Characterization of aerosol chemical composition from urban pollution in Brazil and its possible impacts on the aerosol hygroscopicity and size distribution

Characterization of aerosol chemical composition from urban pollution in Brazil and its possible impacts on the aerosol hygroscopicity and size distribution

Accepted Manuscript Characterization of Aerosol Chemical Composition from Urban Pollution In Brazil and Its Possible Impacts on the Aerosol Hygroscopi...

1MB Sizes 0 Downloads 32 Views

Accepted Manuscript Characterization of Aerosol Chemical Composition from Urban Pollution In Brazil and Its Possible Impacts on the Aerosol Hygroscopicity and Size Distribution Gerson P. Almeida, Antônio T. Bittencourt, Marçal S. Evangelista PII:

S1352-2310(19)30046-9

DOI:

https://doi.org/10.1016/j.atmosenv.2019.01.024

Reference:

AEA 16513

To appear in:

Atmospheric Environment

Received Date: 6 March 2018 Revised Date:

3 January 2019

Accepted Date: 5 January 2019

Please cite this article as: Almeida, G.P., Bittencourt, A.T., Evangelista, M.S., Characterization of Aerosol Chemical Composition from Urban Pollution In Brazil and Its Possible Impacts on the Aerosol Hygroscopicity and Size Distribution, Atmospheric Environment, https://doi.org/10.1016/ j.atmosenv.2019.01.024. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT 1

Abstract

2

We studied the effect of aerosols inorganic chemical composition on the aerosol hygroscopicity of urban

3

pollution in Brazil, where biofuels have been used in large scale. We applied size segregated inorganic chemical

4

composition analysis using ISORROPIA II model and κKöhler theory to determine the hygroscopicity parameter

5

(κ) of submicrometer aerosols measured in São Paulo city. The size dependence of organic and black carbon (BC)

6

mass were estimated by chemical mass balance and mean observed values. Results showed ultrafine mode particles with diameter smaller than 100 nm with a relatively K2SO4 and

8

Na2SO4 large amount inducing further growth by diffusive condensation and coagulation of low-volatile organic

9

compounds. The process could lead to modifications of aerosol size distribution and also to formation of more

10

active Cloud Condensation Nuclei (CCN) due to the formation of aerosols with considerably increase of

11

hygroscopicity ( > 40 %). The contribution from BC can decreases up to 40% of the observed hygroscopicities

12

values of particles around 100 nm in diameter.

SC

RI PT

7

Moreover, we present a parameterization based on aerosol mass fraction to accurately predict κ derived

14

from data of Aerosol Mass Spectrometer (AMS) collected in urban pollution in Brazil. Results are compared to

15

hygroscopicity derived from observations of the pollution plume downwind Manaus, on the northern region of

16

Brazil. Both cases were analogous indicating that, despite the fact of receiving influences of organic components

17

from the forest, the pollution plume of Manaus shows the same characteristics of hygroscopicity, and can be

18

modeled following the same parameterization.

19

22

Keywords: Urban pollution, Brazil, Biofuels, Aerosols, Chemical Composition, Hygroscopicity, CCN

1. Introduction.

23

TE D

20 21

M AN U

13

The hygroscopic properties of atmospheric aerosol particles are of major importance in

25

describing its life cycle and the related direct and indirect effects on climate. Such properties

26

define particles processes to take up water under saturated environments, developing into cloud

27

droplets at supersaturated conditions (e.g. McFiggans et al., 2006). Hygroscopic properties are

28

directly related to the chemical composition of individual aerosol particles, and can be derived

29

from the volumetric fractional composition of organic and inorganic compounds (Petters and

30

Kreidenweis, 2007).

AC C

31

EP

24

The influence of water uptake of the organic aerosol fraction, which can contribute 20–

32

90% of atmospheric fine aerosol mass, is much lower than the one from inorganic compounds,

33

though still not completely understood. However, the importance of hygroscopic properties of

34

atmospherically relevant inorganic salts is significantly recognized. The high hygroscopicity

35

values of inorganic compounds imply that even small fluctuation on those compounds can be

36

crucial to allow a particle to be activated or not as a CCN at a given supersaturation condition.

37

Keeping that in mind, it is of special importance that one considers urban air, where most

38

hydrophobic particles are produced from combustion (Swietlicki et al., 2008), and, afterwards

ACCEPTED MANUSCRIPT 39

they could undergo interactions with long-range transported background particles, producing

40

complex aerosols. Particles produced from combustion can also contain soot (black carbon, BC), which

42

extinguishes direct solar radiation. Whether those particles are hygroscopic or not can

43

substantially influence their removal and life cycle in the atmosphere. Since pollutants are also

44

transported, influences from large urban areas can be expected on a regional, continental, and

45

global scale (Molina and Molina, 2004; Lawrence et al., 2007; Kunkel et al., 2013). Therefore,

46

the assessments of their properties are of major interest due to their potential influences in large

47

areas. The theme becomes more intriguing if we consider that the eminent fossil fuel shortage

48

has motivated the search for alternatives fuels, such as biofuels.

SC

RI PT

41

It is important to highlight that there are no specific studies of biofuel effects on aerosol

50

hygroscopicity and size distribution characteristics. Some initial inferences on the theme were

51

presented by Salvo et al (2017), considering the characteristics of Brazilian fuel, which is the

52

most developed and integrated biofuels program in the world (Sorda et al., 2010), and use a

53

blend unique, being diesel with 7% of bio-diesel burned by heavy-duty vehicles and gasoline

54

with 27% of anhydrous ethanol (gasohol), as well as hydrated ethanol by flex-fuel light-duty

55

vehicles. The emissions by these fuel blend burnings can produce aerosols chemical

56

composition not similar to those observed in other countries. In the US, for example, all

57

gasoline powered vehicles sold can run on fuels with gasoline and only 10% ethanol. On the

58

other hand the EU aims to have 10% of the transport fuel of every EU country come from

59

renewable sources such as biofuels by 2020.

TE D

M AN U

49

In this context, this study characterized the hygroscopicity parameter of the inorganic

61

aerosol fraction collected during the winter 2012 in São Paulo, Brazil. Emphasis is placed on

62

describing the major aerosol chemical inorganic components and evaluating its possible

63

impacts of the solute contributions to the hygroscopicity and aerosol size distribution.

64

Considering that São Paulo is a very large urbanized polluted area, where biofuels are used in

65

large scale, and not dissimilar from other urbanized areas around the country, this study helps

66

to clarify important properties of aerosols from polluted areas in Brazil.

AC C

EP

60

67

The use of size segregated aerosol chemical composition for determination of

68

hygroscopicity has been previously reported in the work of Liu et al. (2014), with which this

69

study shows similarities. Nevertheless, while in the work of Liu et al. (2014) they intend to

70

determine hygroscopicity as a function of synoptic transport pattern, our emphasis is on the

71

characteristics of urban pollution marked by the presence of products from the biofuels

ACCEPTED MANUSCRIPT 72

combustion, with a focus on providing information for further study that can lead towards the

73

mechanism about particles formation.

74 75

2. Location Description, Sample Collection, and Experimental Methods

76

78

Data used in this study was collect from two sites in Brazil. One in São Paulo and the other on a more remote location in Brazil, close to the city of Manaus.

RI PT

77

The Metropolitan Area of São Paulo (MASP), at 23.50 S, 46.60 W, is located in the

80

southeastern region of Brazil, and consists of 39 municipalities, including São Paulo City,

81

capital of São Paulo State. MASP is around 50 km from the Atlantic Ocean, 800 m of altitude,

82

and is among the megacities worldwide reaching more than 22 million inhabitants with 8

83

million vehicles burning a mix of fuels (gasoline, ethanol, diesel and biodiesel). The region is

84

also affected by industrial emissions, which results in complex sources of aerosols and its

85

precursors, causing serious air pollution conditions (CETESB, 2017).

M AN U

SC

79

The aerosol sample site in MASP, Armando Salles de Oliveira campus of the University

87

of São Paulo, is a green park (7.4 km2) surrounded by streets and avenues with intense light-

88

and heavy-duty vehicles. The samples were collected each 24 h using a cascade Micro-Orifice

89

Uniform Deposit Impactor (MOUDI, model 100, MSP Corporation; Marple et al., 1986), at the

90

roof of the Institute of Astronomy, Geophysics and Atmospheric Science (IAG), about 12 m

91

above mean ground level, from 15th August to 5th September 2012, winter characterized by dry

92

period (Vieira-Filho et al., 2016). The particles sampling was performed through eleven stages

93

with 50% cut off diameters (D50) each: < 18, 10, 5.6, 3.2, 1.8, 1.0, 0.56, 0.32, 0.18, 0.1 and

94

0.056 µm, followed by an after-filter as the last stage (0.020 µm).

EP

TE D

86

In general, particles of sizes larger than the cut-size of certain stage but smaller than a

96

cut-size of the previous stage are present in all stage. This indicates that masses observed at all

97

stages are composed from particles of sizes smaller and larger than the cut off diameter,

98

probably with chemical composition skewed toward larger sizes. For simplicity we assume

99

here that the mean chemical composition represents values of cut off diameter. As will be

100

shown, this assumption seems to be valid because the chemical composition described below

101

undergoes processes isolated by size ranges whose major influences are close to the cutting

102

diameters

AC C

95

103

Mass concentrations were obtained gravimetrically by employing an electronic high-

104

precision microbalance with 1 µg sensitivity (Mettler Toledo MX5), before and after sampling

ACCEPTED MANUSCRIPT 105

on polycarbonate filters, in a room with controlled temperature and humidity, 22±2 ºC and

106

45±3 %, respectively (Vieira-Filho et al., 2016). Water-soluble ions extraction in 10 mL deionized water (18 MΩ.cm) was performed with

108

a continuous mechanical stirring for 60 min, followed by micro-filtration in MILLEX

109

membrane, 0.22 µm pore size (Vieira-Filho et al., 2016), which allows the determination of

110

individual ions masses. All the solutions were kept frozen until analysis by ion chromatography

111

(IC) with conductivity detection (Metrohm model 851). The conditions for anions analyses

112

were: anionic column Metrosep A-Supp5 (250mm x 4mm), eluent of Na2CO3 4.0 mmol L-1 /

113

NaHCO3 1.0 mmol L-1; flow rate of 0.7 mL min-1; Metrohm suppressor system using

114

regenerant solutions of H2SO4 50 mmol L-1and deionized water under 0.8 mL min-1flow rate.

115

Cationic Metrosep column C2 150 (150x4 mm) Metrohm, tartaric acid as eluent 4 mmol L-1 /

116

dipicolinic acid 0.75 mmol L-1, 1.0 mL min-1 flow rate and Metrohm electronic suppression

117

system conditions were used to determine cations. The quantification was performed using

118

external calibration curve from standard concentrations for the ions. Anions measured included

120

SC

M AN U

119

RI PT

107

acetate (   ), chloride (Cl−), nitrate ( ), oxalate (  ), and sulfate (  ). The cations measured were ammonium (NH4+), calcium (  ), potassium (K+), magnesium

(  ), and sodium (Na+) (Vieira-Filho et al., 2016). The detection limit (DL) values were

122

calculated from parameters obtained by the analysis, using the least squares method of the

123

calibration curve (y = a + bx) and correspond to the blank signal (or linear coefficient) plus 3

124

times the standard deviation of the "blank" (sy/x), that is, DL = a + 3sy/x (Miller and Miller,

125

1988). The lowest DL value (0.5 µmol L-1) were observed for potassium, ammonium, nitrate,

126

sulfate and oxalate, the highest (2.4 µmol L-1) for magnesium.

EP

127

TE D

121

The ion mass obtained from the IC measurement were then used as input for the ISORROPIA‐II

thermodynamic

equilibrium

code

(Fountoukis

and

Nenes,

2007)

129

(http://nenes.eas.gatech.edu/ISORROPIA) to provide a realistic combination of anions and

130

cations forming the inorganic compound according to the measured ion mass, which also

131

allows the definition of its volume. According to the inorganic volume fraction Köhler Theory

132

analysis and κ‐Köhler theory are applied (section 3).

AC C

128

133

Elemental analysis was performed by EDXRF - Spectrometer EDX 700HS; Shimadzu

134

(Miranda et al., 2012). The filter was submitted to EDXRF, and spectra accumulated for 900 s

135

under the following conditions: Al filter, vacuum as X-ray path, 10-mm diameter collimator,

136

10–20 keV energy range, 50 kV tube voltage, an Rh X-ray tube, and a Si(Li) detector. We

137

analyzed the elements Al, Si, P, S, Cl, K, Ca, Ti, V, Cr, Mn, Fe, Ni, Zn, Ga, Br, Zr, and Pb. The

ACCEPTED MANUSCRIPT 138

spectra were reduced withWinQXAS software, available from the website of the International

139

Atomic Energy Agency (http://www.iaea.org/OurWork/ST/NA/NAAL/pci/ins/xrf/pciXRFdown.php).

140

The elemental analysis allowed the determination of total mass and volume of insoluble

141

material, like SiO , Al O , Fe O and others. The combination of the ion mass with elemental

142

mass defines the total inorganic mass.

144

Blank filters were analyzed to evaluate ions and elements by IC and EDXRF,

RI PT

143

respectively, followed by discount for each parameter evaluated in all samples.

The experiment site in Amazonas, North of Brazil, was in Manacapuru, in a farm

146

located 60 km southwest of Manaus (3.20 S, 60.60 W), a place that represents a time travel of 4

147

to 6 hours for the pollution plume. The Manaus metropolitan area has population estimated at

148

2.5 million, with about 650 thousand vehicles (2014/15), which could be compared to 1/10 of

149

MASP. During the years of 2014/15, a comprehensive aerosol field campaign was conducted

150

(Martin et al, 2016) on the site. Details about the experiments, including calibration and data

151

control are presented in different articles (Martin et al, 2016; Mei et al., 2013a; Mei et al.,

152

2013b; Thalman et al., 2017; and de Sá et al., 2017), and will concisely be described

153

hereinafter.

M AN U

SC

145

The aerosol sampled 10 m above the ground was first dried by a poly-tube Nafion

155

(Perma Pure, model 224 MD-110), reaching RH < 40%. A CCNC (Droplet Measurement

156

Technologies, Boulder, CO) was coupled to an Differential Mobility Analyzer (DMA) and a

157

condensation particle counter (CPC, TSI Inc., 3010) was employed to measure CCN activation

158

fraction of size-selected particles (Frank et al., 2006; Mei et al., 2013a; Moore et al., 2010;

159

Petters et al., 2007) and to determine aerosol hygroscopicity properties. The aerosol was

160

submitted to steady state charge distribution inside a Kr-85 aerosol charger (TSI, model 3077A)

161

before DMA measurements. After size selection by the DMA, particles were split in two and

162

simultaneously characterized by a CPC and a CCNC (Thalman et al., 2017). The seven particle

163

diameters (51, 75, 94, 112, 142, 171, and 222 nm) through which the DMA classified particles

164

size were submitted to the CCNC, being super saturation (SS) changed at each diameter by

165

stepping the flow rates and temperature gradient. Adequate statistics were considered from

166

1500 particles counted on the CPC, given SS a minimum of 30 s or up to a maximum of 120 s.

167

This implied that the measurement cycle through the seven particles sizes ranged 1 – 2 h,

168

depending on the particle number concentrations. The SS of the CCNC was calibrated using

169

ammonium sulfate particles (Mei et al., 2013b). At each operational set point, ranging from

170

0.075 – 1.1 %, the temperature fluctuation was considered, due to the course of the day.

AC C

EP

TE D

154

ACCEPTED MANUSCRIPT 171

The hygroscopicity of the individual particle sizes was derived from the activation

172

curves as described in Mei et al. (2013a). Initially, a lognormal function is used to fit the curves

173

of activated fraction (Ra) as a function of SS given in percent as

174 $% − $% ∗  1 + !" # +, 2 (2)*

175

RI PT

  ) =

being E the fraction of the particles active as CCN at a given SS; SS* the characteristic super

177

saturation and σs the coefficient related to the slope of the function, σs is associated to the

178

dispersion of values in which particles are activated as CCN. This function is used to represent

179

the physical phenomena of particle activation.

SC

176

Activation curves were determined by taking into account the influence of the multiple

181

charged particles using size distribution information from the SMPS measurements and

182

calculus of the influence of the charged particle activation curves measured twice or three times

183

for each size (Rose et al., 2008; Mei et al., 2013a; Mei et al., 2013b). The SS* is retrieved at

184

50% of the maximum point at which E is activated. The SS* is then used to calculate the

185

particle hygroscopicity according to κ-Köhler theory (Köhler, 1936; Petters and Kreidenweis,

186

2007) as follows:

TE D

187

M AN U

180

-=

4/ 2712 ∗

with

189

/=

190

and being Dp the particle diameter, and Mw, ρw and σw the molecular weight, density and surface

191

tension of pure water, respectively, with σw = 0.072 J m-2.

6784

,

AC C

192

34 54

EP

188

The chemical composition was quantitatively evaluated with High-Resolution Time-of-

193

Flight Aerosol Mass Spectrometer (HR-ToF-AMS, hereafter AMS; Aerodyne, Inc., Billerica,

194

Massachusetts, USA). The design principles and capabilities of this instrument are described in

195

DeCarlo et al. (2006) and Canagaratna et al., (2007).

196

The AMS inlet sampled aerosols from 5m above ground level. Organic, sulfate,

197

ammonium, nitrate, and chloride particulate matter (PM) mass concentrations were obtained

198

from “V-mode” data every other 4 min (de Sá et al., 2017). The choice of ions to fit the mass

199

resolving of instruments was aided by the “W-mode” data, which were collected for 1 day

200

every 5 days. Data analysis was performed using SQUIRREL (1.56D) and PIKA (1.14G) from

ACCEPTED MANUSCRIPT 201

the AMS software suite. Positive-matrix factorization was applied to the time series of the

202

organic component of the high-resolution mass spectra (Ulbrich et al., 2009).

203

For our analysis we focus on the measurements carried out during the first Intensive Operating Period (IOP1) from February 1, to March 31, 2014, period in which Manacapuru did

205

not receive substantial influence of biomass burning plume and the site experienced conditions

206

ranging from nearly natural to heavily polluted.

207 208

3. Hygroscopicities Used in κ-Köhler Theory

209

RI PT

204

The κ–Köhler theory can be applied to a multicomponent particle supposing internal

211

mixing rules defined by Zdanovskii, Stokes and Robinson (ZSR) (Stokes and Robinson, 1966;

213

Zdanovskii, 1948) by weighting hygroscopicity parameters -9 of each component according to their volume fractions, :9 , as

214 215

M AN U

212

SC

210

- = ∑ :9 -9 ,

with the corresponding -9 value for inorganic compounds, organic matter (OM), BC, insoluble crust debris and condensed water. The -9 values (Fountoukis and Nenes, 2007) of individual

217

components are listed in Table 1. Although the mixing rule for different particle components

218

refers to volume fractions, the mass fractions can be used as a first-order approximations

219

(Petters and Kreidenweis, 2007, 2008) if the densities of individual components are comparable

220

to the overall particle density. In general this assumption is realistic for particles consisting

221

mostly of organics and sulfate (Gunthe et al., 2009).

TE D

216

A relatively large number of components need to be considered in the composition of

223

atmospheric aerosols. The most abundant inorganic ions are sulfate (SO42−), followed by nitrate

224

(NO3−), and ammonium (NH4+), present in the aerosol formation as compounds H2SO4 and the

225

inorganic salts, such as NH4NO3, (NH4)3H(SO4)2 (letovicite, LC), (NH4)2SO4 and NH4HSO4. In

226

contrast with many places where they are found in trace amounts (van Pinxteren et al. 2009),

227

sodium and potassium salts, in MASP, needs to be considered in the fine mode, as will be

AC C

EP

222

228

further discussed. Insoluble material, like SiO , Al O and Fe O (oxides), also were taken into

229

account in the aerosol composition and volume fraction, as well as the condensed water. We

230

assume that elemental carbon (BC), crust debris, and oxides have -9 values equals to 0. Table 1

231

presents inorganics salts that are considered in the aerosol composition of this study.

232

Water-soluble organic components (WSOC) also are important in determining the CCN

233

activities of particles (Novakov and Penner, 1993; Saxena et al., 1995; Facchini et al., 1999).

234

WSOC is also associated with a major fraction of secondary organic aerosols (SOA), which is

ACCEPTED MANUSCRIPT formed by oxidation of volatile organic compounds (VOCs) followed by condensation on

236

existing particles or nucleation. Oxalic acid is the most abundant dicarboxylic acid in ambient

237

aerosol (Kawamura et al., 1996, Kawamura and Sakaguchi, 1999; Yu et al., 2005). It is formed

238

from oxidation of VOCs and aqueous phase chemistry in cloud droplets, as well as from

239

primary emissions from fossil fuel combustion, and biomass burning (Norton et al., 1983;

240

Kawamura and Kaplan, 1987; Kawamura et al., 1996; Warneck, 2003; Kawamura and Yasui,

241

2005; Sorooshian et al., 2006; Yu et al., 2005). The κoxalic ranges from 0.27 to 0.36

RI PT

235

242

In this study, the mean κorg value is estimated at 0.12, according to Thalman et al. (2017)

243

who measured κorg influenced by urban pollution downwind Manaus. The quite common value

of -<=> shown by Thalman et al. (2017) indicates that the biofuels used in Brazil do not induce

245

a significant modification of hygroscopicity on organic secondary aerosols from those in other

246

places. Also the organic fraction density of 1.2 g.cm-3 was taken into account in both

247

hygroscopicities and volume fraction.

M AN U

248

SC

244

249

4. Results

250

4.1 Aerosol Mass Distribution and Major Chemical Components

251

The PM mass size showed a bimodal distribution, in a coarse fraction (D50 from 3.2 to

253

10 µm) and in fine fraction (D50 from 0.18 to 0.56 µm), being the maximum concentrations

254

(3.71 ± 2.91 µg m-3) observed in ultra-fine particles, D50 < 0.020µm (Vieira-Filho et al., 2016).

255

Likewise, a bimodal distribution also was observed for PM sampled during a polluted period

256

(4–6 August 2012), while the highest concentrations occurred in fine fractions during

257

unpolluted period (1–5 September 2012) under postfrontal conditions in MASP (Albuquerque

258

et al., 2012). The ions distribution were segregated in two groups, in fine mode (D50 < 1µm)

259

with higher concentration of SO42-, NH4+, K+ and Na+; and in coarse mode (D50 > 1µm) in

260

which NO3-, Cl-, Na+, Ca2+ and Mg2+ were predominant (Vieira-Filho et al., 2016).

EP

AC C

261

TE D

252

Since for hygroscopicity calculus the aerosol fractions considers D50 < 0.56 µm (560

262

nm), the six fractions data are analyzed and discussed as follows. Average mass and chemical

263

composition distributions for fine fractions highlight unidentified compounds followed by

264

sulfate in modes 180 – 560 nm (Fig. 1). The fractions 100 – 560 nm presented a 10.9±6.3

265

µg m−3 mean mass concentration. Regarding the aerosol mode composition, from 15 August to

266 267 268

 5 September, 2012, the average mass concentrations of SO  , NO , Na , K , and NH were

0.44, 0.044, 0.027, 0.053 and 0.065 µg m−3, respectively. Sulfate was the most abundant

inorganic ion and together with NH , K , Na , and NO reached more than 96% of the total

ACCEPTED MANUSCRIPT

269

ion mass concentration (Tab. 2). In the D50 < 0.56 µm size range, Cl represents a very small

fraction, except for 20 nm, where it presents almost the same amount of mass of K . The two

271

organic components measured also represent small fraction of the total mass and are less

272

concentrated than inorganic components: acetate has the maximum concentration of 0.03

273

µg m−3 at 20 nm and values below 0.01 µg m−3 for larger diameters. Oxalate has a mean

274

concentration of about 0.03 µg m−3for all sizes. The chemical analysis shows that about 20% of

275

the particulate mass concentration in all diameters was composed of inorganic matter.

RI PT

270

Due to non determination of OM or BC during this sampling campaign, both were

277

considered as the unidentified compounds, totalizing 8.8 µg m−3 on average in the fraction 100

278

nm to 560 nm. This value was comparable to OM + BC (8.1 µg m−3) measured during

279

experiments in October 2012, using an Aerosol Chemical Speciation Monitor (ACSM) in

280

conjunction with a Multiangle Absortion Photometer (MAAP, Almeida et al., 2014). The BC

281

measurements reported elsewhere in MASP (Ynoue and Andrade, 2004, Albuquerque et al.,

282

2012, Miranda et al., 2012) accounted for ~40% of the fine fractions total mass. The

283

assumption of the undetermined mass as organic is also consistent with previous measurements

284

of organic matter in São Paulo, showing that it composes about 40% of fine particulate matter

285

mass (Castanho and Artaxo, 2001). Another study showed that the organic fraction was

286

somewhat larger, reaching about 50% of the aerosol mass (Almeida et al., 2014). The important

287

results from tunnel studies showed the contribution of the nitrogen and oxygen compounds to

288

fine aerosol organic compositions, which were attributed to vehicular emissions complexity in

289

MASP (Oyama et al., 2016).

TE D

M AN U

SC

276

The unidentified mass (OM and BC), as one of the main components of atmospheric

291

aerosol particles, was rather important for the ultrafine particles, being predominant in all

292

fractions, ranging from 65 to 85%. Also, for particles up to 100 nm, the OM and BC fractions

293

were even higher, reaching 85%. As the particle sizes increased, the fraction of OM+BC in the

294

particulate mass concentration decreased slightly. For particles larger than 560 nm, the

295

OM+BC mass fractions reached 67%.

AC C

296

EP

290

297

4.2 Determination of κ from the Measured and Estimated Particle’s Chemical

298

Composition

299 300

The ISORROPIA II model provides the chemical composition for particulate matter in

301

different sizes, which allows the estimation of κ mean values considering the contribution of

302

undetermined mass as OM with BC and without BC. To be consistent with observation, when

ACCEPTED MANUSCRIPT 303

we consider the presence of BC we use the mean size distribution values observed in previous

304

BC measurements, which have shown a maximum concentration of BC around 100 nm.

305

The κ values have one or two modes depending on the supposed composition of the undetermined mass. If the undetermined mass is composed of only OM, the largest values of κ

307

are observed at 100 nm (0.20). In the other hand, if we consider the presence of BC according

308

to their mean values observed, there can be two hygroscopic modes with mean κ values around

309

0.20. The two modes are produced by a decrease in hygroscopicity at 100 nm due to the peak of

310

BC mass concentration around this diameter. The BC concentration can reaches up to 50% of

311

the mass, on average, leading to a decrease of up to 40% in hygroscopicity in this size range.

312

For illustration, in Fig. 2 we present the estimated mean κ values considering the undetermined

313

mass as being only OM and also OM combined with BC at its maximum observed

314

concentration. For the other diameters the presence of BC causes smaller shift in κ values. For

315

180 nm the hygroscopicity can decrease by 30%. In all cases, variances are close to 0.05,

316

indicating the variance of the particle’s chemical compositions, which primarily result from

317

different amount of contributions from the inorganic sources.

SC

M AN U

318

RI PT

306

A remarkable character of the chemical composition is the relatively large concentration

319

of K  SO and Na SO for all the diameters. In fact, K  SO was the most abundant inorganic

320

component, contributing to, on average, about 5% of total volume of aerosols at 56 nm. The second most abundant component was NH HSO , followed by NH ) SO , and Na SO ,

TE D

321 322

respectively. One should also consider the fact that both K  SO and Na SO have large

323

densities and hygroscopicities.

K  SO and Na SO had the largest contribution to the mass at 56 nm, while NH ) SO

325

tended to increase its participation on the mass of particles of larger sizes. In the fine mode (D50

327 328 329 330

 < 100 nm) K+ and Na+ ensured SO  and NO completely combination, which avoided acid

particles formation. For 320 and 560 nm particles, the inclusion of NH and SO (or H SO ), in

AC C

326

EP

324

which SO predominates (or H SO ), leading to slightly acidic particles formation, commonly observed as NH HSO and LC.

K  SO , Na SO , and NH ) SO altogether, explained 51% and 35% of the

331

hygroscopicity for 56 and 100 nm particles sizes, respectively. K  SO and Na SO contribute

332

with 67 and 55% of those values, respectively, indicating their substantial importance on the

333

definition of κ values of those particles.

334

Fine mode particles seemed to be produced initially with the accumulation of relatively

335

large amount of K  SO (and probably Na SO ) followed by further growth by diffusive

336

condensation and coagulation of low-volatile organic compounds (Tissari et al., 2008). In this

ACCEPTED MANUSCRIPT 337

case, the process could impact the aerosols size distribution, promoting faster growth of

338

ultrafine mode particles, decreasing their concentration (Salvo et al., 2017). OM is the

339

predominant component for particles in these modes, contributing, on average, to more than

340

77% for the total mass and with 48% of the κ values.

341

Particles in the accumulation mode (D50 > 100 nm) seemed to grow from smaller

342

particles by both coagulation and condensation, undergoing growth by heterogeneous chemical reaction, probably by humid processes. K  SO , Na SO , NH ) SO and NH HSO or LC

344

altogether, contribution to 52% and 67% of the hygroscopicity for 180 and 320 nm particles

345

RI PT

343

sizes, respectively. The contribution of K  SO and Na SO of those particles were 39% and 43% for the values of inorganic κ, which indicates that, although NH ) SO is more

347

incorporated and predominates on the determination κ for particles of these range size, the

348

contribution of potassium and sodium sulfate is still substantial.

350 351

The hygroscopicities observed here were smaller than those observed in Liu et al.

M AN U

349

SC

346

(2014) due to two important facts: Liu et al. organic fraction of fine mode particles with large hygroscopicity values (-DEF > 0.2) and the organic fraction of accumulation mode with still

352

larger hygroscopicity values (-DEF > 0.3), being much more enriched with inorganic

353

compounds.

356 357 358 359

4.3 Parameterization of κ to AMS Data from Urban Pollution Impacted by Biofuels Off line techniques like the one used with MOUDI furnish practically all chemical information necessary to determine the hygroscopic behavior of particles from any size.

EP

355

TE D

354

In online techniques, nevertheless, as in the use of AMS, the speciation information is not complete. As a result, the κ values need to be determined using a parameterization scheme

361

based on available information (Gunthe et al., 2009; Dusek et al., 2010; Thalman et al., 2017).

362

In those cases the missing information is solved using a priori composition, which can be not

363

AC C

360

suitable in some occasions. For example, SO  can supposedly be completely or partially

364

neutralized with NH , or considered as H SO . The parameterization scheme can also assumes

365

that the volumetric fraction be replaced by the mass fraction, considering that all chemical

366

components have approximately the same density.

367

According to our study this kind of parameterization can produce a bad estimation of

368

hygroscopicity for three reasons. Firstly, due to large fraction of SO  associated with

369

potassium or sodium, which contrasts with the supposition that the aerosols can be acidic, as

370

observed by Almeida et al. (2014). Secondly, even if a total combination is assumed in the

ACCEPTED MANUSCRIPT

371

form of NH ) SO , the corresponding κ value is lower than the one from sodium or potassium sulfate. Finally, both sodium and potassium sulfate have a larger density than NH ) SO,

373

which can produce bad estimations of the inorganic component fraction inside the sample. This

374

implies that the current AMS-based parameterization schemes are not suitable for representing

375

aerosol hygroscopicity derived from Brazilian urban pollution, which indicates the need to set

376

up for a new parameterization to accurately predict the hygroscopicity parameter κ from the

377

chemical composition derived from AMS.

RI PT

372

An AMS analysis consider the mass concentration of inorganic ionic species such as

378

 SO  , NO , and NH and OM. To represent the contribution of the mass fraction of any of

380

those ions to the total hygroscopicity, we consider the κ value obtained taking into account the

381

volume fraction contribution of all salts formed by that ion as a function of its mass fraction. In

383 384 385 386 387

 our case, we plot the κ value obtained from the X + SO  (or X + NO ) volume fraction,

 -L M< , as a function of the mass fraction of SO  (or NO ), in which X stands for NH ) , N

K  , Na and H (Fig. 3 only for SO  ).

M AN U

382

SC

379

The individual inorganic species contribution to κ undergoes variation in all different sizes. Table 3 shows the slopes (O) and the correlation coefficient of the regression lines

 between κ obtained using the volumetric fraction of X + SO  and X + NO and the mass

 fraction of SO  and NO , respectively. As we can see, the mass fraction of these two ions is

389

highly correlated with the volumetric κ. For smaller sizes, however, the slopes are larger than

TE D

388

390

 the commonly attributed value of 0.6 for SO  and NO , which indicate major participation of

391

other components rather than the commonly assumed ammonium sulfate and ammonium

nitrate. Considering the size dependency of -M
393

linear regression relationship based on the least square method as:

395 396 397

-2E = OM
AC C

394

EP

392

where -2E represents κ calculated from the parameterization function, OT represents the

individual value of - for specie x as a function of the diameter and "T represents mass fraction of species x. To take into account the variations of O as a function of the diameter, OR), their

398

values were fit to a polynomial curve. NH does not appear on the parameterization because its

399

mass fraction is relatively small and its contribution to the hygroscopicity is taken into account

400

 on its combination with SO  and NO . This procedure avoids the noise associated with NH

401

measurements on AMS due its very high detection limit.

402

The parameterization above is of the same kind of those of Rose et al. (2011) and Liu et

403

al. (2014) and is based on the fact that the mass fractions is used as a first-order approximations

ACCEPTED MANUSCRIPT

404 405 406 407 408

to the volume fractions. In the case of Rose et al. (2011), OM
into account the fraction of NH and consider OPVWN = 0.6. O
Since the chemical composition defines the particle hygroscopicity, different measurement techniques should provide good agreement if the source of the aerosol is the

410

same. Takin Brazil as an example, the fuel source is almost the same all over the country:

411

gasohol (gasoline + ethanol anhydrous) and hydrated ethanol burned by light-duty vehicles

412

(mainly, flex fuel) and diesel with 7% biodiesel burned by heavy-duty vehicles (buses and

413

trucks). To test the applicability of this parameterization we involved a data set from a more

414

recent and complete study, developed in Manaus, a city far from MASP, where a

415

comprehensive aerosol field campaign was conducted.

SC

RI PT

409

Values of - derived from measurements in Manaus and those derived from chemical

417

composition in São Paulo are shown in Fig. 4a. To clarify, we present data considering both

418

absence and presence of BC on the samples of São Paulo. In Fig. 4b we show values of the

M AN U

416

parameterized-predicted -2E from the particle’s chemical composition obtained from AMS,

420

and from DMA+CPC+CCNC obtained both in Manaus. For comparison, we also included the

421

values obtained using the parameterization of Rose et al (2011).

422

TE D

419

In Fig. 4a, the values of Manaus are intermediate to those of São Paulo. The mean value observed in São Paulo at 56 nm is only 8 % larger than that observed at 51 nm in Manacapuru

424

when we consider the presence of BC on the sample, which suggest that the chemical

425

composition defining the hygroscopicity at that stage is representative of particles around the

426

EP

423

cut off diameter. The -2E values derived from chemical composition (Fig. 4b) presents

agreement within 10% with the DMA+CPC+CCNC – derived κ. The agreement of -2E shows

428

even the trend of growth of its value in smaller sizes, which indicates the consistency of the

429

parameterization and its applicability to air measurements affected by emission in Brazilian

430

urban areas. For particles in the accumulation size range ( d > 100 nm) there is a good

431

agreement between the observations and the parameterization of Rose et al. (2010). The values

432

of Rose’s parameterization are close to our parameterization at larger sizes ( d > 200 nm),

433

showing an increase trend due to the relatively large fraction of inorganic mass. At this size

434

range, both parameterizations represent approximately the same chemical composition, with a

435

large fraction of compounds with the same density. In contrast, the values of the

436

parameterization of Rose et al (2011) overestimate substantially (>30%) the observations at

AC C

427

ACCEPTED MANUSCRIPT 437

small sizes (d <90 nm), where there is the largest number of particles. The cause of this

438

overestimation is due to the relatively large addition of sulfate at smaller sizes and mainly to its

439

quite lower density when compared to sodium and potassium sulfate.

440 441

5. Discussion and Conclusions

442

In this study, aerosol hygroscopic properties were investigated based on the particle’s

RI PT

443

chemical composition analyzed from ambient aerosol samples collected daily using a cascade

445

impactor during the winter 2012, at an urban site, in São Paulo, Brazil. We evaluated variations

446

on chemical composition and its possible impacts of aerosols hygroscopicity as a function of

447

size. As a result, a parameterization for hygroscopicity κ was developed based on correlations

SC

444

448

 of the volumetric determined κ value and the mass fraction for SO  and NO and the assumed

449

OM.

451

Particles of all size investigated here were composed of inorganic, with SO  , NH ,

M AN U

450

K , Na , and NO together made up more than 96% of the total inorganic mass concentration.

452

A relatively large amount of K and Na is regularly observed in the fine mode of all the

453

samples. For fine particles, inorganic ions were not much less abundant than for larger

454

particles, although OM was estimated as about 77% of the total mass of small particles. The presence of sodium and potassium in the chemical composition of the ultrafine

TE D

455

fraction of the aerosols of São Paulo has long been recognized (Ynoue et al., 2004, for

457

example). Their sources, however, has only recently been on the focus of the investigation

458

(Vieira-Filho et al., 2013), leading to the conclusion that the megacity itself is predominantly

459

involved in the process of emission and removal of the aerosol and the influence of external

460

components is limited to circumstantial events. The evidence of vehicular contribution (direct emissions), comes from the fact that the

AC C

461

EP

456

462

matrices used in biofuels production can assimilate several inorganic components from the soil

463

and water during its development, and also during industrial processing. Consequently, those

464

constituents can be part of the final composition. Biodiesel, for example, can reach as much as

465

28 mg of Na kg-1(Oliveira et al. 2009), while Ethanol can have as much as 1.5 mg Na L-1

466

(Oliveira et al. 2002). On the other hand, Brazilian distilleries generate huge volumes of

467

potassium-rich vinasse, a sub product of ethanol that is used for sugar cane fertiirrigation to

468

reduce chemical fertilizers. This fact allows us to conclude that, although there are no

469

measurements, potassium must also be a constituent of ethanol.

ACCEPTED MANUSCRIPT 470

The presence of BC can cause dramatic shift in hygroscopicity. The larger impact can

471

be at 100 nm, since observations indicate that the amount of BC can, on average, reaches up to

472

50% of the mass at that size, which can reduce κ in about 40%. The mean κ values is around

473

0.20 in both 56 and 180 nm size fraction.

474 475

Despite the relatively low mass contribution of inorganic compounds, they contribute

most to the observed hygroscopicity. In fact, the four major inorganic compounds (K  SO ,

Na SO , NH ) SO and NH HSO ) accounted for more than 55% on average of the κ value

477

for smaller particles. This contribution is very important for ultrafine particles to act as CCN.

478

For particles of around 56 nm, for example, the contribution of inorganics can increase the

479

hygroscopicity from around 0.12 to about 0.21. Most of that increase is done by the

RI PT

476

contribution of K  SO and Na SO . Due to the inorganic enrichment, the critical

481

supersaturation for those particles could decrease from 0.9% to 0.6%, which can allow a

482

substantial increase of the number of particles to be able to nucleate as cloud droplets at a lower

483

supersaturation.

M AN U

484

SC

480

The -2E values obtained from the parameterization scheme agreed well with those

485

obtained from observation close to Manaus, and data points ranged around 15% deviation for

486

particles < 100 nm diameters, demonstrating the quality of the parameterization. In addition,

487

the concentration of NH was not necessary on the parameterization, which avoids the noise

associated with measurements of NH on AMS due to its very high detection limits. The new

489

parameterization is also able to correct bad estimations from the parameterization of Rose et al

490

(2011), for example. In this context, the presence of sodium and potassium is a modifying

491

component of the hygroscopicity characteristics of the observed aerosols, producing relatively

492

high addition of sulfate in diameters in which lower values would be expected. Such

493

modification might have activated more CCN at a given supersaturation than in environments

494

not influenced by the same aerosol source. These aspects also influence a particles density,

495

which is main cause of parameterization overestimation proposed by Rose et al (2011).

EP

AC C

496

TE D

488

The small difference observed among the -2E values might have come from daily

497

variability in both sites and also due to the fact that chemistry composition for MOUDI

498

represented a large range of diameters, while in the Manaus arrangement the measurement are

499

made at a very narrow range of diameters, which prevent one to not have a good definition of

500

which component influences more at a given diameter or if there are abrupt properties changes

501

as a function of the size, as is shown in Manaus arrangement. In spite of these facts, even the

502

large values of κ for the smallest sizes measured were evidenced using the parameterization

503

proposed. The agreement seems to be much more caused by the same fuel burning than to the

ACCEPTED MANUSCRIPT 504

analysis system, once online techniques cannot currently provide as much detailed speciation

505

information as may be available from offline techniques.

506

Despite the similarities of chemical composition of the smallest particles, aerosols observed in Liu et al. (2014) have much larger hygroscopicities than those described here. For

508

the smallest particles, this is due to the larger values of the corresponding associated organic

509

hygroscopicity, while for larger particles the discrepancies is accentuated by the large inorganic

510

enrichment.

511

RI PT

507

Observations described above seem to agree with previous studies of biomass burning.

Accordingly, ultrafine K  SO rich particles (ash) would be formed either as homogeneous

513

nucleation or as heterogeneous nucleation and condensation on the existing soot particles. First,

514

the alkali sulfate vapors would condense via gas phase sulfation reaction between alkali

SC

512

hydroxides and chlorides and gaseous sulfur (presumably SO ) (Jokiniemi et al.,1994; Iisa et

516

al., 1999; Jensen et al., 2000; Sippula et al., 2007,). However, the recent findings have shown

517

that soot and ash materials are primarily as separate particles (Tissari et al., 2008), indicating

518

that soot particles are not likely to act as seeds for the condensing ash species. In addition, the

519

additional growth of particles seems to be mainly determined by OM condensation. In all cases,

M AN U

515

the fine particles are mainly formed of potassium salts (K  SO , KCl, K  CO ), with

521

thermodynamics imposing the formation of alkali sulfates instead of chlorides and likewise

522

formation of chlorides instead of carbonates. Thus, when there are any excess of sulfates in the

523

gas phase, almost no chlorine will be found in the particulate phase.

TE D

520

In any case, if the submicron aerosols studied here are formed by gas-to-particle

525

conversion, this will produce particles with spherical aggregate-like morphology, which cannot

526

result from breakup processes.

528

The chemical analysis of particles shown here is an important step towards

AC C

527

EP

524

understanding biofuels impact on particle pollution in urban air. The use of ethanol-rich fuel

529

blends inducing the formation of new particles composed of K  SO and Na SO is a feasible

530

candidate to explain the significant reduction in ultrafine – specifically < 50 nm diameters

531

nanoparticles – levels in São Paulo. Salvo et al. (2017), for example, analyzed aerosol,

532

meteorological, traffic, and consumer behavior data and find, empirically, that ambient number

533

concentrations of those nanoparticles rise by one-third during the morning commute when

534

higher ethanol prices induce drivers in São Paulo to substitute to gasoline use. The conclusion

535

is also supported by the data observed during October 2014 (Almeida et al., 2014), where it is

536

possible to verify that when the amount of potassium and sodium increase on the samples, the

537

aerosol size distribution tends to becomes bimodal due to the decrease of particles in ultrafine

ACCEPTED MANUSCRIPT 538

mode. Conversely, when less soluble particles are formed, a relatively larger amount of

539

ultrafine particles are observed.

540

The studies by Dusek et al. (2006) and Zhang et al. (2011) found that particle size is most important aspect defining aerosol ability to act as CCN. A few others studies,

542

nevertheless, revealed that the chemical composition is also critical (Almeida et al., 2014;

543

Ervens et al., 2005; Lance et al., 2004; McFiggans et al., 2006; Nenes et al., 2002; Wang et al.,

544

2008). Considering that aerosol fine mode ( d < 100 nm ) has a larger contribution to the CCN

545

concentration than the coarse-mode (Nakajima et al., 2001), aerosol index derived from urban

546

pollution in Brazil can impact cloud formation and properties in two folds: increasing its

547

concentration and hygroscopicity.

SC

RI PT

541

548 549

Acknowledgements

The authors would like to acknowledges, with thanks, Shannon Capps and Athanasios

551

Nenes, who kindly provided the Isorropia II Model for the use in this work. Part of this

552

research has been supported by FAPESP (2008/58104-8, Project NUANCE – SPS - Narrowing

553

the uncertainties on aerosol and climate changes in São Paulo State). We thank to CAPES

554

(Projeto MODELAGEM and PROEX, Post-Graduation Program of Meteorology, IAG/USP)

555

and CNPq for the student grants provided. We acknowledge the Office of Biological and

556

Environmental Research of the Office of Science of the United States Department of Energy,

557

specifically the Atmospheric Radiation Measurement (ARM). We also acknowledge two

558

anonymous reviewers who gave important suggestions on the improvement of the text.

559

561 562

References

Albuquerque, T. T, Andrade, M. F., Ynoue, R. Y., 2012. Characterization of

AC C

560

EP

TE D

M AN U

550

563

atmospheric aerosols in the city of São Paulo, Brazil: comparisons between polluted and

564

unpolluted periods, Environ. Monit. Assess., 184, 969–984, doi:10.1007/s10661-011-2013-y.

565

Almeida, G. P., Brito, J., Morales, C. A., Andrade, M. F., Artaxo, P., 2014.: Measured

566

and modelled cloud condensation nuclei (CCN) concentration in São Paulo, Brazil: the

567

importance of aerosol size-resolved chemical composition on CCN concentration prediction,

568

Atmos. Chem. Phys., 14, 7559–7572, doi:10.5194/acp-14-7559-2014.

569

Canagaratna, M. R., Jayne, J. T., Jimenez, J. L., Allan, J. D., Alfarra, M. R., Zhang, Q.,

570

Onasch, T. B., Drewnick, F., Coe, H., Middlebrook, A., Delia, A., Williams, L. R., Trimborn,

571

A. M., Northway, M. J., DeCarlo, P. F., Kolb, C. E., Davidovits, P., and Worsnop, D. R.:

ACCEPTED MANUSCRIPT 572

Chemical and microphysical characterization of ambient aerosols with the aerodyne aerosol

573

mass spectrometer, Mass Spectrom. Rev., 26, 185–222, https://doi.org/10.1002/mas.20115,

574

2007.

576 577 578 579

Castanho, A. D. A., Artaxo, P, 2001. Wintertime and summertime São Paulo aerosol source apportionment study. Atmospheric Environment, 35: 4889-4902. Cetesb (2017) Relatório de qualidade do ar no Estado de São Paulo, 2016 (in Portuguese—www.cetesb.sp.gov.br) Accessed December 2017.

RI PT

575

Christensen, K. A., and Livbjerg H. (1996) A Field Study of Submicron Particles from the Combustion of Straw, Aerosol Science and Technology, 25:2, 185-199,

581

DOI:10.1080/02786829608965390

582

SC

580

de Sá, S. S., Palm, B. B., Campuzano-Jost, P., Day, D. A., Newburn, M. K., Hu, W., Isaacman-VanWertz, G., Yee, L. D., Thalman, R., Brito, J., Carbone, S., Artaxo, P., Goldstein,

584

A. H., Manzi, A. O., Souza, R. A. F., Mei, F., Shilling, J. E., Springston, S 832 . R., Wang, J.,

585

Surratt, J. D.,Alexander, M. L., Jimenez, J. L., and Martin, S. T.: Influence of urban pollution

586

on the production of organic particulate matter from isoprene epoxydiols in central Amazonia,

587

Atmos. Chem. Phys. Discuss., 2016, 1-58, 2016.

M AN U

583

DeCarlo, P. F., Kimmel, J. R., Trimborn, A., Northway, M. J., Jayne, J. T., Aiken, A.

589

C., Gonin, M., Fuhrer, K., Horvath, T., Docherty, K. S., Worsnop, D. R., and Jimenez, J. L.:

590

Field-deployable, high-resolution, time-of-flight aerosol mass spectrometer, Anal. Chem., 78,

591

8281–8289, https://doi.org/10.1021/ac061249n, 2006

592

TE D

588

Dusek, U., Frank, G. P., Curtius, J., Drewnick, F., Schneider, J., Kürten, A., Rose, D., Andreae, M. O., Borrmann, S., Pöschl, U., 2010. Enhanced organic mass fraction and

594

decreased hygroscopicity of cloud condensation nuclei (CCN) during new particle formation

595

events, Geophys. Res. Lett., 37, L03804, doi:10.1029/2009GL040930, Dusek, U., Frank, G. P., Hildebrandt, L., Curtius, J., Schneider, J., Walter, S., Chand,

AC C

596

EP

593

597

D., Drewnick, F., Hings, S., Jung, D., Borrmann, S., Andreae, M. O., 2006. Size matters more

598

than chemistry for cloud-nucleating ability of aerosol particles. Science, 312(5778), 1375–

599

1378. https://doi.org/10.1126/science.1125261

600

Ervens, B., Feingold, G., & Kreidenweis, S. M. (2005). Influence of water-soluble

601

organic carbon on cloud drop number concentration. Journal of Geophysical Research, 110,

602

D18211. https://doi.org/10.1029/2004JD005634

603

Facchini, M. C., Fuzzi, S., Zappoli, S., Andracchio, A., Gelencsér, A., Kiss, G.,

604

Krivácsy, Z., Mészáros, E., Hansson, H.-C., Alsberg, T., Zebühr, Y., 1999. Partitioning of the

ACCEPTED MANUSCRIPT 605

organic aerosol component between fog droplets and interstitial air, J. Geophys. Res. Atmos.,

606

104, 26821–26832.

607

Fountoukis, C., Nenes, A., 2007. ISORROPIA II: a computationally efficient

608

thermodynamic equilibrium model for K+–Ca2+–Mg2+–NH4+–Na+–SO42−–NO3−–Cl−–H2O

609

aerosols, Atmos. Chem. Phys., 7, 4639-4659, https://doi.org/10.5194/acp-7-4639-2007. Gunthe, S. S., King, S. M., Rose, D., Chen, Q., Roldin, P., Farmer, D. K., Jimenez, J.

611

L., Artaxo, P., Andreae, M. O., Martin, S. T., Pöschl, U., 2009. Cloud condensation nuclei in

612

pristine tropical rainforest air of Amazonia: size-resolved measurements and modeling of

613

atmospheric aerosol composition and CCN activity, Atmos. Chem. Phys., 9, 7551–7575,

614

doi:10.5194/acp-9-7551-2009.

617 618 619

SC

616

IBGE-Cidades, 2018: https://cidades.ibge.gov.br/brasil/am/manaus/panorama (January 19, 2018).

Iisa, K., Lu, Y., Salmenoja, K., 1999. Sulfation of potassium chloride at combustion

M AN U

615

RI PT

610

conditions. Energy & Fuels 13, 1184–1190.

Jensen, J.R., Nielsen, L.B., Schultz-Moller, C., Wedel, S., Livbjerg, H., 2000. The

620

nucleation of aerosols in flue gases with a high content of alkali – a laboratory study. Aerosol

621

Science and Technology 33, 490–509.

Jokiniemi, J., Lazaridis, M., Lehtinen, K., Kauppinen, E.J., 1994. Numerical simulation

623

of vapour–aerosol dynamics in combustion processes. Journal of Aerosol Science 25, 429–446.

624

TE D

622

Kawamura, K., Kasukabe, H., Barrie, L.A., 1996. Source and reaction pathways of dicarboxylic acids, ketoacids and dicarbonyls in arctic aerosols: one year of observations.

626

Atmospheric Environment 30 (10–11), 1709–1722

EP

625

Kawamura, K., Sakaguchi, F., 1999. Molecular distributions of water soluble

628

dicarboxylic acids in marine aerosols over the Pacific Ocean including tropics. J. Geophys.

629

Res. 104, 3501–3509.

630

AC C

627

Kawamura, K., Yasui, O., 2005. Diurnal changes in the distribution of dicarboxylic

631

acids, ketocarboxylic acids and dicarbonyls in the urban Tokyo atmosphere, Atmos. Environ.,

632

39, 1945 – 1960, doi:10.1016/j.atmosenv. 2004.12.014

633 634 635 636

Kawamura, K.,Kaplan, I. R., 1987. Dicarboxylic acids generated from thermal alteration of geopolymers. Geochim. Cosmochim. Acta 51, 3201–3207. Kunkel, D., H. Tost, and M. G. Lawrence, Aerosol pollution potential from major population centers, Atmos. Chem. Phys., 13, 4203–4222, 2013. doi:10.5194/acp-13-4203-2013

ACCEPTED MANUSCRIPT 637

Lance, S., Nenes, A., & Rissman, T. A. (2004). Chemical and dynamical effects on

638

cloud droplet number: Implications for estimates of the aerosol indirect effect. Journal of

639

Geophysical Research, 109, D22208. https://doi.org/10.1029/2004JD004596

640

Lawrence, M. G., Butler, T. M., Steinkamp, J., Gurjar, B. R., and Lelieveld, J.: Regional pollution potentials of megacities and other major population centers, Atmos. Chem. Phys., 7,

642

3969– 3987, doi:10.5194/acp-7-3969-2007, 2007.

643

RI PT

641

Liu, H. J., Zhao, C. S., Nekat, B., Ma, N., Wiedensohler, A., van Pinxteren, D.,

Spindler, G., Müller, K., Herrmann, H., 2014. Aerosol hygroscopicity derived from size-

645

segregated chemical composition and its parameterization in the North China Plain, Atmos.

646

Chem. Phys., 14, 2525-2539, https://doi.org/10.5194/acp-14-2525-2014,

647

SC

644

Martin, S. T., Artaxo, P., Machado, L. A. T., Manzi, A. O., Souza, R. A. F., Schumacher, C., Wang, J., Andreae, M. O., Barbosa, H. M. J., Fan, J., Fisch, G., Goldstein, A.

649

H., Guenther, A., Jimenez, J. L., Pöschl, U., Silva Dias, M. A., Smith, J. N., Wendisch, M.,

650

2016. Introduction: Observations and Modeling of the Green Ocean Amazon

651

(GoAmazon2014/5), Atmos. Chem. Phys., 16, 4785-4797.

652

M AN U

648

McFiggans, G., Artaxo, P., Baltensperger, U., Coe, H., Facchini, M. C., Feingold, G., Fuzzi, S., Gysel, M., Laaksonen, A., Lohmann, U., Mentel, T. F., Murphy, D. M., O’Dowd, C.

654

D., Snider, J. R., Weingartner, E., 2006. The effect of physical and chemical aerosol properties

655

on warm cloud droplet activation, Atmos. Chem. Phys., 6, 2593–2649, doi:10.5194/acp-6-

656

2593-2006,

657

TE D

653

McFiggans, G., Artaxo, P., Baltensperger, U., Coe, H., Facchini, M. C., Feingold, G., et al. (2006). The effect of physical and chemical aerossol properties on warm cloud droplet

659

activation. Atmospheric Chemistry and Physics, 6(9), 2593–2649. https://doi.org/10.5194/acp-

660

6-2593-2006

Mei, F., P. L. Hayes, A. Ortega, J. W. Taylor, J. D. Allan, J. Gilman, W. Kuster, J. de

AC C

661

EP

658

662

Gouw, J. L. Jimenez, and J. Wang (2013), Droplet activation properties of organic aerosols

663

observed at an urban site during CalNex-LA, J. Geophys. Res. Atmos., 118, 2903–2917,

664

doi:10.1002/jgrd.50285.

665

Mei, F., Setyan, A., Zhang, Q., and Wang, J.: CCN activity of organic aerosols observed

666

downwind of urban emissions during CARES, Atmospheric Chemistry and Physics, 13, 12155-

667

12169, 2013b.

668

Miranda, R.M., Andrade, M.F., Fornaro, A., Astolfo, R., Andre, P.A., Saldiva, P., 2012.

669

Urban air pollution: a representative survey of PM2.5 mass concentrations in six Brazilian

670

cities.. Air Qual Atmos Health (2012) 5: 63. https://doi.org/10.1007/s11869-010-0124-1

ACCEPTED MANUSCRIPT Molina, M.J., L.T. Molina, 2004. Megacities and atmospheric pollution. J. Air Waste

671 672

Manag. Assoc., 54. pp. 644-680.

673

Nakajima, T., Higurashi, A., Kawamoto, K., & Penner, J. E. (2001). A possible

674

correlation between satellite-derived cloud and aerosol microphysical parameters. Geophysical

675

Research Letters, 28(7), 1171–1174. https://doi.org/10.1029/2000GL012186

RI PT

Nenes, A., Charlson, R. J., Facchini, M. C., Kulmala, M., Laaksonen, A., & Seinfeld, J.

676 677

H. (2002). Can chemical effects on cloud droplet number rival the first indirect effect?

678

Geophysical Research Letters, 29(17), 1848. https://doi.org/10.1029/2002GL015295

Norton R. B., Roberts J. M., Huebert B. J., 1983.Tropospheric oxalate. Geophys. Res.

679

Lett. 10, 517– 520.

SC

680

Novakov, T., Penner, J.E., 1993. Large Contribution of Organic Aerosols to Cloud-

681

Condensation Nuclei Concentrations. Nature, 365, 823-826.

683

http://dx.doi.org/10.1038/365823a0

M AN U

682

684

Oliveira, A. P.; Okumura, L. L.; Gomes Neto, J. A.; Moraes, M., 2002. Evaluation of

685

analyte additions method for sodium determination in fuel ethanol by flame atomic emission

686

spectrometry Eclet. Quím. [online]., 27, 285-291, 2002 (in Portugues )..

687

http://dx.doi.org/10.1590/S0100-46702002000200024.

Oliveira, A. P.; Villa, R.D.; Antunes, K.C.P.; Magalhães, A.; Castro e Silva, E.; 2009.

TE D

688 689

Determination of sodium in biodiesel by flame atomic emission spectrometry using dry

690

decomposition for the sample preparation,Fuel, 88, 764–766. Oyama S., Andrade M. F., Herckes P., Dusek U., Röckmann T., Holzinger R., 2016.

691

Chemical characterization of organic particulate matter from on-road traffic in São Paulo,

693

Brazil, Beatriz Atmos. Chem. Phys., 16, 14397–14408, doi:10.5194/acp-16-14397-2016

EP

692

Petters, M. D. and Kreidenweis, S. M.: A single parameter representation of

694

hygroscopic growth and cloud condensation nucleus activity, Atmos. Chem. Phys., 7, 1961–

696

1971, doi:10.5194/acp-71961-2007, 2007.

AC C

695

Petters, M. D. and Kreidenweis, S. M.: A single parameter representation of

697 698

hygroscopic growth and cloud condensation nucleus activity Part 2: Including solubility,

699

Atmos. Chem. Phys., 8, 6273–6279, doi:10.5194/acp-8-6273-2008, 2008.

700

Petters, M. D., Kreidenweis, S. M., 2007. A single parameter representation of

701

hygroscopic growth and cloud condensation nucleus activity, Atmos. Chem. Phys., 7, 1961–

702

1971.

703 704

Rose, D., Gunthe, S. S., Su, H., Garland, R. M., Yang, H., Berghof, M., Cheng, Y. F., Wehner, B., Achtert, P., Nowak, A., Wiedensohler, A., Takegawa, N., Kondo, Y., Hu, M.,

ACCEPTED MANUSCRIPT 705

Zhang, Y., Andreae, M. O., and Pöschl, U.: Cloud condensation nuclei in polluted air and

706

biomass burning smoke near the megacity Guangzhou, China – Part 2: Size-resolved aerosol

707

chemical composition, diurnal cycles, and externally mixed weakly CCN-active soot particles,

708

Atmos. Chem. Phys., 11, 2817–2836, doi:10.5194/acp-11-2817-2011, 2011.

709

Salvo, A., Brito, J., Artaxo, P., and Geiger, F. M., Reduced ultrafine particle levels in São Paulo’s atmosphere during shifts from gasoline to ethanol use. DOI: 10.1038/s41467-017-

711

00041-5, Nature Communications, 2017.

712

RI PT

710

Saxena, P., L.M. Hildemann, P.H. McMurry, J.H. Seinfeld, 1995. Organics alter

hygroscopic behavior of atmospheric particles Journal of Geophysical Research, 100,1995,

714

pp. 18755-18770.

716 717 718

Sippula, O., Hyto¨ nen, K., Tissari, J., Raunemaa, T., Jokiniemi, J., 2007. Effect of wood fuel on the emissions from a top-feed pellet stove. Energy & Fuels 21, 1151–1160. Sorda, G., Banse, M., Kemfert, C., 2010. An overview of biofuel policies across the

M AN U

715

SC

713

world, 38, 6977–6988.

Sorooshian, A., Varutbangkul, V., Brechtel, F. J., Ervens, B., Feingold, G., Bahreini, R.,

720

Murphy, S. M., Holloway, J. S., Atlas, E. L., Buzorius, G., Jonsson, H., Flagan, R. C., Seinfeld,

721

J. H., 2006. Oxalic acid in clear and cloudy atmospheres: Analysis of data from International

722

Consortium for Atmospheric Research on Transport and Transformation 2004. Jour. Geophys.

723

Research, 111 (D23), 17.

724 725

TE D

719

Stokes, R. H. and Robinson, R. A.: Interactions in Aqueous Nonelectrolyte Solutions .I. Solute-Solvent Equilibria, J. Phys. Chem., 70, 2126–2131, doi:10.1021/j100879a010, 1966. Swietlicki, E., Hansson, H. —C., Hämeri, K., Svenningsson, B., Massling, A.,

727

McFiggans, G., McMurry, P. H, Petäjä, T., Tunved, P., Gysel, M., Topping, D., Weingartner,

728

E., Baltensperger, U., Rissler, J., Wiedensohler, A., Kulmala, M., 2008. Hygroscopic properties

729

of submicrometer atmospheric aerosol particles measured with H-TDMA instruments in

730

various environments—a review. Tellus, 60B, 432–469.

AC C

731

EP

726

Thalman, R., de Sá, S. S., Palm, B. B., Barbosa, H. M. J., Pöhlker, M. L., Alexander, M.

732

L., Brito, J., Carbone, S., Castillo, P., Day, D. A., Kuang, C., Manzi, A., Ng, N. L., Sedlacek

733

III, A. J., Souza, R., Springston, S., Watson, T., Pöhlker, C., Pöschl, U., Andreae, M. O.,

734

Artaxo, P., Jimenez, J. L., Martin, S. T., and Wang, J.: CCN activity and organic

735

hygroscopicity of aerosols downwind of an urban region in central Amazonia: seasonal and diel

736

variations and impact of anthropogenic emissions, Atmos. Chem. Phys., 17, 11779-11801,

737

https://doi.org/10.5194/acp-17-11779-2017, 2017.

ACCEPTED MANUSCRIPT 738

Tissari, J., Lyyra¨nen, J., Hyto¨ nen, K., Sippula, O., Tapper, U., Frey, A., Saarnio,

739

K.,Pennanen, A.S., Hillamo, R., Salonen, R.O., Hirvonen, M.-R., Jokiniemi, J., 2008.Fine

740

particle and gaseous emissions from normal and smouldering wood combustion in a

741

conventional masonry heater. Atmospheric Environment 42, 7862–7873. Ulbrich,I.M.,Canagaratna,M.R.,Zhang,Q.,Worsnop,D.R.,and Jimenez, J. L.:

742

Interpretation of organic components from Positive Matrix Factorization of aerosol mass

744

spectrometric data, Atmos. Chem. Phys., 9, 2891–2918, https://doi.org/10.5194/acp-92891-

745

2009, 2009.

RI PT

743

van Pinxteren, D., Brüggemann, E., Gnauk ,T., Iinuma, Y., Müller, K., Nowak, A.,

746

Achtert, P., Wiedensohler, A., and Herrmann, H.,2009. Size- and time-resolved chemical

748

particle characterization during CAREBeijing-2006: Different pollution regimes and diurnal

749

profiles, J. Geophys. Res., 114, D00G09, doi:10.1029/2008JD010890.

SC

747

M AN U

Vieira-Filho, M., Pedrotti, J., Fornaro, A., 2016. Water-soluble ions species of size-

750 751

resolved aerosols: Implications for the atmospheric acidity in São Paulo megacity, Brazil,

752

Atmospheric Research 181, 281–287.

Vieira-Filho, M.S., Pedrotti, J.J., Fornaro, A., 2013. Contribution of long and mid-range

753

transport on the sodium and potassium concentrations in rainwater samples, São Paulo

755

megacity, Brazil. Atmospheric Environment, 79, 299-307

TE D

754

Wang, J., Lee, Y. N., Daum, P. H., Jayne, J., & Alexander, M. L. (2008). Effects of

756

aerosol organics on cloud condensation nucleus (CCN) concentration and first indirect aerosol

758

effect. Atmospheric Chemistry and Physics, 8(3), 9783–9818. https://doi.org/10.5194/acpd-8-

759

9783-2008

EP

757

Warneck, P., 2003. In-cloud chemistry opens pathway to the formation of oxalic acid in

760 761

the marine atmosphere, Atmos. Environ., 37, 2423–2427, 2003. Ynoue, R. Y., Andrade, M. F., 2004. Size-Resolved Mass Balance of Aerosol Particles

AC C

762 763

over the São Paulo Metropolitan Area of Brazil. Aerosol Science and Technology, 38(S2):52–

764

62, 2004. DOI: 10.1080/02786820490466756 Yu, J. Z., Huang, S. F., Xu, J. H., Hu, M., 2005. When aerosol sulfate goes up, so does

765 766

oxalate: Implication for the formation mechanisms of oxalate, Environ. Sci. Technol., 39, 128 –

767

133.

768 769

Zdanovskii, A. B.: Novyi Metod Rascheta Rastvorimostei Elektrolitov V Mnogokomponentnykh Sistemakh .1, Zhurnal Fizicheskoi Khimii, 22, 1478–1485, 1948.

ACCEPTED MANUSCRIPT Zhang, Q., Quan, J., Tie, X., Huang, M., & Ma, X. (2011). Impact of aerosol particles

771

on cloud formation: Aircraft measurements in China. Atmospheric Environment, 45(3), 665–

772

672. https://doi.org/10.1016/j.atmosenv.2010.10.025

AC C

EP

TE D

M AN U

SC

RI PT

770

ACCEPTED MANUSCRIPT 1

Table 1. The inorganics salts that are considered in the inorganic aerosol composition along

2

with its measured hygroscopicity. κ were taken from Fountoukis and Nenes (2007).

SC

RI PT

kappa 0.61 0.70 0.67 0.93 0.60 0.24 0.63 0.98 0.78 1.10 0.00 0.76 0.82 1.24 0.87 0.89 0.69 0.53 0.59 0.81

M AN U

Inorg. salt (NH4)2SO4 NH4HSO4 NH4NO3 NH4Cl (NH4)3H(SO4)2 (LC) MgSO4 Mg(NO3)2 MgCl2 CaCl2 Ca(NO3)2 CaSO4 Na2SO4 NaHSO4 NaCl NaNO3 KCl K2SO4 KNO3 KHSO4 H2SO4

Table 2.The average particulate inorganic chemical composition from measurements in São

5

Paulo during the winter of 2012.



0

0.113

0.078

0.041

0.864

0.008

Total mass (µgm-3) 3.58

0.107

0.078

0.017

0.725

0.006

3.49

0.066

0.066

0.012

0.437

0.004

2.96

0.029

0.034

0.008

0.176

0.004

1.81

0.017

0.013

0.018

0.009

0.074

0.003

0.62

0.033

0.017

0.027

0.017

0.124

0.025

3.84

0.030

180

0.026

100

0.017

20

EP

320



AC C

Diameter   (nm) 560 0.045

56

TE D

3 4





6 7

Table 3. γ for SO = 1.784 × 10   − 1.205 × 10   + 0.8167 with  is defined as  

8

% = 0.99, while γ for NO is () = −8.341 × 10 +   + 1.309 × 10  − 4.969 +

9

1.012, with % = 1.00. Values are valid for  from 50 to 560 nm. Diameter (nm) 56 100

 

r2 (SO  ) 0.76 0.92 0.72 0.97

ACCEPTED MANUSCRIPT 180 320 560

0.65 0.62 0.7

0.93 0.93 0.89

10

() 0.77 0.67 0.49 0.49 0.87

r2 (NO ) 0.91 0.97 0.62 0.16 0.79

RI PT

Diameter (nm) 56 100 180 320 560

AC C

EP

TE D

M AN U

SC

11

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

1

Figure 1) mean aerosol mass concentration for 6 different stages during the period of

3

investigation. Ox+Ac represent Oxalate and acetate.

AC C

EP

TE D

2

4 5

Figure 2. Hygroscopicity derived from size segregated aerosol mass in São Paulo, Brazil,

6

during the winter 2012.

represents inorganic compounds + OM, while

represents

ACCEPTED MANUSCRIPT inorganic compounds + OM + BC. Points represent mean values while bars represent standard

8

deviation. The points representing inorganic compounds + OM + BC illustrate the situation

9

when the BC mass are at its maximum fraction observed.

M AN U

SC

RI PT

7

AC C

EP

TE D

10

11

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

12

Figure 3. Volumetric hygroscopicity of X + SOଶି ସ (y-axis) derived as a function of the mass

14

fraction of SO4 (x-axis) for different sizes a) 180 nm; b) 100 nm; and c) 56 nm. (Slope of

15

ି regression line and correlation coefficients are shown in Table 3 for SOଶି ସ and NOଷ ). The

16

figures corresponding to NOି ଷ are not shown.

AC C

EP

TE D

13

17

ACCEPTED MANUSCRIPT

M AN U

SC

RI PT

18

19 20

Figure 4. Comparison of Hygroscopicity a) measured in Manaus (using DMA+CPC+CCNC

21

and represented by

22

and considering the undetermined mass as being OM, represented by

23

represented by

TE D

) and ߢ௣௔௥ values (section 4.1, obtained from AMS data from Manaus and represented by ). Points represent mean values while bars represent standard deviation.

26

parameterization from Rose et (2011).

AC C

25

27

; and OM+BC,

), and b) measured in Manaus (using the DMA+CPC+CCNC and represented

EP

24

) and chemically derived from data in São Paulo (using MOUDI and IC

represents the

ACCEPTED MANUSCRIPT Characterization of the inorganic chemical composition and Aerosols hygroscopicity in urban environments in Brazil Modifications of aerosol size distribution on urban environments due to biofuels New parameterization of aerosol hygroscopicity based on aerosol mass fraction from AMS in Brazil

AC C

EP

TE D

M AN U

SC

RI PT

Mechanism of particles formation in environments impacted by biofuels

ACCEPTED MANUSCRIPT Conflicts of Interest Statement Manuscript Title: Characterization of Aerosol Chemical Composition from Urban Pollution In Brazil and Its Possible Impacts on the Aerosol Hygroscopicity and Size Distribution

SC

RI PT

On behalf of the authors whose names are listed immediately below I certify that we have NO affiliations with or involvement in any organization or entity with any financial interest (such as honoraria; educational grants; participation in speakers’ bureaus; membership, employment, consultancies, stock ownership, or other equity interest; and expert testimony or patent-licensing arrangements), or non-financial interest (such as personal or professional relationships, affiliations, knowledge or beliefs) in the subject matter or materials discussed in this manuscript.

Marçal S. Evangelista Marcelo S. Vieira-Filho

AC C

EP

TE D

Adalgiza Fornaro

M AN U

Antônio T. Bittencourt

Dr. Gerson Paiva Almeida