Airborne measurements of HCl from the marine boundary layer to the lower stratosphere over the North Pacific Ocean during INTEX-B

Gas phase HCl was measured from the marine boundary layer (MBL) to the lower stratosphere from the NASA DC-8 during ﬁve science ﬂights (41 h) of the Intercontinental Chemical Transport Experiment-Phase B (INTEX-B) ﬁeld campaign. In the upper troposphere/lower stratosphere (UT/LS, 8–12 km) HCl was observed to range 5 from a few tens to 100 pptv due to stratospheric inﬂuence with a background tropospheric level of less than 2 pptv. In the 8–12 km altitude range, a simple analysis of the O 3 /HCl correlation shows that pure stratospheric and mixed tropospheric/stratospheric air masses were encountered 30% and 15% of the time, respectively. In the mid troposphere (4–8 km) HCl levels were usually below 2 pptv except for a few cases of 10 stratospheric inﬂuence and were much lower than reported in previous work. These data indicate that background levels of HCl in the mid and upper troposphere are very low and conﬁrm its use in these regions as a tracer of stratospheric ozone. How-ever, a case study suggests that HCl may be produced in the mid troposphere by the dechlorination of dust aerosols. In the remote marine boundary layer HCl levels were 15 consistently above 20 pptv (up to 140 pptv) and strongly correlated with HNO 3 . Cl atom levels were estimated from the background level of HCl in the MBL. This analysis suggests a Cl concentration of ∼ 3 × 10 3 atoms cm − 3 , which corresponds to the lower range of previous studies. Finally, the observed HCl levels are compared to predictions by the Real-time Air Quality Modeling System (RAQMS) to assess its ability to characterize the impact of stratospheric transport on the upper troposphere. of


Introduction
Hydrochloric acid (HCl) is produced in the troposphere and stratosphere by different mechanisms. In the troposphere the major source of HCl is thought to be dechlorination of sea-salt aerosol by acids such as HNO 3 and H 2 SO 4 (Erickson, 1959a, b; Kerminen has been suggested to lead to enhanced ozone production (Finley and Saltzman, 2006;Tanaka et al, 2003;Chang et al., 2002;Spicer et al., 1998).
Direct observations of HCl in the MBL and lower troposphere are limited in terms of frequency and geographical coverage. However, HCl has been measured in a variety of locations (Graedel and Keene, 1995). These measurements indicate HCl mixing 15 ratios in the remote MBL (0-200 m) of 100-300 pptv with levels decreasing to 50-100 pptv in the remote marine free troposphere (1 km-6 km; Vierkorn-Rudolph et al., 1984). In the urban influenced troposphere, ppbv levels of HCl have been reported in various locations (Keene et al., 2007;Graedel and Keene, 1995). Keene et al. (1999) calculated the global tropospheric budget of HCl, based on the data from Graedel 20 and Keene (1995), and reported that additional sources are needed to explain the distribution of HCl in the troposphere. However, this conclusion was based on data from analytical methods (filter techniques), which have been identified to potentially have positive artifacts such as NOCl, ClNO 2 , ClNO 3 and chlorinated aerosols. Therefore, most of these studies report HCl*, which includes these and potentially other species. Introduction  , 1997Jobson et al., 1998;Arsene et al., 2007). These estimates range from 720 atoms cm −3 (Wingenter et al., 1999) to 10 5 atoms cm −3 (Singh et al., 1996a;Wingenter et al., 1996).
In the stratosphere HCl is produced primarily by the reaction of Cl radicals with CH 4 (Lin et al., 1978). The source of stratospheric chlorine is the photodissociation of 5 chlorofluorocarbons (CFCs) (Molina and Rowland, 1974). HCl is the most abundant form of inorganic chlorine in the stratosphere due to its long photochemical lifetime (∼30 days at 20 km, Webster et al., 1994). However, HCl can be lost via heterogeneous processes in the stratosphere (Hanson et al., 1994;Hanson and Ravishakara, 1991, 1993, and 1994Tolbert et al., 1988). Lelieveld et al. (1999) reported a mean 10 mixing ratio of ∼450 pptv of HCl at 12.3 km in the late Arctic winter (Feburary 1995;Kiruna, Sweden) from observations with a quadrupole mass spectrometer. Webster et al. (1994) also reported in situ HCl measurements in the stratosphere using a tunable diode laser spectrometer integrated on the NASA ER-2 research aircraft during the Stratospheric Photochemistry, Aerosols, and Dynamics Expedition (SPADE) mis-15 sion. These research flights covered a latitude range of 15-60 • N and altitudes below 20 km in spring and fall of 1992 and 1993. HCl levels of 500 pptv to 1ppbv of HCl were observed over a pressure range of 50 to 70 mb. In addition, this study also found that model predictions of the HCl fraction of inorganic chlorine (Cl y =HCl, ClO, and ClONO 2 ) in the stratosphere were systematically higher than observations although the model 20 predicted Cl y within the uncertainty of the measurement.
Remote sensing has been used to measure the global distribution of HCl in the stratosphere. The Halogen Occultation Experiment (HALOE) indicated ∼1 ppbv of HCl at 10 mb with no obvious variation as a function of season or latitude (Russell et al., 1996). 25 Recently, Marcy et al. (2004)  Area Cirrus Experiment (CRYSTAL-FACE) mission. They also found the observed relationship to be consistent with the Integrated Massively Parallel Atmospheric Chemistry Transport (IMPACT) model of stratospheric chemistry (Rotman et al., 2004). Consequently, they proposed using the observed HCl-O 3 ratio to calculate the fraction of stratospheric ozone in an air parcel. However, this method relies on the assumption 5 that HCl in the free troposphere is only of stratospheric origin which is inconsistent with previous observations (e.g. Graedel and Keene, 1995). In this work we present observations of HCl from the NASA DC-8 during the Anchorage, AK deployment of the Intercontinental Chemical Transport Experiment-Phase B (INTEX-B). This phase of the mission consisted of five flights over the North Pacific as 10 shown in Fig. 1. Each science mission consisted of level flight legs and multiple spiral vertical profiles from the MBL to the upper troposphere (UT). The comprehensive vertical coverage allows us to examine HCl levels over the entire troposphere. These data are analyzed using correlations with other measured species and a 3-D chemical transport model to probe our understanding of the sources and distribution of HCl in 15 the troposphere.

Instrumentation
The chemical ionization mass spectrometer (CIMS) integrated on the NASA-DC 8 aircraft for the INTEX-B campaign was identical to that used in INTEX-North America 20 (NA) and is described in detail in Kim et al. (2007) and Slusher et al. (2004). The instrument comprises an inlet, a flow tube ion molecular reactor, a collisional dissociation chamber (CDC), an octopole ion guide, and a quadrupole mass filter. SF − 6 ion chemistry is utilized to selectively ionize HCl and SO 2 (R1 and R2) in the CIMS flow tube reactor (Huey, 2007;Slusher et al., 2001;Huey et al., 1995). 25 HCl + SF − 6 →SF 5 Cl − + HF R1a was chosen to detect HCl, as it has the highest yield (∼45%), is more selective than R1c, and was less prone to interference than R1b. The sensitivity of the CIMS to 5 HCl was not calibrated during the science flight. The sensitivity to SO 2 was periodically measured (every 2.5 min) by standard additions of 34 SO 2 to the inlet (R2). Periodic background checks to both SO 2 and HCl were performed using an activated charcoal and nylon wool scrubber (∼every 10 min). The temporal variations of the HCl and SO 2 product ions during a science flight are presented in Fig. 2. 10 The sensitivity of HCl relative to SO 2 was assessed post mission by a series of laboratory tests over the pressure and the humidity conditions encountered on the DC-8. These experiments demonstrate that the sensitivity ratio of HCl and SO 2 is a strong function of water vapor (Fig. 3) at dewpoints below −15 • C. The sensitivity to SO 2 has a negative correlation with humidity due to the hydrolysis of the product ion F 2 SO − 2 .

15
However, the sensitivity to HCl was found to be essentially independent of humidity. For this reason, the HCl sensitivity at dewpoints below −15 • C was calculated relative to the SO 2 standard addition. However, at dewpoints above −15 • C, the strong correlation between the sensitivity of HCl and the reagent ion signal ( 34 SF − 6 ) was utilized to obtain the HCl sensitivity. An estimated uncertainty of 33% was obtained from con-20 sidering potential errors in the gas standards (SO 2 , 5% and HCl, 10%), the dew point measurement (5%) and the post-mission calibrations (18%) combined with the statistical error at the 2σ level of the in flight calibration (1 s average, 25%). The lower limit of detection (LLOD) for HCl was estimated to be 2 pptv for a signal to noise ratio of one with the noise defined as 2σ of the background signal for a 30 s average. 25 In earlier work we have pointed out that ozone and in particular water vapor are potential interferences for detection of any ambient species using CIMS and SF − 6 ion 3568 Introduction  (Slusher et al., 2001). For this reason, the selectivity of the HCl measurement was further assessed by taking in-flight mass spectra in both the stratosphere (high O 3 ) and the MBL (high water vapor) (Fig. 4). These results showed low background levels in the region of the product ion (162 amu) and exhibit the expected isotope signature of SF 5 Cl − ( 35 Cl: 37 Cl). These results indicate that the HCl measurement 5 at SF 5 Cl − is viable over a wide range of atmospheric conditions. These observations were also confirmed with laboratory tests that demonstrated that SF 5 Cl − detection channel is essentially immune to interference from water vapor and ozone. However, the product channel Cl − (HF), R1b, has significant interferences due to water vapor at even moderate dew points. and 60 • N were assimilated. The assimilation accounts for the trace gas retrieval sensitivities by convolving the model first guess ozone profile with the OMI and TES averaging kernels and a priories. Since the TES L2 quality flags have been developed primarily for tropospheric retrievals we restricted assimilation of TES CO to below the local 25 tropopause while TES O 3 was assimilated below 10 mb. The RAQMS HCl predictions were not constrained with observations. The RAQMS meteorological forecasts are initialized from NOAA Global Forecasting System (GFS) analyses at 6-hour intervals.

Results and discussion
All reported data and analyses are based on a 1 min merged dataset unless otherwise noted. The median and the mean attitude profiles of HCl for five science flights of the INTEX-B mission are presented in Fig. 5, along with the median profiles of O 3 and HNO 3 . In addition, in Table 1, we report the statistics of the vertical distribution of HCl.

5
In general, we found high HCl episodes (up to 140 pptv) in the upper troposphere (8-12 km) interspersed with observations of low levels near our detection limit of 2 pptv. Although the median value of the highest altitude bin in the figure indicates a low value, it should be noted that this value represents only 10 min of data from one flight leg.
In the MBL, HCl levels above 20 pptv were routinely observed. In the mid troposphere 10 (4-8 km), HCl was measured below 15 pptv more than 90% of the time. However, one case of strong stratospheric influence was identified by HCl enhancement along with enhanced O 3 . There was also evidence for one episode of dechlorination in the mid troposphere. The vertical distribution of HCl in this study is significantly different from that of Keene et al. (1999) and Graedel and Keene (1995). High levels of HCl in the upper troposphere were strongly associated with stratospheric influences. This is illustrated in Fig. 6a  with Be-7, another stratospheric tracer measured on the DC-8 with a 15 min integration time (Dibb et al., 2003). The method of Marcy et al. (2004) was employed to assess the extent of stratospheric influence on the upper troposphere during the Anchorage deployment of INTEX-B. This method utilizes the O 3 -HCl correlation (Fig. 7a) and sets a stratospheric 5 end member (O 3 : 160 ppbv and HCl: 30 pptv). Above the end member the air is categorized as pure stratospheric. Air parcels with HCl below detection limit are categorized as pure tropospheric air. In between these limits the air is characterized as a mixture of both. The analysis suggests that in the upper troposphere (8-12 km) during INTEX-B pure stratospheric air was sampled ∼30% of the time and air with significant 10 stratospheric influence (i.e. detectable HCl) was observed ∼15% of the time.  influenced air (O 3 >120 ppbv) and pure tropospheric air (HCl<17 pptv). The results predict a higher fraction than observations for stratospheric influenced air (∼44%) and underestimate the amount of pure tropospheric air (∼20%) suggesting that RAQMS overestimates stratosphere-troposphere exchange processes. However, in general the RAQMS model appears to do a reasonable job of capturing the broader features of 10 stratospheric impact on the upper troposphere given the relatively coarse horizontal resolution (2×2 degree) of the analysis.

The MBL and the Lower Troposphere (0-4 km)
The median levels of HCl ( Fig. 5 and Table 1) increase at altitudes below 4 km and reach up to 20 pptv in the MBL (z<1 km). These observations are much lower than 15 recent measurements in the relatively clean Hawaii MBL of 30-250 pptv (Pszenny et al., 2004). Figure 8a and b shows the correlation of HCl with SO 2 and HNO 3 in the MBL (z<1 km), respectively. Both species show a good correlation with HCl except for a few outliers associated with either volcanic influence, from the Veniaminof volcano, located in the Aleutian Island chain, or anthropogenic pollution near the western U.S. coast.

20
These positive correlations with HNO 3 and SO 2 , the major precursor of H 2 SO 4 , are consistent with HCl production from the acidification of sea-salt aerosol (assuming that HCl has a relatively short lifetime in the MBL). However, at very low levels of HNO 3 and SO 2, significant levels of HCl were still observed (∼20 pptv). These low levels of HCl in the absence of HNO 3 and SO 2 could be 25 produced by the reactions of Cl atoms with VOCs. This is supported by enhanced ibutane/n-butane ratios (median=0.7), which is consistent with Cl atom oxidation (Jobson et al., 1994) Interactive Discussion needed to produce this amount of HCl is conducted using the assumption that HCl is in steady state. Three HCl loss pathways: aerosol uptake, oceanic deposition, and reaction with OH, are considered. Lifetimes for each loss process are estimated based on the INTEX-B data summarized in Table 2. These assumptions give the equation below for calculation of the Cl atom number density.
The input parameters (Table 2) are based on observations from the INTEX dataset and rate constants from the JPL compilation (Sander et al., 2006). A dry deposition rate of 1.6 cm s −1 is estimated based upon Kerkweg et al. (2006). An accommodation coefficient of 0.15 is used to calculate HCl loss to aerosol (Sander et al., 2006). 10 The average result of the calculation is 2.8×10 3 atoms cm −3 . Probably the most uncertain of these parameters is the dry deposition rate of HCl. Dry deposition rates are expected to range from 0.2-2.0 cm s −1 over a water surface (Seinfeld and Pandis, 2006). This indicates the derived value is probably an upper limit to the Cl atom number density as the dry deposition rate is only likely to be slower. The derived es-15 timate (2.8×10 3 atoms cm −3 ) is in the lower range (Singh et al., 1996b;Rudolph et al., 1996Rudolph et al., , 1997Jobson et al., 1998;Wingenter et al., 1999) of previous studies and is not compatible with higher estimates of greater than 10 5 atom cm −3 (Singh et al., 1996a;Wingenter et al., 1996). The most recent reports of Cl atom levels are toward the higher end of the range; 6×10 3 −4. Interactive Discussion studies (e.g. Spicer et al., 1997) and recent observations (Keene et al., 2007). Large enhancements of HCl were not observed in the polluted air mass (09:02 to 09:05 PM in the Fig. 9). However, the sampling duration was very short (∼3 min) and was over a limited geographic area, though.
3.3 The mid troposphere (4-8 km) 5 Figure 10 presents the correlation between O 3 and HCl in the mid-troposphere (4-8 km). Although HCl levels in the mid-troposphere were usually low (<2 pptv 55%, and <15 pptv 90%), significant levels of HCl were observed that were associated with stratospheric influence and had a similar ratio of O 3 to HCl as at higher altitudes (∼7.5 km). However, Fig. 10 also shows that HCl enhancements (more than 20 pptv) 10 can be observed with no stratospheric influence (i.e. no enhancement in O 3 and incompatible back trajectories). To investigate the origin of the non-stratospheric HCl in the mid troposphere we examined two aircraft spirals in similar geographical locations with contrasting HCl. These profiles are shown in Fig. 11 with spiral 1 having undetectable HCl and spiral 2 with significant HCl in the mid troposphere.

15
The transport of HCl to the mid troposphere from the MBL could explain the enhancement in spiral 2. However, RAQMS backtrajectory analyses show that this air mass had resided in the mid troposphere for ∼5 days without any influence from either the stratosphere or the MBL. Moreover, chemical tracers such as O 3 for the stratosphere, and CH 3 I and CH 3 NO 3 for the MBL were not enhanced in the mid troposphere during spiral 20 2. In fact they are very similar levels to those observed in the non-enhanced spiral 1. In addition, ambient temperature profiles demonstrate that the mid troposphere during both spiral 1 and spiral 2 was stratified (Fig. 12). Consequently, we are skeptical that the HCl in the mid troposphere has a recent MBL origin. However, there are high levels of non-volatile aerosol (T >350 • C) in the mid troposphere that correspond with 25 enhanced scattering at 450 nm in spiral 2 (Fig. 12). The properties of the nonvolatile aerosol (e.g. the ratio of refractory aerosol to total aerosols and the aerosol depolarization) all indicate that it is primarily dust of Asian origin. There is evidence from 3574 Introduction

Tables Figures
Back Close

Printer-friendly Version
Interactive Discussion recent field studies that dust particles can absorb significant amounts of chlorine when passing through the MBL (Sullivan et al., 2007;Ooki and Uematsu, 2005;Zhang and Iwasaka, 2001). This dust could undergo dechlorination later by exposure to strong acids such as nitric and sulfuric. High levels of HNO 3 and SO 2 are observed in spiral two that are consistent with the air mass having been in contact with urban areas such 5 as Shanghai, China as indicated by a 7-day back trajectory analysis in Fig. 11b). For these reasons, we speculate that the mid-tropospheric HCl in this case is produced by dechlorination of dust particles activated by the oxidation of anthropogenic pollution. It is doubtful if this mechanism is a large source of HCl to the atmosphere. However, this mechanism should be recognized as a potential interference to using HCl as a 10 stratospheric tracer in the free troposphere. In addition, the production of HCl from dust particles provides a mechanism to transform the chemical composition of the dust aerosol.

Summary
Airborne measurements of HCl during the Anchorage deployment of the INTEX-B field 15 mission provide a unique dataset from the MBL to the lower stratosphere over the North Pacific Ocean. In the upper troposphere (z>8 km), HCl serves as a good tracer for recent stratospheric influence due to its very low background concentration (less than 2 pptv) and it well defined relationship with stratospheric ozone. A simple analysis using the HCl/O 3 correlation illustrates that ∼50% of the air above 8 km (up to 20 12 km) was either stratospheric (∼30%) or recent stratospheric influenced air (∼15% and strongly correlated with HNO 3 . This is consistent with dechlorination of sea-salt aerosols by gas phase acids as the major source of HCl in the MBL. One sampling leg (∼15 min) in a polluted coastal boundary layer (south of Seattle, WA) did not show significant enhancements of HCl relative to the remote MBL which is in contrast with other studies. The background level of HCl in the MBL was used to estimate average 5 Cl atom number density of 3×10 3 atoms/cm 3 , which is consistent with the lower range of previous studies. In the mid troposphere (4-8 km), HCl was usually below our detection limit of 2 pptv, which is consistent with recent in situ measurement of HCl in the upper troposphere by Marcy et al. (2004). On a few occasions HCl associated with enhanced O 3 was detected due to recent stratospheric influences. In addition, enhanced HCl not of stratospheric origin was detected in the mid troposphere. This HCl appears to have been produced by dechlorination of Asian dust aerosols.
The measured HCl profiles in this work indicate above the MBL that background tropospheric levels of HCl are very low (<2 pptv). This is consistent with the findings of 15 Marcy et al. (2004) but is inconsistent with the profile of Keene et al. (1999). However, profiles obtained in this study are over a limited geographic region (Northern Pacific Ocean) which may be a reason for the disagreement. Consequently, observations by the method presented in this paper over a wider geographic range would be useful to sort out this difference.