Influence of NO_x on the Physical and Chemical Properties of Isoprene SOA

Abstract

Isoprene is a significant source of secondary organic aerosol (SOA) in the atmosphere. This study investigates the physicochemical properties of isoprene-derived SOA formed through ozonolysis and photooxidation under varying NO_x conditions in an environmental chamber. SOA produced by dark ozonolysis and under low NO_x conditions had a density of 1.35–1.38 g cm⁻³ and an organic-to-carbon (O:C) ratio of 0.89–0.97. It was relatively volatile, consisting of semi-volatile organic compounds (SVOCs, 40%) and low-volatility organic compounds (LVOCs, 52%), with a small fraction of extremely low-volatility organic compounds (ELVOCs, ~7%); its vaporization enthalpy (ΔH_vap) was 90–106 kJ mol⁻¹. Under high NO_x conditions (isoprene/NO_x ratios = 1.2–6.8, with isoprene units in ppbC), SOA exhibited lower density (1.26–1.29 g cm⁻³) and lower O:C ratios (0.62–0.72). It was also less volatile than SOA formed under dark ozonolysis and low NO_x conditions; volatility decreased with decreasing isoprene/NO_x ratio, while ΔH_vap increased from 65 to 95 kJ mol⁻¹. SOA formed under very high NO_x conditions (isoprene/NO_x ratio = 0.6) was characterized by a higher density (1.34 g cm⁻³) and O:C ratio (0.88). However, it was the least volatile, comprising 68% LVOCs and 32% ELVOCs, and had the highest ΔH_vap of 114 kJ mol⁻¹. At low isoprene/NO_x ratios (0.6–1.2) yields were suppressed (0.6%) in comparison to those (6.8%) at higher isoprene/NO_x ratios (5–7).

1. Introduction

Isoprene (2-methyl-1,3-butadiene, C₅H₈) is the most abundant nonmethane hydrocarbon emitted into the atmosphere with a global flux of approximately 535 Tg/year [1]. Isoprene reacts with atmospheric oxidants and the lower-volatility oxidation products are transferred to the particulate phase, forming SOA [2,3]. Isoprene SOA formation has been studied both in field measurements and laboratory experiments with a focus on chemical composition and yield [4,5,6,7,8,9,10,11,12]. Nitrogen oxides (NO_x = NO + NO₂) strongly influence SOA formation through their impact on organic peroxy radical (RO₂) chemistry [5,13,14,15]. However, relatively few studies have examined how NO_x affects the physicochemical properties of isoprene SOA.

Although biogenic SOA density has been broadly investigated [16,17,18], data specific to isoprene SOA density remains limited and inconsistent. Kroll et al. [19] estimated a density of~1.4 g cm⁻³ under high NO_x conditions using Differential Mobility Analyzer (DMA) and Aerosol Mass Spectrometer (AMS) data. Later, Kroll et al. [5] reported densities of 1.25 ± 0.01 g cm⁻³ and 1.35 ± 0.05 g cm⁻³ for low and high NO_x, respectively, by using the AMS vacuum aerodynamic diameter and the Scanning Mobility Particle Sizer (SMPS) mobility diameter. The same technique was applied by Xu et al. [13] who found isoprene SOA densities 1.3 g cm⁻³ and 1.4 g cm⁻³ for low and high NO_x experiments correspondingly. Clark et al. [14] using an Aerosol Particle Mass Analyzer (APM) and an SMPS measured the isoprene SOA densities at various temperatures, and they found values of 1.35 g cm⁻³, 1.4 g cm⁻³ and 1.32 g cm⁻³ for ozonolysis, low and high NO_x conditions respectively at 300 K. These discrepancies indicate that there is an uncertainty regarding the isoprene SOA density at various conditions and highlight the need for more systematic investigation.

Volatility is another underexplored parameter of isoprene SOA. In the ambient, isoprene SOA volatility distribution has been estimated in three studies, all of them during the Southern Oxidant and Aerosol Study (SOAS) that took place in the southeastern United States. During this campaign, isoprene epoxydiols IEPOX SOA (i.e., low NO_x isoprene SOA products) were identified. Based on the Filter Inlet for Gases and AEROsols coupled to a Chemical-Ionization Mass Spectrometer (FIGAERO-CIMS) signals of C₅H₁₂O₄ and C₅H₁₀O₃, Lopez-Hilfiker et al. [20] suggested that the IEPOX SOA had a very low saturation concentration of C* = 10⁻⁴ μg m⁻³. Hu et al. [21] using ambient and thermodenuded AMS measurements, Positive Matrix Factorization (PMF) analysis and the technique of Faulhaber et al. [22], estimated an average volatility of C* = 5.2 × 10⁻⁵ μg m⁻³ for the IEPOX SOA. Kostenidou et al. [23] found an IEPOX SOA saturation concentration of C* = 1.05 ± 0.3 μg m⁻³ and a vaporization enthalpy of 63 ± 15 kJ mol⁻¹ using AMS measurements (both ambient and thermodenuded), PMF analysis and applying the dynamic mass transfer model of Riipinen et al. [24]. These results suggest large uncertainties in the volatility distribution of low NO_x isoprene SOA.

Smog chamber studies on isoprene SOA volatility are also limited. Xu et al. [13] performed thermodenuder measurements on isoprene photooxidation SOA at low and high NO_x conditions. Using the thermograms, they found that low NO_x SOA was more volatile than high NO_x SOA and they concluded that at high NO_x conditions the volatility exhibits a non-linear dependence on NO_x levels. Clark et at. [14], deploying a thermodenuder at 100 °C and size selected particles, found that ozonolysis and low NO_x isoprene SOA was more volatile compared to high NO_x SOA. Kleindienst et al. [7] reported an effective vaporization enthalpy of 42 kJ mol⁻¹ for isoprene ozonolysis SOA, while Kleindienst et al. [8] found that the effective vaporization enthalpy from isoprene photooxidation SOA in the absence and presence of NO_x were 38.4 and 43.2 kJ mol⁻¹, respectively. However, no comprehensive study has yet addressed the volatility distribution of isoprene SOA across ozonolysis, low, and high NO_x regimes.

In this work we conducted smog chamber isoprene ozonolysis, low and high NO_x photooxidation experiments at varying NO_x concentrations. SOA was chemically characterized, and at the same time the density, volatility distribution, vaporization enthalpy and yield were quantified. In addition, the effect of NO_x concentration on the above properties was investigated. To our knowledge, this is the first study to systematically investigate how isoprene SOA density, volatility distribution, and vaporization enthalpy vary with NO_x concentration across a range of isoprene/NO_x ratios. While previous chamber studies have reported SOA densities and vaporization enthalpies under high- and low NO_x conditions, they did not examine their evolution as the isoprene/NO_x ratio changes. Regarding volatility, only one previous chamber study [13] examined the mass fraction remaining (MFR) as a function of NO_x concentration. Here, we go beyond MFR measurements by deriving the volatility distribution and vaporization enthalpy of isoprene SOA as a function of the isoprene/NO_x ratio. A list of abbreviations used throughout this manuscript is provided in the abbreviations section.

2. Materials and Methods

The experiments were conducted at the environmental chamber facilities of the Foundation for Research and Technology-Hellas (FORTH-ASC) in Patras. Isoprene reacted with ozone in the dark or with OH radicals produced either by H₂O₂ or by HONO photolysis under UV illumination ($j_{{NO}_2}$ = 0.59 min⁻¹). SOA formation started within a few minutes to approximately 2.5 h after the initiation of photooxidation or ozonolysis.

2.1. Experimental Set Up

A 10 m³ Teflon environmental chamber was used for the experiments. The relative humidity (RH) was consistently maintained below 20% to minimize the influence of aerosol liquid water and aqueous-phase chemistry and to isolate the effect of NO_x levels on SOA formation and properties. Τhe temperature ranged between 21 °C and 25 °C. Prior to each experiment, the chamber was flushed with purified dry air for 12–15 h and then filled with clean air containing fewer than 1 particle cm⁻³. Isoprene (Sigma-Aldrich, St. Louis, MO, USA; purity ≥ 99.5%) was subsequently injected. In the dark ozonolysis experiment, ozone was generated using an ozone generator (Azco Industries Inc., Cincinnati, OH, USA; HTU500ACPS), without OH scavenger added. For the OH photooxidation experiments, either an H₂O₂ solution (35 wt% in H₂O₂, Sigma-Aldrich, St. Louis, MO, USA) or a HONO solution, prepared as described in Kostenidou et al. [25] was introduced into the chamber via a bubbler. In each HONO experiment, the bubbling time varied, resulting in different OH radical and NO_x concentrations produced through HONO photolysis. For the photolysis experiments, UV lamps (Osram GmbH, Munich, Germany; model L36W/73) were used as the light source. The experimental parameters are summarized in

Table 1. Experimental conditions and resulting SOA properties for the seven chamber experiments. The table reports the initial concentrations of isoprene, O₃, and NO_x, as well as the corresponding NO_x/isoprene ratios. In addition, the estimated initial OH concentrations, the maximum SOA mass concentrations formed during the experiments, the average SOA density, CE values, and the corresponding SOA yields are shown.

	Experiment
	1 a	2	3	4	5	6	7
Initial isoprene (ppb)	200	150	100	100	100	100	100
Initial O3 (ppb)	770	-	-	-	-	-	-
H2O2	-	Yes	-	-	-	-	-
HONO	-	-	Yes	Yes	Yes	Yes	Yes
Initial NOy (ppb) b	0.43	<1	73	100	226	425	900
NOx/Isoprene (ppb/ppb) c	0.215	0.13	0.73	1	2.26	4.25	9
Isoprene/NOx (ppbC/ppb) c	465	37.5	6.8	5	2.2	1.2	0.6
Initial OH concentration (molecules cm−3)	NA	8.05 × 105	2.53 × 106	3.47 × 106	9.58 × 106	1.37 × 107	2.53 × 107
Maximum SOA mass concentration (μg m−3) d	14.7	12.8	7.3	9.1	1.3	0.6	0.3
Average SOA density (g cm−3)	1.38 ± 0.02	1.35 ± 0.02	1.28 ± 0.06	1.26 ± 0.05	1.29 ± 0.06	1.26 ± 0.06	1.34 ± 0.01
CE	0.99 ± 0.06	0.7 ± 0.1	1.04 ± 0.09	1.08 ± 0.11	1.05 ± 0.15	0.94 ± 0.06	0.96 ± 0.26
Yield (%) e (Yield range (%))	3.0 (3.0–3.2)	3.6 (2.2–4.6)	6.4 (6.1–6.9)	6.1 (5.9–6.7)	1.3 (1.2–1.3)	0.5 (0.5–0.5)	0.6 (0.5–0.6)

^a Dark ozonolysis experiment without OH scavenger. ^b The reported NO_y concentrations were measured using a chemiluminescence analyzer with a molybdenum converter and may include contributions from HONO. ^c The reported isoprene/NO_x ratios are calculated using the measured NO_y concentrations (see note above) and should be interpreted as operational indicators of the oxidation regime. ^d Using the SMPS data and applying the density found (no particle wall-loss corrections). ^e After applying for particle wall-loss corrections.

.

The size-resolved chemical composition and mass concentration of the SOA were measured using an Aerodyne High-Resolution Time-of-Flight Aerosol Mass Spectrometer (HR-ToF-AMS; Aerodyne Research Inc., Billerica, MA, USA) [26]. A Scanning Mobility Particle Sizer (SMPS; classifier model 3080, DMA model 3081, CPC model 3775, TSI Inc., Shoreview, MN, USA) provided measurements of particle size, number, and volume distributions.

A thermodenuder (TD) [25,27,28] was employed to measure the mass fraction remaining (MFR), providing information about the volatility of isoprene SOA. The aerosol flow alternated every 3 min between a bypass line and the thermodenuder line. In both cases, the sample was analyzed by the HR-ToF-AMS and the SMPS. The thermodenuder was operated at temperatures between 25 °C and 250 °C, with a centerline residence time of 14 s. Τhermodenuder measurements began after the SOA mass concentration had reached its maximum value.

Volatile organic compound (VOC) concentrations were monitored using a Proton Transfer Reaction Mass Spectrometer (PTR-MS, Ionikon Analytik GmbH, Innsbruck, Austria) for 48 different m/z values, with a resolution of 11 s. The inlet tube and reaction chamber were maintained at 60 °C, while the drift tube pressure was 2.2 mbar, and the applied voltage was 600 V. NO_x and O₃ concentrations were measured using the corresponding analyzers (Teledyne Instruments Inc., San Diego, CA, USA; models T201 and 400E, respectively). T201 employs a molybdenum catalytic converter to convert NO₂ to NO prior to detection. Under these conditions, other oxidized nitrogen species such as HONO may also be partially converted to NO in the converter and contribute to the reported NO₂ signal. Therefore, during the HONO photolysis experiments, the reported NO₂ (and consequently NO_x) concentrations may include significant contributions from HONO present in the chamber. Therefore, the reported NO_x values should be interpreted as NO_y-like measurements. Gas standards (Air Liquide, Paris, France) were used for calibration.

2.2. Data Analysis

2.2.1. AMS Data Analysis

HR-ToF-AMS data were analyzed using standard HR-ToF-AMS softwares, SQUIRREL v1.60A and PIKA v1.20A, with Igor Pro 9 (Wavemetrics, Portland, OR, USA). Elemental ratios were estimated using the method of Canagaratna et al. [29] and the fragmentation table of Aiken et al. [30].

2.2.2. OH Calculation

OH concentrations were calculated based on isoprene decay by tracking the m/z 69 signal in PTR-MS and using a reaction rate constant of 1.01 × 10⁻¹⁰ cm³ molecule⁻¹ s⁻¹ for OH [31]. For the photooxidation experiments, the OH concentration ranged between 8 × 10⁵ and 2.5 × 10⁷ molecules cm⁻³.

2.2.3. Particle Wall-Loss Corrections

Particle wall-losses in the smog chamber were corrected using the approach described by Wang et al. [32] which accounts for size-dependent particle correction. The wall loss rate constant (k) was determined using dry polydisperse ammonium sulfate particles. These measurements were performed between experiments.

2.2.4. SOA Density and Collection Efficiency (CE)

The density of SOA and the CE of the HR-ToF-AMS (both bypass and thermodenuded) were estimated by applying the algorithm of Kostenidou et al. [6] which matches AMS mass distributions to SMPS volume distributions.

2.2.5. Thermodenuder Loss Corrections

To correct for particle losses inside the thermodenuder, we followed the procedure suggested by Louvaris [28]. Ammonium sulfate particles were passed through the bypass and TD line at different temperatures. For the particle losses quantification, we compared the bypass and TD SMPS distributions for each size bin and temperature.

2.2.6. Mass Fraction Remaining

MFR was calculated based on SMPS measurements. The TD volume concentrations were first corrected for particle losses inside the thermodenuder, and then the volume fraction remaining (VFR) was determined. VFR was converted to MFR using the bypass and the TD SOA densities measured for each experiment.

2.2.7. Volatility Distributions

SOA volatility was estimated using the dynamic mass transfer model of Riipinen et al. [24]. The model simulates particle evaporation inside the thermodenuder using experimental parameters as inputs, including the loss-corrected MFR, thermodenuder temperature and residence time, the average bypass particle size and SOA concentration as well as the accommodation coefficient (α_m). This model outputs the predicted MFR, the SOA volatility distribution at 298 K and the effective vaporization enthalpy by minimizing the error between model predictions and measurements. This constrained optimization reduces the number of free parameters and therefore limits the possibility of overfitting the thermodenuder data. In this analysis we investigated a 6-bin solution with a variable mass fraction value for each bin in the range of C* = 10⁻⁶ to 10 µg m⁻³ and assuming a mass accommodation coefficient of unity. The optimal bin range for each experiment was determined using the error analysis method proposed by Karnezi et al. [33], which identifies an ensemble of parameter combinations that reproduce the measured thermograms and uses them to derive uncertainty ranges.

2.2.8. Yield Calculation

SOA yields were calculated from SMPS volume concentrations, corrected for particle losses on the chamber walls and converted to mass using the experimentally determined SOA density. For both low and high NO_x conditions we used the final mass concentrations. The yield (Y) was calculated using the following equation:

Y = C_SOA/ΔVOC

(1)

where C_SOA is the SOA mass concentration and ΔVOC is the mass concentration of VOC (isoprene) that reacted during photooxidation.

3. Results and Discussion

Isoprene oxidation proceeds through different pathways depending on NO_x levels. Under low NO_x conditions, RO₂ radicals primarily react with HO₂, leading to the formation of IEPOX and subsequent SOA products such as 2-methyltetrols and 2-methyl alkenetriols. Under high NO_x conditions, RO₂ reacts mainly with NO forming intermediates such as methacrolein-derived peroxy radical (MACRO2), which can subsequently react with NO₂ to form methacryloyl peroxy nitrate (MPAN) and lead to products such as 2-methylglyceric acid (2-MG), oligomers and organic nitrates. At higher NO_x concentrations, MACRO2 reacts preferentially with NO rather than NO₂, resulting in fragmentation into smaller, more volatile species and/or the formation of organic nitrate compounds, and at the same time the formation of lower amounts of 2-MG. These pathways influence the oxidation state, volatility distribution, and density of the resulting SOA. Scheme S1 shows a schematic representation of the main chemical pathways. It should be noted that in this study the reported NO_x concentrations in the HONO experiments correspond to NO_y-like measurements due to potential HONO interference. However, HONO photolysis leads to the formation of NO and OH, and therefore the isoprene/NO_x ratio can still be interpreted as an indicator of the prevailing RO₂ reaction regimes.

3.1. HR-ToF-AMS Mass Spectrum and O:C Ratio

The high-resolution mass spectra for experiments 1, 2, 3 and 6 and for experiments 4, 5 and 7 are shown in Figure 1 and Figure S1, respectively. In the high NO_x cases (experiments 3–7) m/z 44 (i.e., CO₂⁺ fragment) was the dominant peak. This could be attributed mainly to 2-MG (and its oligomers) as well as to organic nitrate compounds formed under high NO_x conditions through the pathway of MACRO2 + NO₂ that produces MPAN [4,6]. The fraction of m/z 44 (f44) increased from 0.11 to 0.22 with increasing initial NO_x levels, indicating a greater contribution from 2-MG, its oligomers, and organic nitrates at higher NO_x. In experiments 6 and 7, where NO_x levels were particularly high, the SOA mass concentration remained low (0.3–0.6 μg m⁻³). This is likely due to MACRO2 reacting preferentially with NO rather than NO₂ leading to fragmentation into smaller and more volatile species (e.g., hydroxyacetone and methylglyoxal) and/or to formation of organic nitrate compounds, thereby suppressing 2-MG formation [9]. This is supported by the increasing aerosol NO₃/organics ratio in the formed aerosol with NO_x levels (Figure S2). Aerosol NO₃ was measured using the HR-ToF-AMS. The NO₃ signal (i.e., NO⁺ and NO₂⁺ fragments) represent total particulate nitrate, including both organic and inorganic nitrate species. The average NO₂⁺/NO⁺ ratio was low (0.17 ± 0.03), indicating that the nitrate signal is dominated by organic nitrate species. This interpretation is further supported by the low RH conditions (<20%) and the absence of ammonia, which limit the formation of inorganic nitrate. Other characteristic peaks include m/z’s 58 (C₃H₆O⁺), 75 (C₃H₇O₂⁺) and 85 (C₄H₅O₂⁺), which were also reported under high NO_x by Chhabra et al. [10] and Xu et al. [13]. Their relative intensities decreased with increasing NO_x, indicating changes in SOA chemical composition. Notably, higher NO_x concentrations also suppressed particle growth, resulting in smaller particle sizes. Figure S3 presents the mode diameter obtained by the number size distributions as a function of the isoprene/NO_x ratio.

At low NO_x levels, (experiment 2), the mass spectrum was characterized by m/z’s 29 (COH⁺), 43 (C₂H₃O⁺) and 44 (80% CO₂⁺ and 20% C₂H₄O⁺). m/z 43 (C₂H₃O⁺) was more abundant than m/z 44, likely due to the IEPOX products (2-methy alkenetriols and 2-methyl-tetrols) and the reduced presence of 2-MG. Similar trends were observed by Chhabra et al. [10] under low NO_x conditions.

For the ozonolysis case (experiment 1) m/z 43 and m/z 44 had similar intensities. As no OH scavenger was used, it is expected that approximately 50% of the SOA products originated from HO₂^. channel [7]. Thus, alongside acidic compounds from isoprene ozonolysis (e.g., pyruvic acid, glycolic acid, acrylic acid, succinic acid, etc.) [11], IEPOX-derived products and 2-MG contributed to SOA load.

The O:C ratio as a function of the isoprene/NO_x ratio for each experiment is depicted in Figure 2a. The O:C ratio was higher (0.9–1.0) in experiments 1 and 2, where IEPOX products dominated, and lower (0.62–0.72) in high NO_x experiments, (3–6), (isoprene/NO_x ratios 1.2–6.8), where 2-MG, its oligomers, and nitrates were more prominent. However, in experiment 7 (isoprene/NO_x = 0.6), the O:C increased again to 0.88. This could be explained by an increased role of MACRO2 + NO reactions at high NO_x, promoting fragmentation, reducing 2-MG formation and favoring organic nitrate production. The smaller fragmentation products are more volatile and tend to remain in the gas phase. As a result, a larger fraction of the compounds that partition to the particulate phase is likely composed of organic nitrate species.

3.2. Density and AMS CE

Figure 2b shows the average SOA density calculated for each experiment. Isoprene SOA density was highest during dark ozonolysis (1.38 ± 0.02 g cm⁻³) and at low NO_x conditions (1.35 ± 0.02 g cm⁻³). In HONO experiments at NO_x levels 73–425 ppb (isoprene/NO_x ratios 1.2–6.8), SOA density decreased to (1.26–1.29 g cm⁻³). Interestingly, at the highest NO_x (900 ppb, isoprene/NO_x = 0.6), the density increased again to 1.35 ± 0.01 g cm⁻³. Reported values of isoprene SOA density in previous chamber studies span a relatively wide range (1.25–1.40 g cm⁻³) depending on the oxidation conditions. For example, Clark et al. [14] reported lower SOA densities in the presence of NO_x compared to low NO_x conditions, consistent with the decrease observed in our experiments at high NO_x levels. In addition, they found similar densities for ozonolysis SOA. However, Xu et al. [13], Kroll et al. [5] and Kroll et al. [19] reported higher SOA densities under high NO_x conditions compared to those in the present study. Part of this discrepancy may arise from methodological differences. For instance, Kroll et al. [5] and Kroll et al. [19] used seed particles in their high NO_x experiments, and their approach assumes a constant particle density with particle size. In addition, Xu et al. [13] estimated SOA density by comparing mobility and aerodynamic diameters, but the corresponding size distributions were not explicitly matched. Overall, our results for low NO_x cases fall within the range reported in the literature, whereas for high NO_x conditions our densities are in a lower regime (Figure 3).

The results of the present work reveal a non-linear dependence of SOA density on the isoprene/NO_x ratio. This behavior suggests that RO₂ chemistry plays a key role in determining SOA density. Products associated with the IEPOX pathway (e.g., 2-methyltetrols and 2-methyl alkenetriols) under low NO_x conditions, as well as products formed through RO₂ + NO reactions (likely organic nitrate compounds), may lead to relatively high particle densities. In contrast, products formed through RO₂ + NO₂ reactions may generate compounds with different molecular structures (e.g., 2-MG, 2-MG dimers, and 2-MG organonitrates), resulting in lower densities at intermediate NO_x levels. Figure 2 compares isoprene SOA density to that of other biogenic SOAs, and key constituent compounds identified in biogenic SOA systems [16,17,18,25,34,35]. Isoprene SOA density was similar to that of norpinic acid, terebic acids and α-humulene SOA, but lower than α-pinene, β-pinene, and d-limonene SOA, and higher than β-caryophyllene SOA.

CE did not exhibit a trend, and it was around 1 in all experiments except for the low NO_x experiment where it was 0.7 ± 0.1. CE and density for thermodenuded and bypass SOA were similar during TD sampling (Figures S4 and S5).

3.3. Volatility

3.3.1. Thermodenuder Measurements

Figure 4 shows the measured MFR as a function of the thermodenuder temperature. SOA from experiments 1 and 2, where there was no NO_x present, was more volatile as half of the SOA evaporated at 75 °C. At high NO_x conditions SOA was less volatile since half of the SOA was evaporated at higher temperatures (90–115 °C). This behavior shows that the IEPOX products are more volatile than the MACRO2 products. Similar trends were observed by Xu et al. [13] and Clark et al. [14]. Our results suggest an inverse relationship between SOA volatility and NO_x concentration within the range of conditions investigated, with lower NO_x (higher isoprene/NO_x ratios) producing less volatile SOA. At lower NO_x levels 2-MG and its oligomers contributed more, while high NO_x favors fragmentation. Xu et al. [13] reported similar but more complex behavior, with volatility decreasing, reaching a minimum, then increasing with NO_x. The thermodenuded mass spectra at the temperature where half of the SOA was evaporated were similar to the corresponding bypass ones with angle thetas ranging between 3.6 and 15.6 degrees (R² = 0.930–0.996). The angle theta is described in Kostenidou et al. [36].

3.3.2. Volatility Distributions and Vaporization Enthalpy

Figure S6 depicts the measured versus the modeled MFR for each experiment. The optimal 6-bin solution for each experiment is shown in Table S1. Figure S7 shows the estimated volatility distribution for all experiments. Figure 5 presents vaporization enthalpy and volatility composition at 298 K versus the isoprene/NO_x ratio, based on the volatility basis set (VBS) framework proposed by Donahue et al. [37]. Under ozonolysis and at low NO_x, SOA consisted of ~40% SVOCs, ~52% LVOCs, and ~7% ELVOCs. At high NO_x SVOCs decreased with NO_x (or as the isoprene/NO_x ratio decreased) and at very high NO_x (very low isoprene/NO_x ratio) their contribution was practically zero. ELVOCs became important as NO_x increased (or as the isoprene/NO_x ratio decreased) and reached around 32% at very high NO_x levels. The presence of EVOCs may be explained partially by low-volatility nitrates and dinitrates [38] supported by increasing NO₃/Organics ratios along with NO_x levels (Figure S2). ΔH_vap was 90–106 kJ mol⁻¹ under low NO_x and ozonolysis. For high NO_x conditions ΔH_vap varied from 65 to 114 kJ mol⁻¹ increasing with NO_x.

3.3.3. Sensitivity Analysis of Accommodation Coefficient

In the above analysis we assumed no resistance to SOA evaporation in the thermodenuder and a mass accommodation coefficient (α_m) equal to unity. To examine the sensitivity of the results to this assumption, additional simulations were performed using α_m values of 0.1 and 0.01. The resulting volatility distributions and vaporization enthalpies for each experiment are presented in Table S2 in the Supplement. The corresponding modeled MFR curves are shown in Figure A1 (Appendix A), while the volatility composition and the corresponding ΔH_vap values are presented in Figure A2 (Appendix A).

The predicted thermograms for α_m = 0.01 were inconsistent with the measurements for all experiments, while α_m = 0.1 reproduced the thermograms within experimental uncertainty only for experiments 3 and 4. Despite these differences, the derived volatility distributions remained relatively similar to the base case (α_m = 1), whereas the inferred vaporization enthalpies (ΔH_vap) were more sensitive to the assumed accommodation coefficient. For example, in experiment 3 the SVOC and LVOC fractions changed from 34% and 52% in the optimal solution to 43% and 45% for α_m = 0.1, while ΔH_vap increased by approximately 16 kJ mol⁻¹. Overall, these sensitivity tests indicate that the estimated volatility distributions are relatively insensitive to the assumed accommodation coefficient, whereas ΔH_vap is more strongly affected. Similar behavior was reported by Kostenidou et al. [23], although it differs from the results of Riipinen et al. [24].

3.4. Yields

Figure 6 shows SOA yields calculated for each experiment. For high NO_x cases at low isoprene/NO_x ratios (0.6–1.2) yields were low (0.6%), while at higher isoprene/NO_x ratios (5–7), yields reached 6.8%. For low NO_x experiments (<1 ppb NO_x) yields were ~3.5%. This non-linear trend has been observed previously [6,13]. At high NO_x levels (lower isoprene/NO_x ratio) yields were suppressed, due to fragmentation and formation of small compounds. Recently Xu et al. [12] proposed that NO_x could influence the formation of low, extremely low and ultralow volatility organic compounds (LVOCs/ELVOCs/ULVOCs) by altering the RO₂ fates, which are the critical compounds for the formation of the particulate phase. Combining the volatility and yield results, we observe that the contribution of LVOC and ELVOC fractions increases at lower isoprene/NO_x ratios, where the SOA yield is suppressed (Figure S8). This suggests that the small amount of SOA formed under low isoprene/NO_x conditions consists predominantly of low-volatility compounds. At higher isoprene/NO_x ratios, a larger fraction of semi-volatile compounds (SVOCs) is produced; a fraction of these species condenses and partitions into the particulate phase, increasing the SOA yield while simultaneously reducing the relative contributions of the LVOC and ELVOC fractions.

Our low NO_x yields align with Kroll et al. [5], Chhabra et al. [10], Xu et al. [13] and Clark et al. [14], (Figure S9) where the conditions were similar (i.e., absence of (NH₄)₂SO₄ seeds, similar isoprene and NO_x concentrations). Yields in our study are lower than those reported by Kroll et al. [15], likely because they employed ammonium sulfate seeds, which enhance SOA formation by capturing early oxidation products. High NO_x yields derived from this study are in good agreement with Xu et al. [13] and Clark et al. [16], where the conditions were similar (i.e., no ammonium sulfate seeds, similar isoprene and NO_x levels) (Figure S10). However, our yields are lower than Kroll et al. [5], since they used (NH₄)₂SO₄ seeds. These results highlight the importance of chemical pathways (ozonolysis, low and high NO_x photooxidation) and the associated products—acids, IEPOX + HO₂ products, MACRO2 + NO₂ products, and MACRO2 + NO products—in defining isoprene SOA physicochemical properties. Our findings suggest that RO₂ chemistry modulates isoprene SOA properties in a non-linear way.

4. Conclusions

This study examined the influence of NO_x concentration on the physicochemical properties of isoprene-derived SOA formed via ozonolysis and photooxidation in an environmental chamber. Isoprene SOA produced under dark ozonolysis and low NO_x conditions exhibited higher densities, higher O:C ratios, and greater volatility, dominated by semi-volatile and low-volatility organic compounds. In contrast, SOA formed under high NO_x conditions displayed lower density and O:C ratios over a broad range of isoprene/NO_x ratios, accompanied by reduced volatility and increasing vaporization enthalpy, indicating enhanced formation of lower-volatility material. At very high NO_x levels, SOA density and O:C ratios increased again, while volatility was minimized and extremely low-volatile compounds became a substantial fraction of the aerosol, consistent with the formation of organic nitrates and multifunctional products.

Although the MFR curves suggest that SOA formed at higher NO_x levels evaporates more readily, the application of the thermodenuder modeling framework indicates that this SOA actually contains a larger fraction of low-volalite compounds. This result highlights that MFR measurements alone are insufficient to constrain SOA volatility distributions and that modeling is required for a more comprehensive interpretation of the thermodenuder data, an approach that has not been applied in previous studies of isoprene SOA volatility. SOA yields exhibited a pronounced non-linear dependence on NO_x, with strong suppression at low isoprene/NO_x ratios due to enhanced fragmentation pathways.

Overall, these results demonstrate that NO_x-driven RO₂ chemistry exerts a critical and non-linear control on isoprene SOA density, volatility distribution, vaporization enthalpy, and yield, highlighting the importance of accurately representing NO_x-dependent pathways in atmospheric aerosol models. Future work should further investigate the role of additional atmospheric variables, such as relative humidity and aerosol phase state, on the physicochemical properties of isoprene SOA across a wider range of atmospheric conditions.