You are on page 1of 12

Applied Geochemistry 18 (2003) 97–108

www.elsevier.com/locate/apgeochem

Aqueous geochemistry in the Udden pit lake,


northern Sweden
Madeleine Ramstedta,*, Erik Carlssonb, Lars Lövgrena
a
Inorganic Chemistry, Department of Chemistry, Umeå University, SE-901 87 Umeå, Sweden
b
Division of Applied Geology, Luleå University of Technology, SE-971 87 Luleå, Sweden

Received 3 April 2001; accepted 17 January 2002


Editorial handling by B.A. Kimball

Abstract
The Udden pit lake in northern Sweden was studied from June 1998 to February 1999 in order to increase knowl-
edge of the geochemistry in lakes created as a result of decommissioning open pit mines. The vertical water profile in
the lake was sampled on 4 different occasions, in June, August, September and February. Water samples were analysed
for total concentrations of Fe, As, Cu, Cd, Zn, Pb, Al, Ca, K, Mg, Na, Mn, S, Cl, N and P. Temperature, concentra-
tion of dissolved O2, conductivity, pH, and redox potential were measured in situ at different depths. Four layers could
be observed in the lake during summer, and 3 layers during winter. A thermocline was observed during summer at a
depth of 5 m and on all 4 occasions a halocline was observed at a depth of 20 m, and a redoxcline at  35 m. Oxygen
concentration decreased dramatically at a depth of 20 m. pH increased downwards in the lake from 4.8 at the surface
to 6.4 at the bottom of the lake. Geochemical processes occurring in the lake, the origins of the layers, the metal con-
centrations and the anion concentrations are discussed in this article. # 2002 Elsevier Science Ltd. All rights reserved.

1. Introduction results in a pronounced stratification and with time, the


development of an anoxic bottom layer (Miller et al.,
Decommissioning of open pit mines in order to pre- 1996; Tempel et al., 2000). Doyle and Runnells (1997)
vent acid mine drainage often involves flooding the open stated that one of the most important variables con-
pit and thereby creating an artificial lake. In many cases trolling water circulation in lakes is the relative depth
waste rock is deposited in the pit prior to flooding. By (the relation between the depth and the diameter of the
flooding it is expected that O2 will be prevented from lake). Other important variables are the length of free wind
diffusing into mine wastes and rock walls, leading to a action and the wave fetch (Klapper and Schultze, 1997).
minimised oxidation of sulphide minerals and a subse- The distribution and speciation of trace metals in
quently minimised release of heavy metals and acidity stratified lakes are influenced by 3 major geochemical
(MEND, 1989). However, oxidation can still occur in processes; cycling across redox boundaries, sorption and
the absence of O2 but with Fe(III) as the oxidising agent coprecipitation on/with particles and formation of metal
(Herbert, 1999). complexes in aqueous solution with inorganic ligands
These artificial lakes are called pit lakes and usually (e.g. OH, HS) and organic matter (Balistrieri et al.,
have a small diameter to depth ratio compared to nat- 1994). Iron and Mn are important metals for controlling
ural lakes. This, together with the saline waters, often the dissolved concentration of trace metals in oxic
waters since particles of Fe (oxy)hydroxides and Mn
(oxy)hydroxides act as scavengers for trace metals.
* Corresponding author. Fax: +46-90-786-9195. Adsorption of metal ions onto Fe (oxy)hydroxides
E-mail address: madeleine.ramstedt@chem.umu.se increases with pH (Dzombak and Morel, 1990), while
(M. Ramstedt). elements like As, forming oxo-anions, are preferably
0883-2927/03/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0883-2927(02)00068-9
98 M. Ramstedt et al. / Applied Geochemistry 18 (2003) 97–108

adsorbed at lower pH (Miller et al., 1996; Lützenkirchen phide minerals at Udden were pyrite (FeS2), pyrrhotite
and Lövgren, 1998). The objective of the present work (Fe1-xS), sphalerite (ZnS), chalcopyrite (CuFeS2), galena
was not only to study and describe the aqueous geo- (PbS) and arsenopyrite (FeAsS). The average grade of
chemistry in a pit lake by field sampling and in situ the ore is shown in Table 1. The mining in Udden star-
measurements but also to discuss the obtained results in ted in 1971 with an open pit. After 3a of mining the
terms of geochemical processes such as weathering of surface deposit was depleted and the mining operations
rock walls, adsorption and coprecipitation of heavy continued underground. The mining ended in 1991. A
metals on Fe and Mn (oxy)hydroxides, sulphide pre- total of 690 000 ton ore and 1000 000 ton of waste rock
cipitation and oxidation of organic material. was excavated from the open pit (Boliden, 1980).
The major waste rock deposits are situated south of
the pit lake and cover an area of 12 ha (Fig. 1). The
2. Site description waste rock consists of approximately 1 wt.% S and 25
wt.% rocks with a buffering ability. The calculated
The Udden open pit is situated approximately 70 km concentration of calcite is less than 0.5 wt.% of the
west of Skellefteå in northern Sweden (Fig. 1). The buffering rock. The waste rock deposit has been covered
Udden sulphide ore deposit was part of the Skellefte with till (Boliden, 1991). The decommissioning pro-
district volcanic series. The bedrock in the area consists gram, based on flooding the open pit, started in 1991.
mainly of tuffite, some amphibolite and quartz-sericite All buildings, including the shaft tower, were demol-
rocks, containing quartz and sericite with subordinate ished and the ground at the northern side of the open pit
amounts of chlorite. The Udden ore deposit was part of was levelled out. The excavated material and the waste
a greater vein, which is called Kedträskfältet. This vein material from the buildings were deposited into the pit
also includes 3 other ore deposits: Kedträsk, Åsen and (Boliden, 1990) together with waste rock from the
Svansel (Boliden, 1955, 1971, 1980). The principle sul- closely located Kedträsk mine. The waste rock from

Fig. 1. The location of the Udden pit lake in Sweden. Sampling localities marked with A, B and C.
M. Ramstedt et al. / Applied Geochemistry 18 (2003) 97–108 99

Kedträsk had previously been stored in open air and 3.1. In situ measurements
was thus significantly oxidised (T. Göransson, Boliden,
pers. comm.). The average grade of the ore in Kedträsk Conductivity, temperature, concentration of dissolved
is also shown in Table 1. The surrounding rock has a O2 (DO), pH and redox potential were measured in situ
SiO2 content higher than 52 wt.% (Boliden, 1982). The with a probe (Hydrolab MiniSonde) (The MiniSonde
composition of the waste rock from Kedträsk was measures redox as a potential between a Pt electrode
unknown but, considering the composition of the ore, and a AgCl reference electrode immersed in a KCl
the waste rock most likely had a high sulphide content. solution. Conductivity is measured as a current between
The shafts leading down to the underground mine in 2 electrodes held at a fixed potential. Dissolved O2 is
Udden were covered with blocks of concrete prior to the measured as a current resulting from the electrochemical
remediation. Natural sand and tailings were disposed of reduction of O2 at a Au membrane. For measuring pH
in the western part of the open pit. All surrounding the probe has a glass electrode.). The probe was cali-
areas were sown with grass. Due to a high inflow of brated in the laboratory prior to the measurements and
groundwater the pit was filled within 2a (Boliden, 1990). the measurements were performed at 3 locations in the
Today, the pit lake is approximately 390 m long and Udden pit lake, points A, B and C (Fig. 1). In June
150 m across (Fig. 1). The maximum depth of the measurements were done only at point A, in August and
lake was found to be 50 m using a sonic altimeter. September, points A and B and in February, point C.
Because of p the large
ffiffiffi p ffiffiffiffiffiffi relative depth (18%) (relative The DO data were recalculated into O2 saturation (%)
depth ¼ 50zm = A0 where zm is maximum depth and according to the equation given by Weiss (1970).
A0 is surface area. Doyle and Runnells, 1997) the lake is
possibly meromictic. 3.2. Water samples
The only inflows to the lake are rainwater, surface
runoff and groundwater. There is no outflow of surface Water samples were collected using a metal free
water except at the northern side, where the lake over- Ruttner sampler at different depths at two locations in
flows into a small creek during periods when the water the Udden pit lake, points A and B. Point A corre-
surface is elevated. A site-monitoring program collects sponds to the deepest part of the lake. Some duplicate
samples for analysis of the water quality when surface samples were collected at point B to study possible var-
water flows out from the lake. This happened in 1993, iations in metal concentration towards the shore. Sam-
1994, 1995 and 1998. ples were collected at depths from 0.5 to 45 m.
Replicates were taken at a depth of 10 m at point A.
Equipment blanks and travel blanks were collected. The
3. Methodology recovery of Cu, Cd, Pb and As in spiked samples was
between 89 and 96%. Water samples were collected on 3
The water quality at different depths in the Udden pit different occasions during the period June–September
lake was examined by collecting water samples and per- 1998; 22–26 June, 3–10 August and 7–14 September. To
forming in situ measurements with a probe. Measure- estimate natural background concentrations in the area,
ments were carried out on 4 different occasions during samples were also collected in a nearby natural lake,
1998 and 1999. Water samples were collected in June, Lake Ljusträsket.
August and September for analysis of metals and major Water samples were collected in 50 ml acid washed
anions. The in situ measurements were performed on polypropylene tubes. Two samples were taken at each
the same occasions with additional measurements in depth, one filtered immediately through a 0.22 mm
February. The transparency of the water was estimated MILLEX1-GS MILLIPORE filter, to remove the par-
by measuring the Secchi-depth with a Secchi disk and ticulate phase, and one which was not filtered. Both
water binoculars. samples were acidified with concentrated suprapure
HNO3 to obtain pH <1. The tubes were filled to the top
Table 1 to avoid, as much as possible, prolonged exposure to
Average grade of the ore in Udden and Kedträsket air. Samples collected for the determination of S, Cl, N
Element Udden Kedträsket and P were neither filtered nor acidified and the samples
collected for analysis of N and P were collected in tubes
Zn (wt.%) 4.72 2.86 that had not been acid washed. The sample tubes for
Cu (wt.%) 0.41 0.31 collection of S and Cl samples were not completely filled
Pb (wt.%) 0.35 0.22 since they were to be stored frozen until analysis. The
As (wt.%) 0.1 0.22 other samples were stored in the dark at 4  C until ana-
Ag (g/ton) 41 21
lysis. TOC samples were collected in September at the
Au (g/ton) 0.8 0.4
S (wt.%) 25.6 35
depths 3, 10, 30, 45 m and in L. Ljusträsket at 3 m. They
were collected in 15 ml polyethylene tubes, preserved
100 M. Ramstedt et al. / Applied Geochemistry 18 (2003) 97–108

using 200 ml 1.2 M HCl and stored in the dark at 4  C. cooling from September to February. In February, the
None of the samples had any smell and no particulate uppermost part of the water column showed lower tem-
phase or colour could be observed during the sampling, peratures than the rest of the lake.
except for a few white particles below a depth of 30 m in During summer a thermocline and a halocline were
September. developed at 5 m. The uppermost part of the water col-
umn had a higher degree of O2 saturation during sum-
3.3. Chemical analysis mer due to contact with the atmosphere and mixing by
wind and waves. The conductivity increased stepwise
The different samples collected were analysed accord- downwards in the lake from 1 to 2 mS/cm. Between
ing to Table 2. Samples analysed using inductively cou- June and September, a halocline was found at the ther-
pled plasma mass spectrometry (ICP–MS) were diluted mocline and another one was found at a depth of 20 m.
with Millipore water (Super Q plus) and acidified. The In February there was only one halocline at a depth of
concentration of As was corrected for interference with 20 m. According to Levy et al. (1997), the distribution
ArCl. The total concentration of Fe was determined of O2 and other constituents in a lake may be affected
after reduction of Fe3+ in the samples. For reduction, by the circulation patterns. In Udden, dissolved O2 was
0.3 ml of 6 M HCl / 1.2 M NH2OH–HCl was added to found down to 20 m, thus indicating a possible seasonal
15 ml of sample and the solution was kept at 60  C turnover of the top 20 m in the lake. The February
for 45 min and 5 ml of the reduced sample were mixed measurements, showing a rather homogenous top 20 m,
with 5 ml phenanthroline in 40 ml of acetate buffer. The support this assumption.
absorbance was measured at 510 nm using UV/vis pH showed the lowest value, 4.8, at the surface of
spectrometry. the lake. Below the halocline (at 20 m) pH increased to
The August samples were also analysed with ICP–  5.5 and remained rather constant down to 35 m,
AES by SGAB Analytica, Luleå, Sweden for validation where it increased to  6.4. Pit lakes created from mines
and to obtain data for additional elements. (All data are where sulphide ore has been extracted are often acidic.
not shown in this article, for a complete dataset see Two examples of this are the Spenceville Cu pit in
Ramstedt et al., 2002). Nevada USA, with a pH around 2.5 (Levy et al., 1997),
and the Berkeley pit in Montana USA with a pH
around 3 (Davis and Ashenberg, 1989). The reason for
4. Results and discussion the relatively high pH in Udden is thought to be the
rather strong buffering ability of the surrounding rock,
4.1. In situ measurements of physical parameters the shaft tower and the blocks of concrete covering the
underground mine. Another important factor is that
All physical parameters measured in situ, except for comparatively small areas of sulphurous rock are
temperature and conductivity, appeared to be rather exposed at Udden, thus making the continuous produc-
constant during the period June–February (Fig. 2). The tion of acid rather limited (Nils Eriksson, Boliden, pers.
temperature was constant at 5  C below 20 m but comm.). The relatively low pH at the surface of the lake
showed some variation in the upper 20 m of the water is likely caused by oxidative weathering of mainly pyrite.
column. During summer the surface water was heated to From the in situ measurements, 4 distinct layers could
a temperature of 16  C and was cooled during autumn be defined in the pit lake during summer and autumn, as
and winter. In the water column between 5 and 20 m, a result of differences in temperature and salinity (Doyle
warming was observed from June to September and and Runnells, 1997). The layers are indicated in Fig. 3

Table 2
Techniques used for analysis of different elements on different sampling occasions during 1998 in Udden

Analysis Elements Sampling period

ICP–MS As, Cu, Cd, Zn, Pb June and August


UV/vis Fe June, August and September
ICP–AESa Li, Na, K, Ca, Mg, Mn, Al, Si, Ba, Sr, Co, Cr, August
Mo, Ni, S, V, B, Zn, Cu, Cd, As, Pb, Fe
Ion chromatographyb S, Cl June and August
Continuous flow analysis—UV/visc N, P June, August and September
Pt- catalysed high temperature combustionc C (TOC) September
a
The analyses were performed at SGAB, Luleå, Sweden.
b
The analysis was performed at SLU, Umeå, Sweden.
c
The analysis was performed at UMF, Norrbyn, Sweden.
M. Ramstedt et al. / Applied Geochemistry 18 (2003) 97–108 101

and will be referred to as layer I, II, III, IV in the fol- dation of organic material requires an oxidising agent
lowing text. In Udden, layer I reaches from the surface and, since there is no mixing vertically, O2 and other
down to the thermocline situated at a depth of approxi- oxidising agents will be depleted. The redox potential is
mately 5 m. Layer II reaches between  5 m and the thus expected to decrease towards the bottom of the
halocline at  20 m. Layer III is defined as the section of lake in a similar manner as was observed in Udden. The
water between  20 and  35 m. In this layer the O2 redox level in the 4 different layers is thought to be
saturation decreases dramatically to reach a value of 0.3 governed mainly by dissolved O2 in layer I and II and
mg/l at 35 m depth. Layer IV stretches from  35 m and by Fe in layer IV. Which species govern the redox level
downwards. It is anoxic and the redox potential in this in layer III is yet an unresolved question as is the small
layer is lower than in the other 3. In February the but consistent increase in Eh at 20 m. One should bear in
uppermost layering, i.e. the thermocline, had dis- mind that redox potentials measured in natural systems
appeared. The different layers found in the Udden pit may be difficult to interpret due to the fact that natural
lake can be compared with redox layers found in natural waters seldom are in internal equilibrium, and the redox
meromictic lakes reported in the literature, e.g. Hall potential measured should rather be viewed as an indi-
Lake in Washington, USA. Hall Lake is 15 m deep with cation of the overall redox state of the water. Apparent
3 redox layers (oxic, suboxic and anoxic layer) which are redox levels can be calculated based on equilibria
each approximately 5 m deep (Balistrieri et al., 1994). between different redox ‘‘indicator’’ species e.g. dis-
The redox potential (Eh) measured was rather stable solved O2, Fe(II)/Fe(III), HS-/SO2- 4 etc. (Lindberg and
at 500 mV down to a depth of approximately 30 m. In Runnells, 1984). From these apparent levels, saturation
layer IV Eh decreased with depth to a value of  300 mV indices can be calculated for the redox reactions in
close to the bottom of the lake. The redox readings in question. This approach will result in a constant in
June were unstable and are thus excluded. The degra- which all uncertainties are accumulated such as errors in

Fig. 2. Physical parameters measured in situ during the period June 1998–February 1999.
102 M. Ramstedt et al. / Applied Geochemistry 18 (2003) 97–108

redox measurements and non equilibrium status of The total concentration of P was below 5 mg/l on all 3
the water. Such calculations, with respect to Fe2+/ sampling occasions. Doyle and Runnells (1997) stated
FeOOH(s) equilibria, will be reported in a later section that meromictic lakes generally are oligothrophic and
of this paper. have low biological productivity, which also was
observed in Udden. Simultaneously with this study of
4.2. Nutrients and organic carbon the geochemistry of the Udden pit lake, an investigation
of the phytoplankton community in the two lakes was
The total concentration of N increased with depth made (not reported here). The amount of algae was
through the upper 3 layers of the lake, but appeared to much lower ( 1/10) than in L. Ljusträsket and some of
decrease downwards in layer IV (Fig. 4). The con- the plankton species found in Udden are species that are
centration of N in Udden was much higher compared to known to be able to survive in waters with high metal
L. Ljusträsket; the concentration near the surface was content (Sofie Absér, pers. comm.).
twice as high, and at a depth of 30 m almost 5 times The concentration of total organic C was measured in
higher than the surface water in L. Ljusträsket (surface the September samples from each layer of the lake. In
concentrations:  0.5 mg/l in Udden and 0.2 mg/l in L. layer I, at 3 m, the concentration was 0.6 mg/l. At 10 m
Ljusträsket). Although none of the values can be regar- it was slightly less, 0.4 mg/l. Further down in the lake
ded as high for surface waters one should bear in mind the content of organic C increased and at 30 and 45 m it
that the Udden pit lake has mainly been filled by a large was 1.0 and 2.7 mg/l, respectively. A comparison of
inflow of groundwater, which in this area generally these concentrations with the concentration in Lake
exhibits very low N concentrations. A possible source of Ljusträsket shows that the content of organic C in
the high total concentration of N is diffusion of remnant Udden was low, 0.6 mg/l compared with 5.8 mg/l at 3 m.
undetonated explosives from the rock walls and the The low content of particles and total organic C was also
deposited waste rock. The concentration of N decreases reflected in the water transparency. The Secchi depth
below 30 m, which can possibly be explained by deni- was 9 m in June and 12 m in September and August.
trification, i.e. reduction of NO
3 to elemental N2, which The lower transparency of the water in June is thought
in turn can escape the lake in the gaseous phase. to be due to an increased amount of algae at that time.

Fig. 3. Visualisation of the observed layers in the lake. The


layers are numbered from the surface and downwards. Data
from measurements in August 1998 (O2 sat=saturation in %,
cond=conductivity in mS/cm, Eh=redox potential in mV). Fig. 4. Concentration of N at different depths in Udden.
M. Ramstedt et al. / Applied Geochemistry 18 (2003) 97–108 103

4.3. Main elements at that depth and presumably the white particles
observed in September were small particles of gypsum.
For most water samples no significant difference in
concentration could be found comparing the filtered 4.4. Iron, manganese and trace metals
(0.22 mm) and the non filtered samples from the same
depth. Nor could any significant difference be found The concentrations of Fe in mine waters are often
between sampling points A and B. These two observa- high due to the slow oxidation of Fe(II) in acidic water
tions indicate that the water is well mixed horizontally (Nordstrom and Alpers, 1999) and the origin is often
and that the metals present in the water either occur as considered to be weathering of the rock walls (e.g. Davis
soluble species or the colloidal phase passes through the and Eary, 1997; Miller et al., 1996; Levy et al., 1997).
filter. Previous studies by e.g. Kimball et al. (1992) have Bacteria increase the oxidation rate of sulphide miner-
shown that particulate material can pass through the als, especially at low pH (MEND, 1995), and since bac-
filters and, thus, we have chosen to include discussions teria are ubiquitous, the oxidation of rock walls in the
considering possible sorption onto a colloidal particu- Udden pit lake can be considered as micro-organism
late phase although we do not have any data supporting mediated.
this. The concentration profile for Fe in Udden corre-
The profiles for SO2 4 and Cl concentrations sponds well with what could be expected from the redox
strongly resemble those for the conductivity measure- profile, with a rather constant concentration in the
ments (Figs. 2 and 5). The concentration of Cl upper 20 m and a sharp increase in Fe below 35 m
increased downwards in Udden but appeared to be (Fig. 6). It can be expected that Fe(III) dominates under
rather stable within each layer. The origin of this Cl is oxidised conditions, while Fe(II) is stable at greater
probably weathering of Cl-containing silicate minerals. depths with anoxic conditions. However, the high Fe
The largest difference in concentrations was found at the concentrations found in the filtered samples from the
border between layers I and II and between layers II and upper 20 m, 3 mg/l, are much higher than the
III. The concentration of Cl was 5 times as high as in thermodynamic solubility of Fe (oxy)hydroxide in the
L. Ljusträsket at the surface, and 9 times as high in the
bottom water. The SO2 4 concentration profile showed
exactly the same pattern but with a higher concentration
difference between the pit lake and Ljusträsket (Udden
140 – 400 mg/l; L. Ljusträsket 0.7 mg/l). The origin of
SO24 is obviously oxidation of sulphide containing
minerals in the rock walls and waste rock.
The authors attempted to distinguish between SO2 4
and reduced S species in the water samples. However,
samples collected in June and August for analysis of Cl
and SO2 4 , although stored in a freezer until analysis,
showed signs of oxidation in the sense that an Fe oxide
precipitate had formed. This might indicate that the
analysed concentration of SO2 4 possibly represents the
total concentration of S.
The concentrations of Na, K, Mg and Ca increased
downwards in the lake and the depth profile resembles
that of the conductivity. Calcium was the dominating
cation with concentrations of 140 mg/l at a depth of 3 m
and increased levels downwards in the water column to
a concentration of 440 mg/l at 45 m depth. The second
most common cation in the lake was Mg with con-
centrations between 24 and 45 mg/l. The positive
charge, formed by those major cations, is mainly
balanced by SO4 ions. The high concentrations of Ca
and SO2 4 implies that formation of gypsum is possible.
Thermodynamic calculations (using equilibrium con-
stants from Ball and Nordstrom (1991) adjusted with
respect to temperature and ionic strength) result in
saturation indices close to zero in the fourth layer of the Fig. 5. Concentrations of Cl and SO2
4 at different depths in
lake, which indicates that gypsum is likely to be formed the pit lake.
104 M. Ramstedt et al. / Applied Geochemistry 18 (2003) 97–108

current pH-range. This can be caused either by a pro- (oxy)hydroxide and crystalline goethite (constants from
nounced disequilibrium explained by slow oxidation Ball and Nordstrom, 1991) under the assumption that
kinetics or by the presence of colloidal Fe(III) (oxy)- Fe was present in the aqueous phase as Fe(II). SI for
hydroxides which have passed the filter ( <0.22 mm). At amorphous Fe (oxy)hydroxide, calculated for condi-
a depth of 20 m, the aqueous concentration of Fe tions at the depths 40 and 47 m, was 1.2 at both depths.
decreases to below the detection limit and remains low SI for goethite was 1.8 at 40 m and 1.7 at 47 m. These
down to layer IV where the redox conditions are such SI-values indicate a super-saturation at the bottom of
that Fe(II) is the stable form. the pit lake. Similar results have been reported by
The distinct decrease in redox potential below a depth Nordstrom and Alpers (1999) and Holmström et al.
of 32 m is reflected in a sharp increase in dissolved Fe. (2000). There are several possible explanations for
This observation agrees with previous findings by Ball deviations of SI from zero, e.g. presence of colloidal Fe
and Nordstrom (1989). One possible explanation for (oxy)hydroxides and existence of mixed Fe (oxy)hydr-
this observation is reductive dissolution of Fe (oxy)- oxide phases with unknown solubility. Another uncer-
hydroxide. Such a reduction can be the result of oxida- tainty is related to temperature effects. Since no H
tion of dissolved organic matter (Stumm and Morgan, value is available for the reductive dissolution of
1996). The concentration of total organic C at 45 m in Fe(OH)3 no attempt was made to correct this calcula-
September was of the same order of magnitude as the Fe tion with respect to temperature. Finally, uncertainties
concentration i.e. 2.7 mg C/l (0.225 mM), and 27.2 mg in the redox measurements may result in significant
Fe/l (0.49 mM). For each mole of organic C oxidised by deviation of SI from the ideal value of zero. However,
FeOOH, 4 moles of Fe2+ are formed. Thus, the Fe in the calculations indicate that the origin of the high con-
solution could be a result of degradation of natural centrations of Fe in layer IV might be reductive dis-
organic matter. Alkalinity titrations, performed without solution of FeOOH(s) to form Fe(II).
complete absence of O2, gave alkalinity values of 0.01 The observed aqueous concentration of Mn in Udden
mM at 0.5 and 5 m, 0.45 mM at 25 m and 2.3 mM at 45 was between 4 and 9 mg/l and increased downwards
m. Since a possible oxidation of reduced species, such as through the water column in the same way as the con-
Fe(II) and Mn(II), by dissolved O2 would result in centrations of Ca, Mg, K and Na (Fig. 6), i.e., Mn is
release of protons, these values represent a minimum apparently not influenced by changes in redox potential
alkalinity and should not be considered as the actual in Udden, unless the oxidation results in formation of
alkalinity of the lake. However, the alkalinity trend also colloidal sized particles. The Mn present in the lake is
supports the hypothesis of reductive dissolution of most likely in the form of Mn2+ ions, since Mn(III) and
FeOOH. Another possibility is reduction of Fe(III) in
secondary precipitates during oxidation of reduced S
species, such as pyrite, since it is well known that Fe3+
ions are actively involved in sulphide oxidation (e.g.
Moses and Herman, 1991; Herbert, 1999). A third
option might be dissolution of Fe-containing sulphide
minerals without involvement of redox processes.
Saturation indices (SI) have been calculated based on
the aqueous concentrations of Fe in layer IV and the
corresponding pH and Eh values, in order to examine
wheter reductive dissolution (1) of Fe (oxy)hydroxides is
a likely origin of Fe2+ in layer IV.

Fe2þ þ 2H2 O $ FeOOHðsÞ þ 3Hþ þ e ð1Þ

SI is defined according to Eq. (2).

IAP
SI ¼ log ð2Þ
K

where IAP is the ion activity product and K is the equi-


librium constant for formation of the solid phase. SI for
a solution in equilibrium with a particular solid phase is
ideally zero, while super-saturation is reflected by posi-
tive values and under-saturation by negative values. In Fig. 6. Concentration of Mn, Fe and As at different depths
this case, SI was calculated for both amorphous Fe during June, August and September.
M. Ramstedt et al. / Applied Geochemistry 18 (2003) 97–108 105

Mn(IV) (oxy)hydroxides dissolve at lower pH. Besides low through layers III and IV. Comparing these con-
being more thermodynamically stable than Fe(II), the centrations with guideline concentrations for surface
oxidation of Mn2+ is known to be a considerably slower water provided by the Swedish Environmental Protec-
process than Fe2+ oxidation (Stumm and Morgan, 1996). tion Agency (Naturvårdsverket), shows that the Pb
The concentration profile for As in Udden shows concentrations would be classified as low concentrations
similarities to the profile for Fe (Fig. 6), i.e. low con- (Naturvårdsverket, 1999).
centrations down to a depth of 35 m and increasing The Cu concentration increased from  80 to  140
concentrations with depth in layer IV. This observation mg/l at the interface between layers I and II (Fig. 7). The
may be explained by a release of adsorbed As from sur- concentration is rather stable within layer II but
faces of colloidal Fe (oxy)hydroxide or could be a con- decreases dramatically in the upper parts of the suboxic
sequence of reductive dissolution of FeAsO4 and/or layer III and even more in layer IV.
oxidative dissolution of FeAsS by Fe(III) originating The concentrations of Cd and Zn are higher in layer
from secondary Fe minerals in the deposited waste. II compared to layer I (Fig. 7). The concentrations
Arsenic can exist either as As(V) or as As(III), either increase downwards through the second layer, drop
way it is present as oxyanions, HnAsO3n3 or slightly just at top of the suboxic layer III and decrease
HmAsO4m3 (n, m=0,1,2,3). downwards in the same layer and finally, in layer IV, the
Both forms can be adsorbed onto surfaces of Fe concentrations decrease to a level equal to or below the
(oxy)hydroxides, a process which is to some extent pH concentration at the surface. The dip in the concentra-
dependent. Arsenic has been shown to adsorb onto Fe tion curves for Zn and Cd at approximately 20 m
(oxy)hydroxide within a wide pH range (pH 3–10), appears at the same depth as the temporary rise in the
although the adsorption is strongest at lower pH (Lüt- redox profiles. Nickel concentrations are highest in layer
zenkirchen and Lövgren, 1998). The pH of the Udden II ( 450 mg/l) and decrease towards the bottom of the
pit lake (pH 4.5–6.5) is within the range of pronounced lake similarly to Cd, Zn, Pb and Cu (Fig. 7).
As adsorption and, thus, if colloidal Fe (oxy)hydroxides Metals expected to precipitate as sulphides under
are present, the origin of As is likely to be the reductive anoxic conditions are e.g. Pb, Cu, Zn, Cd and Ni. Thus,
dissolution of these. the aqueous concentrations of these metals could be
The concentration of Pb was rather constant around expected to decrease across an oxic–anoxic interface.
1 mg/L through the first two layers and decreased in the Since the solubility of metal sulphides is increasing in
upper part of layer III, where the O2 concentration also the order CuS < PbS CdS <ZnS  NiS < FeS< MnS
decreased (Figs. 2 and 7). The concentration remains (Nordstrom and Alpers, 1999), CuS can be expected to

Fig. 7. Concentration of Cu, Pb, Ni, Zn and Cd at different depths in the pit lake.
106 M. Ramstedt et al. / Applied Geochemistry 18 (2003) 97–108

form before CdS and ZnS. At high concentrations of centrations are correlated with pH. The strong correla-
sulphide, aqueous sulphide and bisulphide complexes tion between conductivity and the different metals
may be formed, which would lead to increased metal mentioned is not surprising since the conductivity is a
solubility (Balistrieri et al., 1994). However, the measure of charged ions and particles.
observed redox potentials in Udden, even at the greatest
depths, are all in a range unfavourable for SO24 reduc- 4.5. Temporal trends
tion. Thus, the concentration of sulphides can be
expected to be low and the precipitation of metals as The Udden pit lake has been sampled by Boliden AB
metal sulphides should therefore be limited and the on several occasions since 1993 as part of a monitoring
formation of bisulphide complexes nonexistent. On the programme. A decrease in metal concentrations from
other hand, the redox potential measured is an average 1993 to 1998 can be noticed (Table 3). This may be the
redox level and will mainly be influenced by the high result of a number of processes. Some of the metal ions
amounts of Fe in solution. Thus, sulphide species might that were leached from the deposited waste material and
have been formed in the bottom water without affecting the rock walls after flooding may have adsorbed onto
the measured redox potential. Furthermore, sulphide settling mineral particles or precipitated as secondary
minerals, such as pyrite (FeS2), may dissolve to a certain minerals. Furthermore, the weathering of the pit walls
extent in contact with water and form free metal ions and waste rock can be expected to be substantially
and sulphide ions, as well as other reduced S anions. reduced. The observed trend is an indication of a suc-
Slow oxidation kinetics would allow this sulphide to cessful decommissioning of the pit mine. However, it
react with heavy metals in the water and form solid should be noted that the exchange of water between the
metal sulphides. It seems likely, from the data obtained, pit and its surrounding soil and bedrock has not been
that the metals in the suboxic and anoxic region are to a quantified. There is no inflow of surface water into the
large extent associated with solids which are removed lake besides the annual precipitation, which in this area
from the water column by settling. However, the is fairly low, 518 mm in 1999 (P. Lindström, Boliden,
authors do not have any data supporting precipitation pers. comm.). The only other sources of water are
of metal sulphides and pH is too low for formation of groundwater and the increased amount of surface runoff
metal oxides. The mechanism behind trace metal during the spring when the snow melts. The rapid filling
removal from the water column in Udden at large of the abandoned pit implies a large flux of ground-
depths is yet to be resolved. water, however, the topography of the area does not
Correlation coefficients between the different elements indicate the presence of a large hydraulic gradient after
and the different physical parameters were calculated. the pit was filled. The extent to which decreased metal
High correlation coefficients were obtained between concentrations can be explained by an outflow of water
conductivity, Ca, K, Na, Mg, S, Si, Mn, and Sr con- through the ground is an open question. However, the
centrations. The redox potential strongly correlates with stability in e.g. temperature, redox potential, SO2
4 and
Cd and Fe concentrations, and Al, Ba and Cu con- Cl concentrations suggests a rather low flux of water.

Table 3
Water quality data for surface water collected in Udden pit lake previously and during 1998

Date Cu (mg/l) Zn (mg/l) Pb (mg/l) Cd (mg/l) Cond (mS/m) SO2-


4 (mg/l) pH (units)

Surfacea 93–11–16 0.17 51.6 62 0.08 160 976 4.7

Surfacea 94–08–03 0.17 81.1 10 0.13 184 1270 5.8

Surfacea 95–05–29 0.04 14.8 18 0.024 66 303 5


Surfacea 95–06–09 0.11 2.58 6 29.5 932 2.7
Surfacea 95–08–16 0.12 67.3 8 0.11 174 1081

Surfacea 98–05–19 0.03 2.15 18 34.5 172 4.3


Surfacea 98–08–12 0.10 35.0 26 98.0 746 5.9
0.5 m 98–06–25 0.076 33 0.66 0.033 93 159 5.14
98–08–06 0.086 34 1.2 0.036 95 148 4.93
98–09–14 107 4.87

Surface=surface water.
Cond=conductivity.
a
=Data collected by the monitoring programme at Boliden (unpublished data).
M. Ramstedt et al. / Applied Geochemistry 18 (2003) 97–108 107

4.6. Comparison with other pit lakes depth of  20 m, below which they decreased. The Cd
and Zn concentration profiles showed similar patterns
In comparison with other pit lakes described in the to Cu but with a less striking increase at  5 m. The
literature, the water quality of the Udden pit lake is decrease of Cu, Cd, Pb and Zn may be explained by
relatively good in terms of pH and heavy metal content. precipitation of the metals as sulphides, although it
However, the concentrations for most metals are much should be noted that the measured redox potential at
higher than background concentrations and guideline this level is too high to support SO4 reduction. The
concentrations given for lake and surface water by dominating anion in the lake is SO2
4 and the concentra-
Naturvårdsverket (1999). The extent to which the higher tion of SO24 increases downwards in the water column
pH can be explained by the acid neutralising effect of but appears to be rather stable within each layer
the concrete, in e.g. the shaft tower, or a high buffering observed. Close to the bottom of the lake the con-
ability of the surrounding rock cannot be deduced from centration of SO24 is at a level corresponding to the
the present data. The content of Fe, Cu, Zn, Cd and solubility of gypsum.
SO24 is much higher than in natural lakes due to dis- This investigation is a description of the prevailing
solution of old weathering products and weathering of geochemical conditions in the Udden pit lake. For an
sulphide minerals. Comparing the concentrations found improved and more complete understanding it would be
in Udden to those in Berkeley pit and in the Spenceville necessary to also include a detailed hydrological study
Cu pit, the metal concentrations in the Udden pit lake to form the basis for mass balance calculations. With
generally appear to be lower. Comparing the different more knowledge of sediment composition, sediment
pit lakes mentioned in Miller et al. (1996), it can be depth and composition of the colloid phase together
observed that pit lakes with higher pH tend to have with a detailed hydrological study, it would be possible
lower aqueous metal concentrations. This can be due to to gain a better understanding of different processes in
several mechanisms: decreased solubility of both pri- the lake.
mary and secondary mineral phases and increased
adsorption to particle surfaces, mainly secondary Fe
(oxy)hydroxides. The oxidation rate of sulphides is also Acknowledgements
higher in acidic waters. Comparison of the results from
Udden with measurements made in Hall Lake (Balis- The authors would like to thank Nils Eriksson, Johan
trieri et al., 1994) shows a similar behaviour of 3 ele- Ljungberg, Staffan Sjöberg, Kristina Axe, Julia Sheals
ments measured in both lakes (Cu, As, Zn). Hall lake is and two anonymous referees for valuable comments on
a natural lake and most metals in this lake precipitate the manuscript and Milan Vnuk for help with the site-
below the anoxic border where sulphide is generated. map. The field study was financed by Boliden Mineral
AB. This work was also supported by funds from the
Foundation for Environmental Strategic Research
5. Conclusions (MISTRA) provided to the senior author (L.L.) through
the research programme MiMi (Mitigation of the
Four distinct layers could be observed in the Udden Environmental Impact of Mining waste).
pit lake from the measurements performed during sum-
mer and autumn 1998. The layers seem to be established
due to temperature gradients and differences in salinity References
in the water column. A thermocline and a markedly
increased salinity were found at a depth of  5 m during Balistrieri, L., Murray, J.W., Paul, B., 1994. The geochemical
summer, separating layer I from layer II. Layers II and cycling of trace elements in a biogenic meromictic lake.
III are separated due to a change in salinity at a depth Geochim. Cosmochim. Acta 58, 3993–4008.
of  20 m. A redoxcline seems to have been formed at a Ball, J.W., Nordstrom, D.K., 1989. Final Revised Analyses of
depth of 35 m as a result of this layering. The strati- Major and Trace Elements from Acid Mine Waters in the
fication of the lake leads to depletion of oxidising agents Leviathan Mine Drainage Basin, California and Nevada—
and the oxidation of organic material may occur differ- October 1981–October 1982. US Geological Survey Water-
ently in the 4 layers. The metal concentrations were Resources Investigations Report 89–4138, Menlo Park Cali-
fornia.
rather stable throughout the entire period, but differ-
Ball, J.W., Nordstrom, D.K., 1991. WATERQ4F—User’s
ences in concentration could be observed at different Manual with Revised Thermodynamic Database and Test
depths. Iron and As showed a pronounced increase Cases for Calculating Speciation of Major, Trace and Redox
below 35 m where Fe(II) is thermodynamically stable. Elements in Natural Waters. US Geological Survey Open-
Lead concentrations decreased downwards in the lake. File Report 90–129.
The concentrations of Cu increased significantly at a Boliden, 1955. Borehole protocols for the Udden mine, hole. #7
depth of  5 m and remained at that level down to a (in Swedish) address: Boliden SE-936 81 Boliden, Sweden.
108 M. Ramstedt et al. / Applied Geochemistry 18 (2003) 97–108

Boliden, 1971. Borehole protocols for the Udden mine, hole. Levy, D.B., Custis, K.H., Casey, W.H., Rock, P.A., 1997. The
#69 (in Swedish) address: Boliden SE-936 81 Boliden, Swe- aqueous geochemistry of the abandoned Spenceville copper
den. pit, Nevada County, California. J. Environ. Qual. 26, 233–
Boliden, 1980. Mine description of the Udden mine (in Swed- 243.
ish) address: Boliden SE-936 81 Boliden, Sweden. Lindberg, S.E., Runnells, D.D., 1984. Ground water redox reac-
Boliden, 1982. Prospect reports for ‘‘Uddenblocket’’ 1982–10– tions: an analysis of equilibrium state applied to Eh meas-
20 (in Swedish) address: Boliden SE-936 81 Boliden, Sweden. urements and geochemical modeling. Science 225, 925–927.
Boliden, 1990. Decommissioning plans for the Udden pit mine Lützenkirchen, J., Lövgren, L., 1998. Experimental study of
(in Swedish) address: Boliden SE-936 81 Boliden, Sweden. arsenite adsorption to goethite. In: Proc. V. M. Goldschmidt
Boliden, 1991. Koncessionsnämnden för miljöskydds beslut Conference, Toulouse, France, pp. 927–928.
1991–06–25 angående efterbehandlingsplan av Udden- MEND, 1989. Subaqueous Disposal of Reactive Mine Wastes:
gruvan. address: Boliden SE-936 81 Boliden, Sweden. An Overview, MEND Report 2.11.1a-a. Library and Doc-
Davis, A., Ashenberg, D., 1989. The aqueous geochemistry of umentation Services Division, CANMET, Ottawa, Ontario,
the Berkeley pit, Butte, Montana, USA. Appl. Geochem. 4, Canada.
23–36. MEND, 1995. Review of in Pit Disposal Practices for the Pre-
Davis, A., Eary, L.E., 1997. Pit lake water quality in the wes- vention of Acid Drainage—Case Studies, MEND Report
tern United States: An analysis of chemogenetic trends. 2.36.1. Library and Documentation Services Division,
Mining Eng., June, 98–102. CANMET, Ottawa, Ontario, Canada.
Doyle, G.A., Runnells, D.D., 1997. Physical limnology of Miller, G.C., Lyons, W.B., Davis, A., 1996. Understanding the
existing mine pit lakes. Mining Eng. December 76–80. water quality of pit lakes, Environ. Sci. Technol. 30, 118–123.
Dzombak, D.A., Morel, F.M.M., 1990. Surface Complexation Moses, C.O., Herman, J.S., 1991. Pyrite oxidation at cir-
Modeling Hydrous Ferric Oxide. John Wiley and Sons. cumneutral pH. Geochim. Cosmochim Acta 55, 471–482.
Herbert, R.B. Jr., 1999. Sulphide Oxidation in Mine Waste Naturvårdsverket, 1999. Naturvårdsverket Bedömningsgrunder
Deposits—A Review with Emphasis on Dysoxic Weathering. för miljökvalitet - Sjöar och vattendrag, Rapport 4913. Nat-
MiMi – Report, 1999:1. Stockholm (ISSN 1403–9478, ISBN urvårdsverkets förlag, Stockholm (ISBN: 91–620–4913–5).
91–89350–06–5). Nordstrom, D.K., Alpers, C.N., 1999. Geochemistry of acid
Holmström, H., Salmon, U.J., Carlsson, E., Paraskev, P., mine waters. In: Lesher, C.M. (Ed.), Reviews in Economic
Öhlander, B., 2000. Geochemical investigation of sulphide- Geology, Vol. 6, The Environmental Geochemistry of
bearing tailings at Kristineberg, northern Sweden, a few Mineral Deposits, Part A: Processes, Techniques, and Health
years after remediation. In: Carlsson, E. (Ed.), Geochemical Issues. Chap. 6, Society of Economic Geologists. Colorado.
Effects of Soil Cover Remediation on Sulphide-rich Tailings Ramstedt, M., Carlsson, E., Lövgren, L., 2002. A Study of the
at the Kristineberg Mine, Northern Sweden, Licentiate thesis, Aqueous Geochemistry in the Udden Pit Lake—Complete
Luleå Univ. of Technology, Sweden (ISSN: 1402–1757). Dataset. MiMi-Report, 2002:1. Stockholm (ISSN 1403-9478,
Kimball, B.A., McKnight, D.M., Wetherbee, G.A., Harnish, ISBN 91-89350-19-7).
R.A., 1992. Mechanisms of iron photoreduction in a metal Stumm, W., Morgan, J.J., 1996. Aquatic Chemistry. John
rich, acidic stream (St Kevin, Gulch, Colorado, USA). Wiley & Sons, New York.
Chem. Geol. 96, 227–239. Tempel, R.N., Shevenell, L.A., Lechler, P., Price, J., 2000.
Klapper, H., Schultze, M., 1997. Sulfur acidic mining lakes in Geochemical modeling approach to predicting arsenic con-
Germany—ways of controlling geogenic acidification. In: centrations in a mine pit lake. Appl. Geochem. 15, 475–492.
Proc. 4th Internat. Conf. Acid Rock Drainage, Vancouver, Weiss, R.F., 1970. The solubility of nitrogen, oxygen and argon
BC, Canada, pp. 1727–1744. in water and seawater. Deep-Sea Res. 17, 721–735.

You might also like