SUPPLEMENTARY INFORMATION

for

Polluted dust promotes new particle formation and growth

Wei Nie1,2,6, Aijun Ding1,6, Tao Wang2,3, Veli-Matti Kerminen4, Christian George5, Likun Xue3, Wenxing Wang2, Qingzhu Zhang2, Tuukka Petäjä4, Ximeng Qi1,6, Xiaomei Gao2, Xinfeng Wang2, Xiuqun Yang1,6, Congbin Fu1,6 and Markku Kulmala4

1Institute for Climate and Global Change ResearchSchool of Atmospheric Sciences, Nanjing University, Nanjing, 210093, China, 2Environment Research Institute, Shandong University, Jinan, China, 3Department of Civil and Environmental Engineering, The Hong Kong Polytechnic University, Hong Kong, China, 4Division of Atmospheric Sciences, Department of Physics, University of Helsinki, Helsinki, Finland, 5Université de Lyon, Lyon, F-69626, France; Université Lyon 1, Lyon, F-69626, France; CNRS, UMR5256, IRCELYON, Institut de Recherches sur la Catalyse et l’Environnement de Lyon, Villeurbanne, F-69626, France, 6Collaborative Innovation Center of Climate Change, Jiangsu Province, China.

Correspondence and requests for materials should be addressed to A.D () or T.W ()

  1. Site description, field campaign and measurement techniques

Themeasurement site was at a meteorological station at summit of Mount Heng (27°18′ N, 112°42′ E, 1269 m a.s.l.) in Hunan Province, Southern China, which is about 500 km and 900 km from the South China Sea and the East China Sea, respectively (Fig. 3). Mt. Heng is located in a remote region with few anthropogenic emissions but abundant nature biogenic emissions (Fig. 3b and 3c). The closest city is approximately 50 km to the south (Detailed information can be found in references1-3).

The field campaign was conducted from March to May 2009. A comprehensive suite of trace gases, aerosols, meteorological parameters, and cloud and rain water was measured/collected. The dataset deployed in this paper includes continuous data of SO2,O3, PM2.5, PM10, sulfate, calcium, BC in PM2.5, and particle number size distribution (10 nm - 10000 nm); and filter based metal elements (Fe and Ti) in PM2.5 andsize resolved sulfate andnitrate in TSP.Measurement techniques are briefly descript as follows.

A pulsed UV fluorescence analyzer (TEI model 43C) was used to measure SO2, and a UV photometric analyzer (TEI model 49i) to measure O3.ATapered Element Oscillating Microbalance (TEOM, Thermo Electron Corporation, East Greenbush, NY, USA) was employed to continuously measure PM2.5 and PM10. An Aethalometer (Magee Scientific, Berkeley, California, USA, Model AE-21) was deployed to measure BC at a wavelength of 880 nm. A Wide-range Particle Spectrometer (WPSTM, MSP Corporation model 1000XP) was used to measure particle number size distribution in the range of 10 – 10000 nm.A four-channel particle sampler (Thermo Andersen Chemical Speciation Monitor, RAAS2.5-400, Thermo Electron Corporation) was used to collect PM2.5 filter samples with Teflon filters (Teflon™, 2 µm pore size and 47 mm diameter, Pall Inc.) at a flow rate of 16.7 liters per minute (LPM). A Micro-Orifice Uniform Deposit Impactor (MOUDI, MSP Company) was used to collect size resolved particles with aluminum substrates (MSP Company) at a flow rate of 30 LPM. There are eight size ranges for the impactor to collect particles:18 μm (inlet), 10-18 μm, 5.6-10 μm, 3.2-5.6 μm, 1.8-3.2 μm, 1-1.8 μm, 0.56-1 μm, 0.32-0.56 μm, and 0.18-0.32 μm4. The filter samples were stored at -5 ºC in order to minimize artifacts. The metal elements in PM2.5 were measured using X-ray fluorescence (XRF). The water-soluble ionswere analyzed using ion chromatography (IC) (Dionex 90)5.

  1. Size distribution of particle surface area

Figure S1 Mean size distribution of particle surface area during the daytime (10:00 - 14:59 local time) of 7th, 25th and 26th April at Mt. Heng. The mean value of total particle surface area concentrationsare816 μm2/cm3and 619 μm2/cm3in the dust event days of 25th and 26th April respectivelyand mostly concentrated in coarse mode particles (more than 1 μm). The value in a non-dust event day of 7th April was 99.5 μm2/cm3, and mostly concentrated in accumulation mode (0.1 – 0.5 μm).

  1. CALIPSO measurement on 25 April

Fig. S2 CALIPSO measured vertical profiles of 532 nm total attenuated backscatter and aerosol subtype on 25 April., 2009. Red open cycle marked the location around the summit of Mt. Heng. The figures (including the map) were provided by NASA,

  1. Method to calculate particle formation rates, growth rates and condensation sink

Calculation of the formation rate, growth rate, and condensation sink is followed the critical Kulmala et al.6. Briefly, the formation rate is represented as:

The second item is the coagulation losses by large particles and third item is the growth out of the considered size range. The Fourth item in the equation is the additional losses which aren’t considered in this study. The growth rate of NPF events can be expressed as:

In this equation, dp1 and dp2 are the representative particle diameter at time t1 and t2. For calculation, dp1 and dp2 is defined as the center of size bin and t1, t2 are the times when the concentration of this size bin gets the maximum. The condensation sink (CS) is the value to describe the speed that condensable vapor molecules condense on the existing aerosol. It is expressed as:

Where D is the diffusion coefficient of the condensing vapors, and βm is a transition-regime correction7.GR can be expressed in another way as:

Where mv is the molecular mass of the condensing vapor, ρ is particle density, Cvapor is vapor concentration. Integrating the equation, vapor concentration Cvapor can be expressed as8:

In this equation, α is the mass accommodation coefficient and λ is the vapor mean free path. Meanwhile, the variation of vapor concentration Cvapor can be written as a balance equation:

Where CS is condensation sink and Q is the vapor source rate. Assuming a pseudo steady state situation,is zero. Then we can get:

  1. HONO formation during the dust storms

Fig. S3 shows the nitrite concentrations in the 4 TSP samples during the dust storm of 25-26 April at Mt. Heng. The result showed pretty high nitrite concentration during this dust event, with the average value of 2.5 μg m−3, which was much higher than that in the non-dust days (the mean value was 0.3 μgm−3). In addition, the enhanced nitrite formation was only observed in the daytime samples (with the highest concentration in the daytime sample of 26th April). These results indicate a dust involved photochemical pathway to yield nitrite and HONO. The TiO2photocatalysis of NO2 was considered the most possible mechanism (detailed analysis can be found in reference 1). The photolysis of enhanced HONO would produce additional OH radical to promote the oxidation of SO2 to form H2SO4.

Fig.S3 Nitrite concentrations in total suspendedparticulates (TSP) collected during the dust storm of 25-26 Aprilat Mt. Heng (Fig. 8 in reference1).

  1. Lagrangian dispersion modeling and VOCs simulation

The transport and dispersion simulations were made using a Lagrangian dispersion model, called the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model developed in the Air Resource Laboratory of the National Oceanic and Atmospheric Administration9. The model calculates the position of particles by mean wind and a turbulence transport component after they are released at the source point for forward simulation or receptor for backward run.

In this work, the LPDM simulations followed the method developed by Ding et al.10. Briefly, the model was used to conduct hourly backward particle dispersion simulations. In each simulation, particles were released at the site and backward in time for a 7-day period. The hourly position of each particle was calculated using a 3-D particle, i.e. horizontal and vertical, method. The air concentrations were calculated according to the particle number distribution. We calculated the air concentration at 100 m to get a “footprint” retroplume of the released air particles, which represents the distribution of the surface probability or residence time of the simulated air mass. We used the ARL format GDAS data, which were converted from the original 3-hour archive data of the Global Data Assimilation System of the National Center for Environmental Prediction ( The GDAS has a horizontal resolution of 1 degree by 1 degree and 23 vertical levels from the surface to 100 hPa.

Fig. S4Temporal variations of particle number size distributions (10-500 nm), PM10, simulated total anthropogenic VOC and simulated monoterpene during observed dust episodes at Mt. Heng, and the mean values of simulated total anthropogenic VOC and simulated monoterpene during the whole campaign.

The simulation result overall showed lower concentrations of both anthropogenic VOC and monoterpeneduring dust episodes than the average values (Fig. S4). In specific, bothconcentrations of anthropogenic VOC and monoterpene decreased sharply at the beginning of both dust events because of the dilution of dust plumes; then increased to some extent probably due to the influence of other plumes, e. g. anthropogenic emissions from eastern China. As low volatile organic vapors are the principle contributor to the particle growth, the largely decreased values of VOC in 21 April are probably the reason why no typical growth was observed on that day (Fig. S4).

  1. Meteorological factors, photosensitive species and anthropogenic pollutants during dust events

Several conditions of photosensitive components (e. g. Fe2O3, TiO2), light, water and reactive species are needed to induce heterogeneous photochemical processes11,12. For example, in the laboratory work of photo induced particle nucleation, Fe2O3 is the photosensitive median, H2O is the precursor to produce OH radicals, SO2 is the reactive species to form H2SO411. For dust event, despite of abundant containing of metal oxides, it tend to weaken the solar radiation by scattering of sunlight, reduce humidity by absorbing ambient water, and “clean” the air by bringing the pollution free air. Therefore in the real atmosphere, heterogeneous photochemical processes are rarely observed in the fresh dust plumes.

During the dust episodes of this study, there were relative strong sunlight (with maximum solar radiation above 800 W/m2) and high RH (generally more than 60%) even in the strongest dust event day of 25April; sharply increased metal elements, which mostly exist in the form of the oxides; elevated anthropogenic pollution concentrations comparing to those in non-dust days (Fig. S5). (Detailed comparison of anthropogenic pollutants between dust and non-dust days can be found in Table 1 of reference 1, which showed anthropogenic pollutants increased 0.4-1.2 folds in dust days.) These results suggest the air masses arrived Mt. Heng during the dust episodes were mixed plumes of dust and anthropogenic pollutions, which provides photosensitive medians and reactive species, as well as sufficient solar radiation and water to promote the heterogeneous photochemical processes, e. g. HONO production and new particle formation.

Fig. S5 Temporal variations of solar radiation, RH, Fe, Ti, sulfate and BC in PM2.5 at Mt. Heng during 20 to 29 April 2009.

  1. Secondary coating on dust particles

As showed in Fig. S6, sulfate and nitrate were mostly concentrated in coarse mode particles during the dust events, suggesting the formation of secondary coating on the surface of dust particles. This kind of coatings modified the surface nature of the dust (from hydrophobic to hydrophilic), thus significantly enhanced their CCN capacity1 and provided sufficient water to induce heterogeneous photochemical processes. This may be also one reason of the observed high RH during strong dust storms (Fig. S5). In addition, as nitrate is reactive and photosensitive species, the combined effect of nitrate and metal oxides (e. g. TiO2) under illumination can produce O313 and promote the oxidation of VOCs by releasing NO3 radical14, and finally contribute to the particle growth.

Fig. S6 Size distribution of sulfate and nitrate on 25 April at Mt. Heng.

  1. Fine sulfate formation during the dust eventof 25-26 April

Fig. S7 Temporal variations of sulfate, and sulfate to BC ratios in PM2.5 during 25-26 April.

Evident diurnal cycles were observed for PM2.5 sulfate during the dust event of 25-26 April, with the peaks appearing around the noon time accompanied with the NPF (Fig. S4, Fig. S5 and Fig. S7), indicating the formation of fine sulfate during the strong dust event. To exclude the influence of transport and the development of boundary layers, we took BC as a benchmark. The results showed a similar pattern of sulfate/BC ratios as that of sulfate, suggesting a daytime formation of fine sulfate.

  1. Ammonia and ammonium during the dust event of 25-26 April

Beside sulfuric acid, ammonia gases are another possible precursor of the NPF. In this study, although gas-phase ammonia was not measured, our previous study 3 has demonstrated Mt. Heng was in ammonia-poor environment.In the aerosol phase, ammonium was not enough to neutralize the acidic species (sulfate and nitrate) in the fine mode particles (PM2.5). During NPF periods in the dust storms of 25-26 April, it was in the same situation that fine mode ammonium was not enough to neutralize sulfate (see Fig S8). This suggests the gas-phase ammonia concentrations were very low (because ammonia gas is efficiently taken up by acidic particles). Therefore, we believe ammonia would have little effect on nucleation.

Fig. S8 Neutralization extent of sulfate by ammonium in PM2.5 during the daytime of 25-26 April

Additional reference

  1. Nie, W. et al. Asian dust storm observed at a rural mountain site in southern China: chemical evolution and heterogeneous photochemistry. Atmos. Chem. Phys.12, 11985-11995 (2012).
  2. Sun, M. et al. Cloud and the corresponding precipitation chemistry in south China: water-soluble components and pollution transport. J. Geophys. Res.115, D22303 (2010).
  3. Gao, X. et al. Aerosol ionic components at Mt. Heng in central southern China: abundances, size distribution, and impacts of long-range transport. Sci. Total Environ.433, 498-506 (2012).
  4. Nie, W. et al. Comparison among filter-based, impactor-based and continuous techniques for measuring atmospheric fine sulfate and nitrate. Atmos. Environ.44, 4396-4403 (2010).
  5. Wu, W. S. & Wang, T. On the performance of a semi-continuous PM2.5 sulphate and nitrate instrument under high loadings of particulate and sulphur dioxide. Atmos. Environ.41 (2007).
  6. Kulmala, M. et al. Measurement of the nucleation of atmospheric aerosol particles. Nat. Protoc.7, 1651-1667 (2012).
  7. Fuchs, H. & Sutugin, A. G. High-dispersed aerosols - topics in current aerosol research (Volume 2), Hidy, G.M. & Brock, J.R. (eds.) 1-60 (Pergamon, 1971).
  8. Kulmala, M. et al. On the growth of nucleation mode particles: source rates of condensable vapor in polluted and clean environments. Atmos. Chem. Phys.5, 409-416 (2005).
  9. Draxler, R. R. & Hess, G. An overview of the HYSPLIT_4 modelling system for trajectories, dispersion and deposition. Aust. Meteorol. Mag.47, 295-308 (1998).
  10. Ding, A., Wang, T. & Fu, C. Transport characteristics and origins of carbon monoxide and ozone in Hong Kong, South China. J. Geophys. Res.118, 9475-9488 (2013).
  11. Dupart, Y. et al. Mineral dust photochemistry induces nucleation events in the presence of SO2. Proc. Natl. Acad. Sci.USA109, 20842-20847 (2012).
  12. Cwiertny, D. M., Young, M. A. & Grassian, V. H. Chemistry and photochemistry of mineral dust aerosol. Annu. Rev. Phys. Chem.59, 27-51 (2008).
  13. Monge, M. a. E. et al. Ozone formation from illuminated titanium dioxide surfaces. J. Am. Chem. Soc.132, 8234-8235 (2010).
  14. Styler, S. A. & Donaldson, D. J. Photooxidation of atmospheric alcohols on laboratory proxies for mineral dust. Environ. Sci. Technol.45, 10004-10012 (2011).