Insights into HONO sources from observations during a solar eclipse

Nitrous acid (HONO) is a major, and often the dominant, precursor to primary OH radical production in the daytime boundary layer, driving the removal of many primary pollutants and formation of secondary species such as ozone and many aerosol components. A number of photochemical HONO production mechanisms have been proposed, alongside homogeneous gas-phase reactions, to account for field observations of daytime HONO. The range of production mechanisms show varying dependencies upon precursor species such as NO 2 , available surfaces for heterogeneous reactions, and dark / photoenhanced aspects. Here, we exploit measurements of HONO and related species during a near-total solar eclipse as a natural perturbation to the atmospheric photochemistry to assess the characteristics of the production mechanisms occurring at an urban background location. Little variation in HONO abundance was observed in response to changing light levels during the eclipse, pointing to relatively balanced photochemical source and (well-understood) sink terms. We employ a series of simple kinetic simulations to explore the consistency of different potential source mechanisms with the observations, finding evidence for a dominant role for photochemical processing of traffic-derived NO 2 upon surfaces producing HONO, alongside indications of a smaller contribution from direct vehicular emissions. Other mechanisms involving dark heterogeneous reactions were not, in isolation, consistent with the observations. The critical role of NO 2 , ultimately derived overwhelmingly from local road traffic emissions at this location, points to significant future reductions in daytime HONO production with vehicle fleet evolution and reduction of tailpipe emissions. Nitrous acid (HONO) is a major, often the dominant, boundary layer precursor to the key daytime atmospheric oxidant OH; however sources of HONO are poorly understood, with a number of candidate formation mechanisms advanced. Here, we use a natural perturbation – a solar eclipse – to identify the photochemical factors governing HONO formation, from their variation with this large-scale change in solar intensity. We demonstrate that HONO is not controlled by a single source, but that photoenhanced production (i.e. reactions accelerated by sunlight) are a major component of HONO production, suggesting a dominant role for photochemical processing of traffic-derived NO 2 upon surfaces producing HONO, alongside a smaller contribution from direct vehicular emissions. In both cases, association with (anthropogenically derived) NO 2 indicates that HONO formation may fall with future vehicle fleet evolution and tailpipe emissions reduction. include emissions 9, 10 , biomass burning 11-13 , microbial activities in 14, 15 associated with 16 . Laboratory studies have shown that photoenhanced HONO production can occur following NO 2 uptake to surfaces including soot 17 , aromatic species 18 , humic acids 19, 20 and TiO 221 . Photoenhanced NO 2 heterogeneous reactions on ‘urban grime’ on building surfaces 22 have also been shown to generate appreciable levels of HONO. Further potential sources are photolysis of nitric acid 23 , nitrophenols 24 and of particulate nitrate 25-27 .

Nitrous acid (HONO) is a major, often the dominant, boundary layer precursor to the key daytime atmospheric oxidant OH; however sources of HONO are poorly understood, with a number of candidate formation mechanisms advanced. Here, we use a natural perturbationa solar eclipse -to identify the photochemical factors governing HONO formation, from their variation with this large-scale change in solar intensity. We demonstrate that HONO is not controlled by a single source, but that photoenhanced production (i.e. reactions accelerated by sunlight) are a major component of HONO production, suggesting a dominant role for photochemical processing of traffic-derived NO 2 upon surfaces producing HONO, alongside a smaller contribution from direct vehicular emissions. In both cases, association with (anthropogenically derived) NO 2 indicates that HONO formation may fall with future vehicle fleet evolution and tailpipe emissions reduction.

Introduction
Atmospheric chemical processing in the sunlit troposphere is driven primarily by the OH radical, which initiates the removal of most organic compounds, and drives the formation of ozone, and secondary organic and inorganic aerosols. 1,2 While the dominant primary source of OH in the free troposphere is the photolysis of ozone and subsequent reaction of electronically excited oxygen atoms with water vapour, a series of recent field campaigns have shown that the photolysis of nitrous acid (HONO) is a major / the dominant primary OH precursor in the continental boundary layer. 3,4 HONO +hν (λ < 440 nm) → NO + OH (R1) HONO is formed from the slow homogeneous reaction between OH and NO, and in this sense acts as a photolabile reservoir for OH (and NO) with R1 and R2 forming a null cycle; however observed levels of HONO during daytime are typically one order of magnitude higher than R1 and R2 in isolation predict, indicating the presence of additional HONO formation mechanisms or emissions (and hence net OH production). [5][6][7] The dark formation of HONO through the heterogeneous hydrolysis of nitrogen dioxide (R3 below) has been known for a number of years 8  Many field and laboratory studies have been carried out to investigate the sources of HONO in different environments, e.g. urban, rural, coastal, forest and vehicle tunnels. 5,9,[28][29][30][31][32][33][34] . In a recent study, Tong, et al. 35 performed measurements at urban and suburban environments in Beijing during winter, and observed that direct emission and homogenous gas phase sources made a larger contribution in urban areas, while heterogeneous sources were suggested to be more significant away from urban centres. A recurrent challenge remains quantitatively reconciling daytime HONO concentrations in many urban environments with the current known sources. [5][6][7] Here, we take advantage of a near-total solar eclipse (~90% attenuation) as a natural shortterm perturbation to atmospheric photochemistry to explore the chemical processes affecting HONO abundance. The reduction in photolysis frequencies during an eclipse is effectively uniform over a large area (relative to the chemical lifetime and hence spatial footprint of HONO, NOx and related species), of relevance to the challenges of heterogeneity. 36,37 Comparison of observed temporal behaviour during the eclipse therefore provides a unique test of our understanding of HONO photochemistry. We report the temporal variation of HONO, NO x , O 3 and aerosol characteristics during the eclipse, and use these measurements to explore the nature of sources of HONO which are consistent with their behaviour.
stripping coil, into an acidic solution and is derivatived into an azo dye. The light absorption at 550 nm of the azo dye is then measured with a spectrometer using an optical path length of 2.4 m. The LOPAP was operated and calibrated according to the standard procedures 39 , with a sampling height of 3m above ground level, and data acquired at 5 minute time resolution.
Baseline (zero) measurements, obtained by sampling zero air, were taken at frequent intervals (8 hours). The detection limit (2σ) under the instrument operating conditions of these measurements was determined to be 6 pptV. In addition to the LOPAP, co-located measurements of NO, NO 2 and NO x (Thermo Scientific 42c -Mo convertor for NO 2 measurement, hence potentially responding to NO y interferences), and ozone (Thermo Scientific 49i) were performed. An optical particle spectrophotometer (TSI 3330) measured the particle number size distribution, and converted from particle number concentration (PNC) to total surface area (TSA) via the TSI AIM software (Version 9), which assumes particle sphericity. NO x , O 3 and particle data were obtained at a 1 min time resolution.
Meteorological data (relative humidity, solar intensity, air temperature, and wind speed and To explore the variation in HONO (and other species) abundance anticipated across the eclipse timeframe (ca. 2 hours) a series of simple simulations were performed representing different scenarios (mechanisms) for potential atmospheric processes / emissions that may form and remove HONO. We do not attempt to reproduce the detailed gas-or heterogeneous chemistry using a formal model or chemical mechanism -as many of the species that would be required were not measured (notably, no VOC data were available, precluding modelling HO x sources or sinks). However, with some assumptions (below), over the limited duration of the eclipse event, the time evolution of HONO can be calculated from its initial abundance, the time variation of various possible source term(s), and its chemical removal which is dominated ( 95%, see below) by photolysis. This allows us to explore the response of potential HONO source(s), parameterised in terms of the measured variation of sunlight, NOx abundance and aerosol particle parameters, by comparison with the observed HONO variation over the course of the eclipse.
The attenuation of photolysis frequencies was approximated using values obtained from the TUV model 41 for clear-sky conditions, with the reduction in j values modelled as a 1- sine wave. We note that this is an approximation, both to the actual photolysis frequencies and the geometric coverage of the solar disk, but one which is acceptable in the context of the  Figure S1); while the optmised parameters vary slightly the conclusions regarding which mechanism(s) are consistent/inconsistent with the data are unchanged). All other species were set to their actual observed levels for the relevant point in time. It is important to note that we do not assume HONO is in steady state / equilibrium with its production and removal terms. For each source scenario, we repeat the analysis under two conditions -firstly for the actual (eclipse influenced) reduction in solar intensity (photolysis frequencies), and secondly, for comparison, a hypothetical scenario where no eclipse occurred and clear-sky photolysis (as derived from the changing SZA) applied throughout the 2 hour time period. The resulting predicted and observed HONO concentrations are shown in Figure 2, and discussed below.

Observations
During the eclipse, weather conditions were calm (wind speed of 0.5 ± 0.4 m/s) with a mean relative humidity of 76 ± 6 % and temperature of 6.7 ± 1.7 o C, which are typical for the time    18,19 To explore the effect of such reactions, a HONO source which scaled with the product of j NO2 and the NO 2 mixing ratio was explored in Case 4 (Figure 2d), where a significantly better fit was observed, indicating that a source term which scaled with j NO2 x [NO 2 ] could account for the majority of HONO production in this environment.

Evaluation of HONO source scenarios
Direct emission of HONO from vehicles is another potential source in urban areas. 9,33,50,51 In an urban area, particle total surface area (TSA) can be used as a reasonably conserved tracer for vehicle emissions. 54 Case 5 ( Figure 2e) considers a scenario where the HONO source scales with the TSA; however this also does not match the measured HONO levels during the eclipse, implying that a source scaled to particle TSA alone -or direct vehicular emission of HONO in isolation -cannot explain the observed HONO concentrations in this environment.
Similar results were observed when particle number concentration was used as a metric of traffic particulate matter emissions, as opposed to TSA (not shown). However, HONO production proportional to the product of j NO2 and TSA was a better match to observed (Case 6, Figure 2f). not reproduce the observed behaviour, indicating that such heterogeneous aerosol conversion of NO 2 to HONO, without photoenhancement, could not, in isolation, account for HONO production in this environment. Case 8 showed significantly closer reproduction of the measured HONO evolution under the eclipse conditions, giving the best fit (lowest RMS residual) of all the scenarios considered, and indicating that a mechanism dependent upon solar insolation (j NO2 ), NO 2 abundance and aerosol surface area is able to reproduce the observed HONO behaviour across the eclipse. The similarity of eclipse-and non-eclipse simulations reflects the balanced photolytic dependence -as observed for the actual HONO.
While we obtain the optimal fit for case 8, in comparison to (e.g. ) case 2, the geometric surface area of ground surfaces significantly exceeds that of aerosol particles (by at least a factor of 16-fold, assuming an aerosol surface area of 300 μm 2 cm -3 and boundary layer height of 200m, which may be representative of the measurement location early in the morning -low winds, under an eclipse condition), and hence it may be likely that ground surface-mediated photoenhanced conversion of NO 2 dominated HONO formation at this location, potentially augmented by contributions from aerosol surfaces.
In reality, it is likely that a combination of mechanisms occur in parallel, and addition of the OH + NO reaction, with OH levels assumed equal to those measured 42 and scaled to the relative solar intensity expected during the eclipse (which may underestimate OH, given the importance of non-photolytic sources such as alkene ozonolysis at this location in wintertime found previously by Heard, et al. 42 resulted in a slight improvement to the statistical fitalthough increasing degrees of freedom would be expected to improve agreement. As a further evaluation of the performance of scenario 8, we applied this analysis approach to data from the days preceding and following the eclipse event (i.e. 19 th and 21 st March respectively); the simulation was able to satisfactorily reproduce the observed HONO levels (Supplementary Info, Figure S2). We note that in Cases 3, 4, 7 and 8 the NO x would vary due to changing PSS in the (hypothetical) non-eclipse case. This approximation has the effect of biasing the non-eclipse (hypothetical) simulations high (as the real NO 2 would be lower, in the absence of the eclipse-derived photolysis attenuation). A further analysis was undertaken to derive the cumulative HONO production with time, i.e.

Insight from temporal variation of HONO production
to remove the effect of the loss of HONO through photolysis under both the real (eclipse) and hypothetical (non-eclipse) conditions. This quantity -hypothetically conserved HONO, [HONO] Conserved ) was calculated using Equation 2: [HONO] Conserved = C n ) Where the [HONO] Conserved estimates the total HONO concentration that has been formed / emitted up to each point in time, derived from the observations after accounting for photolysis. HONO removal through reaction with OH is neglected, as above. C n and C n-1 represent the observed HONO at time t n and t n-1, while ∆t is the time difference. A modest difference in HONO conserved between the eclipse and without-eclipse conditions can be seen ( Figure 3a). Concentrating on the actual (eclipse) condition, the conserved HONO shows an approximately monotonic rise, indicating an approximately constant emission rate. This is at variance with traffic-related emissions, which would be expected to show the characteristic "rush hour" behaviour, i.e. peaking at 8 -9 am. Figure 3b shows the HONO conserved /NO x ratio -an approximately constant ratio would be expected if direct emissions dominated production (of both species), assuming NO x is conserved, however in fact significant variation with time is observed (the chemical NO x lifetime is estimated to be 17.5 hours with respect to OH + NO 2 , using the measured OH from Heard, et al. 42 , indicating that NO x can be considered to be approximately conserved on our timescale of 3.5 hours). The lack of a traffic / rush-hour pattern in the HONO conserved and HONO conserved/ NO x ratio further supports the inference that direct vehicular/combustion emissions, in isolation, are not the dominant source of HONO at this location, with some additional, photochemical / heterogeneous atmospheric chemical term is required to account for the observed behaviour. However, the results from Scenarios 1-8, which point to involvement of NO 2 in all HONO production mechanisms with a degree of involvement with the observed behaviour, reinforce the ultimate importance of vehicle emissions (as the overall source of NO x ).

Concluding Remarks
HONO, NO x , O 3 and particle number size distributions were measured at an urban background location over the course of a near-total solar eclipse in Birmingham, UK. The observed NO, NO 2 and O 3 responded as anticipated from the well-understood atmospheric

Conflicts of interest
There are no conflicts to declare

Figure 2
Comparison between measured (red points) and calculated HONO mixing ratio during the solar eclipse, using actual eclipse photolysis frequencies (solid blue line) and (hypothetical) non-eclipse photolysis frequencies (dotted blue line). Shading indicates eclipse duration from first to last contact. Source scenarios as defined in Table 1.