Introduction
NO and NO2 (NOx) are important catalysts in the
photochemical production of ozone (O3) playing a significant role
in the oxidation of volatile organic compounds (VOCs) and subsequently have
an adverse effect on air quality. In the daytime, NOx is
primarily removed by the hydroxyl radical (OH) to form nitric acid
(HNO3), which is subsequently lost by wet deposition, becoming a
major component of acid rain. At night, the OH radical is not a significant
oxidant as photolysis stops, enabling the reaction between NO2 and
O3 to form significant levels of the nitrate radical
(NO3) (Atkinson, 2000). NO3 can accumulate at night or further
react with NO2, leading to the formation of N2O5
(Brown et al., 2003; Brown and Stutz, 2012). This equilibrium can lead to the
reaction of NO3 with VOCs at night forming organic nitrates or act
as an important intermediate for heterogeneous reaction on aerosols as
N2O5 produces NO3- and NO2+ in the aqueous phase
(Hallquist et al., 1999, 2000; Wagner et al., 2013). In the presence of
chlorine, which is assumed in models to predominantly come from sea salt
(Baker et al., 2016), nitryl chloride (ClNO2) can be formed and
released into the gas phase from the aerosol surface (Osthoff et al.,
2008). ClNO2 formation thereafter acts as a nighttime radical
reservoir due to its stability at night.
At sunrise, ClNO2 is rapidly photolysed, liberating the highly
reactive chlorine atom subsequently converting it into ClO,
HOCl and ClONO2 depending on the available sunlight,
O3, HOx and NOx levels via
the following reaction pathways (R1–R11).
ClNO2+hν→Cl+NO2Cl+O3→ClO+O2ClO+NO2→ClONO2ClO+HO2→HOCl+O2ClONO2+hν→Cl+NO3ClONO2+hν→ClO+NO2HOCl+hν→Cl+OHClONO2+H++Cl-→Cl2+HNO3HOCl+H++Cl-→Cl2+H2OClO+NO→Cl+NO2OH+HCl→Cl+H2O
The liberated chlorine will predominantly react with VOCs, with the pathways
listed (Reactions R2–R11) representing alternative routes to loss of the chloride
radical, and contribute to daytime photochemical oxidation, competing with OH
and perturbing standard organic peroxy radical abundance
(ROx = OH + HO2 + RO2),
O3 production rate, NOx lifetime and
partitioning between reactive forms of nitrogen (Riedel et al., 2014). The
rate constants for the reaction of chlorine atoms with a number of VOCs is
around 200 times larger than the equivalent reaction with OH (Tanaka et al.,
2003); therefore, its abundance, fate and cycling can significantly alter
standard daytime oxidation pathways. The oxidation of VOCs by chlorine atoms
is thought to be significant in the early hours of the day while OH mixing
ratios are low and chlorine atom production is high through the photolysis of
ClNO2, as well as feeding into the standard
HOx / NOx cycles via production of peroxy
radicals from reactions with alkanes. Additional Cl2 photolysis and HCl
reaction with OH can also produce chlorine atoms throughout the day but at
lower rates.
Saturated hydrocarbons are usually oxidised by reaction with OH or a chlorine
atom to form an organic peroxy radical (RO2), and H2O or HCl
depending on the oxidant (Reactions R12 and R13), which is the dominant pathway for
chloride–VOC (Cl–VOC) reactions. In a heavily polluted environment such as Beijing,
the RO2 favours further reactions with NO to form an oxygenated volatile
organic compound (HO2 and NO2) or an alkyl nitrate
(RONO2). Specifically, acyl peroxy radicals can also react with
NO2 to form acyl peroxy nitrates (APNs) such as peroxy acetyl
nitrate (PAN).
RH+OH⟶O2RO2+H2ORH+Cl⟶O2RO2+HCl
RO2+NO→OVOC+HO2+NO2RO2+NO→RONO2
RO2+NO2↔APN
Addition of the chlorine atom to an unsaturated VOC can also occur
and then continue on the similar reaction pathway as denoted by
Reactions (R12)–(R15). These pathways result in the production of unique
chlorine atom chemistry markers which have been previously investigated to
indicate the extent of chlorine atom oxidation reactions (Riemer et al.,
2008; Keil and Shepson, 2006). The utilisation of these compounds, such as
2-chloro-peroxypropionyl nitrate (2-Cl PPN) and
1-chloro-3-methyl-3-butene-2-one (CMBO), as chlorine atom chemistry markers
relies on the abundance of the chlorine atom, the VOC precursor,
HOx, NOx and O3 and competing
pathways for chlorine atom reactions. Riedel et al. (2014) calculated that
up to 10 parts per trillion (ppt) Cl–VOCs are formed as a result of chlorine
atom addition to alkenes and can therefore provide a number of potential
periods of dominating active Cl chemistry (Wang et al., 2001).
The production of chloroperoxy radicals via chlorine atom addition can lead
to the formation of semivolatile oxidation products which have been observed
for both biogenic (Cai and Griffin et al., 2008) and anthropogenic
emissions (Huang et al., 2014; Riva et al., 2015) in controlled
laboratory studies. Chlorine-initiated oxidation of isoprene could also
represent a significant oxidation pathway due to its rapid reaction rate
compared with OH (Orlando et al., 2003) resulting in gas-phase products
such as chloroacetaldehyde and CMBO, a unique tracer for atmospheric chlorine
atom chemistry (Nordmeyer et al., 1997). Furthermore, reactions of the
chlorine atom with isoprene or its secondary organic aerosol (SOA) derived
products could serve as an atmospheric chlorine sink (Ofner et al., 2012).
D. Wang et al. (2017) revealed chlorine-initiated oxidation of isoprene can
produce SOA yields up to 36 %, with products similar to that of OH
isoprene oxidation, compared with the 15 % yield from standard oxidation
calculated by Liu et al. (2016), although this is known to be a factor of 2
higher than utilised in standard climate models. This SOA formation from
chlorine-initiated oxidation presents a large knowledge gap in the
literature, which to date is limited by measurement capabilities.
This complex system results in a large uncertainty in the global budget of
chlorine atoms of ∼ 15–40 Tg Cl yr-1 calculated by indirect
means (Allan et al., 2007; Platt et al., 2004), which is further limited
by the ability of measurement techniques to accurately quantify short-lived
species at low mixing ratios. Our knowledge of the Cl budget therefore
depends on the accurate measurement of its precursors, namely ClNO2
and major reaction pathways of the chlorine atom upon liberation in the
daytime. Measurements to date show that the mixing ratio of ClNO2
vary geographically from below limits of detection to hundreds of ppt (Mielke
et al., 2015; Phillips et al., 2012; Bannan et al., 2015) and up to 3
parts per billion (ppb) (Tham et al., 2014; Riedel et al., 2014; Liu et
al., 2017) in heavily polluted urban areas. To date, the majority of these
measurements have been performed in the United States, with more recent
research in Europe, China, etc. (Tham et al., 2014; Wang et al., 2016; X.
Wang et al., 2017; Z. Wang et al., 2017; Liu et al., 2017). A major
factor in the variation of ClNO2 mixing ratios is the availability
and abundance of aerosol chloride which can vary significantly, although it
is predominantly present as sodium chloride from sea salt.
Iodide adduct ionisation has previously been applied to measure inorganic
halogens in ambient air (Osthoff et al., 2008; Riedel et al., 2012;
Thornton et al., 2010; Le Breton et al., 2017) using mass spectrometers
with quadrupole mass analysers. This technique involves periodically changing
the tuning of the spectrometer to allow transmission of a particular mass ion
to the detector. Several species are therefore often “chosen” for detection
in order to achieve high-enough time resolution. Recent developments and
availabilities of a time-of-flight chemical ionisation mass
spectrometer (ToF-CIMS) have enabled the simultaneous measurement of all
detectable ions by an ionisation technique via high-frequency full mass
spectral collection. The high resolving power (3500) of this technique also
enables much lower limits of detection for species which may have the similar
mass to a compound that is much more abundant via multi-peak fitting. This
technique has previously been applied for the measurement of ClNO2
and Cl2 (Faxon et al., 2015) and recently for Cl–VOCs (D. Wang et
al., 2017) in the gas phase. In this study, a ToF-CIMS utilising the Filter
Inlet for Gas and AEROsols (FIGAERO) is deployed at a site in semi-rural
Beijing, China, to measure the gas- and particle-phase precursors
(ClNO2, N2O5) and selective halogen-containing species
at high time frequency and resolving power to further our understanding of
the chlorine atom budget in this region and its potential fate.
Experimental
Site description
The data presented here were collected during the inter-collaborative field
campaign, within the framework of a Sino-Swedish research project
(“Photochemical Smog in China”) aimed to further our understanding of the
episodic pollution events in China through gas- and particle-phase
measurements with numerous analytical instruments. The laboratory setup on
the Changping University campus of Peking University was situated at a semi-rural site
40 km northwest of Beijing close to Changping (40.2207∘ N,
116.2312∘ E). The general setup has previously been described by Le
Breton et al. (2018).
All instruments sampled from inlets set up in a laboratory 12 m high from
13 May to 23 June 2016. The site has a small town within its vicinity and
some small factories within 5 km. A high-resolution time-of-flight aerosol
mass spectrometer (HR-ToF-AMS) was utilised to measure the mass mixing ratios
and size distributions of nonrefractory species in submicron aerosols,
including organics, sulfate, ammonium and chloride (DeCarlo et al., 2006;
Hu et al., 2013). The setup of this instrument has been previously
described by Hu et al. (2016). Photolysis rates were measured by a
commercial spectroradiometer for O3, NO2, HCHO, HONO and
H2O2 (Metcon UF CCD); the instrument was calibrated by a high-power
halogen lamp after the field campaign. The photolysis rates of other related
species were scaled by the recommendation of the Jet Propulsion
Laboratory (JPL) kinetic evaluation report (Burkholder et al., 2015). Before
the campaign, the instrument was calibrated through comparison with a chemical
actinometer utilised in 2014 (Zou et al., 2016), agreeing within 10 %.
The surface albedo is normally 0.05 at the ground near the site. Upwelling
radiation is neglected as is represents an insignificant fraction of the
downwelling values.
An Ionicon Analytik high-sensitivity proton transfer mass
spectrometer (PTR-MS) as described by de Gouw and Warneke et al. (2007) provided
supporting precursor VOC measurements. Detailed information about the PTR-MS
measurements can be found in Yuan et al. (2012, 2013). In brief, 28 masses
are measured throughout the campaign at 1 Hz. Zero air, which was produced
by ambient air passing through a platinum catalytic converter at
350 ∘C (Shimadzu Inc., Japan), was measured for 15 min every 2.5 h
to determine the background. Aromatic masses (m/z 79 for benzene, m/z 93
for toluene, m/z 105 for styrene, m/z 107 for C8 aromatics and m/z 121
for C9 aromatics), oxygenated masses (m/z 33 for methanol, m/z 45 for
acetaldehyde, m/z 59 for acetone, m/z 71 for methyl vinyl ketone plus methacrolein and m/z 73 for
methyl ethyl ketone), isoprene (m/z 69) and acetonitrile (m/z 42) were calibrated by
using EPA TO15 standard from Apel-Riemer Environmental Inc., USA. Formic acid
(m/z 47), acetic acid (m/z 61), formaldehyde (m/z 31) and monoterpenes
(m/z 81 and m/z 137) were calibrated by permeation tubes (VICI, USA). The
uncertainties of most species are below 10 %, which is detailed in the
previous work (Liu et al., 2015).
ToF-CIMS setup
Gas- and particle-phase ambient species were measured using an iodide ToF-CIMS
(resolving power of 3500) coupled to the FIGAERO inlet (Lopez-Hilfiker et
al., 2014). The setup for this campaign has previously been described by Le
Breton et al. (2018). Briefly, the iodide ionisation scheme was utilised to
acquire non-fragmented ions of interest by passing ultra-high-purity N2 over a
permeation tube containing liquid CH3I (Alfa Aesar, 99 %) and
through a Tofwerk X-ray ion source type P (operated at 9.5 kV and
150 µA) to produce the iodide reagent ions. The ionised gas was then carried out
of the ion source and into the ion-molecule reaction (IMR) chamber, which was
heated to 40 ∘C to reduce wall loss, through an orifice
(Ø = 1 µm). The inlet lines were 2 m long and composed of
copper tubing (12 mm) for the aerosol inlet and Teflon tubing (12 mm) for
the gas sample line. Particles were collected onto a
Zefluor® polytetrafluoroethylene membrane filter at the same
rate as the gas inlet line sampling (2 slm).
FIGAERO was operated in a cyclic pattern; 25 min of gas-phase
measurement and simultaneous particle collection, followed by a 20 min
period during which the filter was shifted into position over the IMR inlet
and the collected particle mass was desorbed.
Calibration
In the field, formic acid calibrations were performed daily, utilising a
permeation source maintained at 40 ∘C. A dry N2 flow
(200 sccm) was passed over the permeation source and joined a 2 slm N2
flow line directed towards the inlet. The mixing ratio of the flow was
determined by mass loss in the laboratory after the campaign. The sensitivity
of the ToF-CIMS to formic acid was found to be 3.4 ion counts
per ppt Hz-1 for 1 × 105 iodide ion counts.
N2O5 was synthesised by mixing 20 ppm O3 with pure
NO2 (98 %, AGA gas) in a glass vessel and then passing the mixture
through a cold trap held at -78.5 ∘C by dry ice. The
N2O5 was transferred to a diffusion vial fitted with a capillary
tube (i.d. 2 mm). The N2O5 diffusion source was held at a
constant temperature (-23 ∘C), and the mass loss rate was
characterised gravimetrically for a flow rate of 100 sccm. The same flow was
added to a dry nitrogen inlet dilution flow of 2 slm to calibrate the CIMS.
ClNO2 measurements were quantified by passing the N2O5
over a wetted NaCl bed to produce ClNO2. The decrease in
N2O5 from the reaction with NaCl was assumed to be equal to the
mixing ratio of ClNO2 produced (i.e. a 100 % yield).
Conversion of N2O5 to ClNO2 can be as efficient as
100 % on sea salt, but it can also be lower, for example, if
ClNO2 were to convert to Cl2 (Roberts et al., 2008). For NaCl,
the conversion efficiency has however been as low as 60 % (Hoffman et
al., 2003). In this calibration, we have followed the accepted methods of
Osthoff et al. (2008) and Kercher et al. (2009) that show a conversion yield
of 100 % and have assumed this yield in the calibrations of this study.
The lower detection limit of the CIMS to N2O5 and
ClNO2 was found to be 9.5 and 1.2 ppt, respectively, for 1 min
averaged data. The error in the individual slope of the calibration results
yields a total uncertainty of 30 % for both N2O5 and
ClNO2. These sensitivities for N2O5 and
ClNO2 (9.8 and 1.6 ion counts per ppt Hz-1 for
1 × 105 iodide ion counts) were applied relatively to that of
formic acid. The other inorganic halogens reported in this work are reported
in ion counts. Other acids identified by CIMS which are reported in the
literature are given the sensitivity of N2O5 to provide a
minimum concentration so no concentrations are overestimated.
(a) Average mass spectrum for the whole measured
range. (b) Average mass spectrum for the region that contains
all gas-phase nighttime species utilised in this work. A high-resolution
spectral fit for ClO and ClNO2 is displayed with corresponding
multi-peaks with 0.5 atomic mass units (AMU) (c and d). The black line
represents the total fit from all peaks. The grey line represents the mass
spectral raw data.
CIMS and cavity enhanced absorption spectrometer (CEAS) 1 min averaged data of N2O5 with
corresponding correlation plot (a), campaign and diurnal
deviation (panels b and c, respectively). The red
highlighted periods represent data collected on 3 June where a different
correlation gradient was observed between CIMS and CEAS. The box-and-whisker
plot represents the diurnal difference for the campaign between the CEAS and
CIMS measurements (d). Panel (c) shows the y intercept of
the line of best fit and M is the gradient.
A post campaign calibration of chloroacetic acid (99 %, Sigma Aldrich)
was utilised to characterise a sensitivity factor for a Cl–VOC. The
calibration was performed using the same method as for formic acid and gave a
sensitivity of 1.02 ion counts ppt-1 Hz when normalised to
1 × 105 I- ion counts. This similar sensitivity to that of
the Cl–VOC to that of ClNO2 could imply a relative sensitivity may
be appropriate to constrain the mixing ratios of all Cl–VOCs, although
further work is required to confirm this, and therefore the paper reports
all Cl–VOC measurements in units of ion counts.
Model setup
The European Monitoring and Evaluation Programme (EMEP) Meteorological
Synthesizing Centre – West (MSC-W) chemical transport model (Simpson et al.,
2012, 2017) driven by meteorology from the Weather Research and Forecasting
Advanced Research (WRF-ARW) model (Skamarock et al., 2008) was utilised to
support source analysis of the particulate chloride. The model was run on two
nested domains (0.5 and 0.1667∘ resolution, respectively) with
biomass burning emissions from the two databases, the Fire Inventory from
NCAR (FINN) and Global Fire Assimilation System (GFAS), and anthropogenic
emissions from the Multi-resolution Emission Inventory for China (MEIC;
http://meicmodel.org/, last access: 1 March 2018). Two versions of the
model, one getting emissions from open biomass burning from FINN (Wiedinmyer
et al., 2011) and one getting them from GFAS (Kaiser et al., 2012), were run
for the entire period of the Changping measurement campaign.
Results and discussion
Peak identification and quantification
Peak fitting was performed utilising the Tofware peak fitting software for
molecular weights up to 620 AMU. The standard peak shape was fitted a peak
on the spectra until the residual was less than 5 %. Each unknown peak
was assigned a chemical formula using the peak's exact mass maxima to five
decimal places and also isotopic ratios of subsequent minor peaks. An
accurate fitting was characterised by a ppm error of less than 5 and
subsequent accurate fitting of isotopic peaks. The analysis here focuses on
species identified in the mass spectra considered to possibly play important
roles with respect to the nighttime chlorine reservoir and several other key
nighttime oxidants: ClNO2, HCl, Cl2, ClO, HOCl, OClO,
ClONO2, N2O5 and Cl–VOCs. Figure 1 displays the average
mass spectra for the measurement campaign and the peak fitting applied for
ClO and ClNO2. All species were represented by a dominant peak with
a multi-peak fit, although a number of coexisting peaks were present for
much of the campaign. This signifies the importance of high-resolution fit
data and the need for high-resolution measurements. A quadrupole CIMS may not
be able to resolve the peak adjacent to ClO at m/z 178 (dominant peak is
IC6F3HO3-) and the second dominant peak for the
ClNO2 fit (cluster of HNO3 with water) would result in a
10 % overestimation.
N2O5 measurements
The CIMS and a cavity enhanced absorption spectrometer (CEAS) measured
N2O5 (H. Wang et al., 2017) simultaneously from 13 May 2016 to
6 June 2016. However, given the use of FIGAERO, the CIMS alternated
measurements between gas and particle phases and thus did not generate a
completely continuous gas-phase time series. Here, the CEAS is utilised to
validate the CIMS N2O5 (at m/z 235) measurements and also
instrument stability. The CEAS utilised a dynamic source by mixing
NO2 and O3 to generate stable N2O5 for
calibration (H. Wang et al., 2017). The source was used to calibrate the
ambient sampling loss of N2O5 in the sampling line, filter, the
pre-heater cavity and optical cavity. This was performed pre- and
post-campaign. During the campaign, the reflectivity of the high-reflectivity
mirror was calibrated daily and the filter changed hourly. The simultaneous
measurements of N2O5 can be shown in Fig. 2 for 1 min averaged
data. The time series shows good agreement for both background mixing ratios
during the day (below 10 ppt) and high nighttime mixing ratios (up to
800 ppt), excluding one night. The highest N2O5 levels observed
by both the CEAS and CIMS were observed on 3 June, although the CEAS reports
880 ppt, whereas the CIMS reports 580 ppt. If included in the analysis, the
R2 is 0.71 and when excluded it is 0.76. To date, the reason for this
deviation during that night is not known, but it should be stressed that
N2O5 measurements are delicate and highly dependent on sampling
condition, e.g. the RH. Nevertheless, excluding this night from the
comparison, a slope of 0.85 is observed and an offset of 0.9 ppt. The
diurnal profile in Fig. 2 represented the difference between the two
measurements throughout the campaign. The largest error between the two
measurements occurs at night during the higher levels of N2O5,
although averaging at 4 ppt (representing 11 % error on the average
campaign concentration). Differences could arise from a number of various
factors. Inlet differences such as the CIMS-heated IMR (to 40 ∘C to
reduce wall loss), residence time and ambient NO2 can all change
thermal decomposition and wall loss rates between the instruments, which is
determined for the CEAS in H. Wang et al. (2017) but not for the CIMS in
this work. Also, the separate inlets were facing in different directions
within the same laboratory, possibly enabling local wind patterns to affect
the mixing ratios reaching each instrument.
Mean diurnal profiles of the inorganic halogens detected by the CIMS
from the 23 May to 6 June with average J rate for ClNO2 as guide
for photolysis. ClNO2 and HCl mixing ratios are on the left
y axis and the other inorganic halogens on the right y axis displayed in
ion counts.
The CEAS data were further utilised to assess any sensitivity changes for the
CIMS that daily carboxylic acid calibrations did not account for. A time
series of hourly factor differences between the CIMS and CEAS was implemented
into these data to weight the measurements to a normalised sensitivity. The
high level of agreement (R2 of 0.76) from low mixing ratio measurements
and a species with a short lifetime from different inlets confirms the
accuracy and reliability of the CIMS measurements for this campaign.
Generally, N2O5 was detected throughout the campaign with a
clear diurnal variation peaking at night and rapidly falling to below limits
of detection in the daytime as a result of photolysis of N2O5
and NO3. The campaign mean nighttime mixing ratio was 121 ppt with
a standard deviation of 76 ppt. The maximum mixing ratio of
N2O5 observed was 880 ppt on 3 June. This range of mixing
ratios lies within the recently reported values in the literature but not at
the extreme mixing ratios as observed in Germany (2.5 ppb) (Phillips et
al., 2016) or Hong Kong (7.7 ppb) by Wang et al. (2016). Although the
mean mixing ratios do not increase significantly during the pollution
episodes, the maximum mixing ratios detected overnight increase by up to a
factor of 4. Further analysis of N2O5 nighttime chemistry was
performed by Wang et al. (2018), who calculated an average steady-state
lifetime of 310 ± 240 s and mean uptake coefficient of
0.034 ± 0.018.
Inorganic chlorine: abundance, profiles and source
Abundance and profiles
Mean diurnal profiles of HCl, Cl2, ClONO2, HOCl, ClO and
ClNO2 are displayed in Fig. 3 from data between 23 May and 6 June.
HCl exhibited a standard diurnal profile increasing in mixing ratio
throughout the day and peaking at 16:00 LT
which then fell off slowly at night. The mean HCl campaign mixing ratio was
510 ppt (standard deviation (σ) 270 ppt) and the maximum HCl mixing
ratio was 1360 ppt on 30 June. Cl2 exhibited a diurnal profile peaking
in both nighttime and daytime. High mixing ratios were observed at night
followed by a sharp loss at sunrise and a general build-up throughout the
day. The campaign mean mixing ratio was 0.65 ppt (σ 0.5 ppt) and the
maximum mixing ratio was 4.2 ppt on 4 June just before midnight. This agrees
well with recent urban measurements of Cl2 in the US where Faxon et
al. (2015) observed a maximum of 3.5 ppt and Finley et al. (2006) observed
up to 20 ppt in California. Up to 500 ppt Cl2 have recently been
reported in the Wangdu County, southwest of Beijing (Liu et al., 2017).
Although the mixing ratios we report here are significantly lower, as
detailed later, their source may be of similar origin, which is indicated to
be from power plant emissions.
The diurnal profile of HOCl peaked during the daytime via its main formation
pathways via reaction of ClO and HO2 and Cl with OH. Interestingly,
the ClO in this work exhibits a nighttime diurnal peak, contradicting known
formation pathways via Cl reaction with O3 and the photolysis of
ClONO2. The complexity continues as ClONO2 also peaks
during the night, given that its main known formation pathway is via reaction
of ClO (produced at sunrise via ClNO2 photolysis) with
NO2. The misidentification of ClONO2 and ClO is not
thought to be a possible reason for these discrepancies due to the low number
of mass spectral peaks that have maxima at night and the mass defect of
chlorine making the peak position unique to chlorine-containing molecules.
IMR chemistry is also not a possible source as these reactions would occur
throughout the day, therefore skewing all of the data and not just the
nighttime levels, although there is a possibility that ClONO2 can
be formed in the IMR by reactions between ClO and NO2. It is
hypothesised that in extremely high OH and HO2 mixing ratios, all ClO is
rapidly converted to HOCl, limiting the formation on significant levels of
ClO and subsequently ClONO2. Khan et al. (2008) suggest that Cl
atoms of around 2 × 104 molecules cm-3 could be present
at night via analysis of alkane relative abundance. Although a formation
mechanism is not proposed, it provides further evidence that ClO formation at
night is possible and may represent an unknown reaction pathway, which
would agree with the measurements presented in this work.
Correlation of particulate Cl- from the AMS measurements
and CO colour coded by the SO2 mixing ratio and size binned by
increasing benzene mixing ratio. A wind rose plot illustrates the wind
direction and particulate Cl- mixing ratio colour coded by
SO2 mixing ratio.
ClNO2 exhibited a similar diurnal profile as N2O5,
peaking at night and lost during daylight due to photolysis. The campaign
mean nighttime mixing ratio was 487 ppt. The maximum mixing ratio observed
was 2900 ppt on 31 May, similar to that previously measured at semi-rural
site in Wangdu (up to 1500 ppt) (Liu et al., 2017), Mount Tai (2000 ppt)
(Z. Wang et al., 2017), but lower than that in Hong Kong (4 ppb) (Wang et
al., 2016).
Source of chloride
The high levels of ClNO2 indicate a local significant source of
chlorine to support these observations. The dominant source of chlorine atoms
for ClNO2 production within models, such as the Master Chemical
Mechanism (MCM), is from sea salt. However, the site is situated 200 km from
the Yellow Sea, and therefore this origin would have a low probability. The
mean AMS chloride mass loading was 0.05 µg m-3 for the
campaign with a maximum of 1.7 µg m-3. The Cl- from
the AMS appears to be correlated strongly with CO and SO2, possibly
originating from power plants or combustion sources. It should be noted that
the AMS data do not include refractory aerosol and also have a cut-off size
larger than the anticipated size of sea salt particles. Instead, the high
Cl- observed appears to originate from mainland areas to the site
(Fig. 4) rather from the nearest coast, further supporting a anthropogenic
source. Tham et al. (2016) observed a strong correlation of aerosol chloride
with SO2 and potassium from measurements done during the same
season in 2014 in Wangdu (a semi-rural site 160 km southwest of Beijing) and
suggested contributions to fine chloride from burning of coal and crop
residues. The latter was also supported by satellite fire spot count data
(Tham et al., 2016). Riedel et al. (2013) have previously reported high
ClNO2 mixing ratios observed from urban and power plant plumes
measuring high mixing ratios of gas-phase Cl2. The correlation with
SO2 indicates coal burning as a potential source of particulate
chlorine which is known to be a significant source of PM in the Beijing
region (Ma et al., 2017), and the correlation with CO and benzene could be an
indicator of biomass burning (Wang et al., 2002). To support this analysis,
Fig. S1 in the Supplement displays a wind rose plot in which radial and
tangential axes represent the wind direction and speed (km h-1). The
colour bar represents the PM2.5 concentration. We could see that during
the campaign, the severe pollution was from the south and southwest, with
little contribution from the east part. Therefore, it is likely that little
contribution of the chloride was from the ocean.
In order to test the hypothesis of biomass burning as a source of particulate
chlorine, biomass burning emissions and transport utilising the EMEP MSC-W
chemical transport model driven by meteorology from the WRF-ARW model
(Skamarock et al., 2008) were used. Neither of the two biomass burning
databases used (FINN and GFAS) contained data on chlorine emissions, so
instead the biomass burning emissions of CO (CObb) were tracked and compared
with the total mixing ratio of CO (COt) at the Changping site. CO was chosen
since the measurements in Changping had shown a strong correlation between CO
and ClNO2 and because CO could be expected to be co-emitted with
chlorine for both biomass burning and industrial combustion.
ClNO2 gas- and particle-phase campaign time series (1 h
averaged) (a) and average diurnal profiles (b).
The peak fitting for ClNO2 and the SO3 interfering mass
at 207–208 AMU (c) and the desorption profile for the counts
attributed to the high-resolution ClNO2 peak (d).
Diurnal profile of N2O5, ClNO2 and
j(ClNO2) for the campaign highlighting the persistence of
ClNO2 past sunrise and the expected rapid photolysis of
N2O5.
Figure S2 (the Supplement) shows the time series of the measured ClNO2
mixing ratios at the Changping site, as well as the modelled mixing ratios of
COt and CObb. CObb is shown for calculations using either the FINN or the
GFAS database, while for clarity the COt is only shown using the FINN
database. It is clear that mixing ratios of CObb are very low compared with COt
(Fig. S2 in the Supplement). The two pollution episodes on 18–23 May and
28 May–5 June are to some extent visible in all time series, but for the
biomass burning CO series the second episode is much less pronounced.
Nighttime averages of the mixing ratios shown in Fig. S2 in the Supplement
were calculated for each night for the time period 18:00–08:00 LT
(UTC + 8), roughly corresponding to the period when ClNO2 is
not destroyed by photolysis. Nights with a significant amount of missing data
for the measurements were excluded. Figure S2 in the Supplement shows scatter
plots of these averages of ClNO2 against the averages of the other
species including their linear fits. The R2 values for these fits were
0.48, 0.04 and 0.21 for COt, CObb FINN and CObb GFAS, respectively. The fact
that mixing ratios of CObb are so much smaller than COt according to the
model, combined with the much better correlation for COt than for CObb,
strongly suggests that industrial emissions are the dominant source of
chlorine, rather than biomass burning. To further investigate the source of
chloride, the model was also run to calculate sea salt levels instead of CO.
This resulted in a poor correlation between sea salt and the ClNO2
(Fig. S4 in the Supplement). The absolute levels of sea salt calculated by
the model were also very low, unlikely to be able to produce the observed
mixing ratios of ClNO2 as observed by CIMS.
Particle-phase ClNO2
A particle desorption profile was observed in the high-resolution data for
ClNO2. The count increase at this 1 AMU mass can be attributed to
two sources: SO3 and ClNO2 as shown in Fig. 5. The
SO3 peak is predominantly found in the particle phase and is below
the limit of detection (LOD) in the gas phase. During initial analysis of
these data, SO3 interfered with the ClNO2 peak fitting
and attributed its counts to ClNO2 in the particle phase as its
33S ion is only 0.005 AMU away from the ClNO2 peak. Upon its
inclusion into the peak list and utilisation of the Tofware feature which
constrains isotopes and reallocates the signal appropriately, ClNO2
remains to indicate a strong desorption profile. The diurnal cycle of this
desorption correlates well with the ClNO2 gas-phase profile,
indicating a correct assignment of the counts to particle-phase
ClNO2. The desorption profiles with respect to temperature also
exhibit a thermogram structure and not, for example, a gas-phase leak into the
system which could have accounted for the correlation with the gas-phase time
series. This suggests the possible presence of ClNO2 in the
particle phase. Another possible explanation could be the deposition of
ClNO2 from the gas phase onto the filter as the ambient air flows
through FIGAERO.
(a) Steady-state calculation of inorganic halogens'
contribution to chlorine atom production. (b) Relative mean diurnal
profiles of calculated chlorine atom mixing ratio calculation from this
work (Beijing) and measurements in the UK (London; Bannan et al., 2015) and a
marine site (Weybourne Atmospheric Observatory; Bannan et al., 2017). The
steady-state OH production rate from Beijing is also displayed to illustrate
relative mixing ratios of oxidants.
If we assume the analysis and collection technique is correct, we see an
average particle- to gas-phase partitioning of 0.07, with a maximum of 0.33
and a minimum of 0.009. The average mixing ratio of ClNO2 collected
onto the filter during desorption is 13 ppt with a maximum of 120 ppt.
Previous modelling studies assume all ClNO2 is in the gas phase due
to the low Henry's law constant; e.g. for the TexAQS II campaign, they
calculated that 0.1 ppb in the gas phase would yield 0.54 ppt in the
particle phase (Simon et al., 2009). However, these data suggest that a
non-negligible amount of the chlorine associated with ClNO2 is not
liberated from the particle phase, assuming that no additional
ClNO2 is formed by thermally driven reactions. The slope of the
particle- to gas-phase CIMS data is calculated to be 0.048, a factor of 96
higher than using Henry's law coefficient to estimate the particle mixing
ratio.
ClNO2 daytime persistence and Cl liberation
Both ClNO2 and N2O5 are photolytically unstable, with
studies reporting lifetimes on the order of hours for ClNO2
depending on the solar strength (e.g. Ganske et al., 1992; Ghosh et al.,
2011). Nocturnal ClNO2 removal pathways have generally been
reported to be negligible, with ClNO2 being assumed to be
relatively inert (Wilkins et al., 1974; Frenzel et al., 1998; Rossi, 2003;
Osthoff et al., 2008), but the work of Roberts et al. (2008) and Kim et
al. (2014) would suggest that this may not be strictly true. However, given
that the average diurnal profile does not show the importance of nocturnal
removal pathways in this study, observed losses are attributed solely to
photolysis, with J(ClNO2) controlling the lifetime.
Rapid photolysis can be observed for N2O5 in Fig. 6 showing a
near instant drop below LOD, whereas the ClNO2 mixing ratio not
only persists for up to 7 h but also shows evidence of an increase in
mixing ratio at 07:00 UTC (Fig. 6). This is observed throughout the campaign
and has been frequently observed in the previous study in Wangdu (Tham et
al., 2016). The breakdown of the nocturnal boundary layer and inflow of air
masses from above, carrying pollution from nearby industry/industries is a likely
cause of this persistence of possible increase of ClNO2. Liu et
al. (2017) also observed high daytime mixing ratios of ClNO2
(60 ppt) at the Wangdu site which they attribute to a possible oxidation
mechanism due its correlation with O3 and Cl2 providing a
daytime formation pathway to maintain mixing ratios against its rapid
photolysis.
Consistent with past measurements and the measurements of this study,
ClNO2 is expected to provide a significant source of Cl during
daytime hours, presenting a potentially significant source of the reactive Cl
atom during the day. Its rapid photolysis rate and elevated mixing ratios
enable Cl to compete with OH oxidation chemistry, the known dominant
daytime radical source. Here, a simple steady-state calculation will be used
to determine the Cl atom mixing ratio summarised as follows but detailed
within the Supplement:
(ss1)Cl2+hν→Cl+Cl(ss2)ClNO2+hν→Cl+NO2(ss3)ClONO2+hν→ClO+NO2(ss4)HOCl+hν→OH+Cl(ss5)OClO+hν→O+ClO
(ss6)OH+HCl→Cl+H2O(ss7)Cl+O3→ClO+O2(ss8)Cl+CH4equivalent→HCl+products[Cl]SS=2J1[Cl2]+J2[ClNO2]+J3[ClONO2]+J4[HOCl]+J5[OClO]+k7[OH][HCl]}/{k7[O3](ss9)+k8[CH4]equivalent},
where [CH4]equivalent represents the reactive VOC present as if
it were reacting as CH4.
Bannan et al. (2015) were able to use this steady-state approach to compare
the relative loss via reaction with OH compared with Cl atoms. Although this
approach is an estimation, it was shown to produce results comparable with
those of the more rigorous MCM approach, although we do acknowledge that large
errors will be present in the radical species calculations, which is detailed
in the supporting information. Steady-state calculations reveal a sharp rise
of chlorine atoms produced at sunrise peaking at
1.6 × 105 molecules cm-3 around 07:00 LT which then
gradually decreases, contributing to Cl atom production until 14:00 LT (Fig. 7a).
Supporting Cl2, ClONO2, OClO, HOCl and HCl measurements
by CIMS report that chlorine atoms can sustain a relatively high production
rate until 15:00 LT as evidenced by the daytime build-up of HCl and Cl2.
ClNO2 on average contributes 78 % of the chlorine atoms
produced from inorganic halogens with 13 % from Cl2. ClNO2
also represents over 50 % of the chlorine atoms until midday. After
approximately 15:00 LT, Cl2 and HCl become the more dominant Cl atom sources. On the night
where the highest ClNO2 mixing ratios were measured, 90 % of
the chlorine atoms originated from ClNO2 photolysis until 14:00 LT, and
HCl and Cl2 then became main contributors until 16:00 LT (up to 80 %).
ClONO2, HOCl and OClO appear to be insignificant contributors to
chlorine atom production throughout the campaign compared with
ClNO2, HCl and Cl2.
To put these chlorine atom mixing ratios into a more global perspective, data
collected by the University of Manchester from a marine site and an urban
European site have been compared in Fig. 7b. Bannan et al. (2015, 2017)
previously utilised a box model to calculate Cl atom mixing ratios during the
campaign so that the rate of oxidation of VOCs by Cl atoms could be compared
with oxidation by measured OH and measured ozone. The simple steady-state
calculation described previously will be used to determine the Cl atom mixing
ratio for this measurement study. The results show that both at the UK marine
and urban site maximum chlorine atom mixing ratios are more than an order of
magnitude lower than the mean of Beijing. It should be noted that the only
source of Cl in the UK studies was ClNO2, but given the dominance
of ClNO2 in this study, the measurements presented here suggest a high
importance of the chlorine chemistry for the Asian air chemistry. Studies of
chlorine radical production in Los Angeles by Riedel et al. (2012) and
Young et al. (2012) indicate that the high production rate in Beijing is
somewhat typical of urban sites, although HCl and ClNO2
contribution to radical production is the same, whereas here we see very
little chloride radical production from HCl in comparison with
ClNO2.
Although this study does not reach the scope of characterising O3
and ROx production from chlorine atom chemistry,
statistics are often reported with ClNO2 morning chemistry via
modelling simulations, and we can put into perspective the mean and maximum
mixing ratios relative to other studies. Tham et al. (2016) recorded a
maximum ClNO2 mixing ratio of 2070 ppt from a plume originating
from Tianjin, the closest megacity to Beijing, and report a 30 % increase
in ROx production and up to 13 % of O3
production. Liu et al. (2017) observed peak mixing ratios up to 3 ppb and
similar diurnal mixing ratios which they calculated contribute to a 15 %
enhancement of peroxy radicals and 19 % O3 production. Wang et
al. (2016) report up to 4.7 ppb of ClNO2 in Hong Kong and
calculated a maximum increase of 106 % of HOx in the
morning and an enhancement of O3 production the next day by up to
41 %. Therefore, it is evident that this work supports similar studies in
Asia that conclude that chlorine atom oxidation significantly contributes to
atmospheric oxidation via ROx and O3 production.
Although several studies have demonstrated a non-negligible impact of
chlorine oxidation chemistry (e.g. Oshoff et al., 2008; Riedel et al., 2014;
Sarwar et al., 2014), the impact of Cl chemistry varies significantly between
various areas and atmospheric conditions; e.g. Bannan et al. (2015, 2017)
deemed the impact from chlorine atom chemistry to be relatively low with
respect to O3 production and competing with OH radicals for VOC
oxidation.
VOC oxidation by chlorine atoms
Steady-state calculations of OH (as described by Whalley et al., 2010)
estimate that campaign average maximum mixing ratio was
7 × 106 molecules cm3 (Fig. 7b), 6 times greater than
the maximum chlorine atom mixing ratio and 14 times higher than the average
chlorine atom mixing ratio. Pszenny et al. (2007) report estimated OH to
chlorine atom ratios, from VOC lifetime variability relationships, of 45 to
199 along the east coast of the United States. Although the ratio appears
much larger than calculated in this work, here we present not only
significantly higher mixing ratios of ClNO2 which appear to
be a consistent conclusion from measurements in Asia, but also the chlorine
within this study appears to originate from an anthropogenic origin rather
than marine, possessing the ability to supply a much larger reservoir of
halogens to be liberated through photolysis.
The relative oxidation rate of the chlorine atom and OH to VOCs can vary
greatly. Rate coefficients for reaction of Cl atoms with some volatile
organic compounds have been shown to be up to 200 times faster than the
comparable reaction with OH. The ratio reported here is significantly less
than this each day; Cl can subsequently dominate VOC oxidation for some
fraction of the day. Here, the diurnal maxima of the chlorine atom and OH
differ by 5 h, enabling chlorine atoms to dominate VOC oxidation earlier
in the day before OH mixing ratios have built up. The relative oxidation rate
of VOCs to OH and the chlorine atom also varies greatly, creating a
difference for various VOCs. If an average reaction rate for alkenes and
alkanes to Cl and OH is calculated, it is possible to generalise the
significance of each oxidation pathway to qualitatively asses the
contribution chlorine atoms have on oxidation chemistry. It can be seen in
Fig. 8 that alkenes are much more likely to be oxidised by OH than Cl,
although a significant contribution (15 %) is attributed to chlorine
chemistry. Although significant if evaluated on a global level, Liu et
al. (2017) estimated that Cl atoms oxidise slightly more alkanes than OH
radicals in a similar region of China, implying the increased scale of
chlorine oxidation in China. Alkanes are known to have a much higher Cl-to-OH
relative reaction rate than alkenes, and Cl contribution to oxidation is
higher than OH until midday. The contribution to oxidation remains almost
equal for the remainder of the day due to the persistence of ClNO2
and also relatively high levels of Cl2 and HCl. This analysis is
representative of that by Bannan et al. (2015) who report contributions of
alkene and alkane oxidation by Cl up to 3 and 15 %, respectively, from
ClNO2 mixing ratios peaking at 724 ppt.
Mean diurnal time series of alkene (pink) and alkane (blue) relative
reaction rate (arbitrary value) with the chlorine atom (dashed) and OH
(solid).
Campaign time series of isoprene, isoprene epoxydiol (IEPOX), CMBO and steady-state
production rate of chlorine atoms and OH.
Mean diurnal profiles of isoprene (right y axis) and its OH
oxidation product (IEPOX) and chlorine atom oxidation products CMBO,
C5H9ClO2 and C5H9ClO3 (left y axis).
This significant oxidation of VOCs by chlorine atoms will result in different
products to those of OH oxidation as illustrated and that neglecting the
contributions made by Cl atoms will significantly underestimate the degree of
chemical processing of VOCs in this study and other environments where there
is a source of Cl atoms. Evidence of the proposed Cl oxidation of VOCs is
validated through detection of selected Cl-induced oxidation products by the
ToF-CIMS, all of which are displayed in Table 1.
Isoprene oxidation by the chlorine atom
1-Chloro-3-methyl-3-butene-2-one (CMBO, C5H6ClO), a unique marker
of chlorine chemistry, has previously been measured at mixing ratios up to
9 ppt by offline gas chromatography in Houston, Texas (Tanaka et al.,
2003), and in laboratory studies of chlorine–isoprene oxidation (D. Wang et
al., 2017). CMBO exhibited a campaign maximum of 21 and mean of 34 ion
counts (near similar ppt mixing ratio if the chloroacetic acid calibration
sensitivity is applied) exhibiting a near-typical diurnal profile with
abundance rising sharply after sunrise, at the same rate as the chlorine atom
production, but maintaining mixing ratios past noon longer than that of
isoprene and the chlorine atom.
Campaign time series of benzene and 4-chlorocrotonaldehyde (CCA) with supporting calculations
of OH and the chlorine atom production rates.
Mean campaign diurnal profiles of benzene (grey) and CCA in the
particle (dashed red) and gas phases (solid red).
The daily maxima of CMBO varied throughout the campaign and can be explained
by the relative mixing ratios of its precursors: the chlorine atom and
isoprene. Its mixing ratio throughout the campaign followed similar
intensities to its precursors, and Fig. 9 highlights its dependence on both Cl
atom and isoprene mixing ratios. The production rate of Cl and mixing ratio
of isoprene were relatively low from 24 to 27 May
(1.6 × 105 molecules cm-3 s-1 Cl and 0.5 ppb
isoprene), which resulted in relatively low CMBO mixing ratios. An increase
in isoprene and Cl on 28 to 30 May was subsequently mirrored by the CMBO
levels as qualitatively expected. On closer inspection of 30 and 31 May, the
mixing ratio of CMBO was lower than expected on 30 May due to higher
chlorine atom and isoprene mixing ratios compared with 31 May. This could
be explained by anticipated higher OH mixing ratio as calculated by the
steady-state model, which is also further represented by higher mixing ratios
of IEPOX (isoprene epoxydiol, i.e. OH oxidation products) on 30 May.
This illustrates how the ToF-CIMS can identify isoprene oxidation products of
two competing oxidation pathways. The high levels of IEPOX on 28 May can
also possibly describe the relatively high levels of CMBO in the particle
phase due to an already well-oxidised air mass. CMBO may also not be unique
to only isoprene–chloride reactions and therefore have alternative sources
not represented in this data set.
Identified Cl–VOC reaction products, nomenclature of Cl–VOC and
precursor compound.
Cl–VOC
Potential nomenclature
Precursor
CHClO
formyl chloride
formaldehyde
C2H3ClO
chloroacetaldehyde
acetaldehyde
C3H5ClNO5
chloro PPN
PPN
C3H5ClNO5
chloro PAN
PAN
C3H5ClO
chloroacetone
acetone
C2H3ClO2
chloroacetic acid
acetic acid
CHClO2
chloroformic acid
formic acid
C4H7ClO
chloromethyl vinyl ketone
isoprene
or butanal
C5H6ClO
CMBO – chloro-3-methyl-
isoprene
3-butene-2-one
C5H9ClO2
–
isoprene
C5H9ClO3
–
isoprene
C3H5ClO
propanoyl chloride
1,3-butadiene
C8H9Cl
chloroethyl benzene
aromatic
Further daily oxidation rates can be probed via analysis of the related
isoprene oxidation products observed by the CIMS. Figure 10 depicts the
diurnal time series of the precursor itself and several Cl–VOC products and
IEPOX. CMBO mixing ratios rise rapidly after sunrise due to the low mixing
ratio of OH and high production rate of the chlorine atom. The secondary and
tertiary products, C5H9ClO2 and C5H9ClO3
(also measured in the laboratory by D. Wang et al., 2017), increased in
mixing ratio at a much slower rate but appear to peak later in the day
(16:00 LT), whereas CMBO peaked around 10:00 LT (similar to the
ClNO2 peak time) and fell off, due to its further oxidation to form
the secondary and tertiary products. IEPOX mixing ratios increased slowly
after sunrise and peaked later in the day, as expected due to the
availability of OH and competition from the chlorine atom chemistry. The
similar time series of the secondary and tertiary products to IEPOX were also
reported by D. Wang et al. (2017) and were suggested to be ideal tracers of
SOA production.
Anthropogenic Cl–VOC production
A similar unique chlorine oxidation marker in urban coastal areas, has been
reported in the literature for 1,3-butadiene: 4-chlorocrotonaldehyde (CCA)
(Wang et al., 2000). No measurements of 1,3-butadiene were made during
this field campaign, although due to its common source to benzene (automobile
exhausts (Ye et al., 1998), we present a comparison of the CCA measured by
CIMS and benzene measurements made by the PTR-MS. The intensity of CCA in
both the gas and particle phases reflects well the abundance of its precursors.
The maximum mixing ratio of the chlorine atom coincides with a high mixing
ratio of benzene and subsequently CCA on 30 May, whereas very low levels of
CCA were observed for the beginning of the campaign (Fig. 11).
The diurnal time series of benzene (Fig. 12) indicates high mixing ratios in
the early hours of the day, possibly associated with high anthropogenic
activity or an inflow of urban air masses from downtown Beijing. The mixing
ratio falls off throughout the day and almost perfectly anti-correlates with
the CCA gas-phase diurnal profile which increases from sunrise and peaks at
15:00 LT. The particle-phase CCA diurnal time series steadily builds up
throughout the day and does not peak until late in the evening, providing
evidence of SOA production from the chlorine oxidation of anthropogenic
pollutants.
Conclusions
A FIGAERO ToF-CIMS was utilised in Beijing to assess the liberation of
chlorine atoms via inorganic halogen photolysis. A suite of inorganic
halogens was detected, namely ClNO2, reaching mixing ratios up to
2900 ppt, which is suggested to have an anthropogenic origin due to the
particulate chlorine correlation with SO2, benzene and CO.
ClNO2 was identified in the particle phase at higher ratios with
respect to its gas-phase component than expected, which may only prove to be
significant at such elevated mixing ratios as observed in east Asia.
ClNO2 mixing ratios above LOD persisted up to 7 h past sunrise,
attributed to the lifetime of ClNO2 at these high mixing ratios and
a possible inflow of heavily polluted air masses from the downtown urban
area. Supporting Cl2 and HCl mixing ratios proved to be significant
contributors to chlorine atom production via steady-state calculations.
Compared with data attained from European-based campaigns, these mixing
ratios exceed marine and urban environments by at least an order of
magnitude.
This high mixing ratio of chlorine atoms resulted in a steady-state
calculated OH : Cl ratio down to a factor of 6, suggesting Cl chemistry may be
able to dominate alkane oxidation until midday but contribute significantly
to alkene oxidation throughout the day (15 % on average). This enabled
significant mixing ratios of Cl–VOCs to be formed, providing the first ambient
high-time-resolution measurements of specific Cl–VOC species simultaneously
measured in the gas and particle phases. The measured unique markers of
chlorine chemistry for both biogenic and anthropogenic precursors provide
quantitative and qualitative data to probe the extent of chlorine atom
chemistry and how they compete with OH. Simultaneous measurements of the VOC
precursors via PTR-MS, and IEPOX, Cl–VOCs with the CIMS provides rich
information on SOA formation pathways via both OH and chlorine atom
oxidation. Multistep oxidation products of Cl–VOCs were also identified and
can provide partitioning information and SOA formation rates and lifetimes.
The results highlight deficiency in chlorine atom chemistry descriptions
within models possibly due to a lack in quantification and identification of
Cl–VOC products in the gas and particle phases. This work provides instrumental
capability to probe the competition between OH and Cl oxidation chemistry and
quantify their effect on ozone and SOA formation.