Volume 111, Issue D21
Aerosol and Clouds
Free Access

Nitric acid condensation on ice: 1. Non-HNO3 constituent of NOY condensing cirrus particles on upper tropospheric

B. Gamblin

B. Gamblin

Laboratory for Atmospheric and Space Physics, University of Colorado, Boulder, Colorado, USA

Department of Chemistry and Biochemistry, University of Colorado, Boulder, Colorado, USA

Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, Colorado, USA

Search for more papers by this author
O. B. Toon

O. B. Toon

Laboratory for Atmospheric and Space Physics, University of Colorado, Boulder, Colorado, USA

Department of Atmospheric and Oceanic Sciences, University of Colorado, Boulder, Colorado, USA

Search for more papers by this author
M. A. Tolbert

M. A. Tolbert

Department of Chemistry and Biochemistry, University of Colorado, Boulder, Colorado, USA

Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, Colorado, USA

Search for more papers by this author
Y. Kondo

Y. Kondo

Research Center for Advanced Science and Technology, University of Tokyo, Tokyo, Japan

Search for more papers by this author
N. Takegawa

N. Takegawa

Research Center for Advanced Science and Technology, University of Tokyo, Tokyo, Japan

Search for more papers by this author
H. Irie

H. Irie

Frontier Research Center for Global Change, Japan Agency for Marine-Earth Science and Technology, Atmospheric Composition Research Program, Yokohama, Japan

Search for more papers by this author
M. Koike

M. Koike

Department of Earth and Planetary Physics, University of Tokyo, Tokyo, Japan

Search for more papers by this author
J. O. Ballenthin

J. O. Ballenthin

Air Force Research Laboratory, Space Vehicles Directorate, Hanscom Air Force Base, Massachusetts, USA

Search for more papers by this author
D. E. Hunton

D. E. Hunton

Air Force Research Laboratory, Space Vehicles Directorate, Hanscom Air Force Base, Massachusetts, USA

Search for more papers by this author
T. M. Miller

T. M. Miller

Air Force Research Laboratory, Space Vehicles Directorate, Hanscom Air Force Base, Massachusetts, USA

Search for more papers by this author
A. A. Viggiano

A. A. Viggiano

Air Force Research Laboratory, Space Vehicles Directorate, Hanscom Air Force Base, Massachusetts, USA

Search for more papers by this author
B. E. Anderson

B. E. Anderson

NASA Langley Research Center, Hampton, Virginia, USA

Search for more papers by this author
M. Avery

M. Avery

NASA Langley Research Center, Hampton, Virginia, USA

Search for more papers by this author
G. W. Sachse

G. W. Sachse

NASA Langley Research Center, Hampton, Virginia, USA

Search for more papers by this author
J. R. Podolske

J. R. Podolske

NASA Ames Research Center, Moffett Field, California, USA

Search for more papers by this author
K. Guenther

K. Guenther

NASA Dryden Flight Research Center, Edwards, California, USA

Search for more papers by this author
C. Sorenson

C. Sorenson

NASA Dryden Flight Research Center, Edwards, California, USA

Search for more papers by this author
M. J. Mahoney

M. J. Mahoney

Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California, USA

Search for more papers by this author
First published: 08 November 2006
Citations: 3

This is a companion to DOI:10.1029/2005JD006049.

Abstract

[1] Measurements of NOY condensation on cirrus particles during the SOLVE-I field campaign are analyzed and segregated based on altitude. Significant amounts of NOY were found on the upper tropospheric ice particles; therefore condensation on ice appears to be an important method of NOY removal from the gas phase at the low temperatures of the Scandinavian upper troposphere. For the data set collected on 23 January 2000, NOY condensation on cirrus particles has different properties depending on whether the ice particles are sampled in the upper troposphere, where HNO3 does not dominate NOY, or in the lower stratosphere, where HNO3 does dominate NOY. Nitric acid becomes enriched in the gas phase as NOY condenses on upper tropospheric ice crystals, indicating that a non-HNO3 component of NOY is condensing on upper tropospheric ice particles much faster and at higher concentrations than HNO3 alone on this day. It is unclear which non-HNO3 constituent of NOY is condensing on upper tropospheric ice particles, although N2O5 is the most likely species. This condensation of a non-HNO3 component of NOY is not universal in the upper troposphere but depends on the conditions of the air parcel in which sampling occurred, notably exposure to sunlight.

1. Introduction

[2] In the upper troposphere, global chemistry transport models underpredict the concentration of NOY by 30–80% when compared to the observed mean [Thakur et al., 1999], while they overpredict the concentration of gas phase nitric acid by up to 400% when compared to observed data [Jacob et al., 1996; Thakur et al., 1999]. This discrepancy between theory and observations suggests an additional loss mechanism for nitric acid must be considered. Lawrence and Crutzen [1998] suggested condensation of HNO3 on settling ice particles is a significant loss mechanism for nitric acid in the upper troposphere. When sedimentation of these cirrus particles occurs, the atmosphere within a cloud can become denitrified. This loss could account for some of the excess nitric acid in modeling studies that do not include HNO3 removal on ice clouds. Removal of nitric acid and other NOY species (NO3, 2N2O5, PAN, HONO, HO2NO2, ClONO2, etc.) from upper tropospheric altitudes affects the NOX (NO + NO2) budget, which in turn affects ozone production. Since many NOY species are interconvertible, removal of one species would suppress production of other NOY species. As NOY containing particles reach warmer, lower altitudes, the particles can evaporate, releasing the NOY and hence increasing NOY at these lower altitudes.

[3] Weinheimer et al. [1998] present observations of NOY on ice in a wave cloud at temperatures from 205–220 K. Their calculations show that all of the gaseous HNO3 (assumed to be 15% of NOY) was depleted by condensation on ice in a relatively large surface area wave cloud in under 2 min. By contrast, a study by Meilinger et al. [1999] found very little HNO3 present on a cold (T = 196 K) ice cloud. Kondo et al. [2003] found NOY does not condense appreciably on warmer temperature cirrus ice clouds (T > 215 K), while it does condense on colder clouds (T < 215 K). Similarly, Ziereis et al. [2004] observed coverages greater than one percent of a monolayer for temperatures below 217 K, while the observed coverages were substantially reduced at warmer temperatures.

[4] Tabazadeh et al. [1999] note that atmospheric HNO3 vapor pressures are an order of magnitude less than HNO3 pressures used in laboratory measurements of nitric acid adsorption on ice by Abbatt [1997] and Zondlo et al. [1997]. Using extrapolations of these laboratory data to such lower HNO3 pressures led Tabazadeh et al. [1999] to suggest that little HNO3 would condense on ice clouds. Additionally, Tabazadeh et al. [1999] suggested that most cirrus particles do not precipitate over a great depth of the atmosphere so that nitric acid would be released very near the point of initial absorption. To explain the relatively low observed abundance of HNO3, alternative nitric acid loss mechanisms have been proposed, including HNO3 condensing on organic aerosols, reacting on mineral aerosols or forming liquid ternary aerosols [Dentener et al., 1996; Tabazadeh et al., 1998; Irie et al., 2004].

[5] Kondo et al. [2003] examined SOLVE-I data, collected above the Arctic Circle during the winter of 1999–2000, and demonstrated that a substantial fraction of total NOY can be present on ice crystals. We have reexamined these data and find that the behavior of NOY condensation on ice is different for ozone concentrations less than 100 ppbv than for [O3] > 100 ppbv. [O3] is a surrogate altitude scale. We label the data sampled at [O3] < 100 ppbv as “lower altitude” or “upper tropospheric” data, while the data taken at [O3] > 100 ppbv is referred to as “higher altitude” or “lower stratospheric” data. Generally, “lower altitude” also means that NOYcontains only a small fraction of HNO3 while “higher altitude” means that NOY is dominated by HNO3. It is found that at lower altitudes ([O3] < 100 ppbv) substantial amounts of NOY occur on the ice particles. However, gas phase nitric acid abundance is not correlated on all days with particulate NOY abundance. We demonstrate that at upper tropospheric altitudes, gas phase HNO3 becomes relatively enriched in regions where clouds are present on some days, suggesting an alternative component of NOY is condensing on tropospheric cirrus ice particles. Subsequently, we discuss what other NOY components could be condensing on upper tropospheric ice particles and the implications of this theory for the atmospheric budget of NOY.

2. Data Analysis

[6] HNO3 and NOY concentrations were measured in situ on NASA's DC-8 aircraft flown in the upper troposphere and lower stratosphere during the SOLVE-I (SAGE III Ozone Loss and Validation Experiment I) [Newman et al., 2002] campaign. These data allow us to study nitric acid condensation on cirrus ice clouds in the upper troposphere. A chemiluminesence detector was used to obtain gas phase and total NOY data [Kondo et al., 1997, 2003]. The Kondo et al. [1997, 2003] chemiluminesence detector was estimated to have a precision of 4 parts per trillion by volume (pptv) for an NOY value of 1000 pptv and an absolute accuracy of 10% for the given NOY values [Kondo et al., 2003]. Particulate NOY (PNOY) was obtained by subtracting gas phase NOY from total NOY and then dividing this value by the instrument enhancement factor, where the enhancement factor varied by only 26% over the entire flight (from approximately 110 to 150). Data from a Chemical Ionization Mass Spectrometer (CIMS) [Miller et al., 2000; Ballenthin et al., 2003], were used to obtain gas-phase nitric acid concentrations. The CIMS instrument has an accuracy of ±30% for stratospheric HNO3 mixing ratios and a precision of ±10–15% [Ballenthin et al., 2003]. A Forward Scattering Spectrometer Probe (FSSP-300) [Baumgardner et al., 1992], which measures particle size diameters less than 20 μm, was used to obtain aerosol surface area density. Information on particle size distribution and condensed water content for cirrus events during the SOLVE-I mission can be found in Hallar et al. [2004]. In situ ozone concentrations were determined using a nitric oxide chemiluminescence technique (see http://cloud1.arc.nasa.gov/solve and http://cloud1.arc.nasa.gov/solveII/instrument_files/O3.pdf). An external path diode laser hygrometer (DLH) was used to obtain water vapor concentrations [Diskin et al., 2002; Podolske et al., 2003] and a Microwave Temperature Profiler [MTP] was used to compare flight altitude and tropopause altitude [Denning et al., 1989]. The DC-8's Data Archive and Distribution System (DADS) meteorological data were used for ambient temperature and pressure. Temperature calibrations of the MTP using radiosondes near the DC-8 flight track suggest the DADS outside air temperature was 0.8 K too warm (M. J. Mahoney, personal communication, 2004). Therefore 0.8 K was subtracted from temperature data in our analysis. The MTP data have a 15 s observing cycle, whereas all other data sets have a 1 Hz response time.

[7] When choosing data to study from this field campaign, we looked for segments of each flight where the aircraft was at cruising altitude and sampling cirrus cloud particles. To this end, we did not analyze data collected during the ascent or descent of NASA's DC-8 aircraft. We analyzed several dates during the SOLVE-I mission but report predominantly on data from 23 January 2000, as data from this date exhibit behavior in the upper troposphere that is unexpected. Additionally, the only nitrogen containing species measured during this field campaign were HNO3, NOY and NOX. Therefore, when we attempt to understand the behavior of other NOY species not measured, we accomplish this by subtracting the gas-phase HNO3 concentration and the gas-phase NOX concentration from the gas-phase NOY concentration to obtain the bulk concentration of gas-phase NOY species other than HNO3 or NOX.

[8] Figure 1 is an overlay of the DC-8 flight track on an AVHRR infrared satellite image from the NOAA 14 satellite showing cloud cover and cloud temperature (K) (http://cloud1.arc.nasa.gov/solve/). This image depicts the extensive horizontal cloud coverage over Scandinavia and Greenland on 23 January 2000. Along the flight path, the DC-8 sampled cloudy regions from 60 km wide to hundreds of km wide. The cloud temperatures shown in Figure 1 are warmer than the temperatures sampled by NASA's DC-8 aircraft. This is likely due to the cloud tops being optically thin such that the temperatures measured were located deeper into the cloud layer.

Details are in the caption following the image
Flight track of the DC-8 on 23 January 2000 overlaid on an advanced very high resolution radiometer (AVHRR) infrared image from the NOAA 14 satellite showing cloud cover and cloud temperature (K).

[9] During the flight on this date, the pilots of the DC-8 were intentionally trying to cruise as high as possible to avoid flying in cloud. However, cloud tops often slightly exceeded their maximum flight altitude. This is can be seen in Figures 2 and 3, showing the nadir and zenith profile respectively of the aerosol scattering ratio collected from NASA's Lidar Atmospheric Sensing Experiment (LASE) system on board the DC-8 [Browell et al., 1997]. The aircraft altitude (km) and the time (UT, hours) at the flight location are also shown in Figures 2 and 3. We will be examining cirrus microphysical data from these cloud tops. Figure 2 illustrates the vertical extent of these cloud features, where several clouds are at least 4–5 km thick.

Details are in the caption following the image
Nadir profile of aerosol backscatter measured by Lidar Atmospheric Sensing Experiment (LASE) over the 23 January 2000 DC-8 flight during the SOLVE-I mission (http://asd-www.larc.nasa.gov/lidar/sol/data/slv18n_asr.gif).
Details are in the caption following the image
Zenith profile of aerosol backscatter measured by LASE over the 23 January 2000 DC-8 flight during the SOLVE-I mission http://asd-www.larc.nasa.gov/lidar/sol/data/slv18z_asr.gif).

[10] Figure 3 illustrates that clouds only extend slightly above the aircraft in a few regions, thus no significant sedimentation or particle redistribution is likely to have occurred to bring NOY containing particles from above the aircraft down to the altitude where the aircraft was sampling. There are some PSCs near 24 km, over Scandinavia, but there is no evidence that precipitation is occurring from them. Therefore this analysis assumes any particles sampled at the time of flight generally existed in the cloud near the sampling altitude throughout their lifetime.

[11] Figure 4 shows the aircraft altitude relative to the tropopause altitude, ozone concentration, surface area density, gas phase NOY, gas phase NOX and particulate NOY observed during the data analysis period. Figure 4a shows the data were collected near the local tropopause, where the dashed line indicates a difference of zero between the pressure altitude and the tropopause altitude. Figure 4b shows ozone concentration (ppbv) as a function of time. The dashed line in Figure 4b indicates an ozone value of 100 ppbv.

Details are in the caption following the image
(a) Difference between the pressure altitude and the tropopause altitude (km) of NASA's DC-8 aircraft on 23 January 2000 versus universal time (s). The horizontal, dashed line indicates where this difference is zero. Also included are time series plots of (b) O3 (ppbv), (c) measured surface area density (μm2/cm3), (d) NOY (pptv), (e) NOX (pptv), and (f) PNOY (pptv). The horizontal, dashed line in Figure 4b indicates where the ozone concentration is 100 ppbv. The data in Figures 4b–4f indicate data that were analyzed for this study. In this data set we removed data sampled during the DC-8's ascent or descent and in conditions where NAT particles would be stable. In addition, we removed data where the surface area density was less than 10 μm2/cm3.

[12] We wished to study the behavior of nitric acid in cirrus clouds; therefore we required a method to determine what data points came from a cirrus event. To ensure we examined only data obtained from a cirrus cloud, we binned the data by 10 s time segments and excluded any time segments containing data points where the surface area density was less than 10 μm2/cm3. The aircraft was skimming in and out of the top most layer of clouds (Figures 2 and 3 and personal observation), hence the surface area was highly variable. If a data point with a large surface area is near one where the surface area is too low (< 10 μm2/cm3), the data could have been collected when the aircraft was entering or exiting a cloud and hence any surrounding data would be suspect. Thus that time segment (the instrument lag time) was not analyzed. Finally, we removed any nitric acid trihydrate (NAT) particles by excluding from the data set any instances where the CIMS HNO3 pressure was greater than the NAT equilibrium pressure.

[13] It is important to note, however, that data from all cirrus events during the SOLVE mission cannot be grouped together, because different processes occur depending on the composition of the atmosphere and the meteorological conditions at the time of cloud formation.

[14] After excluding data as described above, we obtained what we refer to in this paper as the “data set,” indicated by the data points shown in Figures 4b–4f for 23 January 2000. Our data set is organized chronologically with Universal Time (UTC); during each 10 s interval we have a combination of sampled data and results calculated from this sampled data. For example, on 23 January 2000, at 51200 s, we have one value for the ambient pressure (mbar), temperature (K), nitric acid pressure (pptv), etc., as well as calculated values obtained from the sampled data, such as surface coverage (molecules/cm2) and particulate NOY (molecules/cm3).

3. NOY Surface Coverage

[15] Figure 5 shows plots of NOY surface coverage as a function of HNO3 gas pressure in Torr. The data have been segregated by ozone concentration (a surrogate altitude scale), where data collected at ozone levels above 100 ppbv are considered to be from a more stratospherically influenced air parcel (Figure 5a) and data sampled at [O3] < 100 ppbv are assumed to be from the upper troposphere. The value of 100 ppbv O3 is often used to indicate the tropopause region because the slope of [O3] as a function of altitude changes dramatically at this value because of ozone's stratospheric source [e.g., Pan et al., 2004].

Details are in the caption following the image
NOY surface coverage versus HNO3 gas pressure (Torr). (a) Only the [O3] > 100 ppbv SOLVE data or data sampled in what is considered to be stratospherically influenced air. (b) Data sampled at ozone levels below 100 ppbv which are considered to be from a more tropospheric-like air parcel. SOLVE data of NOY surface coverage are shown as differently colored “plus” symbols, where the blue data points represent the coldest temperature field data (197–199 K), the green points indicate data collected at temperatures of 199–201 K, and the yellow points represent temperatures of 201–203 K. The warmest data (203–205 K) are colored in red. In Figure 5 the surface area used in the calculation of surface coverage has been increased by a factor of 2 relative to the measurements.

[16] Surface coverage was obtained for the data set by first subtracting the Kondo et al. [2003] gas-phase NOY data from their total NOY data, to obtain particulate NOY. Each particulate NOY value was then divided by the corresponding FSSP-300 surface area and the NOY instrument enhancement factor at that particular time, to obtain a value for surface coverage in units of molecules/cm2. These lower altitude SOLVE surface coverages are organized into four different temperature bins, as denoted by the differently colored data points, which correspond in color to the “Field Data” legend on the right side of the bottom panel.

[17] The FSSP-300 is known to underrepresent cirrus cloud surface areas [Heymsfield et al., 1990], because its size range is limited to particles smaller than 20 μm. Kondo et al. [2003] conducted a preliminary analysis of the surface area measurement error by estimating the surface area for any particles greater than 20 μm using the number density measured by the FSSP-300 and the total volume of particulate water obtained from the University of Colorado closed-path laser hygrometer (CLH) [Hallar et al., 2004] assuming a lognormal size distribution. In this study, Kondo et al. [2003] suggest the total surface areas measured by the FSSP-300 were low by at least a factor of 2 and masses were low by a factor of 5. Hallar et al. [2004] discuss a detailed comparison of ice water contents obtained from the CLH and the FSSP-300. The Hallar et al. [2004] study suggests that the FSSP-300 may have measured ice water content low by a factor of approximately 10. The data shown in Figure 5 account for a factor of 2 correction in the surface area, whereas Figure 6 shows a factor of 10 correction in the surface area.

Details are in the caption following the image
NOY surface coverage versus HNO3 gas pressure (Torr), as seen in Figure 5. In Figure 6 the surface area used in the calculation of surface coverage has been increased by a factor of 10 relative to the measurements.

[18] It can be seen in Figures 5 and 6 that surface coverage data show no dependence of NOY surface coverage on HNO3 gas pressure or temperature. Figures 5a and 6a show that the HNO3 values at ozone levels greater than 100 ppbv (stratospherically influenced air) separate based on temperature. This pattern may be a coincidence, but is discussed further by Gamblin et al. [2006].

[19] Nearly all the SOLVE NOY surface coverage data in Figures 5 and 6 are submonolayer, where one monolayer of HNO3 is indicated by the dashed line at 5 × 1014 molecules/cm2 [Hudson et al., 2002]. Since most of the data are submonolayer, it is relevant to determine whether NOY condensation on ice is a significant process and if this process differs depending on location in the upper troposphere or lower stratosphere. In this vein, the frequency of finding different amounts of NOY on ice particles sampled during the SOLVE mission on 23 January 2000 is illustrated in Figure 7. Again, these data were separated based on ozone concentration where [O3] > 100 ppbv represents the stratospherically influenced air (Figure 7a) and [O3] < 100 ppbv represents the tropospherically influenced air (Figure 7b). Figure 7 shows the percent of data points sampled at specific temperatures that have more than a certain percent of NOY on the particle. The three different temperature bins are differentiated by color, as shown in the temperature legend on the right in Figure 7b). This plot was obtained by first dividing particulate NOY by the sum of particulate NOY and gas-phase NOY to obtain the particulate fraction. Next the number of data points found between different particulate fraction values was counted. For example, Figure 7b shows that 45% of the data points between temperatures of 203–205 K had more than 30% of NOY condensed on the particles and about 20% of the points had more than 40% of total NOY on the particles. Total NOY was determined by summing the gas-phase NOY concentration with the particulate NOY abundance (which was first divided by the enhancement factor for the NOY instrument) (NOY + equation image), and not taken directly from the NOY instrument data. Kondo et al. [2003] show similar data including both the SOLVE and BIBLE field campaigns. Because significant amounts of NOY were found on the ice particles, condensation on ice appears to be an important method of NOY removal from the gas phase at the low temperatures of the Scandinavian upper troposphere. Conversely, the histogram in Figure 7a, a plot of only the higher ozone/lower stratospheric data, is significantly different from Figure 7b in that it shows much less condensed NOY. The percentages shown in Figure 7a are consistent with the Meilinger et al. [1999] data where very little HNO3 is found in the higher altitude ice cloud.

Details are in the caption following the image
Percentage of points (y axis values) having more than a certain percentage of NOY on the particle (x axis values). The blue line corresponds to data collected between 197 and 199 K, the green lines represent data between 199 and 201 K, the yellow lines represent data between 201 and 203 K, and the red lines denote the warmest temperature data collected between 203 and 205 K. As seen in Figures 5 and 6, the data set was segregated based on ozone concentration (a surrogate altitude scale) to produce histograms in Figure 7. (a) Data sampled above ozone levels of 100 ppbv are considered to be particles sampled in the lower stratosphere. (b) Data sampled at ozone levels below 100 ppbv are considered to be from tropospheric air parcels.

[20] The difference between Figures 7a and 7b suggests the process of NOY condensation on ice at upper altitudes may be somehow different from the process occurring at lower altitudes. The difference in behavior of NOY condensation on lower stratospheric and upper tropospheric ice particles is not likely to be due to differences in the clouds. Indeed the cloud deck sampled was more or less continuous along the flight path, as shown by Figure 1 where the satellite image shows the cloud layer extending for tens to hundreds of kilometers horizontally and the lidar data in Figure 2 show the cloud layers extending for several kilometers below the aircraft. Instead, it is suggested that the differences between the panels depicting upper tropospheric data and the panels depicting lower stratospheric data are due to differences in the partitioning of NOY species at different altitudes.

4. Cocondensation of Other NOY Species on Ice

[21] We next examine vertical profiles of various constituents of total NOY and ratios of these constituents to NOY and total NOY. Figure 8 shows the fraction of total NOY (not including NOX) due to HNO3 (in blue) as a function of ozone abundance (ppbv). Also shown in Figure 8 is the fraction of total NOY−NOX due to NOY species other than HNO3 or NOX (in red) as a function of ozone abundance. Figure 8 depicts how the ratio of the condensable components of NOY to total NOY−NOX behaves as a function of altitude. As discussed previously, total NOY was calculated to account for the enhancement efficiency and not taken directly from the NOY instrument data. It can be seen in Figure 8 that at altitudes where [O3] > 100 ppbv, HNO3 is the dominant component of NOY, composing 50–100% of total NOY−NOX, while at altitudes where [O3] < 100 ppbv, NOY appears to be composed of a mix of species. Figure 9 represents the same relationships described in Figure 8, instead using data collected on 9 March 2000. On this date, it can be seen that NOY is dominated by HNO3 in the troposphere and the stratosphere, in contrast to the trend seen for 23 January 2000, where NOY in the troposphere is not dominated by any particular species. While Figures 8 and 9 demonstrate how the ratio of the condensable components of NOY to total NOY−NOX behaves as a function of altitude, Figure 10 compares gas phase HNO3 and PNOY concentrations as a function of altitude for the data collected on 23 January 2000.

Details are in the caption following the image
Fraction of total NOY (not including NOX) due to HNO3 (blue) and due to NOY-HNO3-NOX (that part of NOY which is not due to HNO3 or NOX, in red) as a function of ozone abundance for 23 January 2000. Total NOY was determined by summing the gas-phase NOY concentration with the particulate NOY abundance (which was first divided by the enhancement factor for the NOY instrument). The blue-black and red-black larger circles represent the mean of the total NOY-NOX fraction for HNO3 and NOY-HNO3-NOX, respectively, at the mean value of the ozone data when grouped into sextiles.
Details are in the caption following the image
Fraction of total NOY (not including NOX) due to HNO3 (blue) and due to NOY-HNO3-NOX (that part of NOY which is not due to HNO3 or NOX, in red) as a function of ozone abundance for 9 March 2000. The blue-black and red-black larger circles represent the mean of the total NOY−NOX fraction for HNO3 and NOY−HNO3−NOX, respectively, at the mean value of the ozone data when divided into eighths.
Details are in the caption following the image
Ozone concentration (ppbv), a surrogate altitude scale, versus HNO3 gas (blue) and PNOY (red) concentrations. The dashed line separates regions of tropospherically influenced air (below 100 ppbv O3) from regions of more stratospherically influenced air (above 100 ppbv O3), and the blue-black and red-black circles indicate the HNO3 or PNOY concentration at the mean value of ozone data, respectively, when separated into sextiles.

[22] Figure 10 is a plot of HNO3 gas concentration (pptv) (in blue) and PNOY concentration (pptv) (in red) versus ozone abundance (ppbv). As expected, HNO3 gas is tightly correlated with ozone since both have primarily stratospheric sources. Previous studies have assumed HNO3 is the primary or major constituent of particulate NOY. If HNO3 were the primary substance condensing on ice and did so faster than competing species, it is expected that PNOY would also show a tight positive correlation with ozone. At concentrations less than 100 ppbv ozone (below the dashed line) PNOY seems to be independent of HNO3 concentration. Above 100 ppbv ozone, high concentrations of HNO3 are seen, while the corresponding PNOY is actually negatively correlated with ozone and HNO3.

[23] Figure 11, a plot of ozone concentration (ppbv) versus surface area density (μm2/cm3), shows the anticorrelation of PNOY seen in Figure 10 cannot be due to surface area since surface area is relatively unrelated to ozone. It is possible that the concentration of PNOY falls off at higher altitudes because temperature was found to be slightly higher there than at lower altitudes, as seen in Figure 12, a plot of ozone abundance (ppbv) versus temperature (K). Unfortunately, laboratory data disagree about the temperature dependence of nitric acid condensation on ice particles, where some studies suggest little temperature dependence and others suggest a strong temperature dependence (Tabazadeh et al. [1999] and Hudson et al. [2002], respectively).

Details are in the caption following the image
Ozone concentration (ppbv) versus measured surface area density (μm2/cm3). Circles indicate the surface area at the mean value of ozone data when grouped into sextiles. The dashed line separates regions of tropospherically influenced air (below 100 ppbv O3) from regions of more stratospherically influenced air (above 100 ppbv O3).
Details are in the caption following the image
Ozone concentration (ppbv) versus temperature (K). Circles represent the temperature at the mean value of ozone data when grouped into sextiles. The dashed line separates regions of tropospherically influenced air (below 100 ppbv O3) from regions of more stratospherically influenced air (above 100 ppbv O3).

[24] In Figure 10, there is no obvious correlation between HNO3 and PNOY at altitudes where ozone concentrations are below 100 ppbv. To better understand the processes occurring at these altitudes where [O3] < 100 ppbv, we examined the behaviors of the fraction of total NOY due to HNO3 and the fraction of total NOY which is due to NOY species other than HNO3 or NOX (i.e., the fraction of NOY due to species including but not limited to N2O5, PAN, HONO, HO2NO2, ClONO2, etc.) as the mass of particulate NOY increases because of gases condensing on the particle.

[25] Figure 13 shows the gas phase nitric acid fraction of NOY (the amount of NOY−NOX due to HNO3) versus the particulate fraction of NOY (the amount of total NOY, not including NOX, which is found on the particle) below 100 ppbv ozone (tropospherically influenced air) for data collected on 23 January 2000. The solid circles represent, as a fraction, the amount of gas-phase HNO3 over the total amount of condensable components of NOY. Moving from left to right on the x axis in Figure 13, the ice particle is growing because of NOY condensation (an increasing particulate fraction). As the particle is growing by NOY condensing on it, the HNO3 concentration (solid circles) is becoming enriched relative to the concentration of the other condensable species of NOY. If the gas-phase HNO3 concentration is becoming enriched relative to other condensable NOY species, then HNO3 cannot be what is condensing on the particle. The open circles represent the fraction of NOY−NOX not composed of HNO3 or NOX, which becomes depleted as particles accumulate (as the values on the x axis increase). If HNO3 was condensing on the ice particle, its trend would look more like the open circles, which are being depleted as the particle is growing. The trends observed in Figure 13 suggest a constituent of NOY is leaving the gas phase to condense on the particles, while the majority of HNO3 remains in the gas phase, thus causing NOY to become enriched in HNO3. One cannot conserve HNO3 when gas-phase HNO3 is becoming enriched as the particle is growing, while also at the same time having a sink for HNO3. Therefore Figure 13 shows that HNO3 cannot lead to the observations of NOY surface coverages shown in Figures 5 and 6. As an aside, the possibility of HNO3 or NOY burial in a growing ice particle is treated by Gamblin et al. [2006]. HNO3 burial in a growing ice particle cannot play an important role in this portion of our study, because with particle growth the gas-phase NOY is becoming enriched in HNO3. Burial would be considered a sink for HNO3. To conserve HNO3 which is generally remaining in the gas phase, there cannot exist simultaneously an effective sink.

Details are in the caption following the image
Fraction of gas phase HNO3 to (NOY−NOX) concentration (solid circles, y axis) versus the particulate fraction (fraction of NOY on the particle, x axis) for data sampled on 23 January 2000 at ozone concentrations less than 100 ppbv (air sampled in the upper troposphere). The open circles show the behavior of the gas-phase fraction of NOY not composed of HNO3 versus particulate fraction. X = Pan + HNO4 + N2O5 + HONO + etc. (i.e., X = NOY−NOX−HNO3). The standard deviation of the data is shown as solid lines.

[26] To investigate whether the trends shown in Figure 13 are simply due to a correlation between the HNO3 and NOY instruments, these same trends were plotted for ozone concentrations greater than 100 ppbv, i.e., stratospherically influenced air. If the trends shown in Figure 13 are the result of an instrument artifact, or cloud dynamics, these gas phase fraction trends should be the same whether in air with a high fraction of HNO3 in NOY or in air with a low fraction. Figure 14 examines the behavior of HNO3 gas as it condenses on cloud particles where ozone levels are above 100 ppbv. Within the standard deviation, there is little variation in HNO3 concentration with particle fraction. Further, HNO3 is clearly the dominant gas phase NOY species with much greater values than in the tropospheric case.

Details are in the caption following the image
Fraction of gas phase HNO3 to (NOY−NOX) concentration (solid circles, y axis) versus the particulate fraction (fraction of NOY on the particle, x axis) for data sampled at ozone concentrations greater than 100 ppbv (stratospherically influenced air). The open circles show the behavior of the gas phase fraction of NOY not composed of HNO3 versus particulate fraction. X = Pan + HNO4 + N2O5 + HONO + etc. (i.e., X = NOY−NOX−HNO3). The standard deviation of the data is shown as solid black lines.

[27] Another relevant example is observed on 9 March 2000. As shown in Figure 9, on this day nitric acid is a significant component of NOY in the troposphere. Moreover, as shown from plots of solar zenith angle (SZA) versus hours before sampling (obtained from NASA Goddard Space Flight Center (GSFC) back trajectory analyses), the 23 January air parcels had been in the dark for many hours prior to sampling, whereas on 9 March the air parcels had been exposed to sunlight prior to sampling (Figures 15 and 16, respectively). Figure 17 shows that on March 9 the tropospheric nitric acid did not become enriched in HNO3 as particulate NOY built up. Finally, Figure 18 shows the general correlation of PNOY and HNO3 with ozone in the stratosphere on 9 March 2000. Unlike the January 23 case (Figure 10), the 9 March data in Figure 18 show that PNOY and HNO3 are positively correlated with ozone.

Details are in the caption following the image
Solar zenith angle (SZA) (degrees) versus hours before sampling for 23 January 2000 (obtained from NASA GSFC back trajectory analyses). The dashed line indicates a SZA of 92°. As the x axis decreases from left to right, the air parcel is moving forward in time. The air mass was sampled by the DC-8 when the x axis is zero. Angles below the dashed line indicate the presence of sunlight, while SZAs greater than 92° indicate the absence of sunlight.
Details are in the caption following the image
Solar zenith angle (degrees) versus hours before sampling for 9 March 2000 (obtained from NASA GSFC back trajectory analyses). The dashed line indicates a SZA of 92°. As the x axis decreases from left to right, the air parcel is moving forward in time. The air mass was sampled by the DC-8 when the x axis is zero. Angles below the dashed line indicate the presence of sunlight, while SZAs greater than 92° indicate the absence of sunlight.
Details are in the caption following the image
Ratio of gas phase HNO3 to (NOY−NOX) concentration (solid circles, y axis) versus the particulate fraction (fraction of NOY on the particle, x axis) for data sampled on 9 March 2000 at ozone concentrations less than 100 ppbv (air sampled in the upper troposphere). The open circles show the behavior of the gas-phase fraction of NOY not composed of HNO3 versus particulate fraction. X = Pan + HNO4 + N2O5 + HONO + etc. (i.e., X = NOY−NOX−HNO3). The standard deviation of the data is shown as black lines.
Details are in the caption following the image
Ozone concentration (ppbv), a surrogate altitude scale, versus HNO3 gas (blue) and PNOY (red) concentrations for data collected on 9 March 2000. The dashed line separates regions of tropospherically influenced air (below 100 ppbv O3) from regions of more stratospherically influenced air (above 100 ppbv O3), and the blue-black and red-black circles indicate the HNO3 or PNOY concentration at the mean value of ozone data when separated into eighths, respectively. PNOY data above 200 ppbv O3 come from very small clouds that are removed from the data set when the lag time is increased from 10 to 30 s. Owing to their small size the cloud lifetime of these points may be shorter than for other PNOY data in this plot; therefore kinetics may explain the reduced amount of PNOY on these particles [Gamblin et al., 2006].

[28] We draw the following conclusions from these various data sets. Figures 5 and 6 show a difference in behavior of condensed NOY depending on the ozone concentration (altitude). Furthermore, there is a difference in the amount of condensed NOY on the particles at different altitudes (Figure 7). The question becomes “What is causing this difference in behavior?” Figures 1, 2 and 3 show the cloud layer was very homogeneous; therefore any differences in behavior cannot be due to cloud microphysical properties. Subsequently, Figure 13 shows that as NOY condenses on the particle, HNO3 becomes a larger fraction of NOY. Therefore most of the HNO3 is remaining in the gas phase at lower altitudes ([O3] < 100 ppbv). Figure 14, the same plot as Figure 13 but for [O3] > 100 ppbv, shows that within the error bars, the concentration of HNO3 relative to the concentration of the condensable amount of NOY stays relatively constant as NOY is condensing on the particles. Thus Figures 5, 6, 7, 13 and 14 suggest that the spatial differences in the “HNO3 to NOY fraction” may be the cause of the behavior differences seen in Figures 5, 6, and 7.

[29] Further conclusions can be drawn. First, as is well known, in the lower stratosphere, HNO3 usually dominates NOY (Figures 8 and 9). In this case, nitric acid condensed on ice. The fact that HNO3 can be either a large or small fraction of the PNOY (Figures 10 and 18) may suggest that the condensation is relatively slow so that PNOY varies depending on cloud lifetime (the condensation rate of HNO3 is further discussed by Gamblin et al. [2006]). Otherwise one would expect PNOY to be correlated with HNO3 (or at least with HNO3+PNOy). Second, it is often the case that in the troposphere NOY contains only a small fraction of nitric acid (Figure 8). While nitric acid may also condense in the troposphere, there are conditions in which some other constituents of NOY must condense. This is illustrated by the fact that HNO3 becomes enriched in the gas phase as PNOY increases on 23 January 2000 (Figure 13). Conversely, Figure 17 shows that HNO3 does not become enriched in the gas phase as PNOY increases on 9 March 2000. The only obvious difference between the conditions on 9 March and 23 January is the exposure of the air parcel to sunlight in the day prior to sampling. Assuming therefore that at least one component of NOY other than HNO3 is condensing on lower altitude, upper tropospheric particles collected on 23 January, we next explore which of the possible NOY species could be cocondensing.

5. Condensation of NOY Component Other Than HNO3

[30] Several species could be condensing in addition to, or instead of, HNO3 on the lower altitude, upper tropospheric cloud particles. We seek a condensable component of NOY with an abundance near 50–100 pptv (the abundance of upper tropospheric PNOY in Figure 10). In addition, if the species cocondensing is going to adsorb faster than HNO3, it must have an accommodation coefficient above that of HNO3. Approximate values of 0.01 and 0.005 for the accommodation coefficient of HNO3 on ice (over the temperature ranges 215–235 K and 208–220 K) have been suggested (Hynes et al. [2002] and Hudson et al. [2002], respectively). A detailed comparison of the mass accommodation coefficients for HNO3 is given by Gamblin et al. [2006]. Here we primarily discuss three constituents of NOY which might condense on ice at upper tropospheric conditions: N2O5, PAN (peroxyacetyl nitrate, CH3C(O)OONO2) and PNA (peroxynitric acid, HO2NO2 or HNO4). N2O5 is the most likely species given its reactivity and concentration in the UT/LS region.

[31] It is well known that N2O5 reacts on ice to form nitric acid with an accommodation coefficient of about 0.024 [Leu, 1988]. Vertical profiles of N2O5 derived from limb emission spectra measured by the MIPAS-B instrument inside the polar vortex show over 100 pptv N2O5 at approximately 11 km [Wetzel et al., 1995]. Although N2O5 is rapidly photolyzed, the air sampled by the DC-8 on 23 January 2000 was primarily in darkness. Furthermore, back trajectory analyses of air parcels sampled on that date (NASA GSFC model, Figure 15) lead to regions experiencing polar night for most of the previous 24 hours, thus providing the opportunity for the more photochemically sensitive species such as N2O5 to increase in abundance and making N2O5 a plausible suggestion for the NOY constituent that is condensing along with HNO3.

[32] By contrast, Figure 16 shows back trajectory analyses of air parcels from 9 March 2000. Air parcels sampled on this date experienced mostly daylight during the flight and for the previous 9–12 hours before sampling. This extended period of sunlight would reduce the abundance of most condensable NOY species other than HNO3, which is much more stable in the presence of sunlight.

[33] Because condensation on ice, as described for the 23 January 2000 case, would result in semipermanent removal of N2O5, less NO3, NO2, and NO would be produced in the upper troposphere as a result of N2O5 photolysis when the air parcel is next exposed to sunlight. In addition, N2O5 condensation on falling ice particles would remove HNO3 and NOX to lower, warmer altitudes, where the ice particles would evaporate and thereby cause a repartitioning of NOX and NOY.

[34] PAN and PNA are both known to condense on ice at much higher temperatures and partial pressures than present in the upper troposphere. PAN has been found on ice in varying locations and temperatures and has the necessary concentrations to be a non-HNO3 species of NOY condensing on upper tropospheric cirrus particles [Ford et al., 2002; Munger et al., 1999; Singh et al., 1992; Thakur et al., 1999]. On the basis of molecular adsorption enthalpy calculations using conditions somewhat similar to those measured during the SOLVE campaign, Bartels-Rausch et al. [2002] found that adsorption on ice might be of some importance as a removal mechanism for PAN. Although this reference suggests PAN is much less likely to condense than HNO3, the studies were conducted approximately 13 K above average temperatures measured during SOLVE and the surface areas used were approximately 1000 times larger. In the laboratory, HNO3 adsorption on ice (and presumably that of PAN and other condensable NOY species) behaves very differently at 200 K than at 213 K. It is impossible to tell from the Bartels-Rausch et al. [2002] paper exactly how the species will behave at typical UT/LS conditions measured during the SOLVE mission. The laboratory data of PAN are insufficient to completely exclude it as the species which could be condensing on ice in the upper troposphere. Assuming at typical upper tropospheric conditions that the accommodation coefficient of PAN on H2O ice is greater than 10−2, it is possible, albeit perhaps unlikely, that PAN could be removed by condensation on sedimenting ice particles. In the upper troposphere, PAN should not be sensitive to the sunlight exposure of the air parcel over short periods of time.

[35] Peroxynitric acid (PNA) has an ambient concentration of approximately 0.2 ppbv in the upper troposphere/lower stratosphere and an accommodation coefficient of 0.15 ± 0.10 on solid H2O ice at temperatures near 195 K [Li et al., 1996]. The Li et al. [1996] study states that HNO4 is rapidly adsorbed on water ice surfaces at 190 K, resulting in only condensed phase species. Although PNA can be removed by rapid photolysis, most of the SOLVE-I mission data studied on 23 January 2000 were obtained in darkness, thus PNA could be another non-HNO3 component of NOY condensing on ice in the upper troposphere. According to Li et al. [1996], if HNO4 were to rapidly condense on ice, its products could either “result in a significant shift in the NOY partitioning toward more active NOY species (i.e., NO, NO2, or HONO) or toward less active NOY species such as HNO3.” It has been shown that HNO4 will decompose on room temperature glass or in solution to form HONO [Zhu et al., 1993; Logager and Sehested, 1993], a species which is readily photolyzed to yield NO. Thus condensation on ice could destroy reservoir HNO4 and, once the sun rises over the Arctic Circle, reform NOX. Alternatively, if HNO4 were to rapidly condense on H2O ice particles and not decompose, these particles could sediment, falling to lower, warmer altitudes, where they could evaporate and act as a source for HNO4. When the sun rises, this species could rapidly photolyze to NO2. Therefore a heterogeneous chemical reaction by PNA on ice would provide an upper tropospheric source of NOX and adsorption of PNA on ice would repartition NOX to lower altitudes.

[36] There are other possible constituents of NOY such as alkyl nitrates or other organic nitrates that may be cocondensing on the upper tropospheric ice particles. While they generally appear to be too low in concentration (∼2–25 pptv) [Stroud et al., 2001; Blake et al., 2003a, 2003b] to supply the condensed NOY, speciation in the upper troposphere of NOY is generally lacking and few data are available. Generally, we cannot tell what other species are condensing on the ice particles, because the laboratory studies of non-HNO3 NOY species condensing on ice at the atmospheric conditions measured during SOLVE-I are not sufficient. Since the data from 23 January 2000 SOLVE flight were collected in darkness, it is most likely that N2O5 is the non-HNO3 species of NOY condensing on ice. It is known that N2O5 has a sufficient abundance and readily reacts on ice to form HNO3; therefore its role in NOY condensation on ice should be significant.

[37] It has been shown that condensation of NOY on lower stratospheric and upper tropospheric ice particles has different properties depending on the fraction of NOY that is HNO3 and perhaps other factors such as sunlight exposure of the air parcel. This behavior is not likely to be due to the nature of the clouds themselves as the cloud decks sampled were more or less continuous along the flight path and did not have significant differences in temperature or surface area (as measured). Gamblin et al. [2006] propose an additional process affecting the data. Condensation of HNO3 on lower stratospheric cirrus particles may be kinetically limited. Furthermore, they suggest the low accommodation coefficient for HNO3 on ice combined with relatively short-lived clouds causes highly variable, cloud lifetime limited HNO3 uptake on cirrus particles.

6. Conclusions

[38] NOY condenses on ice particles in the upper troposphere to a significant degree. In the upper troposphere HNO3 is often not the dominant component of NOY. On 23 January 2000, in the upper troposphere, nitric acid was found to be enriched in the gas phase in regions where NOY was present on clouds, while the remainder of NOY (not including HNO3 or NOX) was depleted. It is suggested that some constituents of NOY other than HNO3, such as PAN, HNO4 or N2O5, condensed along with HNO3 on the lower altitude, upper tropospheric ice particles on this day. Further atmospheric data on the speciation of condensed phase NOY and laboratory data on the adsorption of PAN and PNA measured under the atmospheric conditions encountered on 23 January 2000 during the SOLVE-I mission would be pertinent, though N2O5 is known to condense and has to be considered the most likely candidate to explain these data. In the lower stratosphere on 23 January 2000, and on 9 March in the upper troposphere and lower stratosphere, most of the NOY is composed of gas phase HNO3 and in these cases HNO3 is not depleted as PNOY increases. Our results suggest that the presence of condensed phase NOY or even nitric acid, is not sufficient evidence that gas phase nitric acid condensed; other species of NOY, most likely N2O5, but possibly PNA or PAN may condense in the upper troposphere and do so more rapidly than HNO3. Additionally, Gamblin et al. [2006] suggest that kinetics may control the condensation of nitric acid so that particulate NOY can be highly variable in the atmosphere.

Acknowledgments

[39] Work performed at the Jet Propulsion Laboratory, California Institute of Technology, was carried out under a contract with the National Aeronautics and Space Administration. B. Gamblin and Owen B. Toon were supported by NASA. The LASE aerosol backscattering ratio images were provided by Edward Browell's Lidar Applications Group, NASA LaRC.