Environ. Sci. Technol. 2001, 35, 600-605
Experimental Study of the Sampling Artifact of Chloride Depletion from Collected Sea Salt Aerosols XIAOHONG YAO,† MING FANG,‡ AND C H A K K . C H A N * ,† Department of Chemical Engineering and Institute for Environment & Sustainable Development, The Hong Kong University of Science & Technology, Clear Water Bay, Kowloon, Hong Kong
optical properties and thus the radiation transfer characteristics of atmospheric aerosols (5). The reactions also provide a sink of atmospheric trace gases. Furthermore, the formation of Cl from the chloride depletion reactions can destroy O3 (6). The sea salt Cl- concentration is defined as (7)
[Cl-ss] ) 1.174[Na+meas]
where [Na+meas] is the measured Na+ equivalent concentration. The percentage of chloride depletion of sea salt aerosols can be estimated as
[Cldep] (%) )
Sampling artifact of chloride depletion from collected sea salt particles was studied, based on simultaneous measurements of size distribution measurements by a 10stage Micro-Orifice Uniform Deposit Impactor (MOUDI) and of PM2.5 measurements by a Compact Porous Metal Denuder Sampler (PMDS) at a coastal site in Hong Kong on May 7, 8, 9, 11, and 29, 1998. The ambient concentrations of SO2, HNO3, HNO2, and NH3 were also measured by the PMDS. PM2.5 measurements by the PMDS, which is equipped with denuders and nylon back filters, are compared with the PM1.8 and PM3.1 measurements by the MOUDI. The percentages of chloride depletion from sea salt aerosols in PM1.8 and PM3.1 were 4-45% higher than that in PM2.5. This suggests that chloride evaporation in PM1.8 and PM3.1 collected on Teflon filters of the MOUDI during sampling was present. From the sum of the contributions of particles on the Teflon and nylon filters of the PMDS, nitrate formation almost completely accounts for chloride depletion in PM2.5 prior to collection since the equivalent ratio of [Na+] to ([NO3-] + [Cl-]) is close to the seawater ratio of 0.85. However, it was found that 2274% of nitrate and 45-86% of chloride in the collected particles on the Teflon filter of the PMDS evaporated during sampling. Excess chloride depletion unexplained by NO3- and nss-SO42- was found in the collected particles on the Teflon filter of the PMDS. Similarly, an amount of 3.727.2 nequiv m-3 of excess depleted chloride (equivalent to 8-55% of total chloride depletion) was found in supermicron particles collected by the MOUDI. In the 1.8-3.1 µm particles, the excess depleted chloride is positively correlated to the chloride evaporated from the deposited particles. Particle-particle interactions are proposed to explain the evaporation of nitrates and chlorides in the PMDS and MOUDI measurements. The observed chloride depletion from seasalt aerosols was partially attributed to sampling artifact due to particle-particle interactions.
Introduction Chloride depletion reactions in sea salt aerosols have been observed in many field measurements (1-4). The substitution of the sea salt chloride by nitrate and sulfate changes the * Corresponding author phone: (852)2358-7124; fax: (852)23580054; e-mail:
[email protected]. † Department of Chemical Engineering. ‡ Institute for Environment & Sustainable Development. 600
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 3, 2001
(1)
[Cl-ss] - [Cl-meas] [Cl-ss]
× 100%
(2)
where [Cl-meas] is the measured Cl- equivalent concentration. The mechanisms of the chloride loss reaction have been widely studied (3, 6, 8-12). The most commonly discussed reaction is the reaction with acids, followed by the evaporation of HCl to the gas phase. Much attention has been given to the formation of NO3-, nss-SO42- (2, 4), and more recently, organic anions (3). Reactions involving the formation of other Cl compounds such as NO2Cl (9, 10), Cl2 and HOCl (6, 11), and BrCl (12, 13) have been reported in the literature. The rate of these reactions strongly depends on the liquid water content of the particles (10). Little reaction of dry sea salt particles has been found (6, 10). The size dependence of chloride depletion has been studied by various researchers (2, 4, 8, 14-16). Most results show that the percentage of chloride depletion decreases with increasing particle size. The thermodynamics and kinetics of HCl phase partitioning (15, 16), the dynamics of the surface reaction such as HNO3 on the sea salt particles, and in-cloud processes (3, 17-19) have been used to explain the size dependence of chloride depletion in sea salt aerosols. So far, most research on chloride depletion has focused on atmospheric reactions, although sampling artifacts of atmospheric aerosols have been generally observed in field measurements. The positive artifacts are possibly originated from the gas-particle interactions, and the negative artifacts are usually a result of the decomposition of semivolatile species such as NH4Cl and NH4NO3 and possible particleparticle interactions (20-22). Using laboratory generated aerosols, Cohen et al. (23) studied the water cycles of aqueous mixed droplets of NaCl and (NH4)2SO4 in an electrodynamic balance and found that 10% of the solute evaporated in a 72-h span. They proposed that the most likely species to volatize are NH3 and HCl. Since atmospheric sulfate particles are likely to be incompletely neutralized, their potential in displacing chloride from sea salt particles may be even higher. In this paper, we investigate the possibility of chloride depletion as a result of particle-particle interactions in sampling. In particular, we will compare the aerosol compositions measured by a Micro-Orifice Uniform Deposit Impactor (MOUDI) (24) and by a Compact Porous Metal Denuder Sampler (PMDS) (25). Chloride depletion from particles collected on the Teflon filter was found in both the MOUDI and the PMDS measurements.
Measurements of Aerosol Composition and Gaseous Concentrations Experiment. The sampling site was located at the Hong Kong University of Science and Technology at Clear Water Bay, a sparsely populated area on the eastern coast in Hong Kong. 10.1021/es000964q CCC: $20.00
2001 American Chemical Society Published on Web 12/30/2000
Sampling was conducted on the roof of the academic building, which is approximately 20 m above the ground, 130 m above sea level, and 500 m from the bay. The site is also about 500 m from a lightly trafficked road and 5 km from the closest industrial area. The sampler inlet was about 1.5 m high and far from any obstacles. The prevailing wind direction was east or southeast during sampling. Sampling was conducted for 24 h on May 7-9, 11, and 29 in 1998. A 20-mm rainfall was recorded on May 10 before the sampling on May 11. A small rainfall (2 mm) was recorded during sampling on May 9. Size distributions of water-soluble ions have been studied at this site (4, 26). Aerosols were collected using a 10-stage MOUDI-100 (MSP Corp.) with 50% cut sizes of 18 (inlet), 9.9, 6.2, 3.1, 1.8, 1.0, 0.54, 0.32, 0.177, 0.097, and 0.052 µm and an after-filter at a flow rate of 30 L min-1. Marple et al. (24) characterized the cutoff characteristics of the MOUDI in detail. The substrates used in the MOUDI were Teflon membrane filters 47 mm in diameter with 2 µm pores (Teflo R2PJ047, Gelman Sciences). The after-filter was 37 mm in diameter with 2 µm pores (Teflo R2PJ037, Gelman Sciences). The PMDS was designed and characterized by Poon et al. (25). It has a 50% 2.5 µm cut impactor upstream of four stages of porous metal pieces coated with different chemicals to absorb HNO3, HNO2, SO2, and NH3, followed by a tandem of Teflon and nylon filters for PM2.5 sampling. The first denuder stage was coated with 0.1% NaCl in a 90% H2O/10% methanol mixture to collect HNO3 and SO2. The second and third stages were coated with 1% Na2CO3/1% glycerol in a 50% H2O/50% methanol mixture to absorb HNO3, HNO2, and SO2. The fourth stage was coated with a 4% citric acid and 2% glycerol-methanol solution to absorb NH3. The collection efficiencies of SO2 and HNO3 in the PMDS have been reported to be 99.6 ( 2.5% and 93.4 ( 2.5%, respectively. The loss of particles in the size range of 0.1-2.5 µm is less than 3%. The mean differences between the measured concentrations of the PMDS and of the annular denuder system (27) were 2.3%, 6.1%, and 2.9% for particulate nitrate, sulfate, and mass, respectively (25). The PMDS was operated at a flow rate of 10 ( 0.5 L min-1, calibrated by a DC-I flow Calibrator (BIOS, USA). Because of the high efficiency of the coated denuders in absorbing gases before particle collection, gas-aerosol interactions are not significant in the PMDS. Chloride and nitrate evaporated from the collected particles on the Teflon filter are collected by the nylon filter (22). Although NO2 might react with the nylon filter and result in the positive artifact of nitrate because the coated denuders cannot efficiently absorb NO2, there is no report of such reactions in the literature. To extract the water-soluble inorganic ions including SO42-, NO3-, Cl-, Ca2+, Mg2+, Na+, K+, and NH4+ from the Teflon filters, 0.2 mL of methanol was used to prewet the filters, and 10 mL of ultrapure water (specific resistance g 18.1 MΩ-cm) was used in ultrasonic bath for 20 min. To extract Cl- and NO3- from nylon filters, 10 mL of anion eluent (1.8 mM Na2CO3 + 1.7 mM NaHCO3) was used (27). The ionic concentrations of the aqueous extract were determined by ion chromatography (Dionex LC20) with an electrochemical detector (ED 40). An AS4A-SC (4 mm) column and an eluent of 1.8 mM Na2CO3 and 1.7 mM NaHCO3 were used for anion detection. A CS12 column and an eluent of 20 mM MSA were used for cation detection. The detection limits (in nequiv m-3) are 0.2 for SO42-, 0.1 for NO3-, 0.2 for Cl-, 0.6 for Ca2+, 0.4 for Mg2+, 0.6 for Na+, 0.1 for K+, and 0.4 for NH4+. Uncertainties of SO42- and NH4+ concentrations in PM1.8, PM3.1, and 1.8-9.9 µm particles, calculated from those of each stage, were (5%. Uncertainties of Na+, Cl-, and NO3- concentrations were (10% in PM1.8, (6% in PM3.1, and (6% in 1.8-9.9 µm particles. The uncertainties of SO42-, NH4+, Na+, Cl-, and NO3- concentrations in PM2.5 are (8%.
TABLE 1. Measured Gaseous Concentrations and Meteorological Conditions during Sampling
date May 7 May 8 May 9 May 11 May 29
HNO3 HNO2 SO2 NH3 (ppbv) (ppbv) (ppbv) (ppbv)
T (°C)
wind wind RH direction speed (%) (deg) (km h-1)
0.07 0.04 0.03 0.08 0.03
30.3 29.0 27.7 31.0 30.4
77 86 92 77 72
0.24 0.14 0.18 0.74 0.16
0.51 0.51 0.57 1.88 0.40
2.30 1.55 1.57 2.42 1.39
70 90 100 120 130
16.1 16.8 23.7 9.4 13.9
TABLE 2. Concentrations of Water-Soluble Ions and Percentages of Chloride Depletion in PM1.8, PM2.5, and PM3.1
date May 7
size
PM1.8 PM2.5a PM3.1 May 8 PM1.8 PM2.5a PM3.1 May 9 PM1.8 PM2.5a PM3.1 May 11 PM1.8 PM2.5a PM3.1 May 29 PM1.8 PM2.5a PM3.1 a
SO42NH4+ Na+ NO3Cl(nequiv (nequiv (nequiv (nequiv (nequiv [Cldep] m-3) m-3) m-3) m-3) m-3) (%) 92.5 96.3 99.6 112.5 110.0 117.7 100.0 109.0 106.7 51.9 51.0 54.6 170.6 174.4 189.8
71.1 52.8 71.7 82.2 101.1 82.2 60.6 58.3 61.1 41.7 35.6 41.7 113.9 107.8 115.6
16.1 21.7 35.2 17.8 32.6 40.0 19.1 39.6 56.5 11.3 22.2 21.7 7.8 15.7 17.4
3.9 12.3 7.3 5.5 23.4 16.1 8.1 15.6 18.7 4.0 11.3 10.6 1.1 12.4 7.1
3.7 10.4 5.6 6.2 14.6 16.1 6.5 32.7 21.7 2.5 16.3 4.5 1.1 6.2 4.2
80.7 59.3 86.4 70.5 61.9 66.0 71.3 30.0 67.5 81.0 37.5 82.4 87.8 66.4 79.4
For particles collected on both Teflon and nylon filters.
Results of Measurements Table 1 lists the meteorological conditions during sampling and the measured gases concentrations. HNO3, HNO2, SO2, and NH3 are in the ranges of 0.03-0.08, 0.14-0.74, 0.401.88, and 1.55-2.42 ppbv, respectively. They are close to rural concentrations (28, 29) and are much lower than concentrations found in urban areas (29-31). The concentrations of NH3 are close to typical coastal concentrations (32). Because a 50% cut size of 2.5 µm is not available in the MOUDI, we compared the PM2.5 measurements in the PMDS with the PM1.8 (summing up the particles collected on and below the 1.8 µm cut stage) and PM3.1 measurements in the MOUDI, as shown in Table 2. Equivalent concentrations are used for all water-soluble ions in this paper unless otherwise stated. The total contributions from both the Teflon and nylon filters were used to calculate the concentrations of water-soluble ions in PM2.5. As a reference, the measured Na+ and Clconcentrations in PM2.5 are between the observed values in Long Beach (an urban coastal site in Los Angeles) and San Nicolas Island (a clean coastal site near Los Angeles) (33). The measured SO42- and NO3- concentrations in PM2.5 are much higher than the background in North Pacific atmosphere (8, 34). They are close to those observed at San Nicolas Island but are smaller than those at Long Beach (33). Overall, our sampling site is principally affected by sea salt aerosols and is slightly polluted by regional pollution. The difference in SO42- concentrations among PM1.8, PM2.5, and PM3.1 is less than 10% because SO42- is a nonvolatile species and is principally distributed in submicron particles. The difference in NH4+ concentration among PM1.8, PM2.5, and PM3.1 is less than 20%. However, the differences of Cl- and NO3- concentrations are much larger. VOL. 35, NO. 3, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
601
In general, the Na+ concentration in PM3.1 is at least 10% larger than in PM2.5. An exception was observed on May 11, which can be explained by the sampling loss of supermicron particles in the MOUDI (24). Howell et al. (34) reported a 50% loss of supermicron particles in the MOUDI (see Figure 3 of ref 34), based on comparison of measurements with a Sierra impactor. In contrast, the PMDS has less than a 3% loss of particles between 0.1 and 2.5 µm (25). Na+ loss in PM3.1 on other days may also exist although the Na+ concentrations in PM3.1 were larger than in PM2.5 on these days. Unlike Na+, the NO3- concentrations in PM3.1 were 6-47% smaller than in PM2.5 except on May 9. The difference in PM3.1 and PM2.5 concentrations can be attributed to the evaporation of nitrate from the collected particles (20-22, 35) and sampling loss. Similarly, the Cl- concentrations in PM3.1 were 33-72% smaller than in PM2.5, except on May 8. However, even if a 50% sampling loss of Cl- in supermicron particles were assumed in the MOUDI measurements, the Cl- concentrations in PM3.1 remained 4% and 55% smaller than those in PM2.5 on May 7 and 11, respectively. Hence, it is concluded that additional chloride depletion exists in PM3.1. The percentages of chloride depletion in PM1.8 and PM3.1 were calculated by eq 2. The percentage of chloride depletion in PM2.5 prior to collection was calculated by
[Cldep]PM2.5 (%) )
1.174[Na+meas]Teflon - ([Cl-meas]Teflon + [Cl-meas]nylon) 1.174[Na+meas]Teflon
× 100% (3)
where [Na+meas]Teflon and [Cl-meas]Teflon are the Na+ and Clconcentrations on the Teflon filter, and [Cl-meas]nylon is the Cl- concentration on the nylon filter. As listed in Table 2, the percentages of chloride depletion in PM1.8 and PM3.1 were greater than those in PM2.5, especially on May 9 and 11. It suggests that chloride has evaporated from the deposited particles in PM1.8 and PM3.1. Chloride evaporation from PM2.5 and PM10 collected on Teflon filters in a dichotomous sampler has been reported by Tsai and Perng (22). On May 11, much lower SO42- concentrations in PM1.8, PM2.5, and PM3.1 were found. However, the SO2 and HNO2 concentrations were about 4 times higher than other sampling days. A 20-mm rainfall, which partially scavenged gaseous species and removed the aerosol particles, was recorded on May 10 before sampling on May 11. The percentage of chloride depletion in PM2.5 on May 11 was only 37.5% because of continuous replenishment of fresh sea salt particles.
Chloride Depletion from Particles Collected on Teflon Filter in PMDS Measurements Role of Nitrate in Chloride Depletion of PM2.5. In PM2.5, nitrate can associate with ammonium, sea salt, and crustal aerosols. We will first investigate if NH4NO3, which could evaporate even when collected on Teflon filters, exists in the atmosphere prior to sampling. In the PMDS measurements, the total nitrate concentrations [TN] and the total ammonium concentrations [TA] are
[TN] ) [HNO3(g)] + [NO3-]
(4)
[TA] ) [NH3(g)] + [NH4+]
(5)
and
where [HNO3(g)] and [NH3(g)] are the gaseous nitric acid and ammonia concentrations and [NO3-] and [NH4+] are the 602
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 3, 2001
FIGURE 1. Chloride depletion in PM2.5: (a) products to form ammonium nitrate for PM2.5; (b) ratio of [Na+]/([Cl-] + [NO3-]); and (c) percentages of chloride and nitrate evaporation in PM2.5. particulate nitrate and ammonium concentrations in PM2.5. The existence of particulate ammonium nitrate can be investigated by comparing [TN]‚[TA] with Kp, the equilibrium constant of the dissociation reactions of ammonium nitrate (36). The dissociation reactions of ammonium nitrate can be described as
NH4NO3(s) S NH3(g) + HNO3(g) and
NO3- + NH4+ S NH3(g) + HNO3(g) If [TN]‚[TA] < Kp, no ammonium nitrate will be present. The measured product [TN]‚[TA], Kp,avg (calculated at the average temperature and RH during sampling), and Kp,min (calculated at the temperature and RH most favorable for NH4NO3 formation during sampling) are compared in Figure 1a. The measured product [TN]‚[TA] was generally smaller than Kp,avg and Kp,min except on May 8 and May 9 when [TN]‚[TA] was close to Kp,min. Therefore, ammonium nitrate is not expected to be present in PM2.5 prior to collection mainly because of the ambient high temperature. The nylon filter does not absorb gaseous NH3, and thus the amount of NH3 evaporated from particles on the Teflon filter cannot be determined. It is possible that a negative artifact of NH4+ existed in PM2.5 but was not detected. Assuming that the evaporation of nitrate and chloride from the collected particles on the Teflon filter was accompanied by the evaporation of ammonium, [TA] would increase by 5-20%, which would still make [TN]‚[TA] generally smaller than Kp,min. Since NH4NO3 does not exist, the perturbation of gas-particle equilibrium as a result of the use of denuders would not promote additional evaporation of HNO3 from collected particles. Furthermore, [Ca2+] (not listed) is less than 5% of [Na+] in PM2.5; therefore, the chloride and nitrate associated with Ca2+ in PM2.5 are negligible. Hence, the nitrate found in the nylon filter is attributed to nitrate
TABLE 3. Concentrations (nequiv m-3) of Chloride Depleted and Ions in PM2.5a date
[Clloss]Teflon
[NO3-loss]Teflon
[Clloss]Teflon,ex
[nss-SO42-meas]
[NH4+meas]
May 7 May 8 May 9 May 11 May 29
4.7 8.1 23.5 13.8 5.3
2.6 6.2 4.5 3.8 9.0
