Indoor ozone/human chemistry and ventilation strategies

Abstract This study aimed to better understand and quantify the influence of ventilation strategies on occupant‐related indoor air chemistry. The oxidation of human skin oil constituents was studied in a continuously ventilated climate chamber at two air exchange rates (1 h−1 and 3 h−1) and two initial ozone mixing ratios (30 and 60 ppb). Additional measurements were performed to investigate the effect of intermittent ventilation (“off” followed by “on”). Soiled t‐shirts were used to simulate the presence of occupants. A time‐of‐flight‐chemical ionization mass spectrometer (ToF‐CIMS) in positive mode using protonated water clusters was used to measure the oxygenated reaction products geranyl acetone, 6‐methyl‐5‐hepten‐2‐one (6‐MHO) and 4‐oxopentanal (4‐OPA). The measurement data were used in a series of mass balance models accounting for formation and removal processes. Reactions of ozone with squalene occurring on the surface of the t‐shirts are mass transport limited; ventilation rate has only a small effect on this surface chemistry. Ozone‐squalene reactions on the t‐shirts produced gas‐phase geranyl acetone, which was subsequently removed almost equally by ventilation and further reaction with ozone. About 70% of gas‐phase 6‐MHO was produced in surface reactions on the t‐shirts, the remainder in secondary gas‐phase reactions of ozone with geranyl acetone. 6‐MHO was primarily removed by ventilation, while further reaction with ozone was responsible for about a third of its removal. 4‐OPA was formed primarily on the surfaces of the shirts (~60%); gas‐phase reactions of ozone with geranyl acetone and 6‐MHO accounted for ~30% and ~10%, respectively. 4‐OPA was removed entirely by ventilation. The results from the intermittent ventilation scenarios showed delayed formation of the reaction products and lower product concentrations compared to continuous ventilation.


| INTRODUC TI ON
The importance of indoor air chemistry has been increasingly recognized over the past decades. [1][2][3][4] Building materials, furnishings and carpeting, cleaning products, personal care products, human activities, as well as outdoor air have been considered major sources of indoor air pollutants. During the past decade, there has been growing recognition that humans themselves considerably contribute to indoor air pollution and impact indoor air chemistry. 5,6 Squalene constitutes approximately 12% (by weight) of human skin lipids; other constituents include fatty acids, glycerides, wax esters, ceramides, and cholesterol esters. 7 Having six carbon-carbon double bonds in its structure, squalene serves as a natural antioxidant to protect our skin from atmospheric oxidants such as ozone (O 3 ), which reacts with the unsaturated sites of the molecule. Squalene is responsible for about half of all unsaturations available in skin lipids, 8 and given the slight differences in reaction probabilities between unsaturations in squalene and fatty acids and their esters, it is responsible for approximately half the ozone uptake.
Ozone is the most abundant oxidant in the indoor environment.
Its major source indoors is outdoor air, entering the buildings via ventilation and infiltration. The indoor-to-outdoor (I/O) concentration ratio of ozone is usually between 0.1 and 0.7, depending on the building air exchange rate (AER) and the properties of the exposed surfaces. 9 Ozonolysis of surface-bound squalene was first investigated on Mediterranean vegetation. 10 The complex reaction of squalene with O 3 produces a large spectrum of oxygenated volatile species, including acetone, geranyl acetone (GA), 6-methyl-5-hepten-2-one In subsequent studies, 4-OPA was shown to be a sensory irritant and capable of inducing inflammatory cytokine expression in vitro. [14][15][16] The derived human reference value for sensory irritation by 6-MHO is 0.3 ppm, and the reference values for airflow limitation are 0.03 and 0.5 ppm for 4-OPA and 6-MHO, respectively. 17 Other volatile secondary products of this reaction include 4methyl-8-oxo-4-nonenal (4-MON), 4-methyl-4-octene-1,8-dial (4-MOD), 1,4-butanedial and carboxylic acids such as 4-oxo-butanoic acid and 4-oxopentanoic (levulinic) acid. In addition to the volatile products, the ozonolysis of squalene also generates a number of long-chained polyunsaturated aldehydes and carboxylic acids such as C27-pentaenal, C22-tetraenal, C17-trienal, C27-pentaenoic acid, C22-tetraenoic acid, C17-trienoic acid, that remain bound to the surface. Surface reactions remove ozone from indoor environments more effectively than gas-phase reactions. 18 The human surface has been shown to be very effective at removing ozone. 12,13,19 The oxygenated and highly oxygenated products of squalene ozonolysis are difficult to measure in indoor air by conventional techniques such as sampling on adsorbent tubes followed by thermal desorption into a gas chromatograph with, for example, mass selective detector (TD/GC-MS). These compounds are often thermally unstable and undergo decomposition in the thermal desorption stage. Even if some of these "stealth compounds" may be suitable for the classical TD/GC-MS analysis, the time resolution is in minutes to hours due to sampling on the adsorbents. Development of monitoring instruments with time resolution in minutes to seconds, based on mass spectrometry, has enabled the determination and concentration-time profiling of a range of volatile organic compounds in air. Proton transfer reaction mass spectrometer (PTR-MS) was used, for example, in the above-mentioned simulated aircraft cabin 11,12 and office experiments. 13 Other studies looked at ozone removal and product formation on clothes, 20 human hair 8 , pieces of clothes soaked with methanol extracts of human skin lipids, 21 on the heterogeneous mechanism of squalene/ozone reaction, 22 and on modeling of the processes with relevance to indoor environments. 18,19 Building ventilation is used to control the thermal environment and the concentrations of indoor air pollutants.
Experimental and modeling studies have demonstrated the impact of ventilation on indoor chemistry and product formation. 18,23,24 The air exchange rate (AER) has only a small impact on chemical reactions that occur on indoor surfaces. However, for indoor reactions in the gas-phase, AER determines the time available for reactions to occur. While higher AER shortens the time available for reactions, it increases indoor ozone concentrations relative to their outdoor level. That is, low AER transports less ozone from outdoors to indoors, but allows more time for chemistry. 1 Energy saving measures have led to the implementation of alternative ventilation

Practical Implications
• In occupied buildings, ventilation strategies do not offer a simple approach for controlling products that may arise from ozone/human chemistry. This reflects the fact that such products are generated from both surface reactions and gas-phase reactions.
• Surface reactions are relatively insensitive to ventilation rates while the gas-phase reactions compete with ventilation.
• Exposure to noxious products of ozone/human chemistry can be reduced by decreasing ventilation during periods with high outdoor ozone levels. When buildings are unoccupied (eg, overnight or on weekends), decreasing or stopping ventilation reduces indoor ozone chemistry and surface accumulation of oxidation products.
• Results from this study can serve as inputs to predictive models of ozone/human interaction in indoor environments.
strategies, such as intermittent ventilation, where the ventilation system operates with minimal or no airflow overnight. Different ventilation strategies are anticipated to affect chemical transformations in indoor air differently.
The objective of this study was to investigate the effect of continuous and intermittent ventilation strategies on occupant-related ozone chemistry. Dynamic and steady-state mixing ratios of oxygenated products of skin oil ozonolysis were measured. To gain a better understanding of simultaneous formation and removal processes, product yields were derived by fitting a series of mass balance equations to the data.

| Climate chamber
The experiments were performed in a 30 m 3 stainless steel climate chamber. In order to remove reactive compounds from the surfaces, the chamber was treated with 375 ppb of ozone during a period of 72 hours prior to the first experiment. During the experiments, the temperature in the chamber was maintained at 23°C. The relative humidity (RH) was not controlled, but was relatively constant around 35%. The air exchange rate (AER) was set to 1 or 3 h −1 (Table 1). It was measured regularly during the experimental month with an Innova 1312 Photoacoustic Multi-gas Monitor (LumaSense Technologies A/S, Ballerup, Denmark) using Freon ® 134a as tracer gas. The chamber's HVAC system contained a particle filter and an activated carbon filter. The background ozone mixing ratio in the chamber was <2 ppb.
Ozone was generated in the HVAC system downstream of the activated carbon filter by delivering air through a Jelight 600 UV ozone generator (Jelight Company Inc). The ozone generation rate was controlled by changing the fraction of the UV lamp that was exposed. The amount of air drawn through the ozone generator (0.5 L min −1 ) was controlled with a flow meter. Ozone was continuously monitored with an Environics Series 300 UV absorption ozone monitor with a time resolution of 1 minute and accuracy of ±2 ppb (Environics Inc). Temperature and relative humidity were continuously monitored with 1-minute time resolution using Wöhler CDL 210 sensors (Wöhler Technik GmbH).

| Experimental conditions
Human occupancy was simulated using recently worn t-shirts. Plain, white, cotton t-shirts (exposed surface area of one t-shirt is approximately 0.85 m 2 ) were washed at 60°C, tumble dried, and stored in zip-locked plastic bags before the experiments. Before a measurement, the same four persons were asked to wear fresh t-shirts overnight and come to the laboratory in them without showering. 11 The four t-shirts were then turned inside out and stretched over the backs of wire mesh chairs in the chamber. Two mixing fans were operating in the chamber.
Two sets of experiments were performed ( Table 1). The first set (Conditions 1-7) examined the effect of two air exchange rates   TA B L E 1 Experimental matrix, for all Conditions except 2, t-shirts placed in chamber at t = 0 h 9. The following morning, ozone generation was started and the AER was increased to 1 h −1 either at the time when the shirts were placed in the chamber (t = 0 hour, Condition 8) or 2 hours before (t = −2 hour, Condition 9).

| ToF-CIMS measurement
The Aerodyne ToF-CIMS was utilized for the measurement of gasphase species produced in the chamber. The work described in this paper used the ToF-CIMS in positive mode with protonated water clusters as reagent ions. A detailed description of the instrument can be found elsewhere. 25 The sample air was brought into the ion molecule region (IMR) at Data were acquired at 1s time resolution and averaged to 1 minute for data analysis using the software Tofware 3.1.1 (Aerodyne Inc).
Post-campaign mass calibration of the spectra was conducted using four constant peaks on the spectra; ( , enabling an accuracy of 5 ppm for peak identification. The average spectral mass resolution (m/Δm) was 3500.
Permeation sources held at constant temperature were utilized for calibration. Compound standards for acetone, 6-MHO, and geranyl acetone (GA) were purchased from Sigma-Aldrich and 5 mL was placed in a small vial with an orifice in the lid. An N 2 flow of 2.5 SLM was passed through a large glass vial holding the permeation source and then partial flow was sampled by the TOF-CIMS with the excess being evacuated to an exhaust. The mixing ratio in the flow was calculated using a mass loss approach (see Supporting Information for a more detailed description). The sensitivities for acetone, 6-MHO, and geranyl acetone were calculated to be 12.77, 4069, and 2785 ion counts per ppb, respectively. We quantified 4-OPA using the sensitivity factor of 6-MHO, ensuring that humidity normalization and total ion count signal were constant for each experiment.

| Experimental procedure
Each experiment was initiated by placing four t-shirts in the chamber. The pressurized chamber door was open for less than 3 seconds, a person entered and the door was closed. After placing the t-shirts on the chairs, the chamber door was open again as the person was leaving. Opening the chamber door may have affected the condition in the chamber. However, no substantial impact on the ozone mixing ratio was observed; ozone mixing ratio decays followed from the beginning the pattern expected after introducing the soiled t-shirts (see Figure 2). The experiment continued for 3 hours with measurements of ozone and the organic compounds with the ToF-CIMS. This time interval was judged to be sufficient for the system to reach at least 95% steady-state mixing ratios of ozone and primary reaction products at both target air exchange rates. During this phase, concentration-time profiles of ozone and the primary and secondary squalene/ozone reaction products were monitored with 1-minute time resolution. After 90 minutes, desorption of particles collected during the preceding 30 minutes of gas-phase measurement was performed using the Filter Inlet for Gases and AEROsols (FIGAERO). 26 We plan to present results from the particle measurements in a future article. After the FIGAERO desorption process, it took approximately 10 minutes for the measured mixing ratios of gas-phase species to reach those values immediately prior to particle desorption. After the experiment, the t-shirts were removed from the chamber and the chamber was flushed with outdoor air (20 h −1 for 20 minutes).
The error for k sur was estimated to be 8%, and we used this value to calculate the uncertainty of k shirt .
All parameters in the equations are known, measured in these experiments or estimated based on the measurements, with the exception of

| Geranyl acetone (GA)
Geranyl acetone is primarily produced through surface reaction of ozone with squalene on the t-shirts when ozone adds to double where Yield GA (unitless) is the yield of geranyl acetone from the primary and the secondary reactions of ozone on the shirts.
Visualization of the formation and loss of geranyl acetone, 6-MHO, and 4-OPA (f -branching ratio; 1°, 2°, 3° -primary, secondary, tertiary reaction; λ -air exchange rate) 2.5.3 | 6-methyl-5-hepten-2-one (6-MHO) 6-MHO is formed as a primary product of ozone/squalene reactions on the surface of the t-shirts when ozone adds to the double bonds 2 or 2′ in the squalene molecule ( Figure 1). It can also be formed through secondary reactions both on the t-shirts and in the gasphase as ozone reacts with geranyl acetone.
Equation (5) can be rewritten for steady-state condition where Yield 6MHO (unitless) is the yield of 6-MHO from primary (squalene) and secondary (C27-pentaenal, C22-tetraenal, C17trienal, C27-pentaenoic acid, C22-tetraenoic acid, C17-trienoic acid, and geranyl acetone) reactions of ozone on the t-shirts, and f 6MHO is the fraction of the gas-phase reaction of ozone with geranyl acetone that produces 6-MHO (branching ratio). The branching ratios for 6-MHO and 4-OPA from the gas-phase reaction of ozone and GA are unknown; we have set the value of f 6MHO to 0.3 in accordance with the findings described by Grosjean and Grosjean (1997). 31 The sum of formation yields of primary carbonyls is close to one, and the branching ratio favors the primary carbonyl with the more substituted associated biradical.

| 4-oxopentanal (4-OPA)
4-OPA is formed as a secondary product from ozone reacting with C27-pentaenal and C22-tetraenal on shirts, as a secondary product of ozone reacting with geranyl acetone in the gas-phase, and as secondary or tertiary product from ozone reacting with 6-MHO in the gas-phase ( Figure 1). Equation (7) can be rewritten for steady-state condition

| Ozone mixing ratio
The rate at which ozone was removed by the stainless steel chamber surfaces (k chamber , 0.15 h −1 ) was around 10% of the removal by all surfaces when four soiled t-shirts were present (k sur , 1.37 h −1 ). Table 2 lists the measured initial and steady-state mixing ratios of ozone for all Conditions. No significant change in ozone mixing ratio was observed during the measurement period for Condition 2, in which no t-shirts were present in the chamber. The steady-state mixing ratios of ozone were between 12 and 37 ppb when t-shirts were present.
The highest value was observed for Condition 6 with an initial ozone mixing ratio of 54 ppb and an elevated ventilation rate (3 h −1 ). At identical initial ozone mixing ratios, the ozone generation rate is three times higher at AER 3 h −1 compared to AER 1 h −1 . Although low, the measured mixing ratios of the ozone-squalene reaction   Table 3 presents the parameters in the model that addresses steady-state ozone mixing ratios. The measured steady-state ozone mixing ratios fit well with the known ozone generation rates and the identified sinks (source-to-sink ratios close to unity; Table 3). Ozone removal rates at steady-state were dominated by AER (λ) and the removal by reactions on the surfaces of the chamber walls and the t-shirts (k sur ) (Equation (2); Table 3). Air exchange was responsible for ~40% of the ozone removal rate at 1 h −1 and for ~70% at 3 h −1 .
The sum of air exchange rate and ozone removal rate by the surfaces accounted for more than 96% of the loss of ozone in the chamber under all conditions. Gas-phase reactions with GA and 6-MHO had a negligible impact on ozone loss compared to ventilation and surface removal. The time to steady-state for each Condition is consistent with the ozone loss rate being dominated by ventilation and surface removal (λ + k sur ).
The worn t-shirts accounted for approximately 90% of the removal of ozone by all surfaces (k shirt /k sur = 1.22 h −1 /1.37 h −1 = 0.89). The rate of ozone removal to a highly reactive surface is limited primarily by the resistance to mass transport across the boundary layer of air adjacent to the surface. 32 This may be the case for the t-shirts (k shirt ), where the transport-limited ozone flux controls the maximum rate of reactive uptake of ozone with skin lipids. This is supported by the observation that, from the introduction of the shirts into the chamber until the end of the measurements, ozone consumption occurred at a nearly constant rate. We also observed that k shirt varied very little from condition to condition. Given the surface area of the four t-shirts (3.4 m 2 ),  The assumption that k shirt is constant during the course of an experiment (approx. 3 hours in total) is supported by the observation that the ozone mixing ratio does not vary, within the ±2 ppb accuracy of the measurement, during the final 100 minutes of an experiment (see Figure 2). However, it should be noted that the assumption that the ozone mixing ratio is at steady-state (ss) is not necessarily valid for prolonged periods of time. After reaching its lowest mixing ratio     Table 4 shows steady-state model parameters for geranyl acetone. The yield (Yield GA ), which is that fraction of ozone removed by the shirts that produces geranyl acetone, has been adjusted to produce source-to-sink ratios near unity. The estimated effective yield for geranyl acetone was close to 12% for Conditions 3, 5, and 6, while it was only 2.5% for Condition 7.
Air exchange rate and gas-phase reactions with ozone are the sinks for geranyl acetone in the chamber. For Condition 3, these sinks are of similar magnitude; this is also the case for Condition  Table 5 shows steady-state model parameters for 6-MHO. The yield (Yield 6MHO ), which is that fraction of ozone removed by the shirt that produces 6-MHO, has been adjusted to produce sourceto-sink ratios near unity. The estimated yield for 6-MHO from surface reactions was between 3.0% and 4.4% for Conditions 3, 5, and 6, while it was only 1.3% for Condition 7.
The highest steady-state mixing ratio of 6-MHO (  (6)) constituted about ~70% percent of the total production of 6-MHO inside the chamber. The gas-phase reaction of geranyl acetone with . The reaction additionally forms secondary and tertiary products such as 4-oxopentanal, 4-oxopentanoic acids, acetone, and hydroxyacetone. Table 6 shows steady-state model parameters for 4-OPA. The yield (Yield 4OPA ), which is that fraction of ozone removed by the shirt that produces 4-OPA, has been adjusted to produce sourceto-sink ratios near unity. The fitted yield for 4-OPA varied among the conditions and was highest for Condition 6 (9.2%) and lowest for Condition 7 (3.8%).
The highest steady-state mixing ratios of 4-OPA (    may be caused, in addition to the different starting ozone mixing ratios at the time the t-shirts were placed in the chamber, by differences in soiling and differences in the extent to which skin lipids have been oxidized before the t-shirts were placed in the chamber.

| LIMITATI ON S OF THE S TUDY
Acetone is known to be a major product of squalene ozonolysis.
However, we were unable to quantify it, since the instrument had poor response to this molecule.
The yields of formation for geranyl acetone, 6-MHO, and 4-OPA in the models were adjusted to achieve source-to-sink ratios close to unity. These yields were similar for Conditions 3, 5, and 6, while they were substantially lower for Condition 7. This latter condition occurred early in the campaign after a power outage. However, the humidity and temperature within the chamber was unaffected by the outage, and the mass calibration of the ToF-CIMS did not change substantially. The total ion count was approximately 10% lower during Condition 7, and the ratio of water dimer to monomer was 20% lower. The signal normalized by the mixing ratio of the reagent ion partially takes this into account. Nonetheless, the accuracy of the measurements depends on many parameters. We decided to still report the data from Condition 7 since it provided valuable information on the dynamics of the product distribution within the chamber.

| CON CLUS IONS
Occupant-related indoor chemistry is complex. In the case of unsaturated species such as geranyl acetone and 6-MHO, ozone can both produce and consume a molecule. Steady-state mixing ratios of squalene-ozone reaction products are a result of an interplay between primary, secondary, and tertiary gas-phase and surface reactions, as well as ventilation. In order to quantify the formation and removal processes, a series of mass balance equations was applied to measurement data obtained for a number of experimental conditions in a climate chamber.
When previously worn t-shirts were in the chamber, more than 96% of the ozone removal was attributable to the combination of surface reactions and ventilation (about 40% at AER = 1 h −1 and 70% at AER = 3 h −1 ), while gas-phase reactions accounted for less than 4% of the ozone removal. The ventilation strategies have only a marginal effect on the ozone removal rate by the t-shirts, which is mass transport limited.
Geranyl acetone is produced only by ozone reactions occurring on the surface of the t-shirts. Ventilation was responsible on average for about 50% of its removal from the gas-phase, while the rest was attributable to further reactions with ozone. Approximately Turning off the ventilation overnight or on weekends may lead to the accumulation of certain pollutants with indoor sources, but could also limit the extent to which ozone-derived products are formed and will delay their generation when the ventilation is turned back on again. In applying such a practice, it should be recognized that ozone concentrations are often at their lowest daily values overnight.
Preventing ozone-initiated chemistry in indoor environments is better achieved by limiting indoor ozone concentration via filtration rather than adjusting ventilation rates.