Total OH Reactivity of Emissions from Humans: In Situ Measurement and Budget Analysis

Humans are a potent, mobile source of various volatile organic compounds (VOCs) in indoor environments. Such direct anthropogenic emissions are gaining importance, as those from furnishings and building materials have become better regulated and energy efficient homes may reduce ventilation. While previous studies have characterized human emissions in indoor environments, the question remains whether VOCs remain unidentified by current measuring techniques. In this study conducted in a climate chamber occupied by four people, the total OH reactivity of air was quantified, together with multiple VOCs measured by proton transfer reaction time-of-flight mass spectrometry (PTR-ToF-MS) and fast gas chromatography–mass spectrometry (fast-GC–MS). Whole-body, breath, and dermal emissions were assessed. The comparison of directly measured OH reactivity and that of the summed reactivity of individually measured species revealed no significant shortfall. Ozone exposure (37 ppb) was found to have little influence on breath OH reactivity but enhanced dermal OH reactivity significantly. Without ozone, the whole-body OH reactivity was dominated by breath emissions, mostly isoprene (76%). With ozone present, OH reactivity nearly doubled, with the increase being mainly caused by dermal emissions of mostly carbonyl compounds (57%). No significant difference in total OH reactivity was observed for different age groups (teenagers/young adults/seniors) without ozone. With ozone present, the total OH reactivity decreased slightly with increasing age.

The correction factor was obtained from injecting known amount of a standard gas under different pyrrole/OH ratios conditions. A factor can be obtained from measured reactivity vs. calculated reactivity for that gas. Total of five standard gases with different rate constants reacting with the S4 OH radical including propane, propene, isoprene, α-pinene and acetaldehyde were tested within the pyrrole/OH range observed during the entire campaign. A linear fit was applied among all the factors derived from the tests, resulting in a linear relationship between the correction factor and pyrrole/OH ratio (f = a[pyrrole/OH]+b), with an error of 31%. Throughout the campaign, the pyrrole/OH ratio ranged from 2.0 -3.0, which increased the OH reactivity by a factor of 1.3 to 2.7.

 NO x interference
Previous studies have also shown that NO x can potentially cause an interference to CRM measurements by producing OH radicals via reactions with HO 2 radicals. Relevant corrections should be applied if the measured conditions have abundant NO x (NO > 10ppb 1 ) and it should be noted that NO has a more significant interference compared to NO 2 at the same level [4][5][6] . NO and NO 2 were continuously monitored by a chemiluminescence NO/NO x analyzer (ECO PHYSICS, model CLD 700 AL). The mixing ratios of NO and NO 2 in the occupied chamber were near or below the detection limit (1 ppb) for most of the time during the entire experimental period. By taking into account the dilution factor of the CRM (1.37), NO and NO 2 levels in the glass reactor would be lower than the detection limit most of the time. Therefore, NO x interference to the CRM could be neglected and no correction was applied. NO and NO 2 have OH rate constants comparable to some VOCs shown in Table S1 (9.70 × 10 -12 cm 3 molecules -1 s -1 and 9.80 × 10 -12 cm 3 molecules -1 s -1 for NO and NO 2 , respectively 7 ). The upper limit of the NO x contribution to the total reactivity (assuming 2 ppb of NO or NO 2 ) would be 0.5 s -1 (3.0% of the total reactivity under ozone-free condition and 1.5% under ozone-present condition). This is comparable to some top ten OH reactivity contributing species ( Table 2). The NOx data were mostly at or below the detection limit and no clear trend can be observed due to human occupancy; the 2 ppb of NOx assumption S5 represents an upper limit estimate rather than a measurement. Therefore, NOx was not included in the calculation of OH reactivity in the study.

 Ozone interference
High level of ozone was also found to cause interference depending on the CRM system 5 . For ozone interference test, different levels of ozone (0 -110 ppb) was introduced to the empty chamber.
The measured OH reactivity interference due to ozone was less than 3 s -1 at the highest ozone level, which is lower than the limit of detection for CRM during this campaign (5 s -1 ). The ozone level in the chamber during the afternoon steady state when occupied was around 37 ppb, resulting in 1 s -1 difference in the measured reactivity which was much smaller than the total uncertainty of the measured reactivity (16 -22 s -1 ). As the ozone interference is negligible and extra uncertainty would be introduced from the correction factor, no correction was applied.

Calculations of precision and total uncertainty of CRM
When a test gas is injected into the CRM at several known concentrations, the corresponding mean reactivity at each concentration can be obtained together with the standard deviation. By plotting the relative reactivity (standard deviation/mean reactivity, ) against the mean reactivity ( ), an exponential curve is obtained that can be fit with the equation: where parameters a, b and c are derived from curve fitting and the relative reactivity defines the precision. For this study, results from test gases mentioned in "Correction factor for not being at pseudo-first-order conditions" were included to derive the fitting results. Based on the measured reactivity of real measurements, the precision can be calculated by applying this equation to each data point.

S6
The accuracy of the CRM is the propagation of uncertainties from the pyrrole standard gas concentration, the dilution factors derived from flow measurements, the OH rate constant of pyrrole, the humidity correction, and the non-pseudo-first-order correction. Detailed numbers can be found in Table S1.

Empty-chamber background
The background obtained from the empty chamber before volunteers entered was in general very stable for calculated reactivity (3 s -1 on average with a variability of 10%). For measured total reactivity, the background was typically around or under the detection limit (5 s -1 ).

Definition of steady-state condition
As the calculated reactivity has less uncertainty than measured reactivity, the steady-state condition was verified by the relative change of calculated reactivity during the 15 minutes before volunteers exited the chamber. This is to avoid any effect left due to requested movements by the volunteers (standing up and stretching) every hour during each experimental period. For all the experiments, the relative changes ((max-min)/mean) were below 5% (0.8-5.0%, mean 2.1%), which is much less than the uncertainty of the calculated OH reactivity (21% -45%, median and mean 29%). Therefore, the time period of 15 minutes is suitable to be considered as steady-state condition.

Adjustment method applied for OH reactivity per person comparison
To be able to compare the OH reactivity per person (s -1 p -1 ) in the ICHEAR chamber experiments with other studies, the OH reactivity per person obtained from other studies were adjusted for the room volume and the air change rate (ACR) applied in the ICHEAR chamber experiments using the following equations: where refers to the OH reactivity per person after the adjustment and refers to the OH reactivity per person before the adjustment, which is obtained from the total OH reactivity divided by the number of occupants in that environment. 22.5 m 3 is the ICHEAR chamber volume and 3.2 h -1 is the ACR for the chamber. The per-person OH reactivity before the adjustment ( ) for the other studies mentioned in Table 3 were estimated using reported values. For the museum gallery room study 70 , is derived from the incremental total OH reactivity (14 s -1 ) during the high occupancy event (compared to the low occupancy condition) divided by the number of occupants (176 on average). For the classroom study 71 and cinema study 72 , the was calculated based on the VOC emission rates (μg h -1 p -1 ) reported in each work. As both studies used PTR-ToF-MS, some masses could only be assigned to chemical formulas instead of specific compounds.
Therefore, we only included masses with specific compound assignments reported in those two studies to calculate the total OH reactivity per person. Those included VOCs were mostly measured during the present experiments as well. The emission rates were first converted to mixing ratios per person (ppb p -1 ) based on the indoor space volume and the ACR reported in each study. Then the total OH reactivity per person was calculated using Eq.1 in the main text. The estimated of those two studies may slightly underestimate the actual values as those masses without a specific S8 compound assignment (accounting for < 20% of the total VOC emission rates) were not included in the calculation.

Potential artifacts from decomposition of hydroperoxides
It has been reported that metal surface can act as a catalyst for the decomposition of organic hydroperoxides, and that this is temperature dependent. 73 It has been further reported that with stainless tubing, the conversion rates for ISOPOOH to formaldehyde at room temperature were not significant (below 10%) but increased to 50% at 160 °C 73 . As the temperature in the ICHEAR chamber study was always less than ~ 31°C, this artifact on the stainless-steel chamber walls is not anticipated to be important. However, for the PTR-ToF-MS instrument, this interference may still exist as the drift tube was heated to 60 °C, converting a small amount of ISOPOOH to other products. Another important factor to consider is that in the ICHEAR chamber study, the main oxidant was ozone instead of OH radicals (which are the major oxidant outdoors). MVK and MACR, rather than ISOPOOH, are the major products of isoprene ozonolysis 74 . Furthermore, the reaction between ozone and isoprene is relatively slow (1.1 x 10 -3 ppb/h at 298 K) 7 ; at the average ozone concentration (37 ppb) used in ICHEAR, the O 3 /isoprene reaction occurs at a rate (0.04 h -1 ) substantially slower than the air change rate (3.2 h -1 ). Hence, production of ISOPOOH via ozone reaction was quite small in ICHEAR. The formation of ISOPOOH, to the extent that it occurred in ICHEAR, was probably from isoprene/OH oxidation as OH radicals can be generated during the ozonolysis of unsaturated compounds. Assuming the extreme scenario that measured MVK/MACR were half decomposed from ISOPOOH, it would increase the calculated reactivity during ozonepresent condition by 0.6 s -1 as ISOPOOH has a faster reaction rate constant compared to MVK/MACR (9.65 × 10 -11 cm 3 molecules -1 s -1 , averaged of (1,2)-ISOPOOH and (4,3)-ISOPOOH) 75 .
The calculated increase of 0.6 s -1 is within the standard deviation of the total calculated reactivity S9 (1.4 s -1 ). Therefore, we judge that the interference should be small. In terms of reactivity measurement, as the PTR-QMS measured the air coming out of the glass reactor and temperature inside the reactor was around 35-40 °C, there should be no interference. S10 * rate constant ranges listed in parentheses refer to the range of rate constants of listed isomeric compounds. **Compounds without mentioning as "PTR calibrated", "fast-GC" nor "Piccaro" refer to the mixing ratios of species measured by PTR-ToF-MS were calculated based on theoretic method using a constant rate coefficient (2.0E-9 cm 3 s -1 ) for the reactions with H 3 O + 68 , except for 6-MHO where a known rate coefficient (3.8E-9 cm 3 s -1 ) was used 69 .

Ozonepresent
Sum of all species 13.5 *isoprene and propanal data were obtained from fast-GC

Ozonepresent
Sum of all species 31.5 #ammonia data was from Picarro. S17 Figure S1. Calculated and measured OH reactivity during ozone-free and ozone-present steady-state conditions of the benchmark experiments with young adults. Error bars represent the total uncertainty of measured OH reactivity and calculated OH reactivity. Experiment 21 is a replicate of Experiment 6.
(a) (b) Figure S2. Top ten species contributing to the total OH reactivity for teenagers, young adults and seniors under (a) ozone-free condition and (b) ozone-present condition. The species marked with red underline represent unique species that do not appear among the top ten species of the other two groups. The pie chart in each plot represents the fractions of the total reactivity attributable to the top ten species (hatched) and remaining species (blank).