1 Rising Importance of Organosulfur Species for Aerosol Properties and Future 2 Air Quality
3 M. Riva1,#,¥,*, Y. Chen1,¥, Y. Zhang1,2, Z. Lei3, N. E. Olson4, H. C. Boyer Chelmo5, S. Narayan5, 4 L. D. Yee6, H. S. Green1,‡, T. Cui1, Z. Zhang1, K. Baumann7, M. Fort7, E. Edgerton7, S. H. 5 Budisulistiorini1,†, C. A. Rose1, I. O. Ribeiro8, R. L. e Oliveira8, E. O. dos Santos9, C. M. D. 6 Machado9, S. Szopa10, Y. Zhao11,§, E. G. Alves12, S. S. de Sá13, W. Hu14, E. M. Knipping15, S. L. 7 Shaw16, S. Duvoisin Junior8, R. A. F. de Souza8, B.B. Palm,14 J. L. Jimenez14, M. Glasius17, A. 8 H. Goldstein6, H. O. T. Pye1,18, A. Gold1, B. J. Turpin1, W. Vizuete1, S. T. Martin13,19, J. A. 9 Thornton10, C. S. Dutcher5, A. P. Ault3,4*, and J. D. Surratt1*
10 Affiliations: 11 1 Department of Environmental Sciences and Engineering, Gillings School of Global Public 12 Health, The University of North Carolina at Chapel Hill, Chapel Hill, NC, USA. 13 2 Aerodyne Research Inc., Billerica, MA, USA. 14 3 Department of Environmental Health Sciences, University of Michigan, Ann Arbor, MI, USA. 15 4 Department of Chemistry, University of Michigan, Ann Arbor, MI, USA. 16 5 Department of Mechanical Engineering, University of Minnesota-Twin Cities, Minneapolis, 17 MN, USA. 18 6 Department of Environmental Science, Policy, and Management, University of California, 19 Berkeley, CA, USA. 20 7 Atmospheric Research & Analysis, Inc., Cary, NC, USA. 21 8 Escola Superior de Tecnologia, Universidade do Estado do Amazonas, Manaus, Amazonas, 22 Brasil. 23 9 Department of Chemistry, Federal University of Amazonas, Manaus, Amazonas, Brazil. 24 10 Laboratoire des Sciences du Climat et de l'Environnement, CEA-CNRS-UVSQ-IPSL, Gif-sur- 25 Yvette, France. 26 11 Department of Atmospheric Sciences, University of Washington, Seattle, WA, USA. 27 12 Environment Dynamics Department, National Institute of Amazonian Research (INPA), 28 Manaus, Brazil. 29 13 John A. Paulson School of Engineering and Applied Sciences, Harvard University, 30 Cambridge, MA, USA. 31 14 Department of Chemistry and Cooperative Institute for Research in Environmental Sciences, 32 University of Colorado, Boulder, CO, USA. 33 15 Electric Power Research Institute, Washington, D.C, USA. 34 16 Electric Power Research Institute, Palo Alto, CA, USA. 35 17 Aarhus University, Dept. of Chemistry and iNANO, 8000 Aarhus C, Denmark.
1 36 18 National Exposure Research Laboratory, US Environmental Protection Agency, Research 37 Triangle Park, NC, USA. 38 19 Department of Earth and Planetary Sciences, Harvard University, Cambridge, MA, USA. 39 40 # Now at the Univ Lyon, Université Claude Bernard Lyon 1, CNRS, IRCELYON, F-69626, 41 Villeurbanne, France. 42 43 ‡ Now at Department of Food Science and Technology, University of California, Davis, Davis, 44 CA, USA. 45 46 † Now at Earth Observatory of Singapore, Nanyang Technological University, Singapore 47 639798, Singapore. 48 49 § Now at School of Environmental Science and Engineering, Shanghai Jiao Tong University, 50 Shanghai, 200240, China. 51 52 ¥ These authors contributed equally to this work 53 54 *E-mail (M. R.): [email protected] 55 *E-mail (A. P. A): [email protected] 56 *E-mail (J. D. S): [email protected]
2 57 Abstract
58 Acid-driven multiphase chemistry of isoprene epoxydiols (IEPOX), a key isoprene oxidation
59 product, with inorganic sulfate aerosol yields substantial amounts of secondary organic aerosol
60 (SOA) through the formation of organosulfur. The extent and implications of inorganic-to-
61 organic sulfate conversion, however, are unknown. Herein, we reveal that extensive consumption
62 of inorganic sulfate occurs, which increases with the IEPOX-to-inorganic sulfate ratio (IEPOX:
63 Sulfinorg), as determined by laboratory and field measurements. We further demonstrate that
64 organosulfur greatly modifies critical aerosol properties, such as acidity, morphology, viscosity,
65 and phase state. These new mechanistic insights reveal that changes in SO2 emissions, especially
66 in isoprene-dominated environments, will significantly alter biogenic SOA physicochemical
67 properties. Consequently, IEPOX:Sulfinorg will play a central role in understanding historical
68 climate and determining future impacts of biogenic SOA on global climate and air quality.
3 69 Secondary organic aerosol (SOA) formed through the oxidation of volatile organic compounds is
70 a major and globally ubiquitous component of atmospheric fine particulate matter (PM2.5: aerosol
71 particles ≤ 2.5 μm in aerodynamic diameter).1 Aerosol chemical composition and
72 physicochemical properties, such as viscosity and phase state, play a central role in terms of the
73 effects of SOA on air quality and climate.2 Understanding how SOA forms and interacts with
74 other gas- and particle-phase species is crucial to accurately evaluating its importance in the
75 Earth’s climate system and adverse effects on public health.
2- - 76 Inorganic sulfate species (e.g., SO4 , HSO4 ) are also a significant PM2.5 component with
77 the capacity to impact atmospheric composition and climate, in part, because of its predicted
78 impact on aerosol acidity, hygroscopicity, visibility and cloud nucleation.1,3 The oxidation of
79 sulfur dioxide (SO2) to sulfuric acid (H2SO4) increases aerosol acidity, which enhances SOA
80 formation.4–7 Sulfur (S(VI)) in aerosols was generally assumed to be primarily present as
2- - 81 inorganic sulfate (SO4 and HSO4 ) ions until more recent studies revealed the presence of
8–12 82 organosulfur components in PM2.5. Despite observations that organosulfur compounds are
83 important contributors to SOA mass in a range of environments globally,13–18 estimations of
84 aerosol acidity and liquid water content typically assume that only inorganic sulfate plays a
85 role.19 Correctly identifying the chemical form of sulfate (i.e., inorganic vs. organic), and
86 representing it accurately in atmospheric models is essential as the different forms lead to
87 different aerosol physicochemical properties that will have different predicted impacts on air
88 quality and climate.2
89 Laboratory studies have demonstrated that acid-driven multiphase chemistry (i.e.,
90 reactive uptake) of isoprene epoxydiols (IEPOX) is key to explaining the chemical form and
91 extent of SOA formation from photochemical oxidation of isoprene 20–23 and field measurements
4 92 have confirmed this is the predominant pathway.24–29 While chamber studies have shown that
93 organosulfur compounds, specifically organosulfates (OS) 30 formed by the reactive uptake of
94 IEPOX with particulate inorganic sulfate, contribute significantly to IEPOX-SOA,31,32 the extent
95 and implications of sulfate conversion to organic forms have remained unknown.
96 Combined laboratory, field, and modeling studies described herein reveal a hitherto
97 unrecognized impact of the acid-driven multiphase chemistry of IEPOX, specifically substantial
98 conversion of inorganic sulfate to organosulfur compounds. Laboratory experiments show for
99 higher IEPOX:Sulfinorg ratios (e.g., Amazon) increase sulfate conversion versus lower ratios (e.g.,
100 southeastern U.S. (SE-U.S.)). Future sulfate reductions in the Northern Hemisphere are expected
101 to greatly increase the fraction of inorganic sulfate converted. This was likely also the case for
102 pre-industrial conditions, when inorganic sulfate was much lower. In sum, these major changes
103 in the SOA chemical composition due to IEPOX reactive uptake govern aerosol physicochemical
104 properties.
105 Uncharted IEPOX conversion of inorganic sulfate to organosulfur. Despite the wealth of
106 studies on the reactive uptake of IEPOX, its reactivity remains poorly constrained.22,23,26,31,33,34
107 We performed controlled chamber experiments in the presence of ammonium bisulfate (ABS)
108 seed particles (pH = 1.5) at ~50 % relative humidity (RH) using atmospherically-relevant ratios
109 of IEPOX:Sulfinorg (Table S1). Fig. 1 shows that immediately following IEPOX addition, rapid
110 conversion of inorganic sulfate is observed under all conditions measured by a particle-into-
111 liquid sampler (PILS) coupled to an ion chromatograph (IC) with 5 minute resolution. This
112 depletion is correlated to IEPOX-OS and oligomeric-OS (quantified by liquid chromatography
113 coupled to electrospray ionization high-resolution mass spectrometry (LC/ESI-HR-MS) from the
114 same PILS samples), which is supported by computational modeling.35 IEPOX-OS accounts for
5 115 most (90-100%) of the converted sulfate within the first 40-60 min under conditions that mimic
116 IEPOX:Sulfinorg ratios relevant to both the SE-U.S. (Fig. 1A) and the Amazon (Fig. 1B). As
117 shown in Fig. 1A and B ~40% of inorganic sulfate injected into the chamber is converted to
118 organosulfur under SE-U.S. conditions, while up to 90% is converted to organosulfur under
119 Amazonian conditions. One hour following IEPOX injection, stabilization of inorganic sulfate
120 commences, indicating that IEPOX uptake is inhibited due to various reasons described further
121 below. One possibility is the presence of organic coating as suggested in recent studies.32,36
122 Meanwhile, the concentrations of IEPOX-OS start to decrease in all experiments. A net
123 reduction (up to 30% in one hour) of the three quantified OS species indicates that IEPOX-OS
124 are not stable and react to yield as yet uncharacterized organosulfur compounds. One potential
125 class of species, sulfur-containing oligomers, were observed below quantifiable levels in the
126 positive ion mode.
127 Interestingly the conversion of inorganic-to-organic sulfate appears to be mainly driven
128 by the IEPOX:Sulfinorg ratio as illustrated in Fig. 1C. Indeed, our results clearly demonstrate that
129 the conversion fraction of inorganic to organosulfur correlate with the initial concentrations of
130 IEPOX and inorganic sulfate. Hence, IEPOX:Sulfinorg ratio is a critical and previously uncharted
131 factor in the conversion of inorganic-to-organic sulfate. High IEPOX:Sulfinorg ratios (>2) are
132 common over large geographic areas globally, as shown in Fig. 1D, especially in equatorial
133 regions and the Southern Hemisphere. Based on our experimental results, we infer that a higher
134 S(VI) fraction is likely in organic forms in those regions, which are undersampled relative to the
135 Northern Hemisphere. Overall, our chamber experiments reveal the central role of IEPOX in the
136 speciation and fate of inorganic sulfate in the atmosphere.
6 137 Concentration of ambient organosulfur. Over the past 30 years, air pollution regulations in the
-1 138 Northern Hemisphere have led to decreases in SO2 concentration of 5.7 % yr in the U.S. and
-1 37 2- -1 139 5.1 % yr in Europe. Recent studies project a reduction of SO4 of ~4.5% yr based on current
38,39 2- 140 efforts. Fig 2A illustrates this trend and presents the SO4 concentration over the last 10
141 years measured at the Great Smoky Mountain Site (Look Rock, TN)40 which is strongly
142 influenced by isoprene chemistry.25 A decrease of inorganic sulfate by a factor of 6 is correlated
143 with a subsequent rise in organosulfur by a factor of 5. In addition, events with high organosulfur
144 fractions tend to correlate with low inorganic sulfate concentrations (Fig. 2B), which is
145 consistent with a previous analysis11 and corroborate our laboratory findings. The increase in the
146 IEPOX:Sulfinorg ratios within the next decades is predicted, using our laboratory experiments, to
147 significantly shift the distribution of inorganic sulfate towards organosulfur (Fig. 1C). Since the
148 pre-industrial period, inorganic sulfate has significantly increased41,42 while the concentration of
43 149 isoprene has remained constant. This trend indicates that IEPOX:Sulfinorg was likely much
150 higher during the pre-industrial period than currently, and possibly had a major role in SOA
151 physicochemical properties in isoprene-dominated areas. While the estimated conversion from
152 chamber studies is an upper limit as the reactive uptake of IEPOX is sensitive to other
153 environmental factors (e.g. NOx, organic coating, RH, etc.), this underappreciated trend is central
154 to understanding future air quality and radiative forcing.
155 Molecular characterization of SOA samples collected in isoprene-rich areas show that OS
156 alone contribute to a substantial mass fraction of S(VI) in PM: on average 8% (up to 25%) in the
157 SE-U.S. and 20% (up to 45%) in downwind Manaus (Figs. 2C and S1). Indeed, OS masses range
158 from 400-1500 ng m-3 (Tables S2-S4), at the high end of previous studies.24,25,30,44,45 While the
159 sum of OS in the SE-U.S. is significantly different from that of downwind Manaus, isoprene-OS
7 160 nevertheless represent the predominant OS in both areas. During the 2013 Southern Oxidant and
161 Aerosol Study (SOAS) campaign, total mass concentrations of organosulfur compounds were
162 also determined by isotope ratio inductively coupled plasma mass spectrometry ((IR-ICP-MS);
163 Fig. S1). The average total sulfur mass and the total inorganic sulfate measured by IR-ICP-MS
-3 164 and IC respectively, differs by 300 ± 200 ngSO4 m . This is an estimate of total organosulfur
165 and provides an organosulfur/total sulfate ratio similar to that previously observed.11,12,46
166 Although total OS + MSA exhibits a strong correlation (r2 = 0.78) with the total mass
167 concentration of organosulfur compounds (Fig. S2), the identified products explain only 50-60%
168 of the total mass of organosulfur quantified in PM2.5 during the 2013 SOAS field campaign. In
169 other words, 16% of the inorganic sulfate is converted into organosulfur (Fig. S2) in excellent
170 agreement with the chamber experiments presented here. Using the fit obtained from all chamber
171 experiments (Fig 1C) and estimating an IEPOX:Sulfinorg ratio of 0.24 based on collocated
172 measurements during the field campaign,47 inorganic sulfate conversion to organosulfur is
173 estimated to be ~15% .
174 Atmospheric impact of the acid-driven reactive uptake of IEPOX. Herein, we demonstrate
175 that high levels of inorganic-to-organic sulfate conversion will significantly alter surface tension,
176 aerosol acidity, morphology, viscosity, and reactivity. These altered properties can then modify
177 climate-relevant aerosol properties, such as enhancing CCN activity due to reduced surface
178 tension.48
179 Because acidity is one of the governing factors of atmospheric multiphase chemistry,5 the
180 response of condensed-phase acidity to inorganic-to-organic sulfate conversion was further
181 investigated. The widely used thermodynamic models assume sulfate only as inorganic S(VI),
182 leading to inconsistencies in charge balance in the condensed phase when a substantial
8 183 proportion of OS are assumed to be inorganic.49,50 While organic compounds can reduce the rate
184 of ammonia-to-particle partitioning, exclusion of organosulfur from thermodynamic models
185 within aerosol may contribute to an even larger discrepancy in acidity. As an example, Fig. S3
186 shows the pH of aerosols when not considering the contributions of IEPOX-OS to acidity from
187 the chamber experiments, compared with a thermodynamic box model constructed to take
188 IEPOX-OS into consideration. The results reveal that a large increase of pH occurs if OS are not
189 considered, suggesting that aerosol pH is higher than currently predicted from measured
190 inorganic sulfate in locations where inorganic-to-organic sulfate conversion is substantial (e.g.,
191 Amazon, future S-E U.S.). In cases where total sulfate (measured by ACSM or AMS) is treated
192 as inorganic sulfate and input to thermodynamic models, aerosol pH could be underpredicted.
193 Acid-driven multiphase chemistry of IEPOX also leads to a modification of the aerosol
194 morphology from a well-mixed sphere to a core-shell structure, shown by phase images from
195 atomic force microscopy (AFM) at ambient temperature and pressure (Fig. 3A-C), as well as
196 scanning electron microscopy (SEM) images (Fig. S4) of SOA collected from chamber
197 experiments. These results are consistent with previous theoretical and semi-empirical studies
51,52 198 predicting frequent phase separation for particles in the SE-U.S. Larger IEPOX:Sulfinorg
199 ratios lead to thicker organic shells (e.g., Amazon) and can be correlated with the relatively
200 larger amounts of particulate IEPOX-OS and corresponding oligomers (sum of OS corresponds
201 to 27% of total inorganic sulfate in Fig 1A and 69% in 1B). Hence, OS and oligomeric-OS may
202 lead to a net modification of the morphology of the OA.
203 Over the course of the experiment, as inorganic sulfate is consumed and OS are formed,
204 the particle phase becomes more viscous and particle heights increase (Fig. 3D). Height images
205 and profiles from AFM were used as a proxy for liquid, semi-solid, or solid phase of the organic
9 206 component, which is largely composed of inorganic sulfate for the seed and organosulfur for the
207 SOA. The observation of more viscous aerosol is supported by model simulations of molecular
208 tracer viscosity based on recent studies,2,53 which suggests that IEPOX-OS has viscosity values
209 of 1-4 orders of magnitudes higher than α-pinene SOA when the RH levels are lower than 70%
210 (Fig. 3E). Hence, IEPOX-OS can significantly increase the viscosity of aerosol particles,
211 compared to isoprene-SOA generated in atmospheric simulation chambers from self-nucleation
212 under high concentrations without the addition of acidic sulfate particles.54
213 IEPOX-SOA exhibits much lower volatility than structure-based vapor pressures of
214 polyols would predict,55 which is likely explained by the large inorganic-to-organic sulfate
215 conversion. Given the high modelled viscosity of IEPOX-OS and likely higher viscosity values
216 of oligomeric-OS, the OS coating would have a stronger reduction effect on multiphase
217 chemistry compared to coatings with α-pinene SOA.36 It has been reported that SOA particles
218 sampled in the Amazon rainforest were liquid as the RH was higher than 80%, suggesting no
219 diffusion limitation on multiphase chemistry at RH>80%.56 Even though the average RH is close
220 to 80% for both the Amazon and SE-U.S., Fig. S5 and previous measurements show that the
221 median daytime RH in those regions is consistently lower than 70% (40-50% during the dry
222 season in the Amazon), concordant with the hours of maximum diurnal IEPOX-SOA
223 production.36,57 Hence, limited diffusion due to IEPOX-derived OS will be more likely to affect
224 multiphase chemistry during daytime in isoprene-rich environments, reducing heterogeneous
225 SOA formation from compounds that are generated by photooxidation reactions.
226 The change in morphology is further supported by measuring the interfacial tension (IFT)
227 depression of the major IEPOX-SOA products in microfluidic platform experiments. Surface
228 tension and IFT are proxy measures of surface concentrations, as tensions will decrease with
10 229 increased bulk-to-surface partitioning of surface-active components in the aqueous droplets. IFT
230 depression was observed in all cases (Fig. 4A and S6), and IFT was lower when ammonium
231 sulfate (AS) was present. IEPOX-OS has a propensity for salting out in the presence of AS,
232 which helps to explain the dependence of shell thickness on the ratio of IEPOX:Sulfinorg. In fact,
233 lowering of the IFT in salty solutions indicates potential salting out58,59 of the organic due to an
234 enhancement of organic activity, driving more organic molecules to the surface. In turn, the
235 salting out effect subsequently alters the SOA physicochemical properties by changing
236 morphology to a core-shell structure rather than a homogeneously mixed particle36 Salting out
237 effects are quantified by combining the Setschenow equation and a two parameter surface and
60 238 IFT model. The setschenow constant (Ks) indicates salting out when positive for organic-
239 inorganic aqueous systems. Fig. 4A shows model treatment for IEPOX-OS in pure water and in
240 salty water and reveals that IEPOX-OS has a propensity for salting out. The reduction of surface
241 tension by IEPOX-SOA may alter the climate properties of aerosols by suppressing the surface
242 tension to enhance cloud droplet formation from aerosols in organosulfur-rich particles, causing
243 larger droplets to form before and during cloud activation.48
244 Fig 4B illustrates the limiting effects of IEPOX-SOA products on aerosol reactivity due
245 to the slight decrease in acidity, the core-shell morphology, and the salting out effect induced by
246 organosulfur compounds. A dramatic decrease of the reactive uptake coefficient of IEPOX
247 (γIEPOX) was observed as atmospheric-equivalent exposure time increases, highlighting the
248 profound modification of IEPOX-SOA on aerosol reactivity, explains the presence of IEPOX in
249 the gas phase (Fig. S7), and the stabilization of inorganic sulfate (Fig. 1A and B) in our
250 laboratory experiments. While previous studies have shown that other organic coatings tend to
251 reduce such multiphase chemical processes,23,32,36 the results presented in Fig. 4B provide direct
11 252 evidence that IEPOX-SOA has a self-limiting effect and formation of an IEPOX-SOA coating
253 prevents further SOA formation mainly during daytime, when RH is lower and aerosol is phase
254 separated with a more viscous shell.
255 In sum, this study demonstrates that acid-driven multiphase chemistry of IEPOX converts
256 a significant fraction of inorganic sulfate to organosulfur within a range of IEPOX:Sulfinorg ratios
257 relevant to most isoprene-dominated environments. We further demonstrate through laboratory
258 and field measurements the substantial conversion of sulfate, underlining the major role of
259 IEPOX in controlling the chemical form of S(VI). Retrospective examination of field data in the
260 SE-U.S. consistently shows that the contribution of organosulfur to S(VI) has been increasing
261 with declining inorganic sulfate. IEPOX reactive uptake results in a core-shell morphological
262 configuration confirmed by microscopic imaging. Through chemical aging, IEPOX-OS
263 transforms into uncharacterized organosulfur compounds, impeding further reactive uptake of
264 IEPOX. In isoprene-dominated areas, IEPOX-OS, and potentially other biogenic/anthropogenic
265 OS, likely govern the physicochemical properties of aerosol as well as the distribution of the
266 inorganic species such as sulfate or ammonium. Consequently, aerosol growth, multiphase
267 reactions, including aging and reactive uptake of other species, and CCN activity changes occur
268 due to surface tension, acidity, hygroscopicity and viscosity modifications. These changes could
269 greatly impact atmospheric composition, air quality and associated health impacts, and the net
270 climate forcing of biogenic SOA formed over isoprene-dominated areas. Hence, changes in SO2
271 emissions at different locations around the world and time periods have major implications for
272 the physicochemical properties of atmospheric fine aerosols and, ultimately, their global
273 radiative forcing.
274
275 Methods
12 276 Smog Chamber Experiments. Experiments were performed in the indoor environmental chamber facility
277 at the University of North Carolina.31,61 The experimental setup and analysis techniques used in this work
278 were described in detail previously. Briefly, experiments were carried out under dark and wet conditions
279 (50 ± 4 %, RH) at 296 ± 1 K in a 10-m3 Teflon chamber. A summary of the experimental conditions is
280 provided in Table S1. Prior to each experiment, the chamber was flushed continuously with clean air for ~
281 24 hours corresponding to a minimum of seven chamber volumes until the particle mass concentration
282 was < 0.01 μg m-3 to ensure that there were no pre-existing aerosol particles. Chamber flushing also
283 reduced VOC concentrations below the detection limit. Temperature and RH in the chamber were
284 continuously monitored using a dew point meter (Omega Engineering Inc.). Acidified ammonium sulfate
285 seed aerosols were injected into the pre-humidified chamber using a custom-built atomizer with an
286 aqueous solution of 0.06 M (NH4)2SO4 (aq) and 0.06 M H2SO4 (aq) until desired total aerosol mass
287 concentration was achieved. After seed injection, the chamber was left static for at least 30 min to ensure
288 that the seed aerosol was stable and uniformly mixed. Then, a known amount of trans-β-IEPOX,
289 synthesized in house,62 was injected into the chamber at 2 L min−1 for 10 min and then 4 L min-1 for 50
290 min by passing high-purity N2 (g) through a heated manifold (60°C).
291 Aerosol size distributions were continuously measured using a differential mobility analyzer
292 (DMA, BMI model 2002) coupled to a mixing condensation particle counter (MCPC, BMI model 1710)
293 in order to monitor aerosol number, surface area, and volume concentration within the chamber. SOA
294 generated from the reactive uptake of IEPOX were collected using a PILS (BMI model 4001) throughout
295 each experiment. The aerosols are sampled through a 2.5 μm cut size impactor at a flow rate of
296 approximately ~ 13 L min-1 with a carbon strip denuder (Sunset Labs) upstream of the impactor to remove
297 organic vapor. Cool sample air flow was mixed adiabatically with a steam flow heated at 98.5-100℃ in
298 the PILS condensation chamber, which allows aerosol particles to grow into collectable size for collection
299 by a quartz impactor plate. Impacted droplets were transferred by a wash-flow at 0.50 – 0.55 mL min-1
300 into a debubbler and the resulting bubble-free sample liquid was delivered through a tubing with an inline
13 301 filter into 2 mL glass vials held on an auto-collector (BMI) with a rotating carousel. Air sampling rate and
302 wash-flow rate were checked and recorded before and after each experiment. Milli-Q water used in the
303 wash-flow was spiked with 25 µM of lithium bromide (LiBr) as an internal standard to correct for dilution
304 caused by condensation of water vapor during droplet collection inside the PILS (information regarding
305 the determination of the dilution factor for samples collected by the PILS are provided in the
306 supplementary information). Under the configuration used in this work, a fine time resolution (up to 5
307 min) was used to collect PILS samples for the subsequent offline chemical analyses by IC and
308 UPLC/ESI-HR-QTOF-MS. The PILS vials were promptly stored under dark conditions at 2ºC after
309 collection and analyzed within 24 h without further pretreatment.
310 Flow Tube Experiments. Measurements of the uptake coefficient of IEPOX (γIEPOX) on aqueous
311 ammonium bisulfate particles shown in Fig. 4B were conducted in an aerosol flow tube coupled to an
312 iodide-adduct chemical ionization mass spectrometry (CIMS), operation conditions have been described
313 previously.63,64 The flow tube, CIMS, as well as aerosol and gas phase IEPOX generation have been
314 described in detail previously.23 Briefly, the flow tube consists of a 6 cm i.d. and 90 cm long Pyrex
315 cylinder having inner walls coated with halocarbon wax to reduce the wall loss of IEPOX gas.
316 Ammonium bisulfate aerosols were generated using a constant output atomizer (TSI Inc., Model 3076)
317 from dilute solutions (0.1 wt %). The atomizer output was diluted and conditioned to about 38% RH by
-1 318 mixing with a 3 L min of humidified ultrahigh purity (UHP) N2 flow before entering the flow tube. The
319 IEPOX gas was generated by flowing a 30 standard cubic centimeters per minute (sccm) of UHP N2 over
320 ~200 μl trans-β-IEPOX solution in ethyl acetate kept in a glass bulb at room temperature (~23 °C), and
321 was added to the flow tube via a movable injector downstream of the aerosol flow inlet. A constant 2 L
322 min-1 of conditioned aerosol flow was drawn through the flow tube by the CIMS and a scanning mobility
323 particle sizer (SMPS) connected to the flow tube exit, which provided real-time measurements of the gas-
324 phase IEPOX concentration and particle size distribution, respectively IEPOX injector was moved in 10
325 cm increments from the bottom 10 cm to the top 70 cm of the flow tube to vary the reaction time between
14 326 IEPOX and aerosol particles. Control experiments without adding particles to the flow tube were
327 performed to determine the effects of reactor walls. The decay of gas phase IEPOX signal versus the
328 injector position (reaction time) in the presence and absence of particles was measured to derive a pseudo-
329 first-order rate constant, kobs and kwall, respectively, for the loss of gas-phase IEPOX in the flow tube. The
330 pseudo-first-order reaction rate constant for IEPOX uptake onto particles, khet, is then the difference
331 between kobs and kwall. The IEPOX signals derived at every three adjacent injector positions (i.e., 10-30,
332 20-40, 30-50, 40-60, 50-70 cm), which correspond to different average reaction times, were used to
333 determine the khet and then γIEPOX as a function of reaction time. Additional information on the data
334 fittings are provided in supplementary information.
335 Measurements of γIEPOX shown in Fig. S8 were conducted using a cylindrical glass flow reactor (1
336 m in length with 8 cm ID) coated also with halocarbon wax (Halocarbon Products Corp.). Aerosols were
337 generated with a constant output atomizer (TSI Inc., Model 3076) from 0.06 M of ammonium sulfate
338 (Sigma Aldrich, ≥99% purity) mixed with an equivalent concentration of sulfuric acid (Sigma Aldrich,
339 ≥98% purity). A flow of 40 psi of purified air was sent through the atomizer to generate a constant aerosol
340 flow of 3 L min-1. Atomized acidic sulfate particles first passed through a Nafion tube (Perma Pure,
341 Model PD-200T-12) in order to be dried to 15% RH before entering into an electrostatic classifier (TSI
342 Inc., Mode 3080L). In the electrostatic classifier and the differential mobility sizer (DMA; TSI Inc.,
343 Model 3080), the sheath flow was selected to be 12 L min-1 and three aerosol mobility diameters were
344 selected, i.e., 60 nm, 100 nm and 200 nm, by varying the voltage inside the DMA column. These particles
345 were then injected at ~ 5.2 L min-1 into the two side inlets at the top of the flow tube and perpendicular to
346 the flow axis. Trans-β-IEPOX (>99%) was delivered to the flow reactor by passing ~0.32 L min-1 of high-
-1 347 purity N2 (g) into a 10 mg mL IEPOX solution dissolved in ethyl acetate. IEPOX was introduced into
348 the aerosol stream through a 1/8-inch stainless steel tube that can be moved from the top to the bottom of
349 the flow tube reactor, altering interaction times between trans-β-IEPOX and the aerosol particles. At the
350 base of the reactor, aerosol flow exited the flow tube in order to measure real-time aerosol size
15 351 distributions and total surface area concentrations (Sa) by using a SMPS consisting of an electrostatic
352 classifier (TSI, Inc., Model 3080) and a condensation particle counter (CPC, TSI, Inc., Model 3022A).
353 The outlet of the flow tube was also coupled to an iodide-CIMS and a RH-temperature sensor (Omega
354 Engineering Inc.) recording the RH every 5 s.
355 Collection of PM2.5 samples. Ambient SOA from two isoprene-dominated environments (SE-U.S. and
356 Amazon forest) were collected onto quartz filters during 3 intensive campaigns: (i) during the 2013
357 Southern Oxidant and Aerosol Study (SOAS) campaign from 1 June to 15 July 2013 at the CTR, AL
358 ground site; (ii) from 18 July through 1 August 2016 from downtown Manaus, Brazil and (iii) during the
359 Green Ocean Amazon (GoAmazon2014/5) field campaign65 during intensive operating period 2 (IOP2).
360 Additional information regarding filter collections are provided in supplementary information.
361 Aerosol Chemical Characterization. Chemical characterization of the PM2.5 samples and the PILS vials
362 was performed by UPLC/ESI-HR-Q-TOFMS (6520 Series, Agilent) operated in the negative ion
363 mode.33,66 Total water-soluble organosulfur compound mass was determined from the difference between
364 total water-soluble sulfur measured by IR-ICP-MS and sulfate-sulfur measured by IC on the same sample
365 aliquot Samples extractions, operating conditions, standard preparation, uncertainty estimates are
366 described in detail in supplementary information.
367 Microscopy Imaging. Aerosol particle samples were collected before, during, and after IEPOX reactive
368 uptake to confirm core-shell structure. A 3-stage microanalysis particle sampler (MPS-3, California
369 Measurements, Inc.) with size cuts of 5.0, 2.5, and 0.4 μm was operated at 2.1 L min-1. Particles were
370 impacted onto carbon-type-b Formvar-coated copper transmission electron microscopy (TEM) grids and
371 silicon wafer substrates (Ted Pella, Inc.). Samples were stored in sealed plastic vials at room temperature
372 prior to analysis. For AFM particles on silicon substrates were imaged in 5x5 μm regions by an AFM
373 (Anasys Instruments) in tapping mode with 75 ± 15 kHz resonant frequency and 1-7 N m-1 spring
374 constant at ambient laboratory temperature (~23 °C), pressure, and RH (~36%). A FEI Helios 650
375 Nanolab-Dualbeam electron microscope equipped with a high angle annular dark field (HAADF) detector
16 376 operated at an accelerating voltage of 10.0 kV, a current of 0.80 nA, and pressures ranging from 10-3 to
377 10-5 Pa analyzed TEM grids. For SEM particles were analyzed orthogonal to the beam on TEM grids and
378 at a 55 degree angle (tilted) on the silicon wafer.
379 Biphasic Microfluidics. Surface and interfacial tensions inform solute surface-bulk partitioning and,
380 therefore, the availability of compounds for interactions with the ambient, such as heterogenous chemistry
381 and water uptake. We measure liquid-liquid interfacial tensions of isoprene tracer 2-methyltetrols (2-MT)
382 and its organosulfate derivative (3-methyltetrol ester sulfate, IEPOX-OS) using a biphasic microfluidic
383 platform. A microfluidic chip is fabricated using standard soft-lithography techniques 67–69 and mounted
384 on an inverted microscope. Two immiscible liquid phases separately enter the device by pressure-driven
385 flow, 70 and droplet breakup is induced by a flow-focusing geometry 71–73 Aqueous solutions containing
386 the SOA are the dispersed (droplet) phase, and silicone oil is the carrier (continuous) phase. The dispersed
387 phase enters a contraction geometry downstream, which induces deformation of the liquid-liquid interface
388 due to an extensional flow field. This deformation, a result of the balance between interfacial tension and
389 deforming force due to flow, is imaged at high speed, and related to material and flow-field properties 74–
390 79 to calculate the interfacial tension between silicone oil and aqueous SOA solutions through the
391 following equation of motion:
5 − = . 2 ̂ +3
392 In the above equation, is a function of the viscosity ratio ̂, is the continuous phase viscosity,
393 is the extensional strain rate in the contraction, is the velocity of the droplet centroid, is
394 the deformation, is the un-deformed diameter of the spherical droplet, and is the interfacial tension