Chemical composition and mixing state of BC-containing particles and the implications on light absorption enhancement

1State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100029, China 2College of Earth and Planetary Sciences, University of Chinese Academy of Sciences, Beijing 100049, China 10 3Center for Excellence in Regional Atmospheric Environment, Institute of Urban Environment, Chinese Academy of Sciences, Xiamen 361021, China 4Institute of Mass Spectrometer and Atmospheric Environment, Jinan University, Guangzhou, 510632, China 5Guangdong Provincial Engineering Research Center for On-Line Source Apportionment System of Air Pollution, Guangzhou, 510632, China 15 6Guangzhou Hexin Analytical Instrument Company Limited, Guangzhou, 510530, China 7Institute for Environmental and Climate Research, Jinan University, Guangzhou 511443, China 8State Key Laboratory of Severe Weather & Key Laboratory for Atmospheric Chemistry, Institute of Atmospheric Composition, Chinese Academy of Meteorological Sciences, Beijing, 100081, China 9Institute of Surface-Earth System Science, Tianjin University, Tianjin 300072, China 20 anow at: State Key Joint Laboratory of Environmental Simulation and Pollution Control, College of Environmental Sciences and Engineering, Peking University, Beijing, 100871, China

SPAMS uses a new aerosol concentration sampling device and improves the transmission efficiency of coarse particles.
Meanwhile, a delayed extraction technology is introduced to improve the mass resolution and increase the hit rate by a factor 95 of 2 ~ 4 for ambient particles (Chen et al., 2020c). Also, the ion signals with high and low intensity are separated by multichannel acquisition technology and detected simultaneously, which makes the system dynamic range more than 40 times of the traditional data acquisition system and improves the detection of ions with low signals greatly (Shen et al., 2018;Zhong et al., 2021).

Data analysis 100
In total, 3 619 038 and 4 655 426 particles were analyzed in Beijing and Gucheng, respectively. The size and chemical composition of each single particle is informed by Computational Continuation Core (COCO) toolkit in MATLAB software. are classified as CC pure type from coal combustion (Zhang et al., 2009;Healy et al., 2010); (3) (Yang et al., 2017); (4) particles internally mixed with more than one source (three sources above) are unified 110 together and named as Mix Source type. The BC-containing particles above are collectively referred to BC fresh . The remain BCcontaining particles are named BC aged and classified by using the ART-2a algorithm with a vigilance factor of 0.75, a learning rate of 0.05 and 20 iterations (Song et al., 1999). Seven particle types are grouped and named based on two principles: (1)  ). Otherwise, they are named BCOC NS when they present comparable peak areas of nitrate and sulfate. The more detailed names of BC-containing particle types are shown in Table S1.
The sources of bulk OA were analyzed by positive matrix factorization (PMF) and five OA factors were identified at both urban and rural sites, including biomass burning OA (BBOA), fossil fuel-related OA (FFOA), cooking OA (COA), less 120 oxidized oxygenated OA (LO-OOA) and more oxidized OOA (MO-OOA) in Beijing and BBOA, coal combustion OA (CCOA), hydrocarbon-like OA (HOA), OOA and aqueous-related OOA (aq-OOA) in GC (Table S1). The detailed PMF analysis of OA in Beijing and Gucheng are presented Xu et al. (2021) and Chen et al. (2021). The PMF analysis was also performed to identify the effect of different mixing state on E abs , by inputting b abs, total , b abs, BCpure and 11 major types of BCcontaining particles derived from HR-SPAMS. The detailed pre-treatment of error matrix, and the selection of factor solutions 125 could be found in previous studies (Petit et al., 2014;Xie et al., 2019b). Then, the E abs of each factor can be calculated as the ratio of b abs, total fi and b abs, BCpure fi in factor i.

BC-containing particles at urban and rural sites
BC-containing particles on average accounted for 62% of the total particles in Beijing, lower than that in Gucheng (73%) and 130 higher than that in winter 2018 in Beijing (55%) (Xie et al., 2020). Similarly, a previous winter study in Beijing also found that 60 -78% of aerosol particles contained BC (Chen et al., 2020a). According to Figs. 1 and S1, we found that the mass spectra of BC-containing particles at the two sites are somewhat similar which are both characterized by C n ± (n = 1-7), explanation is the more nitrogen-containing compounds formed from either aqueous-phase processing due to higher RH (69% vs. 45%) or biomass burning emissions at the rural site (Zhang et al., 2012;Chen et al., 2019). 140 The study in Beijing was divided into two periods, i.e., non-heating period (BJ-NHP) from 28 October to 10 November and heating period (BJ-HP) from 10 November to 1 December, while the observation in Gucheng was performed during heating period. As illustrated in Fig. 2a, the BC-containing particles are dominantly contributed by BC N and BCOC N (~20%) during BJ-NHP, indicating that BC was mainly internally mixed with nitrate at the urban site. Compared with BJ-NHP, the fractions of BCOC S and Mix Source increased significantly during BJ-HP, especially during relatively clean periods, suggesting a 145 considerable change in BC mixing state from non-heating to heating period due to the enhanced primary emissions, e.g., coal combustion. Previous studies showed that BC was mainly mixed with sulfate in winter in Beijing (Chen et al., 2020b;Xie et al., 2020), while this study showed a dominant mixing of BC with nitrate. This was likely due to the fact that coal fuels in Beijing were replaced by clean energies, e.g., natural gas and electricity after 2017 (Zhang et al., 2019). Indeed, the changes in nitrate concentrations were relatively small in winter in Beijing since clean air action although the sulfate concentrations 150 showed large decreases (Zhou et al., 2019;Lei et al., 2020). Comparatively, BCOC N was the major BC-containing particle type accounting for 36% in GC, which was twice than that in BJ-HP indicating that BC particles were dominantly mixed with OC and nitrate in an environment with high RH and intensive primary emissions, e.g., coal combustion emissions. In addition, BB pure and TR pure showed pronounced diurnal cycles in GC ( Fig. S2) compared with the relatively flat diurnal variations of BB pure in BJ, suggesting intensive biomass burning and diesel vehicle emissions at the rural site especially nighttime. 155 Almost all BC aged types showed strong dependence on RH while the number fractions of BC fresh types decreased with increasing RH at both sites indicating that high RH environment was more favorable for BC aging (Zhang et al., 2021). Similar to BC fresh , BCOC S was the only type of aged BC showing decreased fraction as a function of RH in BJ and GC. In fact, the high 160 correlations between BCOC S and CC pure (R 2 = 0.89 and 0.98, in BJ and GC, respectively) highlight that BC emitted from coal combustion could be directly mixed with OC and SO 4 at low RH level, and evolve towards the mixing with OC and NO 3 under high RH conditions. Moreover, the number fraction of BC N increased gradually as the increase of RH and dominated BC particles (~30%) at RH = 90 -100% in BJ, suggesting that the newly formed nitrate that coated on fresh BC played an important role in the formation of severe pollution in urban region. Comparatively, the number fraction of BCOC N increased the most by 165 43%, accounting for more than half of BC-containing particles at high RH levels in GC. This result indicates that the type of BCOC N was more important to aggravate air pollution in rural area. In addition, we found that the fraction of BCOC NS decreased obviously as a function of RH in GC indicating the impact of the transition from photochemical production to aqueous-phase reactions on the mixing state of BC. This result also suggests that aqueous-phase formation of sulfate at high RH level appeared not to affect BC mixing state substantially, consistent with previous studies that aqueous-phase processing 170 did not affect the total E abs in GC (Sun et al., 2021;Zhang et al., 2021). However, BCOC NS in BJ showed relatively stable fractions across different RH levels suggesting the different sources from the rural site. Figure 3 shows the evolution of mixing state of BC-containing particles during two different haze events in Beijing. During the initial stage of haze episode (P0), the contribution of BCOC S started to decrease while the number fraction of BCOC N increased significantly. As a consequence, E abs increased rapidly from ~1.1 to 1.3 in half day. Then, the number fraction of 175 BC N increased while that of BC fresh decreased during P1 period. These results indicated that fresh BC was gradually aged by mixing with nitrate during the evolution of haze episode. BC N increased continually during P2 with high E abs up to 1.4, and finally the mixing state of BC was stabilized as indicated by the relatively stable number fractions of most BC particle types and small changes in E abs. Similar to haze episode 1, the freshly emitted BC was gradually mixed with nitrate and OC causing high E abs (up to 1.5) during P3 and P4. As shown in Fig. 3d the evolution of BC mixing state during most haze events were similar in GC (Figs. 2c and S3), which was characterized by most significant increase in BCOC N rather BC N as the PM increased. Moreover, BCOC N particles were increased more significantly during nighttime while BCOC NS was more significant during daytime. Such differences were mainly due to the 190 enhanced coal combustion pollutants at nighttime (Fig. S2) which were mixed with photochemical products during daytime.

Chemical composition and mixing state in different environment
Overall, our results suggest that the fresh BC particles from biomass burning emissions are more directly mixed with nitrate under high RH conditions, and then mixed with more sulfate during further aging. Comparatively, the fresh BC particles from coal combustion are often mixed with OC and sulfate first, and then mix further with OC and nitrate at high RH level. In addition, our results also demonstrate that the E abs in GC was largely due to coal combustion emissions internally mixed with 195 OC, consistent with our previous study showing a large impact of coal combustion emissions on E abs (Sun et al., 2021).

Effects of chemical composition on Eabs
As shown in the average positive mass spectra of total BC-containing particles (Fig. S2), the peak areas of C n + , OC and metal contributed more than 95% to the total peak area, while the peak areas of NO 3 (46[NO 2 ]and 62[NO 3 ] -) and SO 4 (97[HSO 4 ] -) accounted for more than 80% in the negative mass spectra. To better characterize the relationship between chemical species 200 and E abs , we summed C n ± (n= 1~5, accounting for more than 99% in C n ) peak areas to represent BC and the total of NO 3 and SO 4 peak areas to represent secondary inorganic components coated on BC. In addition, the sum of positive peak areas except C n + was defined as OC + Metal to represent the OC and metal components coated on BC. These peak areas covered almost all of the chemical components coated on BC in total BC-containing particles.

Figures 4a and 4b show the relationship between peak area ratios (measured by SPAMS) and mass concentration ratios 205
(measured by HR-AMS) in Beijing and Gucheng, respectively. As the increase of (NO 3 + SO 4 ) AMS /eBC mass concentration ratio, the (NO 3 + SO 4 )/C n peak area ratio increased first and then gradually became stable at both sites. These results indicated that BC was rapidly aged and internally mixed with secondary inorganic components during the early stage of haze episode, and appeared to be fully aged when the ratio of NO 3 and SO 4 to eBC exceeded ~6. Different from secondary inorganic species, the peak area ratio of (OC + Metal)/C n showed a high dependence on the mass concentration ratio of POA (e.g., the sum of ratio of SOA/eBC also presented a high correlation with (OC + Metal)/C n in BJ (R 2 = 0.81) and GC (R 2 = 0.95). We then used multiple linear regression analysis to quantify the impacts of POA and SOA on BC-coated OC. Our results showed that the average contribution of SOA to the coated OC was nearly twice that of POA (65% vs. 35%) in Beijing, while POA and SOA contributed similarly in Gucheng. respectively. Light absorption enhancement showed strong dependence on (NO 3 + SO 4 )/C n at both sites yet the changes in E abs appeared independent of (OC + Metal)/C n . As shown in Fig. 4c, the ratio of (OC + Metal)/C n still presented high values at E abs = ~1, while the secondary species coated on BC were negligible. These results suggested OC and metals were likely either filled internal void spaces of fresh BC or mainly externally mixed with BC which did not induce light absorption enhancement at the urban site. As the progress of aging, E abs increased significantly mainly due to the increased secondary coating materials. 225 Similar to Beijing, E abs also increased significantly as a function of (NO 3 + SO 4 )/C n in Gucheng. The difference is the high background E abs of ~1.20 in GC when the (NO 3 + SO 4 )/C n ratio was closed to 0. Combined with the peak area ratio of (OC + Metal)/C n was about 2.5 at that time, we found that OC and metals were not only as filler materials but also coated on fresh BC and induced light absorption enhancement at the rural site. After the further aging process in atmosphere, E abs was mainly due to the increased secondary inorganic components coated on BC. Considering that the measurements of BC mixing state 230 by SPAMS are often not available in field campaigns, we predicted the E abs by using Eqs. 1 -6 with aerosol species measured by AMS. The predicted E abs showed overall agreements with the measured values in both BJ and GC, yet the variations at specific NR-PM 1 /eBC were much smaller (Fig. S4). We also estimated the E abs in summer 2017 using the same method and

Effects of mixing state on Eabs
The PMF analysis is used to characterize the effects of different mixing state on E abs (Fig. 5). Note that E abs was not estimated when the factor contributed negligibly to the total BC, such as Factor5 in BJ. As illustrated in Fig. 5, Factor2 is the major type 240 of aged BC in the urban region, accounting for more than 60% of the total BC. This factor was dominated by BC N , BCOC N , BCOC NS and BC NS , and presented a high E abs of 1.38. Comparatively, FactorB is the major type of aged BC in the rural area which was dominated by BCOC N . The E abs was ~1.35 for this factor, which was comparable to that in BJ. As this factor evolved towards FactorA after further aging and internally mixed with a large amount of sulfate, E abs was increased up to 1.41. In Beijing, the relatively fresh traffic emissions are dominant in Factor4 which showed a negligible impact on light absorption 245 enhancement (~1.06), consistent with the results in previous studies (Liu et al., 2017;Sun et al., 2021). Compared with traffic emissions, the relatively fresh biomass burning and coal combustion emissions (Factor3) comprising mainly Mix Source and BCOC S showed a moderate E abs (~1.11) in BJ. Although the mixed fresh primary emissions (FactorD) in GC presented a relatively low E abs (1.06), we found that the FactorC from coal combustion emissions and mixed with much OC and nitrate showed a much higher E abs (~1.31) than that in BJ (Factor3). After more aging, more sulfate could be internally mixed with 250 BC and enhanced the E abs , such as Factor1 with E abs up to 1.42. Overall, E abs shows a similar dependence on the evolution of mixing state of BC-containing particles at urban and rural sites, i.e., fresh BC particles from primary emissions (e.g., biomass burning, coal combustion and traffic) showed the small E abs (1.06~1.11). At relatively high RH level, BC could be directly mixed with nitrate or OC-nitrate (BC N and BCOC N ), accounting for more than 60% of BC, and lead to an increase in E abs Shen, W., Dai, X., Huang, Z., Hou, Z., Cai, W., Du, X., Zhou, Z., Li, M., and Li, L.: Improvement of the dynamic range of data acquisition system in single particle mass spectrometry, J. Chinese Cinemas., 039, 331-336, 2018. Silva, P. J., Liu, D. Y., Noble, C. A., and Prather, K. A.: Size and chemical characterization of individual particles resulting from biomass burning of local Southern California species, Environ. Sci. Technol., 33, 3068-3076, 10.1021/es980544p, 1999 Song, X. H., Hopke, P. K., Fergenson, D. P., and Prather, K. A.: Classification of single particles analyzed by ATOFMS using an artificial neural network, ART-2A, Anal. Chem., 71, 860-865, 10.1021/ac9809682, 1999. Sun, J., Xie, C., Xu, W., Chen, C., Ma, N., Xu, W., Lei, L., Li, Z., He, Y., Qiu, Y., Wang, Q., Pan, X., Su, H., Cheng, Y., Wu, C., Fu, P., Wang, Z., and Light absorption of black carbon and brown carbon in winter in North China Plain: comparisons between urban and rural sites, Sci. Total Environ., 770, 10.1016/j.scitotenv.2020.144821, 2021 Thamban Xie, C., Xu, W., Wang, J., Liu, D., Ge, X., Zhang, Q., Wang, Q., Du, W., Zhao, J., and Zhou, W.: Light absorption enhancement of black carbon in urban Beijing in summer, Atmos. Environ., 213, 499-504, 2019a.