In each experiment, therefore, we included a 5-min sampling period prior to the source period to account for the variation in PM2.5 background concentration. For each indoor experiment, we added a 60-min sampling period following the source period for determining the PM2.5 decay inside the building. The decay rates were determined by the log linear regressions between 1-min PM2.5 concentrations averaged over the 5 monitors versus time during the well-mixed decay periods. For the same source type , the PM2.5 decay rate could reflect the relative strength of air mixing indoors. A higher air change rate will lead to stronger indoor air mixing , enhancing the particle surface deposition . This suggests both the cause and consequence of stronger air mixing could contribute to a higher decay rate. Therefore, given a comparable evaporation loss rate , a larger decay rate could indicate stronger air mixing indoors, which could cause more uniform concentration and a smaller proximity effect. Air mixing is one governing factor that affects the spatial distribution of concentration and exposure close to a source . By examining the decay rates for experiments with the same source type, we can ensure comparisons are based on comparable air exchange and air mixing conditions.To determine the source and environmental characteristics in each indoor experiment, we calculated the average exhalation peak velocity and duration and the decay rate . Table 1 summarizes the statistics of average exhalation peak velocities, average exhalation durations,vertical grow system and decay rates for indoor smoking versus indoor vaping from 16-17 experiments with all the windows and doors closed without fan operating . These base-case experiments had background air velocities below the anemometer’s detection limit – this enabled more accurate determination of exhalation velocities for the two different sources.
The mean of average exhalation peak velocities for indoor smoking was ~2 times as high as that for indoor vaping . The mean of average exhalation durations for indoor smoking was ~70% of that for indoor vaping . The mean decay rate for indoor vaping was higher than the mean decay rate for indoor smoking . Particle losses due to air exchange and particle settling are expected to be comparable for indoor smoking and vaping experiments; the sizable difference was likely due to the higher aerosol volatility for vaping. This finding was consistent with previous studies testing the decay rates of 4 different marijuana sources inside a car chamber and in a residential bedroom . Li et al found PM2.5 particle loss rates for vaping aerosols were >4 times as high as that for – Di-EthylHexyl-Sebacat aerosols with little evaporation. In addition to exhalation pattern, aerosol evaporation could have a significant effect on exposure versus distance from the source. The average air velocities for outdoor experiments ranged from 0.21 to 0.33 m/s. The highest average velocity was recorded when the overhead outdoor umbrella was folded . This could be due in part to less blockage of the air movement. Klepeis et al and Acevedo-Bolton et al measured ground-level air velocities in the backyard of a California home. Their reported average air velocities were comparable to our measured values. These backyard measurements are expected to be affected by eddy currents near buildings. Figures 2 and 2 show examples of the 1-s concentration time series of PM2.5 measured indoors and outdoors at 1 m, 2 m, and 3 m horizontal distances from the participant performing marijuana vaping in the residential property . Unlike the standard indoor experiments that were performed separately with 1-h decay periods , continuous indoor measurements were taken across multiple source periods with only 5 minutes apart. This was to align with the emission sequence of the outdoor time series to allow comparisons between Figures 2 and 2. Here, all concentrations greater than the monitor’s upper limit were replaced with 20 mg/m3 , giving maximum concentrations ~10 mg/m3 . For both the indoor and outdoor experiments, the magnitudes and occurrences of transient concentration spikes – “micro-plumes” – increased with decreasing distances, showing the proximity effect during active emissions .
Striking differences were observed between indoor and outdoor situations. Micro-plumes were much more likely indoors than outdoors. In the indoor environment , aerosols could follow the exhaled airflow, moving toward the monitors that were in front of the vaper. In contrast, aerosol movement outdoors was primarily governed by the wind patterns. The rapidly changing directionality of outdoor air flows near the building made micro-plumes less likely to emerge. The durations of micro-plumes were longer indoors than outdoors. The slower air movement indoors could make emitted plumes linger at a monitoring location. This effect can also be seen from the persistent PM2.5 concentration time series after each source emission period ended indoors. As expected, the more frequent occurrences and longer durations of micro-plumes indoors greatly increased the average concentration and exposure at close proximity to the active emission source. Figure 3 summarizes the time-averaged PM2.5 concentrations over the 5- min source periods at 1, 2, and 3 m distances from the source in all the 35 indoor and outdoor experiments with marijuana smoking and vaping. Figures 3-3 correspond to the condition with all windows and doors closed and without HVAC fan running whereas Figure 3 involves opening a door and two windows and with HVAC fan running . Figures 3-3 correspond to the condition with the umbrella open and above the smoker whereas Figure 3 involves fully closing the overhead umbrella . Each boxplot contains measurements from the 5 SidePak monitors at different angles in front of the smoker with the dashed line representing the mean value and the solid line representing the median. Background concentrations ranged from 1.2 to 6.8 mg/m3 ; they were subtracted from these 5-min PM2.5 averages. Statistics of each boxplot are available in the Supplementary Material . The 5-min PM2.5 concentrations at 1 m were higher and more variable for indoor vaping than for indoor smoking versus 3). However, the levels of indoor vaping decreased more noticeably with distance than for indoor smoking .
This finding could be associated with the difference in exhalation pattern – the exhalation peak velocity for indoor vaping was only ~50% that of indoor smoking. Therefore,indoor vertical garden systems vaping aerosols are expected to have longer time for decay before reaching a given distance. Another consideration involves the aerosol evaporation process – the higher decay rate of the vaping aerosols due to their higher volatility could also result in a greater concentration decrease over distance. The PM2.5 exposures for indoor marijuana smoking were much higher than for indoor tobacco smoking . This could be caused by the higher emission rate for marijuana smoking accompanied with the smaller indoor volume . Another factor was the different monitoring setups – our study used 5 monitors to cover 60o angle facing the smoker, making it more likely to capture the emitted plumes than a single monitor. Similarly, PM2.5 exposures for indoor marijuana vaping were much higher than indoor e-cigarette vaping . This again was likely due to more monitors at each distance and the smaller indoor volume . Both vaping sources had a significant concentration decrease over distance, but the marijuana decrease was smaller . This could be due in part to the lower aerosol volatility of marijuana vaping compared to ecigarette vaping . Figure 3 shows the measurements from the only 3 indoor vaping experiments with the HVAC fan operating in the house . In addition to lowering the 5-min PM2.5 levels , mechanical ventilation greatly reduced the variation of the 5-min PM2.5 averages measured at the 5 different angles at each distance versus 3). In addition, it diminished the pronounced concentration gradient over distance observed without mechanical ventilation operating. As expected, stronger air mixing due to mechanical ventilation made the PM2.5 concentration more uniform in space. The outdoor 5-min PM2.5 levels at each distance were less than 5% of the indoor levels for either smoking or vaping. Therefore, a different vertical scale was needed for Figures 3-3. Again, the varied airflow direction and more rapid plume movement outdoors made the PM2.5 exposures in front of the smoker much lower than indoors. The PM2.5 exposure for outdoor marijuana smoking was higher than for outdoor tobacco smoking: 13 g/m3 at 1 m and 29 g/m3 at 0.8-1.5 m .
In addition to the higher emission rate for marijuana smoking , use of 5 1-m monitors under an outdoor umbrella with the smoker made plume encounters more likely . Most of the outdoor experiments involved the participant smoking or vaping under an outdoor umbrella except for the 3 alternative-case experiments in Figure 3 . In these 3 experiments without an umbrella above the smoker, the lower exposures were likely caused by the less-enclosed setting. This, in combination with the highest recorded average air velocity , could cause greater dispersion of emitted particles near the smoker. For each box plot in the 4 base-case graphs -3 and Figures 3-3, we separated the 5-min averages into two groups based on 1 and 1.5 m breathing heights and calculated the mean for each group. For indoor vaping, the means of the 5-min averages for all the 3 distances were higher at 1 m than at 1.5 m height. This is not surprising as the source was closer to 1 m height. In contrast, the means for all the 3 distances were higher at 1.5 m than at 1 m height for indoor smoking. The difference was greatest at the shortest distance ; the mean at 1.5 m height was ~1.7 times as high as the mean at 1 m height. This might be due to the stronger plume buoyancy created by acombustion source – the burning joint – thus increasing the means at 1.5 m height. The means of the 5-min averages outdoors -4) did not necessarily follow the same pattern observed indoors; for outdoor smoking, the mean at the 1.5 m height was greater at 1 m distance, but the outdoor means at 1 m height became greater at the 2 and 3 m distances. In the presence of outdoor wind, the effect of plume buoyancy could become less noticeable, especially for greater distances from the source. Figures 4-4 show the cumulative frequency distributions of 1-s PM2.5 concentrations collected during 5-min source periods on log-probability graphs for 18 indoor and outdoor experiments with smoking and vaping. Again, the left four graphs corresponded to the base-case experiments indoors -4; with all windows and doors closed; without HVAC fan running and outdoors -4); outdoor umbrella open above the smoker. The right two graphs and 4) corresponded to the alternative-case experiments indoors and outdoors , respectively. Each frequency distribution contains aggregated measurements from the 5 SidePak monitors at different angles . Each graph compared the cumulative frequency distributions at 1, 2, and 3 m distances from the 3 experiments with similar environmental conditions. Indoor experiments that had comparable decay rates were grouped together for each graph: 0.34-0.37 h-1 for smoking, 0.97-1.06 h-1 for vaping, and 6.9-7.8 h-1 for vaping with a door and two windows opened and HVAC fan running. Experiments in each outdoor graph -4) were conducted consecutively with 5 min intervals to minimize the outdoor weather variation. To avoid negative values for the log scale concentrations, the background concentrations were included in these 1-s PM2.5 concentration frequency distributions. Plotting a cumulative frequency distribution on the log-probability graph, one can visualize the frequency of exceeding any given concentration limit. Taking figure 4 as an example, 10% of the concentrations exceeded 1000 g/m3 at 2 m from the source. The frequency increased to ~40% at 1 m and decreased to 0% at 3 m. For the same frequency of exceedance , the concentration limit increased to ~4000 g/m3 at 1 m and decreased to ~150 g/m3 at 3 m. Compared to indoor smoking, the frequency distributions for indoor vaping showed much greater separation at the 3 distances. For example, from 1 to 3 m distance, the frequency of exceeding 1000 mg/m3 dropped ~40% for indoor vaping but only ~10% for indoor smoking. The more noticeable decrease in the frequencies for vaping again could be associated with the longer travel time and the higher decay rate compared to smoking. Turning on the mechanical ventilation system flattened the cumulative frequency distribution at each distance for the middle range of concentrations .