Effects of photochemical oxidation on the mixing state and light absorption of black carbon in the urban atmosphere of China

The relationship between the refractory black carbon (rBC) aerosol mixing state and the atmospheric oxidation capacity was investigated to assess the possible influence of oxidants on the particles’ light absorption effects at two large cities in China. The number fraction of thickly-coated rBC particles (FrBC) was positively correlated with a measure of the oxidant concentrations (OX = O3 + NO2), indicating an enhancement of coated rBC particles under more oxidizing conditions. The slope of a linear regression of FrBC versus OX was 0.58% ppb−1 for Beijing and 0.84% ppb−1 for Xi’an, and these relationships provide some insights into the evolution of rBC mixing state in relation to atmospheric oxidation processes. The mass absorption cross-section of rBC (MACrBC) increased with OX during the daytime at Xi’an, at a rate of 0.26 m2 g−1 ppb−1, suggesting that more oxidizing conditions lead to internal mixing that enhances the light-absorbing capacity of rBC particles. Understanding the dependence of the increasing rates of FrBC and MACrBC as a function of OX may lead to improvements of climate models that deal with the warming effects, but more studies in different cities and seasons are needed to gauge the broader implications of these findings.


Introduction
Black carbon (BC) aerosol is emitted into the atmosphere mainly through the incomplete combustion of fossil fuels and biomass. These particles efficiently absorb solar radiation, and they have been recognized as the second largest contributor to anthropogenic radiative forcing after carbon dioxide (Jacobson 2001, Ramanathan andCarmichael 2008, Bond et al 2013). Estimates of the BC direct radiative forcing have generally ranged from 0.2 to 0.8 W m À2 (Jacobson 2001, Kim et al 2008, Bond et al 2011. Although the importance of BC in the context of global warming has long been recognized, the Fifth Assessment of the Intergovernmental Panel on Climate Change (IPCC) highlighted the fact that there are still large uncertainties in magnitude of the BC direct radiative effects (IPCC 2013).
Variations in the BC direct radiative forcing in different regions can be explained by the differences in the emission, transport, aging, and removal of BC particles (Shiraiwa et al 2008, Schwarz et al 2010, Bond et al 2013. The aging of BC particles as expressed here refers to the transformation of the BC mixing state, namely, the degree to which BC particles become coated with other chemical substances. This is one of Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence.
Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. main factors that have led to large discrepancies in model estimates of radiative effects because those results are affected by whether the BC particles are treated as internally-or externally-mixed with other materials. Fresh BC particles from many sources, including fossil fuel combustion, tend to be hydrophobic, and they are initially externally mixed with other particulate matter, but they become internally mixed and more hydrophilic as coatings by other components accumulate during atmospheric aging (McMeeking et al 2011a, Liu et al 2013. Both physical and chemical processes, including coagulation, condensation, and heterogeneous reactions, can cause changes in the BC mixing state (Kondo et al 2011a, Liu et al 2013, Browne et al 2015. The BC mixing state plays a crucial role in determining the particles' optical properties because light-scattering materials can refract sunlight to the BC cores through a 'lens effect' , and this can amplify the heating potential of BC-containing particles (Lack and Cappa 2010). Both chamber experiments and field observations have shown that the light absorption of BC can be significantly enhanced by internal mixing. For example, Schnaiter et al (2005) investigated dieselsoot particles coated with secondary organic compounds, and they found an absorption enhancement of 1.8-2.1 fold for internally-mixed BC compared with externally-mixed BC. Results presented by Liu et al (2015) clearly demonstrated that coatings can substantially enhance light absorption by BC emitted from mixed fossil fuel and residential solid fuel combustion sources in the UK during winter. On the other hand, Cappa et al (2012) found that the average absorption enhancement was small, only 6%, as BC particles underwent photochemical aging at two sites in California. These discrepancies show that the optical properties of BC particles vary in complex ways with respect to their chemical and physical characteristics, including composition, size, shape, and mixing state.
Photochemical oxidation plays an important role in the evolution of BC mixing state. The term oxidant (OX) as used here is the sum of the ozone and nitrogen dioxide mixing ratios (O 3 þ NO 2 ), and it can be considered as a proxy for atmospheric aging caused by photochemical reactions (Jenkin 2014, Canonaco et al 2015. In the urban atmosphere, OX is a more conserved tracer of photochemical processing than O 3 alone because fresh emissions of NO can react with O 3 to form NO 2 (Herndon et al 2008). In this paper, we report results from wintertime measurement campaigns in Beijing and Xi'an, China. The mixing state of the refractory black carbon (rBC) particles was determined with a single particle soot photometer (SP2) and light absorption was measured with a photoacoustic extinctiometer (PAX). The mixing ratios of OX, calculated from data obtained with O 3 and NO x (NO þ NO 2 ) analyzers, were used to represent the atmospheric oxidation capacity. The objective of the study was to investigate how two important properties of the rBC, that is, the mixing state and mass absorption cross-section, were related to the atmospheric oxidation capacity in the urban atmosphere of China.

Experimental methods
The studies were conducted at two urban sites, one in Beijing and the second in Xi'an, China (see supplemental figure S1 available at stacks.iop.org/ ERL/12/044012/mmedia). Beijing is one of the largest cities in the world (population > 21 million), and it is located in an area bordering the North China Plain and the Inner Mongolia Plateau; it is surrounded by the Yanshan Mountains to the north and Taihang Mountains to the west. The measurements in Beijing were made on the rooftop of the Institute of Remote Sensing Applications (IRSA, ∼18 m above ground level, 40.01°N, 116.39°E, Wang et al 2016a) from 1 to 18 February 2014. The sampling site is surrounded by an educational district and the Olympic Park; to the south is the Datun Road (∼450 m), where the traffic is heavy, while to the east is the Beichen West Road (∼150 m), which has less traffic. Xi'an is located on the Guanzhong Plain at the southern edge of the Loess Plateau, and it is the largest city in Northwest China (population of >8 million). The measurements in Xi'an were made on the rooftop of the Institute of Earth Environment, Chinese Academy of Sciences (IEECAS, ∼10 m above ground level, 34.23°N, 108.88°E, Cao et al 2009) from 8 to 20 February 2013. The Xi'an sampling site is surrounded by residential/commercial areas, and no major industrial sources are located nearby. To the east is the Fenghui South Road (∼20 m), where the traffic is usually light. Both sampling sites are located in urban zones, and they may be considered generally representative of urban conditions. Due to the rapid economic development, population growth, and urbanization over the past decades, both cities often suffer from high loadings of particulate pollutants (Cao et al 2012, Huang et al 2014. Previous studies had shown that coal burning is the most important source for rBC during the winter in Beijing (Schleicher et al 2013) while traffic is the dominant wintertime source for rBC in Xi'an (Wang et al 2016b).
The mixing state of individual rBC particles was determined with a commercially available SP2 instrument (Droplet Measurement Technologies, Boulder, CO, USA). The operating principles of the SP2 have been described in detail elsewhere (Schwarz et al 2006, Gao et al 2007, Cross et al 2010. Briefly, the SP2 uses a continuous intra-cavity Nd:YAG laser (1064 nm) with a Gaussian profile (TEM00 mode) to heat rBC-containing particles to their vaporization point, which causes the emission of an incandescence signal. The peak intensity of the incandescence signal is linearly related to the rBC mass, and it is influenced little by the particles' mixing state or morphology (Slowik et al 2007, Kondo et al 2011b. The time delay between the occurrence of the scattering and incandescence signal peaks is used to determine whether individual rBC particles are internally-mixed with non-rBC materials (Schwarz et al 2006, McMeeking et al 2011b, Perring et al 2011. Time delays occur because the coatings must be removed from the rBC particles before the onset of incandescence. In this study, time delays of ∼3 ms and ∼2 ms were used as the criteria for distinguishing thicklycoated rBC particles from uncoated or thinly-coated ones at Beijing and Xi'an, respectively. These criteria were based on the minima in the bimodal frequency distributions of the time delays for the full study periods at the two sites (see below figure 2). The incandescence signal was calibrated using standard fullerene soot samples (Lot F12S011, Alpha Aesar, Inc., Ward Hill, Massachusetts). The uncertainty in the rBC mass determination was estimated to be ∼25%. More details concerning the IEECAS SP2 operation and calibration can be found in our previous publication (Wang et al 2014).
The particulate light absorption coefficient at λ ¼ 532 nm was measured using a PAX (Droplet Measurement Technologies, Boulder, CO, USA). The PAX uses intracavity photoacoustic technology to directly measure the light absorption of aerosol particles. A laser beam produced by a modulated diode in the acoustic chamber of the instrument heats suspended absorbing aerosol particles, and the resulting pressure wave is detected with a sensitive microphone. Different mixing ratios of NO 2 were used to calibrate the instrument before and after the experiments.
Mixing ratios of O 3 and NO 2 , measured as 5 min averages, were obtained using a Model EC9810 photometric ozone analyzer (Ecotech Pty Ltd, Knoxfield, Australia) and a Model EC9841 NO/NO 2 /NO x (Ecotech Pty Ltd, Knoxfield, Australia) gas-phase chemiluminescence instrument, respectively. Standard reference O 3 and NO gases were used to calibrate the ozone and nitrogen oxides analyzers before and after the experiments. Relative humidity (RH) and temperature (5 min averages) were measured using an automatic weather station (MAWS201, Vaisala, Vantaa, Finland) configured with an RH/temperature probe (Model QMH101). The absolute humidity was estimated using the RH and temperature data.

Results and discussion
3.1. Characterization of OX Significant variations in the time-resolved (5 min averages) in OX were observed over the entire campaign periods at both Beijing and Xi'an, with mixing ratios ranging from 19 to 79 ppb and 10 to 60 ppb and averaging 40 ± 11 ppb and 25 ± 10 ppb at the two sites, respectively (supplemental figure S2). Differences in the daily levels of O 3 precursor gases (NO x ) and volatile organic compounds (VOCs) as well as the O 3 -NO x -VOCs coupled chemistry presumably were the main factors influencing the variations in the day-to-day OX values (Tang et al 2012, Feng et al 2016. Although O 3 contributed only ∼30% to the OX at both sites on average, the O 3 /OX ratios increased to ∼50% in the afternoon when OX reached its daily maximum.
A plot of the 5 min average OX mixing ratios (figure 1) shows that OX exhibited similar diurnal Environ. Res. Lett. 12 (2017) 044012 patterns at Beijing and Xi'an. The daily minimum occurred at ∼08:00 loca time (LT, all time references that follow are given in LT) followed by a sharp increase to a daytime peak in the afternoon at ∼16:00. Then, the OX mixing ratios declined continuously until the early morning of the following day. This diurnal cycle is typical of what has been found in other urban areas (Notario et al 2012), and it is likely driven by variations in local anthropogenic activities, photochemical conditions, and atmospheric boundary layer heights. The low OX values early in the day can be attributed to heavy motor vehicle emissions from the morning rush hour. It can be seen in Supplemental figure S3 that the NO mixing ratios at both sites reached high values at ∼08:00, which is when low O 3 values were observed, mostly likely due to the titration of O 3 by NO (Shaw et al 2015, Wang et al 2015. Increase in OX after sunrise were probably due to the photochemical production of O 3 and to some extent vertical mixing. Low mixing ratios of OX during the night can be explained by the lack of photochemical production and losses through titration reactions and surface deposition .

Effect of photochemical oxidation on rBC mixing state
The number fraction of thickly-coated rBC (F rBC ) is a good indicator of the percentage of rBC particles internally mixed with other substances, and it is calculated as the ratio between the number of thicklycoated rBC particles and total number of measured rBC particles. The mixing state of rBC can vary for particles from different sources; for example, rBC particles from biomass burning are more likely to be coated than those from traffic emissions (Schwarz et al 2008, Liu et al 2014. In addition, F rBC typically is higher for more aged rBC particles than fresh ones because coatings can accumulate through various atmospheric processes over time (McMeeking et al 2011b, Huang et al 2012). As we sampled rBC particles from a complex mixture of sources, the daily variations in F rBC were likely due to changes in the relative contributions of various rBC sources as well as subsequent aging of the particles. Supplemental figure  S2 shows that F rBC ranged from 35% to 92% with arithmetic mean of 59 ± 7% at Beijing and varied from 27% to 76% with average of 55 ± 9% at Xi'an. In comparison with previous studies made with SP2s deployed in other Chinese areas, the F rBC values at Beijing and Xi'an were comparable to those in winter at Jiaxing (Huang et al 2013) and autumn at Qinghai Lake (Wang et al 2014) where the averages were 53% and 50%, respectively. However, the F rBC at Beijing and Xi'an were higher than what was reported for spring/summer in Shanghai (Huang et al 2012) and spring in Xiamen (Wang et al 2016c) where the average values were 41% and 31%, respectively. These differences simply show that there are substantial spatial and temporal variability in the rBC mixing state in China. Figure 1 shows that F rBC exhibited similar diurnal patterns at Beijing and Xi'an with 'two peaks and two valleys' each day. The lower valley in the early morning around 07:00-09:00 can be explained by increases in fresh rBC particles emitted by local rush-hour traffic. At both sites, F rBC increased after sunrise and peaked at 15:00-16:00, and that was followed by a continuous decrease until sunset. The diurnal pattern of F rBC is broadly consistent with the variations of OX, indicating that photochemical oxidation may play an important role in the evolution of the rBC mixing state. The F rBC showed comparable rates of increase from 08:00 to 16:00 at the two sites, 0.8 % h À1 (r ¼ 0.96) at Beijing and 0.6 % h À1 (r ¼ 0.80) at Xi'an. Low F rBC values during the night (19:00-23:00) may be explained by the production of particles from primary anthropogenic emissions, especially those from heating activities at night. As the night progressed, the emitted rBC particles aged in the relatively shallow boundary layer and that led to another peak of F rBC at 03:00-05:00. However, identifying the specific mechanisms involved in the rBC night-time aging process is beyond the scope of this study, and further research is needed to uncover the mechanisms and processes responsible for this internal mixing.
To investigate possible effects of photochemistry on the rBC mixing state, the data collected from 08:00 to 16:00-a period when OX typically increasedwere examined to determine how the fraction of thickly-coated rBC changed with OX. Seventeen out of eighteen sampling days at Beijing and eleven out of thirteen sampling days at Xi'an showed increasing trends in OX from 08:00-16:00 (see vertical yellow and grey shading in Supplemental figure S2) while the remaining days at both sites exhibited decreasing OX trends (see light cyan shades). The F rBC values broadly tracked increases in OX except for the samples from 2, 8, and 16 February at Beijing and 12 and 17 February at Xi'an (see vertical grey shades in figure S2). Those five days were characterized by large variations in meteorological conditions; for example, as shown in figure S2, the absolute humidity between 08:00 and 16:00 sharply decreased on those five days but was rather constant on the other days. These changes in absolute humidity suggest the advection of air masses with different histories and characteristics than what was sampled during the more common conditions. More specifically, the lower humidity indicates either intrusions of free tropospheric air or advection of air from much drier areas, and neither of those cases would be optimal for studying the effects of the local/ regional chemistry. Thus, the data for those five days as well as the three days on which OX decreased were excluded in the following analysis and discussion.
The distributions of the time delays (Δt) between the scattering and incandescence signal peaks were used to investigate potential impact of photochemistry Environ. Res. Lett. 12 (2017) 044012 on the rBC mixing state. Figure 2 shows the frequency distributions of Δt at Beijing and Xi'an for rBCcontaining particles measured at 08:00-09:00 (lowest OX level) and at 15:00-16:00 (highest OX level) for the entire campaign periods. The frequency distributions for Δt from both sites were bimodal: one peak occurred at a time delay of ∼1.2 ms for Beijing and ∼0.8 ms for Xi'an (short-Δt) and the other at ∼5.0 ms for Beijing and ∼3.5 ms for Xi'an (long-Δt); these two peaks represent the thinly-and thickly-coated rBC particles, respectively (see Methods). From 08:00-09:00, a large fraction of the particles clustered at the short-Δt, reflecting the important contribution of thinly-coated rBC to the total rBC population; at 15:00-16:00, when photochemical oxidation processes were more active, a larger fraction of the particles clustered at the long-Δt, indicating that more of the rBC particles were thickly-coated.
To further investigate the impact of oxidant levels on the rBC mixing state, the F rBC data collected between 08:00-16:00 at both sites are plotted against OX mixing ratios in figure 3. It can be seen that F rBC was positively correlated with OX (r ¼ 0.65 for Beijing and 0.72 for Xi'an), indicating greater fractions of coated rBC particles under more oxidizing conditions. The enhancement in thickly-coated rBC observed between 08:00-16:00 can be explained by the photochemical production of secondary substances, including both inorganic and organic non-refractory compounds, which condensed onto the rBC. The slopes of the F rBC versus OX were 0.58% ppb À1 for Beijing and 0.84% ppb À1 for Xi'an, respectively, which may reflect differences in the oxidation rates of the rBC particles at the two sites. That is, a stronger effect at Xi'an compared with Beijing can be attributed to the more rapid increase in OX at Xi'an (2.9 ppb h À1 ) compared to Beijing (1.8 ppb h À1 ).

Effect of photochemical oxidation on MAC rBC
The mass absorption cross-section of rBC (MAC rBC , in m 2 g À1 ) is one of the most informative optical properties of the rBC aerosol because it reflects the relationship between the radiative effects of the rBC particles and their mass. The MAC rBC is calculated by dividing the absorption coefficient (s abs ) measured with the PAX by the rBC mass concentration from the SP2 (MAC rBC ¼ s abs /[rBC]). It is worth noting that the MAC rBC calculated in this way should be considered as an upper limit because even though their contributions are usually small, other chemical species, such as brown carbon, also can contribute to light absorption in mid-visible wavelengths (Laskin et al 2015). Due to the lack of PAX measurements in Beijing, the MAC rBC values could only be calculated for Xi'an, and those results are discussed below. The MAC rBC values generally followed a Gaussian distribution and showed a peak at 12.7 m 2 g À1 (supplemental figure S4). The overall average for the campaign was 14.6 ± 5.6 m 2 g À1 , which is ∼1.9 times higher than the value of 7.8 m 2 g À1 (extrapolated to 532 nm from 550 nm, assuming an absorption Ångström exponent of 1.0) that was suggested to be generally representative of uncoated rBC particles (Bond and Bergstrom 2006). The relatively high MAC rBC at Xi'an suggests that absorption was affected by internal mixing with other materials. Indeed, the internal mixing of rBC does increase the absorption of visible light, partly because the non-absorbing materials act like a lens and therefore refract light toward the absorbing rBC core (Lack and Cappa 2010). Figure 4 shows that a daily minimum in the 5 min average MAC rBC values occurred ∼08:00-09:00; the MAC rBC then started to increase at a rate of 0.6 m 2 g À1 h À1 until attaining the maximum value at 16:00. As day turned to night, the MAC rBC decreased at a rate of 0.2 m 2 g À1 h À1 until early morning. This diurnal pattern in MAC rBC was similar to the variations seen in the OX mixing ratios measured in parallel, and this suggests that the enhancement of the MAC rBC was largely driven by photochemical processes. We plotted the MAC rBC values against the OX mixing ratios between 08:00 and 16:00 for the days when OX was increasing (figure 5) to focus on the interval of increasing OX. This plot shows that MAC rBC was positively correlated (r ¼ 0.71) with the OX mixing ratios, and these results further support the argument that photochemical processes led to increases in MAC rBC . It can also be seen in figure 5 that MAC rBC increased with increasing F rBC values, and although the relationship was not strong, it was statistically significant (r ¼ 0.60, p < 0.05).
Positive correlations among MAC rBC , F rBC , and OX indicate that more oxidizing conditions favor the formation of secondary aerosol coatings on the rBC particles, which in turn enhance the particles' absorption efficiencies. The slope of 0.26 m 2 g À1 ppb À1 obtained from an orthogonal regression is arguably representative of the rate at which photochemical oxidation altered the MAC rBC . Results of a study by Cappa et al (2012) showed that the absorption enhancement increased more weakly with photochemical aging in California compared with our results. This implies that megacities in North America, and likely other parts of the world, may need to be treated differently compared with those in Asia where the concentrations particulate and secondary components are typically quit high (Huang et al 2014). Indeed, additional studies in different areas and other seasons are needed for more reliable and comprehensive assessments of the effects of oxidants on the optical properties and environmental effects-especially the warming potential-of rBC particles.

Conclusions
The mixing state of refractory black carbon (rBC) was measured with a single particle soot photometer (SP2) at Xi'an and Beijing, China, in February 2013 and 2014, respectively. The number fraction of thicklycoated rBC particles (F rBC ), expressed as a percentage, was used to determine the degree to which the rBC particles were internally mixed with other materials. The overall average F rBC values for the campaigns were 59 ± 7% at Beijing and 55 ± 9% at Xi'an, and the percentages of internally-mixed particles increased after sunrise. On average, from 08:00 to 16:00, F rBC increased at a rate of 0.8% h À1 at Beijing and 0.6 % h À1 at Xi'an, and these enhancements in the percentages of coated rBC particles were paralleled by increases in oxidant (OX ¼ O 3 þ NO 2 ) mixing ratios. Strong correlations between F rBC and OX were found between 08:00 and 16:00 at both sites, suggesting that the photochemical oxidation plays an important role in forming the rBC coatings. Regression analyses showed that the rate of increase in F rBC with OX was greater at Xi'an than Beijing (0.84 versus 0.58% ppb À1 ), suggesting that the rBC particles in Xi'an were more rapidly coated that those in Beijing. Light absorption measurements were not made at Beijing, but similar to F rBC , the mass absorption crosssection of rBC (MAC rBC ) at Xi'an tracked the increases in OX. Indeed, the MAC rBC varied linearly with the OX mixing ratios, and therefore the coatings on the rBC that resulted from photochemical reactions evidently increased the particles' absorption efficiencies. The slope of þ0.26 m 2 g À1 ppb À1 for the orthogonal regression between MAC rBC and OX is a measure of the rate at which the MAC rBC increased at Xi'an during the daytime. The mixing state and optical properties of rBC particles are among the most critical parameters used in the global and regional models of radiative forcing. Photochemically-induced changes in F rBC and MAC rBC should be investigated for different areas and seasons, and the results of those studies should be included in climate models to provide more reliable assessments of the warming potential of rBC particles. Browne  F rBC (%) Figure 5. Correlation between the mass absorption cross-section of refractory black carbon (MAC rBC ) and oxidant (OX) mixing ratios. Data points are color-coded with respect to the number fraction of thickly coated rBC (F rBC ).