Laboratory Insights into the Diel Cycle of Optical and Chemical Transformations of Biomass Burning Brown Carbon Aerosols

Transformations of biomass burning brown carbon aerosols (BB-BrC) over their diurnal lifecycle are currently not well studied. In this study, the aging of BB tar proxy aerosols processed by NO3• under dark conditions followed by the photochemical OH• reaction and photolysis were investigated in tandem flow reactors. The results show that O3 oxidation in the dark diminishes light absorption of wood tar aerosols, resulting in higher particle single-scattering albedo (SSA). NO3• reactions augment the mass absorption coefficient (MAC) of the aerosols by a factor of 2–3 by forming secondary chromophores, such as nitroaromatic compounds (NACs) and organonitrates. Subsequent OH• oxidation and direct photolysis both decompose the organic nitrates (ONs, representing bulk functionalities of NACs and organonitrates) in the NO3•-aged wood tar aerosols, thus decreasing particle absorption. Moreover, NACs degrade faster than organonitrates by photochemical aging. The NO3•-aged wood tar aerosols are more susceptible to photolysis than to OH• reactions. The photolysis lifetimes for the ONs and for the absorbance of the NO3•-aged aerosols are on the order of hours under typical solar irradiation, while the absorption and ON lifetimes toward OH• oxidation are substantially longer. Overall, nighttime aging via NO3• reactions increases the light absorption of wood tar aerosols and shortens their absorption lifetime under daytime conditions.


Contents
1. Experimental setup and parameters: 14 Figure S1 & Table S1 15 2. Box-model simulations of heterogeneous reactions for O 3 and NO 3 • in the AFR:       41 Detailed information on the chemical box model development and simulation can be found in our previous publication. 1  52

Box-model simulations of heterogeneous reactions for O 3 and NO 3 radical in the AFR
Eq.2 Eq.3 Where k p and k w are pseudo-first order loss rates to the particles' surface and to the reactor inner wall, respectively. γ eff is 56 effective uptake coefficient (unitless) for gas G, such as O 3 , NO 3 •, NO 2 , and N 2 O 5 . ω (m s -1 ) is molecular speed of gas G. S 57 is the total particle surface area exposed to reactant (nm 2 cm -3 ). S AFR and V AFR are inner surface area and volume of the 58 reactor. D int is the inner diameter of the cylindrical flow reactor we used, that equals to 105 mm. γ p and γ w are uptake 59 coefficient (unitless) to the particulate surface and to the AFR inner wall, respectively. Γ diff describes the gas phase diffusion 60 limitation (unitless) in the surface uptake by particles and reactor inner wall. For the uptake onto monodisperse spherical 61 particles, several methods have been suggested to calculate Γ diff-p . 4,5 The regular method is described as the Fuchs-Sutugin 62 Equation: 63 Eq.4 69 above first-order reaction kinetic k p in Equation 2 should be modified as Equation 6, taking the first-order wall loss of the 70 particles into account, that is the integrated result for each size of particle: Where Ni is number concentration (cm -3 ) for particle of size Di (nm), k wall-p is first-order wall loss rate (s -1 ) for the particles.

73
In our experiments, the wall loss rate for wood tar particles through the AFR can be neglected according to the CPC and 74 SMPS measurements, thus, Equation 6 can be simplified: Γ diff-w describes wall loss of O 3 , NO 3 •, NO 2 , and N 2 O 5 to the AFR, as suggested in Equation 8: For the case where the loss rates of the gases to the reactor wall is not determined by surface reactivity, but by the 79 diffusion through the gas phase (γ w > Γ wall ~7×10 -6 ), the following expression holds: Equation 9 is valid for Peclet numbers in excess of ~20. 6 Figure S2. We did not observe significant size distribution change for wood tar aerosol due to O 3 oxidation, while the surface 96 and volume concentration for the wood tar aerosols increased by 24% and 31%, respectively, after NO 3 • aging. The results 97 demonstrate secondary aerosol formation from NO 3 • reactions, as a result, the equation 2 and 7 need to be modified surface area can be described by Equation 10: Where S 0 indicates the initial particle surface area concentration (unit: nm 2 cm -3 ), k s (unit: s -1 ) simply describes the growth 103 rate of particle surface area. Given that the surface area concentration increased by 24.4% in residence time of 298.2 s, k s 104 was estimated to be 4.17×10 -3 s -1 . The growth of particle surface area implies changes in the reactive surface uptake rates 105 (k p in Equation 7) in the AFR. To make the model feasible, a particle surface area concentration weighted surface uptake 106 rate was proposed, the first-order kinetic of surface uptake in Equation 7 was then modified as pseud-second-order kinetic 107 ( ) in following Equation 11: We found that was not sensitive to the minor size changes of the wood tar particles in reactions, suggesting that we , 111 can keep constant in all reactions. ,

112
The gas loss to the particles and the reactor wall can be described by Equation 12: 114 According to the above equations and parameters summarized from the literature, the first-order wall loss rate of the 115 gaseous species and second-order surface uptake rates of these gaseous oxidants are presented in Table S2.
116  Figure S2. Size distribution changes for wood tar aerosol after NO 3 • heterogeneous reactions in presence of O 3 and NOx (error bar 120 was not presented for clarity).

121
Acetonitrile was used as a solvent to dissolve the wood tar in the atomizer. Although sufficient charcoal denuders were 122 applied downstream of the aerosolization to remove the outgas, residual acetonitrile in that gas phase may still be involved 123 in competitive reactions with NO 3 • in the AFR. We assumed that each denuder has filtration efficiency of 80%, the residual 124 gaseous acetonitrile in the AFR can get to ~5.26×10 13 molec cm -3 (296 K and 1 atm). 12,13 Applying the above simplified 125 kinetic parameters, a chemical box model including gaseous NO 3 • formation and gaseous oxidants uptake by wood tar 126 particles and reactor wall was initialized using the COPASI software (complex pathway simulator, http://copasi.org/), and 127 the result is displayed in Figure S3.

133
Based on chemical box model simulation, it was found that NO 3 • uptake by wood tar particle was about one magnitude 134 order higher than that of N 2 O 5 and O 3 , demonstrating that the NO 3 • reaction is the dominate reaction pathway to oxidize 135 wood tar aerosol. To prove the practical significance of the result, wood tar aerosol aging due to NO 3 • in the laboratory was 136 compared with outdoor nighttime aging of smoke aerosol by NO 3 • in field plume, and the aging content was quantified to 137 equivalent aging time at night (EAN, unit: hour) for ambient biomass burning aerosols. The detailed quantification method can be found in our previous study. 1 The surface uptake of NO 3 • and N 2 O 5 was first normalized to the wood tar particle 139 surface area concentration: of NO 3 • compared to particle phase uptake, and sink of NO 3 • due to homogeneous reactions can be a factor of 100-1000 152 greater than that by particle surface uptake. But, considering the rapid aging and dilution of VOCs and also smoke particle 153 growth due to condensation and coagulation during biomass burning emissions transport, the NO 3 • reactivity due to surface 154 uptake should weigh more in the total reactivity with fire plume transporting. Here, we followed the same method proposed 155 in our previous study (SI, Text S6.2) 1 and assumed a median and constant ratio of 500 for total NO 3 • reactivity to smoke 156 particle uptake. That is: Eq.17 159 Based on Equation 17, the NO 3 • sink rate to the particles can be estimated, and it was then normalized to the surface area 160 concentration as below: Where S aerosol is field smoke particle surface area concentration in biomass burning plume. 166 In this study, the typical ambient NO 2 , O 3 , and smoke particle surface density were taken as 25 ppbv, 35 ppbv, and 2×10 8 167 nm 2 cm -3 , respectively, referring to reference and also our previous work. 1 From the results of chemical box model simulation in Figure S2, NO 3 • exposure was estimated as (2.74±0.62)×10 15 molec 172 cm -3 s. In the field, NO 3 • concentration has large variation from several to hundreds pptv level, and its concentration depends 173 on the air pollution and environment conditions. [17][18][19] In field fire plume, although the NO 3 • can be ultralow, it has a rapid 174 formation rate up to ppbv h -1 level. 14 Referring to a previous study on ambient nitrate radical chemistry as summarized in 175  studies. [31][32][33] In this estimation, we applied the absorption cross sections for both gaseous and particle-bound, specifically 210 azelaic acid particle-adsorbed, 2-nitrophenol from Hinrichs et al. 32 There is a lack of comprehensive gaseous or particle-  Figure S4. phases. 33,[38][39][40][41] But the HONO and OH quantum yields from the photodissociation of 2-nitrophenol and the variation of the 222 upon irradiation by 320-450 nm light (λ max =370nm). 31 Sangwan and Zhu measured the OH quantum yield to be 0.69 at 308 223 nm and 0.70 at 351 nm, while the HONO quantum yields from laser-induced 2-nitrophenil photolysis to be 0.34 at 308 nm 224 and 0.39 at 351nm. 33 The HCO radical formation and NO 2 formation represents two vital channels occur in photolysis of 2-

225
NBA, and 0.5 was recommended as the photolysis quantum yield in photodissociation of 2-NBA in gas, aqueous, and solid 226 states by the near-UV irradiation. 34 Albinet et al. measured an ultralow photolysis rate for 2,4-DNP in solution that was 227 irradiated by 300-500 nm Xenon lamp, the calculated quantum yield was influenced by the acidity of the solution, and it 228 was in the range of (3.4-8.1)×10 -5 . 42 The photolysis quantum yield for 4-NC and 4-NG is not available, we applied the 229 quantum yield of 2-NP for these analogous nitrophenols. Moreover, we assumed these nitroaromatic proxies have high 230 quantum yield around 250 nm, thereby, unity of photolysis quantum yield was assigned for the 250-280/290nm range. The 231 quantum yields for these proxies and applied in this modeling were summarized in Table S4.

232
The final equivalent ambient photolysis times corresponding to the high UV exposure (9.30×10 16 photon cm -2 ) in the PAM-

233
OFR were calculated and summarized in Table S5         We assumed a time-dependent exponential decay for particulate nitrate and particle absorption. Based on the 262 fitted first-order kinetic (k, s -1 ), the lifetime for particulate nitrate and absorption was estimated. See method 263 below:

264
Eq.24 C t and C 0 indicate time-dependent and initial concentration of a parameter. In this study, C t is particle absorption 266 coefficient (C abs_t , Mm -1 ) and particulate organic nitrate (C ON_t , μg m -3 ).
Where V t (μm 3 cm -3 ) and ρ t (g cm -3 ) are wood tar aerosol volume concentration and particle effective density 270 measured by the SMPS and AAC (results in Figure S6 and The fitting results for decay of particulate nitrate and absorption are presented in Figure S7:

302
(MAC, m 2 g -1 ) was calculated from imaginary RIs and particle density. Here we presented wavelength-weighted MAC for convenience that were grouped as UV 303 range (330-400nm) and visible range (400-550nm). AAE was fitted over the measured effective wavelength range of 330-550 nm.