Pollution of soils by the toxic spill of a pyrite mine (Aznalcollar, Spain)

M. Simón, , I. Ortiz, I. García, E. Fernández, J. Fernández, C. Dorronsoro and J. Aguilar
Departamento de Edafología y Química Agrícola, Facultad de Ciencias, Universidad de Granada, C/Fuentenueva s/n, Campus Universidad, 18071 Granada, Spain

1. Introduction
2. Methods
3. Results and discussion
3.1. Characteristics of the tailings and the contaminated soils
3.2. Elemental contents of the tailings, water and contaminated and uncontaminated soils
3.3. Pollution of soils by water and tailings from the toxic spill
3.4. Statistical study
3.5. Pollution of the soils 10 days after the spill and its evolution over time
4. Conclusion

1. Introduction

Pyrite mines have been worked for centuries in the province of Seville (southern Spain), especially in Aznalcollar mining district, on the eastern edge of the Iberian pyrite belt (Carvalho, 1976). The origin of this belt can be found in the volcanic sediments produced by submarine volcanic intrusions, with alternating acidic and basic episodes that emerged during the Hercinian period, from pre-existing pre-Cambrian and Palaeozoic sedimentary materials, which precipitated finally as metallic sulphides on the marine bed. The mineral phase, located mainly in the upper part of the layer formed during the acidic episodes, consists of different sulphides ( Almodovar et al., 1998): pyrite (83.1%), sphalerite (5.4%), galena (2.1%), chalcopyrite (1.4%), arsenopyrite (0.9%) and non-productive materials (7.1%).

In these mines, the processing of the ore consists of grinding, treating of the particles with SO2(g) and slaked lime and finally separation by differential floating of Cu, Pb and Zn at different pH values. The residues from this process are composed mainly of pyrite with minor proportions of other sulphides. The most abundant elements present in this residue are: Pb (0.8–1.1%), Zn (0.5–0.8%) As (0.2–0.5%) and Cu (0.1–0.2%). These residues are stored in very large walled ponds (approx. 1.4 km2). On 25 April 1998 the walls of two contiguous ponds broke open, and approximately 36Å~105 m3 of polluted water (solution phase) and 9Å~105 m3 of toxic tailings (solid phase) spilled into the Agrio and Guadiamar River basin (Fig. 1). The tailings spread some 40 km in a down-river direction, stopping at Puente Don Simón. Meanwhile, the polluted water reached the Guadalquivir River, affecting the National Park of Doñana (proclaimed by UNESCO in 1994 as part of World Heritage).

Nevertheless, the low dry-season flow of the Agrio and Guadiamar Rivers made the rapid construction of a retention dam possible at Lucio del Cangrejo, some 50 km downstream of the ponds, minimizing the damage of the toxic wastes in the wildlife reserve. The total surface area of the zone affected by the toxic spill was approximately 55 km2: 12.3 km2 of the eucalyptus plantation, 11.9 km2 cereals and sunflower, 9.9 km2 pasture, 5.4 km2 rice, 4.9 km2 marshland, 3.0 km2 fruit and olive trees, 2.9 km2 riverbeds, 2.2 km2 cotton, 0.8 km2 riparian vegetation, 0.8 km2 gravel bed, 0.5 km2 meadow and 0.4 km2 field crops.

Prior to the spill, mining activity at Aznalcollar had already acidified and increased the heavy-metal content of the waters of the Agrio and Guadiamar Rivers (Arambarri et al., 1984). The use of this water for irrigation has raised the heavy-metal content in the soils of the zone, and the contents in Cu and Zn have registered values higher ( Ramos et al., 1994) than those found in uncontaminated areas of the world ( Allaway; McGrath and Holmgren).

In the present study, the heavy metals and other associated elements in the polluted water, toxic tailings and contaminated soil were evaluated in comparison with uncontaminated soil in the area of the spill. The aim was to identify and quantify the pollutants involved, to relate the pollution to soil characteristics, and ultimately assess the extent of the damage caused to the soils of this area.

2. Methods

Seven sectors were studied along the basins of the Agrio and Guadiamar Rivers (Fig. 1): near the mine (M), Soberbina (SO), Puente de las Doblas (D), Aznalcazar (A), Quema (Q), Pescante (P) and Los Pobres (LP). In each sector, two square plots were laid out (25 mÅ~25 m), one on the contaminated and the other on uncontaminated soil. At each corner and in the centre of the plots, samples were taken of tailings (in contaminated plots) as well as of soil at 0–10 cm and at 10–30 cm in depth (in contaminated and uncontaminated plots). All samples, categorized according to origin (tailings and the two soil depths), were air dried and screened (2 mm screen size). Next, 250 g of each sample category from the five sampling points per plot, were mixed and homogenized for laboratory analysis. In addition, isolated pools of water from the toxic spill were also sampled and analysed. All samples were taken on 4 and 5 May 1998.

The soils studied, contaminated and uncontaminated, were classified into two categories according to the Soil Survey Staff (1997): Typic Xerofluvent (M, SO, D and Q) and Typic Xerorthent ( A, P and LP). Field descriptions of soils were based on procedures of the Soil Survey Staff (1951). To provide a quantitative assessment of the soil structure, we formulated a structural-development index (SDI) using the equation: SDI=SizeÅ~Grade, where values of the grade are given in Table 1, and the size of the structure take the following values: fine=10, medium=7, coarse=5, very coarse=3. Particle-size distribution was determined by the pipette method after elimination of organic matter with H2O2 and dispersion with sodium hexametaphosphate (Loveland and Whalley, 1991). The pH was measured potentiometrically in a 1:2.5 soil–water suspension. CaCO3 equivalent was determined by method of Bascomb (1961). Total carbon was analysed by dry combustion with a LECO instrument. Organic carbon was determined by the difference between total carbon and inorganic carbon from CaCO3. Iron oxides (Fed) were extracted with citrate–dithionite (Holmgren, 1967) and measured by atomic absorption spectroscopy. In the field, tailing samples were taken in hermetically sealed bags and their moisture content was determined by weight difference after drying the samples at 110°C. A saturated extract of the tailings was prepared ( US Salinity Laboratory Staff, 1954). Samples of the tailings and soils, after being very finely ground (<0.05 mm), were digested in strong acids (HNO3+HF+HCI). In each digested sample, water sample and saturated extract of the tailings, 24 elements were measured by ICP-MS with a PE SCIEX ELAN-5000A spectrometer. For the statistical analysis, the StatView 4.02 program was used.


3. Results and discussion

3.1. Characteristics of the tailings and the contaminated soils

In the tailings the particle-size class (Table 1) was silty loam, the silt content surpassed 70%, and the structure was platy. In addition, carbonate was absent, pH consistently acid, and the organic content was extremely low.

The soils in general were relatively homogeneous with respect to certain properties. All soils were neutral or slightly alkaline (pH between 7.2 and 8.1) and with relatively narrow ranges of organic-carbon content (0.5–1.7%) and Fed (0.8–1.5%). By contrast, there was a broader range of CaCO3 equivalent (0–20%), gravel (0–43%) and particle size of the fine-earth fraction (<2 mm): from clay loam (P and LP) to sandy loam (SO).

3.2. Elemental contents of the tailings, water and contaminated and uncontaminated soils

Of the 24 elements analysed (Table 2), the content in the tailings of Mn, V, Cr, Ni, Be, Y, Th and Sc was similar to or lower than that of the uncontaminated soils. These elements have been excluded as contaminants in this toxic spillage. Ba and Co in the tailings increased two to fourfold with respect to uncontaminated soils, while Sn and Hg increased eight to 12-fold; nevertheless, the content of the contaminated soils proved very similar to that of the uncontaminated soils, and therefore could not be clearly related to soil pollution. Mo and In in the tailings increased 25 to 35-fold with respect to the uncontaminated soils; Zn, Cu and Cd 50–90-fold; and Tl, As, Bi, Pb, and Sb more than 90-fold (Sb 400-fold). Despite the magnitude of their relative increases, Mo and In reached only minor median values in the contaminated soils (0.4 and 0.1 mg kg-1, respectively); similarly, Se, absent altogether from uncontaminated soils, was present in only one of the seven sectors studied (P), with a low content (3.4 mg kg-1). In short, these results indicate that the principal pollutants, in descending order of median concentrations in the contaminated soils, were: Zn, Pb, Cu, As, Sb, Cd, Bi and Tl. The standard deviation of these pollutants (Table 3) increased considerably in the contaminated soils with respect to uncontaminated soils, often surpassing the mean value. This greater dispersion of the datapoints indicates that the contamination had highly irregular effects on the soils of the different sectors.

The elemental contents of the polluted water (Table 4) indicated their potential toxicity. Although the original quantitative chemical composition may have changed by the time of sampling (10 days after the spill), the median concentration in Mn, Ni, Co, Cd and Cu clearly exceeded the maximum allowed in water to be used for irrigation (Crook and Bastian, 1992). By contrast, other elements such as Cr, Mo, Sn and Bi were not detected. The standard deviation of the different elements was relatively low, indicating a certain homogeneity in the chemical composition of this polluted water.


3.3. Pollution of soils by water and tailings from the toxic spill

Because the water from the toxic spill contained no Bi, the total Bi contamination of the soils must have come from the tailings. Thus, the quantity of tailings that penetrated the soil in each sector (Z) can be calculated by the equation:

where TBi is the Bi concentration in the tailings and CSBi and UCSBi are the Bi concentration in the contaminated and uncontaminated soils, respectively, all expressed in mg kg-1.

Eq. (1) presupposes that in each sector the Bi concentration of the uncontaminated soil was the same as the contaminated soil before the spill, an assumption that may introduce an error. In any case, the results ( Table 5) indicate that the penetration of the tailings into the soil was highly irregular, varying considerably from one sector to another according to the characteristics of each soil, especially structure. In general, the penetration of the tailings diminished with soil depth, as structure size increased and grade decreased. The only exception was the sector Puente de las Doblas (D), where the soil had a substantial gravel content which increased in depth (Table 1), allowing greater vertical percolation, as well as lateral infiltration of the spill from the bank of Guadiamar River where the layer richest in gravel reaches the surface. Of all sectors, the greatest penetration of the tailings occurred in the upper 10 cm of Pescante (P) as the result of cotton cultivation, in which ploughing of the original clay loam formed an artificial unaccommodated, strong (grade 3), fine angular blocky structure (0.5–2.0 cm diameter), opening a great number of interpedal voids that allowed the tailings easy entry. Below 10 cm in depth, changed to accommodated, weak (grade 1), very coarse angular blocky structure (10–15 cm diameter), and thus the number interpedal voids diminished, thereby reducing penetration of the tailings. In the other sectors, not having been recently ploughed, less penetration resulted from coarser structure with weaker grade. The least penetration occurred in the sectors Soberbina (SO) and Aznalcazar (A); in the former the soil was structureless sandy loam and in the latter nearly structureless (angular blocks larger than 20 cm diameter) silty clay, with practically no gravel at either site.

From the quantity of tailings that penetrated the soil (Z, in g kg-1) and from the concentration of each element (i) in the tailings (Ti, in mg kg-1), estimates can be made of the quantity of each element (i) (Si) which entered the soil from the solid phase of the spill Eq. (2).

In addition, because the total pollution of the soils was caused both by water and the tailings from the toxic spill, the quantity of each element (i) that entered as part of the solution phase (Wi) can be estimated by the equation:

where (CSi-UCSi, in mg kg-1) is the total contamination of each soil for each element (i).

The results (Table 6) indicate that the range of total contamination for each element was very broad. The percentage of this total belonging to the solid phase of the spill clearly varies from one element to another. In the case of Cu, Zn and Cd, only 20% of the median total contamination of the first 10 cm of the soil penetrated as part of the solid phase of the spill, and thus the remaining 80% must have penetrated as part of the solution phase of the spill. In any case, the range of these percentages was extremely broad, as a consequence of the markedly different quantities of tailings that penetrated each soil (Table 5). As and Sb entered the soil primarily through the solid phase (median of 95%) and to a far lesser degree through the solution phase (5%). Meanwhile, Pb and Tl registered intermediate values, penetration through the solid phase reaching 75 and 85%, respectively. In addition, because the quantity of tailings that penetrated each soil generally decreased with depth, the total contamination, as well as the percentage of each element that entered the soil through the solid phase, tended to diminish between 10 and 30 cm in depth.


3.4. Statistical study

Principal component analysis was performed with Si and Wi values, structural-development index, and the analytical characteristics of the soils (Table 1). Three factors explain more than 80% of the variance ( Table 7).


Factor 1 represents soil pollution from the toxic spill, and thus includes the concentration of all of the elements that penetrated the soil in the solid phase, as well as the concentration of Sb, Tl and Pb that penetrated in the solution phase. The inclusion in this factor of SDI with high load shows structure to be the soil characteristic which best determined the penetration of the tailings, this penetration being greater the smaller the structure size and higher the grade of development. The negative sign of the pH indicates that the pollution tended to acidify the soil, although this trend is not strongly evident (low load) apparently due to the buffering effect of the CaCO3 in most of the soils.

Factors 2 and 3 are interpreted as heavy-metal sorption in the soil. Factor 2, which includes the contents in WZn, WCd, CaCO3, organic carbon and finest-particle soil (silt+clay), all with relatively high loads, shows that the sorption of Zn and Cd was caused by the formation of insoluble precipitates such as carbonates, as well as organic complexation and cation exchange (Alloway, 1995). The increases of the pH tended to favour these sorption processes, although the effect was not pronounced (low load) probably because of the above-noted narrow range of the pH values of the soils. Factor 3, including WAs and Fed with high loads and WCu with low load, reveals that the specific adsorption of As and to a lesser extent Cu was due primarily to the iron oxides (Alloway, 1995).

A correlation matrix was made with Wi and pH values as well as the CaCO3, clay, silt, Fed and organic-carbon contents of the soils. The absolute values of those correlation coefficients, equal or greater than 0.7 (Table 8), show that organic-carbon and CaCO3 content determined the accumulation in the soil of Zn dissolved in the solution phase, and the CaCO3 content the accumulation of Cd. The range of the different parameters should be taken into account when considering these correlations. For example, the sorption of the heavy metals by organic matter (Riffaldi and Levi-Minzi, 1975) and clays ( Farrah and Pickering, 1977) increases at higher pH values; nevertheless, this effect was not appreciable within the narrow pH interval (7.2–8.1) of the soils studied.


3.5. Pollution of the soils 10 days after the spill and its evolution over time

The concentrations in the soil of the elements considered to be pollutants (Table 9) indicates that, in only one of the seven sectors studied (P), all elements exceed the maximum permitted concentrations, for agricultural soils, set by Canada (Sheppard et al., 1992), Belgium ( Stringer, 1990) and Holland (NMHPPE, 1991, in Alloway, 1995), as well as the ecotoxicological level set by Van Den Berg et al. (1993). In D and LP, only Zn and As exceeded the aforementioned maximum values allowed, and in A only Zn. In the other sectors, no element exceeded these maximum values. Nevertheless, in the not-too-distant future, these latter elements could exceed these maximums, given that, as the tailings dry and aerate, a complex process oxidizes the sulphides to sulphates ( Nordstrom, 1982), lowers of the pH ( Stumm and Morgan, 1981) and solubilizes part of the formerly insoluble pollutants ( Rogowski and Caruccio).


This oxidation process and its rate could be appreciated by 4 May 1998. On this date, the tailings differed in moisture content as a consequence of their different thicknesses, the thinner areas drying more rapidly than the thicker ones. The concentrations of water soluble SO42-, Cd2+ and Pb2+ in the tailings on this date, measured in saturated extracts, increased logarithmically with declining moisture (Fig. 2). Given that between 25 April (date of the spill) and 4 May (date of sampling) no rain fell, these solubilized elements, remained in the solution phase of the tailings and, with evaporation, rose by capillary action to the surface, forming a white salty crust. It would be expected that subsequent rains would dissolve the soluble salts, which would then infiltrate the soil, raising the pollution level. This process, and its rapidity (in scarcely 10 days, the driest tailings multiplied their content in soluble Pb and Cd by 10-fold with respect to the wetless tailings), underscores the urgency of removing the tailings from the soil surfaces in these types of spills.


4. Conclusion

The present report identifies the pollutants, their concentrations in the soil at the moment of the study, the primary mode by which each pollutant entered the soil, and the rapid oxidation of the tailings that makes the soils susceptible to a future increase in pollution. This indicates the necessity of monitoring the concentrations using frequent sampling. A monitoring programme, with sampling conducted every 20 days, is currently under way, and will be the subject of future reports.


Allaway, W.H., 1968. Agronomic control over the environmental cycling of trace elements. Adv Agron 20, pp. 235–274.
Alloway BJ, editor. Heavy metals in soils. London: Blackie Academic & Professional, 1995:368..
Almodovar, G.R., Saez, R., Pons, J.M. and Maestre, A., 1998. Geology and genesis of the Aznalcóllar massive sulphide deposits, Iberian Pyrite Belt, Spain. Miner Deposita 33, pp. 111–136. Abstract-GEOBASE  
Arambarri P, Cabrera F, Toca C. Estudio de la contaminación del río Guadiamar y su zona de influencia (Marismas del Guadalquivir y Coto de Doñana) por residuos de industrias mineras y agrícolas. Consejo Superior de Investigaciones Cientificas, editor. Madrid, España, 1984:118..
Bascomb CL. A calcimeter for routine use on soil samples. Chem Ind 1961:1826–1827..
Caruccio, F.T. and Geidel, G., 1978. Geochemical factors affecting coal mine drainage quality. In: Schaller, F.W. and Sutton, P. Editors, 1978. Reclamation of drastically disturbed lands ASA-CSSA-SSSA, Madison, Wisconsin, pp. 128–148.
Carvalho, D., 1976. Consideraçoes sobre o vulcanismo da regiao de Cercal-Odemira. Suas relaçoes com a Faixa Piritosa. Comun Serv Geol Port 60, pp. 215–238.
Crook, J. and Bastian, R.K., 1992. . In: Guidelines for water reuse US EPA, Washington DC, p. 71.
Farrah, H. and Pickering, W.F., 1977. The sorption of lead and cadmium by clay minerals. Aust J Chem 30, pp. 1417–1422.
Holmgren, G.G.S., 1967. A rapid citrate–dithionite extractable iron procedure. Soil Sci Soc Am Proc 31, pp. 210–211.
Holmgren, G.G.S., Meyer, M.W., Chaney, R.L. and Daniels, R.B., 1993. Cadmium, lead, zinc, copper, and nickel in agricultural soils of the United States of America. J Environ Qual 22, pp. 335–348. Abstract-EMBASE  
Loveland, P.J. and Whalley, W.R., 1991. Particle size analysis. In: Smith, K.A. and Mullis, Ch.E. Editors, 1991. Soil analysis: physical methods Marcel Dekker, New York, pp. 271–328.
McGrath, S.P. and Loveland, P.J., 1992. . In: The soil geochemical atlas of England and Wales Blackie Academic and Professional, Glasgow, p. 162.
Nordstrom, D.K., 1982. Aqueous pyrite oxidation and consequent formation of secondary iron minerals. In: Kitrick, J.A., Fanning, D.S. and Hossner, L.R. Editors, 1982. Acid sulfate weathering Soil Sc Soc Am, Madison, WI, pp. 37–56.
Ramos, L., Hernández, M. and González, M.J., 1994. Sequential fractionation of copper, lead, cadmium and zinc in soils from or near Doñana National Park. J Environ Qual 23, pp. 50–57. Abstract-Elsevier BIOBASE | Abstract-EMBASE | Abstract-Compendex  
Riffaldi, R. and Levi-Minzi, R., 1975. Adsorption and desorption of Cd on humic acid fraction of soils. Water Air Soil Pollut 5, pp. 179–184. Abstract-EMBASE | Abstract-INSPEC| Abstract-Compendex  
Rogowski, A.S., Pionke, H.B. and Broyan, J.G., 1977. Modeling the impact of strip mining and reclamation processes on quality and quantity of water in mined areas: a review. J Environ Qual 6, pp. 237–244. Abstract-EMBASE | Abstract-Compendex  
Sheppard, S.C., Gaudet, C., Sheppar, P.I., Cureton, P.M. and Wong, M.P., 1992. The development of assessment and remediation guidelines for contaminated soils: a review of the science. Can J Soil Sci 72, pp. 259–394.
Soil Survey Staff. Soil survey manual, Handbook 18. US Department of Agriculture, Washington DC, 1951:227..
Soil Survey Staff, , 1997. . In: Keys to soil taxonomy. 7th ed Pacohontas Press, Blacksburg, Virginia, p. 545.
Stumm, W.Y. and Morgan, J.J., 1981. . In: Aquatic chemistry: an introduction emphasizing chemical equilibria in natural waters John Wiley & Sons, New York, p. 218.
Stringer DA. Hazard assessment of chemical contaminants in soil. ECETOC Technical. Rep. No. 40. Aneme. Louise 250. Brussels, Belgium, 1990:121..
US Salinity Laboratory Staff. Diagnosis and improvement of saline and alkaline soils, Handbook 60. US Department of Agriculture, Washington DC, 1954:160..
Van Den Berg, R., Dennenman, C.A. and Roels, J.M., 1993. Risk assessment of contaminated soil: proposal for adjusted, toxicologically based Dutch soil clean-up criteria. In: Arendt, F., Annokkée, G.J., Bosman, R. and van der Brink, W.J. Editors, 1993. Contaminated soils '93 Kluwer Academic Publisher, London, pp. 349–364.