Environ Sci Pollut Res DOI 10.1007/s11356-016-6089-3

RESEARCH ARTICLE

Hydrogeochemistry of arsenic pollution in watersheds influenced by gold mining activities in Paracatu (Minas Gerais State, Brazil) Edison Bidone 1 & Zuleica Castilhos 1,2 & Ricardo Cesar 3 & Maria Carla Santos 1 & Ricardo Sierpe 1 & Marcos Ferreira 1

Received: 29 June 2015 / Accepted: 11 January 2016 # Springer-Verlag Berlin Heidelberg 2016

Abstract The aim of this study is to evaluate total arsenic (As) concentrations in drinking water (main pathway of human exposure) and its hydrogeochemical controls in the BMorro do Ouro^ gold mine region, which is the largest gold mine in Brazil, characterized by gold-arsenopyrite association. Arsenic concentration was generally below the detection limit (LOD < 0.5 μg L−1). Thus, water ingestion may not be a significant exposure pathway to local population. Low groundwater As concentrations (<1 μg L−1) are likely due to ore body structural setting, which plunges from 10° to >20°, being readily covered by thick phyllites that are poor in As some hundreds of meters away from the mine. Thirty-five percent of As levels in superficial waters (<0.5 to 40 μg L −1 ) were >10 μg L−1, which is the maximum permissible value for human ingestion. The highest concentrations were found nearby mine facilities and old artisanal mining areas surrounding the mine, decreasing downstream. Undisturbed watersheds showed As concentrations close to LOD. Hydrogeochemical data stress the sorption (adsorption and co-precipitation) of As role, mainly by Fe oxyhydroxides, as a geochemical filter that

Responsible editor: Philippe Garrigues * Edison Bidone [email protected]

1

Department of Environmental Geochemistry, Fluminense Federal University, UFF, Outeiro São João Batista, s/n. Centro, Niterói, RJ, Brazil

2

Center for Mineral Technology, CETEM/MCTI, Cidade Universitária, Av. Pedro Calmon, 900, Rio de Janeiro, RJ, Brazil

3

Department of Geography, Federal University of Rio de Janeiro, UFRJ, Av. Athos da Silveira Ramos, 274–Cidade Universitária, Rio de Janeiro, RJ, Brazil

retains As, attenuating its concentration in both superficial and groundwater. Such minerals are abundant in the region oxisols, sediments, and phyllites and may form stable mineral complexes with As under the pH (mostly neutral) and Eh (reduced environment) conditions found in the field. It has been demonstrated that As(III) (more toxic) and As(V) coexist in the analyzed waters and that As(V) predominates in superficial water. Keywords Arsenic . Drinking water . Superficial water . Groundwater . Contamination . Risk attenuation

Introduction Over the last decades, arsenic (As) contamination has become an issue of concern due to its diverse damage on human health and biota. The toxicity of this metalloid is strongly associated with its chemical form. The inorganic forms (As+3 and As+5) are more toxic than the methylated forms (ATSDR-Agency for Toxic Substances and Disease Registry 2007). The As+3 is the most toxic As form (ATSDR-Agency for Toxic Substances and Disease Registry 2007), since it is usually carcinogenic, teratogenic, and genotoxic (Peakall and Burger 2003; Planer-Friedrich et al. 2007; Rabieh et al. 2008; Matschullat 2000). In Brazil, recent researches on As biogeochemistry and its potential consequences on human health were executed in the BTriangulo Mineiro^ region (Minas Gerais), BVale da Ribeira^ region (São Paulo), and Amazônia (Matschullat et al. 2000; Mello et al. 2006; Figueiredo et al. 2007; Deschamps and Matschullat 2008). Gold deposits are usually associated with sulfides (reduced minerals), which are exposed to weathering due to mining activities. In gold mining areas, sulfides are often oxidized, thus releasing As and other elements, forming sulfates,

Environ Sci Pollut Res

hydroxides, and other oxidized compounds. Then, arsenic can be leached and mobilized by water according to several factors, such as pH, Eh, ionic complex formation, and adsorption (Hutchinson and Meema 1987). Arsenic is found in more than 200 minerals, especially arsenates, sulfides, and sulfosalts, and the most common is the arsenopyrite (FeAsS). Arsenic freshwater concentrations vary by more than four magnitude orders, according to the As source, availability, and local geochemical environment (Prohaska and Stingeder 2005; Amaro et al. 2014). Freshwater As average from 0.4 to 80 μg L−1 (Penrose 1975) can rise to several hundreds of μg L−1 in streams near industrial and mining areas (adjacent to mill tailings or mine wastes), including reported concentrations above 556 μg L−1 (Adriano 2001). The most common As water valence states are arsenate (AsO4−3)–As(V), which is more prevalent in aerobic surface water, and arsenite (AsO3−3)–As(III), which is more likely to occur in anaerobic groundwater (Pettine et al. 1992). Both As(III) and As(V) form protonated oxyanions in aqueous solutions, and protonation degree depends on pH. Arsenic greatest range and highest concentrations—as with other metals and metalloids—are potentially found in groundwaters, as a result of water-rock interactions which are influenced by physical and geochemical conditions (Drever 1982). Such processes determine As mobilization and accumulation in the environment. Adsorption is a common mechanism for controlling As mobility and transport in surface and groundwater systems. Iron (Fe), aluminum (Al), and manganese (Mn) oxyhydroxides and clay minerals are commonly associated with particulate/solid materials and have been shown to be significant As adsorbents. Arsenic adsorption extent is influenced by aqueous phase chemistry, including pH, arsenic speciation, and competing ion presence and concentration (Stollenwerk 2003). Usual groundwater concentrations range from <0.5 to 10 μg L−1. Wells in contaminated areas, e.g., in Bangladesh, can reach 2.500 μg L−1 (Prohaska and Stingeder 2005) or more and more than 7.800 μg L−1 in bore waters in Argentina (Hernández et al. 2014). Paracatu Municipality (Minas Gerais State, Brazil) could be used as a good general model for understanding the impacts related to gold mining activities and arsenic contamination hot spots in southeastern Brazil. BMorro do Ouro^ is located at Paracatu Municipality and is the biggest gold mining site worldwide operating with low contents of gold, whose extraction is nowadays performed by Kinross Company. Huge ore amounts are required to be removed in order to obtain significant gold production (around 15 t year−1, with an average of 0.4 g of gold per ore ton) (Henderson 2006). Local government authorities are concerned about potential intoxication cases via As-contaminated water for human consumption. The present study aims to assess (i) drinking water quality nowadays consumed by Paracatu human population, (ii) contamination levels in surface and groundwaters and alternative

human consumption water sources, and (iii) hydrogeochemical processes controlling As concentrations in aquatic systems. These study hypotheses are (i) mining activities contribute to increase abnormally As concentrations in surrounding aquatic systems, (ii) Fe and Al oxyhydroxides play an important role in As sorption in the environment, and (iii) current water sources are not adequate for human consumption due to high As concentrations.

Material and methods Study area Paracatu Municipality is located at Minas Gerais State (southeastern Brazil) (Fig. 1). Paracatu is 250 km far from Brasília (Brazilian national capital) and is located in the tropical savannah dominium (Bcerrado^), 850–1800 mm/year (average 1300 mm); rainy is season between October and March; and temperature amplitude is between 15 and 35 °C. Paracatu is a predominantly agricultural town with approximately 85,000 inhabitants (2010), although more than 85 % of population lives in urban areas (Souza et al. 2011). Industrial open-pit gold mine activities began in 1976–1977 and are estimated to continue until 2042. Paracatu major mining-related features include an open-pit mine, two process plants, two tailing facilities, and related surface infrastructure (Henderson 2006). The mining area (1.100 ha) is about 2 km far from northern Paracatu; although it is classified as a rural zone, there are human communities bordering the site and, in some places, no more than 10 m separate mining from homes. Gold associated with arsenopyrite characterizes the Morro do Ouro ore deposit. Ore body As varies from around 1.100 mg kg−1 (oxidized zone) to 4.900 mg kg−1 (nonoxidized zone) (Möller et al. 2001). The Morro do Ouro is a topographic prominence in Paracatu, and as a consequence, mining activities may affect surrounding superficial water and groundwater quality through runoff, groundwater fluxes, and atmospheric deposition. Paracatu urban population majority (97 %) is served by treated water, while rural communities consume drinking water from local groundwater sources. Urban area drinking water is mainly obtained from freshwater sources, but it is complemented by urban groundwater sources, according to its availability. The mineralization occurs within a phyllite lithological sequence. Host rocks are not mineralized, have fine texture, and are easily weathered. Its mineralogy does not contain As, but some sulfide minerals may be eventually associated with As occurrences. Northern Morro do Ouro mineralization was interrupted by a fault that broke and raised the mineralized rock, which was completely eroded and leached. Southern phyllite thickness is higher than the northern ones, but they are mineralogically similar between each other. Average As

Environ Sci Pollut Res Fig. 1 Localization of the Rico, Escuro, and Entre-Ribeiros watersheds at Paracatu, Minas Gerais State, Brazil

concentration from southern Morro do Ouro phyllites was 17.22 ± 1.02 mg kg−1 (Almeida et al. 2009). However, arsenic contents can reach more than 500 mg kg−1 in phyllites that are directly in contact to the mineralized zone. They also present low resistance to weathering, and consequently, a thick mass of oxidized and mineralized rocks covered the ore body and its adjacencies. From 1770, these rocks consisted of alluvial gold deposits that were intensively extracted by artisanal gold miners. Between the 1960s and 1990s, some residues were reprocessed by using mercury amalgam (Santos 2012). Sampling period Water sampling was performed during dry season (September 2010 and 2011). Under these climatological conditions, water quality presents the highest interaction with soils, rocks, and pollutant loads, without dilution processes and runoff impacts. Total precipitation in the 2 years were significantly similar (Wilcoxon test n = 12, T = 20, z = 0.76, α = 0.05, p = 0.86), with the same standard distribution (n = 12, α = 0.05, Rs = 0.770, p = 0.003). Drinking water sampling Drinking water sampling was performed according to COPASA drinking water supply system (Paracatu water supply private agency). Thirty-seven samples from the system were sampled, before and after treatment, including the main source (a superficial source), complementary sources from nine artesian wells (located in the urban area), four water reservoirs, and tap water from 23 houses in Paracatu. These selected houses represented critical areas in groundwater urban water system distribution. Water samples (drinking water,

superficial water, and groundwater) were kept in cleaned, acidified (pH < 2 HNO3 suprapure Merck), wrapped in black plastic, and refrigerated until the analysis of polypropylene vials. Groundwater sampling The following 29 artesian wells (70–250 m depth) distributed through the municipality were sampled: municipal supply (9), private human consumption (6), rural communities consumption (4 but 1 from traditional Bquilombola^ community), and groundwater samples from monitoring wells (10) installed inside of the mining sites (2 in the gold mine and 8 in lead and zinc mines). Monitoring and supply wells are at the same lithological unit, phyllites. Monitoring wells sampling followed the low-flow method (USEPA-United States Environment Protect Agency 2010), using a Portable MicroPurge Pump system (QED trademark) operated by compressed gas and a bladder pump control unit (407 model), used with the MicroPurge MP10. Sampling was performed by adopting a low-speed water entry at the capture point and monitoring the flow rate during water pumping to the surface through MicroPurge 3020 electric probe use, in order to continuously measure this parameter in each well. Superficial water Twenty-seven freshwater (27) samples were collected along three watersheds, which are Paracatu River subbasins (Figs. 1 and 2): (i) Córrego Rico watershed—its spring is located inside the open-pit mine site, and its upstream sector is located at Paracatu Municipality. Córrego Espalha is a tributary flowing into Córrego Rico before the urban area, and its basin covers

Environ Sci Pollut Res

Physicochemical parameter monitoring (pH, DO, Eh, EC, T, and total dissolved solids (TDSs)) was performed in monitoring wells. In order to implement them, purged water passed through a continuous flow cell coupled with a MicroPurge MP20 multiparameter probe, in order to measure parameters. As, Fe, S, Mn, and Al analyses

Fig. 2 Schematic representation of the a Rico watershed, b EntreRibeiros watershed, and c Escuro watershed

an important ecological conservation unit, which is essential for Rico watershed water quality sustainability; (ii) Ribeirão Entre-Ribeiros watershed—it is located downstream of the tailing dam and receives its effluents, as well as those from the gold mine hillsides; and (iii) Rio Escuro watershed—it is outside the gold mining direct influence, and partially located into a conservation unit, thus being considered a reference area. Its tributary (Santa Izabel) is the main drinking water source for Paracatu urban population. Waters were shallow at the sampling period, generally with some tens of centimeters. Water samples were directly collected in the stream flow central part by using a Van Dorn bottle. Chemical and physicochemical analyses For superficial waters, physicochemical parameters (pH, electrical conductivity (EC), dissolved oxygen (DO), temperature (T), and oxyredox potential (Eh)) were measured at the field busing a MultiSonda (HANNA, Hi 9828). Suspended particulate matter (SPM, mg L−1) was quantified by gravimetric method. Samples were filtered (0.45-μm pre-weighed cellulose acetate filter). Filters were dried to constant weight and weighed again (APX-200 DENVER analytical balance, 10−4 g precision). For groundwater, the water level drawdown parallel was monitored in order to ensure low-flow adoption.

Superficial water total As concentration was determined by inductively coupled plasma–mass spectrometry (ICP-MS) (42 ICP-MS model, PerkinElmer MS). The detection limit (LOD) was of 0.5 μg L−1. Precision and accuracy (standard reference material (RCM)) were of 90 and 95 %, respectively. Method validation was performed by STD TMDA-70 certified RCM analysis. Before being analyzed, samples were prefiltered in 0.45-μm pore size filter. Groundwater total As concentration was quantified by ICP-MS (Agilent, 7500ce) with nebulizer (Micromist and Cross Flow camera)—EPA 200.8 method (inductively coupled plasma water and waste trace elements determination–mass spectrometry; LOD < 0.1 μg L−1). Precision and accuracy were measured by RCM analysis, being 90 and 95.5 %, respectively. Method validation was performed by NIST 1643e certified RCM analysis. Samples were prefiltered in 0.45-μm pore size filter. The same procedure used to quantify superficial water As was also applied for total Fe, Mn, Al, and S determination. LODs were 10 μg L−1 (Fe), 0.05 μg L−1 (Mn), 1 μg L−1 (Al), and 1 mg L−1 (S). Before being analyzed, all samples (groundwater and superficial waters) were pre-filtered in 0.45-μm pore size filter.

Results Drinking water arsenic Arsenic concentrations in all urban supply water samples (from Santa Izabel river catchment) were lower than LOD (<0.5 μg L−1), including all waters sampled at home taps. In urban areas, rural communities, and private consumption groundwater sources, arsenic concentrations were also below the LOD value (<0.1 μg L−1). The limit for total As concentration in freshwater and groundwater (for human consumption) established by Brazilian law is 10 μg L−1 (resolution no. 357; CONAMA-Brazilian National Council on the Environment 2005). All freshwater and groundwater samples showed As levels below such level. Groundwater arsenic and hydrogeochemical parameters Table 1 shows As concentrations and physicochemical parameters determined in groundwater samples from 10 monitoring

Environ Sci Pollut Res Table 1 Arsenic (As) concentrations, dissolved oxygen (DO), temperature (T), electric conductivity (EC), redox potential (Eh), and total dissolved solids (TDSs) determined in groundwater from monitoring wells

Sampling points

MW01 MW02 MW03 MW04 MW05 MW06 MW07 MW08 MW09 MW010 Mean

Monitoring wells DO (mg/L)

pH

T (°C)

EC (μS/cm)

Eh (mV)

TDS (mg/L)

As (μg/L)

0.56 2.41 3.46 3.70 2.94 2.03 4.62 1.94 2.26 2.01 2.59 ± 1.13

7.3 5.2 6.4 6.4 4.7 7.0 5.9 6.8 6.2 7.1 6.3 ± 0.8

25.4 24.4 29.0 31.3 32.8 26.9 29.9 32.8 29.6 29.9 29.2 ± 2.9

5 23 144 109 16 381 17 572 177 366 181 ± 195

94.0 233.0 −21.5 −40.0 −6.5 −114.7 −53.33 −76.7 −42.2 −161.2 18.9 ± 111.2

– – 72 55 8 191 9 286 89 183 111.63 ± 98.82

<0.1 <0.1 <0.1 <0.1 <0.1 0.29 <0.1 <0.1 0.96 0.35 –

wells. Arsenic concentrations range between <0.1 and <1 μg L−1. The average pH value (6.3 ± 0.8) suggests that water is moderately acidic in some sampling points. EC values (181 ± 195 μS cm−1) can be considered low for groundwater. EC is an indicator of TDS contents and is associated with the sum of cation and anion in solution (essentially Ca, Na, K, Mg, Cl, and some SO4−2). Because of that, TDS can control pH values (positive significant correlation n = 10, α = 0.05, Rs = 0.83). Thus, results indicated low potential for leaching processes and for liberation of chemical elements (host phyllite contains soluble carbonate minerals, 5 % ankerite associated with calcite and siderite) to the water. This observation is supported by the fact that Fe concentrations in 80 % of groundwater from supply wells were below LOD (10 μg L −1 ), and its maximum value was 64 μg L −1 . Furthermore, S concentrations were below LOD (1 mg L−1), and Mn and Al average concentrations were 3.91 ± 5.46 and 8.50 ± 6.91 μg L−1, respectively. All these values can be considered low. In addition to carbonates, the host phyllite contained quartz (40 %) and aluminosilicates (30 % sericite and 15 % chlorite associated with muscovite), refractory carbon material (5 %), and some other accessory minerals (<5 %), Fe and Mn oxyhydroxides and low sulfide contents (mainly pyrite and pyrrhotite) (Almeida et al. 2009). Therefore, mineral leaching product is apparently retained in the rocks. DO and Eh values were low (2.59 ± 1.13 mg L−1 and −18.9 ± 111.2 mV, respectively), suggesting the occurrence of reducing conditions. Since DO contents are directly related to water temperature (29.2 ± 2.9 °C), measured oxygen was strongly subsaturated (Tchobanouglous and Schroeder 1987). The pH and DO are inversely correlated (n = 10, Rs = −0.64, significant to α = 0.05). When plotting As-H2O system pH and Eh diagram values (Cherry et al. 1979), (i) the average pH (6.3 ± 0.8) and Eh (−18.9 ± 111.2 mV) values corresponded to As (III) upper

limit (H3AsO30) and As(V) stability (between H2AsO4− and HAsO42−) and (ii) the Eh-pH pair values showed similar plotting values. When plotting Fe-As-H2O (Langmuir et al. 2006) and FeAs-S-H2O system pH and Eh values (Carageorgos and Melamed 1995), (i) pH and Eh average values defined a position in pyrite and Fe+2 upper stability limit and (ii) individual well pair values showed that 60 % of samples are between pyrite and Fe+2 stability limits. Regarding the other samples, 10 % of them are between pyrite and goethite stability limits, other 10 % samples are between Fe+2 and goethite stability limits, and 20 % are in the goethite stability. Superficial water arsenic and hydrogeochemical parameters Total arsenic concentrations and hydrogeochemical parameters determined in fluvial waters are shown in Table 2. About 35 % of samples showed As concentrations below 10 μg L−1, which is the maximum concentration established by Brazilian law. At the Escuro watershed, As concentrations are close to the 0.5 μg L−1 detection limit. At the Córrego Rico watershed, total As levels range from <0.5 to 40.10 μg L −1 (average concentration = 12.58 ± 12.00 μg L−1; n = 12). As levels attained 23.60 μg L−1 at R1 sampling point (very close to Morro do Ouro mine, without other influence than that from mining). Arsenic concentrations varied from 12.00 to 23.00 μg L−1 at the river segment that crosses the city. The 40.1 mg/kg (R10) value was found around 15 km downstream from the urban area. In this sampling station, government environmental authority (IGAMMG) monitoring data (from 2008 to 2011) showed an average value of 21.9 ± 6.4 μg L−1. From this sampling point, As concentration decreased to around 7 μg L−1 some kilometers before its confluence with Paracatu River. Córrego Espalha upstream tributary of Ribeirão Rico, protected by an

Environ Sci Pollut Res Table 2 Brazil)

Physicochemical parameters and arsenic (As) concentrations in fluvial waters collected at the Paracatu Municipality (Minas Gerais State,

R1 R2 R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 ER1 ER2 ER3 ER4 ER5 ER6 ER7 ER8 ER9 ER10 ER11 E1 E2 E3 E4 Average mean

pH

DO (mg/L)

Eh (mV)

T (°C)

EC (μS/cm)

SPM (mg/L)

Fe (μg/L)

Mn (μg/L)

S (mg/L)

Al (μg/L)

As (μg/L)

6.5 6 5.9 6.7 6.8 6.9 7.8 7.6 7.1 7.8 7.6 7.6 8 7.7 8.1 8.1 7.8 7.8 7.6 4.7 7.8 7.8 7.7 7.3 6.8 7.3 7.4 7.3 ± 0.3

5 6 6.39 5.22 0.5 0.97 4.4 4.41 1.74 5.07 5.89 5.69 6.92 5.72 6.4 7.07 4.94 6.77 6.77 4.72 6.44 6.56 6.2 7.48 6.18 8.2 7.38 5.5 ± 1.87

25.1 33.4 31.9 30 −19.9 −46.6 145 50.2 −17.8 120.2 153 81.3 140.6 71.2 56.6 142.3 64 99.9 99.9 53.6 64.7 47.6 61.2 70.4 100.5 124 83 69.1 ± 51.3

35.5 23.2 21.2 29 27.5 27.3 25 26.7 27.8 27.1 24.2 24.2 25.1 22 23.3 25.4 24.6 24.7 26.4 23.9 25.8 25.8 23.9 26.1 25.3 27.8 27.6 25.8 ± 2.7

72 7 11 63 97 116 301 330 142 255 186 169 415 246 318 377 83 340 220 220 251 252 132 22 36 16 48 175.0 ± 124

22.3 5.0 10 67 11.2 33.5 33.5 11.2 44.6 22.3 13.4 78.1 12.5 40.0 40.0 44.6 80.0 11.2 22.3 33.5 44.6 22.3 22.3 22.3 6.7 21.2 66.9 31.2 ± 21.5

531 125 263 106 140 391 10 10 238 148 138 133 206 63 71 137 341 50 304 329 262 285 177 297 354 175 392 210.2 ± 130

104.1 23.4 12.2 1.1 0.4 1.1 7.3 0.6 12.6 26.3 5.4 6.1 69.1 28.9 22.3 49.3 18.3 0.3 0.6 23 0.9 1.1 15.6 19.7 55.5 25.5 4.2 19.8 ± 24.6

9.1 1 1 2.1 2.3 2.2 1 2.5 2.4 4.5 1 1 9.4 5.8 5.9 4.6 2.3 4.2 1 3.8 1.9 1.2 1 1 1.5 1 1 2.8 ± 2.4

698 26 23 3 9 15 21 4 24 408 36 88 1542 625 648 944 294 5 12 215 29 35 13 29 27 12 907 247.9 ± 394

23.6 0.5 0.5 5.6 12 23.2 0.5 11.9 19.1 40.1 6.8 7.1 5 29.1 22.1 18.9 1.5 19.1 5.2 13.5 0.5 0.6 2.7 0.5 0.7 0.5 0.8 10.1 ± 10.8

OD dissolved oxygen, Eh redox potential, T temperature, EC electric conductivity, SPM suspension total solids, As arsenic

environmental conservation unit, showed <0.5 μg L−1 As concentrations. At the Ribeirão Entre-Ribeiros watershed, As levels were from <0.5 to 29.10 μg L−1 (average concentration = 10.75 ± 10.16 μg L−1; n = 11). The sampling point close to the tailing dam output (ER1) showed 5 μg/L As concentration, suggesting As retention by dam sediments (Bidone et al. 2014). The highest values were found in São Domingos tributary (ER2 and ER3), running to NE on the mine boundary line, which is under influence of creeks coming down the mine hillside and old artisanal mining areas. The concentration decreases downstream. Rio Escuro watershed is outside the gold mine direct influence. As concentrations were from <0.5 to 0.80 μg L−1 (0.63 ± 0.15 μg L−1; n = 4). Santa Izabel River, which is a tributary of Escuro River, consists of the main drinking water source for Paracatu urban population, and its As concentration was <0.5 μg L−1 (sample E1). Arsenic, Fe, Mn, Al, and S concentrations are some orders of magnitude higher than those determined in groundwater.

The pH values are predominantly neutral to alkali (7.3 ± 0.3) and are apparently controlled by dissolved chemicals. The statistical correlation between pH and EC was positive and significant (n = 27, α = 0.05, r = 0.77). EC values were relatively low (175.0 ± 124 μS cm−1, similar to the groundwater EC 181 ± 195 μS cm−1, essentially Ca, Na, K, Mg, Cl, and SO4) even in Rico watershed, which drains Paracatu urban area and receives large amounts of domestic wastes without an adequate treatment. EC values of one order of magnitude higher than those last ones were observed in less degraded areas (R2 and R3 samples, located at Córrego Espalha conservation unit, in Rico watershed, and samples collected in Escuro watershed). Arsenic and S were positively and significantly correlated with EC, with r = 0.43 and r = 0.56, to n = 27 and 0.05, respectively. Arsenic-S correlation was positive and significant (r = 0.73). Fe and EC showed a negative significant correlation, r = −0.45. DO contents can be considered normal for superficial waters (5.5 ± 1.87 mg L−1), and such values suggested intermediary subsaturation (around 30 % or less) in relation to water

Environ Sci Pollut Res

temperature (25.8 ± 2.7 °C). The lowest DO concentrations were detected in Rico watershed. Eh values were generally low (69.1 ± 51.3 mV), especially in Rico watershed, suggesting reducing conditions for these waters. Eh showed positive and significant correlation with OD (n = 27, α = 0.05, r = 0.50) and pH (r = 0.55). Arsenic-OD correlation was negative and significant (n = 27, α = 0.05, r = −0.42). SPM data (average 31.2 ± 21.5 mg L−1) are in agreement with runoff absence in that season and have not shown correlation with any hydrogeochemical parameter. When plotting pH and Eh values for the As-H2O system (Cherry et al. 1979), (i) the average pH (7.3) and Eh (69.1 mV) values were in the limit between H2AsO4−—As(V)—and HAsO42−; (ii) the Eh-pH pair values from each sampling point in the three watersheds were close to the limit between As(III) and As(V), and 30 and 70 % of samples were in the H3AsO30 stability—As(III)—and HAsO42−—As(V), respectively; and (iii) most samples (96 %) that were in the As(III) stability were collected in Rico watershed. When plotting pH and Eh values for Fe-As-H2O system (Carageorgos and Melamed 1995), (i) pH (7.3) and Eh (69.1 mV) average values were inside goethite stability and close to pyrite and Fe+2 stability and (ii) Eh-pH pair values from each sampling point were in the limit of pyrite and Fe+2 (25 % of them were collected in Rico watershed), in the limit of pyrite and goethite (30 %), and in the limit of goethite alkaline stability (45 %).

Discussion Arsenic in drinking water Arsenic concentrations in drinking water were extremely low, more than one magnitude order below the legal criteria. Regarding the water supplied by the municipal authority (i.e., 97 % of the total population) for human ingestion, the results indicated that the toxicological risk associated with the human exposure to As is negligible. This fact that does not support our initial third working hypothesis and can be explained by some lithological and structural attributes. Additionally, a survey could be conducted focusing sparse rural populations that have their water supplied from shallow wells and untreated superficial water. Groundwater, arsenic, and hydrogeochemical parameters Some factors converge to explain groundwater As low concentrations and support its use for domestic supply and others. The first one consists of the fact that host rocks do not have As in their mineralogy. In this context, although As concentrations in non-mineralized phyllites are apparently low, the potential As leaching to groundwater must be carefully analyzed.

Fresh phyllite, i.e., not weathered phyllite (the weathering degree decreases with depth), have low porosity (10 % or less), interstitial micropores (5 to 30 μm) interconnected by joints and cracks, and density between 2.21 and 2.78 t m−3 (mean around 2.70 t m−3) (ABC-Associação Brasileira de Cerâmica 2014). Weathered phyllites have 30 % or more porosity and density around 1.90 t m−3 (Lopes et al. 2007). When considering the average content of 17.22 ± 1.02 mg As kg−1, fresh non-mineralized phyllites from this study area could stock 41.9 g As m−3 rock and weathered phyllites could stock 22.9 g As m−3 rock. Leaching and liberation of only 1 % of this As content to the non-saturated pore water of these two phyllite types could result in As concentrations of around 102 and 103 magnitude orders expressed in μg L−1, respectively. Melamed et al (1996), when investigating As adsorption in soils from Morro do Ouro region, showed maximum loading capacity of around 3.0 mg As/g soil for soil and 2.0 mg As/g for weathered phyllite, while Ladeira et al. (2002) found values between 0.7 and 3.6 mg As/g for soil. Both studies suggest that almost all As contents were associated with oxyhydroxide minerals, mainly Fe oxyhydroxides. These adsorption levels inhibit As transference to the pore water from regional geology non-mineralized phyllites. Oxyhydroxides formation may occur through oxygen transport from recharge in aquifers. Another important aspect is the mineralization of geological setting and structural controls. Gold mineralization host phyllites have been thrust from SW to NE, producing extensive deformation and faults, later filled by ascendant hydrothermal solutions (As, S, Au, Pb, Zn, and SiO2), through an inclined plane structure, 10° to 20° or more, 100 to >160 m thick, traceable over 4 to 6 km along a SW-NE trend, and with around 3 km width (Möller et al. 2001; Henderson 2006). Therefore, off the mineralized area, the ore body is covered by non-mineralized phyllite whose thickness increases with distance, following the dip mineralization. This increase may be of around 20 %, measured by drilling (Möller et al. 2001), i.e., 20 × 100 m terrain. As a result, supply wells, which usually have depths between 70 and 250 m, are collecting aquifer water in rocks with very low As levels. Despite the fact that host phyllites contain Fe, Al, and S of around 7, 2, and 1 %, respectively, such element low concentrations were found in groundwater. This fact suggests that these elements are not being leached from the rocks to the water and are in agreement with physicochemical conditions measured in the groundwater. EC values are low and suggested diluted water occurrence. DO (2.59 ± 1.13 mg L−1) and Eh (−18.9 ± 111.2 mV) concentrations are also low and suggest anoxic conditions. On the other hand, it is also important to note that Fe and Al oxyhydroxides play an important role in As adsorption and its mobility reduction, especially under high Fe:As ratio values, low temperature, and moderately acidic pH. Those conditions are in agreement with this

Environ Sci Pollut Res

study groundwater (Fe:As ratio > 10) and also support low As concentration occurrence. When plotting pH and Eh diagram values for the As-H2O system (Cherry et al. 1979), the results suggested that As(III) and As(V) co-existed in groundwater. Rezende et al. (2013), while determining Paracatu groundwater As(III) and As(V) concentrations by chemical speciation, also observed both forms existence. When considering Fe-As-H2O (Langmuir et al. 2006) and Fe-As-S-H2O (Carageorgos and Melamed 1995) systems, results suggest (i) anoxic condition predominance and sulfide relative stability, with some Fe+2 liberation to pore water, and (ii) arsenic and Fe oxide (goethite) co-existence, consisting of weathering products, due to water percolation in the aquifer. Superficial water, arsenic, and hydrogeochemical parameters The results suggest that mining activities play an important role in the spatial distribution of As concentrations at the Rico and Entre-Ribeiros watersheds, where As concentrations showed a decreasing trend to downstream from the mining site. This observation supports our first working hypothesis. R1 (in the upstream of Rico watershed) and ER2 samples (in São Domingos River, Entre-Ribeiros watershed tributary), both located in the most contaminated river segments, are examples of starting points of these decreasing concentration gradients. Besides Morro do Ouro mine facilities and disposal sites for tailings and overburden rock material, and rock and soil contribution as superficial water As sources, other extra sources may be old artisanal gold mining activity residues and small dams and swamps in rural and urban areas. R8, R9, R10, ER4, and ER8 samples are good examples of these aspects. The sampling was performed at the dry season and reflects a permanent water flux from the non-saturated zone, which can be of Bconstructed soil^ type, i.e., soil mixed with old residues and/or extremely weathered phyllites. This context apparently favors the geochemical mobility of elements. Arsenic, Fe, Mn, Al, and S concentrations in superficial waters were almost one to two orders of magnitude higher than those measured in groundwater. Lower EC in protected areas and noncontaminated superficial water (e.g., Espalha—R2 and R3—and Escuro—E1 to E4—watersheds) suggested that major cations and anions are not released from sensu stricto natural soils. Regional pedology is mainly composed of oxisols, i.e., acid materials characterized by high Fe and Al oxyhydroxide contents, high kaolinite concentrations, and low nutrient content (low fertility). Such soils, when not disturbed, can store As and reduce its transference into surrounding rivers.

It is important to highlight that As can be precipitated with Mn, Fe, and Al oxyhydroxides into bottom sediments, a fact that may reduce fluvial water As content. Thus, arsenic could be deposited in low-energy fluvial segments or could be transported by advective fluxes. Ferreira et al. (2013) also indicated that As presence in river sediments from the study area is present in a less available form, bind to iron oxides, and partially oxidized compounds from the original mineral. The Fe concentration average in sediments was higher than that of soils (6.6 ± 4.5 and 4.6 ± 4.0 %, respectively), confirming the high stability of this metal when combined to oxyhydroxide in fluvial conditions (Bidone et al. in preparation). This fact may be the main reason for high arsenic concentrations in sediments, which would be retained when associated with Fe, reducing their availability and risk; sediment and soil As mean resulted in 1207.36 ± 1408.00 and 280.30 ± 467.50 mg kg−1, respectively (Bidone et al. 2014). Such high capacity of As adsorption in soils and sediments in the three watersheds apparently explains the relatively low As concentration in superficial water, supporting our second working hypothesis. However, some concentrations were above the limit established by Brazilian law (10 μg L−1), 55 and 45 % of samples from Córrego Rico and Ribeirao Entre-Ribeiros watersheds, respectively. Values higher than 500 μg L−1 have been reported in rivers that are adjacent to mines and tailings and sterile deposits (Prohaska and Stingeder 2005). Superficial water mean pH was neutral to alkaline, and Eh mean indicates reducing conditions. This may be explained by the following three simultaneous processes: silicate hydrolysis from phyllite and residue weathering, generating alkaline solutions; carbonate dissolution from regional rocks and residues from old gold mining sites; and aminoacid deamination, with NH4+ and HCO3− solution generation in fluvial segments impacted by domestic wastes (e.g., Rico watershed). The pHEh, pH-EC, and Eh-DO positive and significant correlations seem to support these hypotheses. The pH-Eh correlation suggests that pH is affected by microbiological oxidation of organic matter from bottom sediments, which is also directly related to the reduction of oxides and hydroxides, according to the availability of dissolved oxygen (Sabadini-Santos et al. 2014). Thus, this correlation suggests anaerobic conditions associated with the degradation of amino acids and carbohydrates from domestic wastes (e.g., Rico watershed) and/or accumulation of organic matter from other sources in fluvial segments of low energy (e.g., Espalha and Escuro better conserved watersheds). Higher temperatures can also contribute to reduce water DO levels. During the dry season sampling performed in this research, superficial water level was low and the temperature varied between 21 and 36 °C along the day. Arsenic-S, As-EC, and S-EC positive and significant correlations, as well as Fe-EC negative and significant correlation, point out to sulfide oxidation in oxisols, releasing As and S to superficial water, and Fe retention in soils and sediment.

Environ Sci Pollut Res

The negative and significant correlations between As and DO may indicate As(III) in superficial waters. The pH and Eh values plotting in the As-H2O system (Cherry et al. 1979) also seem to support this hypothesis, especially in Rico watershed, where smaller DO values were found. Rezende et al. (2013) quantified total As, As(III), and As(V) concentrations in nine samples (six of them from Rico watershed) of superficial waters in Paracatu in 2010 and 2011, at the beginning of the rainy season. The results showed As(V) predominance in the analyzed samples, with around 60 % of total As. In the river segment under urban influence and with poor sewage treatment, As(III) tend to increase in the dry season, when anthropogenic organic loads reduce Eh. Part of oxyhydroxides in the sediments can be reduced by releasing more soluble As(III), which remain stable in solution under reducing conditions; however, if water becomes predominantly oxidant in rainy season again, As can oxidize and be adsorb by oxyhydroxides and/or be co-precipitated with them. Arsenate and arsenite may be easily converted under reducing or oxidizing conditions (Wenzel and Blum 1994). When considering pH and Eh measured values in the FeAs-H2O system (Carageorgos and Melamed 1995), results suggested that Fe oxyhydroxides (e.g., goethite and amorphous oxides that were not detected by X-ray diffraction) are present in the river sediments, retaining As highest part. It is possible that superficial water As has sewage contributions (e.g., Rico watershed). Sewage sludge has 2–26 mg kg−1 of As (Kabata-Pendias 2011). These value ranges are similar to non-mineralized phyllites, so incremental effects may be masked.

Conclusions Arsenic drinking water contents did not represent risk on human health. Thus, municipal water sources—fluvial water and groundwater—are adequate for human consumption. As total concentrations in superficial water were relatively low—considering a mining area with abundant arsenopyrite—and virtually absent in groundwater. Groundwater is influenced by rocks containing low As contents due to a structural control; off the mining area limits, the ore rapidly plunges and is covered by a huge deposit of non-mineralized phyllites. Superficial water As sources are related to mining facilities (As concentrations decrease downstream) and degraded areas mixing residues of old artisanal mining, weathered phyllite, and oxisols (in less disturbed areas, arsenic concentration is close to LOD). Untreated domestic sewage seems to exert a secondary influence. Physicochemical parameters suggest As(III) and As(V) coexistence in both superficial water and groundwater, with As(V) predominance in superficial waters. Results and previous studies show strong evidences that Fe and probably Al

oxyhydroxides retain the arsenic by sorption (adsorption and co-precipitation) in phyllites, oxisols, and sediments, reducing its transfer to waters. This mineralogy is quite stable at the study area Eh-pH conditions. Acknowledgments The authors are grateful to CETEM/MCTI (Centro de Tecnologia Mineral do Ministério de Ciência, Tecnologia e Inovação) for funding this study. They also thank REUNI—Programa de Apoio a Planos de Reestruturação e Expansão das Universidades Federais—for the grant of Maria Carla Santos (Doctorate/UFF) and CAPES for the grant of Marcos Ferreira (Doctorate/UFF).

References ABC–Associação Brasileira de Cerâmica (2014) www.abceram.org.br Adriano DC (2001) Trace elements in terrestrial environments: biogeochemistry, bioavailability, and risks of metals, 2nd edn. Springer, New York Almeida BS, Dardenne NA, Thomson B, Brod JA, Guimarães EM (2009) Geoquímica dos filitos carbonosos da Formação Paracatu, Depósito Morro do Ouro, Minas Gerais. In: Almeida BS (ed) Geoquímica dos filitos carbonosos da Formação Paracatu, Depósito Morro do Ouro, Paracatu, Minas Gerais. Dissertação de Mestrado, n. Instituto de Geociências, Universidade Federal de Brasília, DF, Brasil, p 249 Amaro AS, Venecia Herrera BC, Lictevout E (2014) Spatial distribution of arsenic in the region of Tarapacá, northern Chile. In: Litter MI, Nicolli HB, Meichtry JM, Quici N, Bundschuh J, Bhattacharya P, Naidu R (eds) V International Congress of Arsenic in the Environment—As2014. Buenos Aires, Argentina, CRC PressTaylor and Francis Group, Proceedings, pp 54–55 ATSDR-Agency for Toxic Substances and Disease Registry (2007) US. Agency for Toxic Substances and Disease Registry. Toxicological Profile for Arsenic, Atlanta, Georgia Bidone ED, Castilhos ZC, Bertolino LC, Santos MCB, Silva RSV, Cesar RG (2014) Arsenic in soils and sediments from Paracatu, Brazil. In: Litter MI, Nicolli HB, Meichtry JM, Quici N, Bundschuh J, Bhattacharya P, Naidu R (eds) V International Congress of Arsenic in the Environment—As2014. Buenos Aires, Argentina, CRC Press-Taylor and Francis Group, Proceedings, pp 113–115 Carageorgos T, Melamed R (1995) Estabilidade do arsênio proveniente do processamento de arsenopirita para recuperação do ouro: revisão de literatura e estudo de caso. Relatório Técnico RT 01/96, Centro de Tecnologia Mineral, CETEM/MCTI Cherry JA, Shaikh AV, Tallman DE, Nicholson RV (1979) Arsenic species as an indicator of redox conditions in groundwater. J Hydrol 43: 373–392 CONAMA-Brazilian National Council on the Environment (2005) resolution no. 357. http://www.mma.gov.br/port/conama/res/res05/ res35705.pdf Drever JI (1982) The geochemistry of natural waters. Prentice-Hall, Inc., Englewood Cliffs, N.J., 388p Deschamps E, Matschullat J (eds) (2008) Arsênio antropogênico e natural. Um estudo em regiões do Quadrilátero Ferrífero. Fundação Estadual do Meio Ambiente, Belo Horizonte, 330 p. ISBN 97885-61029-00-5 Ferreira MM, Patchineelam SR, Castilhos ZC, Calmano W (2013) Sequential extraction of arsenic in sediments of Paracatu, MG, Brazil. Goldschmidt2013 Conference Abstracts, p. 1080, DOI:10. 1180/minmag.2013.077.5.6 Figueiredo BR, Borba RP, Angélica RS (2007) Arsenic occurrence in Brazil and human exposure. Environ Geochem Health 29:109–118

Environ Sci Pollut Res Alvarez M, Del P, Trovatto MM, González N, Hernández MA (2014) From the presence of arsenic in natural waters to HACRE. In: Litter MI, Nicolli HB, Meichtry JM, Quici N, Bundschuh J, Bhattacharya P, Naidu R (eds) V International Congress of Arsenic in the Environment—As2014. CRC Press-Taylor and Francis Group, Proceedings, Buenos Aires, Argentina, pp 861–863 Henderson RD (2006) KINROSS—paracatu mine. Technical report., p 179 Hutchinson TC, Meema KM (eds) (1987) Lead, mercury, cadmium and arsenic in the environment. Wiley & Sons, New York Kabata-Pendias A (2011) Trace elements in soils and plants, 4th edn. CRC Press Taylor & Francis Group, Boca Raton, p 505 Ladeira ACQ, Ciminelli VST, Nepomuceno AL (2002) Seleção de solos para imobilização de arsênio. REM: Revista da Escola de Minas. Ouro Preto 55(3):215–221, jul. set Langmuir D, Mahoney J, Rowson J (2006) Solubility of amorphous ferric arsenate and crystalline scorodite (FeAsO4.2H2) and their application to arsenic behavior in buried mine tailings. Geochimica et Cosmochimica Act 70:2942–2956 Lopes MC, Lana MS, Ribeiro LFM, Caldeira Silva CH (2007) Geotechnical behavior of weak rocks in Córrego do Sítio mine, Santa Bárbara (MG). REM 40:23–30 Matschullat J (2000) Arsenic in the geosphere—a review. Sci Total Environ 249:297–312 Matschullat J, Borba RP, Deschamps E, Figueiredo BR, Gabrio T, Schwenk M (2000) Human and environmental contamination in the Iron Quadrangle, Brazil. Appl Geochem 15:181–190 Melamed R, Neumann R, Carageorgos T (1996) Estudo de caracterização mineralogica, estabilidade e retenção de arsênio em rejeitos industriais na Rio Paracatu Mineração. Relatório de Projeto RT-29/ 96, Centro de Tecnologia Mineral, CETEM/MCTI. Mello JWV, Roy WR, Talbott JL, Stucki JW (2006) Mineralogy and arsenic mobility in arsenic-rich Brazilian soils and sediments. J Soils Sediments 6:9–19 Möller JC, Batelochi M, Akiti Y, Maxwell S, Borges AL (2001) Geologia e caracterização dos recursos minerais de Morro do Ouro, Paracatu, Minas Gerais. In: Pinto CP, Martins-Neto MA (eds) Bacia do São Francisco: geologia e recursos naturais. SBG/MG, Belo Horizonte, pp 199–234 Peakall D, Burger J (2003) Methodologies for assessing exposure to metals: speciation, bioavailability of metals, and ecological host factors. Ecotoxicol Environ Safety 56:110–121 Penrose WR (1975) Arsenic in the marine and aquatic environments: analysis, occurrence and significance. Crit Rev Environ Control 4: 465–82 Pettine M, Camusso M, Martinotti W (1992) Dissolved and particulate transport of arsenic and chromium in the Po River. Italy Sci Total Environ 119:253–280

Planer-Friedrich B, London J, McCleskey RB, Nordstrom DK, Wallschläger D (2007) Thioarsenates in geothermal waters of Yellowstone National Park: determination, preservation, and geochemical importance. Environ Sci Technol 41(15):5245–5251 Prohaska T, Stingeder G (2005) Arsenic and arsenic species in environment and human nutrition. In: Cornelis R, Crews H, Caruso J, Heumann KG (eds) Handbook of elemental speciation II: species in the environment, food, medicine and occupational health. John Wiley & Sons, Ltda, pp 69–85 Rabieh S, Hirner V, Matschullat J (2008) Determination of arsenic species in human urine using high-performance liquid-chromatography (HPLC) coupled with inductively coupled plasma-mass spectrometry (ICP-MS). J Anal Atomic Spectrometry 23(3):544–549 Rezende PS, Costa LM, Windmoller CC (2013) Total and trivalent inorganic arsenic determination in water samples by selective hydride generation atomic absorption spectrometry. Br J Anal Chem 10: 429–435 Sabadini-Santos E, Bidone ED, Castilhos ZC, Crapez MAC (2014) Microbial activities response to as exposure in soil and sediments surrounding a gold mining area at Paracatu—Brazil. In: Litter MI, Nicolli HB, Meichtry JM, Quici N, Bundschuh J, Bhattacharya P, Naidu R (eds) V International Congress of Arsenic in the Environment—As2014. Buenos Aires, Argentina, CRC PressTaylor and Francis Group, Proceedings, pp 396–398 Santos MJ (2012) O ouro e a dialética territorial em Paracatu, MG: opulência e resistência. Dissertação de Mestrado-Programa de Planejamento e Gestão Ambiental. UCB, Brasília, DF, p 192 Souza KV, Alamino RCJ, Fernandes FRC (2011) Paracatu: o conflito entre o BRio Bom^ e a mineração. In: Fernandes FRC, Enríques MARS, Alamino RCJ (eds) Recursos minerais & sustentabilidade territorial. CETEM/MCTI, Rio de Janeiro, BR, pp 259–282 Stollenwerk KG (2003) Geochemical processes controlling transport of arsenic in groundwater: a review of adsorption. In: Alan H, Stollenwerk WKG (eds) Arsenic in ground water—geochemistry and occurrence ed. Kluwer Academic Publishers New York, Boston, Dordrecht, London, Moscow, pp 67–100 Tchobanouglous G, Schroeder ED (1987) Water quality. Addison-Wesley Publishing Company, USA, p 768 USEPA-United States Environment Protect Agency (2010) Low stress (low flow) purging and sampling procedure for the collection of groundwater samples from monitoring wells. USEPA Region 1, Quality Assurance Unit, Publication EQASOP-GW 001, January 19, 30 p Wenzel WW, Blum WEH (1994) Assessment of metal mobility in soilmethodological problems. In: Metal Speciation and Contamination Soil, Allen HE, Huang CP (Eds.), Lewis Publisher.