1741

Passive treatment systems based on the dissolution of coarse calcite grains are widely used to remediate acid mine drainage (AMD). Unfortunately, they tolerate only low metal concentrations or acidity loads, because they are prone to passivation (loss of reactivity due to coating) and/or clogging (loss of permeability) by precipitates. To overcome these problems, a dispersed alkaline substrate (DAS) composed of a fi ne-grained alkaline reagent (calcite sand) mixed with a coarse inert matrix (wood chips) was developed. Th e small grains provide a large reactive surface and dissolve almost completely before the growing layer of precipitates passivates the substrate, whereas the dispersion of nuclei for precipitation on the inert surfaces retards clogging. Chemical and hydraulic performance of DAS was investigated in two laboratory columns fed at diff erent fl ow rates with natural AMD of pH 2.3 to 3.5 and infl ow net acidity 1350 to 2300 mg/L as CaCO

3. Th e DAS

columns removed 900 to 1600 mg/L net acidity, 3 to 4.5 times more than conventional passive treatment systems. Regardless of the fl ow rate employed, Al, Fe(III), Cu, and Pb were virtually eliminated. Minor Zn, Ni, and Cd were removed at low fl ow rates. High acidity removal is possible because these metals accumulate intentionally in DAS, and their precipitation promotes further calcite dissolution. During 15 mo, DAS operated without clogging at 120 g acidity/m2·d, four times the loading rate recommended for conventional passive systems; DAS may therefore be capable of treating AMD at sites where infl uent chemistry precludes the use of other passive systems.

Passive Treatment of Acid Mine Drainage with High Metal Concentrations

Using Dispersed Alkaline Substrate

Tobias S. Rötting* Newcastle University

Robert C. Thomas CH2M

Carlos Ayora and Jesús Carrera CSIC

Acid mine drainage (AMD) generated by the oxidation of

pyrite and other metal sulfi des is a major cause of water

contamination worldwide (e.g., Kaufmann et al., 1992; Banks

et al., 1997). Acid mine drainage contains high concentrations

of sulfuric acid and metals. Release of contaminants can persist

for decades or centuries after mining has stopped (e.g., Demchak

et al., 2004). Th e problem is particularly severe in metal mining

areas, such as the Iberian Pyrite Belt (SW Spain), the largest pyrite

mining district in the world (Olias et al., 2004). Concentrations

of Al, Fe, Zn, Mn, Cu are observed in hundreds of mg/L with net

acidity several thousands of mg/L as CaCO3 equivalents.

Passive treatment systems may be an economic option for reme-

diation of AMD (e.g., Johnson and Hallberg, 2005). Once installed,

these systems do not need continuous addition of reagents and they

only require infrequent (albeit regular) maintenance. Th is technol-

ogy has been applied successfully at sites with moderate acidity and

metal concentrations (e.g., Ziemkiewicz et al., 2003), primarily at

coal mines with low pyrite contents and in zones of humid climate.

Th e most common passive treatment systems for AMD are

“Anoxic Limestone Drains” (ALD) and “Reducing and Alkalin-

ity Producing Systems” (RAPS). In these systems, the dissolution

of calcite gravel and/or bacterial sulfate reduction produce up

to 300 mg/L (ALD; e.g., Hedin et al., 1994; Cravotta, 2003) or

450 mg/L (RAPS; e.g., Jage et al., 2001; Watzlaf et al., 2004)

alkalinity as CaCO3. However, when exposed to elevated metal

concentrations, ALD and RAPS tend to lose reactivity (“pas-

sivation” or “armoring”) or permeability (“clogging”) due to the

formation of precipitates on the surface of the limestone grains

or in the pore space of the system (e.g., Rees et al., 2001; Watzlaf

et al., 2002, 2004; Lopez Fernandez et al., 2003; Ziemkiewicz et

al., 2003; Rose et al., 2004; Simon et al., 2005). Usually only a

very small fraction of the applied limestone gravel has been con-

sumed when clogging or passivation occurs.

Abbreviations: ALD, Anoxic Limestone Drains; AMD, acid mine drainage; DAS, dispersed

alkaline substrate; EDS, energy dispersive system; ICP-AES, inductively coupled plasma

atomic emission spectroscopy; ICP-MS, inductively coupled plasma mass spectroscopy;

LBOS, limestone-buff ered organic substrate; RAPS, Reducing and Alkalinity Producing

Systems; SEM, scanning electron microscope; TIC, total inorganic carbon.

T.S. Rötting, Hydrogeochemical Engineering Res. and Outreach Group, Sir Joseph Swan

Inst. for Energy Res., 3rd Floor, Devonshire Bldg., Devonshire Terrace, Newcastle Univ.,

Newcastle on Tyne, NE1 7RU, UK; C. Ayora, J. Carrera, (and formerly T.S. Rötting), Inst. of

Earth Sciences Jaume Almera, CSIC, Lluis Solé i Sabarís s/n, 08028 Barcelona, Spain; R.C.

Thomas, CH2M Hill, Northpark 400, 1000 Abernathy Rd., Suite 1600, Atlanta, GA 30328.

Copyright © 2008 by the American Society of Agronomy, Crop Science

Society of America, and Soil Science Society of America. All rights

reserved. No part of this periodical may be reproduced or transmitted

in any form or by any means, electronic or mechanical, including pho-

tocopying, recording, or any information storage and retrieval system,

without permission in writing from the publisher.

Published in J. Environ. Qual. 37:1741–1751 (2008).

doi:10.2134/jeq2007.0517

Received 1 Oct. 2007.

*Corresponding author (Tobias.Roetting@newcastle.ac.uk).

© ASA, CSSA, SSSA

677 S. Segoe Rd., Madison, WI 53711 USA

TECHNICAL REPORTS: HEAVY METALS IN THE ENVIRONMENT

1742 Journal of Environmental Quality • Volume 37 • September–October 2008

Another shortcoming of conventional passive treatment

systems is their high land demand. Design guidelines for RAPS

recommend an acidity load of less than 30 g acidity/m2⋅d (Wat-

zlaf et al., 2004). Due to this constraint, RAPS may need very

large areas to treat high-acidity AMD when fl ow rate is also high.

For example, 1 L/s AMD with 2 g/L acidity (a common value

in the Iberian Pyrite Belt) would require >5700 m2 of RAPS

surface. Furthermore, Zn, Cu, Ni, and other heavy metals can

be toxic to sulfate-reducing bacteria at levels found in AMD in

the Iberian Pyrite Belt (e.g., Utgikar et al., 2003; Cabrera et al.,

2006; Neculita et al., 2007). Th is may inhibit the passive treat-

ment of water with high heavy metal concentrations by RAPS.

Moreover, ALD and RAPS are designed to produce alkalinity

but not to precipitate metals inside the reactor. Th e generated

alkalinity is consumed outside the reactor to oxidize Fe(II) and

precipitate Fe(III)-solids in subsequent sedimentation ponds.

Th e ponds require additional land area, which may exceed that

occupied by the RAPS or ALD itself. Mines discharging AMD

are often located in steep, narrow valleys, so that construction of

passive systems may be restricted by the availability of fl at areas.

Th ese constraints limit the use of ALD and RAPS for treatment

of AMD with high metal concentrations.

Several authors (Watzlaf, 1997; Ziemkiewicz et al., 1997; Cra-

votta and Trahan, 1999) observed that limestone may still dis-

solve despite armoring with Al- or Fe-precipitates. Based on these

results, Th omas and Romanek (2002a, 2002b) developed a novel

limestone-buff ered organic substrate (LBOS) to treat acidic, fer-

ric iron- and aluminum-dominated water (142 mg/L Fe, >99%

Fe(III), 84 mg/L Al, pH 2.4). Th e LBOS consists of a reducing

organic substrate (mainly wood chips and horse manure) mixed

with an additional 25% (v/v) of limestone sand. Th e LBOS tanks

eliminated 90% of infl owing Fe and depleted Al to <0.2 mg/L.

Precipitation of Al- and Fe-compounds within the LBOS did not

negatively aff ect either reactivity or permeability. However, LBOS

was not tested on water with high ferrous iron concentrations,

which is most common at abandoned mines.

In the present study, the dispersed alkaline substrate (DAS)

was developed to treat water with high concentrations of Fe(II),

Fe(III), and Al. It consists of a fi ne-grained alkaline reagent mixed

with an inert, coarse high-surface material, so that the surfaces

of the inert matrix are partially covered with the reactive sub-

stance. No reducing organic matter is used in DAS to maximize

Fe(II)-oxidation and Fe(III)-hydrolysis and precipitation. Th e

small grain size of the alkaline material provides a greater reactive

surface than the same amount of coarse material. Th is should

increase reactivity. Also, a larger fraction of each grain should dis-

solve before the crust of precipitates becomes too thick and im-

pedes further dissolution. Th is should improve the effi cient use

of the alkaline material. Th e coarse particles of the inert material

provide large pores and high permeability and separate individual

limestone grains, so that precipitates do not fi ll the entire pore

space between grains. Th is should retard clogging problems.

Th e present study investigates the chemical and hydraulic

eff ectiveness of the proposed reactive substrate in laboratory

column tests. Th e experimental results were reproduced using

a numerical reactive transport model to explain the geochemi-

cal metal elimination processes.

Materials and MethodsTh e DAS used in this study consists of fresh wood chips

as an inert high surface media mixed with limestone sand as

alkaline material. It is used in down-fl ow reactors with super-

natant open to the atmosphere to maximize Fe oxidation and

minimize Fe(II) mobility in the reactor.

ColumnsTwo columns were fabricated from transparent polymethyl

methacrylate (9.6 cm i.d., height 35 cm) and equipped with

lateral ports for taking 0.1-μm fi ltered water samples (Macro

Rhizon samplers; Eijkelkamp, Giesbeek, the Netherlands) and

measuring pressure (Fig. 1). Both columns contained a perfo-

rated drain pipe and a 2.5-cm layer of quartz gravel (5–8 mm

diam.) at the bottom.

Th e DAS used to fi ll the columns was composed of 25% (v/v)

calcite sand (d10

= 0.3 mm, d50

= 1.2 mm, d100

= 5 mm) and 75%

(v/v) wood chips (maximum length about 2 cm, thickness up to

1 mm). To mix the DAS, fi rst the wood chips were wetted slight-

ly with deionized water, and then the calcite sand was added and

mixed. As a result of this procedure, the calcite particles formed

a thin layer on the surface of the wood chips and adhered suffi -

Fig. 1. Schematic design of laboratory columns.

Table 1. Mean composition of the input acid mine drainage from Monte Romero Mine during the experiment.

Parameter Value/concentration

pH 2.8

mg/L

Al 106

Ca 268

Cd 0.38

Co 1.0

Cu 3.3

Fe(II) 121

Fe(III) 129

Mg 284

Mn 22

Ni 0.84

Pb 0.15

Sulfate 3510

Si 40

Zn 365

Total acidity (as CaCO3) 1727

Rötting et al.: Passive Treatment of Acid Mine Drainage using DAS 1743

ciently to minimize segregation while packing the columns. Both

columns were then slowly saturated with water from the bottom

to expulse air trapped in the substrate.

During the experiment, AMD from Monte Romero Mine

(Huelva province, SW Spain) was fed to the supernatant of

each column and fl owed out of the overfl ow pipe into an

output container that acted as an oxidation and sedimenta-

tion tank. Flow rate was controlled by a peristaltic pump

(Minipuls 3; Gilson, Middleton, WI). Supernatant water level

was controlled by adjusting the height of the overfl ow pipe.

Th e mean composition of the AMD in the input container

feeding the columns is shown in Table 1. Even though on col-

lection in the fi eld the AMD contained 87% Fe(II), and the

solution in the input container was changed frequently, some

oxidation occurred during storage.

Th e columns were operated at diff erent fl ow rates. Th e

high-fl ow column started to operate 15 wk before the low-fl ow

column. Th e high-fl ow column had 18 cm of substrate and 10

cm of supernatant, and ports for water and pressure at 5-cm

intervals (3-, 8-, and 13-cm depth). Th e low-fl ow column had

14 cm of substrate and 7 cm of supernatant, three ports for wa-

ter and pressure at 4-, 9-, and 14-cm depth, and an additional

pressure measuring port at 1.5-cm depth. Th e mean fl ow rate

of the high-fl ow column was 0.47 L/d or 0.067 m3/m2⋅d, that

of the low-fl ow column was 0.14 L/d or 0.021 m3/m2⋅d. Th e

fl ow rates of both columns were changed several times to inves-

tigate the eff ect of diff erent residence times and contaminant

loadings on treatment effi ciency. After each change, conditions

were maintained during at least 4 wk to allow the columns to

respond. Several water samples were taken during each period

to verify that a steady state had been reached.

Th e eff ect of higher Fe(II) concentration and net acidity

on the columns was studied by adding 200 mg/L of Fe(II) (as

FeSO4⋅7H

2O) to the input AMD during Weeks 6 to 13 and

34 to 38 of operation the high-fl ow column and Weeks 19 to

23 of the low-fl ow column.

Because the water had to be shipped >1000 km from the

fi eld site to our laboratory, the columns were not replicated.

Replication would have required transporting twice the

amount of water, which we considered prohibitive. Never-

theless, the two columns are to some extent replicates: Th e

substrate of both columns was mixed and packed following

the same procedure, input water was taken from the same

container during the experiment, and the fl ow rate ranges of

the two columns overlap. In the Results and Discussion sec-

tion, performance of both columns at similar fl ow rates will

be compared to check consistency of the results.

Water Sampling and AnalysisWater samples were collected from the input container, the

supernatant water, the drain pipe, and from the output container

on an approximately weekly basis. Samples from the intermedi-

ate sampling ports were taken monthly. pH was measured using

a Crison glass electrode calibrated with buff er solutions of pH

7 and 2. Redox potential was measured using a Pt combina-

tion electrode (SureFlow; Th ermo Fisher Scientifi c, Waltham,

MA) that was calibrated with standard buff er solutions of 220

and 468 mV. Measurements were corrected to the Standard

Hydrogen Electrode (Eh) to calculate pe. Total (gross) alkalinity

was measured using disposable hand-held titration cells (Titrets;

CHEMetrics, Calverton, VA; range 10–100 or 100–1000 mg/L,

accuracy approximately 5%). In the following, “alkalinity” always

refers to measured gross alkalinity. Filtered samples (0.1 μm) were

acidifi ed with HCl for analysis of ferrous and total Fe by spec-

trophotometer (Stookey, 1970; Fe(III) calculated as diff erence)

and with HNO3 for analysis of major and trace elements. Major

cations (Ca, Mg, Zn, Fe, Mn, Si) and total S were measured by

inductively coupled plasma–atomic emission spectroscopy (ICP-

AES) (Optima 3200 RL; PerkinElmer, Waltham, MA), trace

metals (Ni, Cd, Co, Pb) by inductively coupled plasma–mass

spectroscopy (ICP-MS) (Sciex Elan 6000; PerkinElmer). Copper

and Al were measured with both instruments due to the broad

range of their concentrations. Detection limits were 0.1 mg/L

for S; 0.05 mg/L for Ca, Mg, and Si; 0.02 mg/L for Fe, Zn, and

Mn; 5 µg/L for Al; 1.5 µg/L for Cu and Ni; 0.5 µg/L for Pb; and

0.2 µg/L for Cd and Co. Samples were diluted 1:10 for ICP-AES

and 1:50 for ICP-MS using 1% HNO3. Assuming all S to be

sulfate, charge balance error was usually less than 5%.

Net acidity (mg/L as CaCO3 equivalents) was calculated

according to Kirby and Cravotta (2005) as:

Net acidity = 50,045 × (3cAl

+ 2cFe

+ 2cMn

+ 2cZn

+ 10−pH) − alk [1]

where cX are molar concentrations (mol/L) and alk is

measured gross alkalinity (mg/L as CaCO3 equivalents). Th e

factor 50,045 converts moles of charge into mg of CaCO3

equivalents. Net acidity may be positive (net acidic water) or

negative (net alkaline water). Note that we included Mn and

Zn into the formula due to the high concentrations of these

metals in the AMD treated in our study.

Absolute net acidity elimination was calculated as:

Net acidity elimination = Supernatant net acidity

− Drain pipe net acidity [2]

Relative metal removal r (%) from the supernatant to the

drain pipe or output container was calculated as:

sup out

sup

( )100

c cr

c−

= × [3]

where csup

is supernatant concentration (mg/L) and cout

is drain

pipe or output container concentration (mg/L).

Porosity MeasurementsPorosity of the fi lling material of the high-fl ow column was

determined close to the beginning and end of the experiment

by injecting distilled water with a known concentration of LiBr

tracer. Bromine was measured at the column outfl ow using a

Br−-electrode (Ionplus; Th ermo Fisher Scientifi c). Both satura-

tion and desaturation curves were measured. Initial porosity was

measured gravimetrically while saturating the packed low-fl ow

column with water. It was assumed to be similar to the initial

1744 Journal of Environmental Quality • Volume 37 • September–October 2008

porosity of the high-fl ow column, because both columns fol-

lowed the same packing procedure. On a similar column from

a diff erent experiment porosity was determined both gravimet-

rically and by tracer tests, and diff erences were smaller than 2%.

Hydraulic Conductivity MeasurementsHydraulic conductivity profi les of the substrates were de-

termined by connecting piezometric standpipes to every pres-

sure measuring port. Th en fl ow rate was temporarily raised

to about 1 to 2 cm3/s (≈86–173 L/d) to increase head-loss

across the column to easily measurable values, and the piezo-

metric head diff erences between adjacent ports were read. Th e

hydraulic conductivity K (cm/s) in the depth interval between

two pressure ports was calculated using Darcy’s Law:

QLKA h

=Δ [4]

where Q (mL/s) is fl ow rate, A (cm2) is tank cross-section

perpendicular to fl ow, and L and Δh (cm) are distance and

head-loss between pressure ports, respectively.

Solid SamplesSubstrate and precipitates of the columns were excavated

after the end of the experiments from the top in a glove box

under N2 atmosphere to prevent oxidation of reduced mineral

phases. Samples for microscopy were carefully transferred into

polycarbonate boxes to preserve structure. All samples were

sealed under N2 in LDPE bags, immediately frozen at −30°C,

and freeze dried within 1 wk of sampling.

Th in sections were prepared with epoxy resin (Epofi x; Stru-

ers, Westlake, OH), and polished with diamond paste from

3 µm down to 1 µm. Granular material and thin sections of

the samples were observed under an optical microscope with

refl ecting and transmitting light (LV100POL; Nikon, Tokyo,

Japan) and coupled digital camera (DXM1200F; Nikon), and

under a scanning electron microscope (SEM; model JSM840;

JEOL, Tokyo, Japan) with energy dispersive system (EDS;

model Link; Oxford Instruments, Abingdon, UK). Element

maps were recorded using a Stereoscan S-360 SEM (Leica

Cambridge, Cambridge, UK) with Oxford Link EDS.

Samples for chemical analyses and identifi cation of mineral

phases were ground in a tungsten carbide mill for 1 to 2 min.

Powder diff ractograms were measured using an X-ray diff rac-

tometer (XRD; model D5005; Bruker, Karlsruhe, Germany)

with Cu Lα radiation. Total inorganic carbon (TIC, essentially

the amount of carbonates) was determined by measuring the

volume of CO2 gas produced after addition of 16% HCl to a

known mass of ground solid sample. Chemical compositions

of the solid samples were determined by dissolution in 7.5 M

HNO3 and analysis by ICP-AES. Th e Fe/S and Al/S molar ra-

tios were calculated to obtain information on the nature of the

minerals that precipitate inside the columns. Sulfate concentra-

tion was corrected for sulfate bound in gypsum supposing that

all carbonate was from calcite, and that all Ca in excess of car-

bonate was bound as gypsum with an equal amount of S.

Geochemical ModelingSaturation indices of possible minerals with respect to analyzed

element concentrations were calculated using the code PHREEQC

(Parkhurst, 1995) and the WATEQ4F database (Ball and Nord-

strom, 1991). Th ermodynamic data for schwertmannite was added

according to Bigham et al. (1996). A 1-D reactive transport model

was created using the code RETRASO (Saaltink et al., 2004) to

test the hypotheses on Fe and Al removal.

Results and Discussion

PorosityInitial porosity, measured by gravimetry in the low-fl ow

column, was equal to 65%. A fi rst series of tracer tests was

conducted after 15 wk of operation in the high-fl ow column

(Fig. 2). Mean calculated porosity was 73%. After 63 wk of

operation, a second tracer tests series yielded a mean porosity

of 82%. Th is indicates that porosity of the substrate increased,

and that calcite dissolution created more pore space than was

consumed by formation of precipitates.

It is relevant to observe the increase in asymmetry of both

saturation and desaturation curves. Th is implies that fl ow is be-

coming increasingly channelized. Earliest breakthrough, which

is caused by well connected large pores, occurs fi rst in the

63- to 64-wk test, despite its increased porosity. On the other

hand, the tail becomes heavier and longer for the same test,

which suggests that a signifi cant fraction of pores are poorly

connected. Altogether, the tracer tests imply that the column is

evolving from a homogeneous to a heterogeneous medium.

Water SamplesInput and supernatant pH of both columns (Fig. 3) ranged

normally between 2.5 and 3. In the high-fl ow column, pH

ranged usually between 6.0 and 6.4 at the drain pipe and rose

to 6 to 7 in the output container due to CO2 degassing. Th e

high values of around pH 4 in the supernatant and 8.4 in the

output container were measured during and immediately after

the fi rst tracer test, respectively, which was conducted using

deionized water. In the low-fl ow column (Fig. 3) drain pipe

pH ranged between 6.2 and 6.7 throughout the experiment.

Output container pH rose to over pH 8 during the fi rst 10 wk,

probably due to dissolution of very fi ne calcite grains. After-

Fig. 2. (A) Saturation and (B) desaturation curves with calculated porosity values from the tracer tests conducted in the high-fl ow column near the beginning (15–18 wk) and end (63–64 wk) of the experiment.

Rötting et al.: Passive Treatment of Acid Mine Drainage using DAS 1745

ward, it remained near 7 at fl ow

rates between 0.1 and 0.3 L/d.

Drain pipe measured alkalinity

of the high-fl ow column (Fig. 3)

started at 330 mg/L, but declined

gradually during the fi rst 10 wk,

even though fl ow rate was rather

constant during that period. Af-

terward, drain pipe alkalinity var-

ied inversely to fl ow rate, but was

virtually independent of input net

acidity. In the low-fl ow column

(Fig. 3), drain pipe alkalinity var-

ied between 230 and 300 mg/L,

inversely correlated with fl ow rate.

Again, drain pipe alkalinity was

independent of infl ow net acidity.

Absolute net acidity elimina-

tion of both columns fl uctuated

between 900 and almost 1600

mg/L (mean 1150 mg/L for the

high-fl ow column and 1250

mg/L for the low-fl ow column).

Net acidity elimination was larg-

est when input acidity was maxi-

mum. Relative net acidity elimi-

nation of the high-fl ow column

varied between 61 and 88% with a mean value of about 70%.

Relative net acidity elimination was more stable in the low-

fl ow column, with values around 75% during the fi rst 55 wk

at fl ow rates between 0.1 and 0.3 L/d. During the last weeks

it rose to 100% at the very low fl ow rate of 0.02 L/d.

Depth profi les of the high-fl ow column (Fig. 4) show that

rise of pH and acid neutralization fi rst occurred at a sharp

front near the substrate surface. Later on, the acid neutraliza-

tion front moved downward through the substrate and pH

change became more gradual.

Dissolved Fe in the supernatant was mostly Fe(III), while

Fe inside the substrate and in the drain pipe was mostly

Fe(II). Most of dissolved Fe(III) was precipitated near the

substrate surface. Iron(II) concentration decreased by up to

70 mg/L from the supernatant to the drain pipe, probably

because of oxidation to Fe(III) and subsequent precipitation.

Some Fe(II) may also have precipitated as siderite, because

pore waters within the column were supersaturated with re-

spect to this mineral. Siderite precipitation from AMD in a

carbonate-rich environment was observed by Al et al. (2000).

Usually all Fe(II) which reached the drain pipe was precipi-

tated in the output container after oxidation.

Aluminum precipitated at greater depth than Fe(III). At

fi rst, most of infl owing Al was eliminated in the top 3 cm of the

substrate. However, after 28 wk Al concentrations at 3-cm depth

were higher than input concentrations and decreased consider-

ably at 8-cm depth. Th is indicates that the zone of Al accumula-

tion moved downward through the substrate and that previously

precipitated Al-phases redissolved around 3 cm and precipitated

again below. Some Al always reached the drain pipe, but elimina-

tion was nearly complete in the output container.

Silicon concentration decreased substantially in the same

depth intervals as Al, but only about half of the total Si was

eliminated. Lead had depth profi les similar to Al, showing

redissolution at 3 cm after Week 28 and massive precipitation

below. Lead concentrations in the output container were be-

low 4 μg/L throughout the study.

Zinc, Mn, Cd (Fig. 4, Fig. 5A), Ni, and Co (Fig. 5A) were

not signifi cantly removed in the high-fl ow column. Some Zn,

Ni, and Cd was precipitated in the output container (Fig.

5B), reaching maximum removal values at fl ow rates below

0.3 L/d when pH rose above 6.9 (Weeks 19–32 and 39–48).

A variable fraction of infl owing Cu was gradually eliminated

throughout the substrate. Most remaining Cu was precipi-

tated in the output container.

Calcium concentrations displayed similar trends as pH, show-

ing that acid neutralization was due to limestone dissolution. Cal-

cium increased most in the depth interval where Al and Fe removal

was highest, indicating that calcite dissolution was directly linked

to Al- and Fe-hydrolysis and precipitation. Despite long residence

times (usually between 1.3 and 7 d) pore waters did not reach

calcite saturation in the columns, except for the very last samples of

the low-fl ow column taken at 0.02 L/d (residence time ≈35 d).

Mean sulfate removal from supernatant to drain pipe was

327 mg/L (9%). Redox potential (Eh) at all sampling points

was always above +195 mV, and no other signs of bacterial

sulfate reduction such as H2S-smell or black coloration of

substrate were noticed. Sulfate removal was probably due to

Fig. 3. Evolution of (top) pH, and (bottom) input net acidity, drain pipe net acidity, measured drain pipe alkalinity and fl ow rate at diff erent sampling points of (left) the high-fl ow and (right) the low-fl ow column.

1746 Journal of Environmental Quality • Volume 37 • September–October 2008

precipitation of Al- and Fe-hydroxysulphates and gypsum.

Depth profi les in the low-fl ow column (not shown) were

similar to those of the high-fl ow column. Th e lower fl ow rate

caused the zones of Al- and Fe-precipitation to be narrower

and chemical gradients to be steeper than in the high-fl ow

column. In the low-fl ow column, pH at 4-cm depth was

above 6.0 during the fi rst 41 wk. Underneath, depth pH re-

mained above 6.2 during the entire monitoring period.

Supernatant Fe concentrations in the low-fl ow column (not

shown) were on average 70 mg/L lower than in the high-fl ow

column, even though both columns were fed from the same in-

put container. Due to the lower fl ow rate, residence time in the

supernatant was higher and a larg-

er fraction of Fe(II) was oxidized

to Fe(III) and precipitated before

entering the substrate. Th is shows

that the oxic supernatant is an es-

sential part of the DAS treatment

system when high Fe(II) concen-

trations are present in the input

water. Depth profi les of the other

studied elements were qualitatively

similar to those of the high-fl ow

column shown for Week 24 (when

its fl ow rate was only 0.18 L/d,

Fig. 4). Mean sulfate removal in

the low-fl ow column from super-

natant to drain pipe was 477 mg/L

(13%). Again, no signs of bacterial

sulfate reduction were observed.

Relative metal removal rates (Fig.

5C, 5D), especially of Cd, Zn, Cu,

and Ni, were slightly higher in the

low-fl ow column compared to the

high-fl ow column due to the lon-

ger residence times and higher pH

and alkalinity.

Removal of Mn and Co was

poor in both columns, because

they hydrolyze at a higher pH

than that achieved in calcite

DAS. An alkaline material pro-

ducing higher pH on dissolution

(e.g., MgO) would need to be

applied in DAS to precipitate

these metals (Cortina et al.,

2003; Rötting et al., 2006).

Chemical performance of both

columns was similar when operat-

ing at the same fl ow rate, as can

be seen in the outfl ow gross alka-

linity, which correlates with fl ow

rate (Fig. 3, bottom graphs). For

example, the high-fl ow column

operated at 0.26 to 0.27 L/d dur-

ing Weeks 39 to 43 with an out-

fl ow alkalinity of 200 mg/L; the

low-fl ow column operated at 0.25 to 0.28 L/d during Weeks

36 to 54 with an outfl ow alkalinity of 220 mg/L. For lower

fl ow rates, outfl ow alkalinity increased in both columns—the

high-fl ow column operated at 0.16 to 0.18 L/d during Weeks

19 to 27 with an outfl ow alkalinity of 330 mg/L; the low-fl ow

column operated at 0.13 to 0.14 L/d during Weeks 25 to 32

with an outfl ow alkalinity of 290 mg/L.

Th e removal of diff erent types of metals was also comparable

in both columns when operating at similar fl ow rates. Usually,

the high-fl ow column did not remove any Cd, Zn, and Ni,

whereas the low-fl ow column removed a small portion of these

Fig. 4. Depth profi les of pH and dissolved concentrations (mg/L) of Ca, SO4

2−, Fe(II), Fe(III), Al, Zn, Mn, Si, Cu, Cd, and Pb in the high-fl ow column. The values at −3-cm depth represent the supernatant, the gray shaded area from 0 to 20 cm the depth profi le in the column substrate, and the values at 25-cm depth the output container.

Rötting et al.: Passive Treatment of Acid Mine Drainage using DAS 1747

metals. However, when the high-fl ow column was operated at

low fl ow rates, it started to remove some Cd, Zn, and Ni (see Cd

and Zn in Fig. 4, series “24 wk, 0.18 L/d”). Th ese facts show that

the results of both columns are consistent with each other.

Hydraulic Conductivity Profi lesHydraulic conductivity of the high-fl ow column was

measured for the fi rst time after 2 wk (Fig. 6A). By then,

brownish precipitates had accumulated in the top 3 cm of

the substrate, but hydraulic conductivity was still very high at

depth. With time hydraulic conductivity decreased, fi rst most

notably in the upper part, and later on also in the middle and

to some extent in the lower part of the substrate.

Th e low-fl ow column was equipped with an additional

pressure port at 1.5-cm depth, which revealed that the great-

est head loss did not occur at the substrate surface, but be-

tween 1.5- and 4-cm depth (Fig. 6B). Hydraulic conductivity

remained high near the surface and at the bottom of the col-

umn and decreased only moderately between 4 and 9 cm.

It is interesting to notice that minimum hydraulic conductivity

was similar in both columns, despite the fact that the high-fl ow

column had treated about four times more water than the low-fl ow

column. A broader precipitation band ought to be expected in

the high-fl ow column, which should be further from equilibrium,

than in the low-fl ow column. While this is indeed observed, K

measurements suggest that precipitates fi rst concentrate in a nar-

row band and that this band is then extended downward.

Solid Samples

Observations from Substrate Excavation

In the high-fl ow column, the uppermost 1 cm of excavated

substrate consisted of moderately cemented wood chips covered

with dark brown precipitates and no traces of calcite. Th e next 2

cm contained loose wood chips covered with light brown precipi-

tates and few calcite grains. Just below this “brown” zone, a hard,

yellowish-white precipitate covered the wood chips and calcite

grains. Th e fi rst 2 cm of this “white” layer were heavily cemented

and diffi cult to break. Th e “white” layer had a total thickness of

about 7 cm and contained zones of brown precipitates, especially

near the column wall. From 10- to 15-cm depth, the column

contained loose wood chips and calcite grains with some white

precipitates and no brown zones. Below 15 cm, the substrate ap-

peared to contain much calcite and few precipitates.

Th e same zones were found in the low-fl ow column, even

though the precipitates were softer, and thickness of the lay-

ers was smaller. Th e uppermost 2 cm consisted of slightly ce-

mented wood chips covered with dark brown precipitates. Th e

uppermost pressure sampling port (1.5-cm depth) was located

in the lower part of this “brown” layer. Th erefore, the relatively

high hydraulic conductivity of the uppermost segment of the

profi le (Fig. 6B) corresponds to this layer. Just below this zone,

a yellowish-white precipitate covered the wood chips and calcite

grains, but this layer was not as strongly cemented as the one

found in the high-fl ow column. Th is “white” layer was about 4

cm thick and contained zones of brown precipitates, especially

near the water sampler at 4-cm depth. Th e pressure sampling

port at 4-cm depth was located in the lower part of the “white”

layer. Th is shows that the low hydraulic conductivity of the

second segment of the profi le corresponds to the “white” layer.

From 6 cm to the bottom, the column contained loose wood

chips with much calcite and few precipitates.

Substrate Analyses

Th e brownish precipitates of both columns (up to 1-cm

depth in the low-fl ow and up to 3-cm depth in the high-fl ow

column) contained high concentrations of Fe and some S (Fig.

7A). Practically all calcite had dissolved in this layer. X-ray

diff raction showed the characteristic peaks of goethite. Never-

theless, in fi eld studies schwertmannite is the mineral which

Fig. 5. Box-and-whisker plots of relative element removal (minimum, 25% percetile, median; 75% percentile, maximum) in the high-fl ow column from (A) supernatant to drain pipe and (B) supernatant to output container, and in the low-fl ow column from (C) supernatant to drain pipe and (D) supernatant to output container.

Fig. 6. Average hydraulic conductivity between pressure ports for (A) the high-fl ow and (B) the low-fl ow column.

1748 Journal of Environmental Quality • Volume 37 • September–October 2008

precipitates from Monte Romero AMD and subsequently

transforms into goethite and jarosite (Acero et al., 2006). Molar

Fe/S ratios in these samples range from 8 to 11, suggesting that

a mixture of schwertmannite and goethite was present.

Th e SEM-EDS (Fig. 8) revealed very Fe-rich precipitates

with minor S and Si covering the wood chips and fi lling the

shape of former limestone grains which had dissolved com-

pletely (“limestone ghosts”). Th is shows that Fe precipitates

do not passivate DAS.

Th e “white” layers (from 3- to 10-cm depth in the high-fl ow

column and from 1- to 4-cm depth in the low-fl ow column)

contained high concentrations of Al (Fig. 7A, 7B). In the very Al-

rich samples (“3 cm” in high-fl ow, “1.5 cm” in low-fl ow column)

about half of the initial calcite had dissolved. Sulfate concentra-

tion decreased with depth while calcite increased. XRD revealed

peaks of calcite and gypsum, but no peaks could be attributed to

any Al-bearing compound. Al probably precipitated as an amor-

phous solid. In the low-fl ow column the gypsum-peaks were

most intense in the sample from 1.5-cm depth and decreased

with depth. In the high-fl ow column gypsum peaks were very

weak, probably because most gyp-

sum had been redissolved due to

injection of distilled water during

the fi nal tracer test.

Th e SEM-EDS images of this

layer (Fig. 9) show that calcite

grains and wood chips are covered

by a thick inner layer of gypsum

and a thin outer layer of Al-rich

precipitates with no visible crystal

structure. Th e shape of the gypsum

mass suggests that it fi lls the spaces

formerly occupied by limestone.

Th is indicates that limestone

grains dissolve even when they are

covered by precipitates, as was ob-

served in the Fe zone. Probably the

protons released during Fe(III)-

and Al-hydrolysis and precipita-

tion enhance calcite dissolution, which in turn boosts hydrolysis.

Due to the small calcite grain size used in DAS, a large fraction of

the calcite grains dissolve completely before the crust of precipi-

tates becomes too thick. Th e Al/S ratios of the bulk samples (after

correction for S associated with gypsum) ranged from 5.8 to 9.7,

higher than that of Basaluminite [Al4(SO

4)(OH)

10⋅5(H

2O)],

which has the highest Al/S ratio of known Al-hydroxysulphates

in nature. Th is indicates that both Al-hydroxysulphates and Al-

hydroxides precipitate inside the DAS columns. Th e Al/S ratio

increased with depth, which again suggests that with increasing

pH more hydroxyl (OH−) and less sulfate anions are integrated in

the solid phase. A similar behavior is observed for Fe(III)-miner-

als (Bigham and Nordstrom, 2000).

In the Al-rich samples the concentrations of Zn, Cu, Pb, and

Cd were also highest (Fig. 7A′). X-ray diff raction did not show

peaks for phases containing any of these metals, but this was to

Fig. 7. Chemical composition of bulk solid samples. (A) Major and (A') trace elements in the high-fl ow column and (B) major elements in the low-fl ow column. Wood chips (organic carbon) and structural water are not represented.

Fig. 8. Electron backscatter image of a thin section of Fe-rich precipitates (fe) and a wood chip (w) from the high-fl ow column at 0.5-cm depth. The precipitates outline and fi ll former limestone grains which have dissolved completely (fe-lg, “limestone ghosts”).

Fig. 9. Electron backscatter image (bs) of a thin section from the low-fl ow column at 2-cm depths, and element maps of Ca, Al, and S marked in black. Calcite (cc), gypsum (gy), and Al-hydroxysulfate (Al-S) are labeled.

Rötting et al.: Passive Treatment of Acid Mine Drainage using DAS 1749

be expected due to the low metal

concentrations in the samples.

Th e deeper samples (below

10-cm depth in the high-fl ow

column and below 4-cm depth

in the low-fl ow column) con-

tained little Al and metals. X-ray

diff raction showed only the

peaks of calcite and gypsum.

Reactive Transport ModelA fi nite-element mesh with

410 linear elements and a total

length of 20 cm was used to

represent the depth profi le of the

high-fl ow DAS column. Dis-

solution of calcite was simulated

according to the kinetic rate

law described by Plummer et

al. (1978), with an initial reac-

tive surface area of 4300 m2/m3

calculated from grain-size distri-

bution assuming spherical par-

ticles. Schwertmannite (Fe8O

8(OH)

5.5(SO

4)

1.25, log K = 18.0;

Bigham et al., 1996) and basaluminite were set to precipitate

in equilibrium. As gypsum was oversaturated within the DAS

columns, it was set to precipitate kinetically at a rate of 10−19

mol/m3⋅s. Th e supernatant was set to be in equilibrium with

atmospheric O2 and CO

2.

Oxidation of Fe(II) to Fe(III) was modeled according

to the rate laws of Singer and Stumm (1970). However, to

match the observed Fe(II) depth profi les, the Fe(II) oxidation

rate for pH < 3.5 had to be adjusted to 0.1 mol/m3⋅s, fi ve or-

ders of magnitude higher than the published abiotic rate. Th is

indicates that Fe(II) oxidation is either enhanced by bacteria

(e.g., Johnson and Hallberg, 2003), or by catalytic oxidation

on Fe(III)-hydroxide surfaces (Jarvis and Younger, 2001).

To match observed pH and dissolved Ca concentrations (Fig.

10A, 10B), the reactive calcite surface had to be decreased. Diff er-

ent values had to be assigned to the zone of Fe precipitates (calcite

area 270 m2/m3 from 0- to 3-cm depth) and the zone of Al precipi-

tates and below (calcite area 90 m2/m3 from 3- to 20-cm depth).

Th ese lower-than-calculated calcite areas assigned are consistent

with the assumption of partial armoring by the precipitating solid

phases. Th e degree of passivation appears to depend on the solid.

As observed in the solid samples section, Fe precipitates reduce cal-

cite dissolution only slightly. In the Al zone, passivation is stronger.

Th e thick gypsum layers around limestone grains observed in this

zone suggest that gypsum is the main cause of passivation, as also

described by Booth et al. (1997).

Simulated pe and dissolved species other than Ca are in good

agreement with observations. Simulated precipitated and dis-

solved solids (Fig. 10C) are consistent with observations. Calcite

has dissolved almost completely in the Fe zone, and partially in

the Al zone. Schwertmannite forms massively at the substrate

surface, moderately up to 3-cm depth, and most of the remain-

ing Fe(III) precipitates at the contact with the Al precipitates.

Iron(III) precipitation near the surface is enhanced by oxidation

of Fe(II). Up to 3-cm depth the protons released by Fe(III)-

precipitation are neutralized by dissolution of calcite. At the

contact with the Al zone, they are neutralized by redissolution of

previously precipitated basaluminite. Dissolved Al concentration

is constant in the top 3 cm, but increases above the input value

where basaluminite redissolves, and precipitates below dissolving

calcite. Gypsum starts to precipitate in the Al zone, consistent

with observations. Th e bulk balance of minerals dissolved and

precipitated leads to an increase in porosity of 2%, consistent

with the trend observed in the Br-tracer tests. Discrepancy in the

actual values (9% measured) are probably due to (i) dissolution

of gypsum during the tracer tests (which occupies about 2% of

the substrate volume by the end of the simulation), (ii) uncer-

tainties in the molar volumes of Al and Fe mineral phases, and

(iii) uncertainties in porosity measurements.

Implications for Future Field OperationBoth DAS columns eliminated Al, Fe(III), Pb, and Cu al-

most completely. Th e high-fl ow column operated at a loading

rate of 120 g acidity/m2⋅d, four times the recommended rate

for conventional passive systems. Th e low-fl ow column had

a mean loading rate of 35 g acidity/m2⋅d, similar to the rate

recommended for RAPS. Th is column additionally removed

some Zn, Ni, and Cd. However, this loading rate would re-

quire very large surface areas to treat high-acidity AMD. At

fi eld scale, it will probably be more effi cient to operate DAS at

the loading rate of the high-fl ow column, trading off a slightly

lower metal removal for a considerably smaller treatment area.

While the columns did not clog during 15 mo of opera-

tion, the permeability reduction was signifi cant. Th is implies

that suffi cient freeboard should be left in practice for the

supernatant head to increase, so as to maintain the fl ow rate.

Fig. 10. Depth profi les of reactive transport model results: (A) pH and pe, (B) total dissolved concentrations of primary aqueous species (mol/L solution), and (C) total amount of precipitation (positive values) or dissolution (negative values) of minerals (mol/m3 bulk substrate). Symbols represent values measured in the high-fl ow column after 61 wk of operation; lines represent simulated results.

1750 Journal of Environmental Quality • Volume 37 • September–October 2008

Th e “life time” of DAS until exhaustion or clogging will have

to be determined in fi eld trials, which are currently being per-

formed at Monte Romero Mine (SW Spain).

ConclusionsHigh Fe(III) and Al concentrations can be treated in DAS.

Th is contrasts with ALD and RAPS, where infl owing Fe(III)

and Al concentrations have to be kept low to avoid passiva-

tion or clogging. Aluminum, Fe(III), Pb, and Cu were almost

completely eliminated within both DAS columns. Removal of

divalent metals (e.g., Zn, Mn, Ni, Cd, Co) was poor. It could

be improved using caustic magnesia (MgO) instead of calcite

as a reactive alkaline material in DAS.

Mean net acidity elimination was around 1200 mg/L.

Th ese removal rates are three to four times higher than those

achieved by conventional passive treatment systems. Net

acidity elimination in DAS appears to depend primarily on

infl owing Al and Fe(III) concentrations.

Th e oxic supernatant is an essential part of the treatment

system when high Fe(II) concentrations are present in the

input water. Iron(II) elimination in the supernatant increased

with residence time. Drain pipe alkalinity was inversely cor-

related with fl ow rate. Th is is an important factor if Fe(II)

concentration in the outfl ow is high and has to be oxidized in

subsequent treatment steps.

Iron(III) precipitated in the upper few centimeters of the col-

umns, probably as Schwertmannite which then transformed into

goethite. Aluminum precipitated at greater depth as hydroxysul-

phates, together with gypsum. Calcite dissolved completely in the

zone of Fe precipitation, and partially in the zone of Al precipita-

tion. Hydraulic conductivity profi les indicate that Fe precipitates

reduced DAS permeability only marginally, whereas permeability

reduction by Al solid phases and/or gypsum was more important.

Overall porosity was increased, whereas the tracer tests suggest

that mobile porosity was reduced. Th is is consistent with a con-

centrated permeability reduction in the top part of the Al zone.

Th e reduction was similar in the high-fl ow and in the low-fl ow

column, and did not prevent the columns from functioning for

more than 1 yr without clogging.

Th e DAS reactivity is greater than that of limestone gravel-

based treatment systems, because mixing with a coarse inert

matrix in DAS allows the use of limestone sand, which provides

a higher reactive surface. Due to their small size, the grains are

dissolved before being coated (passivated), and most of the reac-

tive material is consumed. Th e high acidity removal is possible

because Al, Fe(III), and other metals accumulate intentionally

in DAS. Hydrolysis and precipitation release protons, which

promote further dissolution of alkaline reagent. Accumulation

within the reactor also facilitates waste management. Th e large

pores of the inert DAS matrix and the dispersion (separation)

of the calcite grains reduce clogging problems.

AcknowledgmentsTh e authors thank Eva Peleri (ICP-AES), Antonio Padro

(ICP-MS), and the SEM-EDS team from SCT-UB as well

as Josep Elvira (XRD) and Jesus Parga (preparation of thin

sections) from IJA-CSIC for technical assistance. Th is study

was funded by the Spanish Government projects REN-

2003-09590-C04-02, CTM2006-28151-E/TECNO, and

CTM2007-66724-C02-01.

ReferencesAcero, P., C. Ayora, C. Torrento, and J.-M. Nieto. 2006. Th e behavior of

trace elements during schwertmannite precipitation and subsequent transformation into goethite and jarosite. Geochim. Cosmochim. Acta 70:4130–4139.

Al, T.A., C.J. Martin, and D.W. Blowes. 2000. Carbonate-mineral/water interactions in sulfi de-rich mine tailings. Geochim. Cosmochim. Acta 64:3933–3948.

Ball, J.W., and D.K. Nordstrom. 1991. User’s manual for WATEQ4F with revised thermodynamic database and test cases for calculating speciation of major, trace and redox elements in natural waters. U.S. Geol. Surv. Water-Resour. Invest. Rep. 91-183.

Banks, D., P.L. Younger, R.T. Arnesen, E.R. Iversen, and S.B. Banks. 1997. Mine-water chemistry: Th e good, the bad and the ugly. Environ. Geol. 32:157–174.

Bigham, J.M., U. Schwertmann, S.J. Traina, R.L. Winland, and M. Wolf. 1996. Schwertmannite and the chemical modeling of iron in acid sulfate waters. Geochim. Cosmochim. Acta 60:2111–2121.

Bigham, J.M., and D.K. Nordstrom. 2000. Iron and aluminum hydroxysulfates from acid sulfate waters. p. 351–403. Sulfate minerals: Crystallography, geochemistry and environmental signifi cance. Mineralogical Soc. of Am., Washington, DC.

Booth, J., Q. Hong, R.G. Compton, K. Prout, and R.M. Payne. 1997. Gypsum overgrowths passivate calcite to acid attack. J. Colloid Interface Sci. 192:207–214.

Cabrera, G., R. Perez, J.M. Gomez, A. Abalos, and D. Cantero. 2006. Toxic eff ects of dissolved heavy metals on Desulfovibrio vulgaris and Desulfovibrio sp strains. J. Hazard. Mater. 135:40–46.

Cortina, J.L., I. Lagreca, J. De Pablo, J. Cama, and C. Ayora. 2003. Passive in situ remediation of metal-polluted water with caustic magnesia: Evidence from column experiments. Environ. Sci. Technol. 37:1971–1977.

Cravotta, C.A., and M.K. Trahan. 1999. Limestone drains to increase pH and remove dissolved metals from acidic mine drainage. Appl. Geochem. 14:581–606.

Cravotta, C.A. 2003. Size and performance of anoxic limestone drains to neutralize acidic mine drainage. J. Environ. Qual. 32:1277–1289.

Demchak, J., J. Skousen, and L.M. McDonald. 2004. Longevity of acid discharges from underground mines located above the regional water table. J. Environ. Qual. 33:656–668.

Hedin, R.S., G.R. Watzlaf, and R.W. Nairn. 1994. Passive treatment of acid-mine drainage with limestone. J. Environ. Qual. 23:1338–1345.

Jage, C.R., C.E. Zipper, and R. Noble. 2001. Factors aff ecting alkalinity generation by successive alkalinity-producing systems: Regression analysis. J. Environ. Qual. 30:1015–1022.

Jarvis, A.P., and P.L. Younger. 2001. Passive treatment of ferruginous mine waters using high surface area media. Water Res. 35:3643–3648.

Johnson, D.B., and K.B. Hallberg. 2003. Th e microbiology of acidic mine waters. Res. Microbiol. 154:466–473.

Johnson, D.B., and K.B. Hallberg. 2005. Acid mine drainage remediation options: A review. Sci. Total Environ. 338:3–14.

Kaufmann, P.R., A.T. Herlihy, and L.A. Baker. 1992. Sources of acidity in lakes and streams of the United States. Environ. Pollut. 77:115–122.

Kirby, C.S., and C.A. Cravotta. 2005. Net alkalinity and net acidity: I. Th eoretical considerations. Appl. Geochem. 20:1920–1940.

Lopez Fernandez, A., G. Lopez Montenegro, and J. Romero Sousa. 2003. Tratamiento de aguas de Mineria en el rio Odiel. Medio Ambiente 44:52–55.

Neculita, C.M., G.J. Zagury, and B. Bussiere. 2007. Passive treatment of acid mine drainage in bioreactors using sulfate-reducing bacteria: Critical review and research needs. J. Environ. Qual. 36:1–16.

Olias, M., J.M. Nieto, A.M. Sarmiento, J.C. Ceron, and C.R. Canovas. 2004. Seasonal water quality variations in a river aff ected by acid mine drainage: Th e Odiel River (South West Spain). Sci. Total Environ. 333:267–281.

Parkhurst, D.L. 1995. User’s guide to PHREEQC: A computer program for

Rötting et al.: Passive Treatment of Acid Mine Drainage using DAS 1751

speciation, reaction path, advective-transport, and inverse geochemical calculations. U.S. Geol. Surv. Water-Resour. Invest. Rep. 95-4227.

Plummer, L.N., T.M.L. Wigley, and D.L. Parkhurst. 1978. Kinetics of calcite dissolution in CO

2–Water Systems at 5-Degrees-C to 60-Degrees-C

and 0.0 to 1.0 Atm CO2. Am. J. Sci. 278:179–216.

Rees, B., R. Bowell, M. Dey, and K. Williams. 2001. Passive treatment: A walk away solution? Mining Environ. Manage. March 2001:7–8.

Rose, A.W., D. Bisko, A. Daniel, M.A. Bower, and S. Heckman. 2004. An “autopsy” of the failed Tangaskootack #1 vertical fl ow pond, Clinton Co., Pennsylvania. p. 1580–1594. In R.I. Barnhisel (ed.) Joint Conference of the 21st Annual Meetings of the American Society of Mining and Reclamation and 25th West Virginia Surface Mine Drainage Task Force Symposium, Morgantown, WV. ASMR, Lexington, KY.

Rötting, T.S., J. Cama, C. Ayora, J.L. Cortina, and J. De Pablo. 2006. Use of caustic magnesia to remove cadmium, nickel, and cobalt from water in passive treatment systems: Column experiments. Environ. Sci. Technol. 40:6438–6443.

Saaltink, M.W., F. Battle, C. Ayora, J. Carrera, and S. Olivella. 2004. RETRASO, a code for modeling reactive transport in saturated and unsaturated porous media. Geol. Acta 2:235–251.

Simon, M., F. Martin, I. Garcia, P. Bouza, C. Dorronsoro, and J. Aguilar. 2005. Interaction of limestone grains and acidic solutions from the oxidation of pyrite tailings. Environ. Pollut. 135:65–72.

Singer, P.C., and W. Stumm. 1970. Acidic mine drainage. Th e rate-determining step. Science 167:1121–1123.

Stookey, L.L. 1970. Ferrozine: A new spectrophotometric reagent for iron. Anal. Chem. 42:779–781.

Th omas, R.C., and C.S. Romanek. 2002a. Passive treatment of low-pH, ferric iron-dominated acid rock drainage in a vertical fl ow wetland: I. Acidity neutralization and alkalinity generation. p. 723–751. In Proc.

of the 2002 National Meeting of the American Society of Mining and Reclamation, Lexington, KY. 9–13 June 2002. ASMR, Lexington, KY.

Th omas, R.C., and C.S. Romanek. 2002b. Passive treatment of low-pH, ferric iron-dominated acid rock drainage in a vertical fl ow wetland: II. Metal removal. p. 752–775. In Proc. of the 2002 National Meeting of the American Society of Mining and Reclamation, Lexington, KY. 9–13 June 2002. ASMR, Lexington, KY.

Utgikar, V.P., H.H. Tabak, J.R. Haines, and R. Govind. 2003. Quantifi cation of toxic and inhibitory impact of copper and zinc on mixed cultures of sulfate-reducing bacteria. Biotechnol. Bioeng. 82:306–312.

Watzlaf, G.R. 1997. Passive treatment of acid mine drainage in down-fl ow limestone systems. p. 611–622. In J.E. Brandt et al. (ed.) Proc. of the National Meeting of the American Society for Surface Mining and Reclamation (ASSMR), Austin, TX. 10–15 May 1997. Am. Soc. for Surface Mining and Reclamation, Princeton, WV.

Watzlaf, G.R., C.L. Kairies, K.T. Schroeder, T. Danehy, and R. Beam. 2002. Quantitative results from the fl ushing of four reducing and alkalinity-producing systems. 2002 West Virginia Surface Mine Drainage Task Force Symp., Morgantown, WV. 16–17 April, 2002. West Virginia Surface Mine Drainage Task Force, Morgantown, WV.

Watzlaf, G.R., K.T. Schroeder, R.L.P. Kleinmann, C.L. Kairies, and R.W. Nairn. 2004. Th e passive treatment of coal mine drainage Laboratory report DOE/NETL-2004/1202. U.S. Dep. of Energy, National Energy Technol. Lab., Pittsburgh, PA.

Ziemkiewicz, P.F., J.G. Skousen, D.L. Brant, P.L. Sterner, and R.J. Lovett. 1997. Acid mine drainage treatment with armored limestone in open limestone channels. J. Environ. Qual. 26:1017–1024.

Ziemkiewicz, P.F., J.G. Skousen, and J. Simmons. 2003. Long-term performance of passive acid mine drainage treatment systems. Mine Water Environ. 22:118–129.