Hygroscopicity of Inorganic Aerosols : Size and Relative Humidity Effects on the Growth Factor

The hygroscopic properties of inorganic salt particles, including (NH4)2SO4, NaCl, Na2SO4 and NaNO3, are investigated using a self-assembled hygroscopic tandem differential mobility analyzer (H-TDMA) system. The iso-GF (growth factor) curves are derived to illustrate the effects of the initial particle size (D0) and relative humidity (RH) on the GFs. For those salt particles of 100 nm, the GFs measured agreed well with their theoretical Köhler curves. In the size range of 20–200 nm, the GFs of (NH4)2SO4, NaCl and Na2SO4 particles all continuously decrease with D0 increasing below the deliquescence RH (DRH). However, when RH is higher than the DRH, the GFs of those salts aerosols increase with D0 throughout the investigated size range. Similar increase trend of GFs with D0 is also observed for NaNO3 aerosols though they do not exhibit the abrupt deliquescence behavior. From iso-GF curves, it can be clearly observed that the GFs of (NH4)2SO4, NaCl and Na2SO4 particles all increase with the RH while the values decrease with D0 below DRH. And above DRH, the GFs are more sensitive to D0 for particles smaller than 60 nm, while the GFs are more sensitive to RH for particles larger than 80 nm. For NaNO3 aerosols, the iso-GF curves indicate the size-effect becomes more prominent on their hygroscopicity as the RH increases. The iso-GF curves provide a lucid and explicit insight into the hygroscopic growth of salts particles. Through iso-GF curves, we can clearly elucidate the major factor that affects the ultimate particle diameter at ambient atmosphere.


INTRODUCTION
The hygroscopic growth of aerosol particles are of current interest due to their effects on light scattering and absorption properties of particles which thereby affect the air quality, visibility, Earth's radiation and the climate (Charlson et al., 1992;IPCC, 2007).On a global basis, sulfates, nitrates and chlorides contribute the largest to the mass budget of fine atmospheric particles (IPCC, 2007;Li et al., 2009;Shen et al., 2009).These inorganic salt aerosols are hygroscopic by nature, thus their size, phase and subsequently the optical properties would be strongly influenced by the ambient relative humidity (RH).Based on recent studies, the effect of sulfate particles on the annual-average global direct radiative forcing, arising from the fluctuation of atmospheric particles between aqueous and solid state, is estimated up to as much Fax: +86-21-65642080 E-mail address: jmchen@fudan.edu.cn,yangxin@fudan.edu.cn as 24% (Martin et al., 2004;Wang et al., 2008aWang et al., , 2008b)).Thus, close attention has been paid to study the hygroscopic properties of these inorganic salts.
Hygroscopic tandem differential mobility analyzer (H-TDMA) is a common approach to study the hygroscopic properties and phase transitions of atmospheric particles (Liu et al., 1978).Through H-TDMA, primarily hygroscopic growth factors (GFs) are measured, and information about particle phase and phase transitions can be inferred as a derived quantity.Previous works on determining the deliquescence transitions and water content of single salt particles, including sulfates ((NH 4 ) 2 SO 4 ), chlorides (NaCl) and nitrates (NaNO 3 , Ca(NO 3 ) 2 , Mg(NO 3 ) 2 ), are mainly concentrated on that of initial diameter of 100 nm (Cruz et al., 2002;Gysel et al., 2002;Gibson et al., 2006).In general, those sulfates and chlorides particles exhibit pronounced deliquescence behavior in humid air, i.e. their phases remain unchanged until RH reaches the deliquescence RH (DRH), from which they deliquesce abruptly to form a saturated solution droplet by water vapor condensation (Gysel et al., 2002).Otherwise, those nitrates particles which are highly hygroscopic, always gain their water content smoothly at ambient conditions (Gibson et al., 2006).Recently, initial particle sizes for hygroscopicity studies have been developed to a wide-size-scale, ranged mainly from 6 to 200 nm.The hygroscopic properties of salts particles are revealed to be initial particle size dependent.Biskos et al. (2006) found from the hygroscopic properties of 6 to 60 nm NaCl particles that the GFs steadily decreased with decreasing dry particle size below approximately 40 nm at the RH above deliquescence point.Such size-dependent hygroscopic properties of NaCl particles are also demonstrated by Park et al. (2009). Moreover, Russel et al. (2002) proved the similar size-dependent hygroscopic properties of NaCl particles through model calculations.However, the size-dependent hygroscopic properties varied a lot from various chemical species.For atomizer-generated KCl, MgCl 2 and CaCl 2 particles, their GFs are independent from particle size, while distinct size-effect on GFs for atomizer-generated (NH 4 ) 2 SO 4 particles is observed by Park et al. (2009).
Although some efforts have been made to investigate the size-effect of salts particles on hygroscopic properties, still some chemical species being ubiquitous in aerosol particles (such as NaNO 3 and Na 2 SO 4 ) are not taken into consideration.Moreover, since both the size-effect and RH may giantly impact on the GFs, it will be of significance to distinguish the dominant factor from size-effect and RH.To our knowledge, it is still less details till now to distinguish the impact of size-effect and RH on the GF of inorganic aerosols.To consider the impact of size-effect and RH synchronous will provide a new insight into visibility degradation and air quality improvement under ambient atmospheric condition.
In this paper, the hygroscopic properties of several salts particles, (NH 4 ) 2 SO 4 , NaCl, Na 2 SO 4 and NaNO 3 included, are investigated using H-TDMA technique.And the iso-GF curves are put forward to differentiate the impact of size-effect and RH.

EXPERIMENTAL PROCEDURE
The experimental setup consists of an aerosol generator and a set of H-TDMA system (Fig. 1).The setup has been described in detail in a previous paper (Ye et al., 2009).Herein, a brief description of the experimental procedure is presented.Aerosol particles, produced by the aerosol generator (Model 3076, TSI Inc., USA), are firstly dried to a low RH (RH < 5%) and then fed into the first differential mobility analyzer (DMA-1).Through DMA-1, particles of a certain "dry" size are selected.Afterward, the size-selected-particles are exposed to a controlled humid environment, i.e. a multitube Nafion humidifier (Models PD-70T-24ss, Perma Pure Inc.) where RH is regulated in the range of 5% to 90%.Then, a new particle size distribution is measured by the second DMA (DMA-2) with combination of a Condensation Particle Counter (CPC, Model 3771, TSI Inc., USA).And hygroscopic GFs (D P /D 0 ), defined as the ratio of the humidified (D P ) and dry particle mobility diameter (D 0 ), could be deduced.In this system, to avoid unwanted particle dehydration or further hygroscopic growth in DMA-2, the RH is monitored both on the inlet and outlet of DMA-2 with RH modules (Model HMM22D Module, Vaisala Inc.) to make sure the RH difference between the humidified sample and the humidified DMA-2 sheath gas flow is within 1% RH (throughout all the RH range).Both the DMA-1 and DMA-2 are operated with N 2 as sheath gas in 6.5 LPM, and with sample in 1.0 LPM.The particle size mainly covered the range of 10-400 nm.The accuracy of all DMAs is tested using the standard PSL particles, and the test experiment of H-TDMA system is described in detail elsewhere (Ye et al., 2009).

Hygroscopic Growth of D 0 =100 nm (NH 4 ) 2 SO 4 , NaCl, Na 2 SO 4 and NaNO 3 Aerosols
Firstly, experiments with (NH 4 ) 2 SO 4 are performed to test the H-TDMA system.The hygroscopic properties of (NH 4 ) 2 SO 4 have been well understood and characterized by several different techniques (Tang andMunkelwitz, 1993, 1994;Tang et al., 1995;Cziczo et al., 1997;Onasch et al., 1999;Gysel et al., 2002), and thus making it an ideal system to test the H-TDMA system.Fig. 2(a1) presents a series of Gaussian fits to the measured size distributions of 100 nm (NH 4 ) 2 SO 4 aerosols as a function of RH.The fits have been normalized to the peak concentration.It shows that there is no shift in diameter until the DRH is reached, at which point there is a distinct and abrupt increase in size of the particles as the particles undergo a solid to liquid phase transition.Fig. 2(b1) displays the measured hygroscopic growth curve of 100 nm (NH 4 ) 2 SO 4 aerosols.A good agreement is observed between the measured GFs and the theoretical Köhler curves (derived from AIM model).And the measured DRH appears at 80 ± 2% RH, also in good agreement with values reported previously (Tang and Munkelwitz, 1993;Onasch et al., 1999;Gysel et al., 2002).
The hygroscopic behavior of D 0 =100 nm NaCl, Na 2 SO 4 and NaNO 3 aerosols are displayed in Fig. 2(a2-a4).Fig. 2(a2) shows that NaCl aerosols have a deliquescence point like (NH 4 ) 2 SO 4 and occur at 75 ± 2% RH at 25°C, agree well with the values reported previously (Tang and Munkelwitz, 1993).Moreover, as the measured hygroscopic growth curve of NaCl aerosols shows in Fig. 2(b2), good agreement is also observed between the GFs and theoretical hygroscopic curve.For Na 2 SO 4 aerosols, as can be seen from Fig. 2(a3), crystalline Na 2 SO 4 particles remain unchanged until RH reaches the DRH.Then, they deliquesce spontaneously to form a saturated solution droplet by water vapor condensation, gain about 1.7 times its initial diameter.After that, the droplets continue to grow as RH further increases.The hygroscopic properties of Na 2 SO 4 aerosols have been examined previously with several different techniques and are expected to deliquescence at 84% RH at 25°C ( Tang and Munkelwitz, 1994).As Fig. 2(b3) displays, the measured DRH of Na 2 SO 4 aerosols appeared at 82 ± 2% RH, a little lower than the values expected.Such discrepancy may be caused by the slight temperature fluctuates during the experiment where the temperature increase will lead to the decrease of DRH (Tang and Munkelwitz, 1993).
As Fig. 2(a4) presents, unlike (NH 4 ) 2 SO 4 , the NaNO 3 aerosols do not exhibit the deliquescence phenomenon and show continuous growing with RH increasing in this experiment.Literature data (Tang and Munkelwitz, 1994) reported a hysteresis effect of NaNO 3 particles at 25°C with a DRH of 74.5%, and an ERH between 30% and 0.05%.The minimum RH in this experiment (about 5% after the diffusion dryer) is obviously not low enough for the spontaneous crystallization of the NaNO 3 particles.Thus, NaNO 3 aerosols do not completely dry out even at 5% RH.Because of the missing crystallization, the NaNO 3 aerosols in the DMA-1 have a diameter larger than the volume equivalent diameter of the solute NaNO 3 .Thus, the particle diameter measured in the DMA-1 (RH = 5%) should be divided by the theoretical GF at this RH (1.02 at 25°C and 5% RH) to deduce the real diameter D 0 (98 nm) of dry NaNO 3 aerosols.Fig. 2(b4) displays the measured GFs curve and it agrees well with the theoretical hygroscopic growth curve.

Size-effect on the Growth Factor
(NH 4 ) 2 SO 4 , NaCl and Na 2 SO 4 aerosols all exhibit pronounced deliquescence behavior, and their growth factors as a function of D 0 are demonstrated in Fig. 3.As can be seen, below DRH, the GFs of those salt aerosols all continuously decrease with D 0 increasing in the range from 20 to 200 nm at given RH, and the smallest particles show the highest growth factor.Take (NH 4 ) 2 SO 4 aerosols for instance, the GFs decrease from 1.05 for 20 nm aerosols to 1.03 for 200 nm aerosols at 70% RH.However, the theoretical GFs of those inorganic salt particles should be 1.00 below DRH.This interesting finding can be explained from two aspects.On the one hand, the small particle growth just below the DRH is probably caused by water adsorption on particle surface.Several studies (Dia et al., 1997;Romakkaniemi et al., 2001) have testified that the salt surfaces are actually adsorbed by water even though the ambient RH was far below its deliquescence point.On the other hand, although the adsorption of water vapor on small particles is more difficult than that on large particles due to the curvature effect, the relative increment (expressed by GFs) of small particles is still higher than large particles as a result of their high surface to volume ratio.In addition, it is noteworthy that the GF-Size curves gradually lift up with increasing of RH.Probably higher RH that means more moisture in the atmosphere leads the particles to adsorb more water vapor on particle surface.
When RH is higher than the deliquescence point, the GFs of those salts aerosols increase with D 0 throughout the investigated size range.Further analyses reveal the GFs increase pronouncedly for particles in the range of 20-100 nm, and the increase trend relatively slows down for particles in the range of 100-200 nm.As Fig. 3(b1) shows, the GFs increase from 1.52 for 20 nm (NH 4 ) 2 SO 4 aerosols to 1.61 for 100 nm aerosols at 86% RH, corresponding to  0.09 GF increment, while it only increases by 0.02 for aerosols from 100 to 200 nm.Moreover, the diameter increments (Table 1) are corresponded to 10.3, 60.9 and 125.3 nm for those (NH 4 ) 2 SO 4 aerosols of 20, 100 and 200 nm, respectively.It indicates that the large particles are hygroscopic vulnerably and gaining water easier.This might be caused by the fact that larger particles contain relatively more solute than smaller ones.Our results consist with the size-dependent hygroscopicity reported in the previous works (Gysel et al., 2002;Biskos et al., 2006;Park et al., 2009).Kelvin effect is supposed to be the major reason caused the size-dependent GFs.The growth of aqueous droplets in humid air is commonly described by Köhler theory (Kohler, 1936;Gysel et al., 2002;Kreidenweis et al., 2005).At equilibrium, the RH is the ratio of the vapor pressure over a solution droplet to the saturation vapor pressure of water.The relationship between droplet diameter and RH at equilibrium is given by: where a w is the water activity, C ke is the Kelvin correction factor, v w is the partial molar volume of water, σ sol/v is the surface tension of the solution at the composition of the droplet, R is the universal gas constant, T is the temperature and D p is the equilibrium diameter.The Köhler equation states that the RH equals the water activity for a plane solution surface, and that for curved surfaces such as aerosol particles, the water activity is always smaller than the ambient RH.The Kelvin factor therefore causes the hygroscopic growth to decrease with decreasing dry size for particles of identical composition.
Water-soluble aerosol particles grow by condensation of water vapor simply because the water activity of the aqueous solution strives to equilibrate to the RH in the surrounding air in accordance with the Köhler equation.Among the salts investigated, the abrupt deliquescence behavior do not occurred for the NaNO 3 aerosols.Like the hygroscopic growth behaviors of (NH 4 ) 2 SO 4 , NaCl and Na 2 SO 4 above DRH, similar increase trends of GFs with particle size are also observed for NaNO 3 aerosols (Fig. 4).Throughout the entire RH scope, the GFs keep increasing markedly until D 0 reaches 100 nm, and the increase trend of the GFs slows down as the particles further increasing.At 60% and 86% RH, the GFs increase from 1.28 and 1.65 for D 0 = 20 nm particles to 1.32 and 1.85 for D 0 = 100 nm particles, and ultimately the values increase to 1.33 and 1.89 for D 0 = 200 nm NaNO 3 aerosols, respectively.

Size and Relative Humidity Effects on the Growth Factor
The GFs of aerosols are influenced markedly not only by their chemical composition, but also by the initial particle size and ambient RH.To distinguish the dominant impact on GFs from size-effect and RH, iso-GF curves with D 0 and RH as the coordinates are put forward and the corresponding curves of (NH 4 ) 2 SO 4 , NaCl and Na 2 SO 4 aerosols are illustrated in Fig. 5. Below DRH, the GFs increase with RH while they decrease with D 0 .It indicates RH plays a dominant role to augment the particle diameter at ambient atmosphere.Above DRH, the GFs are more sensitive to particle size for particles smaller than 60 nm, while the GFs are more sensitive to RH for particles larger than 80 nm.For NaNO 3 aerosols (Fig. 6), at 80% ≤ RH ≤ 86%, the GFs are easier to increase with D 0 for particles smaller than 60 nm, while they are easier to increase with RH for particles larger than 80 nm.However, at 20% ≤ RH ≤ 70%, it is noted that the GFs are more sensitive to the RH than to D 0 even though the particles are smaller than 60 nm.It indicates that the size-effect of NaNO 3 aerosols on their hygroscopicity becomes more and more prominent as the RH increases.

CONCLUSION
Based on the hygroscopicity investigation of (NH 4 ) 2 SO 4 , NaCl, Na 2 SO 4 and NaNO 3 aerosols, the size-effect is found on the GFs of Na 2 SO 4 and NaNO 3 aerosols besides (NH 4 ) 2 SO 4 and NaCl aerosols (Gysel et al., 2002;Biskos et al., 2006;Park et al., 2009).Nevertheless, much more studies are needed to explore the size-dependent hygroscopicity of other compounds being ubiquitous in atmospheric aerosols and to elucidate the mechanisms causing such size-effect.As is observed in our studies, The GFs and ultimate particle size are strongly influenced by D 0 .And they will subsequently affect the optical properties of particles which will further impact on Earth's radiation and the climate.In this paper, the iso-GF curves are proposed for the first time.Such iso-GF curves provide a lucid and explicit insight into the hygroscopic growth of salts particles.Through iso-GF curves, we can clearly determine the major factor that affects the ultimate particle diameter at ambient atmosphere.Then, it will be of significance to improve the visibility and air quality by controlling the dominant factor impact on aerosol hygroscopicity.
Technology R&D program (No. 2006BAC12B001) of China, the Key Science & Technology Project (No. 108050) from the Chinese Ministry of Education and the International Collaboration Project from the Science and Technology Commission of Shanghai Municipality (No. 09160707700 ).