Elevated Outdoor O₃ Can Sustain New Particle Formation and Growth Indoors

Abstract

Ultrafine particles and submicron aerosol mass negatively impact human health; while indoor primary sources are known, secondary aerosol formation from oxidation of volatile organic compounds (VOCs) is an important, but less understood, exposure source. Previous studies have suggested that indoor new particle formation (NPF) and growth are ozone-limited. However, we find that in an authentic aged unoccupied office in a region with high outdoor O₃ concentrations, sufficient O₃ is transported indoors that NPF and growth are instead limited by the availability of VOCs. Mechanical ventilation continuously supplies O₃ in the summertime and sustains the subsequent growth so that particle growth rates upon the addition of limonene range from 20 to 100 nm h^–1rates 1–2 orders of magnitude greater than those observed outdoors. Additionally, we observe meaningful mass contributions from the growth of newly formed particles. Household use of limonene-containing products in rooms with elevated ozoneeither from ozone-producing remediation and germicidal UV technologies, or during outdoor ozone exceedance eventswill produce ultrafine particles and increase particle mass. The effect of ventilated O₃ from outdoors on particle formation and growth highlights the value of continued efforts to reduce outdoor O₃ pollution for moderating both indoor and outdoor sources of secondary pollutants.

1. Introduction

Despite many observations of new particle formation (NPF) indoors, − lack of understanding of the conditions that drive indoor secondary particle formation inhibits our ability to generalize these results to different buildings or geographical regions. Distinct from primary particle production through cooking, indoor combustion, and other activities, secondary aerosol formation occurs when the oxidation of volatile organic compounds (VOCs) produces lower-volatility molecules that nucleate and grow new particles. This pathway can be an important source of ultrafine particles (UFPs; aerosols with diameters less than 100 nm) and integrated aerosol mass. The links between aerosol mass, specifically criteria air pollutant PM_2.5 (aerosols with diameters less than 2.5 μm), and cardiovascular and respiratory health effects and premature mortality are well established, even if the effect of ultrafine particles, independent of total particle mass, remains under investigation. Particle growth rates are measured in units of nm h^–1, and indoor ultrafine particle growth occurs at rates several orders of magnitude greater than outdoors. ,, The discrepancy is likely due to chemical regime (VOC versus oxidant concentrations), concentration of condensable vapors, different dilution time scales, and effective sinks. A simple paradigm that has been used to describe the difference between indoor and outdoor chemical regimes is that outdoor NPF events are typically limited by VOC mixing ratios, while indoor NPF events are typically limited by O₃ mixing ratios. Further, despite generally lower O₃ concentrations indoors, relatively high VOC concentrations indoors could be responsible for elevating particle formation and growth rates beyond what is observed in outdoor environments where VOCs are emitted from vegetation and there is relatively unlimited O₃. However, there is little investigation into the indoor chemical regime beyond O₃ levels and the observations of NPF under high terpene conditions, and this simple paradigm has not been rigorously tested.

Ozonolysis of monoterpenes is a key player in secondary chemistry indoors and can lead to the generation of ultrafine particles and/or additive growth of existing particles. Ozone is typically the major indoor oxidant, , and indoor O₃ concentration is controlled by ventilation of outdoor air into buildings (air change rate, ACR, h^–1), O₃ loss on indoor surfaces, and, in the case of densely populated indoor spaces, O₃ loss on humans. While typically accounting for less than 10% of ozone loss indoors, , ozonolysis of gas-phase VOCs may be an important source of ultrafine particles. Monoterpenes like α- and β-pinene, limonene, and δ-3-carene are known to produce secondary organic aerosol (SOA) from ozonolysis, and are ubiquitous indoors due to their presence in cleaning products, personal care products, and food. In particular, limonene ozonolysis is a well-studied, highly reactive indoor process that produces many gas-phase and condensed-phase products. ,,,,− Limonene has two unsaturated bonds available for ozonolysis: (1) endothe double bond in the ring, and (2) exothe double bond external to the ring. O₃ addition can occur at either double bond, with a branching ratio of endo to exo of 0.85:0.15, with the endo reaction favoring ring-opened products. Ozonolysis of monoterpenes can also initiate autoxidation mechanisms to form highly oxidized organic molecules with low volatilities that can participate in the formation and growth of new particles.

Much of our understanding of NPF comes from chamber studies or outdoor observations, which may not be representative of the indoor space for several reasons. In particular, chamber studies designed to probe outdoor SOA chemistry are often performed in chemical regimes that do not adequately replicate the relevant atmospheric conditions; for example, extensive analysis of the fate of organic peroxy radicals (RO₂)an important intermediate in VOC oxidationindicates that the pathways accessible in chamber studies may not be the same as those accessible in the global atmosphere. Furthermore, in contrast to outdoor and chamber studies, indoor spaces have a large surface area to volume ratio, and the surfaces can participate in gas-to-surface partitioning, multiphase chemistry, and chemistry within films that may coat the surfaces. At least one study showed that chemistry mediated by the surfaces, rather than gas-phase chemistry alone, may enhance particle formation rates and yields. The lack of authentic surfaces (gypsum board/drywall, paint, wood, carpet, etc.) further separates chamber studies from reality. Indoor studies in authentic indoor spaces are necessary to probe the chemistry under relevant chemical conditions.

Outdoor O₃ concentrations drive the oxidative capacity indoors and vary regionally, seasonally, and daily. Synthesized reports from over 2000 indoor environments found that indoor ozone was typically 4–6 ppb compared to an average outdoor ozone of 20 ppb (indoor-to-outdoor ratio of 0.25). However, in the urban corridor of the Denver Metropolitan Area/North Front Range area from Denver to Fort Collins in Colorado, outdoor summertime O₃ concentrations are much higher, routinely exceeding the EPA 8-h average standard of 70 ppb. , Furthermore, these 8-h averages fail to capture the hourly variability in which O₃ surpasses 70 ppb in the afternoon on days that do not qualify as ozone exceedance days. Thus, in regions with elevated outdoor O₃, indoor O₃ can easily reach mixing ratios of tens of ppb, providing substantial oxidative reagents to form ultrafine particles and challenging the generalization that NPF and growth are limited by the availability of O₃.

Regions with high outdoor O₃ present ideal oxidative conditions for indoor new particle formation if formation rates outpace removal by surface uptake and exfiltration. While ventilation schemes differ across commercial and residential buildings, we used a genuine office space on the Colorado State University campus to investigate the particle growth rates arising from the addition of limonene standards or peeling whole citrus fruit at unperturbed ozone concentrations. We examine how the combination of O₃ and VOCs leads to observed ultrafine particle growth rates, their contribution to aerosol mass indoors, and implications for ozone exceedance regions and ozone-producing cleaning technologies.

2. Materials and Methods

We conducted experiments in an authentic indoor office test space during summer 2024 (June–August) and winter 2025 (January–February). Here, we discuss experiments during which VOCs were added to the office (α-pinene, n = 2; limonene, n = 23; and citrus fruit, n = 3). Before and after the addition experiments, we periodically added O₃ to temporarily increase oxidative capacity and carbon dioxide (CO₂) to calculate air change rates. VOC additions were done after O₃ and CO₂ returned to their ambient concentrations. Other experiments associated with this project are the subjects of separate manuscripts and not discussed here. A brief description of the office space, instruments, and experimental methods is provided below with additional information in Text S1.

2.1. Indoor Office Test Space

Experiments were performed in an authentic office space in the Chemistry Department at Colorado State University, Fort Collins, CO, with instruments located in an adjacent room. The shared wall had a plexiglass window through which inlets were connected for sampling and addition requirements (Figure S1). This room had been used as an administrative office for the previous four decades and had not been recently renovated. The office was unoccupied and unfurnished during all of the experiments described herein unless specifically noted.

Characterization of the room: Dimensions of the room and built-in wooden shelving were measured with a measuring tape. The room volume was 33.5 m³, with a geometric surface area-to-volume ratio (SA/V) of 2.47 m^–1. Indoor surfaces included carpet flooring (10.8 m²), painted drywall (25.4 m²), wooden shelving (36.4 m²), and ceiling tiles (10.4 m²). Since geometric surface area does not account for surface roughness or porosity, we acknowledge that the SA/V metric is a lower limit. Air change rates were determined by the exponential decay of a pulse of CO₂ added to the room (described below). During most of the experiments (n = 20), ventilation was reduced by blocking both the supply and return vents to modulate the ACR to 1.7 ± 0.2 h^–1. The ventilation rate is 1.5 L s^–1 m^–2, which meets the ASHRAE Standard 62.1-2002 for an office space (0.3 L s^–1 m^–2). During these low-ventilation experiments, the average relative humidity at the start of these experiments was 25 ± 5% and the temperature was 27.6 ± 0.4 °C. For the high-ventilation experiments (n = 6), the supply and return vents were unblocked so that the average ACR was 10.1 ± 0.4 h^–1, with relative humidity and temperature of 45 ± 1% and 21.6 ± 0.2 °C. This study focuses on low-ventilation experiments.

2.2. Instrumentation

The office space shared a wall with a second room that contained most of the instruments. The shared wall had a 50 cm × 30 cm rectangular cutout into which a plexiglass window sealed by a rubber gasket was installed. Swagelok bulkhead fitting ports, sealed by gaskets, were mounted in the plexiglass window to act as feedthroughs for instrument sampling or VOC/ozone/CO₂ additions.

We measured particle size distributions using two scanning mobility particle sizers (one nano-SMPS for 5–65 nm particles and one wide-range SMPS for 20–850 nm particles; TSI, Inc.) and particle chemical composition using a high-resolution time-of-flight aerosol mass spectrometer (AMS; Aerodyne Research, Inc.). The nano-SMPS was operated in the test space with no inlet. The wide-range SMPS and AMS shared an unheated inlet from the experiment room to the separate instrument room; a valve switching system allowed the instruments to periodically sample filtered air as well as air in the supply vent.

Monoterpenes and their oxidation products were measured by two chemical ionization time-of-flight mass spectrometers: one with NH₄ ⁺ reagent ion (NH₄ ⁺-CIMS; Tofwerk AG) and another with I^– reagent ion (I^–-CIMS; Tofwerk AG), respectively. For the citrus peeling experiments, we collected thermal desorption tubes for off-line gas chromatography/mass spectrometry (TD-GC/MS) analysis to quantify the relative contribution of limonene to the total monoterpene signal because the NH₄ ⁺-CIMS alone cannot resolve isomers.

We measured trace gases in the room and supply air and used publicly available monitoring data as an outdoor reference. Ozone was measured with a dual-beam UV ozone monitor (2B Tech, model 205) in the experiment room and in the supply. Nitrogen oxides (NO, NO _x ) were measured via chemiluminescence (Thermo Fisher 42i-TL NO–NO₂–NO _x analyzer, calibrated by 2B Tech Calibration Source Model 714). CO₂ was measured via infrared absorption (LICOR LI-840 CO2/H2O gas analyzer). Hourly outside monitor data for PM_2.5, PM₁₀, and ozone were accessed from the Colorado Department of Public Health network.

2.3. Experimental Additions

Three types of additions were performed: (1) “neat” additions of VOCs including limonene, α-pinene, or δ-3-carene; (2) additions of ozone or CO₂; and (3) peeling or cutting of citrus fruit. We describe addition methods below with details on safety considerations in Text S1.

Most experiments presented herein include a series of automated O₃ and CO₂ additions and manual VOC additions, typically following a schedule (in local time) of O₃ addition at 3 AM, CO₂ at 5 AM, VOC at 9 AM, O₃ at 12 PM, and CO₂ at 6 PM.

Addition methods:

• (1) Neat VOC addition via syringe pump:

  Limonene (95 or 190 μL, Sigma-Aldrich, analytical grade ≥99%, CAS 5989-27-5), α-pinene (Millapore Sigma, analytical standard grade ≥99%, CAS 7785-26-4), or δ-3-carene (Sigma-Aldrich, analytical standard grade ≥98.5%, CAS 498-15-7) was loaded into a syringe (Hamilton, 250 μL). Zero air was flowed through 1/4” PFA tubing (1 slpm, ultra-zero grade, AirGas) regulated by a mass flow controller (MKS 1179C; power supply MKS 247D). The VOC (190 μL) was injected via a syringe pump (Chemyx, Inc. Fusion 200) over 30 s (380 μL/min) into a stainless steel heated tee (Swagelok Inc.) heated to 60 °C with heat tape and fitted with a Viton septum. The flow of zero air continued for an additional 30 min before closing valves. We calculate that the maximum limonene concentration during the addition is 500 or 1000 ppb in the absence of ventilation; comparison to household activities is discussed later.

  For some additions, we used isotopically labeled limonene: (4R)-1-methyl-4-(1-methylethenyl-2-¹³ C-2,2,-d ₂-cyclohexane) (TLC Pharmaceutical Standards, 40% n-hexane). Further investigation of the isotopically labeled experiments is the subject of separate papers, and here, we treat all experiments alike, regardless of isotopic labeling.

• (2) O₃and CO₂additions via compressed gas cylinders and ozone generator:

  Ozone was added to the space by passing O₂ (AirGas, Ultra High Purity, CAS 7782-44-7) through an ozone generator (AZCO Industries Ltd., Model HTU-500). A mass flow controller (MKS Inc., 1179A) controlled O₂ flow to 0.70 vlpm. When we used isotopically labeled O₃, an additional 0.140 vlpm flow of ¹⁸O–O₂ (Sigma-Aldrich; CAS 32767-18-3) controlled with a critical orifice (O’Keefe Controls) at 7 psi was introduced to the O₂ line upstream of the ozone generator. The ¹⁸O–O₂ flow was manually opened and closed during a computer-automated O₃ addition. The isotopically labeled experiments are the subject of separate manuscripts.

  The O₃ additions resulted in an average concentration of 61 ± 7 ppb increase in O₃ on top of ambient, preaddition levels, resulting in an average maximum indoor O₃ concentration of ∼80 ppb. These elevated levels are comparable to high-ventilation conditions (therefore high indoor-to-outdoor ratio) or window-opening on a day that exceeds the National Ambient Air Quality Standard of 70 ppb. A leak in the ozone sampling line between 18 June and 9 August 2024 led to an overestimation of ambient ozone and an underestimation of added ozone; see Text S1: Trace Gases for additional information.

  CO₂ (AirGas, Bone Dry grade, CAS 124-38-9) was added into the test office at 15 psi (delivery pressure) through a HEPA filter (Pall Life Sciences, 12144) using a computer-controlled solenoid valve (ASCO Red-Hat, 8262H036) that opened the valves for 90 s, then closed for 30 s; this cycle was repeated three to four times to prevent formation of dry ice and interferences from organofluorine from the tubing walls. The typical maximum CO₂ concentration after the addition was ∼2500 ppm (6 times the background concentration of ∼420 ppm under no perturbations).

• (3) Complex VOC mixture addition via citrus peeling:

  Two people entered the test office and peeled one citrus fruit (lemon or Valencia orange) into a paper or plastic bag to minimize the release of droplets generated by the peeling process. In these experiments, we simultaneously collected air onto Tenax-TA tubes for subsequent off-line analysis by thermal desorption and gas chromatography/mass spectrometry (TD-GC/MS). Details of the speciated monoterpene sampling are in Text S1.

3. Results and Discussion

3.1. O₃ and Limonene Drive Indoor New Particle Formation and Growth

A representative time series of a limonene addition experiment is shown in Figure . The 3 AM ozone addition into the empty, unoccupied room, with limonene below the detection limit, did not induce changes in aerosol number, mass concentration, or size distribution. However, the limonene addition at 9 AM at an ambient indoor ozone concentration of 10.0 ppb rapidly generated particles at the smallest detectable sizes on the wide-range SMPS (∼20 nm) that continued to grow for 2+ hours. Results from the nano-SMPS (5–65 nm) are shown in Figure S2, but we use size distribution from the wide-range SMPS for subsequent analysis presented here. The secondary aerosol growth mode with diameters between 20 and 125 nm is distinct from the existing particle population in both number and mass size distributions (Figure c,e) and contributes several μg m^–3 of organic aerosol, as measured by AMS (Figure d). The observation of particles with diameters less than 100 nm and subsequent growth from 5 nm to >100 nm after the addition of an oxidant or VOC is evidence for nucleation and growth. Upon limonene addition, O₃ in the room decreased by 2 ppb, but quickly recovered. The ozone addition several hours later at 12:30 PM induced particle growth in two size modes: (1) appearance and growth of small particles (diameter <125 nm) at a much lower concentration and (2) growth of the previously generated aerosol population with mean diameter ∼150 nm. At the time of limonene addition, ozone measured in the supply air was ∼40 ppb, and it increased to 90 ppb later in the afternoon (Figure S3).

1.

Trace gas mixing ratios (a), particle number (b) and mass (d) concentrations, and particle number (c) and mass (e) size distributions observed in the office space during a representative summertime limonene and ozone addition experiment. ∼70 ppb O₃ was added at 3 AM and 12:30 PM, and 190 μL of neat limonene at 9 AM. Limonene was detected as C₁₀H₁₆NH₄ ⁺. SMPS-derived mass size distributions assume a density of 1.0 g cm^–3.

We designed our experiments to generate limonene levels that are reasonable for everyday household activities, including direct emission from citrus fruit. Household products (e.g., degreasers, cleaning solutions, candles) have been observed to elevate several-day average limonene concentrations by 10s to 100s of ppb, with acute increases on the order of 1000 ppb. We targeted maximum limonene concentrations of 500 and 1000 ppb and measured maximum limonene ranged from 450 to 1500 ppb, so our experiments are well within the range of observed limonene levels during everyday activities.

Particle formation and growth in the outdoor atmosphere is often driven by acid–base chemistry (sulfuric acid + ammonium/amines), with halogen chemistry (e.g., iodic acid) in remote northern latitudes and coastal regions, or oxidation products of marine sulfur (sulfuric acid, methanesulfonic acid) over remote oceans, or secondary generation of highly oxygenated, low-volatility organic molecules that can induce pure organic nucleation. Here, particles that form and grow larger than ∼100 nm diameter to be observed by the AMS are entirely organic with no contributions from inorganics (pSO₄, pNH₄, pNO₃, pCl), which is consistent with pure organic particle formation and growth. Ozonolysis of limonene at modest limonene levels (several ppb) can increase OH levels to 10⁶ molecules cm^–3, comparable with outside mixing ratios, and OH-initiated autoxidation of limonene , can produce highly oxidized multifunctional compounds (HOMs) that readily condense to the particle phase. While NO _x , in particular NO, remains low, we observe oxidized compounds including organic nitrates in the gas phase (Figure S4) and hypothesize that these processes could drive the formation and growth of the ultrafine particle mode. However, although we see gas-phase organic nitrates, the lack of particulate nitrate (pNO₃) above the detection limit in our observations suggests limited importance of nitrate radical oxidation to particulate mass; NO₃ can be generated indoors by reaction of NO₂ and O₃ from infiltration during outdoor pollution events or combustion cooking and would generate organic nitrate aerosol that would appear as pNO₃ in the AMS. Additionally, we do not observe any pCl that would indicate particle formation from reactive chlorine oxidation (Cl₂ and HOCl) associated with the use of bleach-based cleaning products; , this is unsurprising as bleach was not used in this office during these experiments or in prior months. The alternate hypotheses presented above regarding inorganic contributions or contributions of organic nitrates are not supported by the chemical composition data provided by the AMS.

The product of limonene and ozone concentrations explains the variability in growth rates (Figure a) and maximum organic aerosol (OA) mass (Figure b) during particle formation and growth events, consistent with second-order kinetics. We fit the geometric mean diameter for particles <125 nm over time (Figure S5) and found that the resulting growth rate and maximum OA mass increase linearly with the product of added limonene and preaddition indoor ozone concentrations (Figure ). We chose 125 nm as the upper limit for the geometric mean because it captured the entire growth period of the small particle mode while avoiding the established accumulation mode (diameters centered around 300 nm). For the neat experiments, the limonene concentration was estimated as an upper bound as if the added limonene instantaneously dispersed throughout the room (500 or 1000 ppb). For the citrus peeling experiments in which oranges and lemons emit other monoterpenes (e.g., sabinene, myrcene, α-pinene, β-pinene), we adjusted the C₁₀H₁₆NH₄ ⁺ signal in NH₄ ⁺-CIMS by the TD-GC/MS-derived ratio of limonene to total monoterpene mass (78% for orange; 37% for lemon; Figure S6). Ozone used in this calculation was a 10 min average prior to introduction of limonene and ranged from 4 to 15 ppb. High Pearson correlation coefficients from least-squares linear fits (r ² = 0.57 and r ² = 0.66 for particle growth rate and maximum OA mass, respectively) indicate that (1) particle growth rate and maximum OA mass can be predicted from limonene and ozone concentration, and (2) particle growth is consistent with product formation from a second-order reaction or from later-generation reactions that are rate-limited by the initial O₃ + limonene reaction. The NO mixing ratios were near the detection limit (Figure S7), and we did not experience a wide range of existing PM surface area to test sensitivities to those variables. Here, we used the upper limit of limonene mixing ratio as estimated from the volume of limonene added into the space to remove complications of loss to ventilation, partitioning to surfaces, and chemical reactions. However, the maximum limonene measured by NH₄ ⁺-CIMS shows fair agreement with the targeted limonene mixing ratios (Figure S8), and relationships between growth rates and maximum OA mass are comparable (Figure S9). We also note that a higher ACR results in even higher growth rates for a given limonene and ozone concentration pair (Figure S10), but the effect of changes associated with increased ventilation rate (e.g., humidity and temperature, mixing dynamics within the room, O₃ surface reactions) on particle formation and growth and air change in this data set are difficult to track. Despite some variability in indoor growth rates at one ACR, relationships between growth rate and solely limonene or ozone concentration are poor (Figure S11), indicating that both reagents are needed to drive the growth rate and total OA mass.

2.

(a) Growth rate for particles with diameters <125 nm from wide-range SMPS and (b) maximum OA concentration from AMS versus the product of estimated limonene mixing ratio and O₃ mixing ratio prior to VOC addition for June 16–August 9 experiments with ACR = 1.7 h^–1 shown. Error bars in growth rate reflect 1σ uncertainties for the linear fit. Limonene concentrations from citrus peeling and zesting were calculated by multiplying the NH₄ ⁺-CIMS monoterpene signal by the TD-GC/MS limonene fraction of monoterpenes (see SI). The size of the marker corresponds to the relative limonene mixing ratio. The variability in growth rate and maximum organic aerosol mass is mostly explained by the variability of the product of limonene and O₃ concentration at a given ACR; Pearson correlation coefficients: r ² = 0.57 and r ² = 0.66, respectively.

The particle growth rates for limonene ozonolysis indoors are very high compared to those observed outdoors and comparable to previous studies of limonene indoors. Outdoor growth rates are ordinarily several nm h^–1, − but here, we see growth rates between 20 and 100 nm h^–1 under ambient ozone conditions and realistic limonene concentrations that are more consistent with previous measurements up to 300 nm h^–1 indoors. Furthermore, the link between citrus, ozone, and particles has been previously noted in indoor observations, − ,,, but some studies lack quantification of particle growth rates. The good agreement between the neat experiments and the limonene component from the citrus peeling experiments suggests that other monoterpenes (e.g., sabinene, myrcene, α-pinene, β-pinene) neither contributed to nor suppressed growth within experiment-to-experiment variability.

In contrast to outdoors, indoor spaces contain surfaces and the ratio of surface area to volume is often used to contextualize the differences between indoor and outdoor atmospheric chemistry in terms of reservoirs for condensable material and reaction sites of oxidants. In our test office, the geometric surface area to volume ratio is 2.47 m^–1, which is typical of indoor environments. The specific surface area, which accounts for surface roughness and pores, would be even greater. In comparison, the typical maximum aerosol surface area concentration in our experiments was 40 μm² cm^–3 or 4 × 10^–5 m^–1. By these measurements, the surface area of indoor space is orders of magnitude greater than the surface area provided by the existing particle populationeven during high aerosol concentration periods. Additionally, the high surface area indoors causes some molecules that would typically be considered volatile outdoors to act like semivolatile species with substantial partitioning to the surface reservoir. Ozone reacts with this surface reservoira reaction that causes substantial loss for indoor ozone, accounting for much of the difference between supply air and room levels. Nevertheless, despite the large condensational sink of VOCs and reactive sink for O₃ provided by indoor surfaces, NPF and growth compete effectively to generate particles that substantially impact both number and mass concentration of indoor spaces. This suggests that limonene + O₃ is either producing such a large amount of HOMs that NPF is competitive with partitioning to indoor surfaces and/or the time scale of HOM transport to surfaces is slower than the time scale for nucleation and condensation to sub-100 nm particles.

3.2. Outdoor Air Supplies Oxidants for Indoor Organic-Derived NPF and Growth

Indoor air quality is inextricably linked to outdoor air quality due to infiltration and ventilation of outdoor air that may contain pollutants. Mechanically ventilated buildings that use limited recirculated airincluding the building housing our test office spacetransport outdoor air to the indoor space. Figure compares the composition of the supply air to nearby Colorado Department of Public Health and Environment (CDPHE) monitors, showing how natural variability in outdoor ozone and particle concentration impacts the supply air. Ozone in the supply is directly related to outdoor ozone concentrations with a ratio near 1 (Figure S12), consistent with minimal impact from ventilation ducts on ozone concentrations. As expected, indoor ozone follows a strong diel pattern with maximum ozone concentrations in the afternoon; indeed, this daily outdoor variability generated a dynamic range of ozone concentrations for our experiments. The median and interquartile range of outdoor ozone for summer experiments (June 16–August 9) was 48 and 24 ppb, contrasted with 30 and 12 ppb for the winter experiments (February 12–March 12). Even in the winter, Colorado’s Front Range experiences O₃ concentrations higher than the 20 ppb average reported from a global survey of indoor and outdoor O₃ concentrations. The experiments shown here, where elevated outdoor ozone is transported indoors and can react on surfaces, particles, and gases, are a counterpoint to indoor studies performed in regions with lower O₃ concentrations.

3.

Time series of (a) ozone mixing ratio in the supply air (green) and outdoor air (black), (b) particle mass (d < 850 nm) of the supply air (blue) and outdoor PM_2.5 mass (black), and (c) number concentration of the supply air (red) and room air (gray). Outdoor measurements were taken from nearby CDPHE monitoring sites. Black markers on the bottom axis indicate limonene addition experiment days.

While indoor background particle mass is driven by ventilation of outdoor air, particle formation events do not correspond to increases in outdoor PM_2.5 concentrations (Figure ). We note that PM_2.5 shows a weak diel pattern with overall elevated concentrations in late July and early August due to wildfire smoke. Fort Collins experienced wildfire smoke with long-range transport from western Canada as well as smoke from the local Alexander Mountain fire that began on July 29 approximately 13 miles SW of the building in which we conducted the experiments. We paused experiments during the most intense influence from smoke to avoid confounding factors introduced by both elevated PM and VOC concentrations from the fire. The air supply shows that ozone levels are driven by outdoor air pollution and confirms that the number and mass concentration and appearance of <125 nm particles (spikes in Figure c) were not due to transport of particles from outdoors.

3.3. Particle Growth Rates Are Not Sustained for Certain Monoterpenes and under Winter Conditions

We tested three other monoterpenes (α-pinene, myrcene, δ-3-carene) at similar summertime ambient ozone levels but found that limonene was unique in its ability to generate ultrafine particles indoors with measurable growth rates. The aim of this comparison was to examine ultrafine particle formation potential for different monoterpenes that may also be used as household cleaning products or in personal care products. While the citrus peeling was intended to replicate a real activity, the addition of monoterpenes at the same concentrations allowed us to explore the chemistry of different monoterpenes. Rate constants for the reaction of O₃ with α-pinene, myrcene, and δ-3-carene are 8.7, 47.0, and 3.7 × 10^–17 cm³ molecules^–1 s^–1, respectively, and SOA yields are 41, 11, and 54%, respectively. While α-pinene produced small particles, the particles did not grow beyond the initial formation (a lack of ‘banana’ structure in Figures and S13) despite previously observed SOA formation from ozonolysis of α-pinene in seeded and unseeded chamber studies. , Additionally, δ-3-carene produced a negligible number of particles (<2000 cm^–3) under our experimental conditions, despite showing propensity for NPF. Myrcene produced no particles despite having known O₃-initiated SOA formation. Because of the lack of monotonic increase in diameter in the particle size distribution evolution, particle growth rates could not be determined for experiments with α-pinene, myrcene, and δ-3-carene, despite indoor O₃ levels comparable to those observed during summertime limonene-induced ultrafine particle production. Limonene may be uniquely effective in sustained ultrafine particle growth due to its fast O₃ reactivity, high OH production, which may sustain multigenerational chemistry, efficient HOM production to form condensable products, and possibly enhancements due to surface chemistry.

4.

Trace gas mixing ratios (a), particle number (b), and mass (d) concentrations, and particle number (c) and mass (e) size distributions observed in the office space during a representative summertime α-pinene addition experiment. 190 μL of neat α-pinene was added at 3 PM. The added α-pinene was detected as C₁₀H₁₆–NH₄ ⁺ (b). SMPS-derived mass size distributions assume a density of 1 g cm^–3. While an independent mode of particles is generated, the growth is less organized, and a growth rate cannot be determined.

We also performed limonene additions in the wintertime under lower ozone conditions and after many VOC additions, and we observed only a moderate number of particles upon addition of limonene (Figure S14). Particle production completely stopped as O₃ rapidly dropped to levels below the detection limit. When O₃ was provided by artificial addition, particles were rapidly and abundantly generated. This wintertime absence of O₃-induced production and growth of ultrafine particles contrasts with the summertime experiments in which ozone decreased by a few parts per billion immediately following VOC addition but recovered quickly.

4. Implications

In regions where outdoor ozone routinely exceeds the EPA 8-h average standard of 70 ppb and indoor reactivity is low, indoor O₃ concentrations can be high enough that indoor particle formation requires additional VOC. When sufficient VOC is present, particle formation and growth meaningfully contribute to overall aerosol mass. The threshold for sustained growth will likely depend on building parameters including ventilation and ozone reactivity of the space. Our major findings are as follows:

• (1)At sufficiently low indoor O₃ concentrations, determined by the transport of outdoor ozone and ozone reactivity, NPF and growth are limited by the availability of O₃. Adding VOCs (namely, limonene) quickly depletes indoor O₃, which limits the total ultrafine particle production potential. Subsequent addition of artificially generated O₃ initiates particle production and growth with growth rates greater than 100 nm h^–1 (Figure S15).
• (2)At sufficiently high indoor O₃ concentrations, determined by transport of outdoor ozone and ozone reactivity, indoor NPF and growth are limited by the availability of VOCs. O₃ is continually supplied by ventilation at a rate greater than O₃ losses indoors, and adding VOCs produces ultrafine particles with growth rates of between 20 and 100 nm h^–1 to sizes that contribute several μg m^–3 of organic aerosol. Adding O₃ in the presence of the recently formed particles induces additional formation and growth events with growth rates of 30–45 nm h^–1 as well as growth of the existing particle population. For a given VOC addition quantity, particle growth rate and maximum organic aerosol scale with indoor O₃ concentration.
• (3)Ultrafine particle production in an indoor space can be reduced by reducing VOC or O₃ concentrations.

The observation that large O₃ additions did not generate particles or contribute to particle mass in all cases attenuates the concern that adding any amount of O₃ to a space will generate ultrafine particles; the O₃ and particle generation potential of air-cleaning technologies like germicidal/far-UV (λ = 222 nm) or simple O₃ generators is an emerging question. ,− However, the understanding of the cumulative effect of germicidal or far-UV technologies on particle formation, like our understanding of indoor particle formation in general, is limited by the complexity of the indoor environment. In a building that routinely experiences a near-zero O₃ background, indicating that NPF and growth are limited by the availability of O₃, O₃ generated by far-UV cleaning technologies may react with building material emissions to produce ultrafine particles; however, after the most reactive molecules have outgassed from the new materials, the ultrafine particle production potential will decrease to minimal levels, suggesting NPF formation and growth then becomes limited by availability of VOCs. Our study indicates that the identity of VOCs present plays an important role in the potential for ultrafine particle production, and that limonene appears to be a particularly potent VOC for generating ultrafine particles at ambient ozone mixing ratios.

When deploying ozone-generating technologies to combat biological hazards, understanding whether ultrafine particle production would be limited by O₃ or VOC mixing ratios could inform users and help reduce the particle inhalation risk. In an O₃-limited regime, the O₃ produced by UV technologies would increase the oxidative potential of a building so that particle production may be induced. However, in a VOC-limited regime, such as our summertime conditions, additional O₃ may not induce particle production unless specific VOCs are added or added O₃ would only slightly increase the particle formation potential beyond what will naturally occur with ventilated O₃. A VOC-limited regime could be encouraged by avoiding monoterpene-containing cleaning products concurrently with ozone-generating technologies.

Our results indicate that buildings can shift between O₃- and VOC-limited chemical regimes on daily and seasonal time scales, with implications for air pollution control. In areas of ozone nonattainment across the United States, including rural and urban California, and urban centers in the Midwest, East Coast, and Mountain West (Figure S16), elevated outdoor ozone increases the potential for ultrafine particles that contribute aerosol mass in an indoor space. Among indoor pollutants, ultrafine particles may pose cardiovascular and respiratory health concerns ,, and fine particle mass (often associated with criteria air pollutant PM_2.5) are known for cardiovascular and respiratory health including premature deaths. Furthermore, health effects with fine particle mass may even be amplified in the presence of ozone. We have shown that not only does particle formation indoors initiated by limonene ozonolysis contribute to the number of ultrafine particles, it also contributes to aerosol massthe metric by which outdoor air quality has been clearly linked to health effects. This finding supplements recent work showing that indoor reactions of ozone (i.e., the difference between indoor and outdoor ozone) correlate more strongly to cardiorespiratory metrics than either level individually. ,− Reducing the use of limonene-containing products during ozone exceedance events or reducing O₃ through effective O₃ scrubbing in air handling systems , will reduce indoor NPF potential in high ozone regions. Because outdoor ozone is seasonal, the changes in aerosol mass and number concentrations due to ultrafine particle production will also be seasonal. Continued reduction of outdoor ozone pollution will reduce the primary risk to ozone exposure as well as the secondary risk to ultrafine and fine particle exposure from particles generated indoors.

The data used in this study are openly available in Dryad at DOI: 10.5061/dryad.x95x69q21.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.6c07592.

• Additional details about instrumentation, geometric mean diameter calculations, and analyses (PDF)

L.A.G.: Conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, software, visualization, writingoriginal draft, writingreview and editing; A.G.S., J.H.C., A.B., Mj.R., R.K.R., J.L., and R.G.: data curation, investigation, methodology, writingreview and editing; M.D.W. and D.K.F.: conceptualization, funding acquisition, supervision, writingreview and editing.

The authors declare no competing financial interest.