Separating refractory and non-refractory particulate chloride and estimating chloride depletion by aerosol mass spectrometry in a marine environment

Introduction Conclusions References

that can be ambiguous in regions with significant quantities of sea salt aerosols. This ambiguity arises due to large differences in the sensitivity of the HR-AMS to refractory chloride species (i.e., NaCl) and non refractory chloride species (i.e., NH 4 Cl,HCl,etc.). Using the HR-AMS, the aerosol chloride signal is typically quantified using ion signals for 35 Cl + , H 35 Cl + , 37 Cl + and H 37 Cl + (H x Cl + ). During this study, the highest 10 aerosol chloride signal was observed during sea sweep experiments when the source of the aerosol chloride was NaCl present in artificially generated sea salt aerosols even though the HR-AMS has significantly lower sensitivity to such refractory species. Other prominent ion signals that arise from NaCl salt were also observed at m/z 22.99 for Na + and m/z 57.96 for Na 35 Cl + during both sea sweep experiments and during periods 15 of ambient measurements. Thus, refractory NaCl contributes significantly to the H x Cl + signal, interfering with attempts to quantify non sea salt chloride (nssCl). It was found that during ambient aerosol measurements, the interference in the H x Cl + signal from sea salt chloride (ssCl) was as high as 89 %, but with a study wide average of 10 %. The Na 35 Cl + ion signal was found to be a good tracer for NaCl. We outline a method to 20 establish nssCl in the ambient aerosols by subtracting the sea salt chloride (ssCl) signal from the H x Cl + signal. The ssCl signal is derived from the Na 35 Cl + ion tracer signal and the H x Cl + to Na 35 Cl + ratio established from the sea sweep experiments. Ambient submicron concentrations of ssCl were also established using the Na 35 Cl + ion tracer signal and a scaling factor determined through simultaneous measurements of submi- 25 cron aerosol chloride on filters. This scaling factor accounts for the low vaporization response of the AMS heater to ssCl, although regular calibration of this response is recommended in future applications. It follows that true total particulate chloride (pCl)

Introduction
Atmospheric aerosols can have an adverse effect on human health, and elevated 10 aerosol concentrations have been linked to increased morbidity and mortality (Lippmann et al., 2000;Cai and Griffin, 2006). Aerosols also impact the Earth's climate through direct and indirect effects on the radiative balance (Solomon et al., 2007) and can cause significant reduction of visibility in polluted areas (Watson, 2002) Inorganic aerosols typically constitute 25-50 % of the total aerosol submicron mass. 15 Particulate chloride (pCl) can be a major component in coastal and marine aerosols (Moya et al., 2002, and references therein). Sources of pCl are both primary and secondary in nature, where the former refers to sources that lead to direct emissions of pCl into the atmosphere and the latter occurs as a result of chemical and physical processes including gas to aerosol conversion (Pio and Harrison, 1987 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | tion of ammonium chloride (NH 4 Cl) in the solid (s) and aqueous (aq) phases. NH 4 Cl (s,aq) is in equilibrium with its gaseous precursors hydrochloric acid (HCl) and ammonia (NH 3 ),: HCl (g) + NH 3 (g) NH 4 Cl (s, aq) (R1) Regional sources of ammonia in California arise mainly from dairy farms and automo-5 biles that are equipped with three-way catalytic converters (Neuman, 2003;Docherty et al., 2008;Livingston et al., 2009;Nowak et al., 2012;Hersey et al., 2013). The formation of particle phase NH 4 Cl is dependent on temperature, relative humidity (RH), aerosol chemical composition, and the partial pressures of both HCl and NH 3 . The formation of NH 4 Cl (s,aq) is favorable under conditions of low temperatures and high RH 10 (Pio and Harrison, 1987;Wexler and Seinfeld, 1990;Matsumoto and Tanaka, 1996;Trebs et al., 2005;Ianniello et al., 2011). Furthermore, the partitioning of NH 4 Cl to the aerosol phase is reported to be dependent on the availability of excess NH 3 after the neutralization of H 2 SO 4 to form (NH 4 ) 2 SO4, since the affinity of NH 3 for H 2 SO 4 is higher than its affinity for HCl (Trebs et al., 2005;Ianniello et al., 2011). 15 HCl is emitted directly in the atmosphere from biomass burning, coal combustion and waste incineration (Andreae et al., 1996;McCulloch et al., 1999;Ianniello et al., 2011). Significant amounts of gaseous HCl are also produced from acid displacement of chloride from sea salt aerosols. This acid displacement is driven by nitric acid, sulfuric acid (Finalyson-Pitts and Pitts, 1999;Keene et al., 1999), nitric acid anhydride, 20 N 2 O 5 (McLaren et al., 2004) and organic acids (Laskin et al., 2012). The displacement by HNO 3 and H 2 SO 4 is particularly important during the daytime, since it involves the photo-oxidation of nitrogen and sulfur oxides respectively. At night, however, N 2 O 5 , formed from the reaction of NO 2 and NO 3 , can contribute significantly to acid displacement and subsequent formation of HCl. Some of the HCl can repartition into the 25 aerosol through reaction with ammonia and/or dissolve in hygroscopic aerosols directly in environments with high water content, such as marine environments, especially if the aerosols are non-acidic in nature (Keene et al., 1999;Kim et al., 2008 The HCl released during acid displacement (Finlayson-Pitts and Pitts, 1999; and references therein) is a significant sink of ssCl (Keene et al., 1999). The released HCl can lead to halogen activation since HCl can react with the hydroxyl radical, OH, to form chlorine atoms in the gas phase. Other secondary sources of HCl include the reaction of chlorine atoms with methane and other VOCs (Kim et al., 2008;Mielke 5 et al., 2011). Chlorine atoms are also formed from the photolysis of nitryl chloride, ClNO 2 , and molecular chlorine, Cl 2 Roberts et al., 2008;Riedel et al., 2012).
Aerosol mass spectrometry (AMS) has become a common method for continuous measurement of submicron aerosols . The AMS used in 10 this study contains a critical orifice and aerodynamic lens that allows sampling of submicron aerosols 70 to 700 nm (50 % vacuum aerodynamic cutoff diameter) in diameter at 1 × 10 5 Pa (Liu et al., 2007). Sub-micron aerosols impact a heater in the AMS that is typically set at 600 • C, which is not high enough to efficiently vaporize refractory (R) aerosol components such as sea salt, many mineral oxides, elemental carbon 15 (soot) and metals. For this reason, the AMS is not expected to measure refractory sea salt aerosols in a quantitative way whereas, most non-refractory (NR) aerosol species, such as organics and many inorganic salts (i.e.: NH 4 Cl, NH 4 NO 3 and (NH 4 ) 2 SO 4 ) are efficiently vaporized and ionized upon impact by an electron beam, with subsequent detection in the mass spectrometer. 20 Given these operational conditions, one would assume that literature reported aerosol mass loadings of submicron chloride, via the AMS, would only contain nonrefractory chloride and that refractory components such as NaCl should be absent since NaCl has a melting point of 800.7 • C and a boiling point of 1465 • C (Haynes, 2012), much higher than the vaporizer temperature of 600 • C. However in coastal en- 25 vironments, where there are both sea spray and non-sea spray sources of pCl, uncertainties exist in terms of exactly what is being reported since the vaporization of sea salt NaCl is not zero at 600 • C. In fact, NaCl signals using a HR-AMS have been observed in the South Atlantic marine boundary layer and found to correlate positively with wind  (Zorn et al., 2008). More recently, total sea salt concentration in the submicron range was quantified using an HR-AMS by utilizing the Na 35 Cl + ion signal as a sea salt surrogate (Ovadnevaite et al., 2012). Neither study, however, quantified ssCl. This paper reports measurements of aerosol chloride using the HR-AMS deployed 5 onboard the research vessel Atlantis during the CalNex study in 2010 (Ryerson et al., 2013). The CalNex campaign goal was to measure ambient aerosol concentrations, composition and microphysical properties along the coast of California (Cappa et al., 2012;Massoli et al., 2014). In addition, nascent sea spray aerosols were artificially generated through in situ bubbling of seawater during sea sweep experiments (Bates et al., 2012), and sampled by the HR-AMS. We detected high levels of refractory NaCl in the sea sweep aerosols, as well as in the ambient marine environment throughout the study. This work seeks to provide clarity on the reporting of particulate chloride using AMS instruments. In particular, we propose a way to correct pCl signals for the presence 15 of ssCl signals that can be abundant in areas impacted by sea salt aerosol, such that the more volatile non sea salt chloride (nssCl) can be reported more accurately. This study also proposes a method to establish the percentage of ssCl that is depleted from sea salt aerosols. Both methods utilize the Na 35 Cl + ion signal, but the latter method also utilizes the Na + ion signal. Finally, a method to establish total submicron chloride 20 concentrations is presented. Submicron chloride concentrations are established strictly from HR-AMS data and account for both ssCl and nssCl.

Experimental
The HR-AMS (Aerodyne Research Inc., Billerica, MA, USA) was deployed onboard the R/V Atlantis for the measurements of submicron aerosol concentration and composi-Introduction Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | edu/jimenez-group/ToFAMSResources/ToFSoftware/index.html). In this study we only report data collected in the V-mode. All data were averaged to 10 min to improve the signal to noise ratio, unless otherwise stated. Using the AMS, aerosol mass concentrations (C) for a species X is determined using the following equation (Jimenez et al., 2003;Canagaratna et al., 2007): where MW is the molecular weight of the species, Q is sampling flow rate, N A is Avogadro's number and IE is the ionization efficiency, a dimensionless quantity that measures the ionization and detection efficiency and is defined as the ratio of the ions detected to the parent molecules vaporized. The IE NO 3 was established through calibrations using monodispersed 300 nm diameter NH 4 NO 3 particles during the study. RIE x is the relative ionization efficiency of X relative to that of NO − 3 (Alfarra, 2004). I x, i refers to the ion count rate (in Hz) of ion fragment i that results from the ionization of X . When the RIE of X is not known nor utilized in Eq. (1), the value obtained is referred to as the nitrate equivalent concentration (NEC) (Jimenez et al., 2003;Canagaratna 15 et al., 2007), a proxy of signal intensity. For chloride, a RIE of 1.3 is commonly used (e.g. Lee et al., 2013;McGuire et al., 2014), based on the methodology by Jimenez et al. (2003) and Alfarra et al. (2004). In this study, we introduce and utilize the chloride equivalent concentration (CEC), a surrogate for measured mass loadings that utilizes a RIE of 1.3. This will be used for some species with unknown RIE values such as 20 NaCl. Finally, we define an ion group H x Cl + , which is the sum of ions typically used  .
Ambient submicron aerosol (50 % aerodynamic cut-off diameter, D a , < 1.1 µm at 25 60 % RH) was also sampled using one and two-stage multijet cascade impactors (Berner et al., 1979 and [Cl − ]) and other ions on the filters used in the impactor were determined using Ion Chromatography (IC) (Quinn et al., 1998). The time resolution of these filter measurements was 2-16 h with a frequency of 1-3 per day. In addition to ambient air aerosol measurements, Sea Sweep experiments were conducted periodically while at sea during the CalNex 2010 study to measure artificially generated nascent sea spray aerosols (Bates et al., 2012). Briefly, a Sea Sweep experiment was performed by bubbling clean air just below the ocean surface, while instruments sampled the sea salt aerosols created by bubble bursting above the surface in an enclosure. Ambient particles are prevented from entering the Sea Sweep enclosure with a curtain of particle-free air surrounding the enclosure. Hence, only freshly emit-

Mass spectrum of sea salt aerosols
An average aerosol mass spectrum obtained using the HR-AMS for the Sea Sweep experiments is shown in Fig. 1. The mass spectrum includes the ions for aerosol species typically measured and reported by the AMS including particulate organics (OM), sul-20 fate (SO 4 ), ammonium (NH 4 ) and nitrate (NO 3 ). Figure 1 also shows the H x Cl + ion group, which is typically utilized to determine chloride concentration using an AMS. Also included in the mass spectrum are ions observed that are typical of sea salt aerosol components namely, Na + , NaCl + , Mg + , MgCl + , K + and KCl + . Other ion signals (such as metal ions) were grouped under the "Other" category, while air and water ACPD 15,2015 Separating refractory and non-refractory particulate chloride is 0.32 : 1, in agreement with chlorine's natural isotopic abundances (Numata et al., 2001). The OM signals result from biogenic derived organic matter that is injected into the marine aerosol during bubble bursting (Bates et al., 2012). The dominance of ion signals associated with sodium chloride (i.e., Na + , H x Cl + and NaCl + ions) during the sea sweep experiments was similar to that observed by Ovadnevaite et al. (2012) dur-10 ing sampling of ambient marine aerosols. Even though sodium chloride is not efficiently detected by the AMS operating with a 600 • C vaporizer temperature, large ion signals are still apparent in the mass spectrum given the large amounts of sea salt sampled during these experiments. Even if only a small fraction of NaCl is vaporized, ionized and detected, this fraction appears dominant in the mass spectrum for the sea sweep ex- 15 periments. It is worth noting that the standard AMS data analysis reports total aerosol chloride (pCl) using the H x Cl + mass fragments outlined above, but does not include the NaCl + ions, as NaCl is assumed to be truly refractory. In default AMS applications, it is also assumed that aerosol chloride species are evaporated with unit efficiency. It should be obvious from the sea sweep signals in Fig. 1 that any report of non-refractory 20 chloride mass (nssCl) in areas that contain sea salt aerosols (such as coastal areas) will be compromised, largely due to the vast difference in vaporization efficiency of refractory and non refractory seasalt aerosols. Under those conditions when sea salt contributes to the H x Cl + signal, aerosol chloride (pCl) reported using standard assumptions will underestimate sea salt chloride and overestimate the non-refractory chloride 25 component.

Observation of sea salt chloride (ssCl) in ambient air
During the study, the Na + , Na 35 Cl + and Na 37 Cl + signals were low but detectable with well-resolved peaks indicating that sea salt was sampled and detected by the AMS. The Na 37 Cl + peak at 59.956, however, has significant overlaps with potential interfering ions, CSO + (59.967) and SiO + 2 (59.967), even at the high resolution of this AMS, 5 and was thus not used as a quantitative indicator of sea salt. Figure 2 shows  respectively. Note that these chloride equivalent concentrations should only be treated as relative signals and not true concentrations. The NaCl signal was characterized by slow evaporation in the AMS after sampling high sea salt aerosol during the sea sweep experiments. This slow evaporation led to 15 high background signal for certain ions, which was apparent from the chopper "closed" signal. In the AMS, the beam chopper alternates between blocking position, (closed) and transmitting position (open) of the aerosol beam. The aerosol signature is then properly quantified by subtracting the chopper-closed signal from the chopper-open signal). During this study, the background signals for Na + , H x Cl + and Na 35 Cl + were 20 much higher than the "open -closed" measurement signal after sea sweep experiments. Similar trends were observed in other studies for sodium and lead signals when using an AMS for sampling refractory aerosols (Salcedo et al., 2010;Ovadnevaite et al., 2012). In this study, we found that the time for complete disappearance of refractory sea salt signals was directly proportional to the amount of sea salt sampled and ranged  our observations. This high background signal resulted in a low signal to noise ratio for the ambient measurements that followed the sea sweep experiments; and so we have eliminated these time periods from further analysis. The difference in background signal recovery times for the different components of the same parent compound is not fully understood. 5 The observation of Na + and NaCl + signals in ambient aerosol measurements confirms the detection of refractory sodium chloride by the AMS. Thus, refractory chloride contributes to the H x Cl + ion family, which is typically used to calculate non-refractory chloride (i.e., nssCl in the form of HCl, NH 4 Cl or other volatile chlorinated species) mass concentrations via the AMS. It then follows that the contribution of this refractory 10 component to the H x Cl + signal needs to be corrected before reporting mass concentrations of nssCl. To do this, it is necessary to isolate the nssCl signal from the H x Cl + signal. In the following section, we present a method to establish nssCl concentration from H x Cl + and Na 35 Cl + signals. 15 In order to correctly report nssCl we need to estimate the refractory component of H x Cl + , which we define to be the chloride portion of the H x Cl + signal that is in the form of NaCl. For this estimate, we use an ion (I) that is not a part of the H x Cl + group that results exclusively from the fragmentation and ionization of NaCl. In addition, we establish the H x Cl + /I ratio for NaCl in sea salt aerosol using the HR-AMS. For I, one 20 can consider using Na + or Na 35 Cl + , but while Na 35 Cl + is expected to result exclusively from the ionization of NaCl, Na + could also additionally arise from chloride depleted sea salt aerosol in the form of NaNO 3 and/or Na 2 SO 4 . This is especially true with aged sea salt aerosols, where acid displacement can remove chloride (see Finlayson-Pitts and Pitts, 1999 In order to prove the validity of using the Na 35 Cl + signal as an indicator of sea salt, ambient Na 35 Cl + signal measured using the HR-AMS was compared to ambient submicron aerosol mass concentrations from the filters, which were continuously available during most of the study (Fig. 3). The Na 35 Cl + signal was averaged to the collection time of the filters for this analysis. The total sea salt concentration in Fig. 3 was cal-5 culated from the filter measurements using the following equation (Bates et al., 2012, and references therein):

A method to establish nssCl with sea salt chloride interferences
where [Cl − ] and [Na + ] are the ambient mass concentrations of the Cl − and Na + ions measured on the filters and the coefficient 1.47 is the mass ratio of major non-chloride 10 ions (Na + , K + , Mg +2 , Ca +2 , SO 2− 4 , HCO 2− 3 ) to Na + ions present in sea salt. From  Fig. 3, the mass concentrations of Na + , Cl − and total sea salt are all positively correlated with the HR-AMS Na 35 Cl + signal with correlation coefficients, R 2 , ranging from 0.83 to 0.89 and slopes as indicated in the figure. These correlations suggest that the Na 35 Cl + signal from the HR-AMS can indeed be reasonably used as a tracer for sea 15 salt aerosol in ambient air. The source of scatter in the graph may be a result of the low time resolution of the filter measurements with sampling frequency of 1-3 per day and sampling duration of ∼ 2-16 h and the different particle size cutoff for both measurements. Note that the slopes are lower limits as the aerosols sampled on filters had a cutoff of D a < 1.1µm, while the HR-AMS cutoff size is D a < 700 nm. The HR-AMS 20 only sampled a fraction of the sea salt aerosols that were collected on filters. Figure 4  The 95 % confidence intervals of the regression line, calculated using the standard errors of the slope and intercept, as well as the critical t value at probability (p) of 0.05 and 130 • of freedom (n−2), are also shown in Fig. 4 where the term in parentheses refers to the refractory Cl component of the ambient H x Cl + signal. Note that for the above equation, the intercept was forced to zero and only the H x Cl + to Na 35 Cl + ratio was used. A zero intercept is within error and both the H x Cl + and Na 35 Cl + signals are from the same source in sea sweep experiments (correlation is improved with a forced zero intercept, R 2 = 0.95). Using the last term in 15 Eq.
(2), we can quantify the contribution of refractory H x Cl + to the total H x Cl + signal, i.e. how much of the H x Cl + signal can be attributed to NaCl. This contribution is estimated by (3.9 × Na 35 Cl + )/H x Cl + × 100 %, and the result of this estimation is shown in Fig. 5. Also shown in Fig. 5 is the submicron chloride concentration from filter measurements, which primarily detects NaCl, but not nssCl, due to evaporation issues of volatile chlori-20 nated compounds (such as NH 4 Cl) from the filters. Using the Na 35 Cl + ion fragment and the factor of 3.9, the refractory H x Cl + signal 25 detected by the HR-AMS was low in ambient air during this study. However, the contribution of refractory H x Cl + to the total H x Cl + signal in ambient air (not sea sweeps) was ACPD 15,2015 Separating refractory and non-refractory particulate chloride Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | significant with a maximum contribution of 89 % and an average contribution of 10 %. In this study, periods with high refractory contributions to the H x Cl + signal were also periods of relatively low H x Cl + signals that were not enriched with nssCl.

A method to establish ssCl and pCl from AMS measurements
In this section we will propose an empirical formula to estimate the air concentrations 5 of submicron sea salt chloride ssCl and total submicron ambient chloride pCl. In order to estimate ssCl, we use NaCl + and a scaling factor established using [Cl − ] from filters.
The ratio of submicron chloride from filters to Na 35 Cl + (Fig. 3) was found to be 98.9 ± 6.7, Sea salt chloride concentration is therefore estimated to be: ssCl = 98.9 × Na 35 Cl + assuming that all of the chloride measured on the filters is in the form of NaCl. This assumption is reasonable, since the filter measurements employed here do not detect volatile chlorinated species as mentioned previously. Now that we have a measure of nssCl and ssCl, pCl can be estimated using the following equation: pCl(µg m −3 ) = nssCl + 98.9 × Na 35 Cl + (5) 15 where the first term is nssCl calculated using Eq. (3) and the second term is ssCl calculated using Eq. (4). Figure 6 shows the time series of pCl, nssCl and ssCl concentrations measured by the HR-AMS, as well as submicron [Cl − ] from filters. The lack of sensitivity of the filter measurements to nssCl is apparent from Fig. 6 as an increase in nssCl does not coincide with an increase in [Cl − ]. In this study, the median levels Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | measurements due to its high temporal resolution and since filter measurements may not be sensitive to semi-volatile NH 4 Cl and other volatile chlorinated compounds, as seen in this study. they had used Na 35 Cl + , which is 5.6 times lower than the scaling factor reported here. This difference could be attributed to the higher vaporizer temperature utilized by Ovadnevaite et al. (2012), 650 • C, compared to 600 • C used in this study. Figure

Determining chloride depletion in sea salt using AMS data
We now present a method to establish the percentage of chloride depleted from submicron sea salt in ambient air using HR-AMS ion signal measurements exclusively. Chloride depletion is an important indicator of sea salt aerosol aging and the formation of activated chloride in the form of HCl. Chloride depletion is typically calculated using ACPD 15,2015 Separating refractory and non-refractory particulate chloride Note that this ratio should be viewed as a signal ratio, since chloride equivalent concentrations are used as ion signals in this study. As such, the above ratio cannot be directly compared to the true [Cl − ] to [Na + ] ratio of 1.81. The difference between the two ratios is due to different ionization efficiencies in the HR-AMS for chloride and sodium. 15 Using the aforementioned estimates for expected and observed chloride, the HR-AMS estimate of submicron sea salt chlorine depletion can be written: % Cl depletion = 0.827 × Na + − 3.9 × Na 35 Cl + 0.827 × Na + (7) Figure 9 shows the time series of the percentage chloride depletion in submicron aerosols during the CalNex study for both ambient aerosols and sea sweep experi-20 ments. For ambient air, the percentage chloride depletion was calculated using two methods. The first uses the HR-AMS measurements and Eq. (7) to establish chloride depletions using sea sweep experiments, since filters were not available for these experiments. Hourly averages of the HR-AMS data were used to reduce noise in the Na + signals. From Fig. 9, the chloride depletion during the sea sweep experiments was very close to zero, which is expected since fresh sea salt aerosols were sampled during these experiments and the ratios used in Eq. (7) were 5 derived from these experiments. For ambient data, the HR-AMS method gives reasonable results with ∼ 95 % of the data in the 0-100 % range, ∼ 5 % are less than zero and are attributed to the noise associated with the HR-AMS. Both methods show similar trends (Fig. 9) and are positively correlated, which provides confidence to the method proposed here. The HR-AMS method captures more variations due to the higher time 10 resolution of the measurements. The median chloride depletion observed in this study (HR-AMS method) was 78 % with 5 and 95 percentiles of 0 and 96 %. It is important to note that the method proposed here only considers chloride depletion from NaCl and does not take into account other chlorinated species in sampled aerosols.

15
This study attempts to provide some clarity and presents new methods in the reporting of aerosol chloride with the aerosol mass spectrometer. The issue that arises in the interpretation of different AMS datasets is that the chloride signal, typically measured via the H x Cl + ion group, has contributions from both non-refractory (i.e., NH 4 Cl, HCl, organic chlorides, etc.) and refractory chloride species (i.e., NaCl) albeit with vastly 20 different sensitivities. The AMS is a couple orders of magnitude less sensitive to the detection of refractory sea salt chloride (ie. ssCl) compared to non-refractory chloride at typical operational heater temperatures (600 • C) due to the high boiling point of NaCl. However, the atmospheric loading of ssCl can be much higher than non sea salt chloride (nssCl) in environments condusive to the production of sea salt aerosol. To 25 properly report submicron aerosol chloride in such environments, one needs to know what fraction of the H x Cl + ion group signal arises from the two different types of compo-ACPD 15,2015 Separating refractory and non-refractory particulate chloride nents. In this study we use the Na 35 Cl + ion signal as the tracer for detection of ssCl to help us separate the two signal components and to report separately nssCl and ssCl, as well as total chloride (pCl), which is the sum of nssCl and ssCl. We used data from sea sweep experiments when it was known that aerosols being sampled were 100 % sea salt, to calibrate the H x Cl + /Na 35 Cl + signal ratio (3.9 ± 0.2) 5 for our operational conditions. The nssCl is then obtained by subtraction of the sea salt portion of the H x Cl + signal (Eq. 3). During ambient aerosol measurements in this study, we found that the sea salt chloride contribution to the H x Cl + signal averaged ∼ 10 %, with a maximum of ∼ 90 %. The contribution is expected to be even higher in marine areas with stronger wave breaking and in areas where nssCl levels are low.

10
The ssCl (Eq. 4) was obtained via calibrations of the [Cl] filter /Na 35 Cl + ratio (98.9 ± 6.7) from ambient measurements with the assumption that the filter measurements are not sensitive to non refractory chloride (i.e. NH 4 Cl), which evaporates from the filters. The true submicron total chloride is then obtained by adding ssCl and nssCl (Eq. 5). The two ratios referred to above that were established in this study are more 15 than likely, highly dependent on AMS operating conditions and should be calibrated with individual instruments and conditions. In particular, small changes in the AMS heater temperature, either through changing the filament current or due to instrument temperature drift, are known to result in large changes in the evaporation efficiency of refractory sea salt components. Regular calibrations of these ratios should be part of 20 future applications. Similar to the determination of ssCl, we estimated the total sea salt aerosol concentration in ambient air by multiplying the sea salt surrogate signal, Na 35  Finally, this study proposes a method to establish the relative amount of chloride depletion from NaCl in ambient sea salt aerosol. This method strictly uses HR-AMS measurements and utilizes H x Cl + /Na + and H x Cl + /Na 35 Cl + ratios in sea salt aerosols established from the sea sweep experiments. Using this method, the percentage depletion of chloride in submicron sea salt aerosols was observed to vary between 0-100 % 5 for 95 % of the data, with outliers existing due to instrumental noise. The median submicron sea salt aersosol chloride depletion during the whole study was found to be quite high, 78 %. This is likely attributable to the fact that the measurements during CalNex were taken in California coastal polluted regions characterized by high NO x sources and frequent land breeze outflow events.  . Time series for percentage chloride depletion in submicron aerosols calculated from AMS measurements using Eq. (7). Also shown for comparison is chloride depletion calculated from filter measurements using Eq. (6).