Airborne measurements of the nitric acid partitioning in persistent contrails

Airborne measurements of the nitric acid partitioning in persistent contrails D. Schäuble, C. Voigt, B. Kärcher, P. Stock, H. Schlager, M. Krämer, C. Schiller, R. Bauer, N. Spelten, M. de Reus, M. Szakáll, S. Borrmann, U. Weers, and Th. Peter Deutsches Zentrum für Luftund Raumfahrt, Institut für Physik der Atmosphäre, Oberpfaffenhofen, Germany Institut für Physik der Atmosphäre, Johannes-Gutenberg Universität Mainz, Mainz, Germany Institut für Stratosphärenforschung, FZ Jülich, Jülich, Germany Max-Planck-Institut für Chemie, Mainz, Germany Institut für Atmosphäre und Klima, ETH Zürich, Zürich, Switzerland Received: 22 May 2009 – Accepted: 14 June 2009 – Published: 30 June 2009 Correspondence to: D. Schäuble (dominik.schaeuble@dlr.de) Published by Copernicus Publications on behalf of the European Geosciences Union.


Introduction
Heterogeneous processes such as the uptake of nitric acid (HNO 3 ) in cirrus clouds influence the ozone budget in the tropopause region.Results from a global chemistry transport model indicate that HNO 3 uptake in cirrus ice particles and subsequent particle sedimentation has the potential to remove HNO 3 irreversibly from this region, leading to a large-scale reduction of gas phase HNO 3 concentrations (von Kuhlmann and Lawrence, 2006).Popp et al. (2004) and Krämer et al. (2008) suggest, based on in situ measurements, that tropical convective cirrus with a high ice water content have the highest potential to vertically redistribute HNO 3 .Voigt et al. (2008) show that tropical cirrus clouds at low temperatures can also serve as nuclei for the formation of nitric acid trihydrate (NAT) particles.Kärcher (2005) demonstrates, based on cloud-resolving simulations, that long-lived Arctic frontal cirrus clouds are capable of denitrifying (and partially dehydrating) upper tropospheric air very efficiently at temperatures < 220 K over vertical regions as large as 3 km.Irreversible removal of gaseous HNO 3 reduces the concentrations of the ozone precursor nitrogen oxide (NO x =NO+NO 2 ) and hence net ozone production rates at typical upper tropospheric NO x levels.Meier and Hendricks (2002) address the sensitivity of this process by means of chemistry box model studies and find reductions in local ozone concentrations of up to 14%.
The interaction of HNO 3 and cirrus ice crystals was investigated experimentally during several field campaigns from the tropics to the Arctic (Weinheimer et al., 1998;Schlager et al., 1999;Kondo et al., 2003;Popp et al., 2004;Ziereis et al., 2004;Voigt et al., 2006).Voigt et al. (2006) summarized these measurements in terms of average nitric acid to water molar ratios (µ=HNO ice 3 /H 2 O ice ) in cirrus D. Schäuble et al.: Nitric acid partitioning in contrails ice particles and ice-bound fractions of total nitric acid (φ =HNO ice 3 /HNO tot 3 ).Kärcher and Voigt (2006) explained the inverse temperature trend of the µ and φ data by means of a model describing nitric acid uptake in growing ice crystals (trapping).Trapping refers to the combined effect of mass accomodation and desorption of molecules at an ice crystal surface that grows from the vapor phase by the deposition of water (H 2 O) molecules, allowing for the burial of adsorbed molecules during surface layer growth.A recent update of the trapping model predicts the steady-state amount of HNO 3 trapped in bulk ice particles for a given ice growth rate, assuming Langmuir-type adsorption isotherms, including the underlying adsorption model asymptotically for nongrowing ice particles (Kärcher et al., 2009).Kärcher et al. (1996) simulate the formation of high levels of HNO 3 in jet aircraft exhaust plumes prior to contrail formation.These early HNO 3 concentrations are determined by the NO x emission index and the engine exit plane concentration of the hydroxyl radical (OH) which acts as the primary oxidant.Arnold et al. (1992) and Tremmel et al. (1998) report the detection of gas phase HNO 3 in young aircraft exhaust plumes.Kärcher (1996) demonstrates by means of microphysical simulations that gaseous HNO 3 is efficiently transferred into plume aerosols and ice particles during contrail formation.Gao et al. (2004) measured HNO 3 in ice particles of a WB-57 aircraft contrail in the subtropics at 14-15 km altitude.They demonstrated that HNO 3 was present in contrail particles at temperatures below 205 K, presumably in the form of NAT, but did not explicitly report ice phase HNO 3 concentrations.An experimental quantification of the HNO 3 content in contrail ice particles as a function of mean ice particle size, a proxy for the microphysical age of persistent contrails, is lacking.
In this study, we present and interpret in situ observations of reactive nitrogen (NO y ), H 2 O, as well as microphysical properties of contrails and thin frontal cirrus layers, performed over Germany, the North Sea, and the Baltic Sea during the CIRRUS-III campaign during 23-29 November 2006.During this campaign, the cirrus layers were probed with the Enviscope Learjet for almost 30 min in the temperature range 210-230 K (24, 28, and 29 November 2006) and 37 contrails were unintentionally sampled for more than 200 s at temperatures from 211 to 220 K (24 and 29 November).Contrail ages of up to about twelve hours were estimated based on the NO y data (Schumann et al., 1998).These observations provide an unprecedented data set on the uptake of HNO 3 in persistent contrails.

Instrumentation
The Learjet performed 5 flights at latitudes between 48 • N and 68 • N with instruments measuring NO y , H 2 O, small ice crystal size distributions, and condensation nuclei (CN).The forward-and backward-facing inlets of the NO y -instrument (Ziereis et al., 2000;Hegglin et al., 2006) operated by DLR and ETH Zurich sampled total (gas phase plus enhanced particulate) and gas phase NO y , respectively.Total reactive nitrogen represents the sum of NO x and other nitrogen-containing species in higher oxidation states, NO y =NO x +HONO+HNO 3 +HO 2 NO 2 +2N 2 O 5 +PAN+... .Two converter channels are located in the inlet system.Inside these gold tubes, heated to 300 • C, NO y is converted to NO using CO as the reducing agent (Fahey et al., 1985).The chemiluminescence of the reaction of NO with O 3 is detected with a photomultiplier at a rate of 1 Hz.Before each flight the conversion efficiency of the converters and the sensitivity of the chemiluminescence detector are determined by adding known amounts of NO 2 and NO, respectively.The instrument background is measured every ten minutes during the flight.Sampling of particulate NO y with the forward-facing inlet depends on the ratio of the flow speed outside and inside the inlet tube as well as on the particle size.To calculate the concentration of particulate (in our case essentially ice phase) reactive nitrogen, NO ice y =(NO forw y -NO back y )/EF NO y , we used the maximum enhancement factor EF max NO y =u 0 /u≈119±12 (u 0 = aircraft true air speed, u = inlet flow speed; numerical values for 24 November) (Belyaev and Levin, 1974) for cirrus observations.This approximation may introduce errors of less than 8% in the EF NO y for observed ice crystal mean diameters (d >14µm).In contrails containing smaller ice particles, we used a size-dependent relationship for EF NO y (d) (Belyaev and Levin, 1974).Aerosol particles smaller than 1µm in diameter were sampled with the forward-as well as the backward-facing inlet.Thus, they do not contribute to the NO ice y signal.Larger NO y -containing aerosol particles, if present, are still likely smaller than most ice particles and are therefore associated with lower enhancement factors (∼63% of EF max NO y for d =3µm), resulting in a minor contribution to the NO ice y signal.The NO y detection limit in particles is 0.8 pmol/mol on 24 and 29 November and 0.6 pmol/mol on 28 November, as derived from the difference of the forward-and backwardfacing channel signals outside clouds divided by the enhancement factor.
Nitric acid was not measured directly during CIRRUS-III, except during the flight on 28 November where we used a nylon filter in the aft-facing inlet to retain HNO 3 and thus derive the gas phase HNO 3 /NO y ratio.As the cirrus clouds on 24 and 29 November were detected slightly below the tropopause (ozone mixing ratios < 90 nmol/mol), we employed a value of 0.45±0.2,corresponding to the mean gas phase HNO 3 /NO y ratio at these ozone levels during the flight on 28 November.Talbot et al. (1999) and Neuman et al. (2001) measured both HNO 3 and NO y obtaining mean gas phase HNO 3 /NO y ratios of 0.35 and 0.2-0.5, respectively.However, they sampled at lower altitudes down to 8 and 7 km, respectively.
In young contrails NO x constitutes the major fraction of NO y .A microphysical plume model study (Meilinger et al., 2005) including detailed heterogeneous aerosol and ice phase chemistry indicates that the chemical conversion into HNO 3 is small in contrails in terms of the gas phase HNO 3 /NO y ratio (≈1%) within the first ten hours after emission due to heterogeneous dehoxification and production of HONO.Entrainment of ambient HNO 3 is supposed to be the dominant process controlling the gas phase HNO 3 concentration in such contrails.Thus, we used the ambient HNO 3 concentration as an approximation for the concentration inside contrails.Further, we assumed that HNO 3 accounts for 100% of NO y in ice particles in cirrus clouds as well as in contrails.N 2 O 5 and HO 2 NO 2 are quickly photolyzed during daytime and the uptake of PAN and HONO on ice crystals is probably small compared to HNO 3 (Bartels et al., 2002).
Total water (H 2 O tot = gas phase plus enhanced particulate water) was measured with the forward-facing inlet of the Fast In situ Stratospheric Hygrometer (FISH) (Zöger et al., 1999) using the Lyman-α photofragment fluorescence technique.Water vapor (H 2 O) was detected with the OJSTER instrument (Open-path Jülich Stratospheric TDL ExpeRiment) (Schiller et al., 2008) by measuring the amount of laser light absorption in gaseous water molecules at a wavelength of 1.37µm.The data of the OJSTER instrument were adjusted to the FISH instrument data in order to reach agreement in observations outside of clouds.On average, the OJSTER data were increased by 15 µmol/mol in cirrus clouds.The ice water content (IWC) was derived from IWC=(H 2 O tot -H 2 O)/EF H 2 O .EF H 2 O was around 6 during CIRRUS-III.According to Schiller et al. (2008) the detection limit of the IWC is temperature dependent and averages to ∼0.1 mg m −3 in the range 210-230 K.The uncertainty in IWC is 15%.
The number concentration and size distribution of particles with diameters 2.8-29.2µm were measured with the Forward Scattering Spectrometer Probe FSSP-100 (Borrmann et al., 2000).The particles were grouped into seven size bins and assumed to be spherical and composed of pure ice.The FSSP-100 data were used in combination with CN and NO y measurements to identify contrails and cirrus (Sect.3), i.e. to decide upon the presence or absence of ice crystals.Therefore, possible uncertainties associated with the FSSP-100 data are not critical for this purpose.We detected cirrus clouds from correlated increases in the FISH and FSSP-100 signals: cirrus clouds were observed either if the IWC was above 0.25 mg m −3 or if 0.1 mg m −3 <IWC< 0.25 mg m −3 and the FSSP-100 particle concentration > 0.

Identification of contrails
Contrails were identified from distinct simultaneous shortterm increases in the time series of the concentrations of NO gas y , NO tot y , CN (> 5 nm diameter, not discussed here), and ice particles.The concurrence of these four signals and their strong increase and decrease within few seconds enabled a separation of contrails from cirrus clouds that were not recently affected by aviation.Lower limits of 0.1 nmol/mol for NO gas y and 100 cm −3 for CN were used as thresholds in this identification.These limits correspond to a contrail age of approximately 12 h assuming homogeneous dilution and sampling through the contrail core (t[s] = (525[nmol/mol]/ NO gas y [nmol/mol]) 1.25 , Anderson et al., 1999).As contrails were not the primary focus of the CIRRUS-III campaign, we consider the spatial sampling to be random in terms of angles relative to the contrail axis and distances from the contrail core.
The amplitudes of the instrument peaks decrease with increasing contrail age and, for a given age, with distance from the contrail core.Contrails often exhibit significant spatial inhomogeneities in structure and dilution properties.The detection of contrails older than 12 h as well as the merging of several individual contrails may result in an extended contrail detection signal.Together, this introduces considerable uncertainty in the exact determination of contrail age from plume dilution data which is difficult to quantify.
We regard the probed cirrus clouds as upper tropospheric ice clouds that were not influenced by aircraft emissions within the last 12 h of their lifetime.It cannot be ruled out that part of the cirrus were actually older (age > 12 h) contrails.That being said, most of the observed contrails were embedded in thin cirrus clouds.In 79% of all cases either the FSSP or the FISH signal are > 0 before and after the contrail encounter which suggests the presence of ice crystals.53% of the contrails were located in cirrus that meet the robust cirrus criterion mentioned in Sect. 2. This implies that for statistical reasons our data set does not allow us to compare properties of isolated contrails with those of contrails in cirrus clouds.

Nitric acid partitioning and ice water content in contrails
European Centre for Medium-Range Weather Forecasts analyses indicate that the upper regions of frontal cirrus layers were probed on 24 and 29 November, whereas on 28 November the observations originate from deeper inside cirrus clouds.On 24 November large parts of central Europe were covered by the cirrus.The cirrus layers were much more stretched out with a large North-South extension on 28 and 29 November.As an example of our suite of data, Fig. 1 shows a typical time series of data from the flight on 24 November over Germany.In this and the following section, we further present model results providing explanations of the observed uptake of HNO 3 in contrails and the cirrus layers.
Figure 2 (top panel) illustrates the decrease of the ice phase fraction of total nitric acid (φ) with increasing temperature.This dependence agrees with previous observations but almost all of the observed φ values in cirrus clouds (grey circles) are below the mean of previous measurements (Kärcher and Voigt, 2006).This is probably due to the relatively low IWCs measured during CIRRUS-III (Fig. 2, bottom panel) and the positive correlation between φ and IWC (Krämer et al., 2008).Further, the HNO 3 gas phase partial pressures were higher (4.7×10 −8 ±1.3×10 −8 hPa∼0.2nmol/mol) than the average of the other campaigns (2×10 −8 hPa) (Kärcher and Voigt, 2006).
The uncertainty of the gas phase HNO 3 /NO y ratio (Sect.2) introduces an uncertainty in φ (time resolution 1 s) of about a factor of 2.
On average, contrails were found to have larger φ values than the cirrus layers under similar conditions (black and red squares, respectively).This is further illustrated with the help of probability distributions of φ displayed in Fig. 3a.At temperatures between 211 and 220 K, contrail ice particles contained on average 6% (cirrus 3%) of the total nitric acid.Twelve young contrails with NO gas y > 0.75 nmol/mol, corresponding to ages < 1 h, had even larger ice phase nitric acid fractions (9%).The maximum φ measured in contrails was 22%.
The gas phase equivalent mixing ratio of nitric acid in ice, HNO ice 3 , is 14 pmol/mol in contrails compared to 6 pmol/mol in the cirrus layers.The mean HNO ice 3 in the young contrails is 21 pmol/mol.Our finding that the absolute amount of HNO ice 3 in contrails is larger than in cirrus could be linked to differences in the ice formation process (Sect.5).
Figure 2 (middle panel) shows the temperature dependence of the ice phase molar ratios of nitric acid and water (µ) in contrails (red circles) and in the surrounding cirrus layers (grey circles) that were not recently influenced by aviation.The µ values generally increase with decreasing temperature, as the probability of HNO 3 molecules to escape from the growing ice surfaces after adsorption is reduced at low temperatures leading to more efficient trapping of HNO 3 (Kärcher and Voigt, 2006).
The curves are results from a recent update of the trapping model now accounting for possible surface saturation upon adsorption at high (> 2×10 −8 hPa) HNO 3 partial pressures (Kärcher et al., 2009), causing a weaker temperature dependence of µ than estimated by the previous model (Kärcher and Voigt, 2006).The dashed and solid model curves were computed for HNO 3 partial pressures of 3×10 −8 hPa and 6×10 −8 hPa, respectively, roughly capturing the range of measured values occurring in contrails and cirrus (see above).The trapping model bounds the observed mean µ values in cirrus (black squares) very well at 211-220 K.The model assumes temperature-dependent mean ice particle growth rates (net supersaturations) to estimate steady-state molar ratios.Deviations of individual data points from the mean model curves are likely caused by variable growth/sublimation histories of the observed ice particles.Further, even in cirrus the µ of small ice crystals may be strongly influenced by the composition of the freezing aerosol (Sect.5).As for the data points above 220 K, the low µ suggests that these ice particles have grown at low HNO 3 partial pressures.The relatively high IWC may imply that a considerable number of large ice crystals have sedimented from their formation region depleted in HNO 3 into the measurement region.Despite these non-local effects the data are still captured by the lower limit of trapping estimated in Kärcher and Voigt (2006).
The mean HNO 3 /H 2 O molar ratio in contrails of 4×10 −6 is found to be approximately twice as large as in the cirrus layers (2×10 −6 ), which is illustrated by means of the probability distribution of µ (Fig. 3b) .The increase in mean µ is mainly caused by the data points below 215 K, compare red and black squares in Fig. 2 (middle panel).This difference is roughly consistent with the difference in the mean HNO ice 3 per particle (see above).
The IWC of contrails and cirrus clouds versus temperature is shown in Fig. 2 (bottom panel).Compared to the comprehensive compilation of IWCs by Schiller et al. (2008), based on a large number of in situ measurements at midlatitudes (mean values indicated by the solid curve), the IWC observations on 24 and 29 November below 220 K appear to be slightly lower for the following reasons.First, thin ice clouds with IWCs as low as 0.1 mg m −3 were included in our data set derived from simultaneous FSSP and total water increases.In addition, in the upper region of frontal cirrus layers, where the data were taken, ice nucleation is supposed to take place and sedimentation of large ice particles out of these layers keeps the IWC small.Finally, there may be entrainment of dry air from above and a reduced IWC arises while crossing cloud free patches within the 200 m (1 s) measurement interval.The solid curve marks the average values obtained by probing the whole vertical extension of a population of cirrus clouds (Schiller et al., 2008) and is therefore located above our mean IWCs (black squares) below 220 K. Above 220 K our IWC data were taken deeper inside a cirrus cloud and the respective means are therefore in better agreement with the solid curve.The fact that the trapping model seems to overpredict µ at higher temperatures (curves not shown in Fig. 2, middle panel) may be tied to the impact of sedimentation and more strongly varying ambient conditions throughout the probed cirrus clouds, which is much less pronounced in the upper layers probed at lower temperatures.
The contrails have a slightly larger mean IWC (red squares) than the cirrus clouds, 1.8 and 1.3 mg m −3 , respectively.The IWC of young contrails 2.9 mg m −3 is more than twice the cirrus mean.According to plume dilution estimates derived from large-eddy simulations of aircraft wake vortex development (Gerz et al., 1998), the H 2 O aircraft emissions become unimportant for contrails older than a few minutes.So most of the H 2 O in contrail ice condenses from the ambient air during ice particle growth and differences between contrail and cirrus IWCs are expected to diminish over time.3 /HNO tot 3 ), (middle) molar ratio of nitric acid to water in ice particles (µ=HNO ice 3 /H 2 O ice ), and (bottom) ice water content (IWC) of contrails (red circles) and cirrus clouds (grey circles).Squares are means over 2 K temperature bins.For µ and IWC in contrails the warmest bin is centered at 217.3 K (instead of 217 K) to comprise the measurements at temperatures slightly higher than 218 K. Curves in the middle panel are results from the trapping model evaluated at HNO 3 partial pressures of 3×10 −8 hPa (dashed curve) and 6×10 −8 hPa (solid curve) at temperatures where contrails were detected (Kärcher et al., 2009).The curve in the bottom panel depicts the mean cirrus IWCs based on a large number of in situ measurements at midlatitudes (Schiller et al., 2008).
We additionally report a striking difference between measured contrail and cirrus cloud ice particle number densities in the FSSP-100 size range 2.8-29.2µm in diameter, 1.5 cm −3 and 0.2 cm −3 , respectively.We are aware of the fact that shattering of large (d >50µm) ice crystals at the front end of the FSSP instrument may artificially enhance the measured number concentrations (Heymsfield, 2007;De Reus et al., 2009).However, shattering is probably not affecting the above values to a great extent, as the concentrations of such large particles in the contrails and www.atmos-chem-phys.net/9/8189/2009/Atmos.Chem.Phys., 9, 8189-8197, 2009 in the nucleation layers of cirrus are presumably very small (Schröder et al., 2000).More robust conclusions would be possible when using cloud probes unaffected by shattering and extending the size distribution measurements to diameters > 30 µm.

Nitric acid uptake in developing contrails
Given the limited amount of contrail data, it is unclear whether the systematic differences in ice phase fraction and molar ratios in ice presented in Sect. 4 are representative.Nevertheless, sorting the molar ratio data as a function of the mean ice particle size enables us to study the HNO 3 uptake process in contrails in more detail.
Figure 4 shows the molar ratios (grey circles) along with the mean (black squares) and median (grey squares) values of the contrails probed during CIRRUS-III as a function of the mean ice particle diameter.We associate rough estimates of the contrail age with the mean diameter, as indicated in the figure.The contrail age is calculated according to Sect. 3 and linked to the mean ice crystal diameter with a linear regression for contrails younger than 1 h.In persistent contrails, the mean size increases due primarily to uptake of H 2 O from the gas phase (depositional growth).The measurements show a clear trend of decreasing µ with increasing mean size or age.Young contrails with ages 10-15 min and mean diameters ∼7 µm have a mean µ≈10 −5 , while older contrails with ages > 1 h and mean diameters > 10-15 µm exhibit values closer to the mean ∼2×10 −6 of the cirrus data.
Using campaign-average contrail temperature (216 K) and HNO 3 partial pressure (∼5×10 −8 hPa), the updated trapping model (Kärcher et al., 2009) predicts an average molar ratio µ ∞ ≈2×10 −6 (dash-dotted curve), consistent with the cirrus observations (middle panel in Fig. 2 and Fig. 3b) and the older contrail data in Fig. 4. Increasing the partial pressure would not significantly increase µ because the adsorption of HNO 3 becomes limited by a monolayer surface coverage during trapping.Hence, trapping of low levels of entrained ambient HNO 3 by the growing contrail ice particles is not capable of explaining the high molar ratios exceeding 10 −5 in young contrails.
How are the high molar ratios brought about?We argue that a high concentration of HNO 3 had already entered the ice particles during contrail formation, when they are very small (mean diameters ∼1 µm), leading to very high molar ratios.Subsequent depositional growth increases the size per ice particle, while trapping in young contrails (mean diameters 1-10 µm) only adds a small contribution to the HNO 3 content per particle; both effects cause µ to decrease in this phase of contrail development.Further growth to larger sizes diminishes the role of the initially high molar ratios and trapping takes over the dominant part in determining µ.These processes are illustrated by the model curves as explained below.
Before we quantify the modeled µ dependence shown in Fig. 4 in more detail, we note that similar arguments were used to explain the in situ observations of high molar ratios of ∼5×10 −5 in nascent cirrus ice particles forming and growing at temperatures near 202 K and HNO 3 partial pressures similar to those encountered during CIRRUS-III (Voigt et al., 2007).In this case, it was proposed that sulfate aerosol particles dissolve large amounts of gaseous HNO 3 prior to freezing.Assuming that about half of the dissolved HNO 3 remained in the ice particles after freezing, this led to initial molar ratios ∼0.01.We suggest that the same processes act in contrails.
Freezing aerosol surface areas, ice supersaturations, and HNO 3 partial pressures are higher during contrail formation than during cirrus formation.Contrail ice particles form from plume particles that efficiently dissolve exhaust HNO 3 prior to ice formation (Kärcher, 1996).We estimate the nitric acid to water molar ratio to be µ 0 ≈0.03 in the freezing plume aerosol particles as in Voigt et al. (2007), assuming an equilibrium composition of aqueous sulfuric acid droplets (diameter d 0 =0.75µm) at 215 K and an HNO 3 partial pressure 2×10 −6 hPa (corresponding to 10 nmol/mol) consistent with plume chemistry simulations (Kärcher et al., 1996).This value roughly determines the initial molar ratio of the nascent contrail ice particles.The resulting evolution of µ in the contrail as a function of the mean ice particle diameter d (solid curve in Fig. 4) is given by (Kärcher and Voigt, 2006;Voigt et al., 2007): with the ratio of ice and freezing aerosol particle bulk mass densities κ, the initial ice particle diameter d 0 , and the trapped molar ratio µ ∞ as derived above.The ice particle dilution factor N quickly takes values 1 due to depositional growth.The first part of this relationship describes the initial decrease of µ∝1/d 3 by uptake of H 2 O (dashed curve); it becomes smaller than the second (trapping) contribution for d >10µm, at which point the trapped amount (dash-dotted curve) is approached, µ→µ ∞ .Except during the very short initial growth phase after ice nucleation, cirrus particles are usually large enough to render the aerosol component in Eq. (1) unimportant relative to the trapping component (Kärcher and Voigt, 2006).As shown here, the aerosol component is more prominent in persistent contrails, because mean contrail ice particle diameters stay relatively small owing to the large number density of ice particles compared to cirrus.Given the measurement uncer-tainties, the simple uptake model explains the observed trend in the mean values µ(d) in developing contrails fairly well.Variability in initial and ambient conditions may explain deviations of individual data points from the mean model curve.A set of data points between 10-20 µm following the median values seems to be well fitted by the dashed curve without trapping.We speculate that those measurements might be more representative of supersaturated inner contrail regions in which the local HNO 3 gas phase mixing ratio was small due to inefficient mixing from ambient air.

Summary and conclusions
During the CIRRUS-III field campaign gas phase and ice phase reactive nitrogen, ice water content, and ice crystal size distributions were measured in contrails and cirrus at midlatitudes close to the tropopause.The temperatures and HNO 3 partial pressures were in the ranges 210-230 K and 3-6×10 −8 hPa, respectively.The observed uptake of HNO 3 in ice particles residing in the upper layers of frontal cirrus clouds confirms previous results from airborne field campaigns carried out in polar, midlatitude, and sub-tropical regions (Voigt et al., 2006).
On average the probed contrails contained twice as much ice-bound HNO 3 as the cirrus clouds within 211-220 K, 14 pmol/mol and 6 pmol/mol, respectively.Thus, the mean fraction of total HNO 3 in ice particles was considerably larger in the contrails (6%) than in the cirrus layers (3%).In young contrails (approximate age < 1 h) this fraction was even higher (9%).The measured molar ratios of HNO 3 and H 2 O in contrail ice particles exceeded 10 −5 for small particle sizes, or contrail ages.For older contrails, molar ratios approached the mean value of 2×10 −6 detected in the cirrus layers.Averaged over all detected contrails regardless of age, this caused the mean molar ratios in contrails to be about twice as large as in the cirrus clouds.
Motivated by our study, contrails may be regarded as an atmospheric laboratory to study HNO 3 uptake during ice particle growth.Our data show that ice phase HNO 3 /H 2 O molar ratios decrease during contrail ageing.This dependence was explained by uptake of high levels of HNO 3 into the freezing aerosol particles during ice formation in contrails and subsequent trapping of relatively low levels of ambient HNO 3 in growing contrail ice particles.In young contrails with ages < 1 h or mean diameters < 10 µm, the ice phase HNO 3 concentrations are therefore largely controlled by the jet engine NO x and OH emission indices.More airborne measurements with extended instrumentation are needed to study the dependence of HNO 3 uptake on HNO 3 partial pressure and to better quantify ice particle size distributions in developing contrails.
The results of this study help constrain chemicalmicrophysical models simulating heterogeneous chemistry in persistent contrails in order to constrain the impact of www.atmos-chem-phys.net/9/8189/2009/Atmos.Chem.Phys., 9, 8189-8197, 2009 plume processing of NO x emissions on the chemical production or loss of ozone.Chemistry transport model simulations suggest that net upper tropospheric ozone production rates may decrease by ∼15-18% in the main midlatitude Northern Hemisphere traffic areas due to decreased aircraft-induced NO x perturbations when chemical plume processing effects are accounted for (Kraabøl et al., 2002).Uptake of HNO 3 in contrail ice particles may further reduce ozone production by passivating NO x .

Fig. 2 .
Fig. 2.The temperature dependence of the (top) ice phase fraction of total nitric acid (φ =HNO ice 3 /HNO tot 3 ), (middle) molar ratio of nitric acid to water in ice particles (µ=HNO ice 3 /H 2 O ice ), and (bottom) ice water content (IWC) of contrails (red circles) and cirrus clouds (grey circles).Squares are means over 2 K temperature bins.For µ and IWC in contrails the warmest bin is centered at 217.3 K (instead of 217 K) to comprise the measurements at temperatures slightly higher than 218 K. Curves in the middle panel are results from the trapping model evaluated at HNO 3 partial pressures of 3×10 −8 hPa (dashed curve) and 6×10 −8 hPa (solid curve) at temperatures where contrails were detected(Kärcher et al., 2009).The curve in the bottom panel depicts the mean cirrus IWCs based on a large number of in situ measurements at midlatitudes(Schiller et al., 2008).

Fig. 3 .
Fig. 3. Normalized probability density functions of the ice phase fraction of total HNO 3 φ (a) and the HNO 3 /H 2 O molar ratio in ice µ (b) for contrails (red) and cirrus clouds (transparent).Arrows indicate the means for the temperature range (211-220 K) where contrails occurred.

Fig. 4 .
Fig.4.Molar ratios of nitric acid to water in contrail ice particles µ versus mean ice particle diameter.Black (grey) squares represent means (medians) over 3 µm size bins.For illustration, four data points are displayed with estimated measurement errors.Diameters > 20 µm are uncertain due to the large-size cut-off of the FSSP-100.Approximate contrail ages are inferred from the gas phase NO y data.Curves are model results; solid curve according to Eq. (1), dash and dash-dotted curves indicate limiting cases; see text for details.
Time series of gas phase and total NO y (gas phase plus enhanced ice phase), ice phase HNO 3 volume mixing ratios, ice water content (IWC), number densities of ice particles with approximate diameters between 2.8 and 29.2 µm, as well as altitude and temperature for the flight on 24 November 2006.Gaps in the NO y measurements are due to calibration procedures.