The Indiana University laser-photofragmentation/laser-induced fluorescence (LP/LIF) instrument consists of four primary components: (1) a photolysis laser that fragments HONO into OH and NO; (2) a 308 nm laser for the excitation of the OH radicals; (3) a low-pressure sampling cell; and (4) a sensitive gated detection system that synchronizes photofragmentation of HONO, excitation of OH, and detection of the resulting OH fluorescence. A schematic of the instrumental configuration is shown in Fig. 1.

The absorption spectrum of HONO is shown in Fig. S1 in the Supplement (Burkholder et al., 2019). The strong peak near 355 nm was chosen for photofragmentation because it coincides with the third harmonic of a Nd:YAG laser. A Spectra Physics Navigator II YHP40-355HM laser is used for photofragmentation of HONO, producing approximately 3–4 W of radiation at 355 nm and at a repetition rate of 10 kHz with a pulse width of approximately 20 ns. The OH excitation laser system consists of a Spectra Physics Navigator II YHP40-532 Nd:YAG laser that produces approximately 7–8 W of radiation at 532 nm at a repetition rate of 10 kHz and a pulse width of approximately 20 ns. This laser pumps a dye laser (Sirah Credo, 255 mg L-1 of rhodamine 610 and 80 mg L-1 of rhodamine 101 in ethanol) to produce approximately 40–100 mW of radiation at 308 nm. A small portion of the 308 nm emission is diverted to a low-pressure reference cell for wavelength calibration (Dusanter et al., 2009a). In this cell, a high concentration of OH radicals is produced by the thermal dissociation of water vapor using a hot alumel filament. The resulting OH fluorescence is collected by a Hamamatsu photomultiplier tube (H6180-01) equipped with a band-pass filter centered at 308 nm (Esco Optics). Using the OH fluorescence signal from this cell, the excitation laser is tuned to the Q1(3) transition of OH at 308.1541 nm, a transition that exhibits one of the strongest absorption cross sections around 308 nm (Dusanter et al., 2009a).

The sampling cell is shown in Fig. 2. Ambient air is drawn into the sampling cell through a flat 1 mm diameter pinhole inlet by means of two scroll pumps (Edwards XDS35i) connected in parallel. The cell is maintained at a pressure of 0.25 kPa to reduce quenching of the OH fluorescence by ambient air and thus increase the OH radical fluorescence lifetime. As the sampled air passes through the inlet, it expands into a central aluminum cube and is intersected by the fragmentation and excitation laser emissions. The 355 nm laser emission is propagated to the sampling cell through a 12 m long, 1000 µm fiber-optic patch cord (OZ Optics), which results in approximately 1.5 W of laser power at the entrance of the sampling cell. The excitation laser emission is propagated to the cell through a 12 m long, 200 µm fiber-optic cable (Thorlabs, FG200AEA), which results in approximately 1–4 mW of 308 nm radiation into the sampling cell. After exiting their respective fiber-optic cables, both the 355 and 308 nm laser emissions are spatially combined by a dichroic mirror (Rocky Mountain Instrument Co.) that reflects greater than 90 % of the 308 nm laser beam and transmits greater than 90 % of the 355 nm laser beam. The beams are temporally separated, with the 308 nm pulse entering the detection cell 100 ns after the 355 nm pulse. Upon exiting the detection cell, the beams are spatially separated by a second dichroic mirror, and the power of each beam is monitored using a photodiode (UDT-555UV, OSI Optoelectronics) equipped with interference filters (Esco Optics, Thorlabs).

The fluorescence from the OH radical fragment is collected by an optical train orthogonal to the excitation beam. Two lenses (f=75 mm, CVI Laser Optics) focus the fluorescence onto the detector, and a band-pass filter centered at 308 nm (Barr Associates, transmission 65 %, bandwidth 5 nm, o.d. > 5 at other wavelengths) selectively passes OH fluorescence to the detector and reduces the detection of solar scatter, potential broadband fluorescence of other species, and scatter from the 355 nm laser. A concave mirror (100 mm diameter, 40 cm focal length, Melles Griot) opposite the optical train effectively doubles the solid angle of collection.

The detection system consists of a time-gated micro-channel plate photomultiplier tube (MCP-PMT) (Photek PMT325), a preamplifier/discriminator (F-100T, Advanced Research Instruments), and a high-speed counter (National Instruments, 6024E). A delay generator (Berkeley Nucleonics Model 575) triggers both laser emissions, separated by 100 ns, and also increases the MCP-PMT gain after the 308 nm laser pulse (Fig. 3). Turning on the detector after the 308 nm laser pulse reduces the detection of the intense scattered radiation from the laser pulse while allowing the detection of the OH fluorescence. The gain of the detector is reduced during the laser pulse and switched to the highest gain approximately 70 ns after the laser pulse. The gain is kept high for approximately 550 ns in order to collect most of the OH fluorescence and then is reduced until the next laser pulse. The signal from the MCP is amplified and filtered by a pulse-height discriminator (F100T) that delivers TTL pulses for each detected photon. The photon counter is set with a timing gate to count the fluorescence photons during a 400 ns window while the gain of the detector is high, avoiding potential noise associated with the increase and decrease in detector gain (Fig. 3).

Wavelength modulation is used to tune the 308 nm dye-laser excitation emission on and off resonance with the Q1(3) transition of OH at 308.1451 nm. The net signal from the OH fluorescence is derived by subtracting the off-resonance signal, consisting primarily of solar scatter and some scattered laser radiation that extends into the detection window, from the on-resonance signal. To differentiate OH fluorescence signals due to HONO photofragmentation from fluorescence due to ambient OH radicals, the 355 nm fragmentation laser is cycled on and off with the use of a diaphragm shutter (Thorlabs). When combined with dye-laser wavelength modulation, this creates a complete measurement sequence that allows near-simultaneous measurement of both ambient OH and HONO. Figure 4 shows an example of a typical measurement sequence. Each measurement cycle consists of four 15 s steps: (1) a background signal is established where HONO is photolyzed but the 308 nm laser is tuned off resonance (S1), (2) both ambient OH and the OH fragment from HONO are excited by tuning the 308 nm laser to on resonance (S2), (3) the 355 nm photolysis laser is blocked by a shutter and background signal is re-established by tuning the 308 nm laser off resonance (S3), and (4) the 355 nm laser is still blocked but ambient OH is excited by tuning the 308 nm laser on resonance (S4). The net HONO signal is obtained from the difference between the signals from cycles 2 and 4 (NetHONO=S2-S4), while the net ambient OH signal (NetOH) is obtained from the difference between the signals from cycles 4 and the background signal (Sbkg) which is the average of cycles 3 and 1 (NetOH=S4-Sbkg). Because the band-pass filter rejects scatter from the 355 nm laser, the background signal with and without the fragmentation laser is not significantly different and only varies with fluctuations in the 308 nm laser power.

In addition to the sequence described above, chemical modulation cycles can be used to test for potential interferences in the measurement of ambient OH as performed on FAGE instruments (Mao et al., 2012; Novelli et al., 2014; Rickly and Stevens, 2018). Due to the single-pass laser design of the LP/LIF sampling cell, signals due to laser-generated OH from Reactions (R4) and (R5) are small and are calibrated as a function of laser power, ozone, and water concentrations (Griffith et al., 2016). However, Criegee intermediates formed from the ozonolysis of alkenes (Rickly and Stevens, 2018), the decomposition of ROOOH species (Fittschen et al., 2019), or other unknown interferences could lead to the formation of OH radicals inside the detection cell. Removal of ambient OH through external reaction with a scrubbing agent, such as perfluorpropylene (C3F6) (Griffith et al., 2016; Rickly and Stevens, 2018), allows quantification of all OH formed within the detection cell. The remaining OH signal would be a measurement of the interference that can be subtracted from the total OH signal when the ambient OH is not removed. While the LP/LIF instrument can incorporate chemical modulation cycles to measure potential interferences, the technique was not used in the HONO and OH measurements reported below in order to increase the HONO measurement frequency.

The LP/LIF instrument is automated using a National Instruments multifunction DAQ board (NI USB 6024) and a customized LabVIEW interface program that controls the 355 nm laser shutter and monitors the power of both the 355 nm fragmentation laser and the 308 nm excitation laser. The 308 nm laser is controlled by the Indiana University laser-induced fluorescence – fluorescence assay by gas expansion (IU-FAGE) instrument as described previously (Dusanter et al., 2009a). The output of the 308 nm laser system is split between the LP/LIF instrument and the IU-FAGE instrument, allowing simultaneous measurements of ambient HONO and OH by the LP/LIF instrument with measurements of ambient concentrations of OH and HO2 by the IU-FAGE instrument (Dusanter et al., 2009a).

Several of the previously mentioned measurement techniques utilize a HONO generation source to characterize instrumental response to a known concentration of HONO. Many sources are based on the design of Febo et al. (1995), in which gaseous hydrochloric acid reacts with sodium nitrite to form HONO (Reaction R6). R6HClg+NaNO2s⟶HONO(g)+NaCl(s) This method requires the reaction chamber to be heated to 50 ∘C and can produce mixing ratios of HONO over a wide range (5–20 000 ppb) (Febo et al., 1995) that often require large dilution flows to reach typical outdoor atmospheric concentrations (Lao et al., 2020). Furthermore, the high HONO mixing ratios produced by this approach can disproportionate to form NO and NO2 (Febo et al., 1995; Stutz et al., 2000; Gingerysty and Osthoff, 2020; Lao et al., 2020). As a result, this method typically requires an additional technique to verify both the purity and output concentration of HONO (Pérez et al., 2007; Gingerysty and Osthoff, 2020). While appropriate for a laboratory setting, these limitations, along with the long warmup times needed to ensure stability, can make this method less suitable for calibration in a field setting.

Instead, the sensitivity of the LP/LIF instrument to HONO (RHONO) is determined from (1) the photofragmentation efficiency of HONO by the 355 nm laser (PE) and (2) the sensitivity of the instrument to the measurement of the OH fragment (ROH). 2RHONO=ROH×PE The instrumental sensitivity towards OH (ROH) is determined using the water-vapor photolysis technique, which has previously been described in detail (Dusanter et al., 2008). Briefly, this method relies on the photolysis of water vapor at 184.9 nm to produce a known amount of OH (and HO2). R7H2O+hv184.9nm⟶OH+HR8H+O2+M⟶HO2+M 3OH=HO2=H2O×σH2O×φOH+H×(F×t) As shown in Eq. (3), the concentration of OH and HO2 produced by the calibrator can be calculated from the time-integrated photolysis of water vapor. In this equation, σH2O is the absorption cross section of water at 184.9 nm (6.78 × 10-20 cm2 per molecule), and φOH+H is the unity photo-dissociation quantum yield (Burkholder et al., 2019). F is the photon flux and t is the photolysis exposure time. The quantity (F × t) can be determined via O2 actinometry experiments, as molecular oxygen is also photolyzed at 184.9 nm to form O3(P) and then O3 after reaction with O2. The concentration of ozone produced is also dependent on the product of (F × t). 4O3=O2×σO2×φO3×(F×t)5F×t=[O3]2×O2×σO2 In these equations, φO3 is the quantum yield of ozone from oxygen photolysis (φO3=2) and σO2 is the effective absorption cross section of oxygen at 184.9 nm. Thus, measurements of ozone concentrations can be used to determine the quantity (F × t) if the effective oxygen absorption cross section is known. Previous studies have shown that the effective oxygen absorption cross section at 184.9 nm is dependent on operating conditions (O2 column density, lamp current, and lamp temperature), making it necessary to measure σO2 for each calibration system (Hofzumahaus et al., 1997; Lanzendorf et al., 1997). The dependence of the effective oxygen absorption cross section on the mercury lamp is the result of the overlap between several features of the highly structured Schumann–Runge band and the lamp emission at 184.9 nm that depends on the operating conditions due to line reversal (Lanzendorf et al., 1997) and potential fluorescence of the fused silica envelope (Cantrell et al., 1997).

Using Eqs. (3)–(5), a known concentration of OH can be produced from known concentrations of water vapor and ozone, and the instrumental sensitivity towards OH (Eq. 6) can be derived from the measured fluorescence signal (SOH) and normalized to the power of the 308 nm laser emission (P308) (Dusanter et al., 2008). Typical ROH values vary with ambient water-vapor concentrations due to collisional quenching of excited OH radicals and range from 1.5–4 × 10-8 counts s-1 (cm-3 mW)-1 (Fig. S3), with an estimated uncertainty of 18 % (1σ) (Dusanter et al., 2008). Compared to the IU-FAGE instrument, the OH sensitivity of this detection cell is approximately a factor of 5–20 times lower per milliwatt of laser power due to the multi-pass laser design of the IU-FAGE instrument compared to the single-pass design described here. However, while the single-pass design does not eliminate potential laser-generated interferences, it significantly reduces laser-generated OH from Reactions (R4) and (R5) as there is no beam overlap, and the smaller beam size reduces the potential for double pulsing of the sampled air compared to the multi-pass design at the same laser power. This allows for higher laser powers to be employed in the single-pass instrument, improving the limit of detection with significantly lower laser-generated interferences. 6ROH=SOHOH×P308 A schematic of the calibrator is shown in Fig. S2. The calibrator consists of a rectangular flow reactor made of aluminum (1.27 × 1.27 × 30 cm) equipped with a quartz window on two sides (Dusanter et al., 2008). The light source is a low-pressure mercury lamp (UVP Inc) housed in an aluminum cartridge that is continuously purged with dry nitrogen to prevent light absorption by atmospheric gases as well as helping to stabilize the temperature of the lamp. A 10 SLPM flow of humidified air is used to create a turbulent flow in the reactor. Mixing ratios of water vapor and ozone are monitored in the flow exiting the calibrator using commercial analyzers (Dusanter et al., 2008).

Once a stable concentration of OH and HO2 is produced in the calibrator after the lamp flux and water-vapor concentration have stabilized, the photofragmentation efficiency (PE) of HONO is determined by adding an excess of NO (approximately 800 ppb) to the calibrator to convert the known concentrations of OH and HO2 into HONO through the HO2+ NO → OH + NO2 and OH + NO → HONO reactions. Figure 5 illustrates model simulations of the conversion of OH and HO2 into HONO using the RACM2 mechanism constrained to the concentrations of water vapor and oxygen. After production of OH and HO2 in the illuminated region of the calibrator (first 10 ms), reactions with NO lead to the production of HONO after the approximate 80 ms residence time inside the calibrator. In these simulations, the photolysis of water vapor is adjusted to produce approximately 1 ppb of both OH and HO2 in the calibrator, which in the absence of NO decreases after illumination due to loss from radical–radical reactions and surface loss (Fig. 5a). During typical OH sensitivity calibrations, the loss of radicals in the absence of NO is measured by changing the location of the light source relative to the exit of the calibrator (Dusanter et al., 2008). These measurements indicate that 20 %–30 % of the OH and HO2 radicals produced are lost due to reaction with the calibrator surfaces as well as loss due to the OH + HO2 reaction. Model simulations indicate that a first-order loss rate of 2.6 s-1 is needed to match this observed loss of OH radicals in the calibrator in the absence of NO, and this loss rate has been included in the simulations (Fig. 5a). However, these simulations suggest that during photolysis efficiency calibrations, the excess of NO is sufficient to ensure that reaction with NO is the dominant radical sink, accounting for greater than 95 % of the total loss of OH, with less than 3 % of the OH radicals lost via surface reactions and less than 2 % lost by the OH + HO2 and other radical–radical reactions. Model simulations of this chemistry also suggest that after addition of NO, the OH and HO2 concentrations are negligible and the concentration of HONO is nearly equal to the total OH and HO2 concentrations produced by the calibrator (Fig. 5b).

Figure 6 illustrates a typical measurement of the photofragmentation efficiency. The original signal from the initial amount of OH produced in the calibrator in the absence of added NO is shown in panel (a) (SinitialOH), and the remaining OH concentration after NO is added to the calibrator is shown in panel (b), with the 355 nm photofragmentation laser blocked from entering the detection cell (SremainingOH). This remaining OH signal is likely due to reactant segregation in the turbulent flow of the calibrator preventing all of the OH from reacting with the added NO. When the 355 nm photofragmentation laser is turned on, the increase in the signal relative to the remaining OH reflects the additional OH produced in the detection cell from HONO photolysis (SHONO+remainingOH) (Fig. 6c). Model simulations indicate that reformation of HONO from reaction of the OH fragment with the added NO is negligible due to the reduced concentrations of both OH and NO in the low-pressure detection cell and the short reaction time between the photofragmentation and excitation laser pulses. Additionally, impurities in the NO cylinder appear to be photolyzed by the 355 nm photofragmentation laser, leading to a small production of OH that is observed when the radical source in the calibrator is turned off (Fig. 6d). This impurity must be subtracted from the signal recovered as HONO and is measured by turning the mercury lamp off but keeping the 355 nm laser emission on (Simpurity). This impurity is not observed when the 355 nm laser is blocked from entering the detection cell. The source of this impurity is not clear but could be the result of heterogeneous reactions of NO2 in the calibration system leading to the production of impurity HONO. Experiments employing the use of a scrubber to remove NO2 from the cylinder such as iron (II) sulfate heptahydrate (FeSO4⚫7H2O, Fisher Scientific) did not appear to impact the OH signal due to the impurity, suggesting that production of this impurity may occur inside the NO cylinder. Additional experiments will be needed to identify this impurity.

The HONO photolysis efficiency (PE) of the 355 nm laser can be calculated as the ratio of OH signal recovered as HONO to the net OH and HO2 concentrations produced in the calibrator. This can also be written as the ratio of net HONO signal to the initial OH signal, after corrections to account for the 20 %–30 % OH radical loss due to the OH + HO2 reaction and reaction on the walls of the calibrator based on measurements in the absence of NO as described above (SinitialOH,corr). Because the calibrator produces equal concentrations of OH and HO2, the factor of 2 accounts for the conversion of the produced HO2 to OH when NO is added. 7PE=SHONO+remaining OH-Sremaining OH-Simpurity2Sinitial OH,corr-Sremaining OH Photolysis efficiency measurements are typically performed before and after ambient measurement periods, and variations between measurement periods are likely caused by shifts in the alignment of the 355 nm photolysis laser. Typical PE values for the measurement periods described below were between 0.25 % and 0.34 % for a 355 nm laser power of approximately 1.5 W. Impurities in the added NO that react quickly with OH and compete with reaction of NO, such as NO2, could lead to apparent lower photofragmentation efficiencies by reducing the amount of HONO produced in the calibrator. Model simulations suggest that a 5 % NO2 impurity could reduce the production of HONO by approximately 10 % due to reaction of OH with NO2 instead of NO (Fig. 5b). As a result, the NO added should be of high purity, and chemical scrubbers designed to reduce impurities such as NO2 should be used.

For ambient measurements, the concentration of HONO is determined from the net HONO signal (NetHONO, Fig. 4) and the HONO sensitivity (RHONO). 8HONO=NETHONORHONO=NETHONOROH×PE For typical ROH values ranging from 1.5–4 × 10-8 counts s-1 cm3 mW-1 and PE values between 0.25 %–0.34 %, minimum detectable HONO concentrations (1σ) are typically between 9 and 18 ppt for a 10 min average in the absence of OH (laser power 1.5 W at 355 nm, 1–3 mW at 308 nm). Because the limit of detection depends on the ambient OH signal that is subtracted from the HONO signal, the limit of detection reported here will be higher during the day compared to at night. For the highest sensitivity, 308 nm laser power, and photofragmentation efficiency described above, a daytime maximum concentration of OH of 4 × 106 cm3 would increase the HONO limit of detection by approximately 20 % (10 min average). The overall calibration uncertainty is estimated to be 35 % (1σ), including the uncertainty associated with the OH calibration (18 %, 1σ), and depends on the precision of the photofragmentation efficiency measurement. With the same parameters, the OH limit of detection is typically between 1.1 and 2.2 × 106 cm-3 (S/N=1, 10 min average, laser power 1–3 mW at 308 nm).

The limit of detection for HONO described above is similar to that reported by Dyson et al. (2021) in a laboratory setting using a similar instrument employing a 355 nm laser operating at 10 Hz for photofragmentation of HONO, as well as detection of OH at 308 nm using a dye laser operating at 5 kHz (12 ppt, 50 s average). While the details of the photofragmentation laser in this study were not provided, the lower repetition rate of the 355 nm laser likely leads to a higher pulse energy and a higher photofragmentation efficiency compared to the 10 kHz photofragmentation laser employed in this study. However, the higher pulse energy could lead to photolysis of other ambient species that produce OH and interfere with the measurements of HONO (see below). However, these potential interferences can be minimized in a laboratory setting.

As mentioned above, the large uncertainty associated with the photofragmentation measurements is likely due to shifts in the overlap between the two laser beams as a result of temperature fluctuations impacting the optical alignment. Although this uncertainty is currently large, measurements of HONO were in good agreement with an acetate CIMS instrument during the recent HOMEChem (House Observations of Microbial and Environmental Chemistry) indoor measurement campaign (Wang et al., 2020a). Overall, the measurements of HONO by the LP/LIF instrument agreed with the CIMS measurements to within ±20 %, on average, suggesting that variations of the photofragmentation efficiency over the entire month-long campaign were less than the overall instrumental uncertainty reported here. An example of the measurements during the intercomparison is illustrated below, and a detailed analysis of the intercomparison, including an analysis of the spatial distribution of indoor HONO emissions, will be presented in a future publication.

The LP/LIF instrument was deployed in two locations on the Indiana University campus in Bloomington to measure outdoor concentrations of HONO as a test of the instrument. As HONO was the focus of these measurement periods, chemical modulation cycles were not performed, and the OH measurements presented may be impacted by unknown interferences. During the outdoor measurement periods described below the instrumental sensitivity to OH was ∼ 3 × 10-8 counts s-1 (cm-3 mW)-1, and the measured photolysis efficiency was 0.25 % for a laser power of approximately 1.5 W at 355 nm, resulting in a limit of detection for HONO of approximately 18 ppt (10 min average, 1σ) and a limit of detection for OH of approximately 5 × 105 cm-3 (1 h average, 1σ, 1–3 mW at 308 nm) (Table S1 in the Supplement). The first of two measurement periods occurred during the summer of 2019 at a forested site within the Indiana University Research and Teaching Preserve (IURTP) (39.1908∘ N, 86.502∘ W), located approximately 2.5 km northeast of the IU campus (Lew et al., 2020). The mixed deciduous forest is dominated by emissions of isoprene and monoterpenes and is approximately 1 km away from the nearest high-traffic road. The LP/LIF sampling axis was placed in a small clearing 5 m from the IURTP field lab building, and sampling occurred approximately 0.5 m above a grassy surface to measure potential HONO emissions from the soil. Additional measurements of NO and NO2 were conducted using a commercial chemiluminescence instrument (Thermo Environmental Instruments Inc. Model 42C), and measurements of the photolysis rate constant for NO2 (JNO2) were conducted using a radiometer.

An average of HONO measurements from 4–8 September 2019 is presented in Fig. 7a. At this site, average mixing ratios of NO2 varied from less than 500 ppt at night up to approximately 1 ppb during morning rush hour, while mixing ratios of NO were below the detection limit of the instrument but estimated to be less than 300 ppt based on previous measurements of the NO2/NO ratio at this site (Lew et al., 2020). At night, HONO mixing ratios ranged from approximately 35 ppt to 75 ppt. Maximum observed HONO mixing ratios of approximately 150 ppt occurred during the day, indicating the presence of strong HONO sources that can compete with the loss of HONO due to photolysis during the daytime. Measurements of OH radical concentrations for these days by this instrument using spectral modulation to determine the instrument background reached a maximum value of 6–8  × 106 cm-3, similar to that observed previously by the IU-FAGE instrument at this site after measured interferences were subtracted (Lew et al., 2020). While it is possible that the LP/LIF instrument is also subject to unknown OH interferences, the single-pass laser system minimizes the effect of laser-generated interferences compared to the multi-pass system utilized by the IU-FAGE instrument. In addition, unknown interferences observed using chemical modulation by the IU-FAGE instrument correlated with increases in temperature. Average temperatures during the measurement period were lower than the summer measurement period described in Lew et al. (2020), suggesting that any unknown interferences may not have been detectable.

Nevertheless, because measurements of potential interferences using chemical modulation were not conducted during these days, the measured OH concentrations represent an upper limit to the actual OH concentrations and the [HONO]/[OH] ratio a lower limit. The measured [HONO]/[OH] ratio was relatively constant during the day, resulting in an average value of approximately 1000 at this site. These values are greater than the estimated [HONO]/[OH] ratio assuming steady-state production and loss using Reactions (R2) and (R1) and assuming the loss of HONO by Reaction (R3) is negligible compared to loss by photolysis (Reaction R1). 9HONOOHSS≈kOH+NO[NO]JHONO Because the mixing ratio of NO was below the detection limit of the instrument (less than approximately 500 ppt), the concentration of NO used in this equation was estimated based on previous measurements of the NO/NOx ratio at this site. The value of JHONO was calculated as a function of solar zenith angle (Jenkin et al., 1997) and corrected for cloud coverage according to measured JNO2 values. As illustrated in Fig. 7a, the estimated [HONO]/[OH] ratio using this equation agrees with the measured ratio in the morning and evening but decreased in the afternoon to a value of approximately 10 as photolysis of HONO increased, returning to a value of approximately 1000 in the evening. The difference between the measured and modeled ratio reflects the magnitude of the missing HONO source in this environment, likely due to soil emissions mediated by the microbial community structure at this site (Mushinski et al., 2019).

A second outdoor measurement period occurred on the roof of the Multidisciplinary Science Building II near the center of the Indiana University Bloomington campus. Sampling occurred at a height of approximately 12 m above the ground and 1–5 m away from building surfaces. This measurement site is adjacent to local roads and is influenced by direct emissions from traffic and other anthropogenic sources. Average measurements from 14 and 15 October 2019 are shown in Fig. 7b. At this location, mixing ratios of NO were typically 0.25 ppb at night and increasing to approximately 0.7–1 ppb during the morning and evening rush hours. NO2 varied from 0.5 to 2 ppb during the day and night, higher than that observed at the IURTP site. Overall, HONO mixing ratios were higher than those observed at the IURTP, with maximum mixing ratios of 350–400 ppt at night decreasing to 50–100 ppt during the day. Measured OH concentrations by the instrument were similar to that observed at the forested site, with maximum observed concentrations of 6–8 × 106 cm-3. As at the forest site, chemical modulation experiments to measure unknown interferences were not done during these days; thus these OH measurements represent an upper limit to the actual OH measurements. In contrast to the forest site, the measured [HONO]/[OH] ratio at the urban site varied during the course of the day, decreasing from a value of approximately 104 during the morning to a value of approximately 100 during the afternoon and increasing to a value of 104 in the evening. This trend is similar to that estimated by the steady-state [HONO]/[OH] ratio using Eq. (9). While the ratio estimated by this equation reproduces the measured ratio in the afternoon and evening, it underestimates the measured ratio in the morning, again suggesting that there is an additional source of HONO at this site either from direct emissions or heterogeneous production.

In 2018, the LP/LIF instrument was deployed inside a test house in Austin, Texas, as part of the HOMEChem study (Farmer et al., 2019). HOMEChem was a collaborative field study intended to investigate how household activities influence emissions and chemistry of gases and particles within the indoor environment. During the campaign HONO was measured by the Indiana University LP/LIF instrument as well a high-resolution time-of-flight chemical ionization mass spectrometer (HR-ToF-CIMS) from the University of Toronto (Collins et al., 2018). Several cooking and cleaning perturbation experiments were performed from 5–28 June. During this period the instrumental sensitivity to OH was ∼ 2.75 × 10-8 counts s-1 (cm-3 mW)-1, and the measured photolysis efficiency was 0.34 % (1.5 W at 355 nm), resulting in a limit of detection for HONO of approximately 9 ppt (10 min average, 1σ, 1–3 mW at 308 nm) (Table S1). Results from these experiments are summarized in Wang et al. (2020a).

The LP/LIF sampling axis was placed in the living area of a 111 m2 manufactured home, adjacent to two westward-facing windows. Results from a repeated set of enhanced ventilation experiments, where all of the windows and doors of the house were opened, are presented in Fig. 8. During the ventilation periods the HONO mixing ratio was reduced to approximately 1 ppb due to mixing with outdoor air and returned to a steady state of 3–4 ppb within 20–30 min after ventilation was stopped. The fast return to steady-state concentrations after ventilation ceased indicates that gas-phase HONO is in equilibrium with a reservoir of HONO precursors on interior surfaces (Collins et al., 2018). These values are similar to those measured by the University of Toronto CIMS instrument during this experiment (Wang et al., 2020b), and a more complete instrumental intercomparison of the HOMEChem HONO measurements will be presented in a future publication. Additionally, during this experiment, the air conditioning unit in the house was turned off to minimize variations due to the on–off cycling of the system. As a result, the indoor temperature of the house slowly increased until the unit was turned on at the end of the experiment. The observed increase in HONO as the temperatures increased within the house during the experiment suggests that the equilibrium is temperature dependent (Fig. 8).

Figure 9a illustrates an example of the LP/LIF measurements and the University of Toronto CIMS instrument (Collins et al., 2018) during a cooking event using a gas stove during HOMEChem. During this experiment, mixing ratios of HONO were approximately 2 ppb for several hours prior to the cooking episode. When the gas stove was turned on, mixing ratios of HONO quickly increased to approximately 6 ppb before slowly decaying after the gas stove was turned off. As illustrated in this figure, the LP/LIF measurements of HONO were in excellent agreement with the CIMS measurements during this event, with the measurements agreeing to less than 20 %, and are representative of the overall agreement during the intercomparison, which will be presented in a future publication. These results provide confidence in the accuracy of the LP/LIF instrument and the calibration method.

Potential interferences with outdoor measurements of HONO include species that photolyze at 355 nm, leading to both prompt and secondary production of OH in the detection cell. Possible prompt interferences include HNO3, H2O2, and organic peroxides, while potential secondary interferences include species that could produce OH precursors from photolysis, such as HO2 from the photolysis of HCHO and HO2NO2, which could react to produce OH. A previous analysis of the impact of these species on the atmospheric pressure LP/LIF instrument by Liao et al. (2006a) suggested that the photolysis of typical ambient mixing ratios of these species would not lead to the production of significant OH concentrations at a photofragmentation laser pulse energy of approximately 700 mJ at 355 nm, with 1 ppb of HCHO and 1 ppb HO2NO2 together estimated to produce 0.16 ppt of OH, 1 ppb of H2O2 estimated to produce 1.1 ppt of OH, and 1 ppb of HNO3 estimated to produce 0.03 ppt of OH (Liao et al., 2006a). Given that the LP/LIF instrument described here utilizes a much lower pulse energy (0.2 mJ) and the mixing ratios of potential interfering species are reduced upon sampling at low pressure, it is likely that any OH produced by photolysis of these species would be significantly less than that estimated by Liao et al. (2006a) and would not significantly interfere with outdoor HONO measurements. In addition, the short time interval between the 355 and 308 nm laser pulses likely minimizes these and other secondary interferences.

During HOMEChem, measurements of gas-phase organics were generally higher indoors than outdoors across a broad range of species that were further enhanced during cooking events (Farmer et al., 2019). The cooking events likely increased the concentration of formaldehyde, as combustion from gas stoves can be a significant source of formaldehyde, resulting in indoor mixing ratios potentially greater than 10 ppb (Salthammer et al., 2010; Logue et al., 2014). The agreement of the LP/LIF instrument with the CIMS instrument during these events such as that illustrated in Fig. 9a suggests that interferences from the photolysis of formaldehyde or the potential reduction of the photofragmentation efficiency due to absorption of the 355 nm laser by formaldehyde are likely minimal. Unfortunately, formaldehyde concentrations were not quantified during HOMEChem, and additional measurements together with quantification of formaldehyde concentrations will be needed to confirm these results.

However, higher concentrations of other species that photolyze at 355 nm, leading to the formation of OH, could interfere with measurements of HONO. One possible indoor interference is HOCl, which can be produced during chlorine bleach mopping episodes. Figure 9b shows a bleach mopping experiment during HOMEChem that resulted in the production of approximately 100–200 ppb of HOCl measured by the Colorado State University CIMS instrument (Farmer et al., 2019). During several of these episodes, measurements of HONO by the LP/LIF increased and were correlated with the HOCl measurements while the HONO measurements by the University of Toronto CIMS instrument decreased as expected during these bleach mopping episodes, as the increase in pH likely impacted the surface equilibrium production of HONO (Collins et al., 2018; Wang et al., 2020b). Although the absorption cross section of HOCl is approximately a factor of 40–50 times lower than that of HONO (1.2  × 10-20 cm2) (Burkholder et al., 2019), the indoor mixing ratios of HOCl during several of these mopping episodes were 50–100 times greater than the mixing ratios of HONO during these experiments. Bleach mopping experiments that produced lower mixing ratios of HOCl (less than 12 ppb) resulted in lower interferences in the LP/LIF instrument and better agreement with the CIMS instrument. This interference will be examined in more detail in a future publication. It appears unlikely that outdoor mixing ratios of HOCl would be a significant interference given that HOCl mixing ratios in marine environments are generally less than 1 ppb (Lawler et al., 2011), much lower than the level of indoor HOCl that produced an interference during HOMEChem. While the absorption cross sections of HOBr and HOI are factors of approximately 10 and 30 times greater than that of HOCl at 355 nm (Burkholder et al., 2019), outdoor ambient mixing ratios of HOBr and HOI are much lower than ambient HOCl, with measured values of HOBr less than 26 ppt in the arctic marine boundary layer (Liao et al., 2012) and measured mixing ratios of HOI less than 70 ppt in the marine boundary layer (Tham et al., 2021). Thus, it is unlikely that outdoor mixing ratios of HOBr and HOI would significantly interfere with LP/LIF measurements of HONO. However, these and other potential interferences, both prompt and secondary, will need to be tested in the laboratory.

Because the HONO measurements require the subtraction of the fluorescence signal due to ambient OH, the limit of detection of the instrument will vary with the concentrations of ambient OH as discussed above. The worst limit of detection will likely occur around solar noon when elevated photolysis frequencies lead to large production rates for OH and a short lifetime for HONO. In addition, while the current single-pass laser system with excitation at 308 nm using a high repetition rate laser reduces laser-generated OH from the photolysis of ozone, unknown interferences leading to the formation of OH radicals inside the detection cell, such as Criegee intermediates formed from the ozonolysis of alkenes (Rickly and Stevens, 2018), could interfere with measurements of OH in addition to impacting the HONO limit of detection. These and other unknown interferences can be measured through the external removal of ambient OH through the chemical modulation technique described above, similar to that used by the IU-FAGE instrument (Griffith et al., 2016; Rickly and Stevens, 2018; Lew et al., 2020). In the absence of an interference, this method could also be used to improve the limit of detection of HONO through the removal of the ambient OH background signal.