Interactive comment on “ The detection of nocturnal N 2 O 5 as HNO 3 by alkali-and aqueous-denuder techniques ”

This paper presents co-located data on measurements of N2O5 and ‘HNO3’ for an upland site in central Europe, and shows that a significant fraction of ‘HNO3’ measured using a rotating wet denuder might have originated from sampling N2O5. The paper is clearly written, but some statements are made without justification (see below). A distinction also needs to be made more clearly about the ‘fitness for purpose’ of different measurement techniques – if an estimate of dry deposition is required (for example) the inclusion of N2O5 as ‘HNO3’ is probably irrelevant, given that both molecules would be expected to deposit at rates limited only by turbulence. If, however, measurements are made to elucidate processes occurring in the atmosphere at short time scales (e.g.


Introduction
Nitric acid (HNO 3 ) is an important fraction of gas-phase oxidised nitrogen and has been monitored and measured for a number of years as part of national and trans-national government sponsored efforts to understand the important issue of ecosystem acidification and transboundary air pollution (Sutton et al., 2011).HNO 3 can be considered to be the end-point for the gas-phase chemical processing of NO x (NO 2 + NO) via the reaction of NO 2 with the hydroxyl radical (OH) in the troposphere, the heterogeneous hydrolysis of N 2 O 5 , the thermodynamic repartioning of particle-phase nitrate, and the reaction of the nitrate radical, NO 3 , with dimethyl sulphide (DMS) (Yvon et al., 1996) and a number of VOC species (Finlayson-Pitts and Pitts, 1997).HNO 3 is efficiently deposited on surfaces (Huebert and Robert, 1985) or can partition into the particulate phase in the presence of alkaline dust, sea salt or excess ammonia (Wexler and Seinfeld, 1992).However, not all NO x is lost from the atmosphere via formation of HNO 3 .
Dinitrogen pentoxide (N 2 O 5 ) is an important fraction of oxidised nitrogen which is frequently relegated as a minor player with respect to its contribution to N deposition.Methodological descriptions of monitoring techniques for HNO 3 rarely mention N 2 O 5 (e.g.Tang et al., 2009;Bytnerowicz et al., 2001;Markovic et al., 2012;Dong et al., 2012;Thomas et al., 2009) and do not consider the possibility that it may cause measurement interferences despite its ubiquity and reactivity.This may be for a number of reasons, including the technical challenges in its detection and, in the main, nocturnal occurrence.It is also true that, until relatively recently, the vast majority of N 2 O 5 was thought to end up as HNO 3 /particle nitrate and, therefore, it is possible that its role was considered interchangeable with HNO 3 .This may be true for wet and dry deposition; however, these are not sufficient reasons for its neglect when considering the mechanisms at work within the environmental nitrogen cycle for, as we shall see, the mechanisms of N 2 O 5 production, its fate, and its temporal occurrence differ markedly from HNO 3 .
N 2 O 5 is formed via a number of steps beginning with the reaction of NO 2 with O 3 : In the daytime NO 3 is rapidly photolysed reforming NO or NO 2 , or reacts with NO to reform NO 2 .However, during the night, when photolysis frequencies are low, and NO is removed by reaction with O 3 , NO 3 can react with NO 2 to form N 2 O 5 , which is in thermal equilibrium with its precursors: With the equilibrium given by K eq is strongly dependent on temperature, varying with ∼ exp (11 000/T ) , with cold conditions favouring the formation of N 2 O 5 .At an NO 2 mixing ratio of 1 ppbv and temperature of 270 K, the ratio of N 2 O 5 to NO 3 is about 300 : 1 which drops to a ratio of approximately 1 : 1 at a temperature of 295 K (Brown and Stutz, 2012).Losses of N 2 O 5 can be divided into direct and indirect losses.Indirect losses are processes resulting in the loss of NO 3 , e.g.via reaction with biogenic VOCs, or NO, which drive the equilibrium (1) to the right.Direct losses of N 2 O 5 are mainly heterogeneous hydrolysis on particles (Brown and Stutz, 2012) and other surfaces, or the reaction with aqueous particle chloride (Finlayson-Pitts et al., 1989), the total uptake efficiency depending strongly on the particle composition (Bertram et al., 2009;Chang et al., 2011): ClNO 2 is photolysed in the morning following production, reforming NO x in addition to the Cl radical.Ambient measurements of the formation of ClNO 2 are relatively recent and scarce; however, they show the importance of the chloride channel of the heterogeneous loss of N 2 O 5 in the marine (Osthoff et al., 2008;Riedel et al., 2012), continental North American (Thornton et al., 2010;Mielke et al., 2011), and Western European environments (Phillips et al., 2012).
A number of studies and monitoring networks have used data derived from denuder sampling techniques for both the estimation of acid deposition and the investigation of gasparticle dynamics of nitrogen in the atmosphere.In North America, the Clean Air Status and Trends Network (CAST-NET) (http://epa.gov/castnet/javaweb/index.html)measures nitric acid data by both nylon filter packs (Bytnerowicz et al., 2001) and, more recently, is assessing the performance of the MARGA (Monitor for Aerosols and Gases in Ambient Air, Metrohm Applikon BV, Netherlands) system for the measurement of fluxes and concentrations reactive gases and aerosols (Cowen et al., 2011).In Asia, the Acid Deposition and Monitoring Network in East Asia (EANET) (http: //www.eanet.cc/index.html)uses denuder methods to monitor nitric acid concentrations and estimate acid deposition.In the UK, the United Kingdom Eutrophying and Acidifying Pollutants (UKEAP) network uses monthly alkaline-denuder samplers to measure concentrations of nitric acid and estimate acid deposition.Across Europe as a whole, the NitroEurope project (Sutton et al., 2011) and European Monitoring and Evaluation Programme (EMEP) under the Convention of Long-range Transboundary Air Pollution use network of filter-pack systems and denuders in addition to a number of intensive monitoring periods (Aas et al., 2012) using higher time-resolution techniques to estimate acid deposition and study the processes controlling nitrogen chemistry and dynamics in the atmosphere.Data from these programs is freely available.
Denuder methods for the detection of nitric acid using the higher time resolution rotating annular denuder type measurements are becoming more common.Recent datasets obtained with MARGA-type systems and used for the evaluation of models and the investigation of gas-particle partitioning of nitrate have been reported by a number of authors (Aan de Brugh et al., 2012;Twigg et al., 2011;Aas et al., 2012;Makkonen et al., 2012;Schaap et al., 2011).The hourly time resolution of these datasets has a number of advantages over the monthly and weekly data available from monitoring networks, allowing researchers to investigate in more detail the chemistry and dynamics of nitrogen in the lower atmosphere.

Methods
We present data measured during August 2012 from the Taunus Observatory, Kleiner Feldberg, to the NW of Frankfurt, Germany.The Taunus Observatory is situated at the summit of the Kleiner Feldberg, 825 m above sea level, just north of the heavily populated Rhein-Main area of southwestern Germany.The site has been described in detail by Handisides (2001).A MARGA 1S annular denuder system was deployed on top of the observatory roof, approximately 3 m from a cavity ring-down absorption spectrometer (CRDS) system for the measurement of N 2 O 5 , previously deployed at this location in both 2008 and 2011 (Crowley et al., 2010b;Phillips et al., 2012).The instruments were deployed as part of a larger study of the nocturnal chemistry of NO 3 and N 2 O 5 .

N 2 O 5 by off-axis cavity ring-down spectroscopy
The N 2 O 5 system is a two-channel off-axis cavity ring-down system (OA-CRDS) which detects the NO 3 radical in one channel and the sum of N 2 O 5 and NO 3 via thermal dissociation in the second channel.The instrument has been deployed on a number of previous occasions and is described in detail by Crowley et al. (2010b).The concentration of N 2 O 5 is calculated by subtraction and total uncertainty is estimated at 15 %.

Monitor for Aerosols and Gases in Ambient Air
(MARGA) The MARGA system is a commercialised version of the GRAEGOR analyser developed by ECN, Netherlands (Thomas et al., 2009;ten Brink et al., 2009), and produced by Metrohm Applikon BV, Netherlands.The methodology and instrument description is contained in references above and so a brief description of the method follows.The MARGA was operated as per the manufacturer instructions as it would usually be operated in the field.The air sample was drawn, at 1 m3 h −1 , through a short, < 25 cm length of high density polyethylene (HDPE) tube, through Teflon coated PM 10 cutoff cyclone into a wet rotating denuder (WRD).The WRD consists of two concentric glass tubes wetted with IC grade water.The gases are efficiently denuded in the WRD, but the low mobility of particles prevents them from diffusing to the walls of the WRD.The air sample then enters the steamjet aerosol collector (SJAC) where the water soluble particle fraction is condensed using steam and collected.The samples from both the WRD and SJAC are analysed online every hour using ion chromatography.The detection system is continuously calibrated by the use of an internal standard of LiBr.
In addition, before starting the ambient measurements standard solutions were injected to check the retention times and the analytical system.The MARGA system measures very reactive analytes and therefore care must be taken with the inlet systems as significant losses of gases such as HNO 3 can occur.Our inlet system including cyclone were cleaned after approximately 2 weeks of operation.

Co-located MARGA HNO 3 and OA-CRDS N 2 O 5
The time series of gas-phase HNO 3 detected as NO − 3 , referred to subsequently as the nitrate anion, in the ion chromatograph and N 2 O 5 , both as N-equivalent mass, are plotted in Fig. 1, bottom panel.N 2 O 5 mixing ratios were variable with peak concentrations of approximately 800 pptv.The nocturnal portion of the MARGA time series exhibits clear portions, see Fig. 2, with a distinct similarity to the N 2 O 5 series and on certain days, e.g.16/17 August, the nocturnal MARGA signal is dominated, > 90 %, by signal most likely arising from the hydrolysis of N 2 O 5 in the denuder.This short series of data was obtained in the high summer with nocturnal temperatures ranging between 15 and 25 Celsius.Even in these summer conditions, the average campaign contribution of N 2 O 5 to the nocturnal MARGA HNO 3 signal is 17 % assuming a detection efficiency of 1 for N 2 O 5 and each N 2 O 5 molecule captured resulting in 2 NO − 3 ions detected.It is possible that in colder regions during winter, with longer nights and lower NO 3 loss rates, that a large N 2 O 5 mixing ratio relative to HNO 3 will be encountered.

Alkaline denuder and filter methods
The absorption of constituents of ambient air by denuding technology followed by determination of the nitrate anion concentration is a frequent method for the determination of ambient HNO 3 concentrations.For example, the Nitro Europe (NEU) monitoring network using the DELTA denuder system (Tang et al., 2009) to determine monthly HNO 3 concentrations from which deposition maps are developed and against which dispersion models are tested (Flechard et al., 2011).The acid gases are denuded by passing ambient air through alkali-coated glass tubes and subsequently tested for nitrate anion by ion chromatography.The efficient removal of N 2 O 5 (and NO 3 ) is likely to occur on the surface of alkalicoated denuder tubes resulting in the detection of two NO − remove HNO 3 for analysis are likely to suffer from N 2 O 5 detection artefacts in ambient conditions, though we expect that this would be favoured at high relative humidity resulting in the presence of water leading to N 2 O 5 hydrolysis on the filter surface.

Ambient measurements of N 2 O 5
N 2 O 5 is not a commonly measured trace gas owing to the technically difficult nature of its detection.Brown and Stutz (2012) present a comprehensive review of the current state of knowledge with respect to NO 3 and N 2 O 5 in the atmosphere.Ambient point measurements are usually achieved via absorption spectroscopy such as cavity ring-down spectroscopy (CRDS) (Schuster et al., 2009;Simpson, 2003;Brown et al., 2001).Mixing ratios of N 2 O 5 are highly variable (e.g.Matsumoto et al., 2005;Ayers and Simpson, 2006;Osthoff et al., 2006;Brown et al., 2003Brown et al., , 2007)).In regions with high concentrations of NO or VOCs, the steady-state concentration of N 2 O 5 can be below the detection limit, e.g.< 1 pptv, of current analytical instrumentation (e.g.Rinne et al., 2012).However, in situations with long NO 3 lifetimes and low particle surface areas, atmospheric mixing fractions can exceed several ppbv (e.g.Phillips et al., 2012) tying up a significant proportion of NO x .It is possible that during the cold, long, polluted northern hemisphere night, with relatively high concentrations of NO x and O 3 and low concentrations of BVOCs, that N 2 O 5 makes up a significant proportion of (HNO 3 + 2N 2 O 5 ).

The likely significance of the detection of N 2 O 5 as HNO 3
The chemical processes which control the formation and loss of N 2 O 5 and HNO 3 are not the same and therefore analytical techniques reporting N 2 O 5 as HNO 3 are systematically incorrect.HNO 3 will be produced in the presence of NO 2 in periods, usually the daytime, with sufficient OH radical concentrations and, to lesser extent owing to the lower reactivity, at night via NO 3 reactions with VOCs.The resulting HNO 3 may partition into the aerosol phase or be lost via dry and wet deposition.Further chemical transformations do not occur in atmospherically relevant timescales.Conversely, N 2 O 5 is produced via Reaction (R2) which, owing to the rapid photolysis of the nitrate radical, only occurs with sufficient rate during periods of low sunlight, i.e. the night.N 2 O 5 losses are, to some extent, similar to HNO 3 .N 2 O 5 is lost into the aqueous particle phase via hydrolysis producing HNO 3 /NO − 3 and additionally lost via dry and wet deposition.Although daytime deposition velocities have been assumed to be similar (Zhang et al., 2012) and turbulence limited, N 2 O 5 being present only at night, will be lost to dry deposition in addition to losses on aqueous aerosol surfaces; the final deposition rate via hydrolysis being mediated via the losses of particle NO − 3 .In addition, there is a second chemical pathway open to N 2 O 5 ; N 2 O 5 can react on the surface of Cl − containing aqueous particles, e.g.sea salt or NH 4 Cl, and produce ClNO 2 and HNO 3 .ClNO 2 is not soluble and will leave the aqueous particle system.The nightime losses of ClNO 2 are negligible and result in the survival of NO x in an unreactive form until the morning.At daybreak ClNO 2 is photolysed, liberating an active chlorine radical and NO 2 .The presence of this loss pathway for N 2 O 5 will prolong the atmospheric lifetime of NO y , and consequently the atmospheric transport distance, as it cycles back via the HO x -NO x chemical cycles.In addition to the recycling of NO x via ClNO 2 , N 2 O 5 may itself recycle NO x .The rapid photolysis of NO 3 at sunrise will drive the thermolysis equilibrium towards the formation of NO x .Measurements of HNO 3 which are sensitive to N 2 O 5 will therefore, during periods of significant mixing ratios of N 2 O 5 , over-estimate the concentrations of HNO 3 and overestimate the contribution of HNO 3 to the total nitrate deposition.
The consequences of the detection of N 2 O 5 as an artefact within analytical systems are also dependent on the use to which the data is put, and the characteristics of the dataset.For example if, in high time-resolution data, the diurnal pattern of HNO 3 concentration can be observed, then interpretations and comparison with models should not be affected during daylight hours or periods where NO 3 lifetimes are short.However, when weekly or monthly integrated concentrations of HNO 3 are measured and data is analysed on a seasonal basis, then some periods during the year may be affected to greater extents than others, for example in regions with low BVOC emissions and longer periods of darkness.We are not aware of long-term seasonal datasets reporting the simultaneous contribution of N 2 O 5 and HNO 3 to the total concentration of NO y .It is therefore important that the response of the various nitric acid sampling systems to dinitrogen pentoxide is quantified and published.

Conclusions
Aqueous denuders are very likely sensitive to N 2 O 5 in addition to HNO 3 , and it is likely that alkaline denuders are at least partially sensitive to N 2 O 5 .The main question remaining is to what extent are HNO 3 datasets using these techniques affected.The detection of the sum of N 2 O 5 + HNO 3 as HNO 3 in systems using alkaline or aqueous denuder acid gas removal may result in the misrepresentation of seasonal and geographic pattern of the contribution of HNO 3 to gasphase oxidised nitrogen deposition.The detection of N 2 O 5 is a systematic error and it is possible that in some cases, in high time-resolution datasets, a large proportion of the HNO 3 signal is due to the detection of N 2 O 5 .Data presented here show that a significant portion of nightime HNO 3 measured via an aqueous annular denuder system is N 2 O 5 .The warm-temperature summertime during which these measurements were obtained may be a lower limit of the effect as higher VOCs and warmer temperatures both militate against the formation of N 2 O 5 .The campaign average of nocturnal N 2 O 5 contribution was 17 %, of the HNO 3 signal.Previous modelling work by Riemer et al. (2003) found that the inclusion of N 2 O 5 heterogeneous processes in chemical transport models "causes remarkable changes in the nocturnal concentrations of nitrogen containing species" illustrating the importance of N 2 O 5 in the anthropogenic nitrogen cycle.In addition, the confusion of attempting to compare model output of HNO 3 with measures which are essentially the sum of two separate chemical entities with differing lifetimes and environmental fates may be partly the reason why modelled surface HNO 3 concentrations sometimes fail to agree with measurement data when model performance seems to be good for other forms of reactive N, such as NO 2 and NH 3 (Fowler et al., 2012).The standard network detections systems which use alkali and/or aqueous denuder systems should be tested for the effect of erroneous N 2 O 5 detection, possible in addition to possible artefacts arising from newly observed molecules such as ClNO 2 .
Measurements of N 2 O 5 should be made with a wide seasonal and geographical reach to assess the extent of monitoring bias and also to help validate the difficult task of modelling the contribution of NO 3 and N 2 O 5 to the NO y budget.It is also important that the chloride production channel of heterogeneous N 2 O 5 loss should be correctly included in models assessing reactive N deposition and chemical transport of reactive nitrogen.The uncertainties associated with this process are still relatively large and consequently more work is needed to ascertain its importance, both with respect to the nitrogen cycle, but also for the effect of chlorine activation on atmospheric radical cycling in the troposphere.It is also important that the sensitivity to the N 2 O 5 is determined in all denuder-type systems before datasets of HNO 3 are published.

Fig. 1 .
Fig. 1.Bottom panel: time series of N 2 O 5 measured by CRDS and "HNO 3 " measured by MARGA during August 2012 from the Taunus observatory, Kleiner Feldberg, Germany.Top panel: nocturnal contribution of measured N 2 O 5 to the MARGA HNO 3 signal and the calculated remaining contribution of HNO 3 when data was measured concurrently.There was no N 2 O 5 data available on the night of the 13 August and partial data available on the nights of the 14/22/23 August.

Fig. 2 .
Fig. 2. Expanded time series of the effect of N 2 O 5 on the detection of HNO 3 by an aqueous denuder technique.Left panel: 7 August to 12 August 2012.Right panel: 16 to 17 August 2012.