Online Density and Shape Measurement of Single Black Carbon Aerosol Particles in a Heavily Polluted Urban Area

ABSTRACT 
 
Black carbon (BC) aerosol imposes adverse effects on atmospheric visibility, climate, and health. The particle density and morphology are often needed to investigate the mixing state and aging process of BC particles. A method, combining an aerodynamic aerosol classifier (AAC), a differential mobility analyzer (DMA), a single-particle soot photometer (SP2) and a single particle aerosol mass spectrometer (SPAMS), was developed to determine the density and dynamic shape factor (χ) of ambient BC particles with three different aerodynamic diameters (Da, 200 nm, 350 nm, and 500 nm) in Shanghai, China, a typical urban area. The BC particles were either classified as “BC-dominated particle” which is mainly made of black carbon or “BC-mixed particle” which is a mixture of both BC and non-BC substances. The results showed that BC-dominated particles whose BC core mass (~2.2 fg) was almost equal to particle mass (~2.3 fg) were observed in particles with 200 nm Da. The morphology of these BC-dominated particles was near-spherical (χ ≈ 1.02), indicating that they had undergone rapid morphology modification from the initial highly irregular morphology to near-spherical shape. Most BC particles with 350 nm or 500 nm Da were BC-mixed particles. Combining the effective densities (1.62–1.77 g cm–3) and average single particle mass spectra of particle, the ammonium sulfate and ammonium nitrate were found to be the main secondary substances of these BC-mixed particles, indicating that condensation of inorganic species such as nitrates and sulfates could play a significant role in the aging process of fresh BC in Shanghai. Generally, the morphology and density information of single BC particle is crucial to identify the mixing state and aging process of BC aerosols.

as catalyzing SO2 oxidation, which could play a critical role in driving the formation, and aging of 48 regional haze (Zhang et al., 2020). In addition, it can affect climate directly by scattering and 49 absorbing solar radiation, and indirectly by influencing cloud formation through acting as cloud aerosol has been evaluated by epidemiological studies for their adverse health effects (Zhang et al., 52 2015). It is found that the dose-response relationship between mortality and BC aerosols was much 53 higher than that between mortality and particulate matter (PM) concentration (Janssen et al., 2011). 54 55 BC particles are mostly formed from incomplete combustion of carbon-containing fuels, such as 56 coal, diesel engines, biofuels, and biomass (Steinfeld, 1998; Moffet and Prather, 2009). Many in-57 situ measurements and laboratory studies show that freshly emitted BC particles exist as aggregates 58 of small carbon spherules (Bond et al., 2013;Soewono and Rogak, 2013). The mixing state of 59 freshly emitted BC particle begins to transform during fuel burning process and then keeps 60 changing during atmospheric aging (Jacobson and Seinfeld, 2004; Zhang et al., 2008; Yuan et al.,

A C C E P T E D M A N U S C R I P T
5 et al., 2007). A deep understanding of BC particle's morphology and density is important for better 80 evaluation of its mixing state, aging process and climate impact. 81 Multiple aerosol instruments can be used in tandem to simultaneously obtain the information on 82 particle shape and effective density. For the determination for effective density, many investigations 83 use a differential mobility analyzer (DMA) that can measure particle's electrical mobility diameter 84 ( ) and the single particle mass spectrometer (SPMS) instruments that can measure particle's 85 vacuum aerodynamic diameter ( ), such as single-particle aerosol mass spectrometer (SPAMS) 86 is the Cunninghan slip correction 89 factor, p is the true density of particle, is the volume equivalent diameter of particle. Only 90 if the particle morphology is spherical, namely the dynamic shape factor is equal to 1, the is 91 equal to according to Eq. (2), the particle effective density ( ) is equal to particle density 92

A C C E P T E D M A N U S C R I P T
7 tandem AAC-DMA configuration to measure the effective density of liquid dioctyl sebacate 115 droplets via Eq. (4), where the was determined by DMA, the particle mass ( ) was 116 determined by combing Eq. (6)-(8), is the viscosity of air, is the particle relaxation time, and 117 is the particle mobility. 118 Recently, Peng et al. (2021) firstly proposed AAC-SPAMS system to achieve the measurement of 122 effective density ( ) of aspherical particle. In short, based on the approximation that = , the 123 volume equivalent diameter ( ) of particle can be calculated via Eq. (1) and Eq. (9). Then the 124 effective density of particle can be obtained via Eq. (10). 125 Therefore, most existing tandem systems usually determine the effective density instead of particle 130 8   and Olfert (2014), measurements of  and  alone are not sufficient to determine the shape  133 factor, only if the particle density is known (or can be assumed), it is possible to determine 134 via Eq. (11) based on the measured particle mass ( ), then the dynamic shape factor in transition 135 regime ( ) can be measured via Eq. (2). 136

A C C E P T E D M A N U S C R I P T
In addition, for the differential mobility analyzer (DMA) and the single particle mass spectrometer 138 (SPMS) tandem system, like the work of Zelenyuk et al. (2006), they also measure the dynamic 139 shape factor of polystyrene latex (PSL) spheres agglomerates with known density (1.05 g cm -3 ) via 140 Eq. (3) based on the approximation that = . Zhang et al. (2016) combined a single-particle soot 141 photometer (SP2) and a volatility tandem differential mobility analyzer (VTDMA) to measure the 142 dynamic shape factor of the BC cores (density:1.8 g cm -3 ) within the BC-containing particle via 143 Eq. (2) and Eq. (11), where the BC core mass was measured by SP2. Thus, it can be known that for 144 the ambient particle whose density is unknown, using existing tandem online system to measure 145 their dynamic shape factor without assumption is challenging. 146

147
The goals of this study were to investigate the density, shape and the chemical composition of 148 ambient BC-containing particle in Shanghai (a typical urban environment in China) by adopting an

A C C E P T E D M A N U S C R I P T
9 analyzer (DMA), a single-particle soot photometer (SP2) and a single particle aerosol mass 151 spectrometer (SPAMS). It has been shown that BC particles play a significant role in the PM 152 pollution in Shanghai (Wei et al., 2020). The method adapted in this study can easily identify the 153 "BC-dominated particle" mainly made of BC from all the BC-containing particles by comparing 154 the particle mass ( ) and mass of BC core in particle ( ). This identification method for BC-155 dominated particle has not been reported before. Then the BC-dominated particle density instead 156 of effective density can be known, thus the measurement for their dynamic shape factor can be 157 achieved further. Meanwhile, for "BC-mixed particle" made of BC and non-BC substances, their 158 effective densities can also be investigated. The results of this study can provide helpful information 159 for estimating the mixing state and aging process of BC aerosols in Shanghai. In this study, the particles with known aerodynamic diameter ( ) were selected by aerodynamic 166 aerosol classifier (AAC, Cambustion Ltd., UK). AAC consists of two concentric cylinders rotating 167 at the same rotational speed and the same direction. The classifier speed controls the centrifugal 168 force applied to particles, the total classifier axial flow controls the residence time ( ) of particle. 169 Once the aerosols enter the sheath flow through a slit in the inner cylinder wall, the particles

A C C E P T E D M A N U S C R I P T
10 than the setpoint of AAC remain entrained in the sheath flow due to insufficient radial trajectory, 172 and particles with larger than the setpoint of AAC will impact and adhere to the outer surface 173 of the classifier due to excessive radial trajectory. Hence, only particles within a narrow range 174 follow the correct trajectory and pass through the AAC classifier with the sample flow. More 175 detailed description for AAC is in the previous study (Tavakoli and Olfert, 2013). Briefly, the SP2 uses a continuous Nd:YAG intra-cavity laser beam to heat a BC particle to its 181 vaporization temperature of ~4000K, at which point, detectable amounts of incandescent light are 182 emitted, the intensity of the incandescent light is linearly proportional to the BC mass (Moteki et  183 al., 2010). Therefore, using black carbon particle standards to calibrate the incandescence signal of 184 SP2 is required. In this study, Aquadag® black carbon particle standard (Acheson Inc., USA, Ultra-185 fine graphite), one of a commonly used BC reference material was used to calibrate SP2 ( shown in the Text S1 of Supporting Information. 188

11
Here, a single-particle aerosol mass spectrometer (SPAMS, Hexin Ltd. China) was adopted to 190 obtain the chemical information of single BC particle. More detailed information on the SPAMS 191 has been described elsewhere (Li et al., 2011). The particles in the size range of 0.2-2.0 μm are 192 drawn into the vacuum through an aerodynamic focusing lens, each particle is accelerated to a size-193 dependent aerodynamic velocity. Subsequently, particles with the specific velocity pass through 194 two lasers that are fixed at a 6 cm distance, the scatter light is collected by two photomultiplier 195 tubes (PMTs). Afterwards, each single particle arrives at the ion source region, a pulsed desorption-196 ionization laser (Qswitched Nd :YAG; 266 nm) is triggered. Both positive and negative mass 197 spectra for each single particle were recorded by a bipolar time-of-flight spectrometer. The single-198 particle mass spectra were converted to a list of peaks at each m/z by setting a minimum signal 199 threshold of 30 arbitrary units above the baseline. The resulting peak lists together with other 200 SPAMS data were imported into YAADA (version 2.11; www.yaada.org), a software toolkit for 201 single-particle data analysis written in MATLAB (version R2014b). According to the similarities 202 of the mass-to-charge ratio and peak intensity, the BC particles were identified from the ambient 203 particles by utilizing a clustering method called "adaptive resonance theory neural network" (ART-  (Tavakoli and Olfert, 2013), was maintained at 16.5, corresponding to a 1/10 aerosol to sheath flow 232 ratio during the whole study. The power of the desorption-ionization laser of SPAMS was set to 233 ~0.5 mJ per pulse. Before conducting any measurements, aerosols were dried (RH < 20%) by silica 234 diffusion dryers. 235

Estimation for particle mass and BC mass 242
For particle selected by AAC with a fixed set value ( ) and DMA with a fixed set value ( ), 243 the particle mass ( ) can be obtained by Eq. (6)-(8), like the previous study (Tavakoli and Olfert, 244 2014). Meanwhile, for a group of AAC-DMA selected BC particles, the mass of BC content ( ) 245 within each individual particle was further measured by the SP2. Thus, the BC mass distribution, 246 namely the number distribution of these AAC-DMA selected particles based on the BC mass, can 247 be obtained. If there existing a peak in the BC mass distribution, then the Gauss fitting was found 248 to be an appropriate function to do peak-fitting (R 2 ＞0.90). The peak center was exactly the mass 249 of BC content of particles under this peak. 250

Measurement for "BC-dominated particle" and "BC-mixed particle" 251
Comparing the BC mass ( ) measured by SP2 and the particle mass ( ) measured by AAC-252 DMA system, if in a single particle was nearly equal to , such a BC particle was referred 253 to "BC-dominated particle". If in a single particle was far less than , such a BC particle 254 was referred to "BC-mixed particle". The shape and density of these two types of BC particles were 255 investigated separately. For BC-dominated particle, their density should be very close to the 256 material density of black carbon as we assume that there are no internal voids within BC particle. 257 The value of 1.8 g cm -3 had been commonly used as the material density of BC particle The location of Peak 1 was fitted to be ~2.2 fg, which was close to the particle mass (~2.3 fg), thus, 286 it can be known that the particles in Peak 1 were almost made of BC and named as BC-dominated 287 particles. Here, was not strictly equal to , which could be partly due to the existence of 288 non-BC components or the accuracy of measured by AAC-DMA-SP2 (~5%). For particles 289 in Peak 2, the mass of BC content ( ) within BC particle was apparently less than the particle 290 mass, (~2.3 fg), thus these particles with a fair amount of non-BC components was BC-mixed 291 particle. Their effective density was derived as 1.75 g cm -3 via Eq. (4), suggesting that ammonium 292 nitrate (1.73 g cm -3 ) and/or ammonium sulfate (1.77 g cm -3 ) were probably the main coating 293 substances (Levy et al., 2013).

A C C E P T E D M A N U S C R I P T
18 In terms of the density of BC-dominated particle, as introduced in "METHODS", their density can 295 be estimated as 1.8 g cm -3 , along with the known particle mass, their dynamic shape factor ( ) was 296 calculated as ~1.02 via Eq. (2) and Eq. (11). The value of indicated that the morphology of BC-297 dominated particle almost made up by BC was near-spherical. It is acknowledged that the shape of demonstrated that in Beijing the fresh BC particles underwent rapid morphology modification from 303 a highly fractal structure to a near-spherical shape within 2.3 hours by quantifying the dynamic 304 shape factor of BC particles under particle-free ambient gases at a steady flow rate in a chamber. 305 Moffet and Prather (2009) also found that within 3 h after sunrise photochemical activity results in 306 the rapid conversion of fresh non-spherical soot to aged spherical coated soot particles. The 307 collapse in structure of BC particle may be related to the condensation of water and other gas-phase 308 In addition, traffic was found to be a possible contributor to these BC-dominated particles based 315 on Fig. 5, which shows that in many sampling days the diurnal variation of the number ratio of BC-316 dominated particles to BC-containing particles (the aggregation of BC-dominated particles and 317 BC-mixed particles) presents an ascending tendency during the morning (red shaded area) and 318 evening rush hours (grey shaded area). The calculation method for the number of three types of 319 particles (Non-BC Particle, BC-dominated particle, and BC-mixed particle) are shown in Text S3.

A C C E P T E D M A N U S C R I P T
20 for almost all of the particles observed in the size ranges of 100-300 nm in Shanghai, which was 322 consistent with our results that these BC-dominated particles in Mode ( =200 nm, =135 nm) 323

Aerosol with 350 nm 336
In addition to the study for ambient BC particle with 200 nm , the morphology and density of 337 particles selected by AAC at 350 nm were measured, too. Figure. 6 shows that the location of 338 the largest peak (~75%) in the average electrical mobility distribution of the particles selected by 339 AAC at 350 nm setpoint was ~260 nm, thus the particles in Mode ( =350 nm, =260 nm) were 340 studied by SP2 and SPAMS. The mass of particles in this mode was ~14.9 fg. Fig. S1A shows the 341 BC mass distribution of the particles in this mode, indicating that the BC mass was smaller than 342 the particle's mass, thus most BC particles in this mode were BC-mixed particles. The effective 343 density can be calculated as ~1.62 g cm -3 via Eq. (4). Meanwhile, the average mass spectra of these 344 BC-mixed particle were shown in Fig. 7

Aerosol with 500 nm
the electrical mobility size distribution of these particles in Fig. 8 was ~359 nm, thus, the 360 particles in Mode ( =500 nm, =359 nm) were mainly studied. The mass of particles in this 361 mode was ~42.7 fg. According to the BC mass distribution of particles in this mode in Fig.S1B, 362 most BC particles were found to be BC-mixed particles. The effective density was calculated to 363 be ~1.77 g cm -3 . Meanwhile, Figure. 9 showed that the average mass spectra of these BC-mixed 364 particles measured by SPAMS presented high intensity peaks of HSO4 -, NO3 -, NO2with weaker 365 negative carbon ion signals (Cn -) than the BC-mixed particles in Mode ( =350 nm, =260 366 nm), indicating the existence of ammonium sulfate (density:1.77 g cm -3 ) or ammonium nitrate 367 (density:1.73 g cm -3 ), and the effective density was within the density range of these coating 368 materials. In this study, the density and morphology of ambient BC particles in Shanghai, China were 380 investigated by measuring the BC particles in three typical modes: Mode ( =200 nm, =135 381 nm), Mode ( =350 nm, =260 nm) and Mode ( =500 nm, =359 nm). The mass of BC 382 particles ( ) can be computed directly from measured aerodynamic diameter ( ) and electrical 383 diameter ( ), which were readily determined by an AAC and a DMA, respectively. Then, the 384 mass of black carbon content ( ) of these AAC-DMA selected particles can be measured by 385

A C C E P T E D M A N U S C R I P T
25 SP2. By comparing the particle mass ( ) and the mass of BC content ( ) in particle, the BC 386 particles were classified into "BC-dominated particle" mainly made of BC and "BC-mixed particle" 387 made of both BC and non-BC substances. 388 The results showed that a fraction of BC-dominated particles whose dynamic shape factor ( ) 389 was ~1.02 were detected in Mode ( =200 nm, =135 nm), which means that their shape was 390 near-spherical, potentially indicating that morphology modification from the initial highly irregular 391 morphology to near-spherical shape was rapid. The increasing number ratio of BC-dominated 392 particles to BC particles during commuting hours suggests traffic emission are the primary source 393 of these BC-dominated particles. Meanwhile, for larger particles in Mode ( =350 nm, =260 394 nm) and Mode ( =500 nm, =359 nm), most BC particles were found to be BC-mixed particles. 395 The effective densities of particles in Mode ( =350 nm, =260 nm) and in Mode ( =500 396 nm, =359 nm) were respectively ~1.62 gcm -3 and ~1.77 gcm -3 , along with their average single 397 particle mass spectra information, it can be deduced that during the sampling periods condensation 398 of inorganic species such as nitrates and sulfates could play a significant role in the aging process 399 of fresh BC in Shanghai.