Intercomparison of methods for sampling and analysis of atmospheric mercury species

Abstract

An intercomparison for sampling and analysis of atmospheric mercury species was held in Tuscany, June 1998. Methods for sampling and analysis of total gaseous mercury (TGM), reactive gaseous mercury (RGM) and total particulate mercury (TPM) were used in parallel sampling over a period of 4 days. The results show that the different methods employed for TGM compared well whereas RGM and TPM showed a somewhat higher variability. Measurement results of RGM and TPM improved over the time period indicating that activities at the sampling site during set-up and initial sampling affected the results. Especially the TPM measurement results were affected. Additional parallel sampling was performed for two of the TPM methods under more controlled conditions which yielded more comparable results.

Full text

Atmospheric Environment 35 (2001) 3007–3017
Intercomparison of methods for sampling and analysis
of atmospheric mercury species
J. Munthe
a,
*, I. W.
angberg
a
, N. Pirrone
b
,
(
A. Iverfeldt
a
, R. Ferrara
c
,
R. Ebinghaus
d
, X. Feng
e
,K.G
(
ardfeldt
e
, G. Keeler
f
, E. Lanzillotta
c
,
S.E. Lindberg
g
,J.Lu
h
, Y. Mamane
i
, E. Prestbo
j
, S. Schmolke
d
, W.H. Schroeder
h
,
J. Sommar
e
, F. Sprovieri
b
, R.K. Stevens
k
, W. Stratton
l
, G. Tuncel
m
, A. Urba
n
a
IVL Swedish Environmental Research Institute, P.O. Box 47086, S-402 58 G .
oteborg, Sweden
b
CNR-Institute for Atmospheric Pollution (CNR-IIA), c=o: UNICAL, 87036 Rende, Italy
c
CNR-Institute of Biophysics (CNR-IB), Via S. Lorenzo 26, 57127 Pisa, Italy
d
GKSS Research Centre, Institute of Physical and Chemical Analysis, Max-Planck-Street, D-21502 Geesthacht, Germany
e
Inorganic Chemistry, Department of Chemistry, G .
oteborgUniversity, S-412 96 G .
oteborg, Sweden
f
The University of Michigan Air Quality Laboratory, Ann Arbor, MI 48109, USA
g
Environmental Sciences Division, Oak Ridge Laboratory, P.O. Box 2008, Oak Ridge, TN 37831, USA
h
Environment Canada, Atmospheric Environment Service, 4905 Dufferin Street, Downsview, On Canada M3H 5T4
i
Environmental and Water Resources Engineering, Technion, Israel Institute of Technology, Haifa 32000, Israel
j
Frontier Geosciences Ltd., 212 Pontius North, Seattle, WA 98109, USA
k
FL Dept. Environmental Protection at US Environ. Protection Agency, Res. Triangle Park, NC 27711, USA
l
Department of Chemistry, Earlham College, National Road West, Richmond, IN 47374, USA
m
Middle East Technical University, Department of Environmental Engineering, Inonu Bulvari, 06531 Ancara, Turkey
n
Institute of Physics, A. Gostauto 12, Environmental Research Department, 2600 Vilnius Lithuania
Received 20 March 2000; received in revised form 25 October 2000; accepted 28 October 2000
Abstract
An intercomparison for sampling and analysis of atmospheric mercury species was held in Tuscany, June 1998.
Methods for sampling and analysis of total gaseous mercury (TGM), reactive gaseous mercury (RGM) and total
particulate mercury (TPM) were used in parallel sampling over a period of 4 days. The results show that the different
methods employed for TGM compared well whereas RGM and TPM showed a somewhat higher variability.
Measurement results of RGM and TPM improved over the time period indicating that activities at the sampling site
during set-up and initial sampling affected the results. Especially the TPM measurement results were affected.
Additional parallel sampling was performed for two of the TPM methods under more controlled conditions which
yielded more comparable results. #2001 Elsevier Science Ltd. All rights reserved.
Keywords: Mercury; Measurement; Atmosphere; Speciation
1. Introduction
Pollution of the environment by mercury remains a
major concern in many regions. Measures to control
releases to air and water have been implemented in
many countries in Europe and North America. Despite
this, mercury levels in freshwater fish remain elevated
and further cuts may be warranted. Recently, an
international agreement on emission controls of mercury
has been signed by 32 countries from Europe, the USA
and Canada, within the framework of the LRTAP
*Corresponding author.
E-mail address: john.munthe@ivl.se (J. Munthe).
1352-2310/01/$ - see front matter #2001 Elsevier Science Ltd. All rights reserved.
PII: S 1352-2310(01)00104-2

convention of the UN-ECE (UN-ECE, 1998). As the
requirements on industry and utilities to reduce mercury
emissions to the atmosphere increases, the demand for
quantitative descriptions of source–receptor relation-
ships increases.
A particular aspect of mercury is that it exists in the
environment in a number of different chemical and
physical forms with different behaviour in terms of
transport and environmental effects (Schroeder and
Munthe, 1998). Large research efforts have been put
into the identification and quantification of these species
over the last decades (Braman and Johnson, 1974;
Brosset, 1982, 1987; Brosset and Lord, 1995; Stratton
and Lindberg, 1999).
In the atmosphere, the main three forms of Hg are:
elemental Hg vapour ðHg0Þ, reactive gas phase Hg
(RGM) and particulate phase mercury (TPM). Of these
three species, only Hg0has been tentatively identified
with spectroscopic methods (Edner et al., 1989) while the
other two are operationally defined species, i.e. their
chemical and physical structure cannot be exactly
identified by experimental methods but are instead
characterised by their properties and capability to be
collected by different sampling equipment. RGM is
defined as water-soluble mercury species with sufficiently
high vapour pressure to exist in the gas phase. The
reactive term refers to the capability of stannous chloride
to reduce these species in aqueous solutions without pre-
treatment. The most likely candidate for RGM species is
HgCl2and possibly other divalent mercury species.
TPM consists of mercury bound or adsorbed to
atmospheric particulate matter. Several different com-
ponents are possible; Hg0or RGM adsorbed to the
particle surface, divalent mercury species chemically
bound to the particle or integrated into the particle itself
(Brosset, 1987). Another species of particular interest is
methylmercury (MeHg) due to the high capacity of this
species to bioaccumulate in aquatic foodchains and to
its high toxicity. The presence of MeHg in the atmo-
sphere and its importance in the overall loading of
aquatic ecosystems has been demonstrated in a number
of publications (Bloom and Watras, 1989; Hultberg
et al., 1994; Brosset and Lord, 1995). Since MeHg is only
present at low pg=m3concentration levels in ambient
air, it is not an important species for the overall
atmospheric cycling of Hg, but mainly due to its toxicity
and capacity for bioaccumulation.
Sampling and analysis of atmospheric Hg is often
made as total gaseous mercury (TGM) which is an
operationally defined fraction defined as species passing
through a 0:45 mm filter or some other simple filtration
device such as quartz wool plugs and which are collected
on gold, or other collection material. TGM is mainly
composed of elemental Hg vapour with minor fractions
of other volatile species such as HgCl2;CH3HgCl or
ðCH3Þ2Hg. At remote locations, where TPM concentra-
tions are usually low, TGM makes up the main part
ð>99%Þof the total mercury concentration in air.
The different mercury species are ubiquitous in the
atmosphere with ambient TGM concentrations aver-
aging about 1:5ng=m3in the background air through-
out the world (Iverfeldt, 1991; Slemr and Langer, 1992).
Higher concentrations are found in industrialised
regions and close to emission sources. RGM and TPM
vary substantially in concentration typically from 1 to
600 pg=m3depending on location (Keeler et al., 1995;
Stratton and Lindberg, 1995). The major sources of
RGM and TPM (as well as Hg0Þin urban locations are
fossil fuel combustion and incinerators but secondary
formation via reactions with Hg0may also be important.
RGM is more water-soluble and has a high dry
deposition rate usually assumed to be comparable to
that of nitric acid (e.g. Petersen et al., 1995). Because
RGM compounds are water-soluble, they are efficiently
removed from the atmosphere during rain events and
have an atmospheric lifetime in the order of days or a
few weeks (Schroeder and Munthe, 1998). Elemental Hg
on the other hand has a relatively long lifetime of 0.5 to
2 yr due to its low solubility in water and slow removal
rate from the atmosphere via deposition and transfor-
mation to water-soluble species (Lindqvist et al., 1991;
Slemr and Langer, 1992).
In the last few years, new automated and manual
methods have been developed to measure TGM
(Tekran, 1998; Urba et al., 1999), RGM (Xiao et al.,
1997; Stratton and Lindberg, 1995; Sommar et al., 1999;
Feng et al., 2000) and TPM (Keeler et al., 1995; Lu et al.,
1998). These developments make it possible to determine
both urban and background concentrations of RGM,
TPM and TGM. Accurate determinations of emissions
and ambient air concentrations of different Hg species
will lead to an increased understanding of the atmo-
spheric behaviour of Hg and to more precise determina-
tions of source–receptor relationships. This information
linked with other data can be used to assess the various
pathways of human exposure to mercury (EPA report to
Congress, U.S. EPA, 1996).
Denuders have been used in a variety of air pollution
studies to collect reactive gases for subsequent analysis,
such as ammonia, nitric acid and sulphur oxides (Ferm,
1979; Possanzini et al., 1983; Ferm, 1986). Denuders
were also used to remove reactive gases to prevent
sampling artefacts associated with aerosol collection
(Stevens et al., 1978). Gold-coated denuders were
developed for removal of Hg vapour from air but were
not applied to air sampling (Munthe et al., 1991).
Potassium chloride (KCl) coated tubular denuders
followed by silver coated denuders were used by Larjava
et al. (1992) to collect HgCl2(RGM) and elemental Hg
emissions from incinerators and by Xiao et al. (1997),
Sommar et al. (1999) and Feng et al. (2000) for gaseous
divalent mercury in ambient air.
J. Munthe et al. / Atmospheric Environment 35 (2001) 3007–30173008

For particulate Hg, a variety of different filter
methods have been applied such as Teflon or Quartz
Fibre filters (Keeler et al., 1995; Lu et al., 1998; Berg
et al., 2001). Before analysis, these filters undergo a wet
chemical digestion usually followed by reduction–
volatilisation of the mercury to Hg0and analysis using
cold vapour atomic absorbance spectrometry (CVAAS)
or cold vapour atomic fluorescence spectrometry
(CVAFS). Recently, a collection device based on small
quartz fibre filters mounted in a quartz tube}the AES
Mini trap}was designed. The mercury collected on
the filter can be released thermally, followed by gold
trap amalgamation and CVAFS detection (Lu et al.,
1998).
The purpose of the intercomparison exercise described
here was to test and evaluate measurement methods to
be used within the two research programmes Mercury
Over Europe (MOE) and Mediterranean Atmospheric
Mercury Cycle System (MAMCS) funded by the
European Commission DGXII. The MOE and
MAMCS projects are devoted to an increased
understanding of the atmospheric cycling of different
Hg species in Europe and include co-ordinated
measurement campaign at 10 sampling sites over
Europe.
2. Experimental
2.1. Site location
The measurement campaign was performed in Tus-
cany, Italy, approximately 100 km south of Pisa. The
site is situated in a hilly area 25 km from the coastline.
The sampling equipment was set up on the edge of a
small courtyard facing a valley with some olive tree
plantations and forests. All sampling was conducted
roughly within a 10 m distance.
2.2. Measurement techniques
A number of different sampling techniques were
employed for different mercury species. A summary of
the methodology used is given in Table 1. More detailed
descriptions of the applied methods are given below.
2.3. TGM measurements
2.3.1. Tekran gas phase mercury analysers
Two groups used Tekran Gas Phase Mercury
Analysers (Model 2537A) for measurements of TGM.
The pre-filtered sample air stream is passed through gold
cartridges where the mercury is collected. The mercury is
then thermally desorbed and detected in an integrated
CVAFS detector (Tekran, 1998). The instrument utilises
two gold cartridges in parallel, with alternating opera-
tion modes (sampling and desorbing=analysing) on a
predefined time base of 10 min. A sampling flow rate of
1:5 l min1was used. Under these conditions, a detec-
tion limit of roughly 0:15 ng=m3was achieved. A 47 mm
diameter Teflon pre-filter protects the sampling car-
tridges against contamination by particulate matter.
The accuracy and precision of this instrument has
recently been assessed in measurements intercompar-
isons performed at an urban=industrial site in Windsor,
Ontario, Canada (Schroeder et al., 1995a), a remote site
in north-central Wisconsin, USA (Schroeder et al.,
1995b), and at a remote marine background station in
Ireland (Ebinghaus et al., 1999). The instrument and the
internal permeation sources were calibrated prior to the
experimental work by manual calibration. The proce-
dure was adopted from Dumarey et al. (1985).
2.3.2. Gardis analysers
Two groups used Gardis analysers for measurements
of TGM. The Gardis instrument is based on gold
amalgamation and AAS detection (Urba et al., 1995).
Table 1
Applied methods for sampling and analysis of atmospheric mercury species
Method Mercury species Analytical method
Tekran Total gaseous mercury (TGM) CVAFS, semi-continuous
Gardis Total gaseous mercury (TGM) CVAAS, semi-continuous
Manual gold trap Total gaseous mercury (TGM) CVAAS
Charcoal adsorbents-INAA Total gaseous mercury (TGM) INAA
Teflon filters Total particulate mercury (TPM) Acid digest., SnCl2-CVAFS
Miniature quartz fibre filters Total particulate mercury (TPM) Thermal desorption, CVAFS
Cellulose acetate filters Total particulate mercury (TPM) Acid digest., SnCl2-CVAAS
Glass fibre filters Total particulate mercury (TPM) Acid digest., SnCl2-CVAFS
Mist Chamber with 0:1 M HCl Reactive gaseous mercury (RGM) SnCl2-CVAFS
Tubular KCl-coated denuders Reactive gaseous mercury (RGM) Thermal desorption, CVAFS
Annular KCl-coated denuders Reactive gaseous mercury (RGM) Thermal desorption, CVAFS=Acid
rinse-SnCl2-CVAFS
J. Munthe et al. / Atmospheric Environment 35 (2001) 3007–3017 3009

The Gardis instrument operates with ambient air as
carrier gas and does not require Argon or Helium for
detection. The sampling is run at about 1 l min1with
sampling times of 10 min. Under these conditions, a
detection limit of about 0:1ng=m3is achieved. A 25 mm
diameter PTFE membrane is used to protect the
analyser gas inlet from contamination by aerosol
particles. This instrument has been utilised in previous
intercomparison excercises (Urba et al., 1999; Ebinghaus
et al., 1999).
2.3.3. Manual gold trap method
The manual gold trap method is based on gold trap
amalgamation and subsequent analysis using CVAFS
(Brosset, 1987; Bloom and Fitzgerald, 1988). Samples
were collected on 10 cm traps consisting of a 6 mm
quartz tube with a mixture of small pieces (1–2 mm) of
gold wire and quartz glass pieces. The airflow was kept
at 0:5 l min1. All samples were analysed on-site using a
Brooks Rand CVAFS instrument.
2.4. RGM measurements
2.4.1. Mist chamber
The method of operation of the mist chamber was
adopted from Stratton and Lindberg ð1995Þ. The mist
chamber is filled with an aqueous solution containing
0:1 M HCl. The sampling flow rate is kept at 12 l min1.
In order to allow intercomparison with the tubular
denuder techniques, the sampling time was extended to
21 h in which case the acidic solution had to be
replenished to avoid complete evaporation. The mist
chamber was shielded from sunlight with aluminium
foil. Analysis of the collected RGM is made via direct
reduction using stannous chloride, amalgamation on
gold traps and CVAFS detection.
2.4.2. Tubular denuders
Tubular denuders consist of 6 mm quartz tubes
coated with KCl. The method is described in Sommar
et al. ð1998Þ. During sampling, the denuders are heated
to approximately 458C by means of a heating band to
avoid water vapour condensation. During this cam-
paign, three denuders were run in parallel. A sampling
flow rate of 1 l min1was employed. Analysis was made
using thermal desorption and CVAFS detection. The
denuders were heated to 4508C and purged with N2.
Mercury released from the denuder was collected on a
gold trap which was then analysed using CVAFS.
2.4.3. Annular denuders
Annular denuders for sampling of RGM consist of a
15 mm outer diameter quartz tube with an inner,
enclosed 8 mm tube. Air is pulled through the space
between the two tubes. Both the inner surface of the
outer tube and the outer surface of the inner tube are
coated with KCl. The set-up employed in this study was
a recently developed automated semi-continuous RGM
based on the KCl coated annular denuders coupled to
an automated TGM instrument. The RGM is quantita-
tively collected in the annular denuder at a sampling
flow rate of 5 l min1. These denuders are then heated to
5008C which converts the RGM to elemental mercury
which is detected by the Tekran CVAFS instrument.
A separate set of annular denuders was also used
without coupling to the analytical instrument. The
RGM collected on these denuders was determined by
extracting the denuders with ultrapure water (MQ)
followed by SnCl2reduction and CVAFS.
2.5. Particulate mercury
Three different methods for smapling TPM were
employed AES Mini traps, 47 mm Teflon filters, 47 mm
glass fibre filters and 47 mm cellulose acetate filters.
2.5.1. AES mini traps
The AES Mini traps consist of a small quartz fibre
filter enclosed in a 8 mm glass tube. A detailed
description of this smapler can be found in Lu et al.
(1998). The traps used in this study were modified and
instead of a quartz thread a 1:5 cm piece of silicon
tubing was used (W.
angberg et al., 2001). The sampling
time for the AES Mini traps was 21 h at a flow rate of
around 3–4 l min1. All samples were analysed on-site
using a Brooks Rand CVAFS instrument.
2.5.2. Cellulose acetate filters
The cellulose-acetate filters (47 mm diameter, 0:45 mm
pore-size), were pre-treated for 20 days in acid solution.
The sample volumes were 8.6–10:6m
3. Filters were
mineralised by means of a microwave oven (MILE-
STONE 1200) using a solution of 2 ml of nitric acid
(MERCK Selectipur) and 6 ml of hydrogen peroxide
(Carlo Erba RPE) in a Teflon vessel. Mercury was
determined by CVAAS after pre-concentration on gold
trap.
2.5.3. Glass fibre filters
Particulate samples were collected using pre-fired
glass fibre filters (Gelman Type A=E) in acid cleaned
Teflon filter packs at a nominal flow rate of 10 l min1.
Glass fibre filters which are acid extracted are pre-fired
at 5008C for a minimum of 1 h prior to use in sampling.
Filter packs, forceps, Teflon sample storage vials, petri
dishes and other field sampling equipment are rigorously
acid-cleaned in the 5 step, 11 day process described in
Landis and Keeler (1997). Extraction and analysis is
performed in a class 100 clean room. The samples are
extracted in a 10% solution of a nitric acid in Teflon
vials using a microwave digestion procedure (Keeler
et al., 1995). After extraction, the solution is oxidised
J. Munthe et al. / Atmospheric Environment 35 (2001) 3007–30173010

with BrCl for 1 h, converting all forms of Hg present
into the inorganic, þ2 oxidation state. The sample is
reduced with NH2OH and SnCl2is added to convert the
Hg2þto Hg0, which is volatile and liberated from
solution, by purging with Hg-free N2and subsequently
analysed using the dual-amalgamation CVAFS method.
The method detection limit for total particulate mercury
is about 1 pg=m3for a 24 h sample at the flow rates
employed.
2.5.4. Teflon filters
Particulate samples were collected using 47 mm
Teflon filters in acid cleaned Teflon filter packs at a
nominal flow rate of 10 l min1. The samples are
extracted in a 7 : 3 nitric acid : sulphuric acid solution
at 808C in Teflon vials. Sample analysis is made as for
glass fibre filters.
3. Results
In Table 2, the results of all applied methods for
TGM, TPM and RGM are summarised. The observed
concentration ranges for TGM are similar to what has
been reported for Northern Europe (Iverfeldt, 1991;
Ebinghaus et al., 1999; Schmolke et al., 1999) and the
Mediterranean basin (W.
angberg et al., 2001). For TPM,
the observed average concentrations in Tuscany are
higher than what is expected. This is due to a few
extreme results obtained in the beginning of the
campaign which were probably affected by local
conditions at the site. See discussion below. Observed
RGM concentrations are similar to what was found in
the following campaigns in MOE and MAMCS
(W.
angberg et al., 2001). No previously reported data
for RGM or TPM in this region have been found.
3.1. Total gaseous mercury (TGM)
TGM was measured from the start of the intercom-
parison exercise using all four methods. Only one of the
groups using a Gardis instrument reported complete
data. The Gardis instrument was run at a 10 min
sampling frequency. The Tekrans were first operated
at 5 min intervals, which was later changed to 10 min.
All reported concentrations were converted to 0:5h
averages to allow comparison. Irregular and variable
data was obtained during the start-up phase of the
measurements. Gaps in the data set also occurred when
one of the Tekran instruments (CNR-IIA) was used for
analysis of RGM samples collected on the denuders. The
use of the CNR-IIA Tekran for RGM sampling and
analysis also led to significant change in response and a
deviation of measured TGM values (see below).
In Fig. 1, the results of the measurements of total
gaseous mercury (TGM) are presented. The results
indicate that the three employed methods yield compar-
able results. Additional samples were collected on
activated carbon and analysed using INAA. The results
of these measurements yielded highly uncertain results
due to problems with high blank values in the activated
carbon batch and poor counting statistics. Daily
averages for the five measurement days were;
4:36 1:03;7:21 0:86;5detection limit; 2:27 0:3
1 and 2:50 2:22. The two latter results agree well with
the average levels presented in Fig. 1. For a more
detailed analysis of the data, linear slopes and correla-
tions were calculated for the three automated TGM
instruments employed. Only data from time periods with
near complete data sets were used in this evaluation.
In Table 3, the linear slopes and correlations are
presented. The CNR IIA instrument was only operated
according to the recommended procedures during day 1.
From day 2, this instrument was also used for
experimental work on sampling and analysis of RGM
using annular denuders. This resulted in a significant
shift in slope from 0.985 for the first day to 1.11–1.13 for
day 2–4, in comparison to the results form the GKSS
instrument. The most likely explanation is that the use
of RGM denuders caused a shift in detector response
which was not compensated for when using the
instrument for TGM measurements between the denu-
der experiments. The correlation between the two
Tekran instruments remained high during the whole
campaign. In contrast to these results, the correlation
between the GKSS Tekran (which was only used
for TGM measurements) and the CNR IB Gardis
instruments improved continuously over the 4 day
measurement period and reached acceptable values
ðS¼1:03;R2¼0:56Þduring the last day.
The manual gold trap method was run at 2 h sampling
times. In Fig. 2, the results obtained with this method
are compared with 2 h averages representing the same
time period from the three other methods (GKSS-
Tekran; CNR-IIA–Tekran and CNR-IB–Gardis).
In Fig. 2, it is clearly demonstrated that the overall
variability between the methods decreases during the
sampling campaign, despite the shift in response in the
CNR IIA Tekran. In the last four samples, all individual
Table 2
Average, median and range of observed concentrations of
TGM, TPM and RGM in Tuscany, June
TGM TPM RGM
ðng=m3Þðpg=m3Þðpg=m3Þ
Average 1.98 56 22
Median 1.93 25 22
Maximum 3.38 314 41
Minimum 1.28 13 3
J. Munthe et al. / Atmospheric Environment 35 (2001) 3007–3017 3011

results are within one standard deviation of the overall
average value.
During the fourth measurement campaign at R .
orvik,
SW Sweden, the manual method employed by IVL and a
Tekran operated by GKSS, were intercompared. The
results from this intercomparison is presented in
Fig. 3.
3.2. Reactivegaseous mercury (RGM)
The results of the measurements of RGM are
presented in Table 4. Here, a somewhat larger variability
in the results obtained with the different methods is
evident. To some extent, this is caused by using different
sampling times but mainly reflects the analytical
difficulties for this operationally defined species. Based
on these results, both denuders and mist chambers were
judged to be acceptable alternatives for the sampling of
RGM but further evaluation was needed after the
completion of the first sampling campaign. One of the
mist chamber results (number 6) was judged to be an
outlier caused by the evaporation of the acidic solution
during the prolonged sampling time. The remaining
results (except number 2 and 5 where only one method
was used) are summarised in Fig. 4. The results are
satisfactory considering the differences in sampling
principle and methods of analysis.
As for TGM, the variability in RGM results is higher
in the beginning of the campaign than in the final three
samples. This indicates that optimisations of the
sampling set-up made during the initial phase of the
campaign were beneficial for the measurements.
Fig. 1. TGM intercomparison results. The methods employed are: GKSS and CNR-IIa ¼Tekran automated analyser (Gold trap
CVAFS); IVL ¼manual gold trap collection and CVAFS detection; CNR-IB ¼GARDIS automated analyser (Gold trap CVAAS).
Table 3
Linear slopes and correlation coefficients for three automatic TGM methods
Day 1 Day 2 Day 3 Day 4 Total
GKSS-CNR-IIA S¼0:985 S¼1:12 S¼1:12 S¼1:13 S¼1:09
R2¼0:42 R2¼0:78 R2¼0:83 R2¼0:96 R2¼0:74
CNR-IIA-CNR-IB S¼0:89 S¼0:82 S¼0:86 S¼0:91 S¼0:86
R2¼0:26 R2¼0:61 R2¼0:43 R2¼0:59 R2¼0:15
GKSS-CNR-IB S¼0:88 S¼0:92 S¼0:97 S¼1:03 S¼0:96
R2¼0:05 R2¼0:01 R2¼0:23 R2¼0:56 R2¼0:29
J. Munthe et al. / Atmospheric Environment 35 (2001) 3007–30173012

3.3. Total particulate mercury (TPM)
An even larger variability is found in the results for
TPM, presented in Fig. 5. Most likely, these results were
influenced by the local conditions (dust, vehicles, etc.) at
the sampling site, especially during the first two samples.
In order to further test the comparability of the AES-
Mini traps and the Teflon filters, a second intercompar-
ison was performed in Sweden. The results of this
campaign are presented in Fig. 6.
The results from this second intercomparison for
TPM suggest that both methods give comparable results
Fig. 2. Two-hour average TGM results from four different methods. Error bars indicate 1 standard deviation for the individual
method over the time period.
Fig. 3. Intercomparison of manual and automated (Tekran) methods for total gaseous mercury (TGM). Slope and correlation is given
in the chart.
J. Munthe et al. / Atmospheric Environment 35 (2001) 3007–3017 3013

within a factor of two. However, the correlation between
the two methods is poor and no systematic bias can be
detected. Based on these results, it is not possible to
select one method as more accurate than the other.
Despite the relatively high variability observed in the
first intercomparison campaign, the AES Mini
trap method was selected for general sampling at all
MOE and MAMCS sites but Teflon filters and other
sampling methods will also be employed in parallel
during the campaigns to ensure the reliability of the
results.
4. Discussion
4.1. TGM
As has been demonstrated in earlier intercomparisons,
the Tekran, Gardis and manual gold trap methods all
yield comparable results (Ebinghaus et al., 1999).
However, a higher variability than expected between
Tekran and Gardis instruments was found during the
first days of the sampling campaign which indicates that
activities at the sampling site have influenced the results.
Table 4
Results from the RGM measurements
Date Start time
(h : min)
Stop time
(h : min)
Sample no. Tubular denuder Annular denuder
thermal
Mist
chamber
Annular denuder
rinse
27=6 17 : 27 22 : 26 1 30 41
27–28=6 12 : 00 09 : 00 2 4
28–29=6 12 : 00 09 : 00 3 20 7 24 37.5
29–30=6 11 : 00 09 : 00 4 10 33238
30=6 09 : 50 22 : 15 5 35
30=6–1=7 10 : 15 09 : 00 6 23 10 15.5 3 37
1–2=7 10 : 08 09 : 00 7 22 3 14.1 16
2–3=7 10 : 00 08 : 05 8 15 510
Fig. 4. Reactive gaseous mercury (RGM) intercomparison results.
J. Munthe et al. / Atmospheric Environment 35 (2001) 3007–30173014

Significant deviation in instrument response in one of
the Tekran instruments was caused by experimental
work with RGM denuders but is not judged to be a
serious problem when following established measure-
ment procedures (Tekran, 1998). All three TGM
methods are based on gold amalgamation and spectro-
scopic detection of Hg. The fourth method employed,
activated carbon collection}INAA detection, is poten-
tially of great value since both the pre-collection and the
detection methods are different from the others. Due
to unfortunate blank problems, comparable data were
only obtained in the last two samples, and in one
case with nearly 100% uncertainty, which demonstrates
the technical difficulties associated with this method.
Nevertheless, the activated Carbon}INAA technique
must be considered as a valuable reference method
Fig. 5. Total particulate mercury (TPM) intercomparison results.
Fig. 6. Results of intercomparison of AES mini traps and teflon filters.
J. Munthe et al. / Atmospheric Environment 35 (2001) 3007–3017 3015

especially considering the more and more common use
of gold pre-collection and CVAFS detection for Hg
measurements.
4.2. RGM
Methods for sampling and analysis of RGM have
only recently become available. Considering this, and
the fact the RGM is an operationally defined species, the
agreement between methods based on KCl-coated
denuders and the mist chamber method is quite
acceptable. Neither of the methods can be considered
as a routine monitoring technique for RGM but both
are highly valuable tools for research-based measure-
ments aimed at an increased understanding of the
atmospheric behaviour of mercury species.
The annular denuders have the advantage over the
tubular denuders that higher flow rates can be used and
thus a larger mass of RGM can be collected. In
combination with an automated Hg analyser such as
the Tekran, a vast number of research applications can
be envisaged. For example Stevens (2000) tested a
modification of the automated Tekran Inc Hg analyser.
This modification consisted of coupling a temperature
programmed quartz KCl coated annular denuder
coupled to the 2537 Tekran analyser. Air is sampled
through the annular denuder at 10 and 1:5 l min1is
directed into the Tekran Hg analyser to measure Hg0at
5 min intervals as described above. After 1–2 h sampling
is discontinued and clean air is brought through the
annular denuder while it is rapidly heated to 5008C.
During the heating cycle the HgCl2is converted to Hg0
and analysed down stream with the 2537 Tekran
analyser at 2 h intervals. This system can provide
automated measurements of both elemental Hg and
RGM at ambient concentration levels.
4.3. TPM
Measurements of TPM have been performed at a
number of occasions during the last decades. Although
the methods employed in the earlier days have been
discarded, it is clear that further development is needed
before we can completely understand the nature of
TPM. The initially poor agreement between Teflon filter
measurements and AES-mini trap measurements were
substantially improved in the second intercomparison
performed (Fig. 6) but further studies of the chemical
and physical properties of TPM are warranted before a
more complete description of this species can be made.
Acknowledgements
This work is part of the Mediterranean Atmospheric
Mercury Cycle System (MAMCS) (Contr. No. ENV4-
CT97-0593) and Mercury Over Europe (MOE) (Contr.
No. ENV4-CT97-0595) projects funded by EU-DGXII
Commission on Environment and Climate Programme
and is part of the European Land–Ocean Interaction
Studies (ELOISE) network (Ref. No. 205).
References
Berg, T., Bartnicki, J., Munthe, J., Lattila, H., Hrehoruk, J.,
Mazur, A., 2001. Atmospheric mercury species in the
European Arctic: measurements and modelling. Atmospheric
Environment 35, 2569–2582.
Bloom, N.S., Fitzgerald, W.F., 1988. Determination of volatile
mercury species at the picogram level by low-temperature gas
chromatography with cold vapor atomic fluorescence detec-
tion. Analytica Chimia Acta 209, 151–161.
Bloom, N.S., Watras, C.J., 1989. Observations of methylmer-
cury in precipitation. Science of the Total Environment
87=88, 199.
Braman, R.S., Johnson, D.L., 1974. Selective absorption tubes
and emission technique for determination of ambient forms
of mercury in air. Environment Science and Technology 8,
996–1004.
Brosset, C., 1982. Total airborne mercury and its origin. Water
Air and Soil Pollution 17, 37–50.
Brosset, C., 1987. The behaviour of mercury in the physical
environment. Water Air and Soil Pollution 34, 145–166.
Brosset, C., Lord, E., 1995. Methylmercury in ambient air.
Method of determination and some measurement results.
Water Air and Soil Pollution 82, 739–750.
Dumarey, R., Temmerman, E., Dams, R., Hoste, J., 1985. The
accuracy of the vapour-injection calibration method for the
determination of mercury by amalgamation=cold-vapour
atomic adsorption spectrometry. Analytica Chimica Acta
170, 337–340.
Ebinghaus, R., Jennings, S.G., Schroeder, W.H., Berg, T.,
Donaghy, T., Guentzel, J., Kenny, C., Kock, H.H., Kvietkus,
K., Landing, W., Munthe, J., Prestbo, E.M., Schneeberger,
D., Slemr, F., Sommar, J., Urba, A., Wallschl .
ager, D., Xiao,
Z., 1999. International field intercomparison measurements
of atmospheric mercury species at Mace Head, Ireland.
Atmospheric Environment 33, 3063–3073.
Edner, H., Faris, G.W., Sunesson, A., Svanberg, S., 1989.
Atmospheric atomic mercury monitoring using differential
absorption LIDAR technique. Applied Optics 28, 921.
Feng, X., Sommar, J., G(
ardfeldt, K., Lindqvist, O., 2000.
Improved determination of gaseous divalent mercury in
ambient air using KCl coated denuders. Fresenius Journal of
Analytical Chemistry 366, 423–428.
Ferm, M., 1979. Method for determination of atmospheric
ammonia. Atmospheric Environment 13, 1385–1393.
Ferm, M., 1986. A Na2CO3coated denuder and filter for
determination of gaseous HNO3and particulate NO
3- in the
atmosphere. Atmospheric Environment 20, 1193–1201.
Hultberg, H., Iverfeldt,
(
A., Lee, Y.-H., 1994. Methylmercury
input=output and accumulation in forested catchments and
critical loads for lakes in southwestern Sweden. In: Watras,
C.J., Huckabee, J.W. (Eds.), Mercury Pollution}Integration
J. Munthe et al. / Atmospheric Environment 35 (2001) 3007–30173016

and Synthesis. CRC Press Inc., Lewis Publishers, Boca
Raton, FL, USA.
Iverfeldt,
(
A., 1991. Occurrence and turnover of atmospheric
mercury over Nordic countries. Water, Air and Soil Pollution
56, 251–265.
Keeler, G., Glinsorn, G., Pirrone, N., 1995. Particulate mercury
in the atmosphere: its significance, transport, transformation
and sources. Water, Air and Soil Pollution 80, 159–168.
Landis, M.S., Keeler, G.J., 1997. Critical evaluation of a
modified automatic wet-only precipitation collector for
mercury and trace element determinations. Environmental
Science and Technology 29, 2123–2132.
Larjava, K., Laitinen, T., Vahlman, T., Artmann, S., Siemens,
V., Broekaert, J.A.C., Klockow, D., 1992. Measurements and
control of mercury species in flue gases from liquid waste
incineration. International Journal of Analytical Chemistry
149, 73–85.
Lindqvist, O., Johansson, K., Aastrup, M., Andersson, A.,
Bringmark, L., Hovsenius, G., Hakanson, L., Iverfeldt,
(
A.,
Meili, M., Timm, B., 1991. Mercury in the Swedish
environment}recent research on causes, consequences and
corrective methods. Water, Air and Soil Pollution 55, 23–32.
Lu, J.Y., Schroeder, W.H., Berg, T., Munthe, J., Schneeberger,
D., Schaedlich, F., 1998. A device for sampling and
determination of total particulate mercury in ambient air.
Analytical Chemistry 70, 2403–2408.
Munthe, J., Xiao, Z., Schroeder, W.H., Lindqvist, O., 1991.
Removal of gaseous mercury from air using a gold coated
denuder. Atmospheric Environment 24A, 2271–2274.
Petersen, G., Iverfeldt,
(
A., Munthe, J., 1995. Atmospheric
mercury species over central and northern Europe; model
calculations and comparison with observations from the
Nordic air precipitation network for 1987 and 1988. Atmo-
spheric Environment 29, 47–67.
Possanzini, M., Febo, A., Liberti, A., 1983. New design of a
high-performance denuder for the sampling of atmospheric
pollutants. Atmospheric Environment 17, 2605–2610.
Schmolke, S., Schroeder, W.H., Munthe, J., Kock, H.H.,
Schneeberger, D., Ebinghaus, R., 1999. Simultaneous mea-
surements of total gaseous mercury at four sites on a 800 km
transect: spartial distribution and short time variability of
total gaseous mercury over central Europe. Atmospheric
Environment 33, 1725–1733.
Schroeder, W.H., Keeler, G., Kock, H., Roussel, P., Schnee-
berger, D., Schaedlich, F., 1995a. International field inter-
comparison of atmospheric mercury measurement methods.
Water, Air and Soil Pollution 80, 611–620.
Schroeder, W.H., Lamborg, C., Schneeberger, D., Fitzgerald,
W.F., Srivastava, B., 1995b. Comparison of a manual
method and an automated analyser for determining total
gaseous mercury in ambient air. In: Wilken, R.-D., F .
orstner,
U., Kn.
ochel, A. (Eds.), Proceedings of the 10th International
Conference on Heavy Metals in the Environment. CEP
Consultants Ltd., Publisher, Edinburgh, UK, Vol. 2, pp.
57–60.
Schroeder, W.H., Munthe, J., 1998. Atmospheric mercury}an
overview. Atmospheric Environment 29, 809–822.
Slemr, F., Langer, E., 1992. Increase in global atmospheric
concentrations of mercury inferred from measurements over
the Atlantic Ocean. Nature 355, 434–437.
Sommar, J., Feng, X., G(
ardfeldt, K., Lindqvist, O., 1999.
Measurements of fractionated gaseous mercury concentra-
tions in Northwestern and Central Europe, 1995–1999.
Journal of the Environment Monitor 1, 435–439.
Stevens, R.K., 2000. Automated mercury speciation methods
and demonstrations. Proceedings from Workshop on Source
Emissions and Ambient Air Monitoring of Mercury,
September 13–14, Bloomington, MN, EPA=625=R-00=002,
pp. 82–83.
Stevens, R.K., Dzubay, T.G., Russwurm, G.M., Rickel, D.,
1978. Sampling and analysis of atmospheric sulfates and
related species. Atmospheric Environment 12, 55–68.
Stratton, W.J., Lindberg, S.E., 1995. Use of a refluxing mist
chamber for measurement of gas-phase mercury(II) species in
the atmosphere. Water, Air and Soil Pollution 80, 1269–1278.
Tekran, 1998. Model 2357A–Principles of Operation. Tekran
Inc.
U.S. EPA, 1996. Mercury Study Report to Congress: SAB
Review Draft. Report No. EPA-452R-96-001. U.S: Environ-
mental Protection Agency, Research Triangle Park, North
Carolina, USA.
UN-ECE, 1998. Protocol on heavy metals. Convention Long-
range Transboundary Air Pollution, United Nations Eco-
nomic Commission for Europe.
Urba, A., Kvietkus, K., Schomolke, S., Munthe, J., 1999.
International field intercomparison and other measurements
of total gaseous mercury at Preila, Lithuania, during 1996–
1997. In: Borrel, P.M., Borrel, P. (Eds.), Proceedings of
EUROTRAC Symposium ’98. WIT press, Southampton, pp.
364–368.
Urba, A., Kvietkus, K., Sakalys, J., Xiao, Z., Lindqvist, O.,
1995. A new sensitive and portable mercury vapor
analyzer GARDIS-1A. Water, Air and Soil Pollution. 80
1305–1309.
W.
angberg, I., Munthe, J., Pirrone, N., Iverfeldt,
(
A., Bahlman,
E., Costa, P., Ebinghaus, R., Feng, X., Ferrara, R.,
G(
ardfeldt, K., Kock, H., Lanzillotta, E., Mamane, Y.,
Mas, F., Melamed, E., Nucaro, E., Osnat, Y., Prestbo, E.,
Sommar, J., Spain, G., Sprovieri, F., Tuncel, G., 2001.
Atmospheric mercury distribution in Nothern Europe and in
the Mediterranean region. Atmospheric Environment 35,
3019–3025.
Xiao, Z., Sommar, J., Wei, S., Lindqvist, O., 1997. Sampling
and determination of divalent mercury in air using KCl
coated denuders. Fresenius Journal of Analytical Chemistry
358, 386–391.
J. Munthe et al. / Atmospheric Environment 35 (2001) 3007–3017 3017