You are on page 1of 16

Analytica Chimica Acta 479 (2003) 233–248

Selective extractions to assess the biogeochemically


relevant fractionation of inorganic mercury in
sediments and soils
Nicolas S. Bloom∗ , Eve Preus, Jodie Katon, Misti Hiltner
Frontier Geosciences Inc., 414 Pontius Ave North, Seattle, WA 98109, USA
Received 15 August 2002; received in revised form 2 December 2002; accepted 4 December 2002

Abstract
Here we present a new method for sequential selective extractions (SSEs) for Hg in geological solids, validated with ex-
tensive quality assurance procedures. Mercury was separated into fractions which “make sense” biogeochemically, rather
than being identified by specific compounds. Experiments elucidated the effects of extraction time, solids-to-liquid ra-
tio, and alternate solvents in natural samples, reference materials, and pure compounds. Compounds tested included HgS
(red and black), HgCl2 , Hg0 , Hg2 Cl2 , HgSe, HgO, Hg(II) adsorbed on goethite, Hg–humate, and gold amalgamated Hg.
Based on these findings, a five-step sequence of extractions was established to separate the compounds into biogeochemi-
cally distinct categories. The fractions and leaching media were as follows: F1 (deionized water), F2 (0.01 M HCl + 0.1 M
CH3 COOH), F3 (1 M KOH), F4 (12 M HNO3 ), and F5 (aqua regia). Method blanks and method detection limits (MDLs)
of 0.1–5 ng/g were obtained for the various analytical fractions, depending on the reagent concentrations used. Preci-
sion ranged from 2 to 8% for the major fractions in a sample, but increased to 2–40% for fractions making up <5%
of the total. Recovery of total Hg by the sum of species in reference materials showed that the accuracy of the method
ranges from 90 to 105%. Methylation potential, determined by anoxic incubation sample aliquots with biologically ac-
tive sediments, showed that inorganic Hg extracted in the F3 fraction is most strongly correlated with methylation po-
tential. In most natural and sediment incubation samples, the majority of Hg present was found either in the F3 or F5
fractions.
© 2002 Elsevier Science B.V. All rights reserved.
Keywords: Mercury; Speciation; Sediment; Sequential extraction

1. Introduction systems, because the particular distribution of metal


compounds and their interaction with the native ma-
Geochemists have long sought the elusive goal of trix under aqueous conditions determine their environ-
quantifying forms (speciation) of mercury in soils and mental mobility and bioavailability [1]. A variety of
sediments. Solid phase chemical speciation is critical approaches have been attempted to divulge the spe-
to understanding and modeling metal-contaminated ciation of Hg in solids, including sequential selec-
tive extractions (SSEs) [2–7], thermal volatilization
∗ Corresponding author. Tel.: +1-206-622-6960; [8], wave dispersive X-ray microprobe spectroscopy
fax: +1-206-622-6870. (WD/XMP) [9], and extended X-ray absorption fine
E-mail address: nicolasb@frontiergeosciences.com (N.S. Bloom). structure spectroscopy (EXAFS) [10].

0003-2670/03/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0003-2670(02)01550-7
234 N.S. Bloom et al. / Analytica Chimica Acta 479 (2003) 233–248

Inherent methodological limitations have slowed 2. Experimental


progress toward the development of a practical solid
phase speciation scheme for mercury in environmen- 2.1. Sequential selective extraction materials
tal samples. Those methods capable of making accu-
rate assertions concerning the ligands binding to Hg The sequential extraction procedure uses the fol-
(such as EXAFS and WD/XPS) have such high limits lowing five solutions, in the same order as presented.
of detection (>100 ␮g/g Hg) as to render them use- Also included are instructions for preparing 0.2N BrCl
less for all ambient, and most contaminated samples. solution [17], since this is required as an oxidizer
Conversely, extraction-based methods, though capa- of the extracts, prior to total Hg determination. All
ble of quantifying at very low levels (<0.005 ␮g/g), solutions were prepared from reagent-grade chemicals
inherently confound in situ compound-specific infor- and were pre-tested and found to be sufficiently low
mation to the point that only operationally defined in Hg (less than 10 ng/l) before use. All analytical
“speciation” is possible. Selective extraction proce- procedures were conducted using ultra-clean sample
dures [11–13] are commonly used to study Fe, Mn, handling [18] to avoid contaminating low-level sample
and relatively “typical” transition metals such as Cu, extracts, the lab, and cross-contamination of high-level
Ni, Co and Zn. These are not appropriate for Hg samples.
work however, due to the numerous and diverse Hg
species with unique physical and chemical properties. 2.1.1. 0.2N bromine monochloride (BrCl) solution
Mercury-specific methods that have been utilized give Twenty-seven grams of KBr crystals were added
contradictory results on ambient samples [1,14]. This to a 2.5 l bottle of low Hg reagent-grade concentrated
confusion is due both to insufficient QA validation HCl. Using a clean magnetic stir bar, the contents
using pure compounds, and to over-interpretation of were stirred 1 h in a fume hood. Slowly, 38 g of
the specificity of the methods. pre-analyzed, low Hg KBrO3 were added to the acid
Although this dilemma remains today, we have mixture while stirring. When all of the KBrO3 is
refined and extensively validated a solid phase analyt- added, the solution should change from yellow to red
ical scheme based upon selective extractions. Instead to orange. The bottle was loosely capped, and the
of species-specific information, this method provides solution stirred for another hour before tightening the
differentiation of Hg compounds into behavioral lid. This process generates copious quantities of free
classes, including (a) water soluble (F1), (b) ‘hu- halogens (Cl2 , Br2 , BrCl) which are released from
man stomach acid’ soluble (F2), (c) organo-chelated the bottle, so KBrO3 was added slowly in a fume
(F3), (d) elemental Hg (F4), and (e) mercuric sul- hood.
fide (F5), while still providing sufficiently low limits
of detection for background environmental stud- 2.1.2. Fraction 1 (F1), deionized water
ies. Of equal importance, the method was devel- A 2.0 l Teflon bottle was filled with reagent-grade
oped to be easily implemented by most trace metals double deionized water, and then purged overnight
laboratories. with high purity argon at a flow rate of 100 ml/min.
Initially, several published selective extraction The bottle was kept completely full, with the cap
schemes were cross-checked for selectivity using 10 tightly closed until use. In cases where Hg0 is not be-
pure Hg compounds dispersed in kaolin clay. Based ing measured, unpurged reagent-grade deionized wa-
on these findings, extraction kinetics, and effects of ter is acceptable.
solid-to-liquid ratio, a five-step composite method
was developed explicitly for the fractionation of Hg 2.1.3. Fraction 2 (F2), pH 2, 0.1 M CH3 COOH
compounds. Afterwards, the finalized method was + 0.01 M HCl
compared with other published techniques, such as A 2.0 l Teflon bottle was half filled with reagent-
the US EPA toxic characterization leaching protocol grade deionized water, to which 12 ml of glacial acetic
(TCLP) [15] and an in vitro artificial human stom- acid and 1.6 ml of concentrated HCl were added. The
ach extraction [16], using a variety of samples from bottle was then filled to the neck with deionized water,
contaminated sites. and shaken to homogenize.
N.S. Bloom et al. / Analytica Chimica Acta 479 (2003) 233–248 235

2.1.4. Fraction 3 (F3), 1 M KOH by adding 1.25 ml of 0.2 M BrCl before the rinse was
A 2.0 l Teflon bottle was half filled with reagent- added. The F3 fraction, owing to its high acid neutral-
grade deionized water, and 132 g of reagent-grade izing capacity, is oxidized by adding 10.0 ml of BrCl
KOH pellets (85% KOH, 15% water) were added solution.
with constant swirling. After dissolution, the bottle As a rinse step, the extraction vials containing the
was filled to the neck with reagent water and shaken sediment pellet were refilled with the same extrac-
to homogenize. tant, shaken vigorously to resuspend the sediment,
re-centrifuged, and filtered. The rinse was then added
2.1.5. Fraction 4 (F4), 12 M HNO3 to the extract from the same sample and the combined
To a 2.0 l Teflon bottle, 500 ml of reagent-grade sample diluted to 125 ± 1 ml with deionized water,
water was added, followed by the slow addition of as marked by the base of the bottle neck. After the
concentrated HNO3 up to the neck of the bottle. The rinse step, the next sequential extractant was added,
bottle was then capped and shaken to homogenize. the vials shaken to resuspend the sediment, and the
This reagent was stored in the dark to avoid the for- 18 h extraction and rinse procedure was repeated for
mation of brown NO2 . It is critical that the HNO3 and each extraction fraction except the last. After oxida-
Teflon bottle used are trace chlorine-free. tion with BrCl and dilution to volume, all selective
extracts can be stored indefinitely in glass bottles with
2.1.6. Fraction 5 (F5), aqua regia Teflon-lined caps until analysis.
This reagent is produced in situ at the final extrac- For the aqua regia extraction (F5), the reagents were
tion step. The procedure generates copious-free Cl2 added to the sediment pellet directly in the origi-
gas, so it must be conducted in a fume hood. Ten nal 40 ml vial. The samples were allowed to digest
milliliters of concentrated HCl was added to the sedi- overnight at room temperature in loosely capped vials,
ment pellet remaining in the 40 ml vial. After swirling and then diluted to 40.0 ml with deionized water. Aqua
the sample to dislodge the sediment, 3.0 ml of concen- regia extracts all known Hg species in most environ-
trated HNO3 was added, and the vials were loosely mental matrices, so it can be used for the direct de-
capped. As nitrosyl chloride (NOCl) and free Cl2 are termination of total Hg on sediments and soils, using
generated, the solution turns a dark orange color. sample aliquots of up to 2.00 g, dry weight basis [19].
In the case of crystalline metal ores such as bauxite
2.2. Selective extraction procedure and hematite, a significant residue of Hg remains after
aqua regia digestion [20]. In such cases, a sixth step
Extractions were carried out using 0.40 ± 0.04 g incorporating the dissolution of the final pellet in a
of dry solid, or the equivalent mass of wet sedi- mixture of HF + HNO3 + HCl at elevated temperature
ments, in 40.0 ml trace metal clean borosilicate vials and pressure may be warranted.
with Teflon-lined caps. All materials should be finely
ground or sieved to <100 ␮m prior to extraction. Each 2.3. Analytical procedure for total mercury
step of the extraction procedure except for the final
(F5) step included a rinse between each new extractant. Total Hg in each oxidized fraction extract was
An initial aliquot of extractant was added to the solids determined using SnCl2 reduction, purge and trap
for 18±4 h with end-over-end tumbling at 30 rpm. The gold amalgamation preconcentration, and cold vapor
vials were then centrifuged at 3000 rpm for 20 min, atomic fluorescence spectrometry (CVAFS) detection
and the supernatant liquid decanted for filtration. [17,19,21]. This method has recently been validated
For the first three fractions, centrifuged extracts and promulgated for aqueous samples as US EPA
were filtered through disposable 0.4 ␮m nitrocellu- Method 1631 [22]. A sample aliquot was pipetted
lose filter units. The last two fractions (F4 and F5) into a pre-purged bubbler containing 100 ml of 1%
do not employ a filtration step because those solu- HCl in deionized water, and 0.3 ml of 25% SnCl2
tions dissolve cellulose nitrate membranes. The ex- solution. For the 12 M HNO3 and aqua regia digests,
tractant from fractions F1, F2, and F4 were placed into no more than 5.0 ml of solution should be analyzed,
clean 125 ml borosilicate glass bottles and oxidized due to potential signal inhibition from the high acid
236 N.S. Bloom et al. / Analytica Chimica Acta 479 (2003) 233–248

concentration of these extracts. The Hg(II) in the bub- 24 h to allow gas phase equilibration. For our analysis,
bler is converted to Hg0 , which was then purged with the headspace gas was injected into an argon stream
N2 (20 min at 300 ml/min) onto a gold-coated sand passing through a gold-sand trap, which was then ana-
trap. Mercury collected on the trap was then ther- lyzed using the dual amalgamation/CVAFS procedure.
mally desorbed (2.5 min at 450 ◦ C) under argon flow The presence of Hg0 in the headspace at equilibrium
onto a similar “analytical” trap, which is part of the concentrations (15.54 ng/ml at 22.0 ◦ C), confirms the
CVAFS detector. The Hg collected on the analytical presence of free liquid elemental Hg in a sample. In
trap was then thermally desorbed (1.5 min at 450 ◦ C) cases where free elemental Hg was confirmed, the to-
under argon flow into the atom cell of the CVAFS tal Hg measured in the F4 fraction was assumed to
detector. Signals were quantified by chart recorder largely represent the Hg0 content of the sample.
or peak integrator. The instrument was calibrated by
reducing and purging known masses of pure Hg(II) 2.5. Analytical procedure for methyl mercury
aqueous standard solution (0.05–10 ng Hg), made
up by dilution of certified NIST-3133 solution. The Sediments and soils (0.5 g aliquots) were first ex-
instrumental detection limit is approximately 0.1 pg tracted from a KBr/H2 SO4 /CuSO4 mixture into 10 ml
Hg, although method detection limits (MDLs) are of CH2 Cl2 by vigorous shaking for 1 h. After cen-
ultimately determined by the reagent blanks. trifugation to separate the aqueous and organic lay-
ers a 2.0 ml subaliquot of the CH2 Cl2 layer was then
2.4. Analytical procedure for elemental mercury back-extracted by solvent volatilization into 60 ml of
pure water. This procedure avoids positive analytical
Elemental Hg (Hg0 ) was directly determined on the artifacts associated with the distillation extraction pro-
unpreserved deionized water extract (F1) prior to de- cedure [23,24]. Aliquots of the final extract were ana-
canting or filtering (to avoid volatilization losses) us- lyzed by aqueous phase ethylation, purge and trap onto
ing the same instrumentation as described for total Carbotrap, isothermal GC separation, and CVAFS de-
Hg. The only difference was that the extract aliquot tection [25,26]. The system was calibrated by ethyla-
was added to a bubbler containing pre-purged 1% HCl tion of known masses (5–200 pg) of CH3 Hg contained
in deionized water, with no addition of SnCl2 . The in a 10 ␮g/l standard, which was calibrated for total
reduction reaction is so sensitive that a bubbler that Hg against NIST-3133 as described above. The ab-
has never been in contact with SnCl2 must be used solute detection limit of the system is approximately
to purge out the Hg0 , which is intrinsically volatile. 0.1 pg Hg as CH3 Hg, which results in a method detec-
The analytical system was calibrated using the same tion limit of approximately 0.005 ng/g Hg under the
Hg(II) standards, reduced with SnCl2 , in separate bub- conditions described.
blers, as described for total Hg. Because Hg0 reaches
saturation in water at approximately 50 ␮g/l, values 2.6. Toxic characteristic leaching procedure (TCLP)
measured at or near this concentration in the F1 frac-
tion suggest that the sample contains free liquid metal- The TCLP procedure employs a solution of 0.5 M
lic mercury. In these cases, the total Hg measured sodium acetate and 0.5 M glacial acetic acid (pH of
in the F4 fraction gives a good estimate of the total 4.9 ± 0.1) to simulate leachability of solids within the
elemental Hg in the sample. If a significant amount moderately acidic environment of a municipal landfill
(greater than 1 ml) of the F1 extract is used for Hg0 [6,15]. Because only finely powdered samples were
analysis, a volume correction factor must be intro- analyzed for this study, sample heterogeneity was
duced in the calculation of the total Hg in the F1 not a significant problem, thus allowing a miniatur-
fraction. ized version of the TCLP procedure to be employed.
Confirmation of the presence of free liquid Hg0 can Instead of extracting 100 g of solids with 2.0 l of so-
be alternatively determined by the syringe removal of lution, 2.0 g of solids were extracted with 40.0 ml of
an aliquot (0.10–5.00 ml) of the headspace gas over a solution, maintaining the same 20:1 solution to solids
5–20 g aliquot of sediment in a closed glass jar. The ratio as the standard procedure. Samples were ex-
sample must remain at a known temperature for at least tracted for 18 ± 4 h in 40 ml borosilicate glass vials,
N.S. Bloom et al. / Analytica Chimica Acta 479 (2003) 233–248 237

at room temperature by end-over-end tumbling at (0.01–3 g) of contaminated soils or pure compounds


30 rpm. After extraction, the samples were centrifuged dispersed in kaolin to obtain a final total Hg concen-
for 20 min at 3000 rpm, and the supernatant solution tration of about 5 ␮g/g (wet basis). The bottles were
was vacuum filtered through disposable 0.4 ␮m nitro- tightly sealed and placed on a roller at approximately
cellulose filter units. The filtrates were then oxidized 5 rpm at room temperature (18–25 ◦ C). After 3 weeks,
and preserved by the addition of 2% by volume of the samples were removed from the roller and peri-
0.2 M BrCl solution, and later analyzed for total Hg odically shaken until aliquotting for selective extrac-
by CVAFS as described above. tions and CH3 Hg analysis. Aliquots were analyzed
at 1-, 3-, and 22-week intervals. Incubations were
2.7. In vitro simulated mammalian stomach conducted in duplicate, and mean results reported.
extraction In almost all cases, the relative percent difference in
concentrations between the duplicates was less than
The in vitro procedure was created to simulate 25% for measures that were greater than 10 times the
bioaccessability of metals contained in geological MDL.
solids within the acidic environment of the mam- Another set of incubations were performed for 19
malian gastrointestinal tract. This extraction proce- weeks. Several Hg compounds were added to three
dure has been successfully compared to in vivo ani- sediment types, marsh and sandy sediments from
mal bioavailability results for Pb and As, but not yet Green Lake (Seattle, WA), and sandy marine sedi-
for Hg [16]. Finely powdered aliquots (1.0 ± 0.05 g) ments from Puget Sound. Solids were added as above,
of the test solids were extracted with 100 ml of sim- but were not duplicated and were only shaken weekly.
ulated stomach acid solution, a mixture of 0.4 M The Green Lake sediments contained 11 ng/g total Hg,
glycine and 0.7 M HCl acid in reagent water. Samples <100 ␮g/g SO4 2− , and 0.2% TOC in the sandy area;
were then tumbled end-over-end at 30 rpm for 1.0 h in and 19 ng/g total Hg, <100 ␮g/g SO4 2− , and 11.1%
125 ml high density polyethylene (HDPE) bottles, at TOC in the marshy area. The Puget Sound sandy
37 ± 2 ◦ C. After extraction, the samples were syringe sediment contained 8 ng/g Hg, 1200 ␮g/g SO4 2− , and
filtered through disposable 0.4 ␮m cellulose acetate 0.2% TOC.
filters. The filtrates were oxidized and preserved by
adding 1% by volume of 0.2 M BrCl solution, and
were later analyzed for total Hg by CVAFS as de- 3. Results and discussion
scribed previously.
3.1. Effect of extraction time
2.8. Natural sediment methylation assay
To investigate the effects of extraction time on
To assess their methylation potential, various sub- the amount of Hg removed with each extractant,
strates were mixed with a natural low Hg, high or- several experiments were conducted with pure Hg
ganic matter freshwater sediment, and incubated in compounds, natural samples, and certified reference
sealed glass bottles. The sediment was taken from the materials (CRMs). Because these experiments gener-
0–10 cm layer of a forest pond in the Cascade Moun- ally lead to similar conclusions, only one is illustrated
tains (Washington), sieved through a 1.4 mm screen, (Fig. 1). In this example, a composite soil from an
and kept at room temperature in an aquarium over- historic gold mine tailings (GMT) pile was extracted
lain with aerated pond water. After homogenization, with constant agitation using each of the test ex-
the sediment was mixed with overlying water to con- tractants at a 1:100 solids-to-liquid ratio for periods
tain ∼20% solids. This slurry, pH of 6.4, contained ranging from 1 to 48 h. GMT was chosen because it
192 ng/g of total Hg, 1.44 ng/g of methyl Hg, <1.6 ␮g/ had previously shown significant leachability in each
g of acid volatile sulfide, 480 ␮g/g of sulfate, and of the fractions. This experiment was performed us-
4.8% total organic carbon (TOC; dry weight basis). ing simultaneous independent extractions, rather than
In a 125 ml glass jar with Teflon-lined caps, 100 g of the sequential extractions employed in the final ana-
Cascade pond slurry were mixed with a small amount lytical procedure. As a result, the Hg culled by each
238 N.S. Bloom et al. / Analytica Chimica Acta 479 (2003) 233–248

3.2. Effect of solids-to-liquid ratio

The effect of the solids-to-liquid ratios on the


amount of Hg extracted in each fraction was also
investigated using several matrix and extractant com-
binations. Again we chose to use GMT as an ex-
ample, because it shows significant extractability
in all fractions (Fig. 2). In most cases, a smaller
solid-to-extractant ratio resulted in greater amounts
of Hg leached to a particular fraction. This effect
was exaggerated with the weaker extractants and
compounds of moderate to low solubility, such as
Hg2 Cl2 , HgO, and Hg0 . For compounds of very high
or very low solubility (e.g. HgCl2 , HgS) and for the
stronger extractants (F3–F5), the amount of mercury
Fig. 1. Amount of mercury extracted by each extractant solution
extracted was more independent of the solids-to-liquid
from gold mine tailings (141 ␮g/g Hg) as a function of extraction ratio.
time (liquid-to-solid ratio 1:100). Between 16 and 48 h, variance The amount of Hg leached in each fraction tended
from the average concentration was 3% for F1, 2% for F2, 7% to reach a plateau at solid-to-extractant ratios around
for F3, and 11% for F4. 1:100, although actual leachability is also a function
of the specific compound solubility, its concentration
in the solid, and the adsorptive (chemical and sur-
extractant should be approximately equal to the sum face area) nature of the soil matrix. Given the goal of
of mercury obtained by each preceding extractant in developing a practical analytical tool, this set of in-
the standard procedure. teractions is too complex to resolve for each sample
For the majority of species and extractants tested, encountered. In theory, it is best to utilize the low-
there was an initial period of rapidly changing so- est solids-to-liquid ratio possible to leach each species
lution concentration, which slowed after 12–24 h. In fully. In practice, the ratio is limited by real-world sam-
most cases, the concentration of Hg in solution con- ple heterogeneity and the ability to accurately weigh
tinued to increase with time. In samples with low Hg
and high organic matter, the concentration of Hg in
F3 was initially high and decreased over time, pos-
sibly due to re-adsorption of Hg by humic materi-
als. For very soluble or insoluble compounds (such as
HgCl2 or HgS) there was a point where the amount of
mercury extracted remained constant for longer peri-
ods. In all other cases, the concentration continued to
change, albeit at an increasingly slower pace, over the
entire length of the experiment. Because 70–90% of
the change occurred within the first 12–24 h of leach-
ing for most analyte/matrix combinations, an extrac-
tion time of 18 ± 4 h was chosen for each phase. This
allows the extractions to proceed according to a sim-
ple daily rhythm, provides consistency with the EPA
TCLP and EP-Tox extraction procedures, and oper-
ates over a time period where moderate differences in
Fig. 2. Effect of liquid-to-solid ratio on the amount of mercury
the extraction time produce minimal differences in the extracted from gold mine tailings (141 ␮g/g Hg) using an 18 h
amount of analyte extracted. extraction time.
N.S. Bloom et al. / Analytica Chimica Acta 479 (2003) 233–248 239

very small sample masses. Thus, a ratio of 1:100


was adopted, requiring the use of 400 mg sample to
40 ml of extractant. This ratio also enables the method
to reach sub-ppb detection limits for most fractions,
a feat that would not be possible at higher dilution
factors.

3.3. Effect of analyte concentration

Related to the effect of the solids-to-liquid ratio


is the effect the concentration of a given solid phase
Hg species has upon its ultimate leachability. We
measured the Hg leached in each fraction at vari-
ous concentrations of HgCl2 (very soluble), Hg2 Cl2
(slightly soluble) and HgS (insoluble) compounds
suspended in kaolin (Fig. 3). Overall, as the concen-
tration of total Hg decreased, it was more difficult to
assign likely Hg species to observed leaching patterns.
While the results generated from low-level samples
(<100 ␮g/g total Hg) represent the biogeochemical
behavior of the samples (i.e. water solubility, bioavail-
ability), assignment of an extraction fraction to a
particular Hg species class becomes somewhat more
ambiguous.
In general, as the Hg concentration of a sample
decreases, more mercury is found in the F3 and F4
extraction phases, regardless of the species initially
added. At low concentrations it appears as though
soluble species (such as HgCl2 ) are re-adsorbed Hg
on the clay substrate, which pushes some of the Hg
present to subsequent extraction fractions. Alternately,
small amounts of highly insoluble species, such as
HgS, may be oxidized to soluble forms in the ex-
traction, for example by trace Cl2 in the F4 step. Al-
though small in an absolute sense, these appear as
larger and larger proportions as the total Hg concen-
tration decreases. Elemental Hg is particularly prob-
lematic at low concentrations, because free Hg0 has
an equilibrium solubility of about 50 ␮g/l in aque-
ous solution (5 ␮g/g under the conditions used in this Fig. 3. Effect of sample concentration on mercury extracted in
each fraction using (a) HgCl2 (very soluble), (b) Hg2 Cl2 (slightly
study). Thus, at decreasing concentrations, the pro- soluble) and (c) HgS (insoluble).
portion of the total Hg found in the F1–F3 fractions
increases significantly. Using this extraction scheme,
a sample containing 20 ␮g/g of Hg0 would be indis- 3.4. Method detection limits, blanks and CRM
tinguishable from a sample containing 5 ␮g/g each of recoveries
HgCl2 , HgO, Hg–humate and Hg0 , unless confirma-
tory identification for the presence of free Hg0 is also We performed a method blank, CRM recovery,
made. and MDL study similar to 40 CFR 136 (US Federal
240 N.S. Bloom et al. / Analytica Chimica Acta 479 (2003) 233–248

Table 1
Example of method blank and estimated method detection limit (eMDL) study, method detection limit (MDL) study with standard
NIST-1646, and recoveries for a number of certified reference materials
Sample name Mercury concentrations (ng/g)

F1 F2 F3 F4 F5 Sum Percent recovery

Blank-1 0.09 0.12 1.63 3.8 0.60


Blank-2 0.06 0.13 1.55 1.5 0.89
Blank-3 0.06 0.08 1.35 1.5 0.67
Blank-4 0.06 0.10 1.29 1.1 0.35
Mean 0.07 0.11 1.45 1.6 0.63
S.D. 0.02 0.02 0.16 1.6 0.22
eMDL 0.05 0.07 0.48 4.8 0.67
MDL study
NIST-1646 rep 1a 0.40 0.04 21.0 17 1.2 39 106
NIST-1646 rep 2 0.45 0.22 21.5 17 2.3 41 111
NIST-1646 rep 3 0.40 −0.06 19.5 15 2.9 38 102
NIST-1646 rep 4 0.34 0.23 20.6 16 1.9 39 106
NIST-1646 rep 5 0.40 0.03 20.0 16 2.8 39 105
NIST-1646 rep 6 0.35 0.04 19.0 17 3.1 40 107
NIST-1646 rep 7 0.36 0.07 15.6 21 2.8 40 107
NIST-1646 rep 8 0.31 0.82 21.0 17 3.9 43 117
Mean 0.38 0.17 19.8 16 2.6 40 108
S.D. 0.04 0.28 1.9 1.7 0.8 1.7
MDL 0.13 0.83 5.6 5.1 2.4
CRM recoveriesb
NIST-2709 8.73 1.69 143 830 400 1400 99
NIST-2710 218 15.1 1470 12000 19000 32000 102
NIST-2711 61.6 2.40 205 3700 2100 6000 96
NIST-2781 34.2 1.84 1040 1600 420 3000 84
B31 @ 2 1100 2350 1530 35000 240000 280000 116
B31 @ 19 2190 15300 6410 14000 4100 42000 72
B26 @ 2.5 8290 7750 6400 21000 590000 640000 88
Nyanza (REFSED) 5.20 0.45 3310 1000 230 4600 95
Eu No 580 4.33 20.0 2010 49000 82000 130000 105
S.D.: standard deviation.
a 50% NIST-1646 (certified value = 63 ± 11 ng/g; typical value = 75 ± 4 ng/g) + 50% deionized water.
b Samples B31, B26 and REFSED are not true CRMs, but inter-laboratory reference soils from Hg-contaminated sites.

Register) by quantifying eight replicate extractions illustrate the typical detection limits obtained when
of a very low-level marine sediment (Table 1). The the method is applied to low-level samples.
MDLs were calculated as t × S.D. of the eight repli-
cates, where t = 2.96 for n − 1 degrees of freedom. 3.5. Pure compound speciation fingerprints
Because the speciation was not evenly distributed in
this natural sample, some of the fractions had con- Once the leaching parameters were established, a
centrations that were too high or too low to meet the series of 10 pure Hg compounds dispersed in pow-
(1–3)× of the MDL concentration criteria mandated dered kaolin were selectively extracted using this
by strict adherence to 40 CFR 136. We have found method to establish “extraction fingerprints” for each
on subsequent MDL studies that the values measured compound, and to help identify leachability classes
for all fractions vary by a factor of 3 from those re- (Fig. 4). All the compounds except the Hg–humics and
ported here, so that this data set serves adequately to Hg–Au amalgam were introduced as the powdered
N.S. Bloom et al. / Analytica Chimica Acta 479 (2003) 233–248 241

eral class, with virtually all of these species leached


in the sum of the F1 + F2 extractions. As will be seen
later, these compounds were also fully dissolved by
operationally defined bioavailability measures such as
TCLP and the in vitro human stomach simulation.
At concentrations of up to 20 ␮g/g, elemental Hg
appears in all fractions except F5 at concentrations of
5 ␮g/g, probably because it has a water solubility of
approximately 50 ␮g/l. At higher concentrations, all
of the remaining Hg0 appears in the F4 fraction. Thus,
if 50 ␮g/l Hg is present in F1–F3, a high proportion
of mercury in F4 may indicate high native Hg0 in
the sample. Other confirming measures, such as the
microscopic or macroscopic observation of balls of
liquid Hg0 , or the presence of saturated Hg0 in the
Fig. 4. Extraction fingerprint showing the percent of total mercury F1 fraction or headspace (see above) are required to
extracted in each fraction for 10 mercury compounds suspended support this conclusion.
in kaolin. The concentration of mercury (␮g/g) in each compound Mercury associated with humic organic matter and
is as follows: HgSe, 1.02; HgAu, 0.1; Hg2 Cl2 , 6.7; HgSO4 , 4.70; Hg2 Cl2 appears largely in the F3 fraction. Except in
HgS, 4.3; m-HgS, 9.6; Hg0 , 44.2; HgO, 3.6; HgCl2 , 2.6; Hg–humic,
0.05; CH3 Hg, 0.02 ng/g.
the case of purely inorganic minerals and industrial
products (calomel), Hg2 Cl2 was not expected at sig-
nificant levels in environmental samples, so we inter-
solid dispersed into kaolin powder at concentrations pret a high Hg content in the F3 fraction as an indi-
between 2600 and 44,000 ␮g/g. Because we were un- cation of organo-chelated Hg(II). Samples with high
able to successfully prepare a Hg–humic mixture that humic matter content often have a high proportion of
mimicked real-world organo-chelated Hg (i.e. rela- Hg in F3, and typically color the extractant liquid dark
tively insoluble in the F1 and F2 fractions), the figure brown. In addition to humic complexed forms, Hg as-
shows the results for Hg in a historically contami- sociated with living and dead biota are predominantly
nated natural swamp sample (Nyanza Site, Sudbury, seen in the F3 fraction as evidenced by ancillary ex-
MA, USA) with >80% organic carbon content and periments in which 80% of the Hg in biological tissue
a Hg concentration of 5 ␮g/g. The Hg–Au amalgam (dogfish liver reference material, DOLT-2 was used)
was prepared by adsorbing gas phase Hg on <200 was extracted in F3. Methyl Hg in natural sediment
mesh gold powder, and then suspending that powder samples and biota was also found in the F3 fraction,
in kaolin, with a final concentration of 10 ␮g/g Hg. but CH3 Hg usually comprises a very small proportion
While pure compounds such as these are probably not of the inorganic Hg in sediments, typically 0.01–1% of
found in the natural environment, they may model the the total. When a sample contains a significant amount
behavior of similar naturally occurring compounds. of CH3 Hg, this amount can be subtracted from F3 for
As would be expected, compounds with very low an estimate of organically bound Hg(II).
solubility, such as cinnabar and metacinnabar were not These extractions were carried out at high total Hg
significantly leached by any of the solutions, but were concentrations (1000–10,000 ␮g/g), as might be found
fully solubilized by aqua regia. Similar behavior was in mine tailings or industrially contaminated sites.
expected for the Au–Hg amalgam, although the man- Based upon the effect of Hg concentration on Hg dis-
ner of introducing Hg to the gold may have caused tribution among the fractions (Fig. 3), it seems likely
some of it to remain on the surface where it could be that the fractionation profiles would become more am-
easily leached. Further investigation of naturally oc- biguous for F1–F3 at lower total Hg concentrations.
curring amalgam particles may be necessary to resolve However, when natural low-level samples were ana-
this discrepancy. Compounds with relatively high wa- lyzed, typically 80–95% of the Hg was found in the
ter solubility (HgCl2 , HgSO4 , HgO) behaved as a gen- F3 fraction, unless it was collected downstream of a
242 N.S. Bloom et al. / Analytica Chimica Acta 479 (2003) 233–248

mercury or gold mining site, in which case most of the headspace or F1 extract, we have detected sub-
the Hg was found in the F5 fraction. This discrepancy stantial quantities (5–30%) of the total as volatile Hg.
between standards and real samples is probably an ar- Thus, the thermal volatilization method may substan-
tifact of using pure kaolin as an “inert” carrier for the tially overestimate the amount of Hg0 present in natu-
standards. ral samples. We infer from measurements of 1 M KOH
extractable Hg and/or TOC that this volatile Hg is a
3.6. Elemental and volatile mercury result of the breakdown of organo-complexed Hg(II),
although at some chlor-alkali sites, it may have re-
The accurate determination of Hg0 in solids is sulted from HgCl2 volatilization. Because of the ubiq-
problematic, and generally must be diagnosed using uity and magnitude of the error associated with this
a weight of evidence approach. The F1 (deionized method, it is not recommended for the routine quan-
water) extraction can be used to determine levels of tification of Hg0 in environmental samples.
Hg0 in the samples which are less than approximately Previous research shows that all free Hg0 present
5 ␮g/g under the conditions of this extraction, assum- in a sample will dissolve in the cold 12 M HNO3
ing that Hg0 is not adsorbed on particles or oxidized (F4) step [4]. Although this measurement is poten-
to Hg(II). All samples exhibiting saturated Hg0 in tially confounded by other compound classes such as
the F1 fraction have contained macroscopic or micro- Hg(I), and Hg bound up in amorphous organo-sulfur,
scopic liquid Hg0 . Even if values of less than 50 ␮g/l Hg–Ag amalgams, or crystalline Fe/Mn oxide phases,
Hg in F1 are interpreted ambiguously, mercury at it may be interpreted as an estimate of total Hg0 in
water saturation levels can be used to confirm the cases where Hg0 was confirmed by saturation of the F1
presence of significant Hg0 levels in soil samples. fraction or visual identification. Interestingly, in cases
Historically, Hg0 has been defined as that which where small balls of liquid Hg0 were added to natural
is volatilized upon heating the sample for 5 days at soils and sediments, we observed a rapid loss of Hg0 in
150 ◦ C [2]. To investigate the validity of this approach, the headspace. In these same samples, the bulk of the
we looked at the volatility of Hg0 and HgS (investi- Hg remained in the F4 fraction, even after periods of
gated by Revis) and several other mercury compounds up to 6 months. This suggests that in the natural envi-
(Fig. 5). As in Revis’ experiment, we found that vir- ronment microscopic balls of Hg0 are quickly coated
tually all of the Hg0 and none of the HgS (cinnabar with a gas-impervious layer, preserving the mass of
or metacinnabar) was volatilized. Unfortunately, we the droplets from further oxidation and/or diffusion.
also found high rates of volatilization for HgCl2 and
Hg bound to humic matter. In every case where we 3.7. Comparison with other extraction procedures
have applied thermal volatilization to environmental
samples that contained no detectable Hg0 in either Because selective extractions are used to make as-
sertions about bioavailability, we compared the cur-
rent procedure to two other extractions commonly
used to assess “bioaccessable” Hg. Bioaccessability
refers to the amount of the compound that can poten-
tially dissolve from the sample and become an aque-
ous exposure route to an organism. This differs from
bioavailability, the fraction of the compound of inter-
est that actually enters the organism, which can only
be determined using in vivo tests [1]. Although ex-
periments have yet to be performed with Hg, previous
work with Pb and As have shown a very good corre-
lation between bioaccessability measured in vitro and
bioavailability measured in vivo. In this study, the sol-
Fig. 5. Percent of mercury volatilized in compounds heated at uble Hg determined as the sum of the F1 and F2 extrac-
150 ◦ C for 5 days. tion fractions compares quite favorably with both the
N.S. Bloom et al. / Analytica Chimica Acta 479 (2003) 233–248 243

Table 2
Comparison of TCLP and in vitro methods to extract “bioavailable inorganic Hg” with the sum of F1 (deionized water) and F2 (acetic
acid/HCl) extracted mercury
Sample description Lab code Total Hg (␮g/g) TCLP Hg (␮g/l) Percent of total Hg leached

F1 + F2 TCLP In vitro

Marine sediment NIST-1646 0.075 <0.02 1.5 <1.3 <7.8


Moderately terrestrial soil NIST-2711 6.02 <0.02 1.1 <0.02 0.02
Smelter flyash SFA 7540 67.3 0.005 0.018 0.01
63.4% (HgCl2 + HgO), 36.3% HgS MMS 3520 113000 64.6 64.6 64.3
Gold mine tailings GMT 141 93.8 5.9 1.3 10.0
Highly contaminated soil NIST-2710 31.7 0.8 0.52 0.04 0.01

Hg extracted by the TCLP extraction and the in vitro NIST-2710 was extracted by F2 , compared to <1%
simulated human stomach extraction test (Table 2). typically found in our F2 fraction. The increased Hg
Small differences observed between the methods were found in F2 resulted in less Hg observed in the F4 and
likely a result of between-method variations in ex- F5 fractions, suggesting that a significant amount of
traction time, solids-to-liquid ratio, and temperature. Hg in this CRM may be bound up in crystalline Fe/Mn
Overall, the sum of F1 + F2 showed the greatest ex- oxides. Further investigation of the behavior of this
traction, probably due to the low solids-to-liquid ratio extractant with environmental samples is warranted.
and pH, while TCLP averaged the lowest extraction,
probably due to the methods higher pH and extraction 3.8. Relationships of speciation to methylation
ratio. The simulated stomach test operated at a more potential
aggressive pH and temperature, though this was offset
by the short extraction time of 1 h, compared to 18 h Besides employing selective extractions to assess
for F1 + F2 and TCLP. the bioaccessability of solid phase Hg to mam-
Recent reports have indicated that sodium py- malian ingestion, the method can be used to infer
rophosphate (0.5 M Na4 P2 O7 at pH 10.5) may be a
more accurate and robust measure of organo-complexed Table 3
metals than the 1 M KOH extraction at pH 14 [27]. Results of an extraction sequence using glycine/HCl instead of
To assess the differences between the two extractants acetic acid/HCL for fraction 2 and pyrophosphate in place of 1N
for Hg analysis, a comparison was performed using KOH for fraction 3 on various standards
the entire extraction scheme, but substituting the pH Sample Percent of total mercury
1.5 glycine solution for F2 and the 0.5 M Na4 P2 O7 F1 F2 F3 F4 F5
solution for F3 (Table 3). Although these two sets
of extractants gave significant differences for their NIST-2710a 0.5 0.4 2.3 47.8 49.0
NIST-2710b 0.7 0.0 4.5 37.5 57.2
particular fractions, overall the conclusions about
the biogeochemical availability of most environmen- NIST-2711a 0.0 14.1 6.9 72.4 6.6
NIST-2711b 1.0 0.0 3.4 61.2 34.3
tal samples would be similar with either extraction
scheme, when data are compared as percentages of (Hg0 )a 2.7 1.3 0.0 95.7 0.2
the total Hg in the extracted sample. (Hg0 )b 0.1 0.2 0.3 96.7 2.8
In another study, the classic extractant for total Fe+ HgCl2 a 98.3 1.6 0.0 0.1 0.0
Mn oxides (25% CH3 COOH plus 1N HN2 OH (F2 ) HgCl2 b 96.5 3.2 0.2 0.1 0.0
[12]) was substituted for the F2 extractant of the cur- Hg2 Cl2 a 7.2 12.2 43.7 36.1 0.8
rent method on a series of deep soil cores. Almost none Hg2 Cl2 b 0.8 1.5 53.3 43.8 0.5
of the Hg was extracted in the Tessier F2 fraction. In- HgSa 0.0 0.1 0.0 0.5 99.4
stead it was found in the F3 and F4 fractions, as would HgSb 0.0 0.0 0.0 0.1 99.9
be expected from organic-rich samples. Conversely, a The substituted procedure.
in that extraction 40% of the Hg in control sample b The procedure presented in this paper.
244 N.S. Bloom et al. / Analytica Chimica Acta 479 (2003) 233–248

bioavailability of Hg to sediment bound methylating Hg methylated following anoxic incubation (Fig. 6).
organisms. We tested this by mixing various Hg com- All of the incubations were spiked such that the total
pounds and natural high mercury sediments with a Hg in the samples was increased from approximately
biologically active sediment, and then measured the 0.06 to 5 ␮g/g (wet weight basis), and were incubated
anaerobically, with periodic shaking.
Samples spiked with readily bioavailable com-
pounds such as HgCl2 show rapid distribution of Hg
to a similar species distribution as the initial sediment,
as well as a high degree of methylation. Although less
soluble compounds, such as HgS, are more inert, they
appeared to slowly convert to the speciation of the
ambient sediment, and were also methylated, albeit to
a lesser degree. All of the incubated Hg-contaminated

Fig. 7. (a) Distribution of Hg among the fractions in fine grained


lacustrine sediment (mud), sandy marine sediment (marine), sandy
lacustrine sediment (sand), and highly organic swamp sediment
Fig. 6. Change in mercury fraction profile of mud spiked with (swamp) spiked with 5 ␮g/g HgS, gold mine tailings, or HgCl2
5 ␮g/g of (a) HgCl2 , (b) gold mine tailings (GMT) and (c) HgS after a 1-week incubation. (b) Net methyl mercury in the same
during a 143-day incubation. samples after a 1-week incubation.
N.S. Bloom et al. / Analytica Chimica Acta 479 (2003) 233–248 245

3.9. Speciation of real-world samples

As an example of the utility provided by the se-


quential selective extractions method described in this
paper, we present extraction profiles for 11 varied
Hg-contaminated samples taken from the natural en-
vironment (Fig. 8). Table 4 briefly describes these
samples, and includes the total Hg, methyl Hg, TCLP
leachable Hg, and headspace Hg0 for each. The first
two samples were collected from the tailings pile of a
19th century gold mine (California Sierra Mountains)
that utilized Hg extraction for gold recovery. A set of
three samples was collected at different soil depths
near the retort area of a cinnabar mine in central Cal-
Fig. 8. Extraction fingerprints of 11 naturally occurring sediments. ifornia. The third set of samples was taken from three
See Table 4 and text for sample descriptions and ancillary data. industry-contaminated soils. The first of these was
from a mercury spill onto arid soil, the second a sedi-
ment core from the flood plain of the East Fork, Poplar
samples showed intermediate patterns, although Creek (Oak Ridge, TN), which was heavily contami-
methylation increased steadily over time. nated by dissolved Hg(NO3 )2 as part of the Cold War
Fig. 7 shows the result of incubating the same hydrogen bomb production [14], and the third a highly
Hg compound with different sediment types, sandy, contaminated sample of soil collected under the cell
muddy, and swamp sediment from a freshwater lake house of an abandoned mercury cell chlor-alkali plant.
and sandy sediments from a marine estuary. For all The last three samples shown were collected at the
Hg compounds, the sandy sediments resulted in the abandoned Red Devil Mercury Mine in southwestern
lowest CH3 Hg production and the swamp and mud Alaska [28].
the highest. Even pure HgS showed measurable net These sample sets are too limited to provide detailed
methylation, suggesting that any Hg compound can geochemical interpretation, but they illustrate the wide
be methylated if discharged into the aquatic environ- range of Hg speciation found at contaminated sites,
ment. The constraint on the final CH3 Hg concentra- and the impact of aging on the speciation observed
tion may be more a function of reaction kinetics and today. For example, consider the case of two routes
the receiving sediment composition than of the Hg of off-site Hg contamination from a chlor-alkali plant
compound initially deposited there. (data not shown). Mercury was deposited both to the

Table 4
Ancillary data for 11 natural sediment profiles
Sample Description Hg0 (␮g/m3 ) Total Hg Methyl Hg TCLP Hg
(␮g/g) (␮g/g) (␮g/l)
GM-1 Gold mine tailings (deep) 72.7 41.7 nd 407
GM-2 Gold mine tailings (surface) 5.6 635 nd 28.3
CM-1 HgS mine soil, retort area (surface) 3.3 511 0.01 0.73
CM-2 HgS mine soil, retort area (10 cm) 0.1 860 nd 0.14
CM-3 HgS mine soil, retort area (210 cm) 18000 7180 0.03 2900
CS-1 Hg0 spill impacted arid soil (90 days) 1240 37.2 nd 0.39
CS-2 Oak Ridge Hg(NO3 )2 floodplain soil (50 years) nd 546 0.03 nd
CS-3 Chlor-alkali plant cell area top soil 16800 73300 0.01 18900
CS-4 Red Devil Mine sediment 1 19.7 19900 9.85 nd
CS-5 Red Devil Mine sediment 2 8880 78400 7.11 nd
CS-6 Red Devil Mine sediment 3 2.3 3990 0.65 nd
246 N.S. Bloom et al. / Analytica Chimica Acta 479 (2003) 233–248

floodplain as aqueous Hg(II), and to upland soils as tion. Samples from the floodplain of East Fork Poplar
Hg0 adsorbed to foliage, and then deposited to the Creek are similar in age (30–50 years) and deposi-
soil as litterfall. Today, ∼90% of the Hg at both sites tion environment to those from the chlor-alkali plant
is bound to soil humic matter. Yet the fraction in the site, but >70% of the Hg, originally deposited from
methylated form is approximately 1000 times higher dissolved Hg(II) is now present as cinnabar (HgS),
in the upland forest than in the floodplain site. Differ- rather than organo-chelated Hg. The East Fork Poplar
ences like this beg for further research and explana- Creek soils contain methyl Hg fractions approximately

Fig. 9. Methyl mercury (as % of total mercury) correlated with each fraction in Thai marine sediments with unique mercury fingerprints
(n = 18).
N.S. Bloom et al. / Analytica Chimica Acta 479 (2003) 233–248 247

50 times lower than corresponding samples from the ral samples and laboratory sediment incubations, we
chlor-alkali floodplain site. These differences may be have identified that the F3 (organo-chelated) fraction
due to the difference in total Hg concentration between is most strongly correlated with sediment methyla-
the sites. One explanation is that inorganic Hg added tion potential. We have also shown that when added
to sediments or soils readily converts to cinnabar at to aquatic sediments, all Hg compounds will tend to
high concentrations but at lower levels complexation convert to an ambient Hg speciation profile dictated
by organic matter interferes with the formation of a by the specific biogeochemical properties of the re-
pure macroscopic crystalline phase. ceiving sediment. The question does not appear to be
if the more recalcitrant forms, such as HgS will be
3.10. Relationship of in situ inorganic Hg speciation methylated, rather, how long it will take to reach equi-
to methylation librium. Although more research is needed to quantify
the kinetic relationships in various climates and sed-
The relationship of in situ Hg speciation to methy- iment types, this finding has significant implications
lation is difficult to accurately represent because each for the management of Hg-contaminated sediments,
sediment type has a different methylation potential. soils, and tailings.
To examine this relationship, we looked at biogeo-
chemically similar surface marine sediments from the
Gulf of Thailand, which range in total Hg and specia- Acknowledgements
tion due to historic contamination by HgCl2 and Hg0
from natural gas extraction. Data relating the percent We would like to thank Alfred Rodarme for his
of methyl Hg found in these sediments to each of the skill in preparing the figures for this text, as well as
inorganic Hg extraction fractions obtained by selec- Sharon Goldblatt and Jennifer Parker for their helpful
tive extraction are presented in Fig. 9. The fraction of criticisms and editorial comments.
methylated mercury in these sediments shows a fairly
strong positive correlation with the F3 fraction, and
a negative correlation with the F4 and F5 fractions. References
No relationship was apparent between methyl mercury
and total mercury in these sediments. Although very [1] A. Davis, N.S. Bloom, S.Q. Hee, Risk Anal. 17 (1997) 557.
[2] N.W. Revis, T.R. Osborne, G. Holdsworth, C. Hadden, Water
limited in nature, these data suggest that eventually
Air Soil Pollut. 45 (1989) 105.
the use of solid phase Hg speciation may prove to be [3] N.W. Revis, T.R. Osborne, D. Sedgely, A. King, Analyst 114
a powerful tool in predicting the relationship between (1989) 823.
Hg inputs to different ecosystems, the risk of methy- [4] E. Miller, D. Dobb, D. Cardenas, D. Heithmar, US EPA
lation, and eventual release to the food chain. EMSL-LV Internal Report, April 1994.
[5] H. Sakamoto, T. Tomiyasu, N. Yonehara, Anal. Sci. 8 (1992)
35.
[6] K.L. Willett, R.R. Turner, J.J. Beauchamp, Hazard. Waste
4. Conclusion Hazard. Mater. 9 (1992) 275.
[7] D. Wallschläger, M.V. Desai, M. Spengler, R.D. Wilken, J.
We have developed and validated a sound sequential Environ. Qual. 27 (1998) 1034.
selective extraction scheme that gives unique insights [8] H. Biester, M. Gosar, S. Covelli, Environ. Sci. Technol. 34
(1999) 3330.
into the biogeochemistry of mercury in the environ-
[9] M.O. Barnett, Doctoral Dissertation, School of Public Health,
ment. Although selective extractions cannot identify University of North Carolina, Chapel Hill, NC, 1998.
all specific compounds, they do provide much-needed [10] C.S. Kim, J.J. Rytuba, G.E. Brown, J. Synch. Rad. 6 (1999)
information on the biogeochemical behavior of var- 648.
ious classes of Hg compounds under environmental [11] R. Chester, M.J. Hughs, Chem. Geol. 2 (1967) 262.
[12] P.G.C. Campbell, A. Tessier, in: Proceedings of the Inter-
conditions. The method has detection limits in the
national Conference on Heavy Metals in the Environment,
0.1–5 ng/g range for all fractions, which, unlike other vol. 1, Geneva, 1989.
methods, allows application to pristine sediments and [13] J.M. Harrington, M.J. Laforce, W.C. Rember, S.E. Fendorf,
soils. By applying the method to contaminated natu- R.F. Rozenweig, Environ. Sci. Technol. 32 (1998) 650–656.
248 N.S. Bloom et al. / Analytica Chimica Acta 479 (2003) 233–248

[14] M.O. Barnett, L.A. Harris, R.R. Turner, T.J. Henson, R.E. [21] N.S. Bloom, W.F. Fitzgerald, Anal. Chim. Acta 208 (1988)
Melton, R.J. Stevenson, Water Air Soil Pollut. 80 (1995) 151.
1105. [22] US EPA Method 1631, Mercury in Water by Oxidation, Purge
[15] US EPA Method 1311, Toxicity Characteristic Leaching and Trap, and Cold Vapor Atomic Fluorescence Spectrometry,
Procedure, Revision 0, 1992. US EPA 821-R-95-027, 1999.
[16] M.W. Ruby, A. Davis, T.E. Link, R. Schoof, S. Eberle, C.M. [23] N.S. Bloom, J.A. Coleman, L. Barber, Fresenius Z. Anal.
Sellstone, Environ. Sci. Technol. 30 (1996) 422. Chem. 358 (1997) 371.
[17] N.S. Bloom, E.A. Crecelius, Mar. Chem. 14 (1983) 49. [24] H. Hintelmann, R. Falter, G. Ilgen, R.D. Evans, Fresenius Z.
[18] N.S. Bloom, Environ. Lab. March–April (1995) 20. Anal. Chem. 358 (1997) 363.
[19] N.S. Bloom, R. Brunette, in: Proceedings of the [25] L. Liang, M. Horvat, N.S. Bloom, Talanta 41 (1994) 371.
EPRI/DOE/EPA Combined Utility Air Pollutant Control [26] N.S. Bloom, Can. J. Fish Aquat. Sci. 46 (1989) 1131.
Symposium, Washington, DC, USA, 1999. [27] G.E.M. Hall, P. Pelchat, Water Air Soil Pollut. 99 (1999)
[20] N.S. Bloom, E.M. Prestbo, C.L. Dobbs, in: Proceedings of the 217–223.
POST-CSI XXXII Conference on Environmental and Health [28] N.S. Bloom, M. Hiltner, in: Proceedings of the Assessing
Aspects of Mining, Refining and Related Industries, Kruger and Managing Mercury from Historic and Current Mining
National Park, South Africa, 2001. Activities Conference, San Francisco, CA, November 2000.

You might also like