Article  

Nitrogen-oxide-based functional groups are an area of significant interest in atmospheric oxidative chemistry. Organic nitrates are formed through reactions between volatile organic compounds (VOCs), of which the global majority are biogenic in origin (Seinfeld and Pankow, 2003; Perring et al., 2013) and NOx (=NO+NO2) or NO3 (Ng et al., 2017), which is predominantly anthropogenic in origin (Seinfeld and Pandis, 2006). The two major organic-nitrate products of these reactions are alkyl nitrates (ANs) of the form RONO2 and peroxy nitrates (PNs) of the form ROONO2. These organic nitrates play an important role in regulating ozone in the troposphere by serving as temporary reservoirs of NO2 (Buhr et al., 1990; Thornton et al., 2002). Equilibrium partitioning of high molecular-weight, low-volatility organic molecules occurs, causing some organics to condense onto existing particles (Jimenez et al., 2009). These secondary organic aerosols (SOAs) consist primarily of the highly oxidized products of VOC + oxidant reactions because of their increased molecular weight and higher polarity. Lower nighttime temperatures decrease volatility even further, leading to increased partitioning into the particle phase (Fry et al., 2013). Warmer temperatures, deposition, and chemistry within the particles change the equilibrium, resulting in the release of NO2. Because of long residence times of SOA, significant quantities of NO2 can be transported away from source regions by wind in reservoir form (Perring et al., 2013; Browne et al., 2013; Wolfe et al., 2007; Kim et al., 2014; Zare et al., 2018).

Different classes of organic nitrates dissociate in distinct temperature ranges, based upon the inherent stability of the molecules. At residence times of 30-90 ms in quartz tubes, peroxy nitrates (PNs, RO2NO2) dissociate at approximately 150 ∘C, alkyl nitrates (ANs, RONO2) at 350 ∘C, and nitric acid (HNO3) at 600 ∘C (Day et al., 2002). The dissociation temperatures are dependent on residence times, but there seems to be very little dependence on what constitutes the R group (Hao et al., 1994; Kirchner et al., 1999). This is useful for the detection of total peroxy and alkyl nitrates (ΣPNs and ΣANs, respectively) because they can be dissociated as a class, with identical detection efficiency regardless of the chemical nature of the R group. Reaction (R1) shows that the thermal dissociation of each class of organic nitrates results in one NO2 and a hydrocarbon-containing X group.

where X=RO2, RC(O)OO, RO, or OH.

PNs serve as a temporary reservoir of NO2 in the atmosphere because the equilibrium between formation and dissociation is rapid. For example,

has a Keq of 2.2×10-12 cm3 molecules−1, resulting in a PN lifetime at 20 ppb NO2 of 0.56 s at 298 K and 1 bar (Atkinson et al., 2006; NASA/JPL Data Evaluation, 2015). In contrast, ANs and HNO3 predominantly serve as sinks of NO2, with spatial transport scales that depend on their meteorology-dependent deposition lifetimes (Horowitz et al., 2007).

Previous studies of organic nitrates have been done by measuring specific nitrates (Wolfe et al., 2007; Horowitz et al., 2007; Parrish and Fehsenfeld, 2000; Surratt et al., 2006; Lee et al., 2016) or by looking at the sum of nitrates using thermal-dissociation NO2 measurements (Zellweger et al., 1999; Day et al., 2002; Hargrove and Zhang, 2008; Paul et al., 2009; Rollins et al., 2010; Sobanski et al., 2016). The instrument described in this paper has drawn on aspects of three different thermal-dissociation nitrate measurement strategies in the literature. The general oven and flow plan was based on the thermal-dissociation-laser-induced fluorescence (TD-LIF) instrument built by the Cohen group at UC Berkeley (Day et al., 2002). Instead of LIF, the NO2 detection device in the instrument described here is a commercial cavity ring-down spectrometer (CRDS). Once interferences are characterized and absorption cross sections are known, the CRDS does not require in-line calibration by an authentic standard gas cylinder during sample measurement, as discussed in Paul et al. (2009). Gas-particle partitioning measurements using a switchable charcoal denuder were incorporated from Rollins et al. (2010).

The benefit of using a CRDS over chemiluminescence (CL) detection of NO2 is its selectivity. The (partial) thermal dissociation of multiple unstable nitrate compounds like ANs, PNs, and N2O5 into NO2 by the CL heating process and molybdenum catalyst has been well documented (Wooldridge et al., 2010). The CRDS can make direct measurements of NO2, unlike CL, which uses a metal catalyst to turn NO2 into NO and back-calculates NO2 concentration by subtraction. The CRDS does not require heating or a catalyst and is therefore more selective. LIF can be tuned to a specific spectroscopic transition like the CRDS and can be run at lower cell pressures that reduce recombination (see Sect. 3.7 below), but laser power becomes limiting for the measurement of low concentrations and requires delicately aligned multipass optical cells to achieve low limits of detection for NO2. The downsides of the CRDS come from the expense and delicateness of the instrument.

Since high molecular-weight oxidation products can condense into the particle phase, it is valuable to be able to make both gas and particle phase measurements. Denuders work by using diffusion to separate gases from liquid- or solid-phase particles. Higher diffusion rates for gases means that they are more readily absorbed into the walls of a charcoal denuder, leaving behind the particle phase. The fraction of gas removed depends on residence time in the denuder and the surface area available to diffusing gas molecules. The diffusion coefficient of NO2 is reported to be 0.154 cm2 s−1 (Williams et al., 2012) and 0.070 cm2 s−1 for n-propyl nitrate (Paul et al., 2009). According to previous studies using charcoal denuders, the denuder removed the majority of particles with diameters <0.1 µm (Glasius et al., 1999) as well as all semivolatile organic gases.