Isoprene hydroxynitrates (IN) are tracers of the photochemical oxidation of
isoprene in high NO
Hydroxynitrates from OH-initiated isoprene oxidation (high
NO
Isoprene (C
Methods to quantify organic nitrates include infrared spectroscopy (IR),
thermal-dissociation laser-induced fluorescence (TD-LIF) spectroscopy,
chemiluminescence, gas chromatography (GC)-based separation and detection
techniques, and mass spectrometry (MS) (Rollins et al., 2010; Tuazon and
Atkinson, 1990; Sprengnether et al., 2002; Day et al., 2002; O'Brien et al.,
1995; Beaver et al., 2012; Lee et al., 2014a; Lockwood et al., 2010; Paulot
et al., 2009; Giacopelli et al., 2005; Grossenbacher et al., 2004; Patchen
et al., 2007; Hartsell et al., 1994; Kwan et al., 2012; Teng et al., 2015).
IR, TD-LIF, and chemiluminescence can only measure total organic nitrates
because they respond solely to the nitroxy functional group (Day et al.,
2002; Rollins et al., 2010; Tuazon and Atkinson, 1990; Sprengnether et al.,
2002; O'Brien et al., 1995; Hartsell et al., 1994). GC- and MS-based methods
can speciate organic nitrates and have been employed previously to quantify
IN in both laboratory and field studies (Lockwood et al., 2010; Patchen
et al., 2007; Giacopelli et al., 2005; Paulot et al., 2009; Lee et al.,
2014a; Grossenbacher et al., 2004; Beaver et al., 2012; Kwan et al., 2012).
For MS-based techniques, the fragile O–NO
Here we present a comprehensive laboratory and field study of the formation
of IN from the isoprene reaction with OH. In the summer of 2013, we quantified
ambient IN in rural Alabama for 6 weeks during the Southern Oxidant and
Aerosol Studies (SOAS,
A chemical ionization mass spectrometer (CIMS) was used to measure IN
concentrations during the chamber experiments and the SOAS field study. The
instrument is similar to the one described by Liao et al. (2011), which uses
I(H
Two authentic standards, 4,3-IN and 1,4-IN (a mixture of
The IN gas-phase sample for CIMS calibration was prepared by evaporating an
IN
The 1,4-IN calibration was conducted following the same procedures. Since
the 1,4-IN standard contained a mixture of
As we were unable to synthesize the 1,2-IN standard in the condensed phase, a relative method was used, where the CIMS was interfaced with a GC equipped with an electron capture detector (ECD, Fig. 1) to determine the CIMS sensitivity of 1,2-IN relative to 4,3-IN. A mixture of the eight IN isomers was generated by irradiation of a mixture of isoprene, isopropyl nitrite, and NO. The IN mixture was cryo-focused at the head of a 4 m Rtx-1701 column that separated the IN isomers, and the effluent was split into two fused-silica deactivated transfer columns, directed simultaneously to the CIMS and the ECD.
As the CIMS was operated with water addition to the sample gas before
ionization, the GC-ECD/CIMS setup enabled direct observation of the
influence of water vapor to the sensitivity of the two dominant IN isomers.
Figure 2 shows the GC-ECD/CIMS chromatograms with and without water added to
the CIMS. The change in retention time was the result of change in initial
oven temperature setting, which had little influence on the elution
temperature of IN. 1,2-IN and 4,3-IN were the dominant IN isomers and 1,2-IN
eluted before 4,3-IN, according to a recent study using the same stationary
phase (Nguyen et al., 2014b). 1,2-IN and
4,3-IN are expected to have the same ECD sensitivity, because the ECD has
similar response to all mononitrates and the hydroxyl group in
hydroxynitrate has no influence on ECD sensitivity (Hao et al.,
1994). Therefore, the CIMS sensitivity of 1,2-IN relative to 4,3-IN was
calculated as the ratio of the CIMS signal intensity to the corresponding
ECD signal intensity, for the pair of isomers. The calculated relative CIMS
sensitivity was 0.37(
The CIMS sensitivities toward alkyl alcohols and alkyl nitrates are both
around 5 orders of magnitude smaller than its sensitivity toward the
isoprene hydroxynitrates. Hence, it is the combination of the OH group and
the NO
GC-ECD/CIMS setup for the CIMS sensitivity of 1,2-IN relative to 4,3-IN.
GC-ECD/CIMS chromatogram with water
Seven experiments were conducted in the 5500 L Purdue photochemical reaction
chamber (Chen et al., 1998) to determine the yield of IN from
OH-initiated isoprene oxidation in the presence of NO
The IN concentration was measured continuously during each experiment with
the CIMS. Chamber air was sampled through a 5.2 m long inlet, made of 0.8 cm
ID heated (constant 50
Isoprene and its oxidation products, methyl vinyl ketone (MVK) and
methacrolein (MACR), were quantified with a proton-transfer reaction linear
ion trap mass spectrometer (PTR-LIT MS), with measurement precision of 3 ppb
and accuracy of
One wall loss experiment was conducted by keeping the IN isomers produced from isoprene oxidation in the dark chamber and sampling the chamber air with CIMS periodically for 4 h. No significant IN loss was observed, so no wall loss correction was applied for IN measurement.
During SOAS, the CIMS was used to measure ambient IN concentrations
continuously from 26 May to 11 July 2013 at the Centerville
(CTR) site (32.90
Initial conditions for IN yield experiments.
A zero-dimensional (0-D) model based on the Master Chemical Mechanism
(MCMv3.2) (Jenkin et al., 1997; Saunders et al., 2003) was used to
investigate the production and loss of IN in the chamber and in the SOAS
field study. The mechanism was updated for recent experimental and
theoretical studies of isoprene chemistry, including the interconversion of
isomeric isoprene RO
For the IN observations during SOAS, our analysis is focused on the
production and loss of IN. Therefore, the 0-D model for the SOAS data
analysis was constrained to the observed concentrations of the major species
involved in the IN chemistry, including isoprene, HO
Because the 0-D model does not take into account the changes in IN
concentration as IN was transported to and out of the measurement site both
vertically and horizontally, the ratio of total IN concentration to the sum
of methyl vinyl ketone (MVK) and methacrolein (MACR) was used to compare the
model results with observations. Major sources of MVK and MACR include
isoprene ozonolysis (Grosjean et al., 1993) and OH-initiated
isoprene oxidation (Liu et al., 2013). Because IN, MVK, and MACR
are produced simultaneously in the isoprene photochemical oxidation process,
the ratio [IN]
Isoprene data from the PTR-ToF-MS (Misztal et al., 2015) were used to constrain the model and its MVK
The IN yield was calculated from the production of IN relative to the loss
of isoprene, using data obtained in the photochemical reaction chamber
experiments. The isomer-weighted IN sensitivity is expected to change during
each experiment, as IN isomers are lost to OH consumption with different
reaction rate constants. To account for the change in IN isomer distribution
during each experiment, an iterative method was applied to derive a
self-consistent set of total IN yield, IN isomeric distribution, and
isomer-weighted IN sensitivity (Supplement Sect. 3.1). IN loss by OH oxidation was
corrected (Atkinson et al., 1982) with an isomer-weighted
rate constant to account for the difference in OH reactivity for different
isomers (Lee et al., 2014b). The correction factor
was around 25 % by the end of each experiment. Figure 3 shows the results
from the IN yield chamber experiments. The average IN yield was 9 %, based
on the slope of
The relative uncertainty for isoprene concentrations is 17 % based on
instrument calibration. The uncertainty for IN concentrations is caused by
both the uncertainty in the CIMS sensitivity for each IN isomer and the
uncertainty in the relative abundance of the IN isomers. Through a
sensitivity test on the RO
IN and isoprene data for chamber experiments. An average yield of 9 % was obtained from data of the seven experiments.
Figure 4 shows the temporal profile of total IN mixing ratio observed during the SOAS field study with an averaging 10 min time resolution. In general, fast IN production was observed after sunrise. On average, the concentration rose to peak around 70 ppt at 10:00 CDT (Fig. 5) and then decreased to a minimum around 10 ppt by 06:00 the next day, as a result of vertical mixing, boundary layer expansion, dry deposition, and further oxidation. IN concentrations were significantly lower from 4 to 8 July, due to wet deposition and less photochemical reactivity caused by continuous rain events.
In contrast to the IN average diurnal profile (Fig. 5), the diurnal
profiles for isoprene, OH, and NO
IN observed during SOAS.
IN diurnal average from 28 May to 11 July. The blue shade indicates
day-to-day variation (1
Diurnal average of OH (
Isoprene RO
The calculated diurnal average of the
The
As shown in Fig. 7b, the production rates of IN and MVK
The early morning increase in IN concentration could imply significant
contribution from downward mixing of accumulated IN in the residual layer
(RL), as the inversion is broken up after dawn (Hastie
et al., 1993). When the earth surface cools in the evening, the remnants of
the upper daytime boundary layer are isolated from the lower region near the
ground, and the RL forms. The RL contains the same amount of isoprene, IN,
and NO
It is worth mentioning that the nighttime ground-level IN production from
NO
Due to limited availability of overlapping data for model input from
multiple instruments, ambient data for the following 12 days were used: 14
June, 16 June, 22–23 June, 25 June–1 July, and 3 July. For each day, only the
daytime chemistry (05:00–19:00) was simulated, when photochemical
reactivity was high. The observed IN and MVK
Figure 8a shows the temporal profiles of the modeled and observed
[IN]
Simulated and observed [IN]
As shown in Fig. 8c, the modeled results deviated from observations from
10:00 to 12:00 for all the three yields applied. During this period,
the simulated [IN]
The model overestimation in the afternoon can be caused collectively by
measurement uncertainties for model input, uncertainties in the IN loss
rates for OH oxidation and deposition, uncertainties in ambient IN (25 %)
and MVK
We also considered that an underestimated IN photolysis rate could be one of
the reasons for the model–observation discrepancy. The photolysis rate for
IN was set to be identical to the photolysis rate for alkyl nitrates in
MCMv3.2, but IN isomers have double bonds and hydroxyl groups, which could
increase the IN absorption cross section and enhance the photolytic
reactivity for IN. When the IN photolysis rate was increased by 5 times for
the 9 % yield, or 12.5 times for the 12 % yield, the simulated
[IN]
Despite the discrepancy in absolute values, the simulated
[IN]
Although the simulated [IN]
The residual layer IN concentration before mixing (6:00) was estimated
with the 0-D model, using the same initial input as the ground-level
observation on the previous day at 20:00. The chemical processes involved
are IN production from isoprene oxidation by NO
Modeled IN in the residual layer and IN observed near ground before
dawn the next day. The model includes IN production from isoprene oxidation
by NO
Figure 9 shows the simulated IN concentration in the RL and IN observed near
ground before dawn, assuming the RL was completely stable at night with no
depositional loss for IN from the RL. The simulated IN concentration in the
RL before dawn was greater than the concentrations measured at ground level
by up to one order of magnitude, indicating the IN stored in the RL
overnight may be a significant ground level IN source during inversion
breakup. This high IN concentration above the NBL is the result of IN
produced during the previous day, which is present with the high
concentration in the RL as it is formed, and zero deposition removal
overnight. The NO
The calculated residual layer IN does not take into account the altitude-dependent IN concentration caused by OH oxidation, as well as possible IN concentration change caused by advection. Therefore, the actual IN concentration may be very different from the calculated results. This is reflected in a comparison of the large RL IN excess relative to surface IN on 26 and 27 June (Fig. 9), with simultaneous model overprediction of daytime IN on these 2 days (Fig. 8a). Hence, detailed three-dimensional chemical transport models are needed to fully elucidate the production and storage mechanisms of IN in the ambient environment.
OH oxidation was the most important daytime sink for BVOCs during SOAS. As
the
As IN was consumed by OH, it would also undertake both NO and HO
Experimental studies by Jacobs et al. (2014) suggest that OH
addition to IN can invoke IEPOX formation with a yield of 13 % at
atmospheric pressure, which simultaneously releases NO
For the RO
Diurnal averages of IN and ISOPOOH
Possible oxidation mechanism for 1,2-IN with
The RO
A range can be estimated for the NO
Our chamber experiments indicate a 9(
Our field measurements and model simulations suggest that in the southeast
US, the formation of organic nitrates in the boundary layer is controlled by the
availability of NO
While IN were produced and destroyed in the morning through high NO
During the past 15 years, NO
We thank the organizers of the SOAS study, especially Ann Marie Carlton. We appreciate help from Jozef Peeters at the University of Leuven in elucidating the uncertainties associated with the current LIM1 mechanism. We acknowledge funding from the National Science Foundation (NSF) grant 1228496 and US Environmental Protection Agency (EPA) STAR grant 83540901. Edited by: A. E. Perring