Comparison of measured and modeled [NO] and [NO2] values
following O(1D) + N2O and NO + O3 reactions
Figure compares modeled and measured NO mixing ratios
obtained following 80 s residence time in the reactor at the operating
conditions described in Sect. . The corresponding integrated
OH exposures are approximately 2.6 × 1011 and 2.4 × 1012 molec cm-3 s,
respectively, in the absence of added N2O.
Symbols are colored by [N2O], which ranged from 0 to 3 %.
Measured [NO] ranged from 0 to 10.4 ppb and increased with increasing
[N2O], as expected, at both [H2O] = 0.07 and 1 %. The mean
ratio of modeled to measured [NO] was 0.94 ± 0.19 at [H2O] = 0.07 %
and 3.85 ± 2.33 at [H2O] = 1 %.
Scatterplot comparing measured and modeled [NO] at 80 s residence
time in the PAM oxidation flow reactor;
I254 = 4 × 1015 ph cm-2 s-1,
[O3] = 1 ppm, [H2O] = 0.07 and 1 %, and
[N2O] = 0 to 3 %.
Symbols are colored by [N2O], with 1 : 2, 1 : 1, and 2 : 1 lines shown for reference.
Error bars represent ±60 % uncertainty in model outputs
and ±40 % precision in replicate [NO]
measurements at fixed [N2O].
NO2, which is formed by the NO + O3 reaction, is more
straightforward to measure under these conditions because NO2 reacts
approximately 500 times slower than NO with O3.
Thus, a comparison of modeled and measured [NO2] provides
additional method evaluation with less uncertainty than [NO] measurements.
Figure compares corresponding modeled and measured
NO2 mixing ratios obtained during the same experiments described in
Fig. 1. As expected, [NO2] increased with increasing [N2O]
because of faster NO + O3 reaction rate from increasing [NO]. At
[H2O] = 0.07 %, measured [NO2] ranged from 0 to 291 ppb,
whereas at [H2O] = 1 %, measured [NO2] ranged from 0 to 59 ppb.
[NO2] was lower in the latter case because additional OH was formed from
O(1D) + H2O reactions (Sect. ), which increased
the rate of the OH + NO2 reaction.
The mean ratio of modeled to measured [NO2] was 0.72 ± 0.39 at
[H2O] = 0.07 % and 1.05 ± 0.50 at
[H2O] = 1 %. These results, combined with results shown in
Fig. , suggest that an uncharacterized H2O- or
HNO3-related artifact negatively biased the measured [NO] values at
[H2O] = 1 % and that the photochemical model described in
Sect. may be used to evaluate a wider range of reactor
operating conditions. The model also constrains mixing ratios of radical
species such as HO2 that are difficult to measure directly or require
additional measurement techniques .
Optimal reactor operating conditions for O(1D) + N2O reactions
To investigate optimal operating
conditions for NOx generation, we implemented the model described
in Sect. over operating conditions
I254 = 3.2 × 1013 to
6.4 × 1015 ph cm-2 s-1, [O3] = 0.5 to
50 ppm, and [H2O] = 0.07 to 2.3 % at 22 ∘C,
respectively, as a function of [N2O] = 0 to 5 %.
These values span the nominal range of operating conditions that can be
achieved with the PAM reactor. To facilitate independent evaluation of the
effects of [O3] and I254 on [NO], we restricted our analysis to
conditions that use only 254 nm photolysis. Using both 185 and 254 nm
photolysis provides additional sources of O(1D) and OH from
N2O and H2O photolysis at 185 nm, respectively, at the
expense of independent control of [O3] and I254.
Figure shows the modeled steady-state [NO] in the reactor as
a function of [N2O] = 0 to 5 %, assuming a mean residence
time of 80 s, [H2O] = 1 %, and [O3] = 5 ppm. In
addition, Figs. S1–S3
in the Supplement show modeled NO : HO2 and OH : NO3 ratios as a function
of input [N2O], with I254, [O3], and [H2O] each varied individually, while other input conditions are fixed.
The following observations that are obtained from Figs. and S1–S3 were used to identify the optimal operating conditions:
At fixed [O3], [H2O], and [O(1D)], [N2O] and [NO] increase with increasing I254 (Figs. and S1).
At fixed I254, [H2O], and [N2O], increasing O3 increases the production and loss
rates of NO from O(1D) + N2O and NO + O3
reactions, respectively. The relative importance of NO + OH,
NO + O3, and NO + NO3 reactions, which depend on
[N2O] and [O3], further influence [NO]:
At [N2O] ∼ 1 %, increasing [O3] from 0.5 to 5 ppm increases
[NO] because the reaction rate of NO + OH decreases relative to
NO + O3 (Fig. S2a).
At [N2O] > 1 %, increasing [O3] from 5 to 50 ppm decreases [NO] because the
reaction rate of NO + NO3 increases relative to
NO + O3 (Fig. S2a).
At fixed I254, [H2O], and [N2O], increasing [O3] decreases [NO] : [HO2]
and [OH] : [NO3] by increasing NO2 and NO3 formation from NO + O3 and NO2 + O3 reactions.
At fixed I254, [O3], and [N2O], increasing [H2O] increases [OH] : [NO3]
by increasing OH production from H2O + O(1D) reactions (Fig. S3).
Scatterplot comparing measured and modeled
[NO2] at 80 s residence time in the PAM reactor;
I254 = 4 × 1015 ph cm-2 s-1, [O3] = 1 ppm,
[H2O] = 0.07 and 1 %, and [N2O] = 0 to 3 %.
Symbols are colored by [N2O], with 1 : 2, 1 : 1, and 2 : 1 lines shown for reference.
Error bars represent ±60 % uncertainty in model outputs
and ±20 % precision in replicate [NO] measurements at
fixed [N2O].
The relative importance of these operating conditions is situationally
dependent on the relative OH, O3, and NO3 rate constants of
the target species and photochemical age.
To demonstrate proof of principle, we present NO3--CIMS spectra
of isoprene and α-pinene oxidation products in the following sections.
NO3--CIMS spectra of isoprene oxidation products
Figure shows NO3--CIMS mass spectra of
products generated from the oxidation of isoprene (C5H8) that
cluster with NO3- ions to form NO3--species
adducts. Ion signals are plotted as a function of mass-to-charge ratio
(m/Q). NO3- adduct formation is a relatively low-energy
process that does not result in fragmentation of the analyte
.
Thus, the measured ion signals are directly related to the chemical formulas of individual species that are generated in the reactor.
Ion signals corresponding to isoprene oxidation products shown in
Fig. were colored based on classification in ion groups
containing two–five carbon atoms with zero (C4H4,6,8O4-7 and
C5H6,8,10,12O3-8), one (C2-3H3,5NO5 and
C5H7,9,11NO6-11), and two (C5H10N2O8-10)
nitrogen atoms, where we assumed that nitrogen atoms were associated with
nitrate functional groups and not heterocyclic compounds. We also assumed that
nitrate functional groups are formed from RO2 + NO or
RO2 + NO2 reactions (Sect. ). To examine
changes in relative contributions of C4H4,6,8O4-7,
C5H6,8,10,12O3-8, C5H7,9,11NO6-11, and
C5H10N2O8-10 ions as a function of added NOx, we
made two simplifying assumptions: (1) the NO3--CIMS had the same
sensitivity to all species that were detected, and (2) HNO3 generated
in the reactor did not alter the relative selectivity of the CIMS to
different classes of oxidation products, as has been observed in some cases
.
Modeled steady-state [NO] as a function of
[N2O] input to the PAM reactor at I254 = 0.032×1015,
0.64 × 1015, and 6.4 × 1015 ph cm-2 s;
[H2O] = 1 %; [O3] = 5 ppm; and mean residence time = 80 s.
Error bars represent ±60 % uncertainty in modeled [NO] .
To generate spectra shown in Fig. , the reactor was
operated at I254 = 4.2 × 1013 and
2.1 × 1015 ph cm-2 s-1,
[H2O] = 1 %, and [N2O] = 0 and 3 %. As shown
in Figs. S4 and S5, corresponding OH exposures ranged within
(5.6–6.3) × 109 (Fig. a and c; calculated
> 43 % of isoprene reacted) and within
(0.43–1.4) × 1012 molec cm-3 s
(Fig. b and d; calculated ∼ 100 % of isoprene
reacted), respectively. At low OH exposure, the OH suppression at “high
NOx” relative to “low NOx” was comparatively minor
(11 %), whereas at high OH exposure, the OH suppression at high
NOx relative to low NOx was larger (69 %).
At the high-NOx OH exposure of
4.3 × 1011 molec cm-3 s, isoprene can react with
OH up to 43 times in the reactor. This presumably exceeds the number of OH
reactions (followed by RO2 + NO reactions) that are necessary to
fragment or condense oxidation products to the point where they are no longer
detected with NO3--CIMS. Thus, it is unlikely that OH
suppression at high OH and high NOx significantly
affected the NO3--CIMS spectra shown in Fig. .
To aid interpretation of results shown in Fig. ,
Fig. summarizes several known isoprene + OH reaction
pathways that are terminated by reactions of RO2 with HO2,
NO, or NO2. As will be discussed in the following sections, these
pathways yield multigenerational oxidation products with chemical formulas
corresponding to the major ions that are plotted in Fig. .
NO3--CIMS mass spectra of
isoprene oxidation products generated in the PAM reactor at
[H2O] = 1 %, [O3] = 5 ppm, and mean residence
time = 80 s:
(a) I254 = 4.2 × 1013 ph cm-2 s-1,
[N2O] = 0 %;
(b) I254 = 2.1 × 1015 ph cm-2 s-1,
[N2O] = 0 %; (c)
I254 = 4.2 × 1013 ph cm-2 s-1,
[N2O] = 3.2 %; and
(d) I254 = 2.1 × 1015 ph cm-2 s-1,
[N2O] = 2.9 %. NO3--CIMS mass spectra of the
same compounds detected at the SOAS ground site in Centreville, Alabama, USA,
during (e) “low-NO” and (f) “high-NO” conditions (see
text for additional details; C5H6O5-7 ions removed from SOAS
spectra due to larger contributions from α-pinene + OH oxidation
products; Fig. ). “Cx” or “Ox”
indicates number of carbon or atoms in labeled ions (not including oxygen
atoms in nitrate functional groups).
NO3--CIMS spectral features observed at low-NOx conditions
C4-5H4-12O3-8 ions comprised 93 and 97 % of the signals
at low and high OH exposure (Fig. a and c, respectively).
The C5H7-11NO6-11 signals that were observed here may be due to
background NOx in the reactor (Sect. ). The signal
at m/Q = 230, C5H12O6 (NO3- omitted for
brevity here and elsewhere), was the largest signal detected at both low and
high OH exposures at low-NOx conditions.
Figure suggests this species is a second-generation
oxidation product generated from two reactions with OH and two
RO2 + HO2 termination reactions and is typically associated with isoprene SOA formation and
growth under low-NOx conditions . Signals
in Fig. b and d are approximately 10 times higher than in
Fig. a and c because additional OH exposure produces higher
yields of multigeneration oxidation products that are detected with
NO3--CIMS.
Previously identified multigeneration isoprene oxidation products such as
C5H10O5, C5H12O5, and C5H10O6
were also detected at
significant intensity under low-NOx conditions. These species are
formed after two reactions with OH, one RO2 + HO2
termination reaction, and one RO2 + RO2 termination
reaction (Fig. ). When the OH exposure was increased from
6.3 × 109 to
1.4 × 1012 molec cm-3 s, the signal at
C5H12O6 increased by a factor of 10, and the signal at
m/Q = 246, C5H12O7, increased by a factor of 5. At high OH
exposure, C5H12O7 was the second-largest peak in the spectrum.
These highly oxygenated isoprene oxidation products are likely also important
in SOA formation processes. C5H10O7 is a proposed tri-hydroperoxy
carbonyl product formed after one reaction with OH, two hydrogen shifts, and
one RO2 + HO2 termination reaction as shown in
Fig. .
We hypothesize that C5H10O7, C5H12O7, and
C5H10O8 are more prominent in our spectra than in other studies because
NO3- is more selective to highly oxidized species than other reagent ions
.
Simplified reaction scheme summarizing known isoprene + OH
reaction pathways yielding multigeneration oxidation products. Four peroxy
radical (RO2) isomers are generated following initial OH addition to
isoprene: 1,2-RO2, 4,3-RO2, 1,4-RO2, and 4,1-RO2.
The 1,2-RO2 and 4,3-RO2 isomers follow the same reaction pathways, yielding chemical formulas with green text
that were detected with NO3--CIMS. The 4,1-RO2 isomer
yields C5H10O7, also detected with NO3--CIMS.
Chemical formulas with red text may be generated through the methacrolein
(MACR) channel but were not detected with
NO3--CIMS.
NO3--CIMS spectral features observed at high-NOx conditions
Following addition of N2O at ∼ 3 % mixing ratio, the
NO3--CIMS spectra changed significantly at low and high OH
exposures (Fig. b and d). The signals of
C4-5H4-12O3-8 oxidation products decreased, although the
C4H4,6,8O4-7 : C5H6,8,10,12O3-8 ratio increased,
presumably due to decomposition of alkoxy (RO) radicals generated from
C5 RO2 + NO reactions into C4 products. The
C2-3H3,5NO5 (peroxyacetyl nitrate (PAN) and peroxypropionyl
nitrate (PPN)), C5H7,9,11NO6-11, and C5H10N2O8-10
signals increased. At low OH exposure, C2-3H3,5NO5,
C5H7,9,11NO6-11, and C5H10N2O8-10 signals
constituted 2, 38, and 7 % of the NO3--CIMS signals,
respectively (Fig. c). The largest signal in this spectrum
was m/Q = 259, C5H11NO7. Fig. suggests
this compound is a second-generation oxidation product that is formed after
two reactions with OH, one RO2 + NO termination reaction, and one
RO2 + HO2 termination reaction . The
signal observed at m/Q = 288, C5H10N2O8, is a
second-generation oxidation product that is formed after two reactions with OH and two RO2 + NO
termination reactions (Fig. ) .
Other ion signals associated with dinitrate species included m/Q = 304,
C5H10N2O9, and m/Q = 320, C5H10N2O10.
Related signals were detected at m/Q = 351 and 367 (not shown), which
we assume represent (HNO3NO3-)C5H10N2O8 and
(HNO3NO3-)C5H10N2O9 because we are not aware of other
feasible (NO3-)C5 adducts at these mass-to-charge ratios.
NO3--CIMS mass spectra of α-pinene oxidation
products generated in the PAM reactor at [H2O] = 0.07 %,
[O3] = 5 ppm, and mean residence time = 80 s:
(a) I254 = 1.8 × 1015 ph cm-2s-1, [N2O] = 0 %;
(b) I254 = 1.8 × 1015 ph cm-2s-1, [N2O] = 3.2 %.
(c) NO3--CIMS mass spectra of the same compounds
detected at the SOAS ground site in Centreville, Alabama, USA during
“high-NO” conditions shown in Fig. f (note:
C5H7NO6-11 signals in SOAS spectra also contributed from
isoprene + OH oxidation products). “Cx” or “Ox”
labels indicate number of oxygen atoms in corresponding signals (not
including oxygen atoms in nitrate functional groups).
At high OH exposure, the same C5H7,9,11NO6-11 and
C5H10N2O8-10 species observed at low OH exposure were
detected, albeit at higher concentrations and at higher dinitrate : nitrate. This
is presumably due to higher NO : HO2 achieved at higher I254
and fixed [N2O] (Figs. , S2, S5–S6).
C2-3H3,5NO5, C5H7,9,11NO6-11, and
C5H10N2O8-10 signals made up 0.3, 33, and 56 %,
respectively, of the NO3--CIMS spectrum shown in
Fig. d, where C5H10N2O8 was the largest signal
that is detected.
To demonstrate our ability to mimic atmospheric NOx-dependent
photochemistry, Fig. e and f show
C4H4,6,8O4-7, C5H6,8,10,12O3-8,
C2-3H3,5NO5, C5H7,9,11NO6-11, and
C5H10N2O8-10 ion signals detected in NO3--CIMS
spectra at the SOAS ground site in Centreville, Alabama, USA. The spectra
shown were obtained on 25 June 2013 (07:30–11:00) and 4–5 July 2013
(12:00–00:00) which represented periods with sustained high and low NO
mixing ratios of 0.53 ± 0.17 and 0.024 ± 0.025 ppb,
respectively, measured at the site. Figure a, c, and e
indicate that adding N2O to the reactor increases the similarity
between the composition of isoprene oxidation products generated at lower
photochemical age in the reactor (Fig. a and c) and under
low-NO ambient conditions (Fig. e). Likewise,
Fig. b, d, and f indicate that adding N2O to the
reactor increases the similarity between the composition of isoprene
oxidation products generated at higher photochemical age in the reactor
(Fig. b and d) and at high-NO ambient conditions
(Fig. f). (HNO3NO3-)C5H10N2O8-9
adducts were also observed in Fig. f (not shown).
Influence of acylperoxy nitrates from RO2 + NO2 reactions
Acylperoxy nitrates
(APNs) may be generated from reactions of aldehydic, biogenic VOC oxidation
products with OH followed by RO2 + NO2 termination
reactions (e.g., ). PAN (C2H3NO5) and PPN
(C3H5NO5) are minor components (< 2 %) of the spectra shown
in Fig. c–f.
A comparison of Fig. c and e suggests that yields of
PAN and PPN are not enhanced in the reactor compared to atmospheric conditions.
Additional APNs may be generated following the OH oxidation of methacrolein,
a first-generation isoprene oxidation product. Methacryloyl peroxy nitrate
(MPAN, C4H5NO5) is a second-generation oxidation product formed
after one methacrolein + OH reaction and one
RO2 + NO2 termination reaction .
C4-hydroxynitrate-PAN (C4H6N2O9) is a third-generation oxidation
product formed through the methacrolein channel after three reactions with
OH, two RO2 + NO termination reactions, and one
RO2 + NO2 termination reaction .
Neither C4H5NO5 nor C4H6N2O9 was detected in the
laboratory and ambient NO3--CIMS spectra shown in
Fig. c–f. Either these compounds were oxidized or
thermally decomposed prior to detection, or their signals were below
detection limit. C4H7NO5, which is formed after one
methacrolein + OH reaction and one RO2 + NO termination
reaction , was detected (Fig. ).
Taken together, these observations suggest that yields of APNs were not significantly enhanced in the reactor compared to atmospheric conditions.
Influence of isoprene + NO3 reactions
Based on the calculated isoprene + OH and isoprene + NO3
reaction rates (Figs. S5–S6), we assume that isoprene + NO3
reactions had a minor influence on the NO3--CIMS spectra shown
in Fig. c and d. This assumption is further supported by
the similarity between laboratory and ambient NO3--CIMS spectra,
the latter of which was obtained during the daytime and thus with minimal
NO3 exposure (07:30–11:00 for the high-NO spectra shown in
Fig. f). Specific operating conditions different than those
used in this study could increase the relative influence of
isoprene + NO3 reactions. In this hypothetical situation,
enhanced yields of C5H7NO5, C5H8N2O8, and
C5H10N2O8 might occur following two reactions with NO3
. In addition, C5H10N2O9 may be generated
from one isoprene + NO3 reaction followed by one
RO2 + HO2 termination reaction .
All four of these ions are detected in the spectra shown in
Fig. , although C5H8N2O8 (not shown in
Fig. ) is present at 0.5 % of the intensity of
C5H10N2O8. If C5H8N2O8 : C5H10N2O8 is
significantly different under NO3-dominated conditions, this ratio
could distinguish the relative rates of isoprene + OH and
isoprene + NO3 reactions. Otherwise, it is not clear that the
expected product distributions are significantly different whether isoprene
is oxidized by OH or NO3 in the presence of NOx.
NO3--CIMS spectra of α-pinene oxidation products
Figure shows NO3--CIMS mass spectra of products
generated from the oxidation of α-pinene (C10H16). Ion
signals corresponding to α-pinene oxidation products were colored
based on classification in C5H6,8O5-7,
C6-9H8,10,12,14O6-12, C10H14,16,18O5-14, and
C19-20H28,30,32O9-18 ion groups containing zero nitrogen
atoms; C2-3H3,5NO5, C5H7NO6-11,
C6-9H9,11,13,15NO5-10, and C10H15,17NO4-14
ion groups containing one nitrogen atom; and a
C10H16,18N2O6-13 ion group containing two nitrogen atoms. As
was the case with isoprene oxidation products, we assumed nitrogen atoms
present in α-pinene oxidation products were associated with nitrate
functional groups formed from RO2 + NO or
RO2 + NO2 reactions. Additionally, we again assumed that the
NO3--CIMS had the same sensitivity to all species that were
detected and that HNO3 generated in the reactor did not alter the
relative selectivity of the CIMS to different classes of oxidation products.
Because the oxidation pathways leading to α-pinene-derived HOMs are
significantly more complex than those leading to isoprene-derived HOMs, the
analogous figure to Fig. for α-pinene-derived HOMs
is beyond the scope of this paper.
To generate the spectra shown in Fig. , the reactor was operated
at I254 = 1.8 × 1015 ph cm-2s-1,
[H2O] = 0.07 %, and [N2O] = 0 and 3.2 %. In
this experiment, lower [H2O] was used to minimize [OH] and facilitate
closer comparison with spectra from previous NO3--CIMS studies
of α-pinene + O3 oxidation products generated at
low-NOx conditions . As shown
in Fig. S7, corresponding OH and O3 exposures ranged within
(0.13–1.1) × 1011 molec cm-3 s and within
(7.2–9.3) × 1016 molec cm-3 s for the low- and
high-NOx conditions, respectively. To first order, at OH and
O3 exposures of 1.3 × 1010 and
7.2 × 1015 molec cm-3 s that are attained at
[N2O] = 3.2 %, α-pinene should react once with each
oxidant in the gas phase. Thus, at the highest [N2O] used, yields of
second-generation (or later) α-pinene + OH oxidation products
detected with the NO3--CIMS were minimized relative to
α-pinene + O3 first-generation oxidation products, as
desired . However, a potential consequence of using
O(1D) + N2O reactions to study the
NOx dependence of chemical systems similar to those examined by
is that RO2 may be produced from
α-pinene + NO3 reactions in addition to
α-pinene + O3 or α-pinene + OH reactions
(Sect. and Fig. S7).
NO3--CIMS mass spectral features observed at low-NOx conditions
C5H6,8O5-7, C6-9H8,10,12,14O6-12,
C10H14,16,18O5-14, and C19-20H28,30,32O9-18
ion groups comprised 5, 36, 46, and 4 % of the signal
detected at low-NOx conditions (Fig. a),
respectively, assuming equal CIMS sensitivity and transmission to all detected species.
The C10 monomers and C19-20 dimers compounds that were
observed are often associated with atmospheric new particle formation events
. The prominent C10H14,16O7-9 signals
detected at m/Q = 308, 310, 324, 326, 340, and 342 in our measurements
were dominant signals in previous laboratory and field experiments influenced
by the ozonolysis of α-pinene emissions
. Other signals that
were observed correspond to C5-9 species that were generated
following carbon–carbon bond cleavage of the C10 backbone
. The remaining ∼ 10 % of the signal was
classified into C2-3H3,5NO5, C5H7NO6-11,
C6-9H9,11,13,15NO5-10, and C10H15,17NO4-14
ion groups and may be due to background NOx in the reactor
(Sect. ).
NO3--CIMS mass spectral features observed at high-NOx conditions
As was the case with NO3--CIMS spectra of isoprene oxidation
products, the addition of N2O to the reactor significantly changed
the mass spectrum of α-pinene oxidation products
(Fig. b). At [N2O] = 3.2 %, organic nitrates
and dinitrates comprised 65 % of the total ion signal. We observed
reduction in C6-9H8,10,12,14O6-12,
C10H14,16,18O5-14, and C19-20H28,30,32O9-18
signals, along with increases in C5H6,8O5-7,
C5H7O6-11, C6-9H9,11,13,15NO5-10,
C10H15,17NO4-14, and C10H16,18N2O6-13
signals. The C10 dinitrates may originate from two
α-pinene + OH reactions followed by two RO2 + NO
reactions, but they may also include contributions from one
α-pinene + NO3 reaction followed by one
RO2 + NO reaction. The largest signal in Fig. b
was observed at m/Q = 240, C5H6O7. The largest organic
nitrate signals in this spectrum were at m/Q = 329,
C8H13NO9, followed by C10H15NO9 (m/Q = 355),
C10H16N2O9 (m/Q = 354), and C10H15NO8
(m/Q = 339).
Figure c shows C5H6O5-7,
C6-9H8,10,12,14O6-12, C10H14,16,18O5-14,
C19-20H28,30,32O9-18, C2-3H3,5NO5,
C5H7NO6-11, C6-9H9,11,13,15NO5-10,
C10H15,17NO4-14, and C10H16,18N2O6-13
signals detected with NO3--CIMS at the Centreville site during
the SOAS campaign. The spectra shown here were obtained during the sampling
period shown in Fig. f and, given the large number of
compounds, may include contributions from HOM precursors other than
α-pinene. A comparison of Fig. a–c indicates that
adding N2O to the reactor increases the similarity between the
composition of α-pinene oxidation products generated in the reactor
and under high-NO ambient conditions, especially in regards to the
enhanced C5H6O5-7, C6-9H9,11,13,15NO5-10,
C10H15,17NO4-14, and C10H16,18N2O6-13
signals.
Normalized NO3--CIMS signals of
C5H6,8,10,12O3-8, C5H7,9,11NO6-11, and
C5H10N2O8-10 isoprene oxidation products generated in the PAM
reactor at I254 = 2.1 × 1015 ph cm-2 s-1,
[H2O] = 1 %, [O3] = 5 ppm, and mean residence
time = 80 s as a function of modeled NO : HO2.
For each of the species classes, signals were normalized to the maximum signal.
Proposed structures for C5H12O6, C5H11NO7, and
C5H10N2O8 signals are shown as representative compounds for each
species class (Fig. ).
Representative error bars indicate ±1σ uncertainty in
NO3--CIMS signals and ±85 % uncertainty in NO : HO2.
Detection of acylperoxy nitrates from RO2 + NO2 reactions
Figure b and c indicate that PAN (m/Q = 183,
C2H3NO5) and PPN (m/Q = 197, C3H5NO5) are formed at
lower yields (< 0.4 %) than were observed with isoprene
(Fig. c and d), suggesting that PAN and PPN formation from
reaction of α-pinene-derived RO2 with NO2 is not
enhanced in the reactor compared to atmospheric conditions.
C9H13NO6 and C10H15NO6-8 are APNs generated
following OH oxidation of pinonaldehyde, a major first-generation oxidation
product of α-pinene, with termination by RO2 + NO2
reaction . All four compounds are detected in the
reactor and ambient NO3--CIMS spectra shown in
Fig. b and c, with C10H15NO6-8 signals among
the largest in the spectra.
If these signals represent APNs, they appear to be important in both laboratory and atmospheric conditions.
Influence of α-pinene + NO3 reactions
Our calculations suggest that α-pinene + NO3 reactions
compete with α-pinene + OH reactions at the experimental
conditions used to generate the NO3--CIMS spectrum shown in
Fig. b (Fig. S7). If this were the case, enhanced yields of
C10H15NO6 would be anticipated from
α-pinene + NO3 reaction to generate pinonaldehyde,
followed by pinonaldehyde + NO3 reaction and
RO2 + NO2 termination . Other
minor α-pinene + NO3 products detected with CIMS include
C10H15NO5, C9H13NO6, C10H16N2O7, and
C10H15NO9 . We hypothesize that, if
α-pinene + NO3 reactions influence the spectrum shown in
Fig. b, C10H15NO6 : C10H15NO8
should be higher in Fig. b than in Fig. c.
Instead, the C10H15NO6 : C10H15NO8 ratio was
0.12 in the reactor and 0.28 at the Centreville site during a daytime period
(07:30–11:00) with presumably negligible NO3 influence.
Normalized NO3--CIMS signals of (a)
C5H6,8O5-7, C6-9H8,10,12,14O6-12,
C10H14,16,18O5-14, and C19-20H28,30,32O9-18, and
(b) C5H7NO6-11, C6-9H9,11,13,15NO5-10,
C10H15,17NO4-14, and C10H16,18N2O6-13
α-pinene oxidation products generated in the PAM reactor at
I254 = 1.8 × 1015 ph cm-2 s-1,
[H2O] = 0.07 %, [O3] = 5 ppm, and mean residence
time = 80 s as a function of modeled NO : HO2.
For each of the species classes, signals were normalized to the maximum signal.
Representative error bars indicate ±1σ uncertainty in
NO3--CIMS signals and ±85 % uncertainty in modeled NO : HO2.
Dinitrates (C10H16,18N2O6-13) shown in
Fig. b may originate from two α-pinene + OH
reactions followed by two RO2 + NO terminations, or one
α-pinene + NO3 reaction followed by one
RO2 + NO termination. Given comparable calculated OH and
NO3 reaction rates under these conditions (Fig. S7e), we hypothesize
that the majority of dinitrate signals should originate from
α-pinene + NO3 reactions if their yields are not
oxidant-dependent. If this were the case,
C10H16,18N2O6-13 : C10H15,17NO4-14 should
be larger in Fig. b than in c. However,
C10H16,18N2O6-13: C10H15,17NO4-14 was 0.23
in the reactor and 0.61 at the Centreville site. Thus, while the calculated
α-pinene + NO3 oxidation rate is significant (Fig. S7e),
it is not clear that α-pinene + NO3 oxidation products
significantly affect the spectrum shown in Fig. b. This may
be due to significantly lower organic nitrate yields from
α-pinene + NO3 than from α-pinene + OH
reactions in the presence of NO .
Transition from RO2+HO2- to RO2+NO-dominant regimes observed in isoprene and α-pinene oxidation products
Figures and show normalized
signals of the representative groups of
isoprene and α-pinene oxidation products as a function of increasing NO : HO2,
which may be influenced by NO + HO2, NO + RO2, and HO2 + RO2 reactions in the reactor.
For each group of compounds, signals obtained at a specific
NO : HO2 were normalized to the maximum observed signal.
NO : HO2 is correlated with the relative branching
ratios of RO2 + HO2 and RO2 + NO reactions that
govern the distribution of oxidation products observed in Figs. and .
As is evident from Figs. and ,
different ion families were characterized by different trends as a function
of NO : HO2. The normalized signals of C5 (isoprene),
C6-10 (α-pinene), and C19-20 (α-pinene)
species decreased monotonically with increasing NO : HO2. In Fig.
, the abundance of C19-20 dimers decreased
significantly faster than the C6-10 species. Because dimers are
products of RO2 + RO2 self-reactions, their yield is
quadratic with respect to [RO2] and therefore was more affected by
competing RO2 + NO reactions than species formed from
RO2 + HO2 reactions.
The normalized signals of C5 (isoprene) and C10
(α-pinene) organic nitrates reached their maximum values at
NO : HO2 ≈ 0.5–2 prior to decreasing. Maximum signals of
C6-9 organic nitrates (α-pinene) were obtained at
NO : HO2 = 2.2, and maximum signals of C5 (isoprene)
and C10 (α-pinene) dinitrates were obtained at
NO : HO2 = 2.1 and 2.2. The formation of dinitrates was favored
when RO2 + NO ≫ RO2 + HO2, as expected,
and regardless of whether RO2 was formed from oxidation of
α-pinene by OH, O3, or NO3. We hypothesize that
NO : HO2 ≫ 1 favored
RO2 + NO → RO + NO2 fragmentation
reactions that led to formation of smaller, more volatile
C5H6-8O5-7 and C5H7NO6-11 α-pinene oxidation
products , whose signals
continuously increased with increasing NO : HO2, along with other
products not detected with NO3--CIMS. This pathway apparently
competed with RO2 + NO → RO2NO
reactions that led to formation of C5 isoprene dinitrates,
C6-C10 α-pinene nitrates, and C10 α-pinene
dinitrates.
Isoprene oxidation products such as C5H9NO7 and
C5H11NO7 contain one peroxide and one nitrate functional group,
and C5H9NO8 contains two peroxides and one nitrate functional
group. The formation of these species, as well as C6-10
α-pinene-derived organic nitrates, was favored at NO : HO2
≈ 0.5–2, where the relative rates of RO2 + NO and
RO2 + HO2 reactions were similar. This correlation
suggests that the C6-10 α-pinene organic nitrates detected
with NO3--CIMS contained a combination of peroxide and nitrate
functional groups, whereas C5 (isoprene) and C10
(α-pinene) dinitrates contained fewer functional groups that were
specifically formed from RO2 + HO2 reactions.