PRELIMINARY EVALUATION OF ACID MINE DRAINAGE IN MINAS GERAIS STATE, BRAZIL

A despeito da importância economica de atividades mineralogicas para o estado de Minas Gerais, perceber-se que elas causam significativo impacto ao ambiente. Drenagem acida e um dos aspectos ambientais considerados pela atividade em funcao de seus impactos. Ela tem origem a partir da oxidacao de sulfetos metalicos que acidificam as aguas, com possibilidade de mobilizar elementos toxicos e poluir os recursos hidricos. Este trabalho teve como objetivo realizar uma avaliacao preliminar do problema em Minas Gerais. O trabalho teve inicio a partir de um levantamento bibliografico junto a Fundacao Estadual de Meio Ambiente (FEAM) para identificar mineracoes cujo minerio esta associado a sulfetos. Amostras dos substratos foram retiradas para determinar o potencial de acidez (AP) e de neutralizacao (NP). AP foi avaliado por dois procedimentos: pelo teor total de sulfetos e pela oxidacao com peroxido seguida da titulacao acida. O NP foi avaliado pelo teor equivalente de carbonato de calcio. Lâminas petrograficas foram preparadas para a descricao mineralogica com enfase em sulfetos e carbonatos. O balanco acido-base (ABA) foi determinado pela diferenca entre AP e NP, enquanto o potencial de geracao de aguas acidas foi obtido pelo valor de ABA e o volume total de material de cada atividade mineraria. Os resultados permitiram identificar quatro substratos com potencial para gerar drenagem acida no estado de Minas Gerais. Juntas, estes materiais representam um potencial de geracao de cerca 3,14 a 10,37 bilhoes de metros cubicos de agua a pH 2,0 ou 31,4 a 103,7 bilhoes de metros cubicos de agua a pH 3,0. Estes valores significam custos entre US$ 7,8 e 25,9 milhoes para neutralizar a acidez com o uso de calcario. Estas estimativas sao provavelmente subavaliadas, uma vez que algumas mineracoes nao foram incluidas no levantamento e, para outras, as amostras superficiais podem nao representar a realidade. Uma avaliacao mais rigida do potencial de drenagem acida em nivel estadual requer estudos futuros. Estes estudos precisariam incluir um numero maior de amostras e mineradoras nao incluidas neste trabalho. Eles poderiam considerar, ainda, a cinetica de geracao de acidez por meio de procedimentos de intemperismo simulado.

further studies, including a larger number of samples.Such investigations should consider other mining operations beyond the scope of this study as well as the kinetics of the acid generation by simulated weathering procedures.Index terms: acid mine drainage, acid-base accounting; sulfide minerals.

INTRODUCTION
Mining is one the most important traditional and economical activities in Brazil, which accounts for considerable economic and social benefits.Nevertheless, mining is considered to cause a number of negative environmental impacts.Acid mine drainage (AMD) is one of the currently most discussed environmental problems.It is initiated during the mining process, where a variety of metallic sulfides is exposed to the atmosphere and spontaneously oxidizes in the presence of O 2 (Geidel & Caruccio, 2000).Sulfide materials may also be exposed by highway construction and other large scale excavations (Skousen, 2000).The products of such reactions are not only highly soluble, but also strongly acidic.Thus, high sulphate and iron contents make the water drainage acid.If potentially toxic elements are present, such as Al, Mn, Cu, Zn, Pb, Hg, Cd, etc., they are mobilized in the drainage water, increasing the risk of toxic metal incorporation in biological systems, promoting biomagnification by the trophic chain (Mello & Abrahão, 1998).Among the sulfide minerals, pyrite and marcasite (FeS 2 ) are dominant in the state's mining regions, but other metals may be complexed with these sulfides, forming chalcopyrite (CuFeS 2 ), covellite (CuS), galena (PbS), sphalerite (ZnS), and arsenopyrite (FeAsS).
In Brazil, AMD studies are recent, and the problem has been reported in coal mines in the states of Rio Grande do Sul and Santa Catarina (Soares, 1995).There are many mines in the state of Minas Gerais where sulfides are associated with manganese, gold, nickel, zinc, lead and uranium, with the potential to generate acid mine drainage (Santana Filho, 2000).
The pH values of acid drainage (AD) in mined areas is usually below 3.5, with total amounts of soluble metals of over 50 mg L -1 (Ziemkiewics et al., 1977).In fact, high metal concentrations are common in acid drainage from coal mines, where average values of Fe and Mn in solution can vary from 20 to 300 mg L -1 and sulfate from 20 to 2,000 mg L -1 (Richards et al., 1993).
An important aspect to control the occurrence of AD is an accurate prediction of the acid-producing potential of the material.Several methods of determining the acidity potential (AP) have been proposed, for example: Smittenberg et al. (1951); Neckers & Walker (1952); Murthy & Sharada (1960); Pons (1964); Petersen (1969);Aspiras et al. (1972); Caruccio (1975);and Sorenson et al. (1979).In Brazil, Soares et al. (1997) used the "aqua regia" (HNO 3 :HCl 1:3) digestion as a simplified method to evaluate the total sulfide content in coal mining samples from Rio Grande do Sul.Each of these methods measures either S or Fe associated with iron sulfides and therefore provides an indirect estimate of the potential acidity by the stoichiometric equation for pyrite oxidation.The constraints on such estimates are that sulfides other than pyrite may be present, which produce acidity in a different stoichiometry in relation to that of pyrite, as described by Singer and Stumm (1970).Such estimates would therefore require a detailed description and an accurate evaluation considering all sulfides present in the sample, which is a very laborious and demanding determination.To overcome this problem, the acidproducing potential can be measured directly.A rapid oxidation technique has been proposed with 30 % H 2 O 2 that oxidizes iron sulfides (Yoneda, 1961;Grube Jr. et al., 1971;Sobek et al., 1978), followed by a direct titration of the produced acid.
The occurrence of AMD is not only determined by the presence of sulfides since it is very common to find carbonates associated with sulfides, which may reduce or neutralize the acidity generated by sulfide oxidation.Acid-Base Accounting (ABA) is a widely used method to estimate the potential of acid generation (Sobek et al., 1978).It is calculated by the difference between the acidity (AP) and neutralization potential (NP).In the case of negative values, the neutralization potential of the substratum is higher than its potential of acidity and AMD is, therefore, not expected.
With regard to the fact that Brazilian environmental regulations are yet incipient concerning AMD aspects, the scope of the problem needs to be evaluated.It is therefore important to identify suitable simple analytical methods for such an assessment.The "aqua regia" digestion appears to be a simple enough option for AMD studies.
The present study aimed to compare direct measures of the acid-producing potential technique as described by O' Shay et al. (1990) with the total sulfide oxidation evaluated by the "aqua regia" digestion technique.Furthermore, the potential of some mining activities in the state of Minas Gerais, Brazil to generate acid drainage was outlined in a preliminary evaluation.

MATERIAL AND METHODS
In a thorough bibliographic survey at the Minas Gerais Environmental Agency (FEAM), sites where mining activities are associated with sulfides were identified.Seven large mining sites were listed whose activities have the potential to generate acid drainage, including uranium, nickel, gold, and zinc exploitations.The second step consisted in the sampling of the different substrates including ore, overburden, and tail (Table 1) according to the peculiarities of each mine.In all situations, at least three simple surface (0-0.2 m) samples were taken to compose the final sample.
After each sample had been dried and ground (0.149 mm), analyses were carried out to determine the pH and electrolytic conductivity (EC) in water (1:2.5 v/v), the acidity potential (AP) and neutralization potential (NP).AP was evaluated by two procedures: total sulfide content after HCl + HNO 3 mixture (3:1 v/v) digestion (Soares, 1995) and oxygen peroxide (30 %) oxidation followed by acidity titration (O' Shay et al., 1990).NP was evaluated as calcium carbonate equivalent by titration of the remaining HCl after reaction with a ground sample (Embrapa, 1979).Based on the chemical analysis, the acid-base accounting (ABA) was determined by the difference of AP and NP.The generation potential of acid drainage was obtained from ABA values and the total amounts of the respective materials of each mining activity.This potential was stated as acid water volume at pH 2.0 or 3.0.Petrographic thin sections were produced after impregnating representative rock samples from each site.The main minerals were described, focusing on the presence of sulfite/carbonate.Sulfites were identified based on the following criteria (Winchell, 1933;Kerr, 1977): • Pyrite (FeS 2 ) was identified by cubic to octahedral crystals, opaque, showing a brilliant metallic luster, creamy white color under reflected light and yellowish colors under dimmed reflected light.
• Pyrrhotite (FeS) was identified by hexagonal prism, opaque crystals, produced marked rotation of the polarization plane under R. Bras. Ci. Solo, 30:365-375, 2006 reflected light; Color yellow to copper, darker than pyrite; cream-colored under vertical reflected light; usually associated with chalcopyrite.
• Arsenopyrite (FeAsS) with octahedral to subhedral prismatic crystals, silvery white to grayish, white color under vertical reflected light, commonly intergrown with chalcopyrite and especially pyrite.
• Chalcopyrite (CuFeS) was identified by tetrahedral crystals, often striated, of brassyellow color, light yellow under vertical reflected light; there was no visible plane rotation under polarized reflected light.
In the overall discussion the mining companies were identified by letters rather than by names, hence preserving their legal rights and to avoid ethical issues.

pH, electrolytic conductivity, sulfur and carbonate contents
According to the pH values, the samples were classified in two groups.The first one had low to very low pH values (2.6 to 5.4) and included samples from sites A, B, and C (Table 2), with exception of site B overburdens which presented lower sulfide content and the presence of carbonates.The group formation was corroborated by the microscopic observation of thin sections under high magnification (Figure 1).The low pH suggests the presence of sulfides and, possibly, AMD.The second group, including samples from sites D, E, F, and G, showed high pH values (> 7.0) due to the substantial presence of carbonates.Among all sites, the highest (9.5) as well as lowest (5.5) pH values were observed in tails from a gold mining site.
The high pH values associated with the presence of carbonates are in agreement with the geology of each area.Samples from sites F and G showed high pH values due to the dominance of carbonates in the limestone and dolomites of the Bambuí Group.Samples from sites D and E showed high pH values ascribed to the presence of carbonates associated to the Banded Iron Formation (BIF) of the Rio das Velhas Supergroup.In such cases, not only calcium (calcite) and magnesium (dolomite), but also iron (siderite) are expected.It is worth mentioning that the presence of siderite in a sample can overrate its neutralization potential (Caruccio and Geidel, 1996).Higher pH values than expected in a saturated CaCO 3 solution in equilibrium with atmospheric CO 2 are probably associated with the presence of salts or hydroxides, arising from the mineral concentration process.
The EC results showed greater variability than pH.Values ranged from 0.04 dS m -1 in ore B1 from site C to 17.56 dS m -1 in tailing from site D. The highest EC values are expected in tailing samples due to the mineral concentration process.In general, EC values are higher in samples containing sulfides and are associated with the AMD process.Sulfide oxidation involves solubilization reactions that increase the ionic activity and, consequently, raise EC values.

Acid base accounting
The different ABA values resulting after subtracting the acidity (AP) from the neutralization potential (NP) (Table 3) only allowed an identification of the AMD potentiality in overburden samples from  case, AP estimates of the total S content, after "aqua regia" digestion, were calculated on the basis of pyrite oxidation stoichiometry, as described by Singer & Stumm (1970).On the other hand, estimates from H 2 O 2 oxidation by the method of O´Shay et al. ( 1990) are probably underestimated, due to incomplete sulfide oxidation, as verified by Corrêa et al. (2002).An evaluation of the true potential of generating acid water, the costs and environmental risks must consider the peculiarities of each situation.The actual values probably range somewhere between the estimates of both criteria.

AMD estimates
The ABA values, in mols per metric ton (Table 3) were converted to liters per metric ton of acid water at pH 2.0 or 3.0.The total acid water volumes were calculated from the ABA values versus the mass (in metric tons) of each material (Tables 4 and 5).The calculations were only carried out for samples with an acid drainage potential (i.e., with positive ABA values), taking the mean values of the same materials.Thus, the estimates of acid drainage reached 13.3 x 10 9 m 3 of acid water at pH 2.0, or     132.8 x 10 9 m 3 at pH 3.0 according to the calculations, and considering the total S content extracted by "aqua regia" (Table 4).On the other hand, the hydrogen peroxide criteria presented estimates of around 5.9 x 10 9 m 3 with pH 2.0 or 58.9 x 10 9 m 3 with pH 3.0 (Table 5).The real contribution to the acid drainage generation depends on the different mitigation measures adopted by local companies, as mentioned below.
In the case of site A, mining activities have ceased and acid water arises from the sampled overburdens.Thus, the estimate of acid drainage for site A is between 2.42 and 5.28 x 10 9 m 3 of acid water at pH 2.0 or 24.2 to 52.8 x 10 9 m 3 at pH 3.0 (Tables 4  and 5).It is known that the company mitigates the problem by damming and neutralizing the acid water.Besides, the overburden piles were covered with clayey soil to seal them.Site B has a potential for acid drainage generation of around 2.24 x 10 9 m 3 of acid water at pH 2.0 or 22.4 x 10 9 m 3 at pH 3.0, according to the lowest estimate criterion (Table 5).Nevertheless, ore processing at this mine includes a plant for sulphuric acid production, which decreases acid generation by 68.1 %, from 421,767 L t -1 in the ore to 58,700 L t -1 in tailing, according to the lowest estimate at pH 2.0.Calculations of the acid water generation should therefore take into consideration that 46 % of the ore is converted into tailing (Brandt, 1996).Thus, 5,207,066 t of ore produce around 2,395,250 t of tailing and the acid generation potential would be between 185 and 230 millions m 3 of water at pH 2.0 or 1.85 to 2.30 x 10 9 m 3 at pH 3.0 (Table 6).Another consideration is that approximately a third part of  the tails is destined for landfills of the underground mine, jointly with the overburdens.In this situation it is possible that, depending on the physical conditions and mixture quality, the overburden with a negative ABA will further neutralize the acidity tailing causes.Such considerations would suggest a more optimistic estimate of the acid drainage potential at site B (Table 6).
At site C, only ores were sampled, since there were no overburden piles.In this case, the AMD potential depends on the quantities and composition of the ores to be processed, as well as on the sulfide recovery rate during the process.The proportion of each ore type in the blend depends on the gold price, which is rather variable.Nevertheless, a gold price of US$ 325.00 per oz can be assumed.In this case, the blend composition would be 72 % of ore B2 (227,054,880 t) and 24 % of ore B1 (88,299,120 t) to be exploited and processed from 1998 to 2014.Hence, the potential for acid drainage generation would be as shown in Table 7.
Considering a minimum recovery rate of 58 % for sulfides at the "flotation" process, the acid drainage potential at the tailing dam would be only 42 % of the estimate in Table 7.The difference corresponds to the material to be settled in pools, where environmental control measures are strict.The AMD estimate for site C was 211 x 10 6 to 4.53 x 10 9 m 3 of acid water at pH 2.0 or 2.11 to 45.3 x 10 9 m 3 at pH 3.0 (Table 7).These values are lower than the ones established by Pinto and Nepomuceno (1998).The reason for this difference is probably the surface sampling done at the "mining front".Under such conditions, the B2 samples could be partially oxidized and might not represent the sulfide body as deeper probe sampling would.
At site D, an old tail dam (tail 5) is a potential acid drainage generator (Table 3), estimated similarly by both criteria at 376 to 399 x 10 9 m 3 of acid water at pH 2.0 (Table 4 and 5  The AMD generation potential for Minas Gerais State is presented in table 8, but it should be kept in mind the limitations and constraints discussed above for each mine.The costs for neutralization with lime, which is the cheapest and most abundantly available neutralizing product, were estimated at US$ 5.00 t -1 .Considering this cost, AMD represents a potential cost of US$ 7.8 to 25.9 millions to neutralize acidity.

CONCLUSIONS
1.The estimates of the AMD generation potential differ according to the adopted criteria.They are higher when calculated based on the total content of sulfides after "aqua régia" digestion than with the acidity potential (AP) measured after peroxide oxidation.The AP calculated from the total sulfide content is probably overestimated due the presence of other sulfides such as arsenopyrite, calcopyrite, sphalerite, and pentlandite, whose stoichiometry of acid generation differs from that of pyrite.On the other hand, the peroxide analysis probably underestimated AP due to incomplete sulfide oxidation.
2. The results identified four mining sites with potential to generate acid drainage in Minas Gerais.Jointly, these activities represent a potential to generate around 3.14 to 10.37 billions of m 3 of water at pH 2 or 31.4 to 103.7 billions of m 3 of water at pH 3. This, in turn, represents costs of US$ 7.8 and 25.9 millions for acidity neutralization with limestone.These quantities are probably underestimated because some mines were not included in the survey, and for other mines, surface samples may not represent the whole picture.A more reliable evaluation of the acid drainage potential at state level would require further studies.These studies must include a larger number of samples from mining operations not considered in this study.They should further consider the kinetics of the acid generation by simulated weathering methods.

Figure 4 .
Figure 4. Photomicrograph under reflected light with parallel polarizers showing anhedral crystals of pyrite (py) and ground mass of sphalerite (sl) at the center.

Figure
Figure 6.Photomicrograph under reflected light with parallel polarizers.In the picture euhedral arsenopyrite (apy) crystal wrapped in anhedral mass of pyrrhotite (po).

Table 1 . Description of different materials sampled at each mining site
(samples from sites B and C), sphalerite (site A) and pentlandite (site B), as well as arsenopyrite (sites A, C, and E) besides pyrrhotites.The presence of such a variety of sulfide minerals, identified by petrographic observation (Figures2 to 6), may have resulted in overestimated ABA values since the acid generation reactions from these sulfides probably have an stoichiometry other than pyrite.In this

Table 5 . Acid water generation potential with pH 2.0 and pH 3.0, as estimated by the H 2 O 2 digestion criteria for different materials sampled at each mining company Acid water volume (L t -1 ) Acid water volume (1.000 L) Company Sample pH 2 pH 3 Mass of material (t) pH 2 pH3
).It is known that the dam is lined with clayey soil and environmental control is probably increasing.Nevertheless, the AMD