Our measurement campaign took place between 22 January and 3 February 2022 at Halley VI Research Station (75∘34′5′′ S, 25∘30′30′′ W), which is located on the Brunt Ice Shelf, Antarctica, at 32 m above mean sea level (Fig. ). This work was carried out in the Clean Air Sector (CAS), 1.5 km south of the main station buildings, avoiding the influence of pollution from station generators and vehicles. The instrument to detect atmospheric HONO was housed in a container at ground level, 10 m north of the CAS laboratory. The average wind speed was 10 ms-1, and reached up to a maximum of 26 ms-1 . The dominant wind direction during the campaign was east (see Fig. ), and the air temperature was between -13 and +1 ∘C, with a mean of -4 ∘C. All times are given in UTC, where local noon and midnight were at 14:00 and 02:00, respectively.

HONO was detected using a long-path absorption photometer (LOPAP; QUMA Elektronik & Analytik GmbH) which has been described in detail elsewhere . Briefly, the instrument works by first collecting HONO in a stripping coil, housed in a temperature-controlled external sampling unit, by a fast chemical reaction in an acidic (pH of 0) sulfanilamide solution (reagent 1, 1 gL-1, lower than originally proposed; see ). HONO is initially converted into NO+ which forms a diazonium salt by reaction with sulfanilamide. Due to the fast chemical reaction, much shorter gas–liquid contact times are applied (four-ring coil) compared to other wet-chemical HONO instruments (typically ≥10-ring coils), which require physical solubility equilibrium. This approach minimises sampling of interferences. In addition, the acidic sampling conditions slow down most known interfering reactions, which are faster under the neutral to alkaline conditions typically used in other wet-chemical instruments . The solution is then pumped via a 3 m long temperature-controlled reagent line to the main instrument, in which an azo dye is formed by reaction with a 0.1 gL-1 N-(1-naphthyl)-ethylenediamine dihydrochloride (NED) solution (reagent 2). The dye is detected in long-path absorption tubing (path length of 5 m) by a spectrometer (Ocean Optics SD2000) at 550 nm. The dye concentration can be related to the atmospheric HONO amount fraction by carrying out calibrations with nitrite solutions of known concentrations and knowing the sample air-to-liquid flow rate ratio.

The sampling unit is made up of two stripping coils in series such that HONO and some interferences are taken up in the first coil, followed by only interferences in the second. The interferences are assumed to be taken up to the same small extent in both channels so that the HONO amount fraction can be calculated by subtracting the signal in channel 2 from that in channel 1 . The instrument has been studied for the effect of various possible interfering species, including NO, NO2, O3, peroxyacetyl nitrate (PAN), HNO3, and even more complex mixtures of volatile organic compounds (VOCs) and NOx in diesel engine exhaust fumes . The instrument gave good agreement with the differential optical absorption spectroscopy (DOAS) technique under complex urban and smog chamber conditions . However, comparison of the instrument under pristine polar conditions is still an open issue. In the present study, the average interference was 40 % of the channel 1 signal, highlighting the importance of using a two-channel instrument, in excellent agreement with other studies of LOPAP instruments under polar conditions and at high mountain sites .

During the campaign, the LOPAP was calibrated every 5 d using nitrite solutions of known concentration (2×10-3 and 8×10-4 mgL-1). To maximise the instrument sensitivity, the gas-to-liquid flow rate ratio was optimised: the gas flow rate was set to 2 Lmin-1 (298 K, 1 atm) with the internal mass flow controller and checked frequently using a flow meter (DryCal DC-Lite), and the liquid flow rate through the stripping coil for each channel was regularly measured volumetrically and was between 0.15 and 0.18 mLmin-1 during the measurement period. Baseline measurements were made every 6 h using a flow of pure nitrogen (99.998 %; BOC) at the instrument inlet. The detection limit (3σblank) was 0.26 pmolmol-1 for the measurement period. The average response time (90 % of final signal change) was (8 ± 1.5) min.

The LOPAP sampling unit required adaptation for use in cold polar environments. The sampling unit box and the reagent lines between this and the main instrument have been coated in Armaflex insulation: no HONO emission from such insulation materials has been detected . The stripping coil and tubing to the sampling unit are temperature-controlled by a flow from a water bath (Thermo Haake K10 with DC10 circulator). The temperature of the sampling unit was kept at +18 ∘C.

The instrument's external sampling unit was mounted 0.4 m above the snow, 2.4 m from the container entrance, and pointed into the dominant wind direction (east). For most of the campaign, the sampling unit was stationary, except for a period of 12 h between 15:00 on 31 January and 03:00 on 1 February 2022 when the height was changed in regular intervals between 0.24 and 1.26 m above the snow in order to measure the HONO gradients needed to estimate vertical fluxes. An automatic elevator, built in-house, was used to raise and lower the sampling unit every 15 min (travel time of 1 min), meaning there was no human involvement in moving the sampling unit, and no tubing was needed to sample at different heights. Such tubing provides an artificial surface for HONO formation . The elevator is depicted in Fig. . The LOPAP data were shifted to account for the time delay ((17 ± 2) min) between gas intake and the observed absorption signal. This is determined from the average of all abrupt concentration changes (start/stop of blanks) and defined as the time between concentration change and the 50 % response of the instrument.

The surface ozone amount fraction was measured simultaneously from the CAS lab by UV absorption (Thermo Scientific Model 49i Ozone Analyzer). Data were collected at a 10 s interval, quality controlled, and then averaged to 1 min for this analysis. Instrument limit of detection (LOD) was taken to be 3σ of 2 h of 10 s measurements of zero air. This was calculated to be 0.38 nmolmol-1. The analyser inlet pointed east and was located at 8 m above the snow surface.

For the HONO flux density calculation the wind speed and direction was measured with a 2D sonic anemometer (Gill Wind Observer 70) located 1.5 m south of the sampling unit and 1 m above the snow. The temperature gradient was measured with two thermometers (TME Ethernet Thermometer) mounted on the vertical post of the elevator at 0.05 and 1.26 m above the snow surface. During the campaign, the incoming shortwave solar radiation (300–2800 nm) was measured by a net radiometer (Kipp & Zonen; CNR4) located at the main station (1.5 km from the HONO sampling site). The ozone column density was measured with a Dobson spectrophotometer also at the main station.

Surface snow samples were collected from the top 3 cm of snow in the clean air sector on 6, 25, and 31 January 2022. The samples were collected using clean sampling procedures (wearing clean room suits, gloves, and masks) and transferred into 50 mL polypropylene tubes with screw caps (Corning CentriStar), which had been rinsed with UHP water and dried in a class 100 clean laboratory in Cambridge prior to field deployment. The samples were transported back to the UK at -20 ∘C where they were melted and analysed for major ions including nitrate using Dionex Integrion ICS-4000 ion chromatography systems with reagent-free eluent generation. A Dionex AS-AP autosampler was used to supply sample water to 250 µL sample loops on the cation and anion instruments. Anion analyses were performed using a Dionex Ionpac AG17-C (2 µm, 2×50 mm) guard column and AS17-C (2×250 mm) separator column. A 3.5–27 mM potassium hydroxide eluent concentration gradient was used for effective separation of the analytes. Calibration was achieved using a range of calibration standards prepared from Sigma-Aldrich standards (1000 µgg-1) by a series of gravimetric dilutions. Measurement accuracy was evaluated using European reference materials ERM-CA408 (simulated rainwater) and CA616 (groundwater) which were all within 5 %. The LOD was 2 ngg-1.

The flux-gradient method was used to determine the HONO flux density following a similar approach as done previously for NOx in Antarctica . By measuring the HONO amount fraction at two heights, the concentration gradient can be found and is related to the flux density by 1F=-Kcdcdz for which Kc is the turbulent diffusion coefficient (in m2s-1) of a chemical tracer. In the atmospheric boundary layer, Kc may be approximated by the eddy diffusion coefficient for heat, Kh . It should be noted that a negative gradient in amount fraction will result in a positive flux density, equivalent to emission from the snow.

Monin–Obukhov similarity theory (MOST) is used to parameterise fluxes in the surface layer, about 10 % of the depth of the atmospheric boundary layer , where turbulent fluxes are assumed to be independent of height. The flux density can be calculated by 2F=κu*[c(z1)-c(z2)]∫z1z2ΦhzLdzz, where κ is the von Karman constant (set to 0.4); u* is the friction wind velocity, found from wind speed measurements; and c(z) is the HONO amount fraction at height z. ∫z1z2Φh(zL)dzz is the integrated stability function for heat, a function of zL where L is the Obukhov length. The full derivation of Eq. () is in Appendix A.

The application of MOST requires certain conditions to be met : (a) the flux density is constant between the two measurement heights, (b) the lower inlet height is above the surface roughness length, (c) the upper measurement height is within the surface layer, and (d) the measurement heights is far enough apart for the detection of a significant difference in amount fraction.

For (a), the chemical lifetime (τchem) with respect to photolytic loss was compared to the transport time (τtrans) between the two measurement heights. If τchem is much larger than τtrans, then the flux density can be assumed to be constant. τchem found from the inverse of the photolysis rate coefficient, J(HONO), was between 10 and 80 min. The transport time can be estimated by 3τtrans=(z2-z1)∫z1z2dzKh=(z2-z1)∫z1z2ΦhzLdzzκu*. The transport time between 0.24 and 1.26 m above the snow for the flux measurement period at Halley was between 16 and 29 s. In all cases, the lifetime was significantly longer than the transport time, meaning the flux density can be assumed to be constant between the two heights.

The lower measurement height was 0.24 m, which is significantly above the surface roughness length of (5.6 ± 0.6) ×10-5 m measured previously at Halley .

During the Antarctic summer, the boundary layer height at Halley is regularly stable, making it difficult to define . Previous analysis of sodar (sound detection and ranging) measurements has suggested that the boundary layer at Halley in summer is consistently above 40 m . The equations of both and have been used to estimate the mixing height at Antarctic locations (South Pole – ; Dome C – ). Though they are unlikely to predict the height accurately, these equations can provide a useful estimate of the minimum boundary layer height. For the period in question, this is calculated to be 75 and 95 m (, and , respectively). When temperature profiles recorded by daily weather balloon launches during the measurement campaign show a temperature inversion , this was above 100 m. Therefore, the upper measurement height, 1.26 m, was very likely within the surface layer.

A t test confirmed that the amount fraction difference between the two heights (Δy) was significant (p<0.01).

All of the above criteria were satisfied for the measurement period, so MOST was used to calculate the flux density by the method described above.

The rate coefficient of photochemical reactions can be calculated from 4J=∫λ1λ2σ(λ,T)φ(λ,T)F(λ)dλ, where σ and φ are the absorption cross-section and quantum yield for the photolysis reaction of interest, functions of wavelength (λ), and temperature (T). F is the actinic flux derived from the TUV radiation model over the wavelength range 300 to 1200 nm using measured ozone column density, a surface albedo of 0.95, and assuming clear-sky conditions . The calculated J values were then scaled by the ratio of measured and modelled incoming shortwave solar radiation to account for non-clear-sky conditions (see Fig. ). It is noted that the wavelength ranges of modelled and measured radiation are not exactly the same, but the contribution of wavelengths >1200 nm is expected to be small.

HONO amount fractions measured at Halley were between <0.3 and 14 pmolmol-1 (Fig. ), with a mean of 2.1 pmolmol-1 .

These HONO amount fractions are some of the lowest ever observed in Antarctica (see Table ) and are only the second series of HONO observations at an Antarctic coastal ice-shelf location. When measurements were attempted once before at Halley, it was thought that the HONO amount fractions were overestimated . The HONO data collected in this study support this suggestion, as the mean is 2.1 pmolmol-1 compared with around 7 pmolmol-1 measured in January–February 2005 . During this measurement period the average interference was 40 % of the channel 1 value, and occasionally >100 %, showing that HONO would be significantly overestimated without the two-channel sampling unit of the LOPAP.

At other coastal sites, the HONO amount fractions are close to those seen in this study. At Browning Pass (Fig. ), HONO amount fractions were <5 pmolmol-1, though the site conditions are unlike Halley, since the snow composition and pH may be affected by rock outcrops nearby . Small amount fractions were also observed at Dumont D'Urville (DDU), but this was attributed to the fact the site had no snow cover . Higher amount fractions were observed in continental air masses, likely due to emissions from the snowpack on the continent.

HONO amount fractions at inland Antarctic locations are predominantly higher than those seen at Halley. At the South Pole, mean HONO amount fractions of 6.3 pmolmol-1 were measured by LIF . Dome C HONO amount fractions, measured using a LOPAP, were found to be higher than in most other studies (mean ca. 30 pmolmol-1) . The higher HONO and NOx amount fractions can be explained by specific conditions on the high Plateau during summer, which include 24 h sunlight, a shallow and frequently stable boundary layer, and very low temperatures leading to low primary production rates for HOx radicals . This causes a non-linear HOx–NOx chemical regime where the NOx lifetime increases with increasing NOx as proposed previously . Together, these factors support increased air–snow recycling and the accumulation of NOy in the regional boundary layer.

An interference of HNO4 in the LOPAP has been suggested . The LOPAP's response to HNO4 has been investigated in both the laboratory with an HNO4 source and in the field at Dome C by placing a heated tube at the instrument inlet to decompose HNO4. Both showed that the LOPAP partially measures HNO4 as HONO with approximately 100 pmolmol-1 HNO4 leading to a HONO interference of 15 pmolmol-1, but further investigation is needed to systematically quantify this effect . In any case, Dome C is expected to have a much higher HNO4 amount fraction than Halley due to the fact its lifetime is controlled by thermal decomposition: R12HNO4+M→HO2+NO2+M. The rate coefficients for thermal decomposition of HNO4 are in Table . An average HNO4 lifetime with respect to thermal decomposition of 12 min was calculated from 1k12 at Halley. This can be compared to an average lifetime of 21 h during January 2012 at Dome C (mean temperature of -31 ∘C). As a further check on this interference, the steady-state concentration of HNO4 at Halley was calculated. The method for this is detailed in Appendix B; the concentrations of HO2, NO2 and OH from the CHABLIS campaign were used. The average steady-state amount fraction was 0.05 pmolmol-1. Using the estimate of , this suggests that the interference is likely <0.01 pmolmol-1, well below the detection limit of the LOPAP.

The median diurnal cycle of HONO amount fraction shows a maximum at solar noon (Fig. ), with a peak-to-peak amplitude of 2 pmolmol-1. Previous observations of HONO at Halley also showed a diurnal cycle but with a larger day-to-night variation , which is likely an overestimate of the true variation in HONO amount fractions. The diurnal cycle observed at Browning Pass compares well with that observed here: the maximum variation was 1 pmolmol-1 . The diurnal cycle of HONO amount fraction at Dome C showed a double peak due to the presence of a strong diurnal cycle in the boundary layer height; the amount fraction dipped at midday when the boundary layer height showed a strong increase . The boundary layer height at Halley does not show such a diurnal pattern , so the variation is predominantly caused by the photochemical production of HONO peaking at solar noon.

The measured flux density is plotted in Fig.  with the data needed for its calculation: the wind speed and temperature used to find the turbulent diffusion coefficient for heat (Kh) and the measured amount fraction gradient. The HONO amount fraction gradient is steep: the amount fraction decreases by about half between 0.24 and 1.26 m above the snow. The flux density varies between 0.5 and 3.4×1012 m-2s-1 from the snow surface, mainly driven by the amount fraction gradient, and appears to decrease between solar noon at 14:00 UTC and solar midnight at 02:00 UTC, suggesting a photochemical snowpack source. In the Antarctic, HONO fluxes have previously only been measured at Browning Pass, where larger values were found (mean upwards flux density of 4.8×1012 m-2s-1). However, occasionally the flux was downwards, equivalent to deposition. This is likely caused by the atypical snow composition: the snow is only weakly acidic and occasionally alkaline . More HONO flux measurements are available from the Arctic: flux densities up to 1014 m-2s-1 from the snow have been observed . However, these Arctic sites are more polluted and therefore there are more HONO precursors in the snow (NO2, nitrate, organics). used measurements of the NOx flux density at Dome C and the HONO to NOx production rate ratio measured in a snow photolysis experiment in the laboratory to estimate a HONO emission flux density between 5 and 8×1012 m-2s-1, larger than that observed here, likely due to the higher snow nitrate concentrations at Dome C.

HONO formation in the snowpack, driving the flux to the boundary layer above, is typically attributed to nitrate photolysis. A HONO flux density from this reaction can be compared to that measured in this study to determine the source of HONO. The production rate (per area) above snow of reactive nitrogen from snow nitrate photolysis, Psnow(NOy), can be estimated using the following equation: 5Psnow(NOy)=∫0∞J(NO3-)[NO3-]dz, where J(NO3-) is the nitrate photolysis rate coefficient, a function of depth in the snow pack, z. This can be approximated by J0(NO3-)e-zze , where J0(NO3-) is the photolysis rate coefficient at the snow surface. The e-folding depth (ze) is between 3.7 and 10 cm (7 cm is used here; ). [NO3-] is the nitrate number concentration (in units of cm-3). To derive HONO production from nitrate photolysis, a HONO yield coefficient Y(HONO) is included , and we also explicitly show the conversion from nitrate mass fraction to number concentration: 6Psnow(HONO)=∫0∞J0(NO3-)e-zzew(NO3-)ρsnowNAM(NO3-)Y(HONO)dz. w(NO3-) is the snow nitrate mass concentration (mass per mass of snow); ρsnow is the snow density, taken as 0.3 gcm-3 ; NA is Avogadro's number; and M(NO3-) is the nitrate molar mass. J0(NO3-) was found from Eq. () using NO3- absorption cross-sections and quantum yields for ice , and it is 1.2×10-7 s-1 at noon for the measurement period (assuming clear-sky conditions).

As detailed in Sect. , the surface snow nitrate mass fraction was measured, the mean value is (78.2 ± 18) ngg-1. A yield of 100 % gives a HONO production rate of 1.9×1012 m-2s-1 at a light intensity corresponding to local noon, below the maximum measured flux density of 3.4×1012 m-2s-1. However, the yield is unlikely to be as high as 100 %.

There are two product channels for nitrate photolysis (Reactions  and ); if it is assumed that all nitrite produced in Reaction () is converted to HONO, Y(HONO) would be 10 % because Reaction () dominates over Reaction () by a factor of 9 . This gives a noon HONO production rate of only (0.19 ± 0.13) ×1012 m-2s-1. This is lower than the measured flux density (see Fig. ). However, it has been suggested that the rate of nitrite production from nitrate photolysis could be as high as that of NO2 implying Y(HONO) is greater than 10 %. Furthermore, the HONO yield from nitrate photolysis may also include a contribution from NO2 reacting on photosensitised organics (Reaction ), and to a lesser extent from NO2 disproportionation (Reaction ), which would further increase Y(HONO) and bring the HONO production rate from nitrate photolysis closer to the measured flux density. As well as the stated uncertainties in ze and w(NO3-), there is uncertainty in the quantum yield for nitrate photolysis in snow: an error of 50 % has been calculated from . These uncertainties are represented by the red shading in Fig. .

Other assumptions made when carrying out this calculation include that the light attenuation in snow is exponential and that the snow density is 0.3 gcm-3, with a nitrate mass fraction that does not change with depth in the snow pack. It has also been assumed that all HONO produced will be released from the snow immediately; snowpack-produced HONO can be vented via wind pumping from the open snow pore space into the air above . The largest gradient in HONO amount fraction with height observed occurs during a wind speed increase (see Fig. ) suggesting such wind pumping does occur at Halley. The degree of wind pumping will be affected by snow permeability, which is related to snow porosity . During this measurement period, the snow was fresh and therefore more porous and likely more permeable than aged snow.

Photochemical reaction of NO2 on organic surfaces in the snow is a commonly suggested HONO formation mechanism (Reactions  and ; ). There is likely to be significant photosensitised organic matter present in the snow at Halley. found fulvic acid mass concentrations between 16 and 400 µgL-1 in coastal snow in east Antarctica, and found the total organic carbon (TOC) concentration of surface snow was 88 to 928 µgL-1 along a transect to the coast in east Antarctica with the higher values measured nearer the coast, which they attributed to marine sources associated with sea spray. have highlighted that these studies could overestimate the organic matter content due to their sampling method and measurement technique. They suggest that the organic matter at coastal Antarctic sites could be lower, comparable to inland sites like Dome C (3–8 µgL-1). suggest that this could still lead to significant HONO production. Assuming Dome C and Halley snow have similar organic content, a HONO flux density can be estimated based on the HONO:NOx emission ratio measured in a laboratory study of Dome C snow and the measured NOx flux density at Halley . The HONO:NOx ratio is temperature dependent: the highest temperature studied by is -13 ∘C, which is below the Halley air temperature for the flux measurement period. An emission ratio of 0.77 and NOx flux density of 7.3×1012 m-2s-1 give a HONO flux density of 5.6×1012 m-2s-1, close to the measured value. NO2 has been measured previously at Halley (see Table ): amount fractions were lower than at other sites, regularly <10 pmolmol-1, which would limit HONO production via this mechanism. However, in interstitial air in snow blocks at Neumayer station (a similar coastal ice-shelf location), NO2 amount fractions were found to be higher than in ambient air, up to 40 pmolmol-1 . In their laboratory study of this HONO production mechanism, estimate that such an NO2 amount fraction, with a snow TOC concentration of 10 to 1000 µgL-1, would lead to a flux density of 3×1012 to 4×1012 m-2s-1. This estimate agrees well with the measured HONO flux density from the snow at Halley, provided that all HONO produced is also emitted into the atmosphere. To investigate this possible source further the snowpack must be analysed in more detail for the presence of such photosensitised species.

In order to assess the consistency of the measured HONO amount fractions and flux density, a simple box model calculation was undertaken. The change in the atmospheric HONO amount fraction over time can be written as the sum of the main sources and sinks: 7d[HONO]dt=k4[NO][OH]+Pss(HONO)h-J(HONO)[HONO]-k5[OH][HONO], which takes into account the production of HONO from the snow (Pss(HONO)), as well as HONO formation through NO and OH combination (Reaction ) and loss through reaction with OH (Reaction ). This can be simplified as the rates of Reactions () and () are typically slow, especially under remote conditions, meaning that in this model, the HONO budget is dominated by emission of HONO from snow nitrate photolysis and atmospheric photolysis of HONO itself. Rearranging and simplifying Eq. () gives 8Pss(HONO)=h×d[HONO]dt+J(HONO)[HONO].

For simplification, it is assumed that the emitted HONO is homogeneously mixed in a layer of height h. The boundary layer height at Halley can be hard to define , but for the current measurement period, the boundary layer height is likely above 40 m . These calculated flux densities, for h between 10 and 50 m, reflect the shape of the measured flux density well showing peaks at noon, though the calculated flux densities appear to decrease to 0 at night, which the measurements do not (Fig. ).

The assumption of a constant HONO amount fraction up to height h above the snow surface is the largest source of uncertainty in this simple box model. Steep gradients in HONO amount fraction are expected, caused by the ground surface source, the turbulent transport, and the photolytic loss of HONO in the atmosphere. The gradients were confirmed in the present study, for which the HONO amount fraction decreased to ca. half between 0.24 and 1.26 m height (see Fig. b). However, these gradients can only be described correctly by a 1D model approach, which is out of the scope of the present study. Errors in the flux density calculation caused by deviations from MOST are also a possibility, and this comparison is further limited by the flux density measurements only being possible for one 12 h period.

For completeness, we tried including reactions involving NO and OH (Reactions  and ) in the simple box model represented by Eq. (). Due to the lack of concurrent observations, previous measurements of other gases during the CHABLIS campaign in 2004 and 2005 at Halley were used for further calculations. Specifically, OH and NO data for the days of January and February 2005 corresponding to the days in 2022 on which HONO was measured (22 January to 3 February) have been used here. NO amount fractions were low, with a mean of 5.7 pmolmol-1, but showed a diurnal cycle peaking between 19:00 and 20:00 UTC, 5 h after solar noon (14:00 UTC) . The mean OH concentration was 3.9×105 cm-3 with an average noontime maximum of 7.9×105 cm-3 . These data, along with other gases that were measured in the campaign, are summarised in Table . The reaction rate coefficients used in these calculations are summarised in Table . Using these NO and OH concentrations, a new value of Pss(HONO) was calculated. As expected, the inclusion of these reactions does not make a large difference; the flux density calculated by this method is on average 3 % smaller than that from Eq. (), though it is also occasionally larger.

If the flux density from the snow is ignored, the photostationary-state (PSS) HONO amount fraction can be calculated. This assumes HONO is solely formed in the air through Reaction () and lost through Reactions () and (). 9d[HONO]dt=0=k4[NO][OH]-J(HONO)[HONO]-k5[OH][HONO] Using NO and OH CHABLIS data again, an average photostationary-state HONO amount fraction of 0.07 pmolmol-1 was calculated. This showed a diurnal cycle with a maximum at solar noon and minimum at night. However, this calculation is only valid at the HONO measurement height of 0.4 m if it is assumed there are no gradients in the NO and OH amount fractions which were measured at higher altitudes, 4.5 to 6 m above the snow. HONO was found to have a steep amount fraction gradient, which suggests that by 4.5 to 6 m above the snow, the amount fraction could be close to the photostationary-state amount fraction.

The inclusion of HONO formation through a dark reaction of NO2 on surfaces (Reaction ) in the PSS calculation raised the HONO amount fraction to 0.3 pmolmol-1, which is still significantly lower than that measured. For this reaction, k10=1.0×10-5 s-1 was estimated from the nighttime increase in the HONO:NOx ratio in the average diurnal cycle observed at Halley (see Table ) . Clearly an additional source is required to raise the HONO amount fraction above stationary-state levels.

The HONO:NOx amount fraction ratio can provide a check on the HONO data: under steady-state conditions of HONO and NOx sources and sinks, and assuming that all NOx is produced by HONO photolysis as an upper limit, the HONO:NOx ratio should approach that of their lifetimes (τ(HONO):τ(NOx)) .

Using the HONO data collected and the NOx amount fraction for the same time period in 2005 , the ratio of the amount fractions was calculated. This is between 0.15 and 0.35 and shows no diurnal cycle. This is significantly lower than other studies , supporting that our measurements are comparatively free from interferences. Only during measurements in Utqiaġvik, Alaska (formerly Barrow, Alaska) were even lower HONO:NOx ratios of 0.06 observed, also using the LOPAP technique . However, the 2022 HONO measurements were made significantly closer to the snow surface than the 2005 NOx measurements (0.4 m compared to 6 m). The steep gradient in HONO that has been observed suggests that the HONO amount fraction at 6 m above the snow will be considerably lower. This would further reduce the ratio, which still supports that these measurements are relatively free from interferences.

The ratio of HONO to NOx lifetimes is 0.07 at night (80min:19 h, τ(HONO) calculated for loss by photolysis and τ(NOx) for loss by reaction with OH; ). This is a factor of 2–4 lower than the measured HONO:NOx ratio. The daytime ratio of lifetimes (12min:6 h; ) is 0.03 and is even lower compared to the range measured because the HONO measurements were made close to the snow surface (the HONO source) where the steady-state conditions are not fulfilled. However, found that the NOx lifetime was reduced by halogen processing (BrNO3 and INO3 formation and heterogeneous uptake). A reduced NOx lifetime would improve the agreement with the observed HONO:NOx ratio.

HONO photolysis is an OH source; when the measured HONO amount fraction is higher than the calculated photostationary-state amount fraction, HONO is a net source of the OH radical. The importance of this net source compared to that from ozone at Halley is plotted in Fig. .

As part of the CHABLIS campaign, calculated the contribution of various OH sources to the HOx budget at Halley. It was found that the halogen species contributed the most; Halley's coastal location means there are high halogen concentrations . The halogen species HOI and HOBr show a strong diurnal pattern and contribute up to 89 and 14 pmolmol-1h-1 OH at solar noon respectively. The OH production by HOBr is comparable to that calculated for HONO from the measurements made here (Table ). However, it must again be highlighted that the halogen measurements were made higher above the snow (4 to 5 m; ) than those of HONO and that the HONO OH source estimate is only valid at 0.4 m where the HONO was measured. At higher altitudes above the snow, HONO will approach the PSS amount fraction leading to a lower contribution to the OH radical source. Thus, the contribution of HONO to the OH radical initiation will be limited only to the lowest part of the mixing layer.

The surface ozone amount fraction recorded during this measurement period is comparable with that measured during the CHABLIS campaign (mean of 14 nmolmol-1 here, 10 nmolmol-1 during January 2005) and lower than at other Antarctic sites due to the high halogen concentrations. However, at the other sites in Table , HONO is a more important OH source than ozone. The low temperatures at these sites (mean: Utqiaġvik -26, DC -31 ∘C) mean the extremely dry air limits OH production via Reactions () and (). At Halley, a warmer coastal location, the water vapour concentration is not limiting, and OH formation through ozone photolysis still dominates. Nevertheless, HONO photolysis as a source of OH cannot be discounted. The HONO amount fractions that were initially thought to be present at Halley, detected during the CHABLIS campaign , would lead to an overestimation of the HOx budget . Besides conceptional problems when comparing box model calculations with HONO measurements close to the snow surface, these HONO measurements should make the HOx budget at Halley easier to rationalise.