ARTICLE IN PRESS

Available at www.sciencedirect.com

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 9 1 – 1 0 0

0043-1354/$ - see frodoi:10.1016/j.watres

�Corresponding auE-mail address:

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

Mercury speciation analyses in HgCl2-contaminated soilsand groundwater—Implications for risk assessment andremediation strategies

A. Bollen�, A. Wenke, H. Biester

Institute of Environmental Geochemistry, University of Heidelberg, Im Neuenheimer Feld 236, 69120 Heidelberg, Germany

a r t i c l e i n f o

Article history:

Received 22 January 2007

Received in revised form

11 July 2007

Accepted 11 July 2007

Available online 18 July 2007

Keywords:

Mercury

Speciation

Contamination

Soil

Groundwater

Remediation

nt matter & 2007 Elsevie.2007.07.011

thor. Tel.: +49 6221 548209a.bollen@ugc.uni-heidelb

a b s t r a c t

Since the 19th century, mercury(II)chloride (HgCl2) has been used on wood impregnation

sites to prevent wooden poles from decay, leaving behind a legacy of highly contaminated

soil/aquifer systems. Little is known about species transformation and mobility of HgCl2 in

contaminated soils and groundwater. At such a site the behaviour of HgCl2 in soils and

groundwater was investigated to assist in risk assessment and remediation. The soil is low

in organic carbon and contains up to 11,000 mg Hg/kg. Mercury (Hg) concentrations in

groundwater decrease from 230 to 0.5mg/l within a distance of 1.3 km. Hg species

transformations in soil and aqueous samples were analysed by means of solid-phase Hg

pyrolysis and CV-AAS. In aqueous samples, Hg species were distinguished between ionic/

reactive Hg and complex-bound Hg. Potential mobility of Hg in soils was studied through

batch experiments. Most Hg in the soil is matrix-bound HgCl2, whereas in the aquifer

secondary formation to Hg0 could be observed. Aqueous Hg speciation in groundwater and

soil solutions shows that an average of 84% of soluble Hg exists as easily reducible,

inorganic Hg species (mostly HgCl2). The proportion of complex-bound Hg increases with

distance due to the transformation of inorganic HgCl2. The frequent occurrence of Hg0 in

the aquifer suggests the formation and degassing of Hg0, which is, in addition to dilution,

an important process, lowering Hg concentrations in the groundwater. High percentage of

mobile Hg (3–26%) and low seepage fluxes will result in continuous Hg release over

centuries requiring long-term groundwater remediation. Results of soluble Hg speciation

suggest that filtering materials should be adapted to ionic Hg species, e.g. specific resins or

amalgamating metal alloys.

& 2007 Elsevier Ltd. All rights reserved.

1. Introduction

The antiseptic effect of mercury(II)chloride (HgCl2) has been

known since the 19th century and was widely applied by the

wood preservation industry. The process of kyanizing, named

after John Kyan who patented this process in England 1832,

consists of steeping wood in a 0.66% HgCl2 preservative

solution to prevent the wood from decay (Schondorf et al.,

r Ltd. All rights reserved.

; fax: +49 6221 545228.erg.de (A. Bollen).

1999). On these wood impregnation sites, improper storage of

treated wood or leakage of dip basins often led to a severe

contamination of the environment, especially of soils and

groundwater. As HgCl2 is toxic and highly soluble and can be

easily transformed (e.g. reduced to Hg0) it possess a large risk

to the environment. The behaviour of HgCl2 in soils and

groundwater, both low in organic matter, has rarely been

investigated (Biester, 1994; Schondorf et al., 1999) and will be

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 9 1 – 1 0 092

focus of this study. Once deposited into the soil, mercury (Hg)

is subject to a wide array of chemical and biological

transformation processes such as Hg0 oxidation, and Hg2+

reduction or methylation depending on soil pH, temperature,

and soil humic content (Schuster, 1991; Stein et al., 1996). The

formation of organic Hg2+ complexes is known to be the

dominating process, which is largely due to the affinity of

Hg2+ and its inorganic compounds to sulphur (S)-containing

functional groups (Schuster, 1991). In soils low in organic

matter, most Hg can be found as reactive, ionic Hg species e.g.

HgCl2 or Hg(OH)2 which can be transformed easily into more

toxic forms such as methylmercury or Hg0 (Skyllberg et al.,

2006).

Therefore, the speciation and mobility of Hg in soils is

essential in determining the potential environmental risk of

contaminated sites. Hg-binding forms and transport beha-

viour in soils influences the release into other environmental

compartments, e.g. groundwater or atmosphere, and gives

important information for human and ecological health

concerns (Boening, 2000). Accurate assessment of the species

and mobility of Hg in soils and groundwater is also necessary

to determine the need for, and type of, remediation actions

that are required. For example, the type of filter material for

groundwater remediation depends on the Hg species in the

groundwater; ionic Hg can be retained through amalgamating

filters (Biester et al., 2000; Huttenloch et al., 2003), whereas

organic-bound Hg demands other filter materials such as

activated carbon (Krishnan et al., 1994).

Fig. 1 – (A) Groundwater Hg contamination plume and location

impregnation site. (B) Hg concentration (lg/l) and Hg species dis

(Source: Harres Pickel Consult, 2006, modified).

1.1. Site description

Wooden telegraph poles, trellis support poles for vineyards,

and sleepers for railway tracks were treated with HgCl2solution against fungal attack at a former wood impregnation

plant in Southern Germany. The site covered an area of

almost 9 ha and during the 60 years of operation (1904–1965)

an estimated amount of 10–20 tons of Hg has been released

into the local soils and the aquifer (Schondorf et al., 1995).

Currently a residential area of 8 ha, the site is contaminated

with up to 11,000 mg/kg Hg and a groundwater plume with a

maximum Hg concentration of 230mg/l and a width of 100 m

persists 1.3 km down gradient (Fig. 1A). Under a 1–3 m thick

artificial filling consisting of redeposited loess/loess loam and

building rubble, homogenous loess is encountered down to

5–6 m depth representing the top layer of the natural soil

profile. These slightly clayey, calcic silt sediments possess a

comparatively low content of organic carbon (0.8%) (Table 1).

The underlying fluvial loose gravel deposits, which form the

upper aquifer, contain even less organic material (usually

below 0.2%). The unconfined groundwater table is encoun-

tered at a depth of 6–11 m below surface level and possesses a

very low content of dissolved organic carbon (DOC) (median

0.57 mg/l). The aquifer consists of highly permeable sand and

gravels and has a groundwater gradient of 0.7–1%. This results

in a hydraulic conductivity of 3�10�3 m/s and a high flow

velocity of 3–10 m/d. The base of the aquifer consists of

weathered gravels of low permeability, which form the

of wells (m) and bores (K) in the vicinity of the former wood

tribution in groundwater samples from groundwater wells

ARTICLE IN PRESS

Table 1 – Medians (7 standard deviation) of physicochemical parameters in the soil and groundwater of the study site

pH (�) Corg (%) Eh (mV) O2 (mg/l) DOC (mg/l) Cl- (mg/l)

Groundwater 6.6270.45 454755 6.0171.06 0.5770.04 2071

(n ¼ 103) (n ¼ 97) (n ¼ 107) (n ¼ 4) (n ¼ 8)

Soil horizons

Artificial filling 7.9070.06 2.5072.20

(n ¼ 4) (n ¼ 4)

Loess cover 7.8970.10 0.8570.47

(n ¼ 6) (n ¼ 8)

Unsaturated zone 7.5570.20 0.2170.14

(n ¼ 6) (n ¼ 10)

Saturated zone 7.2770.38 0.0770.05

(n ¼ 6) (n ¼ 7)

Aquitard 7.3270.26 0.1670.09

(n ¼ 6) (n ¼ 8)

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 9 1 – 1 0 0 93

aquitard (Schondorf et al., 1999). Further physicochemical

parameters of the soil, aquifer and groundwater are shown in

Table 1. Seepage modelling for this site estimated that due to

high Hg concentrations in soil and low seepage fluxes there

will be a continuous Hg release from the unsaturated soil

zone into the aquifer for at least the next few hundred years

(Harres Pickel Consult, 2006, unpublished report).

The aim of this study was to investigate the fate of HgCl2 in

a soil–groundwater system poor in organic matter. Main focus

lies on the exchange of Hg between soil, soil solution and

groundwater depending on the Hg species. Therefore, the Hg

distribution in soils was investigated to localize entry points

into the groundwater and hot spots of Hg contamination. Hg

speciation analysis in solid and aqueous phases was con-

ducted to highlight potential mobile and reactive Hg and Hg

transformation processes. To what extent Hg speciation

analysis reveals important information regarding the risk

assessment and remediation technologies of a contaminated

site will be answered based on the results.

2. Material and methods

2.1. Soil sampling

Soil samples were taken by 10 bores located on the former site

and downgradient areas (�1.2 km) (Fig. 1A). Results shown

here are from 6 selected bores (E3, B28, B27, E4, B25, and E5).

All bores were drilled down to the bottom of the aquifer

(�11 m below surface level) and soil samples were taken in

0.5–1 m steps. The samples were collected in brown glass

bottles and stored at 5 1C until analysed.

2.2. Groundwater sampling

During the past 10 years, 28 groundwater wells were installed

which fully penetrate the shallow groundwater down to the

base of the aquifer. For this study a groundwater monitoring

covering a period of 1 year at monthly intervals was

performed in which more than 150 groundwater samples

were taken. Results of 7 wells (B3, B28, B10, P19, P20, B14, and

B24) forming a transect downgradient from the contamina-

tion hot spot to the end of the contamination plume will be

shown here (Fig. 1A). The groundwater was sampled after

purging the wells using a submersible pump and a Teflon

hose. The samples were stored in brown glass bottles at 5 1C

until analyses. For the analysis of total Hg the samples were

acidulated with HNO3 and K2Cr2O7, whereas the samples for

Hg speciation were not pre-treated.

2.3. Total Hg in soil

Total Hg content was determined in 146 soil samples by cold

vapour atomic absorption spectroscopy (CVAAS) after diges-

tion of samples in aqua regia (HCl:HNO3; 3:1) at 160 1C for

3 h (DIN EN 13346, standard German method). Concentrations

are expressed as mg/kg dry weight. Results were validated

by analysing standard reference material (Montana Soil

NIST 2711).

2.4. Hg solid-phase speciation analysis

Hg species in soil and aquifer material were determined by

means of solid-phase thermo desorption. This method is

based on thermal decomposition or desorption of Hg com-

pounds from solids at different temperatures and continuous

determination of released Hg by atomic absorption spectro-

metry (AAS). The solid soil samples are heated in a pyrolysis

detection unit of an AAS (Perkin Elmer AAS 3030). A platinum

coil heated to 800 1C is located at the furnace outlet to

decompose all released Hg species to Hg(0) for measurement

by AAS. Measurements were carried out at a heating rate of

0.5 1C/s and a N2 gas flow of 300 ml/min. A detailed descrip-

tion of the apparatus can be found in Biester and Scholz

(1997). The results are depicted as Hg thermo-desorption

curves, which show the release of Hg vs. temperature.

Depending on the total Hg content, 2–250 mg of sample

material was used for the measurements. Results were

compared with thermo-desorption curves of standard Hg

materials produced by previous studies (Biester, 1994; Biester

and Nehrke, 1997).

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 9 1 – 1 0 094

2.5. Operational defined total Hg and Hg speciation inaqueous phase

Hg species in soil solution and groundwater were operational

defined as (modified after Brosset, 1987; Meili et al., 1991)

�

Hgtot: total soluble Hg

�

Hgaq0 : elemental Hg

�

HgIIa: inorganic, reactive Hg such as Hg2Cl2, HgCl2 or HgO

�

HgIIb: Hg bound to humic compounds

�

Hgpart: Hg bound to particles.

Total soluble Hg (Hgtot) was determined by EPA method

1631. After oxidation of organic Hg fractions with BrCl (0.2 N)

to inorganic Hg2+, the Hg was measured after subsequent

reduction with SnCl2 by means of cold vapour atomic

absorption spectroscopy (CVAAS). Whereas Hg0 was mea-

sured without reduction, HgIIa was determined by the

addition of SnCl2 as a reducing agent. The operational defined

fractions were determined following the scheme shown in

Schondorf et al. (1999).

2.6. Hg in soil solution

Mobility and transport behaviour of Hg in soil was determined

through batch experiments. Batch experiments were per-

formed on 146 soil samples. Five grams of soil sample were

shaken with 50 ml deionized water (1:10 ratio) for 24 h

(according to German DIN 38414 S4). Samples were then

centrifuged and the leachates were analysed for Hg specia-

tion (Hgtot, HgIIa, HgIIb and Hgpart). Hg0 was not measured as

shaking presumably vaporized most of the Hg0 after opening

the sampling bottles. Therefore, the HgIIa fraction may also

contain unknown amounts of Hg0.

2.7. Major soil characteristics

The total carbon content in soil was determined by infra-red

(IR) detection of CO2 after combustion of the homogenized

dried and ground sample (0.5 g) in a high-frequency induction

furnace. Inorganic carbon was calculated as total carbonate in

the samples determined by a ‘carbonate bomb’ (Muller and

Gastner, 1971) and, after subtracted from total carbon, leaves

the organic carbon (Corg) fraction.

Soil pH was measured by mixing 50 ml of 0.01 M CaCl2solution with 20 g of fresh sample material. Soil pH, after

equilibrating for 1 h, was determined using a glass electrode.

3. Results and discussion

3.1. Hg distribution in soil and aquifer

Results of total Hg concentrations in the contaminated soil

and aquifer are similar to and within the same range as

measured in previous studies (Biester and Scholz, 1997;

Schondorf et al., 1999).

Hg concentrations ranged between 3 and 11,000 mg/kg

(median: 7.5 mg/kg). Highest concentrations were found in

the area of the former treatment hall where HgCl2 was spilled

directly into the ground (E3) and in areas where wooden poles

were stored after treatment (B28, B27) (Fig. 2A). Bore E3 shows

Hg concentrations of 11,048 mg/kg at a depth of 2 m, which

gradually decreases to 23 mg/kg with increasing depth. In the

aquifer material of E3 only 2–3 mg/kg Hg were found. In bore

B28 which is located �200 m downgradient of E3 highest Hg

concentrations (353 mg/kg) were found in deeper layers

(6–8 m). The aquifer material of this bore shows Hg concen-

trations between 18 and 162 mg/kg. Within the distance of

B28–B27 (�140 m) Hg concentrations in the aquifer and the

capillary fringe decrease down to 2–8 mg/kg. This value stays

more or less constant (median 6 mg/kg) in the bores further

downgradient. Hg concentrations in the aquifer exceeding the

German threshold value of 2 mg/kg (UVM-BW, 1998) could still

be found at E5 (4–6 mg/kg), �830 m downgradient of the point

source at E3. At B24 Hg concentrations decrease to local

background values of 0.4 mg/kg. The several metres thick

loess layer between the topsoil and the aquifer shows, in

most cases, no Hg contamination. Thus, the majority of the

area does not interchange directly between the contaminated

topsoil layers and the aquifer. At E3 contamination of all

layers could be detected. The loess horizon is low in organic

carbon and the adsorption sites are limited to mineral

surfaces of e.g. clay minerals or sesquioxides. Due to the

high Hg concentrations, adsorption sites are assumed to be

saturated resulting in high amounts of mobile HgCl2 in the

soil. Hg can easily infiltrate through the loess layers down into

the aquifer.

The Hg distribution in the soil and aquifer clearly indicates

that there is a point source of mobile Hg rather than being

evenly distributed across the site. In particular, Hg enters the

aquifer only from the hot-spot area around E3, where further

transport within the aquifer takes place in groundwater flow

direction, forming a contamination plume in the aquifer with

a total length of 1.3 km.

3.2. Solid-phase Hg speciation

The behaviour of HgCl2 in soil is poorly known. Nevertheless,

it has been previously shown that Hg has a strong tendency to

form complexes with Cl�, OH�, S2� and S-containing func-

tional groups of organic ligands. Under standard temperature

and pressure conditions that occur in the soil and aquifer

environment, Hg should be present in three oxidation states.

The most reduced species is Hg0; the other two forms are

ionic Hg22+ and under oxidizing conditions Hg2+. Hg2

2+ is not

stable under environmental conditions since it dissociates

into Hg0 and Hg2+ (Schuster, 1991). It has been shown that Cl�

concentrations are an important factor for adsorption of Hg2+

in soils (Hahne and Kroontje, 1973; Wang et al., 1991).

Chloride forms hydroxide complexes with Hg2+ at Cl�

concentrations above 10�9 mol/l; HgCl2 forms above

10�7.5 mol/l (Hahne and Kroontje, 1973). Thus, with increasing

Cl� concentrations, the mobility of Hg also increases.

Schuster (1991) stated that chloride may be regarded as one

of the most mobile and persistent complexing agents for Hg.

Thus, HgCl2 spilled into soils might persist in its original form

or, more likely, will be bound as a Hg-chloride complex or

other reactive Hg2+ species to the soil matrix. The stability of

HgCl2 mainly depends on the pH and redox conditions in the

ARTICLE IN PRESS

Fig. 2 – (A) Hg concentrations (mg/kg) in soil of the research area. (B) Percent soluble Hg (%) in soil and percentage of inorganic,

reactive Hg species (%) (italics) in leachates of the batch experiments. The right part is situated below the former industrial

site; the left part adjoins this area. Figure not to scale.

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 9 1 – 1 0 0 95

soil/aquifer system (Fig. 4). HgCl2 can also be easily reduced

by other metals or organic matter in the soil despite its

stability under oxidizing conditions. However, in the presence

of organic matter in the soil, Hg seems to bind predominantly

to the organic fractions as Hg exhibits a great affinity for

S-containing functional groups, which are frequently found

in organic substances (Mierle and Ingram, 1991; Haitzer et al.,

2003). Benoit et al. (2001) reported formation constants (KOC)

ranging between 1010 and 1012 for Hg2+ complexed with

organic matter. Both Skyllberg et al. (2000) and Haitzer et al.

(2003) reported even higher KOC-values of 1023–1024. According

to these high adsorption/complexation coefficients all Hg

should be bound to organic matter in the soil. The soil in the

present study site contains a very low amounts of organic

carbon (0.07–2.5%; median: 0.2%) which is between 0.01 and

0.37 mmol/kg C. Agricultural areas have a C:S ratio of 130

(Scheffer and Schachtschabel, 2002), thus only 8�10�5–3�

10�3 mmol/kg S would be available for sorption sites. As very

high Hg concentrations in the soil up to 11,000 mg/kg

( ¼ 55 mmol/kg) can be found in some areas, only a small

proportion of the Hg could be humic bound.

Results of solid-phase thermal desorption indicate matrix-

bound Hg is the predominant Hg species in the soil (Fig. 3A).

Most of the samples showed a single peak that occurs at a

temperature range between 150 and 250 1C, suggesting that

matrix bonding predominates. Sample desorption curves do

not match Hg release curves of standard materials such as

HgCl2, Hg0, HgS, or any other specific Hg compound (Fig. 3A).

Due to the comparatively low amount of organic carbon

(0.07%), it is assumed that the HgCl2 in this soils is mostly

ARTICLE IN PRESS

Temperature [°C]

0 100 200 300 400 500 600

Temperature [°C]

0 100 200 300 400 500 600

Extinction

0

200

400

600

800

1000

1200

1400

soil sample(s) Hg(0) HgCl2

Hg bound to humic acids HgS

Extinction

0

200

400

600

800

1000

1200

Temperature [°C]

0 100 200 300 400 500 600

Extinction

0

100

200

300

400

500

Temperature [°C]

0 100 200 300 400 500 600

Extinction

0

200

400

600

800

1000

1200

1400

Fig. 3 – Thermo release curves (black) of Hg species in soil. The grey lines indicate release curves of standard Hg compounds.

(A) Hg sorbed to soil matrix (main species) (soil and aquifer samples of various bores and depths); (B) HgCl2 (first peak) and Hg

sorbed to soil matrix (second peak) (soil sample of E3, depth: 6 m); (C) Hg0 (first peak) and Hg bound to soil matrix (second

peak) (aquifer sample of B28, depth:11 m); (D) Hg bound to soil matrix (first peak) and HgS (second peak) (aquifer sample of

B25, depth: 11 m).

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 9 1 – 1 0 096

associated with mineral soil compounds e.g. iron/aluminium

oxides and hydroxides or clay minerals. With solid-phase

thermal desorption, however, it is impossible to distinguish

between those different matrix-bound Hg species and de-

mands a grouping of these species.

Free HgCl2 could only be detected in highly contaminated

soil samples from the upper layers of bores E3 and B28

(Fig. 3B). This confirms the assumption that soluble HgCl2 was

the original species and that it has been transformed, during

transport, into weakly bound Hg species sorbed to the soil

matrix.

In aquifer soil samples the secondary formation of Hg0

could be observed (Fig. 3C). In some samples release of Hg0

during pyrolysis was restrained due to diffusion processes

within the sample causing in a delayed peak signal. Due to its

relatively high standard potential (E0 ¼ 0.65 V) (Bisogni, 1989)

Hg2+ can be easily reduced to Hg0. It can be reduced abiotically

by either humic substances (Allard and Arsenie, 1991) or

microbial processes (Barkay et al., 1992), or by other metals

such as iron in soil (Lovely, 1995). Microbial reduction is

probably unlikely in our soils because of the toxicity and high

concentrations of Hg. Due to the very low amounts of organic

matter throughout the whole soil profile, it is most likely that

Hg2+ is reduced by other metals in the aquifer, which have a

lower standard potential than Hg. Soil gas measurements

proved that reduction to Hg0 takes place, with concentrations

ranging between 100 and 1000 ng/m3 Hg0; especially in highly

contaminated areas in the upper soil layers and in the

unsaturated soil zone (Schondorf et al., 1999).

The formation of S-bound Hg was also found in some

aquifer samples (Fig. 3D), but is restricted to local areas where

it is most likely that peat lenses causing anoxic conditions

ARTICLE IN PRESS

Fig. 4 – Inorganic Hg species as a function of pH and redox

potential (from Schondorf et al., 1999). The cross-hatched

area indicates the Hg stability field of the study site.

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 9 1 – 1 0 0 97

prevail (Revis et al., 1989; Barnett et al., 1995). As the Hg

release curve does not fit standard HgS, Hg precipitates in this

case rather as organo-sulphides or meta-cinnabar or as a

mixture of all three species.

3.3. Hg concentrations in groundwater

Results of Hg concentrations in groundwater of 7 wells are

presented in Fig. 1B. All wells exceeded background values for

Hg in groundwater o0.1 mg/l for this area. At B3, which is the

closest well to bore E3 (100 m distance), Hg concentrations of

54mg/l were found. B28 shows Hg concentrations of 81mg/l.

The Hg concentrations increases downgradient to 161mg/l

(B10) and are highest at P19 with 229mg/l. P19 is located 340 m

downgradient of the Hg entry point at E3. Hg concentrations

of 121mg/l (P20) could still be detected in 740 m distance from

E3. Further on Hg concentrations in groundwater decrease

over a relatively short distance of 130 m to 2.5mg/l (B14). Hg

concentrations above the detection limit (0.1 mg/l) could still

be found at 1.3 km downgradient of E3 at bore B24 (0.5mg/l).

The decrease in Hg concentration is most likely due to

dilution processes by incoming non-contaminated ground-

water at the edges of the plume as well as due to the

formation and degassing of Hg0 (see the later section). The

location of highest Hg concentrations differs between

groundwater and soil. Several circumstances result in this

discrepancy: Firstly, the Hg is not transferred vertically into

the aquifer, but rather slopes in the groundwater flow

direction. Furthermore, E3 is most likely not the only Hg

entry point for high amounts of Hg and there are assumingly

still other undiscovered Hg hot spots in the area resulting in

high Hg concentrations in the groundwater. Groundwater

monitoring over the last 10 years shows that the extent and

direction of the Hg plume in groundwater did not change

significantly. This is most likely due to dilution processes at

the edges of the plume as well as due to degassing of Hg0,

which results in a decrease of Hg concentrations as reported

in the following section.

3.4. Hg speciation in groundwater

Once, after entering the groundwater inorganic, reactive Hg

species can undergo various adsorption and transformation

processes. However, only sparse data on groundwater con-

tamination with Hg and Hg transformation processes in

aquifer systems are presently available.

Many studies revealed that in the presence of dissolved

organic matter (DOM)—such as humic or fulvic acids—Hg

seems to bind predominantly to the organic fraction (Mierle

and Ingram, 1991; Haitzer et al., 2003). The groundwater of

this study site contains very low amounts of DOC (0.5 mg/l)

(Table 1). Therefore it is more likely that the Hg predominates

as inorganic Hg species such as HgCl2, HgOHCl, Hg2Cl2, or Hg0.

The occurrence and stability of theses species mainly

depends on Eh and pH conditions in the groundwater. The

pE–pH diagram shown in Fig. 4 was calculated based on

chemical groundwater data (does not include DOC) derived

from various measurements of the study site’s aquifer of an

earlier study by Schondorf et al. (1999). In those groundwater

samples, the pH ranged between 6.17 and 7.07 and Eh

between 399 and 509 mV (Table 1). Therefore, the inorganic

Hg species Hg0 and HgCl2 should theoretically be the

dominant inorganic Hg species in the groundwater. Indeed,

equilibrium calculations at a temperature of 11 1C

( ¼ groundwater temperature) found that Hg0 should predo-

minate (86%), whereas HgCl2 (7%) and Hg(OH)2 (4%) were of

minor importance (Schondorf et al., 1999).

However, our results for Hg speciation in the groundwater

(Fig. 1B) revealed that on average most of the Hg (84%) exists

as reactive, inorganic Hg species such as HgCl2. The occur-

rence of Hg(OH)2 is unlikely due to the high Cl� concentra-

tions in the groundwater (20 mg/l) (Table 1). Only 4% was

found to be Hg0. Ancillary species were 7% organic-bound Hg

and 5% Hg bound to particles. The high amounts of Hg0

calculated by Schondorf et al. (1999) could not be affirmed by

the present investigation. Possible explanations could be

deviation of natural systems to thermodynamic equilibrium

calculations, inaccuracies in measuring Eh and pH, or Hg0

losses during the sampling and analysis of the samples. More

presumably, Hg0 could have degassed out of the groundwater

into the soil air due to its low aqueous solubility (60mg/l)

and its high vapour pressure (729 Pa m3/mol) (Schroder

and Munthe, 1998). Hg0 was found in aquifer solid samples

(Fig. 3C) and in soil gas measurements (Schondorf et al., 1999),

which supports this assumption. HgCl2 can be reduced to Hg0

either by humic substances (Allard and Arsenie, 1991) or most

likely in this case through bacterial reduction or reducing

metals (Barkay et al., 1992; Baldi et al., 1993). Aquifer and

groundwater have the potential to reduce HgCl2 to Hg0;

however, only small amounts of Hg0 were detected. The fast

flowing groundwater velocity and no strongly reducing

conditions in the groundwater lead assumingly to a slow

HgCl2 reduction.

The spatial distribution of Hg species in the groundwater

(Fig. 1B) shows that the proportion of reactive, inorganic Hg is

highest at the point where the Hg is entering the aquifer

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 9 1 – 1 0 098

(close to the area of former treatment hall). This supports the

results from the Hg species investigations in solid-phase

samples, as HgCl2 and soluble matrix-bound forms were also

found in this area (E3). HgCl2 is quite soluble (74 g/l) and thus

very mobile and can be transported over long distances.

Additionally, the association between the Hg2+ ion and

chloride influences the adsorption behaviour of Hg (Hahne

and Kroontje, 1973; Wang et al., 1991). With increasing

chloride concentrations desorption of Hg bound to soil

material or organic matter takes place and Hg will be

mobilized.

Fig. 1B also shows that with increasing distance from the Hg

source the proportion of organic Hg (humic bound) in the

groundwater increases. The elution of mobile HgCl2 out of the

soil through the fast flowing groundwater and the degassing

of Hg0 left more stable soil matrix complexed and humic-

bound Hg species behind. Additionally, during transport

DOM–Hg binding and retention through other transformation

and adsorption processes took place. Such could be the

precipitation of insoluble HgS species (Schuster, 1991), which

was also detected in solid-phase samples (Fig. 3D) or the

binding to biofilms (Wagner-Dobler et al., 2000).

3.5. Hg mobility

Groundwater concentrations showed that high amounts of Hg

are released from the soil and aquifer into the groundwater.

Prediction of Hg release from the soil and aquifer into

the groundwater is generally difficult but can be estimated

by means of batch and column experiments. The amount of

Hg which can be potentially leached from the soil was

estimated by batch experiments. Results showed a median

of 0.4 mg/kg of leachable Hg and range between o0.1 and

510 mg/kg. Only a few samples from bores E3 and B28 showed

elevated Hg concentrations in leachates ranging between

20 and 510 mg/kg. Those samples correlate with highest total

Hg concentrations in soil (Fig. 2A). The percentage of

leachable Hg ranges between 1% and 26% (median: 3%) and

is shown in Fig. 2B. Highest leachability was found in the

unsaturated soil zone at bores E3 and B28 ranging between

19% and 26% at a depth of 4–6 m below ground level. This part

contains high amounts of soluble HgCl2 (Fig. 3B), which can be

easily leached out of the soil resulting in the high Hg

concentrations in the leachates. In these parts inorganic,

reactive Hg species in the soil solution and groundwater are

expected due to the high solubility of HgCl2 (74 g/l). Analysis

of the Hg species distribution in the leachates confirms that

reactive inorganic Hg species (HgIIa+Hg0) are the predominant

form (90–100%) of leachable Hg (Fig. 2B).

In the aquifer and in the capillary fringe the amount of

mobile Hg decreases to 1–9% (�1–30mg/l Hgtot). Furthermore,

with increasing distance from the entry point within the

contamination plume the portion of reactive inorganic Hg in

the leachates decreases from 100% to 65% (median). Fluctuat-

ing water table and groundwater flow have leached the easy

soluble Hg species out of the soil, leaving more stable soil

matrix complexed and humic-bound Hg species behind.

A comparison of the Hg concentrations found in the ground-

water and those measured in the leachates of the batch

experiments shows that the location of the highest Hg

concentrations in groundwater is at P19 (229mg/l), whereas

E3 shows the highest Hg concentrations in soil solutions

(9000mg/l). Thus, the Hg entering the groundwater at E3 is

diluted by incoming non-contaminated groundwater and is

rapidly transported further downgradient.

The disadvantages of batch techniques have been widely

discussed (Communar et al., 2004). The breakdown of soil

aggregates during sample agitation, the relatively small soil/

solution ratio and differences in mass-transfer and hydro-

dynamic conditions often result in inappropriate estimates of

the degree of adsorption/desorption (Communar et al., 2004).

Thus, batch experiments are a good and easy initial tool to

estimate the potential risk of a contaminated site but they

give no information about the actual leaching rate under

natural conditions. Column experiments, however, can over-

come these limitations and estimates the leaching from the

soil more accurate. For this study, column experiments were

also performed showing a similar Hg species composition as

the batch experiments, whereas the leachates contained

10–17% (unsaturated zone) and 4–8% (saturated zone) less

Hg (data not published).

3.6. Implications for groundwater remediation

The high percentage of mobile Hg and low seepage fluxes will

result in a continuous Hg release over the next few centuries

from this contaminated site and as such the site requires

some form of remediation. However, there are some restric-

tions concerning the remediation of the site. A ‘‘classical’’

excavation of the contaminated soil is in most parts

impossible due to the residential area. Therefore groundwater

remediation is planned; however the high flow velocity in the

aquifer requires appropriate remediation technology. Pump

and treat systems (P&T) are known to have several disadvan-

tages (Voudrias, 2001). Especially for high flowing ground-

water systems, the capacity of a P&T to treat the total volume

of groundwater is not often sufficient. In addition, there are

several physical and chemical processes such as desorption

and diffusion processes of the contaminant that affect P&T

remediation (EPA, 1994). Alternatively, reactive barrier tech-

nology can overcome such problems. Such barriers are

installed in the path of migrating groundwater. For both

technologies of groundwater remediation, the filter material

has to be adapted to the existing situation in the aquifer.

Requirements for such filter materials are, for example, high

retention and selectivity for the contaminant, sufficiently

high hydraulic conductivity and long-term persistence

(Dahmke et al., 1996; Roehl et al., 2000). Previous investiga-

tions into location of the groundwater plume at this

contaminated site showed that the borders of the plume are

well defined and a PRB could easily be installed. The best

location would be as close as possible to the Hg entry point for

2 reasons: Firstly, the contamination of larger soil and

groundwater areas would be reduced and secondly the

possibility of Hg species transformation is limited.

The predominant occurrence of inorganic, reactive Hg

species (HgIIa) found in the groundwater close to the entry

point suggests the use of specific resins or reducing and

amalgamating filtering materials such as metal alloys.

Studies have shown that reactive Hg(II) can be easily reduced

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 9 1 – 1 0 0 99

to metallic Hg (Hg0) and retained by metals such as tin (Biester

et al., 2000) or copper (Huttenloch et al., 2003). The retardation

in such filters consists of two steps: the reduction of Hg2+ to

Hg0 on the metallic surface and further on the formation of a

Hg-metal amalgam. The investigated metals seem to be a very

effective low-cost opportunity concerning the remediation of

Hg; however, they only have been investigated in lab-scale so

far. Furthermore, humic-acid-bound Hg is not retained by

these filters (Biester et al., 2000). Thus, the Hg transformation

observed in the groundwater with increasing distance from

the entry point restricts the position of a metal alloy reactive

barrier close to the point source or demands the use of other

filter materials such as active carbon or Hg-selective poly-

meric resins.

As the soils of the contaminated site are highly polluted

with Hg (concentration exceeding 11,000 mg/kg) and the soil

surface is partly sealed, the Hg is leached in small amounts

into the groundwater and leaching will persists far into the

next centuries. Therefore, a combination of groundwater and

soil remediation techniques would be the most promising

way to remediate this site. Biester and Zimmer (1998)

demonstrated that matrix-bound Hg or HgCl2 in soils can be

easily transformed into stable Hg compounds such as

metacinnabar (HgS) or insoluble organic Hg–S compounds

by treating contaminated soil with polysulfide solutions.

These immobilization treatments could be performed in

areas with high Hg contamination (e.g. E3) to prevent or

reduce further leaching into the groundwater.

Thus, the use of amalgamating filters and the immobiliza-

tion of the Hg in the hot spots would be the best remediation

strategies for this contaminated site and field-scale investiga-

tions have to be subject to further studies.

4. Conclusion

This study showed that the determination of Hg species in

soils and groundwater of a HgCl2 contaminated site is an

essential tool to determine the Hg risk at this site. In soils low

in organic matter most HgCl2 is bound to the inorganic soil

components and results in potentially high mobility of

inorganic reactive Hg species. These species pose a high risk

as they can be easily transformed in the groundwater into

other Hg species, such as organic Hg or Hg0. Redox conditions

in the groundwater indicate that HgCl2 should be reduced to

Hg0 and solid-phase Hg speciation measurements show that

reduction takes place in the aquifer and the capillary fringe.

Low seepage fluxes and high mobility are leading to a high Hg

release over long time periods. Despite high flow velocities

the Hg contamination plume has a relatively short length.

We assume that this is due to the dilution and the formation

of Hg0, which has a solubility product 8�10�7 lower than that

of HgCl2. In addition, degassing of Hg0 will further reduce

Hg concentrations in the groundwater. Furthermore, the

identification of Hg species also gives important information

on the type of remediation technology, which should be

applied on a contaminated site. Results of soluble Hg

speciation suggest that filtering materials should be adapted

to ionic Hg species, e.g. specific resins or amalgamating metal

alloys.

Acknowledgements

We thank Thomas Schondorf from Harres Pickel Consult AG,

Freiburg, Germany for cooperation. This study was supported

by the Regional Council of Baden-Wurttemberg, Germany.

R E F E R E N C E S

Allard, B., Arsenie, I., 1991. Abiotic reduction of mercury by humicsubstances in aquatic system—an important process for themercury cycle. Water Air Soil Pollut. 56, 457–464.

Baldi, F., Parati, F., Semplici, F., Tandoi, V., 1993. Biological removalof inorganic Hg(II) as gaseous elemental Hg(0) by continuousculture of a Hg-resistant Pseudomonas putida strain FB-1. WorldJ. Microbiol.Biotechnol. 9, 275–279.

Barkay, T., Turner, R., Saouter, E., Horn, J., 1992. Mercurybiotransformations and their potential for remediation ofmercury contamination. Biodegradation 3, 147–159.

Barnett, M.O., Harris, L.A., Turner, R.P., Henson, T.J., 1995.Characterization of mercury species in contaminated flood-plain soils. Water Air Soil Pollut. 80, 1105–1108.

Benoit, J.M., Mason, R.P., Gilmour, C.C., Aiken, G.R., 2001. Constantsfor mercury binding by dissolved organic matter isolates from theFlorida Everglades. Geochim. Cosmochim. Acta 65 (24), 4445–4451.

Biester, H., 1994. Application of a Temperature ControlledPyrolysis Method for the Detection of Hg-species in Soils andSediments (Moglichkeiten der Anwendung eines temperatur-gesteuerten Pyrolyseverfahrens zur Bestimmung der Bin-dungsform des Quecksilbers in Boden und Sedimenten).Heidelberger Geowissenschaftliche Abhandlungen Band, Hei-delberg, vol. 75, 156pp.

Biester, H., Nehrke, G., 1997. Quantification of mercury in soilsand sediments—acid digestion versus pyrolysis. Fresenius J.Anal. Chem. 358, 446–452.

Biester, H., Scholz, C., 1997. Determination of mercury bindingforms in contaminated soils: mercury pyrolysis versussequential extractions. Environ. Sci. Technol. 31 (1), 233–239.

Biester, H., Zimmer, H., 1998. Solubility and changes of mercurybinding forms in contaminated soils after immobilizationtreatment. Environ. Sci. Technol. 32 (18), 2755–2762.

Biester, H., Schuhmacher, P., Muller, G., 2000. Effectiveness of mossytin filters to remove mercury from aqueous solution by Hg(II)reduction and Hg(0) amalgamation. Water Res. 34 (7), 2031–2036.

Bisogni Jr, J.J., 1989. Using mercury volatility to measure redoxpotential in oxic aqueous systems. Environ. Sci. Technol. 23,823–831.

Boening, D.W., 2000. Ecological effects, transport, and fate ofmercury: a general review. Chemosphere 40, 1335–1351.

Brosset, C., 1987. The behaviour of mercury in the physicalenvironment. Water Air Soil Pollut. 34, 145–166.

Communar, G., Keren, R., Li, F., 2004. Deriving boron adsorptionisotherms from soil column displacement experiments. SoilSci. Soc. Am. J. 68, 481–488.

Dahmke, A., Lensing, H.J., Schafer, D., Schafer, W., Wust, W., 1996.Prospects for the use of artifical and natural geochemicalbarriers for groundwater remediation (Perspektiven zur Nut-zung kunstlicher und naturlicher geochemischer Barrieren beiGrundwasserkontamiationen). Geowissenschaften 5, 186–195.

EPA, 1994. Methods for Monitoring Pump-and-Treat Performance.Environmental Protection Agency EPA/600/R-94/123, Wa-shington, DC.

Hahne, H.C.H., Kroontje, W., 1973. Significance of pH and chlorideconcentration on behaviourof heavy metal pollutants: mer-cury(II), cadmium(II), zinc(II), and lead(II). J. Environ. Qual. 2(4), 444–450.

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 9 1 – 1 0 0100

Haitzer, M., Aiken, G.R., Ryan, J.N., 2003. Binding of mercury(II) toaquatic humic substances: influence of pH and source ofhumic substances. Environ. Sci. Technol. 37 (11), 2436–2441.

Harres Pickel Consult, 2006. Groundwater contamination in thearea Parksiedlung, Bad Krozingen, remediation investigationOctober 2004—October 2005 (Grundwasserschadensfall imBereich Parksiedlung Bad Krozingen, SanierungsuntersuchungOktober 2004—Oktober 2005), unpublished report.

Huttenloch, P., Roehl, K.E., Czurda, K., 2003. Use of copper shavings toremove mercury from contaminated groundwater or wastewaterby amalgamation. Environ. Sci. Technol. 37 (18), 4269–4273.

Krishnan, S.V., Gullet, B.K., Jozewicz, W., 1994. Sorption ofelemental mercury by activated carbons. Environ. Sci. Tech-nol. 28 (8), 1506–1512.

Lovely, D.R., 1995. Bioremediation of organic and metal contami-nants with dissimilatory metal reduction. J. Ind. Microbiol.Biotechnol. 14 (2), 85–93.

Meili, M., Iverfeldet, A., Hankanson, L., 1991. Mercury in thesurface water of Swedish forest lakes—concentrations, spe-ciation and controlling factors. Water Air Soil Pollut. 56,439–453.

Mierle, G., Ingram, R., 1991. The role of humic substances in themobilization of mercury from watersheds. Water Air SoilPollut. 56, 349–357.

Muller, G., Gastner, M., 1971. The Karbonat-Bombe, a simpledevice for the determination of the carbonate content insediments, soils and other materials. Neues Jahrbuch furMineral. /Monatsheft 10, 466–469.

Revis, N.W., Osborne, T.R., Sedgley, D., King, A., 1989. Quantitativemethod for determing the concentration of mercury(II)sulphide in soils and sediments. Analyst 114 (7), 823–826.

Roehl, K.E., Czurda, K., Simon, F.G., Meggyes, T., 2000. Long-termperformance of permeable reactive barriers used for theremediation of contaminated water. In: ETCA Workshop onthe Protection of European Water Resources, Venice, 06/21-06/23/2000, pp. 179–185.

Scheffer, F., Schachtschabel, P., 2002. Textbook of Soil Science(Lehrbuch der Bodenkunde). Spektrum Akademischer Verlag,Heidelberg-Berlin, 62 pp.

Schondorf, T., Heinrichsmeier, K., Biester, H., Achstetter, U., 1995.New ways and experiences during the exploration of amercury contaminated soil in a residential area (Neue Wege

und Erfahrungen bei der Erkundung einer Bodenbelastungdurch Quecksilber in einer Wohnsiedlung). TerraTech 5, 31–35.

Schondorf, T., Egli, M., Biester, H., Mailahn, W., Rotard, W., 1999.Distribution, bioavailibility and speciation of mercury incontaminated soil and groundwater of a former woodimpregnation plant. In: Ebinghaus, R. (Ed.), EnvironmentalScience—Mercury Contaminated Sites. Springer, Berlin–Hei-delberg, pp. 181–206.

Schroder, W.H., Munthe, J., 1998. Atmospheric mercury—anoverview. Atmos. Environ. 32 (5), 809–822.

Schuster, E., 1991. The behaviour of mercury in the soil withspecial emphasis on complexation and adsorption processe-s—a review of the literature. Water Air Soil Pollut. 56, 667–680.

Skyllberg, U., Xia, K., Bloom, P.R., Nater, E.A., Bleam, W.F., 2000.Binding of mercury(II) to reduced sulphur in soil organicmatter along upland peat soil transects. J. Environ. Qual. 29 (3),855–865.

Skyllberg, U., Bloom, P.R., Qian, J., Lin, C.M., Bleam, W.F., 2006.Complexation of mercury(II) in soil organic matter: EXAFSevidence for linear two coordination with reduced sulphurgroups. Environ. Sci. Technol. 40 (13), 4174–4180.

Stein, E.D., Cohen, Y., Winer, A.M., 1996. Environmental distribu-tion and transformation of mercury compounds. Crit. Rev.Environ. Sci. Technol. 26 (1), 1–43.

UVM-BW, 1998. Administrative regulations of treshold values forremediation of contaminated sites by the ministry of envir-onmental health of Baden–Wurttemberg (Gemeinsame Ver-waltungsvorschrift des Ministeriums fur Umwelt und Verkehrund das Sozialministeriums uber Orientierungswerte fur dieBearbeitung von Altlasten und Schadensfallen) as amendedon 03/01/1998.

Voudrias, E.A., 2001. Pump-and-treat remediation of groundwatercontaminated by hazardous waste: can it really be achieved?Global Nest: Int. J. 3 (1), 1–10.

Wagner-Dobler, I., Lunsdorf, H., Lubbehusen, T., van Canstein,H.F., Li, Y., 2000. Structure and species composition ofmercury-reducing biofilms. Appl. Environ. Microbiol. 66 (10),4559–4563.

Wang, J.S., Huang, P.M., Liaw, W.K., Hammer, U.T., 1991. Kinetics ofthe desorption of mercury from selected freshwater sedi-ments as influenced by chloride. Water Air. Soil Pollut. 56,533–542.