Chemie der Erde 69 (2009) S2 91–107 Geochemistry of natural wetlands in former uranium milling sites (eastern Germany) and implications for uranium retention Angelika Scho¨ner a, , Chicgoua Noubactep b , Georg Bu¨chel c , Martin Sauter b a Martin-Luther-Universita¨t Halle-Wittenberg, Ingenieurgeologie, Von-Seckendorff-Platz 3, 06120 Halle, Germany b Geowissenschaftliches Zentrum der Universita¨t Go¨ttingen, Angewandte Geologie, Goldschmidtstraße 3, 37077 Go¨ttingen, Germany c Friedrich-Schiller-Universita¨t Jena, Angewandte Geologie, Burgweg 11, 07749 Jena, Germany Received 6 July 2007; accepted 12 December 2007 Abstract Discharge from former uranium mining and milling areas is a source of elevated uranium contents in wetlands worldwide. In this work, the efficiency of organic-rich wetland environments for entrapment and accumulation of uranium was assessed using hydrogeochemical field studies of natural small-sized wetlands in Thuringia and Saxony, Germany. The objective was to estimate if artificial wetlands can be used in a similar way: as a sustainable ‘passive’ treatment methodology. Worldwide, a dozen such systems have been implemented for uranium-bearing mine waters as experiments, primarily aiming at uranium reduction and precipitation. Pore water and solid phase samples were collected from the upper decimetres of substrate profiles in minerotrophic ‘volunteer’ wetlands and natural fen-type wetlands that facilitate uranium accumulation. Elemental analyses, correlation techniques and sequential chemical extraction were applied to evaluate retention mechanisms (e.g., reduction, mineral sorption). No process was dominant but the bulk of uranium is retained in moderately labile forms, predominantly as operationally defined organically bound or acid soluble (‘specifically adsorbed’) phases. Macrophyte intracellular uranium accumulation (‘phytoaccumulation’) is not responsible for the high uranium concentrations in the wetland substrates. Although there is no evidence for stable U(IV) mineralisation via ‘reductive precipitation’, high accumulation efficiency of the wetlands results from processes involving species regarded as more labile. According to the findings, the previous concept for treatment wetland that was commonly designed for uranium reduction needs modification. r 2007 Elsevier GmbH. All rights reserved. Keywords: Wetland; Uranium; Trace metal; Uranyl; Treatment wetland; Uranium reduction; Sequential extraction; Hydrogeochemistry; Peat 1. Introduction 1.1. Uranium in the environment Mean uranium concentrations of the upper crust are about 2 mg/kg (e.g., Ballenweg, 2005). Natural uranium consists of radioactive isotopes that are predominantly alpha-emitting. Additionally, uranium as a heavy metal is chemically toxic, therefore posing a health risk when incorporated, especially in aqueous species. In an aquifer, uranium represents a hazard to the environment and human health, at least if present in elevated concentrations. In a provisional guideline, the World Health Organisation proposes a threshold value of 15 mg of uranium per litre for drinking water quality (WHO, ARTICLE IN PRESS www.elsevier.de/chemer 0009-2819/$ - see front matter r 2007 Elsevier GmbH. All rights reserved. doi:10.1016/j.chemer.2007.12.003 Corresponding author. E-mail address: [email protected] (A. Scho¨ner).
17
Embed
Geochemistry of natural wetlands in former uranium milling sites (eastern Germany) and implications for uranium retention
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
ARTICLE IN PRESS
0009-2819/$ - se
doi:10.1016/j.ch
�CorrespondE-mail addr
Chemie der Erde 69 (2009) S2 91–107www.elsevier.de/chemer
Geochemistry of natural wetlands in former uranium milling sites
(eastern Germany) and implications for uranium retention
Angelika Schonera,�, Chicgoua Noubactepb, Georg Buchelc, Martin Sauterb
Discharge from former uranium mining and milling areas is a source of elevated uranium contents in wetlandsworldwide. In this work, the efficiency of organic-rich wetland environments for entrapment and accumulation ofuranium was assessed using hydrogeochemical field studies of natural small-sized wetlands in Thuringia and Saxony,Germany. The objective was to estimate if artificial wetlands can be used in a similar way: as a sustainable ‘passive’treatment methodology. Worldwide, a dozen such systems have been implemented for uranium-bearing mine waters asexperiments, primarily aiming at uranium reduction and precipitation.
Pore water and solid phase samples were collected from the upper decimetres of substrate profiles in minerotrophic‘volunteer’ wetlands and natural fen-type wetlands that facilitate uranium accumulation. Elemental analyses, correlationtechniques and sequential chemical extraction were applied to evaluate retention mechanisms (e.g., reduction, mineralsorption). No process was dominant but the bulk of uranium is retained in moderately labile forms, predominantly asoperationally defined organically bound or acid soluble (‘specifically adsorbed’) phases. Macrophyte intracellular uraniumaccumulation (‘phytoaccumulation’) is not responsible for the high uranium concentrations in the wetland substrates.Although there is no evidence for stable U(IV) mineralisation via ‘reductive precipitation’, high accumulation efficiency ofthe wetlands results from processes involving species regarded as more labile. According to the findings, the previousconcept for treatment wetland that was commonly designed for uranium reduction needs modification.r 2007 Elsevier GmbH. All rights reserved.
consists of radioactive isotopes that are predominantlyalpha-emitting. Additionally, uranium as a heavy metalis chemically toxic, therefore posing a health risk whenincorporated, especially in aqueous species. In anaquifer, uranium represents a hazard to the environmentand human health, at least if present in elevatedconcentrations. In a provisional guideline, the WorldHealth Organisation proposes a threshold value of 15 mgof uranium per litre for drinking water quality (WHO,
ARTICLE IN PRESSA. Schoner et al. / Chemie der Erde 69 (2009) S2, 91–10792
2005). In order to characterise the hazard potential in anatural environment, it is essential to know the mainuranium species present (dissolved, suspended, orimmobilised in the sediment). U(IV) species showgeometrical and electrochemical characteristics similarto thorium, with very low solubility under reducingconditions (Dybek, 1962). Such characteristics arechemically contrary to most heavy metals.
In natural aquatic systems uranium is stable as U(IV)or U(VI), depending on redox conditions, and emergesin dissolved, particulate, organic and inorganic species.Oxidised hexavalent U(VI) species are highly solubleand transported as the uranyl ion ðUO2þ
2 Þ, usuallycomplexed with major anions. Therefore, uranium is anubiquitous element particularly in oxygenated surfacewater. The global average seawater concentration of Uis about 0.3 mg/l (Li, 1982), which is similarly suggestedfor river water (Palmer and Edmond, 1993). InGermany, random groundwater and drinking watersamples revealed U concentrations from 1 and to morethan 100 mg/l (BfFR, 2007; Holler et al., 2005; Merkel,2006) deducible primarily from geogenic input (Baier,2004; Schonwiese, 2007). Such point to a necessity ofexpanded water purification measures not only inregions influenced from former uranium mining suchas parts of Saxony and Thuringia, eastern Germany,with special focus on cost-effective methods.
1.2. Natural wetlands accumulating uranium
Natural wetlands, i.e., habitats frequently watersaturated and highly productive, act notoriously aspotential sinks for hazardous trace elements (Schellet al., 1989; Owen et al., 1992; Cole, 1998). Hence,wetlands enable the enrichment or even elimination oftoxic elements and metals from discharges that are eitheranthropogenically contaminated or geogenically loaded.In some cases wetlands emerge as potentially recover-able deposits with elevated contaminant concentrations(Owen and Otton, 1995).
Since 1945, a number of large-sized wetlands havebeen investigated in the vicinity of uranium-rich sourcerocks in Russia and the western USA. Many wetlandshave exposed high uranium enrichment as compared togeogenic concentrations. As a rule, U concentrationsin wetland substrates are categorised as highly enrichedif ranging from 100 to 1000mg/kg dry weight, andvery highly enriched if exceeding 1000mg/kg (Owenet al., 1992). By comparing inflow (mg/l) and substrate(mg/kg) uranium concentrations, enrichment factorswere estimated for the wetland substrates, ranging from500 up to 2 million (Kochenov et al., 1965; Lopatkina,1967; Idiz et al., 1986; Owen et al., 1992). Even inGermany with its limited area (and therefore less spacefor wetland development), natural wetlands exhibit
elevated uranium concentrations (Landgraf et al.,2002; Seidel, 2002; Schoner, 2006). The occurrencesare linked to disturbed landscapes such as formeruranium mining areas and, in this particular case,mostly the wetlands are recent formations. Regardinglong-term efficacy, the development of surficial uraniumdeposits is not necessarily just a matter of geologicaltime-scales, but also involves the retention capacity.
1.3. Implications of uranium accumulation
As of yet, a limited number of studies has investigatedthe processes of uranium extraction from aquatic solutionin wetland substrates (e.g., Akber et al., 1992; Payne et al.,1998), but with critical gaps in field-based research. On theother hand, for water purification purposes first attemptsto precipitate uranium in constructed wetland-like systemswere started, e.g., in Australia (Shinners, 1996), Slovenia(Veselic et al., 2001), Bulgaria (Groudev et al., 1999), andGermany (Gerth et al., 2000). Worldwide, a dozenuranium treatment wetlands have been proven, but withambiguous efficiency. The mechanisms of uranium en-trapment and accumulation are, as yet, unexplored, as isalso the long-term stability of uranium species occurring inwetland substrates.
Uranium mobility is controlled in large part by thecomposition of the groundwater, which contains naturaland contaminant U(VI) complexing ligands such ascarbonate, phosphate and sulphate ions, and organicchelators (e.g., Dybek, 1962; Langmuir, 1978). Uranium(IV) minerals are known to be barely soluble underreducing conditions. Hence, ‘reductive precipitation’ ofuranium minerals is regarded as a major and sustainablemechanism. This mechanism is claimed as the target ofmany constructed wetland-type systems (Hallett et al.,1997; Groudev et al., 2000; Veselic et al., 2001), inparticular microbial-mediated reduction of uranyl ionsas described by Lovley et al. (1991). An open question iswhether this mechanism is significant in natural wet-lands and, if so, whether such can be applied as a modelfor designing artificial wetlands.
Adsorption by plants or inorganic substrates is alsobelieved to be important (e.g., Duff et al., 1999).Basically, initial fixation may be due to adsorption,rather than due to reductive precipitation (Zielinskiet al., 1987). Regarding uranium fixation on solids, mosteffective sorption of the uranyl cation occurs at slightlyacidic to neutral pH conditions (Langmuir, 1978;Duff and Amrhein, 1996). Fe oxides, such as goethite(a-FeOOH) and ferrihydrite (Fe2O3 �xH2O), are impli-cated as being among the most important inorganicadsorbent phases for U(VI) due to their ubiquity inwaters and sediments (especially in mining areas) andtheir high surface areas (Ames et al., 1983; Hsi andLangmuir, 1985; Bruno et al., 1995; Gabriel et al., 1998).
ARTICLE IN PRESSA. Schoner et al. / Chemie der Erde 69 (2009) S2, 91–107 93
Furthermore, organic carbon (OC) is a major uraniumbinding partner in soils and sediments (Swanson andVine, 1958; Zielinski et al., 1987; Gruau et al., 2000),especially in wetlands. Although uranium sorptionon wetland sediments may be severe, it is normally notan irreversible process. Unless accompanied by co-precipitation or occlusion, re-mobilisation of uranium isexpected (e.g., Duff and Amrhein, 1996).
1.4. Aim and scope of the study
This study aimed at elucidating mechanisms foruranium fixation in natural wetlands by: (1) characteris-ing hydrogeological parameters in natural wetlands ineastern Germany; (2) collecting sediment samples atrelevant sites and characterising them with hydrogeo-chemical methods; and (3) discussing the results oflaboratory investigations and available published data.The results may help to improve current treatmentsystem design and, therefore, the efficacy of artificialwetlands for uranium remediation.
In more detail, a pilot survey of wetlands was aimedat finding relevant sites with evidence for secondaryuranium accumulation derived from surficial tributary.Samples of 22 stratified substrate horizons from threerelevant wetlands sites were investigated by comprehen-sive field and laboratory studies. Here, the elementalcomposition of the aquatic phase and solid phase(screening of substrates and macrophytes) is presented,as well as the statistical evaluation using correlationtechniques. As many competing reactions can occurbetween U(VI), surfaces, and ligands, promoting bothfixation and re-mobilisation reactions, the relationshipsfrom U concentration to other dissolved elements oraquatic parameters were estimated in order to outlinegeochemical processes that control the mobilisation anddissemination of uranium in the wetland environment.The association of uranium with different soil fractionswas characterised by sequential chemical extraction toestimate the strength of uranium binding.
2. Area description
2.1. Wetland locations for pilot survey
In the study area in eastern Germany, a state-ownedcompany, Wismut, was in charge of uranium miningand milling (1946–1991) and, since 1989, of remediation.The mined uranium ore at the Ronneburg mine districtcontained U in concentrations of 200–300mg/kg onaverage (Schuster, 1995). Natural wetlands (Fig. 1) weresurveyed in the vicinity of the former operation sitesRonneburg, Seelingstadt (Federal State of Thuringia),and Crossen (Federal State of Saxony), which arelocated some 60 km south of Leipzig between the cities
of Gera and Zwickau. As affected by mine-relatedactivities, the availability of sufficient water supplyprovides the natural or deliberate establishment ofsmall wetlands (‘volunteer wetlands’). Most of thesewetlands are fed by waste-rock runoff water andsubsequently developed in creek beds, natural streamchannels or at the bottom of heaps or dams and werecolonised by wetland vegetation. Secondarily, naturalfen-type wetlands were also sampled, being anthropo-genically influenced only recently through uraniumprocessing.
A pilot survey of over 20 small-sized wetlandsexamined the hydrogeological setting, the OC concen-trations, and the uranium concentrations (inflow andoutflow, wetland substrates, background sediments).
2.2. Area description for detailed investigations
Three wetlands could be used successfully foradvanced investigations of redox conditions and bond-ing forms. Located in the vicinity of tailing impound-ments (IAA), hereinafter these wetlands are labelledwetland Helmsdorf (WLHe), wetland Culmitzsch(WLCu), and wetland Zinnborn (WLZ), respectively(see Fig. 1). A summary of uranium concentrations,background values, enrichment and exposure conditionsindicates that these three wetlands show uraniumaccumulations resulting most probably from biochem-ical cycling.
A summary of the general wetland characteristics isshown in Table 1, including ranges of OC and Uconcentrations. Two of the three surveyed wetland areas(Helmsdorf and Zinnborn) are situated on Rotliegendsediments (Permian) without uranium content in thebedrock. Here, the only source for uranium input isfrom former milling activity. In contrast, wetlandCulmitzsch developed on phyllitic shale (Ordovician),which was mined with medium to high uraniumconcentrations of some 10–100mg/kg (Wismut GmbH,1999).
In Helmsdorf and Culmitzsch the unconsolidatedwetland substrates consist of inorganic detritus anddegraded plant material to a depth of up to 1m. Thesubstrate layers of Helmsdorf are water saturatedalmost throughout the year, whereas in Culmitzsch thesubstrate is covered by some decimetres of free watertable. Helmsdorf and Culmitzsch are located adjacent toconstructed earth-fill dams. These dams stabiliseU-extraction waste in the tailing impoundments, i.e.,highly contaminated, carboniferous, fine-grained mate-rial. The wetlands may have developed since the onset ofore milling activities, 45 years ago at most. Diffuseseepage from the dams has washed fine-grained sedi-ments into the ponds. In the Helmsdorf area, tailingswith 49–270mg/kg U were deposited, the tailings water
Fig. 1. Map showing the locations of sampled wetlands in Thuringia and Saxony, eastern Germany. Each sampled wetland is shown
by a grey circle. Geographic coordinates of investigation area edges in the format degree, minutes, seconds. NW edge: 50153057.2800N
12104036.3700E; NE edge: 50154015.6000N 12129020.9000E; SW edge: 50142001.4600N 12105004.0800E; SE edge: 50142016.4100N
12129040.4500E.
A. Schoner et al. / Chemie der Erde 69 (2009) S2, 91–10794
containing mean 7mg/l U (Wismut GmbH, 1999). OfIAA Culmitzsch tailings, U concentrations are reportedwith a mean 82mg/kg in the solid and 7mg/l in water.The earth-fill dam adjacent to the wetland was built ofspoil heap material containing up to 17mg/kg U(Wismut GmbH, 1999).
The third selected wetland, Zinnborn, is one out ofmany small ponds in a pre-existing boggy forest (fen).This wetland area contains up to 50 cm of peat with animpermanent water freeboard. Temporarily, substrateoxygenation is expected. The area developed approxi-mately 500m downstream from IAA Helmsdorf andfurther tailing impoundments (IAA Dankritz). Onlyafter deposition of the tailings has uranium-loadedseepage been feeding diffuse springs, which noworiginate in WLZ.
Combining results from inorganic sulphur speciationand sequential extraction (Schoner, 2006) allows for adetailed redox characterisation of the wetland environ-ments according to Berner (1981). For Zinnborn, aprimary oxic environment is deducible mainly fromassociations of monosulphide sulphur to reduced Mnand Fe. Helmsdorf and Culmitzsch exhibit sulphidicenvironmental conditions with elevated pyritic sulphurand Mn.
3. Material and methods
3.1. Sampling and sample preparation
Of the wetland ponds, free water, respectively supernatant
water, as well as aboveground inflow and outflow were
collected in PE vessels. Aliquots were filtered through
0.45 mm membrane filters (cellulose acetate), some stabilised
with concentrated HNO3 (suprapure).
The unconsolidated, waterlogged, mainly organic-rich wet-
land substrates were intersected vertically. Following a
stratification (22 strata, see Table 2) based on visual distinction
of the material (organic matter content), substrate samples
were excavated consecutively and stored under an argon
atmosphere. Pore water was extracted by centrifugation, and
filtered and acidified after pH and conductivity measurements.
Additional wetland substrate samples at distances of 30, 50,
150, and 450 cm downstream from the profile sampling points,
resembling horizontal sections (0–15 cm of depth), allow a
description of the chemical heterogeneity. Terrestrial back-
ground sediments were collected close to the wetlands (0–15 cm
of depth).
All substrate samples were preserved physically in darkness
at �20 1C. For sediment analyses and extractions, aliquots
were defrosted gently, thoroughly homogenised, oven-dried
(60 1C) to weight constancy, and ground with ceramic mortar
A. Schoner et al. / Chemie der Erde 69 (2009) S2, 91–107 97
Plant samples were processed in fuming HNO3 in a closed-
vessel microwave digestion (CEM, Mars 5, Germany) and
analysed by ICP-OES and ICP-MS.
For plant and sediment digestion and extraction, respec-
tively, accuracy was controlled by processing standard
reference material (IAEA-SOIL-7) and sample duplicates or
triplicates. Corrections were made using blank values of the
pure chemicals. All chemicals were analytical grade. Chemical
solutions for digestion, apart from HF, were with additional
distillation.
rsfrom
centrifugationindicatedwith*
mg/l)
Pb(m
g/l)
Cr(m
g/l)
Cd(
0009
o0.0082
0.006
o0.0
0009–0.002
o0.0082
0.004–0.008
o0.0
0014
0.011
o0.0002
o0.0
001
0.002–0.003
o0.005
o0.0
0017
0.003
0.003
0.0
001
0.002–0.004
0.002–0.005
o0.0
NO
3(m
g/l)
NO
2(m
g/l)
PO
4(m
g/l)
HCO
3
(mg/l)
Har
(mm
o1.5
0.042
0.09
572.25
18.3
o1.5
o1.5
o0.05–9.7
–17.7
50.351
o0.05
831.95
15.3
––
o0.05–0.87
–12.2
60
o0.01
o0.05
190.5
2.9
8–67
–o0.05
–2.5
totalhardness;
–¼
insufficientvolumeforan
3.3. Correlation techniques on elemental analyses
The statistical evaluation of pore water data was carried out
by Spearman rank correlation and linear regression assessment
to show associations of U and potential ligands, such as CO3
and OC that may be important at the prevailing circumneutral
pH values. Measured data were normalised to the concentra-
tions of refractory elements, i.e., Si in the case of dissolved
concentrations, to elucidate alterations of pore water chem-
istry. Normalising each single substrate element to refractory
Zr may enable one to highlight influences deduced from
background concentrations.
Table
4.
Elementalconcentrationsin
watersamples(surface
waterfrom
profile
location;pore
wate
Culm
itzsch,andZinnborn
(WLHe,
WLCu,andWLZ)
Fe(m
g/l)
Mn(m
g/l)
U(m
g/l)
As(m
g/l)
Zn(m
g/l)
Cu(m
g/l)
Ni(m
g/l)
Co(
WLHe a
qu
1.07
1.94
2.03
0.049
o0.008
o0.0044
o0.012
o0.
*WLHe1–10aqu
0.068–0.499
3.28–13.3
0.096–2.43
0.017–0.237
o0.008
o0.0044
o0.012
o0.
WLCuaqu
0.158
1.72
6.17
o0.0001
0.050
0.003
0.008
0.
*WLCu1–5aqu
0.098–2.310
4.25–5.60
0.748–4.69
0.061–0.274
o0.047
o0.002
0.015–0.043
o0.
WLZaqu
0.057
0.245
0.426
0.003
0.025
0.004
0.006
0.
*WLZ1–7aqu
0.052–0.188
0.142–0.267
0.133–1.16
0.001–0.006
o0.0025–0.089
0.002–0.003
0.008–0.010
0.
Al(m
g/l)
K(m
g/l)
Ca(m
g/l)
Sr(m
g/l)
Ba(m
g/l)
Si(m
g/l)
Cl(m
g/l)
F(m
g/l)
SO
4(m
g/l)
WLHe a
qu
o0.0048
6.91
422
0.303
0.024
7.3
365
o1
1730
*WLHe1–10aqu
o0.0048
12.3–90.0
404–441
0.308–0.471
0.019–0.208
o5–12
395–523
o1
1465–2000
WLCuaqu
0.020
20.6
245
0.835
0.054
2.6
289
0.94
3192
*WLCu1–5aqu
o0.042
8.72–18.6
229–268
0.349–0.437
0.078–0.094
8.5–16.5
124–162
0.56–0.96
1384–1576
WLZaqu
0.079
20.5
58.1
0.323
0.050
1.7
143
0.5
332
*WLZ1–7aqu
0.045–0.311
20.8–133
53.6–70.1
0.328–0.405
0.056–0.198
o1.5–2.7
132–313
0.3–0.7
318–426
Electricconductivity(cond.)andpH
from
laboratory
measurementonpore
waters
after
centrifugation;hardness¼
schem
atisedto
zero
(values
4100)upto
fourplaces(values
o0.001),apart
from
lower
accuracy
ofmeasurement.
4. Results
4.1. Water
The high ion load of the water samples (Table 4) isshown by a high electric conductivity (mean1.37–3.18mS/cm). Pore water uranium concentrations(Table 1) have median values of 0.7mg/l in Helmsdorf,1.9mg/l in Culmitzsch, and 0.4mg/l in Zinnborn. InCulmitzsch, the surface water on top of the sampledprofile displays the highest uranium concentration of6.2mg/l. From metals that may interfere with uranium(by means of aqueous complexation), Mn shows highconcentrations (p13.3mg/l in Helmsdorf, p5.6mg/l inCulmitzsch), whereas Fe concentrations are low.
Sulphate concentrations are up to 2000mg/l inHelmsdorf, 3192mg/l in Culmitzsch, and 426mg/l inZinnborn (Table 4). From major anion proportions, theHelmsdorf water has been classified as a calcium–sodium–sulphate chemical type with pH ranging from7.0 to 8.0 and a surface water alkalinity of 572mg/lHCO3 (very hard). Culmitzsch provides a sodium–sulphate water with pH ranging from 7.4 to 8.2 and analkalinity in the range of 832mg/l HCO3 (very hard), thelatter deduced from surface water. In Zinnborn, asodium–sulphate–chloride chemical type was deter-mined with pH ranging from 6.6 to 6.9 and a surfacewater alkalinity of 191mg/l HCO3 (hard). Culmitzschprovides an aquatic organic content with mean 16.8mg/lDOC, Zinnborn with mean 19.6mg/l, and Helmsdorfwith mean 34.4mg/l.
A. Schoner et al. / Chemie der Erde 69 (2009) S2, 91–10798
4.2. Plants and sediments
Plants grown on the substrates show U concentra-tions ranging from 20 (Carex) to 400 (Characea)mg/kgroot dry weight (values not displayed). In the shoots, Uconcentrations are much lower, from 5 (Eleocharis) to100 (Characea) mg/kg U on a dry weight basis.
Wetland substrates from temporarily or permanentlyflooded environments consist normally of semiterrestrialto subhydric soils, which are sediments in geologicalterms. In the case of Helmsdorf and Zinnborn,hydromorphic soils with peat mosses developed as histicgleysols to histo-humic gleysols, and subhydric soils(sapropel) in Culmitzsch (see Table 1). Many of theinorganic components are rock fragments derived fromtailing ponds, as well as primary and secondaryminerals, which may originate from authigenic sourcesand particulate transport. X-ray diffraction (XRD)analyses confirmed the presence of abundant quartz,as well as clay minerals (illite, chlorite and/or kaolinite)and feldspar in many strata (Schoner, 2006). InHelmsdorf, pyrite and calcite are also detectable, thelatter near-surface, concordant with high values forTIC. Uranium minerals were not detectable with XRD,nor with X-ray photoelectron spectroscopy (XPS) orscanning electrode microscopy and energy dispersiveX-ray analysis (SEM/EDX) (Schoner, 2006).
Mean Fe concentrations in Helmsdorf are around50000mg/kg, in Culmitzsch 4500 and Zinnborn8000mg/kg (Table 5). The same high concentrationswere analysed for the metals Mn (maximum 18500mg/kg in Helmsdorf) and Al (max. 95000mg in Culmitzsch,and 109000mg/kg in Zinnborn). Uranium in Helmsdorfsubstrate samples has a mean concentration of 389mg/kg, ranging from 4 to 1117mg/kg. Culmitzsch substratesamples have a mean U concentration of 182mg/kg,ranging from 62 to 443mg/kg. Zinnborn substratesamples have a mean U concentration of 2104mg/kg,with range from 152 to 7562mg/kg.
The measured OC concentrations of the wetlands areelevated (mean 12wt%, maximum 26wt%), but clearlybelow histosols typical for wetlands (48–53wt%;Naucke, 1974). The Helmsdorf strata show 1–20wt%of OC, Zinnborn 8–26wt%, and Culmitzsch only1–3wt% (Table 6). Provided that OC is stabilisedcompletely in humic substances, organic matter maybe up to 51wt% (Zinnborn). Helmsdorf displays thehighest concentrations of total S (mean 1.7wt%) and N(mean 1wt%). Compared to Helmsdorf, S concentra-tions are half as much in Culmitzsch, and only one-thirdas much in Zinnborn. The atomic ratio of organic Cto S (atomic weight proportions, compare Friese et al.,1998) is very narrow in Culmitzsch (with low values,ranging from 3 to 10), relatively narrow in Helmsdorf(values ranging from 10 to 26), and particularly wide inZinnborn (high values for C/S, ranging from 73 to 116).
U (mg/kg dry weight) is provided for comparison. In Zinnborn, values for TIC and OC were calculated based on the Scheibler method. (Values in
weight percent; C/N and C/S ratios from atomic weight proportions with OC, TN, and TS values; T ¼ total).aProvided that OC is stabilised completely in humic substances.bProvided that TIC is mineralised completely as calcite minerals.cThe inorganic S fraction in Culmitzsch is in the range 60–85wt% of TS; therefore, C/S is not indicative with respect to soil formation processes.
A. Schoner et al. / Chemie der Erde 69 (2009) S2, 91–107 99
The organic C/N ratios are wide in Zinnborn (25–34)and narrow in Culmitzsch (11–17).
Substrate samples from near-surface horizontal sec-tions, following the flow path of each wetland, showminor textural differences, but relatively high fluctua-tions for elemental compositions (data not shown). Nodistinct concentration trends were found, which mayimply water amelioration. Yet, regarding the sampledsubstrates from vertical and horizontal sections, ele-mental concentrations inside each wetland are in thesame range, resembling the internal distribution andconcentration tendencies. Terrestrial sediments close tothe wetlands Culmitzsch and Zinnborn, unaffected bythe wetland environment, have much lower U concen-trations (Culmitzsch:o20mg/kg; Zinnborn: 62mg/kg).Solid phase U shows a high concentration (360mg/kg)in the background sample from Helmsdorf.
4.3. Correlation
In Helmsdorf water samples, only weak positivecorrelations were found (Table 7) between U and SO4
(rS, normalised ¼ 0.36) and Cl, respectively (rS, normalised ¼
0.43). A positive correlation between U and PO4
(rS, normalised ¼ 0.60) was found in Culmitzsch. InZinnborn, for U there are strong positive correlationswith F (rS, normalised ¼ 0.82), DOC (rS, normalised ¼ 0.77)and, to a lesser extent, SO4 (rS, normalised ¼ 0.61) andDIC (rS, normalised ¼ 0.51), respectively.
Normalised substrate data for Helmsdorf and forZinnborn show comparable Spearman rank correlationarrays (data not shown) with high (rs, normalised40.70) orvery high (rs, normalised40.90) correlation coefficients formost of the elements. At Helmsdorf generally highcorrelation coefficients were calculated for U, thehighest with Fe and Mn (rs, normalised40.98). Uraniumand OC are strongly correlated (rs, normalised ¼ 0.79) onlywith normalised data. From the high correlationcoefficients of non-normalised data (rs40.86), inHelmsdorf strong interrelations are deduced for U withCu and Mg, respectively, and weak correlations for Uwith most heavy metals. Regarding non-normaliseddata, in Zinnborn U is strongly correlated with Ca(rs ¼ 0.93), and Cd as well as P (rs40.79). Whennormalising the substrate elemental concentrations to
ARTICLE IN PRESS
Table 7. Spearman correlation coefficients rS for uranium in relation to selected anions/ligands from pore water analyses in
Helmsdorf (WLHe, n ¼ 10), Culmitzsch (WLCu, n ¼ 5) and Zinnborn (WLZ, n ¼ 7)
U correlation with WLHe: rS WLCu: rS WLZ: rS WLHe: rS, normalised WLCu: rS, normalised WLZ: rS, normalised
Cl 0.13 0.90 �0.07 0.43 �0.60 0.54
F – 1.00 0.02 – �0.30 0.82
SO4 0.03 0.90 0.11 0.36 �0.80 0.61
PO4 �0.09 0.60 – �0.36 0.60 –
DIC 0.67 �0.70 0.23 0.35 �0.89 0.51
DOC 0.70 0.70 0.23 0.20 �0.22 0.77
DIC and DOC from eluates; normalised data to Si; – ¼ measured values below detection limit.
A. Schoner et al. / Chemie der Erde 69 (2009) S2, 91–107100
Zr, for Zinnborn some elements are less stronglycorrelated, e.g., U and Al. There is a strong correlationfor U/OC both with normalised and non-normalisedconcentrations (rs40.82). In contrast, a weak correla-tion for U/TIC is obvious only when regarding the non-normalised data (rs ¼ 0.50). Note that, while non-normalised data reveal no relation of U and Fe, theseelements are completely correlated regarding normaliseddata (rs ¼ 1.00). U is strongly positively correlated withCr and Cd (rs40.87) in wetland Culmitzsch, as well asnegatively with As, Ni, Ba, Ca (rs ¼ �0.80). Only smalldifferences occur between normalised and non-normal-ised correlation coefficients.
4.4. Sequential extraction of uranium
Fig. 2 sums up the percent of U in the differentextraction steps, in vertical profiles displaying thesubstrate U distribution for each stratum. The graphicaldisplay of relative amounts was preferred, since thisstudy focuses on the U distribution in differentextraction steps and not on the total release.
Uranium is distributed predominantly in the verylabile (water soluble, exchangeable) and moderatelylabile fractions (acid soluble and organically bound).The organically bound fraction is dominant in Zinnbornand Helmsdorf (47–64wt% of total U on average). Instrata consisting of moderately degraded peat mosses(approx. 5–20 cm below surface in Helmsdorf and up to60 cm depth in Zinnborn), U was liberated in largeportions in the pyrophosphate extraction step fororganics (up to 77wt% of total U, ranging from 60 to6100mg/kg U). One of the peat horizons contradicts thispartitioning in liberating elevated U portions in theexchangeable fraction, whereas only 27wt% of total Uis organically bound. Only a small percentage ofuranium is stored as ‘residue’ in Helmsdorf and Zinn-born, with the exception of two strata in Helmsdorf(WLHe8, WLHe10), which generally show very lowbulk U concentrations (o5mg/kg, see Table 6).Similarly, in one stratum of Zinnborn 43wt% of U isstored as ‘residue’.
Culmitzsch contains 2wt% of OC on average,compared to 10wt% of U released from the ‘organicallybound’ soil fraction. In the uppermost strata, most ofthe uranium is distributed in very labile phases (31wt%water soluble, 33wt% exchangeable on average).Between 15 and 40 cm depth, the sediments compriseonly low U concentrations (o100mg/kg), distributedin the ‘residue’ fraction (up to 44wt% and increasingwith depth).
5. Discussion
5.1. Environmental conditions pertaining to U
accumulation
The general chemical input to the wetlands isgoverned by various sources and pathways. In termsof uranium retention in the substrates, particulatespecies facilitate a more stable fixation, whereas entrap-ment of aqueous (dissolved or colloidal) species is notnecessarily followed by immobilisation processes.
The investigated wetland pore waters show character-istics of tailing drainage from soda-alkaline uraniumleaching (compare Section 2.2), with very high electricconductivity, circumneutral pH and elevated concentra-tions of Na and SO4. Similarly, the wetland substratesmirror elemental spectra deducible from the tailings, i.e.,ore milling residues and silica-rich overburden. Aqueoustransport with in situ precipitation is taken to be animportant pathway, especially for pollutants that aremobile at circumneutral pH values. Thereby, elevatedpore water concentrations of the anions SO4 (Helmsdorfand Culmitzsch), Cl (Helmsdorf), and NO3 (Zinnborn)carry a risk for undesirable U re-mobilisation anddissemination through aqueous complexation (compareSection 5.2). The same argument applies to elevatedHCO3 concentrations in Helmsdorf and Culmitzsch,and DOC in all three wetlands. The wetlands are rich inDOC (mean 16.8–34.4mg/l), comparable to typicalDOC in bogs (10–50mg/l) and contrast sharply withcommon groundwater DOC (0.5–1.5mg/l) (Sigg andStumm, 1995).
ARTICLE IN PRESS
1015
235
315
15
159
1020
95
157
5
9
19
10
80
6
12
0
1
2
276
5
0% 20% 40% 60% 80% 100%
18
25
40
45
60
68
75
Zinnborn
dep
th [
cm]
2
2
1
18
1
18
5
2
4
17
140
64
00
51
4
2
90
0
22
31
30
31
12
8
120
12
3
2
2
1
1
00
0
2
18
0% 20% 40% 60% 80% 100%
3
7
14
18
28
5
12
16
23
41
Helmsdorf
dep
th [
cm]
14
159
17 4
12
41
48
10
6
1
7
7
3
5
1
2
2
2
2
22
0% 20 % 40% 60% 80% 100%
8
14
22
31
42 11
Culmitzsch
dep
th [
cm] 14
47
267
293
65
water soluble
exchangeable
acid soluble
organically bound
bound to poor crystallineiron oxides
bound to high crystallineiron oxides
residue
36 283
16811087
58 7 75
2528
968
6123
1077646
10 11 31
15108
23
156
72 104
119
20
10
3
159
02
700
10 0
281
471
28415
329
33 4
84
29 38 64
62 126
147 303
148 357
93 3
15 4
86
56
72
15
Fig. 2. Sequential extraction of substrates from wetlands
Helmsdorf, Culmitzsch, and Zinnborn. The diagrams sum up
the percent of uranium in the different extraction phases,
presenting the uranium distribution of the wetland substrate
strata in vertical profiles. The concentrations of the residue
(step 7) are calculated from digestions of aliquots. Data in the
bars give the absolute uranium concentrations in mg/kg dry
weight. The thickness of the bars indicates schematically the
thickness of the sampled strata. Vertical axis not to scale.
A. Schoner et al. / Chemie der Erde 69 (2009) S2, 91–107 101
The refractory Ti in a concentration range of majorelements indicates airborne/waterborne particle input oreven input from underground weathering. Comparable
correlation coefficient arrays for substrate elements ofHelmsdorf and Zinnborn corroborate a compositionresembling a primary solid phase, deducible either fromparticulate input or host rock weathering. Additionally,the rank correlation technique suggests some impor-tance of environmental conditioning for Zinnborn. Inthis wetland, new formation of humic substances isdeduced from the calculated organic C/N and C/S ratios(Nriagu and Soon, 1985; Scheffer and Schachtschabel,2002) and, to a lesser extent, in Helmsdorf. Soilformation makes a minor contribution to the substratesin Culmitzsch. Here, the C/S ratio points to irrelevanceof organic C–S or C–O–S bonds (Nriagu and Soon,1985), whereas the relatively low OC content suggestsmajor minerogenic input of C, N, S, and O (Hakansonand Jansson, 1983). Different elemental relations in eachstratum of wetland Culmitzsch probably indicatealterations in input pathways.
5.2. Uranium distribution and correlation
The wetlands reveal elevated aqueous uranium con-centrations and very highly enriched substrate concen-trations with enrichment factors up to 22,000 (mg/kgcompared to mg/l, compare Section 1.1). Tailingmaterial is the only contamination source in Helmsdorfand Zinnborn, though with various input pathways. Insitu bedrock weathering may contribute via uraniferoushost rock in Culmitzsch. Secondary U enrichment in thewetlands is emphasised by comparing to the localbackground. The terrestrial sediments close to thewetlands Culmitzsch and Zinnborn exhibit less than11%, and 3%, respectively, of the mean U concentra-tions in the wetland substrates. For wetland Helmsdorf,background and substrate samples with analogous Uconcentrations suggest a major importance of air-transported particles from the tailing ponds.
Maximum concentrations of U in the pore waterswere determined at 4.7mg/l, and at 6.2mg/l in thesurface water. Compared to inflow concentrations, themeasured outflow is diminished, but still elevated ascompared to threshold values (compare Section 1.1) orthe runoff value permitted for the Wismut company(0.5mg/l, Kunze et al., 2002). Although U is correlatedwith SO4 and Cl in Helmsdorf, these anions are notrelevant as complexing agents at the pH valuesmeasured. For Culmitzsch in the near-surface strata,formation of uranyl phosphate complexes is taken tooccur. As PO4 concentrations are very low, thesecomplexes are not regarded as important for Utransport. In Zinnborn, carbonate complexes may havecontributed to uranium input as well as humiccomplexes to some extent. Again, the correlation of Uwith SO4 and with F is not indicative for transport incomplexes because pH is not in the appropriate range.
ARTICLE IN PRESSA. Schoner et al. / Chemie der Erde 69 (2009) S2, 91–107102
Apart from a minor interrelation of U and (organic) Cin Zinnborn the statistical evaluation of the data setindicates that one cannot elucidate prevailing processesinfluencing the fate of aqueous uranium in the threewetlands.
The mean U concentrations in the Zinnborn substratesamples (2104mg/kg) are five times higher than Helms-dorf (389mg/kg) and exceed Culmitzsch 6.5 times(182mg/kg). In Zinnborn, the strata rich in OC(WLZ1 to WLZ4, see Table 2) have enormous highuranium enrichment with maximum values of almost1wt%, i.e., p10,000mg/kg U in dry weight. OC mayprimarily, but not exclusively, provoke U accumulationthrough in situ complexation. By comparing normalisedto non-normalised data, the high U/OC correlation forZinnborn suggests importance of biosorption or bioac-cumulation processes, or even formation of humicsubstances. Correlation of U with TIC points possiblyto solid inorganic uranyl carbonate complexes. InHelmsdorf, the high correlation for U and OC (normal-ised data) implies linear relations deducible from thegeogenic input, assuming refractory organic matter insubstrate particles. A possible source is the ore extrac-tion waste derived from U-rich hard coal, which wasmined in the Dresden-Freital area.
The deduced primary relation U/Fe (normalised toZr) in Zinnborn could not be proven by any of thegeochemical investigations, including sequential extrac-tion (see Section 5.3). Regarding U in Culmitzsch, notrends with depth or environmental conditioning areindicated. None of the wetlands reveals statisticalinterrelations between non-normalised values for Uand Fe or Mn as a possible effect of secondary sorptionor co-precipitation processes. Such are in contrast topublished presumptions on the importance of theseprocesses (e.g., Klessa, 2000). Similarly, there is no clearrelation of U/OC for the entire wetland substratesamples. However, there is good evidence suggesting acomplex interrelation between uranium and OC content.Only at higher concentrations are solid OC and Ucorrelated linearly and positively, indicating an involve-ment of OC in the accumulation processes.
Nevertheless, the results do not allow for deducingdirectly the prevailing processes of U retention in thesediments. Such imply that U is not primarily associatedwith ligands, which might build inorganic bonds to U ormeet the same fate in the substrates. From bulk chemicaldata there is no evidence for the dominance of a singleprocess such as sorption to Al hydroxides or co-precipitation with Fe minerals that may lead to thehigh U substrate concentrations. Several processescontribute to uranium fixation.
Most of the measured contaminants in the substrates,notably concerning U, As and different metals (Zn, Cu,Pb, Cd, Mo), show concentration maxima in the upperparts of the investigated profiles. Regarding uranium
vertical allocation, the highest concentrations occur at asubstrate depth of approximately 5–25 cm, below thathorizon proportionally decreasing up to 75–90%. Thevertical U distribution is independent of OC content ormacroscopic consistency. In wetland Helmsdorf theorganic matter content (mean 25wt%) increases to thedepth of stratum WLHe8 (41wt%), declining consider-ably until WLHe10. In wetland Zinnborn, the elementswith major organic binding fractions (C, N, S, P) aremainly accumulated in the uppermost 45 cm (WLZ1 toWLZ4, mean 35wt% organic matter), similar topredominant Sphagnum spp. mosses. These observationslead to the conclusion that between water and wetlandsubstrate a contact zone of intense exchange is limited tohorizons close to the surface, implying a distinct flowpattern. From natural wetlands, uranium accumulationslimited to near-surface sediments have been describedrepeatedly (Pardi, 1987; Tixier and Beckie, 2001) and,similarly, so has the retention of Ni and Co from miningstockpile leachate (Eger and Lapakko, 1988). Thepreferential reaction zone within the peat layer ofZinnborn may result from ponding groundwater indeeper parts that is limited to diffusive exchange,whereas in the upper parts a convective exchange issuggested, causing higher fluctuations in elementalcomposition. Results from several investigations sup-port this suggestion (Schoner, 2006).
5.3. Uranium fractionation and possible speciation
Uranium distribution is inferred from an operation-ally defined sequential extraction procedure. The step-wise sequence (see Table 3) stands for increasingbonding strength of U to the soil, which to some extentmay be attributed to different processes – such assorption or incorporation into Fe oxides or evenreductive U precipitation. Stable U(IV) minerals (i.e.,crystalline) are soluble only in the last extraction step(‘residue’), when digested with HF. Therefore, theamount of available U indicates the sensibility of theecosystem to water erosion processes or to fluctuationsin, e.g., temperature, discharge chemistry, or flow rate.Uranium is partitioned in different compartments of thesubstrates, including minerogenic and biotic reservoirs,the latter comprising macrophytes and micro-organisms.
Wetland Culmitzsch contains little U and OC butcomprises good conditions for sulphate reduction,as deduced from redox conditions and SI calculations(Schoner, 2006). From dissolved uranium enteringthe wetland, U(IV) mineral precipitation is likely.However, according to sequential extraction, onlyinsignificant uranium reduction and mineralisation isindicated. The relative importance of residual uraniumfixation increases in the deeper strata (15–40 cm), with
Table 8. Sequential extraction of soils and sediment with
different procedures: phases where uranium was distributed
predominantly (published data)
Fraction References Extracted
material
Exchangeable Seidel et al. (2002) Wetland
sediments
Organically bound
and exchangeable
Kaplan and Serkiz
(2001), Sowder et al.
(2003)
Wetland
sediments
Organic phase Read et al. (1993),
Braithwaite et al.
(1997)
Wetland
sediments
Organic-sulphidic
phase
Howe et al. (1999);
Beckers (2005)
Sediments;
floodplain
sediments
Organic-sulphidic
phase or crystalline
Fe oxides
Czegka et al. (1998) River
sediments
Fe and Mn oxides Coetzee et al. (2002) Wetland
sediments
Residue and
carbonates
Schultz et al. (1998) Marine
sediments
Residue and bound
to poor crystalline Fe
oxides
Schonbuchner
(2003)
Spoil heap
material
Residue Dhoum and Evans
(1998)
Contaminated
soils
A. Schoner et al. / Chemie der Erde 69 (2009) S2, 91–107 103
simultaneous decrease of more labile U fractions. Suchmay indicate diagenetic redistribution in more stablephases. However, it is more likely that mobilisable U iswashed out from the substrates, especially in deeperstrata underlain with more porous sediments, resultingin the relative enrichment of residual uranium. Alto-gether, in Culmitzsch the high extraction of uranium inthe water soluble fraction has negative implications fortransport, bioavailability, and toxicity. The relativelyhigh amounts of acid soluble U may include, at least,carbonate minerals or other U phases, specificallysorbed (e.g., to clay minerals). Despite the favourableredox environment, the entire observations for wetlandCulmitzsch contradict any effectiveness in long-termuranium fixation.
In the organic-rich strata of Zinnborn and Helmsdorf,U is predominantly associated with the organic soilfraction. When digested with sodium pyrophosphate athigh pH, uranium was mobilised simultaneously.Obviously, uranium is bound relatively weakly andtherefore re-solubilises, e.g., when the mosses aredegraded by micro-organisms. No correlation betweenthe organically bound U and OC was found. In stratacontaining less OC, high U partitioning in exchangeableor acid-soluble fractions implies adsorption or solidphase complexes. As only a few percent of uranium areextractable with water, the chemical processes largelysupplement the physical entrapment. Uranium is mainlyaccumulated due to physical adsorption or specificsorption (complexing reactions) and may be availableto the aquatic environment under changing conditions.Regarding the more sustainable process of uraniumaccumulation in the residual phase, particulate transportor in situ precipitation of U minerals are to beconsidered, as discussed earlier. If at all, reductiveprecipitation of secondary U minerals may haveoccurred to a very small extent, as appropriate redoxconditions have not yet been established in the wetlands(Schoner, 2006).
In comparison to the high amount of organicallybound U in the substrates, sampled plant speciesgrowing directly on the wetland substrates have anaverage U content less than 12wt% of that of thewetlands substrates, suggesting that living plant materialis not responsible for the enrichment of uranium in thesewetlands. The phytoextracted U concentrations are inthe range of typical literature data (e.g., Vandenhoveet al., 2006). Of the measured root concentrations,values from Urtica and Characeae are clearly aboveaverage, yet not signifying uranium hyperaccumulation.Calculated transfer factors from root to shoot rangefrom 0.09 to 0.51 and are relatively high as compared totypical data from controlled down-scaled experiments(Bergmann et al., 2006).
From sequential extraction of uranium in wetlandsoils and similar sediments, both the results of this study
and earlier published data demonstrate that moderatelylabile fractions prevail in most cases (Fig. 2). Wesurveyed results of different sequential extractionprocedures, which were carried out to determinepartitioning of uranium in soils or sediments (Table 8).Kaplan and Serkiz (2001) and Sowder et al. (2003)employed a modified Miller extraction protocol as usedin this study, thereby revealing data for U fractionationsimilar to this study, and as well from wetlandsediments. Coetzee et al. (2002) found most uraniumbound to FeOOH and MnOx complexes, indicating aless weak association. On the other hand, they usedextractants that did not enforce such distinct pH shiftsas the modified Miller protocol. For the Miller protocolSowder et al. (2003) supposed that the application ofpyrophosphate to dissolve organics at pH 10 mayrelease uranium phases aggressively and presumablynon-specific. As such is not disproved so far, one shouldbear in mind that the organic phase may have beenoverestimated. More stable associations were mostlyfound in sediments not comparable to wetland sub-strates, because, for example, river sediments or spoilheap material may contain less OC, but distinctamounts of alluvial deposits including uranium mineralphases. Only for marine sediments (Schultz et al., 1998)are strong reducing conditions evident, demonstrablyresulting in precipitation of U(IV) minerals.
ARTICLE IN PRESSA. Schoner et al. / Chemie der Erde 69 (2009) S2, 91–107104
5.4. Implications for uranium retention in wetlands
High uranium enrichment in natural wetlands sug-gests that constructed wetlands may be successful interms of water purification. The implied flow pattern inthe natural wetlands provides a cost-effective construc-tion scheme, requiring some decimetres of organic-richsubstrate material at most. The role of wetlandsubstrates is connected to organisms (macrophytes andmicro-organisms), offering physical and chemical reme-diation effects.
Macrophytes mainly contribute indirectly to wetlandsystems, via rhizosphere effects (compare Schoner,2006), and directly in providing compost material.Cation exchange on such solid organic material, e.g.,Sphagnum moss as a major wetland plant species andsubstrate component, is very effective in uraniumsorption (e.g., Titayeva, 1967). Organically bound Uwas confirmed via sequential extraction procedures inthis study. However, biosorbed uranium may bereleased if the biomass is degraded.
Field and laboratory studies indicate the minorimportance of macrophytes using phytoextraction pro-cesses (Noller et al., 1994). Phytoremediation conceptsin wetlands may act as a temporary uranium sink andmay increase the transfer to the solid state. In wetlandswith toxic elements the solely favoured contribution isrhizostabilisation, i.e., bioaccumulation by plant roots(Raskin et al., 1994). However, usually up to 99wt% ofthe retained uranium and metal contaminants arelocated in the substrate and are not incorporated tothe plant material (Eger and Lapakko, 1988 andreferences herein; Eger and Wagner, 2003).
Only a few percent of the bulk uranium concentra-tions are retained stably in solid phases, i.e., uraniumminerals. In none of the investigated wetlands andliterature studies could (bio-)chemical reduction ofuranium be confirmed as the principal or even importantretention process, although such has often beenassumed.
In the investigated substrates, uranium is predomi-nantly distributed in labile to moderately labilefractions (adsorption, sorption, complexation) and isre-mobilisable to more than 80% on average. Addition-ally, the varying results of this study indicate thaturanium is bound to multiple sites, for some horizons inapproximately equal quantities, so that any environ-mental change resulting in pH shifts could releaseuranium, as simulated in different chemical extractionsteps.
The fate of such re-mobilisable uranium depends onthe chemical composition and size of the U species andphases. Hence, uranium will be either flushed out of thewetland or, ideally, retained in close vicinity. Thereported high accumulation rates of the wetlandsubstrates support the latter case.
6. Conclusions
The efficiency of constructed wetlands is limited anddependent on both the actual elimination mechanisms andthe wetland size. Only with substantial knowledge of thekey remediation processes may wetlands be optimised. Inthis context, we surveyed over 20 natural wetlands in theformer German uranium milling areas, three in consider-able detail. Uranium is enriched in the substrate strata, asderived from superficial inflow and particulate transport.Highest U concentrations are established close to thesurface and in organic-rich strata. With the statisticalevaluation of the data set, prevailing processes foruranium transport or retention cannot be elucidated.Sequential extraction results indicate a high lability ofuranium. High amounts of organically bound uraniumpoint to biosorption processes. Plant intracellular accu-mulation does not contribute significantly to U accumula-tion. Uranium sorption to secondary Fe minerals in thesubstrate is negligible, as well as (bio-) chemical reduction.The results of the study do not support major uraniumstorage through mineralisation.
With respect to uranium partitioning, release of U isdeducible from external disturbances in wetland systems(e.g., seasonal or hydrological fluctuations, changes inEH or pH values, or varying microbiological activities).Nevertheless, in the examined substrates high U enrich-ment factors imply that several processes contribute totheir effectiveness, finally enabling sustainable storingmechanisms. In the course of sediment aging, theinitially loosely entrapped uranium is transferred tomore stable species.
No expertise exists on long-term effectiveness and re-mobilisation impacts of uranium in wetlands. Conversely,a regulatory approved application of wetland systemsusually depends on stable performance for the proposedperiod. For ‘close-to-nature’ systems such as wetlands,some compromises (including seasonal variability) areinevitable. We encourage discussions on the establishmentof extended periods for demonstrating environmentalcompliance. In order to fulfil remediation targets, reliabledischarge values over particular time periods may be moreappropriate rather than single values.
Acknowledgements
Wetland sampling was kindly authorised by WismutGmbH/Wisutec GmbH. This study was partly sup-ported by the Ministry of Culture, Thuringia. ICPmeasurements were carried out by Dr. D. Merten and I.Kamp, and CNS analyses by C. Luge at the Institute ofGeography, FSU Jena. The authors gratefully acknowl-edge Dr. E. Bozau and an anonymous reviewer forhelpful comments on the manuscript, and Dr. I. Lerchefor improving the English.