key: cord-0715917-uyrn281a
authors: Zimmerman, Alyssa; Petters, Markus D.; Meskhidze, Nicholas
title: Observations of new particle formation, modal growth rates, and direct emissions of sub-10 nm particles in an urban environment
date: 2020-08-07
journal: Atmos Environ (1994)
DOI: 10.1016/j.atmosenv.2020.117835
sha: 50cb4deed3617027af79080cbf5182f9935f9c32
doc_id: 715917
cord_uid: uyrn281a

Ultrafine particles with diameters less than 100 nm suspended in the air are a topic of interest in air quality and climate sciences. Sub-10 nm particles are of additional interest due to their health effects and contribution to particle growth processes. Ambient measurements were carried out at North Carolina State University in Raleigh, NC between April to June 2019 and November 2019 to May 2020 to investigate the temporal variability of size distribution and number concentration of ultrafine particles. A mobile lab was deployed between March and May 2020 to characterize the spatial distribution of sub-10 nm particle number concentration. New particle formation and growth events were observed regularly. Also observed were direct emissions of sub-10 nm particles. Analysis against meteorological variables, gas-phase species, and particle concentrations show that the sub-10nm particles dominated number concentration during periods of low planetary boundary layer height, low solar radiation, and northeast winds. The spatial patterns observed during mobile deployments suggest that multiple temporally stable and spatially confined point sources of sub-10 nm particles are present within the city. These sources likely include the campus utility plants and the Raleigh-Durham International Airport. Additionally, the timing of data collection allowed for investigation of variations in the urban aerosol number size distribution due to reduced economic activity during the COVID-19 pandemic.

To investigate the spatial and temporal variability of aerosols in Raleigh, three different 159 instrumentation configurations, detailed in Table 1 , were used during measurements taken from 160

April 19 to June 4, 2019 and again from November 15, 2019 to May 15, 2020. Instrumentation 161 spatial distribution of sub-10 nm particles throughout Raleigh. This system consisted of 170 instrumentation for aerosol number concentration measurements and continually sampled from 171 the Jordan Hall laboratory, until it was transported to a vehicle for mobile deployments. 172 processing was used for CPC-3. The sample flow rate was continuously measured and 193 coincidence was corrected through live-time counting performed by the manufacturer's 194 algorithm. This method divides the number of counted particles by the time between electrical 195 pulses and the flow rate to provide an accurately determined true counting rate. Due to 196 instrumentation failure, CPC-4 was replaced with CPC-5 (TSI 3772) in May 2020. The 197 difference in the counts between CPC-3 and CPC-4/5 provided a particle number concentration 198 for diameters between 2.5-11 nm (N 2.5-11nm , 10 nm upper limit when CPC-5 was used). was used during mobile deployments 5 and 6. 200

All raw 10 Hz data from the SMPS system and corresponding CPC were inverted using 202 the method described in Petters (2018) . The inversion method includes mapping the time varying 203 electric field to a corresponding mobility diameter (S. C. Wang & Flagan, 1990) . Particle counts 204 are binned and inverted to the total concentration by accounting for the SMPS transfer function, 205 multiple charges, and diffusional broadening of the transfer function. The resulting concentration 206 and particle diameter data from the SMPS were binned into 83 logarithmically spaced bins 207 spanning a size range of 5-55 nm. The binned SMPS data was then averaged onto fixed 5 min. 208 and 30 min time grids to facilitate time-series analysis. 209 Figure 1 presents the temporal variability in the 5-55 nm spectral number density during 210 had an initial mode between 10 to 20 nm and the particles also grew to larger sizes. Finally, if 227 there were nucleation mode particles present, resulting in an increase in the ultrafine particle 228 number concentration, but growth was not continuous or sustained, this event was classified as 229 Class C. More specifically Class C events were characterized by modal growth lasting < 4 hr., or 230 significant modal shrinkage of particles. Examples of Class A, B, and C NPF events are shown 231 in Figures S5, S6 , and S7. The dates and times of all Class A and B NPF events are detailed in 232 Tables S1 and S2. A slide deck analyzing the normalized spectral number density of all observed 233 nucleation events is included in the archived dataset. 234

Visual analysis of the size distribution data also revealed the presence of PB events. 235 These events were characterized by continuous bursts of nucleation mode particles that persisted event, if multiple bursts of nucleation mode particles continued for more than 6 hr and no modal 239 growth was observed. The 30-minute average of 5-10 nm particles was visually analyzed for 240 each PB event to estimate a threshold concentration that was exceeded by each burst in an event. 241

Due to variations in strength of events and background number concentration, each event was 242 assigned a separate threshold concentration (shown in Table S3 ). The threshold concentrations 243 were determined by the maximum 30-minute average number concentration of the weakest 244 particle burst during each event. The threshold values were then used to estimate the PB event 245 duration. The start and end times of each event were estimated as when the 30-minute average 5-246 10 nm number concertation surpassed and receded the threshold concertation. 247

The growth rates were calculated for Class A and B NPF events using the maximum 249 concentration method of Kulmala et al. (2012) . The detection time of each event was initially 250 estimated from visual inspection of normalized number spectral density plots. For each scan 251 during the event, the modal diameter was calculated from the maximum number concentration 252 and then plotted as a function of time. The growth rates of each event were derived using a linear 253 regression of the modal diameters across three size ranges of 5-10 nm, 10-20 nm, and 20-30 254 nm. The first size range was selected based on the lower limit of the SMPS instrument. 255

Considering that the initial mode diameter of Class B events was approximately 10 nm, the 256 second size range captures the initial growth of nucleated particles during these events. The 257 second and the third size ranges allow for comparison of growth rates between the Class A and B 258 events. Due to the 5 nm lower limit in the size distribution data, the growth rates calculated 259 between 5-10 nm were used to estimate the approximate start time of each Class A event.

average growth rates between 1.5 and 3 nm are nearly half of the growth rates measured above 3 262 nm. Considering this variability in growth rates below 10 nm, the Class A event start times are 263 reported within a possible range based on one standard deviation of the 5-10 nm growth rates. 264

Class B start times are reported as a single time, rather than a range, because the initial mode 265 detected was greater than the lower limit of the SMPS and the initial appearance of these 266 particles can be accurately estimated. 267

To investigate potential sub-10 nm particle sources, the mobile system was deployed 6 269 times between March and May 2020. During mobile sampling, the inlet for the particle counters 270 was placed out the rear passenger side window. A 1000-Watt lithium battery power station was 271 used to operate the CPCs, which provided approximately 3.5 hr of sample time. Time series of 272 the driving route coordinates were recorded using a mobile phone application. The particle 273 counts were then averaged on a time grid, ranging between 6 to 7.5 sec depending on the mobile 274 deployment, to match the coordinate dataset. Details of the individual deployments are shown in 275 Table 2 . Two different routes were driven during the various deployments. The first, shown in 276 Figures S8 and S9 was designed to maximize distance traveled around Raleigh and highlight any 277 concentration gradient present between downtown and the suburbs. This route was used in 278 mobile deployments 1-3. The second route (Figs. S10 and S11) was designed locally around the 279 NCSU campus and was used in mobile deployments 4-6. 280

One-minute data for temperature, pressure, wind speed and direction, precipitation, and 282 solar radiation were acquired from the North Carolina Climate Office weather station located on Mesoscale Forecast System (NAM) at a 12 km spatial resolution. Hourly data for CO, NO 2 , O 3 , 285 PM 2.5 , and PM 10 were acquired from the Environmental Protection Agency's (EPA) AQS API 286 system for the Millbrook School site (AQS ID: 37-183-0014). The Millbrook School site is in the 287 Northeast suburbs of downtown Raleigh and is 12.4 km NE of the sampling site. Hourly gaseous 288 pollutant data for SO 2 was unavailable at any nearby monitors, and therefore was not analyzed. instruments. The average integrated SMPS number concentration between 5-10 nm (N 5-10nm ), 296 and 10-40 nm (N 10-40nm ) were 1.6×10 3 cm -3 and 4.3×10 3 cm -3 . These values show that on average, 297 N 5-10nm accounts for ~27% of less than 40 nm sized particle number concentration. For periods 298

where the SMPS and dual CPC system was available, there are concurrent time series number 299 concentrations. For example, N > 11 nm and N 10-40nm correlate. The daily average number 300 concentrations > 2.5, and >11/10 nm from CPCs 3 and 4/5 were 5.3×10 3 cm -3 and 1.1×10 4 cm -3 , 301 respectively. The average N 2.5-11nm computed from the difference in the two CPCs was 5. Carolina was under a state of emergency and all K-12 public schools, restaurants, and bars were 316 mandated to close. After these executive orders, vehicular traffic on Western Blvd. decreased to 317 approximately 1.4×10 4 vehicles per day (i.e., by ~ 67% of the average daily traffic) (Fig. S12) . 318

On March 23, gatherings of 50 or more people were banned, and service sector businesses were 319 mandated to close. On March 30 the statewide stay at home order began. During the stay at home 320 order, local traffic counts were reduced by ~75% of the average daily traffic. Human movement 321 to retail and recreation, grocery and pharmacy, transit stations, and the workplace were down 322 approximately 50%, 28%, 77%, and 45%, respectively according to the Wake County 323 community mobility report. As the stay at home order persisted into late April and early May, 324 movement to parks increased from 5 to 47%. As shown in Figure 3 Wang et al., 2020). Here, the PM 2.5 concentration also decreased during this period (Fig. S14) , 348 but the concentration generally remained within the statistical fluctuation of the last 10 years. 349

New particle formation events occurred during nearly all months of measurement. Figure  351 The start time of NPF events varies greatly based on the class of event. The average start 361 and detection times presented in Table 3 show that Class A NPF events were generally detected 362 during the morning hours, while Class B events were detected during or after midday. The 363 estimated average start time of all Class A events was 9:30 ± 01:00 and the average detection 364 time was 11:00 ± 1:15. The average minimum and maximum start times reported in Table 3  365 represent the starting time range estimated based on one standard deviation of the 5-10 nm GRs. 366

All Class A event times are detailed in Table S1 . The general start time of all Class A events was 367 also pronounced in the diurnal variation of particle number concentration plots shown in Figure  368 5a. The average N 5-10nm and N 10-40nm increases first between 06:00 and 09:00 likely due to rush-369 hour traffic, and again between 09:00 and 12:00 correlating with the average detection time of 370 the nucleated particles. The onset of a Class A NPF event increases the N 5-10nm by almost a factor 371 of two compared to the concentration on a non-NPF day (see Fig. 5b ). The morning traffic 372 signature was also present in the N 10-40nm diurnal variation. Figure 5a shows that the 373 concentration increase due to NPF was delayed on average by ~60 min. from the onset of the Table S2 and Fig. S15) . 381

The particle GRs were broken down into three different size ranges (5-10 nm, 10-20 nm, 382 and 20-30 nm) to analyze how particle growth varies with modal size. Little variation was 383 observed between size ranges in the Class A GRs (5-10 nm: 2.3 ± 1.2 nm hr -1 ; 10-20 nm: 3.1 ± 384 2.2 nm hr -1 ; 20-30 nm: 2.6 ± 1.2 nm hr -1 ). Particle growth rates during Class B events were 385 slightly greater than Class A events. The average GRs were 3.7 ± 2.9 and 3.3 ± 2.3 nm hr -1 for 386 10-20 nm and 20-30 nm, respectively. The Class B GR values measured during this study were 387 comparable to rates previously measured on the NCSU campus (10-25 nm GR: 1.6-3.9 nm hr -1 ; to non-event days. O 3 concentrations were also elevated during Class B event days; however, 405 PM 2.5 levels were comparable to non-event days. 406

Particle burst events play a significant role in shaping the near-surface particle 408 concentration and size distribution in Raleigh. These bursts of sub-10 nm particles were observed 409 during all hours of the day and their duration lasted anywhere between 7 to 93 hr. Figure 4 The PB events observed were characterized by a factor of two increase in N 5-10nm due to 416 continuous bursts of sub-10 nm particles (example event shown in Fig. S18 ). However, when 417 compared to a regular day, there was no change in the N 10-40nm (Figure 5c ). The average N 5-10nm 418 during all events ranged between 7.7×10 3 -6.9×10 4 cm -3 . No modal growth of the particles was 419 measured during PB events, suggesting that the particles were produced by an isolated source. 420

The average mode diameter of all PBs was between 5-7.5 nm suggesting that particles had a the mode diameter would drop below 5 nm causing the N 5-10nm to peak over 5×10 4 cm - On three measurement days (Jan. 17 and 21; and Feb. 21, 2020) a Class B NPF event and 431 a PB event were detected simultaneously (Fig. S19) . As newly nucleated particles grew 432 characteristically of an NPF event, the pre-existing PB sub-10 nm particles persisted throughout 433 the growth event maintaining a constant mode diameter. The decoupled nature of these two 434 events also suggests that the PB particles likely originated from a local source and had similar 435 sizes when reached the detection instrument. Figure S19 also shows that there was little 436 coagulation between the particles produced by two distinct sources, i.e., ones produced by a 437 mesoscale NPF event and by a local source. The large spatial extent of the NPF event allowed 438 for particles to grow as they advected toward the measurement instrument. Providing that the 439 direct measurement of the local pollutant persisted, detection of the PB event particles continued 440 throughout the growth event. The PB event particles likely contributed to the growth event, albeit 441 without measurable modal growth. 442 Wind analysis on PB days shown in Figure 6 reveals that the wind was blowing from the 443 NNE and NE 40% of the of time. Whereas, the winds out of the SSW were most frequent on 444 exceeded 2×10 4 cm -3 . Subtle peaks in S and E winds also correlated with average N 5-10nm 446 concentrations greater than 1×10 4 cm -3 . Some PB events occurred during overcast weather and 447 periods of rainfall, which had no impact on removal of the sub-10 nm particles (Table S3) . 448

Additionally, both the PBLH and solar radiation remained lower on PB days compared to regular 449 days (Fig. S16 ). Low PBLH, solar radiation, and the strong correlation with NE winds on PB 450 days suggests that aerosol dispersion processes were weak, allowing for direct measurements of 451 a pollution plume created by a local source. Studies that also observed PB events suggest weak *The average number concentration reported for mobile deployments 5 and 6 is between 2.5-10 823 nm as CPC-5 was used during this deployment. 824 

Observation of new particle formation 587 in subtropical urban environment

Effects of Grades and Other Loads on

On-Road Emissions of Hydrocarbons and Carbon Monoxide

A New Coincidence Correction Method for

Atmospheric new particle formation at the research station Melpitz, Germany: connection 609 with gaseous precursors and meteorological parameters

Observations of new particle production in the atmosphere of a moderately polluted site 613 in eastern England

Emissions from an 616 International Airport Increase Particle Number Concentrations 4-fold at 10 km

Characterization of

Modeling and measurements of urban aerosol processes on the neighborhood scale in

Atmospheric new particle formation and growth: review of field observations

Dispersion 642 of particle numbers and elemental carbon from road traffic, a harbour and an airstrip in 643 the Netherlands

Characteristics of the atmospheric particle formation events observed at a borel 655 forest site in southern Finland

Exposure 657 to sub-10 nm particles emitted from a biodiesel-fueled diesel engine: In vitro toxicity and 658 inflammatory potential

Impact 661 of nucleation on global CCN

Possible Wintertime Sources of Fine Particles in an

Emission of Ultrafine Particles from Natural Gas Domestic Burners

Anthropogenic and Natural Radiative Forcing. Cambridge, United Kingdom and New

Comparison of atmospheric new particle formation events in three Central European 682 cities

Global analysis of continental boundary layer new particle formation based on long-term 686 measurements

Effects of 689 continental boundary layer evolution, convection, turbulence and entrainment, on aerosol 690 formation

variations of ultrafine particle concentration in urban Gwangju, Korea: Observation of 699 ultrafine particle events

Respiratory effects are 702 associated with the number of ultrafine particles

A language to simplify computation of differential mobility analyzer 706 response functions

Observation and Analysis of Particle 709

Nucleation at a Forest Site in Southeastern US

An Observational Case Study on the Influence of Atmospheric Boundary-Layer

Dynamics on New Particle Formation. Boundary-Layer Meteorology

Spatial and vertical extent of nucleation events in the Midwestern USA: insights from the

A study of ambient fine 727 particles at Tianjin International Airport, China. Science of The Total Environment

Size distribution and hygroscopic properties of aerosol particles from dry-season biomass 731 burning in Amazonia

Frequent nucleation events at the high altitude station of Chacaltaya (5240 m a

Effect of restricted 745 emissions during COVID-19 on air quality in India

Sources and concentration of 748 nanoparticles (<10nm diameter) in the urban atmosphere

Atmospheric sulphuric acid and aerosol formation: implications from atmospheric 752 measurements for nucleation and early growth mechanisms

Air Quality Study: Description and Relation to Key Meteorological, Gas Phase

Aerosol Parameters Special Issue of Aerosol Science and Technology on Findings from 757 the Fine Particulate Matter Supersites Program

Particle Number and Size Distribution from a Diesel and new particle formation events at an urban site in Hong Kong

Severe air pollution events not 769 avoided by reduced anthropogenic activities during COVID-19 outbreak. Resources, 770 Conservation and Recycling

Scanning Electrical Mobility Spectrometer

Characteristics 775 of regional new particle formation in urban and regional background environments in the

Horizontal homogeneity and vertical extent of new particle formation events. Tellus B: 780 Chemical and Physical Meteorology

The Los Angeles International Airport 783 as a source of ultrafine particles and other pollutants to nearby communities

Growth of nucleation mode particles in the summertime Arctic: a case study

Strong 794 atmospheric new particle formation in winter in urban Shanghai

Growth rates of nucleation mode particles in Hyytiälä during 2003&minus;2009: 798 variation with particle size, season, data analysis method and ambient conditions

Characteristics of aerosol 802 size distributions and new particle formation in the summer in Beijing

A hypothesis for growth of fresh atmospheric nuclei

Aircraft emissions and local 816 air quality impacts from takeoff activities at a large International Airport

Daily average integrated SMPS concentration for 5-10 nm (red) and 10-40 nm (blue) 832 size range

Vertical ticks are spaced one week apart. Horizontal lines show monthly averages for 5-10 and 834 10-40 nm particles

Diurnal profiles of median planetary boundary layer height (PBLH, left) and number 837 concentration of 10-40 nm particles (right). Colors correspond to time period before and after

Shaded areas represent the 25 th and 75 th 839 quartiles of the data. Pre-COVID-19 encompasses days from spring 2019 and winter/spring 840 2019/2020 prior to March 14, 2020. COVID-19 days include days between March 14