How Does Personal Hygiene Influence Indoor Air Quality?

Humans are known to be a continuous and potent indoor source of volatile organic compounds (VOCs). However, little is known about how personal hygiene, in terms of showering frequency, can influence these emissions and their impact on indoor air chemistry involving ozone. In this study, we characterized the VOC composition of the air in a controlled climate chamber (22.5 m3 with an air change rate at 3.2 h–1) occupied by four male volunteers on successive days under ozone-free (∼0 ppb) and ozone-present (37–40 ppb) conditions. The volunteers either showered the evening prior to the experiments or skipped showering for 24 and 48 h. Reduced shower frequency increased human emissions of gas-phase carboxylic acids, possibly originating from skin bacteria. With ozone present, increasing the number of no-shower days enhanced ozone-skin surface reactions, yielding higher levels of oxidation products. Wearing the same clothing over several days reduced the level of compounds generated from clothing-ozone reactions. When skin lotion was applied, the yield of the skin ozonolysis products decreased, while other compounds increased due to ozone reactions with lotion ingredients. These findings help determine the degree to which personal hygiene choices affect the indoor air composition and indoor air exposures.

Tables: S1, S2, S3, S4, S5, S6, S7 Figures: S1, S2, S3, S4, S5, S6, S7, S8 1.Quantification of VOCs measured by PTR-ToF-MS A standard gas mixture containing 17 compounds (Apel-Riemer Environmental Inc.) was used for the calibration during the experiments.In total, three calibrations under varying humidity levels were performed.6-methyl-5-hepten-2-one (6-MHO) and 4-oxopentanal (4-OPA) were calibrated after the campaign in the lab using gas cylinders (Westfalen AG).Table S4 shows the compound list with their limits of detections and total uncertainty.Based on the calibration results, an experimental transmission curve was obtained to quantify other measured masses using the theoretical method 1-3 by using a constant proton transfer reaction rate coefficient of 2.5 × 10 -9 cm 3 molecule -1 s -1 .The uncertainty of these species mainly comes from the difference of the exact rate coefficient, which is within ~ 50%. 2,4 order to filter out the VOC species emitted due to human occupancy; a filter was applied based on the mixing ratios.Only VOCs with the difference between steady-state levels in the occupied chamber and the empty chamber larger than two times the standard deviation of the empty chamber level were considered to be associated to human occupancy.In addition, a unified mass list was created based on filtered VOC species from the four hygiene experiments to study the common human-associated VOCs.
2. Steady-state level determination for species not reaching steady state in the last 15 minutes Due to the different chemical and physical properties of VOCs analyzed, the time needed to reach the steady state condition varies.In particular, when ozone was present in the chamber, some VOC species were newly generated in the chamber, not necessarily reaching steady state before the end of the experiments.For example, it was found that 4-OPA, a secondary squalene ozonolysis product, was often found to be taking longer to reach steady state compared to 6-MHO (primary and secondary product) 5 .In this study, 4-OPA kept increasing until the volunteers exited the chamber (Figure 1b).In order to identify any other species that similarly did not reach steady state in the last 15 minutes before volunteers left the chamber, Pearson correlation analysis was performed between 4-OPA and all other species over the time period when ozone was present in the chamber.For those species significantly correlating with 4-OPA (R 2 ≥ 0.90, p ≤ 0.01), we assume they too may not have reached steady-state in the last 15 minutes.In these cases, we applied sigmoidal fitting to those species to estimate the steady-state values using OriginLab (OrigninPro 2021b).The Gompertz function was found to be the best fitting for the selected time period, where those 4-OPA correlated species could be all successfully fitted.Figure S1 shows an example of sigmoidal Gompertz fitting for 4-OPA, where the estimated steadystate mixing ratio was slightly higher than the 15-minute-averaged value (2,4%).For other species that may not have reached steady state, the steady-state mixing ratios derived from the fitting results were mostly around or below 5% higher compared to the last 15-minute-averaged values, as shown in Table S3.Only one species C5H6O2H + (m/z 99.0441) showed much higher steady-state mixing ratio than the last 15-minute-averaged level (14% to 21%).In order to further improve the calculation of the emission rates and the ozone product yields that are based on steady state condition, sigmoidal fitting derived steady-state mixing ratios during the ozone-present condition were used for compounds listed in Table S3.As a result, the total emission rate (ER) only increased by 0.5% ~ 1.5% (83 µg h -1 p -1 to 141 µg h -1 p -1 ) for experiments included in this study (Table S1), compared to using the last 15-minute average mixing ratios.The increase due to steady-state fitting is much less than the total ER difference between benchmark and its replicate (1331 µg h -1 p -1 between Exp. 1 on Day 1 and Exp. 5 on Day 4).Therefore, we can conclude that for most of the VOCs measured, the steady-state condition has been reached.

Mass balance model
• First-order ozone removal rate coefficient by occupants (  ) During steady state, the ozone emission rate is identical to the ozone removal rates from all pathways.In our study, the main ozone removal pathways in the chamber are 1) ventilation; 2) gas-phase reactions; 3) occupants (mainly skin and clothing); 4) chamber surface.Therefore, the ozone emission rate (ER, ppb h -1 ) can be described as the following Eq.S1.
3 can be derived from the measured ozone level in the supply air multiplied by the air change rate ( = 3.2 ℎ −1 ).[ 3 ]  is the ozone mixing ratio during the steady state.
[]  refers to the mean mixing ratio (ppb) of O3-reactive VOCs over the steady-state period.Here, we included 9 VOCs containing a carbon-carbon double bond (see Table S5), which were the main O3-reactive compounds 6 . ( 3 ) represents the gas-phase second-order reaction rate coefficient of VOC i reacting with ozone (listed in Table S5). ℎ is the removal rate of chamber surface.As the experiments were performed in the same chamber that our previous study was performed, the same value of  ℎ (0.17 h -1 ) was used 7 .The first-order ozone removal rate coefficient by occupants (  ) can be derived from Eq. S1.The mass balance model terms for ozone removal in the hygiene experiments are listed in Table S7.
6 and  4 are the surface yields of 6-MHO and 4-OPA, respectively.The second-order rate coefficients of VOCs with ozone in Eq.S1, S2 and S3 are listed in Table S5.0.5 and 2 refer to the branching ratios of 6-MHO or 4-OPA generated from their precursors reacting with O3.Values were taken from our previous study 6 .

Variability calculation for benchmarks and no-shower experiments
When comparing the relative change between no-shower experiments and benchmark experiments, the intra-group variability (same volunteers performing replicate experiments) should be considered to identify if the difference is significant.Therefore, we evaluated the variability (reproducibility) using the benchmark and its replicate experiment (Exp.  and converted to the unit of ppb -1 h -1 .For MVK (methyl vinyl ketone) and MACR (methacrolein), as an averaged number was used in Zannoni et al. 6 , values were taken from preferred numbers from IUPAC 9 .*Breath compounds acetone, methanol and isoprene were subtracted from the total emission rate.** Acetone was not shown here as it was also influenced by the breath emission.# There is significant difference between the relative change of one-day no shower experiment and the relative change of two-days no shower experiment.Figure S8.Time series of species in Figure 4b showing higher ozone reaction product yields when lotion was applied on skin compared to the no-lotion condition (benchmarks from hygiene experiments).

Figure S3 .
Figure S3.Time series of selected major breath-borne VOCs measured during the experiments with different hygiene levels.The shaded gaps represent the time period where PTR-MS measured breath.

Figure
Figure S4.(a) Total emission rates excluding the top three species (breath compounds: acetone, methanol and isoprene) and fractional contributions from other strongly emitted species and the rest of the species grouped into six chemical family groups during the hygiene experiments under ozone-free condition.(b) Delta emission rates (the difference between the ER during steady state with ozone present and absent) and fractional contributions from the strongly emitted species (species that appear among the top 10 contributors in all four experiments) and from rest of species grouped into six groups during the hygiene experiments under ozone-present condition.

Figure S5 .
Figure S5.Total whole-body ERs and fractional contributions from top 10 species and other grouped species during the experiment with lotion on skin under (a) ozone-free condition and (b) ozone-present condition.

Figure S6 .
Figure S6.Time series of species for which the lotion was a strong source during the benchmark experiment, its replicate and the lotion-on-skin experiment.The presented species were the most abundant VOCs observed in the lotion-only experiment, showing no clear impact of ozone.The shaded gaps represent the time periods where PTR-MS measured breath.

Figure S7 .
Figure S7.Time series of acetaldehyde, major species originating from the lotion and potential lotion-ozone reaction products (C9H18O and C9H16O) together with ozone mixing ratios in the chamber during lotion-only experiment.

Table S1 .
5% for the total delta emission rate under ozone-present condition.For major skin ozonolysis products, the overall yield change was mostly below or around 10%.These results indicate good reproducibility of our experiments (experiment vs. replicate).Similarly, we calculated the absolute change for the listed parameters in TableS6between no-shower experiments and benchmark experiments (Exp.1 and Exp.5), which was then normalized by the benchmark value correspondingly.Thus, we obtained a relative change for each no-shower experiment, to be compared with the relative change across all benchmark experiments.Details of the experimental conditions (Exp.No. 2 was an experiment with higher temperature, which was not included in the study).
percentage.As shown in TableS6, the mean relative change across all benchmarks was less than 1% for the total emission rate (excluding top three exhaled compounds) under ozone-free condition and less than 1.

Table S2 .
Ingredient table of the lotion and the amount used in this study

Table S3 .
Difference (%) between steady-state values derived from fitting using sigmoidal Gompertz function and from the average values of last 15-minute measurements over the ozonepresent period for species significantly correlating with 4-OPA (R 2 ≥ 0.90, p ≤ 0.01).Numbers in red indicate that fitting results are lower than last-15-min steady-state values.

Table S4 .
Compounds calibrated to standard gas cylinders with average limit of detection (LOD) and total uncertainty together with standard deviation (std).

Table S5 .
Main O3-reactive compounds 6 included in Eq.S1 for calculating the first-order ozone removal rate coefficient by occupants (  ): their rate coefficient with ozone ( ( 3 ) ), and steady-state mixing ratios during the hygiene experiments.
*The rate coefficient was taken from TableS1in Zannoni et al.

Table S6 .
Mean change in percentage for benchmark experiments compared to their replicate experiments (including Exp. 1 on Day 1 and Exp. 5 on Day 4 in this study and four pairs of replicate benchmark experiments from the previous ICHEAR study 8 ), and for no-shower experiments (Exp.3 on Day 2 and Exp. 4 on Day 3) compared to the benchmarks (Exp. 1 on Day 1 and Exp. 5 on Day 2) for various parameters reported in the study.Values in bold indicate that the mean relative change of no-shower experiments compared to the benchmarks for that parameter is significantly different from the mean relative change across overall benchmarks.

Table S7 .
Mass balance model sink terms for ozone removal during the hygiene experiments.Sinks OPA mixing ratio and red solid line represents the fitting curve.SSmeasured is the steadystate value averaged from measured data over the last 15 minutes before volunteers exited the chamber (shaded area).SSfitting is the steady-state value estimated from the fitting function.