Journal of Geochemical Exploration 172 (2017) 120–132

Contents lists available at ScienceDirect

Journal of Geochemical Exploration

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

Acid rock drainage prediction: A critical review

Bernhard Dold
Division of Geosciences and Environmental Engineering, Luleå University of Technology (LTU), Luleå, Sweden

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

Article history: Acid rock drainage (ARD) prediction is a very important issue in order to predict and prevent environmental pol-
Received 2 February 2016 lution associated with mining activities. Nowadays, simple tests are widely applied and established in the mining
Revised 25 September 2016 and consulting business for ARD prediction. These tests have many known errors and problems, as that they do
Accepted 30 September 2016
not account for the complexity of the mineral assemblage of an ore deposit, and therefore are not able to predict
Available online 6 October 2016
the geochemical behavior accurately. This critical review has the aim of first, highlighting the geochemical pro-
Keywords:
cesses associated to the problems of ARD prediction. Secondly, the errors and limitations of the standard static
Acid mine drainage and kinetic tests are highlighted. The currently applied calculation factor of 31.25 for sulfide acid potential calcu-
Prediction lation overestimates the carbonate neutralization potential by 100% in its geochemical assumptions. Thus, the
Static and kinetic test calculation factor 62.5, based on the effective carbonate speciation at neutral pH, is recommended. Additionally,
Acid-Base Accounting (ABA) standard ABA procedure ignore the acid potential of Fe(III) hydroxides and/or sulfates and do not distinguish be-
Mining tween different carbonate minerals. This can be critical, as for example siderite can be a net acid producing car-
Pollution bonate. Therefore, it is crucial to count on accurate quantitative mineral data in order to be able to accurately
Sustainability
predict ARD formation and potential liberation of hazardous trace elements to the environment.
In many modern mining operations, quantitative mineral data is nowadays produced in order to enhance the re-
covery of the extraction process by the incorporation of geometallurgical information (e.g. quantitative mineral-
ogy, mineral liberation, textural information, grain size distribution). Thus, the use of this very same existing data
for ARD prediction can increase importantly the precision of ARD prediction, often without additional costs and
testing. The only requirement is the interdisciplinary collaboration between the different divisions and data ex-
change in a modern mining operation.
© 2016 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
2. Acid rock drainage and the importance of mineralogy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
3. Acid rock drainage (ARD) formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
3.1. Acid liberation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
3.2. Acid neutralization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
3.2.1. Silicates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
3.2.2. Hydroxides and hydroxide sulfates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
3.2.3. Carbonates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
4. Standard ARD prediction and its limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
4.1. A historical background and mineralogical considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
4.2. ABA calculation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
4.3. Kinetic testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
5. Sample selection criteria for ARD prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
6. Suggested characterization methodology for ARD prediction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
7. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131

E-mail address: [email protected].

http://dx.doi.org/10.1016/j.gexplo.2016.09.014
0375-6742/© 2016 Elsevier B.V. All rights reserved.
 B. Dold / Journal of Geochemical Exploration 172 (2017) 120–132 121

1. Introduction As a consequence, the amount of the waste material produced increased

dramatically. Additionally, the phenomena of sulfide oxidation associat-
Human history is an accumulation of conflicts about geo-resources ed with the mining process could be first observed with the formation of
(e.g. land, soil, water, metal, oil-gas). Therefore, mining has always acid mine drainage in coal mining (Colmer and Hinkle, 1947), which
had an essential role in the development of human history since the now is the main environmental challenge of the metal mining industry
Neolithic Period. This has not changed today and most of the conflicts today (Dold, 2014).
we observe nowadays have essentially the very same origin. Mining As in the last decades environmental pollution has reached levels,
provides the metal supply of modern high-tech society. Although in which are directly critical for human health in some parts of the world
daily life it is not very evident for most of the world's population, mining (Rockstrom et al., 2009), the environmental and sustainable aspects
industry is the basis of economic wealth and technological development have gained an important role in the last decades.
(Matsumoto et al., 2016), as can be observed in the industrial revolution In order to get the dimension of the problem, the world production
in Europe (late 17th century) and the current information and energetic of some base metals through the importance of each exploited ore de-
revolution world-wide. Although doubtless an essential industry, posit type is analyzed (Laznicka, 1999, 2014; Singer, 1995). Plumlee
today's mining is facing severe problems for development, as the nega- (1999) highlighted the relationship between ore deposit type and
tive aspects of the mining activity are perceived to affect more and more final ARD composition, which allows the behavioral prediction for
people in a globalized world. Thus, the mining industry is struggling to most deposit types with respect to ARD formation, assuming that a
obtain the social license in many parts of the world in order to develop good knowledge of their mineralogical composition is available.
new projects (Owen and Kemp, 2013). This is mainly due to the fact that Giant (and super-giant) metallic ore deposits supply between 60
most stakeholders, without direct benefit from the mining activity, per- and 95% of global metallic resources on land, with several individual de-
ceive mining as a threat for their health and the environment, as for ex- posits monopolizing the supply (Laznicka, 2014), from which Cu has the
ample by the formation of acid mine drainage (AMD) or more general highest number of super-accumulations followed by Au, Pb, Mo, Sb, and
acid rock drainage (ARD) and subsequent water pollution (Fig. 1). As Sn (Laznicka, 1999). This is mainly due to the geochemical behavior of
metal extraction technologies have evolved, the amounts of mine the different metals during enrichment processes in the earth crust.
wastes produced have increased tremendously, a trend, which will in- Cu, Au and Mo are mainly extracted from porphyry systems (60% of
crease even more in the future. Hence, more precise ARD prediction the world copper production is from porphyry Cu-Mo and Cu-Au de-
methods are required to tackle this problem. ARD is the result of the ox- posits), and therefore the extraction of these elements is mainly done
idation sulfide-rich rocks, exposed by natural or anthropogenic process- in huge mining operations (Gerst, 2008; Mudd et al., 2013). Thus,
es to the atmosphere, process, which leads to high element when considering these elements and the other ore deposit types,
concentrations in and acidification of the water. This process can result which host them (Fig. 1), it can be stated that most of the giant or
in a pollution of surface and groundwater resources. In this review the super-giant deposits (except some carbonate hosted, skarn or placer de-
more general term ARD is used, as this applies also for any sulfide con- posits) will be very prone to produce ARD (Fig. 1).
taining material extracted and deposited to oxidizing conditions, like for ARD prediction is therefore an important issue due to the fact that
example in road, tunnel, or construction activities, or in natural oxida- ARD is one the principal environmental issue facing the mining industry
tion processes. nowadays (Fig. 2). Due to the increasing social pressure, governments
In ancient times the metal supply was ensured by oxide ore deposits and state agencies increasingly implement new regulations and laws
exposed on the surface of the earth crust (e.g. the gossan in the Iberian to regulate mine waste management around the world, like for example
Pyritic belt from Roman time, or the silver gossans at Cerro de Pasco, the new mine closure law in Chile (in place since November 2012). Con-
Peru and Potosi, Bolivia pre-Inca and during the Spanish conquistador sequently, an accurate characterization of mine waste is required in
era (Abbott and Wolfe, 2003). Since the development of the flotation order to evaluate the geochemical stability and predict if ARD might
technique in the early 20th century (Bergh and Yianatos, 2011), it be- be an issue in the future that needs to be addressed and mitigated in
came possible to extract and separate metal sulfides from the primary time.
zone of the ore deposit. This technological change, increased the dimen- There are mainly two types of tests applied to predict the behavior of
sion of mining, as now low-grade ores could be economically exploited. a rock sample concerning ARD production: The static tests, which eval-
uates if or not a sample will produce an acidic environment; and the ki-
netic tests, which evaluates, when the acidification will take place and
which elements are liberated from the sample. In order to obtain this in-
formation, there are several tests widely used in the consulting and
mining business and some are seen as standard methods.
The standard methods applied in ARD prediction are listed, ex-
plained and reviewed in several reports, books, and papers
(Bouzahzah et al., 2014; Chotpantarat, 2011; Jacobs et al., 2014;
Lottermoser, 2010; Parbhakar-Fox and Lottermoser, 2015; White and
Jeffers, 1994; White et al., 1999) and will therefore not be all listed
and addressed again in this review. It is also widely accepted that
these tests have several limitations and sources of errors
(Parbhakar-Fox and Lottermoser, 2015; White et al., 1999).
Parbhakar-Fox and Lottermoser (2015) have recently highlighted
many of these problems associated with ARD prediction in an excellent
review of ARD prediction. Here I will focus on some additional, funda-
mental errors (calculation factors) and limitations (only sulfides as pro-
ton source) of the current standard approaches. Most of these
Fig. 1. Percentage of the different giant copper deposits by ore types to global copper fundamental problems and errors are known and some were highlight-
production (n = 221). Giant Cu-Mo and Cu-Au porphyry deposits represent about 60% ed 25 years ago (Norecol Environmental Consultants Ltd., 1991). How-
of the world copper production. The in yellowish to reddish colored ore types tend to
produce ARD, while the bluish colored skarn and carbonate hosted deposits might have
ever, many of these errors and problems have been tolerated or
enough carbonates to neutralize the acid potential of the ore (modified after Laznicka, ignored through time and it has reached a point where these test
2014). methods are established as standards in the mining and consultant
122 B. Dold / Journal of Geochemical Exploration 172 (2017) 120–132

Fig. 2. Some examples of acid mine drainage. Above left: AMD pond with schwertmannite/goethite precipitating as sediment, Quiulacocha tailings impoundment, Cerro de Pasco, Peru.
Above right: Collection channel for the AMD solutions from Excelsior Waste Rock Dump, Cerro de Pasco, Peru. Below left: Solid AMD in form of efflorescent salts precipitating at Pan
de Azúcar tailings impoundments, Argentina. Below right: Schwertmannite terraces forming in the AMD effluent from Rio Tinto mine, Spain.

industries. They appear in guidelines and regulations, making it difficult

to argue against them and promote new, more powerful approaches.
Therefore, it is important to summarize these drawbacks once more in
a comprehensive review and promote a modern use of mineralogical
and geochemical data for ARD prediction.
In modern mining operations the importance of quantitative miner-
alogical characterization to increase the extraction efficiency by daily in-
corporation of mineralogical data for decision making in the mining
process has been recently shown (Cruz et al., 2012), and together with
the development of automated quantitative mineralogy gives nowa-
days new tools to improve the mining process (Gottlieb et al., 2000). Al-
though this data exists in many modern, advanced mining operations,
only sporadically it is also used for ARD prediction.
This review has the aim of summarizing and explaining the main
problem of standard ARD prediction and convince the reader that the
use of the often existing quantitative mineralogy used originally for
geometallurgical purpose in a mining operation, could improve impor-
tantly also the ARD prediction, often without additional costs and
testing.

2. Acid rock drainage and the importance of mineralogy

The goal of acid rock drainage prediction is very simple. It deals with
sulfides (SAP) metal-hydroxides (sulfates) sulfates carbonates (CaNP) feldspars (FNP)
the mineralogical composition of a geological sample (Fig. 3). This sam-
ple can contain different ratios of different minerals or mineral groups py fh-gt gy ca an
(e.g. silicates, carbonates, sulfides, sulfates, oxides). There are minerals
which can produce acidity (e.g. sulfides through oxidation or cp sh sd pls
sfs
Fe(III)hydroxides and/or sulfates through equilibrium reactions),
gn jt kfs
which means they liberate protons (H+) into solution and elements,
which can hydrolyze to complexes (e.g. Fe, Al, Mn) in solution and by
doing so deprotonize water, resulting also in a liberation of protons Fig. 3. Schematic mineral composition of a rock samples and the different minerals and
into solution; i.e. resulting in an acidification. If a solution is acidic it in- mineral group producing acidity or neutralizing protons. Yellow colors indicate acid
creases the mobility of heavy metals (e.g. divalent cations Cu2+, Zn2+, potential (AP), blue neutralization potential (NP), grey and white, none of both.
Abbreviations: py = pyrite; cp = chalcopyrite; gn = galena; fh = ferrihydrite; gt =
Cd2+, Pb2+ among others), which can pollute water resources, rivers,
goethite; sh = schwertmannite; jt = jarosite; gy = gypsum; sfs = sulfo-salts; ca =
lakes, sea, or groundwater. calcite; sd = siderite; an = anorthite; pls = plagioclase; kfs = K-feldspar. SAP =
On the other hand, there are minerals in a rock, which can neutralize sulfide acid potential; CaNP = carbonate neutralization potential; FNP = feldspar
these protons (e.g. carbonates, hydroxides, and silicates) and control neutralization potential.
 B. Dold / Journal of Geochemical Exploration 172 (2017) 120–132 123

the pH as buffers at certain values (Langmuir, 1997; Stumm and 3. Acid rock drainage (ARD) formation
Morgan, 1996). If a solution is alkaline, the mobility of elements,
which are stable as oxyanions for example SO24 −, AsO34 −, MoO24 −, In this section, the geochemical and mineralogical basics necessary
CrO24 − among others, is increased (Langmuir, 1997; Stumm and to understand the ARD prediction methodology are revised. For a
Morgan, 1996). more comprehensive review of the geochemical processes leading to
In order to predict if a rock (sample) will produce an acid, neutral or ARD formation in mine waste, the reader is referred to the two open ac-
alkaline environment and/or liberate and possibly mobilize certain ele- cess publications (Dold, 2010, 2014) or other reference therein (Jambor
ments, it is necessary to know as exact as possible how much of each and Blowes, 1994; Plumlee and Logsdon, 1999).
mineral is present in the sample (and its composition; i.e. the trace ele-
ment composition). Thus, any prediction technique has the goal to 3.1. Acid liberation
quantify the reactive minerals in the sample. From an analytical point
of view, this is a very difficult task. There exist only two traditional ana- Sulfide minerals form and are stable under reducing conditions. Dur-
lytical techniques to characterize a mineral properly (i.e. define it's crys- ing mining these minerals can be exposed to oxidizing conditions, and
talline parameters): Optical microscopy and X-ray diffraction (XRD). subsequent, these minerals will undergo oxidation. This is usually
Both techniques have strong limitation for quantification. In case of op- highlighted in the literature with the example of pyrite oxidation, as
tical microscopy, a semi-quantification can be done by point counting, this mineral is the most common sulfide and the one, which produces
which is a very slow and labor intensive technique by highly qualified 4 mol of protons through its oxidation process (Colmer and Hinkle,
mineralogist, thus very expensive. X-ray diffraction has a high detection 1947; Dold, 2014; Nordstrom, 2000; Singer and Stumm, 1970). This as-
limit (2–5%), so that it is not suitable for detection of trace minerals. sumption is used in most ARD prediction method. Following Eq. (1), py-
There are techniques available to lower this detection limit like differen- rite will oxidize in the presence of atmospheric oxygen and water,
tial XRD (Dold, 2003a), but are difficult to apply. Trace minerals can be producing Fe(III)hydroxides, sulfate and 4 mol of protons per mole of
most relevant for the environmental impact prediction, as clearly pyrite oxidized by oxygen. Other sulfide minerals will produce different
shown in the case of marine pollution associated to the tailings deposi- amounts of protons during oxidation under different geochemical con-
tion in fjords at the Black Angel mine, Greenland (Elberling et al., 2002; ditions (Table 1). It can also be observed that mainly sulfide minerals
Perner et al., 2010; Poling and Ellis, 1995). containing iron produce net acidity via its oxidation, with exception of
The development of mineral quantification by the Rietveld method some lesser common sulfides like molybdenite, enargite, and stibnite.
with XRD (Epp, 2016), give today the possibility of a semi-quantifica- The liberated Fe2+ can oxidize automatically at pH N 4 or when cata-
tion of the mayor mineral phases in an mineral assemblage. lyzed by bacterial activity at pH b 4 (e.g. Acidithiobacillus spp. or
Do to these drawbacks for mineralogical quantifications; alternative Leptospirillum spp.), a proton neutralizing process (Eq. (2)). The
faster and cheaper methods were needed for mineral quantification,
specifically for ARD prediction. Subsequently, geochemical methods
based on quantification of element concentrations in the sample and as- Table 1
sumption for their mineral association in order to quantify indirectly the Overview of the oxidation reactions of the most relevant sulfide minerals via atmospheric
oxygen, considering the hydrolysis of ferric iron and the final amount of protons liberated.
mineral content, were developed (e.g. Acid-Base Accounting (ABA);
Below, the oxidation via ferric iron is shown with the liberated amounts of protons. The
(Sobek et al., 1978), with modifications (Lawrence et al., 1989; equations have to be seen as idealized and partial reactions with different end products
Skousen et al., 1997), or correlations for quantification by the reactivity and amounts of protons liberated may occur in nature.
of the mineral assemblage (e.g. paste-pH, Fizz test, NAG) were drawn
Sulfide mineral Formula Moles H+
(White et al., 1999).
3+
Parallel to this development of ARD prediction methods, due to the Oxidation via oxygen + hydrolisis of Fe
Pyrite (FeS2) FeS2 + 3.75O2 + 3.5H2O = Fe(OH)3 + 2SO2−4 + 4
mentioned analytical limitations of the mineralogical techniques, a 4H+
technique was needed for geosciences in general to quantify the mineral Arsenopyrite FeAsS + 2O2 + 3H2O = Fe(OH)3 + SO2− 4 + 3
content of a sample in an automated and fast way, with higher accuracy (FeAsS) HAsO2−4 + 3H+
and lower detection limit. This breakthrough was reached with the de- Chalcopyrite CuFeS2 + 4O2 + 3H2O = Cu2+ + Fe(OH)3 + 2
(CuFeS2) 2SO2− + 2H+
velopment of the SEM-EDX based automated quantitative mineralogy 4
Pyrrhotite x = 0.1: Fe(0.9)S + 2.175O2 + 2
technique, also known as QEMSCAN® or Mineral Liberation Analysis (Fe(1 − x)S) 2.35H2O = 0.9Fe(OH)3 + SO2− 4 + 2H+
(MLA) (Gottlieb et al., 2000). Enargite Cu3AsS4 + 8.75O2 + 2.5H2O = 3Cu2+ + 4
In order to predict ARD for a rock sample, it is necessary to quantify (Cu3AsS4) HAsO2−4 + 4SO2−4 + 4H+
the acid liberating minerals and calculate the moles of protons poten- Sphalerite ZnS + 2O2 + = Zn2+ + SO2− 4 0
(ZnS)
tially liberated, which is called the Acid Potential (AP). On the other Galena (PbS) PbS + 2O2 + = Zn 2+
+ SO42−
0
hand, the amount of minerals, which can neutralize the produced pro- Covellite (CuS) CuS + 2O2 + = Zn2+ + SO2− 4 0
tons, are quantified, the so called Neutralization Potential (NP). The dif-
Oxidation via ferric iron
ference between both indicates if the sample has enough buffer capacity Pyrite (FeS2) FeS2 + 14Fe3+ + 8H2O = 15Fe2+ + 2SO2− + 16/2a
4
to neutralize all protons potentially produced (i.e. stays circum-neu- 16H+
tral), or it will produce more protons than it can neutralize, which Arsenopyrite FeAsS + 13Fe 3+
+ 8H2O = 14Fe 2+ 2−
+ SO4 + 15/2a
means that it will ultimately produce acidic conditions. This is called (FeAsS) HAsO2− 4 + 15H+
Chalcopyrite CuFeS2 + 16Fe3+ + 8H2O = Cu2+ + 17Fe2+ + 16/0a
the Acid-Base Accounting (ABA). If there is a lack of acid buffer to main-
(CuFeS2) 2SO2− 4 + 16H+
tain the system neutral, this buffer can be added, usually in form of car- Pyrrhotite x = 0.1: Fe(0.9)S + 7.8 Fe3+ + 4H2O = 8.7Fe2+ + 8/0.2a
bonates (e.g. calcite or dolomite or sometime as lime). Therefore, the (Fe(1-x)S) SO2−4 + 8H+
calculations are done as tCaCO3/1000 t, in order to know how much car- Enargite Cu3AsS4 + 35Fe3++20H2O = 3Cu2++HAsO2− 4 + 39/4a
bonates are needed to maintain the material neutral. (Cu3AsS4) 35Fe2+ + 4SO24 + 39H+
3+ 2+ 2− +
Sphalerite ZnS + 8Fe + 4H2O = 8Fe + SO4 + 8H 8/0a
As a result traditional, widely used standard prediction methods are (ZnS)
applied, with many drawback and problems, and on the other hand, Galena (PbS) PbS + 8Fe3+ + 4H2O = 8Fe2+ + SO2− 4 + 8H+ 8/0a
high-tech analytical techniques are available at a competitive price for Covellite (CuS) CuS + 8Fe3+ + 4H2O = 8Fe2+ + SO2− 4 + 8H+ 8/0a
the end-user. Thus, it can be predicted that it will be just a matter of a
Amount of protons produced, considering if the amount of ferric iron has to be pro-
time, when this change in the approach of ARD prediction will be ac- duced first through oxidation in the system, which neutralizes one mole of protons per
cepted as new standard. mole of ferric iron (Eq. (2)).
124 B. Dold / Journal of Geochemical Exploration 172 (2017) 120–132

produced ferric iron can then hydrolyze and produce Fe(III) hydroxides, 3.2.1. Silicates
process which liberates 3 mol of protons per mole of iron (Eq. (3)). Among the silicates, quartz is considered as close to inert, and the
Therefore, hydrolysis is the main acid producing process, so that any lib- dominant silicates for neutralization are the feldspars (Feldspar Neu-
erated element during weathering able to de-protonate the water mol- tralization Potential = FNP). Here are some important differences in
ecule and hydrolyze (Ionic Potential (IP) between 4.6 and 5.6) can the reaction kinetics to consider. While anorthite has acceptable relative
produce an excess of protons in solutions (e.g. Al, Fe, Mn), i.e. lower reactivity in weathering processes, i.e. a relatively good FNP, plagioclase
the pH. This is the reason that granites will produce acidic soils, al- (albite100–30) or orthoclase (K-feldspar) have very low relative reactivi-
though they do not contain sulfides. ty (Table 2) and are therefore in general not considered in the evalua-
tion of the overall neutralization potential (NP) of a rock unit (See
Table 2), as equilibrium cannot reach in the contact time of the water
þ
FeS2 þ 15=4O2 þ 7=2H2 O⇒FeðOHÞ3 þ 2SO2−
4 þ 4H ð1Þ with the mineral. In some cases high amount of intermediately fast
weathering silicates like biotite and chlorite (~20%) for example from
potassic and propylytic alterations, might add significant NP.
Fe2þ þ 1=4O2 þ Hþ ⟺Fe3þ þ 1=2H2 O ð2Þ

3.2.2. Hydroxides and hydroxide sulfates

þ
Different metal hydroxides can have contrasting roles in a geochem-
Fe3þ
þ 3H2 O⟺FeðOHÞ3 ðsÞ þ 3H ð3Þ ical system. For example they can act as buffer by receiving protons, or
de-protonate the water molecule due to hydrolysis (Eq. (3)) or they
If pyrite is oxidized by ferric iron (Eq. (4)), 16 mol of protons are lib- can liberate hydroxide ions into solution (Ca(OH)2; portlandite), lead-
erated per mole of pyrite oxidized. In this case it is crucial to have in ing to alkaline pH conditions. The principal hydroxide buffers in the
mind that if the 14 mol of Fe3+ in Eq. (2) have to be produced in the sys- ARD environment or acid soils are dominated by the most abundant
tem (i.e. in the fresh mine tailings impoundment or waste rock dump), metal cations with the valence 3 +; i.e. Fe3 + and Al3 +. This is due to
14 mol of protons will be neutralized (Eq. (2)), so that this reaction ef- the ability of three valence cations to hydrolyze, given to their high
fectively produces 2 mol of protons, as in the case with oxidation via at- ionic potential (IP) between 4.65 (Fe3 +) and 5.61 (Al3 +), forming
mospheric oxygen. However, if a ferric iron-rich solution enters into a solid hydroxide minerals like ferrihydrite, goethite, schwertmannite,
sulfide-rich system, then effectively 16 mol of protons are produced, jarosite-alunite, and gibbsite. These minerals represent buffers, which
as for example in the case of the Quiulacocha tailings impoundment control the pH at ~4.3 (Al(OH)3; gibbsite), ~3.5 (Fe(OH)3; ferrihydrite,
and the Excelsior Waste-rock dump, Cerro de Pasco, Peru (Dold et al., goethite), ~ 2.5–3.5 (schwertmannite), ~ 2 (jarosite). Thus, metal hy-
2009). droxides represent important buffers in the acid pH ranges (Fig. 4).

þ 3.2.3. Carbonates
FeS2 þ 14Fe3þ þ 8H2 O⇒15Fe2þ þ 2SO2−
4 þ 16H ð4Þ
As in ARD prediction, the most important question is to predict if the
system maintain neutral pH or goes acid, therefore, the most important
Based on these assumptions it is generally established that one mole target minerals for the NP are the Ca-Mg carbonates, as they have a fast
of pyrite produces 4 mol of protons through its oxidation and the asso- reactivity and tend to buffer the system a circum-neutral pH values
ciated precipitation of Fe(OH)3 (Eq. (1)). This is used to calculate the AP (Table 2; Fig. 4). Therefore, it is crucial to apply the geochemistry of
of a rock unit based on its sulfur content and expressed as pyrite the carbonate system (Fig. 5) properly, in order to calculate and predict
equivalent. the NP accurately and the final ABA.
Another proton source, which is not considered by the standard ARD Calcite, and magnesite (−logK 8.48, −logK 8.24) are fast dissolving
prediction tests, is the group of Fe(III) hydroxides and Fe(III) hydroxide carbonates, (Table 2), while dolomite has lower solubility (− logK
sulfates (e.g. jarosite-alunite group, schwertmannite) together 17.09). Therefore, calcite is generally considered in calculations for NP
with metal chlorides and sulfates (e.g. eriochalcite, chalcanthite, and results are expressed as its equivalent. Calcite is the key buffer to
rhomboclase), which might be the source of important amounts of acid- maintain the pH at circumneutral conditions and is therefore the most
ity in ARD systems. The protons might be liberated due to dissolution- important neutralizing agent in the mining environment. As long as cal-
equilibrium reaction or due to mineral transformation due to meta-sta- cite is available as a buffer in the system, its acid dissolution will liberate
bility of the secondary mineralogy. bicarbonate (HCO− 3 ) to the solution (Fig. 5) (Sherlock et al., 1995) and is
Dissolution-equilibrium. able to neutralize one mole of protons per mole of calcite dissolved (Eq.
(8)). This is also true in an open system in the presence of CO2(atm),
RhomboclaseðdehydratedÞ : HFeðSO4 Þ2 þ 3H2 O
þ
¼ FeðOHÞ3 þ 2SO2−4 þ 4H ð5Þ Table 2
Relative reactivity in acid-neutralization capacity of minerals (after (Sverdrup, 1990).a

Relative
Transformation : jarosite⇒goethite : KFe3 ðSO4 Þ2 ðOHÞ6 ⇒3FeOðOHÞ reactivity
þ Kþ þ 2SO2− 4 þ 3H
þ
ð6Þ Group Typical minerals (pH 5)

1. Dissolving Calcite, dolomite, magnesite, aragonite, brucite, 1.0

2. Fast Anorthite, olivine, garnet, diopside, wollastonite, 0.6
Schwertmannite⇒goethite : Fe16 O16 ðOHÞ10 ðSO4 Þ3 weathering jadeite, nepheline, leucite, spodumene
þ 3.Intermediate Enstatite, augite, hornblende, tremolite, 0.4
þ 6H2 O⇒16FeOðOHÞ þ 3SO2−
4 þ 6H ð7Þ
weathering actinolite, biotite, chlorite, serpentine, talc,
epidote, zoisite, hedenbergite, glaucophane,
anthophyllite
3.2. Acid neutralization
4. Slow Plagioclase (Ab100–Ab30), kaolinite, vermiculite, 0.02
weathering montmorillonite, gibbsite
The produced protons will interact with the gangue minerals of the 5. Very slow K-feldspar, muscovite 0.01
rock and may undergo neutralization processes. Here, three mineral weathering
groups are the most relevant as potential neutralization agents: carbon- 6. “Inert”a Quartz, rutile, zircon 0.004

ates, hydroxides, and silicates. a

“Inert” is a relative expression, as in geological element cycling no mineral is inert.
 B. Dold / Journal of Geochemical Exploration 172 (2017) 120–132 125

It is additionally important to consider the differences in carbonate

mineralogy, as they might have different reaction kinetics, solubility
and for example siderite might even produce net acidity under certain
conditions. Siderite is frequently found in porphyry copper systems
(Dold and Fontboté, 2001). Thus a mineralogical approach is also essen-
tial for the estimation of the NP.
Siderite may act as a neutralizer under certain conditions, and under
other conditions as an acid producer. The following dissolution reaction
may take place and buffer the system at pH around 5:

FeCO3 þ 2Hþ ⟺Fe2þ þ H2 CO–3 ð11Þ

Combined with ferrous iron oxidation and ferrihydrate formation,

the overall siderite dissolution-oxidation reaction is:

4FeCO3 þ O2 þ 10H2 O⟺4HCO–3 þ 4FeðOHÞ3 þ 4Hþ ð12Þ

This indicates that under elevated pH, where bicarbonate is stable,

Fig. 4. Typical buffers and their pH in an ARD system (carbonates buffers and metal the total reaction from dissolution of siderite to the formation of
hydroxides-sulfates buffers). ferrihydrate gives a net acid production of one mole hydrogen ion per
mol siderite dissolved. However, under more acidic conditions, where
carbonic acid is stable, there will be no net acid production. But, it has
which enhance the dissolution of carbonates (Eq. (9)). For a detailed re- to be considered that when ferrous iron is liberated, at low pH, it can
view of the carbonate system refer to (Appelo and Postma, 2010; be mobilized even under neutral pH condition and, and under oxidizing
Stumm and Morgan, 1996). conditions downstream, it will oxidize and hydrolyze while liberating of
protons and producing ARD far from the source.
CaCO3 þ Hþ ⟺Ca2þ þ HCO–3 ð8Þ If ferrous iron is present in a solution containing bicarbonate, the for-
mation of siderite may occur by the following reaction:

CO2ðatmÞ þ H2 O þ CaCO3 ⇒Ca2þ þ 2HCO−

3 ð9Þ Fe2þ þ HCO–3 ⟺FeCO3 þ Hþ ð13Þ

If the pH is below pK1 (6.3), then carbonic acid is the dominant spe- This reaction buffers the pH at around 5–5.5 and its reaction path
cie in solution (Eq. (10)). may be an alternative to ferrous to ferric iron oxidation and the consec-
utive precipitation of Fe(III)oxyhydroxides. Geochemical studies by sev-
eral authors have shown that in tailings impoundments and AMD
CaCO3 þ 2Hþ ⟺Ca2þ þ H2 CO03 ð10Þ
affected aquifers the waters are frequently close to saturation or even
oversaturated with respect to siderite (Blowes et al., 1991, 1994;
Therefore, the carbonate system is able to neutralize 2 mol of pro- Morin and Cherry, 1986), and siderite precipitation could be observed
tons at low pH (b6.3), but it is important to consider, that at pH b 6.3 (Dold et al., 2009).
the calcite buffer is destroyed and the aqueous carbonate species in so- Before entering into details of the prediction techniques, it is helpful
lution become the only neutralization agent left in the system. There- to highlight the ARD prediction problem with three simplified
fore, if calcite is still available as a buffer in the system, the dominant examples.
carbonate specie will be bicarbonate, as calcite buffers the pH around
7. This is the crucial consideration in acid base accounting calculations, A.) Pyrite is the most abundant sulfide mineral in the Earth's upper
and will be highlighted below in the discussion of the calculation factor. crust and produces 4 mol of protons per mole of pyrite oxidized
(Eq. (1)). Calcite is the most soluble carbonate and buffers at neu-
tral pH, so that at these geochemical conditions one mole of pro-
[ C O 3 2 −] T O T = 1 0 0 . 0 0 m M tons can be neutralized per mole of calcite (Eq. (5)). If the molar
0 weight of pyrite (120 g/mol) and calcite (100 g/mol) is set as 1:1
H 2CO3 H C O 3− C O 32−
for simplification (accepting an error of 16.7%), it can be assumed
that a rock with N20 wt.% of pyrite will always produce ARD,
-2 even if the rest of the sample is pure calcite (20% × 4 = 80% cal-
cite needed to neutralize the protons from pyrite oxidation).
Log Conc.

Thus there are ore deposit types, which will always produce
-4
ARD due to their mineral composition (e.g. massive sulfide ore
bodies).
B.) Porphyry copper deposits have usually only 2 wt.% pyrite in aver-
-6
age, but due to the fact that the exploited ore has no or very little
carbonate content (around 8 wt,% of calcite would be needed)
-8 and the silicate assemblage has very low reactivity due to hydro-
thermal alterations (mainly K-Feldspar), the mine waste of this
2 4 6 8 10 12
deposit type is very prone to producing ARD (Dold and
pH Fontboté, 2001).
C.) A granite contains no sulfides and no carbonates, it is a rock only
Fig. 5. Distribution of solute species in the aqueous carbonate system. Log concentration composed of silicates and trace amounts of other mineral types;
vs. pH for 100 mM CO2− 3 . The pK1 is at 6.3 and pK2 at 10.3. i.e. quartz, feldspars, micas, and some accessory minerals as for
126 B. Dold / Journal of Geochemical Exploration 172 (2017) 120–132

example hornblende, apatite, sphene, among others (Takaya, ABA procedures remains the assumptions leading to the calculation fac-
2014). However, a weathered granite will most likely form an tors applied to the geochemical data. While it is generally accepted that
acidic environment (pH around 4.3–5.5) in the saprolite or soils one mole of oxidized pyrite produces 4 mol of protons, there is disagree-
(Liu et al., 2016); i.e. it produces ARD. This is due to the liberation ment between the amounts of protons neutralized by calcite. The fol-
of Al, and Fe through weathering process and the subsequent hy- lowing two Eqs. (14) and (15) are usually presented as the base for
drolysis of these elements, in combination with the lack of car- the calculation factor of 31.25 (e.g. GARD Guide (Verburg et al., 2009)
bonates and a feldspar assemblage of low relative reactivity. etc.) for the AP of a sample applied to the total sulfur or sulfide sulfur
concentrations of a sample.

4. Standard ARD prediction and its limitations CaCO3 þ 2Hþ ⟺Ca2þ þ CO2 þ H2 O ð14Þ

4.1. A historical background and mineralogical considerations

CaCO3 þ 2Hþ ⟺Ca2þ þ H2 CO3 ð15Þ
The so-called static testing of a rock sample for its potential to pro-
duce ARD tries, in a simplified geochemical way, to predict if a sample This factor results from the assumptions that one mole of calcite can
has an excess of AP or enough of NP to maintain the sample neutral. neutralize two moles of protons. As pyrite oxidation produces 4 mol of
As a consequence, this has direct effects on the liberation and mobility protons per mole of oxidized pyrite, and pyrite contains two moles of
of associated elements as stated above. This is normally done by a sim- sulfur, the final assumption is that the protons produced by the oxida-
plified geochemical quantification of sulfur in the bulk sample, which is tion of one mole of sulfur is neutralized by one mole of calcite. This
assumed to be associated to the sulfide mineralogy (expressed as pyrite leads to the following calculation:
equivalent) and the carbonate mineralogy quantified by titration
(expressed as calcite equivalent). 100ðmole weight calciteÞ =32ðmole weight sulfurÞ  10ðtCaCO3=1000tÞ ¼ 31:25:
Most of the standard methods applied in this field were developed
for other needs and were used for ARD prediction in sulfide metal min- The factor 10 at the end is to present an applicable unit how much
ing without considering the complexity of a sulfidic mineral assem- carbonate has to be added to the ARD forming material in tons of
blage. This is, for example, the case for the most known ABA test, the tCaCO3/1000 t (or g/kg or ‰) to buffer the system (sometimes the factor
so called Sobek test (Sobek et al., 1978), which was developed originally is presented without this adjustment as 3.125 (White et al., 1999).
for coal mining. Therefore, the assumption that all sulfur is associated to If it is accepted that calcite buffers the system at circum-neutral pH,
pyrite was more or less correct in this geological environment. This test and the general goal of mine waste management is to maintain the pH
was then widely applied in the last decades also for ABA calculations in of the material neutral, then the carbonate specie present would be bi-
metal sulfide mining, where sulfur can be associated to many different carbonate (at pH 7 N 85% HCO–3 (Appelo and Postma, 2010); see Fig. 5
minerals, which produce different amounts of acidity or produce no and Eq. (8)). Thus, Eq. (8) has to be considered for ABA calculations in-
acidity at all (e.g. anhydrite-gypsum or sulfide, like sphalerite or galena, stead of Eq. (10) (or 14 and 15), latter overestimating the NP of calcite
see Table 1). This problem was noted by (Lawrence et al., 1989), and by about 100%. Therefore, a calculation factor of 62.5 should be applied,
several intents for improvement for the AP determination (Lawrence as two moles of calcite are needed to neutralize the acidity from one
et al., 1989), as well for the NP estimation were published (Lawrence mole of sulfur:
and Scheske, 1997; Paktunc, 1999). For example, a separation of the sul-
fate mineralogy from the sulfide mineralogy was attempted through the 2  100ðmole weight calciteÞ =32ðmole weight sulfurÞ  10ðtCaCO3=1000tÞ ¼ 62:5:
dissolution of sulfates with HCl (Lawrence et al., 1989), which is now
better known as the “Modified Sobek test”. Both tests assuming that Therefore, the product of the sulfur concentrations multiplied with
only the sulfides are responsible for proton liberation. This assumption the factor 62.5 instead of 31.25 gives the AP for ARD prediction, by in-
has to be expanded, as Fe(III)oxyhydroxides and Fe(III)oxyhydroxide cluding the amount of protons possibly produced (assuming pyrite ox-
sulfates like schwertmannite and jarosite, ferrihydrite and goethite, idation as the only protons source) and the amount of protons able to be
minerals which are often present in the ore geology, tend to acidify neutralized by the carbonates (assuming the presence of calcite) at neu-
the solution due to equilibrium reactions (Alarcon et al., 2014; Dold, tral pH. As discussed above, it has to be considered a neutral pH for the
2010), so that this mineralogy must also be included in an ABA (Dold carbonate NP, as only when the carbonate buffer is depleted the pH will
and Weibel, 2013). Additionally, the HCl leach may attack some acid sol- decrease. This is different, if the amount of calcite has to be calculated to
uble sulfide or Fe(III) hydroxide sulfate minerals and has therefore a low neutralize ARD. In this case, when the solution is acidic (pH b 6.3), cal-
selectivity. cite can neutralize two moles of protons per mole of calcite and there-
A more selective separation of sulfur associated with the oxide frac- fore the factor 31.25 is correct (i.e. for treatment). This is contrary to
tion from the sulfide sulfur fraction can be done by a hot NH4-oxalate ARD prediction, were it is evaluated if the system goes from neutral to
leach (Dold and Fontboté, 2001), or by sequential extraction proto- acidic conditions in the future (Prediction). In a strict sense, before
cols established for a specific ore mineralogy e.g. (Dold, 2003a, ABA can be properly calculated, the paste-pH should be performed to
2003b), data which then gives the possibility of performing a high- evaluate the correct calculation factor based on the dominant carbonate
resolution ABA, including the quantification of Fe(III)hydroxides speciation.
and Fe(III)hydroxide sulfates and their respective AP. The Modified The next step is the quantification of the NP in order to be able to cal-
Sobek test (Lawrence and Scheske, 1997) can be seen as a two-step culate the final ABA. For standard ABA only the carbonate neutralization
extraction method, were only the sulfur content is analyzed. potential (CaNP) is considered. There are several approaches to quantify
As mentioned above, there can be a substantial difference if calcite or carbonates in a sample:
siderite is present as carbonate. For this reason an additional correction The easiest and cheapest method is the macroscopic quantification
for siderite might be necessary (Skousen et al., 1997). by a geologist in a hand specimen. However, this is a qualitative analisis
as the detection of calcite by eye is not easy and usually HCl 10% is used
4.2. ABA calculation to detect calcite in the field. This qualitative quantification follows the
same principles as applied in the Fizz-test (Bouzahzah et al., 2015b)
In addition to all of the problems related to the mineralogical consid- and can be very arbitrary, and is therefore not recommended for quan-
erations mentioned above, the key problem associated with standard tification of NP. Sobek et al. (1978) boiled the sample in HCl and later
 B. Dold / Journal of Geochemical Exploration 172 (2017) 120–132 127

titrate with NaOH in order to quantify the carbonate species in the sam- times to report the evolution of pH, Eh and dissolved elements like sul-
ple. (Lawrence et al., 1989) expose the sample at room temperature to fate and Fe, Cu, Zn, Cd, Pb, As, Cr, among others.
HCl for 24 h and then titrate with NaOH. A modified method for NP de- Although widely applied in the mining and consulting business for
termination (NP(pH 6) method) was developed taking in account to de- ARD prediction, these kinetic tests often do not give conclusive results
termine the “effective” NP or the calcium carbonate equivalent available as shown in a study at the Andina porphyry copper mine, Chile
in the sample to maintain the pH above 6, as discussed above (White et (Weibel et al., 2011). In this example, although all samples were classi-
al., 1999). This method implies titration with 1 N sulfuric acid until pH 6 fied by ABA and mineralogical studies as ARD producing (Fig. 8), in the
is reached, so that it might be the most realistic for carbonate NP applied standard humidity cell tests (ASTM D5744–96), no acidification
determination. due to sulfide oxidation could be observed. From 78 samples only 11.5%
Another possibility is to quantify the organic and mineral carbon by reached final pH between 3.5 and 4.5, because of equilibrium reactions
coulometric titration (E.g. Ströhlein® CS 702 or a LECO® CN-Analyzer). caused by the presence of Fe(III)hydroxides (these samples were locat-
The mineral carbon concentrations are assumed to be associated to the ed from the upper part of the ore deposit, suffering partial natural oxida-
carbonate minerals like calcite or dolomite and are expressed as CaCO3. tion) and not due to sulfide oxidation during the experiment. The
If siderite is of concern, addition of H2O2 is applied to promote Fe ox- observed sulfate liberation was associated to anhydrite-gypsum disso-
idation before back-titration to account for the above mentioned “sider- lution present in the primary ore mineral assemblage. 64% of the sam-
ite problem” in ARD prediction (Skousen et al., 1997). The carbonate ples maintained their circum-neutral pH (N6.3) and 14% reached final
quantification has then to be expressed in tCaCO3/1000 t, or g/kg, or pH between 5.5 and 6.3 buffered by siderite. The evolution of the acidity
‰ in order to be able to calculate the final ABA with the AP. In general concentrations, confirmed low ARD generation, were 73% of the sam-
term, a good knowledge of the carbonate mineralogy is needed for ples remained stable, 10.2 had scarce acidity liberation, and 16.6% had
ARD prediction. a fast contribution associated to the samples containing Fe(III) hydrox-
Although there are many different ways to present the ABA results, ides (Weibel et al., 2011).
the most common is the sulfide net acid potential (SNAP) or net neu- Some unpublished test reports from high sulfide deposit types show
tralization potential (NNP) a fast response of ARD formation in the recommended time frame.
However, the scarce literature published on humidity cell tests of coal
NNP ¼ NP−AP: (Banerjee, 2014), carbonate containing sulfide rich tailings from Joutel
tailings (Benzaazoua et al., 2004b) and of porphyry copper ores shows
clearly that during the recommended time frame, mostly no acidifica-
In the latter case, if the result is negative, this means that the mate- tion and increased element liberation could be observed, even if the
rial will form ARD, as there is an excess of AP (sulfides = pyrite) in re- ABA clearly suggests that the material should become acidic
lation to the carbonate content. If the result is positive, there is (Benzaazoua et al., 2004a; Filipek et al., 1999; Weibel et al., 2011). In a
enough NP (carbonates = calcite) to maintain the material neutral. In field study at the Talabre porphyry copper tailings at Chuquicamata,
some guidelines the range between −20 and +20 is called the “uncer- Chile, it was shown that the system need 3 to 5 years to acidify from
tain zone”. However, with good knowledge of the mineralogy, this is not the alkaline flotation conditions (pH 9.1) to a pH of 3.5–4 (Dold et al.,
an area of uncertainty. For example, if a material has 1 wt.% pyrite 2004; Smuda et al., 2014). This is in the order of data shown in the
(0.53 wt.% S) and no NP, the ABA will be −16,7 (calculated with the fac- GARD Guide of a cell test from an unknown ore type, where after
tor 31.25), but −33.4 with the factor 62.5, and this material will clearly 130 weeks the acidification and element release was observed. Thus, it
produce ARD. A granite without any sulfides and in absence of carbon- can be stated that the recommended time frame is not suitable for
ates will produce ARD although the NNP will be 0. Material from the ox- many types of mineral assemblage, and a flexible set-up in time of
idation zone of an ore deposit or tailings impoundment has an NNP of these tests has to be proposed in order to ensure to document the geo-
zero as there are no sulfides and carbonates left due to sulfide oxidation. chemical behavior properly.
However, this zone has a pH ranges between 2 and 4 due to the pres- Several parameters control the kinetics of sulfide oxidation, like
ence of Fe(III)hydroxides sulfates (Dold and Fontboté, 2001). microtextures (Parbhakar-Fox et al., 2013), textural relation-ships be-
In some cases the regulations apply a criterion for the ABA calcula- tween the minerals (Kwong, 1993), grain size, humidity, temperature,
tions ranging from 1:1.2 to 1:3 (Nevada, California and Montana, bacterial activity, availability of oxidants (e.g. oxygen or ferric iron)
respectively)(White et al., 1999), in order to ensure enough buffer ca- among others. Specifically it has shown by modifications of such cell
pacity. However, if the standard calculation factor of 31.25 was applied, tests, that a decrease in grain size, an increase of temperature and hu-
a criterion of 1:2 would represent simply the correct factor of 62.5 and midity results in an increase of the oxidation rates (Dold and Weibel,
no additional safety would be available. Only if a factor of 1:3 was ap- 2013; Dold et al., 2011). Too high water saturation might lower the ox-
plied, there would be some excess of buffer capacity for additional idation rates (Bouzahzah et al., 2015a).
safety. Thus, there are different possibilities to enhance the kinetics in the
laboratory. Although the wet-dry cycle was introduced to simulate en-
4.3. Kinetic testing hanced weathering cycles, the bacterial community oxidized most effi-
ciently in a modified cell test, under constant experimental conditions.
Above, the static tests were discussed in order to predict if a sample Best response was reached at constant 40 °C and 95% humidity during
has the potential to acidify the geochemical system or if it will maintain the whole cycle and only flushing on the 14th day. The results have
neutral or even go alkaline. Kinetic testing is used to investigate in shown a more rapid response in element liberation and pH drop than
which time frame this will occur. with the ASTM D5744-96 approach (Dold and Weibel, 2013; Dold et
The standard tests (e.g. humidity cells; ASTM D5744-96) use col- al., 2011), which was run in parallel.
umns or cells, were the samples are exposed to oxidation, and is flushed Finally, in light of all the problems of the prediction the kinetics of
with deonized water after a certain time frame and the solution is ana- ARD formation, it can be argued whether or not it is necessary to use ki-
lyzed for the solutes, representing the reaction products of the oxidation netic testing. If the mineralogy and geochemistry suggest clearly that
process. In the mostly applied ASTM D5744-96 humidity cell test, a one ARD will be an issue in the future of an ore deposit, immediate action
kilo sample with a grain size of b6.3 mm is exposed to 3 days humid air, must be taken to prevent ARD formation and to be able to control all ef-
3 days dry air and the seventh day the system is flushed with water fluent in future. Once ARD is formed it is extremely difficult to control it,
which is then analyzed for physical-chemical parameters and element thus prevention from the beginning of any mining activity must be the
concentrations. These cycles are recommended to be repeated 20–25 goal.
128 B. Dold / Journal of Geochemical Exploration 172 (2017) 120–132

For example, it is easy to predict, only on the base of mineralogical or diorites do not have sulfides, thus their original AP is zero. They usu-
composition that a porphyry copper tailings impoundment will first ally also do not have any carbonates, so that their CaNP is also zero. An-
produce a sulfate plume (due to equilibrium reactions with gypsum-an- desite or diorites contain plagioclases like albite - anorthite, latter a
hydrite), and later ARD (due to sulfide oxidation), both infiltrating and feldpar with good FNP, meaning that this original rock suite would be
polluting the groundwater (Dold, 2014). With this knowledge preven- able to neutralize some protons. However, the metals, sulfur, and acidity
tion measures must be in place, before operation starts. In this regard, that were provided to the system during hydrothermal alteration, re-
it is incomprehensible that there are still many mining countries, with sulted in the precipitation of sulfide minerals like for example pyrite,
new legislation for environmental protection in place, which still do chalcopyrite, enargite, and arsenopyrite, all of which are potential acid
not demand a basal impermeabilization for mine waste deposits in producers. These processes are also accompanied by a strong change
order to prevent infiltration of contaminated plumes into the of the gangue mineralogy. For example, the plagioclases are altered to
groundwater. muscovite (quartz-sericitic alteration) or to potassic feldspar (potassic
alteration) or kaolinite (argillic alteration), minerals with very low reac-
5. Sample selection criteria for ARD prediction tivity kinetics and buffer capacity (Table 2) and therefore not helpful for
ARD neutralization. Keeping this in mind, for ARD prediction of an ore
It is most critical for ARD prediction, to select the correct samples for body it is key that for sample selection procedure the hydrothermal al-
this purpose. In a mine waste environment, surficial samples from old teration mineralogy classification is used, and not the original petrolog-
oxidized mine tailings or waste rock dumps do not represent the full ical or lithological classification (e.g. diorite or andesite), as sometimes
AP and NP, as they might have been affected by weathering processes. observed.
Therefore, samples must be extracted from depth, e.g. by drilling, in - Thus, using the original petrological classification as sample selec-
order to obtain samples for ARD prediction. tion criteria might also mislead geochemical modeling interpretations.
Also the geological and petrological classification as selection criteria For example, it is sometimes observed, that samples for ABA analysis
is most relevant for correct ARD prediction. As discussed above, the vast are selected from petrological units like andesites or diorites, but no fur-
majority of base metal production has its origin in igneous ore deposit ther mineralogical characterization is done and only the standard ABA
types. N 90% of the giant metal accumulations on earth relied on water data is used for prediction and setup for a geochemical model to predict
as formation agents (Laznicka, 1999), such that hydrothermal processes future ARD quality. This gives way to the possibility of selecting samples
are essential for most ore forming processes. Hereby, metal rich hydro- outside the mineralized area (often no location of the samples in the ore
thermal solutions interact with the host rock units and physio-geo- deposit is given), and to assume that the feldspar specie is anorthite,
chemical changes (e.g. T, P, pH, redox) induce mineral precipitation which has acceptable neutralization properties. If equilibrium modeling
and alteration. These processes can completely change the original min- uses the FNP without considering the reaction kinetics of the different
eral composition of the host rock, i.e. adding elements or minerals and/ feldspar species, unrealistic predictive geochemical modeling is the re-
or altering minerals or transporting elements away (Fig. 6), so that sult and as a conclusion, it maybe stated that the unit will not produce
sometimes only textural signs can be used for classification of the orig- ARD.
inal rock unit (e.g. porphyric texture). For example when hydrothermal These examples highlight that it is crucial to understand which min-
fluids interact with andesites or diorites along an active subduction eralogy will be exposed to a specific geochemical condition in order to
margin, the alteration completely changes the original mineralogy and predict ARD formation, or more generally, to predict mineral stability
a Cu-Mo-Au porphyry deposit may form. From the ARD perspective, and subsequent element liberation processes and changes in the geo-
this mineralogical change is dramatic and crucial. Unaltered andesites chemical system.

Fig. 6. A) Typical suite of andesitic flows intruded by a dioritic porphyry along the South American west coast. The ABA of these rock units would be zero. B) Change in mineral composition
due to hydrothermal alteration with complete change of the ABA to negative values (excess of sulfides). Only in the external part of the ore deposit, the precipitation of calcite in the
propylitic alteration zone adds CaNP to the original rock units.
 B. Dold / Journal of Geochemical Exploration 172 (2017) 120–132 129

6. Suggested characterization methodology for ARD prediction in the different methodologies and what data is given by the different
techniques, these and the subsequent differences in the NNP are
The key information to predict ARD or any element liberation of a shown in Fig. 7 and Table 3.
geological material, is the mineralogical composition of the sample. Although a good calibration and quality control for each ore and spe-
The mineralogy is the pool of potentially liberated and neutralization cial care for sample preparation is necessary for accurate quantification
of protons through the above presented geochemical processes, and of the mineralogy, these relatively new techniques are able to produce
the subsequent liberation of elements to the environment in form of the necessary data in reasonable time frames and competitive prices
ARD or also as neutral to alkaline solutions. Variables like climate, mi- (Nowadays an analysis by Automated Quantitative Mineralogy costs
crobial interactions, grain size, textures, relationships between minerals in the same range as a mineralogical interpretation of a thin section by
(e.g. electrochemical bridges), play also a important role in these pro- a qualified mineralogist). Additionally they can quantify trace mineral
cesses, but mainly control the kinetics of the processes leading to ARD, amounts, not detectable with any other analytical technique (Menzies
not the overall acid-base balance. Therefore, the efforts should be con- et al., 2015). This can be specially relevant if some trace elements are as-
centrated in identifying the minerals present in a deposit, their quanti- sociated to soluble minerals like metal sulfates or chlorides. However, in
fication, their relations and whether these minerals have potentially order to be able to predict trace element behavior, a more detailed geo-
hazardous trace elements associated. chemical characterization should be additionally performed on the min-
In most cases, a simple overview of the ore deposit type and trace el- erals. This can be done for example by sequential extractions, where the
ement composition gives a good first estimate if it will produce ARD and trace element concentrations of reactive mineral groups can be quanti-
which trace elements might be critical for the environment, and actions fied (Dold, 2003b). Additionally, sequential extractions have advantage
can be taken for ARD control or prevention (e.g. impermeabilization of in terms of turnaround time and prices in relation to some of the other
the waste deposition sites and solution control), or further investiga- techniques and give data for high- resolution acid-base accounting. Al-
tions conducted. though, no prices for the different analysis in commercial labs can be
The first step should always be a thorough revision of the literature given here, as they change with country and demand, the standard
of the ore deposit. Publications in scientific journals (e.g., Mineralium ABA tests are very expensive in relation of the information they deliver.
Deposita, Economic Geology, Ore Geology Reviews, Journal of Geo- For example the modified Sobek test (Lawrence), which is essentially a
chemical Exploration, among others) usually have excellent data on two step sequential extraction in which only sulfur is analyzed with an
the mineralogical characterization of the ore deposits. However, only additional titration for the CaNP costs usually half or even more of the
seldom they do present quantitative mineralogical analysis data. But price, which costs a seven step sequential extraction with analysis of
as shown above, with the information of the ore deposit type and the 31 elements in each leach. Thus, the geochemical information obtained
specific minerals present in an ore, a good first estimate of the ARD be- by sequential extractions is much cheaper and gives much more infor-
havior can be given; see also Plumlee (1999). mation on the sample in relation to simple standard ABA procedures.
The second step, should be a characterization to obtain quantitative Additional advanced analytical tools for geochemical characterization
mineralogical data (can nowadays be done by automated quantitative might be necessary like SEM-EDS, microprobe or Laser-ICP-MS, where
mineralogical analysis; e.g. QEMSCAN® or MLA). trace metal concentration of specific minerals can be detected and
Many modern mining operations already use automated quantita- quantified and therefore their liberation under specific geochemical
tive mineralogy and mineral liberation analysis for geometallurgical conditions can be predicted.
purposes to increase the recovery of their extraction process (Cruz et In Fig. 7 only quantitative mineralogy is able to account on the differ-
al., 2012). The very same data can and should be used for ARD predic- ent sulfide minerals, which produce different amounts of protons
tion (Dold and Weibel, 2013), and no additional standard ABA tests through their oxidation. Also the AP from Fe(III)hydroxides and sulfates
are needed at no additional costs. As mentioned above, the standard can be quantified, as well the differences in NP between calcite and sid-
ABA tests have to be seen only as a very simple, rough approach to at- erite and the different feldspars can be considered. Sobek would ac-
tempt to quantify the ARD susceptible mineralogy through the use of count any sulfur as sulfide sulfur in form of pyrite and siderite as full
geochemical data. In order to highlight which minerals are considered CaNP. It also does not account the AP from iron hydroxides and sulfates.

Automated Quantitative Mineralogy Sobeck Lawrence Skousen

e.g. QEMSCAN - MLA

sulfides (SAP) metal-hydroxides (sulfates) sulfates carbonates (CaNP) feldspars (FNP)

py fh-gt gy ca an

cp sh sd pls
sfs
gn jt kfs

Fig. 7. Schematic illustration which minerals are detected and/or considered in the different techniques and calculated as acid potential or neutralization potential. Yellow colors indicate
acid potential, blue neutralization potential and grey and white none of both. Abbreviations: py = pyrite; cp = chalcopyrite; gn = galena; fh = ferrihydrite; gt = goethite; sh =
schwertmannite; jt = jarosite; gy = gypsum; sfs = sulfosalts; ca = calcite; sd = siderite; an = anorthite; pls = plagioclase; kfs = K-feldspar. SAP = sulfide acid potential; CaNP =
carbonate neutralization potential; FNP = feldspar neutralization potential.
130 B. Dold / Journal of Geochemical Exploration 172 (2017) 120–132

Table 3
Overview of a typical sample from a porphyry copper deposit and the results of the different standard ABA approaches in relation to the quantitative mineralogy by QEMSCAN®.

Quantitative Sobek Lawrence Skousen

mineralogy (siderite
e.g. QEMSCAN corrected)

31.25 62.5 31.25 62.5 31.25 62.5 31.25 62.5

Mineralogy wt.% wt.% wt.% SAP SAP SAP SAP SAP SAP SAP SAP
S S (‰CaCO3)

Pyrite 53.3 1 0.53 16.66 33.31 16.66 33.31 16.66 33.31 16.66 33.31
Chalcopyrite 34.87 1 0.35 5.45** 10.9* 10.90 21.79 10.90 21.79 10.90 21.79
*
Sphalerite 32.86 0.5 0.16 0 0.00 5.13 10.27 5.13 10.27 5.13 10.27
Galena 13.38 0.5 0.07 0 0.00 2.09 4.18 2.09 4.18 2.09 4.18
Gypsum– 23.52 3 0.71 0 0.00 22.05 44.10 0 0 0 0
Anhydrite
Total SAP 16.66 33.31 56.83 113.66 34.78 69.56 34.78 69.56
CaNP
(‰)*
wt.% wt.% wt.% (‰CaCO3) ‰
C C
Calcite 12 0.5 0.06 5.30 5.30 5.30 5.30 5.00 5.30 5.30 5.30
Siderite 10.3 6 0.62 0 0.00 54.57 54.57 54.57 54.57 0.00 0.00
Total CaNP 5.30 5.30 59.87 59.87 59.57 59.87 5.30 5.30

ABA (NNP) –11.36 –28.01 3.04 –53.79 24.79 –9.69 –29.48 –64.26
* Calculation factor for CaNP is 88.3.
** For chalcopyrite only 2 mol of H+ is considered factor is 31.25/2 or 62.5/2.
Green: ABA calculated based on quantitative mineralogy. Orange: ABA calculated after mostly applied prediction protocols.

Lawrence improve in the sense that it does not account gypsum or an- for all 3 standard tests the factor 62.5 instead of 31.25 were applied, in
hydrite as sulfide sulfur, but the AP from sulfosalts and Fe(III)hydroxide all cases the NNP would be negative. The same trend can also be ob-
sulfates is also not considered. Lawrence does consider siderite as full served in a data set of 200 samples from the Andina porphyry copper
NP. Skousen accounts for the siderite problem and corrects the NP. mine in Fig. 8. This clearly shows the importance of a correct mineral
In Table 3 the different ABA (NNP) interpretations resulting through quantification and correct geochemical calculations in order to properly
the characterization of a sample by the different test methods are predict the ARD potential.
highlighted: quantitative mineralogy with QEMSCAN, Sobek (Sobek et However, it is interesting to observe that sometimes, although a
al., 1978), Lawrence (Lawrence et al., 1989), and siderite corrected strong data set of quantitative mineralogy and geochemical data exists
(Skousen et al., 1997). in a project, responsible managers spend additional money for the stan-
This hypothetical sample, typical for a porphyry copper deposit, dard ABA tests, which illustrates how well established these tests are in
shows which of the minerals or mineral groups are addressed in the dif- the mining industry and state agencies, despite all the above-mentioned
ferent methodologies. On the left side is the mineralogy detected by mi- drawbacks and errors.
croscopy and quantified by QEMSCAN. From the detected sulfide
minerals pyrite is producing 4 mol of protons by its oxidation, chalcopy-
rite 2 mol, sphalerite and galena zero, and the sulfates gypsum-anhy- 7. Conclusions
drite also produce zero. These differences can be considered, if
quantitative mineralogy is available, while the Sobek method assumes Standard static acid rock drainage prediction methodologies are
all sulfur containing minerals as acid producers and thereby overesti- based on simple geochemical tests with the goal to quantify potential
mates the AP strongly. With respect to carbonate neutralization poten- acid production and NP of a geological sample. Due to the simplicity of
tial, Sobek strongly overestimates this value, as siderite is also included. the approach and it's basic assumptions, these tests have many prob-
The Lawrence method takes into account sulfates like gypsum, which do lems and drawback, not able to account for the complexity of the min-
not produce acidity, but also overestimates the sulfur AP, as it considers eral assemblage of an ore deposit, and therefore not able to accurately
sphalerite and galena as acid producers. Lawrence also overestimates predict ARD formation. Also, standard kinetic testing is not able to pro-
the CaNP as siderite is considered as a neutralizer. Even if Lawrence is duce concordant data for ARD prediction in the proposed time frames.
corrected for siderite (Skousen), it overestimates the AP, and applies On the other hand, mineral characterization and automated mineral
the wrong calculation factor as shown by the difference with the stan- quantification has experienced in the last decades important develop-
dard calculation coefficient of 31.25 and the suggested 62.5 (see Section ment (e.g. QEMSCAN® and MLA). These data are increasingly produced
4.2). in modern mining operation, nowadays mainly for geometallurgical
As a result, with quantitative mineralogy it is predicted that the sam- purposes to increase the recovery from the ore. It can be predicted
ple has an NNP of −28.1 tCaCO3/1000 t (‰). Calculated with the wrong, that this data will be available in the near future in most competitive
but widely used factor (31.25) the conclusion is that it lies in the uncer- mining operations. Thus, it is suggested to use the very same data of
tainty area with tendency to ARD foramtion (− 11.36‰). In contrast, quantitative mineralogy for ARD prediction with no additional costs
Sobek and Lawrence would predict that the sample is not ARD produc- and testing, instead the very limited and problematic standard ARD pre-
ing. Only if corrected for siderite, the NNP is also negative (−29.48). If, diction tests.
 B. Dold / Journal of Geochemical Exploration 172 (2017) 120–132 131

Samples

Fig. 8. Left: Blue bars: Percentage of the 78 samples plotting into the generator, uncertain (−20–20) and non-generator ABA from the Andina porphyry copper deposit (Weibel et al.,
2011), obtained by the modified ABA (Lawrence et al., 1989) and calculated with 31.25. Red bars: high-resolution ABA obtained from sequential extraction data (Dold, 2003a, 2003b)
from 200 samples from Andina and calculated with 62.5 (Dold et al., 2011). Right: Quantitative mineralogy of the same 200 samples obtained by QEMSCAN®, showing clearly that
calcite are very minor traces and only siderite is available as carbonate NP, thus, all samples are clearly ARD producers.

However, if standard testing is applied for ARD prediction, following Blowes, D.W., Reardon, E.J., Jambor, J.L., Cherry, J.A., 1991. The formation and potential im-
portance of cemented layers in inactive sulfide mine tailings. Geochim. Cosmochim.
improvements should be applied: Acta 55, 965–978.
Bouzahzah, H., Benzaazoua, M., Bussiere, B., Plante, B., 2014. Prediction of acid mine drain-
1. Based on the paste-pH of the sample, the calculation factor of the sul- age: Importance of mineralogy and the test protocols for static and kinetic tests. Mine
fide AP should be chosen: If the paste pH is N6.3 the factor 62.5 must Water Environ. 33, 54–65.
Bouzahzah, H., Benzaazoua, M., Bussière, B., Plante, B., 2015a. ASTM normalized humidity
be applied (i.e. for ARD prediction). If the paste pH is b6.3, 31.25 is cell kinetic test: protocol improvements for optimal sulfide tailings reactivity. Mine
the correct factor for calculation, i.e. for ARD treatment. Water Environ. 34, 242–257.
Bouzahzah, H., Benzaazoua, M., Plante, B., Bussiere, B., 2015b. A quantitative approach for
2. The separation of the sulfate sulfur from the sulfide sulfur is the cor-
the estimation of the “fizz rating” parameter in the acid-base accounting tests: a new
rect way, as applied by Lawrence et al. (1989). However, the use of adaptations of the Sobek test. J. Geochem. Explor. 153, 53–65.
HCl is not very selective, as many sulfides may suffer partial dissolu- Chotpantarat, S., 2011. A review of static tests and recent studies. Am. J. Appl. Sci. 8,
400–406.
tion. A leach with 0.2 M NH4-oxalate, pH 3, at 80° for 2 h has shown to
Colmer, A.R., Hinkle, M.E., 1947. The role of microorgamisms in acid mine drainage. Sci-
completely dissolve all ferric iron oxides and hydroxides sulfate min- ence 106, 253–256.
erals present in sulfide ore deposits (Dold, 2003b), and is therefore Cruz, J., Bustos, C., Martínez, C., Suazo, H., 2012. Daily Mineralogical Control of Andina Di-
recommended as more selective leach for this approach. The use of vision Concentrator CODELCO CHILE. Geomet2012.
Dold, B., 2003a. Dissolution kinetics of schwertmannite and ferrihydrite in oxidized mine
sequential extractions can improve importantly the resolution and samples and their detection by differential X-ray diffraction (DXRD). Appl. Geochem.
precision of the ARD prediction (Dold, 2003b; Dold and Weibel, 18, 1531–1540.
2013). Dold, B., 2003b. Speciation of the most soluble phases in a sequential extraction proce-
dure adapted for geochemical studies of copper sulfide mine waste. J. Geochem.
3. For the determination of the NP, the Modified NP(pH 6) Method Explor. 80, 55–68.
(White et al., 1999) is the most realistic. It considers the real pH con- Dold, B., 2010. Basic concepts in environmental geochemistry of sulfide mine-waste man-
ditions were the neutralization processes take place for ARD predic- agement. In: Kumar, S. (Ed.), Waste Management, pp. 173–198 (http://www.
intechopen.com/books/show/title/waste-management).
tion, and considers the carbonate speciation in the system, and is Dold, B., 2014. Evolution of acid mine drainage formation in sulphidic mine tailings. Min-
therefore recommended. erals 4 (2), 621–641.
Dold, B., Fontboté, L., 2001. Element cycling and secondary mineralogy in porphyry cop-
per tailings as a function of climate, primary mineralogy, and mineral processing.
Acknowledgements J. Geochem. Explor. 74, 3–55.
Dold, B., Weibel, L., 2013. Biogeometallurgical pre-mining characterization of ore de-
posits: an approach to increase sustainability in the mining process. Environ. Sci.
I would like to thank Simon Catchpole and Robert Pooler for their
Pollut. Res. 20, 7777–7786.
thorough revisions, suggestions and comments, which strongly im- Dold, B., Vogt, M.-L., Spangenberg, J.E., Kobek, M., 2004. Hydrogeochemistry of the active
proved the manuscript. The author acknowledges the constructive com- tailings impoundment Talabre, Chuquicamata, Chile. Geochim. Cosmochim. Acta 68,
ments of the anonymous reviewers to improve the manuscript. A450.
Dold, B., Wade, C., Fontbote, L., 2009. Water management for acid mine drainage control
at the polymetallic Zn-Pb-(Ag-Bi-Cu) deposit of Cerro de Pasco, Peru. J. Geochem.
References Explor. 100, 133–141.
Dold, B., Weibel, L., Cruz, J., 2011. New Modified Humidity Cells Test for Acid Rock Drain-
Abbott, M.B., Wolfe, A.P., 2003. Intensive pre-Incan metallurgy recorded by lake sedi- age Prediction in Porphyry Copper Deposits. EnviroMine, Santiago de Chile.
ments from the Bolivian Andes. Science 301, 1893–1895. Elberling, B., Asmund, G., Kunzendorf, H., Krogstad, E.J., 2002. Geochemical trends in
Alarcon, R., Gaviria, J., Dold, B., 2014. Liberation of adsorbed and co-precipitated arsenic metal-contaminated fiord sediments near a former lead-zinc mine in West Green-
from jarosite, schwertmannite, ferrihydrite, and goethite in seawater. Minerals 4 land. Appl. Geochem. 17, 493–502.
(2), 603–620. Epp, J., 2016. X-ray diffraction (XRD) techniques for materials characterization. Materials
Appelo, C.A.J., Postma, D., 2010. Geochemistry, Groundwater and Pollution. 2nd edition. Characterization Using Nondestructive Evaluation (NDE) Methods, pp. 81–124.
CRC, Balkema, Leiden. Filipek, L.H., VanWyngarden, T.J., Papp, C.S.E., Curry, J., 1999. A multi-phase apprach to
Banerjee, D., 2014. Acid drainage potential from coal mine wastes: environmental assess- predict acid production from porphyry copper-gold waste rock in an arid montane
ment through static and kinetic tests. Int. J. Environ. Sci. Technol. 11, 1365–1378. environment. In: Filipek, L.H., Plumlee, G.S. (Eds.), The Environmental Geochemistry
Benzaazoua, M., Bussière, B., Dagenais, A.M., Archambault, M., 2004a. Kinetic tests com- of Ore Deposits, Part B: Case Studies and Research TopicsReviews in Economic Geol-
parison and interpretation for prediction of the Joutel tailings acid generation poten- ogy. Society of Economic Geologists, pp. 433–444.
tial. Environ. Geol. 46, 1086–1101. Gerst, M.D., 2008. Revisiting the cumulative grade-tonnage relationship for major copper
Benzaazoua, M., Bussiere, B., Dagenais, A.M., Archambault, M., 2004b. Kinetic tests com- ore types. Econ. Geol. 103, 615–628.
parison and interpretation for prediction of the Joutel tailings acid generation poten- Gottlieb, P., Wilkie, G., Sutherland, D., Ho-Tun, E., Suthers, S., Perera, K., Jenkins, B.,
tial. Environ. Geol. 46, 1086–1101. Spencer, S., Butcher, A., Rayner, J., 2000. Using quantitative electron microscopy for
Bergh, L.G., Yianatos, J.B., 2011. The long way toward multivariate predictive control of process mineralogy applications. JOM 52, 24–25.
flotation processes. J. Process Control 21, 226–234. Jacobs, J.A., Lehr, J.H., Testa, S.M., 2014. Acid Mine Drainage, Rock Drainage, and Acid Sul-
Blowes, D.W., Ptacek, C.J., Jambor, J.L., 1994. Secondary minerals and acid mine-water fate Soils: Causes, Assessment, Prediction, Prevention, and Remediation.
chemistry. In: Jambor, J.L., Blowes, D.W. (Eds.), Short course handbook on environ- Jambor, J.L., Blowes, D.W., 1994. Environmental geochemistry of sulfide mine-wastes. In:
mental geochemistry of sulfide mine-waste. 22. Mineralogical Association of Canada, Jambor, J.L. (Ed.), Short Course Series. Mineralogical Association of Canada, Nepean,
Nepean, pp. 367–379 p. 438.
132 B. Dold / Journal of Geochemical Exploration 172 (2017) 120–132

Kwong, Y.T.J., 1993. Prediction and Prevention of Acid Rock Drainage from a Geological Plumlee, G.S., Logsdon, M.J., 1999. The Environmental Geochemistry of Ore Deposits. Part
and Mineralogical Perspective. MEND, p. 47. A: Processes, Techniques, and Health Issues. Reviews in Economic GeologySociety of
Langmuir, D., 1997. Aqeous Environmental Geochemistry. Prentice Hall. Economic Geologist.
Lawrence, R.W., Scheske, M., 1997. A method to calculate the neutralization potential of Poling, G.W., Ellis, D.V., 1995. Importance of geochemistry: the Black Angel lead-zinc
mining wastes. Environ. Geol. 32, 100–104. mine, Greenland. Mar. Georesour. Geotechnol. 13, 101–118.
Lawrence, R.W., Poling, G.W., Ritcey, G.M., Marchant, P.B., 1989. Assessment of Predictive Rockstrom, J., Steffen, W., Noone, K., Persson, A., Chapin, F.S., Lambin, E., Lenton, T.M.,
Methods for the Determination of AMD Potential in Mine Tailings and Waste Rock, Scheffer, M., Folke, C., Schellnhuber, H.J., Nykvist, B., de Wit, C.A., Hughes, T., van
Tailings and Effluent Management. Pergamon Press, New York, Halifax, pp. 317–331. der Leeuw, S., Rodhe, H., Sorlin, S., Snyder, P.K., Costanza, R., Svedin, U., Falkenmark,
Laznicka, P., 1999. Quantitative relationships among giant deposits of metals. Econ. Geol. M., Karlberg, L., Corell, R.W., Fabry, V.J., Hansen, J., Walker, B., Liverman, D.,
94, 455–473. Richardson, K., Crutzen, P., Foley, J., 2009. Planetary boundaries: exploring the safe
Laznicka, P., 2014. Giant metallic deposits-a century of progress. Ore Geol. Rev. 62, operating space for humanity. Ecol. Soc. 14.
259–314. Sherlock, E.J., Lawrence, R.W., Poulin, R., 1995. On the neutralization of acid rock drainage
Liu, W., Liu, C., Brantley, S.L., Xu, Z., Zhao, T., Liu, T., Yu, C., Xue, D., Zhao, Z., Cui, L., Zhang, Z., by carbonate and silicate minerals. Environ. Geol. 25, 43–54.
Fan, B., Gu, X., 2016. Deep weathering along a granite ridgeline in a subtropical cli- Singer, D.A., 1995. World-cass base and precious-metal deposits - a qantitative-analysis.
mate. Chem. Geol. 427, 17–34. Econ. Geol. Bull. Soc. Econ. Geol. 90, 88–104.
Lottermoser, B.G., 2010. Mine Wastes. Characterization, Treatment and Environmental Singer, P.C., Stumm, W., 1970. Acid mine drainage: the rate-determining step. Science
Impacts, third edition. 167, 1121–1123.
Matsumoto, S., Shimada, H., Sasaoka, T., 2016. The key factor of acid mine drainage (AMD) Skousen, J., Renton, J., Brown, H., Evans, P., Leavitt, B., Brady, K., Cohen, L., Ziemkiewicz, P.,
in the history of the contribution of mining industry to the prosperity of the United 1997. Neutralization potential of overburden samples containing siderite. J. Environ.
States and South Africa: a review. Nat. Resour. 7, 445–460. Qual. 26, 673–681.
Menzies, A., Álvarez, E., Belmar, M., Belma, A., Marquardt, M., Albornoz, A., Vargas, M., Smuda, J., Dold, B., Spangenberg, J.E., Friese, K., Kobek, M.R., Bustos, C.A., Pfeifer, H.-R.,
Dold, B., 2015. Quantification of trace REE-minerals using automated mineralogy. 2014. Element cycling during the transition from alkaline to acidic environment in
Chilean Geological Congress, La Serena. Chile. an active porphyry copper tailings impoundment, Chuquicamata, Chile. J. Geochem.
Morin, A.K., Cherry, J.A., 1986. Trace amoumts of siderite near a uranium-tailings im- Explor. 140, 23–40.
poundments, Elliot Lake, Ontario, and its implications in controlling contaminant mi- Sobek, A.A., Schuller, W.A., Freeman, J.R., Smith, R.M., 1978. Field and laboratory methods
gration in a sand aquifer. Chem. Geol. 56, 117–134. applicable to overburden and mine soils. EPA 600/2-78-054, p. 203.
Mudd, G.M., Weng, Z., Jowitt, S.M., 2013. A detailed assessment of global Cu resource Stumm, W., Morgan, J.J., 1996. Aquatic Chemistry. 3rd ed. Wiley, New York.
trends and endowments. Econ. Geol. 108, 1163–1183. Sverdrup, H.U., 1990. The Kinetics of Base Cation Release Due to Chemical Weathering.
Nordstrom, D.K., 2000. Advances in the hydrogeochemistry and microbiology of acid Lund University Press, Lund.
mine waters. Int. Geol. Rev. 42, 499–515. Takaya, Y., 2014. Which constituent mineral is dominant in granite weathering? A solu-
Norecol Environmental Consultants Ltd., 1991. New Methods for Determination of Key tion-sided approach through a laboratory experiment. Geoderma 230-231, 204–211.
Mineral Species in Acid Generation Prediction by Acid Base Accounting. MEND, p. 67. Verburg, R., Bezuidenhout, N., Chatwin, T., Ferguson, K., 2009. The global acid rock drain-
Owen, J.R., Kemp, D., 2013. Social licence and mining: a critical perspective. Res. Policy 38, age guide (GARD guide). Mine Water Environ. 28, 305–310.
29–35. Weibel, L., Dold, B., Cruz, J., 2011. Application and limitation of standard humidity cell
Paktunc, A.D., 1999. A method to calculate the neutralization potential of mining wastes; tests at the Andina porphyry copper mine. SGA Biennal Meeting, Antofagasta, Chile.
discussion. Environ. Geol. 38, 82–84. CODELCO, Chile.
Parbhakar-Fox, A., Lottermoser, B.G., 2015. A critical review of acid rock drainage predic- White III, W.W., Jeffers, T.H., 1994. In: Alpers, C.N., Blowes, D.W. (Eds.), Chemical predic-
tion methods and practices. Miner. Eng. 82, 107–124. tive modeling of acid mine drainage from metallic sulfide-bearing waste rock. Envi-
Parbhakar-Fox, A., Lottermoser, B., Bradshaw, D., 2013. Evaluating waste rock mineralogy ronmental Geochemistry of Sulfide Oxidation, Washington, pp. 608–630.
and microtexture during kinetic testing for improved acid rock drainage prediction. White III, W.W., Lapakko, K.A., Cox, R.L., 1999. Static-test methods most commonly used
Miner. Eng. 52, 111–124. to predict acid mine drainage: practical guidelines for use and interpretation. In:
Perner, K., Leipe, T., Dellwig, O., Kuijpers, A., Mikkelsen, N., Andersen, T.J., Harff, J., 2010. Plumlee, G.S., Logsdon, M.J. (Eds.), The Environmental Geochemistry of Ore Deposits.
Contamination of arctic Fjord sediments by Pb-Zn mining at Maarmorilik in central Part A: Processes, Techniques, and Health Issues. Reviews in Economic Geology,
West Greenland. Mar. Pollut. Bull. 60, 1065–1073. pp. 325–338.
Plumlee, G.S., 1999. The environmental geology of mineral deposits. In: Plumlee, G.S.,
Logsdon, M.J. (Eds.), The Environmental Geochemistry of Ore Deposits. Part A: Pro-
cesses, Techniques, and Health Issues. Reviews in Economic Geology, pp. 71–116.