Environ. Sci. Technol. 2008, 42, 793–799
CCN Activity and Hygroscopic Growth of Organic Aerosols Following Reactive Uptake of Ammonia E. DINAR,† T. ANTTILA,‡ AND Y . R U D I C H * ,† Department of Environmental Sciences, Weizmann Institute of Science, Rehovot 76100, Israel, and Research and Development, Finnish Meteorological Institute, 00101 Helsinki, Finland
Received July 27, 2007. Revised manuscript received November 20, 2007. Accepted November 20, 2007.
Recent field observations suggest that ammonium salts of organic acids may be very important in accounting for aerosols’ properties in many environments. In this study we present laboratory experiments and calculations on the influence of ammonia reaction with organic aerosol components and its effect upon their (1) subsaturation hygroscopic growth (HG) and (2) supersaturation cloud condensation nuclei (CCN) activity. By using adipic acid (slightly soluble), citric acid (soluble), and di(ethylene glycol) monovinyl ether (DEGMVE, nonacidic compound) aerosols we show the feasibility and importance of atmospherically relevant acid–base neutralization by ammonia for different organic species. It is suggested that the formation of ammonium salts due to reaction of ammonia with slightly soluble organic acids (such as adipic acid) can affect the CCN activity and hygroscopic growth of aerosols with a significant organic component. It is further confined that the reaction involves carboxylic groups, it requires presence of water in the aerosol, and that the effects are stronger for less soluble organic acids.
Introduction Ammonia(NH3) is the dominant volatile base in the atmosphere. As such, it has an important role in atmospheric chemistry. Ammonia’s atmospheric concentrations have increased during the last two decades, spanning 6 orders of magnitude depending on location, sources, and ambient conditions. Background levels of ammonia are less than 50 ppt in remote areas (1), and up to 8 ppb ammonia and about 5 ppb of particulate ammonium were measured in the Amazon during the biomass burning season (2). Moreover, higher values of ammonia are usually measured near locations with extensive farming with concentrations reaching a few ppm (3). In the last two decades ammonia concentration in urban environments has also increased due to over-reduction of nitrogen oxide compounds in catalytic converters placed in automobiles exhaust and industrial and power station chimneys (4). Ammonia concentrations often exhibit strong daily and annual variations which depend mainly on temperature, relative humidity (RH), rainfall, and * Corresponding author email:
[email protected]. † Weizmann Institute of Science. ‡ Finnish Meteorological Institute. 10.1021/es071874p CCC: $40.75
Published on Web 01/04/2008
2008 American Chemical Society
winds. Ammonia governs the degree of neutralization of acidic inorganic species present in aerosols (5) by reacting with sulfuric (H2SO4) and nitric (HNO3) acids, to form particulate ammonium sulfate (and its derivatives) and ammonium nitrate salts (6, 7). The resulting ammonium salts substantially influence the chemical and physical properties of the aerosol in addition to shifting the thermodynamic equilibrium between the gaseous and the condensed phases of the precursors (7). Under very clean conditions, ammonia may also be involved in new particle formation through ternary nucleation with water and sulfuric acid (8–10). A decade ago it was thought that ammonia concentrations are too low to fully neutralize the acidity from sulfur dioxide (SO2) and nitrogen oxides (NOx) emissions (5). Today, due to an increase in atmospheric ammonia concentrations (6) and efficient reductions in SO2 and NOx emissions, ammonia levels may be the main factor that controls PM2.5 levels (7). The chemical and physical properties of aerosols determine processes such as cloud droplet formation, modification of cloud droplet properties, heterogeneous chemistry, scattering and absorption of solar radiation, and visibility degradation (11–13). Recent laboratory and modeling studies emphasize the importance of the organic fraction in overall particle properties, and hence their role in these atmospheric processes (14–21). Bilde and Svenningsson discussed that addition of any small fraction of soluble material will increase the hygroscopicity of insoluble aerosols (22). The organic component in fine atmospheric aerosols ranges from ∼20 to 90 wt% (21, 23–26), and it is composed of hundreds to thousands of individual species (23, 27) which cover a wide range of molecular forms, solubility, reactivity, and physical properties. This large variety renders complete characterization impossible. Among the organic species aerosols contain C1 to C34 organic acids that are considered to be the main contributors for aerosols acidity in regions where inorganic acids are scarce (26). Although much is known about the role of ammonia in neutralizing inorganic acids (7), less is known about neutralization of organic acids in aerosols by ammonia and its possible effects on their hygroscopic properties. Field observations and modeling studies suggest that ammonium salts of organic acids must be included in aerosol ion balance calculations (9, 10, 19). Mircea et al. (19) and Metzger and Lelieveld (28) recently showed that accounting for ammonium organic salts in the aerosol inventory is necessary for accurate prediction of aerosol hygroscopic growth and CCN activity. Specifically it was shown that accounting for the formation of ammonium adipate can have a significant effect on the growth factor, as adipic acid (consist of slightly less than 20% by mass) but the solubility of ammonium adipate is about 20 times that of adipic acid (19). In this study we focus on the possible changes in hygroscopic properties of aerosols containing organic acids following interaction with gas-phase ammonia. Organic aerosols were exposed to ammonia under different conditions (concentration and relative humidity) and the changes in hygroscopicity under supersaturation (SS) conditions (CCN activity) and at subsaturation conditions (hygroscopic growth) were monitored. We use three surrogates for atmospheric compounds to probe the basic processes that possibly occur in organic aerosols: adipic acid which was found to be a major part of biomass burning aerosols and other organic aerosol, citric acid, which is also found in aerosols, and finally di(ethylene glycol) monovinyl ether (DEGMVE), a surrogate for multifunctional organics that do not contain any carVOL. 42, NO. 3, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
793
boxylic group. The atmospheric implications of the ammoniaparticulate organic reactions are discussed in light of our findings.
Experimental Section Aerosol Generation and Handling. All organic aerosols were prepared by atomizing dilute aqueous solutions at concentrations of 10-30 mg L-1, generating a polydisperse distribution of droplets with a mean diameter of ∼0.3 µm. The particles-containing flow (0.6 SLM) equilibrated in a 10 L bulb for about 16 min before passing through two silica gel column dryers (RH < 3%). Following a neutralizer, a monodisperse aerosol was selected using a differential mobility analyzer (DMA-1) operating with 6 SLM dry (RH < 3%) sheath flow, and 0.6 SLM sample flow. The atomizer and sheath flows are of dry, particle-free, pure nitrogen. Ammonia Generation. Gas-phase ammonia in various concentrations was generated by bubbling nitrogen (1–8 cm3 min-1) through a highly diluted (1–15 mM) ammonium hydroxide solution. The ammonia concentration in the gas was monitored using a BIONICS ammonia gas detector (0–75 ppm), model TX-2460DP-D with an ammonia gas sensor GS-246ODP-H. The ppm-level flow was further diluted to the desired ppb level by precisely controlling the mixing between the ammonia-containing nitrogen flow and the aerosol flow (∼0.5 SLM). CCN Activation Experiments. The experimental setup used in this work is described in Dinar et al. (29). The setup was slightly modified by adding an additional humidity control unit and an ammonia reactor. The monodisperse aerosol flow was humidified bypassing through the inner shell of a water-filled Nafion humidifier (MD-70, Perma Pure LLC). The RH in the Nafion flow (∼0.5 SLM) was controlled by changing the temperature of the water circulating in the outer shell of the humidifier. Following the Nafion tube, the particles were mixed with a water-saturated nitrogen flow (1–8 cm3 min-1) containing 12.5 or 50 ppm ammonia, yielding ammonia concentrations between 25 and 800 ppb. The combined humid flow was directed to a 1 L bulb (2 min residence time) where reaction occurred. The emerging flow RH was reduced to ∼80% by adding dry nitrogen flow and then spilt; 0.3 SLM was directed to a CPC for determination of condensation nuclei (CN) number concentration, while the remainder (0.1–0.2 SLM) was injected at the center of a thermal gradient diffusion cloud chamber (TGDCC) (29). The supersaturation at the center of the chamber could be varied between 0.05 and 2%. The aerosol flow was confined to the center of the TGDCC using a saturated nitrogen sheath flow (0.8–0.9 SLM). Activated particles were sized and counted by an aerodynamic particle sizer (APS, TSI-3321) situated immediately at the exit of the TGDCC. The CCN activation efficiency for any given dry diameter (∼80% RH in our experiments) is the ratio between the CN number concentration entering the TGDCC (determined by a CPC) to the droplet number concentration measured by the APS (droplets >0.5 µm). By changing the injected particles’ dry size (dry particle spectrum), a plot of the percent of activated particles versus dry diameter is produced. It was previously shown that a sigmoidal fit adequately simulates these experiments and that the critical activation diameter (Dc) for a symmetric distribution of particles is the diameter where 50% of the particles activate (D50) (30, 31). In all experiments, the particles number density in the TGDCC ranged between 10 and 100 particles cm-3. The supersaturation in the TGDCC was chosen based on the aerosol type used. For more soluble species, a lower SS was chosen, whereas a higher SS was used for less soluble aerosols. This produced more accurate activation curves and, hence, D50 values. Activation experiments were conducted for adipic acid and di(ethylene glycol) monovinyl ether (DEGMVE) 794
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 3, 2008
particles. The SS chosen for adipic acid particles was 0.64%, and for DEGMVE particles it was 0.52%. Hygroscopic Growth Experiments. All the experiments under sub saturation (SubS) conditions were conducted using humidified tandem differential mobility analyzer (HTDMA) setup, described in Dinar et al. (32). The hygroscopic growth factor (GF) is the relative size increase of particles (of one fixed monodisperse size) due to water uptake below saturation. It is defined as follows: GFRH )
DRH , D0
where D0 is the particle dry diameter, and DRH is its diameter at a specific RH. The adaptations in the setup for the ammonia reactions are detailed below. After the first size selection of the dry aerosol, the particles were directed to the humidification and reaction section as outlined above and the size change was monitored by the SMPS (see the CCN Activation Experiments section). This setup allows control of the dry particle size (50 or 60 nm always) and number concentration (1000–10000 particles cm-3), ammonia concentration (>25 ppb), residence time in the reactor and the reactor RH. Following the humidifier-ammonia reaction section, the size distribution of the particles was measured using a humidified SMPS. SubS hydration growth curves were obtained by measuring changes in particle diameter as a function of RH. The average mean diameter of the aerosols at the lowest RH was taken as the initial size, D0. It was measured at the beginning and at the end of each run. For all GF measurements, the experimental standard deviation did not exceed ( 0.008 which corresponds to a change in the monodisperse spectra mean diameter of ( 0.5 nm (at a GF90 ) ∼1.23). As ammonia partitioning between the gas (NH3) and the condensed phase (NH4+) depends on the water content of the aerosol (33), the neutralization reaction experiments were conducted only under wet conditions with saturated particles which were then dried to the measured RH. As such, all our HG measurements should be considered as dehydration/ evaporation experiments. Experiments were conducted at 22 ( 1 °C. The differences between the RH of the incoming sample to the RH of the humidified DMA sheath flow was within e1% RH (over the entire RH range). For each sample, the growth factor was measured in 0.3–5% RH intervals from ∼10 to ∼96% RH (with the exception of pure adipic acid particles which were measured up to a RH of 99%) and the experiments were repeated for 3–5 times. HG measurements were conducted for adipic acid, citric acid, and Di(ethylene glycol) monovinyl ether (DEGMVE) particles. CCN Activity Calculations Based on the Köhler Theory. The activation diameters of adipic and citric acids aerosols and their ammonium salts were calculated using the Köhler theory in a similar manner to that described in our previous study (29), Table 1 presents the input parameters used for the four compounds. The input parameters values were taken from the studies of Broekhuizen et al. (34), Henning et al. (35) and Topping et al. (36). The calculations were performed for five different supersaturations ranging from 0.2 to 1.03% and for varying mass fractions of the salt resulting from the neutralization of the acid by ammonia (Tables 2 and 3). For pure adipic acid, it was found that the calculated and measured activation diameters agree when using a water solubility of 42 g/L which is higher than the value reported in the literature (25 g/L, see Henning et al. (35)). The former value was used consistently in the calculations, but we note that use of the literature value for the solubility increases the activation diameters by 25-50%, depending on the supersaturation. Citric acid, in contrast, was assumed to be infinitely water soluble in accordance with its high water solubility (600 g/L, MSDS). Also, the salts formed by the
TABLE 1. Input Parameters As Used in the Ko1 hler Equation for Both Acids and Their Conjugated Ammoniated Salts
compound
Van’t Hoff factor
density (g cm-3)
solubility (g L-1)
molecular weight (g mol-1)
adipic acid ammonium adipate citric acid ammonium citrate
1 2a 1 2a
1.36 1.36b 1.54 1.54b
42 ∞ ∞ ∞
146.14 174.15 192.12 234.16
a It is assumed that monoammoniated salts are formed as a result of the ammonia uptake. b In the absence of measurement data, we assume that the salts have the same density as the corresponding acid.
TABLE 2. D50 Values Calculated for Five Different Supersaturation Conditions and for Particles with Different Mass Fractions of Citric Acid in the Particlea mass fraction of pure citric acid supersaturation (%) 0.2 0.52 0.64 0.8 1.03
1 131 68 59 51 42
0.75 124 65 56 48 40
0.5 119 62 55 46 39
0.25 115 61 52 45 38
FIGURE 1. Activation curves of adipic acid particles measured at 0.64% SS. The black squares (0) present activation of pure adipic acid aerosols, with a D50 ) 151 nm. Adipic acid particles following reaction with ammonia (O,
