Interactive comment on “ Chamber-based insights into the factors controlling IEPOX SOA yield , composition , and volatility ”

This manuscript describes laboratory experiments using the Filter Inlet for Gases and Aerosols/Chemical Ionization Mass Spectrometry (FIGAERO-CIMS) technique that aim to investigate the nature of the components of IEPOX-derived secondary organic aerosol (IEPOX-SOA). Specifically, the work addresses the inconsistency between GCMS approaches that have identified semi-volatile molecular components and volatility measurements that have indicated that the bulk of IEPOX-SOA must be made up of much lower volatility molecular components. The main claim is that a desorption signal that corresponds to C5H12O4 has two maxima, one that arises from a semi-volatile


Introduction
Aerosols less than 1 µm in diameter play particularly important roles in the Earth's radiative balance and air quality, a large fraction of which is organic carbonaceous material of biogenic origin (~70%) (Hallquist et al., 2009).Isoprene, with a global emission rate of 500 TgC year -1 (Guenther et al., 2012), has the potential to form significant quantities of secondary organic aerosol (SOA).The ability of a volatile organic compound (VOC) to form SOA depends on either of two factors: the efficiency of its oxidative conversion to lower volatility products that can partition to the condensed phase, or the reaction of gas-phase oxidation products in the condensed phase to form products which remain in the condensed phase.Each of these two SOA formation mechanisms have been heavily studied in the case of isoprene.
The atmospheric oxidation of isoprene under low NO conditions typically proceeds by OH radicals, leading to the formation of first generation isoprene hydroxy hydroperoxides (ISOPOOH) in high yield (70%) (Paulot et al., 2009).The remaining double bond of isoprene can then undergo another OH addition, leaving a carbon-centered radical adjacent to a hydroperoxide moiety.This radical can internally rearrange to form an epoxy diol (IEPOX) at a yield of ~70-80% (St Clair et al., 2016) or undergo addition of O2 to form a peroxy radical.The peroxy radical undergoes bimolecular reactions to form closed shell hydroperoxide or nitrate products, or unimolecular H-shifts to form carbonyl-and epoxide-containing products (D'Ambro et al., 2017b;Paulot et al., 2009).The bimolecular peroxy radical reaction products have been shown to be of sufficiently low volatility to partition to the aerosol-phase (D'Ambro et al., 2017a;Liu et al., 2016), and IEPOX has been shown to react in aqueous acidic particles, forming SOA.
Commonly measured species from IEPOX reactive uptake include 2-methyltetrols (Lin et al., 2012;Surratt et al., 2006;Surratt et al., 2010;Wang et al., 2005), C5-alkene triols (Lin et al., 2012;Surratt et al., 2006;Surratt et al., 2010;Wang et al., 2005), organosulfates (Lin et al., 2012;Surratt et al., 2007a;Surratt et al., 2007b;Surratt et al., 2010), 3-methyltetrahydrofuran-3,4-diols (3-MeTHF-3,4-diols) (Lin et al., 2012), and oligomers (Lin et al., 2012;Lin et al., 2014;Surratt et al., 2010).Recently, studies have called into question whether these commonly measured monomeric products of IEPOX multiphase chemistry exist in the particle-phase as measured.We showed previously that components of organic aerosol in the Southeast U.S. with compositions of C5H10O3 and C5H12O4 desorbed at much higher temperatures, and therefore much lower volatilities, than would be expected based on their composition (Lopez-Hilfiker et al., 2016).We concluded that the measured compositions arise from lower volatility material in the SOA that thermally decomposes rather than evaporating as the native species, and that, based on the relative abundance of these lower volatility components, IEPOX SOA as a whole is typically comprised of very low volatility material.When IEPOX was reacted in bulk solutions and analyzed via nuclear magnetic resonance, no evidence was found for the production of C5-alkene triols or 3-MeTHF-3,4-diols, and although a second isomer of the MeTHF diol was observed (3-MeTHF-2,4-diols), the formation rate from IEPOX was calculated to be so slow relative to nucleophilic addition that its formation would be limited to situations where the aerosol had low water content (Watanabe et al., 2018).These findings were further corroborated by comparing a novel chromatography technique that does not involve heating to traditional GC/EI-MS analysis of IEPOX SOA compositions, finding that alkene triols and 3-MeTHF-3,4-diols were in fact formed via thermal decomposition of 2-methyltetrol sulfates and 3-methyltetrol sulfates, respectively (Cui et al., 2018).
An additional challenge to understanding IEPOX SOA is that there remains a large gap in carbon closure resulting from IEPOX reactive uptake to aqueous acidic aerosol particles.
Although the reactivity of IEPOX in acidic particles is high (Gaston et al., 2014), the SOA yield per reactive loss of IEPOX to particles is relatively low and varies greatly depending on aerosol composition, from approximately 3 to 21% (Riedel et al., 2015).This disconnect may present an inconsistency in models that simulate both IEPOX and IEPOX-derived SOA.If such low yields are indeed realistic, models that adjust the rate of IEPOX reactive uptake to match the IEPOX SOA tracer concentrations without accounting for the lower yield may not correctly simulate IEPOX distributions where reactive uptake is a dominant sink for IEPOX (Gaston et al., 2014).
In this work we seek to understand the nature of products formed via the reactive uptake of IEPOX in aqueous acidic particles and the flow of carbon between the gas-and particlephases.We present measurements from the Pacific Northwest National Laboratory (PNNL, Richland WA) environmental chamber during the 2015 SOA Formation from Forest Emissions Experiment (SOAFFEE) campaign.Batch and continuous flow mode experiments were performed with authentic trans-β-IEPOX, which is the dominant isomer (Bates et al., 2014), and wet acidic seed to study the products of IEPOX uptake and the resulting aerosol properties.The properties of commonly measured IEPOX uptake products, the 2-methyltetrols (C5H12O4) and C5-alkene triols or 3-MeTHF-3,4-diols (C5H10O3), such as volatility and solubility, are examined in the context of experiments utilizing isothermal evaporation of the formed SOA.The effect of acidity versus liquid water content on the products formed is also discussed, along with implications for modeling atmospheric IEPOX and its conversion into SOA.

Experimental Methods
Experiments were performed at the Pacific Northwest National Laboratory (PNNL) as part of the Secondary Organic Aerosol Formation from Forest Emissions Experiments (SOAFFEE) campaign held during the summer of 2015.PNNL's 10.6 m 3 fluorinated ethylene propylene (FEP) environmental chamber has been described in detailed previously (Liu et al., 2012).The chamber was run in both batch mode and continuous flow mode.In batch mode the Aliquots of an authentic trans-β-IEPOX standard, synthesized according to Zhang et al. (2012), dissolved in ethyl acetate were injected into a glass bulb which was connected to the chamber via ~10 cm of ¼" OD polytetrafluoroethylene (PTFE) tubing to achieve 2 ppb in steady state and at the beginning of batch modes.The bulb and transfer line were heated to 30-40 o C and a 100-300 sccm flow of zero air passed over the IEPOX to vaporize and carry it into the chamber.Typically, we measured ~ 10 µg m -3 of IEPOX prior to aqueous seed addition.Given calibration uncertainties of ~ +/-30% from repetitive injections and analytical errors, and potential losses on chamber and sampling surfaces, we estimate an uncertainty of approximately +/-50% for gas phase IEPOX concentrations and other tracers.An ammonium bisulfate solution acidified with additional H2SO4 was atomized to generate wet, polydisperse acidic ammonium sulfate seed which were added to the chamber at concentrations sufficiently large that condensation of IEPOX onto the particles was competitive with the chamber walls.Continuous flow experiments were conducted at 50% RH, while the RH of batch mode experiments was either 30% or 50%.The seed surface area concentration was 6.0 ± 0.3 10 8 nm 2 cm -3 for all experiments.The average NH4 + concentrations measured by the AMS were 1.1, 2.4, and 2.7 μg m -3 at steady state, 30% RH batch, and 50% RH batch, respectively.The average SO4 2- concentrations were 4.1, 13.0, and 10.4 μg m -3 at steady state, 30% RH batch, and 50% RH batch, respectively.The NH4 + /SO4 2-mole ratio of chamber aerosol as measured by the AMS during these experiments was typically between 0.7 and 1.5, implying some excess NH3 present in the chamber was partitioning to the seed, but that the aerosol remained acidic prior to injection of IEPOX.The corresponding aerosol pH range determined from the E-AIM aerosol thermodynamics model was 0.3 to 1 (Wexler and Clegg, 2002).
A suite of online gas and particle-phase instrumentation was used to monitor concentrations throughout the experiments.Aerosol number and volume concentrations were measured with a scanning mobility particle sizer (SMPS, TSI model 3936), O3 and NO/NO2/NOx concentrations were monitored with commercial instrumentation (Thermo Environmental Instruments models 49C and 42C, respectively).An Aerodyne high-resolution time-of-flight mass spectrometer (HRToF-AMS) was utilized to measure bulk submicron organic and inorganic aerosol composition.
A high-resolution time-of-flight chemical ionization mass spectrometer (HRToF-CIMS) using iodide adduct ionization was deployed for the detection of both semi-and low-volatility organic compounds (Lee et al., 2014).The HRToF-CIMS was used to monitor the evolving concentrations of both the precursor (IEPOX) and reaction products in both the gas-and particlephases when coupled to a Filter Inlet for Gases and AEROsols (FIGAERO), from here on referred to as FIGAERO-CIMS.The coupling, optimization, and operation of this combination has been described in detail previously (Lopez-Hilfiker et al., 2014) and is nearly identical to previous operations (D 'Ambro et al., 2017a;D'Ambro et al., 2017b).Briefly, the FIGAERO is an inlet manifold that allows for semi-continuous measurements of both gases and aerosols with approximately hourly resolution.Aerosol was collected on a 24 mm PTFE filter for 43 minutes at 2.5 L min -1 , during which the gases were measured in real time.Following collection, programmatically heated ultra-high purity (UHP) N2 was passed over the filter, while the temperature was ramped from ambient to 200 o C at a rate of 10 o C min -1 in order to thermally desorb compounds from the particle-phase to the gas-phase to be carried into the CIMS for detection.After the temperature was ramped, it was held at 200 o C for 50 minutes to allow species to desorb and signals to return to background levels.Particle blanks were conducted approximately every fourth desorption during continuous flow mode or at the beginning and end of each batch mode experiment by inserting a secondary filter upstream of the primary FIGAERO filter in order to get a measure of the gas adsorption artifact on the primary filter.Gas zeros were conducted every 2 minutes by over blowing the CIMS pinhole flow with UHP N2.
The specific coupling to a HRToF-CIMS with iodide ions allows detailed molecular analysis of hundreds of oxygenated organic compounds via a clustering, fragmentation-free ionization process.
The FIGAERO-CIMS was also utilized to perform isothermal evaporation experiments as have been described previously (D'Ambro et al., 2018).In normal operation, the programmatic thermal desorption is begun immediately after moving the FIGAERO filter under the heating tube.During isothermal evaporation experiments however, the aerosol is instead exposed for one hour to a stream of ambient temperature UHP N2 humidified to 50% RH by using a water bubbler and two mass flow controllers (Figure 1).After the hour exposure, the N2 flow is reverted to its normal dry state and the programmatic heating proceeds as normal.See Figure 1 for the experimental setup.The RH of 50% during evaporation periods was chosen to match that of the chamber to keep the phase state of the collected aerosols the same (i.e.so as to not drive efflorescence).Only the isothermal evaporation portion of the experiment was humidified; during the thermal desorption the bubbler was isolated using two solenoid valves.Passing excess humidified N2 over the aerosol and into the instrument resulted in a constant dilution of the vapor-phase, which allowed for any semi-volatile material to evaporate and also be carried into the CIMS for detection.

Chamber-generated IEPOX SOA Composition and Volatility
Our primary goals with this experiment were to assess whether chamber-generated IEPOX SOA had a composition and volatility similar to that inferred from field measurements using the FIGAERO-CIMS of various IEPOX tracers and to test whether IEPOX SOA as a whole or some of its components were semi-volatile as expected from mechanistic and kinetic considerations.Overall, the estimated mass yield of OA from IEPOX exposure to aqueous acidic seed was generally less than unity, of order 0.5 to less than 0.25, somewhat higher than estimates from Riedel et al [2015].Mean organic aerosol (OA) mass concentrations generated when IEPOX and aqueous acidic seed were 2.5 μg m -3 for steady state conditions, 5.5 μg m -3 at the beginning of the 30% RH batch experiment, and 4.5 μg m -3 at the beginning of the 50% RH batch experiment.The OA to sulfate ratio was observed to evolve during a batch experiment (e.g., decreasing by 20% from the peak), and given the uncertainties associated with potential vapor wall losses of IEPOX and its reaction products, we refrain from quantitatively interpreting the SOA yield behaviors in detail.
Regardless of the conditions, the uptake of an authentic IEPOX standard onto acidic seeds in these experiments results in a rather simple observed particle-phase composition upon thermal desorption.As shown in Figure 2, a few compositions dominate the average particlephase mass spectra, most predominantly C5H12O4 and C5H10O3.Both of these species have been repeatedly shown to be major components of IEPOX SOA (Lin et al., 2012;Surratt et al., 2006;Surratt et al., 2010;Wang et al., 2005), although the relative abundances could change with time or conditions.In ambient aerosol in the Southeast U.S., the same FIGAERO-CIMS instrument detected C5H12O4 and C5H10O3 in SOA, and these tracers correlated with and explained ~50% of the IEPOX SOA mass derived from factor analysis of aerosol mass spectrometer (AMS) data (Lopez-Hilfiker et al., 2016).Our laboratory chamber experiments starting with an authentic IEPOX standard and acidic seed without photochemical oxidants therefore support the use of these FIGAERO-CIMS compositions as tracers of IEPOX SOA in atmospheric particles.As these two compositions are such a large component of the particle-phase signal (97.5%) measured by FIGAERO-CIMS in chamber generated IEPOX SOA, the properties of the corresponding SOA are presumably similar to their properties.that the abundance and variability of the lower Tmax (semi-volatile) mode was consistent with an organic compound having the measured molecular composition and c* of the 2-methyltetrol undergoing equilibrium gas-particle partitioning, while the higher Tmax mode and that of the C5H10O3 arose from the decomposition of accretion products.

Insights into Volatility via Isothermal Evaporations
The above considerations of chamber generated IEPOX SOA suggest that a large fraction (the high Tmax mode) should be relatively stable against evaporation upon dilution of the gasphase, while the lower Tmax mode of the C5H12O4 (2-methyltetrol) component should respond to dilution by evaporating from the particle-phase.To test this hypothesis, we conducted isothermal evaporation experiments using IEPOX SOA generated in the chamber.
Figure 4 shows an example ion signal time series during an isothermal evaporation experiment for C5H12O4.C5H12O4 is detected in the gas-phase during particle collection (middle panel, blue shaded area) when chamber air containing IEPOX, acidic aqueous sulfate particles, and IEPOX SOA was being continuously sampled by the FIGAERO-CIMS.The gas-phase sampling included scanning of the dilution ratio, which resulted in varying signals, as well as periodic zeros resulting in occasional significant short-duration drops in signal.We show the last 15-minute portion of the cycle in the gas-phase (blue shaded region) when dilution was held constant, but zeros are still visible as the 2 dips in signal.The chamber was at steady state and the changing gas-phase signal is due to conditioning of the IMR and inlet tubing.During the isothermal evaporation period (yellow shaded area), C5H12O4 is also detected when a continuous flow of humidified UHP N2 passes over the particles collected from the chamber on the FIGAERO filter and into the mass spectrometer, consistent with C5H12O4 evaporation from the collected particles at room temperature (i.e.without heating).Finally, during the temperatureprogrammed thermal desorption, another pulse of C5H12O4 was detected corresponding to components in the remaining SOA that desorbed at elevated temperature (green shaded area).The mass concentration of C5H12O4 measured during a normal temperature-programmed thermal desorption is compared to that measured during the isothermal evaporation and subsequent desorption (Figure 4, bottom).Mass closure is achieved to within the experimental error, driven by variance in normal temperature-programed thermal desorptions due to chamber conditions and the water vapor effect on CIMS sensitivity (Lee et al., 2014).The observed behaviors, namely detectable gas-phase concentrations of C5H12O4 in the chamber and isothermal evaporation of C5H12O4 from collected particles indicate that (i) C5H12O4 is produced from IEPOX reactive uptake, as expected given that the 2-methyltetrol is predicted to be a major product (Eddingsaas et al., 2010), and (ii) a portion of the detected C5H12O4 behaves as a semivolatile organic compound (SVOC), being present in both the gas-and particle-phases, and evaporating promptly from the particle-phase in response to dilution of the surrounding organic vapors at room temperature.an effective Henry's law constant.Assuming no particle-phase diffusion limitations but accounting for FIGAERO mass transfer limitations (Schobesberger et al., 2018), we predict a c* of 5-15 µg m -3 for the portion of the C5H12O4 thermogram that evaporates.Utilizing COSMOtherm ( 2018) with the BP_TZVPD_FINE_18 parameterization as described previously (Kurtén et al., 2018), a Henry's law constant of 4.910 8 -1.110 10 M atm -1 is predicted if all conformers are used (4.910 8 M atm -1 ) or if the number of internal H-bonds is minimized (1.110 10 M atm -1 ), which compares well to the value calculated from the observed decay of C5H12O4 during the evaporation (1.810 8 M atm -1 ).These estimates do not include the likelihood of a salting-out effect expected for the 2-methyltetrol (Waxman et al., 2015), which would further lower the Henry's Law constant.Whether Raoult's or Henry's law is the appropriate framework for interpretation depends on whether IEPOX reactive uptake results in a phaseseparated organic medium, for example an organic coating, or a homogeneous aqueous solution, and the competitive partitioning of the 2-methyltetrol between two such regimes.Regardless, as we show below, the semi-volatile nature of the 2-methyltetrol, a major product of IEPOX reactive uptake, will cause it to partition strongly to the gas-phase under typical atmospheric conditions outside of cloud.As to the widening of the thermograms, Schobesberger et al. (2018) showed that a shallowing of the low temperature side of the thermogram and a broadening of the higher temperature tail correspond to thermal decomposition from a larger suite of bonds with different dissociation energies, consistent with continued formation of a variety of accretion products during the isothermal evaporation period.
The above evidence supports the previous assertions of Lopez-Hilfiker et al. ( 2016) that the lower Tmax mode of the C5H12O4 thermogram corresponds to a semi-volatile component, very likely the 2-methyltetrol, and further support the conclusion that IEPOX SOA in ambient aerosol is very to extremely low volatility.The isothermal evaporation experiments presented above provide an explanation as to why the ambient SOA contained such a relatively small fraction of the low Tmax (semi-volatile) 2-methyltetrol component in that it likely had evaporated to maintain gas-particle equilibrium.Furthermore, the high Tmax tracers do not decay in abundance during the evaporation experiments, but rather slightly increase with time at the highest temperatures (>120 o C), indicating ongoing accretion chemistry leading to lower volatility components.

Effect of RH and Acidity on IEPOX SOA Characteristics: Mechanistic Insights
We performed two time-dependent "batch mode" chamber experiments using IEPOX and acidic aqueous seed particles, one at 30% and the other at 50% RH.By operating in batch mode as opposed to continuous flow mode, we are able to temporally resolve the formation of SOA.
By varying the RH, we simultaneously varied the liquid water content relative to sulfate, and therefore also acidity.Three sequential thermal desorptions of C5H12O4 obtained over the course of experiments (~10 hrs total) at 30% (left) and 50% (right) RH are shown in Figure 6, top.At 30% RH, the lower Tmax (higher-volatility) mode grows in rapidly and is clearly visible in the first desorption (black line).The first desorption occurred after 43 minutes of particle collection, which began immediately after the seed was injected into the chamber and IEPOX uptake was initiated, resulting in collected aerosols having a variety of ages and thus the median age of 22 min is assumed.However, this mode then does not grow significantly larger as the experiment progresses.During the 2 nd and 3 rd desorptions, 2 hrs 22 min and 4 hrs 22 min respectively, after the initial exposure, the higher Tmax (lower-volatility) mode is visible and dominates the thermogram.Along with the shape, the Tmax of the C5H10O3 and each mode of the C5H12O4 vary slightly in time, the two of which are likely related.It has been shown previously that when the IEPOX-derived organosulfate (C5H12SO7) is deposited on and desorbed from the FIGAERO filter, it decomposes into both C5H12O4 and C5H10O3.The corresponding Tmax of both are colocated and highly dependent on acidity, with higher acidity leading to lower Tmax's (Lopez-Hilfiker et al., 2016).This dependence on the inorganic aerosol components, present in much larger excess in these experiments than our previous FIGAERO experiments, could be the cause of the shifts of the lower Tmax modes.Alternatively, the shift could be due to the increasing complexity of the SOA as it evolves in time leading to different interactions between particle components which affects volatility.
From these observations we can draw two conclusions regarding the mechanisms that give rise to the dominant components of IEPOX SOA, illustrated in Figure 7. First, the low Tmax, semi-volatile C5H12O4 exists in the aerosol from the first desorption and thus is likely formed promptly from IEPOX uptake, consistent with the formation of 2-methyltetrol via nucleophilic attack by water of the protonated epoxide ring (see Figure 7).That this semi-volatile mode is more prominent in the higher RH experiment, i.e. higher liquid water content and therefore higher H2O-nucleophile content relative to sulfate, further supports this interpretation.
Additionally, the higher liquid water content supports a greater amount of 2-methyltetrol remaining partitioned in the aerosol via Henry's Law, consistent with offline filter analysis (Riva et al., 2016).Second, the higher Tmax (lower volatility) modes are mostly produced more slowly over time, indicating a second or higher generation product of IEPOX uptake, as these modes are mainly observed 2.5 hours after IEPOX uptake has largely ended.Thus, if the higher Tmax modes  Another possible mechanism of organosulfate formation, as well as sulfate ester oligomers, is via SN2 reactions where one of the 2-methyltetrol -OH groups is protonated to make H2O the leaving group while bisulfate, sulfate, or an organosulfate is the substituting group (Figure 7).At 30% RH, the particle acidity is higher due to less dilution of the sulfate, which would result in higher organosulfate concentrations and acid catalyzed accretion chemistry (Jang et al., 2002), consistent with the observation of a more prominent higher Tmax (lower volatility) mode compared to the 50% RH experiment.The broader higher Tmax mode at 50% RH indicates that there is likely an array of compounds breaking apart to give rise to this specific composition, consistent with a greater variety of oligomerization reactions occurring due to the dilution of sulfate and higher 2-methyltetrol concentrations.Previous work has identified several non-sulfur containing polyol species, both monomers and oligomers, in IEPOX SOA (Lin et al., 2012;Surratt et al., 2010).

Summary & Atmospheric Implications
To place our findings into context, we present results from a simple conceptual model simulating IEPOX (initially 2 ppb) reactive uptake to form the corresponding 2-methyltetrol and organosulfate at yields of 90 and 10%, respectively, with an uptake coefficient of 0.05 based on Gaston et al. (2014), and an atmospherically relevant total surface area (2.510 -6 cm 2 cm -3 ) and volume (1.610 -11 cm 3 cm -3 ).We also include a loss of gas-phase 2-methyltetrol and IEPOX due to reaction with OH (Atkinson, 1987).The processes in the model are simplified from the reaction scheme discussed above, i.e. it does not include particle-phase processes, but its purpose is to capture the salient factors that control the reactive uptake and partitioning.The chosen branching between 2-methyltetrol and organosulfate yields from IEPOX reactive uptake to aerosol is somewhat arbitrary and only to illustrate the behavior of the system.The Henry's Law constant found via COSMOtherm for the 2-methyltetrol is used to simulate gas-particle partitioning of the 2-methyltetrol, while the organosulfate is a proxy for all low volatility products, including the promptly formed organosulfates, sulfate esters from further accretion, and polyol oligomers.
The simulated loss of IEPOX and formation of IEPOX SOA are shown in Figure 8.
IEPOX is almost completely consumed after 1 hour of reaction, corresponding to rapid formation of SOA.The composition of the aerosol changes significantly as a function of time.Initially, the SOA is composed primarily of 2-methyltetrol.However, despite the relatively high Henry's law constant, much of the 2-methyltetrol evaporates into the gas-phase to maintain equilibrium with the gas-phase 2-methyltetrol which is subjected to loss by gas-phase bimolecular reactions with the hydroxyl radical (OH).This behavior supports our findings herein that on a typical aerosol lifetime, the dominant IEPOX reactive uptake product, 2-methyltetrol, will be a small component of IEPOX SOA, and organosulfates and other low volatility material, including oligomers of 2methyltetrol, will dominate, albeit at a smaller overall SOA yield (Figure 8) per IEPOX reacting on aerosol.

Figure 2 .
Figure 2. Average mass spectrum of IEPOX-derived SOA at a total OA concentration of 5 μg m - 3 .

Figure 3 .
Figure 3. Thermal desorption profiles of chamber aerosol (blue) and calibrations with authentic standards (gray, dashed) of the two major particle-phase compounds detected in the chamber: C5H12O4 (top) and C5H10O3 (bottom).The C5H10O3 thermogram, in contrast, is monomodal, suggesting a single component giving rise to its desorption.However, the Tmax is much higher than the corresponding authentic cis-3-MeTHF-3,4-diol standard, synthesized according toZhang et al. (2012), (and also that of an alkane triol standard).This standard desorbs completely from the FIGAERO filter in seconds without heating (gray dashed line, Figure3, bottom).The Tmax of C5H10O3 desorbing from IEPOX SOA is 90 o C, the same as the higher Tmax mode of the C5H12O4 component, and thus implies a SOA component with an effective c* of at most 0.005 µg m -3 , indicative of thermal decomposition during desorption.For comparison, if the structure is assumed to be a C5-alkene Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2019-271Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 8 April 2019 c Author(s) 2019.CC BY 4.0 License.

Figure 4 .
Figure 4. Schematic of the isothermal evaporation process.Top: relative humidity (blue), N2 flow (green), and temperature (red).Middle: C5H12O4I-during a 1-hour isothermal evaporation experiment shaded by the phase of the experiments: simultaneous real-time gas-phase sampling and offline aerosol collection (blue), isothermal evaporation where compounds are measured as they evaporate off the filter (yellow), temperature-programmed thermal desorption (green), and cool down of the heating tube (gray).Bottom: mass concentration of C5H12O4 measured during a normal desorption (pink, left), versus the isothermal evaporation + desorption (yellow and green, right) for the same chamber conditions.

Figure 5 .
Figure 5. Thermograms obtained from prompt desorption of the aerosol (blue) and after one hour of evaporation (lavender, shaded) of the two major particle-phase compounds detected in the chamber: C5H12O4 (top) and C5H10O3 (bottom).
. Phys.Discuss., https://doi.org/10.5194/acp-2019-271Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 8 April 2019 c Author(s) 2019.CC BY 4.0 License.The second, higher Tmax mode of the C5H12O4 thermogram and the single Gaussian-like thermogram of C5H10O3 (Figure 5, bottom) do not change significantly after 1-hour of exposure to UHP N2.In the case of the C5H10O3 thermogram, nearly 95% of signal remains after the 1hour evaporation period.In both cases, there is a slight broadening of the thermogram and a measurable increase in material desorbing at the highest temperatures (120+ o C).Previous isothermal evaporation experiments with the FIGAERO have shown that, for α-pinene ozonolysis SOA, the physical age of the aerosol had a controlling role in the volatility of the bulk SOA and individual desorbing compounds (D'Ambro et al., 2018), consistent with the additional hour of non-oxidative aging in this system resulting in an increase in lower-volatility material.

Figure 6 .
Figure 6.Sequential desorptions of C5H12O4 (top) and C5H10O3 (bottom) during batch mode experiments at 30% RH (right) and 50% RH (left).In the 50% RH batch experiment, the lower Tmax (higher-volatility) mode of the C5H12O4 thermogram also dominates in the first desorption, but in contrast to the experiment at 30% RH, this lower Tmax mode continues to grow and is the dominant portion of the thermogram for all desorptions.While the higher Tmax mode is observed after the 1 st desorption, it is much broader and has an ambiguous peak, unlike at 30% RH.In the 30% and 50% RH experiments, both modes of the thermogram are observed by the 2 nd desorption 2.5 hours after IEPOX uptake starts, but the relative abundance and shape of the two modes differ with RH.The thermogram shape also changes as a function of time since IEPOX injection in the steady state experiments, although due to the range in aerosol ages present within the chamber at a given time during steady-state experiments, it is more straightforward to define this feature as a function of time in batch mode measurements.The corresponding thermograms of C5H10O3 in each of the two batch mode experiments are shown in Figure6, bottom.Most obvious is that the amount of C5H10O3 desorbing, relative to the 2-methyltetrol, is highest in the 30% RH experiment, when sulfate and hydronium ion concentrations are highest.Further work could be done to understand the evolution of IEPOX SOA components as a function of time, but a fairly stable set of products and volatility are reached within a few hours.
are from the thermal decomposition of an organosulfate product, as suggested by Lopez-Hilfiker et al. (2016) and as Cui et al. (2018) demonstrate, our experiments suggest it is unlikely to form solely from nucleophilic addition of (bi-)sulfate to protonated IEPOX, as that reaction should occur concurrently with 2-methyltetrol formation.

Figure 7 .
Figure7.Schematic of IEPOX reactive uptake and particle-phase processes.White region denotes semi volatile species that actively partition between the gas-and particle-phases, light green denotes species that are of lower volatility, and dark green outline denotes a coating.

Figure 8 .
Figure 8. Model results for the major gas-and particle-phase species of IEPOX reactive uptake for typical atmospheric aerosol and typical IEPOX mixing ratios.