Intercomparison of four different cascade impactors for fine and ultrafine particle sampling in two European locations

Abstract

Due to the need to better characterise the ultrafine particles fraction and related personal exposure, several impactors have been developed to enable the collection of ultrafine particles (<100 nm). However, to the authors’ kno wledge there have been no field campaigns to-date intercomparing impactor collection of ultrafine particles. The purpose of this study was two-fold: 1) to assess the performance of a number of conventional and nano-range cascade impactors with regard to the particle mass size distribution under different environmental conditions and aerosol loads and types, and 2) to characterise aerosol size distributions including ultrafine particles using impactors in 2 European locations. The impactors used were: (i) Berner low-pressure impactor (BLPI; 26 nm - 13.5 μm), (ii) nano-Berner low-pressure impactor (nano-BLPI; 11 nm - 1.95 μm) and (iii) Nano-microorifice uniform deposit impactor (nano-Moudi; 10 nm-18 μm), and (iv) Personal cascade impactor Sioutas (PCIS; <250 nm - 10 μm). Taking the BLPI as an internal reference, the best agreement regarding mass size distributions was obtained with the nano-BLPI, independently of the aerosol load and aerosol chemical composition. The nano-Moudi showed a good agreement for part icle sizes >320 nm, whereas for particle diameters <320 nm this instrument recorded larger mass concentrations in outdoor air than the internal reference. This difference could be due to particle bounce, to the dissociation of semi volatiles in the coarser stages and/or to particle shrinkage during transport through the impactor due to higher temperature inside this impactor. Further research is needed to understand this behaviour. With regard to the PCIS, their size-resolved mass concentrations were compar able with other impactors for PM1, PM2 and PM10, but the cut-off at 250 nm did not seem to be consistent with that of the internal reference.

Full text

1
Intercomparison of four different cascade impactors for
1
fine and ultrafine particle sampling in two European
2
locations
3
4
A. S. Fonseca1,2,*, N. Talbot3,4, J. Schwarz3, J. Ondráček3, V. Ždímal3, J.
5
Kozáková3,4, M. Viana1, A. Karanasiou1, X. Querol1, A. Alastuey1, T. V. Vu5, J. M.
6
Delgado-Saborit5, R. M. Harrison5, †
7
1 Institute of Environmental Assessment and Water Research (IDÆA-CSIC), Barcelona,
8
08034, Spain
9
2 Universidad de Barcelona, Facultad de Química, Barcelona, 08028, Spain
10
3 Institute of Chemical Process Fundamentals of the ASCR, v.v.i. (ICPF), Prague, 165 02,
11
Czech Republic
12
4Charles University in Prague, Faculty of Science, Institute for Environmental Studies,
13
Prague, 128 43, Czech Republic
14
5University of Birmingham, Division of Environmental Health & Risk Management,
15
Birmingham, B15 2TT,UK
16
†Also at: Department of Environmental Sciences / Center of Excellence in Environmental
17
Studies, King Abdulaziz University, PO Box 80203, Jeddah, 21589, Saudi Arabia
18
*Correspondence to: A. S. Fonseca (ana.godinho@idaea.csic.es)
19
20
Abstract
21
Due to the need to better characterise the ultrafine particles fraction and related personal
22
exposure, several impactors have been developed to enable the collection of ultrafine particles
23
(<100 nm). However, to the authors’ knowledge there have been no field campaigns to-date
24
intercomparing impactor collection of ultrafine particles. The purpose of this study was two-
25
fold: 1) to assess the performance of a number of conventional and nano-range cascade
26
impactors with regard to the particle mass size distribution under different environmental
27
conditions and aerosol loads and types, and 2) to characterise aerosol size distributions
28
including ultrafine particles using impactors in 2 European locations. The impactors used
29
were: (i) Berner low-pressure impactor (BLPI; 26 nm - 13.5 μm), (ii) nano-Berner low-
30
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

2
pressure impactor (nano-BLPI; 11 nm - 1.95 μm) and (iii) Nano-microorifice uniform deposit
1
impactor (nano-Moudi; 10 nm-18 μm), and (iv) Personal cascade impactor Sioutas (PCIS; <
2
250 nm - 10 μm).
3
Taking the BLPI as an internal reference, the best agreement regarding mass size distributions
4
was obtained with the nano-BLPI, independently of the aerosol load and aerosol chemical
5
composition. The nano-Moudi showed a good agreement for particle sizes >320 nm, whereas
6
for particle diameters <320 nm this instrument recorded larger mass concentrations in outdoor
7
air than the internal reference. This difference could be due to particle bounce, to the
8
dissociation of semi volatiles in the coarser stages and/or to particle shrinkage during
9
transport through the impactor due to higher temperature inside this impactor. Further
10
research is needed to understand this behaviour.
11
With regard to the PCIS, their size-resolved mass concentrations were comparable with other
12
impactors for PM1, PM2 and PM10, but the cut-off at 250 nm did not seem to be consistent
13
with that of the internal reference.
14
Keywords: Mass size distribution; Chemical characterization; Ultra-fine particles; Cascade
15
Impactors; Nanoparticles; Ultrafine particles
16
17
1 Introduction
18
Used in numerous areas of air quality research, cascade impactors are established, relatively
19
simple, and robust instruments. They collect airborne aerosols and segregate them into a
20
number of aerodynamic sizes for subsequent determination of mass size distribution, chemical
21
and/or physical properties (Hitzenberger et al., 2004; Seinfeld and Pandis, 2006). The
22
mechanical principle behind size impaction employs the known quantities of Stokes number
23
and slip correction factors to derive particle inertia, therefore ascribing a stopping distance in
24
accordance to particle size (Hinds, 1999). Particulates are collected onto substrates, frequently
25
made of quartz, polytetrafluoroethylene (PTFE; best known as Teflon), polyethylene
26
terephthalate (commonly abbreviated PET, otherwise known as Mylar), polycarbonate or
27
aluminium (Howell et al., 1998; Schaap et al., 2004; Tursic et al., 2008). The choice of
28
substrate is dependent on the type of impactor, sampling conditions and analytical techniques
29
intended to be carried out (Fujitani et al., 2006). A variety of cascade impactor designs have
30
appeared since May (1945) first reported on an initial design to sample coarse aerosols (>2.5
31
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

3
µm). Since then, sampling size fractions for traditionally designed commercially available
1
cascade impactors allowed for particle collection from coarse to fine fractions (<2.5 µm), for
2
example 10 µm - 0.034 µm for the Berner low-pressure impactor (BLPI) (Hering et al., 1978;
3
Berner and Luerzer, 1980; Hillamo and Kauppinen, 1991) and size cuts as small as 0.056 µm
4
for the micro-orifice uniform deposit impactor (Moudi) (Marple et al., 1991).
5
However, epidemiological studies have evidenced the need to focus on ultrafine particles
6
(UFP; Dp<100 nm), due to their possibly larger impacts on health when compared to coarser
7
particles (Oberdörster, 2000; Oberdorster et al., 2005). Recently, due to the growing need to
8
better characterise the UFP fraction, the second generation of Moudi impactors (Model 122
9
and Model 125 Nano-Moudi-II™, MSP Corp., Shoreview, MN, USA), both available in the
10
rotating version (122-R and 125-R) and in the non-rotating version (122-NR and 125-NR) and
11
nano-BLPI (not commercially available) were introduced, both adaptions of the original
12
Moudi (Marple et al., 1991) and BLPI impactors (Hering et al., 1978; Berner and Luerzer,
13
1980; Hillamo and Kauppinen, 1991), modified to enable the collection of UFP down to 11
14
nm. Also, the need to better understand and characterise personal exposure led to the
15
development of portable, light-weight impactors such as the personal cascade impactor
16
sampler (PCIS; Misra et al, 2002).
17
Due to the physical principle of particle collection associated with all impactors sampling
18
artefacts can occur, including particle bounce, particle blow off, and particle wall loss (Wall et
19
al., 1988; Schwarz et al., 2012). These artefacts vary according to the impactor type (Hillamo
20
and Kauppinen, 1991; Howell et al., 1998; Štefancová et al., 2011) loads, composition of the
21
aerosol sampled (Huang et al., 2004; Sardar et al., 2005; Fujitani et al., 2006; Crilley et al.,
22
2013), and the type of substrate used (Fujitani et al., 2006; Nie et al., 2010). Also, because
23
long sampling time is required for having enough mass of the finest UFP for chemical
24
analysis may produce sampling artefacts of volatilization or absorption.
25
As well as those previously described, the sampling and accurate sizing of UFP/nanoparticles
26
also present challenges. There is a need to produce a fast flowing jet of air onto an impactor
27
plate, creating the inertia allowing for collection of the smallest size fractions producing a
28
high pressure differential at the lowest cut sizes. This pressure drop changes the vapour
29
pressure in the bulk which can then enhance volatilisation (Hering and Cass, 1999). Attempts
30
to address this issue were successfully carried out by decreasing the pressure drop over a
31
reduced number of stages (Marple et al., 1991; Štefancová et al., 2011). Moreover, the low
32
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

4
mass of UFP requires a greater collection concentration which then increases the possibility
1
of mass overloading on the larger fractions. The commercially available Nano-Moudi-
2
II™seeks to reduce jet velocity, pressure drop, particle bounce, re-entrainment and
3
evaporative loss by incorporating micro-orifice nozzles (up to 2000 as small as 50 µm in
4
diameter in the 10 L/min Model 125 and up to 6 000 of 50 µm diameter in the 30 L min-1
5
Model 122). The rotating Nano-Moudi-II™versions (Model 122-R and 125-R) have internal
6
embedded stepper motors for the rotation of the sampling stages, thereby spreading the
7
sample over the filter to reduce build-up (Marple et al., 2014). However, as will be described
8
below, this spreading of the sample may lead to new uncertainties and complications.
9
Cascade impactors have been deployed in a diverse array of measurement campaigns utilising
10
their versatility, characterising size-fractionated chemical composition of urban aerosols
11
(Sardar et al., 2005; Schwarz et al., 2012), particle volatility (Hering and Cass, 1999; Huang
12
et al., 2004), vapour-particle phase partitioning (Delgado-Saborit et al., 2014), influence of
13
relative humidity (Štefancová et al., 2010), indoor - outdoor relationship (Smolík et al.,
14
2008), archive contamination (Mašková et al., 2015), metals in particles collected near a busy
15
road (Lin et al., 2005; Karanasiou et al., 2007; Ondráček et al., 2011), size-segregated
16
emission particles in a coal-fired power station (Tursic et al., 2008), whilst extensive
17
theoretical investigations and experimental characterization of cascade impactors tended to
18
focus on the performance of one type of cascade impactor (Biswas and Flagan, 1984; Wang
19
and John, 1988; Štefancová et al., 2011; Jiménez and Ballester, 2011; Marple et al., 2014).
20
Howell et al. (1998) carried out an intercomparison of ‘traditional’ BLPI and Moudi
21
impactors during a field campaign. Field campaigns usually provide a greater variation of
22
conditions than controlled laboratory based conditions, offering a more robust analysis of
23
comparable instrumentation. Another notable intercomparison study was conducted by
24
Pennanen et al. (2007) who tested a modified 4-stage Harvard high-volume cascade impactor
25
against a reference 10-stage BLPI in 6 different European locations over different seasons.
26
The authors note the implicit effects on individual impactors of meteorology and aerosol
27
composition. Other studies have run two or more impactors in tandem measuring
28
simultaneously indoors and outdoors (Smolík et al., 2008; Mašková et al., 2015), to cover
29
extended particle size distributions (Geller et al., 2002), or characterise artefacts caused by
30
particle volatility (Huang et al., 2004; Schaap et al., 2004) or changes in size distribution due
31
to different relative humidity (Štefancová et al., 2010).
32
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

5
To the authors’ knowledge there has been no field campaign to-date intercomparing impactor
1
collection efficiency of UFP. As a result, this paper seeks to address this by assessing the
2
performance of a number of conventional and nano-range impactors, namely Berner low-
3
pressure impactor (BLPI, 25/0.018/2, Hauke, Austria), nano-Berner low-pressure impactor
4
(nano-BLPI, 10/0,01, Hauke, Austria), nano-microorifice uniform deposit impactor (Nano-
5
Moudi-II™, MSP Corp., Shoreview, MN, USA Model 125R; U.S. Patent # 6,431,014B1) and
6
Personal cascade impactor Sioutas (SioutasTM PCIS, SKC Inc.; Misra et al, 2002), by means
7
of two intercomparison exercises, one in Prague, during winter 2015, and other in Barcelona
8
during summer 2015. The aim of the campaigns was to test the instruments’ performance
9
under different environmental conditions and aerosol loads and types. Our work reports on the
10
impactor performances not only with regard to different particle size distributions but also
11
aerosol composition and meteorology.
12
2 Methodology
13
2.1 Sampling sites and sampling set-up
14
2.1.1 Prague
15
The field intercomparison initially took place in outdoor air (6th-23rd February 2015) and it
16
was subsequently moved indoors (23rd February 2015 - 2nd March 2015) in Prague, Suchdol at
17
the Institute of Chemical Process Fundamentals (ICPF), Academy of Sciences of the Czech
18
Republic (ASCR) compound (50°7'36.47"N, 14°23'5.51"E, 277 m.a.s.l). Suchdol is a
19
residential area in north-western Prague, about 6 km from the city centre. It is recognized as a
20
suburban background site with residential houses and a university campus interspersed
21
between plenty of green spaces. The traffic flow is moderate along one major 2-lane road
22
(average traffic of 10000-15000 vehicles day-1) with regular bus services. Due to its location
23
on a plateau above the river Vltava there are not many contributory roads alongside (Figure
24
S1). Detailed information of the area where the impactors were located were previously
25
provided by Smolík et al. (2008) and Hussein et al. (2006).
26
Outdoor sampling consisted of 3 weekend sampling periods (6 - 9th, 13 - 16th and 20th - 23rd
27
February 2015), and 2 week-day samplings, (10 - 12th and 17 - 20st).
28
In addition, indoor samples were also collected during 2 week-day samplings (23rd - 25th and
29
25th - 27th February 2015) and a final 3-day weekend sampling period (27th February 2015 -
30
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

6
2nd March 2015). This resulted in a total of 5 valid outdoor samples (three weekend and two
1
week-day) and two valid indoor samples (one weekend and one week-day). For both outdoor
2
and indoor sampling, the weekend runs started on the preceding Friday between 11:00h-
3
13:00h local time and finished at 9:00h local time on the following Monday. The week-day
4
samplings started between 11h00-14h00 and terminated at 9h00. The sample duration in
5
Prague was defined based on the experience from previous research (Smolík et al., 2008;
6
Štefancová et al., 2011). Based on ambient PM concentrations it was considered that samples
7
should be collected over no more than 72 hours, to avoid substrate overload.
8
2.1.2 Barcelona
9
The Barcelona intercomparison was conducted exclusively outdoors at an air quality
10
monitoring station at IDAEA-CSIC located in an urban background site in the southwest of
11
Barcelona (41°23′14″ N, 02°06′56″E,, 78 m.a.s.l) from 18th May to 3rd July 2015 (Figure S2).
12
The sampling site, described in detail by Reche et al. (2015), is influenced by vehicular
13
emissions from one of the city’s main traffic avenues (Diagonal avenue), located at
14
approximately 200 m from the site and with a mean traffic density of 90 000 vehicles day L
15
min-1 (Amato et al., 2015). Even though the site is officially classified as urban background, it
16
is located in a city with very high road traffic and influenced by the emissions of one of the
17
largest arterial roads of the city.
18
Outdoor sampling in Barcelona consisted of 4 valid week-day sampling runs, each run
19
accounting for 96h (4 days duration sampling time). The runs started every Monday between
20
10:00h-12:00h local time and finished on Fridays around 14:00h-16:00h local time. The
21
sample duration in Barcelona was set longer than in Prague since the averages of particle
22
mass collected during a sampling less than 4 days would not be sufficient for further chemical
23
analysis. Indoor intercomparisons were not carried out due to the absence of an appropriate
24
location for indoor air sampling.
25
2.2 Instrument set-up and experimental specifications
26
In the present study, the mass size distribution of the aerosol was measured by different types
27
of cascade impactors:
28
 A Berner low-pressure impactor (BLPI, 25/0.018/2, Hauke, Austria; (Berner et al., 1979;
29
Preining and Berner, 1979) which collects particles onto PET foils (Mylar 13 μm thick)
30
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

7
(flow rate 24.8 L min-1). The impactors separated particle mass into 10 size fractions. The
1
cut diameters of the stages were 0.026, 0.056, 0.1, 0.16, 0.25, 0.43, 0.86, 1.73, 3.425, and
2
6.61 μm (Štefancová et al., 2011). The impactors were equipped with inlets with the cut-
3
point calculated as 14 μm.
4
 A modified BLPI (denominated as nano-BLPI, 10/0.01, Hauke, Austria) collecting
5
particles on PET foils (Mylar 13 μm thick) (flow rate 17.2 L min-1) from 0.01 μm to 1.95
6
μm in 8 size stages. The aerodynamic cut diameters of stages 1 to 8 were 0.011, 0.024,
7
0.039, 0.062, 0.095, 0.24, 0.49, 1.0 μm, and the inlet cut-point was calculated as 1.95 μm.
8
Given that the nano-BLPI is a custom made instrument, the design parameters of each of
9
its impaction stages are shown in Table S1 in the supporting information.
10
 A nano-microorifice uniform deposit area impactor (Nano-Moudi-II™, MSP Corp.,
11
Shoreview, MN, USA Model 125R; U.S. Patent # 6,431,014B1) equipped with PTFE
12
filters (with diameters of 47 mm) was used to collect size-resolved aerosol samples. This
13
impactor effectively separated the particulate matter into 13 stages with nominal cut
14
diameters of 0.010, 0.018, 0.03, 0.06, 0.10, 0.18, 0.32, 0.56, 1.0, 1.8, 3.2, 5.6, 10 μm and
15
the inlet cut-point as 18 μm when operated at an inlet flow rate of 10 L min-1.
16
 Three personal cascade impactor samplers (SioutasTM PCIS, SKC Inc; Misra et al, 2002)
17
operating with a flow rate of 9 L min-1 at a pressure drop of 11 inches of H2O (2.7 kPa).
18
Particles can be separated in the following aerodynamic particle diameter ranges: <0.25;
19
0.25 to 0.5; 0.5 to 1.0; 1.0 to 2.5; and >2.5 μm. The collection substrates were 37 mm
20
PTFE filters (Pall) or quartz fibre filters (Pall) for the < 0.25 μm stage and 25 mm PTFE
21
filters (Pall) for the 0.25-2.5 μm and >2.5μm stages. Two of the PCIS deployed in Prague
22
separated particle mass in all of the 5 size fractions while another unit collected particles
23
only at 3 of the stages (< 0.25 μm; 0.25-2.5 μm and >2.5 μm). In order to facilitate
24
interpretation of the data, a lower cut diameter of 30 nm was assumed for the last filter
25
stage of particles < 0.25 μm (quasi-UFP).
26
All the cascade impactors were loaded with uncoated substrates to avoid possible
27
interferences in future chemical analysis (mainly, determination of organics), so the particle
28
bounce that might occur during dry collection has to be considered excepting for the case of
29
BLPI which foils were coated with a thin layer of vacuum grease (Apiezon L, Apiezon
30
products, M&I Materials Ltd, Manchester, England) to ensure adherence of deposited
31
particles and reduce the artefact of bounce.
32
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

8
For the Prague winter intercomparison, the abovementioned six different impactors were
1
deployed simultaneously in both outdoor and indoor sampling periods. The cascade impactors
2
and their inlets were positioned outside above the roof of ICPF building, 285 m.a.s.l. The
3
nano-Moudi, in order to protect its electrical components, was kept inside an air-conditioned
4
cabin with a temperature continually lower than 20˚C and a metal pipe (about 300 cm long)
5
was extended through the roof of the building. With regard to indoor sampling, the impactors
6
were placed inside Laboratory of Aerosol Chemistry and Physics experimental hall on the 2nd
7
floor where office and other experimental activities take place. In both campaigns (indoor and
8
outdoor), the pump exhausts were extended far of the sampling spots in order to avoid
9
sampling artefacts.
10
For the Barcelona summer intercomparison, the same cascade impactors were deployed
11
(except for the PCIS) at the urban background monitoring site located in IDAEA-CSIC (78
12
m.a.s.l; South West part of the city) within the University Campus and they were positioned
13
under a plastic shelter to protect them from rain while allowing free ventilation. All the
14
impactor pumps were placed 5 m distance from the impactors whilst long tubes (10 m) were
15
connected to the exhausts to avoid contamination of the samples.
16
The error in the sampling flow rate and sampled volume in both campaigns was < 5%. Thus,
17
it is assumed that flow rates did not affect the particle size cut-offs. The uncertainty in the
18
particle mass concentration determination was < 15% except in some cases for the smallest
19
stages of nano-BLPI and nano-Moudi impactor which reached mass value deviations > 20 %
20
(standard deviation).
21
The specifications of the campaigns and the impactors deployed in the intercomparison study
22
are summarized in Table 1. The BLPI was used as internal reference for the size distribution
23
in this study as it was calibrated with the method described by Hillamo and Kauppinen (1991)
24
for the fine stages and by Štefancová et al. (2011) for coarse stages. For the intercomparison,
25
the modal pattern of aerosol mass size distribution was divided into four size groups: (i) PM10
26
(Dp<10 μm), (ii) PM2 (Dp<2 μm), (iii) PM1 (Dp<1 μm) and (iv) PM0.25 (Dp<0.25 μm) particles.
27
Approximate lower cut points for those selected size fractions are shown in Table S2 in the
28
supplementary information.
29
30
31
32
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

9
Table 1. Impactors deployed in Prague and Barcelona and their specifications.
1
Impactor
type
BLPI
nano-BLPI
nano-Moudi
PCIS (5 stages) c
PCIS (3 stages) d
Number of
samplings in
Prague
5x outdoor (3x
weekend-days +
2x week-days)
2 x indoor
(1xweekend-
days + 1x week-
days)
5x outdoor (3x
weekend-days +
2x week-days)
2 x indoor
(1xweekend-
days + 1x week-
days)
5x outdoor (3x
weekend-days +
2x week-days)
2 x indoor
(1xweekend-
days + 1x week-
days)
5x outdoor (3x
weekend-days + 2x
week-days)
2 x indoor
(1xweekend-days + 1x
week-days)
5x outdoor (3x
weekend-days + 2x
week-days)
2 x indoor (1xweekend-
days + 1x week-days)
Number of
samplings in
in Barcelona
4 x outdoor (4 x
week-days)
4 x outdoor (4 x
week-days)
4 x outdoor (4 x
week-days)
N/A
N/A
Flow rate
(L min-1)a
24.8
17.2
10
9
9
Sampling
substrates
PET foils
(MYLAR) 13
μm thick
PET foils
(MYLAR) 13
μm thick
PTFE 47 mm
37 mm PTFE filters
(Pall) < 0.25 μm stage
and 25 mm PTFE
filters (Pall) for the
0.25-2.5 μm and 2.5-10
μm stages
37 mm quartz-fibre
filters (Pall) < 0.25 μm
stage and 25 mm PTFE
filters (Pall) for the
0.25-2.5 μm and >2.5
μm stages
Nº Stages
10
8
13
5
3
Lower cut
sizes (μm) b
0.026
0.011
0.01
0.03
0.03
0.056
0.024
0.018
0.25
0.25
0.10
0.039
0.032
0.50
2.50
0.16
0.062
0.056
1.00
0.25
0.095
0.10
2.50
0.43
0.24
0.18
0.86
0.49
0.32
1.73
1.0
0.56
3.42
1.00
6.61
1.80
3.20
5.60
10
Inlet cut-
point (μm)
14
1.95
18
10
>2.5
a Volumetric flow rate at 20°C and ambient pressure
2
b All sizes are aerodynamic equivalent diameters
3
c Two units deployed; A cyclone was installed ahead which cut PM10
4
d One single unit deployed
5
N/A – Not available
6
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

10
2.3 Sample conservation and gravimetric analysis
1
Particle mass concentrations on impactor substrates were gravimetrically determined by pre-
2
and post-weighing the Mylar foils and filters (PTFE and quartz fiber) with a Sartorius M5P-
3
000V001 electronic microbalance in Prague and a Mettler MT5 electronic microbalance in
4
Barcelona, both with a ±1 μg sensitivity. Blank samples (1 per sample) were collected per
5
each impactor type in both intercomparison (Prague and Barcelona) for each of the sampling
6
periods. The deviation of mass values due to varying conditions was corrected with the help
7
of the corresponding blanks.
8
All samples were equilibrated for a period of 24 hours before weighing in a temperature and
9
relative humidity controlled room. The electrostatic charges of the filters were removed using
10
an U-shaped electrostatic neutralizer (Haug, type PRX U) in Prague and a zerostat anti-static
11
instrument (Z108812-1EA, Sigma-Aldrich Co. LLC.) in Barcelona. Each sample was
12
weighed three times with an accuracy of mass determination of ± 2 μg. After weighing, the
13
sampled foils and filters were stored in the freezer at -18 °C.
14
2.4 Ion chromatography analysis
15
Ion chromatography analysis were only carried out for the Prague samples and for the BLPI,
16
nano-BLPI and nano-Moudi impactors with the aim to support the interpretation of the
17
particle mass size distributions data. The PCIS filters were not analysed due to the differences
18
observed for the finest size fraction with the other impactors, as will be discussed below.
19
The whole nano-Moudi impactor samples were extracted in 7 ml of ultrapure water. In case of
20
the Berner impactors, approximately 1/3 of each foil with samples from each stage was cut
21
out and number of aerosol spots on cut piece was calculated. The ratio between cut and total
22
number of spots at each impactor stage was used to recalculate results to overall ion amount
23
on each stage. All samples were then extracted with 7 ml of ultrapure water, sonicated for 30
24
min in ultrasonic bath and shaken for 1 hour using a shaker. The extracts were then analyzed
25
using a Dionex 5000 system both for cations (Na+, NH4+, K+, Ca2+ and Mg2+) and anions
26
(SO42−, NO3−, Cl−) in parallel. An IonPac AS11-HC 2 x 250 mm column was used for anions
27
using hydroxide eluent, IonPac CS18 2 x 250 mm for cations using methane sulfonic acid
28
solution as an eluent. Both anion and cation set-up were equipped with electrochemical
29
suppressors. External calibration was done using NIST traceable calibration solutions.
30
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

11
3 Results
1
3.1 Meteorological data and mean aerosol concentrations in outdoor air
2
Table 2 displays the meteorological data (ambient temperature, relative humidity, ambient
3
pressure and wind speed), the mean and standard deviations (±σ) of aerosol concentrations for
4
Prague and Barcelona and season during sampling with BLPI.
5
Table 2. Meteorological data and mean daily aerosol concentrations in outdoor air in Prague
6
from 6th to 23rd February 2015 and in in Barcelona from 18th May to 3rd July 2015.
7
Sampling
site
Temperature
(ºC)
Relative
humidity
(RH, %)
Barometric pressure
recalculated to sea
level (mbar)
Wind
Speed
(km h-1)
Mean
PM14*
(µg m-3)
Min
Max
Min
Max
Prague
(winter)
-3.4±2.6
3.9±3.3
51±15.4
92±2.1
1023±9.4
12.5±6.6
34.6 ± 15.8
Barcelona
(summer)
18±3.3
26±3.3
39±9.9
85±7.1
1018±3.1
12±2.6
15.2 ± 2.1
During the winter campaign in outdoor air from 6th to 23rd February 2015 in Prague, the daily
8
maximum average temperature was 3.9±3.3 ºC and the minimum average temperature was -
9
3.4±2.6 ºC. The relative humidity varied in the range of 51-92% from day to day.
10
As expected, higher temperatures during summer were monitored in Barcelona from 18th May
11
to 3rd July 2015 (minimum of 18±3 ºC and maximum of 26±3 ºC). However, slightly lower
12
RH (minimum of 39±10 % and maximum of 85±7%), similar pressure (1018±3 mbar) and
13
wind speed (12±3 km h-1) values were recorded. The results imply that aqueous particles may
14
have been collected on an impaction stage different from the stage where they ought to be
15
collected due to the flow-induced relative humidity changes during the day (Fang et al., 1991;
16
Štefancová et al., 2010). Aqueous particles can shrink due to evaporation caused by pressure
17
drop through the impactor and/or grow due to condensation caused by aerodynamic cooling.
18
Also, a distortion of the size distribution due to bounce-off should not be neglected for
19
Barcelona given that foils were not greased prior to sampling.
20
In Prague, the mean PM14 mass concentration measured outdoors (with BLPI) was 34.6 ±
21
15.8 µg m-3 whilst in Barcelona (with BLPI) it was 15.2 ± 2.1 µg m-3 (Table 2), in agreement
22
with previous results from 2008 winter campaign in ICPF (Schwarz et al. 2012; PM14=34 µg
23
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

12
m-3) and of the same order of magnitude as PM10 from a 2014 summer campaign in the
1
monitoring station at IDAEA-CSIC (PM10=19.6 µg m-3). The reason of higher averages of
2
particle mass concentrations in winter in Prague than in summer in Barcelona are due to
3
higher emissions (mainly due to coal and biomass burning used for residential heating) and
4
meteorological conditions such as the lower mixing heights of the boundary layer or even
5
temperature inversions occurring in Prague (Schwarz et al., 2012).
6
3.2 Average particle mass concentrations per stage for the different
7
impactors
8
To estimate the cumulative mass concentration for the different size ranges in each of the
9
impactors, the integrated curve of the measured particle mass size distributions was
10
determined by Eq. 1:
11
  


   Eq. (1)
12
Where, M is the estimated mass concentration for each impactor stage i, Dpi-1 and Dpi are
13
respectively the lower and upper cut-off diameters of the impactor stage i
14
The cumulative curves of the particle mass size distributions from Prague (indoor and
15
outdoor) and Barcelona are shown in Figures 1 and 2, respectively.
16
Results show that the nano-BLPI behaved similarly to the internal reference considered for
17
this work (BLPI), especially for particles larger than 250 nm. Outdoors and indoors, the nano-
18
Moudi was in agreement with the BLPI for particles larger than 320 nm (independent of the
19
aerosol load and type). However, for particles below 320 nm, the particle mass concentration
20
of the nano-Moudi tended to be higher than for the BLPI, especially during winter in Prague.
21
In indoor air, the nano-Moudi cumulative curve of the mass size distributions was closer to
22
the curve obtained for the BLPI impactor.
23
While in Prague, the nano-Moudi mass size distributions for particles >1 μm were lower than
24
the rest of the impactors, in Barcelona, this trend was not so evident (Figure 1 and 2). This
25
different behaviour could be ascribed to a number of causes: (a) in outdoor air the effect of
26
particle bounce and/or the shrinkage of semi volatile compounds may have caused a shift in
27
particle mass towards the lower sizes of the nano-Moudi, especially in winter in Prague;
28
and/or (b) indoors, the mechanism of the nano-Moudi of spreading the sample (rotating
29
plates), with the increase in temperature, both in indoor air and inside the nano-Moudi shell,
30
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

13
could favour particle dissociation/evaporation from the PTFE filters and thus result in lower
1
mass loads across the lower size ranges, and thus the nano-Moudi curve would appear to be
2
closer to the internal reference BLPI. This effect would not be so prominent in outdoor air,
3
given that the instrument does not reach such high temperatures. Nie et al. (2010) also
4
attributed the loss of volatile compounds to the increase of the temperature inside the
5
MOUDI. However, nitrate concentrations were low in indoor air (see sections below), and
6
therefore the volatilization of this species would have had a low impact on particle mass
7
(leaving only the organic fraction to account for this). Further research is necessary to clarify
8
the different behaviours observed.
9
The size-fractionated average mass concentrations (PM0.25, PM1, PM2 and PM10) collected by
10
each impactor along with standard errors deviation (±σ) in the respective size fractions, using
11
data from a total of 5 experiments outdoors and 2 indoors in Prague, and a total of 4 valid
12
samples outdoors in Barcelona are summarised in Figure 3. Approximate cut points for the
13
selected size fractions are shown in Table S2 in the supporting information. However, it is
14
important to take into account that some differences in the results could be partially attributed
15
to the differences in the real cut points for the selected size fractions.
16
The average PM14 mass concentrations and corresponding standard deviation obtained using
17
the internal reference (BLPI) in Prague outdoors were 34.6 ± 15.8 µg m-3. In Barcelona, the
18
PM14 mass concentrations and standard deviation in summer were 15.2 ± 2.1 µg m-3.
19
Comparison of independent data from Grimm (corrected with high volume sampling) and the
20
impactors with PM1 and PM10 size cuts, was carried out for the outdoor campaign in
21
Barcelona (4 samples). A slope of 0.98 and a R2 of 0.7 was obtained for the PM14 for BLPI
22
with PM10 from an online laser spectrometer (corrected with regard to reference
23
instrumentation) whereas for PM1, a slope of 0.7 and a better fit of the data was obtained
24
(R2=0.9). Similarly to BLPI, the nano-BLPI shows a slope of 0.7 and a R2 of 1 for the cut
25
point PM1. The mass differences detected for PM1 suggest that impactors sampling artefacts
26
such as particle blow off, particle wall losses and/or particle bounce occurred.
27
28
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

14
1
Figure 1. Cumulative mass concentrations measured by the six impactors in Prague: (a)
2
outdoors and (b) indoors. Error bars indicate the standard deviation (±σ).
3
4
Figure 2. Cumulative mass concentrations measured by the three impactors in Barcelona,
5
outdoors. Error bars indicate the standard deviation (±σ).
6
As shown in Figure 3, the largest relative difference between the average mass concentrations
7
collected with the three impactors (PCIS, nano-BLPI and nano-Moudi) and the internal
8
reference (BLPI) was calculated for the PM0.25 size fraction measured outdoors in Prague by
9
PCIS and nano-Moudi, when concentrations were larger by 354 and 126 %, respectively. The
10
best agreement between the three impactors and the internal reference was obtained in the
11
Barcelona summer campaign.
12
Intercomparisons between the nano-BLPI impactor and the reference BLPI indicate an overall
13
good agreement with absolute differences in mass concentrations per size fraction being
14
<30%, independent of the aerosol type. A consistent underestimation of the particle mass
15
concentrations for the PM0.25 size fractions was obtained with the nano-BLPI for all
16
campaigns and locations (Figure 3). This consistent underestimation was in the order of 5 and
17
22% outdoors in Barcelona and Prague, respectively, and 10% indoors in Prague, for PM0.25.
18
As for PM1, a slight overestimation of mass concentrations with regard to the BLPI was
19
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

15
obtained by the nano-BLPI in both sampling campaigns outdoors. The largest deviation in this
1
size fraction was obtained in Prague outdoors (15%) whereas the smallest difference was
2
obtained in Barcelona (5%). Similar to the PM0.25 fraction, the PM1 and PM2 concentrations
3
obtained indoors by the nano-BLPI were lower (12 and 15%, respectively) than those of the
4
BLPI.
5
As for the nano-Moudi, it consistently measured lower PM1 and PM2 concentrations in all
6
campaigns (max difference obtained indoors for PM1 = 31% and PM2 = 30 %). These
7
differences can be explained by the difference in the cut points given that PM1 and PM2
8
fractions from the BLPI are actually 0.86 μm and 1.7 μm, respectively. For quasi-UFP mass
9
concentrations were significantly higher (126%) in Prague outdoors, whereas the
10
disagreement with the BLPI was reduced in Barcelona outdoors (14%). Finally, in indoor air,
11
concentrations registered by the nano-Moudi were lower (30%) than the BLPI, in Prague.
12
13
Figure 3. Average mass concentrations collected by the impactors in approximate size
14
ranges: PM0.25, PM1, PM2 and PM10, during winter Prague campaign (top) and summer
15
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

16
Barcelona Campaign (bottom). Error bars indicate the standard deviation (±σ) of the
1
measured concentrations. The PCIS data correspond to PCIS 3 from Figure 1, where more
2
particle size fractions were collected.
3
Finally, the portable PCIS were only used in Prague during winter given the differences
4
obtained with regard to the BLPI for the quasi-ultrafine size mode PM0.25 (354%). A similar
5
pattern was observed for indoor air, although with a relatively smaller, but still high
6
difference (75%). A possible reason for the discrepancies observed regarding the PM0.25
7
fraction could be ascribed to the different pressure drops across the impactor stages
8
originating from different flow rates (e.g., PCIS 9 L min-1 vs. BLPI 24.8 L min-1). The higher
9
pressure drop in the stationary impactors (e.g., BLPI) may increase the probability of
10
volatilisation of semi-volatile species during prolonged sampling, and could contribute to an
11
underestimation of the PM0.25 when compared to the PCIS (Sioutas, 2004).
12
The differences with regard to the coarse fractions were much smaller when compared to the
13
quasi-UFP fractions (<[±42%] and <[±27%] in outdoors and indoors, respectively). In
14
outdoor air, the PCIS showed consistently higher PM1, PM2 and PM10 concentrations (42, 14
15
and 4%, respectively). Similar results were reported by Sioutas (2004) where an average ratio
16
PCIS to Moudi (Model 110, MSP Corp, Minneapolis, MN) of 2.02 (± 0.59) and 1.21 (± 0.35)
17
was reported for an aerodynamic size range < 0.25 μm and 2.5-1 μm, respectively. However,
18
in indoor air a consistently underestimation (12, 16 and 21 % for PM1, PM2 and PM10), was
19
observed.
20
In summary, for the aerosols and sampling conditions in this work, the PCIS provided
21
comparable size-resolved mass concentrations for PM1, PM2 and PM10 while the cut-off at
22
250 nm did not seem to be consistent with the internal reference BLPI. In order to fully
23
understand these phenomena, a more systematic evaluation might be required. For this reason,
24
data from PCIS will not be discussed in the following sections.
25
3.3 Aerosol mass size distributions
26
3.3.1 Particle size distribution in outdoor air
27
The average particle mass size distributions obtained in the outdoor intercomparison study
28
(Prague and Barcelona) can be found in Figure 4.
29
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

17
1
Figure 4. Average mass size distributions obtained outdoors: (a) winter in Prague and (b)
2
summer in Barcelona.
3
As can be seen, the particle mass size distributions are very different depending on the season
4
and sampling location. During winter in Prague (outdoors), the mass size distributions have a
5
predominantly fine mode, with the coarse mode being almost negligible (by all impactors).
6
The maximum mass concentration obtained in the fine size fraction mode was between 0.4-
7
0.9 μm, whereas in summer in Barcelona, this maximum was shifted towards smaller size
8
fractions between 0.2 and 0.4 μm. In addition to the different aerosol types, this shift to lower
9
sizes might be caused by a lower average relative humidity during sampling in Barcelona that
10
could have caused the particle drying (Tables 2) and therefore, be a reason for particle bounce
11
(Fang et al., 1991; Štefancová et al., 2010).
12
While in Prague during winter the coarse mode was mostly insignificant, in Barcelona during
13
summer the mass size distributions were clearly bimodal, with larger coarse mode
14
concentrations (Figure 4). The coarse mode obtained may be due to mineral and marine
15
aerosol contributions in the study area (Querol et al., 2008).
16
The majority of mass concentrations were found in the accumulation mode (PM1) for both
17
campaigns (7.9 ± 0.7 µg m-3 and 22.9 ± 9.8 µg m-3 according the internal reference BLPI in
18
summer Barcelona and winter Prague, respectively). With the increase in mass there was an
19
increase in agreement between the impactors, where the closest agreement was observed
20
(between 200-600 nm) (Figure 4).
21
Figure 4 reveals that the nano-Moudi recorded higher particle mass concentrations in the
22
ultrafine range (<100 nm) than the reference BLPI during winter in Prague (5 samples in total
23
outdoors). Although differences were smaller, the same is true for the Barcelona summer
24
campaign (4-week sampling, Figure 4). As previously mentioned, to protect the electrical
25
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

18
components of the nano-Moudi during winter campaign in Prague outdoors, it was kept inside
1
a climate controlled cabin with a temperature continually lower than 20˚C. At these
2
temperatures dissociation of ammonium nitrate can still occur at a slow rate (Smolík et al.,
3
2008). In addition, during the sampling, an increase of temperature inside the nano-Moudi
4
shell was detected due to the internal mechanism of spreading the sample (rotating plates)
5
which generates heat. It is therefore likely that the internal temperature in the nano-Moudi
6
was higher than that of the cabin and thus led to particle volatilisation (Štefancová et al.,
7
2010). The lower nitrate and chloride concentrations in the accumulation mode on the nano-
8
Moudi filters (see below) would support this interpretation. It is also known that a 5ºC
9
difference between the PTFE filter (of the type used in the nano-Moudi) and sampling
10
temperature may accelerate the dissociation of ammonium nitrate on PTFE filters up to 20%
11
(Hering and Cass, 1999). The BLPI and nano-BLPI have no internal warming mechanisms
12
and were located outdoors in Prague and Barcelona, so it is expected that lower volatilisation
13
would occur in these scenarios. However, drying of particles before they are deposited on a
14
substrate may happen also in the BLPI and nano-BLPI due to higher pressure drops (at
15
equivalent sizes) in comparison with the nano-Moudi. This would increase the driving force
16
for evaporation at those stages, which would encourage particle shrinkage. All these previous
17
facts (temperature, RH, high surface area) appear to enhance the evaporation of semi-volatiles
18
(and dissociation of ammonium nitrate) and therefore particle shrinkage during transport
19
through the nano-Moudi. Also, the residence time of particles inside the nano-Moudi low
20
pressure stages is longer due to the lower volumetric flow rate in this instrument. All of this
21
could thus explain the mass size distributions from the nano-Moudi being skewed towards
22
smaller particle fractions during the Barcelona and Prague campaigns (Figure 4). It should be
23
stated that the rotation of the impaction plates and the nozzle plates of the nano-Moudi was
24
specifically designed to achieve a uniform deposit on the collection substrates and therefore,
25
eradicate the particle bounce-off artefact (Marple et al., 2014) that may otherwise occur.
26
Particle bounce-off would only be expected when collecting particles in dry conditions such
27
as in Barcelona (< 50% RH) (Table 2) or indoors. Finally, the overall internal volumes in the
28
low pressure stages seem similar in all of the impactors tested; however, this would need
29
experimental confirmation.
30
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

19
3.3.2 Particle size distribution in indoor air
1
In Prague, indoor concentrations were lower than outdoors mainly due to a change in weather
2
conditions resulting in cleaner air masses during sampling periods (Figure 4 and Figure 5).
3
Reduced penetration efficiency and faster settling times probably explain the lower indoor
4
coarse mode mass obtained (Figure 5; Hussein et al, 2007). Once again, the nano-BLPI
5
measured similar mass concentrations to the reference BLPI (Figure 3) while the nano-Moudi
6
recorded notably lower mass from fine to coarse modes. In addition, the nano-Moudi size
7
distribution showed a slight shift towards larger particle sizes (Figure 5). The difference
8
between the BLPIs and the nano-Moudi could suggest that the latter underestimated mass
9
during this campaign for all particle cut sizes. Initially this would appear to reduce the
10
possibility of volatility losses being responsible for this difference, as ammonium nitrate
11
dissociates readily indoors thereby causing equal losses to all impactors (Lunden et al., 2003).
12
However, because of the way the sample is spread across the substrate in the nano-Moudi, as
13
described above, the ammonium nitrate collected would be more prone to volatilization than
14
that collected on the other impactors. Therefore it could be considered that the mechanism of
15
the nano-Moudi of spreading the sample (rotating plates), with the increase in temperatures,
16
both indoors and inside the nano-Moudi shell, could enhance dissociation/evaporation from
17
the nano-Moudi PTFE substrates. This conclusion can be supported by Figures 6 and 7, which
18
show significantly lower mass concentrations of major species of ammonium nitrate with the
19
nano-Moudi, in comparison with the BLPI.
20
A number of sources of uncertainty in this interpretation should be taken into account:
21
a) Increased uncertainty in the mass determination due to lower mass concentrations and
22
shorter sampling times
23
b) No blank correction available for nano-Moudi IC data
24
c) No uncertainty calculations for mass determinations available for nano-Moudi,
25
possibly resulting in negative mass concentrations in the lower stages
26
d) Only 2 valid samples available for indoor air (for all impactors)
27
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

20
1
Figure 5. Average mass size distributions in Prague during winter in indoor air.
2
3.3.3 Size distribution of inorganic ions
3
Figures 6 and 7 show the particle mass size distributions of major (SO42-, NO3- and NH4+) and
4
minor (Cl-, Na+, K+, Mg2+ and Ca2+) aerosol constituents for the winter campaign in Prague in
5
outdoor and indoor air, respectively. In the winter in Prague, the mass size distributions of
6
components have a predominantly fine mode (< 1 μm), with the coarse mode being almost
7
negligible in winter in Prague (by all impactors) but highly significant in Barcelona during
8
summer, such as the case for BLPI.
9
While the fine mode was dominant for the particle mass concentration and all the
10
predominant aerosol constituents (SO42-, NO3- and NH4+) for both indoor and outdoor air
11
during winter in Prague, the average mass size distributions for minor species (Cl-, Na+, K+,
12
Mg2+ and Ca2+), were clearly multimodal (Figures 6 and 7). Similar mass size distributions of
13
these species were obtained by the nano-BLPI and the reference BLPI both outdoors and
14
indoors in Prague. However marked differences in the mass size distributions of these species
15
were observed with the nano-Moudi impactor. In outdoor air there is a clear decrease of NO3-
16
and NH4+ concentrations measured with the nano-Moudi, confirming the interpretations
17
provided in the previous sections. This is also visible, even if less pronounced, indoors. In
18
addition, outdoors in Prague, the mass size distributions obtained by the BLPI showed that
19
Ca2+, Na+ and Mg2+ were dominated by coarse modes and for the case of K+, the fine mode is
20
the dominant one (suggesting biomass combustion as a possible emission source). As for Cl−,
21
the mass size distributions were clearly bimodal. The nano-Moudi outdoors had different size
22
distributions from the BLPI for Cl-, Na+, Ca2+ and Mg2+. Only for K+ the size distribution is
23
similar. Mass size distributions of Cl- and Na+ may have been influenced by filter
24
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

21
contamination. The Ca2+ peak detected at around 100 nm obtained by the nano-BLPI in
1
outdoor air may possibly be ascribed also to filter contamination, although no specific data
2
are available to support this interpretation. Similar peaks at 10 and 50 nm were observed
3
indoors with the nano-Moudi and nano-BLPI which may suggest bounce, contamination or
4
blank variability.
5
6
Figure 6. Average mass size distributions for different ionic species (left: SO42-, NO3- and
7
NH4+ and right: Cl-, Na+, K+, Mg2+ and Ca2+) during winter in outdoor air in Prague.
8
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

22
1
Figure 7. Average mass size distributions for different ionic species (left: SO42-, NO3- and
2
NH4+ and right: Cl-, Na+, K+, Mg2+ and Ca2+) during winter in indoor air in Prague.
3
4 Conclusions
4
This work aimed to assess the performance of four conventional and nano-range impactors,
5
by means of two intercomparison exercises in Prague, during winter 2015 and in Barcelona
6
during summer 2015. The aim of the campaigns was to test the instruments’ performance with
7
regard to the particle mass size distributions under different aerosol compositions resulting
8
from different emission sources, meteorology, seasons, and air mass origins. All the cascade
9
impactors were loaded with uncoated substrates excepting for the case of BLPI which foils
10
were coated.
11
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

23
Taking the BLPI as an internal reference, the best agreement regarding mass size distributions
1
was obtained with the nano-BLPI, especially for particles larger than 250 nm. The nano-
2
Moudi showed a good agreement for particle sizes >320 nm, whereas for particle diameters
3
<320 nm this instrument recorded larger mass concentrations than the internal reference.
4
Different particle effects may have caused the differences regarding particle mass
5
concentrations collected in indoor and outdoor air by the nano-Moudi. Particle volatilisation
6
may have occurred due to the internal rotating mechanisms which heat the impactor casing
7
up. Decomposition of ammonium nitrate and chloride, as evidenced by the lower nitrate and
8
chloride concentrations in the accumulation mode, is probably also enhanced in the nano-
9
Moudi due to the spreading of the sample on the whole filter surface, in comparison with
10
thick individual spots of material obtained with the BLPI and nano-BLPI impactors. Further
11
research is needed to clarify this issue. With regard to the PCIS, their size-resolved mass
12
concentrations were comparable with other impactors for PM1, PM2 and PM10, but the cut-off
13
at 0.25 μm was not consistent with that of the internal reference.
14
In Barcelona, the sampling took place under dry conditions (< 50% RH) and thus, particle
15
bounce would be expected since some particles (depending on composition) could get dry.
16
Inversely, bounce can be probably neglected for the Prague outdoor intercomparison since the
17
RH was always >50 % indicating the presence of droplet aerosols that tend to adhere to the
18
impaction substrate. To avoid such an effect impactor substrates should always be greased
19
especially in areas with low humidity.
20
Aerosol mass size distributions were assessed for the Prague and Barcelona campaigns.
21
During winter in Prague (outdoors), the mass size distributions showed a predominantly fine
22
mode, with the coarse mode being almost negligible (by all impactors). However, in
23
Barcelona, the coarse size fractions showed larger mass concentrations, evidencing the higher
24
influence of mineral and marine aerosols.
25
This study concludes that comparability between the different types of impactors assessed
26
was dependent on particle size. Specifically, the influence of the differences in impactor
27
construction (number of jets, flow, vapour pressure, etc.) on UFP mass concentrations should
28
be further addressed. In addition, further research is necessary with regard to the particle
29
processes (evaporation, bounce, etc.) behind the differences in particle mass observed across
30
size fractions in this study.
31
32
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

24
Acknowledgements
1
The research leading to these results received funding from the European Community’s
2
Seventh Framework Program (FP7-PEOPLE-2012-ITN) no. 315760 (HEXACOMM project).
3
It was also supported by Charles University in Prague, under the project GA UK no. 274213
4
and the Spanish MINECO, under the frame of SIINN, the ERA-NET for a Safe
5
Implementation of Innovative Nanoscience and Nanotechnology, in the framework of
6
ERANET-SIINN project CERASAFE (id.:16).
7
8
References
9
Amato, F., Alastuey, A., Karanasiou, A., Lucarelli, F., Nava, S., Calzolai, G., Severi, M.,
10
Becagli, S., Gianelle, V.L., Colombi, C., Alves, C., Custódio, D., Nunes, T.,
11
Cerqueira, M., Pio, C., Eleftheriadis, K., Diapouli, E., Reche, C., Minguillón, M.C.,
12
Manousakas, M., Maggos, T., Vratolis, S., Harrison, R.M., Querol, X. (2015).
13
AIRUSE-LIFE+: a harmonized PM speciation and source apportionment in 5
14
Southern European cities. Atmos. Chem. Phys. Discuss., 15(17), 23989-24039. doi:
15
10.5194/acpd-15-23989-2015
16
BcnMap. (2015). Barcelona Map, Ajuntament de Barcelona, from
17
http://w20.bcn.cat/Guiamap/Default_en.aspx#x=27601.01&y=83987.71&z=0&w=980
18
&h=574&base=GuiaMartorell
19
Berner, A., Luerzer, C. (1980). Mass size distributions of traffic aerosols at Vienna. The
20
Journal of Physical Chemistry, 84(16), 2079-2083. doi: 10.1021/j100453a016
21
Berner, A., Lürzer, C., Pohl, F., Preining, O., Wagner, P. (1979). The size distribution of the
22
urban aerosol in Vienna. Science of The Total Environment, 13(3), 245-261. doi:
23
http://dx.doi.org/10.1016/0048-9697(79)90105-0
24
Biswas, P., Flagan, R.C. (1984). High-velocity inertial impactors. Environmental Science &
25
Technology, 18(8), 611-616. doi: 10.1021/es00126a009
26
Crilley, L.R., Ayoko, G.A., Jayaratne, E.R., Salimi, F., Morawska, L. (2013). Aerosol mass
27
spectrometric analysis of the chemical composition of non-refractory PM1 samples
28
from school environments in Brisbane, Australia. Science of The Total Environment,
29
458–460, 81-89. doi: http://dx.doi.org/10.1016/j.scitotenv.2013.04.007
30
Delgado-Saborit, J.M., Stark, C., Harrison, R.M. (2014). Use of a versatile high efficiency
31
multiparallel denuder for the sampling of PAHs in ambient air: gas and particle phase
32
concentrations, particle size distribution and artifact formation. [Research Support,
33
Non-U S Gov't]. Environ Sci Technol, 48(1), 499-507.
34
Fang, C.P., McMurry, P.H., Marple, V.A., Rubow, K.L. (1991). Effect of Flow-induced
35
Relative Humidity Changes on Size Cuts for Sulfuric Acid Droplets in the
36
Microorifice Uniform Deposit Impactor (MOUDI). Aerosol Science and Technology,
37
14(2), 266-277. doi: 10.1080/02786829108959489
38
Fujitani, Y., Hasegawa, S., Fushimi, A., Kondo, Y., Tanabe, K., Kobayashi, S., Kobayashi, T.
39
(2006). Collection characteristics of low-pressure impactors with various impaction
40
substrate materials. Atmospheric Environment, 40(18), 3221-3229. doi:
41
http://dx.doi.org/10.1016/j.atmosenv.2006.02.001
42
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

25
Geller, M.D., Kim, S., Misra, C., Sioutas, C., Olson, B.A., Marple, V.A. (2002). A
1
Methodology for Measuring Size-Dependent Chemical Composition of Ultrafine
2
Particles. Aerosol Science and Technology, 36(6), 748-762. doi:
3
10.1080/02786820290038447
4
Hering, S., Cass, G. (1999). The Magnitude of Bias in the Measurement of PM25 Arising
5
from Volatilization of Particulate Nitrate from Teflon Filters. Journal of the Air &
6
Waste Management Association, 49(6), 725-733. doi:
7
10.1080/10473289.1999.10463843
8
Hering, S.V., Flagan, R.C., Friedlander, S.K. (1978). Design and evaluation of new low-
9
pressure impactor. I. Environmental Science & Technology, 12(6), 667-673. doi:
10
10.1021/es60142a004
11
Hillamo, R.E., Kauppinen, E.I. (1991). On the Performance of the Berner Low Pressure
12
Impactor. Aerosol Science and Technology, 14(1), 33-47. doi:
13
10.1080/02786829108959469
14
Hinds, W.C. (1999). Aerosol technology : properties, behavior, and measurement of airborne
15
particles. New York: Wiley.
16
Hitzenberger, R., Berner, A., Galambos, Z., Maenhaut, W., Cafmeyer, J., Schwarz, J., Müller,
17
K., Spindler, G., Wieprecht, W., Acker, K., Hillamo, R., Mäkelä, T. (2004).
18
Intercomparison of methods to measure the mass concentration of the atmospheric
19
aerosol during INTERCOMP2000—influence of instrumentation and size cuts.
20
Atmospheric Environment, 38(38), 6467-6476. doi:
21
http://dx.doi.org/10.1016/j.atmosenv.2004.08.025
22
Howell, S., Pszenny, A.A.P., Quinn, P., Huebert, B. (1998). A Field Intercomparison of Three
23
Cascade Impactors. Aerosol Science and Technology, 29(6), 475-492. doi:
24
10.1080/02786829808965585
25
Huang, Z., Harrison, R.M., Allen, A.G., James, J.D., Tilling, R.M., Yin, J. (2004). Field
26
intercomparison of filter pack and impactor sampling for aerosol nitrate, ammonium,
27
and sulphate at coastal and inland sites. Atmospheric Research, 71(3), 215-232. doi:
28
http://dx.doi.org/10.1016/j.atmosres.2004.05.002
29
Hussein, T., Glytsos, T., Ondráček, J., Dohányosová, P., Ždímal, V., Hämeri, K., Lazaridis,
30
M., Smolík, J., Kulmala, M. (2006). Particle size characterization and emission rates
31
during indoor activities in a house. Atmospheric Environment, 40(23), 4285-4307. doi:
32
http://dx.doi.org/10.1016/j.atmosenv.2006.03.053
33
Hussein, T., Kukkonen, J., Korhonen, H., Pohjola, M., Pirjola, L., Wraith, D., Härkönen, J.,
34
Teinilä, K., Koponen, I.K., Karppinen, A., Hillamo, R., Kulmala, M. (2007).
35
Evaluation and modeling of the size fractionated aerosol particle number
36
concentration measurements nearby a major road in Helsinki &ndash; Part II: Aerosol
37
measurements within the SAPPHIRE project. Atmos. Chem. Phys., 7(15), 4081-4094.
38
doi: 10.5194/acp-7-4081-2007
39
IPR. (2015). Geoportal Praha, Prague geographic data in one place, Prague Institute of
40
Planning and Development (IPR Praha), from http://www.geoportalpraha.cz/en/maps-
41
online#.VlIWWLerR1s
42
Jiménez, S., Ballester, J. (2011). Use of a Berner Low-Pressure Impactor at Low Inlet
43
Pressures. Application to the Study of Aerosols and Vapors at High Temperature.
44
Aerosol Science and Technology, 45(7), 861-871. doi:
45
10.1080/02786826.2011.566900
46
Karanasiou, A.A., Sitaras, I.E., Siskos, P.A., Eleftheriadis, K. (2007). Size distribution and
47
sources of trace metals and n-alkanes in the Athens urban aerosol during summer.
48
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

26
Atmospheric Environment, 41(11), 2368-2381. doi:
1
http://dx.doi.org/10.1016/j.atmosenv.2006.11.006
2
Lin, C.-C., Chen, S.-J., Huang, K.-L., Hwang, W.-I., Chang-Chien, G.-P., Lin, W.-Y. (2005).
3
Characteristics of Metals in Nano/Ultrafine/Fine/Coarse Particles Collected Beside a
4
Heavily Trafficked Road. Environmental Science & Technology, 39(21), 8113-8122.
5
doi: 10.1021/es048182a
6
Lunden, M.M., Revzan, K.L., Fischer, M.L., Thatcher, T.L., Littlejohn, D., Hering, S.V.,
7
Brown, N.J. (2003). The transformation of outdoor ammonium nitrate aerosols in the
8
indoor environment. Atmospheric Environment, 37(39–40), 5633-5644. doi:
9
http://dx.doi.org/10.1016/j.atmosenv.2003.09.035
10
Marple, V., Olson, B., Romay, F., Hudak, G., Geerts, S.M., Lundgren, D. (2014). Second
11
Generation Micro-Orifice Uniform Deposit Impactor, 120 MOUDI-II: Design,
12
Evaluation, and Application to Long-Term Ambient Sampling. Aerosol Science and
13
Technology, 48(4), 427-433. doi: 10.1080/02786826.2014.884274
14
Marple, V.A., Rubow, K.L., Behm, S.M. (1991). A Microorifice Uniform Deposit Impactor
15
(MOUDI): Description, Calibration, and Use. Aerosol Science and Technology, 14(4),
16
434-446. doi: 10.1080/02786829108959504
17
Mašková, L., Smolík, J., Vodička, P. (2015). Characterisation of particulate matter in
18
different types of archives. Atmospheric Environment, 107, 217-224. doi:
19
http://dx.doi.org/10.1016/j.atmosenv.2015.02.049
20
May, K.R. (1945). The Cascade Impactor. Journal of Scientific Instruments, 22(12), 247.
21
Misra, C., Singh, M., Shen, S., Sioutas, C., Hall, P.M. (2002). Development and evaluation of
22
a personal cascade impactor sampler (PCIS). Journal of Aerosol Science, 33(7), 1027-
23
1047. doi: 10.1016/s0021-8502(02)00055-1
24
Nie, W., Wang, T., Gao, X., Pathak, R.K., Wang, X., Gao, R., Zhang, Q., Yang, L., Wang, W.
25
(2010). Comparison among filter-based, impactor-based and continuous techniques for
26
measuring atmospheric fine sulfate and nitrate. Atmospheric Environment, 44(35),
27
4396-4403. doi: http://dx.doi.org/10.1016/j.atmosenv.2010.07.047
28
Oberdörster, G. (2000). Pulmonary effects of inhaled ultrafine particles. Int Arch Occup
29
Environ Health, 74(1), 1-8. doi: 10.1007/s004200000185
30
Oberdorster, G., Oberdorster, E., Oberdorster, J. (2005). Nanotoxicology: an emerging
31
discipline evolving from studies of ultrafine particles. [Research Support, N I H ,
32
Extramural
33
Research Support, U S Gov't, Non-P H S
34
Research Support, U S Gov't, P H S
35
Review]. Environ Health Perspect, 113(7), 823-839.
36
Ondráček, J., Schwarz, J., Ždímal, V., Andělová, L., Vodička, P., Bízek, V., Tsai, C.J., Chen,
37
S.C., Smolík, J. (2011). Contribution of the road traffic to air pollution in the Prague
38
city (busy speedway and suburban crossroads). Atmospheric Environment, 45(29),
39
5090-5100. doi: http://dx.doi.org/10.1016/j.atmosenv.2011.06.036
40
Pennanen, A.S., Sillanpaa, M., Hillamo, R., Quass, U., John, A.C., Branis, M., Hunova, I.,
41
Meliefste, K., Janssen, N.A., Koskentalo, T., Castano-Vinyals, G., Bouso, L., Chalbot,
42
M.C., Kavouras, I.G., Salonen, R.O. (2007). Performance of a high-volume cascade
43
impactor in six European urban environments: mass measurement and chemical
44
characterization of size-segregated particulate samples. [Evaluation Studies
45
Research Support, Non-U S Gov't]. Sci Total Environ, 374(2-3), 297-310.
46
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.

27
Preining, O., Berner, A. (1979). Aerosol Measurements in the Submicron Size Range. EPA
1
report, EPA-600/2-79-105. Washington, DC: EPA.
2
Querol, X., Alastuey, A., Moreno, T., Viana, M.M., Castillo, S., Pey, J., Rodríguez, S.,
3
Artiñano, B., Salvador, P., Sánchez, M., Garcia Dos Santos, S., Herce Garraleta, M.D.,
4
Fernandez-Patier, R., Moreno-Grau, S., Negral, L., Minguillón, M.C., Monfort, E.,
5
Sanz, M.J., Palomo-Marín, R., Pinilla-Gil, E., Cuevas, E., de la Rosa, J., Sánchez de la
6
Campa, A. (2008). Spatial and temporal variations in airborne particulate matter
7
(PM10 and PM2.5) across Spain 1999–2005. Atmospheric Environment, 42(17), 3964-
8
3979. doi: http://dx.doi.org/10.1016/j.atmosenv.2006.10.071
9
Reche, C., Viana, M., Brines, M., Perez, N., Beddows, D., Alastuey, A., Querol, X. (2015).
10
Determinants of aerosol lung-deposited surface area variation in an urban
11
environment. [Research Support, Non-U S Gov't]. Sci Total Environ, 517, 38-47.
12
Sardar, S.B., Fine, P.M., Mayo, P.R., Sioutas, C. (2005). Size-Fractionated Measurements of
13
Ambient Ultrafine Particle Chemical Composition in Los Angeles Using the
14
NanoMOUDI. Environmental Science & Technology, 39(4), 932-944. doi:
15
10.1021/es049478j
16
Schaap, M., Spindler, G., Schulz, M., Acker, K., Maenhaut, W., Berner, A., Wieprecht, W.,
17
Streit, N., Müller, K., Brüggemann, E., Chi, X., Putaud, J.P., Hitzenberger, R.,
18
Puxbaum, H., Baltensperger, U., ten Brink, H. (2004). Artefacts in the sampling of
19
nitrate studied in the “INTERCOMP” campaigns of EUROTRAC-AEROSOL.
20
Atmospheric Environment, 38(38), 6487-6496. doi:
21
http://dx.doi.org/10.1016/j.atmosenv.2004.08.026
22
Schwarz, J., Štefancová, L., Maenhaut, W., Smolík, J., Ždímal, V. (2012). Mass and
23
chemically speciated size distribution of Prague aerosol using an aerosol dryer--the
24
influence of air mass origin. Sci Total Environ, 437, 348-362. doi:
25
10.1016/j.scitotenv.2012.07.050
26
Seinfeld, J.H., Pandis, S.N. (2006). Atmospheric Chemistry and Physics: From Air Pollution
27
to Climate Change, 2nd edition. J. Wiley, New York.
28
Sioutas, C. (2004). Development of New Generation Personal Monitors for Fine Particulate
29
Matter (PM) and its Metal Content. NUATRC Research Report No. 2.
30
Smolík, J., Dohányosová, P., Schwarz, J., Ždímal, V., Lazaridis, M. (2008). Characterization
31
of Indoor and Outdoor Aerosols in a Suburban Area of Prague. Water Air Soil Pollut:
32
Focus, 8(1), 35-47. doi: 10.1007/s11267-007-9141-y
33
Štefancová, L., Schwarz, J., Maenhaut, W., Chi, X., Smolík, J. (2010). Hygroscopic growth of
34
atmospheric aerosol sampled in Prague 2008 using humidity controlled inlets.
35
Atmospheric Research, 98(2–4), 237-248. doi:
36
http://dx.doi.org/10.1016/j.atmosres.2010.04.009
37
Štefancová, L., Schwarz, J., Mäkelä, T., Hillamo, R., Smolík, J. (2011). Comprehensive
38
Characterization of Original 10-Stage and 7-Stage Modified Berner Type Impactors.
39
Aerosol Science and Technology, 45(1), 88-100. doi: 10.1080/02786826.2010.524266
40
Tursic, J., Grgic, I., Berner, A., Skantar, J., Cuhalev, I. (2008). Measurements of size-
41
segregated emission particles by a sampling system based on the cascade impactor.
42
[Research Support, Non-U S Gov't]. Environ Sci Technol, 42(3), 878-883.
43
Wall, S.M., John, W., Ondo, J.L. (1988). Measurement of aerosol size distributions for nitrate
44
and major ionic species. Atmospheric Environment (1967), 22(8), 1649-1656. doi:
45
http://dx.doi.org/10.1016/0004-6981(88)90392-7
46
Wang, H.-C., John, W. (1988). Characteristics of the Berner Impactor for Sampling Inorganic
47
Ions. Aerosol Science and Technology, 8(2), 157-172. doi:
48
10.1080/02786828808959179
49
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2015-1016, 2016
Manuscript under review for journal Atmos. Chem. Phys.
Published: 19 January 2016
c
Author(s) 2016. CC-BY 3.0 License.