Photochemical Processing of Aqueous Atmospheric Brown Carbon

. Atmospheric Brown Carbon (BrC) is a collective term for light absorbing organic compounds in the atmosphere. While the identiﬁcation of BrC and its formation mechanisms is currently a central effort in the community, little is known about the atmospheric removal processes of aerosol BrC. As a result, we report a series of laboratory studies of photochemical processing of BrC in the aqueous phase, by direct photolysis and OH oxidation. Solutions of ammonium sulfate mixed with 5 glyoxal (GLYAS) or methylglyoxal (MGAS) are used as surrogates for a class of secondary BrC me-diated by imine intermediates. Three nitrophenol species, namely 4-nitrophenol, 5-nitroguaiacol and 4-nitrocatechol, were investigated as a class of water soluble BrC originating from biomass burning. Photochemical processing induced signiﬁcant changes in the absorptive properties of BrC. The imine-mediated BrC solutions exhibited rapid photo-bleaching with both direct photolysis and OH 10 oxidation, with atmospheric half-lives of minutes to a few hours. The nitrophenol species exhibited photo-enhancement in the visible range during direct photolysis and the onset of OH oxidation, but rapid photo-bleaching was induced by further OH exposure on an atmospheric timescale of an hour or less. To illustrate atmospheric relevance of this work, we also performed direct photolysis experiments on water soluble organic carbon extracted from biofuel combustion samples and observed 15 rapid changes in optical properties of these samples as well. Overall, these experiments indicate that atmospheric models need to incorporate representations of atmospheric processing of BrC species to accurately model their radiative impacts.


Introduction
There is increasing awareness of the importance of light absorbing organic compounds in the at-20 mosphere (Kirchstetter et al., 2004;Chen and Bond, 2010;Lack et al., 2012;Bahadur et al., 2012;Laskin et al., 2015). Highly variable in sources and identity, this class of poorly characterized organic compounds has been collectively termed Atmospheric Brown Carbon (BrC) (Andreae and Gelencser, 2006). BrC significantly alters the traditional view that organic carbon interacts with solar radiation via only scattering (Chung and Seinfeld, 2002). In the visible range of solar radiation, 25 BrC absorption can affect the direct radiative effect of organic carbon (Feng et al., 2013;Lin et al., 2014;Liu et al., 2014). In particular, Feng et al. (2013) have shown that defining a fraction of organic aerosol as strongly light-absorbing BrC in a global chemical transport model can shift the direct radiative effect of organic carbon from net cooling to net warming. Meanwhile in the near UV range, BrC absorption may affect the flux of short-wavelength radiation that is crucial for driving 30 atmospheric photochemistry (Jacobson, 1999). Motivated by such atmospheric impacts, the characterization of the sources, molecular identity and processing of BrC is a central effort in the aerosol chemistry community.
Atmospheric BrC arises from multiple sources, including primary emission from biomass burning (BB) (Andreae and Gelencser, 2006;Alexander et al., 2008;Chen and Bond, 2010;Lack et al., 2012; 35 Kirchstetter and Thatcher, 2012;Saleh et al., 2014), as well as anthropogenic emissions (Bond, 2001;Zhang et al., 2011). Recently, BB has been reported as the dominant source of BrC observed at a location in Southeast US (Washenfelder et al.). The chemical composition of BB organic aerosol is highly complex, and the optical properties of BB organic aerosol also vary significantly with source fuels and burning conditions (Chen and Bond, 2010). Such complexity significantly hinders the separation, analyses, and molecular identification of BB BrC. BB BrC is at times considered to belong to Humic Like Substances (HULIS) (Hoffer et al., 2004;Graber and Rudich, 2006) and more recently a class of compounds categorized as extremely low volatility organic compounds (Saleh et al., 2014). Nitrophenols represent a class of speciated BrC species in BB plumes (Vione et al., 2009;Einschlag et al., 2009) and have been often employed as molecular tracers for BB 45 (Iinuma et al., 2010;Kitanovski et al., 2012). However, the contribution of nitrophenols to the total absorption of BB BrC is small, with the majority of organic chromophores unspecified (Mohr et al., 2013;Desyaterik et al., 2013).
Secondary formation of BrC in the atmospheric aqueous phases (i.e. cloud, fog and aerosol liquid water) has also been proposed. Photooxidation of aromatic compounds in the aqueous phase gives 50 rise to colored organic compounds with absorption spectra similar to those of HULIS (Chang and Thompson, 2010;Gelencser et al., 2003). Recently, a type of BrC arising from aldehydes reacting with nitrogen containing nucleophiles (e.g. ammonia, amines and amino acids) has been investigated extensively in the laboratory (Bones et al., 2010;De Haan et al., 2009Shapiro et al., 2009;Noziere et al., 2009;Sareen et al., 2010;Yu et al., 2011;Zarzana et al., 2012;Kampf et al., 55 2012; Sedehi et al., 2013;Powelson et al., 2013;Updyke et al., 2012;Nguyen et al., 2013;Laskin et al., 2014;Flores et al., 2014). Since the formation mechanism of this type of BrC involves an imine or a Schiff's base intermediate, this class of BrC is herein referred as "Imine BrC". Although imine intermediates do not absorb at the actinic range, they undergo subsequent reactions to form nitrogen-containing organic chromophores (Lee et al., 2013;Kampf et al., 2012;Yu et al., 2011). It 60 is generally believed that formation of individual chromophores with very low concentrations leads to the color (Nguyen et al., 2013). Imine BrC typically takes days to form in the bulk laboratory solution (Noziere et al., 2009;Shapiro et al., 2009;Sareen et al., 2010;Lee et al., 2013). However, studies have also shown that droplet evaporation may significantly accelerate the rate of such reactions, giving rise to rapid formation of BrC (De Haan et al., 2010;Zarzana et al., 2012;Nguyen et al., 65 2012;Lee et al., 2013;Galloway et al., 2014). While Imine BrC has received an enormous amount of attention by laboratory studies, it has not been reported in ambient measurements.
Finally,we note that a recent study has also suggested that charge transfer complexes between different functional groups may be responsible for absorption in the visible range (Phillips and Smith, 2014). We did not perform experiments targeted to this potential third class of BrC species.

70
Studies from the past decade (Blando and Turpin, 2000;Ervens et al., 2011) have indicated atmospheric aqueous phases (e.g. cloudwater and aerosol liquid water) as important reaction media, where organic compounds can be processed, leading to formation and further aging of secondary organic aerosol (SOA). Imine BrC, forming in the aqueous phase, can undergo subsequent photochemical processing. A previous study has observed rapid photolysis of components in the mixture 75 of methylglyoxal and ammonium sulfate, implying rapid photolysis of Imine BrC (Sareen et al., 2013). More recently, Lee et al. (2014b) investigated aqueous-phase processing of several classes of BrC and observed rapid decay of color (photo-bleaching). To date, there is no systematic investigation of the effect of OH oxidation on Imine BrC. BB BrC, on the other hand, can also be subject to aqueous-phase photochemical processing, given that BB particulate matter can be hygroscopic 80 (Petters and Kreidenweis, 2007;Petters et al., 2009) and a significant fraction of BB BrC belongs to water soluble organic carbon (WSOC) (Iinuma et al., 2007;Chen and Bond, 2010;Zhang et al., 2011Zhang et al., , 2013Washenfelder et al.). In particular, Zhong and Jang (2014) have recently observed changing optical properties of BB particles during photochemistry in chamber experiments, with more rapid changes observed under higher relative humidity.

85
As the ambient BrC is highly complex in nature, investigating the behavior of surrogate compounds or mixtures of BrC in the laboratory with reduced complexity can be highly valuable. In this study, nitrophenols are chosen as surrogates for BB BrC species. Certain nitrophenols exhibit relatively high Henry's law constants (Schwarzenbach et al., 1988) and have been observed from cloudwater samples (Luttke and Levsen, 1997;Luttke et al., 1999;Harrison et al., 2005;Desyaterik 90 et al., 2013), hence their aqueous-phase chemistry can be highly relevant to the atmosphere. Previous studies have investigated aqueous phase UV photolysis (Chen et al., 2005;Zhao et al., 2010), OH oxidation (Einschlag et al., 2003(Einschlag et al., , 2009Vione et al., 2009), as well as heterogeneous oxidation (Knopf et al., 2011;Slade and Knopf, 2014) of nitrophenols, but a clear connection to their optical properties has not been made. For surrogates of Imine BrC, solutions of glyoxal or methylglyoxal 95 mixed with ammonium sulfate are chosen because the precursor compounds are highly relevant to atmospheric aqueous phases, and this type of Imine BrC has been investigated by previous studies (see references listed previously).
In this study, we systematically investigate how atmospheric photochemical processing mechanisms affect Imine BrC and nitrophenols (as surrogates of BB BrC) in the aqueous phase, focusing 100 on changes in their absorptive optical properties. The dual objectives are (1) to quantify the rates of direct photo-bleaching and/or photo-enhancement under realistic radiation condition, and (2) to evaluate the atmospheric importance of BrC oxidative processing, with a particular focus on OH oxidation. We perform these experiments quantitatively under known light and OH exposures, so as to establish which processing mechanism is likely to dominate in the atmosphere. To tie our laboratory 105 experiments to ambient conditions, we also performed direct photolysis experiments on the WSOC extracted from biofuel combustion particles.

Preparation of BrC Solutions
The experimental procedures for Imine BrC and nitrophenols are illustrated in Fig. 1. Solutions of 110 ammonium sulfate mixed with glyoxal (GLYAS) or methylglyoxal (MGAS) were chosen as laboratory surrogates to represent Imine BrC. Stock solutions (200 mL in volume) were made by mixing ammonium sulfate (1.5 M, Sigma Aldrich) with either 0.5 M of glyoxal (Sigma Aldrich, 30 % in water) or 0.2 M of methylglyoxal (Sigma Aldrich, 40 % in water) in 250 mL glass jars. All the solutions were prepared using deionized water (18 mΩ-cm) with total organic carbon less than 1 parts 115 per billion (ppb). The stock solutions were sealed and placed in the dark under room temperature for 2 to 3 months. During this time, the color of the solutions turned dark yellow and eventually dark brown, consistent with previous studies (Shapiro et al., 2009;Sareen et al., 2010;Lee et al., 2013).
Although 2 to 3 months is much longer than typical atmospheric aerosol lifetimes, our previous work has shown that the absorption spectra of Imine BrC obtained this way closely resembled those 120 obtained from droplet evaporation occurring on the time scales of seconds or less (Lee et al., 2013).
The experimental solutions were created by diluting the concentrated stock solutions, typically by a factor of 200, to concentrations that optimize the UV-Vis detection at 400 nm (see next section).
Three nitrophenol compounds (4-nitrophenol (4NP), 5-nitroguaiacol (5NG) and 4-nitrocatechol (4NC)) were chosen to represent primary BB BrC (structures shown in Fig. 2) and are investigated 125 individually. 4NP and 4NC have been detected from BB affected cloudwater samples (Desyaterik et al., 2013) while 5NG has been previously used in the laboratory as a model compound for BB or-ganic matter (Knopf et al., 2011). Commercial standards of these compounds were purchased from Sigma Aldrich and were used without further purification. Individual stock solutions (1 mM) were created every few days, and the experimental solutions were made by diluting the stock solution to 130 4 to 15 µM depending on the nitrophenol species and the type of experiment. This range of concentration matches that of nitrophenols detected in cloudwater (Desyaterik et al., 2013).

Direct Photolysis and OH Oxidation Experiments
Direct photolysis and OH oxidation experiments were conducted separately (Fig. 1). Direct photolysis experiments were performed with a Suntest CPS photo-simulator (Atlas) equipped with a Xe It is crucial to measure the steady state concentration of OH radicals ([OH]ss) in the OH oxidation experiments in order to make sound environmental implications. An aerosol chemical ionization 160 mass spectrometer (Aerosol-CIMS) was employed for this purpose. The experimental setup is similar to that in one of our previous studies (Zhao et al., 2012). Briefly, the experimental solution is constantly atomized with a TSI constant output atomizer (model 3076). The aerosol flow is intro-duced through a heated metal line (100 • C), where organic compounds volatilize to the gas phase and are detected by a quadruple CIMS equipped with iodide water cluster reagent ion (I(H 2 O) − n ).

165
The I(H 2 O) − n reagent ion detects oxygenated organic compounds by forming iodide ion clusters (Aljawhary et al., 2013;Lee et al., 2014a;Zhao et al., 2014). The [OH]ss was estimated by tracking the pseudo 1st order decay of a reference compound with known OH reactivity. For the Imine BrC, unreacted glyoxal or methylglyoxal in the solutions were used as the tracer compounds because their mono-hydrates are detectable by the I(H 2 O) − n reagent ion (Zhao et al., 2012). The OH oxidation rate (3) it reacts with OH rapidly, with a second order rate constant of 1.9 × 10 9 M −1 s −1 (Hoffmann et al., 2009).

Direct Photolysis of WSOC from Biofuel Combustion
The biofuel combustion samples were collected in Henan Province, China (Li et al., 2007).
Agricultural residues, typically used as biofuels in the local area, were burned in an improved stove 180 commonly used in the area. A detailed description of particle collection and the physical properties of the generated particles are provided in Li et al. (2007). Briefly, particles were withdrawn from the stove, and the PM 2.5 fraction was collected on quartz filters after dilution. The quartz filters were baked under 450 • C before collection, and the samples were stored frozen after collection.
Organic carbon (OC) and elemental carbon contents of each filter sample were measured following 185 a method originally developed at the Environment Canada's laboratory in Toronto for measuring δ 13 C of OC/EC (Huang et al., 2006) and later improved by Chan et al. (2010) to be used as the standard OC/EC measurements in the aerosol baseline measurements in Canada. In the current work, we investigated the WSOC from two filter samples, collected from burning of kaoliang stalks and cotton stalks, respectively. A quarter of the filter was extracted in 10 mL of deionized water by 190 constant shaking for 30 min. The extracts were used as the experiment solution after filtration with a 0.2 µm syringe filter. We extracted the same filter a second time and found that the absorption in the second extract was less than 10 % of the first extract. However, it is difficult to estimate the extraction efficiency of total organic carbon. The filtered extract was illuminated with the same solar simulator, and its absorption was monitored with the same waveguide capillary spectrometer mentioned in 195 Sect. 2.2. Oxidation by OH radicals was not performed for these samples due to limited amount of sample volume.

Light Absorption by BrC
Absorption spectra of the BrC solutions are displayed in Fig To provide more quantitative values, we also obtained the wavelength dependent mass absorption initialized by compounds such as imidazole and imidazole-carboxaldehyde. We examined the pres-ence of this type of reaction by varying the initial concentration of the Imine BrC. However, the concentration of Imine BrC did not affect its photolysis rate constant (Sect. S3). This indicates that photosensitized reactions either did not take place in our reaction system, or were not indicated by the color change.

OH Oxidation of Imine BrC
Rapid photo-bleaching was observed also during the OH oxidation experiments. Figure 4a shows As listed in Table 1, the k II OH values for the GLYAS and the MGAS systems are determined to be (2.1 ± 1.1) × 10 10 and (1.2 ± 0.3) × 10 9 M −1 s −1 , respectively. The uncertainty represents SD from between 3 and 4 experimental replicates. We note that the k II OH value for the GLYAS system is essentially diffusion limited.

260
We estimate the atmospheric half-life (τ 1/2 ) of Imine BrC against direct photolysis and aqueous phase OH oxidation based on the observed absorbance change at 400 nm (Table 1). The τ 1/2 values were obtained by extracting the time when the signal reached half of its original value, and the uncertainty represents the range obtained from three replicates. Since the photon flux in the solar simulator is similar to that in the ambient atmosphere (see Sect. S1), the experimentally determined 265 τ 1/2 values, 90±12 and 13±3 min for the GLYAS and MGAS systems, directly reflect the photolytic τ 1/2 of these Imine BrC species in the ambient atmosphere. These τ 1/2 values are on the same order as another type of Imine BrC generated from Limonene SOA and ammonia vapor (Lee et al., 2014b), implying that rapid photolysis will be a common characteristic for this type of BrC. The OH oxidation half-lives are estimated by assuming an ambient cloudwater [OH]ss of 1 × 10 −13 M 270 which represents the upper band of OH in remote cloudwaters . This [OH]ss, together with the k II OH determined in the previous section (Sect. 3.2.2), yields OH oxidation τ 1/2 of 5 and 98 min for the GLYAS and the MGAS solutions, respectively. The rapid bleaching implies that the daytime lifetime of Imine BrC is likely very short in the atmosphere, leading to relatively low concentrations. Knowing that droplet evaporation can lead to rapid formation of Imine BrC on

285
The difference in the major removal mechanisms for GLYAS and MGAS arises from the additional methyl group on methylglyoxal as compared to glyoxal, as we propose in Fig. 5. The methyl group prevents the carbonyl functionality from hydrating into its geminal diol which does not absorb actinic radiation. On the other hand, H-abstraction from a methyl hydrogen is expected to be slower than from the tertiary hydrogen on the geminal diol. In Fig. 5, we use imidazole carboxaldehyde, proposed 290 as a major product in the GLYAS solution (Kampf et al., 2012;De Haan et al., 2010;Yu et al., 2011), as an example to demonstrate this concept.

Direct Photolysis of Nitrophenols
The spectral change of a 4NC solution during a direct photolysis experiment is shown in Fig. 6, color 295 coded by illumination time, with the inset illustrating the change at different illumination times. The change exhibits wavelength-dependence, with a decrease of absorption between 300 and 380 nm but an increase of absorption at 260 nm and above 400 nm. The spectral change is likely due to a combination of 4NC decay and formation of one or more reaction products. Similar trends were also observed for 4NP and 5 NG (Sect. S4). The most noteworthy observation for all the nitrophenols is a photo-enhancement of absorption at wavelengths longer than 400 nm, i.e. in the visible range.
Since the photo-enhancement at 420 nm was the most significant for all the three nitrophenols, we conducted a series of experiments to better characterize the absorbance change at this wavelength.
Formation of color at 420 nm is first order with respect to the precursor nitrophenols, confirmed by altering their concentrations. The discussion below is primarily based on the results from 4NC, while 305 the results of 4NP and 5 NG are included in Sect. S5.
The effect of OH radical is examined first. Previous studies have shown that nitrite ion can be produced during UV irradiation of nitro-aromatic compounds, via photo-induced nucleophilic substitution reactions (Nakagawa and Crosby, 1974;Dubowski and Hoffmann, 2000;Chen et al., 2005).
Nitrite is a photolytic source of OH radical and can potentially affect our direct photolysis exper-  Fig. 7, which does not exhibit significant difference from the experiment without the OH scavenger (blue trace). For 4NP, the OH scavenger reduced but did not completely remove the color formation at 420 nm (Sect. S5). We conclude that photo-enhancement is indeed induced by direct photolysis even without OH radicals present.
Effects of the solution pH are also examined because the light absorption of phenolic compounds 320 is pH dependent, with phenolate being a better absorber than phenol. Phenolate contains additional lone-pair electrons that can participate in the conjugation system, leading to more efficient light absorption. The absorption spectra of the three nitrophenols at various solution pH values are shown in Sect. S6. Light absorption of 4NP and 4NC at 420 nm increased significantly at higher solution pH due to formation of phenolate, but 5NG did not exhibit pH dependence. A meta-nitrophenol 325 compound, such as 5NG, is known to be less acidic than para-and ortho-nitrophenols (i.e. 4NP and 4NC). It is likely that the 5NG phenolate did not form in the range of pH investigated.
The absorbance (420 nm) time profiles of 4NC at two additional solution pH (i.e. pH 4 and 3) are displayed in Fig. 7. The photo-enhancement is more significant at higher solution pH. This is perhaps due to the fact that the products formed also exhibit pH dependent light absorptivity. 4NP 330 and 5NG exhibit unique trends of pH dependence as shown in Sect. S5.
We determined the effective 1st order rate coefficient of photo-enhancement (k I direct ) for 4NC by fitting the observed absorbance at 420 nm to a 1st order growth curve. The k I direct values determined for 4NC are summarized in Table 2. Photo-enhancement in the cases of 4NP and 5NG exhibited stronger linearity, which made fitting to 1st order growth curve difficult. Instead of k I direct , we report 335 an absorbance-based rate constant for these two compounds, and the details are provided in the SI Section S5.

OH Oxidation of Nitrophenols
Oxidation by OH radicals induced rapid bleaching of all nitrophenols investigated, but the decay of absorbance was not monotonous. The spectral change of 4NC during an OH oxidation experiment is 340 shown in Fig. 8a while the time profile of absorbance at 420 nm is shown in Fig. 8b. Results for 4NP and 5NG can be found in the Sect. S7. All the experiments were performed at pH 5 and in duplicate to confirm reproducibility. For all the three nitrophenols investigated, the absorbance exhibited initial increase, followed by decay at longer illumination time.
The initial color formation observed in the current study exhibits similarities with several previous 345 investigations of BB BrC. Gelencser et al. (2003) and Chang and Thompson (2010)  We propose that the observed trend during OH oxidation is due to initial functionalization followed by ring-cleavage reactions. Previous studies (Sun et al., 2010) have shown that OH oxidation leads to hydroxylation of the aromatic ring, in analogy to the gas phase (Atkinson, 1990). The additional hydroxyl group is electron donating, with its lone pair electrons contributing to the conjugation 360 and leading to enhanced absorption. We note that oligomeric products have also been reported from OH oxidation of phenolic compounds (Sun et al., 2010;Chang and Thompson, 2010). In particular, Chang and Thompson have observed significant enhancement of absorption, and they proposed that the absorption is attributed to HULIS produced from phenol OH oxidation. To simulate cloudwater chemistry, we used nitrophenol concentrations orders of magnitude lower than those used in Chang 365 and Thompson and so we consider the formation of oligomers less important in our system.
To quantitatively assess the formation and decay rate of color, we applied a kinetic model framework based on the absorbance at 420 nm (Fig. 9a). The OH radical concentration is assumed to be in steady state at 3.2 × 10 −13 M which is the average of measured [OH]ss using the Aerosol CIMS method. The nitrophenol precursor (NP) follows a prescribed pseudo 1st order decay with a rate 370 constant, k I NP , which is estimated based on 4NP OH reactivity reported by Einschlag et al. (2003). A colored product (CP) is formed from NP with a pseudo 1st order rate constant k I color , but simultaneously undergoes photo-bleaching with another pseudo 1st order rate constant k I bleach . Although NP can likely give rise to more than one CP species, the colored products are lumped into a single compound for simplicity. The sum of absorbance from NP and CP is treated as the total absorbance 375 of the solution. We found the optimal combination of k I color and k I bleach values that minimizes the sum of the squared difference between the modeled and the observed absorbance changes. We note that k I color and k I bleach are absorbance-based rate constants and should not be confused with concentrationbased rate constants. If the identity and molar absorptivity of CP are characterized in future studies, these absorbance-based rate constants can be converted into concentration-basis.

380
The results for one 4NC experiment are shown in Fig. 9b. The two shaded areas in Fig. 9b represent modeled absorbance due to the precursor, 4NC, and the CP, respectively. The red trace is the absorbance change during the experiment shown in Fig. 8. Results for 4NP and 5NG, along with detailed model conditions are included in the SI Section S8. For all the three nitrophenols, this model captures the initial increase and later decay of color, but the time at which the absorbance reaches its 385 maximum and the decay rate at the end of the experiment are more difficult to match. This is perhaps due to the fact that nitrophenols form multiple generations of colored products, giving rise to a more dynamic evolution of absorbance than the current model framework can produce. Nevertheless, the model represents a novel effort to estimate the rates of photo-enhancement and bleaching during OH oxidation of nitrophenols. The optimal k I color and k I bleach values for the three nitrophenols are listed in 390   Table 3. Since these values are all psuedo-1st order rate constants, their corresponding second order rate constants (k II color and k II bleach ) are also calculated using Eq.
(2) and provided in Table 3. The values reported in Table 3 are the average of two replicates performed for each nitrophenol. Relative errors are roughly 10 % for 4NP and 5NG, and 15 % for 4NC.

395
Our results indicate that the photo-bleaching by OH oxidation is rapid and presents the dominant fate for BrC represented by nitrophenols. As the [OH]ss in our experiment (3.2 × 10 −14 M) is roughly that cloudwater in remote areas , the light absorptivity of nitrophenols is expected to reach its maxima and to be bleached within one hour of in-cloud time. On the other hand, photo-enhancement during direct photolysis is much slower, with color forming over a time 400 scale of hours. This observation agrees with Vione et al. (2009) where they also determined radical chemistry as the dominant sink of 4NP compared to direct photolysis. That being said, this trend may not apply to all nitrophenols. For instance, dinitrophenols represent an interesting group of compounds to investigate in the future, as the additional nitro group deactivates OH radical reactions (Einschlag et al., 2003) but enhances light absorption (Schwarzenbach et al., 1988).

Direct Photolysis of WSOC from Biofuel Combustion Samples
A change in absorptivity was observed when WSOC from biofuel combustion samples was exposed to simulated sunlight. Results for the kaoliang stalk sample and the cotton stalk sample are shown in Fig. 10a and b, respectively. Their absorbance changes at three wavelengths (350, 400 and 420 nm) are also shown in Fig. 10c and d, respectively. WSOC from the two samples exhibited different 410 trends, with the kaoliang stalk sample showing a temporary photo-enhancement shortly after the initiation of illumination, and the cotton stalk sample exhibiting monotonous photo-bleaching. The trends for the sample at different wavelengths demonstrate the complexity of the real biomass burning samples. Our results provide qualitative evidence that the optical properties of WSOC extracted from BB BrC can change upon photochemistry.

4 Conclusions and Atmospheric Implications
The overall conclusion from this work is that because atmospheric brown carbon species are organic chromophores and susceptible to photochemical degradation, their optical properties are altered by aqueous-phase photochemical processing with both photo-enhancement and photo-bleaching possibly occurring. In particular, Imine-mediated BrC, arising from aqueous-phase reactions between 420 carbonyl compounds and nitrogen-containing nucleophiles, undergoes rapid photo-bleaching via both direct photolysis and OH oxidation. Such rapid photo-bleaching may indicate a reason why Imine BrC has not yet been observed in ambient samples. Bleaching of glyoxal-ammonium sulfate (GLYAS) BrC was predominantly driven by OH oxidation whereas that for methylglyoxalammonium sulfate (MGAS) was driven by direct photolysis. Three species of nitrophenols were 425 investigated as an important subset of biomass burning BrC. Photo-enhancement of absorption was observed when the nitrophenol species are illuminated with simulated sunlight, as well as during the initial stages of OH oxidation. Although such photo-enhancement can potentially magnify the direct radiative effect of nitrophenols, photo-bleaching of nitrophenols with further OH exposure was observed to be also rapid. This is the first investigation of OH oxidation induced effects on the optical 430 properties of BrC, demonstrating its importance in determining the atmospheric significance of BrC.
Lastly, a study of biofuel BrC species illustrated that the optical properties of ambient samples are also rapidly altered. These findings are in general agreement with prior studies that have also seen evidence for photo-bleaching (Lee et al., 2014b;Zhong and Jang, 2014;Sareen et al., 2013).
Using atmospherically relevant light levels and aqueous OH concentrations, the timescales for 435 these changes are all rapid, i.e. on the order of an hour or less. This indicates the atmospheric concentrations of BrC species will be highest during the night, when their atmospheric significance for shortwave radiative forcing is zero. For example, in the case of the Imine BrC species, they may form slowly during the night in cloud or aerosol water and then will decay away rapidly in the morning. It is expected that during the daytime their steady state concentrations will be highest in regions 440 where there is considerable droplet evaporation proceeding. Biomass burning BrC emitted during the night time will be stable. Upon sunrise, photochemistry can induce photo-enhancement, but the BrC concentration will also fall with further photochemical processing. The magnitude of photoenhancement and bleaching is likely dependent to the BrC components, as well as OH exposure.
We conclude that atmospheric models that include only source functions and depositional loss rates 445 for BrC-bearing organic aerosol will misrepresent the radiative impacts of these particles, requiring additional parameterizations for photo-bleaching and photo-enhancement.
While this study provides fundamental information on the behavior of BrC during photochemical processing, it is also subject to several limitations. Choices of a few single BrC species may limit the atmospheric implications one can make from this study. As the trends observed in the WSOC 450 of biofuel combustion samples are distinct from those observed in Imine BrC and nitrophenols (see Section 3.4), the current study also illustrates the importance for a more systematic investigation for ambient BrC from different origins. The kinetic information obtained from this study is based on the changes of the bulk light-absorptivity. Molecular-level investigations should be performed in the future to convert the absorbance-based rate coefficients into concentration-based ones. Whereas 455 this paper has focused upon aqueous phase processing of water-soluble BrC , a significant fraction of BrC is water-insoluble (Chen and Bond, 2010;Zhang et al., 2013;Washenfelder et al.), with heterogeneous oxidation likely affecting its atmospheric lifetime. It will be important to also assess the rates of heterogeneous oxidation of BrC species in particles via interactions with gas phase oxidants and to study direct photolysis in aerosol particles.        The shaded areas in (b) are the contributions from a newly formed colored product (CP) and the decaying 4NC, respectively. The red line follows data from an experiment.