Bedroom Concentrations and Emissions of Volatile Organic Compounds during Sleep

Because humans spend about one-third of their time asleep in their bedrooms and are themselves emission sources of volatile organic compounds (VOCs), it is important to specifically characterize the composition of the bedroom air that they experience during sleep. This work uses real-time indoor and outdoor measurements of volatile organic compounds (VOCs) to examine concentration enhancements in bedroom air during sleep and to calculate VOC emission rates associated with sleeping occupants. Gaseous VOCs were measured with proton-transfer reaction time-of-flight mass spectrometry during a multiweek residential monitoring campaign under normal occupancy conditions. Results indicate high emissions of nearly 100 VOCs and other species in the bedroom during sleeping periods as compared to the levels in other rooms of the same residence. Air change rates for the bedroom and, correspondingly, emission rates of sleeping-associated VOCs were determined for two bounding conditions: (1) air exchange between the bedroom and outdoors only and (2) air exchange between the bedroom and other indoor spaces only (as represented by measurements in the kitchen). VOCs from skin oil oxidation and personal care products were present, revealing that many emission pathways can be important occupant-associated emission factors affecting bedroom air composition in addition to direct emissions from building materials and furnishings.


SI
Instruments were situated in the basement/garage area of the H3 study site to minimize interference with the occupants' daily lives.Researchers were able to maintain and retrieve data from the instruments without entering the house and disrupting the normal pattern of occupancy.Air sampling lines and in-situ low-cost sensors were placed inside the residence as indicated in Figure S1.

CO2 Measurement Correction
The LICOR LI-820 instrument used to measure CO2 in the sampling manifold did not record measurements higher than 2000 ppm.The Netatmo CO2 sensors measured a broader range, including levels above 3000 ppm.Data from the Netatmo sensors, which were placed around the house, were used for analysis in this paper.A correction factor is applied to improve the accuracy of the Netatmo measurements to match the calibrated LICOR instrument.For the main bedroom, the data collected by the LICOR instrument from November 17 through December 06, 2021, was compared to Netatmo data, averaged over matching 30-minute sample periods.
To ensure that the LICOR measurements near 2000 ppm also did not interfere with the analysis, any LICOR measurements over 1800 ppm and Netatmo measurements over 2000 ppm were removed for determining the CO2 measurement correction.The 'linortfit' function in MATLAB was applied to determine the best fit and correlation, as shown in Figure S2.Bolded values indicate those with CO2 accumulation < 1000 ppm-h overnight.† Indicates time from the night before if the occupants went to bed before midnight.
Based on the 1000 ppm-h selection criterion, the following nights were removed from further analysis: Nov 17, 19, 22, 24, 29, and Dec 04.The small negative accumulation value reported for Nov 19 may be a result of the manipulation experiments conducted by researchers while regular occupants were away.The remaining 12 nights, in which occupant bioeffluents materially affected the bedroom CO2 levels overnight, were used in this paper for analysis of VOC composition, concentrations, and emissions in the occupied bedroom.Here, the duration includes the decay period of the CO2 concentration profiles until the return to steady state, which continued after occupants left the bedroom.The duration used for air-change rate calculations was for the period of increasing CO2 concentrations only, and the duration used in emission rate calculations was for the time periods that occupants were in the bedroom only.

Nighttime VOC Concentrations at H3
Table S4 presents summary statistics on three-hour periods of peak concentrations of compounds at three sampling locations over the twelve occupied nights prescribed by the filtering techniques discussed in the previous section using the sleeping period durations listed in Table S3.We removed 25 compouThe ratio of the average 3-h peak bedroom and average 3-h peak kitchen concentrations is also reported.Values meeting the threshold Cbed/Ckitch ≥ 1.30 are indicated in bold.Data interpolated over the PTR-TOF-MS calibration periods (01:00-04:00 every morning) were included.We report 214 of the 239 detected VOC signals to focus on compounds not related to water clusters, the primary ion, or with a mass below m/z = 33.033(methanol).Of the 214 signals, 94 showed an enhancement ratio of 1.30 or higher.For all those compounds, the corresponding outdoor concentration was either too low to be reported or resulted in an indoor-tooutdoor ratio of ≥ 3.0 (with one exception).a Each concentration entry represents the average value of 12 nighttime determinations in which each night is represented by the 3-h period producing the highest average bedroom concentration.Kitchen and outdoor average values were computed for the same time periods as the 3-h peaks in the bedroom.Bolded values indicate a Cbed/Ckitchen ≥ 1.30.ND indicates "not determined," which reflects an average concentration value below the reporting limit of ~5 ppt.

Breath Emission Tracers
Figure S4.Diel plots of mean (a) CO2, (b) isoprene, and (c) acetone concentrations in the bedroom (green), kitchen (orange), and outside (red) for the twelve nights and corresponding days as enumerated in Table S3.
Carbon dioxide is generally a reliable tracer of human breath emissions.To probe that assumption, a correlation analysis like that applied for CO2 measurements was also performed with respect to both isoprene and acetone.They are the most abundant VOC bioeffluents detected and the bestcorrelated with CO2.The comparison between correlations with respect to CO2 (x-axis) and with respect to isoprene (y-axis) is displayed in Figure S5, which is color-coded in the same manner as in Figure 3.The analysis for isoprene revealed that 33 compounds had R ≥ 0.70, but only those that overlap with the results from the analysis for CO2 are considered in the figure .With the obvious exception of CO2 itself and C7H7 + , every compound considered has an equal or higher R value with respect to isoprene, highlighted by the fact that they are all on or above the 1:1 line.This better correlation with isoprene suggests that, at least in this residence, isoprene provides a more reliable breath emission tracer than CO2.One possible reason for this finding is that the isoprene signal does not encounter significant interference from outdoor signals at night, whereas the CO2 signal is elevated outdoors at night, likely due to a combination of respiration from soils and plants and accumulation of emissions into a more stable atmosphere (than during daytime) from burning carbonaceous fuels.
The higher number of compounds showing R ≥ 0.70 with respect to isoprene than with respect to CO2 offers further evidence of isoprene being more reliable as a VOC bioeffluent tracer.No compounds appeared to be strongly anti-correlated with isoprene.The correlation analysis for acetone revealed that fewer compounds have values of R ≥ 0.70 than noted for CO2 or isoprene, indicating that it is not as reliable a bioeffluent tracer.This outcome is expected given acetone's broader range of potential sources.No VOC signals had an R value greater than 0.70 or less than -0.70 with respect to temperature, but nine compounds showed strong correlation with relative humidity.Of the more abundant compounds, only acetone had a relevant R value (0.71) with respect to relative humidity.Hexanal and some alkyl fragments also had high R values, indicating a potential role of relative humidity in the fragmentation of some VOCs during PTR analysis.

Bedroom VOC Emission Rates
Table S5.Average upper (air exchange with outdoors) and lower (air exchange with kitchen) bounds for the emission rates (µg h -1 ) of all 99 VOC species included in this analysis in the bedroom over 12 nights at H3 using full occupancy period.

Skin Emission Signals
Skin oil contains ~10% squalene which reacts with ozone to form specific products, including 6-MHO and 4-OPA [1]. Figure S6 shows the time series in the bedroom for ozone and these two squalene-ozonolysis products.There was no evidence to suggest that ozone concentrations were directly related to temperature or relative humidity in the bedroom.The average nighttime ozone concentration in the bedroom was low, 0.9 ± 0.5 ppb, but there was clear enhancement of ozone indoors during the day when ozone was also high outdoors.Aside from two notable peaks in the kitchen on Dec 01, 2021, as seen in panel (a) of Figure S6, the kitchen and bedroom concentration profiles are similar.Because skin oil oxidation products clearly have a different formation pathway than emissions from breath and can therefore be considered to have different sources within the human exposome, it is not appropriate to consider isoprene as a reliable tracer for this source category.Therefore, a correlation analysis with respect to ozone was undertaken to determine whether a skin emission tracer could be established.concentration profiles appearing similar.This finding is consistent with their production occurring mainly when ozone is elevated in the house during the day rather than during nighttime bedroom occupancy when the ozone level is very low.Peaks in ozone lead to faster production of 6-MHO and 4-OPA, as indicated by peaks occurring at the same time in all three profiles.

Figure S7
. Diurnal profile, using the 12 nights and their corresponding days examined per Table S3, of D5 siloxane in the H3 bedroom.The top and bottom edges of the blue boxes represent the 75 th and 25 th percentiles, respectively, and the red lines inside the boxes represent the medians.The black whiskers extend distances of 1.5 times the interquartile range from the box edges.The red symbols denote outliers.
Cyclic volatile methyl siloxanes (cVMS, D3-D6) are emitted from some personal care products [2].The D5 siloxane signal dominated the total observed cVMS contributions to indoor air composition, as expected from previous work.Average nighttime concentrations in the bedroom for D3-D6 siloxanes were 0.057 ± 0.018 ppb, 0.040 ± 0.015 ppb, 1.08 ± 0.63 ppb, and 0.101 ± 0.096 ppb, respectively.The peak levels of D3, D4, and D6 never exceeded 1 ppb in the bedroom during the time it was monitored, but D5 concentrations show peaks as high as ~ 20 ppb that occur most mornings.Figure S7 shows the D5 diurnal cycle in the bedroom, with a peak occurring at 09:00 and decaying throughout the morning.There is no evidence of emissions of D5 at night during sleeping periods.This pattern is likely the result of occupants regularly applying personal care products in the morning after waking up.There is no evidence of correlation between cVMS

Figure S6 . 3 LayoutFigure S1 .
Figure S1.House 3 (H3) layout.Orange, green, and black stars indicate sampling line inlet locations, and the gray star indicates a tracer release location.Additional sampling locations not shown are in the attic (above main floor), garage (same elevation as lower level), and outside.Additional tracer release locations not shown are in the attic and garage.

Figure S2 .
Figure S2.Scatter plot (blue) of bedroom CO2 measurements (ppm) made by LICOR (x-axis) and Netatmo (y-axis) sensors with orthogonal least squares best fit line (orange).The equation for line of best fit is also provided.

FigureFigure S3 .
Figure S3 compares the time series of the original Netatmo data, the LICOR data, and the corrected Netatmo data in the main bedroom over one of the last three weeks of the H3 campaign.The corrected Netatmo data (green) is used for all analyses reported in this paper.

Figure S5 .
Figure S5.Scatter plot of compounds with values of R ≥ 0.70 with respect to CO2 (x-axis) and isoprene (y-axis).Some markers may overlap.The black line indicates a 1:1 relationship for the correlation coefficients.

Figure S6 .
Figure S6.Time series of (a) ozone, (b) 6-MHO, and (c) 4-OPA in the bedroom (green), kitchen (orange), and outdoors (red) at H3 over one week.Midnight of each date included is indicated by tick marks on the x-axis.
However, there was no evidence of correlation or anti-correlation with respect to ozone as the R values of VOC signals ranged from -0.35 to 0.41.Average 6-MHO and 4-OPA concentrations in the bedroom over the examined nighttime periods were 0.74 ± 0.20 ppb and 0.29 ± 0.07 ppb, respectively.As seen in panels (b) and (c) of Figure S6, indoor concentrations of both oxidation products are consistently enhanced indoors, with kitchen and bedroom

Table S4 .
Summary statistics on VOC signals in the bedroom at H3 6. (page S13) -Breath Emission Tracers a. Figure S4.Diel plots of CO2, isoprene, and acetone at H3 b. Figure S5.Scatter plots of VOC R values with respect to isoprene and CO2 7. (page S15) -Bedroom VOC Emission Rates a.

Table S5 .
Average upper and lower bounds for bedroom VOC emission rates 8. (page S17) -Skin Emission Signals a.

Table S1 .
Composition of Standard Gas 1 used to calibrate the PTR-TOF-MS at H3.

Table S2 .
Composition of Standard Gas 2 used to calibrate the PTR-TOF-MS at H3.

Table S4 .
Mean VOC concentrations (ppb) in the bedroom, kitchen, and outdoors during twelve selected occupied nights.a