Journal of Contaminant Hydrology 111 (2010) 6582

Contents lists available at ScienceDirect

Journal of Contaminant Hydrology

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j c o n h yd

Geochemical and environmental controls on the genesis of soluble

eforescent salts in Coastal Mine Tailings Deposits: A discussion based on
reactive transport modeling
S.A. Bea a,, C. Ayora a, J. Carrera a, M.W. Saaltink b, B. Dold c
a
b
c

Institute of Environmental Assessment and Water Research (IDAEA), CSIC, c/Lluis Sol Sabars, s/n, 08028 Barcelona, Spain
Technical University of Catalonia (UPC) c/Jordi Girona 1-3, 08034, Barcelona, Spain
Institute of Applied Economic Geology University of Concepcin, Victor Lamas 1290, 4070386 Concepcin, Chile

a r t i c l e

i n f o

Article history:
Received 19 April 2009
Received in revised form 30 November 2009
Accepted 11 December 2009
Available online 24 December 2009
Keywords:
Reactive transport modeling
NaClCuCl2 system
Pitzer
Tailings
Multiphase ow
Unsaturated zone
Water-soluble eforescences

a b s t r a c t
Water-soluble eforescent salts often form on tailings in hyperarid climates. Their high
solubility together with the high risk of human exposure to heavy metals such as Cu, Ni, Zn, etc.,
makes this occurrence a serious environmental problem.
Understanding their formation (genesis) is therefore key to designing prevention and
remediation strategies.
A signicant amount of these eforescences has been described on the coastal area of Chaaral
(Chile). There, highly soluble salts such as halite (NaCl) and eriochalcite (CuCl22H2O) form on
4 km2 of marine shore tailings. Natural occurrence of eriochalcite is rare: its formation requires
extreme environmental and geochemical conditions such as high evaporation rate and low
relative air humidity, and continuous Cl and Cu supply from groundwater, etc. Its formation
was examined by means of reactive transport modeling.
A scenario is proposed involving sea water and subsequently a mixture of sea water/freshwater
in the groundwater composition in the formation of these eforescences. The strong
competition from other halides (i.e. halite and silvite (KCl)) for the Cl may inhibit the
precipitation of eriochalcite. Therefore, the Cl/Na ratio trend N 1 is a key parameter in its
formation. Cation-exchange between Na+ and other major ions such as K+, Ca2+, Mg2+ and
Cu2+ in the clay fraction of tailings is proposed to account for realistic Cl/Na ratios.
With regard to preventing the formation of eriochalcite, a capillary barrier on the tailings
surface is proposed as a suitable alternative. Its efciency as a barrier is also tested by means of
reactive transport models.
2009 Elsevier B.V. All rights reserved.

1. Introduction
Water-soluble eforescent salts often form on tailings in
arid climates. Their formation on the surface increases the
risk of human exposure to heavy metals such as Cu, Ni, and Zn
through wind transport. Signicant amounts of eforescence
salts (mainly halite (NaCl), eriochalcite (CuCl22H2O) and
gypsum (CaSO42H2O)) form on 4 km2 of a porphyry copper
Corresponding author. University of British Columbia, 6339 Stores Road,
Vancouver, Canada V6T 1Z4, BC. Tel.: +1 604 827 5607; fax: +1 604 822 9014.
E-mail address: sbea@eos.ubc.ca (S.A. Bea).
0169-7722/$ see front matter 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.jconhyd.2009.12.005

tailings deposit in the coastal area of Chaaral (Chile, Fig. 1A).

In 1983, the Chaaral contamination case was classied by
the United Nations Environmental Programme (UNEP) as one
of the most serious cases of contamination in the Pacic area.
Despite the fact that Cl and Cu are widespread elements in
the Earth, eriochalcite (also termed antofagastite, eriocalcite or
erythrocalcite) is a highly soluble salt, commonly obtained in a
laboratory, but rarely occurring under natural conditions. In fact,
eriochalcite has never before been reported as eforescence in
tailings (Dold, 2006). Its natural occurrence is mainly associated with fumaroles of volcanoes, e.g. Mt. Vesuvius (Italy)
or associated with copper mineralization, e.g. Mina Quetena,

66

S.A. Bea et al. / Journal of Contaminant Hydrology 111 (2010) 6582

Fig. 1. (A) Location map of Chaaral's tailings. (B) A sketch of geochemical zones in Chaaral tailings. Three geochemical zones are observed: (1) an evaporation
zone mainly formed by eforescences, (2) an oxidation zone characterized by secondary ferric mineralogy (i.e. goethite and K-jarosite), (3) a primary zone
preserving the initial mineralogy of the tailings.

west of Calama, Chile. In the latter case, eriochalcite is

accompanied by other copper halides such as blandylite
(CuB2O4CuCl24H2O) and atacamite (CuCl23Cu(OH)2) in a
leached zone above massive iron suldes(Palache and
Foshag, 1938). Previous works dealing with the environmental problem in Chaaral include those devoted to the
impact on the local and regional marine fauna (Castilla,
1983; Farina, 2001; Lee and Correa, 2005; Lee et al., 2001),
and the study of mobilization of contaminant elements such
as Cu, Zn and Ni, and oxyanions such as As and Mo (Dold,
2006). Environmental and geochemical conditions are particular for formation of the eriochalcite. A range of water
activity between 0.61 and 0.68 was measured in a solution
saturated in halite and eriochalcite (Filippov et al., 1986). As
a consequence, their natural occurrence is restricted to
extremely arid climates such as the Atacama region in Chile.
Moreover, a continuous supply of Cl and Cu is necessary.

Thus, this limits their formation, for instance, to coastal areas

(Cl supply) and to cupric tailings (Cu supply). A priori, these
particular environmental and geochemical conditions are
important for the precipitation of eriochalcite. However, its
rare natural occurrence suggests complex mechanisms in its
formation.
Reactive transport modeling in a vadose zone is relatively
uncommon. This may be attributed to the lack of codes
necessary to deal with multiphase ow associated with
geochemical reactions, and also to the difculty encountered
in obtaining reliable sets of eld or experimental data to
compare with model results. Therefore, the majority of earlier
works to develop reactive transport models for the vadose
zone of mine tailings have assumed time-invariable boundary
conditions and thermohydraulic properties and have approached it as a single reactor (Bain et al., 2000; Gerke et al.,
1998; Gerke et al., 2001; Mayer et al., 2002; Wunderly et al.,

S.A. Bea et al. / Journal of Contaminant Hydrology 111 (2010) 6582

1996). In those studies which have taken uctuating boundary conditions or thermohydraulic properties into account,
a simplied reactive transport problem (usually with pyrite
as the only mineral phase present in the system) has been
considered (Lefebvre et al., 2001a; Lefebvre et al., 2001b; Xu
et al., 2000). In some cases, the developed models have been
used as a tool for predicting the long-term evolution of sulde
wastes. The objective of this paper is to propose the probable
mechanisms in the origin of eriochalcite, so as to design the
best remediation measures. In nding the hypothesis that
best explains the observed eforescences, we will shed light
on the climatic and geochemical conditions that favor precipitation of eriochalcite. To accomplish this, different scenarios
based on the predominance of a particular groundwater and
solute source were simulated. The results of the calculations
were compared with chemical analysis of interstitial water
and sequential extractions of the solid phase from boreholes
in Chaaral (Dold, 2006). The analysis of the multiphase ow,
dissolution/precipitation of the eforescent crust, and sulde
oxidation was carried out by means of reactive transport
modeling. The atmospheric boundary conditions set in the
model account for the hyperarid features of the climate.
2. Site description
2.1. Ore geology
The Chaaral deposits were otation tailings from the El
SalvadorPotrerillos mining district (Fig. 1A). The Potrerillos
porphyry copper deposit was exploited until 1959. Production in El Salvador started that year. Therefore, it can be
assumed that the lower portion of the Chaaral tailings comes
from Potrerillos and that they are overlain by tailings from
El Salvador (Dold, 2006).
The Potrerillos deposit (1.8 106 t Cu) is located in the
Atacama Desert, 120 km east of Chaaral. It is characterized
by a granodiorite intrusion affected by three hypogene alteration and mineralization events and an important supergene
alteration. The early potassic and propylitic alterations are
responsible for the K-feldspar, biotite, chlorite, quartz, ankerite,
and anhydrite content. Associated suldes are chalcopyrite,
bornite, pyrite, molybdenite, minor enargite, sphalerite, and
galena (Camus, 2003).
The El Salvador deposit (5.7 106 t Cu) is located about
100 km east of Chaaral. Primary mineralization is characterized by granodiorite with quartz veins and a matrix of
K-feldspar, hornblende, biotite, minor oxides, anhydrite and
suldes (Gustafson and Hunt, 1975; Gustafson and Quiroga,
1995).
2.2. Mineralogy of tailings
Tailings grain size varies between silty/clayey to sandy
toward the sea. A sketch of geochemical zones in the Chaaral
tailings is shown in Fig. 1B. The tailings surface presents
abundant greenish-blue and white eforescent salts. XRD
analysis revealed that these were mainly halite (NaCl) and
eriochalcite (CuCl2 2H2O) (Dold, 2006).
The unsaturated zone is characterized by oxidation. Its
depth varies between 0.73 and 1.88 m, with pH ranging from
2.5 to 4, underlain by a primary zone of neutral pH. The

67

oxidation zone is characterized by secondary ferric mineralogy with a typical yellowish-brown color (dominated by
K-jarosite) with ochre to orange streaks dotted with goethite.
Gypsum is also present. Clay mineralogy is made up of illite,
kaolinite and a vermiculite-type mixed-layer, the latter resulting from biotite alteration (Dold, 2006; Dold and Fontbote,
2001). Mineralogy from the primary zone consists of quartz,
plagioclase, muscovite, biotite, kaolinite, magnetite and rutile.
In addition, pyrite with minor chalcopyrite and chalcocite are
also observed.
Sequential extractions are widely used for exploration
purposes and for the study of element speciation in tailings. A
seven-step sequence was applied to samples collected at
Chaaral (Dold, 2003). Sequential extractions for Cu, Na, Ca
and Mg, in samples corresponding to the rst meter of
borehole CH1 are shown in Table 1. They showed a high level
of Cu enrichment in the water-soluble fraction in the
evaporation zone at the top of the tailings. This is consistent
with the abundance of eriochalcite described in the previous
section. Thus, eriochalcite content was estimated to be
around 30 mol m 3. However, a large fraction of Cu was
associated with clay minerals (probably adsorbed on illite or
as an interlayer ion-exchange cation in the vermiculite-type
mixed-layer), or adsorbed to the surface and/or co-precipitated with Fe(III)-oxides and hydroxides (see Cu(ex) and Cu
(ox) in Table 1, respectively). It is important to note that only
a very small amount of Cu remains in the sulde fraction,
indicating that most of the original chalcopyrite and chalcocite has oxidized and dissolved in the last thirty-two years
(see Cu(sph) in Table 1).
Na is also present in the water-soluble fraction (mainly as
halite) at the top (around 700 mol m 3, see Na(sol) in
Table 1). However, a signicant portion of the Na is found in
the residual phase (around 440 mol m 3, see Na(r) in
Table 1). The sequential extractions indicate a halite/
eriochalcite ratio in eforescences of about 20.
2.3. Hydrological setting
The west coast of South America, from northern Peru to
Central Chile, is extremely arid (Fuenzalida, 1950). Aridity
results from a combination of subsidence generated by a
permanent high-pressure area over the Pacic Ocean and the

Table 1
Sequential extractions in mol for Cu, Na, Ca and Mg, in samples
corresponding to the rst meter of borehole CH1 (Dold, 2003). sol = soluble
fraction (Step I), ex = exchangeable fraction (Step II), ox = Fe(III) oxides
and hydroxides fraction (Steps III and IV), sg = organic matter and Cusupergene fraction (Step V), sph = primary suldes fraction (Step VI), r =
residual fraction (Step VII) and t = total. A halite/eriochalcite ratio to 20
could be estimated according to Na(sol)/Cu(sol).
Step

Cu

Na

Ca

Mg

I (sol)
II (ex)
III and IV (ox)
V (sg)
VI (sph)
VII (r)
Total (t)

33.77
10.18
9
0.4
0.68
1.01
53.63

684
16.39
20.57
6.82
20.59
442.15
1190.6

152.56
20.48
7.85
4.4
10.62
63.9
259.8

59.0
6.45
33.16
6.45
47.95
120.73
273.64

68

S.A. Bea et al. / Journal of Contaminant Hydrology 111 (2010) 6582

atmospheric stability induced by the cold northward-owing

Humboldt Current.
The climate in the coastal area of the Atacama region is
classied as arid with abundant clouds (Fuenzalida, 1950;
Fuenzalida, 1965). On an annual average 28% of the days are
cloudy, 60% are partially cloudy and only 12% display clear
skies. A cloudy sky means less solar radiation on the tailings
surface and reduced evaporation.
Average precipitation of 30 mm per year was recorded for
Chaaral (Larrain et al., 2002). Therefore, the hydrological
system of coastal tailings deposits at Chaaral is mainly
controlled by tides, an inow of brine originating from Salar
de Pedernales, and leakage from freshwater supply which
feeds the village of Chaaral (Dold, 2006; Wisskirchen et al.,
2006). The temperature in this coastal area is characterized
by low variation between summer and winter and between
day and night. (Thompson et al., 2003) provided a temporal
record of temperature and relative humidity in Pan de Azcar
Park, located 20 km north of Chaaral. There, ve meteorological stations were established at different elevations
(between 210 m and 800 m) and distances from the Pacic
Ocean. Temperature and relative humidity were measured
between June, 1999 and October, 2000 in a station located
close to the Pacic Ocean. The variation in temperatures for
the entire period does not exceed 10 C (approximately 21 C
and 13 C for summer and winter, respectively), and the
Vapor Pressure Decit (VPD) was highest during the summer
(at 0.8 to 1.0 kPa), though not particularly high for desert
ecosystems. Peak temperatures are observed in the months of

January and February. The predominantly WSW winds form

sand dunes, up to 1.5 m high, which migrate over the tailings
surface toward the village of Chaaral. The portion of the
tailings surface not covered by dunes displays abrasion
marks, concordant with the predominant wind direction.
2.4. Hydrogeology
The distribution of major ions (see Na, K, Ca, Mg, Cl and
SO4 in Fig. 2A) suggests that three different groundwater
types were mixed in these tailings deposits (Fig. 2B)
(1) Brine ows through the deepest portion of tailings. It
may have originated at Salar de Pedernales, 150 km
east of Chaaral (Dold, 2006). Therefore, it ows below
the surface along its riverbed in the El Salado valley
and discharges into the ocean.
(2) A lens of freshwater overlies the brine. These freshwaters may have originated from leaks in Chaaral's
freshwater supply (Wisskirchen et al., 2006).
(3) Sea water, less dense than the brine described above.
In Chaaral, the ocean level varies approximately 1 m
during the tidal cycle.
2.5. Aqueous chemistry
Three piezometer nests (CH1, CH2 and CH3) were
installed to a maximum depth of 9 m, along a prole in the
southern part of the tailings deposits, directly in the west of

Fig. 2. Aqueous chemistry distribution and hydrogeology. (A) Vertical proles of pH and major ions (K, Ca, Mg, Cl and SO4) corresponding to borehole CH1 (its
location is shown in Fig. 1A). (B) Hydrogeology: three different water types are identied in the hydrogeologic system: (1) a brine in the deepest part of the
tailings, (2) a lens of freshwater, and (3) sea water.

S.A. Bea et al. / Journal of Contaminant Hydrology 111 (2010) 6582

the Chaaral village (see Fig. 1A). The 20 pore-water samples

were analyzed for major chemical components (Na, K, Mg, Ca,
Cl, NO2, NO3, and SO4), and trace metals (Fe, Al, Cd, Cu, Mn,
Zn, Pb, V, Cr, Co, Ni, As, Se, Sr, Mo, Ba, Ti, and U) by (Dold,
2006).
Vertical proles for pH and major chemical components
corresponding to borehole CH1 are shown in Fig. 2A. Porewater pH ranged between 2.6 and 4 in the oxidation zone, and
increased sharply to neutral values below the oxidation front.
Furthermore, redox potential (Eh) reached 600 mV in the
oxidation zone and dropped to 200 mV in the reducing
environment of the saturated zone. Major ions such as Ca, Mg,
Na, Cl and SO4 increase their concentrations in the oxidation
zone of these tailings, whereas K decreases its concentration
in the same zone.
3. Conceptual and numerical model
A conceptual model for the thermohydraulic and geochemical processes involved in the formation of eriochalcite
is shown in Fig. 3.
3.1. Thermohydraulic model
Several atmospheric processes are considered in the
model: rain, evaporation, radiation and heat exchange
between tailings/atmosphere. Water ux (jw, kgr s 1)
through the tailings surface results from
w

jw = P + E + jg + jsr

69

where P is rainfall, E evaporation, jw

g ux of water vapor and
jsr of the surface runoff. The evaporation rate in Eq. (1) is
given by an aerodynamic diffusion relation according to
a

E = f va ; za ; z0 e e

where ea and et are vapor pressures in the atmosphere and

pore tailings, respectively, f is a constant of proportionality,
which takes into account the wind speed (va, m s 1), the
roughness length (z0, m), and the screen height at which
va and ea were measured (za, m). Vapor pressure in the
atmosphere (ea) is computed from its relative humidity
(Hr) and temperature (Ta) through the psychometric law
(Edlefson and Anderson, 1943). However, the vapor pressure in pore tailings (et) depends on its temperature (Tt),
salinity (through water activity), and the capillary properties of tailings.
On the other hand, the energy ux (Je, J m 2 s 1) consists
of sensible heat (Hs), latent heat (Hc) and radiation (Rn)
uxes. As with evaporation, Hs is a function of air temperature, wind speed and the specic heat of the gas. Hc is equal to
the evaporation rate times the latent energy of vapor. Rn is
calculated according to direct solar short-wave radiation,
long-wave atmospheric radiation, the albedo of the tailings
surface, etc. The calculation of the direct solar short-wave
radiation takes into account seasonal variations, local latitude
and the cloud index (this index represents completely clear to
completely cloudy conditions). The daily variation in Rn is not
considered in the model for the purposes of simplicity.

Fig. 3. Flux, transport, geochemical and energy balance conceptual model for Chaaral tailings.

70

S.A. Bea et al. / Journal of Contaminant Hydrology 111 (2010) 6582

The low relative humidity during the summer causes evaporation, making water ow upwards. Solute concentrations
in the upowing water are dramatically increased, with saturation resulting in some salts forming an eforescent crust.
As evaporation progresses, pore-water content is reduced,
preventing liquid ux from rising to the surface. Therefore, the
evaporation front is displaced slightly downwards, and eforescences continue to grow within the pore volume. This
causes the tailings top to become increasingly cemented by
salts. The associated reduction in permeability favors an increase in downward vapor diffusion. Vapor condenses at depth
and concentrations decrease.
The model is conceptualized as a 1D clayey tailings according to grain size described for borehole CH1 by (Dold,
2006). Physical parameters and constitutive laws used in the
simulations are shown in Table 2. The 1D column is open
at the top and bottom. Atmospheric boundary conditions are
applied at the top, whereas prescribed gas and liquid pressures (approximately 0.1 MPa) are applied at the bottom in
order to simulate the groundwater level.
As for the atmospheric parameters, maximum and minimum
mean values for temperature (Ta) and relative humidity (Hr)
are calculated over a period of thirty-two years (19752007)
from data reported by (Thompson et al., 2003). The relative
humidity varies between 0.63 and 0.74 and the temperature

Table 2
Parameters and the most important constitutive laws used in the simulations. Sl is the pore saturation (m3 m 3), Pl, Pg and e are the liquid, gas and
vapor pressures, respectively (MPa), W is the molecular weight of water
(kg w mol 1), R is the universal gas constant, T the temperature in Kelvin, aw is
the water activity in the pore-water, and l is the liquid density (kg m 3).
Parameter or constitutive law

Tailings

Porosity,
Intrinsic permeability (m2)
Retention curve a
P0 (MPa)
n
Liquid relative permeability,

Constant, C
Residual saturation, Srl
Maximum saturation, Sls
Thermal conductivity,

0.4
10 14



 1 n
P P
S1 = 1 + gP0 l 1n
0.002
0.19
krl = krl CSl
1
1
0.01
0.99
1S
1
= Ssat
+ sat l
1
liq

sat = sol

sol b (W m 1 K 1)
gas (W m 1 K 1)
liq (W m 1 K 1)
Solid density, sol (kg m 3)
Solid specic heat (J kg 1 K 1)
Saturated vapor pressure, es (MPa)
A (MPa)
B (K)
Psychrometric law c
Evaporation by aerodynamic diffusion (E )
Wind speed, va, (m s 1)
Stability factor,
Screen height, za (m)
Roughness length, z0 (m)
Karman's constant, k
a
b
c

dry = sol gas

2
0.026
0.6
1700
789
 
eS = aw Aexp B
T
136,075
5239.7


Pl Pg
e = es exp RT
W
1
2
E = hk vai2 tv av
lnzza
4
0
1
2
0.01
0.4

Van Genuchten model (Van Genuchten, 1980).

Computed from mean values of mineral phases in tailings.
Edlefson and Anderson (1943).

between 21 C and 13 C from summer to winter, respectively. Thus, sinusoidal functions between these values are
simulated. On the other hand, the mean value for temperature (17 C) is set at the bottom of the model.
As laid out above, the wind speed and a clear/cloudy sky
affect the evaporation rate on the tailings surface. Thus, a
constant wind speed of 4 m s 1 and a partially cloudy sky are
considered in the model.
3.2. Geochemical model
In parallel to the reduction of liquid saturation, oxygen
(O2(g)) diffuses downwards, activating oxidation reactions.
Oxidation of pyrite (FeS2) and chalcopyrite (CuFeS2) produces sulfate, metals and acidity. The initial pathway consists
of oxidation of the disulde to sulfate by O2(aq). The second
pathway for sulde oxidation is by reaction with Fe(III). Both
pathways are fast. They yield a low pH and Fe(II), which may
become oxidized by O2(aq) to Fe(III). This oxidation is known to
be very slow at low pH (Singer and Stumm, 1970) except in the
presence of microorganisms, which can increase the rate of Fe
(III) production by up to six orders of magnitude.
The decrease in pH produced by pyrite and chalcopyrite
oxidation increases the dissolution rate of accompanying
silicates, which become the main source of alkalinity in these
environments.
The evolution of tailings impoundments is also controlled
by the precipitation of secondary phases into the pores
of waste materials and/or over their surface. Thus, the SO4,
Fe(III) and K generated from dissolution are consumed
by gypsum (CaSO42H2O), K-jarosite (KFe(SO4)2(OH)6) and
goethite (FeO(OH)).
Furthermore, cation-exchange could occur on the clay
fraction of these tailings (e.g. on the vermiculite-type mixedlayer and illite). Major cations such as Ca2+, Mg2+, Cu2+, Na+
and K+ could be exchanged onto the clay fraction, thus modifying the major cation relationships.
Evaporation trends for water samples collected in Chaaral
are simulated using CHEPROO code (Bea et al., 2009). The ionic
strength of water samples ranged from around 0.5 to 15 M, thus
requiring the Pitzer ion interaction approach (Bea et al., in
press). The evaporation trend for a sample located in the oxidation zone, 0.3 m below the tailings surface (sample CH1-1), is
shown in Fig. 4A. Here, evaporation progress is dened as the
ratio between the initial and the remaining mass of water
after the evaporation step. Halite precipitates when the evaporation progress is approximately equal to 6. This sample
presents a Cl/Na ratio higher than 1. Therefore, precipitation
causes Na concentrations to decrease and Cl to increase but
with a lower slope (geochemical divide). Copper speciation
changes during evaporation. Cu2+ is the predominant
species at the onset of evaporation whereas CuCl+ is predominant after halite precipitation (Fig. 4A1). Eriochalcite
precipitates after evaporation progress equal to 150, when
the ionic strength of the solution is 8 M and water activity
0.7. At this point, the total Cu in solution decreases and Cl
concentration continues to increase.
The previous evaporation trend implies that the key
parameter for eriochalcite to form is the Cl/Na ratio of the
starting solution. For a Cl/Na ratio less than 1, Cl decreases
when halite precipitates but Na does not decrease, and the

S.A. Bea et al. / Journal of Contaminant Hydrology 111 (2010) 6582

71

Fig. 4. Evaporation trends of samples CH1-1 (Cl/Na ratio higher than 1), and CH14 (Cl/Na ratio less than 1). (A1 and B1) Evolution of the concentration for Cl, Na
and cupric species (i.e. Cu2+, CuCl+ and CuCl2). (A2 and B2) Evolution of water activity and ionic strength. (A3 and B3) Evolution of the saturation indices for halite
(NaCl) and eriochalcite (CuCl22H2O). Evaporation is dened as the ratio between initial and the remaining mass of water after the evaporation step.

solution never reaches eriochalcite saturation. This is the case

of the sample CH14, for instance (Fig. 4B, located 0.3 m
below the water-table). The Cl/Na ratio of the pore-water
samples from the three boreholes is plotted in Fig. 5. Besides
sea water, only three samples display a Cl/Na ratio higher
than 1. Two of them are located in the upper part of the
oxidation zone, below the evaporation front. The third is
located in the brine, in the deepest part of the tailings. The
higher Cl/Na ratio could be due to: (1) the exchange of Na+ by
Ca2+, Mg2+, Cu2+ and K+ onto the clay fraction, (2) the
inuence of the marine aerosol on the tailings surface, and

(3) the precipitation of Na secondary phases (e.g. mirabilite,

Na2SO410H2O). In this work, the above-mentioned processes are evaluated in the proposed alternative scenarios.
In this study, the numerical nite element model CODEBRIGHT (Olivella et al., 1994) and the object-oriented module
CHEPROO (Bea et al., 2009) are applied to simulate ow and
reactive transport through tailings. The rst solves the
multiphase ow processes in the unsaturated zone while
the second solves the geochemical processes. Both codes are
coupled by direct substitution of the chemical equations into
the transport equations (Saaltink et al., 1998).

72

S.A. Bea et al. / Journal of Contaminant Hydrology 111 (2010) 6582

Fig. 5. Cl/Na ratio for water samples collected in boreholes CH1, CH2 and
CH3. Besides sea water, only three samples display a Cl/Na ratio higher than
1. Two of them are located in the upper part of the oxidation zone, below the
evaporation front. The third is located in the brine, in the deepest part of the
tailings.

Table 3
Reactions considered in the model. Equilibrium constants are taken of EQ3
data base (Wolery and Daveler, 1992) at 25 C.
Reaction
Homogeneous
+
2
HSO
4 H + SO4
MgOH+ + H+ H2O + Mg2+
OH + H+ H2O
Fe2 + 0.25O2(aq) + H+Fe3+ + 0.5H2O
CuCl+ Cu2+ + Cl
CuCl2(aq) Cu2+ + 2Cl
Cation-exchange
CuX2 + 2Na+ 2NaX + Cu2+
MgX2 + 2Na+ 2NaX + Mg2+
CaX2 + 2Na+ 2NaX + Ca2+
KX + Na+ NaX + K+
Dissolution/precipitation
+2
2
Pyrite + 3.5 O2(aq) + H2O 2 SO
+ 2H+
4 + Fe
+2
2
Pyrite + 14Fe3+ + 8H2O 2SO
+ 16H+
4 + 15Fe
2+
Chalcopyrite + 4O2(aq) 2 SO2
+ Cu2+16H+
4 + 17Fe
2+
Chalcopyrite + 16Fe3+ + 8H2O 2SO2
Cu2+ + 16H+
4 + 17Fe
Albite + 4.2H+ 2.95 SiO2(aq) + 1.05 Al3+ + 0.95Na+ + 0.05Ca2+
+ 2.1 H2O
Anorthite + 8H+ 2 Al3+ + 2SIO2(aq) + Ca2+ + 4H2O
Eriochalcite Cu2+ + 2Cl + 2H2O
Halite Na+ + Cl
Sylvite K+ + Cl
Biotite + 10H+ 3Al3+ + 3SiO2(aq)Fe2+ + K+ + 2Mg2+ + 6H2O
Muscovite + 10H+ 3Al3+ + 3SiO2(aq) + K+ + 6H2O
Geothite + 3H+ Fe3+ + 2H2O
Quartz SiO2(aq)
Gypsum Ca2+SO2
4 + 2H2O
+
Mirabilite SO2
4 + 2Na + 10H2O
+
Kaolinite + 6H 2Al3+ + 2SiO2(aq) + 5H2O
Pickeringite 2Al3+ + Mg2+ + 4SO2
4 + 22H2O
3+
K-jarosite + 6H+ K+ + 2 SO2
+ 6H2O
4 + 3 Fe
3+
(K,Na)-jarosite + 6H+ 0.5Na+ + 0.5 K+ + 2 SO2
+ 6H2O
4 + 3 Fe
Hexahydrite Mg2+ + Fe2+ + SO2
4 + 6H2O
Halotrichite 2Al3+ + Fe2+ + 4SO2
4 + 22H2O
Gas dissolution/exsolution
O2(g) O2(aq)
a
b

Equilibrium constants taken from Haung (1989).

Selectivity coefcients taken from Appelo and Postma (1993).

The domain for reactive transport modeling is 1 m of the

unsaturated zone. The 1D nite element grid consists of 100
vertical elements.
Modeled aqueous species are Cu2+, CuCl+ and CuCl2(aq),

Ca2+, Mg2+, MgOH+, Al3+, Na+, K+, Fe2+, Fe3+, SO2

4 , HSO4 ,

Cl , SiO2(aq), H2O and OH . Virial coefcients for most of the

ions are taken from (Harvie et al., 1984). Halite and
eriochalcite solubility in the NaClCuCl2 system were measured by (Filippov et al., 1986). (Haung, 1989) modeled these
experimental results and provided the best t for the constant
equilibrium of the two minerals and Pitzer virial coefcients
between dissolved ions. The same author included copper
chloride complexes for a better t of experimental solubility.
Modeled exchange complexes are CaX2, MgX2, CuX2, NaX and
KX. Due to the absence of experimental exchange coefcients
in Chaaral clays (mainly vermiculite and illite), we have
tentatively used the selectivity coefcients given by (Appelo
and Postma, 1993), and performed a sensitivity analysis of the
inuence of these data on the nal results.
The main reactions considered in the geochemical model
are listed in Table 3. Homogeneous reactions are assumed to be
in equilibrium except for Fe(II) oxidation, which is modeled
using a modied kinetic expression provided by (Singer and
Stumm, 1970). Once the oxidation rates of suldes have been
dened in the model, the evolution of iron in the pore-water
could only be made to t by calibrating simultaneously the
scaling factor for rates of Fe(II)-oxidation and the precipitation
of goethite and K-jarosite. A factor equal to 107 times the abiotic
Fe(II)-oxidation rate is applied, which indicates that Fe(II)oxidation in the pore-water is biocatalyzed.

a
a

Table 4
Summary of kinetic expressions used in the model.
b
b
b
b

a
a

Reaction process

Rate expression (mol m 2 s 1)

Ref.

Pyrite oxidation
(oxygen path)
Pyrite oxidation
(Fe(III) path)
Chalcopyrite oxidation
(oxygen path)
Chalcopyrite oxidation
(Fe(III) path)
Fe(II) oxidation
Albite dissolution
Anorthite dissolution
Muscovite dissolution
Quartz dissolution
Kaolinite precipitation
Goethite precipitation
K-jarosite precipitation
(K,Na)-jarosite
precipitation
Biotite dissolution

0.11 0.5
aO2(aq) ( 1)
R = 10 8.19a
H+

0.4
R = 10 6.7a0.93
Fe3+aFe2+ ( 1)

R = 10 10.58a0.15
H+ ( 1)

R = 10 6.75a0.43
Fe3+ ( 1)

R = 10 3aO2(aq) ( 1)
R = 10 9.17a0.5
H+ ( 1)
R = 109.17a0.5
H+ ( 1)
R = 10 1223a0.38
H+ ( 1)
R = 10 13.38 ( 1)
0.3
10.77 0.5
R = (10
aH+ + 10 166a
H+ )( 1)
R = 10 11( 1)
R = 10 4( 1)
R = 10 4( 1)

14.1 0.29
R = (10 8.5a0.57
aH+ )( 1)
H+ + 10

e
f
g
h
i
j
j
j

Williamson and Rimstidt (1994).

Acero et al. (2007).
Rimstidt et al. (1994).
d
Modied after Singer and Stumm (1970), calibrated to reproduce
observed iron evolution.
e
Chou and Wollast (1985).
f
Equal to albite.
g
Kalinowski and Schweda (1996).
h
Tester et al. (1994).
i
Nagy et al. (1991).
j
Calibrated to reproduce observed iron evolution.
k
Malmstrom and Banwart (1997).
b
c

S.A. Bea et al. / Journal of Contaminant Hydrology 111 (2010) 6582

Table 5
Initial conditions. Summary of initial volumetric fraction and reactive surface
used in the reactive transport model.
Mineral phase

vol.%

Initial surface area (m2 m 3)

Pyrite
Chalcopyrite
Anorthite
Albite
Quartz
Goethite
Muscovite
Biotite
Kaolinite
Gypsum
Sylvite
Halite
Eriochalcite
Mirabilite
Pickeringite
Hexahydrite
(K,Na)-jarosite
K-jarosite
Halotrichite
Total

0.39
0.32
18
6.8
24
0.18
1.7
2.5
5.8
0
0
0
0
0
0
0
0
0
0
60

130
108
5852
2277
8231
61
576
845
1920
0
0
0
0
0
0
0
0
0
0
20,000

73

each mineral phase. A vermiculite content of 3 wt.% is assumed

in the model. Thus, the total exchange capacity considered in
the calculations is computed from this clay content and the
exchange capacity of Jeffersite vermiculite (Foscolos, 1968).
The chemical composition of the initial and boundary
solutions is shown in Table 6.

3.3. Simulated scenarios

Mineral dissolution reactions are modeled using kinetic laws.

The kinetic rates presented in Table 4 are based on the general
expression presented by Lasaga et al. (1994). Most secondary
phases are considered to precipitate in equilibrium. These include
halite, eriochalcite, sylvite, gypsum, mirabilite, halotrichite,
hexahydrite and pickeringite. K-jarosite, goethite and kaolinite
are considered to precipitate kinetically (see Table 4).
Dissolved oxygen is modeled in equilibrium with the
partial pressure of oxygen gas in pores (pO2 = 0.21 atm).
A mixture of suldes (mainly pyrite and chalcopyrite) and
aluminosilicates (mainly quartz, muscovite, biotite, and plagioclase) is considered as initial mineralogy. Their initial mineral fractions are presented in Table 5. The volumetric fraction
occupied by each mineral is calculated from its weight fraction
by considering its density. Weight fractions are estimated from
sequential extractions and mineralogical descriptions reported
by (Dold, 2006). A total reactive surface corresponding to silty
grain size tailings is considered (104 m2 m 3). It is proportionally distributed in accordance with the volumetric fraction of

Six scenarios (shown in Fig. 6) are simulated to account

for the different genetic hypotheses about the formation of
eforescences. The aim is to obtain a Cl/Na ratio higher than 1
in the pore-water, while using the observed types of water.
Sea water is imposed at the bottom in Scenarios I and II
(SWCE and SW, with/without the cation-exchange, respectively), thus implying a sea water wedge reaching the inner
portion of the tailings, which may well have been the case
during early sedimentation. Mixtures (mixing proportion
= 0.5) of sea water/freshwater and brine/freshwater are
considered for Scenarios III and IV (SFWCE and BFWCE),
respectively. Cation-exchange is also considered in both
scenarios.
There are situations in which the Cl/Na ratio is less than 1
in the original groundwater composition and can be reversed
through cation-exchange, or by adding marine aerosol. Both
cases are tested in Scenarios V and VI (FWCE and FWMASL),
respectively. With regard to the latter, a slow, constant ow
of sea water (20 mm year 1, its chemical composition is
described in Table 6) is imposed at the upper 0.1 m of the
model.

4. Modeling results and discussion

All simulations are run for thirty-two years, beginning
after discharges from the El Salvador were stopped in 1975
(period 19752007).
Scenario I (SWCE) is chosen as the base case. Therefore,
we thoroughly describe its features and results in the following sections. Later, the remaining scenarios are analyzed
and compared.

Table 6
Chemical composition of initial, marine aerosol and boundary waters (mol kg w 1).

pH
Fe
Cu
Zn
Ca
Mg
Na
K
Cl
SO4
O2
Si
Al
Ionic strength

Initial solution

Scenarios I, II and marine

aerosol (sea water)

Scenario III
(0.5 sea-water + 0.5 freshwater)

Scenario IV
(0.5 brine + 0.5 freshwater)

Scenarios V and VI
(freshwater)

7.48

10 4

10 4
10 7
10 7
10 7
5.75 10 1
2.9 10 2
2.510 4

0.347

7.48
6.35 10 6
10 3
7.64 10 8
9.8 10 3
5.2 10 2
5.0 10 1
1.07 10 2
5.75 10 1
2.9 10 2

0.726

7.41
4.8 10 6
2.1 10 5
10 4
1.06 10 2
2.9 10 1
3.5 10 1
7.21 10 3
3.69 10 1
2.73 10 2

0.5

7.45
6.8 10 5
8.0 10 6
9.0 10 5
2.23 10 2
8.7 10 1
1.4 10 2
4.8 10 6
8.7 10 1
2.27 10 2

1.01

7.03
7.5 10 4
9.4 10 6
1.6 10 6
1.1 10 2
6.2 10 3
2 10 1
3.7 10 3
1.6 10 1
2.5 10 2

0.27

74

S.A. Bea et al. / Journal of Contaminant Hydrology 111 (2010) 6582

Fig. 6. Simulated scenarios according to the composition of the water imposed at the bottom. Legend: S = sea water; F = freshwater; B = brine; W = water; MASL =
marine aerosol; CE = cation-exchange.

Fig. 7. Scenario I (SWCE): Vertical distribution of (A) temperature (C), (B) salinity, (C) water activity, and (D) pore saturation, at the last winter/summer.

S.A. Bea et al. / Journal of Contaminant Hydrology 111 (2010) 6582

4.1. Thermohydraulic evolution in Scenario I (SWCE)

Spatial distribution of temperature, salinity, water activity
and pore saturation are shown in Fig. 7. The time evolution of
temperature, evaporation rate and energy balance are plotted
in Fig. 8.
As expected, these parameters uctuate seasonally. The
tailings surface temperature declines in the winter as solar
radiation and air temperature do the same (Figs. 7A and 8).
This temperature: (1) is always higher than the atmosphere
temperature, and (2) increases during the summer (approximately 3 C) and over time (Fig. 8A). Both observations are
explained by the energy balance at the upper boundary
(Fig. 8B). Net radiation is at its maximum and most of the
inowing radiation is used in evaporating water. However,
the mean evaporation rate decreases over time as a
consequence of the increase in salinity of the pore-water.
Thus, the latent heat ux (invested in evaporation) declines
over years. To maintain the heat balance, the sensible heat has
to be increased (Fig. 8B), which requires increasing the
surface/air temperature differential (Fig. 8A). The evaporation is produced during the summer, but in the winter there is
slight condensation (Fig. 8B).

75

During the summer as a consequence of the high level of

evaporation (Fig. 7B), salinity increases in the uppermost
portion of tailings. However, during the winter, salinity
decreases slightly or undergoes a slight increase at the top
of tailings.
An opposite evolution is predicted for water activity
(Fig. 7C). It decreases in the uppermost portion of tailings due
to an increase in salinity during the summer. However, it
increases at the top of tailings during the winter. It decreases
slightly between 0.1 m and 0.5 m below the surface during
the same season.
The water saturation prole (Fig. 7D) shows that the
water saturation reached a steady-state prole. The saturation declined from the initial saturated state to about Sl =0.68
at the top and Sl = 0.99 at the base.
4.2. Evolution of pore-water chemistry in Scenario I (SWCE)
Vertical proles for Cl, K, Ca, SO4, Cu, Na and pH in the last
winter/summer are shown in Fig. 9A to G. In general,
concentrations of ions increase in the oxidation zone,
especially at the top. As expected, seasonal uctuations are
predicted for them. Concentrations of solutes decrease during

Fig. 8. Scenario I (SWCE): (A) the time evolution of temperature in the atmosphere and tailings. (B) The time evolution of evaporation rate and energy uxes in the
tailings.

76

S.A. Bea et al. / Journal of Contaminant Hydrology 111 (2010) 6582

Fig. 9. Scenario I (SWCE): Vertical proles of the main species at the last winter/summer. (A) Cl, (B) K, vertical distribution corresponding to Scenario II (SW) is
also shown, (C) Ca, vertical distribution corresponding to Scenario II (SW) is also shown, (D) pH, (E) SO4, (F) Cu, and (G) Na.

S.A. Bea et al. / Journal of Contaminant Hydrology 111 (2010) 6582

77

Fig. 10. Temporal evolution of eforescences obtained from probable scenarios: (A) Scenario I (SWCE), (B) Scenario III (SFWCE), and (C) Scenario IV (BFWCE).
Vertical distribution of mineral volumetric contents in the last summer: (D) Pyrite and chalcopyrite, (E) geothite and K-jarosite, and (F) gyspum and kaolinite.

the winter in the upper part, but increase slightly below the
evaporation front (between 0.1 m and 0.5 m, e.g. see Cl in
Fig. 9A). Similar behavior is also predicted for Na, K, Cu, Ca,
Mg, Al, Zn, Fe(total) and SiO2(aq). However, SO4 shows

different behavior. It decreases its concentration in the upper

part of the model during the summer (see Fig. 9E).
pH is found to decrease in tailings as a consequence of
sulde oxidation (Fig. 9D). It increases sharply to neutral

78

S.A. Bea et al. / Journal of Contaminant Hydrology 111 (2010) 6582

values below the oxidation front (located 0.8 m below the

surface). pH increases slightly in the upper part of the model
during the winter, and the oxidation front rises 0.1 m in the
same season. The concentration of K decreases within the
oxidation zone due to precipitation of K-jarosite and its
exchange onto the clay fraction (Fig. 9B). Ca concentration
increases mainly due to the exchange with Na and limited
plagioclase dissolution (Fig. 9C).
In general, concentration and pH proles are consistent
with eld observations. The total concentration of Cu
predicted after thirty two years is one order of magnitude
higher than eld observations (Fig. 9F). Other processes (not
considered in this model) could probably explain Cu
depletion in the pore-water. For instance, the adsorption of
Cu onto secondary Fe-oxides/hydroxides (i.e. onto goethite
and K-jarosite, see Cu(ox) in Table 1). Although the trend ts
the experimental one, the predicted concentrations of Cl are
clearly higher than measured. A possible explanation for the
calculated high Cu and Cl concentrations could also be an
insufciently accurate solubility value for eriochalcite in such
complex solutions.
4.3. Precipitation of secondary phases and suldes oxidation in
Scenario I (SWCE)
Eforescences precipitate in the upper part of the model
(Fig. 10A). However, seasonal variation in them is predicted.
This process of formation starts in the summer and partial
dissolution (total in the case of eriochalcite) is produced
during the winter. The total dissolution of eriochalcite was
veried in Chaaral tailings during the winter of 2008.
Eriochalcite was only identied in eforescence samples
collected during the summer.
Halite increased over time, whereas eriochalcite almost
reached a constant value (between 15 and 25 mol m 3).
Thus, the halite/eriochalcite ratio observed during the
summer increased over time from 10 to 70 after thirty-two
years.
Goethite and K-jarosite form at the expense of Fe(II)
oxidation (Fig. 10E). Their precipitated amount is consistent
with that observed in Chaaral (0.07 wt.% K equivalent for Kjarosite). In agreement with the eld observations, gypsum
(CaSO42H2O) and kaolinite (Al2Si2O5(OH)4) form throughout the oxidation zone (Fig. 10F). Kaolinite forms as a product
of the dissolution of aluminosilicates (i.e. muscovite, biotite,
and plasgioclase) in an acidic environment. Gypsum is mainly
predicted in the upper part of the model, and its amount
(150 mol m 3) is consistent with Ca measured in the soluble
fraction of soil samples (see Ca(sol) in Table 1).
Pyrite and chalcopyrite oxidation is predicted in the
model. However, a proportion of chalcopyrite remained
after thirty-two years (Fig. 10E) despite the fact that it was
practically exhausted in the analysis of the tailings (see Cu
(sph) in Table 1). Due to the uncertainty in its initial value, it
is probable that chalcopyrite was overestimated in the initial
condition (1.2 pyrite and 0.8 chalcopyrite equivalent). Based
on sequential extractions and assuming that suldes were
preserved by the oxidation processes below the water-table, a
lower initial proportion of pyrite and chalcopyrite equivalent
could be considered (0.65 pyrite and 0.15 chalcopyrite). Chalcopyrite is exhausted in a simulation taking the latter wt.%

as initial conditions after thirty-two years. After thirty-two

years, all silicate proles remain essentially unchanged from
their initial proles. However, the sum of all mineral volume
changes induces a net mineral volume gain, which results in a
maximum porosity decrease of about 0.3% mainly in the upper
part with virtually no change in other parts (not represented).
4.4. Analysis of scenarios
The precipitation of halite is predicted in all scenarios in
the upper part of the model. However, precipitation of
eriochalcite is only reached in scenarios I (SWCE), III (SFWCE)
and IV (BFWCE) (Fig. 11). The time evolution of halite and
eriochalcite for these scenarios is displayed in Fig. 10A to C.
Notice that cation-exchange is considered in all three scenarios.
In Scenario II (SW), precipitation of eriochalcite is inhibited in several ways. First, the saturation rate of eriochalcite is reduced as a consequence of Cl consumed by halite
precipitation (i.e. geochemical divide, Fig. 4B2). Second,
water activity decreased quickly in the upper part of the
model due to high concentrations of other major ions such as
Mg and K, further preventing the saturation of eriochalcite. As
shown in the previous section, the evaporation rate decreases
for low values of water activity. Due to the high concentrations of K and Mg, the precipitation of other KMg salts such
as silvite (KCl) and hexahydrite (MgFe(SO4)26H2O) is
predicted.
No secondary KMg minerals precipitated in scenarios
with cation-exchange. In these cases, as a result of the exchange between Mg2+ and K+ by Ca2+, hexahydrite and
silvite did not reach saturation.
Other eforescences are associated with halite in the
scenario affected by marine aerosol (Scenario VI (FWMASL),
mirabilite and silvite). None of these associated eforescences have been found in Chaaral.
Eriochalcite is also inhibited in Scenario V (FWCE) because
its Cl/Na ratio continuously decreases in the pore-water. This
is enhanced by the Cl/Na ratio of its boundary water (b1) and
cation-exchange processes.
After thirty-two years, scenarios III (SFWCE) and IV
(BFWCE) predicted larger amounts of halite (Fig. 10B and C,
approximately 1400 and 2200 mol m 3, respectively). However, a similar precipitated amount is predicted for eriochalcite

Fig. 11. Evolution of the saturation index of eriochalcite (CuCl22H2O) as a

function of water activity from different scenarios. Eriochalcite precipitation
is only reached in Scenarios I (SWCE), III (SFWCE) and IV (BFWCE).

S.A. Bea et al. / Journal of Contaminant Hydrology 111 (2010) 6582

in all these scenarios (between 20 and 25 mol m 3). Therefore,

in both scenarios III and IV, the halite/eriochalcite ratio is much
larger than observed.
The time evolution of evaporation/condensation rates
obtained from different scenarios are shown in Fig. 12. All of
them present a similar trend (i.e. the evaporation is produced
during the summer, while there is slight condensation in the
winter). However, some differences are predicted in Scenarios I (SWCE) and II (SW) in spite of the fact that sea water is
imposed in both scenarios. The evaporation rate trend is
mainly controlled by the water activity and eforescence
evolution at the evaporation front. As seen above, different
eforescences are predicted for both scenarios (i.e. halite
eriochalcite and halitesilvitehexahydrite in Scenarios I
(SWCE) and II (SW), respectively). The evolution of these
eforescences is different and consequently, so is the evolution of evaporation rates.
The scenario affected by marine aerosol (Scenario VI,
FWMASL) predicts a higher evaporation rate. The dissolution/
precipitation of eforescences (i.e. halitesilvitemirabilite)
controls the evolution of evaporation rates.
4.5. Sensitivity analysis in Scenario I (SWCE)
In order to gain further insight into the hydrogeochemical
behavior of the Chaaral tailings, and to address parameter
uncertainty, several additional simulation scenarios are run
using Scenario I (SWCE) as the base case. Therefore, the
following parameters are tested: (1) the grain size of tailings,
(2) the radiation on the tailings surface, (3) the wt.% to the
clay fraction, (4) the order of selectivity (i.e. through
selectivity coefcients) for cation-exchange, and (5) the
inuence of the tidal cycle.
The results are compared with data on: (1) eforescence
ratios (i.e. halite and eriochalcite), (2) the pore-water
chemistry, (3) the precipitated amount of secondary phases
described in Chaaral (i.e. gypsum, K-jarosite, goethite,
kaolinite), (4) the formation of eforescences not previously
described in Chaaral (e.g. silvite, hexahydrite, mirabilite),
and (5) the dissolution of suldes (i.e. pyrite and chalcopyrite), and aluminosilicates (i.e. biotite and muscovite).
With regard to sensitivity to grain size (i.e. varying the
retention curve), sandy tailings are modeled. As set out above,

Fig. 12. The time evolution of evaporation rates on the tailings surface from
different scenarios (period 19761980).

79

sandy tailings are characteristic toward the sea. The base case
is repeated here using the same geometry and mineralogy,
but with a modied water retention curve for tailings
(P0 = 0.002 (MPa) and = 0.6 for Van Genuchten parameters
in Table 2). The modication had the effect of reducing the
water retention capacity of tailings. As a consequence of
evaporation, the liquid saturation decreases signicantly in
the upper part, and the evaporation front is located
approximately ve centimeters below the surface. As a
consequence of this, the evaporation rate also decreases
during the summer. Major ions such as Mg, Na and K are
strongly exchanged onto the clay fraction and their concentrations decrease in the pore-water. Cl concentration
increases signicantly at the top, so eriochalcite saturation
is immediately achieved. An amount similar to the base case
is predicted for this eforescence. However, halite saturation
occurs one year later as a consequence of the retardation in
the Na front from groundwater. The halite/eriochalcite ratio
reaches 1 during summers, a much lower ratio than observed.
Only a minor amount of other secondary phases is predicted
(i.e. gypsum, goethite, K-jarosite) due to the lower availability of K and SO4 from the groundwater. A clear sky means
greater solar radiation on the tailings surface and increased
evaporation. It affects mainly the amount of halite and
eriochalcite (the precipitate almost doubled) but not their
relative proportion (halite/eriochalcite). It also increases,
however, the K concentration in the pore-water to much
higher values than observed. Due to the higher evaporation
rate in relation to the precipitation rate of K-jarosite, the K
concentration increases in the pore-water.
Low and high clay contents are evaluated (0.3 wt.% and
30 wt.%, respectively). Each one is implemented in the model
using the CEC (Cation-Exchange Capacity) parameter. A
similar geochemical evolution is obtained for low CEC as in
Scenario II (SW) (scenario without cation-exchange). However, approximately 100 mol m 3 of eriochalcite is predicted
for high CEC. In this case, Na+ is strongly exchanged onto the
clay fraction, increasing the Cl/Na ratio in the pore-water.
Thus, saturation of halite is reached two years later. The
precipitated halite/eriochalcite ratio is approximately 1,
which is much lower than observed.
Each type of clay has its own order of cation-exchange
preference (i.e. order of selectivity). The order of selectivity
used in the base case was typical for illite (K+ N Ca2+ N (Cu2+,
Mg2+) N Na+). However, a different order of selectivity was
suggested by other clays (Foscolos, 1968). This varies
according to selectivity coefcients. Sensitivity analysis was
carried out on the NaCu selectivity coefcient in order to
obtain two orders of selectivity: 1) Cu2+ N K+ N Ca2+ N Mg2+ N
Na+ (high afnity for Cu2+), and 2) K+ N Ca2+ N Mg2+ N Na+ N
Cu2+ (low afnity for Cu2+). The concentrations of Na+ and
Cu2+ were modied in the initial solution in order to obtain
the same exchanged equivalent fractions (i.e. for NaX and
CuX2) as the base case simulation. When Cu2+ is strongly
exchanged onto the clay its availability in solution is depleted,
and a maximum of 15 mol m 3 of eriochalcite is predicted to
form. When the afnity of Cu2+ in the exchange complex is
low the opposite behavior is obtained, and precipitation of
32 mol m 3 of eriochalcite is predicted. Therefore, although
there are differences, the resulting halite/eriochalcite ratios
vary within the same order of magnitude with the selectivity

80

S.A. Bea et al. / Journal of Contaminant Hydrology 111 (2010) 6582

coefcients for Cu2+. In all cases, exchanged copper decreases

with ionic strength, primarily attributed to formation of metal
chlorides (mainly CuCl+, CuCl2(aq)), as was veried experimentally with vermiculite (El-Bayaa et al., 2009).
Regarding sensitivity to the tidal cycle, a tidal oscillation of
0.5 m is imposed at the lower boundary. In this case, a higher
hydrodynamic dispersion is induced by tides and saturation
of halite and eriochalcite is reached much later than for the
base case. However, once saturation is reached, the tidal cycle
did not signicantly affect the evolution and mineral proportion of the eforescences in the upper part of the system.

membrane. Their main physical parameters were (see Table 2

for reference): intrinsic permeabilities, 10 14 and 10 20 (m2),
respectively; retention curve parameter P0, 0.002 and 0.0001
(MPa), respectively; retention curve parameter n, 0.6 and
0.19 (), respectively; and thermal conductivity sol, 0.16
(W m 1 K 1) for both layers.
Regarding the results, evaporation is dramatically reduced thus preventing eforescence formation (see the temporal evolution of the saturation index for eforescences in
Fig. 13B).
5. Concluding remarks

4.6. Remediation alternative

In general, remediation strategies on tailings are concerned with the capacity of the systems to minimize water
and oxygen percolation towards the bulk of tailings. However, our remediation strategy focuses mainly on reducing the
evaporation rate on the tailings surface and then preventing
eforescence precipitation. Therefore, a capillary barrier is
proposed as a suitable alternative (Melchior et al., 1993). This
would consist of two layers overlying tailings (see Fig. 13A):
(1) a coarse sand layer (i.e. drainage layer), and (2) a geomembrane. The latter has two main purposes: (a) to isolate
the capillary barrier from the acid mine drainage system, and
(b) to prevent oxygen percolation toward the tailings. The
former could reduce the evaporation rate as a consequence of
its low water content and thermal conductivity. On the other
hand, this cover design might provide an additional organic
soil layer for the development of vegetation (in the present
conditions, vegetation does not grow on tailings due to the
acidic conditions).
The proposed cover design is evaluated through reactive
transport modeling. Thus, the base case mesh was extended
in order to simulate 0.15 m drainage layer and a thick geo-

Although eforescence formation on tailings is common in

hyperarid climates, natural occurrence of eriochalcite is rare.
According to the modeling exercise described above, this is
due to the fact that several environmental and geochemical
constraints are required for its precipitation. A high level of
evaporation and low relative humidity (Hr) have to be
reached. These parameters control water activity evolution
in the pore-water chemistry. Therefore, eforescences form
mainly during the summer when the relative air humidity is
low and solar radiation is high, whereas their dissolution is
produced during the winter when relative humidity increases
and light condensation is produced on the tailings surface.
The modeled result was veried during a visit to Chaaral
tailings during the winter of 2008. Eriochalcite could only be
collected as eforescence during summer.
With regard to geochemical conditions, the Cl/Na ratio is a
key parameter in the formation of eriochalcite. The strong
competition of other cations for Cl may form halite and
sylvite, and inhibit the precipitation of eriochalcite. Precipitation of halite creates a geochemical divide, allowing either
Cl or Na+ concentrations to increase, depending on whether
the Cl or Na+ concentration is higher in the pore-water.

Fig. 13. Remediation alternative: capillary barrier on tailings. (A) Cover design (Melchior et al., 1993). (B) Temporal evolution of the saturation index for
eforescences on the tailings surface considering and not considering the capillary barrier.

S.A. Bea et al. / Journal of Contaminant Hydrology 111 (2010) 6582

Eriochalcite forms only in those samples with a Cl/Na ratio

higher than 1. Three mechanisms are proposed to obtain a
higher Cl/Na ratio: (1) the exchange of Na+ by Ca2+, Mg2+,
Cu2+ and K+ onto a vermiculite-type mixed-layer, (2) the
inuence of the marine aerosol on the tailings surface, and
(3) the precipitation of secondary Na phases. According to the
mineralogy and solute concentration proles of the different
modeling scenarios, cation-exchange is the most likely mechanism to obtain a higher Cl/Na ratio in the Chaaral tailings.
In fact, it is also supported by analytical evidence: (1) a signicant fraction of Cu, Na, and Ca were exchanged at least in
the oxidation zone of these tailings (e.g. see Cu(ex), Na(ex)
and Ca(ex) in Table 1). In the case of Cu, it may be released
from this vermiculite-type mixed-layer mineral from samples
in the low pH oxidation zone (Dold, 2003), (2) water samples
collected in the oxidation zone (below the evaporation front,
corresponding to boreholes CH1, CH2 and CH3) present a
Cl/Na ratio N 1. In this case, no Namineral phases precipitate in
this zone, and only cation-exchange processes could explain
these ion ratios. With regard to the probable scenario for the
formation of these eforescences, a scenario with two stages is
proposed: a rst stage involving sea water (Scenario I (SWCE),
or a mixture of sea water/freshwater (Scenario III (SFWCE)) in
the groundwater composition, and a second stage with the
intrusion of freshwater. The rst stage is based on the amount
of eriochalcite formed, whereas the freshwater intrusion is
required to explain the observed halite/eriochalcite ratio and
the vertical concentration distribution in the rst 0.5 m above
the water-table.
Sensitivity analysis of different physical and geochemical
parameters is evaluated in the model. The result is particularly sensitive to the clay content, because Na exchanged and,
then the Cl/Na ratio increases with the clay content. As a
result of this, a higher amount of eriochalcite is predicted and
a halite/eriochalcite ratio lower than observed. The model is
less sensitive to the cation-exchange selectivity coefcients.
Thus, although higher afnity for Cu2+ in the exchange
complex decreases the amount of eriochalcite formed, the
variations in the halite/eriochalcite ratio are kept close to the
base case value.
Moreover, the model is also sensitive to grain size. Thus, for
more sandy tailings, the liquid content in the uppermost part of
the model is markedly reduced. Then, eriochalcite precipitates
immediately and a lower evaporation rate is predicted.
Therefore, less capillary transport and precipitation of secondary mineral phases (i.e. gypsum, goethite and K-jarosite) are
predicted in the oxidation zone. However, for consistence
between grain size and clay content, the latter should be
reduced, and eriochalcite formation decreases accordingly.
On the other hand, the model is not particularly sensitive to
the tidal cycle. In this case, major hydrodynamic dispersion in
the oxidation zone is predicted but it did not signicantly affect
the evolution of eforescences in the upper part of the model.
With regard to preventing the formation of eriochalcite, a
capillary barrier on the tailings surface is proposed as a suitable
alternative (previous excavation of precipitated eforescences). Its efciency as a barrier is also tested through reactive
transport (approximately 0.15 m in thickness was considered
in the simulation). Precipitation of eriochalcite is inhibited
because evaporation rate on the tailings surface is dramatically
reduced.

81

Acknowledgments
This work was partially funded by a point action CSIC/
CONYCIT (20082009), the project CTM2007-66724-C02/
TECNO of the Spanish Government, and Swiss National Science
Foundation projects No. 200020-117792/1 and 200021105507/1. Thanks are also to anonymous reviewer for the
valuable comments and suggestions that have signicantly
improved the nal version of the paper.
References
Acero, P., Ayora, C., Carrera, J., 2007. Coupled thermal, hydraulic and
geochemical evolution of pyritic tailings in unsaturated column experiments. Geochimica et Cosmochimica Acta 71, 53255338.
Appelo, C., D. Postma, 1993. Geochemistry, Groundwater and Pollution,
Balkema, A.A. Ed.
Bain, J., Blowes, D.W., Robertson, W., Frind, E., 2000. Modelling of sulde
oxidation with reactive transport at a mine drainage site. Journal of
Contaminant Hydrology 41 (12) 23.47.
Bea, S.A., Carrera, J., Batlle, F., Ayora, C., Saaltink, M., 2009. CHEPROO: a
Fortran 90 object-oriented module to solve chemical processes in Earth
Science models. Computers & Geosciences 35 (6), 10981112.
Bea, S.A., J. Carrera, C. Ayora, . Pitzer algortithm: Efcient implementation of
Pitzer equations in geochemical and reactive transport models, Computers & Geosciences (in press).
Camus, F., 2003. Geologa de los sistemas porfricos en los Andes de Chile.
Santiago de Chile, Servicio Nacional de Geologa y Minera.
Castilla, J.C., 1983. Environmental-impact in sandy beaches of copper mine
tailings at Chaaral. Chile, Marine Pollution Bulletin 14, 459464.
Chou, L., Wollast, R., 1985. Steady-state kinetics and dissolution mechanisms
of albite. American Journal of Science 285, 963993.
Dold, B., 2003. Speciation of the most soluble phases in a sequential
extraction procedure adapted for geochemical studies of copper sulde
mine waste. Journal of Geochemical Exploration 80, 5568.
Dold, B., 2006. Element ows associated with marine shore mine tailings
deposits. Environmental Science & Technology 40, 752758.
Dold, B., Fontbote, L., 2001. Element cycling and secondary mineralogy in
porphyry copper tailings as a function of climate, primary mineralogy,
and mineral processing. Journal of Geochemical Exploration 74, 355.
Edlefson, N., Anderson, A., 1943. Thermodynamics of soil mix- water-wetted
porous media. Water Resources Research 35, 635649.
El-Bayaa, A.A., Badawy, N.A., Abd AlKhalik, E., 2009. Effect of ionic strength on
the adsorption of copper and chromium ions by vermiculite pure clay
mineral. Journal of Hazardous Materials 170, 12041209.
Farina, 2001. Temporal variation in the diversity. Marine Pollution Bulletin.
Filippov, V.K., Charykov, N.A., Fedorov, Y.A., 1986. NaClNiCl2(CuCl2)H2O
systems at 25 C. Zhurnal Neorganicheskoi Khimii 31, 18611866.
Foscolos, A.E., 1968. Cation-exchange equilibrium constants of aluminumsaturated montmorillonite and vermiculite clays. Soil Science Society of
America Journal 32, 350354.
Fuenzalida, P., 1950. Geografa econmica de Chile I. CORFO 188254.
Fuenzalida, P., 1965. Biogeografa. Geografa econmica de Chile, CORFO.
Gerke, H., Molson, J., Frind, E., 1998. Modelling the effect of chemical
heterogeneity on acidication and solute leaching in overburden mine
spoils. Journal of Hydrology 209 (14), 166185.
Gerke, H., Molson, J., Frind, E., 2001. Modelling the impact of physical and
chemical heterogeneity on solute leaching in pyritic overburden mine
spoils. Ecological Engineering 17 (23), 91101.
Gustafson, L.B., Hunt, J.P., 1975. Porphyry copper-deposit at El Salvador,
Chile. Economic Geology 70, 857912.
Gustafson, L.B., Quiroga, J., 1995. Patterns of mineralization and alteration
below the porphyry copper orebody at El Salvador. Chile, Economic
Geology and the Bulletin of the Society of Economic Geologists 90, 216.
Harvie, C.E., Moller, N., Weare, J.H., 1984. The prediction of mineral
solubilities in natural waters: the NaKMgCaHClSO4OHHCO3
CO3CO2H2O system to high ionic strengths at 25 C. Geochimica et
Cosmochimica Acta 48, 723751.
Haung, H.H., 1989. Estimation of Pitzer ion interaction parameters for
electrolytes involved in complex-formation using a chemical-equilibrium
model. Journal of Solution Chemistry 18, 10691084.
Kalinowski, B.E., Schweda, P., 1996. Kinetics of muscovite, phlogopite, and
biotite dissolution and alteration at pH 14, room temperature.
Geochimica et Cosmochimica Acta 60 (3), 367385.
Larrain, H., Velasquez, F., Cereceda, P., Espejo, R., Pinto, R., Osses, P.,
Schemenauer, R.S., 2002. Fog measurements at the site Falda Verde

82

S.A. Bea et al. / Journal of Contaminant Hydrology 111 (2010) 6582

north of Chaaral compared with other fog stations of Chile. Atmospheric Research 64, 273284.
Lasaga, A.C., Soler, J.M., Ganor, J., Burch, T.E., Nagy, K.L., 1994. Chemical
weathering rate laws and global geochemical cycles. Geochimica et
Cosmochimica Acta 58, 23612386.
Lee, Correa, 2005. Effects of copper mine tailings. Marine Environmental
Research.
Lee, Correa, Castilla, 2001. An assesment. Marine Pollution Bulletin.
Lefebvre, R., Hockley, D., Smolensky, J., Gelinas, P., 2001a. Multiphase transfer
processes in waste rock piles producing acid mine drainage I: conceptual
model and system characterization. Journal of Contaminant Hydrology
52 (14), 137164.
Lefebvre, R., Hockley, D., Smolensky, J., Lamontagne, A., 2001b. Multiphase
transfer processes in waste rock piles producing acid mine drainage 2.
Applications of numerical simulation. Journal of Contaminant Hydrology
52 (14), 165186.
Malmstrom, M., Banwart, S., 1997. Biotite dissolution at 25 C: the pH
dependence of dissolution rate and stoichiometry. Geochimica et
Cosmochimica Acta 61, 27792799.
Mayer, K.U., Frind, E.O., Blowes, D.W., 2002. Multicomponent reactive
transport modeling in variably saturated porous media using a
generalized formulation for kinetically controlled reactions. Water
Resources Research 38, 1174.
Melchior, S., Berger, O., Vielhaber, G., Miehlich, G., 1993. Comparison of
effectiveness of different liner systems for top cover. Proceeding of 4th
International Landll Symposium, S. Margherita Di Pula, Cagliari, Itlia,
pp. 225234.
Nagy, K.L., Blum, A.E., Lasaga, A.C., 1991. Dissolution and precipitation
kinetics of kaolinite at 80 C and pH 3 the dependence on solution
saturation state. American Journal of Science 291, 649686.
Olivella, S., Carrera, J., Gens, A., Alonso, E.E., 1994. Nonisothermal multiphase
ow of brine and gas through saline media. Transport in Porous Media
15, 271293.
Palache, C., Foshag, W., 1938. Antofagastite and bandylite, two new copper
minerals from Chile. American Mineralogist 23 (2), 8590.
Rimstidt, J.D., Chermak, J.A., Gagen, P.M., 1994. Rates of reaction of galena,
sphalerite, chalcopyrite, and arsenopyrite with Fe(III) in acidic solutions.
Environmental Geochemistry of Sulde Oxidation 550, 213.

Saaltink, M.W., Ayora, C., Carrera, J., 1998. A mathematical formulation for
reactive transport that eliminates mineral concentrations. Water
Resources Research 34, 16491656.
Singer, P.C., Stumm, W., 1970. Acidic mine drainage. Rate-determining step.
Science 167, 11211123.
Tester, J.W., Worley, W.G., Robinson, B.A., Grigsby, C.O., Feerer, J.L., 1994.
Correlating quartz dissolution kinetics in pure water from 25 C to
625 C. Geochimica et Cosmochimica Acta 58, 24072420.
Thompson, M.V., Palma, B., Knowles, J.T., Holbrook, N.M., 2003. Multi-annual
climate in Parque Nacional Pan de Azcar, Atacama Desert, Chile. Revista
Chilena de Historia Natural 76, 235254.
Van Genuchten, M.T., 1980. A closed-form equation for predicting the
hydraulic conductivity of unsaturated soils. Soil Science Society of
America Journal 44 (5), 892898.
Williamson, M.A., Rimstidt, J.D., 1994. The kinetics and electrochemical ratedetermining step of aqueous pyrite oxidation. Geochimica et Cosmochimica Acta 58, 54435454.
Wisskirchen, C., Dold, B., Spangenberg, J.E., 2006. Hydrogeochemical and
stable isotope study of the watershed of the El Salado valley and its
waters inltrating into marine shore tailings deposit at Chaaral
(northern Chile). Congreso Geolgico Chileno Vol. 2. Symposio Hidrogeologa, pp. 671674.
Wolery, T., Daveler, S., 1992. EQ6, a computer program for reaction path
modeling of aqueous geochemical system: theoretical manual, user's
guide, and related documentation (version 7.0). UCLR-MA-110662 PT IV,
Lawrence Livermore Natl. Lab. Livermore, California.
Wunderly, M.D., Blowes, D.W., Frind, E.O., Ptacek, C.J., 1996. Sulde mineral
oxidation and subsequent reactive transport of oxidation products in
mine tailings impoundments: a numerical model. Water Resources
Research 32 (10), 31733187.
Xu, T.F., White, S.P., Pruess, K., Brimhall, G.H., 2000. Modeling of pyrite
oxidation in saturated and unsaturated subsurface ow systems.
Transport in Porous Media 39 (1), 2556.