlation matrix on standardized data and Varimax rotation was used to maximize the variation expressed by the principal components. Only PC's with eigenvalues >1 were retained. The robust FHCC was designed by Pop and Sarbu (1997) and was successfully applied for classification results of the toxicological responses of 32 in vivo and in vitro test systems to the first 10 MEIC chemicals (Sarbu and Pop, 2000), and modeling and predicting the fate of contaminants in environment (Frentiu et al., 2015; Pourjabar et al., 2014). The FHCC achieves a partition on hierarchical levels of samples and their characteristics. The final Fuzzy partitions contain the groups corresponding to the terminal nods of the binary classification tree according to the membership degree. Compared to CA and PCA, FHCC is a more advanced algorithm as it is able to identify simultaneously qualitative and quantitative characteristics responsible for similarities/dissimilarities within a group of samples, as well as associations of characteristics at each division level of the hierarchy. 3. Results and discussions 3.1. Hydrogeochemical characteristics and summary statistics Table 1 exhibits the hydrogeochemical characteristics and summary statistics (mean, median, minimum, maximum values and standard deviation) for the groundwater samples and the corresponding guideline values where appropriate. For the GWROBA03 aquifer the concentrations of As, Cl, SO2 and PO3 4 4 were compared against TVs in Order 621/2014. The levels of As, Naþ, F and EC in the GW-ROBA18 aquifer, for which TVs are not yet available, were examined in comparison with CMA in current

Table 1 Hydrogeochemical characteristics and summary statistics of groundwater samples (n ¼ 3 parallel samples). Aquifer

GW-ROBA03

1 2 5 17 3 4 6 7 8 9 10 11 12 13 14 15 16 18 Min Max Mean Median Std Skew Kurt MAC EU Legislationa TV GW-ROBA03b TV GW-ROBA18b Law 458/ 2002c

Depth (m)

4 17 30 20 100 100 100 70 45 56 120 120 70 45 70 60 100 100 4 120 76 80 36

Physicochemical parameters As content (mg L1)

Major constituents (mg L1)

As(III)

As(V)

As total

Na

K

Ca

Mg

F

Cl

NO 3

SO24

PO34

HCO 3

9.8 16.5 13.7 0.5 13.5 12.9 14.1 13.8 14.7 11.4 11.4 18.8 9.0 14.1 1.2 3.7 0.6 0.7 0.5 18.8 10.6 12.9 5.9 0.64 0.98

8.2 4.4 38.3 3.4 37.1 11.6 32.6 30.4 51.6 6.1 28.6 20.4 4.9 63.2 7.5 2.5 3.0 2.9 2.9 63.2 20.2 9.8 18.5 0.96 0.06

18.0 20.9 52.0 3.9 50.6 24.5 46.7 44.2 66.3 17.5 40.0 39.2 13.9 77.3 8.7 6.2 3.6 3.6 3.6 77.3 30.8 24.0 22.2 0.55 0.72

97.8 98.6 102 87 82.3 97.2 105 103 111 100 92.9 62.4 77.5 103 82.3 121 85.8 111 62.4 121.0 95.5 97.8 14.0 0.54 0.56

4.13 3.82 1.04 1.25 2.01 1.11 1.56 1.18 1.86 1.68 1.10 0.95 1.15 2.25 1.14 1.03 1.48 0.87 0.87 4.13 1.65 1.25 0.93 1.96 3.26

21.7 38.2 49.1 17.3 31.0 21.8 30.1 36.6 31.2 46.1 27.4 30.0 17.3 36.4 20.0 24.1 14.1 15.0 14.1 49.1 28.2 27.4 10.3 0.50 0.46

8.65 18.4 16.6 5.83 11.1 8.48 11.7 15.7 15.5 15.1 9.88 27.7 6.29 13.1 7.06 11.4 4.99 4.60 4.60 27.7 11.8 11.1 5.8 1.13 1.92

0.15 0.12 0.29 0.07 0.25 0.13 0.28 0.03 0.17 0.16 0.14 0.25 0.03 0.43 0.03 0.97 0.09 0.08 0.03 0.97 0.20 0.14 0.22 2.77 9.18

19 42 28.6 4.96 22.4 18.8 17.8 33.2 15.3 6.70 6.31 5.71 4.22 6.02 5.23 7.50 4.51 6.50 4.2 42.0 14.2 6.7 11.3 1.19 0.62

151 4.91 1.52 0.05 0.05 0.25 2.9 0.05 0.05 0.05 0.05 0.05 0.05 0.14 0.11 0.05 0.22 0.16 0.05 151.0 9.0 0.11 35.5 4.23 17.94 50

87.4 11.6 1.5 0.05 0.05 0.05 3.2 1.7 0.05 0.05 0.05 0.14 0.17 1.2 0.05 0.05 0.05 0.05 0.05 87.4 6.0 0.05 20.5 4.12 17.24

5.91 3.11 3.02 0.96 2.91 4.75 5.19 0.05 4.45 2.88 3.36 6.03 1.82 4.88 1.34 0.81 1.35 1.21 0.05 6.03 3.00 2.91 1.86 0.17 1.19

185 367 395 294 391 346 364 411 429 426 357 392 276 484 290 386 288 355 185 484 358 364 70 0.69 0.96

250 250 250

1.50 1.00

1.2

250 250 250

10 10 10

200

50

Eh (mV)

pH

EC (mS cm1)

TDS (mg L1)

402.2 409.1 395.1 374.2 374.5 380.1 388.8 385.4 382.1 378.9 379.6 406.6 380.4 381.1 369.0 378.9 372.1 374.3 369.0 409.1 384.0 380.1 11.8 1.06 0.15

8.39 8.45 8.45 8.42 8.35 8.41 8.44 8.49 8.38 8.46 8.41 8.33 8.31 8.48 8.37 8.46 8.40 8.43 8.31 8.49 8.41 8.41 0.05 0.47 0.46

821 558 519 398 391 438 426 418 464 482 477 406 345 503 406 516 388 472 345 821 468 438 104 2.40 7.75

566 379 353 267 262 294 286 281 316 333 324 273 232 342 273 351 260 321 232 566 317 294 74 2.39 7.66

6.5e9.5

2500

S. Butaciu et al. / Chemosphere 172 (2017) 127e137

GW-ROBA18

Sample

Notes: a Maximum Allowable Concentration for NO 3 in GW in European legislation (Annex 3 of Groundwater Directive 2006/118/EC). b Threshold Value for GW in Romania (Order 621/2014). c Maximum admitted concentration in drinking water according in Romanian (Law no. 458/08.07.2002 on drinking water quality).

131

132

S. Butaciu et al. / Chemosphere 172 (2017) 127e137

quality standard for drinking water in Romania (Law 458/2002). The NO 3 level was compared with Maximum Allowable Concentration (MAC) in EU legislation (Annex 3 to Directive 2006/118/EC). The TVs for all groundwater bodies in Romania were established using a methodology developed within a pilot project conducted on GW-ROBA03 (Radu et al., 2010) and are stipulated in Order 621/ 2014. The TVs derived from the NBLs for each groundwater body in Romania over the period 1961e2007 and CMA in drinking water (Law 458/2002). The NBLs were calculated as the 90 or 50 percentile results after removing samples suspected to be polluted. When NBLs  CMA, TV was set to be equal to CMA, while when NBLs  CMA, TV was set as 1.2xNBL. Data in Table 1 demonstrate an asymmetric distribution of parameters around the mean. Most of parameters present positive skewness, while four of them (As(III), Naþ, HCO 3 and pH) negative þ  skewness. Nitrate, SO2 4 , EC, TDS, K and F show leptokurtic distribution (kurtosis>3) caused by some outliers significantly different, while the other parameters a platykurtic distribution, which means fewer and less extreme outliers than in the normal distribution. Consequently the median values were taken into consideration for the following discussion. According to data in Table 1, water samples from both GWROBA03 and GW-ROBA18 exhibited elevated concentrations mainly of arsenic and phosphate. Total arsenic was in the range 3.6e77.3 mg L1 (median 24.0) and TV/CMA was exceeded in 72% of samples. The concentration of As (III) species of higher toxicity was within 0.5e18.8 mg L1 (median 12.9) surpassing TV/CMA in 60% of cases, while concentrations of As(V) species were 2.9e63.2 mg L1 (median 9.8) with exceeded guideline values in 50% of cases. In terms of distribution, the As(V) species was dominant accounting for (57 ± 20%) in GW samples with total As>10 mg L1 and 73 ± 18% in the others. In samples with total As content higher than median, the weight of As(V) was 69 ± 11%. Thus irrespective of the level of total As, the distribution of arsenic species is in agreement with the oxidative conditions of the aquifer (372.1e409.1 eV). The Eh-pH diagram of arsenic species showed that the dominant form of As(V) species was HAsO24 , confirmed by PHREEQC speciation modeling (Parkhurst and Appelo, 1999). The elevated contents of total As in GW samples collected from ROBA03 and ROBA18 were found to be similar to those reported for the surroundings, such as Pannonian Basin (eastern Hungary and western Romanian Plain) (0.5e208 mg L1) (Rowland et al., 2011) and Vojvodina (Northeastern Banat, Serbia) (0.5e231 mg L1) (Dimkic et al., 2010). In the same time the As concentrations found by us were much lower than in other groundwaters in other parts of the world (Bangladesh, West Bengal, Cambodia, Vietnam, Indonesia) consistent with local aquifer geology (Chakraborti et al., 2016; Kumar et al., 2016; Sarkar and Paul, 2016; Rahman et al., 2015a; Biswas et al., 2014b; Amini et al., 2008; Anawar et al., 2003; Smedley and Kinniburgh, 2002). Elevated total As concentration and the dominance of As(V) species were also reported in other oxidizing aquifers such as Viterbo area, Central Italy (Sappa et al., 2014), USA (Sorg et al., 2014), North East India (Ghosh and Sar, 2013) and Main Ethiopian Rift (Rango et al., 2013). The concentration of PO3 4 in samples coming from GW-ROBA03 was in the range 3.02e5.91 mg L1 exceeding TV (1.5 mg L1), while in samples from GW-ROBA18 in the range 0.05e6.03 mg L1 also exceeding TV (1.0 mg L1) in 12 out of 14 cases. The concentration of PO3 exceeding TVs in groundwater is consistent with the 4 mineral composition of aquifer in which the XRD analysis identified traces of Ca, K and Na phosphates. The concentrations were found to be similar to those reported for arsenic-rich shallow aquifer in the Bengal Basin (India) (1.25e4.25 mg L1) (Biswas et al., 2014b). Regarding nitrate, its concentration far surpassed MAC (50 mg L1) in a single sample collected from 4 m depth (151 mg L1) which

evidenced anthropogenic contamination from fertilizers used on agricultural fields, since groundwaters seldom contain more than 5e10 mg L1 nitrate. The same groundwater sample exhibited the highest concentration of sulfate (87.4 mg L1), however below TV. The pH of groundwater samples was alkaline (8.31e8.49; 8.41 median) with values in the range of Romanian guideline limits (6.5e9.2) for drinking water (Law no. 458/08.07.2002). A reason for the alkaline pH is the presence of Naþ and HCO 3 as major ions in groundwater and the loamy soil. According the River Basin's Management Plan of the Banat River Basin Administration, GW-ROBA03 is in poor chemical state because of nitrate, while GW-ROBA18 is in good status, which means that the elevated concentrations of phosphate and arsenic species found by us reflect only a local case (ANAR, 2015). 3.2. Groundwater facies The Gibbs diagrams (Gibbs, 1970) for cations (TDS vs. Naþ/ 1 Na þCa2þ) and anions (TDS vs. Cl/Cl þ HCO 3 ) (mg L ) revealed that all samples fall in the rock dominance region that strongly influences the groundwater type and geochemical composition. The leaching study performed on sand and clay samples provided As concentration in leachate in the range 0.03e30 mg L1, while 2 anions (mg L1): 0.01e0.5 F, 0.05e2 NO 3 , 0.05e10 SO4 , 0.05e2 PO3 . These results suggest that the rock-water interaction, namely 4 hydrolysis of silicates and phosphates, dissolution of Ca and Mg carbonates by rain water (pH~5.6) containing CO2 and desorptionadsorption processes from/on clay represent the main pathways  resulting in elevated content of As species, PO3 4 and HCO3 , and alkaline pH. The origin and hydrogeochemical facies of groundwater were identified and grouped by Piper trilinear diagram (Sajil Kumar, 2013; Piper, 1944), which classifies water in six fields for visualizing the relative abundance of the common ions. In our case, the Piper plot (Supplementary material 8) showed that most of groundwater samples (13 out of 18) were confined in the Naþ-HCO-3 facies, typical for fine sediments and oxidizing aquifers, four samples (1; 2; 5; 7) fall in the Naþ-Ca2þ-HCO-3 facies and a single one (11) belonged to Mg2þ-HCO-3 type consistent with coarser layers. Results are in agreement with characteristics of the Quaternary groundwater, in which the major cations are Naþ, Ca2þ and Mg2þ, the main anion is HCO 3 and pH is alkaline. The Stuyfzand Hydrogeochemical Classification System (SHCS) (Stuyfzand, 1989) defined the groundwater samples as oligohaline main type (g) according to Cl concentration in the range 4e42 mg L1, moderately high alkaline type (3) according to HCO 3 concentration in the range 3e8 meq L1, NaþeHCO-3 subtype (16 samples) according to dominant cation and anion and class (þ) according to salinity, which gave (NaþþKþþMg2þ)surplus. Sample (1) had the features of þ 2þ  Namix (HCO 3 , NO3 ), while sample (11) that of Mix (Na , Mg ) HCO . 3 þ

3.3. Chemometric models The CA was the first chemometric method applied to the data matrix. The dendrogram (Supplementary material 9) divided the groundwater chemical parameters into two clusters, C1 and C2, based on similarities among characteristics. The first cluster is attributed to the general features of groundwater (EC, TDS, Eh and HCO 3 ) describing salinity, redox status and alkalinity. The second cluster reveals the more complex relationships among cations  3   (Naþ, Ca2þ, Mg2þ, Kþ) and anions (SO2 4 , NO3 , PO4 , F , Cl ), which could be associated to longtime rock-groundwater interaction or  3  land use activities (SO2 4 , NO3 , PO4 , Cl ) affecting water quality. The presence of arsenic species in this cluster supports the idea of

S. Butaciu et al. / Chemosphere 172 (2017) 127e137

natural origin of arsenic from As bearing Ca- and Mg- rocks as suggested by the associations among Ca2þ, Mg2þ, pH, As(III), As(V) and total arsenic. CA performed on the 18 groundwater samples analyzed within this study (Supplementary material 10) identified two groups (C1, 9 samples) and (C2, 8 samples) based on similar characteristics related to lithology, while sample 1 remained outside. Supplementary material 11 provides the chemical parameters for each cluster and highlights those responsible for the clustering pattern. Cluster C1 encompasses samples of low concentrations of As species and total As, cations and anions, low EC and TDS, while cluster C2 relates to samples with elevated As concentration (>50 mg L1) and high values of the other chemical parameters. C1 is further divided in two subclusters, one (C11) including groundwater samples exhibiting total arsenic below 10 mg L1 and the lowest values of the other characteristics. The other subgroup (C12) covers samples with total As below 50 mg L1 and intermediate values for parameters excepting Mg2þ, Cl and PO3 4 . Sample 1 with elevated As content does not belong to any cluster and differentiates itself by 2 1 1 the highest content of NO 3 (151 mg L ), SO4 (87.4 mg L ), TDS 3 1 1 (566 mg L ), high conductivity (821 mS cm ), high PO4 content (5.91 mg L1), but lowest alkalinity (185 mg L1 HCO 3 ). Factor loadings (Varimax normalized) after rotation describing variability of groundwater parameters show that the five principal components describe 90.2% of total variance (Table 2). The first factor (F1) accounting 29.9% of total variance can be associated to land activities, such as use of NPK fertilizers with higher impact on shallow groundwater. Very probably this is the 2 3 þ reason for which the concentrations of NO 3 , SO4 , K and PO4 are much higher in the groundwater sample (1) collected from only 4 m depth, which is under the anthropogenic contamination risk from fertilizer application, compared to depth groundwater. Cuoco et al. (2015) found that any deviation from the natural HCO 3 -Cl 22 SO4 feature toward SO4 in shallow groundwater was accompanied by an enrichment of NO 3 of anthropogenic origin. The con  centrations of SO2 4 and NO3 increased as HCO3 decreased. The low and relatively homogenous concentrations of anions, namely below 2 maximum allowable (NO 3 )/threshold value (SO4 ) at greater depth suggest their natural origin in the groundwater under study. The lack of correlation between As and SO2 4 in F1 shows that oxidation of sulfides is not involved in the release of As from aquifer. The weak factor loadings of all As species in this factor denote that there

Table 2 Factor loadings after Varimax rotation describing variability of groundwater chemical composition (loading value > 0.75 corresponds to strong relationshipebold face).

F Cl NO 3 PO3 4 2 SO4 Ca2þ Mg2þ Naþ Kþ HCO 3 As(III) As(V) As total Eh pH Conductivity TDS Variance/ %

F1

F2

F3

F4

F5

0.020 0.229 0.961 0.479 0.974 0.044 0.071 0.189 0.797 0.500 0.090 0.080 0.042 0.486 0.020 0.937 0.935 29.9

0.056 0.069 0.070 0.547 0.086 0.415 0.264 0.115 0.063 0.558 0.582 0.959 0.943 0.011 0.049 0.011 0.010 17.9

0.334 0.422 0.127 0.398 0.096 0.413 0.074 0.863 0.086 0.415 0.085 0.128 0.083 0.110 0.908 0.245 0.246 15.1

0.103 0.595 0.073 0.334 0.013 0.683 0.903 0.152 0.236 0.369 0.756 0.108 0.287 0.820 0.108 0.124 0.121 19.8

¡0.852 0.507 0.032 0.197 0.067 0.025 0.193 0.228 0.239 0.244 0.074 0.013 0.008 0.002 0.068 0.131 0.134 7.5

133

is no anthropogenic input of As from human activities and the high concentration in groundwater in the area under study is principally governed by adsorption-desorption on/from mineral phases (clays) and silicates hydrolysis in aquifer. The observations are supported by the relationships in the next factors. Therefore, F2 (17.9%), F3 (15.1%) and F4 (19.8%) describe the natural enrichment of As species in groundwater from the aquifer in the Banat Plain under the in3 fluence of HCO 3 and PO4 and oxidizing conditions. The positive 3 loadings of As species, HCO 3 and PO4 stand for the increase of mobility of As species from aquifer in groundwater under the influence of these anions. Our findings are in agreement with the observation of Sappa et al. (2014), Anawar et al. (2003), Rango et al. (2013) and Kumar et al. (2016) who found positive correlation between As concentration and the combined effect of Naþ, HCO 3 and pH in groundwater from Central Italy, Bangladesh, main Ethiopian Rift Aquifer and Diphu, Assam, Northeastern India. In the case of groundwater in the Banat Plain, hydrolysis of silicates and dissolution of Ca/Mg carbonate due to water-rock interactions lead to pH increase and HCO 3 concentration and further to the release of As species from the aquifer rocks. The alkaline pH in groundwater is consistent with the positive relationship of this parameter with the factor loadings of Naþ and HCO 3 in F3. 3The PO34 eAsO4 positive correlation in F2 suggests that the ion exchange process in aquifer sediment is also responsible for the enrichment of groundwater with arsenate species. The same correlation was found by Biswas et al. (2014b) in shallow aquifers of the Bengal Basin contaminated with arsenic. In another study 3(Biswas et al., 2014a) on the role of PO34 eAsO4 competing ions in enrichment of groundwater with As species, the authors found that in the absence of PO3 4 the modeled concentration of As decreased by more than 90%. F3 and F5 explain the appearance of F in groundwater on a different path than arsenic under the action of HCO 3 and alkaline pH. The associated process is calcium fluoride dissolution under þ 2þ HCO ion exchange to 3 action, precipitation of CaCO3 and Na eCa compensate for the excess of negative charges (Handa, 1975) (Eq. (1)).  þ CaF2ðsÞ þ HCO 3 ⇔CaCO3ðsÞ þ 2F þ H

(1)

The results were confirmed by the XRD analysis which identified traces of fluoride containing minerals in clay samples. Co-occurrence of As species and F in groundwater via different pathways was also signaled in other parts of the world (Kumar et al., 2016). The modeling using FPCA generates stronger grouping of chemical parameters so that the first 3 factors explain more than 99% of variability of the groundwater samples. The threedimensional normal PCA, Varimax rotated PCA and FPCA provide grouping of characteristic parameters (Fig. 2). Spatial grouping of samples using these statistical approaches are presented in Fig. 3. The grouping of parameters in graphics corresponding to Varimax rotated PCA is close to representation of FPCA. The strength of FPCA can be easily emphasized in the spatial grouping of samples that is not affected by outliers. Fig. 3 (FPCA) highlights two groups of samples, with As concentration below 10 mg L1 and elevated level according to the occurrence governed by water-rock interactions. It can be remarked the low influence of the outlier (sample 1) substantially different from the others in terms of high sulfate and nitrate concentration on the grouping of the other samples when FPCA was used. All these approaches fail to provide a simultaneous classification of groundwater samples and associated chemical parameters. Instead the FHCC algorithm allowed for the identification of qualitative and quantitative characteristics responsible for (dis)

134

S. Butaciu et al. / Chemosphere 172 (2017) 127e137

Fig. 2. Three-dimensional normal PCA (a), Varimax rotated PCA (b) and FPCA (c) showing inter-relationship among 17 variables.

Fig. 3. Three-dimensional normal PCA (a), Varimax rotated PCA (b) and FPCA (c) showing the grouping of 18 sampling sites.

S. Butaciu et al. / Chemosphere 172 (2017) 127e137

135

Table 3 The hard fuzzy partition of groundwater samples and their characteristics ranked in descending order of their membership degree. Partition level

Fuzzy divisive partition

Samples Location

0

A

1, …, 18

1

A1

6, 8, 7, 5, 15, 13, 2, 9, 18, 3, 10, 1, 4 17, 16, 18, 14, 12, 15 6, 8, 7, 5, 13, 9, 2, 3, 11,10, 4 1 3, 11 6, 8, 5, 9, 7, 13, 2, 10, 4 5, 9, 6, 13, 2, 7, 10, 4 8

2

3 4

A2 A11 A12 A111 A112 A1121 A1122

Associated characteristics Membership degree range

Characteristics 

0.996e0.536



NO 3,

PO3 4 ,

Membership degree range SO2 4 ,

2þ

Astotal, HCO 3,

As(V), Mg , Naþ, F, pH

0.611 0.738e0.724 0.981e0.839 0.937e0.801 0.938

þ  EC, TDS, SO2 4 , NO3 , K F Naþ, pH pH Naþ

3.4. Chemical conceptual model of groundwater in the Banat Plain A chemical conceptual model describing the natural pathway in the occurrence of As and F species and the anthropogenic factors 2 3 þ in the appearance of NO 3 , SO4 , PO4 and K , respectively was developed based on the hydrogeochemical characteristics of the aquifer and chemometric models (Supplementary material 14). Silicate hydrolysis and carbonate dissolution directly release As species and generate high-pH and NaþeHCO-3 groundwater facies. These conditions in the oxidizing aquifer induce further desorption of arsenate from the clay minerals which occur generally around 8.5 pH (Masscheleyn et al., 1991). The elevated PO3 4 concentration in groundwater of anthropogenic source from NPK fertilizers or natural source from phosphate containing rocks results in an enrichment of arsenate content in local groundwater due to ion

2þ

þ

þ

Ca , Mg , Na , K , F , Cl , HCO 3 , As(III), As(V), Astotal, Eh, pH, EC, TDS  þ þ 2 F , Na , EC, TDS, K , SO4 , NO 3 , pH

0.982e0.566 0.990e0.531

similarities of groundwater samples, and the association of characteristics at each partition level in the hierarchy (Table 3). Membership degrees corresponding to the fuzzy partition history are presented in Supplementary material 12. The groundwater samples were divided into 4 partition levels, of which the first two contain chemical parameters related to naturally arsenic enriched groundwater. Sample partition expresses the enrichment of groundwater with arsenic. Thus, the first group A1 encompasses samples exceeding TV (10 mg L1 As). Their chemical characteristics refer to parameters of geogenic (F, Naþ) and anthropogenic/mixed  origin (Kþ, SO2 4 , NO3 ) with restricted concentration in groundwater or drinking water as well as general features (conductivity, TDS and pH). Under these conditions, quality monitoring of groundwater naturally enriched with As should include both arsenic speciation/total As determination and the hydrogeochemical parameters mentioned above. The second class A2 contains mainly water samples with less than TV and a single one (12) in which TV is slightly surpassed. However, the membership degree of this sample to A2 is small and its position is rather intermediary between A1 and A2. The characteristics associated to 3 2þ  class A2 are As(V), Mg, total As, As(III), HCO 3 , Ca , Cl , Eh and PO4 . The water samples containing As over TV (A1) were further divided into 2 levels (A11, A12). The chemical characteristics linked to A11 suggest the possibility to differentiate the group of water samples naturally enriched with arsenic through the NaþeF-epH association. The partition level A12 contains a single sample (1) substan þ tially different from the others by high SO2 4 , NO3 and K concentration of anthropogenic origin via fertilizers. Two-way joining plot (Supplementary material 13) demonstrates that the simultaneous plotting of samples and chemical characteristics as a function of membership degree to a specific group is very helpful in establishing pattern recognition for groundwater.

2þ

2þ

Ca



, As(III), Cl ,

Eh, PO34

1.000e0.806 0.995e0.772 0.992e0.836 0.959e0.719 0.988 0.985; 0.827 0.827 0.985

exchange processes in aquifer. The dominant species of As is HAsO2 4 according to EhepH conditions in aquifer. The predominance of NaþeHCO-3 groundwater facies is also responsible for the 2 natural occurrence of F from fluorite. The levels of NO 3 , SO4 , þ PO3 4 and K could be linked to the use of NPK fertilizers. 4. Conclusions It was demonstrated the advantage of coupling unsupervised multivariate statistical approaches (CA, PCA, FPCA, FHCC) with classical examination by diagrams to provide advanced modeling of arsenic chemistry and high level of As and other species in local groundwater. This methodology allowed an improved understanding related to sources of inorganic arsenic species and main anions and cations in groundwater in the Banat Plain, and highlighted the hydrogeochemical characteristics describing the system 3 variability. The feature of groundwater (HCO 3 and PO4 and alkaline pH) plays an important role in releasing inorganic arsenic species from aquifer. Land use activities do not influence the concentration of As and F species in local groundwater but result in þ 2 3 the increase of NO 3 , SO4 , PO4 , K from the applied NPK fertilizer. Varimax rotated PCA, FPCA and FHCC statistical approaches showed that occurrence and enrichment of arsenic species in groundwater are mainly of natural sources. Thus water-Ca/Mg rock interaction 3and PO34 eAsO4 ion exchange in aquifer are the main processes responsible for arsenic enrichment in groundwater. Fluoride ion occurs on a different path than As species and is the result of calcium fluoride dissolution under HCO 3 action and alkaline pH. Also, the F - Naþ e pH clustering was found as a feature of groundwater naturally enriched with arsenic in the Banat Plain. Modeling using FHCC has associated to groundwater with elevated arsenic concentration several characteristics with restricted values (F, Naþ,  3 Kþ, SO2 4 , NO3 , PO4 , conductivity, TDS and pH). The FHCC approach was more consistent than CA, Varimax rotated PCA or FPCA in modeling as it allows a rapid identification of parameters associated to arsenic occurrence in groundwater. Acknowledgements This work was supported by a grant of the Romanian National Authority for Scientific Research, CNDIeUEFISCDI, project number PN-II-PT-PCCA-2011-3.2-0219 (Contract no. 176/2012). The authors would like to thank Professor C. Baciu, Faculty of Environmental Science and Engineering, Cluj-Napoca, Romania for the helpful discussion. Appendix A. Supplementary data Supplementary data related to this article can be found at http://

136

S. Butaciu et al. / Chemosphere 172 (2017) 127e137

dx.doi.org/10.1016/j.chemosphere.2016.12.130. References Aksever, F., Karaguzel, R., Mutluturk, M., 2015. Evaluation of groundwater quality and contamination in drinking water basins: a case study of the SenirkentUluborlu basin (Isparta-Turkey). Environ. Earth Sci. 73, 1281e1293. Al-Harbi, M., Al-Ruwaih, F.M., Alsulaili, A., 2014. Statistical and analytical evaluation of groundwater quality in Al-Rawdhatain field. Environ. Prog. Sustain. Energy 33, 895e904. Amini, M., Abbaspour, K.C., Berg, M., Winkel, L., Hug, S.J., Hoehn, E., Yang, H., Johnson, C.A., 2008. Statistical modeling of global geogenic arsenic contamination in groundwater. Environ. Sci. Technol. 42, 3669e3675. ANAR, 2015. Water Management Report (In Romanian). http://www.rowater.ro/ dabanat/GOSPODARIREA%20APELOR (Accessed 12 December 2016). Anawar, H.M., Akai, J., Komaki, K., Terao, H., Yoshioka, T., Ishizuka, T., Safiullah, S., Kato, K., 2003. Geochemical occurrence of arsenic in groundwater of Bangladesh: sources and mobilization processes. J. Geochem. Explor 77, 109e131. Annex 3. To the Commission staff working document accompanying in the report from the Commission in accordance with article 3.7 of the Groundwater Directive 2006/118/EC on the establishment of groundwater threshold values. Information on the groundwater threshold values of the Member States. http:// ec.europa.eu/environment/water/water-framework/groundwater/pdf/com_ swd_annex_iii.pdf. (Accessed 24 December 2015). Biswas, A., Gustaffson, J.P., Neidhardt, H., Kundu, A.K., Chatterjee, D., Berner, Z., Bhattacharya, P., 2014a. Role of competing ions in the mobilization of arsenic in groundwater of Bengal Basin: insight from surface complexation modeling. Water Res. 55, 30e39. Biswas, A., Neidhardt, H., Kundu, A.K., Halder, D., Chatterjee, D., Berner, Z., Jacks, G., Bhattacharya, P., 2014b. Spatial, vertical and temporal variation of arsenic in shallow aquifers of the Bengal basin: controlling, geochemical processes. Chem. Geol. 387, 157e169. Bloom, M.S., Fitzerald, E.F., Kim, K., Neamtiu, I., Gurzau, E.S., 2010. Spontaneous pregnancy loss in humans and exposure to arsenic in drinking water. Int. J. Hyg. Environ. Health 213, 401e413. Bloom, M.S., Surdu, S., Neamtiu, I., Gurzau, E.S., 2014. Maternal arsenic exposure and birth outcomes: a comprehensive review of the epidemiologic literature focused on drinking water. Int. J. Hyg. Environ. Health 217, 709e719. Bonte, M., van Breukelen, B.M., Stuyfzand, P.J., 2013. Temperature-induced impacts on groundwater quality and arsenic mobility in anoxic aquifer sediments used for both drinking water and shallow geothermal energy production. Water Res. 47, 5088e5100. Carrillo-Rivera, J.J., Cardona, A., Edmunds, W.M., 2002. Use of abstraction regime and knowledge of hydrogeological conditions to control high-fluoride concentration in abstracted groundwater: San Luis Potosi basin, Mexico. J. Hydrol. 261, 24e47. Chakraborti, D., Rahman, M.M., Ahamed, S., Dutta, R.N., Pati, S., Mukherjee, S.C., 2016. Arsenic groundwater contamination and its health effects in Patna district (capital of Bihar) in the middle Ganga plain, India. Chemosphere 152, 520e529. Cordos, E.A., Frentiu, T., Ponta, M., Abraham, B., Marginean, I., 2006. Optimization of analytical parameters in inorganic arsenic (III and V) speciation by hydride generation using L-cysteine as prereducing agent in diluted HCl medium. Chem. Spec. Bioavailab. 18, 1e9. Crane, M., Babut, M., 2007. Environmental quality standards for Water Framework Directive priority substances: challenges and opportunities. Integr. Environ. Assess. Manag. 3, 290e295. Crane, M., Kwok, K.W.H., Wells, C., Whitehouse, P., Lui, G.C.S., 2007. Use of field data to support European Water Framework Directive quality standards for dissolved metals. Environ. Sci. Technol. 41, 5014e5021. Cundari, T.R., Deng, J., Pop, H.F., Sarbu, C., 2000. Structural analysis of transition metal ? - X Substituent interactions. Towards the use of Soft computing methods for catalyst modeling. J. Chem. Inf. Comput. Sci. 40, 1052e1061. Cuoco, E., Darrah, T.H., Buono, G., Verrengia, G., De Francesco, S., Eymold, W.K., Tedesco, D., 2015. Inorganic contaminants from diffuse pollution in shallow groundwater of the Campanian Plain (Southern Italy). Implications for geochemical survey. Environ. Monit. Assess. 187, 1e17. Dimkic, M., Djuric, D., Josipovic, J., Jevtic, G., 2010. Solutions for groundwater management in areas affected by high arsenic content: Vojvodina case study. In: Intern. Conf. Transboundary Aquifers: Challenges and New Directions (ISARM). Directive 2000/60/EC of the European Parliament and of the Council establishing a framework for the Community action in the field of water policy, 2000. Off. J. Eur. Communities L327, 1e73. Directive 2006/118/EC of the European Parliament and of the Council of 12 December on the protection of groundwater against pollution and deterioration, 2006. Off. J. Eur. Communities L372, 19e31. Eaton, A.D., Clesceri, L.S., Greenberg, A.E., 2005. Standard Methods for the Examination of Water and Wastewater, Method 2580B Oxidation-reduction Potential Measurement in Clean Water, nineteenth ed. American Public Health Association, Washington DC. 2e74-2-76. Enciu, P., Balteanu, D., Dumitrica, C., 2014. Contributions to the knowledge of quaternary formations in the southwest Romanian Plain. Quatern. Intern 357, 1e12.

Esmaeili-Vardanjani, M., Rasa, I., Amiri, V., Yazdi, M., Pazand, K., 2015. Evaluation of groundwater quality and assessment of scaling potential and corrosiveness of water samples in Kadkan aquifer, Khorasan-e-Razavi Province. Iran. Environ. Monit. Assess. 187, 1e18. Frederick, L., VanDerslice, J., Taddie, M., Malecki, K., Gregg, J., Faust, N., Johnson, W.P., 2016. Contrasting regional and national mechanisms for predicting elevated arsenic in private wells across the United States using classification and regression trees. Water Res. 91, 295e304. Frentiu, T., Butaciu, S., Ponta, M., Darvasi, E., Senila, M., Petreus, D., Frentiu, M., 2014. Simultaneous determination of As and Sb in soil using hydride generation capacitively coupled plasma microtorch optical emission spectrometry e comparison with inductively coupled plasma optical emission spectrometry. J. Anal. At. Spectrom. 29, 1880e1888. Frentiu, T., Petreus, D., Senila, M., Mihaltan, A.I., Darvasi, E., Ponta, M., Plaian, E., Cordos, E., 2011. Low power capacitively coupled plasma microtorch for simultaneous multielemental determination by atomic emission using microspectrometers. Microchem. J. 97, 188e195. Frentiu, T., Ponta, M., Sarbu, C., 2015. Prediction of the fate of Hg and other contaminants in soil around a former chlor-alkali plant using Fuzzy Hierarchical Cross-Clustering. Chemosphere 138, 96e103. Ghosh, S., Sar, P., 2013. Identification and characterization of metabolic properties of bacterial populations recovered from arsenic contaminated ground water of North East India (Assam). Water Res. 47, 6992e7005. Gibbs, R., 1970. Mechanisms controlling world water chemistry. Science 170, 1088e1090. Gurzau, E.S., Gurzau, A.E., 2001. Arsenic in drinkingwater from groundwater in Transylvania, Romania: an overview. In: Chappell, W.R., Abernathy, C.O., Calderon, R.L. (Eds.), Arsenic Exposureand Health Effects IV. Elsevier. Handa, B.K., 1975. Geochemistry and genesis of fluoride-containing groundwaters in India. J. Ground Water 13, 275e281. ISO 10304e1, 2007. Water Quality - Determination of Dissolved Anions by Liquid Chromatography of Ions - Part 1: Determination of Bromide, Chloride, Fluoride, Nitrate, Nitrite, Phosphate and Sulfate. Kelepertzis, E., 2014. Investigating the sources and potential health risks of environmental contaminants in the soil and drinking waters from the rural clusters in Thiva area (Greece). Ecotox. Environ. Saf. 100, 258e265. Kim, K.H., Yun, S.T., Park, S.S., Joo, Y., Kim, T.S., 2014. Model-based clustering of hydrochemical data to demarcate natural versus human impacts on bedrock groundwater quality in rural areas, South Korea. J. Hydrol. 519, 626e636. Kumar, M., Das, A., Das, N., Goswami, R., 2016. Co-occurrence perspective of arsenic and fluoride in the groundwater of Diphu, Assam, Northwestern India. Chemosphere 150, 227e238. Law no. 458/08.07.2002 on drinking water quality (in Romanian). Masscheleyn, P.H., Delaune, R.D., Patrick, W.H., 1991. Effect of redox potential and pH on arsenic speciation and solubility in a contaminated soil. Environ. Sci. Technol. 25, 1414e1419. Mihaltan, A.I., Frentiu, T., Ponta, M., Petreus, D., Frentiu, M., Darvasi, E., Marutoiu, C., 2013. Arsenic and antimony determination in non- and biodegradable materials by hydride generation capacitively coupled plasma microtorch optical emission spectrometry. Talanta 109, 84e90. Miller, J.N., Miller, J.C., 2000. Statistics and Chemometrics for Analytical Chemistry, 4th ed. Prentice Hall, Pearson Education Ltd., London. ^niei (Geological Structure Mutihac, V., 1990. Structura Geologica a Teritoriului Roma of Romania). In Romanian. Tech. Publ. House, Bucharest, 419 pp. Neamtiu, I., Bloom, M.S., Gati, G., Goessler, W., Surdu, S., Pop, C., Braeuer, S., Fitzgerald, E.F., Baciu, C., Lupsa, I.R., Anastasiu, D., Gurzau, E.S., 2015. Pregnant women in Timis County, Romania are exposed primarily to low-level (<10mg/l) arsenic through residential drinking water consumption. Int. J. Hyg. Environ. Health 218, 371e379. Order no. 621/2014 on the approval of threshold values for groundwater in Romania. Panagiotaras, D., Panagopoulos, G., Popoulis, D., Avramidis, P., 2012. Arsenic Geochemistry in groundwater system. In: Panagiotaras, D. (Ed.), Geochemistry Earth's System Processes. InTech Europe, University Campus STeP Ri, Rijeka, Croartia, pp. 27e38. ^nia (Groundwaters from Romania). in Pascu, M.R., 1983. Apele Subterane din Roma Romanian. Tech. Publ. House, Bucharest, 412 pp. Parkhurst, D.L., Appelo, C.A.J., 1999. Users Guide to PHREEQC (Version 2): a Computer Program for Speciation, Batch-reaction, One-dimensional Transport, and Inverse Geochemical Modeling, pp. 99e4259. US Geological Survey Water Resources Investigation Report. Piper, A.M., 1944. A graphic procedure in the geochemical interpretation of wateranalyses. Trans. Am. Geophys. Union 25, 914e928. Pop, H., Sarbu, C., 1997. The fuzzy hierarchical cross-classification algorithm, improvements and comparative study. J. Chem. Inf. Comput. Sci. 37, 510e516. Pourjabar, A., Sarbu, C., Kostarelos, K., Einax, J.W., Buchel, C., 2014. Fuzzy hierarchical cross-clustering of data from abandoned mine site contaminated with heavy metals. Comput. Geosci. U. K. 72, 122e133. Radu, E., Balaet, R., Vliegenthart, E., Schipper, P., 2010. Derivation of threshold values for groundwater in Romania, in order to distinguish point & diffuse pollution from natural background levels. Environ. Eng. Res. 15, 085e091. Radu, E., Radu, C., 2011. Lithological and hydrogeological considerations on the phreatic aquifer from Bega e Timis interfluve, Banat Plain, Romania, with special regard to the unsaturated zone (CC Waters Project). In: Proceedings of IWA Specialist Groundwater Conference, Belgrade, Serbia. ISBN 978 e 86 e

S. Butaciu et al. / Chemosphere 172 (2017) 127e137 82565 e 31 e 4, p.157 e 162. Rahman, M., Al Mamun, A., Karim, M.R., Islam, K., Al Amin, H., Hossain, S., Hossain, M.I., Saud, Z.A., Noman, A.S.M., Miyataka, H., Himeno, S., Hossain, K., 2015b. Association of total arsenic in drinking water, hair and nails with serum vascular endothelial growth factor in arsenic-endemic individuals in Bangladesh. Chemosphere 120, 336e342. Rahman, M.M., Dong, Z., Naidu, R., 2015a. Concentrations of arsenic and other elements in groundwater of Bangladesh and West Bengal, India: potential cancer risk. Chemosphere 139, 54e64. Rango, T., Vengosh, A., Dwyer, G., Bianchini, G., 2013. Mobilization of arsenic and other naturally occurring contaminants in groundwater of the Main Ethiopian Rift aquifer. Water Res. 47, 5801e5818. Romanian National Institute of Statistics, Statistical Yearbook, 2012. Available at http://www.insse.ro/cms/en/content/statistical-yearbook-2012 (Accessed 24 December 2015). Rowland, H.A.L., Omoregie, E.O., Millot, R., Jimenez, C., Mertens, J., Baciu, C., Hug, S.J., Berg, M., 2011. Geochemistry and arsenic behaviour in groundwater resources of the Pannonian Basin (Hungary and Romania). Appl. Geochem 26, 1e17. Sajil Kumar, P.J., 2013. Interpretation of groundwater chemistry using Piper and Chandha's diagrams: a comparative study from Perambalur Taluk. Geosci 54, 12208e12211. Sappa, G., Ergul, S., Ferranti, F., 2014. Geochemical modeling and multivariate statistical evaluation of trace elements in arsenic contaminated groundwater systems of Viterbo area, (Central Italy). SpringerPlus 3, 237. Sarbu, C., Pop, H., 2000. Fuzzy classification of the first 10 MEIC. Chemosphere 40, 513e520. Sarbu, C., Pop, H., 2005. Principal component analysis versus fuzzy principal component analysis. A case study: the quality of Danube water (1985e1996). Talanta 65, 1215e1220. Sarkar, A., Paul, B., 2016. The global menace of arsenic and its conventional remediation-A critical review. Chemosphere 158, 37e49.

137

Sener, E., Sener, S., 2015. Evaluation of groundwater vulnerability to pollution using fuzzy analytic hierarchy process method. Environ. Earth Sci. 73, 8405e8424. Shankar, K., Aravindan, S., Rajendran, S., 2011. Hydrochemical profile for assessing the groundwater quality of Paravanar River sub-basin, Cuddalore District, Tamil Nadu, India. Curr. World Environ. 6, 45e52. Smedley, P.L., Kinniburgh, D.G., 2002. A review of the source, behaviour and distribution of arsenic in natural waters. Appl. Geochem 17, 517e568. Sorg, T.J., Chen, A.S.C., Wang, L.L., 2014. Arsenic species in drinking water in the USA with high arsenic concentrations. Water Res. 48, 156e169. Spanos, T., Ene, A., Simeonova, P., 2015. Chemometric expertise of the quality of groundwater sources for domestic use. J. Environ. Sci. Health Part A 50, 1099e1107. SR EN ISO 12457e1, 2003. Characterization of Waste. Leaching. Compliance Test for Leaching of Granular Waste Materials and Sluges. Part 1: One Stage Batch Test at Liquid to Solid Ratio of 2L/kg for Materials with High Content and with Particle Size below 4 mm. SR EN ISO 9963e1, 2001. Water QualityeDetermination of Alkalinity-Part 1: Determination of Total and Composite Alkalinity (In Romanian). Stuyfzand, P.J., 1989. An accurate, relatively simple calculation of the saturation index of calcite for fresh to salt water. J. Hydrol. 105, 95e107. Surdu, S., Bloom, M.S., Neamtiu, I.A., Pop, C., Anastasiu, D., Fitzgerald, E.F., Gurzau, E.S., 2015. Consumption of arsenic-contaminated drinking water and anemia among pregnant and non-pregnant women in northwestern Romania. Environ. Res. 140, 657e660. U.S. EPA, 2001. National Primary Drinking water regulations; arsenic and clarifications to compliance and new source contaminants monitoring; Final Rule. Fed. Regist. 66 (14), 6975e7066. http://www.epa.gov/sbrefa/documents/pnl14f. pdf (Accessed 24 12 2015). World Heath Organization (WHO, Geneva, Switzerland), 2004. Some Drinking Water Disinfectants and Contaminants Including Arsenic.