Applied
Geochemistry
Applied Geochemistry 21 (2006) 377–403
www.elsevier.com/locate/apgeochem
Rate controls on the chemical weathering of natural
polymineralic material. II. Rate-controlling mechanisms
and mineral sources and sinks for element release from four
UK mine sites, and implications for comparison of
laboratory and field scale weathering studies
K.A. Evans
a
a,*
, D.C. Watkins a, S.A. Banwart
b
Groundwater Protection and Restoration Group, Department of Civil Engineering, University of Sheffield, Mappin St, Sheffield S1 3JD, UK
b
Camborne School of Mines, Tremough Campus, Treliever Road, Penryn, Cornwall TR10 9EZ, UK
Received 9 May 2005; accepted 24 October 2005
Editorial handling by R. Fuge
Abstract
Predictions of mine-related water pollution are often based on laboratory assays of mine-site material. However, many
of the factors that control the rate of element release from a site, such as pH, water–rock ratio, the presence of secondary
minerals, particle size, and the relative roles of surface-kinetic and mineral equilibria processes can exhibit considerable
variation between small-scale laboratory experiments and large-scale field sites.
Monthly monitoring of mine effluent and analysis of natural geological material from four very different mine sites have
been used to determine the factors that control the rate of element release and mineral sources and sinks for major elements
and for the contaminant metals Zn, Pb, and Cu. The sites are: a coal spoil tip; a limestone-hosted Pb mine, abandoned for
the last 200 a; a coal mine; and a slate-hosted Cu mine that was abandoned 150 a ago. Hydrogeological analysis of these
sites has been performed to allow field fluxes of elements suitable for comparison with laboratory results to be calculated.
Hydrogeological and mineral equilibrium control of element fluxes are common at the field sites, far more so than in laboratory studies. This is attributed to long residence times and low water–rock ratios at the field sites. The high water storativity at many mine sites, and the formation of soluble secondary minerals that can efficiently adsorb metals onto their
surfaces provides a large potential source of pollution. This can be released rapidly if conditions change significantly, as in,
for example, the case of flooding or disturbance.
2006 Elsevier Ltd. All rights reserved.
1. Introduction
*
Corresponding author. Present address: Australian National
University, RSES, Building 61, ANU, Mills Road, Canberra
ACT 0200, Australia. Fax: +61 2 61250738.
E-mail address: [email protected] (K.A. Evans).
Mining activities expose large volumes of fresh
rock to atmospheric conditions. Subsequent weathering and release of rock constituents, especially via
0883-2927/$ - see front matter 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apgeochem.2005.10.002
378
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
sulphide oxidation, can lead to acidic solutions and
elevated concentrations of environmentally undesirable elements in surrounding waterways (Younger,
2000; Banks et al., 1997). This global problem is
exacerbated by the lack of an effective framework
for determination of legal liability, which means
that resources for remediation of contaminated
abandoned sites are limited (Younger, 1997). Fortunately, pre-mining environmental assessments are
now required in the majority of countries, and for
this, as well as for cost-effective remediation, it is
necessary to develop the ability to predict the severity and longevity of mining-related contamination
on a site-specific basis. An understanding of the processes affecting weathering is also relevant to climatic modelling, where a better knowledge of
field-scale mineral dissolution rates is necessary to
facilitate assessment of relationships between temperature, precipitation, elevation and weathering
rate (e.g. White and Blum, 1995).
Laboratory assessments are the most convenient method of measuring the contamination
potential of spoil. Detailed studies of the dissolution of individual mineral phases have determined
dissolution mechanisms and relationships between
dissolution rates and pH, temperature, surface
and crystallographic characteristics, and ratedetermining concentrations of reactants that influence kinetic mass action (e.g. .Wieland et al.,
1988; Xie and Walther, 1992; Martello et al.,
1994; Peiffer and Stubert, 1999; Holmes and Crundwell, 2000), but it is difficult to use such work to
account for interactions between the components
of the phases in a polymineralic assemblage. Rudimentary acid–base measurements on polymineralic
materials are popular (e.g. Adam et al., 1997),
because of the relative ease and rapidity with
which they can be undertaken and interpreted.
Such methods measure the total potential of a
sample for contamination, but do not provide
information on the rate at which contamination
is produced, and are thus of limited use (e.g. Jambor, 2000). Batch and column experiments have
also been used (van Grinsven and van Riemsdijk,
1992; Stromberg and Banwart, 1999; Banwart
et al., 2002), to measure rates and to distinguish
rate-controlling mechanisms of mineral dissolution. Column experiments can be more useful than
batch experiments because they involve water:rock
ratios and hydrological solute transport processes
similar to those found in the field. Regardless of
method, however, it is difficult to extrapolate lab-
oratory results to field situations. Laboratoryderived mineral dissolution rates are often 2–4
orders of magnitude faster than those measured
for minerals in the field (e.g. Schnoor, 1990; White
et al., 1996).
A number of factors have been proposed to
account for this discrepancy. These include differing pH, grain size, temperature, hydrology, cation
exchange characteristics, availability of reactants
such as O2, degree of physical and chemical heterogeneity, secondary mineral behaviour, mineral
surface characteristics and proximity to chemical
saturation with respect to the dissolving minerals,
between weathering environments (e.g. Sverdrup
and Warfvinge, 1995; Malmstrom et al., 2000;
White and Brantley, 2003). Algorithms that quantitatively account for some of these factors have
been devised (e.g. Sverdrup and Warfvinge, 1995;
Malmstrom et al., 2000). These are based on the
premise that each of the different parameters that
affect dissolution rates can be considered separately, and a scaling factor calculated for each.
The combination of the scaling factors then allows
extrapolation of laboratory-based weathering rates
to the field. Calculations are based on fundamental
relationships between reaction rate and the physical parameter of interest, and so site-specific calibration is not required for most parameters.
Thus, the algorithms should prove to be robust
and generally applicable. The algorithm of Malmstrom et al. (2000) has been demonstrated to be
successful in a study of the weathering of granitoid
waste rock from the Aitik Mine, Sweden (Malmstrom et al., 2000).
However, application of this type of algorithm
involves the implicit assumption that the same
mechanism determines the rate of element release
in the laboratory and in the field, and that mineral
sources and sinks play the same role in the two
environments. This is not necessarily the case. The
present study investigates rate-determining mechanisms and mineral sources and sinks at four UK
mine sites via a year-long mine-water monitoring
program and bulk and mineralogical analysis of
samples from the sites. Results are related to those
produced by laboratory experiments using results of
hydrogeological analyses of the sites. Results for
one site are then compared to column and batch
experiments on material from the site (Evans and
Banwart, 2006). The implications of results for
prediction and treatment of mine-water related
pollution are discussed.
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
2. Field sites
Locations of all sites are shown in Fig. 1. The
criteria for field sites were that they should: be
hydrologically well defined with point discharges;have a well-documented mining history;
exhibit commonly observed contaminant characteristics; be representative of the selected type of
mining environment; have available data for rainfall, discharge, topology, and geology, and have a
reasonably simple geological structure.
2.1. Quaking Houses
The spoil tip at Quaking Houses, near Newcastleupon-Tyne, County Durham, England (Fig. 1a),
comprises spoil excavated during working of the
Morrison Busty pit between 1922 and 1973. The
tip overlies sand and clay drift deposits, which themselves overlie Carboniferous interbedded mudstones,
sandstones and coal seams (Pritchard, 1997). The
area of the tip is approximately 35 ha and the height
varies from 4.25 to 11 m. The spoil material is a heterogeneous mix of shale, ash, coal, and coal dust,
with scattered cobbles, sandstone boulders, timber,
and traces of red burnt shale. The vast majority of
particles are smaller than 5 cm diameter. Recently
formed secondary minerals are common in samples
from the upper 5 m of the tip. Hydrology of the tip
is simplified by the underlying clay-rich drift deposits
which act as an aquitard, isolating the tip from
groundwater flow (Pritchard, 1997). The discharge
at Stanley Burn (output G, OS grid ref: 41770
55095) is thought to collect water from approximately 10% of the tip, plus relatively uncontaminated water that flows from a surface drain
(Pritchard, 1997).
Significant tip-related pollution was not recorded
between closure of the mine in 1974, and 1986,
when construction of the A693 Annfield Plain
bypass cut through almost the full depth and
breadth of the tip. During construction, drains that
serviced the tip were incorporated into the road
drainage system, the bulk of which discharges into
the nearby Stanley Burn. A subsequent decrease in
pH and increases in Fe, Al, and SO2
4 contents in
the Stanley Burn have been attributed to drainage
from the tip. There have been a number of previous
studies of water quality at the site. Local authority
reports (e.g, Markey-Amey, 1995; Newbegin,
1997) have been complemented by two M.Sc. theses
(Pritchard, 1997; Srour, 1998) and a Ph.D. thesis
379
(Gandy, 2002) with associated publications (Gandy
and Evans, 2002; Gandy and Younger, 2003), which
have developed groundwater flow and chemical
models for the site.
2.2. Grattendale
Limestone hosted Pb/Zn deposits in the valley of
Grattendale in the Peak District, Derbyshire, England (Fig. 1b), were mined up to the end of the
18th Century. Grattendale is in the White Peak
area, which comprises an inlier of Carboniferous
limestone surrounded, except to the extreme SW,
by younger fluvio-deltaic sandstones (Stevenson
et al., 1982). Mississipi Valley Type (MVT) Pb
and Zn ores formed when cooling metal-rich fluids
passed sub-horizontally through the limestones
(Ewbank et al., 1995). Ore minerals include galena,
sphalerite, baryte, and fluorite. Mineralisation was
controlled by lithology and by pre-existing structure, and resulted in a geometry of near vertical or
horizontal sheets, known as rakes and flats respectively, and occasional linear pipes (Ford and Rieuwerts, 1970). Mineralised zones rarely penetrate the
volcanic horizons, known locally as toadstones,
and are thus found mainly within higher levels of
the formation (Ford and Rieuwerts, 1970). Grattendale is a steep sided valley with walls of clean,
fossiliferous (rugose, fragmented corals, and brachiopods) limestone. The Matlock Lower Lava,
known locally as the Grattendale Lava, outcrops
partway up the valley; where it is around 9 m thick
and dips to the NE (Oakham, 1979). Surface expression of mineralisation is rare because most outcropping deposits were extensively worked to a depth of
10–15 m and then filled with earth.
Mining in the White Peak since Roman times has
produced a poorly mapped network of largely interconnected mines and drainage channels, or soughs
(Rieuwerts, 1987). The resultant hydrology is complex; however, Grattendale is effectively hydrologically bounded by topography, geology and soughs
and so presents a relatively simple system. The discharge (OS grid reference: 42083 36078) is continuous year round, with a flow rate between 0.1 and
2 L s1. Mining exposed large areas of fresh sulphide surfaces to relatively oxidising conditions.
Regional S oxidation rates higher than those that
pertained before mining began are indicated by S
isotopes (Bottrell et al., 1999), and analysis of
sediments in cave systems shows fine grained sulphides with a relatively high potential for dissolution
380
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
(a)
(b)
(c)
(d)
Fig. 1. Locations and details of the four field sites. Grid references are to the Ordnance Survey grid.
(Bottrell et al., 2000). However, solubility of mineral
phases bearing Pb, Zn, and Ba, the principal potentially toxic contaminants in the area, are relatively
low under high pH conditions such as those found
in limestone-rich areas because formation of sparingly soluble Pb and Zn carbonates and hydroxides
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
effectively removes most of the metals from solution
(Benevenuti et al., 2000).
2.3. Ynysarwed
The Ynysarwed minewater discharge flows from
an old adit connected to the lower Ynysarwed
workings in the Neath Valley, South Wales
(Fig. 1c). The regional geology consists of faulted
sandstones, mudstones, siltstones and coals of the
Upper and Middle Coal Measures (Westphalian),
which dip at approximately 6 SW (Barclay
et al., 1988). The topography comprises steep sided
narrow valleys, separated by larger areas of relatively flat upland. Upper Coal Measures fluvial
Pennant sandstones, interspersed with a number
of small, largely unworked coal seams, overlie the
Rhondda number two seam, which is the principal
worked seam in the area. Below the coal lie Lynfi
Middle Coal Measures mudstone beds, which show
a marked marine influence (Barclay et al., 1988). It
is this marine influence which gives the anthracitic
Rhondda number 2 coal its high S content (2–4%
pyritic S).
The hydrogeology of the area is dominated by
the highly conductive mine workings. The Ynysarwed discharge (OS grid reference: 28094 20186) is
fed by recharge through the relatively permeable
Pennant Sandstone (approximately 20% of precipitation) and groundwater flows from the Blaenant
system to the west (Younger and Adams, 1997).
Underlying Lynfi mudstones are relatively impermeable and are assumed to prevent downward escape
of water from the system. Flows from the neighbouring unsaturated Upper Ynysarwed, Crynant,
and Lwynon workings, which bound the area under
consideration to the north, are thought to be minor
as all drain to adits well above the groundwater
table (C. Rees, per. comm, 1999; Younger and
Adams, 1997). Piezometer measurements show that
groundwater flow is from NW to SE, that is, from
Blaenant (water table 75 m AOD) to Ynysarwed
(water table 20 m AOD). Prior to the cessation of
mining, the Blaenant works were pumped, while
the lower Ynysarwed workings drained through
the discharging adit. Lower Ynysarwed was closed
in 1938, and although an occasional ferruginous discharge was noted at the adit this was not of any volumetric significance. The Blaenant mines closed in
1991, the pumps were turned off, and groundwater
levels within the workings began to rise. Minewater
discharges were expected to begin at or near the
381
River Dulais, the bed of which passes over the Blaenant workings. However, the coal barriers between
the Blaenant and Ynysarwed mines (Fig. 1c) proved
ineffective and a substantial discharge began at the
Ynysarwed adit in 1993 (Younger and Adams,
1997). Water discharges from the adit portal at
Ynysarwed are up to 36 L s1. Discharges were initially acidic with a pH as low as 3, but are currently
circumneutral. The discharge is reducing, and rich
in dissolved Fe and SO2
(200 and 1800 mg L1
4
respectively), although a gradual decline in Fe concentrations since 1993 has been observed (Ranson
and Edwards, 1997). Precipitation of ochre from
discharging water has adversely affected fish and
invertebrate populations in the Neath Canal, which
receives the discharge, and the discharge was given
11th priority in a 1993 water quality survey of
British minewater discharges. This led to a significant amount of work on both the hydrochemistry
and hydrology (e.g. Younger, 1996; Younger and
Adams, 1997).
2.4. Church Coombe
The Church Coombe minewater discharge, Cornwall, England flows into a small stream (Fig. 1d).
The surrounding area is a geometrically complex
mix of Upper Devonian Mylor slates (killas), and
Late Carboniferous/Early Permian granites which
form part of the Cornubian Batholith. These are
intruded by NNE-WSW striking coarse granite porphyries (elvans) and dolerites (greenstones) (Dines,
1956). Mineralisation was associated with circulation of basinal and meteoric fluids driven by heat
from the intrusion of the Cornubian batholith
(Gleeson et al., 2001). Mines near Church Coombe
are located in a trough of slate between the Carnmellis granite to the south, and the Carn Brea granite to the north. The main ore-bearing body dips at
around 30 S in this region and follows the graniteslate contact in the area shown in Fig. 1d. A number
of ENE-trending sub-vertical off-shoots from the
main body were also worked in the area (Dines,
1956). Principal ore minerals are chalcopyrite and
cassiterite, with a tourmaline/chlorite gangue
(peach). The sub-vertical lodes were worked mainly
for Cu, although Sn and As were also produced
(Dines, 1956). Mining was well established in the
area by 1699, and by 1743 the adit was being used
to produce water power (Hamilton-Jenkin, 1965).
Mining flourished until the 19th Century, but
underground activity had effectively ceased by
382
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
1919. The dumps were reworked for secondary and
accessory minerals in the 1930s (Dines, 1956).
Exploration and mapping of the adit (portal at
OS grid reference: 16916 04076) suggest that it
drains the Wheal Bassett system, which includes
North Bassett and Wheal Bassett (Fig. 1d). Flow
rates are up to 90 L s1. The region is classed as a
minor aquifer, with permeability controlled largely
by the depth of weathering, rather than by the
hydrological properties of the slate or the granite.
The water is circumneutral and contains 1 mg L1
of Cu all year round, and thus potentially poses
some threat to receiving water courses. However,
this is mitigated by subsequent dilution.
3. Methods
3.1. Element release mechanisms
3.1.1. Water sampling and analysis
Sites were monitored monthly (Table 1) for a period of at least 12 months between June 1999 and
June 2001. Quaking Houses samples were taken
from outfall G (Fig. 1a), Grattendale samples from
Wraithe Sough (Fig. 1b), Ynysarwed samples from
the marked adit portal (Fig. 1c), and Church Coombe samples from the adit portal marked in Fig. 1d.
At Quaking Houses, unfiltered acidified (1%
HNO3) and non-acidified samples were analysed
by atomic absorption spectroscopy (AAS) and ion
chromotography (IC) for metals and common
2
anions (F, Cl, NO
2 , Br , NO3 , SO4 ) at the
Department of Civil Engineering, University of
Newcastle. pH, alkalinity, Eh and conductivity were
determined in the field using portable meters. At
Grattendale, Ynysarwed and Church Coombe,
water samples were filtered through 0.2 lm filters
(Schleicher and Schuell) into new Nalgene bottles
which were rinsed in the discharge before use. Acidified (1% HNO3) and unacidified samples were collected at each sampling visit. Acidified samples
were analysed by inductively coupled plasmaatomic emission spectroscopy (ICP-AES) for major
and abundant trace elements at the Assay Office,
Sheffield. Unacidified samples were analysed, in
most cases, by Ion Chromatography (IC: Dionex)
for light alkali metals and alkaline earths (Na+,
K+, Ca2+, Mg2+), and common anions (F, Cl,
3
2
NO
3 , NO2 , Br , PO4 , SO4 ). One or two samples
from each site were screened using inductively coupled plasma-mass spectroscopy (ICP/MS) at the
Centre for Analytical Services, University of Shef-
field, to quantify rarer elements such as Cd, U,
and Th. Significant concentrations of these elements
were not found. Iron-rich samples from Ynysarwed
were not analysed by IC because Fe forms insoluble
compounds inside the ion exchange columns, and,
in any case, it was known that SO2
4 is the dominant
anion (Younger, 1997). pH, redox potential, conductivity and temperature were measured with a
Myron 6P Ultrameter. Alkalinity was measured by
titration with 0.035 N H2SO4 to pH 4.5 using bromocresol green/methyl orange indicator. Flow rate
was also measured at each sampling site. At Quaking Houses and Grattendale this was accomplished
using a bucket and stopwatch. At Ynysarwed, flow
rates were measured electronically with stream
gauging by the Environment Agency. These measurements were affected by the construction of a
treatment plant for the minewater effluent and thus
results from this site are subject to additional
sources of uncertainty, discussed below. At Church
Coombe, flow rates were calculated from the velocity of flow in a channel of known cross-sectional
area.
3.1.2. Interpretation
Relationships between element concentrations
(moles L1), discharge flow rates (L s1) and element fluxes (mole s1; obtained by multiplying concentration and discharge flow rate) were used to
determine the rate-controlling mechanism for each
element. Rate-controlling mechanisms were split
into those that involved principally (a) surface
kinetic-, (b) mineral equilibrium-, and (c) transport-related factors.
3.1.2.1. Surface kinetic (S). If surface kinetic processes control the rate of mineral dissolution then
the flux of elements sourced from the dissolving
mineral is constant, so long as other rate-determining factors such as the quantity of source mineral,
pH, temperature and the concentration of reactants
remain constant (e.g. Wieland et al., 1988). If processes such as secondary mineral precipitation do
not interfere then the flux of the element from the
mine should also be constant. Clearly, temperature,
pH, and other rate-controlling factors are likely to
change in a mine over the course of a year and so
some variation in surface kinetic-controlled element
fluxes would be expected. An approximately constant element flux was thus used to infer that the
principal mineral source or sources for that element
were dissolving at a rate controlled by surface-kinetic
Table 1a
Quaking Houses discharge data
15/10/98
19/11/98
17/12/98
26/1/98
23/2/99
23/3/99
23/4/99
14/5/99
3/6/99
2/7/99
23/7/99
30/7/99
16/9/99
8/10/99
Al
Ca
Fe
K
Mg
Mn
Na
Zn
Cl
SO2
4
3.23
193.2
6.61
100
52.1
3.16
840
0.84
5.48
362.2
5.24
294
126.1
3.72
2493
1.14
5.13
381.9
4.92
297.2
120.9
3.18
2572
1.53
3132
544
8.49
182
11.46
54
46.4
2.37
566
1.09
2049
921
7.734
263.7
2.486
106.6
82.3
3.786
1148
2.405
1691
706
3.854
168.1
2.851
164.2
51.6
0.948
1263
0.948
2062
404
2.85
144.8
2.706
130.2
38.8
1.301
792
0.575
440
242
4.091
196.5
4.39
188.3
65
2.552
1149
1.01
1851
512
7.334
257
7.5
327.3
100.6
4.127
1922.1
1.391
3047
873
8.5
10.18
5.79
9.41
9.89
16.95
5.6
6.18
4.95
565
5.36
171.4
3.73
140.6
54.4
3.56
1074
1.6
2236
552
2621
1227
2446
1257
1581
881
2.24
314
7.14
36.5
98
4.15
2163
0.91
3939
910
pH
Conductivity (lS)
Alk. (mg L1 as CaCO3)
Flow (L min1)
6.46
5765
77
55.2
6.37
7672
67
107.7
6.58
7555
6.48
7130
54
76.3
5.7
7320
6.09
7138
27
89.3
6.14
8010
58
110.5
7.1
4787
107
168
6.45
6880
70
76.36
5.45
11,090
30
131.2
5.22
10,140
<5
20.5
4.2
8760
0
6.2
5970
31
23.8
5.73
12,190
<5
28.6
71.2
39.7
Table 1b
Wraithe Sough discharge data
Conc. (mg L1)
14/10/99
19/11/99
02/12/99
19/01/00
02/02/00
16/02/00
01/03/00
30/03/00
04/05/01
07/06/00
18/08/00
19/09/00
28/10/00
Ca
K
Mg
Na
Si
Pb
Sr
Zn
Cl
F
NO
3
SO2
4
92.40
0.80
34.10
5.85
2.58
0.01
92.60
1.00
31.60
6.66
2.32
0.02
0.20
14.42
1.87
23.83
24.41
0.16
13.38
1.55
25.58
22.79
80.20
0.74
26.04
5.00
2.41
0.06
0.03
0.06
14.05
1.65
26.90
25.23
78.89
0.47
24.64
5.03
2.24
0.01
0.03
0.18
15.19
1.77
35.41
24.96
98.23
1.25
43.67
7.54
3.02
0.04
0.04
0.15
15.71
1.81
28.66
24.80
95.23
1.08
41.53
7.22
4.24
0.02
0.04
0.11
14.87
1.65
27.20
24.78
91.23
1.19
39.39
7.46
3.73
0.01
0.04
0.11
15.65
1.91
29.23
25.04
70.25
0.67
27.32
5.51
2.37
0.00
0.03
0.15
11.04
1.67
30.14
19.46
67.74
0.65
26.03
5.63
2.29
0.00
0.03
0.15
11.16
1.68
25.73
19.35
76.18
0.26
28.84
4.70
2.00
0.00
0.03
0.23
11.04
1.58
25.26
19.87
91.38
0.56
29.24
7.14
2.51
0.08
0.03
0.15
11.21
1.60
27.37
22.01
96.65
0.63
31.89
7.24
2.73
0.00
0.04
0.18
11.14
1.58
27.22
23.65
82.29
0.99
32.09
5.37
2.52
0.00
0.03
0.20
11.64
1.65
pH
Conductivity (lS)
Alkalinity (mg L1)
Temp. (C)
ORP (mV)
Flow rate (ml min1)
8.12
623.30
270.00
9.30
134
142
7.30
612.00
168.00
7.80
205
205
7.57
597.20
192.00
8.90
178
142
7.31
604.00
8.19
606.00
192.00
8.80
16
889
7.37
606.00
161.00
8.40
246
800
6.97
574.00
161.00
9.40
153
1720
7.89
605.00
176.00
8.90
91
875
7.87
588.00
184.00
9.20
109
950
7.69
598.00
200.00
9.40
n/a
845
7.84
621.50
200.00
10.00
97
290
8.08
629.00
330.00
9.30
521
200
7.60
597.00
250.00
9.50
101
548
22.45
383
8.80
205
563
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
Conc. (mg L1)
384
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
Table 1c
Ynysarwed discharge data
Conc.
(mg L1)
17/08/99
20/09/99
13/10/99
15/11/99
10/12/99
26/01/00
22/02/00
28/03/00
26/04/00
16/05/00
28/06/00
Al
Ca
Fe
K
Mg
Mn
Na
Si
As
Ni
Pb
Sr
Zn
SO2
4
<0.05
285.0
173.0
23.1
156.0
3.9
108.0
7.5
<0.01
0.2
0.1
<0.01
<0.1
1839.0
0.0
314.0
170.0
24.9
170.0
3.9
115.0
7.1
0.1
0.2
<0.1
2.2
<0.1
1950.0
<0.01
313.0
169.0
25.2
171.0
0.9
159.0
7.2
0.1
0.2
<0.01
<0.01
0.1
1830.0
0.2
300.0
163.0
25.5
154.0
3.7
152.0
7.8
0.1
3.7
0.1
1.9
<0.1
1980.0
0.2
239.8
150.6
23.3
119.6
3.6
120.0
7.1
0.1
0.2
0.1
1.2
<0.1
1589.2
0.2
241.4
147.6
24.3
120.5
3.5
123.2
7.3
0.1
0.2
<0.1
1.2
<0.1
1595.9
0.8
289.4
154.8
29.1
176.5
4.0
171.3
8.9
0.2
0.3
0.2
1.3
<0.1
2055.9
0.7
238.7
138.1
26.8
121.8
3.7
155.6
8.3
<0.01
0.4
0.1
1.3
<0.1
1911.6
0.8
226.9
131.8
25.8
115.0
3.5
146.9
28.9
<0.01
0.7
0.1
1.2
0.6
1811.7
0.6
288.8
159.5
30.2
144.6
4.1
141.3
7.1
<0.01
0.2
<0.1
1.0
0.1
1520.3
0.6
241.0
139.8
26.9
119.2
3.4
118.6
6.2
0.1
0.2
<0.1
0.9
<0.1
1346.1
pH
Conductivity
(lS)
Alkalinity
(mg L1)
Flow rate
(L min1)
Temp. (C)
ORP (mV)
6.44
2859
6.47
2763
6.38
2750
6.20
2800
6.08
2573
6.37
2709
6.46
2619
6.35
2642
5.98
2539
6.28
2583
6.36
2517
100
160
180
98
112
112
112
96
112
128
140
26.31
24.07
25.97
26.74
29.09
28.43
29.05
28.25
26.52
28.37
27.21
13
56
15
31
15
31
11
5
14
44
11
28
12
44
12
116
13
0
15
n.a.
15
54
processes. The definition of constant flux used here
was that of a standard deviation of the monthly flux
measurements less than 20% of the average value.
The value of 20% is arbitrary, but choices of values
for the constant flux threshold between 15% and
40% give very similar results. Thresholds for the
other categories described below are also arbitrary,
and results show a similar weak sensitivity to the
choice of value.
3.1.2.2. Equilibrium (E). If the concentration of an
element is controlled by chemical saturation with
respect to a mineral then the concentration of that
element is fixed by the presence of the mineral, so
long as factors such as pH, temperature, the chemical activity of other aqueous species, and the activity of the mineral phase remain constant. A solution
may be chemically saturated with respect to a source
mineral present in rock prior to weathering or to
some secondary mineral that incorporates the element of interest. Cation exchange surfaces on clay
minerals also exert some control on the concentrations of elements in solution, although the systematics of this are more complicated. It is unlikely that
factors that affect the equilibrium concentration of
an element are likely to remain constant over the
course of a year at any of the field sites studied.
Thus, an approximately constant concentration of
an element was taken to infer equilibrium control.
Constant concentration was diagnosed when the
standard deviation of the monthly measurements
was less than 20% of the average value, and where
a suitable source/sink mineral was observed, and
calculated to be in equilibrium with the solution
(see below). The second criterion is included because
a constant concentration in a mine discharge may
also result from constant concentration in some
external source.
3.1.2.3. Transport (T). Transport control of element
release occurs when changes in hydrogeological
parameters that affect flow rate, such as the rate of
recharge, the percentage of water saturation in the
unsaturated zone, or water table height, determine
the rate of mineral dissolution and element transport. The transport control described here is similar
to the transport-limited erosion described by Stallard and Eddmond (1987). However, it is slightly
different to the process described by these authors
because it does not include situations in which
Table 1d
Church Coombe discharge data
15/08/00
13/09/00
26/10/00
29/11/00
18/12/00
30/01/01
27/02/01
29/03/01
25/04/01
31/05/01
26/06/01
25/07/01
Al
Ca
Fe
K
Mg
Mn
Na
Si
Cu
Zn
Cl
F
NO
3
SO2
4
0.41
8.13
0.32
2.76
4.92
0.05
13.24
3.23
1.47
0.23
27.57
0.86
12.06
26.74
0.73
11.94
0.04
4.34
5.84
0.08
24.26
4.83
2.09
0.30
28.13
0.86
11.76
28.41
0.45
11.47
0.01
3.96
5.57
0.06
18.28
4.06
1.28
0.22
28.58
0.74
13.28
28.60
0.41
9.97
0.01
0.56
4.05
0.05
18.96
3.39
1.11
0.18
27.00
0.59
13.55
26.65
0.38
10.96
0.02
3.31
4.93
0.05
19.80
3.60
1.08
0.17
27.27
0.64
13.01
26.73
0.39
10.39
0.01
3.28
5.02
0.05
17.89
3.66
1.20
0.19
26.52
0.59
13.91
28.15
0.44
10.05
0.02
3.14
4.76
0.06
18.78
4.28
1.36
0.33
24.61
0.61
14.89
25.04
0.39
11.01
0.03
3.59
4.39
0.05
19.29
3.64
1.14
0.19
24.30
0.60
13.94
27.17
0.39
10.43
0.70
3.26
4.17
0.06
17.19
3.63
1.27
0.22
23.69
0.69
12.44
23.77
0.38
8.69
3.78
13.88
4.65
0.04
19.41
4.19
1.57
0.23
25.79
0.79
13.08
26.08
0.38
9.38
1.06
11.76
5.03
0.04
21.33
4.42
1.67
0.26
24.47
0.83
n.a.
29.25
0.38
9.49
0.72
5.96
5.16
0.05
20.63
4.46
1.74
0.26
25.18
0.81
12.44
32.99
pH
Conductivity (lS)
Alkalinity (mg L1)
Flow rate (L s1)
Temp. (C)
ORP (mV)
5.6
200
11.8
14
12
186
6
190
15
2
11.7
220
5.7
210
15
40
12.3
215
5.4
210
10
60
12
235
6
240
10
45
12
253
5.6
230
10
90
12
261
5.6
250
15
60
12
244
6
240
10
50
12.1
247
6.1
270
10
40
12.2
235
5.3
220
10
14
12.2
215
6.2
210
10
16
12.5
198
5.44
220
10
12
12.6
200
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
Conc.
(mg L1)
385
386
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
mineral equilibrium is approached, although
transport-related processes are implicated in the
attainment of mineral equilibrium. Mechanisms
for transport-controlled element release include:
flushing, where an increase in the height of the water
table dissolves soluble secondary minerals precipitated in previously hydrologically-unsaturated parts
of a hydrogeological system; dilution, where
increased water flow decreases element concentrations and may enhance the rate of mineral dissolution; and changes in the ratio of mobile to
immobile water in the unsaturated zone (Evans
and Banwart, 2006; Banwart et al., 2004), which
affect the efficiency of removal of dissolved mineral
constituents. Physical processes that accompany
transport control are complex (Clow and Drever,
1996) and so the flux–flow characteristics of transport control are variable; however, all are likely to
result in a correlation between flow rate and element
release that could be positive (flushing) or negative
(dilution). Transport control of element release
was inferred when the absolute value of the correlation coefficient between flow rate and element concentration was greater than 0.7.
3.1.2.4. Depletion (D). It may be that the supply of
a mineral may be depleted by dissolution, and that
this process is the main control on the flux of constituent elements of that mineral from the system
of interest. If this is the case then the flux of this element from the system will decrease with time.
Depletion control of an element was inferred when
the absolute value of the correlation coefficient
between element flux and time was less than 0.7.
3.1.2.5. General transport (I). The categories
defined above are end-members in a continuum
of system behaviour that is created by the playoff between mineral reactivity, fluid flow rates,
and other hydrological parameters. There are a
number of situations for which none of the above
criteria would be met, which implies that the ratecontrolling factor for element release varies over
the course of a year. The rate-controlling element
release mechanism in these cases cannot be determined by a simple inspection of element flow and
flux characteristics and is classed as general transport (I).
Figs. 2–5 illustrate examples of the relationships
between element concentrations, fluxes, time, and
discharge flow rate. Inferred rate-controlling element
release mechanisms are summarised in Table 2.
3.2. Mineralogy
3.2.1. Sample collection
Twenty kilograms of each of the main lithologies
was collected from each site. Representative sampling of 20 kg of material from heterogeneous systems containing many tons of material is not
remotely achievable. However, the samples can be
used to identify source and sink minerals present
at the sites.
At Quaking Houses samples were taken from
material retrieved during borehole construction;
and consisted of two types, black coal waste
(QB), and orange clay-rich material from boulder
clay underlying the tip (QO). QB was made by
mixing approximately equal proportions of spoil
taken at 1 m intervals down the tip. Weathering
of the top portion of the tip has caused significant
mineralogical variation with depth (Evans et al.,
2003). Ore-bearing limestone from Grattendale
was collected at a spoil tip 5 km from the adit
portal, and non-ore bearing limestone was collected from Grattendale itself. All samples were
large (1–5 kg), partly weathered limestone boulders. Material from Ynysarwed was collected from
a spoil tip that is known to contain material from
both Blaenant and Ynysarwed mines (C. Rees,
pers. comm.). Samples were almost unweathered
because recent landscaping of the area had
exposed relatively fresh spoil at the surface.
Twenty kilograms of each of pyrite-bearing mudstone (YM), clean, mature sandstone (YS) and
pyrite-bearing coal (YC), in 1–5 kg lumps, were
collected, to allow differences in mineral dissolution rates and composition between lithologies to
be determined. At Church Coombe only one set
of material (CC) was collected from a spoil tip
adjacent to the North Basset Mine. Here, 1–3 kg
samples of a reddish-brown hornfelsed mudstone,
showing evidence of mineralisation, were taken
from a freshly turned spoil tip.
3.2.2. Analysis
Samples were jaw-crushed to a median grain size
of 1 mm and homogenised in an industrial mixer
at the Department of Civil Engineering, Sheffield.
This preparation was necessary for the laboratory
experiments (Evans and Banwart, 2006). Major
and trace element compositions for each of the eight
lithologies were determined at the British Geological Survey, Keyworth, using X-ray fluorescence
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
(a)
(b)
9
I
7
6
5
4
3
2
1400
1200
Concentration (mg L-1)
Concentration (mg L-1)
8
T
1000
800
600
400
200
Quaking House: SO 42-
QuakingHouses: Al
1
Oct
Feb
Jun
0
Oct
Oct
Feb
(d)
0.25
I
0.20
0.15
0.10
E
24
23
22
21
20
Grattendale: SO 42-
Grattendale: Zn
0.05
Oct
Feb
Jun
19
Oct
Oct
Feb
Date
(e)
4.5
180
D
3.5
3.0
2.5
2.0
1.5
160
150
140
130
Ynysarwedd: Fe
Ynysarwedd: Mn
0.5
Aug
Nov
Feb
120
Aug
May
Nov
(h)
2.2
Church Coombes: Cu
1.8
1.6
1.4
1.2
1.0
Aug
15.5
15.0
Concentration (mg L-1)
2.0
T
Feb
T
14.5
14.0
13.5
13.0
12.5
12.0
Nov
Feb
Date
May
Date
Date
Concentration (mg L-1)
Oct
170
Concentration (mg L-1)
Concentration (mg L-1)
(f)
1.0
(g)
Jun
Date
E
4.0
Oct
26
25
Concentration (mg L-1)
Concentration (mg L-1)
Jun
Date
Date
(c)
387
May
Aug
11.5
Aug
Church Coombes: NO 3Nov
Feb
May
Aug
Date
Fig. 2. Concentration versus time for representative elements from the four field sites. (a) Quaking Houses: Al; (b) Quaking Houses: SO2
4 ;
(c) Grattendale: Zn; (d) Grattendale: SO2
4 ; (e) Ynysarwedd: Mn; (f) Ynysarwedd: Fe; (g) Church Coombes: Cu; (h) Church Coombe:
NO
3 . Shaded bands indicate the region of ±20% around the average value.
388
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
(a)
9
8
1400
T
1200
Concentration (mg L-1)
Concentration (mg L-1)
(b)
I
7
6
5
4
3
1000
800
600
400
200
2
Quaking Houses: SO 42-
Quaking Houses: Al
1
0
0
20
40
60
80
100
120
140
160
180
0
20
40
60
Flow Rate (L min-1)
(c)
(d)
0.25
I
100
E
26
Concentration (mg L-1)
Concentration (mg L-1)
80
120
140
160
180
Flow Rate (L min -1)
0.20
0.15
0.10
Grattendale: SO 42-
24
22
20
Grattendale: Zn
0.05
18
0
200
400
600
800 1000 1200 1400 1600 1800
0
200
400
600
Flow Rate (ml min-1)
(e)
4.5
180
D
170
Concentration (mg L-1)
Concentration (mg L-1)
(f)
E
4.0
3.5
3.0
2.5
2.0
1.5
160
150
140
130
1.0
Ynysarwedd: Fe
Ynysarwedd: Mn
0.5
120
23
24
25
26
27
28
29
23
30
24
25
Flow Rate (L min -1)
2.2
Concentration (mg L-1)
T
26
27
28
29
30
Flow Rate (L min -1)
(h)
16
Concentration (mg L-1)
(g)
800 1000 1200 1400 1600 1800
Flow Rate (ml min-1)
15
T
Church Coombes: Cu
2.0
1.8
1.6
1.4
14
13
12
1.2
Church Coombes: NO 31.0
11
0
20
40
60
Flow Rate (L s -1)
80
100
0
20
40
60
80
100
Flow Rate (L s -1)
Fig. 3. Concentration versus flow rate for representative elements from the four field sites. (a) Quaking Houses: Al; (b) Quaking Houses:
2
SO2
4 ; (c) Grattendale: Zn; (d) Grattendale: SO4 ; (e) Ynysarwedd: Mn; (f) Ynysarwedd: Fe; (g) Church Coombes: Cu; (h) Church
Coombe: NO
3.
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
(a)
(b)
1200
1.2e+5
I
Quaking Houses: Al
1.0e+5
r = -0.11
Flux (mg min-1)
Flux (mg min-1)
r = 0.20
800
600
400
200
8.0e+4
6.0e+4
4.0e+4
2.0e+4
0
Oct
0.0
Feb
Jun
Oct
Oct
Feb
Date
Jun
Oct
Date
(d)
14
12
T
Quaking Houses: SO42-
1000
(c)
389
3000
E
I
Grattendale: SO42-
2500
r = 0.00
Flux (mg min-1)
Flux (mg min-1)
10
8
6
4
2000
1500
1000
r = 0.21
2
500
0
Grattendale: Zn
Oct
Feb
Jun
0
Oct
Oct
Feb
Date
(e)
(f)
8000
E
Flux (mg min-1)
Flux (mg min-1)
I
2.6e+5
6000
5000
4000
3000
Nov
Feb
2.4e+5
2.3e+5
r = --0.51
2.1e+5
Ynysarwedd: Mn
1000
Aug
2.5e+5
2.2e+5
r = 0.11
2000
Ynysarwedd: Fe
2.0e+5
Aug
May
Nov
May
80000
T
T
60000
5000
Flux (mg min-1)
Flux (mg min-1)
(h)
7000
4000
3000
2000
1000
r = 0.00
0
Aug
Feb
Date
Date
6000
Oct
2.8e+5
2.7e+5
7000
(g)
Jun
Date
40000
20000
r = 0.05
0
Church Coombes: NO3-
Church Coombes: Cu
Nov
Feb
Date
May
Aug
Aug
Nov
Feb
Date
May
Aug
Fig. 4. Flux versus time for representative elements from the four field sites. (a) Quaking Houses: Al; (b) Quaking Houses: SO2
4 ;
(c) Grattendale: Zn; (d) Grattendale: SO2
4 ; (e) Ynysarwedd: Mn; (f) Ynysarwedd: Fe; (g) Church Coombes: Cu; (h) Church Coombe:
NO
3 . Shaded bands indicate the region of ±20% around the average value.
390
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
(a)
(b)
1200
1.2e+5
T
I
1.0e+5
600
400
Flux (mg min-1)
800
Y Data
Flux (mg min-1)
1000
8.0e+4
6.0e+4
4.0e+4
2.0e+4
200
Quaking Houses: SO42-
Quaking Houses: Al
0.0
0
0
20
40
60
80
100
120
140
160
0
180
20
40
60
Flow Rate (L min-1)
(c)
(d)
14
120
140
160
180
3000
2500
10
Flux (mg min-1)
Flux (mg min-1)
100
E
I
12
80
Flow Rate (L min-1)
8
6
4
2000
1500
1000
2
500
0
Grattendale: SO42-
Grattendale: Zn
0
0
200
400
600
800 1000 1200 1400 1600 1800
0
200
400
600
Flow Rate (ml min-1)
7000
Flux (mg min-1)
(f)
8000
E
2.8e+5
D
2.7e+5
2.6e+5
6000
Flux (mg min-1)
(e)
5000
4000
3000
2.5e+5
2.4e+5
2.3e+5
2.2e+5
2000
2.1e+5
Ynysarwedd: Mn
1000
Ynysarwedd: Fe
2.0e+5
23
24
25
26
27
28
29
23
30
24
25
Flow Rate (L min-1)
(g)
26
27
28
29
30
Flow Rate (L min-1)
(h)
7000
80000
T
T
6000
60000
5000
Flux (mg min-1)
Flux (mg min-1)
800 1000 1200 1400 1600 1800
Flow Rate (ml min-1)
4000
3000
2000
40000
20000
1000
Church Coombes: NO3-
Church Coombes: Cu
0
0
0
20
40
60
Flow Rate (L s-1)
80
100
0
20
40
60
80
100
Flow Rate (L s-1)
Fig. 5. Flux versus flow rate for representative elements from the four field sites. (a) Quaking Houses: Al; (b) Quaking Houses: SO2
4 ;
(c) Grattendale: Zn; (d) Grattendale: SO2
;
(e)
Ynysarwedd:
Mn;
(f)
Ynysarwedd:
Fe;
(g)
Church
Coombe:
Cu;
(h)
Church
Coombe:
4
NO
3.
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
391
Table 2
Element release mechanisms
Field
QH
D
Y
CC
Si
I
I
E
Al
Fe
Mg
Ca
Mn
Na
K
Zn
I
T
I
E
I
E
E
E
T
I
I
I
E
E
I
I
E
E
E
I
I
E
I
I
I
I
I
E
I
Cu
Pb
As
I
I
I
T
S
T
E
E
E
N
Cl
F
I
I
I
E
D
I
T
I: general transport; T: transport control; E: equilibrium control; D: depletion control.
Table 3
Composition of samples
QB
Da
W
YS
YM
YCb
CC
Major elements (wt.%)
SiO2
66.59
TiO2
0.88
Al2O3
16.04
5.9
Fe2O3
Mn3O4
0.05
MgO
1.07
CaO
0.07
Na2O
0.89
K2O
2.72
P2O5
0.13
<0.1
SO3
Cr2O3
0.02
SrO
0.01
ZrO2
0.03
BaO
0.07
NiO
<0.01
CuO
<0.01
ZnO
<0.01
PbO
<0.01
LOI
4.89
Total
99.36
34.83
0.72
20.3
5.01
<0.01
0.5
0.2
0.19
2.04
0.05
<0.1
0.02
<0.01
<0.02
0.07
<0.01
<0.01
<0.01
<0.01
35.48
99.41
1.3
0.01
0.14
0.35
0.11
0.13
37.9
0.13
0.03
<0.01
10
<0.01
0.21
<0.02
14.35
<0.01
0.01
7.28
0.18
24.41
96.54
3.22
<0.01
0.13
0.15
0.05
0.83
53.7
<0.05
<0.05
0.02
<0.1
<0.01
0.04
<0.02
0.09
<0.01
<0.01
0.1
<0.01
41.93
100.26
82.76
0.37
7.65
2.46
0.05
0.59
0.34
0.71
1.63
0.08
<0.1
0.01
<0.1
<0.02
0.03
<0.01
<0.01
<0.01
<0.01
2.82
99.5
44.89
0.85
22.17
4.61
0.11
0.95
0.43
0.47
3.65
0.29
<0.1
0.02
0.02
0.02
0.08
0.01
0.01
<0.01
<0.01
21.05
99.63
2.76
nd
1.83
13.61
0.08
0.78
1.59
<0.05
<0.05
nd
nd
nd
nd
nd
nd
<0.05
<0.05
nd
nd
75.34
95.99
75.28
0.62
11.85
5.22
0.52
0.88
0.03
0.08
2.50
0.01
nd
nd
nd
nd
nd
nd
nd
nd
nd
3.78
101.97
Trace elements (ppm)
Ni
38
Cu
28
Zn
108
As
6
Sr
nd
Ba
nd
Pb
18
42
129
114
131
nd
nd
52
31
164
>2000
<1
>1500
>5000
>1000
4
7
819
2
348
367
88
24
16
73
10
nd
nd
8
52
125
62
13
nd
nd
36
nd
nd
nd
nd
nd
nd
nd
35
<150
232
56
18
361
70
1.28
0.00
0.24
5.36
0.47
0.00
0.22
0.06
0.8
0.00
0.00
0.31
0.99
0.04
0.04
0.18
1.05
0.21
0.21
8.4
0.52
0.00
0.01
0.03
6.7
QO
WSS (% S)
ASS (% S)
Total S (wt.%)
BET (m2 g1)
0.04
0.03
1.90
8.66
2.86
2.86
nd: not analysed.
a
Low totals for D result from incomplete oxidation of sulphides during ashing process and/or the presence of fluorine-bearing minerals.
b
YC was digested in aqua-regia before analysis of solution by ICP-AES. All other samples by XRF (bead and pellet).
(XRF) on glass beads and pellets respectively
(Table 3). Mineral assemblages and textural relationships were identified using X-ray diffraction
(XRD) and scanning electron microscopy (SEM)
at the Department of Materials Engineering, Sheffield University, and optical microscope examination of thin sections. Total S was determined by
LECO combustion analysis at the Assay Office,
392
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
Sheffield, and water- and acid-soluble SO2
were
4
obtained using standard methods (Czerewko,
1997). Disulphide-S was calculated from the difference between total S and the sum of acid plus water
soluble SO2
4 . Organic S was assumed to be negligible (Taylor, 1989). The specific surface area of the
crushed samples was measured by BET analysis at
the Department of Engineering Materials, University of Sheffield.
3.2.3. Mineral modes
Mineral modes (Table 4) were calculated for the
observed mineral assemblage by solving the set of
simultaneous equations that relates bulk composition to mineral modes and compositions (c.f.
Ferry, 1988). This set consists of i equations of
the form:
X
Ni ¼
ni;j pj ;
ð1Þ
j
where Ni is the number of moles of element i per
unit mass, ni,j is the number of moles of element i
in mineral j and pj is the number of moles of mineral j per unit mass. Mineral compositions are
poorly constrained for minerals that exhibit significant compositional variation. This is because the
fine-grained nature of the samples made representative mineral compositions difficult to quantify using
techniques such as electron microprobe on single
grains. For this reason, important mineral composition variables, such as Tschermak’s substitution
in illite, were accounted for by specifying the
exchange component to be an unknown in the set
of equations: for example, the anorthite component
in albite was specified as CaAlNa1Si1 (notation
of Thompson, 1982). The use of exchange components is discussed further by Thompson (1982).
Water-soluble, acid-soluble and disulphide-S were
included as compositional constraints. This allowed
a maximum of 10 variables, including mineral
modes and composition variables, to be determined. The plausibility of results was checked by
a comparison of the mass of the calculated assemblage to measured mass. Ideally, the assemblage
calculated from the analysis of 100 g of material
should weigh 100 g, and incorrect choices of assemblage or mineral formulae will result in deviation
from this value.
Table 4
Mineral modes
(Weight %)
QO
Albite
Illite
Kaolinite
Montmorillonite
Pyrite
Quartz
Galena
Barytes
Sphalerite
Calcite
Dolomite
Siderite
Alunite
Fe(OH)3
Gibbsite
Gypsum
Jarosite
Pb(OH)2/CO3
Zn(OH)2/CO3
6.2
40
Total
X K,ill (set)
X Mg,ill
X Na,mont
X an
Total (g)
7.7
0.2
41
QB
31
34
6.9
3.9
11
D
W
0.07
1.2
2.9
22
7.0
66
0.6
6.2
YS
YM
20
8
51
20
0.6
65
0.2
15
CC
0.8
3.3
1
30
4
15
1
0.04
58
0.1
89
7
0.6
7.4
0.25
5.0
YC
16
4.2
0.04
10
0.6
0.13
6
0.01
0.01
0.01
0.96
0.38
0.9
1.7
0
0
0.7
0.8
0.7
0.2
0.7
0.7
0.2
2.6
3.9
1.8
0.6
0.23
0.99
0.004
97
0.6
0.08
0.58
0.6
0.1
77
101
104
98
85
0.15
42
100
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
3.2.4. Identification of source and sink minerals
If a solution is saturated with respect to a mineral
then the saturation index of that mineral in solution
is 0. The saturation index is the base 10 logarithm of
the observed Ion Activity Product of the stoichiometric mineral dissolution reaction of interest divided
by the conditional thermodynamic constant for the
corresponding stoichiometric reaction for mineral
solubility (e.g. Appelo and Postma, 1993). A positive saturation index indicates that the solution is
supersaturated with respect to the mineral phase, a
negative value, that the solution is undersaturated,
and a value near zero suggests that saturation is
reached, and that equilibrium exists between the
mineral and the aqueous solution. Mineral saturation indices for minerals observed by XRD, SEM
or thin section observation (Table 5) were calculated
using PhreeqC for Windows, v1.0 (Parkhurst and
Appelo, 1999) utilising the PhreeqC database (Ball
and Nordstrom, 1991). This was used to test diagnoses of equilibrium-controlled element release
rates (Table 2). Only solutions with a calculated
Table 5
Mineral saturation indices
Q
Albite
Chlorite
Illite
Kaolinite
Montmorillonite
Pyrite
Quartz
Chalcopyrite
Galena
Sphalerite
Alunite
Fe(OH)3
Gibbsite
Gypsum
Jarosite
Calcite
Dolomite
Siderite
Anglesite
Cerussite
Cuprite
Pb(OH)2
Smithsonite
U
D
Y
CC
U
S
S
U
S+
S+
S+
U
S
U
U
S
S+
S
U
S
U
U
U
S
S
S
S
S
U
U
U
U
U
S
S
U
U
S+
S
S
S
U
S
S
S
S
S
S
S
S
S
U
S
S
U
S
S
S
S
U
S
S
S
S
S
U
U
S
U: undersaturated, SI < 5. S: close to saturation 5 < SI < 1.
S: Saturated: 1 < SI < 1; S+: close to supersaturation,
1 < SI < 5. These relatively wide limits are used to compensate for
errors in the calculated S.I. as a result of poorly known thermodynamic data for the phases of interest.
393
ion balance of ±10% were used to calculate mineral
saturation indices.
3.3. Area-normalised element fluxes
Comparison between laboratory and field element release rates requires some methods of conversion between the element fluxes from field sites
(M T1), and the element fluxes per unit area of
reacting mineral surface obtained from laboratory
experiment (M L2 T1). Element fluxes per unit
area are related to element fluxes via
Fi ¼
Qi
;
V q SSA
ð2Þ
where Fi is the flux of element i in units of
moles m2 s1, Qi is the flux of element i calculated
from field observations (moles s1), V is the volume
of interacting rock in m3, q is the bulk density in
kg m3, and SSA is the reacting surface area in m2
kg1. Uncertainties in the values of V and SSA
are significant. This is because V is controlled partly
by poorly known effects of hydrological focussing
such as fracture flow, and because the SSA of
samples in the field is less than that of the crushed
material for which BET analyses are available.
SSA, which is sensitive to both the freshness of
the mineral surface and to grain size, is difficult to
quantify, even for laboratory samples.
Here, three different approaches are considered.
A maximum value for the SSA, and thus a minimum value for the area-normalised flux, is given
by the use of BET-derived SSA (SSABET). This flux
is described and tabulated (Tables 6 and 7) as FB.
Alternatively, SSABET can be scaled using assumptions of particle size variation between laboratory
and field. The SSA decreases by an order of magnitude for each order of magnitude increase in particle
radius. Flux estimates derived by this method are
referred to here as adjusted BET-derived fluxes
(FABET: not tabulated). However, this method does
not account for the increase in mineral surface reactivity that is caused by crushing of the samples
before BET measurement. A minimum value for
the SSA, and thus a maximum value for the areanormalised flux, is given by a purely geometric
approach to surface area calculation. The simplest
method is to assume that the volume of material
that interacts with infiltrating fluids is divided into
cubic blocks with side length d m. This assumption
gives a total reacting surface area of 6V/d, and a
surface area normalised flux of Qid/6V. The ratio
394
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
Table 6
Area-normalised field fluxes of elements
FB (moles m2 s1)
Si
Al
Fe
Mg
Mn
Ca
Na
K
Zn
Pb
Cu
S
Cl
F
N
FG/FB:d = 0.01
FG/FB:d = 0.1
FG/FB:d = 1
QH flux
QH Ln
flux
QH Ln
r(flux)
4.4E16
2.1E16
7.9E15
1.2E16
1.5E14
1.6E13
1.2E14
4.8E17
35
36
32
37
32
29
32
38
2.1
2.1
2.1
2.1
2.0
2.1
2.1
2.1
1.4E14
1.2E13
32
30
2.1
2.1
8.6 E3
8.6 E4
8.6 E5
D Flux
D Ln
flux
D Ln
r(flux)
Y flux
Y Ln
flux
Y Ln
r(flux)
CC Flux
CC Ln
flux
CC Ln
r(flux)
1.1E14
40
2.2
1.5E13
37
2.2
4.9E13
2.9E14
2.3E15
2.4E16
6.9E18
36
39
41
44
47
2.1
2.2
2.3
2.1
2.4
1.3E15
6.7E17
1.0E14
2.3E14
2.4E16
2.6E14
2.3E14
2.5E15
4.9E18
1.4E18
34
37
32
31
36
31
31
34
40
41
2.1
2.4
2.0
2.0
2.1
2.0
2.0
2.0
5.3
2.5
2.0E18
2.2E19
7.0E20
2.9E18
1.4E20
3.9E18
1.2E17
1.4E18
5.0E20
41
43
44
40
46
40
39
41
44
2.1
2.1
2.7
2.1
2.1
2.1
2.1
2.1
2.1
2.5E14
4.0E14
9.5E15
4.5E14
39
39
40
38
2.1
2.2
2.1
2.2
7.0E14
30
2.0
3.0E19
4.2E18
1.1E17
5.1E19
3.4E18
43
40
39
42
40
2.1
2.1
2.1
2.1
2.1
1.9 E3
1.9 E4
1.9 E5
3.0 E3
3.0 E4
3.0 E5
2.9 E4
2.9 E5
2.9 E6
Table 7
Comparison between laboratory and field results for Ynysarwedd mudstone
Mechanism
Si
Al
Fe
Mg
Mn
Ca
Na
K
S
Field
Column
Batch
I
E
S
I
E
ID
I
ID
E
E
X, T
S
I
X, T
S
E
X, T
S
E
X, T
E
X
S
E
T
S
Release rates (moles m2 s1)
Field
Column
Batch
1.3E15
1.63E14
1.36E12
6.7E17
4.11E15
n/a
FG/FB
1.0E14
2.84E14
n/a
2.3E14
1.98E13
6.80E13
2.4E16
4.90E14
2.1E13
2.6E14
1.82E13
7.6E13
2.3E14
1.15E13
n/a
2.5E15
1.04E13
7.6E12
7.0E14
6.49E13
2.0E12
3 E4a
3 E2
3 E2
I: general transport; ID: indeterminate; D: depletion; E: equilibrium; X: ion exchange; T: transport; S: surface kinetics.
a
Calculated for d value of 0.1 m.
of the maximum flux (FG) to the minimum flux (FB)
is then 6/(dq SSABET). This value is tabulated in
Table 6 for d values of 0.01, 0.1 and 1 m.
V values are calculated from geometric and
hydrological considerations at each field site (see
below). Uncertainties in FB were propagated assuming that the input parameters, Qi, V, q, and SSA,
were uncorrelated. Calculations were made using
the standard expression for propagation of the
uncertainty on x where x = f(a, b, c . . .):
2 2 2
ox
ox
ox
2
ra þ
rb þ
rc þ rx ¼
ð3Þ
oa
ob
oc
Error propagation used a natural log scale because
of the large size of the uncertainties involved. Uncer-
tainties on Qi at Quaking Houses, Grattendale and
Church Coombe were set to the standard deviation
of the flux measurements. Flux uncertainties at Ynysarwed were set to twice the standard deviation of
the measurements to account for the disturbance
to flow rates caused by construction of the mine
effluent treatment plant. Uncertainties in q and
SSA were set to 10% and 20% of their respective
measured values. Note that this strategy does not include systematic error that stem from the inappropriate choice of surface area estimation method;
results that use the same method for surface area
estimation can be compared, but additional uncertainty should be added if those results are to be compared with results that use a different method of
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
surface area estimation. The bulk of the uncertainty
was attributed to V via a standard deviation of ±2
natural log units. Uncertainties on FG are of a similar order of magnitude for any given value of d.
3.3.1. Masses of interacting rock (Vq)
At Quaking Houses the volume of the tip, and
the proportion that interacts with the discharge at
outfall G were calculated from plans of the site,
and from borehole spoil logs (Pritchard, 1997).
The interacting proportion was estimated to be 0.1
of the total volume of the tip (Pritchard, 1997):
1.08 · 105 m3. The contribution of groundwater
from beneath the tip was assumed to be negligible
because the underlying clay has a very low permeability. The average density of the material was
assumed to be 1.8 · 103 kg m3 (Pritchard, 1997).
At Grattendale, the recharge area was calculated
from output flow rate, rainfall figures and evapotranspiration estimates. For an average output flow
rate of 800 ml s1, rainfall of 111 cm a1 (data for
1999–2000 taken from rainmeter at nearby Birchover) and 50% evapotranspiration, the area of
recharge was calculated to be 0.044 km2. The average height of the ground above the level of the adit
portal was estimated, using contours from the geological map (Stevenson et al., 1982), to be 177 m.
The interacting rock volume (7.79 · 106 m3) is the
product of these two parameters. The average
density of the material was assumed to be 2.3 ·
103 kg m3. This was calculated assuming 83% carbonate (q = 2.8 · 103 kg m3) and 17% void, as
either porosity or mined area.
At Ynysarwed, rock interacting with percolating
fluids was assumed to be collapsed mudstones and
coal filling 80% of the previously mined volume of
1.8 · 106 m3 (Younger et al., 1996; Younger and
Adams, 1997). The average density of the material
was assumed to be 1.7 · 103 kg m3. The density is
that of spoil (Pritchard, 1997), which has been
reduced to reflect the presence of 17% (by weight)
of coal with a density of 1.2 · 103 kg m3.
The catchment area for the Church Coombe discharge was identified using a digital elevation
model (DEM) of the region with a 10 m horizontal
grid spacing. The DEM was used with a runoff
algorithm to generate drainage networks (Jenson
and Domingue, 1988), and catchments. The catchment area for the adit at Church Coombe is
2.1 km2. The base of the interacting rock volume
was taken to be at the level of the adit portal
(123 m OD), and the top was taken to be the
395
ground surface. The estimated rock volume is
1.4 · 108 m3. The average density of the material
was assumed to be 2.6 · 103 kg m3, taken from
the lower range of estimates of density for crustal
rocks (Verhoogen et al., 1970).
3.4. Comparison with laboratory work
Rate-controlling mechanisms and area-normalised
element fluxes from Ynysarwed are compared with
rate-controlling mechanisms and area-normalised
element fluxes for batch and column weathering
studies performed on the Ynysarwed mudstone
(YM) material (Evans and Banwart, 2006).
4. Results
4.1. Water sampling and analysis
Representative element release patterns for concentration versus time (Fig. 2), concentration versus flow rate (Fig. 3), element flux versus time
(Fig. 4) and element flux versus flow rate (Fig. 5)
illustrate the wide range of behaviour observed.
Rate-controlling mechanisms are summarised in
Table 2. Table 2 shows that, for just over half
the elements analysed, the rate-controlling mechanism is general transport. This is reflected by the
scatter in many of the plots in Figs. 2–5. Equilibrium-controlled element release is common at
Grattendale (Mg, Ca, Na, S: Figs. 2–5d, Cl and
F), Ynysarwed (Mg, Mn: Figs. 2–5e, Ca, Na, K,
S), and Church Coombe (Si, Fe, Ca, Na, S, Cl).
Transport-controlled element release is rarer, and
is shown at Quaking Houses (Fe, Mn, S: Figs.
2–5b) and Church Coombe (Cu, N: Figs. 2–5g,
and F: Figs. 2–5h). There is no evidence of surface-kinetic controlled release rates. Other noteworthy features include:
• The decline in Fe concentrations with time at
Ynysarwed (Fig. 2f), which is not accompanied
by a significant decrease in Fe flux. This contrasts
with the relatively constant Mn concentrations
(Fig. 2e).
• The contrast between patterns for Cu and N at
Church Coombe. Concentration versus time
plots (Fig. 2g and h) and concentration versus
flow rate (Fig. 3g and h) show opposite trends.
However, the two elements show identical trends
in the element flux versus time (Fig. 4g and h)
and element flux versus flow rate (Fig. 5g and h).
396
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
• The generally positive trend in the element flux–
flow rate plots (Fig. 5). This reflects, to some
extent, the fact that flux is a function of flow
rate, and that the axes are not independent.
Nevertheless, elements whose release rates are
inferred to be equilibrium- or transport-controlled
exhibit stronger correlations than those whose
are not.
to enable small quantities of secondary minerals
such as gypsum to be shown. The total masses of
the calculated assemblage are close to 100 g in the
majority of cases (Table 4), validating the approach
taken here. However, the mass of the calculated
assemblage is significantly less than 100 g for QB,
YM, and YC. QB and YM are clay-rich and underestimation of the calculated mass in these cases is
attributed to the presence of mixed-layer clays of
unknown stoichiometry. The discrepancy for YC
is of less concern as it reflects the absence of C from
the modal calculations.
Most elements have multiple potential sources
and sinks at most sites (Fig. 6). For example, Si at
Quaking Houses could come from albite, illite, kaolinite, montmorillonite or quartz. Lead at Grattendale could come from either primary galena or
from secondary Pb hydroxide, sulphate or carbonate precipitates. Sulphur has multiple sources and
sinks at all of the sites except for Church Coombe.
4.2. Mineralogy
Bulk composition, mineral modes and mineral
saturation indices are shown in Tables 3–5. Mineral
saturation indices exhibit limited seasonal variation
and so the results in Table 5 indicate the predominant values for saturation index over the year-long
study period. Fig. 6 combines a graphical representation of mineral modes with a depiction of the elements present in each mineral. The log scale is used
(b)
100
QO
QB
Wt % Mineral
10
1
0.1
S
Cu
Pb
Zn
K
Na
Ca
Mn
Mg
Fe
Al
Si
100
D
W
S
Cu
Pb
Zn
K
Na
Ca
Mn
Mg
Fe
Al
Si
10
Wt % Mineral
(a)
1
0.1
0.01
0.01
S
Cu
Pb
Zn
K
Na
Ca
Mn
Mg
Fe
Al
Si
10
Wt % Mineral
Wt % Mineral
O3
2/C
O3
CC
H)
2/C
H)
100
(O
S
Cu
Pb
Zn
K
Na
Ca
Mn
Mg
Fe
Al
Si
Pb
(O
Zn
ite
um
ps
Gy
e
lom
Do
lcit
0.01
Ca
ite
0.1
es
1
ler
10
ha
Mineral
(d)
100
YS
YM
YC
Sp
Mineral
(c)
ryt
a
len
Ga
Ba
z
ar t
e
Illit
Qu
ite
ros
Ja
um
ps
Gy
e
sit
bb
Gi
3
H)
(O
Fe
e
nit
Alu
z
ar t
Qu
e
rite
nit
Py
llo
ori
ntm
Mo
e
nit
oli
Ka
e
Illit
ite
Alb
0.001
1
0.1
0.01
e
3
H)
(O
Fe
z
ar t
Qu
rite
Py
nit
oli
Ka
e
Illit
ite
Alb
ite
ros
Ja
um
ps
Gy
e
nit
Alu
te
eri
Sid
ite
e
lcit
lom
Do
z
ar t
Ca
Qu
rite
Py
ite
olin
Ka
e
Illit
Mineral
Mineral
Fig. 6. Mineral modes (left axis) and the relationship between mineral phases and dissolved elements in solution (right axis). (a) Quaking
Houses. QO: Quaking Houses orange; QB: Quaking Houses black. (b) Grattendale. D: ore-bearing limestone; W: ore-free limestone.
(c) Ynysarwedd. YS: sandstone; YM: mudstone; YC: coal. (d) Church Coombe. Spots indicate the presence of an element (RH y axis) in a
mineral (x axis)
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
No mineral source of Cl or N was identified. Modes
of F-bearing minerals such as fluorite and fluor-apatite were not calculated because the atomic number
of F was too low for accurate analysis of this element by XRF. However, it is likely that fluorite
was present at Grattendale, which is close to the
famous Derbyshire Blue John fluorite mines, and
that fluor-apatite was present at Church Coombe
(apatite identified in thin section).
The diagnosis of equilibrium-controlled element
release from concentration-flow rate characteristics
is backed up, in most cases, by calculated chemical
saturation of the solution with respect to at least
one source or sink mineral identified as present
within the rock samples (compare Tables 2 and 5).
A notable exception is Cl at Grattendale and Ynysarwed; this element is likely to have behaved conservatively during fluid–rock interaction within the
mine system, and the concentration is probably
determined externally to the adit systems. The
rate-controlled mechanism is therefore deemed to
be general transport (I). Other exceptions are provided by Na at Gratttendale, Ynysarwed and
Church Coombe, Mg at Ynysarwed, and Ca at
Ynysarwed and Church Coombe. These elements
are commonly found as part of cation exchange
assemblages in clays, and it is likely that cation
exchanger surfaces on clay minerals (illite, kaolinite,
montmorillonite) buffered the aqueous concentrations of these elements. The equilibrium-controlled
element release inferred for F at Grattendale is
likely to result from saturation with respect to fluorite. Solutions are predicted to be supersaturated
with respect to minerals in some cases, for example,
chlorite at Ynysarwed. This results partly from slow
kinetics of mineral precipitation for some minerals,
but may also reflect poorly known thermodynamic
data for the minerals, the effects of mineral solid
solution on mineral stability, and changes in solution parameters such as pH and Eh when the discharges are exposed to the atmosphere.
A number of minerals that are observed to be
present, and that are calculated to be saturated
in the discharges from outflow G at Quaking
Houses (alunite, jarosite, Fe oxy-hydroxides) do
not appear to exert an equilibrium control on the
release of their constituent elements (Fe, K, S,
Al). It is tempting to attribute this to dilution of
the tip discharge by relatively clean rainwater from
the road drains. However, if this were the case then
the saturation indices of the secondary minerals
would drop. This is not observed. It is possible
397
that the secondary minerals change composition
to remain in equilibrium with a seasonally changing fluid composition. If this is the case then the
assumption of relatively constant values for mineral compositions and fluid parameters such as
pH and O2 is invalid at a site such as Quaking
Houses, and mineral equilibrium would be unable
to fix the solute concentrations of the constituent
elements of the secondary minerals.
4.3. Area-normalised element fluxes
BET surface area-normalised field fluxes of
elements (FB: Table 6) range from 1013 to 1020
moles m2 s1. One sigma uncertainties are almost
all close to 2 natural log units, which reflects the
predominance of the contribution of the uncertainty in V to the total uncertainty. Larger uncertainties, such as Fe at Church Coombe, or Zn at
Ynysarwed record scatter in the initial flux data.
Fluxes of Na and Cl from Quaking Houses are
high (1013 moles m2 s1) and have very similar
values. This is likely to be the result of the dissolution of a large uncovered pile of road salt that was
stored unprotected at the tip at the time of the
monitoring program. Adjusted BET-derived fluxes
(FABET: not tabulated) are 1, 2, or 3 orders of magnitude higher than the FBET values, depending on
the estimate of the field grain size. Geometric surface area-normalised element fluxes (FG: Table 6)
are 3–6 orders of magnitude higher, depending
on the assumed value of d.
4.4. Comparison with laboratory work
A systematic variation in element release mechanism as a function of the type of system was
observed (Table 7). Equilibrium-controlled element
release was most common at field scale; transportcontrolled element release was predominant in the
column experiments, and surface-kinetic controlled
element release was observed in the batch experiments. There was also a systematic variation in the
rate of element release (Table 7; Fig. 7). Element
fluxes from the batch experiments were 1–2 orders
of magnitude faster than those from the column
experiments which were, in turn, 1–2 orders of
magnitude faster than BET-normalised field fluxes
(FB). Adjusted BET-normalised fluxes for field grain
sizes that were two orders of magnitude higher than
those in the columns (0.1 m) were similar to the
BET-derived fluxes from the column experiments.
398
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
-10
Si
Si
Log Batch Flux (mol m-2 s-1)
Log Column Flux (mol m-2 s-1)
-10
Al
Fe
-12
Mg
-14
Mn
Ca
-16
Na
K
-18
S
Mg
-12
Mn
Ca
-14
K
S
-16
-18
b
a
-20
-20
-20
-18
-16
-14
-12
-10
Log Field Flux (mol m-2 s-1)
-20
-18
-16
-14
-12
-10
Log Field Flux (mol m-2 s-1)
Fig. 7. Comparison between element fluxes from field (FB) and (a) column experiments and (b) batch experiments. Error bars indicate ±1
standard deviation.
GSA-normalised fluxes (FG) were higher in all cases,
but showed the same trend between batch and column results. Field FG values for d values of 0.1 m
were within error of the column FG values. The other
rock types give similar results (work in progress).
5. Discussion
5.1. Rate-controlling element release mechanisms
The most important result of this study is the predominance of equilibrium- and transport-controls
on element release rates (Table 2) which is observed
at four very different UK mine sites. This suggests
that direct scaling between laboratory-derived surface-kinetic controlled mineral dissolution rates
and equilibrium- or transport-controlled field rates
is unlikely to be appropriate. This conclusion is supported by the comparison between rate-controlling
mechanisms for the Ynysarwed material (Table 7).
Surface-kinetic processes dominate in the batch
experiments, transport and cation-exchange processes control element release in the column experiments, and equilibrium control is most important in
the field.
The occurrence of equilibrium-controlled element release presents something of a paradox; mineral dissolution processes that result in equilibrium
would normally be assumed to have proceeded
more rapidly than those which did not, because relatively rapid dissolution is required to produce solute concentrations that are high enough to reach
chemical saturation with respect to the minerals of
interest. Yet field rates have been shown to be
slower than laboratory-derived rates by 2–4 orders
of magnitude (e.g. Schnoor, 1990; White et al.,
1996). Consideration of surface-area-normalised
fluxes is not helpful in this case, because of the large
uncertainty on the surface-area appropriate for use
in field-based calculations. However, the apparent
contradiction can be resolved if hydrological factors
are considered. Long residence times and low
water:rock ratios are typical of field hydrological
conditions, and these factors favour equilibriumcontrolled mineral dissolution.
The scale of observation is also a controlling factor. At grain scale, solubility equilibrium occurs
when diffusion of solutes away from the surface is
slow compared to the rate of dissolution. At aquifer
scale, equilibrium results when hydrological flushing
is sufficiently slow that mineral dissolution proceeds
to equilibrium over a time-scale shorter than the residence time of water within the hydrological system.
A number of other scale-dependent factors that
influence the rate-controlling mechanism are also
very different in laboratory and field (see below).
5.2. Influences on rate-controlling mechanism
An understanding of the parameters that control
the rate-controlling mechanism of element release
requires consideration of the factors that favour
each mechanism. Equilibrium control is favoured
by factors or processes that lead to high element
concentrations. These can be split into hydrological
factors, such as low water:rock ratios, long residence times, and significant evaporation, and mineralogical factors, such as high mineral reactivity and
high reactive surface area. Field environments tend
to have lower water:rock ratios and longer residence
times than laboratory experiments, but lower mineral reactivities, which result from large particle
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
399
1
Water:Rock Ratio
Equilibrium Control
Equilibrium Control
Kinetic Control
Residence Time
Equilibrium Control
1
Mineral Reactivity
Fig. 8. Schematic illustration of some of the factors that determine the mechanism of element release during weathering.
sizes, and a general decrease in reactivity that results
from prolonged weathering (e.g. White and Brantley, 2003). The results of this study suggest that
the hydrological factors exert a greater influence,
at least at the four field sites examined here. This
analysis suggests that a continuum exists between
field sites with low fluid:rock ratios and high residence times that have a high potential for mineral
equilibrium and those with the opposite characteristics, that have a lower potential for mineral equilibrium (Fig. 8). Spoil tips would in general, be
expected to fall at the low potential for equilibrium
end of this scale, whereas large mined systems such
as Ynysarwed would be expected to fall at the
higher end. This conceptual model is supported by
the results in Table 2.
Transport-controlled element release is associated with sites that have relatively large variations
in hydrological parameters such as flow rate and
the height of the water table. Variations in these
parameters are more extreme at smaller, hydrogeologically isolated sites with low residence times,
and where water storativity is not high enough to
buffer changes in recharge rates driven by temporal
variation in rainfall and other precipitation. Quaking Houses is such a site, and exhibits transportcontrolled element release for more elements than
any of the other sites.
5.3. Mineral sources and sinks
The results of this study emphasize the importance of secondary minerals in the control of element fluxes from mine sites. Secondary minerals
are present and are calculated to exert some control
on element concentrations at each of the four sites
studied here (Table 5, Fig. 6). They have also been
shown to be important at a number of other types
of environment (e.g. Blowes, 1991; Benevenuti
et al., 2000). In such situations it is inappropriate
to interpret field fluxes of polluting elements from
laboratory kinetic studies of primary minerals. It
is instead necessary to identify dissolution/precipitation and adsorption reactions involving secondary
minerals, and quantify rates of element release with
appropriate mathematical expressions.
5.4. Recognition of rate-controlling mechanism
Recognition of rate-controlling mechanism for
element release requires both monitoring of discharge compositions and investigation of the mineralogy of material at a mine site. For example, use of
mineralogy and mineral equilibrium calculations
alone at Quaking Houses would have concluded that
the presence of alunite, Fe oxy-hydroxides, gibbsite,
alunite and jarosite controlled the release of Fe, Al,
S, and K from the site, a result apparently inconsistent with those of the water-quality monitoring
study reported here. Similarly, water analyses alone
could have led to the conclusion that Cl and N were
fixed by the presence of some mineral at Church
Coombe, but mineralogical analysis shows that this
is not the case and that some other mechanism is
required to explain their constant concentrations.
Water-quality monitoring should also be long-term
(at least a year). This allows recognition of monotonic temporal trends in element concentrations
400
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
and/or fluxes, for example, the declining Fe concentrations at Ynysarwed (Younger et al., 1996). More
detailed mineralogical analysis than that presented
here is necessary if the role of cation exchange is to
be understood and quantified at field sites.
5.5. Use of algorithms
The use of scaling algorithms (Sverdrup and
Warfvinge, 1995; Malmstrom et al., 2000) must be
undertaken with caution if there is reason to suspect
that element release rates are controlled by mineral
equilibrium. Application of the algorithms to sites
such as the Aitik mine site are likely to be successful
(e.g. Malmstrom et al., 2000) because the mine spoil
is porous and highly reactive, and water-residence
times are low (Eriksson et al., 1997; Stromberg
and Banwart, 1999). However, these algorithms
would not have been suitable for the majority of
the elements at the mine sites monitored for this
study, and different methods should be used.
Fortunately, the systematics of element concentrations controlled by mineral equilibrium are relatively simple; concentrations remain approximately
constant so long as the mineral assemblage and
other factors such as temperature and pH remain
unchanged. However, exhaustion of the controlling
mineral phase will lead to changes in the concentrations of elements in solution (e.g. Banwart and
Malmstrom, 2001), and the potential for such
changes must be taken into account in any long
term prediction exercise.
5.6. Practical implications
The practical implications of this work for prediction and remediation of contamination at mining
sites are twofold.
1. An integrated study of mineralogical and waterquality parameters is necessary if the mechanisms
that control element release at mine sites are to
be understood. The most poorly known parameters involved in the calculations presented here
are the volume of rock or spoil that interacts with
percolating water, and the effective surface area
of this material. More detailed hydrogeological
analysis of sites (e.g. Rivett and Allen-King,
2003) can be used to constrain the former parameter. Issues with the effective surface area are
more difficult to resolve. The effective surface
area is not equal to either the geometric or
BET-derived surface area, and, in spite of extensive research (e.g. Fischer and Gaupp, 2004;
Metz et al., 2005), there is no commonly used
method for measurement of this parameter. The
strategy recommended here is that effective surface area conventions should be consistent within
any study, and that the uncertainties introduced
by scaling between different particle sizes and/
or mineral reactivities should be incorporated
into error propagation calculations.
2. The presence of soluble secondary minerals such
as alunite and jarosite could provide a significant
source of contamination which can be accessed if
site hydrogeology changes significantly. This is
enhanced if the minerals incorporate toxic metals
(e.g. Winland et al., 1991). One possible example
of this is the change in water quality observed at
Quaking Houses after the construction of the
Annfield Plain by-pass. Options such as dry-capping by landscaping and soil-washing should
therefore be considered carefully.
6. Conclusions
Analysis of element concentration and element
fluxes suggest that equilibrium- and transport-controlled element release mechanisms control the concentrations of major elements and the principal
contaminant metals at the four UK mine sites studied. No evidence was found for surface-kinetic controlled element release. Mineralogical analysis and
mineral equilibrium calculations support this conclusion. However, diagnosis of rate-controlling
mechanism from trends in water analyses can be
incorrect if there is significant variation in the
hydrological characteristics of the site, if there is significant depletion of source minerals, or if the concentrations of conservative elements such as Cl
and N are controlled by processes external to the
system under consideration. Secondary minerals
such as goethite, gibbsite, alunite, jarosite and Pb
and Zn hydroxides, sulphates and carbonates play
an important role in the control of element concentrations from the sites. Comparison of element
fluxes from the mine sites with those from the laboratory requires that the field fluxes should be normalised to surface area. The biggest uncertainties
in such calculations are the volume of rock that
interacts with percolating solutions and the specific
surface area of reacting mineral grains in the field.
A comparison of field- and laboratory-derived
element fluxes for one of the sites shows that surface
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
kinetic mechanisms control the rate of element
release from the batch experiments, while transport
and equilibrium-control dominate in column experiments and in the field. Element fluxes from batch
experiments were 1–3 orders of magnitude faster
than those from column experiments and field measurements. This result presents a paradox, given
that surface-kinetic mineral dissolution would be
expected to proceed more slowly than that controlled by mineral equilibrium. This apparent contradiction can be resolved by the consideration of
hydrological parameters such as water:rock ratio
and water residence time. The implications for practical prediction and remediation of contamination
at mine sites are that water-quality monitoring
should be accompanied by mineralogical analysis,
and that the role of secondary minerals should be
considered carefully.
Acknowledgments
Thanks to Chris Rees, of the Cefn Coed Mining
Museum, Jim Rieuwerts from the Peak District
Mining Historical Society, Staff at Cornwall County
Records Office for help with local data sources. Also
to Adam Jarvis, Patrick Orme and Lesley Batty for
Quaking Houses analysis, to staff at the Assay
Office, Sheffield, to Bev Lane at the Centre for
Materials Engineering at the University of Sheffield
for BET analysis, and to Paul Younger for specific
and general knowledge. This work was funded by
NERC. Thanks also to two anonymous reviewers,
whose reviews have improved this work.
References
Adam, K., Kourtis, A., Gazea, B., Kontopoulos, A., 1997.
Evaluation of static tests used to predict the potential for acid
drainage generation at sulphide mine sites. Trans. Inst. Min.
Metall. 106, A1–A8.
Appelo, C.A.J., Postma, D., 1993. Geochemistry, Groundwater
and Pollution, vol. 1. A.A. Balkema, Rotterdam.
Ball, J.W., Nordstrom, D.K., 1991. WATEQ4F: users manual
with revised thermodynamic data base and test cases for
calculating speciation of major, trace and redox elements in
natural waters. Open-File Report 90-129, US Geol. Surv.,
Denver, Colorado.
Banks, D., Younger, P.L., Arnesen, P.L., Iverson, E.R., Banks,
S.B., 1997. Mine-water chemistry: the good, the bad and the
ugly. Environ. Geol. 32, 157–174.
Banwart, S.A., Malmstrom, M.E., 2001. Hydrochemical modelling for preliminary assessment of minewater pollution. J.
Geochem. Explor. 74, 73–97.
Banwart, S.A., Evans, K., Croxford, S., 2002. Predicting mineral
weathering rates at field scale for mine water risk assessment.
401
In: Younger, P.L., Robins, N.S. (Eds.), Mine Water Hydrogeology and Geochemistry. The Geological Society of London,
London, pp. 137–157.
Banwart, S.A., Zhang, S., Evans, K.A., 2004. Resolving the scale
dependency of laboratory and field weathering rates. In:
Wanty, R.B., Seal, R.R. (Eds.), Proc. 11th Internat. Symp.
WRI-11. Balkema, Leiden, The Netherlands, pp. 1443–1447.
Barclay, W.J., Taylor, K., Thomas, L.P., 1988. Geology of the
South Wales Coalfield. Part V. The Country around Merthyr
Tydfil. British Geological Survey, Keyworth, UK.
Benevenuti, M., Mascaro, I., Corsini, F., Ferrari, M., Lattanzi,
P., Parrini, P., Costagliola, P., Tanelli, G., 2000. Environmental mineralogy and geochemistry of waste dumps at the
Pb(Zn)–Ag Bottino mine, Apuane Alps, Italy. Eur. J. Min.
12, 441–453.
Blowes, D.W., 1991. The formation and potential importance of
cement layers in active sulphide mine tailings. Geochim.
Cosmochim. Acta 55, 965–978.
Bottrell, S.H., Hardwick, P., Gunn, J., 1999. Sediment dynamics
in the Castleton karst, Derbyshire, UK. Earth. Surf. Proc.
Landforms 24, 745–759.
Bottrell, S.H., Webber, N., Gunn, J., Worthington, S.R.H., 2000.
The geochemistry of sulphur in a mixed allogenic–autogenic
karst catchment, Castleton, Derbyshire, UK. Earth. Surf.
Proc. Landforms 25, 155–165.
Clow, D.W., Drever, J.I., 1996. Weathering rates as a function of
flow through an alpine soil. Chem. Geol. 132, 131–141.
Czerewko, M.A., 1997. Diagenesis of mudrocks, illite ‘crystallinity’ and the effects on engineering properties. Unpubl. Ph.D.
thesis. University of Sheffield.
Dines, H.G., 1956. The Metalliferous Mining Region of SouthWest England. Vol. 1. Memoirs of the Geological Survey of
Great Britain. Her Majesty’s Stationary Office, London.
Eriksson, N., Gupta, A., Destouni, G., 1997. Comparative
analysis of laboratory and field tracer tests for investigating
preferential flow and transport in mining waste rock. J.
Hydrol. 194, 143–163.
Evans, K.A., Banwart, S.A., 2006. Rate controls on the chemical
weathering of natural polymineralic material. I. Dissolution
behaviour of polymineralic assemblages determined using
batch and unsaturated column experiments. Appl. Geochem.
21, 352–376.
Evans, K.A., Gandy, C.J., Banwart, S.A., 2003. Mineralogical,
numerical and analytical studies of the coupled oxidation of
pyrite and coal. Min. Mag. 67, 381–398.
Ewbank, G., Manning, D.A.C.., Abbott, G.D., 1995. The relationship between bitumens and mineralisation in the South Pennine
Orefield, Central England. J. Geol. Soc. Lond. 152, 751–765.
Ferry, J.M., 1988. Infiltration-driven metamorphism in northern
New England, USA. J. Pet. 29, 1121–1159.
Fischer, C., Gaupp, R., 2004. Multi-scale rock surface area
quantification – a systematic method to evaluate the reactive
surface area of rocks. Chemie der erde-geochemistry 64, 241–
256.
Ford, T.D., Rieuwerts, J.H., 1970. Lead Mining in the Peak
District. Peak Park Planning Board, Bakewell, Derbyshire.
Gandy, C.J., 2002. Reactive transport modelling of pyrite
oxidation at Quaking Houses, County Durham. Unpubl.
Ph.D. thesis. University of Newcastle-upon-Tyne.
Gandy, C.J., Evans, K.A., 2002. Laboratory and numerical
modelling studies of iron release from a spoil heap in County
Durham. In: Robins, N.S., Younger, P. (Eds.), Minewater
402
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
Hydrogeology and Geochemistry. Geological Society, London, pp. 205–214.
Gandy, C.J., Younger, P.L., 2003. Effect of a clay cap on
oxidation of pyrite within mine spoil. Quart. J. Eng. Geol. 36,
207–215.
Gleeson, S.A., Wilkinson, J.J., Stuart, F.M., Banks, D.A., 2001.
The origin and evolution of base metal mineralising brines
and hydrothermal fluids, South Cornwall, UK. Geochim.
Cosmochim. Acta 65, 2067–2079.
Hamilton-Jenkin, A.K., 1965. Mines and Miners of Cornwall. X.
Camborne and Illogan. Truro Bookshop, Truro.
Holmes, P.R., Crundwell, F.K., 2000. The kinetics of the oxidation
of pyrite by ferric ions and dissolved oxygen: an electrochemical study. Geochim. Cosmochim. Acta 64, 263–274.
Jambor, J.L., 2000. Mineralogy and acid rock drainage. In:
Cotter Howells, J.D., Campbell, L.S., Valsami-Jones, E.,
Batchelder, M. (Eds.), Environmental Mineralogy: Microbial
Interactions, Anthropogenic Influences, Contaminated Land
and Waste Management. Mineralogical Society of Great
Britain and Ireland, London, pp. 141–159.
Jenson, S., Domingue, J., 1988. Extracting topographic structure
from digital elevation data for geographic information
systems analysis. Photogramm. Eng. Rem. Sensing 54,
1593–1600.
Malmstrom, M.E., Destouni, G., Banwart, S.A., Stromberg,
B.H.E., 2000. Resolving the scale-dependence of mineral
weathering rates. Environ. Sci. Technol. 34, 1375–1378.
Markey-Amey, J., 1995. Stanley Burn Dissertation. University of
Sunderland.
Martello, D.V., Vecchio, K.S., Diehl, J.R., Graham, R.A.,
Tamilia, J.P., Pollack, S.S., 1994. Do dislocations and
stacking faults increase the oxidation rate of pyrite? Geochim.
Cosmochim. Acta 58, 4657–4665.
Metz, V., Raanan, H., Pieper, H., Bosbach, D., Ganor, J., 2005.
Towards the establishment of a reliable proxy for the reactive
surface area of smectite. Geochim. Cosmochim. Acta 69,
2581–2591.
Newbegin, D., 1997. Stanley Burn – Monitoring of Pollution.
Durham County Council, Durham.
Oakham, C.D., 1979. The sough hydrology of the WirksworthMatlock-Youlgreave area, Derbyshire. Unpubl. Ph.D. thesis.
University of Leicester.
Parkhurst, D.L., Appelo, C.A.J., 1999. Users guide to PhreeqC
(version 2): a computer program for speciation, batch
reaction, one dimensional transport and inverse geochemical
calculations. Water-Resour. Investig. Rep. 99-4259, US Geol.
Surv., Denver, Colorado.
Peiffer, S., Stubert, I., 1999. The oxidation of pyrite at pH7 in the
presence of reducing and non reducing Fe(III) chelators.
Geochim. Cosmochim. Acta 63, 3171–3182.
Pritchard, J., 1997. Leachate release from a large body of
contaminated land: Morrison Busty Spoil Heap, County
Durham. UnPubl. M.Sc. thesis. University of Newcastleupon-Tyne.
Ranson, C.M., Edwards, P.J., 1997. The Ynysarwed experience:
active intervention, passive treatment and wider aspects. In:
Younger, P.L. (Ed.), Minewater Treatment using Wetlands.
University of Newcastle, Newcastle-upon-Tyne, pp. 151–154.
Rieuwerts, J.H., 1987. History and gazeteer of the lead mine
soughs of Derbyshire. Privately Published.
Rivett, M.O., Allen-King, R.M., 2003. A controlled field experiment on groundwater contamination by a multicomponent
DNAPL: dissolved-plume retardation. J. Contam. Hydrol.
66, 117–146.
Schnoor, J.L., 1990. Kinetics of chemical weathering: a comparison between laboratory and field weathering rates. In:
Stumm, W. (Ed.), Aquatic Chemical Kinetics. Wiley, pp.
475–504.
Srour, S., 1998. Modelling groundwater flow through a body of
contaminated land: Morrison Busty Spoil Heap, County
Durham. Unpubl. M.Sc. thesis. University of Newcastle upon
Tyne.
Stallard, R.F., Eddmond, J.M., 1987. Geochemistry of the
Amazon. 3. Weathering chemistry and limits to dissolved
inputs. J. Geophys. Res. 92, 8293–8302.
Stevenson, I.P., Price, D., Chisholm, J.I., Wedd, C.B., Smith,
E.G., Eden, R.A., Rhys, G.H., 1982. Classical areas of British
Geology: Bakewell, Sheet SK26, 1:25000. Geological Survey
of England and Wales.
Stromberg, B., Banwart, S., 1999. Weathering kinetics of waste
rock from the Aitik copper mine, Sweden: scale dependent
rate factors and pH controls in large column experiments. J.
Contam. Hydrol. 39, 59–89.
Sverdrup, H., Warfvinge, P., 1995. Estimating field weathering
rates using laboratory kinetics. In: White, A.F., Brantley, S.L.
(Eds.), Chemical Weathering Rates of Silicate Minerals.
Mineralogical Society of America, pp. 485–541.
Taylor, R.K., 1989. Report GPC/20410. Mining Department,
National Coal Board.
Thompson, J.B.J., 1982. Composition space: an algebraic and
geometric approach. In: Ferry, J.M. (Ed.), Characterisation
of Metamorphism through Mineral Equilibria, Reviews in
Mineralogy. Mineralogical Society of America, Washington,
USA, pp. 1–31.
van Grinsven, J.J.M., van Riemsdijk, W.H., 1992. Evaluation of
batch and column techniques to measure weathering rates in
soils. Geoderma 52, 41–57.
Verhoogen, J., Turner, F.J., Wieiss, L.E., Wahrhaftig, C., Fyfe,
W.S., 1970. The Earth an Introduction to Physical Geology.
Holt Rinehart and Winston, New York.
White, A.F., Blum, A.E., Schulz, M.S., Bullen, T.D., Harden,
J.W., Peterson, M.L., 1996. Chemical weathering rates of a
soil chronosequence on granitic alluvium. 1. Quantification of
mineralogical and surface area changes and calculation of
primary silicate reaction rates. Geochim. Cosmochim. Acta
60, 2533–2550.
White, A.F., Blum, A.E., 1995. Effects of climate on chemicalweathering in watersheds. Geochim. Cosmochim. Acta 59,
1729–1747.
White, A.F., Brantley, S.L., 2003. The effect of time on the
weathering of silicate minerals: why do weathering rates differ
in the laboratory and field. Chem. Geol. 202, 479–506.
Wieland, E., Wehrli, B., Stumm, W., 1988. The coordination
chemistry of weathering. III. A generalization on the dissolution rates of minerals. Geochim. Cosmochim. Acta 52,
1969–1981.
Winland, R.L., Traina, S.J., Bigham, J.M., 1991. Chemicalcomposition of ochreous precipitates from Ohio coal-mine
drainage. J. Environ. Qual. 20, 452–460.
Xie, Z.X., Walther, J.V., 1992. Incongruent dissolution and
surface-area of kaolinite. Geochim. Cosmochim. Acta 56,
3357–3363.
Younger, P.L., 1996. Technical Audit of the Phase I Feasibility
Study on Long term remediation of the Neath Canal and
K.A. Evans et al. / Applied Geochemistry 21 (2006) 377–403
River Dulais Minewater Discharges. University of Newcastle
upon Tyne, Newcastle-upon-Tyne.
Younger, P.L., 1997. The longevity of minewater pollution: a
basis for decision-making. Sci. Total Environ. 194, 457–466.
Younger, P.L., 2000. Predicting temporal changes in total iron
concentrations in groundwaters flowing from abandoned
deep mines: a first approximation. J. Contam. Hydrol. 44,
47–69.
403
Younger, P.L., Adams, R., 1997. Improved modelling of abandoned coalfields. University of Newcastle-upon-Tyne, Newcastle-upon-Tyne.
Younger, P.L., Pennell, R., Cowpertwait, P.S.P., 1996. Evaluation of temporal changes in the quality of the lower
Ynysarwed minewater discharge, Neath Valley, South Wales.
Report from Nuwater Consulting Services Ltd., Dept. Civ.
Eng. University of Newcastle upon Tyne.