PAPER www.rsc.org/jem | Journal of Environmental Monitoring

Publ

ishe

d on

16

Apr

il 20

09. D

ownl

oade

d by

RM

IT U

ni o

n 20

/08/

2013

17:

57:3

1.

View Article Online / Journal Homepage / Table of Contents for this issue

Comparison of methods for measuring atmospheric deposition of arsenic,cadmium, nickel and lead

Wenche Aas,*a Laurent Y. Alleman,b Elke Bieber,c Dieter Gladtke,d Jean-Luc Houdret,b Vuokko Karlssone

and Christian Moniesf

Received 16th December 2008, Accepted 1st April 2009

First published as an Advance Article on the web 16th April 2009

DOI: 10.1039/b822330k

A comprehensive field intercomparison at four different types of European sites (two rural, one urban

and one industrial) comparing three different collectors (wet only, bulk and Bergerhoff samplers) was

conducted in the framework of the European Committee for Standardization (CEN) to create an

European standard for the deposition of the four elements As, Cd, Ni and Pb. The purpose was to

determine whether the proposed methods lead to results within the uncertainty required by the EU’s

daughter directive (70%). The main conclusion is that a different sampling strategy is needed for rural

and industrial sites. Thus, the conclusions on uncertainties and sample approach are presented

separately for the different approaches. The wet only and bulk collector (‘‘bulk bottle method’’) are

comparable at wet rural sites where the total deposition arises mainly from precipitation, the expanded

uncertainty when comparing these two types of sampler are below 45% for As, Cd and Pb, 67% for Ni.

At industrial sites and possibly very dry rural and urban sites it is necessary to use Bergerhoff samplers

or a ‘‘bulk bottle+funnel method’’. It is not possible to address the total deposition estimation with

these methods, but they will give the lowest estimate of the total deposition. The expanded uncertainties

when comparing the Bergerhoff and the bulk bottle+funnel methods are below 50% for As and Cd, and

63% for Pb. The uncertainty for Ni was not addressed since the bulk bottle+funnel method did not

include a full digestion procedure which is necessary for sites with high loads of undissolved metals. The

lowest estimate can however be calculated by comparing parallel Bergerhoff samplers where the

expanded uncertainty for Ni was 24%. The reproducibility is comparable to the between sampler/

method uncertainties. Sampling and sample preparation were proved to be the main factors in the

uncertainty budget of deposition measurements.

Introduction

Trace elements in atmospheric deposition have always been an

environmental and human health concern. Not only do they accu-

mulate in various parts of the ecosystem (soil, sediment, waters) but

they may transfer to the food chain. Therefore, measurements of

deposition and concentrations of these compounds are performed at

different types of location, and several protocols and regulations

have been established for their analysis.

Deposition measurements of trace elements in continental

regions have usually one of the two following objectives. Either

to study transboundary transport of pollutants and their effect

on ecosystems in remote areas, or in highly industrialized and

densely populated areas, to identify point sources of pollution.

aNorwegian Institute for Air Research (NILU), P.O. Box 100, N-2027Kjeller, Norway. E-mail: waa@nilu.no; Fax: +47 63898050; Tel: +476389 8000bUniversit�e Lille Nord de France, Ecole des Mines de Douai, D�epartementChimie et Environnement, BP10838, F-59508 Douai, FrancecUmweltbundesamt (UBA), D-63225 Langen, GermanydLandesamt f€ur Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen, D-45133 Essen, GermanyeFinnish Meteorological Institute (FMI), FI-00101 Helsinki, FinlandfNational Environmental Research Institute (NERI), DK-4000 Roskilde,Denmark

1276 | J. Environ. Monit., 2009, 11, 1276–1283

As a consequence, different measurement strategies and different

sampling devices are used depending on the scientific objectives.

Trace elements subjected to long range transport should be

measured in remote areas, and usually this is done by a sampling

device constructed to collect rain water only, i.e., a wet only

collector, in addition to air sampling. These measurements are

used to validate models which furthermore estimate the total

deposition load to the different ecosystems.1

Trace element emission sources are usually identified in urban

or industrialized areas. Wet- and dry deposition are generally

both of significant importance, and the deposition of very small

non-sedimenting particles can be neglected in most cases.

Cylindrical gauges are widely used samplers in such measure-

ment programs.

In Europe there are two main international commitments in

addition to several national regulations. In the EU’s 4th

Daughter Directive, it is required that the Member States shall

measure atmospheric deposition of As, Cd and Ni. The 1st as

well as the last Directive3,4 requires Pb measurements in aero-

sols only. Furthermore, the Parties to the Convention on Long-

range Transboundary Air Pollution (CLTRAP) are also

obliged to monitor trace elements (Pb, Cd, Hg, Cd, Pb, Cu, Zn,

As, Cr, Ni) in precipitations and air in accordance with the

EMEP (Cooperative Programme for Monitoring and Evalua-

tion of the Long-range Transmission of Air Pollutants in

This journal is ª The Royal Society of Chemistry 2009

Publ

ishe

d on

16

Apr

il 20

09. D

ownl

oade

d by

RM

IT U

ni o

n 20

/08/

2013

17:

57:3

1.

View Article Online

Europe) monitoring strategy.5 Measurements in urban and

industrialized areas are typically performed by local authorities,

e.g. in parts of Belgium, northern France, Austria and

Germany6 while rural measurements are carried out by cen-

tralised national laboratories.7

The EU Directive2 recommends measuring the total deposition

of trace elements using cylindrical deposit gauges. Where

appropriate, monitoring shall be coordinated with EMEP.

However, the recommendations in the Directive are partly con-

tradictive as EMEP recommends wet only collectors for wet

deposition estimates.8

The EU Directive2 gives maximum allowable uncertainties in

deposition measurements for As, Cd and Ni of less than 70%

uncertainty in the total deposition.2 The uncertainty of deposi-

tion measurements depends on the sampling method, the

methods of analysis in the laboratory, as well as on the deposi-

tion level and meteorological conditions.

To validate and define a standard method that can meet these

requirements, the European Committee for Standardization

(CEN) initiated a working group (CEN/TC264 WG20) that has

conducted an extensive field trial, comparing different sampling

methods under different meteorological conditions and exposure

levels.

The focus of the field trial was to cover all the different aspects

of deposition monitoring. The field validation took into account

every measurement detail including sampling, preparation and

analysis of the samples. Field validation tests were carried out at

four measurement sites, one industrial, one urban and two rural.

The analyses were conducted by five laboratories. The labora-

tories’ performance was tested using different types of inter-

comparison samples. For variation of the analytical method,

both GF-AAS and ICP-MS were employed. Three different

types of collectors were used; wet only, bulk and Bergerhoff type

of gauge. The sampling campaign lasted for five to six months at

each sampling site spanning different seasons. Parallel

measurements and split sample analysis were conducted to esti-

mate uncertainties. The results of the different sampling devices

were tested for comparability.

Experimental

Field sites

Four different sites were selected, one industrial, one urban, one

remote southern and one remote northern site.

The Duisburg site (Germany) (EU code: DENW037) is

located in a residential area with a high population density (51�

310N, 6� 580E). Duisburg harbour, the largest inland harbour in

Europe is located 500 m southwest of the site. The site is sur-

rounded by smaller metal processing factories and huge iron and

steel producing factories are situated in western to north-westerly

direction in a distance of 1 to 2 km. Most frequent wind direc-

tion: south-west. The field validation was carried out between

January 16th and July 4th 2006.

The site in Copenhagen (Denmark) (55� 410N, 12� 340E) is in

a botanic garden, and the influence of nearby sources is minimal.

The site can be characterized as an urban background. The field

intercomparison was carried out between September 5th 2006

and February 20th 2007.

This journal is ª The Royal Society of Chemistry 2009

The Peyrusse site (43� 370N, 00� 110E) is located near the

village of Peyrusse-Vieille (France) in a region where valley

networks are separated by forested hills with slow inclination.

Easterly winds normally prevail. The site is far away from any

urban or industrial sources of pollution and it is part of the

EMEP network (Site FR13). The field intercomparison was

carried out between March 7th and August 22nd 2006.

Birkenes (Norway) (58� 230N, 8� 150E) is a regional EMEP site

(NO01) with minimal local sources. The exposure of trace

elements is mainly from long range transport from the European

continent. The site is situated in a forested area, the terrain is

undulating and the site is located in a clearing with relatively free

exposure to exchange of air masses by wind. The field inter-

comparison was carried out between November 13th 2006 and

May 1st 2007.

Collectors

Three different collectors were used. A wet only collector is

designed to collect wet deposition only, as it is open only during

wet precipitation events. Bulk collectors are openly exposed all

the time. We used two types of bulk collectors. What we define as

the bulk sampler consists of a funnel–bottle combination. From

this setting, there are two methods for deposition estimates: the

‘‘bulk bottle method’’, which is the analysis of the liquid collected

in the bottle, and the ‘‘bulk bottle+funnel method’’, which is the

sum of the liquid collected in the bottle plus the solid residue

collected on the funnel. The other type of bulk collector, here

defined as the Bergerhoff collector is a wide mouthed jar or

bucket, mounted on a post. The collecting surfaces of all the

collectors were made of non-metal containing material, i.e. high

density polyethylene. The height of the sampling orifice was

about 1.5 m above the surface to avoid splashing and contami-

nation from the ground.

Field methods

At each site, two wet-only samplers (from Eigenbrodt), two bulk

samplers (from NILU Products) and four Bergerhoff samplers

were installed. The sampling periods were one week for wet-only

and bulk collectors and four weeks for Bergerhoff collectors, the

latter has a longer sampling period mainly due to a smaller

opening. Additionally, quality control samples such as labora-

tory blanks were analysed. At least 20% of the parallel samples

were shipped to one other laboratory for cross-analysis. Practi-

cally, one of the four Bergerhoff samplers was analysed in one

laboratory and the three others in another one. For the parallel

bulk- and wet-only samplers, one sample was sent to another

laboratory every month. In addition, some additional split

samples, meaning sub-samples from the same sampler, were

analysed at two different laboratories. The wet only- and bulk

sampling bottles, and the Bergerhoff sampling buckets, were

corked and wrapped in plastic bags and sent directly to the

laboratory without transfer to smaller transport bottles.

The precipitation amount was measured by weight. The funnels

of the wet-only and bulk collectors were washed on-site with 1%

nitric acid monthly or weekly (industrial and urban site) and the

rinsing solutions were analysed.

J. Environ. Monit., 2009, 11, 1276–1283 | 1277

Table 1 Analytical detection limits, mg (m2day)�1

Liquid sample Solid sample

ICP-MS GF-AAS ICP-MS GF-AAS

As 0.001 0.17 0.01 2.3Cd 0.0001 0.001 0.003 0.1Ni 0.003 0.01 0.3 1.9Pb 0.004 0.03 0.07 1.6

Publ

ishe

d on

16

Apr

il 20

09. D

ownl

oade

d by

RM

IT U

ni o

n 20

/08/

2013

17:

57:3

1.

View Article Online

Analytical methods

At the laboratory, the wet only and bulk samples were acidified

with concentrated nitric acid (ultrapure) and stored for at least 24

h. Visually non-homogeneous wet precipitation samples were

filtered before analysis to avoid non-reproducible results and

problems with instrumentation. The filtration of solid material

may cause an underestimation of the deposition. Some of these

filters were therefore digested and analysed to evaluate the non-

dissolved metals fraction. The Bergerhoff samples were trans-

ferred into evaporation dishes and evaporated to dryness in

a drying cabinet at 105 �C. The further digestions of the solid

samples (Bergerhoff and filter samples) were carried out using the

microwave assisted acid digestion technique according to EN

14902.9 Analyses were carried out by ICP-MS or graphite

furnace AAS according to EN 149029 including blanks and QA/

QC procedures.

Calculation of deposition

There are different ways to estimate atmospheric deposition. Wet

deposition is defined as the amount of trace elements scavenged

by precipitation. Wet-only collector is probably the best

approach to estimate the wet deposition. This should be

combined or checked with parallel measurement of precipitation

amount using a meteorological rain gauge. A fraction of the trace

elements in the precipitation might get absorbed on the funnel

walls. To assess the importance of this effect, the funnel rinsing

water was analyzed. When using a bulk collector, one will get wet

deposition plus some contribution of dry deposition (i.e. sed-

imenting particles) depending on the meteorology and the

concentration of particles during the sampling period. To esti-

mate the contribution of dry deposition on the funnel wall to the

bulk collectors, the funnels were rinsed with diluted nitric acid

(bulk bottle+funnel method). As for the wet only collector, some

of the trace elements in the bulk precipitation are absorbed on

the funnel walls, and thus add to the contribution from direct dry

deposition. The Bergerhoff collector collects both wet and some

dry deposition (i.e. sedimenting wet and dry particles). However,

one should notice that neither the bulk bottle+funnel method nor

the Bergerhoff collector collect the total deposition as these

collectors will not capture all the small (i.e. non-sedimenting)

particles. The deposition in each wet-only and bulk collector

(bulk bottle method) was calculated in accordance to eqn (1):

Da ¼CaV

tr2 p(1)

where Da is the deposition of element a given in mg (m2 day)�1, Ca

is the concentration of element a given in mg L�1, V is the sample

volume in L (or dm3), r is the radius collector surface in m, t is the

number of days the sampling period lasted.

The deposition on the funnel surface was calculated similarly

using the volume of rinsing solution (in mL). There were no

blank corrections for wet only and bulk collector since the

laboratory blanks were usually below the analytical detection

limits. The detection limits are given in Table 1. For the Ber-

gerhoff samples it was however necessary to correct for blanks

and the deposition was calculated in accordance to eqn (2):

1278 | J. Environ. Monit., 2009, 11, 1276–1283

Da ¼Ba � BbðaÞ

tr2 p(2)

where Ba is the mass of element a in mg, Bb(a) is the mass of

laboratory blanks of element a in mg, r is the radius collector

surface in m, t is the number of days the sampling period lasted.

Data below the detection limits are expressed as a half of the

detection limits in this study.

Calculation of uncertainty

Following the recommendation from ISO and the European

Commission’s working group on Guidance to Equivalence

Demonstration,10,11 the standard uncertainty is estimated using

parallel independent measurements; the general equations are

given in eqn (3) and eqn (4):

uðyÞ ¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiPNj¼1

sð jÞ 2

N

vuuut(3)

where u(y) is the standard uncertainty, N is the number of

samples, s(j) is the standard deviation in each trial j:

sð jÞ ¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiPKk¼1

�yk; j � yRð jÞ

�2

ðK � 1Þ

vuuutwhere yRð jÞ ¼

PKk¼1

yðk; jÞ

K

(4)

y(k,j) is the observed value in parallel k, trial j, K is the number of

parallels in trial j.

For K ¼ 2 the equation becomes identical to what is defined in

A6 in the EN ISO 20988 eqn (5):

uðyÞ ¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiPNJ¼1

�yð1; jÞ � yð2; jÞ

�2

2N

vuuut(5)

where u(y) is the standard uncertainty, y(1,j) is the observed value

in parallel 1, y(2,j) is the observed value in parallel 2, N is the

number of samples.

The relative standard uncertainty is calculated by dividing

the standard uncertainty by the average concentration from

the datasets. The calculations in eqn (3) are used to obtain the

expanded uncertainty eqn (6), an estimate to evaluate the

uncertainty of the measurements.

Up(y) ¼ ku(y) (6)

where Up(y) is the expanded uncertainty, u(y) is the standard

uncertainty, k is the coverage factor.

This journal is ª The Royal Society of Chemistry 2009

Publ

ishe

d on

16

Apr

il 20

09. D

ownl

oade

d by

RM

IT U

ni o

n 20

/08/

2013

17:

57:3

1.

View Article Online

The relative expanded uncertainty is then calculated by using

a coverage factor k, corresponding to a level of confidence of

approximately 95%. A coverage factor (k) of 2 is used if there are

more than 20 parallel samples; otherwise the student t-factor has

been applied.

Reproducibility as defined in ISO 5725-212 has been used to

compare two types of deposition measurements, i.e. bulk bottle

method vs wet only and Bergerhoff vs bulk bottle+funnel.

Comparability of different sampler types

Eqn (5) was also used to compare the results of different types of

samplers, if y1,j is defined as the mean value of sampler type 1 and

y2,j is defined as the corresponding mean value of sampler type 2.

Multiplication with k as shown in eqn (6) leads to the uncertainty

caused by using different sampler types.

Results and discussion

Deposition

Fig.1 shows an overview of the measured deposition with the

different collectors at all the field trial sites, differentiating

between the contribution from the precipitation bottle and the

funnel wash. The parallel samples are averaged for each collector

type. For the Bergerhoff collector the results from the three

parallel samples analysed by the same laboratory were used. It is

clearly confirmed that the deposition at the industrial site in

Fig. 1 Deposition of As, Cd, Ni and Pb at four sites using wet only (WO), bu

collectors.

This journal is ª The Royal Society of Chemistry 2009

Duisburg is much higher than the three others. The deposition at

Copenhagen urban site shows similar values to the two rural

sites. This might be the case for many European cities with no

metallurgic industries nearby.

The deposition level for the three less polluted sites was similar

to the average wet deposition at the EMEP rural sites in 2006.7

The average wet depositions from about 50 EMEP sites were

0.15, 3.0, 0.49 and 3.7 mg (m2 day)�1 for As, Cd, Ni and Pb,

respectively.7 The deposition at Duisburg is comparable to some

of the most polluted sites in the EMEP network for As, Cd and

Ni, but the Pb deposition is more than twice the level of the most

polluted site in the Czech republic.7 It is also higher than the total

deposition observed outside Europe, e.g. in the Pearl River Delta

in China13 and Tokyo Bay,14 which have observed a total Pb

deposition of 27 and 35 mg (m2 day)�1, respectively.

The funnel wash procedure contributes significantly only at

Duisburg, in both the wet only and bulk samplers. At Peyrusse,

the funnel rinsing has somewhat higher contribution than at the

two Scandinavian sites; this can be explained by a drier climate at

Peyrusse. However, at that site, the funnel was rinsed only during

the first three months of the campaign, which might not be

representative for the whole period. The sites differ also when

comparing the wet only and bulk bottle methods. The measured

deposition at Birkenes and Copenhagen is very similar for these

two collector types indicating that sites with frequent rain events

mainly experience wet deposition of trace elements. At Peyrusse

and Duisburg, the deposition in the bulk collector is enhanced,

lk or Bergerhoff (BH) collectors, including funnel wash for WO and bulk

J. Environ. Monit., 2009, 11, 1276–1283 | 1279

Publ

ishe

d on

16

Apr

il 20

09. D

ownl

oade

d by

RM

IT U

ni o

n 20

/08/

2013

17:

57:3

1.

View Article Online

especially at Duisburg where dry deposition has contributed

significantly. The correlation between wet only and bulk bottle

methods at the three less contaminated sites shows no large

systematic difference between the two different types of samplers,

Fig. 2.

The Bergerhoff collectors show sometimes higher and some-

times lower deposition than the two other collector types. This

might be due to somewhat higher uncertainty of Bergerhoff

measurements and problems with overflow of the collector, in

addition to the different digestion procedures. Nevertheless, it

seems that Bergerhoff and bulk bottle+funnel sampling methods

reproduce similar results.

Between laboratory uncertainty

To get statistically significant numbers of data points, parallel

data measurements from all the sites were pooled together, rep-

resenting in total at least 34, 30 and 20 samples for bulk, wet only

and Bergerhoff, respectively.

The uncertainty of analytical methods and sample prepara-

tion, including transport, was calculated from the parallel or

split samples analysed at two different laboratories. To avoid

taking into account the uncertainty of the precipitation volume,

the average precipitation amount was used for the wet-only and

bulk sampler. The between laboratory uncertainty is given in

Table 2.

The standard deviation between the laboratories for the wet

only and bulk measurements are generally below 15% with some

exceptions, especially in the case of Ni. Ni may present sparingly

soluble chemical forms needing complete digestion techniques to

Fig. 2 Comparing deposition of As, Cd, Ni and Pb by wet only bulk bottle m

(circle).

1280 | J. Environ. Monit., 2009, 11, 1276–1283

get it fully soluble. For the Bergerhoff collector, the uncertainty

is higher, mainly impacted by the larger uncertainty of the low

deposition samples. It should be noticed that these estimates are

based on relatively few samples and therefore, the outliers have

a strong effect. Furthermore, the Bergerhoff samples were ana-

lysed using different analytical techniques, GF-AAS and ICP-

MS. GF-AAS is less sensitive than ICP-MS (Table 1), increasing

the uncertainty at low concentrations. In addition, at Birkenes

site, there were occasionally problems with overflow and snow

clogging for the Bergerhoff collectors causing possibly non-

representative samples. At Duisburg, the between laboratory

uncertainty for Bergerhoff was around 10%. This suggests that

the main uncertainty in the Bergerhoff method is sample prep-

aration, digestion and analysis, which becomes relatively more

important at less polluted sites. The analytical or instrumental

uncertainty from the participating laboratory ring test is about

5% suggesting that sample preparation may contribute signifi-

cantly to the analytical uncertainty.

Between sampler uncertainty

The parallel samples at the four sites were combined to one

dataset to estimate the between sampler uncertainty for the

different collectors. The individual precipitation amounts are

included in the deposition estimates of bulk and wet only

collectors and are thus, part of the uncertainty budget. Only in

the cases when one collector deviated considerably, i.e. more

than 25%, the precipitation amounts of the parallel collector was

used. Table 3 shows the between sampler uncertainties obtained

by applying eqn 5.

ethods at three sites. Peyrusse (triangle), Copenhagen (star), and Birkenes

This journal is ª The Royal Society of Chemistry 2009

Table 2 Between laboratory uncertainties, mg (m2 day)�1

Bulk bottle Wet only Bergerhoff

Avg u(y) in % Nr of replicates Avg u(y) in % Nr of replicates Avg u(y) in % Nr of replicates

As 0.5 � 0.1 12% 34 As 0.46 � 0.04 9% 31 As 0.6 � 0.1 21% 21Cd 0.19 � 0.02 11% 34 Cd 0.20 � 0.03 16% 31 Cd 0.29 � 0.08 26% 20Ni 2.6 � 0.5 20% 34 Ni 2.6 � 0.6 24% 30 Ni 6.1 � 0.9 16% 21Pb 10 � 1 14% 34 Pb 11 � 1 11% 31 Pb 17 � 3 20% 21

Publ

ishe

d on

16

Apr

il 20

09. D

ownl

oade

d by

RM

IT U

ni o

n 20

/08/

2013

17:

57:3

1.

View Article Online

In general, the uncertainties are below 20%. The highest

uncertainty is for the bulk sampler and lowest for the bulk bot-

tle+funnel sampler. This may seem somewhat contradictory

since the funnel rinsing procedure is a relatively uncertain esti-

mate. However, the contribution of the funnel rinsing is low for

three of the sites thus this doesn’t contribute much to the total

uncertainty. On the other hand, the higher uncertainty of the

bulk may reflect the impact of the particles left on the funnel that

may vary from sample to sample. Another effect is the averaging

procedure since the bulk+funnel method is an average of four

weekly samples plus one funnel rinsing, and this may smooth out

the outliers and thereby decrease the uncertainty.

The wet only collectors display a lower uncertainty compared

to the bulk collectors. This is most probably due to a lower

influence from more variable dry deposition and problems with

inhomogeneous samples. It seems like the uncertainty is rela-

tively independent of element, which may confirm that uncer-

tainty is mainly associated with the sampling protocol and not

the analytical part.

Uncertainty in deposition

To estimate the overall uncertainty in the deposition, the

different type of collectors were compared for conditions when

they are assumed to measure similar fractions of the atmospheric

deposition. The wet deposition can be measured using wet-only

or bulk bottle methods. The bulk collector is however influenced

by dry deposition during dry periods; the magnitude of this

depends on the site, e.g. sites with frequent dry climatic condi-

tions or closeness to large particle source emissions such as

Table 3 Between sampler uncertainties, mg (m2 day)�1

Bulk bottle

Avg u(y) in % Nr of replicates

As 0.44 � 0.06 15% 78Cd 0.21 � 0.04 20% 78Ni 2.4 � 0.6 24% 77Pb 12 � 3 21% 78

Bergerhoff

Avg u(y) in % Nr of replicates

As 0.6 � 0.1 18% 24Cd 0.27 � 0.03 5% 24Ni 5.7 � 0.7 12% 24Pb 16 � 2 15% 24

This journal is ª The Royal Society of Chemistry 2009

industries. As shown in Fig. 1, the differences are only significant

at the Duisburg industrial site. Consequently, for wet deposition

uncertainty estimation, only the three low level sites are taken

into account. The average of the two parallel collectors is

compared with the average of the two other types of parallel

collectors.

Total deposition when defined as wet deposition plus dry

deposition of particles can not be measured precisely by using

any of these methods. An estimate of the deposition level can

nevertheless be made from the open face collectors, and the

uncertainty can be calculated by comparing the Bergerhoff

collectors and the bulk samplers plus the amount deposited on

the funnel surface (bulk bottle+funnel method). The three Ber-

gerhoff samples analysed at one laboratory were averaged and

compared to the average of the two parallel bulk samples, over

monthly periods, plus the monthly funnel rinsing. Since dry

deposition is only significantly contributing at the industrial site

in Duisburg, this is the only site used for comparing the Ber-

gerhoff and bulk bottle+funnel methods. Table 4 gives an

overview of the uncertainties.

Considering the significant differences in sampling procedures,

the relative uncertainties observed are acceptable, not exceeding

33%, except for Ni when Bergerhoff is compared with the bulk

bottle+funnel method. The reason of this larger discrepancy for

Ni is due to the different work up of the samples, bulk bot-

tle+funnel being analysed as a wet-only sample including acid

washing of the funnels while for the Bergerhoff, the sample goes

through a complete acid digestion. It is commonly known that

a fraction of Ni is bound to large dry particles as silicate or oxide,

which are chemically hardly soluble, and thus this Ni fraction can

Wet only

Avg u(y) in % Nr of replicates

As 0.47 � 0.06 12% 60Cd 0.20 � 0.03 13% 60Ni 2.3 � 0.4 15% 57Pb 11 � 2 15% 60

Bulk+funnel

Avg u(y) in % Nr of replicates

As 0.59 � 0.04 6% 21Cd 0.32 � 0.02 3% 21Ni 3.6 � 0.4 10% 21Pb 22 � 2 8% 21

J. Environ. Monit., 2009, 11, 1276–1283 | 1281

Table 4 Uncertainty in deposition, mg (m2 day)�1

Bulk bottle vs wet only Bergerhoff vs bottle+funnel

Avg u(y) in % Nr of replicates Avg u(y) in % Nr of replicates

As 0.29 � 0.05 19% 47 As 1.4 � 0.2 17% 6Cd 0.08 � 0.02 20% 47 Cd 0.85 � 0.18 21% 6Ni 1.11 � 0.37 33% 45 Ni 14 � 6 46% 6Pb 3.21 � 0.72 22% 47 Pb 60 � 15 26% 6

Publ

ishe

d on

16

Apr

il 20

09. D

ownl

oade

d by

RM

IT U

ni o

n 20

/08/

2013

17:

57:3

1.

View Article Online

not be detected without a complete digestion of the sample. It is

therefore expected that, especially for Ni, the level should be

higher in the Bergerhoff samplers (see Fig. 1).

Reproducibility

In addition to the calculations above, the reproducibility was

evaluated. The same dataset as defined in the deposition calcu-

lations (Table 4) is employed, but months with uncompleted

analysis of any of the parallel samples are not included in the

analysis.

The reproducibilities given in Table 5 are very similar or

somewhat lower than the uncertainties given in Table 3. The

between sampler uncertainties in Table 4 are in principle only

a measure of the repeatability between two different methods,

while the reproducibility is a combination of the repeatability

within each method (collector) and the variance between

methods. Since these uncertainties are similar, it can be

concluded that the main factor in the uncertainty budget is the

sampling and not the methods as such.

Missing deposition in bulk bottle+funnel during dry periods

During dry periods, some material might be deposited in the

collection bottle, which is not accounted for, either because of

too little precipitation or no precipitation at all. The trace

elements deposited in the collector can either originate from

precipitation where some of the volume might have evaporated

or from direct dry deposition. To estimate how important this

phenomenon might be, the deposition in the bottle was analysed

after dry periods in Copenhagen and Duisburg. In the entire field

experiment there were altogether three weeks with no precipita-

tion and four weeks with very little precipitation (<0.6 mm). A

small amount (20 ml) of ultrapure 1% nitric acid was added into

the collector bottle to dissolve the metals.

The deposition is higher, especially Ni and Pb, at the Duisburg

site than at the Copenhagen site. However, when comparing with

the total bulk deposition at both sites for the whole sampling

Table 5 Reproducibility in deposition, mg (m2 day)�1

Bulk bottle vs wet only

Avg SRJ in % Nr of replicates

As 0.30 � 0.05 16% 33Cd 0.08 � 0.01 17% 33Ni 1.24 � 0.32 26% 31Pb 3.51 � 0.61 17% 33

1282 | J. Environ. Monit., 2009, 11, 1276–1283

period, the unaccounted deposition caused by this effect is

insignificant, less than 0.5% for all elements. This means that the

unaccounted fraction during dry periods has not been a source of

large error in estimating the average deposition load during the

field trials.

Loss of trace elements in the filtration of the sample

Precipitation samples from the wet only and bulk collectors,

especially those collected at industrial or at dry remote sites,

may contain non-dissolved material even after acidification.

Such non-homogenous samples should be filtered before

analysis. Non-dissolved metals will not be detected if only the

filtrate is analysed. To check the evidence of non-dissolved

metals in the acidified samples, some filter residues were fully

digested and analysed for metals. Unfortunately, it was not

possible to take fully representative sub-samples of non-

homogenous precipitation samples. Thus this test only indi-

cates the existence/occurrence of non-dissolved metals, but will

not allow the quantification of the non-dissolved metal content

in the whole samples. There is clearly some evidence of non-

dissolved metals in precipitation samples (collected in the

sampling bottle) as well as in the funnel rinsing solution. The

importance of the non-dissolved metal fraction is different for

each site, the industrial site in Duisburg being the most

affected. The non-dissolved metal fraction is mainly observed

for Ni. Checking for non-dissolved metals in precipitation

samples is important when a new station is installed, especially

at industrial sites. At rural sites it may have little impact as

seen from the relatively small uncertainty between samplers at

these sites (Fig. 1).

Conclusions

It has been demonstrated that the different sampling and

analytical methods can in some climatic and geographical

conditions be comparable, while not in others. It is necessary to

use an ICP-MS at the lowest deposition levels; GF-AAS may

Bergerhoff vs bottle+funnel

Avg SRJ in % Nr of replicates

As 1.4 � 0.2 13% 6Cd 0.85 � 0.1 15% 6Ni 14 � 5 33% 6Pb 60 � 11 19% 6

This journal is ª The Royal Society of Chemistry 2009

Table 6 Expanded relative uncertainty in deposition measurements,mg (m2 day)�1

Rural wet condition IndustrialU(y) U(y)

As 37% 41%Cd 40% 50%Ni 67%Pb 45% 63%

Publ

ishe

d on

16

Apr

il 20

09. D

ownl

oade

d by

RM

IT U

ni o

n 20

/08/

2013

17:

57:3

1.

View Article Online

otherwise also be used. It is clear that a different sampling

strategy is needed for rural and industrial sites. The wet only and

bulk bottle methods give comparable results at wet rural and

urban sites where the total deposition originates mainly from

precipitation events. At industrial sites and possibly at very dry

rural and urban sites, it is necessary to use Bergerhoff or bulk

bottle+funnel methods.

Sampling and sample preparation were proved to be the main

factors in the uncertainty budget of deposition measurements.

The overall uncertainty (Table 6) is given as expanded uncer-

tainty, which is defined as the Student t-factor times the

uncertainty given in Table 4. The uncertainties are below the

required 70% given by the EU 4th daughter directive.2 For Ni at

the industrial site, the uncertainty could not be calculated due

to different digestion procedures. A complete digestion is

necessary at sites where deposition of large dry particles is

important. As a minimum value, the between sampler uncer-

tainty for the Bergerhoff, Table 3, gives an expanded uncer-

tainty of 24% for Ni.

Acknowledgements

CEN and the respective national standardisation organisations

are greatly acknowledged for supporting this work, especially

Standards Norway and Rolf Duus for the effort as secretariat.

The additional participants to the CEN working group 20 have

This journal is ª The Royal Society of Chemistry 2009

contributed with valuable discussions and comments. EMEP is

acknowledged for additional support.

References

1 EMEP, Heavy Metals. Transboundary Pollution of the Environment.Moscow, Meteorological Synthesizing Centre – East. EMEP StatusReport 2/2008.

2 EU Directive, 2004/107/EC of the European Parliament and of thecouncil of 15th December 2004 relating to arsenic, cadmium,mercury, nickel and polycyclic aromatic hydrocarbons in ambient air.

3 EU Directive, 1999/30/EC of 22 April 1999 relating to limit values forsulfur dioxide, nitrogen dioxide and oxides of nitrogen, particulatematter and lead in ambient air.

4 EU Directive, 2008/50/EC of the European Parliament and of thecouncil of 21 May 2008 on ambient air quality and cleaner air forEurope.

5 UN-ECE, EMEP Monitoring strategy and measurement programme.2004–2009, EB.AIR/GE.1/2004/5. URL: http://www.unece.org/env/lrtap/emep/Monitoring%20Strategy_full.pdf.

6 L. Th€oni, F. Krieg and U. Siewers, Atmospheric Environment, 1999,33, 337–344.

7 W. Aas and K. Brevik. Heavy metals and POP measurements, Kjeller,Norwegian Institute for Air Research, EMEP/CCC-Report 4/2008.

8 EMEP/CCC. Manual for sampling and chemical analysis. RevisedNovember 2001. Kjeller, Norwegian Institute for Air Research(EMEP/CCC-Report 1/95). URL: http://www.nilu.no/projects/ccc/manual/index.html.

9 European Committee for Standardization (CEN), Standard methodfor the measurement of Pb, Cd, As and Ni in the PM10 fraction ofsuspended particulate matter. EN14902:2005.

10 EU, Demonstration of Equivalence of ambient air monitoring methods.Report by the EC working group on Guidance for the Demonstration ofEquivalence, 2008 URL: http://ec.europa.eu/environment/air/pdf/equivalence_report2.pdf.

11 International Standards Organisation, Air Quality- Guidelines forestimation measurements uncertainty. ISO 20988:2007.

12 International Standards Organisation, Accuracy (trueness andprecision) of measurement method and results – Part 2. ISO5724–2:1994.

13 C. S. C. Wong, X. D. Li, G. Zhang, S. H. Qi and X. Z. Peng,Atmospheric Environment, 2003, 37, 767–776.

14 M. Sakata, Y. Tani and T. Takati, Atmospheric Environment., 2008,42, 5913–5922.

J. Environ. Monit., 2009, 11, 1276–1283 | 1283