Biogenically produced isoprenoids (hydrocarbons comprised of C5H8 units) have global emission rates into the atmosphere surpassing those of anthropogenic hydrocarbons and methane (Guenther et al., 1995, 2012). The biogenic carbon emission flux is dominated by isoprene (C5H8) and monoterpenes (C10H16), which account for approximately 50 and 30 % of the OH reactivity over land, respectively (Fuentes et al., 2000). Furthermore, it has been suggested that the atmospheric oxidation of isoprene, in particular, can buffer the oxidative capacity of forested regions by maintaining levels of the hydroxyl radical (OH) under lower nitric oxide (NO) conditions (Lelieveld et al., 2008). Due to their large abundances, isoprene and monoterpenes also dominate the global budget of secondary organic aerosol (SOA) (Henze et al., 2008). Thus, the accurate representation of detailed chemistry for isoprene and monoterpene is necessary for meaningful simulations of atmospheric HOx (OH + HO2), NOx (NO + NO2), surface ozone (O3), trace gas lifetimes, and SOA.

Unsaturated hydrocarbons like isoprene and monoterpenes are primarily oxidized by OH, O3, and the nitrate (NO3) radical in the atmosphere. OH oxidation is the dominant fate for isoprene, but O3 and NO3 oxidation can dominate reactivity for monoterpenes and sesquiterpenes. Our understanding of the OH-initiated isoprene oxidation mechanism has significantly improved during the last decade, following the first suggestion of the capacity of isoprene to produce SOA (Claeys et al., 2004). The mechanistic developments have been propelled by technological advancements in instrumentation (Hansel et al., 1995; Crounse et al., 2006; Jordan et al., 2009; Junninen et al., 2010), enabling the detection of more complex oxidation products derived from isoprene and other biogenic hydrocarbons. However, the scientific understanding of these biogenic oxidation mechanisms is far from complete. It is outside the scope of this overview to describe comprehensively the isoprene and monoterpene oxidation mechanisms. Rather, we provide a brief background of the oxidation of biogenic hydrocarbons, which includes “state-of-the-science” knowledge, to motivate the study. The mechanisms described here are illustrated in Scheme 1.

The 2014 Focused Isoprene eXperiment at the California Institute of Technology (FIXCIT) is a collaborative atmospheric chamber campaign focused on advancing the understanding of biogenic hydrocarbon oxidation in the atmosphere. The campaign was motivated by the communal need for a tight coupling of field and laboratory efforts toward understanding the mechanistic details responsible for ambient observations, exploring explicit chemistry as driven by the fate of RO2 radicals through well-controlled experiments, and fully characterizing instrumental response to important trace gases using authentic standards to guide data interpretation. To accomplish these goals, a suite of instruments typically deployed for field missions was used to perform parallel measurements at a forested site in Alabama and then in the atmospheric chambers at Caltech. This overview provides an account of the goals and conditions for the experiments performed during the campaign. A key component of FIXCIT is the re-design of “typical chamber experiments” to recreate the ambient atmosphere with higher fidelity so that results from laboratory studies can be implemented in models and used to interpret ambient observations with higher confidence.

Experiments were performed in the Caltech Atmospheric Chamber Facility within a 1 month period in January 2014. The facility contains several in-house gas- and aerosol-phase instruments and an 8 × 5 m insulated enclosure, housing two side-by-side Teflon atmospheric chambers that are suspended from the ceiling. The chambers were manufactured from fluorinated ethylene propylene (FEP) Teflon. The chamber volume was measured regularly by quantitative transfer of highly volatile organics such as isoprene by an externally calibrated GC-FID. Quantitative transfer was checked via injections of a measured quantity of isoprene (checked by gravimetric, volumetric, and FT-IR methods) into a pillow bag with known volume by timing a calibrated mass flow of air into the pillow bag. For most experiments, the chamber volume was between 23 and 24 m3. The spatial configuration of instruments in the chamber facility during FIXCIT is shown in Fig. 1. The instruments, contributors, and identifying abbreviations used in this work are described in Table 1. A total of 320 UV black lamps (broadband λmax ∼350 nm) are mounted on the walls of the enclosure. The lamps are located behind Teflon films so that the heat produced from the operation of the lamps can be removed by recirculating cool air. The interior of the enclosure is covered with reflective aluminum sheets. Light intensities can be tuned to 100, 50, 10, and 1 %. JNO2 was measured to be 7 × 10-3 s-1 at 100 % light intensity. Light fluxes at several locations within the chamber (e.g., center, corner, right, left, high, low) did not vary more than 15 %. Temperature controls in the chamber enclosure are tunable from 10 to 50 ∘C (typically set at 25 ∘C) and did not fluctuate more than 1 ∘C, except during periods when the temperature was explicitly changed or during a 30 min period immediately following a change in the light intensities (up to 2 ∘C increase was observed from switching on 100 % lights.)

The chamber experiments were operated in batch mode throughout the campaign. Temperature and RH were monitored continuously inside the chamber by a Vaisala HMM211 probe calibrated with saturated salt solutions in the RH range of 11–95 %. In the range RH < 11 %, water vapor measurements were provided by the TripCIMS. The chambers were flushed at least 24 h before each use with ultra-purified air (purified in-house via a series of molecular sieves, activated carbon, Purafil^™ media, and particulate filters), at elevated temperature when needed (∼40 ∘C), so that the backgrounds on gas- and particle-phase instrumentation are at baseline levels. As a reference, NO levels before each run were typically less than 100 pptv (from NO–CL measurements) and particle concentrations were less than 0.01 µg m-3. Flushing rates, as balanced by exhaust rates, were typically 250 SD L min-1 (SLM) or ∼0.6 chamber volumes per hour. Chambers were mixed on the timescale of minutes by injecting high-pressure pulses of air during the beginning of experiments.

Chamber 1 was reserved for low-NO experiments, so that the walls did not contact elevated levels of nitric acid and organic nitrates during the lifetime of the chamber, while Chamber 2 was reserved for moderate- to high-NO experiments. Experiments were carried out daily in alternating chambers to allow for the full flushing period of the previously used chamber. Each chamber was characterized separately prior to the campaign for vapor and particle wall loss rates. Typically, wall loss rates for gas-phase species are slightly higher in the high-NO chamber than the low-NO chamber due to the greater acidity of the walls. Particle wall loss rates were not significantly different between chambers. Measurements of the particle wall loss rates were performed by injecting ammonium sulfate (AS) seed aerosols into the chamber and monitoring the decay over the course of 10–24 h. Particles were injected via atomization of dilute salt solutions (e.g., AS 0.06 M) through a 210Po neutralizer and water trap. Measurements of vapor wall loss rates were performed by injecting OVOC standards (e.g., IEPOX, HMHP, etc.) into the chamber. Both particle and vapor wall loss characterizations were performed at several RH conditions (4–85 % RH). These characterizations have been described in more detail previously (Loza et al., 2010; Nguyen et al., 2014).

Organic compounds were injected into the chamber by two methods. (1) For volatile compounds, a measured volume was injected with a micro-syringe through a septum into a clean glass bulb, and the evaporated standard was quantitatively transferred into the chamber by dry purified air. Gas introduction of VOCs (done for isoprene and methacrolein) by filling an evacuated bulb with the chemical vapor, backfilling with nitrogen gas, and characterizing with Fourier transform infrared spectrometry before injecting did not produce significantly different results than volume injection. (2) For semi-volatile compounds, the solid or liquid standard was placed inside a two-neck flask, which was heated by a water bath (35–65 ∘C), and the headspace was carried into the chamber by dry purified air. The ToFCIMS or TripCIMS instruments measured the gas-phase mixing ratio of the semi-volatiles in real time as the compounds entered the chamber, and injection was halted when a satisfactory quantity was introduced. OVOCs were calibrated by the ToFCIMS and TripCIMS by methods described earlier (Paulot et al., 2009a). The desired RH inside the chamber was achieved by flowing dry purified air through a water-permeable (Nafion) membrane humidifier (FC200, Permapure LLC), kept moist by recirculating 27 ∘C ultra-purified (18 MΩ, 3 ppb TOC) water (Milli-Q, Millipore Corp). Particles were atomized into the chamber as described for particle wall loss experiments. When hydrated particles were needed for experiments, particles were injected via an in-line, heated, wet-wall denuder into a chamber that has RH above the efflorescence point of the particular salt (Martin, 2000).

Instruments were connected via sampling lines to both chambers through port holes in the enclosure as shown in Fig. 1. Sampling lines were capped when not in use. Inlet and tubing material were instrument specific, and included stainless steel (GTHOS and ToF-AMS), heated stainless steel and quartz (TDLIF), electro-polished steel and FEP Teflon (NO3CIMS), polyetheretherketone (PEEK) and Teflon (SRI-ToFMS), and perfluoroalkoxy polymer (PFA) Teflon (other instruments).

The duration of each experiment (i.e., the level of oxidation that can be probed) was critically dependent on the net sampling flow rates at which air was withdrawn from the chamber. Sampling strategies were developed to minimize the effective sampling flow rate from each instrument, in such a way that instrument responses were not significantly different than during field campaigns. In many cases, a common high-flow Teflon sampling line was used to minimize the residence time of gases through tubing, and smaller flows were sampled orthogonally by each instrument. In some cases, a duty cycle was used as needed.

Several modifications from field designs were utilized for chamber sampling. The modifications were that (1) the GTHOS detection system was located between the chambers inside of the enclosure to minimize the residence time of HOx inside the instrument (Fig. 1). The detection system was connected to the laser on the outside of the enclosure via a 3 m fiber optic cable fed through the side port hole. The sampling flow rate was similar to field flows (6 SLM); however, the fast-flow inlet was situated horizontally (∼2 m in height) instead of vertically. The inlet was adapted to each bag directly, by attaching it to a Teflon plate that was in turn secured to the chamber walls via a large o-ring. The GTHOS inlet switched from Chamber 1 to Chamber 2 as needed. Chemical zeroing was performed by releasing hexafluoropropene (C3F6) into the inlet as an OH scrubber, and dark zeroing by measuring the difference between online and offline signals. Chemical and dark zeroing methods were used to distinguish between OH present in the chamber or atmosphere (chemical OH) and OH that may have been produced after the gas stream enters the instrument, which is additional to the chemical OH signal; (2) LIF-OHR was diluted a factor of 10 with nitrogen gas (effective flow 6 SLM); (3) NO3CIMS was diluted a factor of 5 with scrubbed zero air (effective flow 2 SLM); (4) GlyLIP and FormLIF both operated at 5 SLM instead of the usual 17 and 10 SLM, respectively; and (5) SRI-ToFMS (1.5 SLM) and GCFID (0.1 SLM) occasionally sampled through a 0.125–0.25′′ OD PFA Teflon tube that was submerged in a cold bath kept at -40 ∘C in order to remove interferences from certain OVOC (see Sect. 2.3).

GC-ToFCIMS, first described in Bates et al. (2014), is an extension of the ToFCIMS. Analyte gas samples were focused with a cold trap onto the head of a RTX 1701 column (Restek) and eluted with a temperature ramping program (30–130 ∘C) in the oven before reaching the ToFCIMS for mass spectrometry detection. GC-ToFCIMS recorded data only when isomer separation was needed, because its operation took the standard scanning mode of the ToFCIMS offline. All other instruments operated normally with the following sampling flows: TDLIF (4 SLM), ToFCIMS and TripCIMS (2 SLM), CRM-OHR (0.5 SLM), NO-CL (1 SLM), and IACIMS (2 SLM). Frequencies of zeroing (with dry N2 or ultrazero air) and calibration (various methods) were instrument specific, with some instruments zeroing once per hour and calibrating once every few hours and others performing zeroing/calibration between experiments.

The experiments performed at FIXCIT can be divided into several categories, each probing one or more specific science questions outlined in Sect. 1.2. Every experiment included successful elements from past studies, but with a special focus on extending to atmospheric conditions. One example is reducing the occurrence of RO2 + RO2 side reactions in chamber experiments, which can lead to yields of atmospherically relevant products that are biased low. Enabled by the high sensitivity of field instruments, photooxidation was performed with precursor mixing ratios as low as 12 ppbv. Certain instruments that required extensive dilution in a chamber setting, e.g., LIF-OHR, had poorer-quality data for low-loading experiments. Experimental durations were typically 4–6 h, with the exception of overnight runs where the majority of instruments sampled briefly to establish starting conditions, then were taken offline during the nighttime and resumed sampling in the morning. The typical reaction time for an overnight experiment was ∼15 h. Experimental details are reported in Table 2. OH concentrations were derived from hydrocarbon decay data from GCFID, SRI-ToFMS, or ToFCIMS, when available, using published rate coefficients (Atkinson et al., 2006; L. Lee et al., 2014; Bates et al., 2014). Otherwise, preliminary GTHOS chemical-zeroing data were used. The following types of experiments were included in the study:

a.

Blank (Exp. 4b and 5b): blank experiments were designed to investigate background signals present in experiments that may have sources other than gas-phase chemistry of the injected hydrocarbon, e.g., from heterogeneous oxidation of residual organics on the chamber walls. OH precursors, such as hydrogen peroxide, were added to each chamber, the UV lamps were turned on, and sampling occurred as usual. Furthermore, the temperatures inside the chambers were ramped from 25 to 35 ∘C to explore the extent to which elevated temperatures change the chamber background signals due to increased volatilization of organics. Blank experiments were performed under dry conditions. Common background compounds produced from heterogeneous wall reactions are formic acid and acetic acid.

Throughout the campaign, several sources of analytical interferences or systematic biases were discovered. Some challenges resulted from the integration of field instruments to a chamber setting, where high concentrations of certain chemicals were used to engineer extremely specific conditions. Thus, these issues do not affect ambient sampling. For example, (1) high NO2 levels in some experiments affected the normal operation of TDLIF because the ΣANs and ΣPNs measurements were determined by subtraction of NO2. When NO2 is much higher than ΣANs and ΣPNs, the measurement by difference contains large uncertainties; (2) high H2O2 for low-NO conditions affected the operation of some CIMS instruments because the ppmv mixing ratios of H2O2 depleted a non-negligible quantity of reagent ions. In order to correct for this, the CIMS instruments needed to calibrate as a function of H2O2 in addition to traditional methods, or account for the true reagent ion signal (which was anti-correlated with H2O2 concentration). High H2O2 also affected GTHOS due to photolysis-derived OH production by the laser. GTHOS corrected for this effect by removing the OH background that was determined by sampling when only H2O2 was present; (3) High formaldehyde, cyclohexane, or H2O2 dominated the OH reactivity for certain experiments. In experiments where ppmv levels of volatile compounds were used, LIF-OHR and CRM-OHR did not operate. In contrast, high ozone and NO levels did not appear to affect the operation of any instruments. Temperature and humidity effects on ion sensitivities have been corrected for by ToFCIMS and TripCIMS as standard procedure. Other CIMS are actively characterizing these effects for analytes of interest.

However, other analytical challenges were not unique to laboratory studies. It was found that chemical artifacts were produced from the decomposition of multifunctional OVOC (e.g., ISOPN, ISOPOOH, IEPOX, and pinonaldehyde) under normal operating conditions in some instruments; thus, possibly affecting ambient sampling and field data interpretation. Scheme 2 shows the proposed decomposition pathways of certain isomers of isoprene-derived OVOC to form MAC and MVK. We are aware of MAC and MVK interference only from the 1,2- and 4,3- isomers of ISOPOOH, the 1,2- and 4,3- isomers and ISOPN, and the beta isomers of IEPOX (i.e., the peroxide, nitrate, and epoxide groups are secondary or tertiary). Unfortunately, these isomers are expected to be the most abundant in the atmosphere, e.g., the beta IEPOXs are estimated to represent more than 97 % of atmospheric IEPOX (Bates et al., 2014). The extent of decomposition and product distribution may also vary based on the operating conditions of the particular analytical method. In general, the decomposition was exacerbated by instruments with harsher sampling conditions, i.e., high ionization energy (e.g., the standard H3O+ mode of SRI-ToFMS), high temperatures, and/or materials incompatible with organics (e.g., the hot stainless steel sample loop and inlet of GCFID). OVOCs from the low-NO isoprene photooxidation have been shown to decompose to MAC and MVK in commercial PTRMS instruments (Liu et al., 2013), but the exact identities of the compounds were unclear. During FIXCIT, it was observed that ISOPOOH, IEPOX, and pinonaldehyde were detected at m/z 71.050 in the SRI-ToFMS in PTR mode (the sum of MAC + MVK). Switchable reagent ions show promise for removing certain biases, but more work is needed to characterize the chemistry that forms interfering ions. Furthermore, we observed that the decomposition interference also affected GCFID, the other commonly used detection method for MAC and MVK in ambient samples. ISOPOOH, IEPOX, and ISOPN were detected as either MAC or MVK in the GCFID, depending on the specific isomer. The interferences may not be localized to this particular GCFID, and a more detailed account is forthcoming (Rivera et al., 2014). Conversion efficiencies of OVOCs to the C4 carbonyls in the Caltech GCFID range in order of ISOPOOH > IEPOX > ISOPN, and can be almost quantitative for ISOPOOH because of the facile cleavage of the weak O–O bond. Lastly, ISOPN were found to be converted to NO with a small yield in the NO–CL and a larger yield in commercial NOx analyzers.

All decomposition-derived artifacts can be avoided by collecting the air sample through a length of tubing submerged in a cold bath (-40 ∘C), which trapped OVOCs that are less volatile than authentic MAC and MVK. Liu et al. (2013) implemented this technique successfully in their laboratory study using SRI-ToFMS, resulting in a lower yield than previously reported for MAC and MVK in the low-NO oxidation of isoprene. Field application may prove more challenging, however, as the trapping is labor intensive and requires careful humidity control to avoid ice buildup and blockage. During FIXCIT, both GCFID and SRI-ToFMS employed trapping techniques at various times to avoid biases in the detection and interpretation of MAC and MVK data.