More Significant Impacts From New Particle Formation on Haze Formation During COVID‐19 Lockdown

Abstract During the COVID‐19 lockdown in 2020, large‐scale industrial and transportation emissions were reduced, but high PM2.5 concentration still occurred. This study investigated the variation of particle number size distribution during the lockdown, and analyzed the characteristics of new particle formation (NPF) events and its potential impact on haze formation. Through measurement conducted in urban Beijing during the first 3 months of 2020, and comparison with year‐over‐year data, the decrease of primary Aitken‐mode particles was observed. However, frequencies, formation rates and growth rates of NPF events remained stable between 2020 and 2019 in the same period. As a result, >25 nm particles produced by NPF events, would play a more important role in serving as the haze formation “seeds” compared to those produced by primary emissions. This finding emphasizes the significance on the understanding of NPF mechanisms when making pollution mitigation policy in the future.


Introduction
This file contains supporting information documenting: Text S1: Calculation of J3,    Table S1: The specific time periods of 2013-2020 adopted in this study Table S2: The specific periods in PRE, LNY and LOCK of 2019 and 2020 Table S3: Information of NPF events in 2019-LOCK and 2020-LOCK Table S4: Instruments used in this study and the parameters they measured.
The formation rate was calculated using the following formula (Cai et al., 2017): (1) Where is the formation rate of particles at size , and is chosen to be 3 nm in this study.
is the upper size bound of the target size range, and 25 nm is adopted for the calculation. [ , ) is the total number concentration of particles in the diameters of [ , ). is the lower bound of each measured size bin, and is lowest size limit detected by measuring instrument. ( , ) is the coagulation coefficient for the collision between the particle at size of and the particle at size of .
is the particle size distribution function which equals to dN/d , and is the growth rate at size of . The second term on the right side of the equation (RSE) represents the loss of particles in the size range of [ , ) due to coagulation scavenging with preexisting clusters or particles (coagulation sink). The third term on the RSE represents the increase of particles in the size range of [ , ) due to coagulation among smaller clusters or particles (coagulation source). The last term on the RSE is the flux of particles growing up to over size .
The growth rate (GR) was obtained by the mode-fitting method (Dal Maso et al., 2005). The particle number size distribution (PNSD) during NPF event days were fitted as the sum of three-mode lognormal distribution. GR was calculated as the variation of the geometric mean diameter Dm of newly formed mode (3-25 nm) in unit internal (Dal Maso et al., 2005): To evaluate the scavenging effects of preexisting particles on condensable vapors, the condensation sink (CS) was calculated as follow (Dal Maso et al., 2005): where D is the diffusion coefficient of the condensing vapor, β is the transition regime correction factor, and , and are the diameter and number concentration in the size class i, respectively.

Text S2.
The H2SO4 concentrations on NPF days are calculated with a pseudo steady state method, assuming the OH and SO2 reaction as the only source and condensation on preexiting particles as the only sink, with the formula as (Zheng et al., 2011): where and indicate the production and the loss rate of sulfuric acid on the surface of atmospheric aerosols.
is estimated from: where k1 is the reaction coefficient between OH radicals and SO2, with a value of 1×10-12 cm 3 s -1 .
[SO2] and [OH] are the levels of reactants in molecules cm-3. is calculated as follow (Freiberg & Schwartz, 1981): where γ is the uptake coefficient of H2SO4 on particle surface, taking a value as 0.73 (Jefferson et al., 1997). S indicates the concentration of particle surface area. is the root mean square velocity of H2SO4 molecules. The concentration of OH radicals are estimated by the following formula (Ehhalt & Rohrer, 2000): where α=0.83, β=0.19, a = 4.1×10 9 , b=140, c = 0.41 and d = 1.7. NO2 levels are in unit of ppb.

Text S3.
The normalized index for certain atmospheric pollutant (e.g. PN25-100, NO2, etc.) is calculated as follows: a. Calculate the daily average of the pollutant concentration around lunar New Year holiday from 2013 to 2020 (Table S1).
In the main text we chose PN25-100, because 25-100 nm particles are mainly derived from primary emission (traffic emission), and have short atmospheric lifetime, as discussed in the manuscript.
b. Get the normalized index of PN25-100 through scaling the daily average of PN25-100 in each year with the 95th percentile as follow: where is the year, ranged from 2013 to 2020, is the day from lunar New Year's day, ranged from -24 to +45 (-24, +45). , indicates the index of PN25-100 in the day of y year. , is the daily average of PN25-100 in the day of y year. ,95th represents the 95th percentile of , in year.
The purpose of this step is to remove the weight of absolute concentration of each year. For example, the primary aerosol emission may be weaker in 2019 compared to that in 2013 in Beijing, due to the Air Pollution Prevention and Control Action Plan (2013-2017). The purpose of using the 95th percentile other than max value is to exclude the influence of special days such as strong atmospheric nucleation and growth days.
The purpose of this step is to reduce the meteorological and weekday-weekend effects, and reflect a common variation of daily average of PN25-100 around lunar New Year holiday.   Text S4.
PM2.5 was measured by a tapered element oscillating microbalance (TEOM, 1400a, Thermo, USA) with a PM2.5 cyclone inlet (Zamora et al., 2019). Tracer gaseous pollutants were continuously detected by a series online monitoring system manufactured by Thermo Electron Corporation (O3 (Model 49i), SO2 (43i-TLE), CO (48i-TLE) and NO-NO2-NOx (model 42i-TLE)). Meteorological parameters including wind speed (WS), wind direction (WD), temperature (T) and relative humidity (RH) were measured by the automatic meteorological station (Met one Instrument Inc). The photolysis frequencies of O3 (JO 1 D) and NO2 (JNO2) were monitored by a spectroradiometer, following the procedure described by (Wang et al., 2019). 99 types of volatile organic compounds (VOCs) were measured by the online gas chromatography and mass spectrometry (Fang et al., 2020). Black carbon (BC) was measured by Aethalometer (Magee Scientific, model AE31), and the concentration of BC at 880 nm was used in this study to reduce the influence of Brown carbon (Kirchstetter et al., 2004).

Text S5.
During the clean days, the strong nucleation occurred in the morning of 4 and 5 February, and the total particle number concentration increased to around 7×10 4 cm -3 (Fig. S2f), which is mainly contributed by the nucleation-mode particles. Then, mean diameter of the nucleated particles was observed to increase continuously in the following 4 days, from ~8 nm to ~90 nm, when PM2.5 increased from 30 μg m -3 to 90 μg m -3 (Fig. S5). The NPF events on 8 February also injected high number concentration of nucleation-mode particles (~4×10 4 cm -3 ) into the atmosphere of Beijing (Fig. S2f). From 8 February to 11 February, the mean diameter continued growing to over 120 nm, and PM2.5 increased to 280 μg m -3 by efficient secondary aerosol formation. The mean diameter and mass concentration fluctuated during the transition of haze periods, which may be influenced by local primary emission, wind speed and planet boundary level. Figure S1. Diurnal cycle of NO2 and PN25-100 from 9 February to 10 March 2020. Figure S2. Timeline with some important time nodes from 1 January to 10 March 2020 and timeseries of meteorology parameters, air pollutants and particle number parameters at PKUERS, including (a) WS, WD, (b) RH, temperature, (c) PM10, PM2.5, BC, (d) SO2, CO, (e) O3, NO2, (f) mean diameter and total number concentration of particles (ranged from 3 nm to 698 nm), (g) PNSD. The gray background represented 3 heavy pollution episodes which was defined as a process when the mass concentration of PM2.5 increased from a level below 50 μg m -3 to a peak value exceeding 200 μg m -3 and then dropped back to a level below 50 μg m -3 .     Table S4. Instruments used in this study and the parameters they measured.