Science of the Total Environment 433 (2012) 323330

Contents lists available at SciVerse ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

From highly polluted Zn-rich acid mine drainage to non-metallic waters:

Implementation of a multi-step alkaline passive treatment system to remediate
metal pollution
Francisco Macas a,, Manuel A. Caraballo a, b, Tobias S. Rtting c, Rafael Prez-Lpez a, d,
Jos Miguel Nieto a, Carlos Ayora d
a
Geology Department, University of Huelva, Campus El Carmen, E-21071 Huelva, Spain
b
Department of Geosciences, Virginia Tech, Blacksburg, VA 24061, USA
c
Department of Geotechnical Engineering and Geosciences, Technical University of Catalonia UPCBarcelona Tech, Jordi Girona 1-3, D2-006, E-08034 Barcelona, Spain
d
Institute of Environmental Assessment and Water Research, CSIC, Jordi Girona 18, E-08034 Barcelona, Spain

a r t i c l e i n f o a b s t r a c t

Article history: Complete metal removal from highly-polluted acid mine drainage was attained by the use of a pilot multi-
Received 30 January 2012 step passive remediation system. The remediation strategy employed can conceptually be subdivided into
Received in revised form 22 June 2012 a rst section where the complete trivalent metal removal was achieved by the employment of a previously
Accepted 24 June 2012
tested limestone-based passive remediation technology followed by the use of a novel reactive substrate
Available online 20 July 2012
(caustic magnesia powder dispersed in a wood shavings matrix) obtaining a total divalent metal precipita-
Keywords:
tion. This MgO-step was capable to abate high concentrations of Zn together with Mn, Cd, Co and Ni below
Acid mine drainage the recommended limits for drinking waters. A reactive transport model anticipates that 1 m 3 of MgO-DAS
Metal pollution remediation (1 m thick 1 m2 section) would be able to treat a ow of 0.5 L/min of a highly acidic water (total acidity
Caustic magnesia of 788 mg/L CaCO3) for more than 3 years.
Brucite 2012 Elsevier B.V. All rights reserved.
Hydrozincite

1. Introduction these waters (Sanchez Espaa and Trevor, 2007), and both systems
play a key role in the natural attenuation of AMD (Bigham and
Acid mine drainage (AMD) is a major cause of water pollution Nordstrom, 2000), and therefore, they must be taken into account
arising from poor management of mine activities, and it can be con- in remediation attempts. Another important AMD constituent in the
sidered as the principal source of inorganic water pollution in mining IPB is Zn which has an estimated discharge to the Atlantic Ocean of
districts (Younger et al., 2002). AMD polluted waters typically show 3500 t/year, an almost 15% of global gross ux of continental Zn to
low pH values, high acidity and high dissolved metal contents. The the oceans (Olas et al., 2006; Sarmiento et al., 2009b). Also other
Iberian Pyrite Belt (IPB) can be considered one of the largest massive divalent metals such as Mn, Cu, Cd, Co, Ni and Pb are present in
sulde districts in the world. This region although mined since 3000 these AMDs, all of them remain in solution until pH values of about
BC (Nocete et al., 2005), experimented its most extensive and intense 8.5 (Cortina et al., 2003), being this an important factor to be consid-
mining period during the 19th and 20th centuries with more than ered in remediation strategies.
100 active mines (Sez et al., 1999). These abandoned and uncon- Given this severe and widespread metal pollution, the IPB is reg-
trolled mines have left an enormous amount of pollution sources dis- arded as a world-class example of AMD polluted region and as the
charging AMD into the surrounding basins (Olas et al., 2004; Sarmiento perfect eld-laboratory to develop and test remediation strategies
et al., 2009a; 2011), particularly in the Odiel basin, where 427 km of its applicable to a wide range of metal and metalloid pollutants, as well as
rivers are affected by AMD (Sarmiento et al., 2009b). other AMD polluted sites around the world where similar acid systems
Iron and Al are two of the main contaminants present in the highly are present and efforts are made for its remediation (e.g. Strosnider
metal-polluted AMD typically found in the IPB, discharging to the and Nairn, 2010; Caraballo et al., 2010; Behum et al., 2011).
Atlantic Ocean an average load of 7900 t/year Fe and 5800 t/year Al Furthermore, the European Water Framework Directive 2000/60/
(Olas et al., 2006). Iron and Al present strong buffering systems in EC exhorts the countries belonging to the European Union to achieve
good ecological and chemical quality of all the European water bodies
by 2015. However, in the context of the Odiel basin, due to its specic
Corresponding author. Tel.: +34 95 921 9834; fax: +34 95 921 9810. characteristics related to an ancient and uncontrolled AMD pollution,
E-mail address: [email protected] (F. Macas). the regional authorities propose 2021 and 2027 as new deadlines for

0048-9697/$ see front matter 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.scitotenv.2012.06.084
324 F. Macas et al. / Science of the Total Environment 433 (2012) 323330

compliance with the Directive, depending on the severity of AMD pol-

lution (BOJA, 2011).
AMD pollution sources in the Odiel basin, and in many other old
mining districts in Europe and in the world, typically correspond to
orphan sites. In addition, the lifetime of this pollution generation
has been estimated to range between tens to hundreds and hundreds
to thousands of years for coal or sulde mining districts, respectively
(Younger, 1997). Taking these considerations into account, passive
treatment systems are proposed as the most feasible and sustainable
remediation option from both economical and environmental points
of view (PIRAMID-Consortium, 2003; Johnson and Hallberg, 2005).
Conventional passive treatment systems (e.g., anoxic limestone
drains, reducing and alkalinity producing systems or limestone sand
reactors) are based on limestone dissolution or a combination of
limestone dissolution and sulfate reducing bacteria activity (Hedin
et al., 1994; Jage et al., 2001; Watten et al., 2005). Their application,
although successful in waters with low to intermediate metal concen-
trations (e.g., AMDs in coal mining districts), is highly discouraged in
highly polluted AMDs because their high Fe and Al concentrations
promote serious problems of clogging and loss of reactivity (Rtting,
2007). Moreover, the pH values reached by limestone dissolution
are insufcient to precipitate divalent metals (Cortina et al., 2003)
and the high Zn and Cu concentrations of these waters may be lethal
for sulfate reducing bacteria (Cabrera et al., 2006).
During the last ten years, the investigations on remediation of
highly metal-polluted AMD in the IPB have been focused on the use
of a novel reactive substrate known as Dispersed Alkaline Substrate
(DAS). The concept of DAS relies on a mixture of a ne-grained alka- Fig. 1. General (A) and detailed (B) eld site location, and schematic view of the DAS
line reagent (e.g., limestone sand or MgO powder) and an inert coarse passive treatment system (C).
matrix (e.g., wood shavings). Limestone-DAS is suitable for trivalent
metal removal like Fe or Al (Caraballo et al., 2009a; 2011a, b), where-
as MgO-DAS efciently removes divalent metals like Zn, Mn, Cd, Co,
(T2) and two further aeration cascades and decantation ponds (D3,
and Ni (Rtting et al., 2006; Caraballo et al., 2009b). In addition, a
D4). Finally, AMD ows through a reactive tank of 1 m 3 in volume
Natural Fe-Oxidizing Lagoon (NFOL) was designed and tested in the
(T3) lled with MgO-DAS reactive mixture (80% (v/v) pine wood
eld to oxidize and remove some of the abundant Fe prior to the
shavings and 20% (v/v) MgO powder). This tank was equipped with
limestone-DAS treatment (Macas et al., 2012).
lateral sampling ports at 1, 5, 10, 15, 20, 30, 40, 60 and 80 cm depth
The aim of the present study is to obtain for the rst time at the
for water sampling and measurement of physicochemical parameters.
IPB a complete metal removal from a highly divalent and trivalent
The MgO-DAS tank outow discharges into the Monte Romero creek.
metal polluted AMD. To this end the hydrochemistry and mineralogy
High purity limestone sand (99.9% CaCO3) was purchased at a local
of the system will be discussed, paying special attention in the pro-
quarry (nominal grain size 36 mm). Pine wood shavings (curved
cess involved in divalent metal removal. The most innovative aspect
akes with a maximum length of about 1.5 cm) were purchased at a
of the work is the implementation on a pilot scale of an MgO-based
local carpentry. MgO powder was supplied by Magnesitas Navarras
passive treatment focused on retention of divalent metals as the last
S.A. (Zubiri, NE-Spain). The bulk chemistry of the caustic magnesia
step of a more complex system capable of treating highly polluted
type used in this study (LIXIMAG) can be observed in Table 1, as
AMD. Moreover, the MgO-based treatment of divalent metal pollu-
well as the average grain size, which is typically less than 0.1 mm.
tion in surface waters would not only be a local case study applied
The chemical composition and physicochemical parameter range
to acid mine water, but also circum-neutral mine water discharges
of the AMD at the mine shaft are shown in Table 2 (see Inow). The
with high divalent metal concentrations (e.g. Zn) can occur in mining
mean ow rate emerging from the mine shaft and owing through
areas where ore suldes are hosted in carbonate rocks, for instance,
the NFOL pretreatment was 1.5 L/s. In order to adjust the inow rate
Mississippi Valley-type leadzinc deposits.
to the pilot scale of the treatment, the inow at the rst limestone-
DAS tank was set to 1 L/min. The residence time (assuming a porosity
2. Materials and methods
of 50% in the reactive tanks) was 1.5 days for T1 and T2, 3 days for

2.1. Field remediation site

The passive pilot plant constructed at Monte Romero abandoned Table 1

mine complex is located in the northern part of the IPB (SW Iberian Bulk composition and particle size distribution of the caustic magnesia used in the T3
Peninsula) (Fig. 1A). The rst step is a natural pretreatment (NFOL), MgO-DAS tank.
constructed near the mine shaft where AMD groundwater emerges. Source: Magnesitas de Navarra S.A., LIXIMAG.
The outow of the NFOL is connected with the DAS passive treatment Compounds (%) Granulometry (%) L.O.I. (%)
system by a pipe (Fig. 1B). A schematic view of the DAS system is
MgO 65.34 >0.5 mm 0.0 14.6
shown in Fig. 1C. The AMD is gravity forced through a downow reac- CaO 9.80 >0.2 mm 9.0 Bulk density (kg/L)
tor of 3 m 3 in volume (T1) lled with limestone-DAS (80% (v/v) pine SiO2 3.09 0.20.1 mm 22.1 1.10
wood shavings and 20% (v/v) limestone sand) connected in series to Fe2O3 2.45 b0.1 mm 68.9
two decantation ponds (D1, D2) of 6 m 3 in volume each with aeration Al2O3 0.38
SO3 4.14
cascades. AMD then ows through an identical limestone-DAS reactor
 F. Macas et al. / Science of the Total Environment 433 (2012) 323330 325

Table 2
Chemical composition and physicochemical parameters of the inow (mine shaft) and outow (discharge of T3) in the passive treatment of Monte Romero. Data obtained from ve
sampling campaigns.

Major elements (mg/L)

Al Ca Cu Fe Mg Mn Si Zn

Inow 7091 193213 2.13.4 237281 219231 1314 3035 342364

Outow b0.2 199767 b0.005 b0.2 200488 b0.2 b0.2 b0.05
FAO 5 n.r. 0.2 5 n.r. 0.2 n.r. 2
WHO 0.2 n.r. 2 0.3 n.r. 0.4 n.r. 3

Minor elements (g/L)

As Cd Co Cr Ni Pb Sr Ti

Inow 395718 447516 591654 3.16.7 581603 123162 187224 711

Outow b2 b1 b1 b1 b1 b1 182445 b1
FAO 100 10 50 100 200 5000 n.r. n.r.
WHO 10 3 n.r. 50 70 10 n.r. n.r.

Physicochemical parameters

pH Eh (mV) Conductivity (mS/cm) Dissolved oxygen (mg/L) Net acidity (mg/L) T (C)

Inow 3.553.61 471503 1.633.76 0.881.55 14041609 18.820.7

Outow 9.3310.35 276335 1.455.26 0.811.86 0 21.626.9

n.r. = no references; FAO = irrigation water recommendations of the Food and Agriculture Organization of the United Nations, and WHO = drinking water limits according to the
World Health Organization. The symbol b followed by a number corresponds to samples where the analyzed value was below the detection limit of the analytical method for that
element.

each decantation pond and 0.5 days for T3; i.e. a total of 15.5 days of method used was designed to estimate major, minor and trace
residence time for the whole system. elements in waters affected by AMD (Tyler et al., 2004). Multi-
elemental standard solutions prepared from single certied standards
2.2. Water and solid sampling supplied by SCP SCIENCE were used for calibration. Certied Reference
Materials (SRM-1640 NIST freshwater-type and inter-laboratory
Water samples were taken every 15 days in the most representa- standard IRMM-N3 wastewater test material, European Commission
tive points of the system: mine shaft, NFOL, inow and outow of the Institute for Reference Materials and Measurements) were also ana-
reactive tanks (T1-in, T1-out, T2-in, T2-out, T3-in and T3-out), and lyzed. Detection limits were: 200 g/L for Al, Fe, Mn, Mg and Si;
the outows of the decantation ponds D1 and D3. Additional water 500 g/L for Ca; 50 g/L for Zn; 5 g/L for Cu; 2 g/L for As and
samples were taken less frequently in the depth prole of T3 reactive 1 g/L for Cd, Co, Cr, Ni, Pb, Sr and Ti, and the analytical error was
tank. In all sampling campaigns the samples were ltered immedi- lower than 5%.
ately after collection through 0.1 m Millipore lters on Millipore Solid samples used for chemical analysis and X-ray diffraction (XRD)
syringe lter holders, acidied in the eld to pH b 1 with Suprapure were crushed and quartered to obtain representative sub-samples,
HNO3 and stored at 4 C in sterile polypropylene containers until which were ground to b0.1 mm in an agate mill. XRD patterns were
analysis. obtained with a Bruker D5005 X-ray Diffractometer with Cu K radia-
Physicalchemical parameters were measured in the eld using tion. Diffractometer settings were 40 kV, 30 mA, a scan range of 365
specic portable meters. The pH and redox potential were determined 2, 0.02 2 step size and 2.4-s counting time per step. To obtain the
using a PH25 meter (Crison) with Crison electrodes, which was bulk chemistry of the T3 reactive tank precipitates, the sub-samples
calibrated with 3 points (4.01, 7.00, 9.21) for pH and checked using were digested with concentrated HNO3. Metal concentrations of the
2 points (240 and 470 mV) for redox potential with Crison standard different digestions were analyzed by ICP-AES.
solutions. Electrical conductivity and temperature were measured Semi-quantitative chemical analysis and imaging of certain areas in
using a CM35 meter (Crison) with 3 calibration points (147 and the precipitates were obtained by the use of a JEOL JSM 5410 scanning
1413 S/cm, and 12.88 mS/cm). Dissolved O2 was measured with electron microscope with an energy dispersive system (SEM-EDS). To
an auto-calibrating Hanna meter. Gross alkalinity was determined gain a more accurate composition of some specic precipitates at the
by CHEMetrics Total Titrets, with a range of 10100 or 100 micro-scale, quantitative chemical analysis was performed with a JEOL
1000 mg/L as CaCO3 equivalents. JXA-8200 SuperProbe Electron Probe Micro-Analyzer (EPMA), using
The drought season in the south of Spain promotes the cease of the tted Wavelength-Dispersive Spectroscopy (WDS) equipment.
the AMD ow from the mine shaft at the end of the summer period.
At this moment, a depth prole of T3 reactive tank was dug out to
study the newly-formed precipitates developed within the tank. 2.4. Reactive transport modeling
Solid samples were taken at 06, 612, 1220, 2030, 3040, 4050,
5060, 6070 and 7080 cm depth. Samples were preserved in A 1-D reactive transport model of the T3 tank was created using the
plastic bags at room temperature in the dark during its transport to code RETRASO (Saaltink et al., 2004) to match the experimental obser-
laboratory, where they were dried at room temperature to prevent vations on water chemistry and precipitates, corroborate the hypoth-
mineral transformation. esized geochemical processes, and extrapolate its future behavior.
Hydrogeochemical speciation and saturation indices were calculated
2.3. Analytical techniques using PHREEQC (Parkhurst and Appelo, 1999). Thermodynamic data
for aqueous, mineral and gas phases were taken from the WATEQ4F
Analyses of water samples were carried out in the Central Research database (Ball and Nordstrom, 1991). Data for hydrozincite was
Services of the University of Huelva by Inductively Coupled Plasma added according to Schindler et al. (1969), while brucite solubility
Atomic Emission Spectrometry (ICP-AES, Jobin Ybon Ultima2). The was updated according to Xiong (2008).
326 F. Macas et al. / Science of the Total Environment 433 (2012) 323330

3. Results and discussion with the Fe and Al precipitates. A detailed explanation of the afore-
mentioned processes and the Fe and Al mineral phases systematically
3.1. Passive treatment system hydrochemistry found in this step of the passive treatment system can be obtained in
Caraballo et al. (2009a; 2011a).
The general hydrochemical behavior of the complete passive In the following decantation ponds (D1 and D2), almost all the
system with respect to the main major elements (Fe, Al, Ca, and remaining Fe is oxidized and precipitated in the form of schwertmannite
Mg), alkalinity and pH is shown in Fig. 2A. leading to some alkalinity consumption (Fig. 2A). A slight increase
An important Fe removal (165 mg/L) occurs between the mine in water pH is also observed as a response of CO2 degassing when the
shaft and T1-in due to the natural pretreatment in the NFOL. This treated water contacts the atmosphere. In the second limestone-DAS
high Fe removal can be attributed to different oxidation processes tank and the third and fourth decantation ponds, the residual Fe and
such as surface catalyzed oxidation of ferrous iron (SCOOFI; Jarvis Al from the previous stage are now completely removed by the same
and Younger, 2001; Younger et al., 2002) and biotic and abiotic oxida- processes mentioned for T1. Due to this preconditioning, the MgO-DAS
tion and subsequent precipitation of schwertmannite (Macas et al., tank (T3) treated an alkaline drainage with no trivalent metals and
2012). Additionally and in accordance to previous studies (Macas high concentration of divalent metals (Zn, Mn, Cd, Co and Ni). This ab-
et al., 2009), 75% of the As present in the NFOL is co-precipitated sence of trivalent metals has been exposed as the best scenario to assure
with schwertmannite. A decrease in AMD pH is observed in the tran- an optimal performance of the MgO-DAS (Caraballo et al., 2009b). As
sit of the water along the NFOL (Fig. 2), moving from 3.6 in the mine expected in the neutral pH conditions observed before T3, divalent
shaft to 2.4 in T1-in. This pH tendency can be explained by the release metal concentration remained constant in the water with the exception
of protons due to Fe hydrolysis and precipitation of schwertmannite. of a slight increase in Zn concentration and a local Cd decrease within T2
This decrease in pH although could be understood as detrimental for (Fig. 2B). Zinc increase along the limestone sections of the system could
the water quality is in fact benecial because it enhances calcite be attributed to various factors: water evaporation in the ponds during
dissolution in the rst limestone-DAS tank (T1). the hot summer, high residence time of the system (inow water does
In T1, due to limestone dissolution, an increase in water pH (from not have to match the water in the different sections of the treatment)
2.4 to 6), gross alkalinity (from 0 to 190 mg/L as CaCO3 equivalents) and possible redissolution of Zn-particulate matter transported by
and Ca concentration (from 221 to 608 mg/L) takes place (Fig. 2A). the wind from the sulde tailings in the proximities of the treatment.
These changes in the water chemistry induce a strong trivalent The process controlling Cd removal within T2 has been described as
metal removal within the tank, as it is reected in the decrease pH-dependent adsorption in previous studies (Caraballo et al., 2009b).
from 100 to 20 mg/L and 90 to 6 mg/L Fe and Al, respectively As shown in Fig. 2B and Table 2, a complete Zn, Mn, Cd, Co and Ni
(Fig. 2A). The residual As not retained in the NFOL as well as all Cu, removal was attained after T3. The hydrochemistry and mineralogy
Pb and Cr present in the AMD are completely removed within the developed within the MgO-DAS reactive tank will be discussed in the
rst limestone-DAS tank due to adsorption and/or co-precipitation next section.
In order to synthesize the enormous water quality improvement
achieved in the different sections of the passive treatment, a modied
Ficklin diagram has been employed. This diagram is commonly
employed to classify AMD waters on the basis of pH and metal con-
centration (Ficklin et al., 1992; Ros et al., 2008). The employed clas-
sication ranges from high acidity (HA) to moderate alkalinity
(MAlk) and from extreme metallic (EM) to low metallic (LM). The
main pollutants, based on their concentration and toxicity, present
on the highly metal polluted waters of Monte Romero has been intro-
duced in the modied Ficklin diagram (Fig. 3). As can be observed,
the initial AMD belongs to the moderate acidity-extreme metallic
class and after the pH decrease in the NFOL evolves to high acidity-
extreme metallic. In the sections of the treatment dominated by lime-
stone dissolution the AMD changes to near neutral-extreme metallic.

Fig. 2. Fe, Al, Ca, Mg (mg/L), alkalinity (mg/L as CaCO3 equivalents) and pH (A), and Zn, Fig. 3. Modied Ficklin diagram for the different steps of the treatment. HA: high acidity,
Mn (mg/L), Cd, Co, Ni (g/L) and pH (B) distribution along the main representative MA: moderate acidity, NN: near neutral, MAlk: moderate alkalinity, EM: extreme metallic,
points of the treatment. HM: high metallic, LM: low metallic. Data obtained from ve sampling campaigns.
 F. Macas et al. / Science of the Total Environment 433 (2012) 323330 327

Table 3
Bulk chemistry (mg/g) of the MgO-DAS lling substrate in the T3 reactive tank after
operation time.

Depth (cm) Zn Mg Ca Mn Fe S Ni Co Cd

06 212.51 61.90 38.97 11.01 6.49 4.02 0.49 0.50 0.21

612 0.98 275.11 24.61 1.12 11.42 3.25 0.14 0.01 n.d.
2030 0.14 263.10 26.59 0.67 10.44 3.41 0.11 0.01 n.d.
3040 0.11 270.36 26.82 0.69 10.62 3.11 0.10 0.02 n.d.
4050 0.13 251.98 24.35 0.67 9.23 3.19 0.10 0.01 n.d.
5060 0.17 259.20 26.34 0.64 10.32 3.68 0.11 0.02 n.d.
6070 0.27 230.57 21.91 0.61 9.18 3.44 0.10 0.02 n.d.
7080 0.26 244.12 24.62 0.62 9.10 3.64 0.11 0.01 n.d.
Initial n.d. 276.41 45.81 0.74 11.18 10.97 0.12 0.02 n.d.

n.d.: not detected.

the unreacted grains of the original substrate, and lling the voids
left by the vegetal cells. These SEM images correspond to some sam-
Fig. 4. Zn, Mg, Ca, Mn, SO4, Cd, Co, Ni and pH hydrochemical depth prole in the ples obtained in the rst 6 cm of the MgO-DAS in T3. From 6 cm
MgO-DAS reactive tank. depth to the bottom, the tank essentially contains unreacted MgO-
DAS as can clearly be deduced if its chemical composition is com-
It is only after the MgO treatment when the water evolves to moder- pared with the one corresponding to the initial substrate (Table 3).
ate alkalinity-low metallic. The high chemical quality obtained in the According to these observations, T3 lifetime with an efcient perfor-
waters at the output of the system is offered in Table 2. As shown, mance could have been maintained during a much longer time if
when water chemistry after the treatment is compared with the the mine shaft had not dried out.
limit values established for metal and toxic element concentration The XRD pattern of the original substrate shows magnesite, fer-
in irrigation (Ayers and Westcot, 1994) and drinking waters (WHO, roan magnesite, periclase, dolomite and portlandite as the main min-
2011), all the analyzed elements easily fulll these recommendations. erals detected. Gypsum was the only crystalline neoformed phase
detected in the rst 6 cm of the reactive material. Although Zn and
3.2. Detailed water chemistry and mineralogy of the MgO-DAS reactive Mn are the main constituents, on the basis of the bulk chemistry
tank (Table 3), no Zn and/or Mn minerals could be precisely detected by
XRD. The poor crystallinity and/or the small size of the crystallites
The inowing drainage in T3 reactive tank showed a pH of 6.5 of some neoformed phases limit the use of conventional XRD on
(Figs. 2 and 4), and after MgO powder dissolution this pH was those environments. From 10 cm to the bottom of the tank, brucite
increased to 10 in the rst 10 cm of the tank prole, maintaining was detected as the only neoformed mineral within the tank, due to
this value to the bottom of the reactive substrate (Fig. 4). both periclase hydration and brucite precipitation.
Slightly under the surface of the reactive material (1 cm depth), EPMA analyses of more than 40 areas of some selected Zn-rich
the pH value is increased to 8.6, promoting a concomitant decrease precipitates show a mean value of 76.22% Zn, with minor contents
in Zn, Co and Ni concentrations by 200 mg/L, 100 g/L and 180 g/L, of Si, S, Mn, Ni and Co (Table 4). The slight Fe amount detected is
respectively. Manganese and Cd are not retained on the surface of probably inherited from ferroan magnesite in the initial substrate
the substrate. The pH value at 5 cm depth is 9.7, leading to a reduc- due to the absence of this element in the inowing water. The sum
tion in Zn concentration to a value of 40 mg/L from the initial of the main constituents (named as total in Table 4) quantied by
430 mg/L in the inowing drainage, as well as for Co and Ni that are EPMA presents an average of 84.88%. The chemical composition
reduced from 850 and 800 g/L to 330 and 220 g/L, respectively. of those Zn-rich precipitates is reliably close to the expected one for
However, Mn and Cd concentrations do not show an important de- hydroxy-carbonates of Zn (hydrozincite type) or Zn-hydroxides.
crease. Finally, at 10 cm depth the pH is stabilized close to a value Rtting et al. (2008) also identied these neoformed mineral phases
of 10, promoting that Zn, Mn, Cd, Co and Ni are abated below detec- in a column experiment treating Zn- and Mn-rich solutions using
tion levels (Fig. 4). This prole evolution of divalent metal concentra- MgO-DAS.
tions is consistent with the pH-dependent variations of those metal In addition to the data presented in this study, a synchrotron
concentrations proposed by Cortina et al. (2003). light-induced micro analysis (combined -XRD and -XRF) has been
Concerning the mineralogy of the neoformed phases within T3, recently performed to these precipitates. Hydrozincite [Zn5(CO3)2
Fig. 5 shows a detailed view of Zn-rich precipitates distribution in (OH)6] and a minor amount of loseyite [(Mn,Zn)7(CO3)2(OH)10] have
the form of massive aggregates surrounding the wood shavings and been the only phases reliably identied as sinks for Zn and Mn, as

Fig. 5. SEM images of the Zn-rich precipitates developed within the MgO-DAS reactive tank.
328 F. Macas et al. / Science of the Total Environment 433 (2012) 323330

Table 4
EPMA analysis (wt.%) of the neoformed Zn-rich phases in the rst 6 cm of depth in the T3 reactive tank. A basic statistic (mean, maximum, minimum and standard deviation for 46
analysis points) is shown.

ZnO (0.08) SiO2 (0.03) MgO (0.02) SO3 (0.03) MnO (0.08) CaO (0.02) NiO (0.08) FeO (0.05) CoO (0.08) Total

Mean 76.22 3.72 2.51 0.92 0.60 0.48 0.17 0.16 0.14 84.88
Max 80.48 5.46 4.66 1,75 1.68 0.86 0.28 0.28 0.24 88.32
Min 68.82 1.51 1.43 0.60 0.26 0.36 0.13 0.10 0.10 79.48
Std. dev. 2.35 0.82 0.78 0.19 0.28 0.08 0.03 0.05 0.03 1.98

Detection limits are presented below each element.

well as for minor amounts of Co and Ni (Prez-Lpez et al., 2011). Even tank. Hydrozincite (Zn5(CO3)2(OH)6) and -Zn(OH)2 were set as
though no Mn3+,4+-rich mineral could be identied, oxidation of Mn2+ potential Zn-sinks, and manganite (-MnOOH) as possible Mn-sink.
to Mn3+ at high pH (Morgan et al., 1967) and subsequent manganite The autocatalytic oxidation of Mn 2+ to Mn 3+ was modeled after
(-MnOOH) cannot be discarded. the kinetic law of Morgan et al. (1967), as cited in Diem and Stumm
(1984):
3.3. Chemical and mineralogical control on MgO-DAS performance and
h i
time evolution d Mn
2
h i h i
2
k1 O2aq OH  Mn
2

In the light of the hydrochemical and mineralogical data acquired

dt h i h i
2
k2 O2aq OH  MnOx  Mn
2
3
in this study, it was decided to obtain a reactive transport model to
fully understand the different processes controlling the MgO-DAS
performance and also to estimate the lifetime that the reactive sub- where [Mn 2+], [O2(aq)] and [OH ] are dissolved concentrations (mol/L)
strate would have if the shaft had not dried out. of Mn 2+, oxygen and OH , respectively, [MnOx] is the concentration
According to the calculations, brucite is slightly supersaturated of Mn +3 reaction solid product (mol/L), k1 is the rate constant for
below 10 cm depth in the tank (SI = 0.20.6), and pH is lightly higher homogeneous oxidation (4 10 12 mol 3 L 3 day 1) and k2 is the rate
than expected for brucite control. As Ca was observed to increase constant for heterogeneous oxidation (10 18 mol 4 L 4 day 1).
with depth in the T3 tank (Fig. 4) and lime (CaO) was detected as a All the phases were allowed to precipitate in equilibrium once
mineral constituent of the initial substrate, it was hypothesized that saturation is reached. The input solution was set to be in equilibrium
some hydrated lime (Ca(OH)2) could be dissolving within the sub- with atmospheric O2.
strate, even though equilibrium was not reached. Moreover, gypsum The values of the initial reactive surface areas of dissolving
(another neoformed phase detected by XRD within T3) was also, to Mg(OH)2 and Ca(OH)2 were calibrated to match the measured depth
some extent, supersaturated throughout the tank, so that gypsum proles of Mg and Ca concentrations, and that of gypsum was adjusted
precipitation could explain the observed decrease of sulfate concen- to reproduce sulfate proles (Table 5).
trations with depth (Fig. 4). Rtting et al. (2008) had also identied Using these parameters, the rest of pore-water concentrations, pH
gypsum precipitated in laboratory MgO-DAS columns treating water and the dissolved and precipitated solids calculated with the model
with a similar composition as the T3 inow. Therefore, the possibility match well with observations along the depth prole of the tank
of brucite dissolution/precipitation, hydrated lime dissolution and (Fig. 6A). Dissolved Mg concentration steeply increases in the rst
gypsum precipitation was incorporated into the reactive transport 4 cm of the substrate due to Mg(OH)2 dissolution. Calcium concentra-
model (Table 5). tion increases more slowly due to Ca(OH)2 dissolution, causing pH to
The kinetic rate law for the dissolution of brucite was derived from increase above Mg(OH)2 saturation below 4 cm depth. Therefore,
data of various authors summarized by Pokrovsky and Schott (2004). dissolved Mg decreases below this depth due to Mg(OH)2 repre-
The dissolution rates shown in g. 8 of their publication were approx- cipitation. The pH increase also causes Zn precipitation, as hydro-
imated by the law: zincite immediately at the substrate surface and as -Zn(OH)2 up to
approximately 3 cm depth (Fig. 6B). Concentration of Mn 2+ starts to
4 0:45 2 1
r 10 aH mol m s 1 decrease below the depth of the Zn-precipitation front, where pH is
high enough to sufciently accelerate Mn 2+-oxidation. The resulting
The amount m of mineral dissolved (mol/s) was calculated as: Mn 3+ immediately precipitates as manganite (Fig. 6B). As manganite
has a low solubility at both alkaline and circumneutral pH (Eary,
m A r 1 2 1999), it does not redissolve even though the Mg(OH)2-dissolution
front moves downwards with time. Due to Ca increase, gypsum
where A is the reactive surface area (m 2/m 3 of porous medium) and precipitates along the depth prole of the tank, and dissolved sulfate
is the saturation of the solution. decreases accordingly.
A nite-element mesh with 410 linear elements and a total length The numerical model was also used to extrapolate the lifetime
of 80 cm was used to represent the depth prole of the T3 MgO-DAS of the MgO tank. Assuming uniform ow, the MgO-DAS in the tank

Table 5
Kinetic-equilibrium reactions and initial reactive surface of the MgO-DAS used in the reactive transport model.

Mineral Initial volume fraction Initial surface area Reaction log K Type of reaction
(m2 of mineral per m3 (m2 mineral per m3
of porous medium) of porous medium)

Brucite 0.2 100,000 Mg(OH)2 + 2 H+ 2 Mg2+ + 2 H2O 17.2 Kinetic

Portlandite 0.012 2.00E + 08 Ca(OH)2 + 2 H+ 2 Ca2+ + 2 H2O 22.8 Kinetic
Gypsum 0 1.00E 12 CaSO4 2H2O Ca2+ + SO42 + 2 H2O 4.58 Kinetic
Hydrozincite 0 10 Zn5(CO3)2(OH)6 + 6 H+ 5 Zn2+ + 2 CO32 + 6 H2O 9.69 Equilibrium
-Zn(OH)2 0 10 -Zn(OH)2 + 2 H+ 2 Zn2+ + 2 H2O 11.75 Equilibrium
Manganite 0 10 -MnOOH + 3 H+ Mn3+ + 2 H2O 0.17 Equilibrium
 F. Macas et al. / Science of the Total Environment 433 (2012) 323330 329

Fig. 6. Depth proles for Zn, Mn, Mg, Ca, and SO4 (mg/L) concentrations and pH (A), and mineral mass change (mol/kg) for brucite, gypsum, manganite, hydrozincite, -Zn(OH)2
and portlandite (B). Symbols indicate experimental data (real) and lines represent data obtained in the reactive transport modeling (model).

(1 m 3) would have been able to treat the Monte Romero AMD at a The widespread use of these passive plants in specic acid dis-
ow rate of 0.5 L/min during approximately 3.4 years before Mn or charges could improve in a few years the ecological and chemical
Zn breakthrough at the tank outlet. However, this modeled hypothe- quality of the Odiel basin. The pilot plant design tested in Monte
sis needs to be corroborated in longer eld trials. Romero abandoned mine is being considered by the regional author-
ities for full scale application at this AMD discharge. The same
NFOL+DAS scheme is proposed to treat other acid discharges in the
4. Conclusions Odiel basin.

The application of the natural pretreatment NFOL to oxidize Fe, Acknowledgments

coupled with limestone-DAS for trivalent metal retention (Fe and Al
mainly) and nally MgO-DAS for Zn, Mn, Cd, Co and Ni elimination, This research was nanced by the Spanish Ministry of Education
has shown to be capable to achieve the complete metal removal and Science through project CGL2010-21956-C02. F. Macas was -
from a highly metal-polluted AMD. nancially supported by the Spanish Government with a FPI PhD
The reactive transport model extrapolations indicate that the 1 m 3 fellowship. The assistance of Cristbal Cantero for EPMA analysis at
MgO-DAS tank is able to treat the Monte Romero AMD with a total the Central Research Services of the University of Huelva is gratefully
acidity of 788 mg/L of CaCO3 at a ow rate of 0.5 L/min during ap- acknowledged. R. Prez-Lpez also thanks the Spanish Minister of
proximately 3.4 years before Mn or Zn breakthrough the tank outlet. Science and Innovation and the Ramn y Cajal Subprogramme
330 F. Macas et al. / Science of the Total Environment 433 (2012) 323330

(MICINN-RYC 2011). Manuel A. Caraballo was nancially supported Nocete F, lex E, Nieto JM, Sez R, Bayona MR. An archaeological approach to regional
environmental pollution in the south-western Iberian Peninsula related to third
by the Spanish Ministry of Education and the Post-doctoral Interna- millennium BC mining and metallurgy. J Archaeol Sci 2005;32:156676.
tional Mobility Subprogramme I+D+i 20082011. The comments Olas M, Cnovas CR, Nieto JM, Sarmiento AM. Evaluation of the dissolved contaminant
and helpful criticisms of an anonymous reviewer and the support load transported by the Tinto and Odiel rivers (South West Spain). Appl Geochem
2006;21:173349.
of the Editors, Dr. Dami Barcel and Dr. Filip M.G. Tack, have consid- Olas M, Nieto JM, Sarmiento AM, Cern JC, Cnovas CR. Seasonal water quality varia-
erably improved the original manuscript and are also gratefully tions in a river affected by acid mine drainage: the Odiel River (South West
acknowledged. Spain). Sci Total Environ 2004;333(1-3):26781.
Parkhurst D, Appelo CAJ. User's guide to PHREEQC (version 2) a computer program for
speciation, batch-reaction, one-dimensional transport, and inverse geochemical
References calculations. USGS Water-Resources Investigations Report 99-4259. Denver, Colo-
rado; 1999.
Ayers RS, Westcot DW. Water quality for agriculture. FAO irrigation and drainage paper Prez-Lpez R, Macas F, Caraballo MA, Nieto JM, Romn-Ross G, Tucoulou R. Mineralogy
29. Rev. 1. Rome, Italy: Food and Agriculture Organization of the United Nations; and geochemistry of Zn-rich mine-drainage precipitates from an MgO passive treat-
1994. ment system by synchrotron-based X-ray analysis. Environ Sci Technol 2011;45:
Ball JW, Nordstrom DK. User's manual for WATEQ4F with revised thermodynamic 782633.
database and test cases for calculating speciation of major, trace and redox elements PIRAMID-Consortium. Engineering guidelines for the passive remediation of acidic and/or
in natural waters. U.S. Geological Survey, water-resources investigation report; 1991. metalliferous mine drainage and similar wastewaters. Newcastle Upon Tyne, UK:
p. 91-183. University of Newcastle; 2003.
Behum PT, Lefticariu L, Bender KS, Segid YT, Burns AS, Pugh CW. Remediation of Pokrovsky OS, Schott J. Experimental study of brucite dissolution and precipitation in aque-
coal-mine drainage by a sulfate-reducing bioreactor: a case study from the Illinois ous solutions: surface speciation and chemical afnity control. Geochim Cosmochim Ac
coal basin USA. Appl Geochem 2011;26:1626. 2004;68(1):3145.
Bigham JM, Nordstrom DK. Sulphate minerals: crystallography, geochemistry, and Ros CA, Williams CD, Roberts CL. Removal of heavy metals from acid mine drainage
environmental signicance. Washington: Mineralogical Society of America; 2000. (AMD) using coal y ash, natural clinker and synthetic zeolites. J Hazard Mater
BOJA. Acuerdo de 2 de noviembre de 2011, del Consejo de Gobierno, por el que se 2008;156:2335.
aprueba inicialmente el Plan Hidrolgico de la Demarcacin Hidrogrca del Tinto, Rtting TS. Dispersed Alkaline Substrate (DAS): A novel option for the passive treatment
Odiel y Piedras. BOJA 2011;216:9-11. of waters with high metal concentrations. Tesis Doctoral, Instituto de Ciencias de la
Cabrera G, Prez R, Gmez JM, balos A, Cantero D. Toxic effects of dissolved heavy Tierra Jaume Almera, CSIC; 2007.
metals on Desulfovibrio vulgaris and Desulfovibrio sp. strains. J Hazard Mater Rtting TS, Cama J, Ayora C, Cortina JL, De Pablo J. Use of caustic magnesia to remove
2006;135:406. cadmium, nickel and cobalt from water in passive treatment systems: column ex-
Caraballo MA, Rtting TS, Nieto JM, Ayora C. Sequencial extraction and DXRD applica- periments. Environ Sci Technol 2006;40:643843.
bility to poorly crystalline Fe- and Al-phase characterization from an acid mine Rtting TS, Ayora C, Carrera J. Improved passive treatment of high Zn and Mn concen-
water passive remediation system. Am Mineral 2009a;94:102938. trations using caustic magnesia (MgO): particle size effects. Environ Sci Technol
Caraballo MA, Rtting TS, Macas F, Nieto JM, Ayora C. Field multi-step calcite and MgO 2008;42:93707.
passive system to treat acid mine drainage with high metal concentration. Appl Saaltink MW, Battle F, Ayora C, Carrera J, Olivella S. RETRASO, a code for modelling re-
Geochem 2009b;24:230111. active transport in saturated and unsaturated porous media. Geol Acta 2004;3:
Caraballo MA, Rtting TS, Silva V. Implementation of an MgO-based metal removal step in 23551.
the passive treatment system of Shilbottle, UK: column experiments. J Hazard Mater Sez R, Pascual E, Toscano M, Almodvar GR. The Iberian type of volcano-sedimentary
2010;181:92330. massive sulphide deposits. Miner Deposita 1999;34:54970.
Caraballo MA, Macas F, Castillo J, Quispe D, Nieto JM, Ayora C. Hydrochemical performance Sanchez Espaa J, Trevor ML. The behaviour of iron and aluminum in acid mine drain-
and mineralogical evolution of a dispersed alkaline substrate (DAS) remediating the age: speciation, mineralogy, and environmental signicance. Thermodynamics,
highly polluted acid mine drainage in the full scale passive treatment of Mina Esperanza solubility and environmental issues. Amsterdam: Elsevier; 2007. p. 13750.
(SW, Spain). Am Mineral 2011a;96:12707. Sarmiento AM, Nieto JM, Casiot C, Elbaz-Poulichet F, Egal M. Inorganic arsenic specia-
Caraballo MA, Macas F, Rtting TS, Nieto JM, Ayora C. Long term remediation of highly tion at river basin scales: the Tinto and Odiel rivers in the Iberian Pyrite Belt, SW
polluted acid mine drainage: a sustainable approach to restore the environmental Spain. Environ Pollut 2009a;157:12029.
quality of the Odiel river basin. Env Poll 2011b;159:36139. Sarmiento AM, Nieto JM, Olas M, Cnovas CR. Hydrochemical characteristics and sea-
Cortina JL, Lagreca I, De Pablo J, Cama J, Ayora C. Passive in situ remediation of sonal inuence on the pollution by acid mine drainage in the Odiel river Basin (SW
metal-polluted water with caustic magnesia: evidence from column experiments. Spain). Appl Geochem 2009b;24:697714.
Environ Sci Technol 2003;37:19717. Sarmiento AM, DelValls A, Nieto JM, Salamanca MJ, Caraballo MA. Toxicity and poten-
Diem D, Stumm W. Is dissolved Mn2+ being oxidized by O2 in absence of Mn-bacteria tial risk assessment of a river polluted by acid mine drainage in the Iberian Pyrite
or surface catalysts. Geochim Cosmochim Ac 1984;48(7):15713. Belt (SW Spain). Sci Total Environ 2011;409(22):476371.
Eary LE. Geochemical and equilibrium trends in mine pit lakes. Appl Geochem 1999;14(8): Schindler P, Reinert M, Gamsjger H. Thermodynamics of metal carbonates 3. Solubility
96387. constants and free formation enthalpies of ZnCO3 and Zn5(OH)6(CO3)2 at 25 de-
EC Decision 2000/60/EC. Council decision of 23 October 2000 establishing a communi- grees. Helv Chim Acta 1969;52(8):232732.
tarian frame of action in the scope of water policy. Off J 2000;L327:1. [22/12/2000]. Strosnider WH, Nairn RW. Effective passive treatment of high-strength acid mine
Ficklin WH, Plumlee GS, Smith KS, McHugh JB. Geochemical classication of mine drainage and raw municipal wastewater in Potos, Bolivia using simple mutual in-
drainages and natural drainages in mineralized areas. Proceedings of the 7th inter- cubations and limestone. J Geochem Explor 2010;105:3442.
national symposium on water rock interaction. Park City, Utah; 1992. p. 3814. Tyler G, Carrasco R, Nieto JM, Prez R, Ruiz MJ, Sarmiento AM. Optimization of major
Hedin RS, Watzlaf GR, Nairn RW. Passive treatment of acid-mine drainage with lime- and trace element determination in acid mine drainage samples by ultrasonic
stone. J Environ Qual 1994;23:133845. nebulizer-ICP-OES (USN-ICP-OES). Pittcon conf., Chicago, USA; 2004.
Jage CR, Zipper CE, Noble R. Factors affecting alkalinity generation by successive alkalinity- Watten BJ, Sibrell PL, Schwartz MF. Acid neutralization within limestone sand reactors
producing systems: regression analysis. J Environ Qual 2001;30:101522. receiving coal mine drainage. Env Poll 2005;137:295304.
Jarvis AP, Younger PL. Passive treatment of ferruginous mine waters using high surface WHO. Guidelines for drinking-water quality. 4th ed. World Health Organization; 2011.
area media. Water Res 2001;35(15):36438. http://www.who.int/water_sanitation_health/publications/2011/dwq_guidelines/
Johnson DB, Hallberg KB. Acid mine drainage remediation options: a review. Sci Total en/index.html [Last accessed January 17, 2011].
Environ 2005;338(1-2):3-14. Xiong Y. Thermodynamic properties of brucite determined by solubility studies and
Macas F, Caraballo MA, Nieto JM, Ayora C, Rotting TS. Iron removal enhancement of their signicance to nuclear waste isolation. Aquat Geochem 2008;14(3):22338.
a two step calcite passive treatment system at the Iberian Pyrite Belt. Geochim Younger PL. The longevity of mine water pollution: a basis for decision-making. Sci
Cosmochim Ac 2009;73(13):A811. Total Environ 1997;194195:45766.
Macas F, Caraballo MA, Nieto JM, Rtting TS, Ayora C. Natural pretreatment and passive Younger PL, Banwart SA, Hedin RS. Mine water hydrology, pollution, remediation.
remediation of highly polluted acid mine drainage. J Environ Manage 2012;104: Dordrecht; Kluwer Academic Publishers; 2002.
93-100.
Morgan JJ, Faust SD, Hunter JV. Chemical equilibria and kinetic properties of manganese in
natural waters. Principles and applications of water chemistry. New York: Wiley;
1967. p. 561624.