Introduction
Atmospheric aerosols have been intensively studied in the last decades
because of their effects on climate, air quality and ecosystems. In many
environments, organic aerosol (OA) constitutes a dominant fraction of
submicron particles mass (Zhang et al., 2007; Jimenez et al., 2009). OA is
made of thousands of individual compounds that can either be emitted
directly into the atmosphere (i.e. primary OA or POA) or formed through
chemical and physical processes (i.e. secondary OA or SOA). Given the
extremely wide range of properties (polarity, vapour pressure, etc.) and
number of compounds, typically only 10–20 % of the OA mass can be
speciated at the molecular level (Seinfeld and Pankow, 2003).
OA is a dynamic component, experiencing both atmospheric oxidation and
reversible partitioning. This processing (usually referred to as ageing) is
generally not completely understood and not well represented in models
(Heald et al., 2010). In the last decade, online instruments like the
Aerodyne Research Inc. aerosol mass spectrometer (AMS) (Jayne et al., 2000;
Canagaratna et al., 2007) have provided new insights into OA chemical
composition and simplified ways of characterizing atmospheric OA ageing. Ng
et al. (2010) showed that OA composition tends to become less variable with
photochemical ageing, regardless of its source, with the most oxidized
spectra resembling that of fulvic acid. Heald et al. (2010) applied the Van
Krevelen diagram (H : C vs. O : C space) to the elemental composition of ambient
and laboratory OA, observing that bulk OA elemental ratios follow a line
characterized by a slope of -1. This implies that OA ageing involves, on
average, the addition of carboxylic acids or equal amounts of hydroxyl and
carbonyl functionalities. By contrast, based on a worldwide data set, Ng et
al. (2011) observed a slope of ∼-0.5 for the oxidation of SOA
which is characteristic of the addition of alcohol, carbonyl and carboxyl
functionalities during the ageing process, or of the addition of carboxylic
groups accompanied by a C–C bond cleavage (molecular fragmentation).
Recently, Holzinger et al. (2013) showed that fragmentation gains importance
over functionalization as the photochemical age of OA increases, as originally
proposed by Kroll et al. (2009). Finally, both ambient observations (Jimenez
et al., 2009; Morgan et al., 2010) and laboratory studies (Massoli et al.,
2010; Lambe et al., 2011) have pointed out that atmospheric ageing lowers OA
volatility and enhances their hygroscopicity, evidencing the importance of
atmospheric processing in determining the OA climate-relevant properties.
In this study, we investigate the atmospheric processing of OA over the Po
Valley basin, taking advantage of the unique observatory of Mt. Cimone, part
of the Global Atmosphere Watch (GAW) network by the World Meteorological
Organization (WMO), a suitable location to study tropospheric background
conditions. As many other mountain sites are close to anthropogenically
impacted areas, Mt. Cimone provides the opportunity to investigate transport
and chemical processing of polluted air masses lifted by convection or by
pressure gradients on the mountain slopes (valley breezes) (Marinoni et al.,
2008; Gilardoni et al., 2009).
The first AMS measurements performed at high-altitude mountain stations were
reported by Hock et al. (2008) and by Lanz et al. (2008). In particular, the
latter study highlighted lower concentrations and higher oxygen content in
aerosols collected at the Hohenpeissenberg (Germany) and Jungfraujoch
(Switzerland) stations compared to measurements performed simultaneously at
low-altitude sites. Sun et al. (2009) and Freney et al. (2011) confirmed
these findings with measurements performed at Whistler Mountain (Canada),
and Puy de Dome (France), respectively, and provided useful insights of
seasonal effects and air mass origin on the physico-chemical properties of
regional aerosol particles measured at elevated sites.
The Mt. Cimone GAW/WMO station is a high-altitude research site located in
the north Italian Apennines, facing the heavily populated and industrialized
Po Valley region. In this study, we present and discuss online submicron
aerosol chemical composition data collected at Mt. Cimone by a high-resolution time-of-flight aerosol mass spectrometer, HR-ToF-AMS (DeCarlo et
al., 2006), during summer 2012, within the EU project PEGASOS and the
Agenzia Regionale per la Prevenzione e l'Ambiente (ARPA) – Emilia-Romagna
SUPERSITO project. The measurements were used to characterize the summer
background aerosol transported into the Po Valley basin area, the vertical
transport of anthropogenic aerosol from the lower troposphere (typical of
summer circulation) and the regional-scale oxidation of OA. Prior to this
study, aerosol chemical composition data for Mt. Cimone station were
reported by Putaud et al. (2004), Marenco et al. (2006) and Carbone et al. (2010, 2014) using offline aerosol characterization
techniques (i.e. filter samples analysed by ion chromatography and organic
carbon analysis). All these papers evidenced the importance of the OA
fraction in submicron aerosol at the site, both in summer and winter.
However, this is the first time that online characterization of atmospheric
aerosol, particularly of OA, was performed with high resolution on a
mountain site at the centre of the Mediterranean climate hotspot region.
Methods
Sampling site
Mt. Cimone is the highest peak of the north Italian Apennines. The top of
Mt. Cimone (44∘11′ N, 10∘42′ E; 2165 m a.s.l.) hosts
the Italian Climate Observatory “O. Vittori” that is part of the GAW
program of the WMO. The station is situated at the southern border of the Po
Valley, which is a highly populated and industrialized area, also
characterized by intense agricultural activities. Anticyclonic conditions
often favour a reduced ventilation within the basin, promoting the build-up
of lower troposphere aerosols and pollutants.
Measurements of atmospheric components carried out at this site are
generally considered representative for the south European free troposphere
(Bonasoni et al., 2000; Fischer et al., 2003). Nevertheless, due to enhanced
vertical mixing occurring during summer months, a daytime influence at Mt. Cimone from the planetary
boundary layer (PBL) has been documented (Fischer et al., 2003; Cristofanelli
et al., 2007). For these reasons, this measurement site can represent a
suitable location to investigate the influence of both local and long-range
transport of polluted air masses on the free troposphere composition
(Marinoni et al., 2008).
Online aerosol chemical characterization
The mass loading and the size-resolved chemical composition of submicron
aerosol particles was characterized online by the Aerodyne HR-ToF-AMS. The
HR-ToF-AMS provides measurements of the non-refractory sulfate, nitrate,
ammonium, chloride, and organic mass in submicron particles (NR-PM1).
During the whole campaign, the HR-ToF-AMS was operating by alternating
between “V” and “W” ion path modes every 5 min. The V mode is
characterized by higher sensitivity and lower mass resolution, while the
W mode provides higher mass resolution, but lower sensitivity. The
concentrations reported here correspond to the data collected in V mode. In
V mode, the instrument also acquires information about size distribution of
particles, or particle time-of-flight, pToF (Jimenez et al., 2003). The AMS
has an effective 50 % cut-off for particle sizes below 80 nm and above 600 nm in vacuum aerodynamic diameter (dva), as determined by the
transmission characteristics of the standard aerodynamic lens (Liu et al.,
2007). Changes in ambient pressure may lead to changes in lens transmission
efficiencies (Liu et al., 2007; Bahreini et al., 2008), but such effects are
not expected to be significant under the pressure conditions typical of Mt. Cimone (Liu et al., 2007; Sun et al., 2009). However, the particle velocity
calibration was adjusted to the altitude (and pressure) conditions of the
Mt. Cimone site before starting the measurements. Ionization efficiency
(IE) calibrations were performed before and after the campaign, and once per
week during the campaign. Filter blank acquisitions during the campaign were
performed at least a couple of times per day to evaluate the background and
correct the gas-phase contribution. All data were analysed using standard
ToF-AMS analysis software SQUIRREL v1.51 and PIKA v1.10 (D. Sueper,
available at: http://cires.colorado.edu/jimenez-group/ToFAMSResources/ToFSoftware/index.html) within Igor Pro 6.2.1 (WaveMetrics, Lake Oswego,
OR). Positive matrix factorization (PMF) analyses of the HR-ToF-AMS data
were performed using the PMF2.exe algorithm (v.4.2) in robust mode (Paatero
and Tapper, 1994). The PMF inputs (mass spectral and error matrices of the
OA component) were prepared according to Zhang et al. (2011). The PMF
solutions were then evaluated with an Igor Pro-based PMF evaluation tool
(PET, v. 2.04) following the method described in Ulbrich et al. (2009) and
Zhang et al. (2011). The HR-ToF-AMS collection efficiency (CE) was
calculated according to Middlebrook et al. (2012) and evaluated against
parallel offline measurements (Fig. S1 in the Supplement). The average CE for the campaign was
0.52 ± 0.06. The propagated, overall uncertainty for the total AMS mass
concentration is 20–35 % (2σ) according to Middlebrook et al. (2012). The aerosol was sampled via a total suspended particle (TSP) aerosol
inlet, which is built according to the EUSAAR/ACTRIS protocol to improve the
collection performances at high altitude. The aerosol was dried to about
40 % by means of a Nafion drier before sampling with the HR-ToF-AMS.
Additional measurements
Ancillary measurements at Mt. Cimone conducted during the campaign included
meteorological parameters and other trace gases (CO, O3, NOx).
Trace gas measurements were carried out by using a common sampling system
designed for reactive gas sampling, characterized by an intake line located
2 m above the roof and 7 m above the ground and consisting of a glass tube
through which the sampled air is passed at a high flow rate (larger than 20 L s-1). Sample air was supplied to the various analysers via a Teflon
manifold pipe (about 1 m long) connected to the glass tube. A particle
filter (changed regularly every 15–30 days) prevented dust, raindrops and
other unwanted material from entering the inlet.
Surface ozone (O3) was continuously measured (1 min time resolution)
by a UV-absorption analyser Dasibi 1108 W/GEN (Cristofanelli et al., 2015).
Carbon dioxide (CO) was measured by a non-dispersive infrared (NDIR) analyser Thermo Tei 49C.
Following Henne et al. (2008), the system and sampling procedures have been
modified to carry out observations in remote conditions usually
characterized by low mixing ratios. During the PEGASOS campaign, NOx
measurements were carried out by a chemioluminescence analyser (Thermo 42C).
This instrument is equipped with a molybdenum converter to determine
NOx, which according to Steinbacher et al. (2007), can overestimate the
NO2 up to ∼ 50 % due to the interference of oxidized
nitrogen compounds (NOy) such as peroxyacetyl nitrate and nitric acid.
The measurement of the aerosol absorption coefficient was obtained by a
Multi-Angle Absorption Photometer (MAAP 5012, Thermo Electron Corporation),
which measures the transmission and the back scattering of a light beam
incident on a fiber filter where aerosol particles are deposited by the
sampling flow. The equivalent black carbon (eqBC) concentration has been
obtained by using a mass absorption efficiency of 6.5 m2 g-1, as
recommended by Petzold et al. (2002).
PM1 offline aerosol samples were collected on quartz filters as
described by Carbone et al. (2014) with a 12 h sampling resolution. Chemical
analysis of main inorganic species was performed via ion chromatography and
carbon elemental analysis (Carbone et al., 2014).
Time series and relative contribution of the main
NR-PM1 components. Time is local (UTC + 2 h).
Results
Online aerosol chemical characterization at Mt. Cimone
Figure 1 shows the time trend of the major non-refractory components of
submicron aerosol (NR-PM1) measured at Mt. Cimone during the campaign,
together with the time-dependent relative mass contribution of the same
aerosol components. NR-PM1 was clearly dominated by OA through the
whole campaign: OA average atmospheric concentration and standard deviation
were 2.8 ± 2.4 µg m-3, for an average mass contribution of
63 %. Sulfate was the second most abundant species with a concentration
of 0.92 ± 0.60 µg m-3 (20 %), followed by ammonium
(0.41 ± 0.33 µg m-3, 9 %) and nitrate (0.33 ± 0.46 µg m-3, 7 %). Chlorine was usually below the detection limit
(80 % of the data points) and contributed less than 1 % to the
NR-PM1 (and was therefore excluded from Fig. 1b). For comparison,
submicrometric aerosol chemical characterization measurements performed in
spring and summer at high (Jungfraujoch) and rural (Hohenpeissenberg) sites
in Europe (Hock et al., 2008; Lanz et al., 2010) report organics ranging
from 43 to 50 %, sulfate ranging from 19 to 26 %, ammonium
contributing between 13 and 11 % and nitrate ranging from 18 to
19 % (see Table 1 for a summary of AMS measurements performed at
background measurements sites).
The measured ammonium concentrations were in equivalent concentrations to
the sum of sulfate, nitrate and chloride, with a slope of 0.99 and a linear
correlation coefficient (R) of 0.99. This means that aerosol particles
measured at Mt. Cimone were neutralized. The average NR-PM1 mass during
the campaign was 4.5 ± 3.4 µg m-3, in fairly good agreement
with PM1 measurements previously performed at the site in the same
season, even though with different measurement techniques (Marenco et al.,
2006; Carbone et al. 2010, 2014). The lowest aerosol
mass concentrations were observed during the first days of the campaign up
to 15 June, when northern Italy was influenced by a low-pressure system,
bringing unstable conditions. Conversely, the highest concentrations were
recorded between 17 and 20 June under high-pressure conditions characterized
by anticyclonic circulation (Decesari et al., 2015), which is
known to favour the stagnation of local pollutants produced within the Po
Valley basin (see Fig. S2 for further details).
Time series of the H : C, O : C and OM : OC ratios. Time is local
(UTC + 2 h).
The results of the elemental analysis (EA) of the organic fraction are
presented in Fig. 2 as time trends of the H : C, O : C and OM : OC ratios.
Average H : C, O : C and OM : OC ratio measured during the campaign are
1.45 ± 0.11, 0.71 ± 0.08 and 2.08 ± 0.10, respectively,
corresponding to an average oxidation state (OSC= 2 × O : C - H : C; Kroll
et al., 2011) of -0.02 ± 0.23. These ratios are indicative of a highly
oxygenated organic aerosol, in agreement with previous AMS measurements at
mountain sites (Freney et al., 2011; Lanz et al., 2010). The values have
been calculated using the Improved-Ambient (I-A) EA method to derive OA
elemental ratios from AMS spectra (Canagaratna et al., 2015).
Summary of AMS measurements at mountain sites published in the
literature. Concentrations are expressed in µg m-3.
Reference
Site/
Altitude
Organics
Nitrate
Sulfate
Ammonium
Chloride
H : C
O : C
OM : OC
season
(m a.s.l.)
Hock et al. (2008)
Hohenpeissenberg/
985
3.4
1.3
1.3
0.7
0.07
–
–
–
spring
(50 %)
(19 %)
(19 %)
(11 %)
(1 %)
Sun et al. (2009)
Whistler Mountain/
2182
1.05 ± 1.03
0.05 ± 0.10
0.58 ± 0.41
0.23 ± 0.16
–
1.66 ± 0.06
0.83 ± 0.17
2.28 ± 0.23
spring
(55 %)
(3 %)
(30 %)
(12 %)
Lanz et al. (2010)
Jungfraujoch/
3580
0.7
0.3
0.4
0.2
< 0.02
–
–
–
spring
(43 %)
(18 %)
(26 %)
(13 %)
(< 1 %)
Freney et al. (2011)
Puy de Dôme/
1465
2.52
1.14
2.4
1.36
0.02
–
–
–
autumn
(34 %)
(15 %)
(32 %)
(18 %)
(0.3 %)
Freney et al. (2011)
Puy de Dôme/
1465
1.24
1.71
1.28
1.08
0.07
–
–
–
winter
(23 %)
(32 %)
(24 %)
(20 %)
(1 %)
Freney et al. (2011)
Puy de Dôme/
1465
15.59
2.33
5.45
3.69
0.06
–
–
–
summer
(57 %)
(9 %)
(20 %)
(14 %)
(0.2 %)
This study
Mt. Cimone/
2165
2.8 ± 2.4
0.33 ± 0.46
0.92 ± 0.60
0.41 ± 0.33
(< 1 %)
1.45 ± 0.11
0.71 ± 0.08
2.08 ± 0.10
summer
(63 %)
(7 %)
(20 %)
(9 %)
The corresponding EA values calculated using the Aiken-Ambient (A-A) method
(Aiken et al., 2008), are 1.32 ± 0.08, 0.58 ± 0.07 and 1.89 ± 0.09, respectively, corresponding to an average
oxidation state (OSC= 2 × O : C - H : C) of -0.16 ± 0.22. The elemental ratios calculated with
the A-A method are reported here just for the purpose of a more direct
comparison with papers published before the introduction of the new I-A
method (2015). Anyway, through all the paper and in the plots, the more
accurate I-A elemental ratios will be reported. Saarikoski et al. (2012)
present results of HR-ToF-AMS measurements in the Po Valley at the site of
San Pietro Capofiume (SPC) during April 2008, showing an average H : C ratio
slightly higher (1.49) and an average O : C ratio slightly lower (0.47) than
those observed at Mt. Cimone, for a resulting average OM : OC ratio of 1.77.
Similarly, OM : OC higher than 1.7 was observed in the outflow plume over
Mexico City and at the mountain site of Altzomoni, above the Mexico City
plateau (Gilardoni et al., 2009). The OM : OC ratio observed at SPC in fall
2011 was 1.6 (Gilardoni et al., 2014). The lower oxidation of the OA
collected at SPC during spring and fall with respect to the present
measurements can be due to (a) the different season or (b) oxidation
processes involving OA during transport from low-altitude sites up to Mt. Cimone. This last aspect will be investigated in Sect. 3.3.
Analysis of the diurnal cycles
The atmospheric concentrations of the major NR-PM1 components present a
clear diurnal cycle with maxima at the early afternoon and minima during the
night (Fig. 3). The concentration daily trend of the NR-PM1
components is the result of the PBL dynamics and valley breezes, as during
the night the site is well above the shallow nocturnal layer forming over
the Po Valley plain and disconnected from the aerosol sources located at the
low altitudes. However, Marinoni et al. (2008) showed that in summer, during
the night, Mt. Cimone station may be affected by polluted air masses
present in the residual layer above the Po Valley. Conversely, during the
day, with the increase of the PBL height, the site is affected by convective
transport from lower altitudes (Schuepbach et al., 2001; Fischer et al.,
2003; Freney et al., 2011) and it is directly connected to the pollution
sources located in the valley, thus experiencing high aerosol
concentrations. During PEGASOS, the time trend of OA (the major contributor
to NR-PM1) correlates with that of specific humidity (SH), which can be
used as a tracer of PBL air at high altitudes (Henne et al., 2005) (Fig. S3). This good correlation strongly supports the hypothesis that the aerosol
transport triggered by the PBL dynamics is the main factor regulating the
NR-PM1 concentrations at Mt. Cimone during the measurement period.
To further investigate the importance of vertical transport from the PBL to
the top of Mt. Cimone during the day, we calculated the daily relative
increase (RI) in SH, following the approach introduced by Prévôt et al. (2000) and Henne et al. (2005) for two different sites in the Alps and
already applied for Mt. Cimone station by Carbone et al. (2014).
RI=SHaftCMN-SHmorCMNSHaftSPC-SHmorCMN
In Eq. (1), SHatf (CMN) is the average specific humidity measured in
the afternoon at Mt. Cimone (12:00 to 18:00 LT), SHmor (CMN) is the
average specific humidity measured during the night at Mt. Cimone (22:00 to
05:00 LT) and SHaft(SPC) is the average specific humidity measured in
the afternoon (12:00 to 18:00 LT) at the rural background station of San
Pietro Capofiume (SPC), located in the Po Valley 90 km north-east of Mt. Cimone at 11 m a.s.l. and considered representative of PBL conditions within
the Po Valley basin. A RI of 1 corresponds to a complete replacement of the
high-altitude air by boundary layer air, while no vertical motion yields
zero relative increase. RI was calculated for each day of the campaign and
the average value was 0.8 ± 0.3, confirming the high influence of
vertical convection during the day at the station in summer.
Daily trends of (a) organics, (b) nitrate, (c) sulfate and
(d) ammonium. Boxes represent median, 25th and 75th percentile; whiskers
indicates 10th and 90th percentile. Time is local (UTC + 2 h).
Similarly to the NR-PM1 components, OA elemental ratios exhibit diurnal
variations (Fig. 4). The O : C and OM : OC ratios have lower values in the
afternoon and maxima at night, with a minimum O : C hourly average of
0.67 ± 0.10 observed between 14:00 and 15:00, and maximum of
0.75 ± 0.08 between 00:00 and 01:00. An opposite trend is observed for
the H : C ratio, with a maximum hourly average of 1.55 ± 0.10 between
14:00 and 15:00, and a minimum of 1.38 ± 0.10 between 00:00 and 01:00.
The daily trends of the O : C and OM : OC ratios are almost coincident
(correlation coefficient 0.999), confirming the results of Pang et al. (2006)
showing that the OM : OC ratio is mainly regulated by the O : C ratio.
Daily trends of the (a) H : C, (b) O : C and (c) OM : OC ratios. Boxes
represent median, 25th and 75th percentile; whiskers indicates 10th and 90th
percentile. Time is local (UTC + 2 h).
These trends highlight the different age of the aerosols measured at the
sampling location in different moments of the day, as a consequence of the
PBL dynamics. In fact, the O : C ratio of OA tends to increase and the H : C
ratio decreases as a function of its atmospheric residence time, because of
the oxidation of reduced species emitted by traffic and combustion and of
SOA formation (Aiken et al., 2008; DeCarlo et al., 2008; Heald et al., 2010;
Chhabra et al., 2011; Ng et al., 2011; Sun et al., 2011b, a). Particles sampled at Mt. Cimone during the day are representative of
an early stage of aerosol atmospheric oxidation, resulting from SOA formed
at lower altitudes in the Po Valley and transported upward by turbulence and
by thermal winds, typically in few hours. By contrast, at night the aerosol
sampled at Mt. Cimone is more processed, as the atmospheric layers affecting
the site at night contain aerosols with an age of several hours (residual
layers) to days (from long-range transport). This is confirmed by the CO / NOx
ratio, often used in the literature as a tracer for air mass photochemical
age (Morgan et al., 2010; Freney et al., 2011). CO / NOx at Mt. Cimone is
159 ± 65 (average ± standard deviation) during the day and
287 ± 168 at night, which is a value representative of aged regional
emissions. The CO / NOx ratio is presented in detail in Fig. S4,
showing clearly that during the night the CO / NOx ratio is
systematically higher than during the day. For these reasons, Mt. Cimone is
an ideal site to investigate the processing of organic aerosol over the Po
Valley basin. Moreover, according to Marinoni et al. (2008), the footprint
of aerosol particles found in the residual layers at night comprises a great
part of central Europe, which is the region where the full oxidation of
organic aerosols that we observe at Mt. Cimone takes place.
Van Krevelen diagram presenting the H : C and O : C ratios of all the
data points collected during the campaign, together with average values
for PBL, TR and FT samples. The solid line represents the fit to the data
(not constrained to H : C = 2). Dashed lines describes oxidation reactions
occurring through addition of carbonyl groups (slope =-2; Heald et al.,
2010), carboxylic acid without fragmentation (slope =-1; Heald et al.,
2010), carboxylic acid with fragmentation (slope =-0.5; Ng et al.,
2011) and alcohol/peroxide (slope = 0; Heald et al., 2010).
Investigation of regional scale organic aerosol ageing
To investigate the oxidation of OA, data collected at Mt. Cimone during the
campaign have been divided based on the position of the station relative to
the PBL height, using SH as a tracer of the PBL evolution. Figure S5 shows
the average daily evolution of the SH at Mt. Cimone during the campaign that
mimics the PBL evolution during the day: the afternoon maximum indicates
that the site is within the PBL, under the influence of moist air coming
from lower altitudes, while the night minimum indicates that Mt. Cimone
station is above the PBL. Following Fig. S5, HR-ToF-AMS measurements
collected between 12:00 and 18:00 have been considered as PBL samples, those
collected between 22:00 and 05:00 have been defined as free troposphere (FT)
samples, while all the samples excluded from the previous two groups have
been considered as transition samples (TR). As expected, PBL samples were
less oxidized (H : C = 1.54 ± 0.06, O : C = 0.69 ± 0.05, OM : OC = 2.05 ± 0.10)
than FT samples (H : C = 1.41 ± 0.09, O : C = 0.74 ± 0.07, OM : OC = 2.12 ± 0.10), with TR samples characterized
by intermediate values (H : C = 1.45 ± 0.09, O : C = 0.72 ± 0.07,
OM : OC = 2.08 ± 0.09), consistent with the average elemental ratios
discussed in Sect. 3.1. The differences between the average elemental
ratios of PBL, TR and FT are statistically significant according to the
t-test (p<0.01). The large standard deviations associated to the
mean values are due to day-by-day variations, as showed by Fig. S6. Figure
S6 also shows that, independently of
the day-by-day variations, O : C is
systematically higher, and H : C systematically lower, in the FT compared to
the PBL air masses, while the TR air masses have intermediate values.
Contribution of organic fragments containing only carbon and
hydrogen (CxHy), organic fragments containing carbon, hydrogen and
one oxygen atom (CxHyO1) and organic fragments containing
carbon, hydrogen and more than one oxygen atom (CxHyOz>1) in
PBL, TR and FT samples.
Figures 5 shows the whole campaign data points in a Van Krevelen diagram
(Van Krevelen, 1950), together with the average H : C and O : C ratios of PBL,
TR and FT samples. The data are lumped within the region delimited by O : C
between 0.5 and 1 and H : C between 1.1 and 1.5. The plot illustrates the
process of OA atmospheric oxidation in the investigated area characterized
by a slope comprised between -0.5 and -1. According to Heald et al. (2010)
and Ng et al. (2011), such an intermediate slope can result from a
combination of reactions adding carboxylic acids to the OA, occurring both
with and without fragmentation of the parent molecules (expected slopes =-0.5 and -1, respectively). Kroll et al. (2011) reported that fragmentation
becomes increasingly important for already oxidized material undergoing
further processing. This can explain the results at Mt. Cimone, where OA
does not resemble recently formed secondary material, in analogy with Ng et
al. (2011). An alternative explanation for a slope tending to -0.5 is given
by equivalent amounts of addition of carboxylic groups and of hydroxyls or
peroxides (Ng et al., 2011).
Size distribution of (a) organics, (b) nitrate, (c) sulfate and
(d) ammonium in PBL, TR and FT samples. Log-normal fits are reported as black
lines.
Pearson correlation coefficients (R) between the time series of the
three PMF factors and several gas-phase and particle tracers measured at Mt. Cimone. T= air temperature,
P= atmospheric pressure, RH = relative humidity, WS = wind speed, UVB = UVB radiation flux.
BC
CO
T
P
RH
WS
UVB
O3
NOx
Nitrate
Sulfate
Ammonium
OOAa
0.54
0.71
0.48
0.42
0.22
-0.23
0.12
0.45
0.70
0.55
0.48
0.57
OOAb
0.35
0.27
0.36
0.25
-0.01
0.02
0.09
0.28
0.18
0.09
0.64
0.43
OOAc
0.41
0.24
0.16
0.26
-0.12
0.12
-0.10
0.58
0.02
0.12
0.49
0.38
The addition of carboxylic functionalities during the OA ageing process is
confirmed by the analysis of the high-resolution mass fragments, showing
that from PBL to FT samples, the average contribution of
CxHyOz>1, attributed to the fragmentation of carboxylic
structures (Aiken et al., 2007; Takegawa et al., 2007; Duplissy et al.,
2011), increases from 35 to 43 %, while both CxHy and
CxHyO fragments decrease (Fig. 6). The mean elemental
compositions calculated for PBL, TR and FT samples fall at distinct
positions along the line of average ageing in the Van Krevelen space (Fig. 5), suggesting that the observed oxidation of OA is dictated by the
different age of the aerosols reaching the station at different times of the
day, as also evidenced by Fig. S6.
Finally, Fig. 7 shows the time-of-flight particle size distributions
(pToF) of the main NR-PM1 components as measured by the HR-ToF-AMS
observed during the three regimes (PBL, TR and RL). For each species, there
is no appreciable size distribution difference between the three sample
subsets (PBL, TR and FT). However, the pToF of organics peak at a slightly
lower dva (293 nm) compared to sulfate, ammonium and nitrate (which
peak between 330 and 340 nm dva), as determined by a log-normal fit of
the size distributions. This result indicates that the organic and inorganic
components in all the sampled air masses are not entirely internally mixed.
Furthermore, the pToF of organics show a tail towards smaller particles sizes
reaching 90 nm, and appreciable amount of mass below 200 nm dva, which
is consistent with previous observations of pToF of organics typically having
lower dva than e.g. SO4, due to different sources and formation
processes. It is possible that a fraction of the organics observed at
dva< 200 nm arises from the growth of smaller
particles (fresh emission typically peak at dva of 80–100 nm) via
condensation processes during transport to high altitude.
OA source apportionment by PMF
In order to further characterize the OA collected at Mt. Cimone, PMF was
applied to the high-resolution OA mass spectra. We screened various
solutions with a number of factors from two to ten. A four-factor solution with
rotational forcing parameter fpeak=0(Q/Qexp=2.3) was chosen,
yielding four different types of OOA, two of which were recombined into one
factor, because of coincident time series and profiles (Fig. S6). The OA
components from the PMF analysis were identified by their mass spectra,
elemental composition (Fig. 8) and diurnal cycles (Fig. 9), as well as by
correlations of their time series with tracers (Table 2). A detailed summary
of key diagnostic plots of the PMF results and a discussion of the factor
solution choices can be found in the Supplement.
High-resolution mass spectra of the three factors extracted by
PMF. Insets in each plot report the results of the elemental analysis (from
the I-A method).
The three resulting factors are all of the oxygenated organic aerosol (OOA)
type and have been defined as OOAa, OOAb and OOAc. No “standard” hydrocarbon
like (HOA) factor (i.e. with m/z43≫m/z44 and
with a significant amount of hydrocarbon-like ions, CxHy) could be
extracted by PMF, similarly to other AMS data sets collected at background
sites (Hildebrandt et al., 2010; Freney et al., 2011), indicating almost no
direct influence of freshly emitted primary aerosols to the observed OA
load. This result is consistent with the highly oxidized character of the
OA, as described previously. Even though m/z 44 (CO2+) dominates
the mass spectra of all the three factors, the OOAa factor has a slightly
higher amount of CxHy ions at m/z 27 (C2H3+), 39
(C3H3+), 41 (C3H5+) compared to the other
factors.
The elemental composition (H : C, O : C) and the OM : OC ratio are also reported
in Fig. 8. OOAc is the most oxidized factor, with an OM : OC ratio of 2.48
vs. 2.03 and 2.13 of OOAa and OOAb, respectively. Consequently, it has
higher O : C (1.02) and lower H : C (1.07) ratios than OOAa (0.67; 1.51) and OOAb
(0.75; 1.44). OOAa average concentration was 1.5 ± 1.7 µg m-3 during the campaign, which accounts for 55 % of the OA. OOAb
presented higher concentrations during the period 17–23 June (2.1 ± 1.5 µg m-3) and low concentrations during the rest of the campaign
(0.27 ± 0.29 µg m-3), for an average concentration of
0.67 ± 1.1 µg m-3 and a contribution of 25 %. The
concentration of OOAc was 0.54 ± 0.40 µg m-3, contributing
20 % on average to the OA.
Time series and diurnal trend of the three factors extracted by
PMF.
When looking at the diurnal profiles of the three factors (Fig. 9), we see
clear diurnal cycles for OOAa and OOAc (but with opposite trends, OOAa having
a maximum at 16:00, and OOAc having a minimum at 14:00) and a less
pronounced diurnal profile for OOAb (slightly higher concentration around
12:00–14:00 than during the rest of the day). The fact that OOAa
concentration is much higher during the day than at night indicates that
this factor derives from sources located within the PBL and is transported
at Mt. Cimone by convection and by thermal winds in daytime, as discussed
above. Conversely, the trend for OOAc with lower concentrations during the
day points to a transport from the free troposphere, or in any case from above
the PBL.
The Pearson correlation coefficients (R) between the three PMF factors and
several external gas-phase and particle tracers are reported in Table 2.
OOAa correlates best with CO, NOx and BC (0.71, 0.70 and 0.54,
respectively), which are attributable to anthropogenic sources located
within the PBL, confirming our interpretation of the OOAa source location.
OOAb shows lower correlation than OOAa against all the tracers, with higher
R values associated with the above enlisted PBL tracers (CO, NOx and
BC), suggesting that this factor was contributed by PBL sources too.
Interestingly, OOAb presents the highest correlation with sulfate, which
suggests a regional character for this OOA component. The highest
correlation for OOAc is with O3, with a very small correlation with all
the PBL tracers. The high correlation of OOAc with O3 is mainly driven
by the coincident daily trends (Fig. S11), showing higher concentration at
night, a typical feature of O3 at high-altitude sites (Fischer et al.,
2003; Cristofanelli et al., 2007). In fact, during summer, air masses richer
in photochemically produced O3 are vented to Mt. Cimone in the
afternoon, leading to an increase of O3 until evening, when O3
observations start to be more representative of the free troposphere.
Pearson correlation coefficients (R) between the profiles of the
three PMF factors and reference high-resolution factor profiles found in the
literature.
Saarikoski et al. (2012)
Mohr et al. (2012)
Crippa et al. (2013)
OOAa
OOAb
OOAc
HOA
BBOA
NOA
BBOA
HOA
LVOA
SVOA
SVOOA
LVOOA
HOA
OOAa
0.93
0.92
0.90
0.75
0.38
0.40
0.62
0.19
0.87
0.81
0.86
0.93
0.50
OOAb
0.94
0.92
0.90
0.70
0.30
0.34
0.61
0.17
0.91
0.81
0.81
0.93
0.45
OOAc
0.97
0.94
0.97
0.58
0.21
0.10
0.31
0.07
0.95
0.76
0.59
0.94
0.32
CO / NOx ratio vs. SH, colour-coded by the contribution of
factors (a) OOAa, (b) OOAb and (c) OOAc. In panel (b) only the data points
corresponding to OOAb maximum contribution period (17–23 June) have been
coloured, to make the plot clearer.
The spectral and elemental features of the three OOA factors are in the
range of others reported in the literature (Mohr et al., 2012; Holzinger et
al., 2013; Sun et al., 2011a). Similar OOA spectra have also been described
by Saarikoski et al. (2012) for the Po Valley site of San Pietro Capofiume.
Correlation analysis with reference high-resolution spectra (Table 3)
suggests that all the OOA components identified can be classified as LV-OOA
(low volatile-OOA). However, OOAa and OOAb present qualitatively similar
spectral features to the SV-OOA (semi volatile-OOA) reported by Freney et
al. (2011), for a similar high-altitude station, and by Hayes et al. (2013),
for photochemically aged aerosol. The f44 (contribution of organic mass
fragments with m/z 44) of OOAa and OOAb is 0.13 and 0.14, respectively,
which is across the SV-OOA/LV-OOA region proposed by Cubison et al. (2011).
Therefore, we believe that OOAa and OOAb can be considered either as highly
oxidized SV-OOA or LV-OOA with a low oxidation level.
In summary, OOAa is clearly attributed to sources or formation processes
located within the PBL that reach the station when vertical transport (PBL
convection and valley/upslope breeze) is maximized. OOAa is the less
oxidized factor retrieved by PMF and likely represents a moderately aged
local OA. OOAb has a less defined diurnal cycle (still with maximum during
the day) and a slightly higher O : C ratio than OOAa. The period of OOAb
maximum contribution coincides with meteorological conditions characterized
by reduced horizontal air motion and dominated by breeze regimes (17–23 June, when an anticyclonic high-pressure system was present over northern
Italy). These conditions favour the accumulation of pollutants within the
PBL and in the residual layers above, because of reduced air circulation.
Collaud Coen et al. (2011) demonstrated that under such meteorological
conditions, air masses from residual layers continue to influence the
Jungfraujoch high-altitude station also during the night, leading to higher
minima in the diurnal aerosol concentration than in other conditions. We
postulate that the same happens at Mt. Cimone. The entire region influenced
by the high-pressure system extends beyond the Po Valley basin, comprising
the great Alpine region. Over the Alps, orographic lifting of PBL air is
responsible for the formation of residual layers at very high altitudes (the
Jungfraujoch is at 3571 m a.s.l.) which can then travel on the top of the
PBL over the surrounding basins (like the Po Valley). We therefore
hypothesize that factor OOAb is a regional component of NR-PM1,
associated with the accumulation and ageing of OA in residual layers when
wind speeds are small throughout the lower troposphere during the period of
enhanced high-pressure conditions. This is also supported by the good
correlation between OOAb and regional sulfate.
Finally, OOAc is the product of prolonged atmospheric processing of OA,
occurring mainly in the free troposphere, and can be considered as
representative of the background FT OA on a spatial scale that comprises all
western European areas upwind of Mt. Cimone. These conclusions are confirmed
by the plots in Fig. 10, showing CO / NOx vs. SH, colour-coded by the
contribution of each factor. Clearly, OOAa is associated with air masses
characterized by reduced photochemical age (low CO / NOx) and strongly
influenced by the PBL (high SH), while OOAc contributes more in air masses
characterized by high photochemical age (high CO / NOx) and less
influenced by the PBL (low SH). OOAb presents intermediate characteristics
(mid- to low-CO / NOx and intermediate SH), consistent with the hypothesis
that OOAb is representative of OA of intermediate age that is accumulated in
the residual layers during the period of high pressure, due to the stagnant
atmospheric conditions.
Conclusions
The chemical composition of non-refractory submicrometric particles was
measured for the first time by a HR-ToF-AMS at the Mt. Cimone GAW/WMO high-altitude
station. Submicrometric aerosol was dominated by the organic
fraction (on average 63 %), with ammonium sulfate as the second contributor,
for an average NR-PM1 mass of 4.5 ± 3.4 µg m-3.
Elemental analysis of the high-resolution AMS data showed highly oxygenated
OA (the campaign-average H : C, O : C and OM : OC were 1.45 ± 0.11,
0.71 ± 0.08 and 2.08 ± 0.10, respectively), suggesting that strong
oxidation and SOA formation processes occur during aerosol transport to high
altitudes. Different stages of OA processing could be identified when
comparing the OA composition during the day, when the station was affected
by the upward transport of PBL air, and the night-time, when the site was in
the free troposphere (FT).
Analysis of the OA elemental ratios in a Van Krevelen space showed that OA
oxidation followed a slope comprised between -0.5 and -1, consistent with
the addition of carboxylic groups to alkyl structures, occurring both with
and without fragmentation of the reagent molecules. The increase of
carboxylic groups during OA processing and ageing is confirmed by the
increased contribution of CxHyOz>1 fragments during
night-time measurements.
Quantitative information on the contributions of more- and less-aged OA
components were achieved by analysing the high-resolution AMS data by
positive matrix factorization (PMF). OOAa (55 %), the least oxidized OOA
factor, was related to sources or formation processes located within the
PBL, reaching the station mainly during the day, when vertical transport is
maximized. OOAb (25 %) was attributed to the accumulation and ageing of OA
in the PBL and in the residual layers above the PBL, due to stagnation over
the great Alpine region. Finally, OOAc (20 %) was interpreted as the
product of prolonged atmospheric processing of OA occurring mainly above the
PBL, and can be considered as representative of background free tropospheric
OA at a continental scale.
This work highlights the important contribution of organic aerosols to the
composition of submicron particles at remote mountain sites. We found that
63 % of the NR-PM1 mass that constitutes the background aerosol
levels for the Po Valley in the summer is accounted for by highly oxygenated
organic matter. No important contribution from primary combustion organic
particles (HOA) was measured, indicating that these compounds were likely
lost during transport, either by evaporation or chemical processing. Most
importantly, in spite of the vicinity of strongly emitting pollution sources
in the Po Valley, only 55 % of the organic matter measured at Mt. Cimone
in the summer could be attributed to sources within the PBL, while the
remaining fraction (45 %) is accounted for by remote upwind sources. This
study confirms the importance of regional-scale physical and chemical
processes and of transboundary transport in determining the background
aerosol composition at rural European sites.