Introduction
Ozone, hydroxyl radicals (OH) and nitrate radicals and halogens atoms can
initiate the oxidation of hydrocarbons such as biogenic terpenes in the
atmosphere (Atkinson, 2000). Although the reactivity of these oxidants toward
a large variety of atmospheric trace gases is well established, ambient
observations have revealed major ambiguities in atmospheric oxidation
chemistry, especially related to OH in locations having high emissions of
biogenic volatile organic compounds (BVOCs) (Di Carlo et al., 2004; Lou et
al., 2010; Nölscher et al., 2012; Lelieveld et al., 2008; Hofzumahaus et
al., 2009; Taraborrelli, et al., 2012). Recently, two additional major
processes contributing to the complexity of atmospheric oxidation have been
revealed: firstly, the auto-oxidation mechanism producing highly oxidized
condensable organic vapours in the gas phase discovered by Ehn et al. (2014)
– such vapours are shown to be essential for formation of secondary organic
aerosol (Kulmala et al., 1998; Riipinen et al., 2011); secondly, the
suggestion that stabilized Criegee intermediates (sCIs), formed by ozonolysis of
biogenic alkenes (Criegee, 1975), might add to the oxidation capacity of the
atmosphere – at least from the point of view of SO2 oxidation and
subsequent formation of sulfuric acid, H2SO4 (Mauldin III et al.,
2012; Berndt et al., 2012; Welz et al., 2012). These findings demonstrate the
incomplete scientific understanding of atmospheric oxidation chemistry. Here,
we focus on the latter of those novel observations.
The sCI formation pathway starts when ozone reacts with the double bond of an
alkene, producing an energy-rich primary ozonide, which very rapidly
decomposes via a concerted ring opening to form a carbonyl oxide, the
so-called Criegee intermediate (CI) (Calvert et al., 2000). The energy-rich
intermediate, CI, either undergoes unimolecular decomposition on a timescale
of 1 ns, yielding OH radicals and other products, or it can be stabilized by
collisions with gas molecules (Kroll et al., 2001). The resulting sCI can still undergo unimolecular decomposition,
leading again to OH radical formation and others, but with a thermal lifetime
thought to be of the order of 1 s depending on temperature and sCI structure
(Kroll et al., 2001). Due to the relatively long lifetime of sCI, bimolecular
reactions of sCIs with several compounds like water vapour, SO2,
carbonyls, organic acids, etc., are also possible (e.g. Neeb et al., 1996,
1997; Johnson, 2001; Welz et al., 2012, 2014; Mauldin III et al., 2012; Berndt et
al., 2012, 2014a, b; Taatjes et al., 2012, 2013). These reactions can
potentially be fast enough to contribute significantly to the atmospheric
oxidation capacity. Thus, some significant gaps in our understanding of
atmospheric oxidation could potentially be filled by sCI chemistry, once the
processes controlling the production and fate of sCIs are properly resolved.
Until recently, the reaction rate coefficients of sCIs with atmospheric
compounds, such as SO2, were thought to be too small (Johnson, 2001) to
cause measurable effects on atmospheric oxidation chemistry, with the
exception of the sCI + water vapour reactions (Hasson, 2003). The
reaction with water vapour was also thought to be the main fate of sCIs in
the atmosphere. However, Mauldin III et al. (2012) recently reported ambient
and laboratory observations strongly suggesting an atmospherically relevant
reaction between sCI and SO2. This was qualitatively supported by the
laboratory experiment of Welz et al. (2012). Welz et al. (2012) studied the
simplest possible Criegee intermediate (formaldehyde oxide, CH2OO) in a
low-pressure (4 torr) laboratory experiment, finding an absolute rate
coefficient for the CH2OO + SO2 reaction of
3.9 × 10-11 cm3 molecule-1 s-1. Mauldin III et
al. (2012) estimated the reaction rate coefficient to be roughly
6 × 10-13 and
8 × 10-13 cm3 molecule-1 s-1 for sCIs from
the ozonolysis of α-pinene and limonene, respectively. Berndt et
al. (2012) investigated experimentally the sCI yields, lifetimes and rate
coefficients for reactions with SO2 for sCIs from the ozonolysis of
selected alkenes including 2,3-dimethyl-2-butene (TME), trans-2-butene and
1-methyl-cyclohexene (MCH). Using an indirect approach based on
H2SO4 measurements, they found typical lifetimes at atmospheric
pressure and an atmospherically relevant humidity of a few hundreds of
milliseconds, and reaction rate coefficients for sCI + SO2 of the
order of 10-13–10-12 cm3 molecule-1 s-1,
depending on the structure of the sCI.
It should be noted that ozonolysis of a single alkene in most cases produces
structurally different types of sCI, including syn- and anti-conformers (for
sCIs with one H atom bound at the C-OO carbon) which might have a different
reactivity. Therefore, the given data for α-pinene and limonene
(Mauldin III et al., 2012) and for trans-2-butene and 1-methyl-cyclohexene
(Berndt et al., 2012) represent average values for the reactivity of all sCIs
arising from the selected alkene.
In summary, both the laboratory measurements by Berndt et al. (2012) and
field observations by Mauldin III et al. (2012) suggest that sCI are roughly a
factor of ∼ 100 more reactive with SO2 than suggested earlier
(Johnson, 2001), but approximately 2 orders of magnitude less reactive than
the close-to-collision-limit rate coefficient by Welz et al. (2012).
Nevertheless, while recent findings agree on the potential significance of
the sCI + SO2 reaction, there is still a considerable uncertainty in
the absolute and relative rate constants obtained by different experimental
approaches.
Understanding the reaction of the sCI + SO2 is highly important from
the atmospheric chemistry and physics point of view. Sulfuric acid plays a
key role in Earth's atmosphere, triggering secondary aerosol formation
(Kulmala et al., 2004; Berndt et al., 2005, Riipinen et al., 2007; Sipilä
et al., 2010; Kerminen et al., 2010), and thus connects natural and
anthropogenic SO2 emissions to global climate via indirect aerosol
effects on radiative forcing. The effect of sCI on SO2 oxidation was
assessed by Boy et al. (2013), who simulated sulfuric acid production at the
SMEAR (Station for measuring atmosphere-ecosystem relations) II boreal forest field station using the reaction rate coefficients
suggested by Mauldin III et al. (2012). Their results supported the experimental
observations by Mauldin III et al. (2012), showing that a significant fraction
(several tens of percents) of ground-level gas phase sulfuric acid
originates probably from sCI-initiated oxidation of SO2. Pierce et
al. (2013) took a step further and studied the role of the sCI + SO2
reaction to global aerosol and cloud condensation nuclei (CCN) concentrations by the global climate
model. They found, in accordance with Boy et al. (2013), that sCIs can
contribute significantly to gas phase H2SO4 in the lower
troposphere above forested areas. However, due to further aerosol dynamical
processes during particle growth to CCN sizes, the influence of sCI on
sulfuric acid concentration was only feebly projected to CCN concentrations,
and thus to radiative forcing. However, Pierce et al. (2013) used reaction
rate coefficients, including the upper limit for the sCI loss rate
(decomposition and reaction with water vapour), obtained by Welz et
al. (2012) for CH2OO. Furthermore, the sCI reaction rate coefficients,
including sCI loss in a reaction with water, may be strongly dependent on the
sCI structure. Therefore, a reassessment of the CCN sensitivity, using
parameters obtained for atmospherically relevant sCIs in atmospheric
conditions, would be warranted.
On top of the reaction with SO2, a further, mostly unresolved question
is whether or not oxidation by sCI has a more general role in atmospheric
chemistry. Earlier studies have probed the reaction of CH2OO with
several atmospheric constituents (see e.g. Fenske et al., 2000). Reaction of
CH2OO with formic acid, HCOOH, yielding to production of
hydroperoxymethyl formate was demonstrated by Neeb et al. (1995, 1996), with
follow-up studies by Thamm et al. (1996) and Hasson et al. (2001b). Neeb et
al. (1997) showed that the rate coefficient for the above reaction is
relatively large (14 000-fold) in comparison to the reaction rate
coefficient for CH2OO + water. An extremely high reactivity of
CH2OO and CH3CHOO toward formic (HCOOH) and acetic acid
(CH3COOH) was observed also by Welz et al. (2014) in a low-pressure
system. Also Taatjes et al. (2012; 2013) showed that sCIs – at least the
simple sCIs – are reactive toward other trace gases in addition to SO2.
These observations imply that reactions of sCI with organic acids might have
atmospheric importance, and further research is urgently required.
In the present study, we investigate experimentally the sCI yield and the
rate coefficient ratio k(loss) / k(sCI + SO2), where the
loss coefficient k(loss) incorporates thermal decomposition of sCI and the
reaction with water vapour,
k(loss) = k(dec.) + k(sCI + H2O) ⋅ [H2O].
This rate coefficient ratio represents (for different temperatures and water
vapour concentrations) the important parameter for understanding the
sCI-controlled oxidation of SO2 to H2SO4 in the atmosphere. This
study comprises reactions of sCIs produced from the ozonolysis of isoprene
and two monoterpenes abundant in the atmosphere, i.e. α-pinene and
limonene. To demonstrate the capability of sCIs playing a more general role
in atmospheric chemistry, we investigated the reaction of acetone oxide
((CH3)2COO, the sCI from TME ozonolysis) with small organic acids.
Methods
NO3-–chemical ionization–atmospheric
pressure interface–time-of-flight mass spectrometer (NO3--CI-APi-TOF)
A NO3--CI-APi-TOF was used in the experiments described here
primarily for the detection of sulfuric acid. The NO3--CI-APi-TOF
comprises a specially designed inlet for chemical ionization at ambient
pressure (CI) and an atmospheric pressure interface (APi) to couple ions to
a time-of-flight mass spectrometer (TOF). The instrument is described in
detail by Jokinen et al. (2012), but the CI part of the system will be
shortly discussed also here. The APi-TOF is well described elsewhere
(Junninen et al., 2010).
The design of the CI inlet is largely based on the original NCAR design
(Eisele et al., 1993; Kurtén et al., 2011; Jokinen et al., 2012). Ions
are produced in a sheath flow concentric to the sample flow by a 10 MBq
241 Am radioactive α-emitter. Minute quantities of nitric acid vapour
are fed into sheath air surrounding the sample inlet flow, resulting in the
formation of NO3-(HNO3)n,n= 0-2 ions. These ions are pushed
into the sample flow, entering the ion–molecule interaction tube at the
centre line, by means of an electric field. The design is virtually
wall-less, and sample wall loss occurs only in the sample inlet tube. The
sample flow in the system is 10 lpm, and the concentric sheath flow where
ions are produced is 20 lpm. Sheath gas is air purified with a particle
filter and an SO2 scrubber.
Concentrations of alkenes (initial and reacted within the residence
time of 39.5 s), OH scavenger and O3, and the reaction rate coefficients
used in the experiments.
Alkene
[Alkene]molecules cm-3
[Propane] molecules cm-3
[O3] molecules cm-3
k(alkene+O3)cm3 s-1
Reacted [alkene]molecules cm-3
α-Pinene
8.0 × 1011
1.64 × 1015
(SO2: (3.2-160) × 1011) 8.2 × 1015
(SO2: (1.6-24) × 1013)
2.2 × 1011
1.1 × 10-16
7.65 × 108
Limonene
1.6 × 1011
1.64 × 1015 (SO2: (3.2-160) × 1011) 8.2 × 1015 (SO2: (1.6-16) × 1013)
2.2 × 1011
2.5 × 10-16
3.48 × 108
Isoprene
1.5 × 1012
1.64 × 1015 (SO2: (3.2-160) × 1011) 8.2 × 1015 (SO2: (1.6-16) × 1013)
2.2 × 1011
1.29 × 10-17
1.68 × 108
Sample ionization in the CI system occurs at atmospheric pressure via proton
transfer between nitrate ions and sulfuric acid and subsequent
HSO4- ⋅ HNO3 adduct formation:
H2SO4+NO3-⋅(HNO3)n,n= 0-2→HSO4-⋅HNO3+n(HNO3),n= 0-2.
The chemically ionized sample is drawn inside the mass spectrometer through a
critical orifice with a flow rate of ∼ 0.8 lpm. The ions are then
guided through the differentially pumped APi using quadrupoles and eventually
to the TOF for m / Q (mass to charge) separation.
A fraction of HSO4- ⋅ HNO3 clusters (Reaction R1) fragment inside
the APi-TOF. The sulfuric acid concentration (in molecules cm-3)
measured with the NO3--CI-APi-TOF is calculated from the measured ion
signals according to
H2SO4=HSO4-+H2SO4NO3-NO3-+HNO3NO3-+HNO3HNO3NO3-×C,
where C is the calibration coefficient. The detection limit for sulfuric
acid monomer is of the order of 104 molecules cm-3, and the error
in determined sulfuric acid concentration is ± 45 % (Berndt et al.,
2012). Data were analysed using TofTools software.
Neglected in the analysis based on Eq. (1) is the potential effect of water
vapour on ion chemistry and thus on the calibration coefficient C. Water
vapour can affect the level of hydration of sulfuric acid, thereby affecting
the charging, the diffusion coefficient and the collision diameter.
Nitrate-water clusters also have different collision diameter than pure
nitrate ions, and clustering could, potentially, change the proton affinity of
the primary ions. Also steric effects may play a role. The CI-APi-TOF
technique is relatively new, and detailed understanding of how relative humidity (RH) affects the
detection does not exist. Experiments on the RH-dependent H2SO4
sensitivity of CI-APi-TOF instruments reveal that the calibration coefficient
C is less influenced by RH in the range 2–65 % and the small changes
observed are within the uncertainty of the measurement; see experimental data
given in the online discussion of this paper and Berndt et al. (2014a). It is
to be noted that the rate coefficient ratios reported here are independent of
the absolute H2SO4 calibration.
Laboratory experiments
Laboratory experiments were conducted in the Leibniz Institute for
Tropospheric Research Laminar Flow Tube (IfT-LFT) at
T = 293 ± 0.5 K, RH = 10–50 %
([H2O] = (0.58–2.89) × 1017 molecules cm-3)
and with a flow of 30 lpm (STP) synthetic air as the carrier gas, resulting
in a total residence time of 39.5 s. The experimental methods are identical
to those reported in Berndt et al. (2014a) but will be described briefly
here.
In the experiments focusing on sCI yields and the relative reaction rate
coefficients, the alkenes, SO2 and the OH radical scavenger (propane)
premixed with the humidified carrier gas were fed at the top of the flow
tube. Ozone diluted with the carrier gas was introduced through an inlet
55 cm downstream of the port for the other reactants. The added propane ensured
scavenging efficiency of 96.9–99.98 %, depending on the experimental
conditions, for OH radicals formed in the ozonolysis. The SO2
concentration was varied in the range
3.2 × 1011–2.4 × 1014 molecules cm-3.
Concentrations of alkenes, propane, ozone, the reaction rate coefficients
used and the concentration of reacted alkenes within the residence
time of 39.5 s are given in Table 1.
In the experiments focusing on the reactivity of sCI (acetone oxide) toward
HCOOH (formic acid) and CH3COOH (acetic acid), with concentrations
ranging between
3.0 × 1010 and 2.0 × 1013 molecules cm-3, were
fed together with other reagents at the top of the flow tube. Concentrations
of the reagents were (unit: molecules cm-3)
[TME] = 4.0 × 1010,
[O3] = 2.2 × 1011,
[SO2] = 3.2 × 1012,
[propane] = 1.64 × 1015 and a relative humidity of
10 %.
In all experiments sulfuric acid was measured using the
NO3--CI-APi-TOF and alkene concentrations we measured with the proton
transfer reaction mass spectrometer (Ionicon PTR-MS) (Lindinger et al.,
1998). The derivation of the parameters of interest from experimental data is
described in the Results and discussion section.
Loss-corrected measured [H2SO4] at the outflow of IfT-LFT
in α-pinene experiments at RH = 10 % and RH = 50 %.
Lines show multivariate least-squares fittings according to Eq. (3), from which
the relative rate coefficients and sCI yield were obtained.
Loss-corrected measured [H2SO4] at the outflow of IfT-LFT
in limonene ozonolysis experiment at RH = 10 % and RH = 50 %.
Lines show multivariate least-squares fittings according to Eq. (3).
Loss-corrected measured [H2SO4] at the outflow of IfT-LFT
in α-pinene experiments at RH = 10 % and RH = 50 %.
Solid lines show multivariate least-squares fittings according to Eq. (3).
Dashed line shows the multivariate least-squares fittings according to
Eq. (5), which accounts for the different behaviour of different sCIs.
Results and discussion
sCI yields and relative rate coefficients
Figures 1–3 show the sulfuric acid concentration measured at the outflow of
IfT-LFT as a function of [SO2] at RH = 10 % and
RH = 50 %. In analysing the experimental data the following reaction
sequence (Reactions R2–R6) was considered.
O3+alkene→y1⋅OH+y2⋅sCI+otherssCI+H2O→products;k(sCI+H2O)sCI+SO2→…→y3⋅H2SO4;k(sCI+SO2)sCI→OH+others;k(dec.)sCI+org.acid→products;k(sCI+acid)
First, ozone, when reacting with alkene, produces Criegee intermediate,
which can either rapidly (picoseconds) decompose and produce OH (with a yield
y1) and other products or be stabilized by collisions with the pressure
gas resulting in formation of sCI with a yield y2 (Reaction R2). The sCI
can react with water vapour (Reaction R3) or with SO2 (Reaction R4). Here we assume that
the H2SO4 formation yield (y3) for reaction (Reaction R4) is unity (see
discussion on the validity of assumption below). The sCI can also thermally
decompose before reacting with other molecules, resulting in the production
of OH and other products (Reaction R5). In addition to the unimolecular decomposition
and reactions with H2O and SO2, sCI can, as we will demonstrate,
react with organic acids (Reaction R6) and potentially with several other atmospheric
constituents.
Since OH formed in reaction (R5) is efficiently scavenged (> 96.9 %,
at highest [SO2]; see Table 1 for propane concentrations
used for different ranges of SO2), the reaction of remaining (< 3.1 %) OH with SO2 can be neglected. Even at highest [SO2] of
2.4 × 1014 molecules cm-3, reaction of OH radicals with
SO2 contributes to less than 10 % of total [H2SO4]. At
[SO2] below 1 × 1014 molecules cm-3, the OH reaction
can be totally neglected.
Another reaction that could be speculated to produce additional
H2SO4 in our system is the reaction of peroxy radicals, RO2,
with SO2. However, there are no clear experimental indications in the
literature suggesting a fast-enough reaction of RO2 + SO2. For
example for the CH3O2 + SO2 reaction, the rate coefficient
is below 5 × 10-17 cm3 molecule-1 s-1 (DeMore
et al., 1997). Theoretical findings also suggest a slow reaction (Kurtén et
al., 2011). Our RO2 concentrations are of the same order or lower than
atmospheric [RO2] ([RO2] cannot exceed alkene conversion), and
therefore the slow reaction of RO2 + SO2 can be disregarded in
our study. Moreover, if RO2 + SO2 were a significant source
of [H2SO4] in atmospheric OH measurements relying on SO2
titration (Petäjä et al., 2009, and references therein), these
measurements would be wrong.
Also carbonyls or acids formed as products in the ozonolysis reaction or in
the reaction of OH with propane could affect the results via the reaction of
sCI + carbonyl/acid in competition with sCI + SO2. Carbonyl
concentration from alkene ozonolysis cannot exceed the reacted alkene
concentration (should be clearly lower). Total OH produced upon ozonolysis
cannot exceed reacted alkene concentration, and thus also carbonyls resulting
from OH + propane cannot exceed reacted alkene concentration. Thus,
maximum carbonyl concentrations in our experiment are in the range of a few
108 molecules cm-3, i.e. 3–6 orders of magnitude lower than
the SO2 concentrations used. The relative reactivity,
k(sCI + carbonyl) / k(sCI + SO2), is clearly below 1 (Taatjes et
al., 2012). Therefore, a potential role of the reaction sCI + carbonyl in
our experimental system can be completely ruled out. The same applies to the
reaction of organic acids with sCI. Acid formation yields are about 5 %
of reacted alkene, and thus the acid concentrations are in the range of
107 molecules cm-3 (4–7 orders of magnitude lower than the
SO2 concentrations). The relative reactivity,
k(sCI + acid) / k(sCI + SO2), is ∼ 3 as shown
later in this work and qualitatively in line with Welz et al. (2014). Thus,
sCI + acid cannot be competitive with sCI + SO2 in this
experiment.
Furthermore, the sCI wall loss could be important for the sCI balance. The
first-order rate coefficient for the diffusion-limited wall loss of sCI can
be estimated according to
kwall-loss = 3.65 ⋅ D / r2, where D is
the diffusion coefficient of sCI and r stands for the tube radius. As the
diffusion coefficient a value of 0.1 cm2 s-1 was assumed
(D(H2SO4) = 0.08 cm2 s-1), resulting in
kwall loss = 0.023 s-1. Thermal decomposition of sCI
is expected to be much more rapid (Welz et al., 2012; Berndt et al., 2012),
making the sCI wall loss negligible in the kinetic analysis.
Results according to Eqs. (3) and (4) from non-linear regression
analysis [H2SO4] = f([SO2]).
Alkene
k(loss) / k(sCI + SO2) molecules cm-3
[H2SO4]sCI molecules cm-3
sCI yield
α-Pinene
(2.4 ± 0.2) × 1012 RH: 10 % (2.0 ± 0.4) × 1012 RH: 50 %
(1.15 ± 0.02) × 108 RH: 10 % (1.13 ± 0.04) × 108 RH: 50 %
0.15 ± 0.07
Limonene
(2.4 ± 0.2) × 1012 RH: 10 % (2.1 ± 0.2) × 1012 RH: 50 %
(9.3 ± 0.1) × 107 RH: 10 % (9.3 ± 0.2) × 107 RH: 50 %
0.27 ± 0.12
Isoprene
(2.5 ± 0.1) × 1012 RH: 10 % (2.1 ± 0.5) × 1013 RH: 50 %*
(9.9 ± 0.1) × 107 RH: 10 % (9.7 ± 0.6) × 107 RH: 50 %
0.58 ± 0.26
* for a “two-sCI” model we get 3.3 × 1013
and 2.6 × 1011 assuming the same total
[H2SO4]sCI ,where the first sCI accounts for
∼ 85 % and the second for ∼ 15 %.
In the absence of organic acid added to the reaction gas, only Reactions (R2)–(R5)
are considered. The fraction of sCI that oxidizes SO2, producing
sulfuric acid (sCIH2SO4 / sCITOT), is equal to the sCI reaction rate
with SO2 (Reaction R4) divided by the sum of all reaction rates (total reaction
rate) of sCI (Reactions R3–R5):
sCIH2SO4sCITOT=ksCI+SO2[SO2]ksCI+SO2SO2+kdec+ksCI+H2O[H2O].
It follows for a given RH and k(loss) = k(dec) + k(sCI + H2O) ⋅ [H2O]
that the total concentration of sulfuric acid produced during the experiment is
H2SO4=11+klossksCI+SO2SO2⋅[H2SO4]sCI,
where [H2SO4]sCI stands for [H2SO4] from sCI
titration; i.e all sCI is converted to H2SO4 in the presence of
high SO2 concentrations via Reaction (R4), making Reactions (R3) and (R5)
negligible. In the analysis we assumed 28 % wall loss in total sulfuric
acid concentration (Berndt et al., 2014a), and the measured values were
corrected for the wall loss before being used in the data analysis.
The yield y2 of sCI from Reaction (R2) can be obtained from the
knowledge of the reacted alkene and [H2SO4]sCI assuming
a H2SO4 yield of unity from Reaction (R4):
y2=[H2SO4]sCIreacted[alkene].
The amount of reacted alkene and ozone was kept very small (less than 1 %
each), allowing us to calculate the reacted alkene concentration according to
Eq. (5) (see also Table 1):
reacted[alkene]=k(O3+alkene)⋅[alkene]⋅[O3]⋅t.
The relative rate coefficients k(loss) / k(sCI + SO2) and
the sCI yield y2 were obtained by least-squares fitting according to
Eqs. (3) and (4), using the experimental data depicted in Figs. 1–3.
The above approach assumes that all sCIs formed from a selected alkene show a
similar reactivity in Reaction (R3)–(R5), i.e. that we are able to describe only
average effects of all sCIs. Ozonolysis of a single alkene can result in the
production of different types of CI and thus sCI. In the case of α-pinene,
possible sCIs include two different isomers, one syn- and one which
can be either a syn- or anti-conformer, with syn
having two different structures possible; all in all sCI from α-pinene
ozonolysis can have four different structures. The same applies for limonene.
Nevertheless, the “one-sCI” approach seems to work well for α-pinene
and limonene, suggesting that one of the possible sCI structures dominates,
or that different sCIs show similar k(loss) / k(sCI + SO2).
With the help of the single-sCI model, Eq. (3), the experimental data are
described reasonably well (Figs. 1 and 2). For isoprene, due to the structure
of the parent alkene, five different sCI structures are possible. In this
case the one-sCI model is too simple for a reliable description of the
measurements; see below.
For monoterpenes, increasing the water vapour concentration by a factor of
5 did not change the results within the experimental uncertainties. This
indicates that thermal decomposition dominates the loss mechanism of sCI
under these conditions and the reaction with water vapour is of less
importance; i.e. k(dec.) ≫ k(sCI + H2O) ⋅ [H2O]
for [H2O] ≤ 2.9 × 1017 molecules cm-3. The
relative rate coefficients and yields from monoterpene ozonolysis experiments
are summarized in Table 2.
The experiments with isoprene showed a different behaviour. At low water
vapour concentration, RH = 10 %, the above approach of the
one-sCI model fits well to the experimental data. At an elevated water vapour
concentration (RH = 50 %), a significant drop in sulfuric acid
concentration is observed and the one-sCI model fails in describing the
measurements. The flaw of the one-sCI model can be explained by the
different reactivity of different sCIs toward water vapour. To account for
the possible differences in the reactivity of different sCIs, Eq. (3) was
extended to a “two-sCI model” considering a different reactivity of
sCII and sCIII in Reactions (R3)–(R5):
H2SO4=11+klossIksCI+SO2ISO2⋅[H2SO4]sCII+11+klossIIksCI+SO2IISO2⋅[H2SO4]sCIII.
Non-linear regression analysis to the experimental data suggests that the
“first sCI” (type I) is responsible on 85 % and the “second sCI”
(type II) on 15 % of the total measured [H2SO4]sCI.
Furthermore, it shows that the relative rates coefficients
k(loss) / k(sCI + SO2) are significantly different between
the two sCIs: 3.3 × 1013 molecules cm-3 for type I and
2.6 × 1011 molecules cm-3 for type II. From our
experiment we cannot draw clear conclusions on what kind of sCI formed from
the isoprene ozonolysis is responsible for type I and type II. It could be
speculated that CH2OO and/or an anti-conformer sCI causes the
strong RH dependence of produced sulfuric acid due to their efficient
reaction with water vapour (Reaction R3) in competition with Reaction (R4). The
relative rate coefficients and yields are summarized in Table 2.
If the fast reaction with water is due to CH2OO, that might have
implications for our understanding of CI chemistry and the inclusion of the
results by Welz et al. (2012) in models, including global chemical transport
models (Pierce et al., 2013). Stone et al. (2014) suggest a relatively slow
water reaction obtained by a technique similar to the approach by Welz et
al. (2012), while the older measurements suggest that the water reaction
dominates in the atmosphere over “all” other reactions; see for example
Hasson et al. (2001a). Recent work by Berndt et al. (2014b) also suggests
that reaction with water (dimer) is relatively fast dominating the
atmospheric fate of CH2OO. The relative rate coefficients
k(loss) / k(sCI + SO2) obtained in this study are close to
those obtained by Berndt et al. (2012) for sCI from the ozonolysis of
trans-2-butene and TME. Berndt et al. (2014a) also showed that significant
differences in the relative rate coefficients
k(sCI + H2O) / k(sCI + SO2) occurred between
syn- and anti-confomers of sCI from trans-2-butene.
Absolute measurements by Taatjes et al. (2012) support this finding as well.
Similar to Berndt et al. (2012, 2014a, b), our analysis of the relative rate
coefficients and sCI yields incorporates the yield of H2SO4 from
the sCI + SO2 reaction – i.e. our investigation is limited to the
channel leading to the formation of H2SO4. However, the yields
cannot be significantly below unity, since otherwise the obtained yields of
sCI should be higher by the same factor. As the sCI yield cannot exceed
unity, we conclude that the yield of H2SO4 from the sCI + SO2
reaction must be, if not unity, at least > 0.2 for monoterpene
sCIs and > 0.5 for isoprene sCIs. However, there is reason to
believe that the H2SO4 yields are much higher than that and thus
very close to unity; our measured sCI yield for α-pinene of
0.15 ± 0.07 assuming a unity H2SO4 yield from
sCI + SO2 is in excellent agreement with a yield of 0.15 recently
determined with an sCI-specific scavenger technique (Drozd and Donahue,
2011). These observations also call into question the stable,
non-SO3-producing, sulfur-bearing secondary ozonides, theoretically investigated by
Kurtén et al. (2011) and Vereecken et al. (2012), as a predominant
product from the sCI + SO2 reaction.
Our results on the relative rate coefficient can be compared to those
calculated from the data reported by Welz et al. (2012), who, as discussed
above, studied the simplest possible Criegee (CH2OO) in a low-pressure
system. They found
k(sCI + SO2) = 3.9 × 10-11 cm3 molecule-1 s-1
to be the lower-end estimation for the lifetime against decomposition of 2 ms –
resulting in upper-end estimation for k(dec) of 500 s-1 – and the
upper-end estimation for reaction coefficient with H2O of
k(sCI + H2O) < 4 × 10-15 cm3 molecule-1 s-1.
Using the upper-end rate coefficients for CH2OO in
k(loss) = k(dec) + k(sCI + H2O) ⋅ [H2O],
for the relative reaction rate coefficients
k(loss) / k(sCI + SO2) follow
< 1.9 × 1013 molecules cm-3 (RH = 10 %)
and < 4.2 × 1013 molecules cm-3 (RH = 50 %), being qualitatively not in contradiction with our results for the
sCIs of the monoterpenes and isoprene; see Table 2. If the reaction of
CH2OO with H2O dominated the loss process
(k(dec) ≪ k(sCI + H2O) ⋅ [H2O]), the
resulting k(loss) / k(sCI + SO2) ratios from Welz et
al. (2012) data would be
< 5.9 × 1012 molecules cm-3 (RH = 10 %)
and < 3.0 × 1013 molecules cm-3
(RH = 50 %), still not contradicting our findings. A more detailed
comparison is impossible because the study by Welz et al. (2012) yielded only
upper limits for the rate coefficients of the sCI decomposition step (R5) and
the reaction of sCI with H2O (Reaction R3). Since pressures in our work
and the study by Welz et al. (2012) were completely different, differences
related to pressure effects may arise. It is to be noted also here that
Reactions (R3) and/or (R5) describe the most important atmospheric loss
processes. For a reliable assessment of the importance of H2SO4
formation in the atmosphere via sCI + SO2, the sCI main reactions
(Reactions R3 and R5) must be characterized very well.
Reaction of sCI with organic acids
The reaction of acetone oxide ((CH3)2COO, sCI from TME ozonolysis)
with small organic acids was investigated by a competitive reaction kinetics
experiment at constant SO2 concentration
(3.2 × 1012 molecules cm-3) and varying the concentration
of the organic acids (Fig. 4).
Experimental data from the competitive reaction kinetics
experiments, sCI + SO2 vs. sCI + acid,
sCI ≡ (CH3)2COO (from TME ozonolysis). The lines show the
best-fit result of the non-linear regression analysis from Eq. (7).
Berndt et al. (2014a) showed that for acetone oxide the reaction with water
vapour (Reaction R3) is of less importance compared with the thermal
decomposition (Reaction R5) for RH ≤ 50 %. Our experiment was
conducted at low RH (10 %), and thus only the thermal decomposition of sCI
was considered together with the reactions of sCI with SO2 (Reaction R4)
and the acids (Reaction R6). From pathways (R4)–(R6) follows
H2SO4=11+kdecksCI+SO2SO2+ksCI+acid[acid]ksCI+SO2SO2⋅[H2SO4]sCI.
The relative rate coefficient k(dec) / k(sCI + SO2) was
determined by Berndt et al. (2014a) to be
4.2 × 1012 molecules cm-3. [H2SO4] stands
again for the loss-corrected sulfuric acid concentration at the IfT-LFT
outflow and [H2SO4]sCI for [H2SO4] from sCI
titration. Results from the non-linear regression analysis
[H2SO4] = f([acid]) from Eq. (7) yield the free parameters
k(sCI + acid) / k(sCI + SO2) and
[H2SO4]sCI; see Table 3. Our measurements reveal an
about-3-times-faster reaction of acetone oxide with the acids compared
with the reaction with SO2.
Relative reaction rate coefficients for reaction of
(CH3)2COO (sCI from TME ozonolysis) with small organic acids and
SO2 based on competitive reaction kinetics experiments.
Acid
k(sCI + acid)/
[H2SO4]sCI,
k(sCI + SO2)
molecules cm-3
HCOOH, formic acid
(2.80 ± 0.32)
2.05 × 108
CH3COOH, acetic acid
(3.43 ± 0.22)
2.05 × 108
Neeb et al. (1997) measured rapid reaction of sCI + HCOOH in comparison
to sCI + water. However, our results cannot be directly compared to Neeb
et al. (1997) due to different water reactivity of CH2OO (studied by
Neeb et al., 1997) and (CH3)2COO investigated here (Berndt et al.,
2014b). However, our result is very similar to the reaction rate coefficients
reported by Welz et al. (2012; 2104) for reactions of CH2OO with
SO2 and HCOOH / CH3COOH. Welz et al. (2012; 2014) studies
demonstrate the relative reactivity of 2.8 and 3.3 for
k(sCI + HCOOH) / k(sCI + SO2) and for
k(sCI+CH3COOH) / k(sCI + SO2), respectively. These
values are stunningly close to our measured values of 2.8 and 3.4, though it
should be kept in mind that our sCI represents (CH3)2COO while Welz
et al. (2012, 2014) data are for CH2OO.
Now we discuss the potential importance of the sCI + acid reaction in the
atmosphere. Mauldin III et al. (2012) estimated the absolute reaction rate
coefficient of sCI + SO2 to be in the range of
6 × 10-13 cm3 molecule-1 s-1 for
α-pinene and limonene. Using this value, the absolute reaction rate
coefficient for sCI + HCOOH and sCI+CH3COOH would be in the range
of a few 10-12 cm3 molecule-1 s-1. The reaction rate
coefficient for OH + CH3COOH is
8 × 10-13 cm3 molecule-1 s-1 and for
OH + HCOOH 4 × 10-13 cm3 molecule-1 s-1.
As shown by Mauldin III et al. (2012), the summertime sCI concentrations are
similar to OH peak concentrations in boreal forest. Therefore, it is
possible that sCI oxidation plays a crucial role for the HCOOH and
CH3COOH budget. However, as stated above, the conformation and structure
of sCI probably have major effects on the sCI reactivity toward acids, and
therefore further investigations are required before our observations can be
generalized to sCIs other than acetone oxide. On the other hand, Welz et
al. (2014) measured for CH2OO + HCOOH / CH3COOH rate
coefficients exceeding 10-10 cm3 molecule-1 s-1,
suggesting that this structurally different sCI (CH2OO) also reacts with
acids rapidly in comparison to the OH radical reaction.
It is important to notice that, even though from sCI point of view SO2
or organic acids were the minor sinks for sCI, the situation can be
completely opposite from the point of view of formation of H2SO4 (Boy
et al., 2013; Pierce et al., 2013) or the loss of organic acids (Welz et
al., 2014). Most likely, the dominating reactions controlling the sCI
concentrations in natural environments are thermal decomposition and
reaction with water vapour. Very high concentrations (several to tens of
ppb) of SO2 or organic acids would be required to alter the sCI
budget significantly. However, the reaction with sCI can still be a
significant or even the main fate of acid or source of H2SO4.
Regarding the role of sCI in atmospheric gas phase
H2SO4 production, the present study is in a reasonable agreement
with the results by Mauldin III et al. (2012). Therefore results by Boy et
al. (2013), who applied Mauldin III et al. (2012) findings in boundary layer
modelling, can be considered valid as well. Pierce et al. (2013) applied Welz
et al. (2012) results in a chemical transport model for studying the role of
sCI in global gas phase H2SO4 burden. Because our relative rate
coefficient between sCI loss and sCI + SO2 are not conflicting with the
results by Welz et al. (2012), our present results validate, form one side,
also the modelling study by Pierce et al. (2013). Thus, our understanding of
the role of sCI in atmospheric H2SO4 production remains unchanged.